Gene Therapy & Molecular Biology Volume 2 Issue A by Niko Boulikas

GENE THERAPY AND MOLECULAR BIOLOGY: FROM BASIC MECHANISMS TO CLINICAL APPLICATIONS

 Volume 2. August 7, 1998 536 pages, color figures

Page

Table of contents

Donald S. Anson (1998) Gene therapy for the mucopolysaccharidoses (review). Gene Ther Mol Biol Vol 2, 1-13.

Joseph R. Kelley and David J. Cole (1998) Gene therapy strategies utilizing carcinoembryonic antigen as a tumor associated antigen for vaccination against solid malignancies. Gene Ther Mol Biol Vol 2, 14-30.

William R. Robinson, Jan Adams, April O'Quinn, and Scott M. Freeman (1998) Vaccine therapy for ovarian cancer using Herpes Simplex virus thymidine kinase (HSV-TK) suicide gene transfer technique: a phase I trial. Gene Ther Mol Biol Vol 2, 31-40.

Thomas A. Gardner, Song-Chu Ko, Chinghai Kao, Toshiro Shirakawa, Jun Cheon, Akinobu Gotoh, Tony T. Wu, Robert A. Sikes, Haiyen E. Zhau, Quajun Cui, Gary Balian and Leland W. K. Chung (1998) Exploiting stromal-epithelial interaction for model development and new strategies of gene therapy for prostate cancer and osteosarcoma metastases (review). Gene Ther Mol Biol Vol 2, 41-58.

Dexi Liu and Young K. Song (1998) Cationic liposome-mediated transfection in vivo (review). Gene Ther Mol Biol Vol 2, 59-68.

Jun Ohkawa, Noriko Yuyama, Shiori Koseki, Yutaka Takebe, Matthias Homann, Georg Sczakiel and Kazunari Taira (1998) The ability of tRNA-embedded ribozymes to prevent replication of HIV-1 in cell culture. Gene Ther Mol Biol Vol 2, 69-82.

Masaki Warashina, Tomoko Kuwabara, Yuka Nakamatsu, Masayuki Sano, Atsushi Shibata, Hideki Shizuku, Hideyuki Takeda, Ryuji Utsunomiya, Jing-Min Zhou, Tadafumi Uchimaru, Jun Ohkawa and Kazunari Taira (1998) Ribozyme-catalyzed trimming reactions and the direct role of Mg2+ ions in the cleavage of RNA. Gene Ther Mol Biol Vol 2, 83-94.

David S. Latchman (1998) Regulation of neuronal differentiation and apoptosis by Brn-3 POU family transcription factors: -potential use in gene therapy (review). Gene Ther Mol Biol Vol 2, 95-101.

103

Scott A. Crist, Sumathi Krishnan, Seong-Su Han, Aysegul Nalca, and Vivek M. Rangnekar (1998) EGR-1 prevents growth arrest by induction of c-myc. Gene Ther Mol Biol Vol 2, 103-117.

118

Abstracts, biographical sketches and presentations of the invited speakers at the Conference on Transcription/ Replication Factors and Protein Phosphorylation (August 7-14, Crete, Greece) and Gene Therapy (August 16-24, Crete, Greece). Photos of article contributors and speakers of the conferences

GENE THERAPY AND MOLECULAR BIOLOGY: FROM BASIC MECHANISMS TO CLINICAL APPLICATIONS

 Volume 2. August 7, 1998 536 pages, color figures

Page

Table of contents

Donald S. Anson (1998) Gene therapy for the mucopolysaccharidoses (review). Gene Ther Mol Biol Vol 2, 1-13.

Dexi Liu and Young K. Song (1998) Cationic liposome-mediated transfection in vivo (review). Gene Ther Mol Biol Vol 2, 59-68.

David S. Latchman (1998) Regulation of neuronal differentiation and apoptosis by Brn-3 POU family transcription factors: -potential use in gene therapy (review). Gene Ther Mol Biol Vol 2, 95-101.

103

Scott A. Crist, Sumathi Krishnan, Seong-Su Han, Aysegul Nalca, and Vivek M. Rangnekar (1998) EGR-1 prevents growth arrest by induction of c-myc. Gene Ther Mol Biol Vol 2, 103-117.

118

Gene Ther Mol Biol Vol 2, 1-13. August 7, 1998.

Gene therapy for the mucopolysaccharidoses (review) Donald S. Anson Department of Chemical Pathology, Women’s and Children’s Hospital, 72 King William Road, North Adelaide, South Australia 5006.

________________________________________________________________________ ______________________________ Correspondence: Donald S. Anson Tel: (61-8) 8204-6373, Fax: (61-8) 8204-7100, E-mail: danson@medicine.adelaide.edu.au

Received 6 April 1998; accepted 17 April 1998 Summary The mucopolysaccharidoses (MPS) are a group of lysosomal storage disorders in which the storage material is glycosaminoglycan. Each MPS is caused by the genetic deciency of a single lysosomal enzyme. Due to the nature of these diseases and the characteristics of the enzymes that are decient most of the MPS are good candidates for gene therapy. Studies in animal models have supported this contention and have shown that several different approaches to gene therapy for the MPS are possible. However, it is also clear that each of these approaches is limited by the currently available technology and that the development of new gene delivery technology is a priority.

I. Introduction The lysosome is a small sub-cellular organelle which is one of the primary sites for the degradation of molecules such as proteins, nucleic acid, mucopolysaccharide and lipids. At least 40 different lysosomal hydrolases that are involved in these processes are known. The protein constituents of the lysosome are synthesised in the endoplasmic reticulum and trafc to the lysosome via the Golgi apparatus. For the soluble lysosomal enzymes trafcking is mediated by the presence of mannose-6-phosphate residues on the enzyme which are specically recognised by the mannose-6-phosphate receptor resulting in targeting to endosomes and then lysosomes (Kornfeld and Mellman, 1989). The acidication of the endosome results in the release and subsequent recycling of the receptor. Other targeting mechanisms also exist, for example the LAMPs (lysosome associated membrane proteins) are targeted via a carboxy terminal tyrosine/glycine motif (Williams and Fukuda, 1990). Extracellular substrates for degradation are delivered to the lysosome via the endocytic pathway while intracellular substrates are delivered via autophagic vacuoles. A large number of genetic diseases have been identied that are caused by deciencies of any one of a number of specic lysosomal enzymes which results in the inability of the lysosome to degrade the substrate normally turned over by that enzyme. This results in the lysosomal accumulation of the undegraded substrate and the clinical development of a lysosomal storage disease. Examples of well known lysosomal storage diseases include Gaucher’s disease (deciency of glucocerebrosidase with resultant storage of glucocerebroside) and Tay-Sachs disease (deciency of hexosaminidase A with resultant storage of GM2 ganglioside). Lysosomal storage diseases that result from deciency of a lysosomal transporter or other proteins that have an indirect effect on lysosomal enzyme activity are also known. Examples of the former include cystinosis, sialic acid storage disease and mucopolysaccharidosis type IIIC. Examples of the latter include sphingolipid activator protein deciencies, I cell disease and multiple sulphatase deciency. The mucopolysaccharidoses (MPS) (Neufeld and Muenzer, 1995) are a group of lysosomal storage disorders (LSD) in which the material that is stored and excreted is glycosaminoglycan (GAG). There are ten known MPS disorders each of which corresponds to a unique single enzyme deciency (Table 1). Each enzyme deciency results in the storage and urinary excretion of one or more GAG types due to the obligatory exolytic nature of the enzymes involved in GAG degradation. In MPS types I and II both dermatan sulphate and heparan sulphate are stored and excreted, in MPS IIIA, IIIB, IIIC and IIID only heparan sulphate is involved, in MPS IVA keratan sulphate and chondroitin-6-sulphate, in MPS IVB only keratan sulphate, in MPS VI only dermatan sulphate (it is thought that chondroitin-4-sulphate, which is also a substrate for NMPSClinical syndromeEnzymeGAG Stored/SecretedGene Isolation (Reference)MPS IHurler/Scheie-LiduronidaseDermatan sulphate, heparan sulphateYes (Scott et al., 1991) MPS IIHunterIduronate-2sulphataseDermatan sulphate, heparan sulphateYes (Wilson et al., 1990) MPS IIIASanlippoSulphamidaseHeparan sulphateYes (Scott et al., 1995) MPS IIIBSanlippoNacetylglucosaminidaseHeparan sulphateYes (Weber et al., 1996)MPS IIICSanlippoAcetyl CoA: -glucosaminideN-acetyltransferaseHeparan sulphateNoMPS IIIDSanlippoN-acetylglucosamine-6-sulphataseHeparan sulphateYes (Robertson et al., 1988)MPS IVAMorquioGalactose-6-sulphataseKeratin sulphate, chondroitin-6sulphateYes (Tomatsu et al., 1991)MPS IVBMorquio-D-galactosidaseKeratin sulphateYes (Morreau et al., 1989)MPS VIMaroteaux-LamyN-acetylgalactosamine-4-sulphataseDermatan sulphate, chondroitin-4-

sulphateYes (Peters et al., 1990)MPS VIISly-D-glucuronidaseDermatan sulphate, heparan sulphate, chondroitin sulphateYes (Guise et al., 1985)Table 1.

acetylgalactosamine-4-sulphatase, the enzyme decient in MPS VI, may be degraded by the action of an alternate enzyme, hyaluronidase) and in MPS VII chondroitin sulphate, heparan sulphate and dermatan sulphate are all stored. All of the MPS with the exception of MPS II are autosomal recessive disorders. MPS II is an X-linked recessive disorder. The genes involved in all of the MPS, with the exception of the gene encoding the Acetyl CoA: -glucosaminide-N-acetyltransferase which is decient in MPS IIIC, have been successfully isolated (Table 1) providing the basic raw material needed for gene therapy for these disorders. All of the MPS are progressive with many affected children appearing normal at birth. Severe cases are usually diagnosed within the rst year or two of life. The clinical symptoms of the MPS vary but generally include several of the following, hepatosplenomegaly, skeletal changes (dystosis multiplex), stiff joints, corneal clouding, hirsutism, respiratory and cardiovascular dysfunction and central nervous system degeneration (Neufeld and Muenzer, 1995). The clinical phenotype of an individual patient with MPS is largely determined by the nature of the storage material and the severity of the enzymatic deciency. For example MPS types IVA, IVB and VI in which the storage material is either keratan sulphate (IVA and IVB) or dermatan sulphate (VI) do not develop CNS pathology or associated symptoms while MPS types IIIA, B, C and D, in which only heparan sulphate is stored, have severe CNS disease and relatively mild somatic features. The tissues affected in each disease can be linked to the type and amount of GAG normally synthesised in the tissue. Established bone pathology and CNS deterioration are usually considered to be irreversible while much soft tissue pathology can be reversed. The most common causes of death in the MPS are due to cardiovascular and respiratory disease. Death normally occurs by or during the teenage years although affected individuals with “mild” disease may, in exceptional cases, live a near-normal life span. A generally applicable treatment for the MPS must therefore be able to deliver replacement enzyme to a large number of sites in the body. On a cellular basis this is a reasonable proposition as the well characterised mannose-6-phosphate lysosomal targeting signal (Kornfeld and Mellman, 1989) and the complementary mannose-6-phosphate receptor provide an efcient mechanism for endocytosis, and the subsequent lysosomal targeting, of enzyme from the peripheral circulation (for example enzyme administered by intravenous injection or enzyme secreted from gene corrected cells after gene therapy). This mechanism provides the theoretical basis for enzyme replacement and gene replacement therapies and has now been well documented experimentally using recombinant sources of enzyme (Oshima et al., 1990; Anson et al., 1992b; Bielicki et al., 1993; Unger et al., 1994; Bielicki et al., 1995; Islam et al., 1993). However some of the cells that need to be treated are only poorly exposed to the peripheral circulation. The prime example is the cells of the central nervous system which are separated from the circulation by the blood brain barrier. Another important example is the cells of the bone growth plate (such as chondrocytes) which are only poorly exposed to the circulation and show high levels of storage in many of the MPS. Enzyme replacement studies have shown that enzyme injected intravenously is very rapidly removed from the circulation with the major proportion being taken up by the liver. Other organs that receive relatively large amounts of enzyme are the spleen, kidney and lung (Sands et al., 1994; Kakkis et al., 1996; Crawley et al., 1997). Dose response studies in the MPS VI cat have demonstrated that the administration of large doses of enzyme from birth are required to have a major effect on moderating the development of

skeletal pathology (Crawley et al., 1997). The amount of administered enzyme reaching the CNS in such studies is very small. The experience with enzyme replacement studies suggests that gene therapy strategies that simply deliver enzyme to the peripheral circulation will not be a completely effective or generally applicable technology for treatment of the MPS. However, such strategies can theoretically be used to treat those MPS in which there is no CNS involvement such as types IVA, IVB and VI, or in the less severe forms of some of the other types in which CNS involvement is also not apparent, for example the Scheie form of MPS type I or mild forms of MPS type II. Most of these strategies are based on the genetic modication of cells that can be easily isolated, cultured and reimplanted such as broblasts and myoblasts. Of these most progress has been made with the use of broblasts.

II. Fibroblast mediated gene therapy for the MPS Fibroblasts can be readily isolated from skin biopsies and grow well in vitro for a limited number of generations before senescing (the so-called Hayick limit). The number of generations that a cell will grow for will vary from individual to individual but is usually between ten and twenty ve (Hayick and Moorhead, 1961). During in vitro culture broblasts are amenable to genetic modication with retroviral or plasmid expression vectors (Veelken et al., 1994; Elder et al., 1997). Retroviral vectors with reasonable titres can be used to effectively transduce 100% of the cells in a primary broblast culture and vectors are available that result in high levels of expression (Miller and Rosman, 1989; Hantzopoulos et al., 1989). Many studies have demonstrated successful in vitro genetic correction of broblasts from MPS patients and the ability of enzyme secreted by gene corrected cells to cross-correct (unmodied) MPS cells (Wolfe et al., 1990; Anson et al., 1992a; Bielicki et al., 1996; Taylor and Wolfe, 1994; Braun et al., 1993). These results have provided a foundation for attempts to reimplant genetically modied broblasts such that they serve as “enzyme factories” in vivo. Most of these studies have been done in laboratory mice, either in naturally occurring mouse models of MPS such as the MPS VII (gusmps) mouse (Moullier et al., 1993) or in nude mice (Salvetti et al., 1995) in which expression of a human protein is tolerated and can be followed with specic reagents (such as monoclonal antibodies). More recently knockout mouse models for MPS I (Clarke et al., 1997) and VI (Evers et al., 1996) have also been generated. These models appear to accurately reect much of the pathology of the corresponding human conditions and can provide a good basis for evaluating forms of therapy although the short time span over which symptoms develop in the mouse (which can be related to the short maturation time of mice) can be a severe limitation when testing some gene therapy protocols. A good example of this is the observation that the skeletal pathology found in the MPS VII mouse develops over the same time frame (approximately 4-6 weeks) as that required for bone marrow engraftment and haematopoietic repopulation; it is therefore impossible to analyse the effect that bone marrow transplantation, or any post-natal gene therapy procedure aimed at the PHSC, has on development of bone pathology as the pathology is already established by the point at which haematopoietic repopulation is complete. Gene transduced broblasts expressing -glucuronidase or -L-iduronidase which have then been reimplanted in the form of neo-organs have been evaluated in the MPS VII (Moullier

et al., 1993), and nude (Salvetti et al., 1995) mice, respectively. The neo-organ is formed by incorporation of the broblasts into a collagen gel containing PTFE bres as a structural matrix. After implantation into the peritoneal cavity the neo-organ becomes vascularised enabling secreted enzyme to enter the circulation. An alternative method of implantation is to deposit a cell mass under the renal capsule (Heartlein et al., 1994). The results of the neoorgan experiments in mice using gene vectors expressing lysosomal enzymes have shown that only low levels of enzyme result from the treatment with most enzyme accumulating in the liver and spleen. In the MPS VII mouse the levels of enzyme reached levels likely to be therapeutic only in these two tissues (Moullier et al., 1993). Expression of -glucuronidase from neo organs has also been evaluated in (normal) dogs. Implantation of one to six neo organs resulted in low levels of enzyme in the liver for at least 340 days (Moullier et al., 1995). One major limitation of the neo organ technology therefore appears to be the low levels of enzyme synthesised by the gene-corrected broblasts after re-implantation. In addition it is now also clear that the reimplanted broblasts have a limited in vivo life span (Kruger et al., 1997), probably due to apoptosis of the implanted cells, and that this is most likely responsible for a signicant proportion of the decline in expression levels seen over time after implantation of gene-corrected broblasts (Moullier et al., 1993, Scharfmann et al., 1991; Hoeben et al., 1993; Naffakh et al., 1995). This suggests that the technological imperative is to develop vectors which direct much higher levels of stable expression in vivo and to modify the apoptotic response of the cells, perhaps by over-expression of genes known to inhibit apoptosis (Kruger et al., 1997). In addition it would be helpful to evaluate this technology in more realistic larger animal models of the MPS and lysosomal storage disorders, such as some of the available cat (Haskins et al., 1992), dog (Schuchmann et al., 1989; Occhiodoro and Anson, 1996; Stoltzfus et al., 1992; Kaye et al., 1992) and caprine (Pearce et al., 1990; Thompson et al., 1992) examples that have been described.

III. Myoblast-mediated gene therapy for the MPS Myoblasts are also considered as potential candidates for gene correction (Blau and Springer, 1995; Miller and Boyce, 1995). In this instance both ex vivo and in situ approaches to gene transduction into myoblasts have been considered (Salvatori et al., 1993; Sajjadi et al., 1994) although neither has been systematically tested with regard to the MPS or any other of the lysosomal storage diseases. There is a single study of myoblast mediated gene therapy, using an ex-vivo approach, in the MPS I dog (Shull et al., 1996). The muscle is a well vascularised tissue, is very metabolically active, and so appears well suited to the synthesis of large amounts of gene product. In addition a large mass of muscle is available for gene correction. The biology of the muscle therefore appears compatible with the requirements for its use as a target for MPS gene therapy in an analogous approach to that outlined above for broblast-mediated gene therapy. However, the ex vivo approach suffers from the poor efciency with which myoblasts can be reimplanted into muscle while the in situ transduction requires vectors that are able to transduce nonreplicating cells. Recent data suggests that recombinant adeno-associated virus vectors are promising vehicles for efcient gene transfer into myoblasts in situ (Fisher et al., 1997). The one study of myoblast-mediated gene therapy in the MPS I dog (Shull et al., 1996) used an ex vivo approach in which cultured myoblasts were transduced with retrovirus carrying the -L-iduronidase gene. Unfortunately the re-implantation of the gene corrected

cells generated an immune response against -L-iduronidase (the MPS I dog carries a null mutation (Stoltzfus et al., 1992) which was correlated with a rapid decrease in the levels of enzyme and the number of myoblasts containing the -L-iduronidase gene. The general applicability of myoblast-mediated gene therapy for the MPS needs to evaluated further especially as there now seems to be a vector, AAV, that is able to efciently transduce myoblasts in situ (Fisher et al., 1997). It is possible that the myoblast may well be more suited for long-term expression of introduced genes than broblasts, especially if it is re-implanted into, or left in, its natural environment, the muscle bre. This, and the use of muscle specic promoter elements, may allow myoblast mediated gene therapy to avoid the problems of down regulation of expression and apoptosis of transduced cells associated with broblast mediated gene therapy (see above).

IV. Bone marrow stem cell mediated gene therapy for the MPS The long established use of bone marrow transplantation both in animal studies and in clinical treatments has demonstrated that in general terms the bone marrow is perhaps the ideal cell population for gene therapy. The biology of the haematopoietic system, in which a small population of very primitive haematopoietic stem cells (HSC) are used as the basis for the continuous generation of extremely large numbers of the variety of mature cells found in the periphery, means that by targeting this small stem cell pool for gene transfer a large population of gene-corrected cells can be continuously generated for the lifetime of the individual. More specically bone marrow transplantation studies in animal models of (Taylor et al., 1992; Birkenmeier et al., 1991; Hoogerbrugge et al., 1988a; Breider et al., 1989; Gasper et al., 1984; Walkley et al., 1994) and patients with (Cowan, 1991; Hopwood et al., 1993; Hoogerbrugge et al., 1995) lysosomal storage disorders provide strong evidence that gene correction of the haematopoietic system is likely to be a viable approach for gene therapy of the LSD in general and the MPS in particular (Walkley et al., 1996). Ideally the development of a screening system for affected newborns (Meikle et al., 1997; Sweetman, 1996) combined with cord blood banking (Broxmeyer, 1995) would provide an opportunity to effect treatment before the development of clinical symptoms. Because of the irreversible nature of the skeletal and CNS pathology in the MPS, treating affected individuals presenting with clinical symptoms is always going to be less than 100% effective. In addition it is clear that the endogenous levels of lysosomal enzymes synthesised and secreted by the haematopoietic system are generally not high enough to be completely corrective, gene therapy therefore must be optimised in terms of expression levels if it is to offer improved efcacy over bone marrow transplantation. One of the animal models that has provided evidence for the potential efcacy of bone marrow transplantation and hence HSC mediated gene therapy is the fucosidosis dog. It is this animal model that we are using, in collaboration with

Figure 1. Effect of bone marrow transplantation on the development of canine fucosidosis. The neurological disability score (y axis) is a quantitative assessment score for the development of the clinical disease associated with deď&#x192;&#x17E;ciency of -L-fucosidase in the English Springer Spaniel. Each line represents an animal that received a transplant of normal allogeneic bone marrow at the age (in months) indicated. The results clearly show that bone marrow transplantation at 3.6 months is almost completely effective at preventing disease progression. Untreated animals usually require euthanisation at approximately three years of age.

the group at Westmead hospital (NSW, Australia), to evaluate this approach to gene therapy with special reference to the treatment of central nervous system pathology. Canine fucosidosis results from a 14 bp deletion at the end of the rst exon of the gene encoding L-fucosidase which in turn results in a frameshift in the reading frame and a truncated protein (Occhiodoro and Anson, 1996). The clinical course of the disease is a progressive central nervous system deterioration manifesting rst as mild hypermetria and ataxia then more overt and pronounced stance and gait defects and nally a severe mental and motor deterioration (Taylor et al., 1987; Taylor and Farrow, 1988). Euthanasia is normally required before approximately 40 months of age. Roseanne Taylor’s studies of allogeneic bone marrow transplantation in the fucosidosis dog (Taylor et al., 1986; Taylor et al., 1988; Taylor et al., 1992) have convincingly demonstrated that this procedure results in signicant enzyme replacement in a wide variety of tissues, including the central nervous system. Enzyme levels of up to approximately 50% of normal result in peripheral tissues and levels of up to approximately 25% of normal in the CNS. The accumulation of enzyme activity in the CNS is signicantly slower than in peripheral tissues. This is thought to be a reection of the slow accumulation of donor derived cells of haematopoietic origin in the CNS (see below). Assessment of the clinical course of the disease shows that if the bone marrow transplant is done before the development of signicant pathology it also has a profound effect on the clinical progression of the disease as measured by an objective score of neurological disability. Animals receiving transplants at 3-4 months are almost completely normalised both in terms of the clinical progression of the disease and in terms of lifespan. In contrast bone marrow at later stages where overt disease pathology was already apparent had little effect on the continuing course of the disease (Fig. 1). In the fucosidosis dog we therefore have an animal model that demonstrates that gene correction of the pluripotent haematopoietic stem cell can result in correction of the central nervous system disease associated with the MPS. It is therefore an ideal model for developing and testing (autologous) stem cell mediated gene therapy which aims to reproduce and improve on the results from allogeneic bone marrow transplantation. It tells us when we need to do the procedure, how much enzyme we need the cells to make and what results we may expect from a positive experiment. Other animal models of lysosomal storage diseases have also provided evidence for the potential efcacy of bone marrow transplantation and have provided some evidence for the mechanism by which bone marrow transplantation results in enzyme replacement in the CNS. In the -mannosidosis cat bone marrow transplantation also halts disease progression and results in clearance of storage material from neuronal tissue (Walkley et al., 1994). In this case bone marrow transplantation was done at an age (8 to 12 weeks) by which time overt clinical symptoms are already apparent. Despite this the procedure appeared to be clinically efcacious. Histological staining for -mannosidase activity revealed the presence of enzyme positive neurones, glial cells and cells associated with blood vessels. Bone marrow transplantation studies in the twitcher (galactosylceramidase deciency/Krabbe’s disease) mouse has provided similar results, BMT results in enzyme replacement in the CNS and decrease of the levels of stored substrate, is partially effective in preventing clinical disease, triples the life span and donor marrow derived cells can be detected in the CNS of transplanted animals (Hoogerbrugge et al., 1988a, b). These were identied as glial cells. Transplantation of bone marrow marked with a retroviral vector containing the human glucocerebrosidase gene has conrmed the penetration of the CNS by bone marrow

derived cells in mice. In this instance immunohistochemical analysis using a monoclonal antibody to human glucocerebrosidase revealed the presence of perivascular and parenchymal microglia of donor origin (Krall et al., 1994). As this study was done in normal mice it suggests that colonisation of the CNS by donor derived cells is not related to the presence of a pathological state, however it is possible that it is related to the effects of the bone marrow transplantation procedure itself. Experiments in animals receiving no radiation or chemical pre-conditioning could resolve this point. Clinical trials of allogeneic bone marrow transplantation in patients suffering from various of the MPS are also somewhat encouraging although with important caveats. Bone marrow transplantation in patients with MPS I for instance, clearly moderates the progression of the disease (Hopwood et al., 1993, Hoogerbrugge et al., 1995) but is equally clearly not curative, especially for skeletal problems, the development of which seem to be little affected by the procedure. Similarly bone marrow transplantation in MPS VI will moderate some of the soft tissue pathology but has no discernible affect on the skeletal pathology (Hoogerbrugge et al., 1995). The success of bone marrow transplantation in the fucosidosis dog has led to this procedure being trialed clinically (Vellodi et al., 1995). Bone marrow transplantation is curative for non-neurological Gaucher’s disease (Ringden et al., 1995, Hoogerbrugge et al., 1995) but in this instance it should be noted that the cells responsible for symptomatology, macrophages, are a haematopoietic lineage. Taken together the results obtained in animal models and clinical trials of bone marrow transplantation clearly show that targeting of the PHSC for gene transfer is likely to be an effective way of achieving enzyme replacement in the CNS in lysosomal storage diseases including the MPS. However, it should also be noted that certain of the lysosomal storage diseases, for example Batten’s disease (Lake et al., 1995) and GM1 gangliosidosis (O’Brien et al., 1990), do not appear to respond to bone marrow transplantation. In the case of Batten’s disease this may be due to the physical nature of the protein involved which may be membrane bound (The International Batten Disease Consortium, 1995) preventing secretion and hence cross-correction of other cells. Other possible reasons for lack of clinical efcacy include low levels of enzyme secretion by haematopoietic cells colonising the CNS or the existence of pre-existing storage and/or pathological damage that cannot be reversed. The reasons for the variability in response of LSD and MPS to BMT need to be studied further and may help dene when this approach to gene therapy will be appropriate. The only convincing HSC mediated gene therapy experiment relevant to the LSD has been done in the MPS VII mouse (Wolfe et al., 1992). In the MPS VII mouse retroviral mediated transfer of the -glucuronidase gene into HSC resulted in signicant enzyme replacement in one animal in bone marrow (26% of normal), spleen and lymph node (6%), thymus and liver (2%) and lung and liver. This resulted in reduced storage in tissues where enzyme was detected. A second animal analysed had lower levels of enzyme and enzyme was only detectable in the bone marrow and spleen (<1% of normal). Experiments addressing HSC transduction in larger animal models and humans have been frustrated by the very low levels of long term HSC transduction obtained (Bienzle et al., 1994; Kiem et al., 1995; Donahue et al., 1996; Xu et al., 1995; van-Beusechem et al., 1995; vanBeusechem and Valerio, 1996; Hanania et al., 1996) indicating the presence of technical limitations in the vector systems used. The low level of transduction of long term

repopulating HSC seen in non-murine animals and in humans appears to result from a basic incompatibility between the vector system used (murine leukemia virus (MLV) based retroviral vector systems) and the biology of primitive HSC. MLV and gene transfer systems based on MLV have been shown to be incapable of infecting/transducing noncycling cell populations (Roe et al., 1993, Miller et al., 1990). In addition it has been convincingly shown that the most primitive HSC populations are extremely quiescent, both in vivo and in vitro (Stewart et al., 1993; Hao et al., 1995; Berardi et al., 1995; Hao et al., 1996). Extensive attempts to overcome this incompatibility between MLV vectors and the biology of the primitive HSC by stimulating the latter with cytokines to induce cell cycling have been made (Donahue et al., 1996; Xu et al., 1995; van-Beusechem et al., 1995; vanBeusechem and Valerio, 1996; Hanania et al., 1996). However, the results of this approach to achieving high levels of gene transfer have been disappointing. Long term haematopoietic cell culture systems have also been used as the basis for transduction protocols (Bienzle et al., 1994; Kiem et al., 1995) and while appear in some instances (Bienzle et al., 1994) to give somewhat better results have not convincingly resulted in efcient transduction of the most primitive HSC. In summary HSC mediated gene therapy appears likely to be a viable approach for the treatment of the MPS but must aim to improve on bone marrow transplantation in terms of the degree of enzyme replacement afforded by the procedure so enzyme replacement is both more complete and more rapid. This will depend on the development of an efcient transduction system for the long term repopulating HSC. In addition it is likely that hyperexpressing vectors will need to be developed. Given that only qualied success has been achieved with bone marrow transplantation in MPS patients we feel that gene therapy procedures must, if at all possible, be evaluated in relevant animal models before clinical use is considered. Our rst attempts to develop stem cell mediated gene therapy in the fucosidosis dog have been based on the use of retroviral vector systems derived from MLV. In an attempt to improve on the generally disappointing results of others in regard to transducing the canine PHSC we used multiple cytokines to stimulate the bone marrow during transduction. These experiments were unsuccessful and will not be discussed here. However, we like many others have come to the conclusion that for this approach to gene therapy to succeed several challenges have to be faced in terms of the gene transfer technology used. It is apparent that for successful stem cell gene therapy we need a vector with certain features including: i. An ability to efciently transduce non-cycling cell populations (the stem cell pool is generally regarded as quiescent). ii. An ability to stably integrate into the host cell genome (so that the transduced gene is retained as the stem cell population divides and expands to produce large numbers of differentiated cells). iii. If transduction is receptor-mediated the vector particle must recognise a receptor expressed at reasonably high levels on stem cells. iv. High level and stable expression of the transduced gene. v. If the vector is viral-based it must be completely defective, of high titre, and demonstrably safe. In my view these criteria can only be met by a viral vector, of the known viruses retroviruses and adeno-associated virus appear to be the obvious contenders. At present it is unclear if adeno-associated virus can be developed into an efcient, helper free

integrating vector system for the PHSC while MLV based retroviral vectors are unable to effectively transduce quiescent cell populations. However in almost all other regards retroviral systems appear to be well suited for transduction of the haematopoietic stem cell. Integration of the virus into the host cell chromosome is a normal part of the viral life cycle, retroviruses appear to be easy to pseudotype with envelopes not only from other retroviruses (Eglitis et al., 1988; Wilson et al., 1989) but also from other viral groups (Ory et al., 1996) suggesting that if a suitable envelope/receptor combination can be found it can be adapted to a retroviral system. In addition the development of third generation retroviral packaging systems have shown that it is possible to produce recombinant retrovirus with a very small probability of contamination with replication competent virus (Markowitz et al., 1988; Cosset et al., 1995). This leaves two problems, the inability of Murine Leukemia virus based vectors to transduce quiescent cell populations (Miller et al., 1990) and the question of what levels of expression can be achieved by retroviral constructs in vivo in haematopoietic lineages. The rst of these appears to be a more fundamental question and in some ways the second is undened until the rst can be overcome, I will therefore disregard it for the present. Is there an answer to the conundrum presented by the incompatible biology of the HSC and MLV? Attempts to manipulate HSC into cycle to facilitate transduction with MLV based vectors have been largely unsuccessful (outside of the laboratory mouse) (Donahue et al., 1996; Xu et al., 1995; van-Beusechem et al., 1995; van-Beusechem and Valerio, 1996; Hanania et al., 1996), there is little effect on transduction efciency either due to inefcient recruitment of the most primitive HSC into cycle and/or due to concomitant induction of cell cycling and differentiation. In vitro studies of human haematopoietic progenitor cells suggest that the most primitive cells are only very slowly recruited into the cell cycle even when cultured with extremely potent combinations of cytokines (Hao et al., 1996). At present this avenue would therefore appear to be effectively closed although new developments in the understanding of stem cell biology and regulation may one day make it viable. Very recently it has been shown that a combination of Flk3 ligand and thrombopoietin appears to support the amplication and self renewal of very primitive haematopoietic progenitors in cord blood (Piacibello et al., 1997) although it is not clear if this includes the true long term repopulating HSC. However this culture system would provide an ideal milieu for efcient retroviral transduction. The other avenue that is being explored by an increasing number of groups including ourselves is to look for alternative retroviruses that do not have the same limitations as Murine Leukemia virus. Members of the lentivirus family appear to be able to infect at least some populations of quiescent cells, a prime example of this is the ability of human immunodeciency virus (HIV) to infect terminally differentiated macrophages in vivo (Koenig et al., 1986). In vitro studies have extended this observation (Lewis et al., 1992) and shown that the ability of HIV to infect non-cycling cell populations is mechanistically linked to nuclear localisation sequences in the matrix protein (Bukrinsky et al., 1993) and Vpr (Heinzinger et al., 1994). Several groups have demonstrated the feasibility of making recombinant viral vectors from HIV and shown that the recombinant virus produced has the same ability to infect non-cycling cells (Reiser et al., 1996; Naldini et al., 1996; Poeschla et al., 1996). It remains to be seen whether safety issues associated with the use of HIV as a gene vector can be convincingly addressed and whether such vectors are more effective in transducing the haematopoietic stem cell pool or if other factors are limiting.

However it is clear that alternatives for vector development do exist and need to be carefully evaluated in relevant animal models and pre-clinical studies. In addition other important factors, such as the distribution of different envelope receptors and mechanisms of gene regulation in vectors need to be investigated at both the basic research and applied levels. Large outbred animal models of human disease will be invaluable in dening the important parameters involved in making this approach to gene therapy a success and in proving its viability as a clinical treatment.

V. Other approaches and future directions A. Direct treatment of the CNS in the mucopolysaccharidoses via gene replacement As the CNS is considered one of the more intractable sites of pathology in the CNS some consideration has been given to developing therapeutic strategies aimed narrowly at correcting CNS pathology. Two approaches have been considered:- introduction of (genetically modied) cells into the CNS or direct gene modication of CNS resident cells. Both appear to be applicable to the treatment of the MPS, however although much generic work has been done on the development of gene transfer vectors suitable for direct transduction of the CNS there is no published work on the use of these systems with MPS genes or in MPS animal models. Transplantation of an immortalised neural progenitor cell line expressing -glucuronidase into the brains of MPS VII mice has been reported and is effective in preventing lysosomal storage (Snyder et al., 1995). Similar results have more recently been demonstrated with broblast implants (Taylor and Wolfe; 1997) although there is clearly a problem with the stability of expression of the transduced gene. B. Prenatal gene therapy It is known that even though in most instances of the MPS there is no overt clinical presentation at birth lysosomal storage already exists and can also be found for a signicant period prenatally. Prenatal gene therapy may therefore be required for completely effective treatment of some severe instances of the MPS. This form of therapy can be approached in several ways. The most immediate technology is that of prenatal allogeneic bone marrow transplantation in the pre-immune phase of foetal development (Cowan and Golbus, 1994, Touraine, 1996) (approximately the rst trimester). Although this is not strictly gene therapy this approach has already been used for the treatment of X-linked severe combined immunodeciency (Flake et al., 1996; Wengler et al., 1996) and could be extended by using allogeneic marrow that has been transduced with a gene vector to boost expression of the required enzyme. The use of neural progenitors to treat the brain (see above) is also applicable to the developing foetus. The other approach is to directly introduce gene vectors into the foetus to allow in vivo gene correction to occur (Clapp et al., 1995). This can either be aimed at the developing haematopoietic system or other tissues.

VII. Conclusion Clinically gene therapy for the MPS has barely reached its infancy. The only ongoing clinical trial at present (July 1997) is for MPS type II in which peripheral T lymphocytes are the target for gene transduction (Whitley et al., 1996). However enzyme replacement studies, bone marrow transplantation in animal models, and in the clinic, and proof of concept experiments for gene correction all suggest that if some clearly dened technical

challenges can be met gene therapy has the potential to be an effective treatment for the MPS. In my view gene transfer into the haematopoietic stem cell is likely to be the most effective and generally applicable approach, unfortunately this approach probably faces the most severe technical challenges. However, there are a number of other approaches that are clearly applicable to many of the MPS disorders that may be easier to develop to the point of clinical applicability. All these approaches need to be further developed and carefully evaluated in animal models before clinical trials proceed. Pre-symptomatic screening of newborns for LSD is also a prerequisite for effective clinical treatment. Screening strategies based on measuring lysosomal markers using ELIZA type assays (Meikle et al., 1997), or storage material by tandem mass spectrometry (Sweetman, 1996), are being developed.

References Anson, D. S., Bielicki, J. and Hopwood, J. J. (1992a). Correction of mucopolysaccharidosis type I broblasts by retroviral-mediated transfer of the human -L-iduronidase gene. Hum. Gene Ther. 3: 371-379. Anson, D. S., Taylor, J. A., Bielicki, J., Harper, G., Peters, C., Gibson, G. J. and Hopwood, J. J. (1992b). Correction of human mucopolysaccharidosis type-VI broblasts with recombinant N-acetylgalactosamine-4-sulphatase. Biochem. J. 284: 789-794. Berardi, A. C., Wang, A., Levine, J. D., Lopez, P. and Scadden, D. T. (1995). Functional isolation and characterisation of human hematopoietic stem cells. Science 267: 104-108. Bielicki, J., Fuller, M., Guo, X-G., Morris, C. P., Hopwood, J. J. and Anson, D. S. (1995). Expression, purication and characterization of recombinant human N-acetylgalactosamine-6-sulphatase. Biochem. J. 311: 333-339. Bielicki, J., Hopwood, J. J. and Anson, D. S. (1996). Correction of Sanlippo A broblasts by retroviral vectormediated gene transfer. Hum. Gene Ther. 7: 1965-1970. Bielicki, J., Hopwood, J. J., Wilson, P. J. and Anson, D. S. (1993). Recombinant human iduronate-2-sulphatase: correction of mucopolysaccharidosis-type II broblasts and characterization of the puried enzyme. Biochem. J. 289: 241-246. Bienzle, D, Abrams-Ogg, A. C., Kruth, S. A., Ackland-Snow, J, Carter, R. F., Dick, J. E., Jacobs, R. M., Kamel-Reid, S. and Dube, I. D. (1994). Gene transfer into hematopoietic stem cells: long-term maintenance of in vitro activated progenitors without marrow ablation. Proc Natl Acad Sci USA 91: 350-354. Birkenmeier, E. H., Barker, J. E., Vogler, C. A., Kyle, J. W., Sly, W. S., Gwynn, B., Levy, B. and Pegors, C. (1991). Increased life span and correction of metabolic defects in murine mucopolysaccharidosis type VII after syngeneic bone marrow transplantation. Blood 78: 3081-3092. Blau, H. and Springer, M. L. (1995). Molecular Medicine. Muscle mediated gene therapy. New Eng. J. Med. 333: 1554-1556. Braun, S. E., Aronovich, E. L., Anderson, R. A., Crotty, P. L., McIvor, R. S. and Whitley, C. B. (1993). Metabolic correction and cross-correction of mucopolysaccharidosis type II (Hunter syndrome) by retroviral-mediated gene transfer and expression of human-iduronate-2-sulfatase. Proc Natl Acad Sci USA 90: 11830-11834. Breider, M. A., Shull, R. M. and Constantopoulos, G. (1989). Long-term effects of bone marrow transplantation in dogs with mucopolysaccharidosis I. Am. J. Pathol. 134: 677-692. Broxmeyer, H. E. (1995). Cord blood as an alterative source for stem and progenitor cell transplantation. Curr. Opin. Pediatr. 7: 47-55. Bukrinsky, M. I., Haggerty, S., Dempsey, M. P., Sharova, N., Adzhubel, A., Spitz, L., Lewis, P., Goldfarb, D., Emerman, M. and Stevenson, M. (1993). A nuclear localisation signal within HIV-1 matrix protein that governs infection of non-dividing cells. Nature 365: 666-669. Clapp, D. W., Freie, B., Lee, W. H. and Zhang, Y. Y. (1995). Molecular evidence that in situ-transduced fetal liver hematopoietic stem/progenitor cells give rise to medullary hematopoiesis in adult rats. Blood 86: 2113-2122. Clarke, L. A., Russell, C. S., Pownall, S., Warrington, C. L., Borowski, A., Dimmick, J. E., Toone and J., Jirik, F. R. (1997). Murine mucopolysaccharidosis type I-targeted disruption of the murine -L-iduronidase gene. Hum. Mol. Gen. 6: 503-511. Cosset, F-L., Takeuchi, Y., Battini, J-L., Weiss, R. A. and Collins, M. K. L. (1995). High-titer packaging cells producing recombinant retroviruses resistant to human serum. J. Virol. 69: 7430-7436. Cowan, M. J. (1991). Bone marrow transplantation for the treatment of genetic diseases. Clin. Biochem. 24: 375-381.

Cowan, M. J. and Golbus, M. (1994). In utero hematopoietic stem cell transplants for inherited diseases. Am J. Pediatr. Hematol. Oncol. 16: 35-42. Crawley, A. C., Niedzielski, K. H., Isaac, E. L., Davey, R. C. A., Byers, S. and Hopwood, J. J. (1997). Enzyme replacement therapy from birth in a feline model of mucopolysaccharidosis type VI. J. Clin. Invest. 99: 651-662. Donahue, R. E., Byrne, E. R., Thomas, T. E., Kirby, M. A., Agricola, B. A., Sellers, S. E., Gaudernack, G., Karlsson, S. and Lansdorp, P. M. (1996). Transplantation and gene transfer of the human glucocerebrosidase gene into + + immunoselected primate CD34 Thy1 cells. Blood 88: 4166-4172. Eglitis, M. A., Kohn, D. B., Moen, R. C., Blaese, R. M. and Anderson, W. F. (1988). Infection of human hematopoietic progenitor cells using a retroviral vector with a xenotropic pseudotype. Biochem Biophys Res Commun 151: 201-206. Elder, E. M., Lotze, M. T. and Whiteside, T. L. (1997). “Methods for generation of genetically modied broblasts for immunotherapy of cancer” in Methods in Molecular Medicine, Gene Therapy Protocols, P. Robbins, ed. (Humana Press, Totowa, NJ). Evers, M., Saftig, P., Schmidt, P., Hafner, A., McLoghlin, D. B., Schmahl, W., Hess, B., von-Figura, K. and Peters, C. (1996). Targeted disruption of the arylsulfatase B gene results in mice resembling the phenotype of mucopolysaccharidosis VI. Proc Natl Acad Sci USA 93: 8214-8219. Fisher, K. J., Jooss, K., Alston, J., Yang, Y., Haecker, S. E., High, K., Pathak, R., Raper, S. E. and Wilson, J. M. (1997). Recombinant adeno-associated virus for muscle directed gene therapy. Nat. Med. 3: 306-312. Flake, A. W., Roncarolo, M. G., Puck, J. M., Almeida-Porada, G., Evans, M. I., Johnson, M. P., Abella, E. M., Harrison, D. D. and Zanjani, E. D. (1996). Treatment of X-linked severe combined immunodeciency by an in utero transplantation of paternal bone marrow. N. Eng. J. Med. 335: 1806-1810. Gasper, P. W., Thrall, M. A., Wenger, D. A., Macy, D. W., Ham, L., Dornsife, R. E., McBiles, K., Quackenbush, S. L., Kesel, M. L., Gillette, E. L. and Hoover, E. A. (1984). Correction of feline arylsulphatase B deciency (mucopolysaccharidosis VI) by bone marrow transplantation. Nature 312: 467-469. Guise, K. S., Korneluk, R. G., Waye, J., Lamhonwah, A. M., Quan, F., Palmer, R., Ganschow, R. E., Sly, W. S. and Gravel, R. A. (1985). Isolation and expression in Escherichia coli of a cDNA clone encoding human glucuronidase. Gene 34: 105-110. Hanania, E. G. et al. (1996). Results of MDR-1 vector modication trial indicate that granulocyte/macrophage colony-forming unit cells do not contribute to posttransplant hematopoietic recovery following intensive systemic therapy. Proc Natl Acad Sci USA 93: 15346-15351. Hantzopoulos, P. A., Sullenger, B. A., Ungers, G. and Gilboa, E. (1989). Improved gene expression upon transfer of the adenosine deaminase minigene outside the t of a retroviral vector. Proc Natl Acad Sci USA 86: 3519-3523. Hao, Q-L., Shah, A. J., Thiemann, F. T., Smogorzewska, E. M. and Crooks, G. M. (1995). A functional comparison of + CD34 CD38 cells in cord blood and bone marrow. Blood 86: 3745-3753. Hao, Q-L., Thiemann, F. T., Petersen, D., Smogorzewska, E. M. and Crooks, G. M. (1996). Extended long-term culture reveals a highly quiescent and primitive human hematopoietic progenitor population. Blood 88: 3306-3313. Haskins, M. E., Otis, E. J., Hayden, J. E., Jezyk, P. F. and Stramm, L. (1992). Hepatic storage of glycosaminoglycans in feline and canine models of mucopolysaccharidoses I, VI and VII. Vet. Pathol. 29: 112-119. Hayick, L. and Moorhead, P. S. (1961). The serial cultivation of human diploid cell strains. Exp. Cell Res. 25: 585-621. Heartlein, M. W., Roman, V. A., Jiang, J. L., Sellers, J. W., Zuliani, A. M., Treco, D. A. and Selden, R. F. (1 994). Long-term production and delivery of human growth hormone in vivo. Proc Natl Acad Sci USA 91: 10967-10971. Heinzinger, N. K., Bukrinsky, M. I., Haggerty, S. A., Ragland, A. M., Kewalramani, V., Lee, M-A., Gendelman, H. E., Ratner, L., Stevenson, M. and Emerman, M. (1994). The Vpr protein of human immunodeciency virus type 1 inuences nuclear localisation of viral nucleic acids in nondividing host cells. Proc Natl Acad Sci USA 91: 7311-7315. Hoeben, R. C., Fallaux, F. J., van Tilburg, N. H., Cramer, S. J., van Ormondt, H., Briet, E. and van der Eb, A. J. (1993). Toward gene therapy for hemophilia A: Long-term persistence of factor VIII-secreting broblasts after transplantation into immunodecient mice. Hum. Gene Ther. 4: 179-186. Hoogerbrugge, P. M., Brouwer, O. F., Bordigoni, P., Ringden, O., Kapaun, P., Ortega, J. J., O’Meara, A., Cornu, G., Souillet, G., Frappaz., D., Blanche, S. and Fischer, A. (1995). Allogeneic bone marrow transplantation for the lysosomal storage diseases. Lancet 345: 1398-1402. Hoogerbrugge, P. M., Poorthuis, B. J., Romme, A. E., van-de-Kamp, J. J., Wagemaker, G. and van-Bekkum, D. W.

(1988a). Effect of bone marrow transplantation on enzyme levels and clinical course in the neurologically affected twitcher mouse. J. Clin. Invest. 81: 1790-1794. Hoogerbrugge, P. M., Suzuki, K., Suzuki, K., Poorthuis, B. J. H. M., Kobayashi, T., Wagemaker, G. and van Bekkum, D. W. (1988b). Donor-derived cells in the central nervous system of twitcher mice after bone marrow transplantation. Science 239: 1035-1038. Hopwood, J. J., Vellodi, A., Scott, H. S., Morris, C. P., Litjens, T., Clements, P. R., Brooks, D. A., Cooper, A. and Wraith, J. E. (1993). Long-term clinical progress in bone marrow transplanted mucopolysaccharidosis type I patients with a dened genotype. J. Inher. Metab. Dis. 16: 1024-1033. Islam, M. R., Grubb, J. H. and Sly, W. S. (1993). C-terminal processing of human -glucuronidase. The propeptide is required for full expression of catalytic activity, intracellular retention, and proper phosphorylation. J. Biol. Chem. 268: 22627-22633. Kakkis, E. D., McEntee, M. F., Schmidtchen, A., Neufeld, E. F., Ward, D. A., Gompf, R. E., Kania, S., Bedolla, C., Chien, S. L. and Shull, R. M. (1996). Long-term and high-dose trials of enzyme replacement therapy in the canine model of mucopolysaccharidosis I. Biochem. Mol. Med. 58: 156-167. Kaye, E. M., Alroy, J., Raghaven, S. S., Schwarting, G. A., Adelman, L. S., Runge, V., Gelblum, D., Thalhammer, J. G. and Zuniga, G. (1992). Dysmyelinogenesis in animal model of GM1 gangliosidosis. Ped. Neurol. 8: 255-261. Kiem, H.-P., Darovsky, B., von Kalle, C., Goehle, S., Graham, T., Miller, A. D., Storb, R. and Scheuning, F. G. (1995). Long-term persistence of canine hematopoietic cells genetically marked by retrovirus vectors. Hum. Gen. Ther. 7: 89-96. Koenig, S., Gendelman, H., E., Orenstein, J. M., Dal-Canto, M. C., Pezeshkpour, G. H., Yungbluth, M., Janotta, F., Aksamit, A., Martin, M. A. and Fauci, A. S. (1986). Detection of AIDS virus in macrophages in brain tissue from AIDS patients with encephalopathy. Science 233: 1089-1093. Kornfeld, S. and Mellman, I. (1989). The biogenesis of lysosomes. Ann. Rev. Cell Biol. 5: 483-525. Krall., W. J., Challita, P. M., Perlmutter, L. S., Skelton, D. C. and Kohn, D. B. (1994). Cells expressing human glucocerebrosidase from a retroviral vector repopulate macrophages and central nervous system microglia after murine bone marrow transplantation. Blood 83: 2737-2748. Kruger, G. G., Jorgensen, C. M., Petersen, M. J., Mansbridge, J. N. and Morgan, J. R. (1997). Use of cloned genetically modied human broblasts to assess long-term survival in vivo. Hum. Gene Ther. 8: 523-532. Lake, B. D., Henderson, D. C., Oakhill, A. and Vellodi, A. (1995). Bone marrow transplantation in Batten disease (neuronal ceroid-lipofuscinosis). Will it work? Am. J. Med. Gen. 57: 369-373. Lewis, P., Hensel, M. and Emerman, M. (1992). Human immunodeciency virus infection of cells arrested in the cell cycle. EMBO J 11: 3053-3058. Markowitz, D., Goff, S. and Banks, A. (1988). Construction and use of a safe and efcient amphotropic packaging cell line. Virology 167: 400-406. Meikle, P. J., Brooks, D. A., Ravenscroft, E. M., Yan, M., Williams, R. E., Jaunzems, A. E., Chataway, T. K., Karageorgos, L. E., Davey, R. C., Boulter, C. D., Carlsson, S. R. and Hopwood, J. J. (1997). Diagnosis of lysosomal storage disorders: Evaluation of lysosome-associated membrane protein LAMP-1 as a diagnostic marker. Clin. Chem. 43: 1325-1335. Miller, A. D. and Rosman, G. J. (1989). Improved retroviral vectors for gene transfer and expression. BioTechniques 7: 980-990. Miller, D. G., Adam, M. A. and Miller, A. D. (1990). Gene transfer by retrovirus vectors occurs only in cells that are actively replicating at the time of infection. Mol. Cell. Biol. 10: 4239-4242. Miller, J. B. and Boyce, F. M. (1995). Gene therapy by and for muscle cells. Trends Genet. 11: 163-165. Morreau, H., Galjart, N. J., Gillemans, N., Willemsen, R., van der Horst, G. T. and d’Azzo, A. (1989). Alternative splicing of -galactosidase mRNA generates the classic lysosomal enzyme and a -galactosidase related protein. J. Biol. Chem. 264: 20655-20663. Moullier, P., Bohl, D., Cardoso, J., Heard, J. M. and Danos, O. (1995). Long-term delivery of a lysosomal enzyme by genetically modied broblasts in dogs. Nat. Med. 1: 353-357. Moullier, P., Bohl, D., Heard, J-M. and Danos, O. (1993). Correction of lysosomal storage in the liver and spleen of MPS VII mice by implantation of genetically modied skin broblasts. Nature Genet. 4: 154-159. Naffakh, N., Henri, A., Villeval, J. L., Rouyer-Fessard, P., Moullier, P., Blumenfeld, N., Danos, O., Vainchenker, W., Heard, J. M. and Beuzard, Y. (1995). Sustained delivery of erythropoietin in mice by genetically modied skin broblasts. Proc Natl Acad Sci USA 92: 3194-3198. Naldini, L., Blomer, U., Gallay, P., Ory, D., Mulligan, R., Gage, F. H., Verma, I. M. and Trono, D. (1996). In vivo gene delivery and stable transduction of nondividing cells by a lentiviral vector. Science 272: 263-266. Neufeld, E. F. and Muenzer, J. (1995). “The Mucopolysaccharidoses” in The Metabolic and Molecular Bases of th Inherited Disease, 7 ed. C. R. Scriver, A. L. Beaudet, W. S. Sly and D. Valle, eds. (McGraw-Hill, New York)

pp. 2465-2494. O’Brien, J. S., Storb, R., Raff, R. F., Harding, J., Appelbaum, F., Morimoto, S., Kishimoto, Y., Graham, T., AhernRindell, A. and O’Brien, S. L. (1990). Bone marrow transplantation in canine GM1 gangliosidosis. Clin. Gen. 38: 274-280. Occhiodoro, T. and Anson, D. S. (1996). Isolation of the canine -L-fucosidase cDNA and denition of the fucosidosis mutation in English Springer Spaniels. Mammalian Genome 7: 271-274. Ory, D. S., Neugeboren, B. A. and Mulligan, R. C. (1996). A stable human-derived packaging cell line for production of high titer retrovirus/vesicular stomatis virus G pseudotypes. Proc Natl Acad Sci USA 93: 11400-11406. Oshima, A., Itoh, K., Nagao, Y., Sakuraba, H. and Suzuki, Y. (1990). -galactosidase-decient human broblasts: uptake and processing of the exogenous precursor enzyme expressed by stable transformant COS cells. Hum. Genet. 85: 505-508. Pearce, R. D., Callahan, J. W., Little, P. B., Klunder, L. R. and Clarke, J. T. (1990). Caprine -D-mannosidosis: characterisation of a model lysosomal storage disorder, Can. J. Vet. Res. 54: 22-29. Peters, C., Schmidt, B., Rommerskirch, W., Rupp, K., Zuhlsdorf, M., Vingron, M., Meyer, H. E., Pohlmann, R. and von-Figura, K. (1990). Phylogenetic conservation of arylsulfatases. cDNA cloning and expression of human arylsulfatase B. J. Biol. Chem. 265: 3374-3381. Piacibello, W., Sanavio, F., Garetto, L., Severino, A., Bergandi, D., Ferrario, J., Fagioli, F., Berger, M. and Aglietta, M. (1997). Extensive amplication and self-renewal of human primitive hematopoietic stem cells from cord blood. Blood 89: 2644-2653. Poeschla, E., Corbeau, P. and Wong-Staal, F. (1996). Development of HIV vectors for anti-HIV gene therapy. Proc Natl Acad Sci USA 93: 11395-11399. Reiser, J., Harmison, G., Kluepfel-Stahl, S., Brady, R. O., Karlsson, S. and Schubert, M. (1996). Transduction of nondividing cells using pseudotyped defective high-titer HIV type 1 particles. Proc Natl Acad Sci USA 93: 15266-15271. Ringden, O., Groth, C. G., Erikson, A., Granqvist, S., Mansson, J. E. and Sparrelid, E. (1995). Ten years experience of bone marrow transplantation for Gaucher disease. Transplantation 59: 864-870. Robertson, D. A., Freeman, C., Nelson, P. V., Morris, C. P. and Hopwood, J. J. (1988). Human glucosamine-6sulfatase cDNA reveals homology with steroid sulfatase. Biochem Biophys Res Commun 157: 218-224. Roe, T., Reynolds, T. C.,Yu, G. and Brown, P.O. (1993). Integration of murine leukemia virus DNA depends on mitosis. EMBO J 12: 2099-2108. Sajjadi, N., Kamantigue, E., Edwards, W., Howard, T., Jolly, D., Mento, S. and Chada, S. (1994). Recombinant retroviral vector delivered intramuscularly localises to the site of injection in mice. Hum. Gene Ther. 5: 693-699. Salvatori, G., Ferrari, G., Mezzogiorno, A., Servidei, S., Coletta, M., Tonali, P., Giavazzi, R., Cossu, G. and Mavilio, F. (1993). Retroviral vector-mediated gene transfer into human primary myogenic cells leads to expression in muscle bers in vivo. Hum. Gene Ther. 4: 713-723. Salvetti, A., Moullier, P., Cornet, V., Brooks, D., Hopwood, J. J., Danos, O. and Heard, J. M. (1995). In vivo delivery of human -L-iduronidase in mice implanted with neo-organs. Hum. Gene Ther. 6: 1153-1159. Sands, M. S., Vogler, C., Kyle, J. W., Grubb, J. H., Levy, B., Galvin, N., Sly, W. S. and Birkenmeier, E. H. (1 994). Enzyme replacement therapy for murine mucopolysaccharidosis type VII. J. Clin. Invest. 93: 2324-2331. Scharfmann, R., Axelrod, J. H. and Verma, I. M. (1991). Long-term in vivo expression of retrovirus-mediated gene transfer in mouse broblast implants. Proc Natl Acad Sci USA 88: 4626-4630. Schuchmann, E. H., Toroyan, T. K., Haskins, M. E. and Desnick, R. J. (1989). Characterisation of the defective glucuronidase activity in canine mucopolysaccharidosis type VII. Enzymes 42: 174-180. Scott, H. S., Anson, D. S., Orsborn, A. M., Nelson, P. V., Clements, P. R., Morris, C. P. and Hopwood, J. J. (1991). Human -L-iduronidase: cDNA isolation and expression. Proc Natl Acad Sci USA 88: 9695-9699. Scott, H. S., Blanch, L., Guo, X-H., Freeman, C., Orsborn, A., Baker, E., Sutherland, G. R., Morris, C. P. and Morris, J. J. (1995). Cloning of the sulphamidase gene and identication of mutations in Sanlippo A syndrome. Nat. Genet. 11: 465-467. Shull, R. M., Lu, X., McEntee, M. F., Bright, R. M., Pepper, K. A. and Kohn, D. B. (1996). Myoblast gene therapy in canine mucopolysaccharidosis I: Abrogation by an immune response to -L-iduronidase. Hum. Gene Ther. 7: 1595-1603. Snyder, E. Y., Taylor, R. M. and Wolfe, J. H. (1995). Neural progenitor cell engraftment corrects lysosomal storage throughout the MPS VII mouse brain. Nature 374: 367-370. Stewart, F. M., Temels, D., Lowry, P., Thraves, T., Grosh, W. W. and Quesenberry, P. J. (1993). Post-5-uorouracil human marrow: stem cell characteristics and renewal properties after autologous marrow transplantation. Blood

81: 2283-2289. Stoltzfus, L. J., Sosa-Pineda, B., Moskowitz, S. M., Menon, K. P. Dlott, B., Hooper, L., Teplow, D. B., Shull, R. M. and Neufeld, E. F. (1992). Cloning and characterisation of cDNA encoding canine -L-iduronidase. mRNA deciency in mucopolysaccharidosis I dog. J. Biol. Chem. 267: 6570-6575. Sweetman, L. (1996). Newborn screening by tandem mass spectrometry (MS-MS). Clin. Chem. 42: 345-346. Taylor, R. M. and Farrow, B. R. H. (1988). Animal model series. Fucosidosis. Comp. Pathol. Bull. 20: 2-5. Taylor, R. M. and Wolfe, J. H. (1994). Cross-correction of -glucuronidase deciency by retroviral vector-mediated gene transfer. Exp. Cell Res. 214: 606-613. Taylor, R. M. and Wolfe, J. H. (1997). Decreased lysosomal storage in the adult MPS VII mouse brain in the vicinity of grafts of retroviral vector-corrected broblasts secreting high levels of -glucuronidase. Nature Medicine 3: 771-720. Taylor, R. M., Farrow, B. R. H. and Healy, P. J. (1987). Canine fucosidosis: Clinical ndings. J. Small Anim. Pract. 28: 291-300. Taylor, R. M., Farrow, B. R. H. and Stewart, G. J. (1992). Amelioration of clinical disease following bone marrow transplantation in fucosidase-decient dogs. Am. J. Med. Gen. 42: 628-632. Taylor, R. M., Farrow, B. R. H., Stewart, G. J. and Healy, P. J. (1986). Enzyme replacement in nervous tissue after allogeneic bone-marrow transplantation for fucosidosis in dogs. Lancet 2: 772-774. Taylor, R. M., Farrow, B. R. H., Stewart, G. J., Healy, P. J. and Tiver, K. (1988). The clinical effects of lysosomal enzyme replacement by bone marrow transplantation after total lymphoid irradiation on neurological disease in fucosidase decient dogs. Trans. Proc. 20: 89-93. The International Batten Disease Consortium (1995). Isolation of a novel gene underlying Batten disease. Cell 82: 949-957. Thompson, J. N., Jones, M. Z., Dawson, G. and Huffman, P. S. (1992). N-acetylglucosamine-6-sulphatase deciency in a Nubian goat: a model of Sanlippo syndrome type D (mucopolysaccharidosis IIID). J. Inher. Met. Dis. 15: 760-768. Tomatsu, S., Fukuda, S., Masue, M., Sukegawa, K., Fukao, T., Yamagishi, A., Hori, T., Iwata, H., Ogawa, T., Nakashima, Y., (1991). Morquio disease: isolation, characterisation and expression of full-length cDNA for human N-acetylgalactosamine-6-sulfatase. Biochem Biophys Res Commun 181: 677-683. Touraine, J. L. (1996). Treatment of human fetuses and induction of immunological tolerance in humans by in utero transplantation of stem cells into fetal recipients. Acta. Haematol. 96: 115-119. Unger, E. G., Durrant, J., Anson, D. S. and Hopwood, J. J. (1994). Recombinant -L-iduronidase: characterization of the puried enzyme and correction of mucopolysaccharidosis type I broblasts. Biochem J. 304: 43-49. van-Beusechem, V. W. and Valerio, D. (1996). Gene transfer into hematopoietic stem cells of nonhuman primates. Hum. Gene Ther. 7: 1649-1668. van-Beusechem, V. W., Bart-Baumeister, J. A., Hoogerbrugge, P. M. and Valerio, D. (1995). Inuence of + interleukin-3, interleukin-6, and stem cell factor on retroviral transduction of rhesus monkey CD34 hematopoietic progenitor cells measured in vitro and in vivo. Gene Ther. 2: 245-255. Veelken, H., Jesuiter, H., Mackensen, A., Kulmberg, P., Schultze, J., Rosenthal, F., Mertelsmann, R. and Lindemann, A. (1994). Primary broblasts from human adults as target cells for ex vivo transfection and gene therapy. Hum. Gene Ther. 5: 1203-1210. Vellodi, A., Cragg, H., Winchester, B., Young, E., Young, J., Downie, C. J., Hoare, R. D., Stocks, R. and Banerjee, G. K. (1995). Allogeneic bone marrow transplantation for fucosidosis. Bone Marrow Transplant. 15: 153-158. Walkley, S. U., Thrall, M. A. and Dobrenis, K. (1996). Targeting gene products to the brain and neurons using bone marrow transplantation: A cell mediated delivery system for therapy of inherited metabolic human disease, in Protocols for gene transfer in neuroscience: Towards gene therapy of neurological disorders. Eds P. R. Lowenstein and L. W. Enquist. (John Wiley and sons Ltd.) Walkley, S. U., Thrall, M. A., Dobrenis, K., Huang, M., March, P. A., Siegal, D. A. and Wurzelmann, S. (1994). Bone marrow transplantation corrects the enzyme defect in neurons of the central nervous system in a lysosomal storage disease. Proc Natl Acad Sci USA 91: 2970-2974. Weber, B., Blanch, L., Clements, P. R., Scott, H. S. and Hopwood, J. J. (1996). Cloning and expression of the gene involved in Sanlippo B syndrome. Hum. Mol. Genet. 5: 771-777. Wengler, G. S. et al. (1996). In-utero transplantation of parental CD34 haematopoietic progenitor cells in a patient with X-linked severe combined immunodeciency (SCIDXI). Lancet 348: 1484-1487. Whitley, C. B., McIvor, R. S., Aronovich, E. L., Berry, S. A., Blazar, B. R., Burger, S. R., Kersey, J. H., King, R. A., Faras, A. J., Latchaw, R. E., McCullough, J., Pan, D., Ramsay, N. K. and Stroncek, D. F. (1996). Retroviralmediated transfer of the iduronate-2-sulfatase gene into lymphocytes for treatment of mild Hunter syndrome (mucopolysaccharidosis type II). Hum. Gene Ther. 7: 537-49

Williams, M. A. and Fukuda, M. (1990). Accumulation of membrane glycoproteins in lysosomes requires a tyrosine residue at a particular position in the cytoplasmic tail. J. Cell Biol. 111: 955-966. Wilson, C., Reitz, M. S., Okayama, H. and Eiden, M. V. (1989). Formation of infectious hybrid virions with Gibbon ape Leukemia Virus and Human T-cell Leukemia Virus retroviral envelope glycoproteins and the gag and pol proteins of Moloney Murine Leukemia Virus. J. Virol. 63: 2374-2378. Wilson, P. J., Morris, C. P., Anson, D. S., Occhiodoro, T., Bielicki, J., Clements, P. R. and Hopwood, J. J. (1990). Hunter syndrome: Isolation of an iduronate-2-sulfatase cDNA clone and analysis of patient DNA. Proc Natl Acad Sci USA 87: 8531-8535. Wolfe, J. H., Sands, M. S., Barker, J. E., Gwynn, B., Rowe, L. B., Vogler, C. A. and Birkenmeier, E. H. (1992). Reversal of pathology in murine mucopolysaccharidosis type VII by somatic cell gene transfer. Nature 360: 749-753. Wolfe, J. H., Schuchman, E. H., Stramm, L. E., Concaugh, E. A., Haskins, M. E., Aguirre, G. D., Patterson, D. F., Desnick, R. J. and Gilboa, E. (1990). Restoration of normal lysosomal function in mucopolysaccharidosis type VII cells by retroviral vector-mediated gene transfer. Proc Natl Acad Sci USA 87: 2877-2881. Xu, L. C., Karlsson, S., Byrne, E. R., Kluepfel-Stahl, S., Kessler, S. W., Agricola, B. A., Sellers, S., Kirby, M., Dunbar, C. E., Brady, O. R., Nienhuis, A. W. and Donahue, R. E. (1995). Long-term in vivo expression of the human + glucocerebrosidase gene in nonhuman primates after CD34 hematopoietic cell transduction with cell-free retroviral vector preparations. Proc Natl Acad Sci USA 92: 4372-4376.

Anson: Gene therapy for the mucopolysaccharidoses Gene Therapy and Molecular Biology Vol 2, page PAGE 13 PAGE 12 PAGE 13 Gene Therapy and Molecular Biology Vol 2, page PAGE 1 PAGE 1

Gene Ther Mol Biol Vol 2, 14-30. August 7, 1998.

Gene therapy strategies utilizing carcinoembryonic antigen as a tumor associated antigen for vaccination against solid malignancies Joseph R. Kelley and David J. Cole Medical University of South Carolina, Dept. of Surgery, 171 Ashley Avenue, 420N CSB, Charleston, SC 29425. ________________________________________________________________________ ______________________________ Correspondence: David J. Cole, MD, Medical University of South Carolina, Department of Surgery, 171 Ashley Ave, Charleston, SC 29425, Tel: (803) 792-1387, Fax: (803) 792-2048; E-mail: coledj@musc.edu Keywords: gene therapy, T-lymphocytes, tumor immunity, vaccination, CEA

Summary Advanced solid malignancies represent a signiď&#x192;&#x17E;cant clinical problem with few effective treatment options. Carcinoembryonic antigen (CEA) is a well deď&#x192;&#x17E;ned tumor-associated antigen on the surface of many solid malignancies that is currently used as a diagnostic and prognostic marker. Recent advances in tumor immunology, the understanding of antigen presentation, and gene transfer vector systems now provide gene therapy strategies in the form of a cancer vaccine which target CEA in an effort to induce a therapeutic immune response. This chapter provides a brief history of cancer vaccine development, discusses current strategies for generating or augmenting CEA-speciď&#x192;&#x17E;c immunity, and focuses on ongoing CEA-based gene therapy approaches for the treatment of solid malignancies

I. Introduction Advances in early detection, surgical technique and chemotherapy have improved treatment options for many solid malignancies such as colorectal, non-small cell lung, and breast cancer. Despite these gains however, the overall mortality rates for these patients have remained relatively stable over the past two decades and continue to account for 60% of annual cancer deaths (Landis 1998). Additionally, once past a surgically resectable stage I or stage II disease, effective radiotherapy and chemotherapy options are limited and survival rates decrease rapidly. Clearly, novel therapeutic approaches are needed for advanced solid malignancies. Gene therapy in the form of a cancer vaccine may provide new treatment options for patients with these advanced stage cancers. An increasing number of human tumors are being shown to display specic tumor associated antigens (TAA) on their surfaces in combination with the major histocompatibility complex (MHC-I). These TAA have the capability of being recognized by cytotoxic T-lymphocytes (CTL) and therefore may function as targets for a tumor specic immune response. Melanoma derived antigenic peptides from proteins such as MART-1, MAGE-1, MAGE-3, gp100, tyrosinase, and catenin have been detected by their ability to sensitize peptide-pulsed target cells to lysis by TIL-derived CTL lines (Kawakami 1994, Van der Bruggen 1991, Kawakami 1994, Cox 1994, Robbins 1996). Furthermore, several different TAAs including HER-2/neu, muc-1, PSA, and CEA have been identied in solid malignancies (Jerome, 1991, Peoples 1995, Ioannides 1993, Tsang 1995, Correale 1997). Although the in vivo signicance of each of these antigens in T-cell immune response against cancer is yet to be dened, it is clear that CTL, which specically recognize tumor antigens in the context of MHC-I, do exist in vivo. Initial clinical work utilizing adoptive transfer of cytotoxic T-lymphocytes capable of specically recognizing TAA led to tumor regression in select patients, demonstrating that a clinically relevant tumor regression can be mediated by TAA specic

CTL (Rosenberg 1986). Moreover, recent work, performed in murine models and in patients with metastatic melanoma utilizing TAA epitopes now provide

evidence that a

in vivo cancer therapy approach based on relevant tumor associated antigens alone can be effective. Recent clinical studies using melanoma derived antigens for vaccination have reported a 42% response rate in patients with advanced disease, closely correlating with the in vitro cytotoxic response to the TAA (Rosenberg 1998). These studies and

others have therefore provided a , , great impetus for furthe r development based cancer and design of TAAvaccines (Mandelboim 1994, Feldkamp 1995, Meleif 1995). Capitalizing on these recent advances in tumor immunology and antigen presentation by combining them with currently available gene transfer vector systems now provides gene therapy vaccination options which may generate or augment an in vivo TAA-specic CTL population. Such approaches have the potential to exploit the exquisite specicity of the immune system to target both primary lesions and metastatic colonies with lower toxicities than current treatment options. Furthermore, an immunomodulatory gene therapy approach would allow a relatively small population of transfected cells to induce a systemic immune response against cancer. This represents a signicant advantage over classical gene replacement therapies which may require the transfer of gene constructs into every cell within a tumor, a feat currently beyond existing vector technologies. Gene therapy in the form of a cancer vaccine would also benet from ease of administration, off-the-shelf availability, lack of a need for ex vivo manipulation, and would allow for the presentation of TAA epitopes away from the immunosuppressive effects of the local tumor site. Consequently, the majority of current cancer gene therapy protocols are now focusing on immunomodulatory approaches to develop a "cancer vaccine" (Sikora 1997).

II. Background There is a long history of clinical trials attempting to immunize cancer patients with various cell preparations and tumor extracts. One early recorded attempt at cancer vaccination occurred in 1777 when Nooth, Surgeon to the Duke of Kent, inoculated himself repeatedly with cancer tissue. Ailbert, physician to King Louis XVIII, subsequently injected himself with breast cancer tissue in 1808. The recorded clinical outcome in both cases was no effect, ill or otherwise (Lyons 1987). In 1893, William Coley vaccinated cancer patients with a bacterial extract (Coley's toxin) to induce a general systemic immune response in hopes that the tumor would be attacked in a nonspecic manner (Coley 1893). More recently, there has been a signicant body of vaccination work performed by several investigators. Mastrangelo's group utilized autologous, enzyme dissociated, irradiated tumor cells combined with the adjuvant bacillus Calmette-Guerin (BCG) as a cancer vaccine following cytoxan treatment. After repeated dosing, positive delayed type hypersensitivity skin tests (DTH) were seen against melanoma (Berd 1986, Berd 1990).

Mitchell's group reported objective responses in 4 of 25 and 5 of 17 patients using a similar approach of an allogenic melanoma cell line lysate with the adjuvant DETOXTM (Mitchell 1988, Mitchell 1990). Other groups have treated solid tumors in colon cancer patients by using enzyme dissociated, live, irradiated tumor cells combined with BCG. In low burden colon cancer patients this method gave delayed hypersensitivity responses against tumor cells and found a reduced relapse rate in some patients (Hoover 1985, O'Boyle 1992). Hollingshead reported that a partially puried "TAA" preparation generated by sonicating allogenic colon cancer cells and combining the extract with complete Freund's adjuvant similarly generated a DTH response (Hollingshead 1985). Although the results of these and many other studies did not provide a signicant clinical benet to the patient with a solid malignancy, they did discover that a cell mediated immune response is more effective in eliciting an antitumor effect than a humoral immune response. As a result, subsequent studies attempted to rene vaccine components in an effort to generate a cell mediated anti-tumor response against tumor associated antigens. The majority of recent advances in this eld have been generated by the discovery of interleukins and subsequent culture of T lymphocytes. Initial work by Yron et al. demonstrated that lymphocytes from murine spleens could be transformed into non-specic cytotoxic cells by incubation with high concentrations of IL-2. These lymphokine activated killer cells or LAK cells were shown to lyse weakly immunogenic tumor cells in vitro (Yron 1980, Rosenstein 1984). Preclinical experiments demonstrated that injection of LAK cells with IL-2 could reduce growth rate and prolong survival in murine models of metastatic lung and liver cancer (Mazumder 1984, Mule 1984, Mule 1985, Lafreniere 1985). Subsequent phase I and II human clinical trials were then conducted administering intravenous IL-2 and LAK cells with a 35% and 21% objective response rate noted in renal cell carcinoma and metastatic melanoma, respectively (Rosenberg 1991, Rosenberg 1992). Treatment with LAK cells and IL-2 also demonstrated activity against colorectal cancer in a small number of patients. Tumor-specic cytotoxic T-lymphocytes (CTL) derived from TIL that could specically recognize and respond to both autologous and allogeneic tumor cells in an MHC restricted manner have now be isolated in vitro (Rosenberg 1988, Rosenberg 1995, Rosenberg 1997). Recent progress in our knowledge of antigen processing/presentation and techniques for the isolation of peptides presented in an MHC-restricted fashion has led to the identication of tumor associated antigens (TAA) recognized by these T-lymphocytes (Topalian 1989, Darrow 1989, Wang 1995, Kawakami 1994, van der Bruggen 1994, Cox 1994, Robbins 1996). Several of the genes encoding for TAA's have been cloned, their class I MHC restricted epitopes described, and in some cases the functional specicity of T-cell receptor heterodimer recognition characterized (Van der Bruggen 1991, Rock 1993, Cole 1994, Cole 1995). These ndings support the concept that the observed CTLmediated tumor regression in vivo can be explained by the T-cell recognition of specic 9 or 10 amino acid peptides bound to MHC class I molecules presented on the surface of cancer cells. This represents a major step forward in cancer immunotherapy, and has provided the reagents to utilize specic tumor-associated antigens or peptides for vaccination therapy (Mandelboim 1994). In order for T-cell based immunotherapy to be of benet to the cancer patient, an

appropriate TAA marker or markers must be selected as a target, and then this target must be presented to the patient's immune system in such a way as to produce a clinically relevant CTL response to the tumor. In the eld of solid malignancies, several different TAA including muc-1, PSA, HER-2/neu, and CEA have been identied. Each of these antigens has advantages and disadvantages as vaccine targets for various cancer types. Prostate specic antigen (PSA) is a cysteine protease, whose expression is normally limited to prostate tissue. This organ specicity allows the use of PSA as both a diagnostic and prognostic marker for prostate cancer, and has lead to a substantial increase in the early detection of this disease. In addition to its role in early detection, PSA is also being investigated as a possible target for prostate-specic cancer vaccines. A recent clinical trial vaccinated prostate cancer patients with dendritic cells pulsed with prostate specic membrane antigen and found a decrease in serum PSA levels, an enhanced cell-mediated immune response against PSA, and a partial regression of cancer in some patients (Salgaller 1998). This same group is continuing to explore the use of a cocktail of prostate antigens along with GM-CSF in efforts to improve PSA-based vaccines. HER-2/neu is a member of the epidermal growth factor receptor family that is overexpressed in 20-40% of intraductal carcinomas of the breast and 30% of ovarian cancers (Berchuck 1990, Kern 1990, Slamon 1987, Yonemura 1991) where it is associated with a poor prognosis (Slamon 1989). The protein is normally expressed during fetal development and is also found at low levels in epithelial cells of many normal tissues (Press 1990). Several approaches to HER-2/neu cancer vaccines are under investigation (Reviewed by Disis and Cheever 1997) and show promise in eliciting therapeutic antibodies against HER-2/neu positive tumors. MUC1 is a large O-glycosylated mucin polypeptide expressed at high levels in many human adenocarcinomas. Cancer-associated MUC1 has an altered pattern of glycosylation which exposes a series of extracellular, antigenic 20 amino acid tandem repeats to the immune system (Gendler 1990, Lan 1990). Various methods are being investigated to develop a MUC1-based cancer vaccine including, recombinant MUC1 expressing vaccinia virus (Hareuveni 1990, Acres 1993), MUC1 cDNA injection (Graham 1996, Pecher and Finn 1996), and recombinant MUC1-derived peptide immunization (Ding 1993, Apostolopoulos 1996, Samuel 1998). A mannan (polymannose)/ MUC1 peptide fusion protein induced a strong CTL response which led to MUC1+ tumor regression in mice (Apostolopoulos 1994). However, when human patients were vaccinated with the mannosylated MUC1 fusion protein they displayed high levels of IgG1 antibodies with a low cell mediated response (Karanikas 1997). This discrepancy has been explained by antibodies to the Gal(1,3) Gal epitope which cross-react with MUC1 (Apostolopoulos 1998). Humans, unlike mice, express high levels of anti-Gal antibodies which may divert MUC1 vaccination to a humoral response. It may therefore be necessary to target antigen presenting cells against MUC1 in vitro to avoid this anti-Gal induced humoral response (Apostolpoulos 1998). Finally, carcinoembryonic antigen (CEA) is an oncofetal protein which is overexpressed in 90% of gastrointestinal, 70% of non small-cell lung, and 50% of breast cancers (2,3,4). As a result of its presence in a large proportion of solid malignancies, CEA is a very attractive target for cancer vaccine therapy. It is one of the few TAAs in solid malignancies with which we have a signicant amount of basic knowledge, in addition to some of the most

mature preclinical and clinical vaccine experience. From these trials, we can derive some initial insight into the immune response to TAA vaccination, and as a result, they will serve as the focus for the remainder of this review. CEA was rst identied by Gold and colleges in 1965 as a fetal antigen which becomes reexpressed in neoplastic cells (Gold 1965). CEA is a 180 kDa cell surface glycoprotein that plays a role in cellular adhesions, cell to cell interactions, and glandular differentiation (Bebchimol 1989, Pignatelli 1990). It is a member of a large family of glycoproteins that are expressed in fetal, normal adult, and malignant tissues (Von Kleist 1972). Several of these family members share antigen cross reactivity with CEA. Non-specic cross reacting antigen (NCA) located on normal neutrophillic leukocytes, normal fecal antigen (NFA-1), and bile antigen (BGP-1) are all weakly cross reactive with anti-CEA antibodies. More strongly cross reactive family members include normal fecal antigen 2 (NFA-2) and the non-specic cross reacting antigen 2 (NCA-2) found in meconium (Von Kleist 1979). In a healthy adult, CEA itself is expressed at very low levels in normal gastrointestinal crypts and in healing intestinal mucosa. Normally, CEA is only weakly antigenic with undetectable anti-CEA antibody levels in normal patient serum (Foon 1995, Schlom 1996). In contrast, the sera of some cancer patients contain CEA-immunoglobulin immune complexes in the thousands of ng/mL yet their tumor remains. The goal of a T-cell based immunotherapy for solid malignancies is to increase the immunogenicity of this natural antigen until a clinically signicant CTL response can be achieved against CEA expressing tumors. Many different approaches have been investigated in an effort to augment this T-cell based immune response to CEA including polynucleotide vaccinations, anti-idiotypic antibodies, peptide pulsed dendritic cells, and recombinant vaccinia virus infection. Each of these techniques shows promise as a potential treatment of solid malignancies. A. Polynucleotide vaccinations It has been proposed that direct DNA immunization might best mimic the circumstances of TAA overexpression by a tumor (Conry 1995). Vaccination by DNA immunization allows for persistent high-level protein expression in vivo and early results have been promising. Myober cells in the mouse were shown to express foreign genes that have been injected into muscle in the form of naked DNA without any cationic lipids, retroviruses or other special delivery systems (Wolff 1990). The duration of gene expression in skeletal muscle using a RSV promoter driving a luciferase reporter exceeded 19 months post injection, even though the foreign plasmid DNA appeared to remain episomal (Wolff 1992). Naked plasmid DNA encoding inuenza A nucleoprotein (NP) delivered to mice by IM injection produced inuenza NP-specic antibodies and CTL response with protection from subsequent challenge with inuenza A virus (Ulmer 1993). DNA-coated microprojectiles have also been used to vaccinate rodents and non-human primates with a variety of HIV-1 encoded antigens and both cellular and humoral immune responses resulted (Coney 1994). More relevant to the treatment of solid malignancies, Conry's group has demonstrated lymphoblastic transformation and lymphokine release to human CEA using intramuscular injection of cDNA for human carcinoembryonic antigen into mice (Conry 1995, Conry 1995b). Furthermore, this naked DNA injection protected animals from tumor challenge

with 2 x 105 syngeneic, CEA expressing tumor cells with no evidence of local toxicity or inammation systemically or at the injection site. These results attest that DNA immunization holds great promise; however, immunization with DNA capable of integrating into the host genome raises signicant safety concerns. The novel approach of mRNA immunization may avoid the possibility of integration and thereby overcome this safety concern. Preliminary studies using CEA have shown that mRNA immunization can generate CEA-specic antibody responses (Conry 1996). Further renements in safety, expression promoters, methods of in vivo transfection, and the concurrent expression of cytokines and/or costimulatory molecules such as IL-2 and B7 will likely improve this direct polynucleotide injection approach to cancer vaccination. B. Anti-idiotypic antibodies

Anti-idiotypic antibodies are another method being investigated as a means of

establishing an anti-CEA immune response. Anti-CEA monoclonal antibodies have been

used primarily for clinical diagnosis of colorectal cancer, either as a tumor marker in serum

to monitor tumor recurrence, or as a means to localize CEA-bearing tumors and metastases

in patients (Hardman 1992). An additional application of antibody technology is the generation of anti-idiotypic antibodies that mimic CEA epitopes. Immunization with a given tumor-associated antigen (CEA) will generate antibodies against this antigen termed Ab1. The variable regions of Ab1 contain determinants known as idiotypes (Id), which are themselves immunogenic. Thus a series of anti-Id antibodies or Ab2 can be generated by injecting Ab1 into naive animals. Some of these Ab2 can effectively mimic the three dimensional structure of the CEA epitope identied by Ab1. Thus administration of Ab2 to cancer patients may generate an immune response to specic epitopes of CEA without generating non-specic cross reacting responses to other family members. An anti-Id antibody designated 3H1 mimics a biologically and antigenically distinct epitope of CEA, but not CEA family members found on normal tissues such as NCA

(Bhattachary-Chatterjee 1990). 3H1 was capable of inducing CEA-specic antibodies in mice and rabbits, and a preclinical study has begun in cynomolgus monkeys (Macaca fascicularis) using aluminum hydroxide precipitated 3H1. Monkeys injected with 3H1 develop specic anti-anti-Id (Ab3) responses that were capable of inhibiting binding of 3H1 to Ab1. In addition, immune sera from monkeys contained Ab3 that bound CEApositive carcinoma lines but not to CEA-negative cell lines. The induction of these antitumor antibodies in monkeys did not cause any apparent side-effects (Chakraborty 1995). The monoclonal antibody 3H1 was then tested as a method for CEA vaccine therapy in 12 human patients with advanced colorectal cancer. Each of the patients received four intracutaneous injections of aluminum-hydroxide-precipitated 3H1. Nine patients demonstrated a CEA specic anti-anti-idiotypic response. Seven out of 12 patients demonstrated idiotypic-specic T-cell proliferation responses and 4 showed T cell proliferation to CEA (Foon 1995). Peptide pulsed dendritic cells comprise a recent approach to cancer vaccines which is receiving a great deal of attention. Dendritic cells (DC) are professional antigen-presenting cells who function to present antigen to naive T cells. In the past, DC have been shown to stimulate both a naive and memory T-cell response in vitro (Inaba 1990, Mahta 1994). Recent studies demonstrated that vaccination of mice with DC pulsed with TAA-derived peptides was highly effective in priming cytotoxic T-lymphocytes responses, and established both a protective and therapeutic anti-tumor immunity in treated animals (Huang 1994, Porgador 1996, Boczkowski 1996, Zitvogel 1996, Paglia 1996). Human studies have found that DC pulsed with TAA proteins can induce a CTL response in vitro (Macatonia 1991, Mahta 1994, Bakker 1995) and can produce a measurable cellular immune response in some B-cell lymphoma patients (Hsu 1996). In the eld of solid tumors, Alters et al have reported that dendritic cells pulsed with the HLA-A2 restricted CEA-derived CAP-1 peptide can generate a CEA-specic CTL response as measured by a restricted T cell receptor repertoire in non-immunized pancreatic, colon, and breast cancer patients as well as in healthy volunteers (Alters 1998). Current studies are attempting to improve on these initial dendritic cell-based vaccines. In one such approach, dendritic cells treated with a proteosome inhibitor or with antisense to TAP-2, transporter associated with antigen presentation, demonstrated an increased density of MHC I expression on their surface which led to a more effective vaccine (Wong 1998). D. Recombinant Vaccinia vaccination One of the best studied and currently popular methods of generating an anti-CEA T cell mediated immune response is the use of a recombinant vaccinia virus. This direct immunologic approach to CEA-bearing tumors was initially developed by the Laboratory of Tumor Immunology and Biology at the NCI using inoculation with a recombinant vaccinia virus (rV-CEA) that expresses the human CEA gene (Kaufman 1991). Vaccinia was chosen for this effort due to an intense inammatory response generated at the site of infection which leads to both a humoral and cell mediated immune response (Bennick 1984,

Moss 1987). Copresentation of a weakly immunogenic protein product at the site of vaccinia viral infection has been shown to elicit a strong "bystander" immune response against a variety of weak antigens (Lathe 1987, Hellstrom 1989). Consequently, vaccinia is currently being investigated for use in immunizations against a wide range of infectious diseases as well as several types of cancer (Mackett 1987, Kierny 1984, Smith 1983, Langford 1986). A recombinant vaccinia expressing the HIV envelope protein has been administered to normal volunteers in phase I trials (Hu 1986, Cooney 1991) and constructs expressing tumor associated antigens have been tested in murine and non-human primate models (Estin 1988, Hu 1988, Bernards 1987, Hareuveni 1990, Hershey 1987, Kawa 1987). A pre-clinical murine model for rV-CEA was initially established using a 2.4 kilobase cDNA segment coding for CEA (Oikawa 1987) inserted into the thymidine kinase gene of a WR (Kaufman 1991) and a NYC (Kantor 1992) strain of vaccinia virus. Cells infected with recombinant virus, rV-CEA, expressed CEA on their surface as detected by the antiCEA monoclonal antibody COL-1. MC-38 murine adenocarcinoma was then transduced with the human CEA gene , causing CEA surface expression at levels comparable to those found on human colon cancer cell lines (Robbins 1991). Immune competent C57B / 6 mice were injected subcutaneously with 2 x 105 MC-38 cells or transduced MC-38 CEA cells. Seven days after tumor transplant, 10 animals with each tumor type were vaccinated with 1 x 107 plaque forming units (PFU) of

either wild-type vaccinia or rV-CEA. Vaccinations were repeated twice at 14

day

intervals.

The

animals

inoculated with rV-CEA showed inhibition of growth of CEA positive tumor. In addition, mice which survived the initial MC-38 tumor challenge due to rV-CEA treatment did not allow growth of MC-38-CEA +tumor when re-challenged (Kantor 1992). The safety of rV-CEA was then tested in the rhesus monkey model (Kantor 1992b) because a successful immune response against CEA could result in an auto-immune colitis against endogenous CEA in gastrointestinal crypts. There is also a risk of auto-immune reaction against the cross reacting fecal antigens (NFA 1 and 2) and bile antigen (BGP-1) resulting in further intestinal and biliary inammation. In addition, the expression of nonspecic cross reactive antigen 1 (NCA) on normal neutrophils holds the possible side effect of leukopenia. Eight monkeys received up to 4 scarications with either 1 x 108 or 5 x 108 PFU of rV-CEA and 4 monkeys received 5 x 105 PFU of control wild type vaccinia. All vaccinated monkeys developed typical local skin reactions, low grade fever, and lymphadenopathy after immunization. All rV-CEA vaccinated animals also exhibited strong anti-CEA responses, with no signs of auto-immune colitis and only minimal non-specic anti-NCA responses. Delayed type IV hypersensitivity responses were seen to intradermal injections of puried CEA in 7 or 8 recipients of rV-CEA, but none of the monkeys treated with wild-type vaccinia, indicating a specic cell mediated immune response. It should also be noted that cancer patients with high serum levels of anti-CEA immunoglobulin immune complexes do not show symptoms of immune complex deposition syndromes (Fuchs 1988). IV. Initial Phase I clinical trials of rV-CEA Based on this preclinical data, an initial phase I study was performed in patients with metastatic adenocarcinoma using an escalating dose administration of the rV-CEA vaccine (Tsang 1995). No Grade III or dose limiting toxicities were demonstrated in the study using doses as high as 1 x 108 PFU per vaccination. The only side effects to vaccination were a local, self-limited reaction at the injection site, lymphadenopathy, and low grade fever. A maximum tolerated dose, therefore, was not dened. Additionally, none of the potential problems of dose limiting leukopenia, auto-immune colitis, or toxic reactions to vaccinia itself were noted. Although a therapeutic response was not realized in this trial, three important facts

emerged. First, a series of HLA-A2 restricted peptides were identied which corresponded to the human major histocompatibility complex (MHC) class I restricted CTL epitopes within CEA. An immunodominant peptide identied in this series was the 9-amino acid (YLSGANLNL) CAP-1 peptide (Tsang 1995). Secondly, in vitro priming of post vaccination peripheral blood lymphocytes with the CAP-1 peptide in combination with IL-2 demonstrated MHC-restricted specic lytic activity against CEA expressing tumor cells in 5 of 5 patients tested (Tsang 1995) (Figure 1). Thus immune recognition of CEA does occur in patients treated with the rV-CEA vaccine but at a sub-clinical level. Finally, presumably due to the high incidence of previous exposure to vaccinia within the population, it was found that an intense anti-vaccinia immune response followed the rst vaccination. This inammation produced neutralizing antibodies and inhibited replication of virus at the second and third administration, thereby limiting the ability of booster inoculations of vaccinia to expand the anti-CEA T cell population. Thus rV-CEA appears to be a self limiting but useful agent in inducing a CEA-specic anti-tumor immunity. Other methods may be needed, however, to boost this initial response to clinically signicant levels. Several different approaches are currently being investigated to augment this initial rV-CEA induced CTL population including the use of various cytokine and costimulatory reagents, avian pox virus vectors, and TAA-derived peptide booster inoculation. One of these "second generation" rV-CEA vaccination approaches combines a Target cells V24 effector T cells CEAHLA-A2Specic LysisB

cells

---EBV-B A2-+-EBV-B A2/CEA+++SW837+--SW837

A2+++SW403+++ Figure 1. MHC class I restricted specic lytic activity of post CEA vaccinated PBL. Patient post vaccination 51 PBL samples (V24) stimulated in vitro with CAP-1 peptide displayed specic lytic activity by standard Cr release assay only against cells expressing both HLA-A2 and CEA. Targets included: autologous B cells (B cells); B cells transformed by EBV to express HLA-A2 (EBV-B A2); EBV transformed B cells expressing both HLA-A2 and CEA (EBV-B A2/CEA); the CEA positive, HLA-A2 negative colon cell line (SW837); the SW837 cell line transformed with HLA-A2 (SW837 A2); and the CEA positive, HLA-A2 positive colon cell line (SW403). Tsang et al. JNCI 87: 982, 1995.

recombinant vaccinia virus with various cytokine reagents. The cytokines Il-2, IFN-, and TNF are produced from the Th1 subset of CD4+ lymphocytes and normally function to induce a cell-mediated immune response. Although other cytokines were ineffective, exogenous IL-2 when added in combination with a TAA-based pox virus vaccine,

enhanced the immunogenicity of the tumor antigen and led to a decrease in pulmonary metastasis in animal models (Bronte 1995). GM-CSF is a potent cytokine which induces the differentiation of hematopoietic stem cells into dendritic cells and then promotes dendritic cell activation and differentiation at the local vaccination site. Studies adding GMCSF to TAA-based cancer vaccines resulted in enhanced TAA immunogenicity (Dranoff 1993) and a recent clinical trial in renal cancer patients demonstrated a signicant increase in DTH response when GM-CSF was added to vaccine formulation (Simons 1997). Recent studies have also focused on the heterodimeric cytokine IL-12, which also functions to shift a Th2 generated humoral immune response to the more effective Th1-based cellmediated immune response. A vaccine composed of the mutated p53 protein combined with IL-12 led to the regression of sarcoma in one animal model (Noguchi 1995), and IL-12 combined with a recombinant vaccinia virus led to a decrease in metastases and a signicant survival benet in a murine model of adenocarcinoma (Rao 1996). E C v,kpx2zV unogeichrasd7-1tlf.ybA m T w B S.95askr4,uin3T M y)1(hB lfct-A edm ow G An alternative method to boost the CEA-specic CTL population involves the use of an avian pox virus. Avian pox viruses are able to infect and express transgene in mammalian cells, but unlike vaccinia virus, avian pox viruses are not able to replicate in human cells. As a result, these vectors do not suffer from the dose limiting inammation and neutralizing antibodies seen with vaccinia, and consequently, avipox vector can be given repeatedly. In addition, avian pox viruses can be safely administered to immunosuppressed patients, a current limitation of vaccinia use. Canary and fowl pox viruses have proven safe in extensive clinical trials as a possible rabies vaccine in both Europe and the United States (Taylor 1991, Taylor 1994, Cadoz 1992, Fries 1996), and a canary pox virus expressing the CEA protein (ALVAC-CEA) has been shown to induce an antibody response, a lymphoproliferative response, and a cytotoxic T lymphocyte response in murine models (Hodge 1997). Moreover, the combination of one rV-CEA vaccination followed by two ALVAC-CEA booster injections resulted in a four-fold increase in CTL activity and prevented tumor formation in 5 of 8 animals (Hodge 1997). A fourth method to augment the initial anti-CEA CTL population is booster vaccination with peptides such as CAP-1. Peptide-based vaccines offer a greater control over the ability to manipulate the immune response than many previous methods. Through the use of clearly dened immunogenic epitopes, peptide vaccines may elicit a CD4+ or CD8+ specic response as determined by the investigator. Peptide boosting also benets from a relative ease of production, chemical stability, off the shelf availability, and lack of infectious or oncogenic potential (Aron and Horowitz, 1992). The initial use of a MHC class I restricted vaccine was reported independently by two groups studying Lymphocytic Choriomeningitis virus and Sendai virus (Schulz 1991, Katz

1991). Work in cancer therapy quickly adopted this approach and animal data from the Laboratory of Tumor Immunology and Biology (LTIB) at the NCI, has shown that subcutaneous immunization of mice with 100 g of short synthetic peptides (Ras5-17) demonstrated a specic T-cell immune response with no noticeable side effects (Peace 1991). The rst use of a peptide vaccine in humans demonstrated that injection of a lipoprotein containing a HLA-A*0201-binding peptide from hepatitis B virus along with a pan HLA-DR binding protein could induce a strong CTL response (Vitiello 1995). Marchand et al. then showed that vaccination of melanoma patients with a MAGE-3 peptide could lead to a partial regression in some patients (Marchand 1995). Thus a TAA derived peptide vaccination may safely and under the proper circumstances, effectively boost a rV-CEA primed CTL population. Extensive experience in microbiology has shown that combining adjuvant reagents with peptide or protein immunogens can prevent tolerance and lead to a productive immunization. The selection of the proper adjuvant for peptide immunization has a profound effect on antigen presenting cell activity at the local site of injection and therefore on the success of the vaccination attempt. Adjuvants may function by affecting the character and number of antigen presenting cells (APC) at the inoculation site, acting as a depot to prolong antigen/ APC exposure, or affecting the pathway by which proteins are processed (Allison 1994, Cole 1997). In the past, reagents such as BCG, Incomplete Freud's Adjuvant, or DetoxTM, have been shown to have an enhancing effect on both the humoral and cellular immune responses when used with vaccines (Ribi 1984). DetoxTM has been used in several clinical trials with minimal side effects limited to u-like symptoms and mild pain at the site of injection. A few patients who received DetoxTM treatment have developed a granuloma at the site of injection but this spontaneously resolved and has not been a dose limiting side effect (Ribi, unpublished data). Therefore, administration of the CAP-1-peptide with the Detox adjuvant reagent may safely stimulate and signicantly expand in vivo the number of CEA specic T-lymphocyte precursor cells present after rVCEA vaccination. An enhanced CEA-bearing tumor T-cell population could then potentially lead to a direct therapeutic anti-tumor immune response. A gene therapy cancer vaccine approach was therefore initiated within the Department of Surgery Molecular Oncology Lab at the Medical University of South Carolina in collaboration with the NCI / LTIB for the treatment of patients with metastatic adenocarcinoma by administration of a rV-CEA vaccine followed by CAP-1 peptide boost in DetoxTM PC adjuvant.

V. MUSC Phase I clinical trial: rV-CEA with CAP-1 peptide boost A phase I clinical trial was designed to investigate the effect of CAP-1 peptide boosting on the CEA-specic precursor T cell population established in patients initially vaccinated with rV-CEA. Because the pilot rV-CEA trial did not establish a maximum tolerated dose 1 x 108 pfu the highest dose tested in the original trial, was chosen as the initial vaccination dose. Additionally, intradermal administration rather than scarication was chosen based on recent data noting equivalent effectiveness for vaccine presentation (Galasso 1977, Wallack 1995). All patients received the rV-CEA vaccination on day 0 and again on week 4. This immunization was followed in four weeks by three rounds of CAP-1 peptide boosting on

week 12, 16, and 20 (Figure 2). As a phase I trial, the study was designed for 12 patients in four groups of three peptide escalations. If grade III toxicity were noted at any peptide dose level the cohort would be doubled. The dose of CAP-1 peptide to be administered was 300 g/mL for the rst three patients and was then escalated to 6000 g/mL in the nal group. At 4 weeks post treatment, patients will be evaluated for complete response (CR), partial response (PR), stabilization of disease (SD), or progression of disease (PD). Follow up is weekly until 28 days after the nal dose, and then monthly until disease progression or until initiation of any new form of therapy. The patient population enrolled on study was dened by diagnosis of a histologically conrmed, CEA+ adenocarcinoma of the gastrointestinal tract, breast, or lung with expected survival of 6-12 months with no concomitant therapy. Due to the CAP-1 MHC restriction, all patients must further demonstrate HLA-A2 expression by tissue typing. Patients were also required to have a Zubrod performance score of 0-1 with serum CEA levels of >10 ng/mL, and normal immunological testing by DTH and CD4/CD8 ratio.

rV-CEA (1.0 x 108 PFU)

8 Figure 2. Treatment Schema. Patients receive 1 x 10 pfu of rV-CEA by intradermal injection on day 0 and again on week 4. This vaccination is followed by CAP-1 peptide boosting on week 12, week 16, and week 20. Patients are followed for signs of response to treatment with immunological responses determined before rV-CEA vaccination, before CAP-1 peptide boosting, and after CAP-1 peptide boosting is complete.

At time of publication, 10 patients have been enrolled in this trial. Although the data is insufcient to draw conclusions as therapy is ongoing, the patient results to date are presented in Table 1. Table 1. Patient prole Patient CharacteristicsNumberAge (average)52.8 yrs SexMale3Female7Primary MalignancyColorectal6Lung2Gallbladder1Unknown1Prior TreatmentChemotherapy5None5Current StatusOngoing7*Off Study 3* 5 patients show clinically stable disease with 2 patients showing progression of disease.

Evaluation of the immune response to a TAA-based cancer vaccine is currently a major hurdle in clinical cancer vaccine trials. Although it is clear that CEA-specic T cells are present after rV-CEA immunization, the clinical response in patients is unpredictable. Physical examination and radiological monitoring are unequivocal measures of response. Short of this however, a meaningful measure of a vaccine's effect on a patients T cell population is also instructive. Due to their exceedingly small numbers, it is rarely possible to measure TAA-specic CTL precursor populations in patient peripheral blood samples (Coulie 1992, Marrocchi 1994, Herr 1994). The assays presently employed in attempts to monitor the immune response to cancer vaccination include delayed type hypersensitivity testing, measurement of T cell precursor frequency by thymidine incorporation and cytokine release, target-specic lysis by chromium release assay, and T cell receptor analysis by gene scan and competitive PCR. Unfortunately, none of these assays give an accurate picture of the T cell response to treatment in and of themselves. Intradermal injection of irradiated tumor cells into a patient before and after treatment elicits a delayed type hypersensitivity response, and this DTH is the in vivo assay most commonly used in clinical trials to follow T cell response to vaccination. Although a DTH assay is technically simple to perform and is generally present in patients displaying a measurable clinical response to treatment, the assay is not predictive of clinical response (Berd 1990). Several in vitro assays are also used to evaluate T cell response to vaccination including cell proliferation, cytokine production, and chromium release assays. These methods measure CTL precursor frequency by culturing TAA pulsed or TAA expressing target cells with patient derived T cells. Limiting dilution techniques allow all of these methods to quantitate the precursor frequency in patient samples. The use of [3H]-thymidine incorporation provides a direct measure of precursor cell proliferation in response to TAA stimulation (Wucherpfennig 1995). The release of cytokines such as IL-2, IFN-, and TNF from patient T cells grown in mixed culture is also used to measure the T cell response to TAA immunization. The Cr51 release assay is the most common in vitro assay used to monitor T cell response to cancer vaccines. However, during a recent clinical trial, patients who underwent complete remission of melanoma as a result of MAGE-3 peptide vaccination did not demonstrate any MAGE-3-specic CTL activity as detected by the chromium release assay (Marchand 1995). The chromium release assay also failed to detect TAA-specic CTL activity in a trial involving a gp100 peptide which was modied to more tightly bind the MHC complex even though this peptide vaccination demonstrated a 41% clinical response rate in patients (Parkhurst 1996).

An alternative measure of T cell response to a TAA-based vaccine is analysis of T cell receptor subtype expression. T cells recognize MHC-restricted antigens through a heterodimeric T cell receptor (TCR) composed of and chains. Somatic recombination between variable (V), joining (J), and diversity (D) genes along with insertion of random N-nucleotides, generates a wide diversity of TCR subtypes in naive T cells. As a result, peripheral blood mononuclear cell samples from non-immunized patients display a roughly equivalent abundance of TCR variable -chain subtypes (TCR-V1 through TCR-V24). In contrast, if a TAA-based vaccine induces a clonal expansion of T cells recognizing the antigen then a subsequent alteration in TCR subtypes expression patterns should result. In fact, TCR screening studies have demonstrated that CTL effector populations can display an oligoclonal expression pattern after peptide immunization and in vitro stimulation (Figure 3). Loftus et al. have shown that TCR-V14, along with V4 and V3 are sharply increased in peripheral blood lymphocytes (PBL)

EMBED Excel.Chart.5 \s Figure 3. Post rV-CEA vaccinated PBL, stimulated in vitro with CAP-1 peptide display an oligoclonal 6 expansion of T cell receptor V family subtypes. Total cellular RNA was isolated from 5 x 10 V24 T cells. Firststrand cDNA was then synthesized from 1 g of total RNA and ampliď&#x192;&#x17E;ed with 25 V oligonucleotides and FITClabeled C oligonucleotide. Labeled PCR products were loaded on a 6% acrylamide sequencing gel and the samples were then run on an ABI 373 sequencer for size and ď&#x192;&#x;ourescence intensity determination. The relative percentages of each V subfamily are represented as histograms.

from peptide stimulated patients (Loftus 1996). Several groups have corroborated these ndings and other studies have shown that V3 and V4 are increased in MART-1 peptide stimulated CTL (Cole 1994, Sensi 1995). These studies have clearly shown TCR changes post vaccination, however, a predictable patient to patient trend in subfamily response has not been observed (Cole 1997). RT-PCR is currently used to identify TCR family subtypes in these analyses, but due to different family-specic annealing temperatures, RT-PCR cannot be used to accurately quantitate various TCR expression levels. The advent of competitive PCR (cPCR) may overcome this difculty and allow quantitation of specic TCR-V chain subtypes within CTL samples derived from PBL and TIL (Uhrberg 1996). In the MUSC phase I clinical trial using rV-CEA vaccination with CAP-1 peptide boosting, patients will be evaluated by several different methods to determine both humoral and cell-mediated responses to treatment. Labs for in vitro testing will be drawn on week 0 before vaccination, on week 12 before CAP-1 peptide boosting, and on week 24 after peptide boosting is completed. Patient sample testing will be divided between the Laboratory of Tumor Immunology and Biology at the NCI and our laboratory at the Medical University of South Carolina (MUSC). The LTIB will evaluate humoral response to vaccination by standard ELISA assay for pre and post treatment levels of CEA, normal cross reactive antigen (NCA), anti-vaccinia, anti-CEA, and anti-NCA antibodies. The level of CEA-anti-CEA immune complexes, CD3, CD4, and CD8 subsets will also be measured. The LTIB will also study T cell precursor frequency in pre and post treatment samples by using limiting dilution assays for [3H]-thymidine incorporation and microtiter ELISA cytokine release as previously described (Abrams 1995). MUSC will monitor T cell receptor family subtype alterations by a combination of gene scanning and competitive PCR techniques to follow any T cell-mediated response to treatment. The optimal dose of peptide will be determined as the lowest level which elicits the highest proliferation or cytotoxic response in all members of a group. Advances in tumor immunology are now combining with gene therapy techniques to provide promising new therapeutic options for the treatment of patients with solid tumors. There are currently several TAA involved in solid malignancy, including PSA, HER-2/neu, MUC1, and CEA which hold potential for future vaccine development. Of these, CEA has received perhaps the most attention as a target antigen for cancer vaccines by numerous methodologies. Previous studies with rV-CEA have proven safe with no evidence of autoimmune or other severe toxicity, and although a clinically relevant response has not yet been achieved, the clear demonstration of a CEA-specic CTL population in vaccinated patients represents a scientic success. Clinical trials using many vaccine strategies are now in progress in an effort to expand this CEA-specic CTL population to clinically benecial levels. A CAP-1 boosting approach to augment the rV-CEA generated anti-CEA CTL population has been initiated, but the effectiveness of the method has yet to be determined. It is clear however, that cancer vaccine-based gene therapy holds tremendous promise and may one day provide an effective treatment for patients with solid malignancy. Acknowledgements We wish to thank Dr. Kwong Y Tsang and Dr. Jeffery Schlom (NCI/ LTIB, Bethesda Maryland) for invaluable scientic input and collaboration on this rV-CEA vaccination with CAP-1 peptide boost clinical trial.

References

Acres RB, Hareuveni, M, Balloul, JM, Kieny, MP, (1993) Vaccina virus MUC-1 immunization of mice: immune response and protection against the growth of murine tumors bearing the MUC1 antigen. J Immunother 14, 136-143. Aichele P, Brduscha RK, Zinkernagel, RM, Hengartner H, Pircher HJ. (1995) T cell priming versus T cell tolerance induced bysynthetic peptides. J Exp Med 182, 261-266. Allison, AC. (1994) Adjuvants and immune enhancement. Int J Technol Assess Health Care 10, 107-120. Alters SE, Gadea JR, sorich M, OÕDonoghue G, Talib S, and Philip R. (1998) Dendritic cells pulsed with CEA peptide induce CEA-specic CTL with restricted TCR repertoire. J of Immunother 21, 17-26. Apostolopoulos V, Pietersz GA, McKenzie IFC. (1994) Murine immune response to cells transfected with human MUC1: immunization with cellular and syntheic antigens. Cancer Res 54, 5186-5193. Appella E, Yannelli JR, Adema GJ., Miki T, Rosenberg SA. (1994) Identication of a human melanoma antigen recognized by tumor-inltrating lymphocytes associated with in vivo tumor rejection. Proc Natl Acad Sci 91, 6458-6462. Arnon R, Horowitz RJ. (1992). Synthetic peptides as vaccines. Curr Opin Immunol 4, 449-453. Bakker ABH, Marland G, de Boer AJ, Huijbens RJ, Danen EA, Adema GJ, Figdor CG. (1995) Generation of antimelanoma cytotoxic T lymphocytes from healthy donors after presentation of melanoma-associated antigenderived epitopes by dendritic cells in vitro. Cancer Res 55, 330-334. Baskar S, Limcher L, Nabavi N, Jones RT, and Ostrand-Rosenberg S. (1995). Major histocompatablity complex class II+B7-1+ tumor cells are potent vaccines for stimulating tumor rejection in tumor-bearing mice. J Exp Med 181, 619-629. Bebchimol S, Fuks A, Jothy S,Beauchemia N, Shirota K, Stanners CP. (1989) Carcinoembryonic antigen, a human tumor marker, functions as an intracellular adhesion molecule. Cell 57, 327-334. Bennick JR, Yewdell JW, Smith GL, Moller C, Moss B. (1984) Recombinant vaccinia virus primes and stimulates inuenza haemagglutinin-specic antigens. Nature 311, 578-579. Berchuck A, Kamel A, Whitaker R, Kerns B, Olt G, Kinney R, Soper J, Dodge R, Clarke-Peterson D, Marks P. (1990) Overexpression of HER-2/ neu is associated with poor survival in advanced epithelial ovarian cancer. Cancer Res 50, 4087-4091. Berd D, Maguire H, Mastrangelo M. Induction of cell-mediated immunity to autologous melanoma cells and regression of metastases after treatment with a melanoma cell vaccine preceded by cyclophosphamide. Cancer Res 46, 2572-2577. Berd D, Maguire H, McCue P, Mastrangelo MJ. (1986) Treatment of metastatic melanoma with an autologous tumor-cell vaccine: Clinical and immunological results in 64 patients. J Clin Oncol 8, 1858-1867. Berd D, Maguire HC Jr, McCue P, Mastrangelo MJ.(1992) Sequential chemo-immunotherapy in the treatment of metastatic melanoma. J Clin Oncol 10, 1338-1343. Bernards R, Destree A, Mckenzie S.Gordon E, Weinberg RA, Panicali D. (1987) Effective tumor immunotherapy directed against an oncogene-encoded produce using a vaccinia virus vector. Proc Natl Acad Sci 84, 6854-6858. Bhattachary-Chatterjee M, Mukerjee S, Biddle W, Foon KA, Kohler H (1990) Murine monoclonal antiidiotypic antibody as a potential network antigen for human carcinoembryonic antigen. J Immunol 145, 2758-2765. Blagosklonny MV, and El-Diery WS (1996) in vitro evaluation of a p53-expressing adenovirus as an anti-cancer drug. Int J Cancer 67, 386-392. Boczkowski D, Nair S, Snyder D, Giloboa E. (1996) Dendritic sells pulsed with RNA are potent antigen presenting cells in vitro and in vivo. J Exp Med 184, 465-472. Brichard V, Van Pel W, Wolfel T, DePlaen E, Lethe B, Coulie P, Boon T. (1993) The tyrosinase gene codes for an antigen recognized by autologous cytolytic T lymphocytes on HLA-A2 melanomas. J Exp Med 178, 489-495. Bronte V, Tsung K, Rao JB, Chen PW, Wang M, Rosenberg SA, and Restifo NP. (1995) IL-2 enhances the function of recombinant poxvirus-based vaccines in the treatment of established pulmonary metastases. J Immunol 154, 5282-5292. Cadoz M, Strady A, Meignier B, Taylor J, Tartaglia J, Paoletti E, Plotkin S. (1992) Immunization with canarypox virus expressing rabies glycoprotein. Lancet 339, 1429-1432. Cai DW, Mukhopadhyay T, Lui T, Fujiwara T, Roth JA. (1993) Stable expression of the wild type p53 gene in human lung cancer cells after retroviral-mediated gene transfer Hum Gene Ther 4, 1017-24. Chakraborty M, Foon KA, Kohler H, Bhattachary-Chatterjee M. (1995) Preclinical evaluation in nonhuman

primates of an anti-idiotypic antibody that mimicks the carcinoembryonic antigen. J Immunother 18, 95-103. Cole DJ, Gattoni-Celli S, McClay EF, Nabavi N, Warner SN, Newton D, Woolhiser C, Wilson M, Vournakis J. (1997). Characterization of a sustained release delivery system for combined cytokine/peptide based vaccination using a fully-acetylated poly-N-acetyl glucosamine matrix. Clin Cancer Res 3, 867-873. Cole DJ, Weil DP, Shamamian P, Rivoltini L, Kawakami Y, Topalian S, Jennings C, Eliyahu S, Rosenberg SA, Nishimura MI. (1994) Identication of MART-1 specic T-cell receptors: T-cells utilizing distinct T-cell receptor variable and joining regions recognize the same tumor epitope. Cancer Res 54, 5265-5268. Cole DJ, Weil DP, Shilyansky J, Custer M, Kawakami Y, Rosenberg SA, Nishimura M. (1995) Characterization of the functional specicity of a cloned T-cell receptor heterodimer recognizing the MART-1 melanoma antigen. Cancer Res 55, 748-752. Cole DJ, Wilson MC, Rivoltini L, Custer M, Nishimura MI. (1997) T-cell receptor repertoire in matched MART-1 peptide-stimulated peripheral blood lymphocytes. Cancer Res 57, 5320-5327. Coley WB. (1991) The treatment of malignant tumors by repeated inoculations of erysipelas. With a report of 10 original cases. 1893: Clinical Orthopedic and Related Research 262, 3-11. Coney L, Wang B, Ugen KE, Boyer J, McCallus D, Srikantan V, Agadjanyan M, Pachuk CJ, Herold K, Merva M, Gilbert L, Deng K, Moelling K, Newman M, Williams WV, Wiener DB. (1994) Facilitated DNA innoculation induces anti-HIV-1 immunity in vivo. Vaccine 12, 1545-1550. Conry RM, LoBuglio AF, Curiel DT. (1996) Polynucleotide-mediated immunization therapy of cancer. [Review] Semin Oncol 23, 135-147. Conry RM, LoBuglio AF, Kantor J, Schlom J, Loechel F, Moore SE, Sumerel LA, Barlow DL, Abrams S, Curiel DT. (1994) Immune response to a carcinoembryonic antigen polynucleotide vaccine. Cancer Res 54, 1164-1168. Conry RM, LoBuglio AF, Loechel F, Moore SE, Sumerel LA, Barlow DL, Pike J, Curiel DT. (1995) A carcinoembryonic antigen polynucleotide vaccine for human clinical use. Gene Therapy 2, 59-65. Conry RM, LoBuglio AF, Loechel F, Moore SE, Summerel LA, Barlow DL, Curiel DT: (1995) A carcinoembryonic antigen polynucloetide vaccine has in vivo antitumor activity. Gene Therapy 2, 33-38. Correal P, Walmsley K, Nieroda C, Zaremba S, Zhu M, Schlom J, Tsang KY. (1997) in vitro generation of human cytotoxic T lymphocytes specic for peptides derived from prostate-specic antigen. J Natl Cancer Inst 89, 293-300. Coulie PG, Sombille M, Lehmann F, Hainaut P, Brasseur F, Devos R, Boon T. (1992) Precursor frequency analysis of human cytolytic T lymphocytes directed against autologous melanoma cells. Int J Cancer 50, 289-297. Cox AL, Skipper J, Chen Y, Henderson RA. (1994) Identication of a peptide recognized by ve melanomaspecic human cytotoxic T-cell lines. Science 264, 716-719. Darrow TL, Slinghoff CC Jr., Siegler HF. (1989) The role of HLA class I antigens in recognition of melanoma cells by tumor-specic cytotoxic T lymphocytes: evidence for shared tumor-antigens. J Immunol 142, 3329-3335. Ding L, Lalani EN, Reddish M, Koganty R, Wong T, Samuel J, Yacyshyn MB, Meikle, A, Fung PYS, TaylorPapadimitriou J, Longenecker BM. (1993) Immunogenicity of synthetic peptides related to the core peptide sequence encoded by the human MUC-1 mucin gene: effect of immunization on the growth of murine mammary adenocarcinoma cells trasnfected with human MUC-1 gene. Cancer Immunol Immunother 36, 9-17.

(

)

. Disis ML, Cheever MA (1997) HER-2/neu protein: a target for antigen-speciď&#x192;&#x17E;c immunotherapy of human cancer. Advances in Cancer Research 71, 343-371. Disis ML, Cheever MA. (1996) Oncogenic proteins as tumor antigens. Curr Opin Immunol 8, 637-642. Disis ML, Gralow JR, Bernhard H, Hand SL, Rubin WD, Cheever MA. (1996b) Peptide based, but not whole protein vaccines elicit immunity to the HER-2/neu oncogenic self protein. J Immunol 156, 3151-3155. Dranoff G, Jaffee E, Lazernby A, Golumbek P, Levitsky H, Brose K, Jackson V, Hamada H, Pardoll D, and Mulligan RC. (1993) Vaccination with irradiated tumor cells engineered to secrete GM-CSF stimulates potent, speciď&#x192;&#x17E;c, and long-lasting anti-tumor immunity. Proc Nat Acad Sci 90, 3539-3543, Estin CD, Stevenson US, Plowman GD, Hu SL, Sridhar P, Hellstrom I, Brown JP, and Hellstrom KE. (1988) Recombinant vaccinia virus against human melanoma antigen for p97 for use in immunotherapy. Proc Natl Acad Sci 85, 1052-1056. Feldkamp M, Vreugdenhil GR, Vierboon MP, Ras E, Melief CJ, Kast WM. (1995) Cytotoxic T lymphocytes raised against a subdominant epitope offered as a synthetic peptide eradicate human papilloma virus type 16 induced tumors. Eur J Immunol 25, 2638-2642. Foon KA, Chakraborty M, John WJ, Sherrat A, Kohler H, Bhattacharya-Chatterjee M. (1995) Immune response to the carcinoembryonic antigen in patients treated with an antiidiotypic antibody vaccine. J Clin Invest 96, 334. Fries LF, Tartaglia J, Taylor J, Kauffman EK, Meignier B, Paoletti E, and Plotkin S. (1996) Human safety and immunogenicity of a canary-pox rabies glycoprotein recombinant vaccine: an alternative poxvirus vector system. Vaccine 14, 428-434. Fuchs, C., Krapf F, Kern P, Hoferichter S, Jager W, Kalden JR. (1988) CEA-containing immune complexes in sera of patients with colorectal and breast cancer-analysis of comlexed immunoglobulin classes. Cancer Immunol Immunother 26,180-184. Galasso GJ, Karzon DT, Katz SL, Krugman S, Neff J, Robbins C. (1977) Clinical and serological study of four smallpox vaccines comparing variations of dose and route of administration. J Inf Dis 135, 131-186. Galili U, Vanky F, Rodriguez L. (1979) Activated T lymphocytes within human solid tumors. Cancer Immunol Immunother 6, 129-133. Gaugler B. Van den Eynde B, Van der Bruggen P, Romero P, Gaforio JJ, De Plaen E, Lethe B, Brasseur F, Boon T. (1994) Human gene MAGE-3 codes for an antigen recognized on a melanoma by autologous cytolytic T

lymphocytes. J Exp Med 179, 921-930. Gendler SJ, Lancaster CA, Taylor-Papadimitriou J, Duhig T, Peat N, Burchell J, Pemberton L, Lalani E-N, Wilson D. (1990) Molecular cloning and expression of the human tumor-associated polymorphic epithelial mucin PEM. Biol Chem 265, 15286-15293. Gold P, Freedman SO. (1965) Specic carcinoembryonic antigens of the human digestive system. J Exp Med 122, 467-481. Graham RA, Burchell JM, Beverly P, Taylor-Papdimitriou J. (1996) Intramuscular immunization with MUC-1 cDNA can protect C57 mice challenged with MUC1-expressing syngeneic mouse tumor cells. Int J Cancer 65, 664-670. Guinan EC, Gribben JG, Boussiotis VA, Freeman GJ, and Nadler LM. (1994) Pivotal role of the B7:CD28 pathway in transplantation tolerance and tumor immunity. Blood 84, 3261-3282. Hardman N, Murray B, Zwickl M, Kolbinger F, Pluschke G. (1992) Application of genetically-engineered antiCEA antibodies for potential immunotherapy of colorectal cancer. Int J Biol Markers 7, 203-209. Hareuveni M, Gautier C, Kierney MP, Wreschner D, Chambon P, Lathe R. (1990) Vaccination against tumor cells expressing breast cancer epithelial tumor antigen. Proc Natl Acad Sci 87, 9498-9502. Hellstrom, K.E. and Hellstrom, I. (1989) Cellular immunity against tumor specic antigens. Advan Cancer Res 12, 167-223. Herr W, Wolfel T, Heike M, Meyer zun Buschenfelde K-H., Knuth A. (1994) Frequency analysis of tumor-reactive cytotoxic T lymphocytes in peripheral blood of a melanoma patient vaccinated with autologous tumor cells. Cancer Immunol Immunother 39, 93-99. Hershey P, Edwards A, Coates A, Shaw H, McCarthy W, Mitton G. (1987) Evidence that treatment with vaccinia melanoma cell lysates may improve survival of patients with stage II melanoma. Cancer Immunol Immunother 25, 257-265. Hodge JW, McLaughlin JP, Kantor JA, and Schlom J. (1997) Diversied prime and boost protocols using recombinant vaccinia virus and recombinant non-replicating avian pox virus to enhance T-cell immunity and anti-tumor responses. Vaccine 15, 759-768. Hollingshead A, Elias G, Arlen M, et al. (1985) Specic active immunotherapy inpatients with adenocarcinoma of the colon utilizing TAA. Cancer 56, 480-489. Hoover H, Surdyke M, Dangel R, Peters LC, and Hanna MG Jr. (1985) Prospectively randomized trial of adjuvant active-specic immunotherapy for human colorectal cancer. Cancer 55, 1236-1243. Hsu FJ, Benike C, Fagnoni F, Liles TM, Crewinski D, Taidi B, Engleman EG, and Levy R.. (1994) Vaccination of patients with B-cell lymphoma using autologous antigen-pulsed dendritic cells. Immunol Today 2, 52-57. H u S L , K o s o w s k i S G , D a l r y m p l e J M . ( 1 9 8 6 ) E x p r e s s i o n o f A I D S v i r u s e n v e l o p e g e n e f o r r e c o m b i n a n t v a c c i n i a v i r u s . N a t u r e 3 2 5 , 2 5 7 - 2 6 5 . Hu SL, Plowman GD, Sridhar P, Stevenson US, Brown JP, and Estin CD. (1988) Characterization of a recombinant vaccinia virus expressing human melanoma-associated antigen p97. J Virol 62, 176-180. Huang AYC, Golumbek P, Ahmadzadeh M, Jaffee E, Pardoll D, Levitsky D. (1994) Role of bone marrow-derived cells in presenting MHC class I-resricted tumor antigens. Science 264, 961-965. Inaba K, Metley P, Crowley MT, Steinman RM. (1990) Dendritic cells pulsed with protein antigens in vitro can prime antigen-specic, MHC-restricted T cells in situ. J Exp Med 172, 631-640. Ioannides CG, Fisk B, Fan D, Biddison WE, Wharton JT, O'Brian CA. (1993) Cytotoxic T cells isolated from ovarian malignant asictes recognize a peptide derived from the HER-2/ neu proto-oncogene. Cell Immunol 151, 225-234. Jerome KR, Barnd DL, Bendt KM. Boyer CM, Taylor-Papadimitriou J, McKenzie IFC, Bast RC, Finn OJ. (1991) cytotoxic T lymphocytes derived from patients with breast adenocarcinomas recognize an epitope present on the protein core of a mucin molecule preferentially expressed by malignant cells. Cancer Res 51, 2908-2916. Kantor J, Irvine K, Abrams S, Kaufman H, DiPietro J, Schlom J. (1992) Anti-tumor activity and immune responses induced by a recombinant carcinoembryonic antigen-vaccinia virus vaccine. J Natl Cancer Inst 84, 1084-1091. K a n t o r J , I r v i n e K ,

A R E S I s v e c g p 6

b r a m s S , S n o g P , O l s e n , G r e n e r J , K a u f m a n H , g g e n s p e r g e r D , a n d c h l o m J . ( 1 9 9 2 ) m m u n o g e n i c i t y a n d a f e t y o f a r e c o m b i n a n t a c c i n i a v i r u s v a c c i n e x p r e s s i n g t h e a r c i n o e m b r y o n i c a n t i g e n e n e i n a n o n h u m a n r i m a t e . C a n c e r R e s 5 2 , 9 1 7 - 6 9 2 5 . Karanikas V, Hwang LA, Pearson J, Ong CS, Apostolopous V, Vaughn H and McKenzie IF. (1997). Antibody and T cell responses of patients with adenocarcinoma immunized with mannan-MUC1 fusion protein. J Clin Invest 100, 2738-2792. Kaufman H, Scholm J, Kantor J. (1991) A recombinant vaccinia virus expressing human carcinoembryonic antigen (CEA). Int J Cancer 48, 900-907. Kawa A, Arakawa S. (1987) The effect of attenuated vaccinia virus strain on multiple myeloma: a case report. Jpn J Exp Med 57, 79-81. Kawakami Y, Eliyahu S, Delgado CH, Robbins PR, Rivoltini L, Topalian SL, Miki T, and Rosenberg SA. (1994) Cloning of the gene coding for a shared human melanoma antigen recognized by autologous T cells inltrating into tumor. Proc Natl Acad Sci 91, 3515-3519. Kawakami Y, Eliyahu S, Jennings C, Sakaguchi K, Kang X, Southwood S, Robbins PF, Sette A, Appella E, Rosenberg SA. (1995) Recognition of multiple epitopes in the human melanoma antigen gp100 by tumorinltrating T lymphocytes associated with in vivo tumor regression. J Immunol 154, 3961-3968. Kawakami Y, Eliyahu S, Sakaguchi K, Robbins PF, Rivoltini L, Yannelli JR, Appella E, Rosenberg SA. (1992) Identication of the immunodominant peptides of the MART-1 human melanoma antigen recognized by the majority of HLA-A2 restricted tumor inltrating lymphocytes. J Exp Med 180, 347-352. Kawakami Y, Eliyahu S, Sakaguchi K, Robbins PF, Rivoltini L, Yannelli JR, Appella E, Rosenberg SA. (1994) Identication of the immunodominant peptides of the MART-1 human melanoma antigen recognized by the majority of HLA-A2-restricted tumor inltrating lymphocytes. J Exp Med 180, 347-352. Kawakami Y, Zakut R, Topalian SL, Stotter H, Rosenberg SA. (1992) Recognition by tumor-inltrating lymphocytes in HLA-A2.1-transfected melanomas. J Immunol 148, 638-643. Kern JA, Schwartz DA, Nordberg JE, Weiner DB, Greene MI, Torney L, and Robinson RA. (1990) Cancer Res 50, 5184-5191. Kieny, M.P., Lathe, R., Drillen, R. Spehner D, Skory S, Schmitt D, Wiktor T, Koprowski H, and Lecocq JP.(1984) Expression of rabies virus glycoprotein from a recombinant vaccinia virus. Nature 312, 163-166. Lafreniere R, Rosenberg SA: (1985) Successful immunotherapy of murine experimental hepatic metastases with lymphokine-activated killer cells and recombinant interleukin-2. Cancer Res 45, 3735-3741. Lan MS, Batra SK, Qui W-N, Metzgar RS, Hollingsworth MA. (1990) Cloning and sequencing of a human pancreatic mucin cDNA. J Biol Chem 265, 15294-15299. Landis SH, Murray T, Bolden S, and Wingo P. A. (1998) Cancer statistics. Ca: a Cancer Journal for Clinicians 48, 6-29. L a n g f o r d C J . ( 1 9 8 6 ) L i v e v i r u s e s f o r t h e d e l i v e r y o f m a l a r i a v a c c i n e s . P a p u a N e w G u i n e a M e d J 2 9 , 1 0 3 - 1 0 8 . Lathe R, Kierny MP, Gerlinger P, Clertant P, Guizani I, Cuzin F, and Chambon P. (1987) Tumor prevention and rejection with recombinant vaccinia. Nature 326, 878-880,. Loftus DJ, Castelli C, Clay TM, Swarcina P, Marincola FM, Nishimura I., Parmiani G, Appella E, Rivoltini L. (1996) Identication of epitope mimics recognized by CTL reactive to the melanoma/melanocyte-derived peptide MART-I. Journal Exp Med 184, 647-657. Lyons AS and Petrucelli RJ. (1987) Medicine An Illustrated History. Abradale Press New York. Macatonia SE, Patterson S, Knight SC. (1991) Primary proliferative and cytotoxic T-cell responses to HIV induced in vitro by human dendritic cells. Immunology 74, 399-406. Mackett, M. (1987) Vaccinia virus recombinants: potential vaccines. ACTA Trop 445, 94-97. Mandelboim O, Berke G, Fridkin M, Feldman M, Eisenstein M, and Eisenbach L. (1994) CTL induction by a

tumor-associated antigen octapeptide derived from a murine lung carcinoma, Nature 369, 67-71. M a n d e l b o i m O , V a d a i E , F r i d k i n M , K a t z - K i l l e l A , F e l d m a n M , B e r k e G , E i s e n b a c h L . ( 1 9 9 5 ) R e g r e s s i o n o f e s t a b l i s h e d m u r i n e c a r c i n o m a m e t a s t a s e s f o l l o w i n g v a c c i n a t i o n w i t h t u m o r a s s o c i a t e d a n t i g e n p e p t i d e s . N a t u r e M e d . 1 , 1 1 7 9 - 1 1 8 3 . Marchand M, Weynants P, Rankin E, Arienti F, Belli F, Parmiani G, Cascinelli N, Bourlound A, Vanwuck R, Humbelt Y, Cannon J-L, Laurent C, Naeyaert J-M, Pagne R, Deraemaeker R, Knuth A, Jager E, Brasseur F, Herman J, Coulie PG, Boon T. (1995) Tumor regression responses in melanoma patients treated with a peptide encoded by gene MAGE-3. Int J Cancer 63, 883-885. Mazumder A, Rosenberg SA: (1984) Successful immunotherapy of NK_resistant established pulmonary melanoma metastases by the intravenous adoptive transfer of syngeneic lymphocytes activated in vitro by interleukin-2. J Exp Med 159, 495-507. Mazzocchi A, Belli F, Mascheroni L, Vegetti C, Parmiani G, and Anichini A. (1994) Frequency of cytotoxic T lymphocyte precursors interacting with autologous tumor via the T-cell receptor: limiting dilution analysis of speciď&#x192;&#x17E;c CTL in peripheral blood and tumor-invaded lymph nodes of melanoma patients. Int J Cancer 58, 330-339. McDonell T, Troncodo P, and Bridbay SM.(1992) Expression of the protooncogene bcl-2 in the prostate and its association with emergence of androgen independent prostate cancer. Can Res 52, 6940-6944. Mehta DA, Markowicz S, Engleman E. (1994) Generation of antigen-speciď&#x192;&#x17E;c CD8+ CTLs from native precursors. J Immunol 153, 996-1003. Melief C, Kast M. (1995) T-cell immunotherapy of tumors by adoptive transfer of cytotoxic T-lymphocytes and by vaccination with minimal essential epitopes. Immunol Rev 146, 167-177. M i l l e r J F , M o r a g a n G , A l l i s o n I . ( 1

9 8 9 ) J . o f C o l d S p r i n g H a r b . S y m p . Q u a n t . B i o l . 2 , 8 0 7 8 1 3 . Mitchell M, Harel W, Kempf Rhu E, Kan-Mitchell J, Boswell WD, Dean G, and Stevenson L. (1990) Active specic immunotherapy for melanoma. J Clin Oncol 8, 856-869. Mitchell M, Kan-Mitchell J, Kempf R, et al.: (1988) Active specic immunotherapy for melanoma: Phase I trial of allogeneic lysates and a novel adjuvant. Cancer Res 48: 5883-5893. Moss, B. Flexner C. (1987) Vaccinia virus expression vectors . Ann Rev Immunol 5, 305-324. Mule JJ, Shu S, Rosenberg SA: (1985) The anti-tumor efcacy of lymphokine-activated killer cells and recombinant interleukin-2 in vivo . J Immunol 135, 646-652. Mule JJ, Shu S, Schwartz SL, and Rosenberg SA. (1984) Adoptive immunotherapy of established pulmonary metastases with LAK cells and recombinant interleukin-1. Science 225, 1487-1489.

Noguchi YE, Richards C, Chen YT, Olds LJ. (1995) Inuence of interleukin-12 on p53 peptide vaccination against established Meth A sarcoma. Proc Natl Acad Sci 92, 2219-23. O'Boyle K, Zamore R, Adluri S, Cohen A, Keneny N, Welt S, Lloyd KO, Oettgen HF, Old LJ, and Livingston PO. (1992) Immunization of colorectal cancer patients with modied ovine submaxillary gland mucin and adjuvant induces IgM and IgG antibodies to sialyated TN Cancer Res 52, 5663-5667. Ohashi PS, Oehen S, Buerki K, Pircher H, Ohashi CT, Odermall B, Malissen B, Zinkernager RM, and Hengartner H. (1991) Ablation of tolerance and induction of diabetes by virus infection in viral antigen transgenic mice. Cell 65, 305-317. Oikawa S, Nakazato M, Kosaki G. (1987) Primary structure of human carcinoembryonic antigen (CEA) deduced from cDNA sequence. Biochem Biophys Res Commun 142, 511-518. Paglia P, Chiodoni C, Rudolofo M, Colobo MP. (1996) Murine dendritic cells loaded in vitro with soluble protein prime CTL against tumor antigen in vivo . J Exp Med 183, 317-322. Pardoll DM. (1998) Cancer Vaccines. Nat Med 4, 525-531. Parkhurst MR, Salgaller ML, Southwood S, Robbins PF, Sette A, Rosenberg SA, Kawakami Y. (1996) Improved induction of melanoma reactive CTL with peptides from gp100 modied at HLA-A*02101 binding residues. J Immunol 157, 2539-2548. Peace DJ, Chen W, Nelson H, Cheever MA. (1991) T cell recognition of transforming proteins encoded by mutated ras protooncogenes. J Immunol 146, 2059-2065. Pecher G, Finn OJ. (1996) Induction of cellular immunity in chimpanzees to human tumro-associated antigen mucin by vaccination with MUC-1 cDNA-transfected Epstein Barr virus-immortalized autolougous B cells. Proc Nat Acad Sci 93, 1699-1704. Peoples GE, Goedegebuure PS, Smith R, Linehan DC, Yoshino I, Eberlein TJ. (1995) Breast and ovarina cancerspecic cytotoxic T lymphocytes recognize the same HER-2/ neu dervied peptied. Proc Nat Acad Sci 17, 432-436. Pignatelli M, Durbin H, Bodmer WF. (1990) Carcinoembryonic antigen functions as an accessory adhesion molecule mediating colon epithelial cell-collagen interactions.Proc Natl Acad Sci 87, 1541-1545. Porgador A, Snyder D, Gilboa E. (1996) Induction of antitumor immunity using bone marrow-generated dendritic cells. J Immunol 156, 2918-2926. Press M, Cordon-Cardo C, and Slamon D. (1990) Expression of the HER-2/neu oncogene in normal human adult and fetal tissues. Oncogene 5, 953-962. R a o J B , C h a m b e r l a i n R S , B r o n t e V ,

C a r r o l l M W , I r v i n e K R , M o s s B , R o s e n b e r g S A , a n d R e s t i f o N P . ( 1 9 9 6 ) I L

1 2 i s a n e f f e c t i v e a d j u v a n t t o r e c o m b i n a n t v a c c i n i a v i r u s b a s e d t u

m o r v a c c i n e s . J I m m u n o l 1 5 6 , 3 3 5 7 3 3 6 5 . Robbins PF, El-Gamil M, Kawakami YF, Loftus D, Appella E, and Rosenberg SA. (1996) A mutated -catenin gene encodes a melanoma-specic antigen recognized by tumor inltrating lymphocytes. J Exp Med 183, 1185-1192. Robbins PF, Kantor JA, Salgaller M, Hand PH, Fernstein PD, and Schlom J. (1991) Transduction and expression of the human carcinoembryonic antigen gene in a murine colon carcinoma cell line. Cancer Res 51, 3657-3662. Rock KL, Rothstein L, Gamble S, and Fleischacker C. (1993) Characterization of antigen-presenting cells that present exogenous antigens in association with class I MHC molecules. J Immunol 150, 438-446. Romani NS, Gruner D, Brang E, Kampgen E, Lenz A, Trockenbacker B, Konwalinka G, Fritsch PO, Steinman RM, and Schuler G. (1994) Proliferating dendritic cell progenitors in human blood. J Exp Med 180, 255-260. Rosenberg S A, Yang JC, Schwartentruber DJ, Hwu P, Marincola FM., Topalian SL. Restifo NP, Dudley ME, Schwartz SE, Spiess PJ, Wunderlich JR, Parkhurst MR, Kawakami Y, Seipp CA., Einhorn JH, White DE. (1998) Immunologic and therapeutic evaluation of a synthetic peptide vaccine for the treatment of patients with metastatic melanoma. Nature Medicine 4, 321-327. Rosenberg SA, Packard BS, Aebersold PM, Solomon D, Topalian SL, Toy ST, Simon P, Lotze MT, Yang JC, Seipp CA, Simpson C, CarterC, Bock S, Schwartzentruber D, Wei JP, White DE. (1988) Use of tumor inltrating lymphocytes and interleukin-2 in the immunotherapy of patients with metastatic melanoma. N Engl J Med 319, 1676-1680. R o s e n b e

r g S A , S p e i s s P , L a f r e n i e r e R A . ( 1 9 8 6 ) A n e w a p p r o a c h t o t h e a d o p t i v e i

m m u n o t h e r a p y o f c a n c e r w i t h t u m o r i n f i l t r a t i n g l y m p h o c y t e s . S c i e n

c e 2 3 3 , 1 3 1 8 1 3 2 1 . Rosenberg SA. (1995) The development of new cancer therapies based on the molecular identiď&#x192;&#x17E;cation of cancer regression antigens. Cancer J Sci Amer 1, 90-100. Rosenberg SA. (1997) Cancer vaccines based on the identiď&#x192;&#x17E;cation of genes encoding cancer regression antigens. Immunol Today 18, 175-182. Rosenberg SA: (1991) Immunotherapy and gene therapy of cancer. Cancer Res 51(suppl): 5074s-5079s. Rosenberg, S.A. (1992) The immunotherapy and gene therapy of cancer. J Clin Oncol 10,180-199. Rosenstein M, Yron I, Kaufmann Y, Rosenberg SA.(1984) Lymphokine-activated killer cells: lysis of fresh syngeneic natural killer-resistant murine tumor cells by lymphocytes cultured in interleukin-2. Cancer Res 44, 1946-53. Roth J, Mukhopadhyay T, Zhang WW, Fujiwara T, and Georges R.. (1994) Gene replacement stategies for the prevention and therapy of cancer. Eur J Can 30: 2032-2037. S a i s o n B e h o m o a r a s T . T o c q u e B , R e y I

, C h a s s i g n e l M , T h u o n g N T , a n d H e l e n e C . ( 1 9 9 1 ) S h o r t m o d i f i e d a n t i s e

n s e o l i g o n u c l e o t i d e s d i r e c t e d a g a i n s t H R A S p o i n t m u t a t i o n i n d u c e s

e l e c t i v e c l e a v a g e o f t h e m R N A a n d i n h i b i t T 2 4 c e l l p r o l i f e r a t i o n . E M

B O J 1 0 , 1 1 1 1 1 8 . Salgaller ML, Tjoa BA, Lodge PA, Ragde H, Kenny G, Boynton A, and Murphy GP. (1998) Dendritic cell based immunotherapy for prostate cancer. Crit Review Immunol 18(1) 109-119. Schlom J, Kantor J, Abrams S, Tsang KY, Panicalli D, Hamilton JM (1996) Strategies for the development of recombinant vaccines for the immunotherapy of breast cancer. Breast Cancer Res Treat 38, 27-39. Schulz M, Zinkernagel RM, and Hengartner H. (1991) Peptide-induced antiviral protection by cytotoxic T cells. Proc Nat Acad Sci 88, 991-993. Sensi M, Traversar C, Radrizzani M, Salvi S, Maccalli C, Mortarini R, Rivoltini L, Farina C, Nicolini G, Wolfel T, Bnichard B, Boon T, Bordingnon C, Anichini A, Parmiani G. (1995) Cytotoxic T lymphocyte clones from different patients display limited Tcell receptors variable gene usage in HLA-A2 restricted recognition of MelanA/Mart-1 melanoma antigen. Proc Natl Acad Sci 92, 5674-5678. Sikora K, and Pandha H, (1997) Gene therapy for prostate cancer. Brit J Urol 79, 64-68. Simons JW, Jaffee EM, Weber CE, Levitsky HI, Nelson WG, Carducci MA, Pardoll DM and Marshall FF. (1997) Bioactivity of autologous irradiated renal cell carcinoma vaccines generated by ex vivo GM-CSF gene transfer. Can Res 57, 1537-1544 S l a m o n D , C l a r k G , W o n g S , L e v i n W , U l

l r i c h A , a n d M c G u i r e W . ( 1 9 8 7 ) , H u m a n b r e a s t c a n c e r : c o r r e l a t i o n o f r

e l a p s e a n d s u r v i v a l w i t h a m p l i f i c a t i o n o f t h e H E R 2 / n e u o n g o g e n e . S

c i e n c e 2 3 5 , 1 7 7 1 8 2 . Slamon D, Godolphin W, Jones L, Holt J, Wong S, Keith D, Levine W, Stuart S, Udove J, Ullrich A, and Press M. (1989) Studies of the HER-2/neu protoncogene in human breast and ovarian carcinomas. Science 244, 707-712. Smith, G.L., Mackett, M. and Moss, B. (1983) Infectious vaccinia virus recombinants that express hepatitis B virus surface antigen. Nature 302, 490-495. Spiess PJ, Yang JC, Rosenberg SA. (1987) in vivo antitumor activity of tumor-inď&#x192;&#x17E;ltrating lymphocytes expanded in recombinant interleukin-2. J Natl Cancer Inst 79, 1067-1075. Taylor J, Tartaglia J, Riviere M, Duret C, Languet B, Chappuis G, and Paoletti E. (1994) Applications of canarypox (ALVAC) vectors in human and veterinary vaccination. Dev. Biol. Stand 82, 131-135. Taylor J, Trimarchi C, Weinberg R, Languet B, Guillemin F, Desmettre P, and Paoletti E. (1991) Efď&#x192;&#x17E;cacy studies on a canarypox-rabies recombinant virus. Vaccine 9, 190-193. Toes RE, Blom RJ, Offringa R, Kast WM, Melief C J. (1996) Enhanced tumor outgrowth after peptide vaccination. J Immunol 156, 3911-3918. T o p a l i a n S L , R i v o l t i n i L , M a n c i n

i M , M a r k u s N , R o b b i n s P F , K a w a k a m i Y , R o s e n b e r g S A . ( 1 9 9 4 ) H u m a n C D 4 +

T c e l l s s p e c i f i c a l l y r e c o g n i z e a s h a r e d m e l a n o m a a s s o c i a t e d a n t i g

e n e n c o d e d b y t h e t y r o s i n a s e g e n e . P r o c N a t A c a d S c i 9 1 , 9 4 6 1 9 4 6 5 . Topalian SL, Solomon D, Rosenberg SA. (1989) Tumor speciď&#x192;&#x17E;c cytolysis by lymphocytes inď&#x192;&#x17E;ltrating human melanomas. J Immunol 142, 3714-3724.

Townsend SE, and Allison JP. (1993) Tumor rejection after direct costimulation of CD8+ T cells by B7transfected melanoma cells. Science 259, 368-370. Traversari C, Van der Bruggen P, Luescher IF, Lurquin C, Chomez P,Van Pel A, De Plaen E, Amar-Costesec A, Boon T. (1994) A nonapeptide encoded by human gene MAGE-1 is recognized on HLA-A1 cytolytic lymphocytes directed against tumor antigen MZ2-E. J Exp Med 176, 1453-1456. Tsang KY, Zaremba S, Nieroda CA, Zhu MZ, Hamilton JM, Schlom J. (1995) Generation of human cytotoxic T cells specic for human carcinoembryonic antigen epitopes from patients immunized with recombinant vacciniaCEA vaccine. J Natl Cancer Inst 87, 982-990. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!Ulmer J, Donnelly J, Parker S, Rhodes GH, Felgnes PL, Dwarki VJ, Gromkowski SH, Deck, RR, Dewitt CM, Irredman A (1993) Heterlolgous protection against inuenza by injection of DNA encoding a viral protein. Science 259,1745-1749. Van der Bruggen P, Szikora JP, Boel P, Wildman C, Somville M, Sensi M, and Boon T. (1994) Autologous cytoyitic T lymphocytes recognize a MAGE-1 nonapeptide on melanomas expressing HLA-Cw*1601. Eur J Immunol 24, 2134-2140. Van der Bruggen P, Traversari PC, Chomez P, Lurquin C, DePlaen E, Van den Eynde B, Knuth A, Boon T. (1991) A gene encoding an antigen recognized by cytolytic T lymphocytes on a human melanoma. Science 254, 1643-1647. Vitello A, Ishiaka G, Grey HM, Rose R, Farness P, LaFond R, Yuan L, Chisari FB, Urze J, Bartholomeuz R, Chesnut RW (1995) Development of a lipopeptide-based therapeutic vaccine to treat chronic HBV infection I: induction of a primary cytotoxic T lymphocyte response in humans. J Clin Invest 95, 341-345. Von Kleist S, Burtin P. (1979) Antigens cross-reacting with CEA. In: Immunodiagnosis of Cancer, edited by Herberman RB, McIntire KR. New York: Marcel Dekkar, Inc, p. 322-341. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!Wang RF, Robbins PF, Kawakami Y, Kang XQ, Rosenberg SA. (1995) Identication of a gene encoding a melanoma tumor antigen recognized by HLA-A31-restricted tumorinltrating lymphocytes. J Exp Med 181, 799-804. White DE. (1998) Immunologic and therapeutic evaluation of a synthetic peptide vaccine for the treatment of patients with metastatic melanoma. Nat Med. 4, 321-327. Whiteside TL, Miescher S, Hurlimman J, et al: (1986) Clonal analysis and in situ characterisation of lymphocytes inltrating human breast carcinomas. Cancer Immunol Immunother 23, 169-178. Wolff JA, Ludtke J, Ascadi G, Chong W, Ascadi G, Jani A, and Felgner PL. (1990) Direct gene transfer into mouse muscle in vivo . Science 247, 1465-1468. Wolff JA, Ludtke J, Ascadi G, Chong W, Ascadi G, Jani A, and Felgner PL.(1992) Long term persistence of plasmid DNA and foreign gene expression in mouse muscle. Human Mol Genetics 1, 363-369. Wong C, Morse M, Nair S. (1998) Induction of primary, human antigen-specic cytotoxic T lymphoctyes in vitro using dendritic cells pulsed with peptides. J Immunother. 21, 32-40. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!Yonemura Y, Ninomiya I, Yamaguchi A, Fushida S, Kimura H, Ohoyama S, Miyazakil I, Endou Y, Tanaka M, and Susaki T (1991) Evaluation of immunoreactivity for erbB-2 protein as a marker of poor short term prognosis in gastric cancer. Cancer Res 51, 1034-1038. Yron I, Wood TA, SpiessPJ, Rosenberg SA (1980) in vitro growth of murine T cells V. The isolation and growth of lymphoid cells inltrating syngeneic solid tumors. J Immunol 125, 238-245. Zitvogel L, Mayordomo JI, Tjandrawan T, DeLeo AB, Clark MR, Lofze MT, and Storkus WJ. (1996) Therapy of murine tumors with tumor peptide-pulsed dendritic cells: dependence on T cells, B7 costimulation, and T helper cell associated cytokines. J Exp Med 183, 87-97.

Kelley and Cole: CEA cancer vaccine gene therapy

Gene Therapy and Molecular Biology Vol 2, page PAGE 29

PAGE 30 PAGE 29 Gene Therapy and Molecular Biology Vol 2, page PAGE 23

PAGE 23

Gene Ther Mol Biol Vol 2, 31-40. August 7, 1998.

Vaccine therapy for ovarian cancer using Herpes Simplex virus thymidine kinase (HSVTK) suicide gene transfer technique: a phase I trial 1 1 1 William R. Robinson , Jan Adams , April O'Quinn , and Scott M. 2 Freeman 1

2 1, 2 Department of Obstetrics and Gynecology, Department of Pathology, Tulane Cancer Center, Tulane University School of Medicine, New Orleans, Louisiana. ______________________________________________________________________________________ _______________Corresponding author: William R. Robinson, M.D., 1430 Tulane Ave, TW-40, Tulane Medical Center, New Orleans, Louisiana 70112, Tel: (504) 584-2805; Fax: (504) 584-1805; E-mail: wrobins@tmcpop.tmc.tulane.edu Received 25 April 1998; accepted 28 April 1998

Summary Genetically altered tumor cells expressing the HSV-TK gene have been used as vaccine therapy for multiple cancers, based on their ability to kill adjacent native cancer cells and activate an antitumor immune response. Our in vitro studies demonstrate that transduction with the HSV-TK system confers ganciclovir (GCV) susceptibility to cultured ovarian cancer cells. A murine tumor model was developed using HSV-TK modied ovarian cancer cells to test efcacy in a preclinical setting. Mice bearing intraperitoneal tumors were injected with gene modied cells and ganciclovir (GCV). The mice were evaluated for survival and immune response by analysis of tumor samples collected post treatment. Murine HSV-TK tumors undergo hemorrhagic tumor necrosis and express a cytokine cascade including TNF-, IL-1, IL-6, IL-2, IFN- and GM-CSF following GCV treatment. Tumor regression occurs much less frequently in immune decient mice than immune competent mice. These studies led to a Phase I trial of intraperitoneal administration of the vaccine in which 18 patients with recurrent chemotherapy resistant ovarian cancer were enrolled. The mean survival of patients in the Phase I trial was 11.9 months. 4/18 patients had responses based on physical ndings or CA-125. One patient died of breast cancer with no evidence of ovarian cancer at 24 months. Toxicities include all patients developing grade I or II temperature elevations without other evidence of infection, seven patients who developed grade I abdominal discomfort or nausea, and one patient with a grade III elevation of kidney function tests. We conclude that the use of an HSVTK modied vaccine is associated with tumor regression in mice, and results in the alteration of the tumor microenvironment, which becomes less immunosuppressive. The use of the vaccine in humans is technically feasible and associated with minimal toxicity. Survival in these heavily pretreated patients is similar to that seen using standard cytotoxic chemotherapy.

I. Introduction Ovarian cancer remains the most lethal female genital malignancy in the United States. It will occur in approximately 26,000 women and result in 14,000 deaths in 1997 (Parker, 1997). Progress in the treatment of ovarian cancer has been limited by the inability of physicians to diagnose the disease at an early stage. Signs and symptoms are vague and infrequent, and no effective screening techniques have been identied. Most cases are therefore widely metastatic at diagnosis. As a result, long term survival (20-30%-5 year) in ovarian cancer patients has improved only minimally since 1980, despite improvements in surgical techniques and new chemotherapeutic agents (Morrow, 1993, Venesmaa, 1994). Treatment of ovarian cancer with standard chemotherapy (usually cis- or carbo-platin and paclitaxel) often results in an initial response. However, in most cases the tumor will re-occur within a few months to a few years. The recurrent tumors are frequently resistant to chemotherapeutic agents, and these patients usually succumb to the disease. The cancer generally remains conned to the peritoneal cavity, and death most commonly results from acute or chronic bowel obstruction. In recent years, interest in alternative therapies for ovarian cancer has grown in response to the slow progress associated with standard therapy. This has been accompanied by rapid increases in our understanding of the molecular etiology of the disease. As a result, investigators have begun to utilize gene transfer techniques in a variety of strategies aimed at specic molecular targets. Some of the most promising approaches include compensation/repair of mutations of the host genome, augmentation of the host immune response, and manipulation of drug sensitivity.

A. Compensation/repair of host mutations A variety of malignant tumors have been associated with alterations of certain sequences of the host genome known as oncogenes and anti-oncogenes (or tumor suppressor genes). Ovarian and breast cancers have been associated with overexpression of oncogene Her-2/neu, and loss of function of the tumor suppressor gene p53. In addition, mutations in the tumor suppressor gene BRCA-1 have been identied in association with familial breast and ovarian cancers. Efforts to alter the function of these sequences may be directed at the level of the DNA, messenger RNA, or the protein product. Curiel and Alvarez have developed an adenovirus vector encoding an anti-erbB-2 sFv directed at abrogating the expression of erbB-2. This strategy was effective in reducing tumor burden and increasing survival in a murine model and is currently being investigated in a human trial (Deshane, 1996).

B. Augmentation of host immune response The ability of tumors to develop mechanisms of escape from immune surveillance appears to be an important part of malignant transformation. Immunotherapy has therefore been considered as an alternative to cytotoxic chemotherapy, but preliminary trials have yielded disappointing results (Gall, 1986, Berek, 1985). This relative immunoresistance appears to result from genetic changes in the tumor which allow neoplastic cells to escape from immune surveillance (Whartenby, 1995, Becker, 1993). Genetic manipulations have been utilized in two ways to augment the host immune response to tumors. Tumor inltrating lymphocytes can be isolated and genetically modied to produce specic cytokines that induce a more effective antitumor response. These lymphocytes can, then, be reintroduced to the patient as passive (or adoptive) immunotherapy. In the second approach, a form of active immunotherapy, tumor cells themselves are genetically modied to express cytokines or co-stimulatory molecules that lead to increased recognition and killing by the host. Rosenberg and associates have developed a human trial based on the use of T-lymphocytes modied to express the MOv-y receptor. This receptor is derived from a monoclonal antibody that recognizes an antigen highly expressed in ovarian cancer. The modied T-lymphocytes are then introduced to the patient (Hwu, 1995). Berchuck and Lyerly (1995) have developed a active immunotherapy trial using IL-2 modied tumor cells. Tumor is collected at surgery, modied to express IL-2, and reintroduced to the patient.

C. Manipulation of drug sensitivity Drug sensitivity in tumor cells can be manipulated to induce selective toxicity of tumor cells to an agent produced within the cell or to an introduced agent. In addition, the multi-drug resistance gene (MDR1) has been used in gene transfer experiments. This gene, identied in a number of tumors treated with chemotherapy, confers a chemoprotective effect when transferred to hematopoietic cells, allowing higher, presumably more effective doses of chemotherapy to be used (Champlin, 1994). The trial described in the current report combines the concepts of drug sensitivity manipulation and augmentation of immune function. Gene transfer techniques using the Herpes Simplex Virus-Thymidine Kinase (HSV-TK) gene as a so-called “suicide gene,” have previously been used to augment immune response in a variety of tumor types (Hasegawa, 1993, Caruso, 1993). Cells carrying this gene are susceptible to the anti-viral drug ganciclovir (GCV), and appear to initiate the “bystander effect,” in which nearby unmodied tumor cells are killed as well. Ovarian cancer would appear to be an appropriate target for this type of therapy, as the disease is generally conned to the peritoneal cavity, allowing for effective use of an agent with a local/regional therapeutic effect. This report describes the in vitro and murine studies and subsequent Phase I human trial of a genemodied vaccine using the HSV-TK suicide gene concept for ovarian cancer. In the murine model, soluble factors associated with the immune response are reported, and tumor response to treatment is described. Toxicity and survival data are presented from the human trial and possible mechanisms of action are discussed.

II. Results A. In vitro studies The KBALB (murine brosarcoma), SKOV-3, and PA-1 (both human ovarian tumor) cell lines were used in this study. The retroviral vector LNL6 (Miller and Rosman, 1990) was transduced into the KBALB line, which was then referred to as KBALB-LNL. The retroviral vector STK (Moolten and Wells, 1990) was constructed from the LNL vector and contains an SV40 promoted HSV-TK gene. This

EMBED MSGraph.Chart.5 Figure 1. Colony Formation in cell lines SKOV-3-STK and PA-1-STK exposed to various concentrations of ganciclovir.

vector was transduced into all cell lines, which were designated as KBALB-STK, SKOV-3-STK, and PA-1-STK. Colony counts were performed 10-14 days following exposure of plated SKOV-3, SKOV-3-STK, and PA-1-STK cells to varying concentrations of GCV as described below. Additional cells from these lines were plated concurrently but not exposed to GCV in order to serve as controls. The number of live colonies was expressed as a percentage of maximal colony formation. The maximal toxic effect on the SKOV-3-STK and PA-1-STK cells was similar, and occurred at a GCV concentration of 5 M. (Figure 1).

B. Murine studies 6 Tumors were established subcutaneously (s.c.) in mice by injecting 1x10 KBALB-LNL or KBALB-STK tumor cells alone or in combination. After three days, the mice were randomly assigned to study (treatment) or control groups. Mice receiving treatment were injected with GCV. Murine tumors derived from KBALB-STK cells were harvested at varying time intervals (1, 2, and 4 days) after treatment with GCV for analysis by RT-PCR. Multiple cytokines, including tumor necrosis factor-alpha (TNF), interleukin (IL)-1, IL-6, granulocyte macrophage-colony stimulating factor (GM-CSF), and interferon-gamma(IFN-) were detected. (Table 2). These factors were expressed sequentially in the following pattern: IL-1, TNF, and IL-6- day 1 through day 4; GM-CSF- days 2 and 4; IFN-day 4 only. The ability of the animals to manifest the bystander effect was then tested in relation to immune competency. Immune competent (Balb/C) and immune decient (nude) mice were injected with subcutaneous tumors consisting of various ratios of HSV-TK gene-modied tumor cells and nonmodied 2 tumor cells (0, 50, or 100%). After three days the tumors measured approximately 10 mm . The animals were then treated with GCV and tumor size was measured. Regression occurred in immune competent mice when the tumor was composed of only 50% HSV-TK gene-modied cells. However, regression occurred in only 25% of immune decient mice with tumors consisting of 100% HSV-TK gene-modied cells, and no regression was seen with tumors consisting of 50% HSV-TK gene-modied tumor cells (Table 1)

C. Phase I human trial Eighteen patients were enrolled to the trial. The mean age of the patients was 57.3 years. 15 patients were Caucasian and three were African-American. 17 had Stage III disease at diagnosis, and one had stage II disease. All patients had received either cis- or carboplatin and paclitaxel. Many received a variety of other cytotoxic agents as well. Entry demographic and clinical characteristics of the patients are summarized in Table 1.

% KBALB-STKImmune CompetentImmune Deficient100% (a)100%25%50% (b)100%0%0%0%0%

Mice Injected with 100% KBALB-STK tumor cells Mice Injected with 50% KBALB-STK tumor cells and 50% KBALB tumor cells Mice Injected with 100% KBALB tumor cells Table 1. Regression of KBALB-STK tumor in Immune-Competent vs. Immune Deď&#x192;&#x17E;cient Mice Treated with Ganciclovir.

No patients were removed from therapy for treatment-related toxicity. All patients had temperature elevations during treatment, including 10 grade II and 7 grade I fevers. These episodes occurred within 36 hours of receiving the intraperitoneal vaccine and were not accompanied by other symptoms. All temperature elevations resolved spontaneously with the use of oral acetaminophen. No delayed fevers or other signs of infection or sepsis were noted. Seven patients had grade I abdominal discomfort and/or nausea. This occurred within 30 minutes to one hour of administration of the vaccine, and was usually described as a feeling of bloating. These symptoms resolved spontaneously within 2-3 hours. All patients tolerated a minimum of 1.5 total liters of uid given through the port to optimize distribution. One patient developed grade I shortness of breath, and one patient developed grade I anemia. Both of these ndings resolved spontaneously. One patient developed a grade III renal toxicity based on elevations of her kidney function tests. This patient had a long history of hypertensive disease and mild kidney dysfunction prior to therapy. She was apprehensive about abdominal discomfort during the treatment and had minimal oral intake during the rst 1-2 days of treatment as a result. Her serum creatinine rose to 3.5 mg/dl during this time. Following slow intravenous uid administration, her serum creatinine fell to 1.1 mg/dl over the next 2-3 days. Her urine output remained stable throughout this episode (Table 2). The mean survival for all patients was 11.9 (range 2-26) months. Kaplan-Meier analysis of survival is plotted in Figure 2. Eight patients (BP, ES1, TA, ES2, JS, MH, RC, DM) had evidence of disease progression during or within one month of treatment. Three patients developed pleural effusions, three had rising serum CA-125 levels, and two developed abdominal tumor masses. The remaining ten patients had no evidence of disease progression during treatment. Three (SV, PN, PA) of these patients had mildly elevated CA-125 levels that remained stable during treatment, but rose approximately 3 months later. Three other patients had CA-125 levels <35 at the beginning of treatment. One (EB) died of breast cancer 24 months following treatment. Another (GR) had laparoscopy performed four months after treatment for symptoms of abdominal bloating and was found to have ascites with small volume disease, and the third (MG) has had slowly rising CA-125 levels but remains asymptomatic after 7 months. Four patients had resolution of physical ndings or decreases in CA-125 levels while receiving treatment. One patient (MM), whose CA-125 levels were consistently <10, had resolution of abdominal bloating and ascites during therapy. She remained asymptomatic for one year before relapsing and dying of disease at 23 months. Two patients had falls in CA-125 levels and died of other causes (myocardial infarction (LI) and pulmonary embolus (CM)) approximately 4-6 months following therapy. The nal patient (SK) experienced a fall in CA-125 levels initially followed by a rebound three months after therapy. The CA-125 levels are summarized in Table 3.

III. Discussion Ovarian cancer, like many other malignancies, appears to result from a complex interaction of acquired (and some inherited) genetic rearrangements. Traditional therapies, including surgery and cytotoxic chemotherapy, are directed at cellular reproduction in a very broad manner. This results in signicant toxicity to normal tissues and a variable degree of therapeutic benet, depending on the type and stage of tumor. Gene transfer technology using viral vectors has been greatly rened in the last decade, allowing for much more precisely directed antitumor effects. In view of the limited

Initials Race AgeStage at Diagnosis Histology

ToxicityMax. Dose Admin. (# of cells)CM W50 IIIbendometriodfever, gr. II3x10 8 PA W43 IIIc*pap. serousfever, gr. II3x10 8 BP65 IIIc*pap. serousfever, gr. I

nausea, gr. I10 8 EB W74 IIIb*pap. serousfever, gr. I10 8 ES1 W66 IIIc*pap. serousabd. dis., gr. I**3x10 8 SV W50 IIIc*pap. serousfever, gr. .II10 9 LI W73 IIb*pap. serousfever, gr. I S.O.B.***10 9 TA W50 IIIc*mucinousfever, gr. I

abd. dis., gr. I** ES2 B49 IIIc*pap. serousfever, gr. II10 9 PN W67 IIIc*pap. serousrenal tox., gr. III fever, gr. I3x10 9 MM B51 IIIc*pap. serousfever, gr. II3x10 9 JS W55 IIIc*pap. serous fever, gr. I

abd. dis., gr. I**10 9 MH W71 IIIc*pap. serousfever, gr. I nausea, gr. I3x10 9 GR W42 IIIc*pap. Serousfever, gr.II anemia, gr. I3x10 9 SK W IIIc*pap. Serousfever, gr. I nausea, gr. I10 10 W60 IIIb*pap. serousfever, gr. II

abd. dis., gr. I** RC W49 IIIc*pap. serousfever, gr. I10 9 DM B54 IIIc*pap. serousfever, gr. II3x10 9 * papillary serous ** abdominal discomfort *** shortness of breath

Table 2. Patient characteristics and toxicities.

EMBED MSGraph.Chart.5 \s Figure 2. Kaplan-Meier analysis of survival of patients treated with PA-1-STK gene-modiď&#x192;&#x17E;ed vaccine.

therapeutic benet associated with current treatments, ovarian cancer would appear to be an appropriate target for clinical trials of the introduction of therapeutic genetic material. Suicide gene therapy, as used in this trial, refers to a process in which chemotherapy-resistant tumor cells are modied to express a gene that renders a new drug sensitivity phenotype to the tumor and under appropriate circumstances will be lethal to tumor cells. It has been previously demonstrated that tumor cells transfected with the HSV-TK gene and exposed to GCV can be killed in vitro (Moolten, 1986). Studies of tumor-bearing animals inoculated with HSVTKpositive cells and treated with GCV showed tumor regression as well (Moolten, 1990). The mechanism by which HSV-TK cells cause cell death is phosphorylation of GCV into a toxic nucleotide analogue which functions as a DNA chain terminator by interfering with DNA polymerase activity. In the preclinical data from the current report, we conrm these ndings by demonstrating that tumor regression occurred when tumors were genetically modied to express the Herpes Simplex virus thymidine kinase gene (HSV-TK) and treated with the anti-viral pro-drug ganciclovir. This anti-tumor effect occurred when only a fraction of the tumor expressed the HSV-TK gene. This is the basis of the “bystander effect,” a complex biological process consisting of three interrelated phases: (i) chemosensitization of some tumor cells, (ii) hemorrhagic tumor necrosis caused by release of soluble factors from the dying HSV-TK gene-modied tumor cells, and (iii) generation of an anti-tumor immune response. In phase 1, the transfer of the HSV-TK gene to tumor cells chemosensitizes the tumor cell to GCV. This is accomplished in the human trial by the intraperitoneal administration of non-native tumor cells that have been transduced with HSV-TK. We have previously demonstrated the ability of tumor cells, introduced to the peritoneal cavity, to “home to” native tumor deposits (Freeman, 1994). The introduced cells in effect become part of the native tumor by their proximity, and thereby sensitize the tumor to GCV. This initiates phase 2, the generation of a generalized hemorrhagic tumor necrosis (HTN) produced by the release of soluble factors from the dying HSV-TK gene-modied tumor cells. The HTN leads to disruption of the tumor blood supply, and thus loss of nutrients which leads to killing of the majority of tumor cells. The nal phase of this process, cytokine production by the dying HSV-TK cells, appears to result in the death of any remaining tumor cells by initiating a host immune response. We have previously demonstrated that HTN stimulates a cellular immune response, leading to a lymphocytic inltration of the tumor (Freeman, 1994), and results in upregulation of a variety of co-stimulatory molecules including ICAM-1, B7-1, and B7-2 (Ramesh, in press). To summarize, we hypothesize that cytokine production by the HSV-TK cells results in the transformation of the tumor microenvironment from immunoresistant to immuno-

Patient InitialDuring TherapyCompletion of TherapyCM16335 31PA120112114BP3854461,588EB10712ES11,3572,710SV308349473LI1197335TA7990398ES2815060677257MM172015JS144237MH600609749GR1111 7SK311158103MG251724233426DM4971,282stimulatory, allowing tumor inltrating lymphocytes, generated by HTN, to have a lethal effect on any remaining viable tumor cells. Thus, a chemotherapy-resistant tumor can undergo complete regression if only a fraction expresses the HSV-TK gene because the augmented host immune response acquires the ability to eradicate any residual cells. Supporting the role of the host immune system in this mechanism, the current data demonstrates that immune decient mice had some anti-tumor effect following treatment with the HSV-TK system, but far less than immune competent mice. We have also demonstrated in the current data that the killing effect of

GCV can be achieved in vitro at a concentration of 5.0 M. Human pharmacokinetic data show that intravenous administration of GCV at the recommended therapeutic dose of 5mg/kg easily exceeds this level (Faulds and Heel, 1990). The results of the human trial reported here demonstrate that intraperitoneal vaccine therapy is technically feasible in this setting. Administration of the vaccine was tolerated by all patients, and no technical problems with the use of the intraperitoneal ports were encountered. The only consistently seen toxicity was fever, which occurred in all patients, but was mild and well tolerated. Nausea and abdominal discomfort associated with infusion of the vaccine diluent were seen in approximately one third of the patients. We found that mild sedation with a benzodiazepene such as Ativan or Valium immediately prior to infusion eliminated these symptoms. As would be expected, survival in these heavily pretreated, presumed chemotherapy-resistant patients was poor. Objective response data was difcult to interpret, as we did not require post treatment histologic verication of the presence of tumor. Most of these patients had undergone multiple surgical procedures and were understandably reluctant to agree to additional operative intervention. Estimation of tumor response was therefore based on physical ndings and CA-125 results. 4/18 patients had some evidence of response, based on these criteria, with a mean survival of just under one year. We feel this is consistent with results from studies of cytotoxic chemotherapy in the salvage setting (Vergote, 1992, Eisenhauer, 1994, Creemers, 1996). In summary, suicide gene therapy as described here results in signicant tumor regression in vitro and in murine studies. The mechanism of action appears applicable to human ovarian cancer based on these studies as well. A Phase I human trial demonstrates that this method of therapy is feasible and well tolerated. Patient survival is similar to that seen using standard cytotoxic chemotherapy. A Phase II trial is indicated to more accurately estimate disease response.

IV. Material and methods A. Cell lines and retroviral vectors The KBALB, (murine brosarcoma) SKOV-3, and PA-1 (both human ovarian tumor) cell lines were used in this study. They were obtained from American Type Culture Collection (ATCC, Rockville, MD) and maintained as described elsewhere (Freeman, 1993). The retroviral vector LNL6 (Miller and Rosman, 1990) was transduced into the KBALB line, which was then referred to as KBALB-LNL. The retroviral vector STK (Moolten and Wells, 1990) was constructed from the LNL vector and contained an SV40-promoted HSV-TK gene. This vector was transduced into all cell lines, which were designated as KBALB-STK, SKOV-3-STK, and PA-1-STK. 3 To determine the in vitro sensitivity of these cell lines to GCV, 10 cells from lines SKOV-3, SKOV-3-STK, and PA-1-STK were plated separately and exposed to concentrations of GCV of either 0, 0.005, 0.05, 0.5, 5.0, or 50 M. 10-14 days later the plates were stained with methylene blue and colonies were counted.

B. Murine studies Female BALB/c mice (Charles River Laboratories, Wilmington, MA) obtained at 5-6 weeks of age were maintained pathogen-free according to established guidelines. BALB/C athymic nude mice (nu/nu) were also obtained from the same source. Tumors were established subcutaneously (s.c.) in all mice by injecting 6 1x10 KBALB-LNL or KBALB-STK tumor cells alone or in combination using a 26-gauge needle. After three days, the mice were randomly assigned to study (treatment) or control groups. Mice receiving treatment were injected with GCV twice a day, for 5-10 doses (150 mg/Kg). Animals not used for survival studies were sacriced on days 1, 2 and 4 after initiation of GCV treatment. Visible tumors were isolated 0 aseptically under sterile conditions, snap frozen, and stored at -70 C. Reverse transcriptase polymerase chain reaction (RT-PCR) was then performed using RNA extracted from the tumors as described elsewhere

(Freeman, 1995a) using RNAzol B (Biotecx Laboratories, Houston, TX). The nal concentration of the extracted RNA was adjusted to 1mg/ml. First strand complimentary DNA (cDNA) was synthesized from total RNA by reverse transcription using 50 picomoles of 3' downstream primer (antisense) for each of the cytokines (TNF, IL-1, IL-2, IL-4, IL-6, IL-10, IFN- and GM-CSF) tested and PCR performed (Freeman, 1995a). The amplied PCR product was detected by agarose gel electrophoresis and conrmed by Southern hybridization.

C. The phase I human trial All patients had a histologically proven diagnosis of ovarian cancer with clinical evidence of recurrent, progressive or residual disease conned to the peritoneal cavity following treatment with combination chemotherapy to include cis-platin or carboplatin and paclitaxel. Southwest Oncology Group (SWOG) performance status for all patients was 0-1. At least six weeks had to have passed since the most recent exposure to chemotherapy, and patients could not have tumor masses larger than 2 cm prior to treatment. Tumor size and location were determined by surgery and/or imaging study. The Tulane Instituitional Review Board and the Food and Drug Administration reviewed and approved this trial. All federal, state and institutional regulations regarding consent were fullled. All patients underwent placement of an intraperitoneal port-a-cath type device. The reservoir of the port was placed in the subcutaneous fat on the chest wall below the breast and secured with permanent sutures. The silicon tubing was tunneled through the subcutaneous tissue of the abdominal wall and inserted through the fascia and peritoneum approximately 3 cm lateral to the umbilicus. The tubing was secured to the abdominal wall fascia with absorbable suture at this point. The intraperitoneal portion of the tubing, including the fenestrated section, was directed into the pelvis. Complete access to the peritoneal cavity was veried at this time. Any adhesions which obstructed access were dissected. The gene-modied human ovarian cancer cell line PA-1-STK was selected for use as the vaccine and tested for contaminants including bacteria, fungi, and viruses. PA-1-STK cells were lethally irradiated prior to administration to eliminate any intrinsic oncogenic potential of the vaccine. The survival of the irradiated vaccine cells was 3-4 days in vitro. The PA-1-STK cells were also tested for and found to be free of replication-competent virus prior to administration (Freeman, 1995b). 1. Study design This Phase I study was designed as an escalating dose trial to determine the maximally tolerated dose (MTD) of use of the PA-1-STK cell line as a vaccine for ovarian cancer. The objectives were: (i) to evaluate the safety and side effects of the treatment, (ii) to determine the technical feasibility of intraperitoneal vaccine administration activated by intravenous ganciclovir, and (iii) to observe for clinical effects on the cancer. All toxicities were graded according to the National Cancer Institute common toxicity criteria. Disease status was documented using physical examination and serum CA-125 levels. Patients were followed until death and survival was plotted using a Kaplan-Meier survival curve. 2. Treatment plan The maximum cell dose each patient received is listed in Table 1. Treatments were planned on 21 day cycles for three total treatments. The dose escalated with each treatment unless any grade 3 or higher toxicity occurred. For grade 3 or 4 toxicity, the dose was not elevated If these grade 3 or 4 toxicities did not resolve within one week, the patient was taken off study. If more than one grade 3 or 4 toxicity occurred in any group, the next lower dose level would be considered the MTD. The PA-1-STK cells were suspended in 500 cc of normal saline and administered through the intraperitoneal port. Additional normal saline (maximum-1500 cc) was then administered to patient tolerance through the port to assure optimal distribution. Ganciclovir was administered intravenously beginning no more than one hour following administration of the vaccine at a dose of 5mg/kg in patients with a creatinine clearance (CrCl) >80. Patients whose CrCl was 50-79 received 2.5mg/kg, and those with

CrCl <50 were excluded. Ganciclovir was given twice daily for seven days following each treatment with the vaccine.

Acknowledgements This work was supported in part by the Brennan Oncology Fund, and by grant #5M01RR05096-06 from the Division of Research Resources, National Institutes of Health.

References Becker JC, Brabltz T, Czerny C, Termeer C, Brocker EB (1993) Tumor escape mechanisms from + immunosurveillance: induction of unresponsiveness in a specic MHC-restricted CD4 human T cell clone by the autologous MHC class II+ melanoma. Int Immunol 5, 1501-1508. Berchuck A and Lyerly H (1995) A phase I study of autologous human interleukin 2 (IL-2) gene modied tumor cells in patients with refractory metastatic ovarian cancer. Human Gene Transfer Protocols, Ofce of recombinant DNA activities. Bethesda, Maryland, National Institutes of Health. Berek JS, Hacker NF, Lichenstein AK, Jung T, Spina C, Knox RM, Brady J, Greene T, Ettinger LM, Lagasse LD (1985) Intraperitoneal recombinant a-interferon for salvage immunotherapy in stage III epithelial ovarian cancer: a Gynecologic Oncology Group study. Cancer Res 45, 4447-4451. Caruso M, Panis Y, Gagandeep S, Houssin D, Salzmann JL, Klatzmann D (1993) Regression of macroscopic liver metastases after in situ transduction of a suicide gene. Proc Natl Acad Sci USA 90, 7024-7028. Champlin F, Kavanaugh J, Deisseroth A (1994) Use of safety-modied retrovirus to induce chemotherapy resistance sequences into normal hematopoietic cells for chemoprotection during the therapy of ovarian cancer:A pilot trial. Hum Gene Ther 5, 1507-1522. Creemers GJ, Bolis G, Gore M, Scarfone G, Lacave AJ, Guastalla JP, Despax R, Kreinberg R, Van Belle S, Hudson I, Verweij J, Ten Bokkel H (1996) Topotecan, an active drug in the second-line treatment of epithelial ovarian cancer:results of a large European Phase II study. J Clin Oncol 14, 3056-3061. Deshane J, Loechel F, Conry R, Siegal G, Alvarez R, King C, Curiel DT (1996) Intracellular antibody directed against erbB-2 mediates targeted tumor cell eradication by inducing apoptosis. Cancer Gene Ther 3, 89-98. Eisenhauer EA, Wim W, Ten Bokkel H, Swenerton KD, Gianni L, Myles J, Van Der Berg MEL, Kerr I, Vermoken JB, Buser K, Colombo N, Bacon M, SantabarbaraP, Onetto N, Winograd B, Canetta R (1994) European-Canadian randomized trial of palitaxel in relapsed ovarian cancer:high dose versus low dose and long versus short infusion. J Clin Oncol 12, 2654-2666. Faulds D and Heel RC (1990) Ganciclovir: a review of its antiviral activity, pharmacokinetic properties, and theraputic efcacy in cytomegalovirus infections. Drugs 39, 597-638. Freeman SM, Abboud AN, Whartenby KA, Packman CH, Koeplin DS, Moolten FL, Abraham GN (1993) The “bystander efect”: tumor regression when a fraction of the tumor mass is genetically modied. Cancer Res 53, 5274-5283. Freeman SM, Mcune C, Robinson WR, Abboud CN, Abraham GN, Angel C, Marrogi A (1995) The treatment of ovarian cancer with a gene modied cancer vaccine: A phase I study. Human Gene Ther. 6, 927-939. Freeman SM, Ramesh R, Marrogi AJ, Jensen A, Abboud CN (1994) In vivo studies on the mechanism of the bystander effect. Cancer Gene Ther 1, 326-332. Freeman SM, Ramesh R, Shastri M, Munshi A, Jensen AK, Marrogi AJ (1995) The role of cytokines in mediating the bystander effect using HSV-TK xenogeneis cells. Cancer Letters. 92, 167-174. Gall S, Bundy B, Beecham J, Whitney C, Homesley H, Lifshitz S, Adcock LL (1986) Therapy of stage III (optimal) epithelial carcinoma of the ovary with melphalan or melphalan plus corynebacterium parvum (a Gynecologic Oncology Group study) Gynecol Oncol 25, 26-30. Hasegawa Y, Emi N, Shimokata K, Abe A, Kawabe T, Hasegawa T, Kirioka T, Saito H (1993) Gene transfer of herpes simplex virus type I thymidine kinase gene as a drug sensitivity gene into human lung cancer cell lines using retroviral vectors. Am J Respir Cell Mol Biol 8, 655-661. Hwu P, Yang J, Cowherd R, Treisman J, Shafer J, Eshhar Z, Rosenberg SA (1995) In vivo antitumor activity of Tcells redirected with chimeric antibody/T-cell receptor genes. Cancer Res 55, 3369-3373. Miller AD, Rosman GJ (1990) Improved retroviral vectors for gene transfer and expression. Biotechniques 7, 980-985. Moolten FL (1986) Tumor chemosensitivity conferred by inserted herpes simplex thymidine kinase genes. Cancer Res 46, 5276-5281. Moolten FL (1990) Mosaicism induced by gene insertion as a means of improving chemotheraputic selectivity

Crit Rev Immunol 10, 203-233. Moolten FL and Wells JM (1990) Curability of tumors bearing herpes thymidine kinase genes transferred by retroviral vectors. J Natl Cancer Inst 82, 297-300. Morrow CP, Curtain JP, Townsend DE (1993) Synopsis of Gynecologic Oncology, 4th ed. New York: Churchill Livingstone, 7. Parker SL, Tong T, Bolden S, Wingo PA (1997) Cancer Statistics, 1997. CA-A Cancer Journal for Clinicians. 47, 5-27. Ramesh, R., A. Munshi, C.N. Abboud, A.J. Marrogi and S.M. Freeman. Expression of co-stimulatory molecules: B7 and ICAM upregulation following treatment with a suicide gene. Cancer Gene Therapy, in press. Venesmaa P (1994) Epithelial ovarian cancer: Impact of surgery and chemotherapy on survival during 1977-1990. Obstet Gynecol 84, 8-11. Vergote I, Himmelmann A, Frankendal B, Scheistroen M, Vlachos K, Trope C. (1992) Hexamethylmelamine as second line therapy in platin-resistant ovarian cancer. Gynecol Oncol 47, 282-286. Whartenby, K.A., C.N. Abboud, A.J. Marrogi, R. Ramesh and S.M. Freeman (1995) Biology of Disease: The Biology of Cancer Gene Therapy, Lab. Invest., 72: p. 131-145.

Gene Ther Mol Biol Vol 2, 41-58. August 7, 1998.

Exploiting stromal-epithelial interaction for model development and new strategies of gene therapy for prostate cancer and osteosarcoma metastases (review) 1 1 1 Thomas A. Gardner , Song-Chu Ko , Chinghai Kao , Toshiro 1 1 1 1 Shirakawa , Jun Cheon , Akinobu Gotoh , Tony T. Wu , Robert A. 1 1 2 2 Sikes , Haiyen E. Zhau , Quajun Cui , Gary Balian and Leland W. 1 K. Chung 1

2 Molecular Urology and Therapeutics, Department of Urology, Orthopedic Research, Department of Orthopedics, The University of Virginia Health Sciences Center, Charlottesville, VA 22908 ______________________________________________________________________________________ ________________ Corresponding Author: Leland W. K. Chung, Ph.D., Molecular Urology and Therapeutics, Department of Urology, HSC Box 422, The University of Virginia Health Sciences Center, Charlottesville, VA 22908, Tel: 804-243-6512; Fax: 804-243-6648. Received 15 May 1998; accepted 20 May 1998.

Results of toxic gene therapy for the treatment of localized and disseminated prostate cancers showed that: (i) Ad-OC-TK expressed high levels of TK in both androgen-dependent and androgen-independent human prostate cancer cell lines; (ii) in parallel with the expression of AdOC-TK in tumor cell lines, the efcacy of Ad-OC-TK toxic gene therapy in target cells is directly correlated with the levels of TK expression in vitro; (iii) in two experimental models of human prostate cancer, C4-2 and PC-3, we demonstrated that Ad-OC-TK, when applied together with ACV, induced tumoricidal effects in vivo. Signicant histomorphologic improvement of human prostate cancer growth in the bone was supported by bone scans in vivo. In the C4-2 model, we obtained evidence that Ad-OC-TK plus ACV diminished serum PSA, which is conrmed by the improvement of histomorphologic appearance of this tumor in the skeleton. Finally, we have focused our effort in the development of combined adenovirus and chemotherapy (i.e. chemogene therapy), the development of a concept of loco-regional delivery of therapeutic genes and drugs, and the exploration of using the homing mechanism to treat prostate cancer skeletal metastasis in vivo. Taking advantage of the reciprocal cellular interaction between prostate cancer and bone stroma, we have developed two novel gene therapy approaches to target prostate cancer growth in the bone. We have achieved for the rst time the use of Ad-OC-TK/ACV as a novel therapeutic agent that can selectively target and induce the killing of both prostate and osteoblast lineage cells.

I. Introduction Molecular therapeutic strategies such as gene therapy are being used with increasing frequency. The exponential expansion of knowledge in the eld of molecular medicine has led to therapy that is based on understanding the molecular events underlying a disease process. Currently, molecular based gene therapy protocols are used predominately for life-threatening diseases like cystic brosis (Boucher et al., 1994; Crystal R.G., 1994), ADA (Blaese R.M., 1995), and cancer (Sanda et al., 1994). Such approaches will rapidly expand into other areas of medicine in the near future. To understand the uses of gene therapy for the treatment of both localized and metastatic prostate cancer and osteosarcoma, our laboratory focused on the development In this review, we will discuss the concepts and the models developed in our laboratory to study the molecular mechanism underlying human prostate cancer progression and metastasis. To understand the molecular basis of bone stromal cell targeting, we also established an osteosarcoma metastatic model in a rodent inoculated with either human or rat osteosarcoma cells. The models will be used as targets for in vivo gene therapy by delivering therapeutic toxic genes using a tissue-specic promoter (Ko et al., 1996). The ability to combine adenoviral gene therapy and chemotherapy, such as the development of chemogene therapy strategy (Cheon et al., 1997) and the systemic delivery of adenoviruses for the treatment of prostate cancer skeletal metastases and osteosarcoma pulmonary metastases will be discussed (Shirakawa et al., 1998). Finally, we will discuss the use and development of an ex-vivo gene therapy that utilizes stably transduced bone stromal cells to deliver the toxic genes and their by-products to the site of prostate cancer skeletal metastasis (Gardner et al., 1998).

II. Prostate cancer growth and metastasis: model development A. Introduction According to 1997 Cancer Statistics, it is predicted that prostate cancer diagnosis will alter the daily life of a man every 3 minutes, and end the life of another man every twelve minutes (Parker et al., 1997). The majority of the morbidity and mortality from this disease is caused by androgen-independent progression, in which tumors develop a metastatic phenotype after an unpredictable period of androgen ablation. The most common anatomical site for these metastases is bone (Franks, 1956). Numerous therapeutic options are under investigation and progress has been made, but none have demonstrated signicant advantages in improving patient survival. New models to study novel therapeutic approaches toward the treatment of metastatic disease are needed. Cancer progression is a multi-step process involving initiation, promotion, and progression, which often are difcult to study directly in human patients and their tissues. For this reason, most of the literature describing human cancer development often cite examples from epidemiological surveys or accidental exposure of human populations to drugs (e.g. diethylstilbestrol and thalidomide), radiation (e.g. the Hiroshima radiation exposure), and carcinogens (e.g. aryl hydrocarbons). Fortunately, signicant conservation of genes and responses to drugs, chemicals, and hormones allows a close monitoring of cancer progression in animal models which facilitated research and achieved signicant levels in understanding the molecular basis of these processes. The benet of understanding the multi-steps of carcinogenesis is apparent. For example, in prostate cancer, an individual’s prostate may undergo initiation of carcinogenesis in his 40’s, but the actual progression and manifestation of clinical diseases will only become apparent in his mid-70’s. This signicant lag time could be a great opportunity for therapeutic intervention, which could signicantly reduce the mortality and morbidity of patients who are predisposed to prostate cancer development. Moreover, understanding the molecular steps of cancer progression could also allow us to assess more accurately the natural history of the disease and hence improve strategies for treating and preventing the disease processes.

B. Osteosarcoma model simulates aberrant osteoblastic growth.

The terminal form of prostate cancer involves the development of an androgen-independent metastatic disease. The morbidity and mortality are derived mostly from the osseous metastases that occur at the end stage of prostate cancer (Franks, 1956). Unlike the majority of bone metastases from other predominantly osteolytic tumors, prostate cancer bone metastases are osteoblastic (Scher and Chung, 1994). Prostate cancer promotes bone deposition and growth where many other cancers promote bone resorption and destruction. Interaction between prostate cancer cells and bone stroma appears to be reciprocal (Chung and Cunha, 1983; Cunha and Chung, 1981; Djakiew et al., 1966), maintaining a symbiotic relationship. To destroy this reciprocal interaction, we focused on the possibility of establishing a condition where individual components of prostate cancer metastasis, i.e. the osteoblastic lesion and prostate epithelium, can be studied separately. With respect to the osteosarcoma model, we used a rat osteosarcoma 17-2.8 (ROS) cell line and a human osteosarcoma (MG-63) cell line as the starting material for subcutaneous inoculation, which consistently formed osteosarcoma in vivo. These cell lines can also be readily grown in vitro to study the molecular interaction between the transduced therapeutic genes and osteosarcoma growth. ROS cells, when injected intravenously in syngeneic animals, formed pulmonary metastases. These characteristics of the ROS cells have been used as a model to study the effect of gene therapy alone and a combination of gene and chemotherapy for the treatment of osteosarcoma pulmonary metastasis (Cheon et al., 1997).

C. LNCaP progression model mimics human androgen-independent prostate cancer progression Dr. Huggins (Huggins and Hodges, 1941) rst demonstrated therapeutic intervention in localized and metastatic prostate cancer by manipulating the host hormonal status through surgical castration and/or administration of an androgen antagonist, diethylstilbestrol. The original discovery that androgen deprivation could lead to symptomatic responses in patients with prostate cancer led to a Nobel Prize for Dr. Huggins. One of the difculties in understanding the tumor biology of prostate cancer is the insidious nature of this cancer, which grows virtually unnoticed and only shows itself symptomatically in the very late stages of the disease. To understand the molecular and cellular basis of androgen-independent progression, our laboratory has developed a mouse model of human prostate cancer progression. In this model, we observed that a marginally tumorigenic LNCaP cell line, when co-inoculated with a human nontumorigenic osteosarcoma cell line, MS, consistently formed PSA-producing tumors in vivo (Thalmann et al., 1994). Upon castration of the athymic hosts, the tumors undergo androgen-independent progression by enhanced proliferation and increased PSA production in the absence of testicular androgen (Thalmann et al., 1994). By employing bone stroma cells as inductors in “selecting” or “inducing” the parental LNCaP cells in castrated animals to acquire androgen independence and metastatic potential, we developed a LNCaP subline, C4-2, which grew in castrated hosts and metastasized to the bone in castrated male hosts. C4-2 cells are an attractive model to study prostate cancer progression for the following reasons: (i) C4-2 cells are capable of growing and metastasizing in castrated athymic mice. (ii) C4-2 cells share a cell lineage relationship with parental LNCaP cells. This cell-lineage relationship allows the detailed biochemical and molecular analysis of genotypic and phenotypic changes of cells during disease progression. (iii) C4-2 cells produce PSA and contain androgen receptor (AR), which allows the study of ligand-dependent and -independent regulation of gene expression in human prostate cancer cell lines both in vivo and in vitro. (iv) The C4-2 cell line exhibits many biochemical and molecular characteristics resembling human prostate cancer. For example, C4-2 cells, when metastasized to bone or when injected intraosseously, produce an osteoblastic response. C4-2 cells produce a protein factor, prostate-specic antigen (PSA)-stimulating autocrine factor (PSAF), which induced human prostate cancer cells to synthesize and secrete PSA (Hsieh et al., 1993). The biological activity of this factor was found to be present in human bone marrow aspirate obtained from men with androgen-independent disease. This attractive model allows us to examine the molecular events that regulate prostate growth and gene expression and to design and test various forms of therapeutic modalities in a pre-clinical model of human prostate cancer metastasis.

One drawback of this in vivo model of prostate carcinogenesis is that the latent period between tumor cell inoculation and the actual development of solid skeletal tumor nodules (e.g. orthotopic or subcutaneous administration of C4-2 cells in athymic mice will take a mean of 6.8 months prior to the observation of tumor metastasis to the skeleton (Thalmann et al., 1994). Recently, we demonstrated that intraosseous administration of tumor cells (C4-2) to athymic mice form reproducibly PSA-secreting tumors in vivo. The parental LNCaP cells injected similarly failed to form tumors in vivo (Wu, 1998). Histomorphologic observation reveals that the tumors formed in the bone appear to be osteoblastic and stain positively by PSA antibody. Using this intraosseous injection of tumor cells in vivo as a model, we found no correlation between serum PSA and circulating prostate cancer cells, as detected by RT-PCR of PSA mRNA (Wu, 1998). This model established for the rst time rapid prostate cancer growth in the bone with a conrmed osteoblastic reaction, which can be used to study the pharmacokinetic relationship with circulating cells in the blood, as well as to study prostate cancer gene therapy.

D. Subcutaneous osseous prostate cancer growth model Stromal epithelial interactions are vital to the development of the prostate gland and the maintenance of homeostasis in the growth and gene expression of the prostate gland (Chung et al., 1993). To study this interaction, we developed a cellular model of human prostate cancer growth and its androgen-independent progression (see above). While this model provides an opportunity to study prostate cancer-bone stromal interaction, the elicited osteoblastic responses in bone sometimes are difcult to discern. Recently, we (Gardner et al., 1998) showed that a mouse pluripotent bone stromal cell line, D1 (Cui Q., 1997; Diduch D.R., 1993), when co-inoculated with an androgen-independent human prostate epithelial cell line C4-2 formed a radio-opaque osteoblastic tumor subcutaneously in athymic mice (Figure 1). This exciting observation established for the rst time an osteoblastic growth of human prostate cancer as subcutaneous deposits. This model provides a mechanism to study molecular events governing the development and maintenance of prostate cancer osseous metastasis. This model was used to study “bystander” cell-kill using drug and/or gene therapy (see below). This model is the rst instance of establishing an osteoblastic human prostate metastasis in the subcutaneous space of an animal. The benet of this new model is that the chimeric tumor can be easily x-rayed, and the response of the tumor to various therapies can be monitored accurately and conveniently. The chimeric nature of the mouse bone stromal cell and the human prostate cancer cell allows us to evaluate the types of autocrine/paracrine growth factors and extracellular matrices derived from each cellular compartment and their actions in conferring growth and differentiation signals to the prostate gland.

Figure 1. Radiographic and histological appearance of subcutaneous osseous metastasis human prostate cancer model. Coinoculation of a mouse bone stromal cell (D1) and androgen-independent human prostate cancer (C4-2) revealed osteoblastic lesion demonstrated by x-ray and histology.

E. Summary The development of in vitro and in vivo models for studying human disease is vital to understanding the molecular mechanisms leading to disease. These model systems can allow the testing of hypotheses, but clinical trials remain the gold standard for the efcacy of a therapy. The elucidation of the molecular mechanism underlying each of the disease states (e.g. prostate cancer, osteosarcoma) will allow the expedient development of novel molecularly-based therapies which can be tested and modied subsequently at the stages of pre-clinical trials. Model systems will facilitate the more rapid development of experimental therapeutics which ultimately will be applied clinically with the potential of curing prostate cancer and its distant spread.

III. Gene therapy approaches to cancer A. Introduction The term ‘Gene Therapy’ in its simplest denition refers to the therapeutic application of genetic materials. Two prototype gene therapy protocols have been chosen for clinical or pre-clinical evaluation for the treatment of cancer. The rst strategy is corrective gene therapy. This involves either replacement of defective genes or inactivation of activated genes in neoplastic cells to restore normal growth control pathways (Boulikas, 1997; Gotoh et al., 1997; Ko et al., 1996). The second strategy is ablative or cytoreductive gene therapy, which is based on the targeted destruction of malignant cells (Bonnekoh et al., 1995; Cheon et al., 1997; Ko et al., 1996; Shirakawa et al., 1998; Tanaka T., 1996; Trinch et al., 1995). Ablative or cytoreductive gene therapy involves the delivery of gene(s) to target cells that catalyses cell-kill or arrest of cell cycle progression through metabolic activation of prodrugs or direct interference with cell survival (Cheon et al., 1997; Gotoh et al., 1997; Ko et al., 1996). The key components of a gene therapy approach include, but are not limited to: (i) the selection of genetic materials which are comprised of therapeutic genes; (ii) the appropriate tissue-specic or universal promoters, where in some instances the promoters may be inducible by a heavy metal, a hormone, or an antibiotic; (iii) the appropriately selected vectors, such as retrovirus (e.g. Moloney leukemia virus and lente virus), adenovirus, adeno-associated virus, liposomes and/or naked DNA; and (iv) the appropriate route of delivery, such as aerosol, intralesional injection, loco-regional perfusion, or systemic administration. Our laboratory has focused on an adenoviral system to deliver therapeutic genes for the treatment of prostate cancer in pre-clinical models. The model systems described above have allowed us to develop and evaluate new therapeutic approaches for the treatment in particular of androgen-independent prostate cancer. Below is a summary of our use of a tissue-specic promoter-directed expression of therapeutic gene, applied alone or in combination with chemotherapy, for the treatment of both localized and disseminated prostate cancer and osteosarcoma in experimental models.

B. Rationale of adenoviral approach for cancer gene therapy The adenovirus has many attractive features for the treatment of cancer, such as its aptitude for infecting a wide range of cell types irrespective of their status of cell cycle progression. Its high infectivity of epithelial cells made this form of virus particularly attractive for the treatment of cancer. The adenovirus is well suited for ablative gene therapy because of the following: (i) the adenoviral genome is well known and is capable of incorporating large foreign genes into the vector; (ii)

adenovirus is not incorporated into the host genome, and thus functions as an episome with much reduced host genome toxicity; and (iii) adenovirus is highly infectious to both dividing and non-dividing cells. Currently, the DNA size limitation of the E1-deleted adenovirus is approximately 7.5 kb (Graham and Prevec, 1991), but as more complete deletion vectors are constructed (e.g. "gutless" version of adenovirus), the DNA size that could be accommodated into adenovirus could be enhanced up to 38 kb. The adenoviral vector also has limitations. The adenoviral proteins can cause a host immune response. This can be benecial to cancer gene therapy, causing a vaccine-like immune response to the tumor, but this immunity also limits the ability for the adenovirus to exert itself over a long period because of the mounting host immune rejection of these foreign adenoviral proteins. The development of neutralizing antibodies can block an initial host response and improve the therapeutic efcacy of this treatment. Current attempts to delete most of the adenoviral genome from the adenovirus will overcome this problem in part. Due to its transient nature of expression in cells as an episome, adenovirus is not the appropriate choice for long-term applications, such as the use of gene therapy for corrective purposes.

C. Vector designs and modes of action of toxic genes The adenovirus subtype 5 has been modied with an E1 deletion (Graham and Prevec, 1991) to allow the insertion of the desired expression cassette. The expression cassette contains a xed region that allows a homologous recombination event to occur in 293 cells and a variable region used to insert a desired DNA sequence. Because this recombination event has deleted the E1a region of the adenoviral genome, the recombinant adenovirus becomes replication-defective. There are two parts of the expression cassette virus that can be engineered: First, the promoters that regulate the transcription of downstream genes. Second, the therapeutic gene(s) of interest which are regulated by the promoters. In the rst category we have employed prostatic-specic antigen (PSA) and osteocalcin (OC) promoters as tissue-restrictive promoters for the delivery and expression of therapeutic genes in target prostatic cancer cells. Because of the highly specic expression of PSA and OC proteins by prostate cancer cells, the promoters of these genes are suitable candidates for the delivery and expression of therapeutic genes in prostate cancer cells. Our studies showed that: (i) PSA (Gotoh et al., 1998) or OC (Cheon et al., 1997; Ko et al., 1996; Shirakawa et al., 1998) promoter-mediated expression of therapeutic genes are equivalent to those mediated by universal promoters, such as CMV and RSV, except that the expression of genes are highly regulated in a tissue and tumor-restricted manner. (ii) Delivery of toxic genes such as thymidine kinase (TK) and cytosine deaminase (CD) genes that are capable of exerting “bystander” effects against the tumor cells have achieved signicant direct as well as “bystander” cell-kill. The molecular mechanisms of TK and CD are as follows. Upon TK expression, this form of enzyme will be able to convert a prodrug, ganciclovir (GCV) or acyclovir (ACV), into a biologically active drug, phosphorylated GCV, which is incorporated into an elongated DNA strand, and interrupts DNA synthesis and causes early chain termination, initiating an apoptotic process. Similarly, CD gene can convert 5uorocytosine (5-FC) into 5-uorouracil (5-FU), which can be incorporated into cellular RNA, and interrupts RNA synthesis in both dividing and non-dividing cells. 5-FU is also considered an inhibitor for TK, thus also interrupting cellular DNA synthesis. Both TK and CD genes are known to exert bystander effects on their neighboring cells. It is proposed that the phosphorylated form of GCV or acyclovir (ACV) can be transported to neighboring cells through gap junctions and interrupt DNA synthesis in neighboring cells. Unlike TK, the CD gene converts 5-FC into 5-FU, which is readily diffusible and can serve as a

direct toxin to neighboring cells without the necessity of gap junctional transfer. Both of these strategies have been demonstrated as efcacious in causing regression of a number of solid tumors, including metastatic colon carcinoma to the liver, gastric carcinoma, and malignant mesothelioma.

D. Viral production and delivery. Recombinant adenovirus containing the selected expression cassette (PSA-TK, OC-TK, CMV-TK, etc.) is produced by co-transfecting a shuttle vector containing the expression cassette (e.g. p!E1sp1B-PSA-P-TK ) and recombinant adenoviral vector (pBHG-11) plasmids in a human fetal kidney 293 cell line, as described (Graham and Prevec, 1995). Recombinant adenovirus was cloned from individual plaques, amplied, and puried by the CsCl centrifugation method. The virus stock was then dialyzed, concentrated, and titered. The plaque-forming unit (PFU) of the viruses was measured by a standard biologic plaque forming assay (Graham and Prevec, 1991). A number of methods of viral delivery can be implemented. For example, adenovirus can be injected intralesionally (Cheon et al., 1997; Ko et al., 1996), loco-regionally by perfusion (Kao et al., 1998), and intravenously (Shirakawa et al., 1998). In our laboratory, we have successfully delivered and expressed adenovirus by all of the above routes in pre-clinical models of cancer growth and metastasis. Some of these results will be illustrated below.

IV. Utilizing a tissue specic promoter to target the growth of prostate cancer and osteosarcoma A. Introduction Several groups, including ours, have constructed vectors containing tissue-specic promoters to restrict the expression of transduced cytotoxic genes to the tissue of interest (Gotoh et al., 1998; Ko et al., 1996; Macri and Gordon, 1994; Shimizu, 1994; Shirakawa et al., 1998; Vile and Hart, 1993). Several studies have described the use of retroviral vectors mediated by tyrosinase promoter for the treatment of melanoma (Vile and Hart, 1993) , albumin promoter for the treatment of hepatoma (Kuriyama et al., 1991; Macri and Gordon, 1994) , myelin basic protein promoter for the treatment of brain tumors (Shimizu, 1994), short PSA promoter for the treatment of prostate cancers (Ko et al., 1996; Pang et al., 1995), and carcinoembryonic antigen (CEA) promoter for gastric carcinoma cells (Tanaka T., 1996) . Our laboratory has developed several adenoviral gene therapy protocols for the treatment of prostate and bone tumors in vivo. Recently, we observed that both prostate cancer cells and their supporting bone stroma expressed a non-collagenase bone matrix protein, osteocalcin (OC) (Ou et al., 1998). Osteocalcin is commonly associated with the turnover of bone cells (McKee et al., 1993; Price, 1985), and is a marker for ossication, which is commonly associated with prostate cancer and its metastases (Arai et al., 1992; Beresford et al., 1984; Shih et al., 1990; Tarle et al., 1989). For these reasons, an ablative gene therapy using the OC promoter to deliver herpes simplex virus-thymidine kinase (TK) (Ishii-Moirta H., 1997) was developed for the treatment of prostate cancer osseous metastasis. In the presence of TK, acyclovir (ACV) or ganciclovir (GCV) will be converted to a toxic guanine analogue capable of disrupting DNA synthesis as described in the preceding section. There are compelling reasons to believe that prostate cancer-bone stromal interaction occurs in vivo, and such communication may contribute to local prostate cancer growth and its distant metastasis (Arai et al., 1992; Curatolo et al., 1992; Ekman and Lewenhaupt, 1991; Shih et al., 1990; Tarle et al., 1989) and associated osteoblastic reactions. Thus, the rationale of this approach is to devise a promoter that will be expressed by both prostate cancer and bone stroma, and use a therapeutic gene that has well-documented bystander effects (Gagandeep S., 1996; Ishii-Moirta H., 1997). By using Ad-OC-TK construct a highly infectious adenovirus, we hope to achieve maximal cell-kill by interrupting cellular communication between the prostate cancer and the bone stroma, and by the direct cytotoxicity exhibited by this version of gene therapy.

Both osteosarcoma and androgen-independent prostate cancers remain major challenges for the orthopedists, urologists and medical oncologists involved in the care of these patients. These seemingly unrelated diseases, however, came together through a molecular analysis of the gene(s) that may be over-expressed in these two forms of cancer during disease progression. Osteocalcin, a noncollagenous Gla protein, was thought to be produced specically in osteoblasts. OC is synthesized, secreted and deposited at the time of bone mineralization (Price, 1985). Interestingly, OC expression was upregulated in several forms of solid tumors, including osteosarcoma and both androgen-dependent and androgen-independent human prostate cancer cell lines. Furthermore, OC expression and secretion were higher in men with metastatic prostate cancer (Coleman et al., 1988; Curatolo et al., 1992), and serum levels of OC were even higher in prostate cancer patients subjected to hormonal deprivation, suggesting that OC may be negatively regulated by testicular androgen (Tarle et al., 1989).

B. Osteocalcin promoter-based tissue-specic gene therapy (Ad-OC-TK) for osteosarcoma 1. Molecular rationale Osteocalcin is a molecular marker present in the serum of patients suffering from either osteosarcoma or prostate cancer. A recent study showed that immunohistochemical staining of osteocalcin was positive in primary osteoblastic osteosarcoma and chondroblastic osteosarcoma specimens, as well as in ve of seven broblastic osteosarcomas. We have shown that strong osteocalcin staining was associated with both primary and metastatic prostate cancer (Ou et al., 1998). Since osteocalcin is the protein most commonly secreted by osteosarcoma cells of the osteoblastic lineage and also a marker of osteoblastic differentiation, we have chosen to use the osteocalcin promoter to achieve tissue-specic expression of the toxic TK gene in rat and human osteogenic sarcoma

Figure 2. Tumor-speciď&#x192;&#x17E;c targeting of osteosarcoma by the osteocalcin promoter (OC). The normal cell does not have the transcriptional factors that are required to activate the osteocalcin promoter to drive thymidine kinase (TK) gene expression. Without the thymidine kinase enzyme, acyclovir (ACV) has no effect on the cell. In contrast, the osteosarcoma cell can activate the osteocalcin promoter and drive transcriptional gene expression of TK. With the thymidine kinase enzyme present, ACV leads to cell death when the cell attempts to divide.

cell lines (Cheon et al., 1997; Ko et al., 1996; Shirakawa et al., 1998) (Figure 2). 2. Results Signicant growth inhibition of rat osteoblastic osteosarcoma (ROS) and a human osteoblastic osteosarcoma (MG-63) occurred when infected with 20 MOIs of Ad-OC-TK and ACV (10 µg/ml). Cells either infected with Ad-OC-TK (20 MOIs) or treated with ACV (10 µg/ml) alone did not exhibit altered growth or morphologic changes during an 8-day observation period. Consistent with their low levels of TK activity, the growth of WH (human bladder cancer) and NIH-3T3 (broblast) cells were not affected by Ad-OC-TK infection, despite the addition of pro-drug ACV in the tissue culture medium. Intralesional injection of Ad-OC-TK followed by ACV administration led to the marked growth inhibition of both ROS and MG-63 subcutaneous xenografts in syngeneic mice. These ndings are further supported by the growth attenuation of ROS pulmonary lesions by systemic Ad-OC-TK as described below.

C. Chemogene therapy for osteosarcoma: combining methotrexate with osteocalcin promoter based tissue-specic gene therapy 1. Molecular rationale To improve the efcacy of the above gene therapy in preclinical osteosarcoma models, we explored the use of methotrexate (MTX) chemotherapy combined with gene therapy. MTX was selected for several reasons. MTX is a proven rst line chemotherapeutic agent for treating osteosarcoma patients (Damron and Pritchard, 1995). MTX, as a competitive inhibitor of dihydrofolate reductase, leads to a decreased accumulation of tetrahydrofolate which then interferes with both purine and pyrimidine biosynthesis (Gorlick et al., 1996), thus diminishing the available nucleotide pool necessary for DNA synthesis in dividing cells. This, combined with the mechanism of TK and ACV, which produces a poisonous phosphorylated ACV (a purine analog known chemically as ACV-triphosphate) through enzymatic catalysis by TK causing DNA chain termination and inhibition of cell division, would allow for enhanced cell-kill. The aim of this study was to investigate the

Figure 3. Proposed mechanism of chemogene therapy. First, MTX decreases the availability of both purines and pyrimidines in all cells. MTX as a competitive inhibitor of dihydrofolate reductase leads to decreased tetrahydrofolate, which then interferes with both purine and pyrimidine biosynthesis, required for DNA synthesis in dividing cells. Second, MTX-induced diminished nucleotide pools are further contaminated by a poisonous phosphorylated ACV (a purine analog known chemically as ACV-triphosphate), which is produced through enzymatic catalysis by TK, and causes DNA chain termination and inhibition of cell division. Thus, the additive effects of chemogene therapy can be attributed to both the efď&#x192;&#x17E;cacy of MTX in decreasing the nucleotide pool size in cells and Ad-OC-TK plus ACV which produce phosphorylated ACV and causes cessation of tumor cell DNA synthesis. The lack of systemic toxicity can be explained by the TK gene expression being controlled by tissue-speciď&#x192;&#x17E;c promoter, thus limiting TK expression only to tumor cells and a bystander tumor cell-kill in localized areas.

possible utility of chemogene therapy in order to combine treatment modalities by maximizing fractions of tumor cell-kill with minimized toxicities. Thus, the additive effects of chemogene therapy can be attributed to both the efcacy of MTX in decreasing the nucleotide pool size in cells and Ad-OC-TK plus ACV which produce phosphorylated ACV and causes more severe cessation of tumor cell DNA synthesis (Figure 3). The lack of systemic toxicity can be explained by the TK gene expression being controlled by tissue-specic promoter, thus limiting TK expression only to tumor cells and a bystander tumor cell-kill in localized areas. 2. Results After determining the IC10 for MTX in both the ROS and MG-63 cell lines, we demonstrated in vitro that low dose MTX (IC10) and Ad-OC-TK plus ACV have additive therapeutic effects as compared to MTX(IC10) or Ad-OC-TK plus ACV treatment alone. In vivo, using a subcutaneous models of murine osteosarcoma, we demonstrated that treatment of Ad-OC-TK plus ACV in combination with low dose MTX chemotherapy inhibited osteosarcoma tumor growth more efciently than either Ad-OC-TK plus ACV or MTX alone. These data suggest that Ad-OC-TK-induced tumor regression was more efcient and signicant when

Figure 4. Treatment of human osteosarcoma (MG-63) with Ad-OC-TK/AVC and MTX chemogene therapy at 180 Days. The untreated mouse has a large ď&#x192;&#x;ank tumor (left panel) and the treated group demonstrated a marked growth inhibition (right panel).

combined with low dose and non-toxic MTX in tumor-bearing animals during a 35 to 45-day study period. At 45 days, 100% and 80% of the animals bearing ROS subcutaneous tumors were alive after gene therapy or chemogene therapy, respectively. Conversely, no animals bearing ROS tumors after PBS or MTX therapy alone were alive at 35 days (Figure 4). The growth inhibition can be demonstrated six months after therapy. In another study, we demonstrated that intravenous Ad-OC-TK plus intraperitoneal ACV signicantly improved pulmonary metastases of osteosarcoma (see below).

D. Systemic delivery of tissue specic gene therapy 1. Introduction Ablative gene therapy for the treatment of cancer continues to gain prominence in preclinical research, but remains limited in clinical application because of an inability to deliver the toxic gene to the tumor cells with specicity. Many vectors (e.g. retroviruses, retroviral producing cells, adenoviruses, liposomes, and others) can deliver genes (therapeutic or ablative) to target cells. Localized delivery and restricted gene expression to the primary tumor have been accomplished via direct injection of therapeutic viruses in animal models (Bonnekoh et al., 1995; Cheon et al., 1997; Eastham et al., 1995; Ko et al., 1996) and clinical trials (Eck et al., 1996; Treat et al., 1996) . This approach is not feasible for the treatment of metastatic disease because of the presence of multiple lesions that would each require separate injection and manipulation. Therefore, alternative approaches to the treatment of metastatic disease with gene therapy must be developed. To study the potential therapeutic efcacy of systemic cancer gene therapy for the treatment of pulmonary metastases, osteosarcoma is an attractive model because a signicant number of these patients eventually develop lung metastasis. Initially, surgical resection of the primary lesion and adjunctive chemotherapy are the mainstay of today’s therapy. For the 20% that present with metastatic disease, 80 % will require additional therapy for relapse; while of the 80% that present with local disease, 35% will require additional therapy for relapse after surgery and adjunctive chemotherapy (O'Reilly, 1996) . Therefore, 44% of patients diagnosed with osteosarcoma will fail conventional rst line therapy. Patients developing recurrent disease usually have a poor prognosis, dying within one year of the development of metastatic disease (Malawer et al., 1993; Naka et al., 1995; Saeter et al., 1995; Ward et al., 1994). New therapeutic approaches that can be applied either separately or in conjunction with current modalities in treating osteosarcoma pulmonary metastases are needed. 2. Mechanistic rationale Systemic delivery of therapeutic genes is attractive for targeting metastatic disease, particularly pulmonary metastases. Because the pulmonary vascular system would be the rst encountered, the adenovirus would be trapped in the lung parenchyma, allowing for higher infectivity. Experimental models using the systemic delivery of liposomal p53 (Lesoon-Wood et al., 1995) and retroviral (Vile et al., 1994) tumor specic TK have been promising. Compared to liposome or retrovirus, adenovirus has several advantages in a systemic delivery strategy, such as its high infectivity in vivo, further aided by the ability to achieve high viral titers through in vitro production. However, a recent report (Brand et al., 1997) demonstrated that systemic administration of adenovirus containing TK under the control of a universal promoter (CMV) supplemented with GCV treatment induced severe hepatotoxic effects. Osteocalcin promoter (OC) has been shown above to be highly effective in directing the transcription of reporter genes in both rat and human osteosarcoma cell lines (Ducy and Karsenty, 1995; Ko et al., 1996) . Since lung epithelium contains the rst capillary bed encountered by therapeutic agents given systemically, several investigators have explored the use of a venous system to deliver therapeutic genes to the lung by cationic liposomes (Lesoon-Wood et al., 1995; Philip et al., 1993; Thierry et al., 1995; Zhu et al., 1993) or retroviral vectors

(Vile et al., 1994) . Since osteosarcoma metastasizes primarily to the lung, and lung vasculature is considered as the rst major capillary bed that a systemically-given therapeutic agent encounters, we designed a strategy to target osteosarcoma pulmonary metastasis by the administration of Ad-OC-TK/ACV in an animal model. 3. Results To prove the principle that a tissue specic promoter regulated gene expression could be achieved with a systemic adenoviral approach. ß-galactosidase reporter gene expression under the transcriptional control of the osteocalcin promoter is specically expressed in osteosarcoma cells rather than the normal lung parenchyma of syngeneic animals bearing pulmonary ROS lesions. In comparison to control animals, systemically delivered Ad-OC-TK plus ACV (via an intravenous route) signicantly retarded the growth of osteosarcoma pulmonary metastases and improved the survival of treated animals. While a limited number of tumor cells in the lung may be infected by Ad-OC-TK, as judged by the immunostaining of a comparable virus that mediates the expression of a reporter gene, ß-galactosidase (ß-gal) Ad-OC-ß gal, a surprisingly potent growth-inhibiting effect by Ad-OC-TK/ACV was noted in osteosarcoma lung metastases. The treated animals bearing ROS pulmonary lesions had markedly less nodules of smaller size and a statistically improved survival. This biologic effect is most likely derived from the existence of close gap junctions between osteosarcoma cells (Donohue and Miller, 1991) which allows the phosphorylated form of ACV to exert its full bystander effect.

E. Osteocalcin promoter-based tissue-specic gene therapy (Ad-OC-TK) for prostate cancer 1. Molecular rationale Prostate cancer’s propensity for the bone environment and the phenotype of osteoblastic growth suggest that it would be susceptible to an osteocalcin promoter based gene therapy. Our laboratory has conrmed the presence of osteocalcin protein by immunohistochemical staining of primary and metastatic human prostate cancer and human prostate cancer cell lines (unpublished data). This nding combined with the clinical ndings of serum osteocalcin elevations in men with metastatic disease would also suggest that the osteocalcin promoter would be active in prostate cancer (Figure 5). Since prostate cancer cells have been shown to interact with surrounding stromal cells, it is reasonable to hypothesize that an osteocalcin promoter based toxic gene would exert an effect on the both the prostate cells and the osteoblastic stromal cells, thus potentiating the bystander effect. 2. Results To assess whether Ad-OC-TK may drive the expression of the TK gene in cells of human prostate cancer, we compared the expressions of TK activities in prostate and

Figure 5. Tumor-speciď&#x192;&#x17E;c targeting of prostate cancer by the osteocalcin promoter (OC). The normal cell does not have the transcriptional factors that are required to activate the osteocalcin promoter to drive thymidine kinase (TK) gene expression. Without the thymidine kinase enzyme present, acyclovir (ACV) has no effect on the cell. In contrast, the prostate cancer can activate the osteocalcin promoter and drive transcriptional gene expression of TK. With the thymidine kinase enzyme present, ACV leads to cell death when the cell attempts to divide.

non-osteoblastic, non-prostatic cell lines after exposure to 20 MOIs of Ad-OC-TK per target cell. The 3 6 mean TK-mediated [ H]-GCV phosphorylation per 10 cells was determined and was designated as the TK activity unit. LNCaP and its androgen-independent lineage-derived sublines, C4, C4-2, C4-2B, and an androgen-independent PC-3 expressed high levels of TK activity in the cell lysates. The TK activity was minimal for several other human prostate cancer cell lines, DU-145 (derived from a brain metastasis), ARCaP (derived from ascites uid), and those of non-osteoblastic and non-prostatic origin. This directly correlated with the amount of in vitro growth inhibition demonstrated by each of these cell lines. Signicant growth inhibition of androgen-independent human prostate cancer cell lines, PC-3 and C4-2, occurred when infected with 20 MOIs of Ad-OC-TK and ACV (10 µg/ml). Cells either infected with AdOC-TK or treated with ACV (10 mg/ml) alone did not exhibit altered growth or morphologic changes during an 8-day observation period. Consistent with their low levels of TK activity, the growth of DU-145, WH and NIH-3T3 cells were not affected by low levels of Ad-OC-TK infection, despite the addition of pro-drug ACV in the tissue medium. PC-3 xenografts were induced by the subcutaneous injection of athymic mice with PC-3 cells. After 3 tumors were palpable (>4 mm ), animals were treated with either PBS alone (n=6), ACV alone (n=6), AdOC-TK alone (n=8), or Ad-OC-TK plus daily ACV intraperitoneal injection (n=8). ACV markedly suppressed the growth of PC-3 tumors during a 45-day observation period following Ad-OC-TK infection; Ad-OC-TK infected PC-3 tumors or ACV treatment of tumor xenografts alone did not signicantly affect the rate of tumor growth. Gross and histological ndings of representative tumors for each group demonstrated a marked treatment effect. Photomicrographs at low magnication illustrate an increased degree of tumor necrosis and migration of lymphocytic cells into the tumor in the Ad-OC-TK plus ACV treatment group versus controls. One week after PC-3 cells were inoculated intraosseously into the marrow space of femurs of nude mice, Ad-OC-TK was injected directly into the femur bone marrow space and followed by 2 weeks of ACV treatment. In the group that received no treatment, X-rays showed that most of the femurs exhibited an intense osteolytic response, while Ad-OC-TK plus ACV treatment clearly inhibited PC-3-induced osteolytic responses with much improved the structure of the femurs. Similarly, we established an intraosseous model for the growth of C4-2 cells, an androgen-independent PSA-producing human prostate cancer cell line. Intraosseous administration of C4-2 but not parental LNCaP cells to the femurs of athymic mice formed reproducible PSA-secreting prostate tumors (Wu, 1998). After intraosseous C4-2 cell inoculation, animals (5 mice/group) were randomized to control (PBStreated), ACV treatment, Ad-OC-TK treatment, and combined Ad-OC-TK and ACV treatment. Serum PSA in mice was followed weekly. Once serum PSA was clearly measurable (> 1 ng/dl), mice were treated with intraosseous injections of adenovirus and intraperitoneal injections of ACV. Serum PSA of both control and treated animals was followed at 1, 3, and 5 weeks. The absolute values of serum PSA (mean + SEM) in animals treated with either ACV, Ad-OC-TK alone, or combined Ad-OC-TK and ACV demonstrated a rise in all groups except for the treatment group. Because of the PSA decline between week 3 and week 5, we calculated the net percentage of PSA elevation [dened as PSA(Wk 5) - PSA(Wk3) / PSA(Wk 3)] was greater in control or ACV-treated mice than Ad-OC-TK alone or the combined Ad-OC-TK and ACV. In fact, the treatment group had negative PSA elevation (-40%) and was lower than that of the Ad-OC-TK treatment alone (+25%). These results were corroborated by the histomorphologic data of the skeletal tumor specimens obtained from control and treated hosts. A characteristic C4-2 intraosseous lesion with viable C4-2 cells conrmed by PSA staining (data not shown) was demonstrated in the control group, while the histomorphologic characteristics of the treatment group demonstrated marked irregularity of the cellular morphology and tumor structure with evidence of many dying tumor cells, not seen in the control group.

F. Potential clinical applications of tissue-specic promoter-mediated gene therapy

Current Phase I trials targeting prostate cancer patients utilize a toxic gene therapy strategy that involves the universal promoter (RSV) or the prostate specic antigen promoter and TK. Both of these promoters may not be suitable for targeting androgen-independent and osseous metastatic patients because: (i) Direct injection of Ad-RSV-TK virus can leak out and infect neighboring normal cells and can damage these cells. Intravenous delivery of Ad-RSV-TK has caused signicant mortality in mice (Brand et al., 1997). In sharp contrast, we demonstrated that systemic delivery of Ad-OC-TK plus intraperitoneal ACV led to improved survival in athymic mice with osteosarcoma pulmonary metastases (Shirakawa et al., 1998). (ii) PSA promoter-mediated toxic genes (e.g. Ad-PSA-TK) may be only killing prostate specic antigen secreting cancer cells but not their supporting osteoblastic cells. There are indications that PSA expression may be greatly reduced in poorly differentiated prostate tumors, a result that seems to be in sharp contrast with OC, whose expression is augmented in metastatic AI prostate cancers (unpublished observation). Therefore, Ad-OC-TK may be the superior agent for targeting end-stage of prostate cancer patients. Rodriguez et al. (1997) reported another strategy to make adenovirus become replication competent under regulation by the short version of the PSA promoter which drives the expression of E1a protein of adenovirus. Since E1a is an essential protein for adenovirus replication, PSA promoter in theory should limit adenoviral replication only in PSA-positive cells and hence induce cytolytic activity in prostate cells only. However, Hitt and Graham reported that “wide variation in E1a expression levels has little effect on virus replication” raising the question of the applicability of this concept (Hitt and Graham, 1990). Further study in suitable animal models capable of adenoviral replication should proceed the initiation of any clinical trials using this approach. Based on the pre-clinical results above, we have proposed a phase I trial for the intralesional treatment of recurrent and metastatic prostate cancer that is currently under FDA review. This study will assess the safety and efcacy of this approach. At the completion of our trial, we hope to have the safety information to propose future trials evaluating the efcacy of this strategy. Subsequently, once safety and efcacy have been established, we will propose a form of regional systemic administration of Ad-OC-TK targeting all forms of prostate cancer.

G. Summary Ad-OC-TK is a tissue specic recombinant adenovirus developed to target osteoblastic cell (i.e.. osteosarcoma) and prostate cancer based on the molecular similarities of these two cancers, that is, their common expression and secretion of OC proteins. In pre-clinical model systems, Ad-OC-TK/ACV demonstrates both in vitro and in vivo tumoricidal activity. We have achieved for the rst time the use of Ad-OC-TK as a novel therapeutic agent that can selectively target and induce the killing of both prostate cancer and cells of an osteoblast lineage. Ad-OC-TK/ACV may be used as the potential gene therapy agent for prostate cancer and osteosarcoma patients who have very debilitating osseous lesions associated commonly with bone pain. By the combination of low-dose MTX plus an osteocalcin promoter-based toxic gene therapy we have developed a novel therapeutic strategy for the treatment of osteosarcoma. We demonstrated that combination treatment is superior to a single drug modality in inducing cell-kill of both rat and human osteosarcoma models (ROS and MG-63). Potentially, chemogene therapy could be a better therapeutic strategy for patients with osteosarcoma as a neoadjuvant, adjuvant, or possibly primary combination therapy. Chemogene therapy could potentially be used to reduce the tumor burden and pain associated with primary metastasis and to eradicate osteosarcoma pulmonary metastasis. Chemogene therapy could ultimately improve both overall survival and the quality of life of patients suffering from osteosarcoma. In summary, we have shown for the rst time that recombinant adenovirus can be given systemically without systemic toxicity to achieve a therapeutic effect on osteosarcoma lung metastasis. Ad-OC-TK/ACV dramatically inhibited the growth of lung nodules and signicantly increased the survival of animals bearing osteosarcoma pulmonary metastases. This approach will open new avenues for targeting pulmonary metastasis using tissue-specic or tumor-specic promoters to guide the expression of therapeutic genes.

For the treatment of prostate cancer, Ad-OC-TK plus ACV could have two signicant effects to cause the shrinkage of prostate cancer growth in the skeleton. First, Ad-OC-TK plus ACV signicantly inhibited the growth of prostate tumor cells in vitro and tumor growth subcutaneously and intraosseously in animal models. Second, Ad-OC-TK plus ACV could signicantly inhibit the growth of tumor supporting bone stroma because of its osteoblastic lineage. Bone stromal cells have been proposed to be important for the adhesion of tumor cells and to maintain their survival, either through cell attachment or by providing tumor cells with soluble growth factors. In addition, bone stromal compartment potentially could secrete important humoral factors (e.g. TGF-), which could protect the host immune response at the site of tumor growth. Ad-OC-TK plus ACV thus could exert not only a direct tumoricidal effect on prostate cancer in the osseous environment, but also block its association with tumor stroma, which presumably plays a vital role in the survival of prostate cancer in the skeleton.

V. Ex-vivo gene therapy using bone homing A. Introduction The importance of stromal-epithelial interaction in prostate cancer has been well established (Chung and Cunha, 1983; Cunha et al., 1987). The reciprocal molecular interactions between prostate cancer (epithelial component) and bone (stromal component) have been well documented (Chung et al., 1991, 1984). These investigations have led to the development of animal models of prostate cancer metastasis and novel therapeutic approaches that have been applied in several preclinical studies. Efforts have also been made to identify putative autocrine and paracrine factors that may be responsible for prostate cancer-bone stroma interaction, and these pathways could have important clinical implications. Based on the theory that prostate cancer cells need the nurturing environment of the bone to survive, we have designed two therapeutic strategies directly targeting the growth of both prostate cancer cells and bone stroma with a therapeutic gene that is also known to exert a bystander effect. Secondly, we have genetically engineered bone stromal cells and have observed the ability of these cells to home to the bone and to exert bystander cell-kill of tumor epithelium in vivo. Our novel therapeutic approach relies on the natural bone homing mechanism of a genetically modied bone cell to deliver therapeutic genes to a target lesion. For example, a bone cell transduced with a toxic gene capable of elaborating a bystander effect to the neighboring cells (i.e. prostate cancer cells) can be an effective therapy for the treatment of prostate cancer bone metastasis. A pluripotent bone stromal cell, D1, was derived from the bone marrow of a mouse and maintains its natural ability to home to the bone after both intravenous and intraosseous injections (Cui Q., 1997; Diduch D.R., 1993). This cell line has been genetically modied to express TK and ß-galactosidase. TK, upon the administration of the prodrug ACV, can convert the prodrug into its active form, killing the D1 cells and resulting in a bystander cell-kill of prostate cancer cells. Bone stromal cells tagged with the ß-galactosidase gene have been shown to home to the bone marrow space, and are widely distributed throughout the bone stroma upon direct intraosseous injection. The strategy of employing genetically engineered bone stromal cells with the potential to home back to the bone, where prostate cancer micrometastasis and lesions may occur, may present an attractive new opportunity for treating patients with bony metastases on an individualized basis.

B. In vitro demonstration of bystander cell kill Osteoblastic cells, D1 and ROS, infected with 20 MOI Ad-CMV-TK and Ad-OC-TK were subsequently co-cultured with LNCaP and its sublines C4-2 and C4-2B. Upon acyclovir administration, D1 cells exhibited a greater bystander growth inhibition on LNCaP than its lineage-related androgen independent sublines. D1 cells transduced with TK but not antisense TK, when co-cultured with LNCaP or its sublines exhibited variable cytotoxicity of the co-cultured tumor cells on plastic dishes. D1-TK, co-cultured with LNCaP or its sublines in the three-dimensional microgravity chamber, demonstrated organoid formation which was altered in size, consistency, and morphology upon administration of ACV. In contrast, the resulting tissue from a 1:1 co-culture of D1-TK and C4-2 in an untreated chamber or in a chamber that was

supplemented with ACV for 8 days revealed a larger amount of tissue mass measured by wet weight and a more compact morphology and tissue architecture as demonstrated by gross tissue/organoids isolated from the untreated chamber. The PSA production measured every 48 hours was minimal in the treated group, while observed to steadily increase in the untreated group.

C. In vivo demonstration of bystander cell kill

Using the subcutaneous co-culture of D1-TK and C4-2 cells, a series of animals bearing chimeric tumors were established as described above. The administration of ACV in the treatment group led to a decreased soft tissue component in D1-TK plus C4-2 chimera and decreased serum PSA production. A signiď&#x192;&#x17E;cant effect on histology was also demonstrated. The genetically engineered D1 cells maintained their ability to preferentially migrate to the bone after several retroviral transfections. At a time point beyond two weeks, after tail vein injection, there was preferential accumulation of D1-TK as demonstrated by X-Gal staining. The same accumulation was demonstrated after tail vein injection or intraosseous injection of D1-TK cells in SCID or athymic nu/nu mice bearing C4-2 intraosseous lesions.

D. Summary This approach is also targeted at disrupting the homeostasis of an osseous prostate cancer metastasis by

inltrating its supportive stroma with cells that can be killed by administration of acyclovir. The D1 cell’s ability to home to and populate an osseous metastasis in an animal model suggests that this approach has potential as an ex vivo form of gene therapy. By combining this bone homing D1 cell with C4-2 in the subcutaneous tissue of an athymic mouse, we were able to (i) generate a novel model to study the bone stroma-prostate cancer cellular interaction and (ii) demonstrate a signicant bystander effect on the growth of prostate cancer cells mediated by the genetically-engineered bone stomal cells. The implications of a bone homing approach are two-fold. (i) The ability of D1-TK cells to exhibit bystander cell-kill in vitro and in vivo in a subcutaneous model mimicking prostate cancer osseous metastasis suggests that the homeostasis of an osseous metastasis may require bone stromal cells and can be disturbed by removing a bone stromal component. (ii) The ability of the D1 cell to maintain its bone homing ability after several ex vivo manipulations suggests that there is a possibility that human bone stromal cells may maintain their skeletal-homing potential. Genetically manipulated bone marrow stem cells have been applied for the treatment of malignancies utilizing autologous bone marrow transplantation. Increased knowledge of the stromal-epithelial interactions of osseous metastasis will allow us to dissect this process and uncover potential new targets for therapy. The bystander effect has not been explained completely. The transfer of toxic metabolites through gap junctions or via the incorporation of apoptotic bodies are the two leading theories (Ishii-Moirta H., 1997; Richards C.A., 1995). Recently, it has been proposed that an immune mediated event may be responsible for the observed cytotoxicity (Gagandeep S., 1996). The pre-clinical models developed in our laboratory may help discern the molecular mechanisms of the bystander effects on prostate cancer growth. A better understanding of the bystander effect will allow us to design and implement more effective therapies. Among many subtypes of prostate cancers, androgen-independence and osseous metastasis have caused signicant mortality and morbidity in patients because there is no available curative therapy. Even after hormonal therapy with the most active agent currently available, a signicant number of prostate cancer patients ultimately develop androgen-independent osseous metastases. To develop new therapeutic modalities for treating end-stage prostate cancer patients, we have explored the possibility of targeting prostate cancer osseous metastasis with toxic gene therapy mediated by an osteoblastic tissue-specic promoter (osteocalcin) and a bone homing mechanism to deliver ablative gene therapy to osseous metastases. The rst approach utilizes an osteoblastic tissue-specic promoter that will restrict the transcription of toxic genes to prostate cancer cells and bone stromal cells. Ablative gene therapy has been demonstrated to exert a bystander effect in achieving maximal cell-kill. In theory, Ad-OC-TK/ACV can exhibit both the expected TK-associated bystander effect on the growth of prostate cancer cells and their supporting bone stromal cells, while also exhibiting an indirect bystander effect by killing the nurturing bone stromal cells and interrupting intracellular communication between prostate cancer cells and bone stroma. The second approach utilizes a natural bone homing mechanism to deliver genetically engineered bone stromal cells to prostate cancer skeletal metastasis, also yielding promising results. This technology needs to be further developed to yield maximal cell-kill at multiple sites of prostate cancer metastases.

VI. Prospects Molecular therapeutics such as gene therapy are being used with increasing frequency. The exponential expansion of knowledge in the eld of molecular medicine has led to therapy based on understanding the molecular pathways of the underlying disease processes. Currently, molecular based gene therapy protocols have been applied predominately for the treatment of life-threatening diseases ( i.e. cystic brosis, ADA, and cancer). With such great potential, molecular approaches will be expanded rapidly into other areas of medicine in the near future. In this review, we have focused our discussion on the concepts and models that we have developed in our laboratory to study the molecular mechanisms underlying human prostate cancer progression and metastasis. These models have been selected and utilized to test the efcacy of various gene therapies using delivery systems containing therapeutic toxic genes, including tumor suppressors and cytotoxic

genes driven by tissue-specic promoters. To understand the use of gene therapy for the treatment of both localized and metastatic prostate cancer, our laboratory focused on the development of animal models that mimic human prostate cancer progression for the exploration of new therapeutic approaches. In the development of animal models, we observed intense reciprocal cellular interaction between prostate cancer cells and bone stroma. We demonstrated that bone stromal cells “select” or “induce” an androgendependent human prostate cancer cell line, LNCaP, to acquire androgen-independent phenotypes particularly in castrated hosts, with resulting LNCaP sublines that exhibit metastatic potential. Results of toxic gene therapy for the treatment of localized and disseminated prostate cancers showed that: (i) Ad-OC-TK expressed high levels in both androgen-dependent and androgen-independent human prostate cancer cell lines; (ii) in parallel with the expression of Ad-OC-TK in tumor cell lines, the efcacy of Ad-OC-TK toxic gene therapy in target cells is directly correlated with the level of TK expression in vitro; (iii) in two experimental models of human prostate cancer, C4-2 and PC-3, we demonstrated that Ad-OC-TK, when applied together with ACV, induced tumoricidal effects in vivo. Signicant histomorphologic improvement of human prostate cancer growth in the bone was supported by bone scans in vivo. In the C4-2 model, we obtained evidence that Ad-OC-TK plus ACV diminished serum PSA, which is conrmed by the improvement of the histomorphologic appearance of this tumor in the skeleton. Finally, we have focused our efforts on the development of combined adenovirus and chemotherapy [i.e. chemogene therapy (Cheon et al., 1997)], the development of a concept of loco-regional delivery of therapeutic genes and drugs, and the exploration of the homing mechanism to treat prostate cancer skeletal metastasis in vivo (Gardner et al., 1998). Taking advantage of the reciprocal cellular interaction between prostate cancer and bone stroma, we have developed two novel gene therapy approaches to target prostate cancer growth in the bone. We have achieved for the rst time the use of Ad-OC-TK/ACV as a novel therapeutic agent that can selectively target and induce the killing of both prostate and osteoblast lineage cells. Our ex vivo approach generated a unique prostate cancer bone growth model, and an osteoblastic reaction was observed when prostate cancer cells were co-inoculated with appropriate bone stromal cells subcutaneously. By introducing genetically engineered bone stromal cells, we observed that the bone stromal cells can confer cytotoxicity to their neighboring prostate cancer cells via a bystander effect. These observations will be developed to improve the delivery of therapeutic genes to the sites of prostate cancer metastases. We anticipate a new array of novel therapeutic approaches that can be applied in the near future to treat prostate cancer in general and its skeletal metastasis in particular.

Acknowledgments Supported by the CaP CURE Foundation (TAG, LWKC), IMClone Scholar for the American Foundation of Urologic Diseases (TAG), NIH Grant #1R29CA74042-01(CK), and NIH Training Grant #5-T32DK07642 (TAG). We also thank our families for supporting our efforts.

References Arai, Y., Takeuchi, H., Oishi, K., and Yoshida O.(1992). Osteocalcin is it a useful marker of bone metastasis and response to treatment in advanced prostate cancer? Prostate 20, 169-177. Beresford, J. N., Gallagher, J. A., Poser, J. W., and Russell, R. G. G. (1984). Production of osteocalcin by human bone cells in vitro. Effect of 1,25-(OH)2D3, parathyroid hormone and glucocorticoids. Metab. Bone Dis. Relat. Res. 5, 229-234. Blaese R.M., C. K. W., Miller A.D., Carter C.S., Fleisher T., Clerici M. , Shearer G., Chiang Y., Tolstoshev P., et al (1995). T lymphocyte-directed gene therapy for ADA-SCID: Initial trial results after 4 years. Science 270, 475-480. Bonnekoh, B., Greenhalgh, D. A., Bundman, D. S., Eckhardt, J. N., Longley, M. A., Chen, S. H., Woo, S. L. C., and Roop, D. R. (1995). Inhibition of melanoma growth by adenoviral-mediated HSV thymidine kinase gene transfer in vivo. J Invest Dermatol 104, 313-317. Boucher, R. C., Knowles, M. R., Johnson, L. G., Olsen, J. C., Pickles, R., Wilson, J. M., Yang, J., and Grossman, M. (1994). Gene therapy for cystic brosis using E1-deleted adenovirus: a phase I clinical trial in the nasal cavity.

The University of North Carolina at Chapel Hill. Human Gene Therapy 5, 615-639. Boulikas, T. (1997). Gene therapy of the cancer: p53, suicidal genes, and other targets. Anticancer Res 17, 1471-1506. Brand, K., Arnold, W., Bartels, T., Lieber, A., Kay, M. A., Strauss, M., and Dorken, B. (1997). Liver-associated toxicity of the HSV-tk/GCV approach and adenoviral vectors. Cancer Gene Ther 4, 9-16. Brand, K., Arnold, W., Bartels, T., Lieber, A., Kay, M., Strauss, M., and Dorken, B. (1997). Liver-associated toxicity of the HSV-tk/GCV approach and adenoviral vectors. Cancer Gene Therapy 4, 9-16. Cheon, J., Ko, S.-C., Gardner, T. A., Shirakawa, T., Gotoh, A., Kao, C., and Chung, L. W. K. (1997). Chemo-gene therapy for osteosarcoma: osteocalcin promoter based suicide gene therapy in combination with methotrexate in murine osteosarcoma model. Cancer Gene Therapy 4, 359-365. Chung, L. W. K., and Cunha, G. R. (1983). Stromal-epithelial interactions: II. Regulation of prostate growth by embryonic urogenital sinus mesenchyme. The Prostate 4, 503-511. Chung, L. W. K., Gao, C., Li, W., and Zhau, H. E. (1993). Stromal-epithelial interaction: implications for prostate tumorigenesis. Cancer Bull. 45, 430-435. Chung, L. W. K., Gleave, M. E., Hsieh, J. T., Hong, S. J., and Zhau, H. Y. E. (1991). Reciprocal mesenchymalepithelial interaction affecting prostate tumour growth and hormonal responsiveness. Cancer Surveys 11, 91-121. Chung, L. W. K., Matsuura, J., and Runner, M. N. (1984). Tissue interactions and prostatic growth. I. Induction of adult mouse prostate hyperplasia by fetal urogenital sinus implants. Biol Reprod 31, 155-163. Coleman, R. E., Mashiter, G., Fogelman, I., Whitaker, K. D., Calef, M., Moss, D. W., and Rubens, R. D. (1988). Osteocalcin: a potential marker of metastatic bone disease and response to treatment. Eur J Cancer Clin Oncol 24, 1211-7. Crystal, R.G., McEluaney, N. G., Rosenfeld, M.A., Chu, C.S., Mastrangeli, A., Hay, J.G., Brody, S.L., Jaffe, H.A., Eissa, N.T., Danel, C. (1994). Administration of an adenovirus containing the human CFTR cDNA to the respiratory tract of individuals with cystic brosis. Nature Genetics 8, 42-51. Cui Q., Wang, G. J., and Balian G. (1997). Steroid induced fat specic gene expression and adipogenesis in cells from bone marrow stroma. J Bone and Joint Surgery 79a, 1054-1063. Cunha, G. R., Donjacour, A. A., Cooke, P. S., Mee, S., Bigsby, R. M., Higgins, S. J., and Sugimura, Y. (1987). The endocrinology and developmental biology of the prostate. Endocrine Rev 8, 338-362. Curatolo, C., Ludovico, G. M., Correale, M., Pagliarulo, A., Abbate, I., Cirrillo, M. E., and Barletta, A. (1992). Advanced prostate cancer follow-up with prostate-specic antigen, prostatic acid phosphatase, osteocalcin and bone isoenzyme of alkaline phosphatase. European Urology 1, 105-7. Damron, T., and Pritchard, D. (1995). Current combined treatment of high-grade osteosarcomas. Oncology 9, 327-350. Diduch D.R., C., M.R., Joyner C., Owen M.E., and Balian G., (1993). Two cells from bone marrow stroma differ in collagen synthesis, osteogenic characteristics, and matrix mineralization. J Bone and Joint Surg 75A, 92-105. Djakiew, D., Tarkington, M., A, and Lynch, J., H. (1966). Paracrine stimulation of polarized secretion from monolayers of a neoplastic prostatic epithelial cell line by prostatic stromal cell proteins. Cancer Res 50, 1966-1974. Donohue, R. E., and Miller, G. J. (1991). Adenocarcinoma of the prostate: biopsy to whole mount. Denver VA experience. Urologic Clinics of North America 18, 449-52. Eastham, J. A., Hall, S. J., Sehgal, I., Wang, J., Timme, T. L., Yang, G., Connell-Crowley, L., Elledge, S. J., Zhang, W.W., Harper, W., and Thompson, T. C. (1995). In vivo gene therapy with p53 or p21 adenovirus for prostate cancer. Cancer Res 55, 5151-5155. Eck, S. L., Alavi, J. B., Alavi, A., Davis, A., Hackney, D., Judy, K., Mollman, J., Phillips, P. C., Wheeldon, E. B., and Wilson, J. M. (1996). Treatment of advanced CNS malignancies with the recombinant adenovirus H5.010RSVTK: a phase I trial. Human Gene Ther 7, 1465-1482. Ekman, P., and Lewenhaupt, A. (1991). Serum tumour markers in human prostatic carcinoma. The value of a marker panel for prognostic information. Acta Oncologica 30, 173-5. Franks, L. M. (1956). The spread of prostatic carcinoma. J. Pathol. 73, 603-611. Gagandeep S., B. R., Green B., Christmas S.E. Klatzmann D., Poston G.J., Kinsella A.R., (1996). Prodrug-activated gene therapy: Involvement of an immunological component in the "bystander effect". Cancer Gene Ther 3, 83-88. Gardner, T. A., Ko, S. C., Kao, C., Cui, Q., Balian, G., and Chung, L. W. K. (1998). Exploiting prostate cancer-bone stromal interaction with gene therapy: Application of osteoblastic tissue-specic promoter and bone homing mechanism to deliver therapeutic toxic genes for the treatment of prostate cancer metastasis. J Urol 59, 4.

Gotoh, A., Kao, C., Ko, S.-C., Hamada, K., Liu, T.-J., and Chung, L. W. K. (1997). Cytotoxic effects of recombinant adenovirus p53, and cell cycle regulators (p21waf1/cip1 and p16INK4) in human prostate cancers. J. Urol. 158, 636-641. Gotoh, A., Ko, S. C., Shirakawa, T., Cheon, J., Kao, C., Miyamoto, T., Gardner, T. A., Ho, L. J., Cleutjens, C. B. J., Trapman, J., Graham, F. L., and Chung, L. W. K. (1998). Development of prostate-specic antigen promoter-based gene therapy for androgen-independent human prostate cancer. J Urol 160, In Press. Graham, F. L., and Prevec, L. (1991). Manipulation of adenovirus vectors, Volume 7, E. J. Murray, ed. (Clifton, New Jersey: The Humana Press, Inc.). Graham, F. L., and Prevec, L. (1995). Methods for construction of adenovirus vectors. Mol. Biotech. 3, 207-220. Hitt, M. M., and Graham, F. L. (1990). Adenovirus E1A under the control of heterolougous promoters: wide variation in E1A expression levels has little effect on virus replication. Virology 179, 667-678. Hsieh, J. T., Wu, H. C., Gleave, M. E., von Eschenbach, A. C., and Chung, L. W. K. (1993). Autoregulation of prostate-specic antigen gene expression in human prostate carcinoma (LNCaP) subline. Cancer Res. 53, 2852-2857. Ishii-Moirta H., A. R., Mullen C.A., Hirano H., Koeplin D.A., Ram Z., Oldeld E.H., Johns D.G., Blaese R.M. (1997). Mechanism of 'bystander effect' killing in the herpes simplex thymidine kinase gene therapy model of cancer treatment. Gene Therapy 4, 244-251. Kao, C., Kaneda, Y., Liu, D., Ko, S. C., Burt, M., Ginsberg, R. J., Chung, L. W. K., and Gardner, T. A. A. f. p. a. t., (Abstract #427). (1998). Locoregional delivery of Ad-CMV-TK combined with ganciclovir is an effective therapy for the treatment of sarcoma pulmonary metastases. In 1st annual Meeting of the American Society of Gene Therapy (Seattle, WA). Ko, A. S. C., Gotoh, G., Kao, C., and Chung, L. W. K. (1996). Tissue targeted toxic gene therapy for an androgenindependent and metastatic human prostate cancer model. J. Urol. 155, 623A. Ko, S. C., Cheon, J., Kao, C., Gotoh, A., Shirakawa, T., Sikes, R. A., Karsenty, G., and Chung, L. W. K. (1 996). Osteocalcin promoter-based toxic gene therapy for the treatment of osteosarcoma in experimental models. Cancer Res 56, 4614-4619. Ko, S.-C., Cheon, J., Kao, C., Gotoh, A., Shirakawa, T., Sikes, R. A., Karsenty, G., and Chung, L. W. K. (1996). Osteocalcin promoter-based toxic gene therapy for the treatment of osteosarcoma in experimental models. Cancer Res. 56, 4614-4619. Ko, S.-C., Gotoh, A., Thalmann, G. N., Zhau, H. E., Johnston, D. A., Zhang, W.-W., Kao, C., and Chung, L. W. K. (1996). Molecular therapy with recombinant p53 adenovirus in an androgen independent, metastatic human prostate cancer model. Human Gene Ther 7, 1683-91. Lesoon-Wood, L. A., Kim, W. H., Kleinman, H. K., Weintraub, B. D., and Mixson, A. J. (1995). Systemic gene therapy with p53 reduces growth and metastases of a malignant human breast cancer in nude mice. Human Gene Ther 6, 395-405. Macri, P., and Gordon, J. W. (1994). Delayed morbidity and mortality of albumin/SV40 T-antigen transgenic mice after insertion of an alpha-fetoprotein/herpes virus thymidine kinase transgene and treatment with ganciclovir. Human Gene Ther 5, 175-182. Malawer, M. M., Link, M. P., and Donaldson, S. S. (1993). Sarcomas of bone. In Cancer Principles and Practice of Oncology, DeVita, Jr., V T, S. Hellman and S. A. Rosenberg, eds. (Philadelphia: J. B. Lippincott Company), pp. 1509-1566. McKee, M., D, Farach-Carson, C. M., C, Butler, W., T, Hauschka, P. V, and and, Nanci A. (1993). Ultrastructural immunolocalization of noncollagenous (osteopontin and osteocalcin) and plasma (albumin and a 2HSglycoprotein) proteins in rat bone. J. Bone Min. Res. 8, 485-496. O'Reilly, R. (1996). NCCN Pediatric osteosarcoma practice guidelines. Oncology 10, 1799-1806. Pang, S., Taneja, S., Dardashti, K., Cohan, P., Kaboo, R., Sokoloff, M., Tso, C. L., Dekernion, J. B., and Belldegrun, A. S. (1995). Prostate tissue specicity of the prostate-specic antigen promoter isolated from a patient with prostate cancer. Human Gene Ther 6, 1417-26. Parker, S. L., Tong, T., Bolden, S., and Wingo, P. A. (1997). Cancer statistics. CA Cancer J Clin 47, 5. Philip, R., Liggitt, D., Philip, M., Dazin, P., and Debs, R. (1993). In vivo gene delivery-efcient transfection of Tlymphocytes in adult mice. J. Biol. Chem. 268, 16087-16090. Price, P. A. (1985). Vitamine-K dependent formation of bone Gla protein (osteocalcin) and its function. Vitamine and Hormone 42, 65-108. Rodriguez, R., Schuur, E. R., Lim, H. Y., Henderson, G. A., Simons, J. W., and Henderson, D. R. (1997). Prostate attenuated replication competent adenovirus (ARCA) CN706: a selective cytotoxic for prostate-specic antigen-

positive prostate cancer cells. Cancer Res. 57, 2559-2563. Sanda, M. G., Ayyagari, S. R., Jaffee, E. M., Epstein, J. I., Clift, S. L., Cohen, L. K., Dranoff, G., Pardoll, D. M., Mulligan, R. C., and Simons, J. W. (1994). Demonstration of a rational strategy for human prostate cancer gene therapy. J Urol 151, 622-8. Scher, H. I., and Chung, L. W. K. (1994). Bone metastasis: improving the therapeutic index. Sem Oncol 21, 630-656. Shih, W. J., Wierzbinski, B., Collins, J., Magoun, S., Chen, I. W., and Ryo, U. Y. (1990). Serum osteocalcin measurements in prostate carcinoma patients with skeletal deposits shown by bone scintigram: comparison with serum PSA/PAP measurements. J Nucl Med 31, 1486-9. Shimizu, K. (1994). Selective gene therapy of malignant gliomas using brain-specic promoters: its efcacy and basic investigations. Japanese J Clin Med 52, 3053-3058. Shirakawa, T., Ko, S. C., Gardner, T. A., Cheon, J., Miyamoto, T., Gotoh, A., Chung, L. W. K., and Kao, C. (1998). In vivo suppression of osteosarcoma pulmonary metastasis with intravenous osteocalcin promoter-based toxic gene therapy. Cancer Gene Ther, In Press. Tarle, M., Kovacic, K., and Strelkov, A. A. (1989). Correlation between bone scans and serum levels of osteocalcin, prostate-specic antigen, and prostatic acid phosphatase in monitoring patients with disseminated cancer of the prostate. Prostate 15, 211-9. Tarle, M., Kovacic, K., and Strelkov-Alrevic, A. (1989). Correlation between bone scans and serum levels of osteocalcin, prostate-specic antigen, and prostatic acid phosphatase in monitoring patients with disseminated cancer of the prostate. Prostate 15, 211-9. Thierry, A. R., Lunardi-Iskandar, Y., Bryant, J. L., Ravinovich, P., Gallo, R. C., and Mahan, L. C. (1995). Systemic gene therapy: biodistribution and long-term expression of a transgenic in mice. Proc. Natl. Acad. Sci. USA 92, 9742-9746. Trinch, Q. T., Austin, E. A., Murray, D. M., Knick, V. C., and Hubber, B. E. (1995). Enzyme/prodrug gene therapy: comparison of cytosine deaminase/5-uorocytosine versus thymidine kinase/ganciclovir enzyme/prodrug systems in a human colorectal carcinoma cell line. Cancer Res. 55, 4808-4812. Vile, R. G., Nelson, J. A., Casteleden, S., Chong, H., and Hart, I. R. (1994). Systemic gene therapy of murine melanoma using tissue specic expression of the HSVtk gene involves an immune component. Cancer Res 54, 6228-6234. Ward, W., Mikaelian, K., Dorey, F., Mirra, J., Sassoon, A., Holmes, E., Eilber, F., and Eckardt, J. (1994). Pulmonary metastases of stage IIB extremity osteosarcoma and subsequent pulmonary metastases. J. Clin. Oncol. 12, 1849-1858. Zhu, N., Liggitt, D., Liu, Y., and Debs, R. (1993). Systemic gene expression after intravenous DNA delivery into adult mice. Science 261, 209-211.

To understand and appreciate the potential uses of gene therapy for the treatment of both localized and metastatic prostate cancer, our laboratory focused on the development of animal models that mimic human prostate cancer progression which allow for the exploration of new therapeutic approaches, particularly the use of gene therapy for the treatment of both localized and disseminated cancers. Reciprocal stromalepithelial interaction occurs when prostate cancer cells metastasize to the bone; both laboratory and clinical evidence suggest that bone microenvironment is vital for the development and survival of prostate cancer cells. This review highlights our efforts to develop novel gene therapy protocols targeting the growth of both prostate cancer and its surrounding bone stroma. A bone homing mechanism was exploited to deliver therapeutic genes to prostate cancer osseous metastases. These models will then be used as targets for gene therapy by delivering therapeutic toxic genes. Bystander cell-kill using adenoviral-mediated expression of thymidine kinase (TK), either regulated constitutively or by an osteoblastic tissue-specic promoter, osteocalcin (OC), was developed. The adenovirus containing TK under the transcriptional control of the OC promoter (Ad-OC-TK) was constructed and tested in several in vitro and in vivo models of human prostate cancer and osteosarcoma. Ad-OC-TK combined with acyclovir (ACV) signicantly inhibited the growth of several osteoblastic cell lines (ROS, MG-63) and prostate cancer cell lines (PC-3, LNCaP, C4-2) in vitro and intraosseous and subcutaneous prostate tumors in vivo. Additionally, we have combined adenovirus and chemotherapy (i.e. chemogene therapy) and the development of systemic and a loco-regional delivery of therapeutic genes for the treatment of cancers in vivo.

A bone stromal cell line, D1, was stably transfected with both b-galactosidase and TK genes to allow for in vitro and in vivo localization and TK expression. The D1 cell line was selected because of its unique ability to localize to the bone upon intravenous injection. D1 cells expressed TK constitutively (D1-TK) and were able to exert strong bystander cell-kill upon the administration of ACV by inhibiting the growth of human prostate cancer cells when grown in vitro in tissue culture, in microgravity chambers, and in vivo as chimeric tumors. In vivo, the potent bystander effect exerted by D1-TK on C4-2 tumor growth was demonstrated radiographically, histologically, and was accompanied by a sharp decrease of serum PSA to a non-detectable level upon ACV administration. We have demonstrated that stromal-epithelial interaction, which is vital to prostate cancer survival, can be interfered with by two novel gene therapy approaches in preclinical models of human prostate cancer. Both adenoviral delivery of TK under transcriptional control by OC and a constitutive expression of TK by bone stromal cells elicit signiď&#x192;&#x17E;cant prostate cancer cell-kill, and warrant further development. Biosketch

TABLE OF CONTENTS II.Prostate Cancer Growth and Metastasis: Model Development A. Introduction C. LNCaP Progression Model Mimics Human Androgen-Independent Prostate Cancer Progression D. Subcutaneous Osseous Prostate Cancer Growth Model E. Summary III.Gene Therapy Approaches to Cancer A. Introduction B. Rationale of Adenoviral Approach for Cancer Gene Therapy B. Vector designs and Modes of Action of Toxic Genes C. Adenoviral Production and Delivery IV.Utilizing Tissue Specic Promoters to Target the Growth of Prostate Cancer and Osteosarcoma A. Introduction B. Osteocalcin Promoter Based Tissue-Specic Gene Therapy (Ad-OC-TK) for Osteosarcoma i. Molecular Rationale ii. Results i. Molecular Rationale ii. Results D. Systemic Delivery of Tissue-Specic Promoter-Driven Gene Therapy for Pulmonary Osteosarcoma i. Introduction ii. Molecular Rationale iii. Results E. Osteocalcin Promoter-Based Tissue-Specic Gene Therapy (Ad-OC-TK) for Prostate Cancer i. Molecular Rationale ii. Results F. Potential Clinical Applications of Tissue-Specic Promoter-Mediated Gene Therapy G. Summary V. Ex-vivo Gene Therapy using Bone Homing A. Introduction C. In Vivo Demonstration of a Bystander Cell Kill D. Summary VI. Summary

Gene Ther Mol Biol Vol 2, 59-68. August 7, 1998.

Cationic liposome-mediated transfection in vivo (review) Dexi Liu* and Young K. Song Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA 15261 ______________________________________________________________________________________ ________________ * Corresponding Author: Tel: (412) 648-8553; Fax: (412) 383-7436; E-mail: dliu@vms.cis.pitt.edu Key words: gene therapy, cationic liposomes, gene delivery, non-viral vector

Summary Cationic liposomes have been widely used as a transfection reagent to introduce gene into cells. Although much has been learned about the factors affecting transfection efciency of liposomes in vitro, understanding on how efcient these lipid carriers are in delivering gene into cells in vivo is still lacking. Recent studies using reporter genes show that signicant level of gene expression can be obtained in different organs including the lung, heart, spleen, liver and kidneys following an intravenous administration of DNA/liposome complexes into mice. In these studies, the cationic lipid to DNA ratio, structure of cationic lipids, liposome composition, and particle size of lipid particles were found to be important in determining the transfection efciency of cationic liposomes. It was also found that gene expression from a single administration is transient but can be maintained by repeated administration. In this paper, we review the data that characterize the in vivo transfection mediated by systemically administered cationic liposome.

I. Introduction Cationic liposomes have become a well established vehicle for introducing genes into cells. Many cationic lipids have been synthesized in the last decade and have been shown to be active in transfecting different type of cells (Behr et al., 1989; Felgner et al., 1987; Felgner et al., 1994; Gao and Huang, 1991; HawleyNelson et al., 1993; Lee et al., 1996; Leventis and Silvius, 1990; Rose et al., 1991; Solodin et al., 1995). In fact, cationic liposomes have become commercially available as transfection reagents and a few liposome formulations have been used in gene therapy clinical trials for treatment of cancer (Nabel et al., 1993; Nabel et al., 1994) and other genetic disorders such as cystic brosis (Caplen et al., 1995; Gill et al., 1997; Porteous et al., 1997; Sorscher et al., 1994). Despite their commercial availability, wide use as gene carriers in preclinical and clinical experiments, the major success of cationic liposome-mediated gene transfer has been limited to in vitro cell culture systems and in vivo at restricted sites where a local regional administration can be applied. Therefore, efforts have been made in the past few years towards the development of an efcient lipid carrier for systemic gene delivery. The major advantage of systemic gene delivery over local injection is that many more sites and a greater number of cells in the body can be targeted. Thus, the successful development of lipid-based carriers for systemic transfection has a great potential for increasing the overall usefulness of lipid carriers in gene therapy. Using a CMV driven expression system containing cDNA of luciferase gene and commonly used cationic lipids, we have systematically examined the factors affecting the transfection efciency of intravenously administered cationic liposomes in mice. In this article we summarize some of our ndings concerning the physicochemical parameters affecting the transfection efciency of cationic liposomes and the characteristics of transgene expression in vivo. We also present our view on the mechanisms involved in cationic liposome-mediated transfection.

II. Physicochemical parameters affecting the transfection activity of cationic liposomes A. Cationic lipid to DNA ratio The ratio of cationic lipid to DNA in DNA/liposome complexes has been shown to be one of the most important factors affecting the transfection efciency. The optimal ratio

Figure 1. Effect of cationic lipid to DNA ratio on liposome transfection activity. DNA/liposome complexes were prepared by mixing different amount of DOTMA liposomes with plasmid DNA (pCMV-Luc) in PBS (see Song et al., 1997, for methods). Each mouse received 25 Âľg of pCMV-Luc plasmid DNA with different amount of DOTMA liposomes in 200 Âľl from tail vein. Luciferase activity was assayed 8 hours after iv injection in the lung (), spleen (), heart (), liver () and kidney (). Error bar represents SEM from three mice.

found for in vitro transfection is in the range of 3.6 to 9 (cationic lipid:DNA, nmol:µg) depending on the types of cationic liposomes, and cell types used (Zhang and Liu, unpublished data). However, for systemic transfection in a mouse model, a much higher cationic lipid to DNA ratio appears to be required for a better transfection into organs such as the lung, heart, liver, spleen and kidneys. As shown in Figure 1, while the level of gene expression in the lung is the highest among all internal organs examined, the level of gene expression in most of organs increases with increasing cationic lipid to DNA ratio. An optimal cationic lipid to DNA ratio for the lung is approximately 36 to 1 or greater under the experimental conditions. Considering the fact that a large number of negatively charged molecules and cellular components exist in the blood, a higher cationic lipid to DNA ratio required for better transfection activity may indicate that the additional cationic liposomes are needed to promote the activity of DNA/liposome complexes. An additional possibility may be that the structure of DNA/ liposome complexes formed at different cationic lipid to DNA ratios are different. For example, it is possible that cells under the articial conditions of cell culture prefer the structural type of DNA/liposome complexes formed at a lower cationic lipid to DNA ratio, while cells in vivo are more sensitive to the complex structure formed at a higher cationic lipid to DNA ratio. To test these possibilities, different amounts of free liposomes were injected intravenously into mice prior to the injection of DNA/lipid complexes prepared at low cationic lipid to DNA ratio (6:1= nmol:µg). Figure 2 shows that, except for the spleen, the level of luciferase activity in all examined organs increased with increasing amounts of free liposomes pre-injected. The pattern and level of gene expression in different organs are very similar to those shown in Figure 1. These results suggest that free liposomes enhance the transfection efciency of DNA/liposome complexes in vivo. The structures of DNA/liposome complexes formed at either low ratio (6:1) or high ratio (36:1) of cationic lipid to DNA are equally active. Therefore, free liposomes play an important role in determining the level of transgene expression following a systemic administration of DNA/liposome complexes.

B. Structure of cationic lipids Since the rst report on cationic liposome-mediated transfection by Felgner and his colleagues (Felgner et al., 1987), many new cationic lipids have been synthesized and shown to be effective in transfecting cells in vitro (for review, see Gao and Huang, 1995). To test whether lipid structure also plays an important role in the transfection of cells, liposomes were prepared using different types of cationic lipids. For all transfections summarized in Figure 3, DNA/liposome complexes were prepared at a cationic

Figure 2. Dose dependent effects of pre-injected free liposomes on the level of gene expression. DNA/liposome complexes were prepared at DOTMA to DNA ratio of 6:1 (nmol:µg) with lipid formulation of DOTMA/Tween 80 (6:2, weight ratio).Twenty ve µg of pCMV-Luc plasmid complexed with DOTMA/Tween 80 formulation were injected 1 minute via tail vein after the animals received different amounts of DOTMA liposomes without Tween 80. Luciferase activity in different tissues was assayed 8 hours after the injection of DNA/lipid complexes. Results represent mean ± SEM of values obtained from 3-6 mice. Lung (), spleen (), heart (), liver () and kidney ().

Figure 3. Effect of cationic lipid structure on transfection activity of liposome-mediated transfection. Luciferase activity was assayed 20 hours after iv injection of 25 µg of pCMV-Luc plasmid complexed with different cationic liposomes in the lung (), spleen (), heart (), liver () and kidney (). Cationic lipid to DNA ratio used was 36:1 (nmol:µg). Error bar represents SEM from three mice.

Figure 4. Size effect on liposome-mediated transfection. Liposomes composed of either DOTMA or DOTAP with different liposome diameter were complexed with pCMV-Luc plasmid DNA at a cationic lipid to DNA ratio of 36:1 (nmol:µg). The size of liposomes and DNA/liposome complexes represents an average size of particles measured by laser light scattering. Luciferase activity was determined 20 hours post-injection in the lung (), spleen (), heart (), liver () and kidney (). Error bar represents SEM from three mice.

lipid to DNA ratio of 36:1. It is evident from Figure 3 that the transfection activity of cationic liposomes varies signiď&#x192;&#x17E;cantly with cationic lipid structure. Between the two types of cationic lipid tested, 3[N-(N', N'-dimethylaminoethane) carbamoyl] cholesterol (DC-Chol), a cholesterol derivative with a tertiary amine as the charged head group (Gao and Huang, 1991), exhibited a low transfection activity in comparison to alkyl chain-based lipids such as N-(2,3-dioleoyloxy)propyl-N,N,N-trimethylammonium chloride (DOTMA) (Felgner et al., 1987) and 1,2-dioleoyloxy-3-trimethylammonium propane (DOTAP) (Leventis and Silvius, 1990). Compared to DOTAP liposomes, the level of gene expression in the lung of animals transfected with DOTMA liposomes is approximately 10-fold higher. As the only structural difference between DOTMA and DOTAP is the linkage between the hydrophilic head group and the alkyl chains, these results suggest that the ether linkage (DOTMA) between the head group and the alkyl chains is superior to the ester linkage (DOTAP). In addition, Figure 3 also shows that the structure of hydrophobic portion of the lipid molecule is also important for the ultimate transfection activity. Liposomes composed of lipids with shorter [1,2-dimyristoyl-3-trimethylammonium propane (DMTAP), C14)] or longer but with unsaturated alkyl chains (DOTAP, C18:1) exhibit better activity than those with long and saturated alkyl chains [1,2dipalmitoyl-3-trimethylammonium propane (DPTAP), C16 and 1,2-distearoyl-3-trimethylammonium propane (DSTAP), C18]. Among the cationic lipids tested, DOTMA liposomes appear to be most active under our experimental conditions.

C. Diameter of lipid particles The particle diameter of liposomes and DNA/liposome complexes is another parameter that was examined for its effect on transfection activity of cationic liposomes. It is apparent in Figure 5 that the diameter of DNA/liposome complexes is directly related to the liposome diameter. For example, the average diameter of DNA/liposome complexes was around 400-500 nm when small sized liposomes (87 nm) were used, compared to a diameter of more than 1 Âľm when the average liposome diameter was around 750 nm.

Figure 5. Effect of neutral lipid on the transfection activity of cationic liposomes. Each mouse received 25 µg of pCMV-Luc plasmid complexed with different liposome composition at cationic lipid to DNA ratio of 36:1 (nmol:µg). The cationic lipid to neutral lipid ratio in liposomes was 1:1 (molar ratio). Luciferase activity was determined 8 hours post-injection in the lung (), spleen (), heart (), liver () and kidney (). Error bar represents SEM from three mice.

This pattern did not seem to depend on the structure of cationic lipids since similar increases in particle size of DNA/liposome complexes were obtained for both DOTMA and DOTAP liposomes. Interestingly, transfection activity of DOTAP liposomes appears to depend on liposome size. An approximately 10-fold increase of luciferase activity in the lung was seen when the size of DOTAP liposomes increased from below 100 to 450 nm or greater. Such size dependence was unique to DOTAP liposomes and was not observed with DOTMA liposomes.

D. Effect of "helper" lipids Inclusion of a neutral lipid such as dioleoylphosphatidyl-ethanolamine (DOPE) into the cationic liposomes has been a common practice in cationic liposome-mediated transfection (Behr et al., 1989; Felgner et al., 1987; Felgner et al., 1994; Gao and Huang, 1991; Hawley-Nelson et al., 1993; Lee et al., 1996; Leventis and Silvius, 1990; Rose et al., 1991; Solodin et al., 1995). In fact, most commercially available cationic liposomes contain DOPE (Gao and Huang, 1995). It is generally believed that, once inside a cell, DOPE in DNA/liposome complexes can facilitate the transfer of DNA across the endosomal membrane and thereby, enhance transfection activity (Farhood et al., 1995; Legendre and Szoka, 1992; Wrobel and Collins, 1995). In addition, several studies have also shown that inclusion of cholesterol into cationic liposomes can enhance the transfection activity at lower cationic lipid to DNA ratios (Bennett et al., 1995; Hong et al., 1997; Y. Liu et al., 1997; Templeton et al., 1997). While it appeared to be true that transfection activity of DOPE-containing liposomes is better than liposomes made of cationic lipid alone under simplied in vitro conditions, it was not clear, however, whether this would hold true in vivo when an optimal cationic liposome to DNA ratio was used. To test whether DOPE and cholesterol have a positive effect on the transfection activity of cationic liposomes in vivo, liposomes containing equal amounts of cationic lipid and DOPE or cholesterol were prepared. Their in vivo transfection activity was then tested in mice using a standard protocol. Figure 5 shows that inclusion of DOPE or cholesterol into DOTMA or DOTAP liposomes does not seem to further enhance the level of gene expression when the cationic lipid to DNA ratio was optimal (36:1, nmol:µg). Furthermore, these results suggest that, despite the fact that DOPE was previously shown to be effective in enhancing the transfection activity of liposomes in vitro, it may not be efcacious to include DOPE as a "helper lipid" into liposomes prepared for systemic gene delivery. These results also indicate that the composition of DNA/liposome complexes optimized under one condition may not be optimal in a different condition. Hence, the optimal composition of the lipid systems for transfection is likely to be condition dependent.

Figure 6. Dose response curve. Various amounts of pCMV-Luc plasmid DNA complexed with DOTMA liposomes at a lipid to DNA ratio of 36:1 (nmol:µg) were intravenously injected into mouse and luciferase activity was determined 20 hours post-injection in the lung (), spleen (), heart (), liver () and kidney (). Error bar represents SEM from three mice.

Figure 7. Time dependent gene expression in different tissues. Luciferase activity in the lung (), spleen (), heart (), liver () and kidney () was assayed at various time point in animals each receiving 25 µg of pCMV-Luc plasmid complexed with DOTMA liposomes (cationic lipid:DNA=36:1, nmol:µg). Error bar represents SEM from three mice.

E. Dose response curve The dose response curve for DOTMA liposomes is shown in Figure 6. It is clear that the level of luciferase activity increased as the injected dose was increased. The highest increase in luciferase activity was obtained in the lung. For example, an approximately 10- to 100-fold increase in luciferase activity was seen in the lung when the injected DNA dose was increased from 10 to 75 µg/mouse. Under these conditions, the level of luciferase activity appeared to be saturated at 50 µg DNA/mouse.

III. Characteristics of transgene expression A. Time dependent gene expression Expression of the transgene in the tissues is transient. The results in Figure 7 show that gene product can be detected as early as 2 hours, reaches the maximal level in the lung around 8-10 hours and decreases to less than 1% of the peak level in 4 days. Among the organs tested, including the lung, spleen, liver, heart and kidney, the level of gene expression in the liver was the most transient. It reaches its highest level of 105 relative light units per mg extracted proteins (RLU/mg) 2-3 hours post injection and dropped to a minimal level in about 48 hours, suggesting that liver may have a higher degradation rate for gene or/and gene product. The relationship between the level of gene expression and the amount of transgene introduced into the lung as a function of time was established using Southern analysis. In these experiments, animals were sacriced at 4, 12, 24, 48, 72, 120 and 168 hours after DNA/liposome complexes were injected. A DNA extract from the lungs was prepared and the relative level of transgene in each sample was analyzed using 32P-labeled full length luciferase gene as a probe. As shown in Figure 8, the amount of luciferase gene detected in the lung decreased with time. Five days after injection, the level of luciferase gene in the lung was below the detectable level for our experimental conditions. These results indicate that the transient gene expression is likely due to the instability of the transgene in transfected cells.

B. Effect of repeated injection While it is expected to observe a transient gene expression in cationic liposome-mediated transfection, it is important to demonstrate whether the level of gene expression can be maintained by repeated administration. In fact, sustaining the level of gene expression by repeated administration is considered as one of the most attractive features for nonviral gene delivery systems. The results in Figure 9 provide direct support to such a prediction. It is evident that a similar level of gene expression in all internal organs was obtained by a repeated administration of DNA/liposome complexes. Interestingly, however, a high level of gene expression may not be achieved if the second injection was performed before approximately

14 days after the ď&#x192;&#x17E;rst injection of DNA/liposome complexes. A period of about two weeks or more between the two injections is needed for an optimal transfection from the second administration.

IV. Discussion It is clear from our work (F. Liu et al., 1997; Song et al., 1997; Song and Liu, in press) and the work of others (Hong et al., 1997; Li and Huang; 1997; Y. Liu et al., 1997; McLean et al., 1997; Templeton et al., 1997; Thierry et al., 1995; Zhu et al., 1993) that cationic liposomes are indeed effective in transfecting cells in vivo by systemic administration of DNA/liposome complexes. Although gene product can be detected in many different organs (lung, spleen, heart, liver and kidneys), the highest level was found in the lung. Such a high level of gene expression in lung is likely due to the fact that pulmonary vasculature is the ď&#x192;&#x17E;rst capillary bed encountered by the DNA/liposome complexes after intravenous injection. DNA/ liposome complexes, once injected into the blood stream via the tail vein, may bind to the endothelial cells lining the capillary bed of the blood vessels in the lung. The embolic effect, resulting from the interaction of DNA/liposome complexes with blood components, and with the negatively charged surface of the endothelial cells of the blood vessel, presumably plays a major role in generating a high level of gene expression in lung.

Figure 9. Effect of repeated injection on the level of gene expression in different tissues. Twenty ve µg of pCMV-Luc plasmid complexed with DOTMA liposomes at lipid to DNA ratio of 36:1 (nmol:µg) were injected intravenously on days 4, 7, 14, 19 and 23 respectively, into mice that had received the same dose and type of DNA/liposome complexes on day zero. The luciferase activity in the lung (), spleen (), heart (), liver () and kidney () was assayed 20 hours after the second injection. In control group (con), mice received only one injection. Error bar represents SEM from three mice. The dose of plasmid DNA injected in this experiment was adjusted to 1.25 mg/Kg based on animal weight.

lung capillary bed. A slower ow would lengthen the exposure time of DNA to the endothelial cells lining the vascular wall and result in a higher level of gene expression. Therefore, DNA/liposome complexes that are capable of being trapped in the lung for an extended period of time will produce a successful transfection. Although many biological factors may be involved, different physicochemical parameters such as cationic lipid to DNA ratio, cationic lipid structure, diameter of the lipid particle and inclusion of helper lipids (Figure 5) may all affect the DNA retention time with the cells to be transfected. It is possible that the DNA retention time with cells, before and after gene transfer occurs, ultimately determines the level of gene expression. Results from biodistribution studies with 125I-labeled plasmid DNA appear to support this hypothesis (F. Liu et al., 1997; Song and Liu, in press). The suppression effect is caused by DNA/liposome complexes but not by free liposomes, the plasmid DNA or gene product (D. Liu, 1997).In summary, the results presented in this paper and those published by other laboratories (Hong et al., 1997; Li and Huang; 1997; Y. Liu et al., 1997; McLean et al., 1997; Templeton et al., 1997; Thierry et al., 1995; Zhu et al., 1993) suggest that the lung is the most transfectable organ by cationic liposomes through intravenous administration. While this may offer an advantage for delivering genes to lung endothelial cells, it also provides a barrier for delivering genes to cells in other organs. Further studies will be required to dene the mechanisms by which cationic liposomes or/and complexes interact with cells in the presence of blood as well as the effect of dynamics of blood ow. Studies are also needed to provide information on how plasmid DNA is transferred to various intracellular compartments of cells in different organs. A better understanding of the nature of these processes, and of how gene transfer efciency is inuenced by multiple physicochemical parameters may allow the development of new strategies for further improvement.

Acknowledgment This work was supported in part by a grant from National Institute of Health CA 72925 and by Targeted Genetics Corporation. Received 19 May 1998; accepted 3 June1998

Reference Behr, J.P., Demeneix, B., Loefer, J.P. and Mutul, J.P. (1989). Efcient gene transfer into mammalian primary endocrine cells with lipopolyamine-coated DNA. Proc. Natl. Acad. Sci. USA 86, 6982-6986. Bennett, M.J., Nantz, M.H., Balasubramaniam, R.P., Gruenert, D.C. and Malone, R.W. (1995). Cholesterol enhances cationic liposome-mediated DNA transfection of human respiratory epithelial cells. Bioscience Reports 15, 47-53. Caplen, N.J., Alton, E.W., Middleton, P.G., Dorin, J.R., Stevenson, B.J., Gao, X., Durham, S.R., Jeffery, P.K., Hodson, M.E., Coutelle, C., Huang, L., Porteous, D.J., Williamson, R. and Geddes, D.M. (1995). Liposome-mediated CFTR gene transfer to the nasal epithelium of patients with cystic brosis. Nature Med. 1, 39-46. Felgner, P.L., Gadek, T.R., Holm, M., Roman, R., Chan, H.W., Wenz, M., Northrop, J.P., Ringold, G.M. and Danielsen, M. (1987). Lipofectin: a highly efcient, lipid-mediated DNA-transfection procedure. Proc. Natl. Acad. Sci. USA 84, 7413-7417. Felgner, J.H., Kumar, R., Sridhar, C.N., Wheeler, C.J., Tsai, Y.J., Border, R., Ramsey, P., Martin, M. and Felgner, P.L. (1994). Enhanced gene delivery and mechanism studies with a novel series of cationic lipid formulations, J. Biol. Chem. 269, 2550-2561. Gao, X. and Huang, L. (1991). A novel cationic liposome reagent for efcient transfection of mammalian cells. Biochem. Biophyhs. Res. Comm. 179:280-285. Gao, X. and Huang, L. (1995). Cationic liposome-mediated gene transfer. Gene Ther. 2, 710-722. Gill, D.R., Southern, K.W., Mofford, K.A., Seddon, T., Huang, L., Sorgi, F., Thomson, A., Mac Vinish, L.J., Ratcliff, R., Bilton, D., Lane, D.J., Littlewood, J.M., Webb, A.K., Middleton, P.G., Colledge, W.H., Cuthbert, A.W., Evans, M.J., Higgins, C.F. and Hyde, S.C. (1997). A placebo-controlled study of liposome-mediated gene transfer to the nasal epithelium of patients with cystic brosis. Gene Ther. 4, 199-209. Hawley-Nelson, P., Ciccarone, V., Gebeyehu, G. and Jessee, J. (1993). LipofecAMINE reagent: a new, higher

efciency polycationic liposome transfection reagent. Focus 15, 73-79. Hong, K., Zheng, W., Baker, A. and Papahadjorpoulos, D. (1997). Stabilization of cationic liposome-plasmid DNA complexes by polyamines and poly(ethylene glycol)-phospholipid conjugates for efcient in vivo gene delivery, FEBS Lett. 400, 233-237. Lee, E.R., Marshall, J., Siegel, C.S., Jiang, C., Yew, N.S., Nichols, M.R., Nietupski, J.B., Ziegler, R.J., Lane, M.B., Wang, K.X., Wan, N.C., Scheule, R.K., Harris, D., Smith, A.E. and Cheng, S.H. (1996). Detailed analysis of structures and formulations of cationic lipids for efcient gene transfer to the lung. Hum. Gene Ther. 7, 1701-1717. Legendre, J.Y. and Szoka, F.C. (1992). Delivery of plasmid DNA into mammalian cell lines using pH-sensitive liposomes: comparison with cationic liposomes. Pharm. Res. 9, 1235-1242. Leventis, R. and Silvius, J.R. (1990). Interactions of mammalian cells with lipid dispersions containing novel metabolizable cationic amphiphiles. Biochim. Biophys. Acta 1023, 124-132. Li, S. and Huang, L. (1997). In vivo gene transfer via intravenous administration of cationic lipid-protamine-DNA (LPD) complexes. Gene Ther. 4, 891-900. Liu, D. (1997) Cationic liposome-mediated gene delivery via systemic administration. J. Liposome Res. 7, 187-205. Liu, F., Qi, H., Huang, L. and Liu, D. (1997). Factors controlling efciency of cationic lipid-mediated transfection in vivo via intravenous administration. Gene Ther. 4, 517-523. Liu, Y., Mounkes, L.C., Liggitt, H.D., Brown, C.S., Solodin, I., Heath, T.D. and Debs, R.J. (1997). Factors inuencing the efciency of cationic liposome-mediated intravenous gene delivery. Nature Biotechnology 15, 167-173. McLean, J.W., Fox, E.A., Baluk, P., Bolton,P.B., Haskell, A., Pearlman, R., Thurston, G., Umemoto, E.Y. and McDonald, D.M. (1997). Organ-specic endothelail cell uptake of cationic liposome-DNA complexes in mice. Ame. J. Phys. 273, H387-H404. Nabel, G.J., Nabel, E.G., Yang, Z.Y., Fox, B.A., Plautz, G.E., Gao, X., Huang, L., Shu, S., Gordon, D. and Chang, A.E. (1993). Direct gene transfer with DNA-liposome complexes in melanoma: expression, biologic activity, and lack of toxicity in humans. Proc. Natl. Acad. Sci. USA 90, 11307-11311. Nabel, G.J., Chang, A.E., Nabel, E.G., Plautz, G.E., Ensminger, W., Fox, B.A., Felgner, P., Shu, S. and Cho, K. (1 994). Immunotherapy for cancer by direct gene transfer into tumors. Hum. Gene Ther. 5, 57-77. Porteous, D.J., Dorin, J.R., McLachlan, G., Davidson-Smith, H., Davidson, H., Stevenson, B.J., Carothers, A.D., Wallace, W.A., Moralee, S., Hoenes, C., Kallmeyer, G., Michaelis, U., Naujoks, K., Ho, L.P., Samways, J.M., Imrie, M., Greening, A.P. and Innes, J.A. (1997). Evidence for safety and efcacy of DOTAP cationic liposome mediated CFTR gene transfer to the nasal epithelium of patients with cystic brosis. Gene Ther. 4, 210-218. Rose, J.K., Buonocore, L. and Whitt, M.A. (1991). A new cationic liposome reagent mediating nearly quantitative transfection of animal cells. Biotechniques 1065, 8-14. Solodin, I., Brown, C.S., Bruno, M.S., Chow, C.Y., Jang, E.H., Debs, R.J. and Heath, T.D. (1995). A novel series of amphiphilic imidazolinium compounds for in vitro and in vivo gene delivery. Biochemistry 34, 13537-13544. Song, Y.K., Liu, F., Chu, S.Y. and Liu, D. (1997). Characterization of cationic liposome-mediated gene transfer in vivo by intravenous administration. Hum. Gene Ther. 8, 1585-1594. Song, Y.K. and Liu, D. (1998). Free Liposomes Enhance the Transfection Activity of DNA/Lipid Complexes in vivo by Intravenous Administration. Biochim. Biophys. Acta in press. Sorscher, E.J., Logan, J.J., Frizzell, R.A., Lyrene, R.K., Bebok, Z., Dong, J.Y., Duvall, M.D., Felgner, P.L., Matalon, S., Walker, L. and Wiatrak, B.J. (1994). Gene therapy for cystic brosis using cationic liposome mediated gene transfer: a phase I trial of safety and efcacy in the nasal airway. Hum. Gene Ther. 5, 1259-1277. Templeton, N.S., Lasic, D.D., Frederik, P.M., Strey, H.H., Roberts, D.D. and Pavlakis G.N. (1997). Improved DNA:liposome complexes for increased systemic delivery and gene expression. Nature Biotechnology 15, 647-652. Thierry, A.R., Lunardi-Iskandar, Y., Bryant, J.L., Rabinovich, P., Gallo, R.C. and Mahan, L.C. (1995). Systemic gene therapy: biodistribution and long-term expression of a transgene in mice. Proc. Natl. Acad. Sci. USA 92, 9742-9746. Wrobel, I. and Collins, D. (1995). Fusion of cationic liposomes with mamalian cells occurs after endocytosis. Biochim. Biophys. Acta 1235, 296-304. Zhu, N., Liggitt, D., Liu, Y. and Debs, R. (1993). Systemic gene expression after intravenous DNA delivery into adult mice. Science 261, 209-211.

Content I.Introduction II.Physicochemical Parameters Affecting the Transfection Activity of Cationic Liposomes A. Cationic lipid to DNA ratio B. Structure of the cationic lipids C. Diameter of lipid particles E. Dose response curve III.Characteristics of Transgene Expression A. Time dependent gene expression B. Effect of repeated administration IV.Discussion

Liu and Song: Cationic liposome-mediated transfection in vivo

Gene Therapy and Molecular Biology Vol 2, page

PAGE 69

PAGE 68 PAGE 69 Gene Therapy and Molecular Biology Vol 2, page

PAGE 66

Gene Ther Mol Biol Vol 2, 69-82. August 7, 1998.

The ability of tRNA-embedded ribozymes to prevent replication of HIV-1 in cell culture Jun Ohkawa1,2,5, Noriko Yuyama2, Shiori Koseki1,2,5, Yutaka Takebe3, Matthias Homann4, Georg Sczakiel4 and Kazunari Taira1,2,5,* 1National Institute for Advanced Interdisciplinary Research, 1-1-4 Higashi, Tsukuba Science City 305-8562, Japan; 2National Institute of Bioscience and Human Technology, 1-1 Higashi, Tsukuba Science City 305-8566, Japan; 3AIDS Research Center, National Institute of Infectious diseases, Toyama, Shinjuku-ku, Tokyo 162-8640, Japan; 4Forschungsschwerpunkt Angewandte Tumorvirologie, Deutsches Krebsforschungszentrum, Neuenheimer Feld 242, D-6900 Heidelberg, Germany, and 5Institute of Applied Biochemistry, University of Tsukuba, Tennoudai 1-1-1, Tsukuba Science City 305-8572, Japan. ______________________________________________________________________________________ ________________ * Correspondence: Professor Kazunari Taira, Institute of Applied Biochemistry, University of Tsukuba, Tennoudai 1-1-1, Tsukuba Science City 305-8572, Japan, Tel/Fax: +81-298-53-4623; E-mail: taira@nibh.go.jp Received 3 June 1998; accepted 15 June 1998 Abbreviations: HIV-1; human immunodeficiency virus, type 1. Key words: AIDS, inhibition, enzyme, catalysis, transcription, cleavage

Summary A trans-acting tRNA-embedded ribozyme targeted to HIV-1 RNA is flanked by cis-acting ribozymes at both its 5’ and its 3’ end so that, upon transcription, the trans-acting ribozyme is trimmed at both its 5’ and its 3’ end, with resultant liberation of a small tRNA-embedded ribozyme. The transacting ribozymes targeted to a different site in HIV-1 RNA were expressed under the control of the SR promoter and were examined in cell culture in a co-transfection transient assay with an HIV-1 infectious molecular clone, pNL4-3, for their ability to suppress HIV-1 replication. Although the extent of inhibition depended on the target site, all constructs caused significant inhibition of the HIV-1 p24 antigen production in culture supernatant. The ribozyme targeted to either 5’ SS (5’ major splicing site) or tat1 site within tat-coding region had the highest inhibitory effect (>80%) when the molar ratio of template DNA for the target HIV-1 RNA to that for the ribozyme was 1:8.

I. Introduction Gene therapy is thought to be the ultimate and radical treatment for several diseases, attributed to genetic disorders. It is reasonable to cure these genetic disorders with genetic approaches, whether it is acquired or congenital. In those cases, AIDS (acquired immunodeficiency syndrome) is one of the most suitable candidates for gene therapy using ribozyme techniques because HIV belongs to the retrovirus, its viral genome consists of RNA, while a ribozyme catalyzes the cleavage of substrate RNA without an aid of proteinaceous apparatus (Altman, 1989; Symons, 1989; Cech, 1990; Dropulic et al., 1992; Scanlon, 1997; Turner, 1997). Therefore, the strategy to overcome the HIV infection using ribozymes as specific restriction enzymes is a hopeful idea to gene therapy for AIDS patients. Indeed, the first demonstration that a hammerhead ribozyme can disarm HIV-1, at least in cells in culture, without any associated detrimental effects (Sarver et al., 1990), has accelerated attempts to use it as an anti-HIV agent (Ojwang, 1992; Altman, 1993; Yu et al., 1993; Yamada et al., 1994a, 1994b; Bertrand and Rossi, 1996, 1997; Scanlon, 1997; Turner, 1997). Since their first discovery, the list of ribozymes with new functions is increasing (Noller et al., 1992; Piccirilli et al., 1992; Bartel and Szostak, 1993; Lorsch and Szostak, 1994; Dai et al., 1995; Wilson and Szostak, 1995; Zhang and Cech, 1997; Vaish et al., 1998). Although there are many kinds of ribozymes with respect to their catalytic mechanisms, hammerhead-type ribozymes (Buzayan et al., 1986; Hutchins et al., 1986; Prody et al., 1986; Foster and Symons, 1987a, 1987b; Symons, 1989) offer particularly attractive possibilities (Taira and Nishikawa, 1992; Eckstein and Lilley, 1996; Scanlon, 1997; Turner, 1997; Kawasaki et al., 1998). These ribozymes work in cis (intramolecularly). However, they have been engineered to work in trans (intermolecularly) both by Uhlenbeck (1987) and by Haseloff and Gerlach (1988). Target RNAs can have any sequence as long as the cleavage site contains NUX, with N being any nucleotide and X being any nucleotide except G (Koizumi et al., 1988; Ruffner et al., 1990; Perriman et al., 1992), although the cleavage efficiency changes dramatically depending on the combination of N and X: in general, GUC offers the best cleavage site (Shimayama et al., 1995). Because of the ease with which the hammerhead ribozyme can be manipulated and of its high activity in vitro, it has been tested and successfully exploited as a regulator of gene expression (Erickson and Izant, 1992; Murray, 1992; Eckstein and Lilly, 1996; Kawasaki et al., 1996, 1998). Since HIV is notorious for its high mutation rate (Preston et al., 1988; Roberts et al., 1988; Peliska and Benkovic, 1992), it is not only difficult to develop vaccines against HIVs but also this genetic variability limits the application of ribozymes for cleavage of HIV RNAs. This is because, once a target site has undergone mutation, the ribozyme targeted to that specific site obviously loses its effectiveness. One way to overcome the problems associated with the mutability of HIV is to use ribozymes that target several conserved sites simultaneously (Taira et al., 1991; Ohkawa et al., 1993). Then, even if one or more sites were to undergo mutation and become resistant to cleavage by a ribozyme, the other conserved sites could still potentially be cleaved by additional ribozymes targeted specifically to those sites. In fact, the use of antisense DNAs targeted simultaneously to different sites has been shown to prevent the development of escape mutants (Lisziewicz et al., 1992) and the effectiveness of a multi-targeting strategy has also been demonstrated in the case of ribozyme-catalyzed reactions (Chen et al., 1992; Weizacker et al., 1992; Ohkawa et al., 1993). We previously proposed the usefulness of self-trimming vectors (Taira et al., 1990) for liberation of multiple ribozymes, each with a different target site (Taira et al., 1991; Ohkawa et al., 1993). Our strategy involves the combination of cis-acting ribozymes with trans-acting ribozymes that have been embedded in tRNA (Cotten and Birnstiel, 1989; Yuyama et al., 1992) so that several trans-acting ribozymes, targeted to HIV (or any other sequence), are trimmed at both their 5’ and 3’ ends by the actions of the cis-acting ribozymes, with resultant liberation of several tRNA-embedded trans-acting ribozymes that should function independently of one another (shotgun type; Ohkawa et al., 1993). These shotgun type ribozyme-activities in vitro should hold promise for similarly efficient ribozyme function in cell culture, provided that each tRNA-embedded 5’- and 3’-trimmed ribozyme is stable and sufficiently active in vivo. However, some earlier reports indicated that, in order to detect ribozyme activities in vivo, a huge excess of ribozymes over the target sequence had to be used (Cameron and Jennings, 1989; Cotten and Birnstiel, 1989). Moreover, when ribozymes were stabilized by being inserted into a long RNA transcript, the ribozyme activity decreased dramatically (Bertrand et al., 1994). Kinetic analysis also indicated that the longer the binding sequence, the lower the activity of the ribozyme, most probably due to formation of an inactive complex (Heidenreich and Eckstein, 1992; Zhou and Taira, 1998). Ribozyme activity in cell culture must be maximized considering the stability of the ribozyme (for this purpose, a longer binding sequence may be preferable) and its kinetic activity (for this purpose, a shorter binding sequence is preferable).Cotten and

Birnstiel (1989), in their pioneering ribozyme design for in vivo usage, have embedded a ribozyme into tRNA anticodon region in order to increase the stability of the ribozyme and to utilize the tRNA's internal promoter. In our present study, as mentioned above, we also designed the ribozyme expression vector in the tRNA embedded form, that did not completely depend on the pol III promoter but it was under the control of SR promoter. We also introduced our self-trimming strategy into this system (Taira et al., 1991; Ohkawa et al., 1993). We expect not only additive but also synergistic effect in vivo after using this shotgun type of ribozyme, targeting different sites simultaneously. Prior to the employment of targets for shotgun type ribozyme, it is very important to evaluate the efficiency of each ribozyme for individual targets in cell culture in order to explore the more effective combination of target sites. We report here the anti HIV-1 activities of five different tRNA-embedded ribozymes, targeting five different sites in HIV-1 RNA, in a transient assay. We could detect significant ribozyme activity in cell culture when the molar ratio of template DNA coding for the target HIV-1 RNA to that for the ribozyme was 1:8.

II. Results A. Ribozyme-expression plasmids and transcription in cell culture In order to examine the anti-HIV activity of tRNA-embedded 5’- and 3-’ trimmed ribozymes in cell culture, each ribozyme unit (Fig. 1) was connected to a mammalian expression vector under the control of SR promoter (Takebe et al., 1988). Figure 2 shows the scheme for the construction of ribozyme expression plasmid. Five relatively conserved regions on HIV-1 RNAs were chosen as the target sites of anti HIV-1 ribozymes (Fig. 3). We previously demonstrated the usefulness of 5’- and 3’-trimmed ribozymes, based on transcription and kinetic studies in vitro (Ohkawa et al., 1993). Since there has been no proof of cis-trimming activities in cell culture, we first examined the self-trimming activities by detecting 5’- and 3’-trimmed ribozymes in total RNA from COS7 cells that had been transfected with the ribozyme expression plasmid, pME18-LTR 1. Trimmed-tRNA-embedded ribozyme was detected directly by Northern hybridization (Fig. 4). Next page:

Figure 1. The 5’- and 3’-trimming plasmid vector, pV3TA-A2, derived from pV3T-A2 (Ohkawa et al., 1993). The tRNA-embedded trans-acting ribozyme is trimmed at both the 5’ and the 3’ end during the transcription reaction.

Figure 2. Scheme for construction of a mammalian ribozyme-expression vector.

Figure 3. Locations and sequences of HIV RNA that can be targeted by ribozymes. Positions of nucleotides relative to the initiation site are indicated by numbers. Cleavage occurs on the 3â&#x20AC;&#x2122; side of the outlined triplets (GUC or GUA).

Figure 4. Detection of tRNA-embedded ribozymes. tRNA-embedded ribozymes trimmed in vivo in COS7 cells were detected by Northern hybridization with 5’ end-labeled synthetic oligonucleotide that was complementary to the ApaI-EcoRV sense strand of the ribozyme targeted at LTR 1. Lane 1 contains total RNA prepared by T7 transcription from the pV3TA-LTR 1 plasmid (Fig. 1; Ohkawa et al., 1993). Overnight transcription produced only fully-trimmed ribozymes. Lane 2 contains total RNA transcribed in vivo in COS7 cells from the pME18-LTR 1 plasmid (Fig. 2) under the control of SR promoter. The total RNA was isolated 24 hours after transfection.

Overnight transcription reaction (<24h) with concomitant self-trimming in vitro, from the pV3TA-LTR 1 plasmid (Fig. 1; under the control of T7 promoter), produced only fully-trimmed ribozymes (lane 1). Judging from the Northern analysis on the total RNAs isolated 24 hours after transfection of COS7 cells with the pME18-LTR 1 plasmid (Fig. 2; under the control of SR promoter), the trimming reaction in cell culture appears much slower (lane 2): the main component was the unprocessed ribozyme. Nevertheless, the tRNA-embedded 5’- and 3’-trimmed ribozymes were at least stable enough in cell culture to be detectable by Northern hybridization, in accord with the previous studies (Cotten and Birnstiel, 1989; Yuyama et al., 1992). The reason for the slower processibility of the transcribed RNA in COS7 cells remains obscure. Large quantities of ribozyme transcripts require higher magnesium ions since ribozymes are metalloenzymes (Yarus, 1993; Uchimaru et al., 1993; Piccirilli et al., 1993; Steitz and Steitz, 1993; Pyle, 1993; Pontius et al., 1997; Lott et al., 1998; Zhou et al., 1996, 1997; Zhou and Taira, 1998). The transcribed RNA in vitro in nuclear extracts from HeLa cells could be processed almost completely upon addition of magnesium ions (5 mM) (data not shown).

B. Inhibition studies on replication of HIV-1 We have confirmed that all five ribozymes transcribed from pV3TA-HIV were active as catalysts since all of them successfully cleaved the HIV-1 RNA in vitro (Fig. 5). No attempts were made to obtain quantitative kinetic parameters because the substrate RNAs used were not the same, as a result of the different target sites (Fig. 3 and Fig. 5). Ribozyme-mediated inhibition of the replication of HIV-1 was then examined in cell culture as described in Materials and Methods according to the procedure outlined in Figure 6 (Sczakiel et al., 1992). In Figure 7A, the relative inhibitory effects of the ribozymes are compared to the results for the control, into which an equal amount of CAT RNA had been introduced in place of the ribozyme plasmid (Homann et al., 1993). We can clearly conclude from Figure 7A that each ribozyme could successfully inhibit the replication of HIV-1 in cell culture. The extent of inhibition was, however, dependent on the target sequence. Among the four target sites chosen, the tat1 site was the most susceptible, with 94-98%

inhibition of replication of HIV-1 in 24 experiments. In order to exclude the possibility that a significant portion of the inhibitory effect shown in Figure 7A originated from the titration of transcription factors by the strong SR promoter, we then constructed a new control plasmid. In case of the newly constructed control plasmid pME18-lac Z, the target sequence of the ribozyme (Fig. 2) was set at a 5’ region of lac Z. In parallel, pME18-5’ SS was also constructed. In Figure 7B, the relative inhibitory effects of the five kinds of ribozymes are evaluated with respect to the new control, pME18-lac Z. The relative inhibitory potentials shown in Figure 7B are pertinent to those shown in Figure 7A and the SR promoter had a little inhibitory effect. Among the five target sites chosen (Fig. 3), the 5’ SS site was the most susceptible one, with 88-94% inhibition of replication of HIV-1. This 5’ SS site turned out to be more susceptible than the tat1 site. Through out these experiments, the molar ratio of the template DNA coding for the target HIV-1 RNA to that for the ribozyme was kept at 1:8. It is to be noted that greater inhibition could be achieved by choosing a higher molar excess of ribozyme template. These results clearly demonstrate that our tRNAembedded ribozyme-expression system under the control of the SR promoter inhibited HIV-1 replication in cell culture.

Figure 5. Ribozyme activities in vitro. Co-transcriptional cleavage of two types of HIV-1 RNA (tat RNA or LTRgag RNA; see details in Figure 1 of Ohkawa et al., 1993) by five kinds of ribozyme-expression plasmids: each target site is shown in Figure 3. Substrate RNAs are indicated by arrows.

III. Discussion A. Ribozyme activities In our previous reports, the ribozyme activity associated with the catalytic sequence against RNA coliphages has been intensively investigated in vitro and in vivo (Yuyama et al., 1992; Inokuchi et al., 1994). RNA coliphages provide systems in which ribozyme activity can be rapidly evaluated in cell culture. We found that (i) a ribozyme designed to cleave the A2 gene (coding the maturation enzyme) of RNA coliphage SP, when transcribed from a plasmid in Escherichia coli, caused failure of the proliferation of progeny phage, and (ii) inactive ribozymes with altered catalytic sequences, which might be expected to form more stable RNA duplex than the active ribozyme, did not have significant inhibitory effects on phage growth. These results indicated that it is mainly the catalytic activity of the ribozyme and not its function as an antisense molecule that

Figure 6. Co-transfection transient assay used for testing the antiviral activity of the ribozyme-expression plasmids. human CD4- epithelioid colon carcinoma cells (SW480) were co-transfected with infectious proviral DNA (pNL4-3) and ribozyme-coding test plasmids (pME18-HIV). The HIV-1 virus initially produced in SW480 cells was amplified in co-cultivated human CD4+ T-lymphoid MT-4 cells that replicate HIV-1 efficiently. The concentrations of HIV-1 antigen p24 were measured 4 days after transfection in cell-free culture supernatants by a polyspecific ELISA.

is responsible for suppressing the proliferation of the RNA phage (Inokuchi et al., 1994). The antisense effect of ribozyme starts to play a role when the binding site becomes longer. In our previous study, the total length of the binding site was set at 20 nucleotides and no antisense effect was recognized (Inokuchi et al., 1994). Having the information that the inhibitory effect of a ribozyme should originate mainly from its cleavage activity (Kawasaki et al., 1998), in the present investigation, we constructed five active ribozymes with the binding sites of 16 nucleotides targeted at five different sites in HIV-1 RNA and their activities in cell culture were compared.

B. Ribozyme trimming system Although all five ribozymes were active in vitro against synthetic substrate RNAs as shown in Figure 5, no attempts were made to obtain quantitative kinetic parameters because (i) the substrate RNAs used were not identical, as a result of the different target sites (Fig. 3 and Fig. 5) and (ii) in general, the kinetic parameters obtained in vitro depended heavily on the length and higher order structure of the substrate RNA used (Heidenreich and Eckstein, 1992; Bertrand et al., 1994). In fact, the highest activity in cell culture exerted by the pME18-5â&#x20AC;&#x2122; SS (Fig. 7B) could not be predicted from the in vitro kinetic data of Figure 5. Similarly, the activity also depends on the length of the ribozyme transcript (Yuyama et al., 1992; Bertrand et al.,

Figure 7. Replication of HIV-1 in the presence of various ribozyme-expression plasmids. The co-transfection transient assay was carried out as described previously (Homann et al., 1993). All constructs (120 ng) were tested by co-transfection in human SW480 cells with infectious proviral HIV-1 DNA (pNL4-3, 40 ng). Either (A) CAT RNA (120 ng) or (B) plasmid that codes for a ribozyme targeted to lacZ (pME18-lacZ, 120 ng) served as the control 100% replication. The bars represent averages of 24 measurements: the average values in % replication are given in numbers above the bars. When we originally examined the activities in vitro of tRNA-embedded 5’- and 3’-trimmed ribozymes against synthetic HIV-1 RNA, five target sites were chosen (Fig. 3). Then all five ribozyme units were separately ligated into the pME18-226HygB vector by the procedure shown in Figure 2. We later realized, upon restriction analysis, that a part of the ribozyme region of 5’ SS was missing from the final pME18-HIV that had been isolated. We therefore used four different kinds of ribozyme constructs in the study shown in (A). The control plasmid pME18-lacZ and the missing pME18-5’ SS were later constructed and the results are shown in (B).

1994). Our previous kinetic analysis in vitro indicated that the shorter ribozyme had the higher cleavage activity (Yuyama et al., 1992), although the effect in cell culture is a combination of the kinetic efficiency (the shorter the better) and the lifetime of the ribozyme (a naked ribozyme will be degraded quickly). It should be noted that when a hammerhead ribozyme (called Rz6) was expressed as a large 3 kb transcript, no cleavage products were observed after incubation with the 1335 nt substrate RNA in vitro, whereas the same Rz6 ribozyme, when expressed as a small 60 nt ribozyme, efficiently cleaved the same substrate RNA (Bertrand et al., 1994). Moreover, in accord with the results in vitro, when the Rz6 ribozyme was expressed in cell culture as a large 2.4 kb transcript from the HIV-1 LTR promoter, there was no inhibition of HIV-1 replication (Bertrand et al., 1994). Since we had a belief long ago that ribozyme transcripts should be as compact as possible, we have been developing a trimming system (Taira et al., 1991; Taira and Nishikawa, 1992; Ohkawa et al., 1993). The accumulating evidence that increasing the size of the flanking regions of the ribozyme would result in a significant reduction of its cleavage efficiency encouraged us to use trimmed ribozymes at their 5’ and 3’ ends. The trimming in cell culture was first demonstrated in COS7 cells (Fig. 4). The trimmed ribozymes are expected to show a significantly higher cleavage efficiency (Bertrand et al., 1994). In fact, our tRNAembedded ribozyme-expression system under the control of the SR promoter inhibited HIV-1 replication in cell culture (Fig. 7). Cameron and Jennings (1989) and Cotten and Birnstiel (1989) demonstrated activities in cell culture of hammerhead ribozymes by transient co-transfection or micro-injection assays. Though both groups successfully demonstrated the usefulness of ribozymes, the extent of inhibition was limited. A huge excess of template DNA coding for the ribozyme over the DNA coding for the target site had to be used for detection of meaningful inhibition. Cotten et al. also showed that short ribozymes were not stable and, therefore, they stabilized the ribozymes by embedding them into tRNA (Cotten and Birnstiel, 1989; Cotten et al., 1989). As a result, RNA polymerase III rather than RNA polymerase II was expected to generate short, correctly folded ribozymes in cells, with stable tertiary structures (Jennings and Molloy, 1987; Cotten and Birnstiel, 1989). However, the pre-tRNA form containing the embedded ribozyme was poorly processed, possibly because the precursor tRNA/ribozyme interacted poorly with the nuclear factors required for both 5’ and 3’ processing and for addition of the 3’ terminal CCA (Cotten and Birnstiel, 1989). When either the target or the ribozyme sequence is inserted into a long RNA, the catalytic activity is expected to be decreased as discussed above, because unfolding of the interacting site may become the rate-limiting step. Therefore, although the activity of a tRNA-embedded ribozyme itself could have been high enough (Cotten and Birnstiel, 1989; Yuyama et al., 1992), that of the precursor tRNA/ribozyme might have been lower. In order not to depend on nuclear factors for both 5’ and 3’ processing, we combined cis-acting ribozymes with each trans-acting ribozyme (Taira et al., 1990) such that the trans-acting ribozyme is trimmed at both its 5’ and its 3’ end by the actions of the cis-acting ribozymes (Taira et al., 1991; Yuyama et al., 1992). The usefulness of such a strategy has been demonstrated in vitro (Ohkawa et al., 1993). Moreover, the results in cell culture, shown in Figure 7, indicate that such a strategy might also be useful in vivo since all five trans-acting ribozymes against HIV-1 RNA inhibited the replication of HIV-1 when the molar ratio of template DNAs for target and ribozyme, respectively, was 1:8. It is to be emphasized that the intention of these assays was not to obtain a large reduction in HIV-1 replication, but to compare the inhibitory potential of each ribozyme construct targeted at different sites. A much greater inhibition could be achieved in our system by choosing a higher molar excess of a ribozyme template.

C. Possible strategies for avoiding "escape" phenomena There are several reports that demonstrate the potential usefulness of ribozymes in the suppression of the proliferation of HIV-1 in cell culture (Sarver et al., 1990; Chen et al., 1992; Dropulic et al., 1992; Ojwang et al., 1992; Yu et al., 1993; Yamada et al., 1994a, 1994b; Bertrand and Rossi, 1996; Zhou et al, 1996; Bertrand et al., 1997; Smith et al., 1997; Li et al., 1998). However, in many cases, the effectiveness of ribozymes was temporary because cells infected with HIV-1 gradually became resistant to the ribozymes (Dropulic et al., 1992). Antisense DNA can also effectively suppress the proliferation of HIV-1 at the early stages of treatment with antisense DNA. However, this suppressive effect is also temporary and, at a later stage, treatment with antisense DNA becomes ineffective (Lisziewcz et al., 1992). These "escape" phenomena may possibly be the result of the high rates of mutation of HIV-1 (Preston et al., 1988; Roberts et al., 1988; Peliska and Benkovic, 1992) or of incomplete shut-down of the expression of HIV, with resultant subsequent breakthrough (Sarver and Rossi, 1993). Since hammerhead ribozymes have high substrate-specificity, mutations in HIV-1 RNA can abolish the effectiveness of the ribozymes (Scanlon, 1997; Turner, 1997; Kawasaki et al., 1998; Zhou and

Taira, 1998). Moreover, some conserved sites are less accessible than others because of the complicated higher-order structure of HIV-1 RNA (Fig. 7). In order to overcome the escape phenomenon associated with HIV-1, several conserved sequences of HIV-1 RNA need to be targeted by multiple ribozymes. Our strategy involves the combination of cis-acting ribozymes with trans-acting ribozymes so that several trans-acting ribozymes, targeted to HIV (or any other sequence), are trimmed at both their 5’ and 3’ ends by the actions of the cis-acting ribozymes, with resultant liberation of several trans-acting ribozymes that should function independently of one another (shotgun type; Ohkawa et al., 1993). Chen et al. (1992) have already demonstrated in cell culture the usefulness of multitarget ribozymes in a system in which several ribozymes, each with a different target site, are connected in tandem. Our present finding that tRNAembedded ribozymes can act in cell culture to inhibit the replication of HIV-1 and our previous finding of the superiority in vitro, in terms of kinetic effectiveness, of shotgun-type multitarget ribozymes relative to the simply connected type of multitarget ribozymes (Ohkawa et al., 1993) encourage us to try to use this system to overcome the "escape" phenomenon. To this end, stable transformants that generate shotgun-type multitarget tRNA/ribozymes are being isolated in our laboratory.

IV. Experimental Procedures A. Construction of Ribozyme-Expression Plasmids The ribozyme-expression vectors were constructed based on the plasmid, pV3TA-A2, as previously reported (Fig. 1, Yuyama et al., 1992), in which Cotten and Birnstiel's tRNA-embedded ribozyme (1989) was combined with self-trimming vector (Taira et al., 1990, 1991) that encodes a trans-acting ribozyme targeted to the A2 gene of RNA phage SP (maturation protein; Inokuchi et al., 1994). The tRNAembedded ribozyme portion of pV3TA-A2 (Fig. 1; ApaI-EcoRV fragment) was replaced by various ribozyme sequences with 8 bases on both substrate-binding arms targeted to relatively well conserved sequences in HIV-1 RNA (Fig. 3). The plasmids generated were designated pV3TA-HIV. During this process, a PstI site was inserted into the stem II loop, replacing the BstPI site of pV3TA-A2, and the newly introduced unique PstI site was used for the confirmation of successful introduction of each transacting ribozyme. As described previously (Yuyama et al., 1992; Shimayama et al., 1993; Sawata et al., 1993; Amontov and Taira, 1996; Kuwabara et al., 1996; Zhou and Taira, 1998), the stem II/loop region is inert in terms of catalysis and, therefore, it can be used for manipulations [According to the report of Tuschl and Eckstein (1993), stem II with 2 base pairs rather than the conventional 4 base pairs has essentially unaltered catalytic activity, independent of the composition of the tetraloop.]. The whole unit within the EcoRI-HindIII fragment in Figure 1 was then inserted into the EcoRI-NotI region of pME18-226HygB (Fig. 2). The resulting ribozyme expression plasmids under the control of the SR promoter (Takebe et al., 1988) were designated pME18-HIV, that includes pME18-LTR 1, pME18-5’SS, pME18-gag1, pME18-tat1, and pME18-tat3 (Fig. 3).

B. Detection of trimmed-tRNA-embedded ribozymes in cell culture tRNA-embedded ribozymes trimmed in cell culture were detected by Northern hybridization procedures. COS7 cells in a 10 cm dish-plate (80% confluent) were transfected with 10 mg of the plasmid that encoded a ribozyme targeted to LTR 1, according to the lipofectin reagent protocol (GIBCO BRL). Total RNA were harvested 24 hours after transfection as described previously (Inokuchi et al., 1994). Total 20 mg of isolated RNA was denatured by glyoxal method and electrophoresed onto 1% agarose gel, then transferred to a nylon membrane (Hybond N, Amersham Inc.). Trimmed-tRNA-embedded ribozyme was detected by Northern hybridization with 5’-end-labeled synthetic oligonucleotide that was complementary to the ApaIEcoRV sense strand of the ribozyme targeted at LTR 1.

C. Cotran s cri p tion al cleavage reactions A mixture for co-transcriptional cleavage reactions, in a total volume of 50 !l, contained 1 pmol of Hind III-linearized pME18-HIVs. In all cases examined, the amount of [32P]-labeled HIV RNA substrate was

kept at 1 pmol. The co-transcriptional cleavage reactions were carried out at 37°C in 40 mM Tris-HCl (pH

8.0), 20 mM MgCl2, 2 mM spermidine, 1 mM DTT, 1 mM NTP (Amersham, UK), 100 units of human placental ribonuclease inhibitor (Takara Shuzo, Kyoto), and 100 units of T7 RNA polymerase (Takara Shuzo).

D. Co-transfection transient assayHuman CD4- epithelioid colon carcinoma cells (SW480) (Leibovitz et al., 1976) in 48-well plates (grown to semi-confluence) were co-transfected with test plasmids (pME18-HIV; 120 ng) and infectious proviral HIV-1-DNA pNL4-3 (Adachi et al., 1986) (40 ng) by the protocol of Chen and Okayama (1987), which is a modified Ca2+ -co-precipitation protocol. Subsequently, transfected SW480 cells were co-cultivated with CD4+ T-lymphoid MT-4 cells, that replicate HIV-1 efficiently (Harada et al., 1985) (2 x 105 per well), and the p24 HIV-1 antigen was quantified in

supernatants from co-cultures 4 days after transfection with a commercially available polyspecific ELISA (Organon). Measured absorbance was transformed into a relative concentration of the antigen. As a control, the identical amount (120 ng) of either CAT RNA (Homann et al., 1993) or a plasmid that encoded a ribozyme targeted to unrelated RNA of lac Z was used in place of the HIV-1 targeted ribozyme expression plasmid. In total, three series of experiments were performed with eight determinations (8 samples were prepared separately so that they represented 8 different experiments) for each construct. Thus, the data shown in Figure 7 are the means of 24 values.

References Adachi, A., Gendelman, H. E., Koenig, S., Folks, T., Willey, R., Rabson, A., and Martin, M. A. (1986). Production of acquired immunodeficiency syndrome-associated retrovirus in human and nonhuman cells transfected with an infectious molecular clone. J. Virol. 59, 284-291. Altman, S. (1989). Ribonuclease P: An enzyme with a catalytic RNA subunit. Adv. Enzymol. 62, 1-36. Altman, S. (1993). RNA enzyme-directed gene therapy. Proc. Natl. Acad. Sci. USA 90, 10898-10900. Amontov, S., and Taira, K. (1996). Hammerhead minizymes with high cleavage activity: a dimeric structure as the active comformation of minizymes. J. Am. Chem. Soc. 118, 1624-1628. Bartel, D. P., and Szostak, J. W. (1993). Isolation of new ribozymes from a large pool of random sequences. Science 261, 1411-1418. Bertrand, E., Pictet, R., and Grange, T. (1994). Can hammerhead ribozymes be efficient tools to inactivate gene function ? Nucleic Acids Res. 22, 293-300. Bertrand, E., and Rossi, J. J. (1996). Anti-HIV therapeutic hammerhead ribozymes: targeting strategies and optimization of intracellular function. In Nucleic Acids Mol. Biol. vol. 10, F. Eckstein and D. M. J. Lilley, eds. (Berlin, Germany, Springer-Verlag), 301-313. Bertrand, E., Castanotto, D., Zhou, C., Carbonnelle, C., Lee, G. P., Chatterjee, S., Grange, T., Pictet, R., Kohn, D., Engelke, D., and Rossi, J. J. (1997). The expression cassette determines the functional activity of ribozymes in mammalian cells by controlling their intracellular localization. RNA 3, 75-88. Buzayan, J. M., Gerlach, W. L., and Bruening, G. (1986). Satellite tobacco ringspot virus RNA: A subset of the RNA sequence is sufficient for autolytic processing. Proc. Natl. Acad. Sci. USA 83, 8859-8862. Cameron, F. H., and Jennings, P. A. (1989). Specific gene suppression by engineered ribozymes in monkey cells. Proc. Natl. Acad. Sci. USA 86, 9139-9143. Cech, T. R. (1990). Self-splicing and enzymatic activity of an intervening sequence RNA from Tetrahymena (Nobel Lecture). Angew. Chem. Int. Ed. Engl. 29, 759-768. Chen, C., and Okayama, H. (1987). High-efficiency transformation of mammalian cells by plasmid DNA. Mol. Cell. Biol. 7, 2745-2752. Chen, C.-J., Banerjea, A. C., Harmison, G. G., Haglund, K., and Schubert, M. (1992). Multitarget-ribozyme directed to cleave at up to nine highly conserved HIV-1 env RNA regions inhibits HIV-1 replication---potential effectiveness against most presently sequenced HIV-1 isolates. Nucleic Acids Res. 20, 4581-4589. Cotten, M., and Birnstiel, M. (1989). Ribozyme mediated destruction of RNA in vivo. EMBO J. 8, 3861-3866. Dai, X., Mesmaeker, A. D., and Joyce, G. F. (1995). Cleavage of an amide bond by a ribozyme. Science 267, 237-240. Dropulic, B., Lin, N. H., Martin, M. A., and Jeang, K.-T. (1992). Functional characterization of a U5 ribozyme: Intracellular suppression of human immunodeficiency virus type 1 expression. J. Virol. 66, 1432-1441. Eckstein, F., and Lilly, D. M. (eds.) (1996). Catalytic RNA, Nucleic Acids and Molecular Biology. vol. 10, (Berlin, Germany, Springer-Verlag) Erickson, R. P. and Izant, J. G. (eds.) (1992). Gene Regulation: Biology of Antisense RNA and DNA (New York, Raven Press). Forster, A. C., and Symons, R. H. (1987a). Self-cleavage of plus and minus RNAs of a virusoid and a structural model for the active sites. Cell 49, 211-220. Forster, A. C., and Symons, R. H. (1987b). Self-cleavage of virusoid RNA is performed by the proposed 55nucleotide active site. Cell 50, 9-16.

Harada, S., Koyanagi, Y., and Yamamoto, N. (1985). Infection of HTLV-III/LAV in HTLV-I-carrying cells MT-2 and MT-4 and application in a plaque assay. Science 229, 563-566. Haseloff, J., and Gerlach, W. L. (1988). Simple RNA enzymes with new and highly specific endoribonuclease activities. Nature 334, 585-591. Heidenreich, O., and Eckstein, F. (1992). Hammerhead ribozyme-mediated cleavage of the long terminal repeat RNA of human immunodeficiency virus type 1. J. Biol. Chem. 267, 1904-1909. Homann, M., Tzortzakaki, M., Rittner, K., Sczakiel, S., and Tabler, M. (1993). Incorporation of the catalytic domain of a hammerhead ribozyme into antisense RNA enhances its inhibitory effect on the replication of human immunodeficiency virus type 1. Nucleic Acids Res. 21, 2809-2814. Hutchins, C., Rathjen, P. D., Forster, A. C., and Symons, R. H. (1986). Self-cleavage of plus and minus RNA transcripts of avocado sunblotch viroid. Nucleic Acids Res. 14, 3627-3640. Inokuchi, Y., Yuyama, N., Hirashima, A., Nishikawa, S., Ohkawa, J., and Taira, K. (1994). A hammerhead ribozyme inhibits the proliferation of an RNA coliphage SP in Escherichia coli. J. Biol. Chem. 269, 11361-11366 Jennings, P. A., and Molloy, P. L. (1987). Inhibition of SV40 replicon function by engineered antisense RNA transcribed by RNA polymerase III. EMBO J. 6, 3043-3047.Kawasaki, H., Ohkawa, J., Tanishige, N., Yoshinari, K., Murata, T., Yokoyama, K. K., and Taira, K. (1996). Selection of the best target site for ribozyme-mediated cleavage within a fusion gene for adenovirus E1A-associated 300 kDa protein (p300) and luciferase. Nucleic Acids Res. 24, 3010-3016. Kawasaki, H., Ecker, R., Yao, T.-P., Taira, K., Chiu, R., Livingston, D. M., and Yokoyama, K. K. (1998). Distinct roles of the co-activators p300 and CBP in retinoic-acid-induced F9-cell differentiation. Nature 393, 284-289. Koizumi, M., Iwai, S., and Ohtsuka, E. (1988). Construction of a series of several self-cleaving RNA duplexes using synthetic 21-mer. FEBS Lett. 228, 228-230. Kuwabara, T., Amontov, S., Warashina, M., Ohkawa, J., and Taira, K. (1996). Characterization of several kinds of dimer minizyme: simultaneous cleavage at two sites in HIV-1 tat mRNA by dimer minizymes. Nucleic Acids Res. 24, 2302-2310. Leibovitz, A., Stinson, J. C., McCombs III, W. B., McCoy, C. E., Mazur, K. C., and Mabry, N. D. (1976). Classification of human colorectal adenocarcinoma cell lines. Cancer Res. 36, 4562-4569. Li, X., Gervaix, A., Kang, D., Law, P., Spector, S. A., Ho, A. D., and Wong-Stall, F. (1998). Gene therapy targeting cord blood-derived CD34+ cells from HIV-exposed infants: preclinical studies. Gene Ther. 5, 233-239. Lisziewicz, J., Sun, D., Klotman, M., Agrawal, S., Zamecnik, P., and Gallo, R. (1992). Specific inhibition of human immunodeficiency virus type 1 replication by antisense oligonucleotides: An in vitro model for treatment. Proc. Natl. Acad. Sci. USA 89, 11209-11213. Lorsch, J. R., and Szostak, J. W. (1994). In vitro evolution of new ribozymes with polynucleotide kinase activity. Nature 371, 31-36. Lott, W. B., Pontius, B. W., and von Hippel, P. H. (1998). A two-metal ion mechanism operates in the hammerhead ribozyme-mediated cleavage of an RNA substrate. Proc. Natl. Acad. Sci. USA 95, 542-547. Murray, J. A. H. (ed.) (1992). Antisense RNA and DNA (New York, Wiley-Liss). Noller, H. F., Hoffarth, V., and Zimniak, L. (1992). Unusual resistance of peptidyl transferase to protein extraction procedures. Science 256, 1416-1419. Ohkawa, J., Yuyama, N., Takebe, Y., Nishikawa, S., and Taira, K. (1993). Importance of independence in ribozyme reactions: Kinetic behavior of trimmed and of simply connected multiple ribozymes with potential activity against human immunodeficiency virus. Proc. Natl. Acad. Sci. USA 90, 11302-11306. Ojwang, J. O., Hampel, A., Looney, D. J., and Wong-Staal, F. (1992). Inhibition of human immunodeficiency virus type 1 expression by a hairpin ribozyme. Proc. Natl. Acad. Sci. USA 89, 10802-10806. Peliska, J. A., and Benkovic, S. J. (1992). Mechanism of DNA strand transfer reactions catalyzed by HIV-1 reverse transcriptase. Science 258, 1112-1118. Perriman, R., Delves, A., and Gerlach, W. L. (1992). Extended target-site specificity for a hammerhead ribozyme. Gene 113, 157-163. Piccirilli, J. A., McConnell, T. S., Zaug, A. J., Noller, H. F., and Cech, T. R. (1992). Aminoacyl esterase activity of the Tetrahymena ribozyme. Science 256, 1420-1424. Piccirilli, J. A., Vyle, J. S, Caruthers, M. H., and Cech, T. R. (1993). Metal ion catalysis in the Tetrahymena ribozyme reaction. Nature 361, 85-88. Pontius, B. W., Lott, W. B., and von Hippel, P. H. (1997). Observations on catalysis by hammerhead ribozymes are consistent with a two-divalent-metal-ion mechanism. Proc. Natl. Acad. Sci. USA 94, 2290-2294. Preston, B. D., Poiesz, B. J., and Loeb, L. A. (1988). Fidelity of HIV-1 reverse transcriptase. Science 242, 1168-1171. Prody, G. A., Bakos, J. T., Buzayan, J. M., Schneider, I. R., and Bruening, G. (1986). Autolytic processing of dimer plant virus satellite RNA. Science 231, 1577-1580. Pyle, A. M. (1993). Ribozymes: A Distinct Class of Metalloenzymes. Science 261, 709-714. Roberts, J. D., Bebenek, K., and Kunkel, T. A. (1988). The accuracy of reverse transcriptase from HIV-1. Science 242, 1171-1173. Ruffner, D. E., Stormo, G. D., and Uhlenbeck, O. C. (1990). Sequence requirements of the hammerhead RNA selfcleavage reaction. Biochemistry 29, 10695-10702. Sarver, N., Cantin, E. M., Chang, P. S., Zaia, J. A., Ladne, P. A., Stephens, D. A., and Rossi, J. J. (1 990). Ribozymes as

potential anti-HIV-1 therapeutic agents. Science 247, 1222-1225. Sarver, N., and Rossi, J. (1993). Gene therapy: A bold direction for HIV-1 treatment. AIDS Res. Hum. Retro. 9, 483-487. Sawata, S., Shimayama, T., Komiyama, M., Kumar, P. K. R., Nishikawa, S., and Taira, K. (1993). Enhancement of the cleavage rates of DNA-armed hammerhead ribozymes by various divalent metal ions. Nucleic Acids Res. 21, 5656-5660. Scanlon, K. J. (ed.) (1997). Therapeutic Applications of Ribozymes, Methods in Molecular Medicine. vol. 11 (New Jersey, Humana Press). Sczakiel, G., Rittner, K., and Pawlita, M. (1992). Human immunodeficiency virus type-1 replication is reduced by intracellular antisense RNA expression. In Prospects for antisense nucleic acid therapy of cancer and AIDS, E. Wickstrom, ed. (New York, Wiley-Liss), pp. 179-193. Shimayama, T., Nishikawa, F., Nishikawa, S., and Taira, K. (1993). Nuclease-resistant chimeric ribozymes containing deoxyribonucleotides and phosphorothioate linkages. Nucleic Acids Res. 21, 2605-2611. Shimayama, T., Nishikawa, S., and Taira, K. (1995). Generality of the NUX rule: kinetic analysis of the results of systematic mutations in the trinucleotide at the cleavage site of hammerhead ribozymes. Biochemistry 34, 3649-3654. Smith, S. M., Maldarelli, F., and Jeang, K. T. (1997). Efficient expression by an alphavirus replicon of a functional ribozyme targeted to human immunodeficiency virus type 1. J. Viol. 71, 9713-9721. Steitz, T. A., and Steitz, J. A. (1993). A general two-metal-ion mechanism for catalytic RNA. Proc. Natl. Acad. Sci. USA 90, 6498-6502. Symons, R. H. (1989). Self-cleavage of RNA in the replication of small pathogens of plants and animals. Trends Biochem. Sci. 14, 445-450.Taira, K., Uebayasi, M., Maeda, H., and Furukawa, K. (1990). Construction of a novel artificial-ribozyme-releasing plasmid. Protein Eng. 3, 691-701. Taira, K., Nakagawa, K., Nishikawa, S., and Furukawa, K. (1991). Construction of a novel RNA-transcript-trimming plasmid which can be used both in vitro in place of run-off and (G)-free transcriptions and in vivo as multisequences transcription vectors. Nucleic Acids Res. 19, 5125-5130. Taira, K., and Nishikawa, S. (1992). Construction of several kinds of ribozymes: their reactivities and utilities. In Gene regulation: Biology of antisense RNA and DNA, R. P. Erickson and J. G. Izant, eds. (New York, Raven Press), pp. 35-54. Takebe, Y., Seiki, M., Fujisawa, J., Hoy, P., Yokota, K., Arai, K., Yoshida, M., and Arai, N. (1 988). SR promoter: an efficient and versatile mammalian cDNA expression system composed of the simian virus 40 early promoter and R-U5 segment of human T-cell leukemia virus type 1 long terminal repeat. Mol. Cell. Biol. 8, 466-472. Turner, P. C. (ed.) (1997). Ribozyme Protocols, Methods in Molecular Biology. vol. 74 (New Jersey, Humana Press). Tuschl, T., and Eckstein, F. (1993). Hammerhead ribozymes: importance of stem-loop II for activity. Proc. Natl. Acad. Sci. USA 90, 6991-6994. Uchimaru, T., Uebayasi, M., Tanabe, K., and Taira, K. (1993). Theoretical analyses on the role of Mg2+ ions in ribozyme reaction. FASEB J. 7, 137-142. Uhlenbeck, O. C. (1987). A small catalytic oligoribonucleotide. Nature 328, 596-600. Vaish, N. K., Heaton, P. A., Fedorova, O., and Eckstein, F. (1998). In vitro selection of a purine nucleotide-specific hammerhead like ribozyme. Proc. Natl. Acad. Sci. USA 95, 2158-2162. Weerasinghe, M., Lien, S. E., Asad, S., Read, S. E., and Joshi, S. (1991). Resistance to human immunodeficiency virus type 1 (HIV-1) infection in human CD4+ lymphocyte-derived cell lines conferred by using retroviral vectors expressing an HIV-1 RNA-specific ribozyme. J. Virol. 65, 5531-5534. Weizacker, F.-v., Blum, H. E., and Wands, J. R. (1992). Cleavage of hepatitis B virus RNA by three ribozymes transcribed from a single DNA template. Biochem. Biophys. Res. Commu. 189, 743-748. Wilson, C., and Szostak, J. W. (1995). In vitro evolution of a self-alkylating ribozyme. Nature 374, 777-782. Yamada, O., Kraus, G., Leavitt, M. C., Yu, M., and Wong-Stall, F. (1994a). Activity and cleavage site specificity of an anti-HIV-1 hairpin ribozyme in human T cells. Virology 205, 121-126. Yamada, O., Yu, M., Yee, J.-K., Kraus, G., Looney, D., and Wong-Stall, F. (1994b). Intracellular immunization of human T cells with a hairpin ribozyme against human immunodeficiency virus type 1. Gene Ther. 1, 38-45. Yarus, M. (1993). How many catalytic RNAs? Ions and the Cheshire cat conjecture. FASEB J. 7, 31-39. Yu, M., Ojwang, J., Yamada, O., Hampel, A., Rapapport, A., Looney, D., and Wong-Stall, F. (1993). A hairpin ribozyme inhibits expression of diverse strains of human immunodeficiency virus type 1. Proc. Natl. Acad. Sci. USA 90, 6340-6344. Yuyama, N., Ohkawa, J., Inokuchi, Y., Shirai, M., Sato, A., Nishikawa, S., and Taira, K. (1992). Construction of a tRNA-embedded-ribozyme trimming plasmid. Biochem. Biophys. Res. Commu. 186, 1271-1279. Zhang, B., and Cech, T. R. (1997). Peptide bond formation by in vitro selected ribozymes. Nature 390, 96-100. Zhou, D.-M., Zhang, L.-H., Kumar, P. K. R., and Taira, K. (1996). The ribozyme mechanism revisited: Evidence against direst coordination of a Mg2+ ion with the pro-R oxygen of the scissile phosphate in the transition state of a hammerhead ribozyme-catalyzed reaction. J. Am. Chem. Soc. 118, 8969-8970. Zhou, D.-M., Zhang, L.-H., and Taira, K. (1997). Explanation by the double-metal-ion mechanism of catalysis for

the differential metal ion effects on the cleavage rates of 5’-oxy and 5’-thio substrates by a hammerhead ribozyme. Proc. Natl. Acad. Sci. USA 94, 14343-14348. Zhou, D.-M., and Taira, K. (1998). The hydrolysis of RNA: from theoretical calculations to the hammerhead ribozyme-mediated cleavage of RNA. Chem. Rev. 98, 991-1026.

Gene Ther Mol Biol Vol 2, 83-94. August 7, 1998.

Ribozyme-catalyzed trimming reactions and the direct role of Mg2+ ions in the cleavage of RNA Masaki Warashina1,2,4, Tomoko Kuwabara1,2,4, Yuka Nakamatsu1,2,4, Masayuki Sano1,2,4, Atsushi Shibata1,2,4, Hideki Shizuku1,2,4, Hideyuki Takeda1,2,4, Ryuji Utsunomiya1,2,4, Jing-Min Zhou1,2,4, Tadafumi Uchimaru3, Jun Ohkawa1,2,4 and Kazunari Taira1,2,4,* 1National Institute for Advanced Interdisciplinary Research, 1-1-4 Higashi, Tsukuba Science City 305-8562, Japan; 2National Institute of Bioscience and Human Technology, 1-1 Higashi, Tsukuba Science City 305-8566, Japan; 3National Institute of Materials and Chemical Research, 1-1 Higashi, Tsukuba Science City 305-8565, Japan; and 4Institute of Applied Biochemistry, University of Tsukuba, Tennoudai 1-1-1, Tsukuba Science City 305-8572, Japan. ______________________________________________________________________________________ ________________ * Correspondence: Professor Kazunari Taira, Institute of Applied Biochemistry , University of Tsukuba, Tennoudai 1-1-1, Tsukuba Science City 305-8572, Japan, Tel/Fax: +81-298-53-4623; E-mail: taira@nibh.go.jp Received 3 June 1998; accepted 15 June 1998 Key words: ribozyme, cleavage, trimming, RNA world, Mg2+ ion, metallo-enzyme

Summary In the hypothetical RNA world divalent Mg2+ ions were exploited for cleavage (or ligation) of ribonucleic acids. Although some Mg2+ ions are involved in forming the tertiary structures of

RNAs, the key Mg2+ ions are directly involved in catalysis not only in the case of hammerhead ribozymes but also in the case of Tetrahymena and, possibly, other types of ribozyme. RNA components bind the indispensable Mg 2+ ions to the phosphodiester bonds that are being broken

(or formed). Our analysis indicates that the chemical cleavage step of reactions catalyzed by the hammerhead ribozyme does not appear to have been perfected and, thus, it seems possible to create RNA-cleaving agents that are significantly more active than the standard hammerhead ribozyme. Moreover, RNA-cleaving mechanisms might converge as one unique and universal mechanism, exploited not only by various kinds of ribozyme but also by artificially created metal-ion-dependent DNAzymes and other RNA-cleaving agents that are yet to be identified.

I. Introduction The term "ribozyme" is derived from the terms ribonucleic acid (RNA) and enzyme, and denotes a type of RNA molecule with catalytic properties. Researchers used to believe that RNA merely acted as an intermediary in the process of genetic-information transfer from DNA to protein molecules. It was not until the publication of work by Altman and Cech that RNA was shown also to play a catalytic role in the cell (Altman, 1989; Cech, 1990). A number of other natural ribozymes and in vitro selected ribozymes (Altman, 1989; Noller et al., 1992; Piccirilli et al., 1992; Symons, 1992; Bartel and Szostak, 1993; Lorsch and Szostak, 1994; Dai et al., 1995; Wilson and Szostak, 1995; Zhang and Cech, 1997; Vaish et al., 1998) have been discovered since the original discoveries of RNase P and the Tetrahymena ribozyme, but from certain standpoints one of the most important discovery has been that of the hammerhead ribozyme (Symons, 1992; Uhlenbeck, 1987; Haseloff and Gerlach, 1988; Sarver et al., 1990; Eckstein and Lilley, 1996; Turner, 1997; Scanlon, 1997; Zhou and Taira, 1998). This ribozyme, first developed by an Australian research team, can act within a single molecule (cis-acting) but has also been engineered in such a way that it acts against other molecules (trans-acting) as well (Symons, 1992; Haseloff & Gerlach, 1988). Until the discovery of ribozymes, attempts to fathom the origin of life were plagued by the classic chickenand-egg paradox -- did nucleic acids carrying the genetic information required to make proteins come first, or did proteins that could synthesize nucleic acids come first? The problem with the hypothesis of protein primacy is that even if one assumes that amino acids could join randomly with one another to form functional proteins, proteins lack a suitable mechanism for transmitting the information on how amino acids should join up to make the next generation of proteins. The discovery of ribozymes, which have enzymatic activities and, at the same time, are capable of carrying genetic information, strongly suggests that nucleic acids, in the form of RNA, were responsible for the origin of life (Gesteland and Atkins, 1993). New research is demonstrating that in several functionally important enzymes, in addition to RNase P, which consist of protein and RNA, the RNA component is in fact, the source of the enzymatic activity (Noller et al., 1992; Sawa and Shimura, 1992; Nitta et al., 1998). Scientists can no longer deny the potential key role of RNA in the origin of life, and new research is constantly revealing that RNA plays a more significant role than anticipated in all life processes (Kobayashi et al., 1993, 1995). At first, we initiated a program of basic research into ribozymes by undertaking molecular orbital calculations (Taira et al., 1990a, 1991a, 1993; Uchimaru et al., 1991, 1992; Storer et al., 1991; Yliniemela et al., 1993; Uebayasi et al., 1994; Zhou and Taira, 1998), and the calculations hint at the positive role of the magnesium ion in catalysis (Uebayasi et al., 1991; Uchimaru et al., 1993). That is, in both Tetrahymena and hammerhead ribozymes, magnesium ions may act as true catalysts. These findings indicated that ribozymes are essentially metallo-enzymes like many other protein enzymes and that the magnesium ions play pivotal roles in catalysis rather than just maintaining the tertiary structures of RNA components (Uebayasi et al., 1991; Uchimaru et al., 1992, 1993; Eckstein and Lilley, 1996; Zhou and Taira, 1998). In agreement with this finding many groups have been able to replace significant portions of hammerhead ribozymes by DNA components (Perreault et al., 1990, 1991; Yang et al., 1990, 1992; Pieken et al., 1991; Williams et al., 1992; Taylor et al., 1992; Paolella et al., 1992; Goodchild et al., 1992; McCall et al., 1992; Hendry et al., 1992; Nishikawa et al., 1991; Taira and Nishikawa, 1992; Shimayama et al., 1992, 1993; Warashina et al., 1997). The groups of Rossi and Jennings, as well as our own, have found that it is also possible to engineer a chimeric DNA/RNA ribozyme that is a better cleaver of RNA than the analogous all-RNA ribozyme (Taylor et al., 1992; Hendry et al., 1992; Shimayama et al., 1992, 1993; Warashina et al., 1997). Moreover, several attempts to generate DNA enzymes by in vitro selection have been successful and single-stranded DNA molecules with enzymatic activity have been isolated, including a Pb2+-dependent DNAzyme with RNA-cleavage activity by Joyce's group (Breaker and Joyce, 1994), a DNAzyme with ligase activity by Szostak's group (Cuenoud and Szostak, 1995), and a DNAzyme with self-cleaving activity by Breaker's group (Carmi et al., 1996). Almost all of the DNAzymes isolated to date by in vitro selection require metal cofactors, such as Pb2+, Mg2+, Zn2+,

Mn2+ or Ca2+ ions (Breaker and Joyce, 1994; Breaker and Joyce, 1995; Faulhammer and Famulok, 1996; Santoro and Joyce, 1997). The Mg2+-binding motif used in the RNA world appears to be conserved even in some of the DNAzymes as is conserved in the protein world (Beese and Steitz, 1991; Steitz and Steitz, 1993). In this report we will review trimming reactions catalyzed by a hammerhead ribozyme, summarizing our data on transcribed ribozymes, and the results of our molecular orbital calculations; finally we suggest that the mechanisms might converge as one unique and universal mechanism, exploited not only by various

kinds of ribozyme but also by artificially created metal-ion-dependent DNAzymes and other RNA-cleaving agents that are yet to be identified.

II. Results and Discussion A. Rates of transcription and cleavage are similar for cis-acting hammerhead ribozymes We previously constructed a novel transcription system that allows trimming of both the 5' and the 3' termini of any RNA transcript by a cis-acting ribozyme (Taira et al., 1990b, 1991b; Yuyama et al., 1992; Ohkawa et al., 1993). The vector consists of a promoter, the "5' Processing Ribozyme", the DNA template (any DNA) to be transcribed, and the "3' Processing Ribozyme" (the pGENE8459 Series of vectors, Fig. 1). In our original construct, designated pGENE8459v3, a trans-acting ribozyme (called "Ribozyme for SFL1"; Fig. 2, Left) was placed between the "5' Processing Ribozyme" and the "3' Processing Ribozyme" (Taira et al., 1991b). To characterize self-processing reactions during transcription and, in particular, to estimate the relative rates of both reactions, the RNAs obtained as products of transcription of the covalently closed circular (nonlinearized), as well as HindIII-linearized, pGENE8459v3 plasmids were analyzed at several time points (Fig. 3). When the RNA transcripts from the HindIII-linearized pGENE8459v3 were examined (after run-off transcription), six products were obtained. The various bands represent the following: Band 1: Initial run-off transcription product (173 nt). Band 2: Product of partial cleavage, produced by the action of the "5' Processing Ribozyme" (120 nt). Band 3: Product of partial cleavage, produced by the action of the "3' Processing Ribozyme" (114 nt). Band 4: Final product (Ribozyme for SFL1; 61 nt) Band 5: Final product (3' Processing Ribozyme; 59 nt) Band 6: Final product (5' Processing Ribozyme; 53 nt) Next page:

Figure 1. The pGENE8459 series of vectors for transcription of RNA. Any DNA template can be inserted between two cis-acting ribozyme sequences. The first ribozyme, called "5' Processing Ribozyme", trims the 5'-region of the transcribed inserted gene and, similarly, the second ribozyme, called "3' Processing Ribozyme", trims the 3'region of the transcribed RNA. It is, thus, not necessary to linearize the DNA template in order to obtain RNA transcripts with defined 5' as well as 3' ends.

Figure 2. Structure of the pGENE8459v3 vector. The "5'-Processing Ribozyme" has a second binding site (right) which is inactive in terms of 5'-end trimming.

Figure 3. Kinetics of self-cleavage and transcription reactions. RNA transcripts were obtained by transcription from the same amount of either circular or HindIII-linearized pGENE8459v3 template. Sampling was performed 30, 60, 140, and 360 min after initiation of the transcription reaction.

The initial full-length RNA transcript of 173 nucleotides (nt), consisting of "5' Processing Ribozyme", "Ribozyme for SFL1", and "3' Processing Ribozyme" (band 1), is ultimately cleaved spontaneously into three fragments (bands 4-6) by the action of both the "5' Processing Ribozyme" and the "3' Processing Ribozyme". Two fragments arising from partial digestion of the full-length RNA are also discernible: the first one (band 2) is a faint band of 120 nt consisting of "Ribozyme for SFL1" and the "3' Processing Ribozyme" produced by the action of the "5' Processing Ribozyme"; and the second one (band 3) is a distinct band of 114 nt consisting of the "5' Processing Ribozyme" and "Ribozyme for SFL1" which was produced from the full-length transcript by the action of the "3' Processing Ribozyme". It is important to note that, despite the equimolar amount of linearized pGENE8459v3 template, more than 10 times higher level of RNA product (band 4) was produced from the DNA template (insert) in Figure 1 when the circular template was used: note the more prominent 61 nt band generated from the circular template. Therefore, the efficiency of transcription is much higher for the circular template, possibly because (i) RNA polymerase prefers a circular template over a linearized template (when both circular and linearized pGENE8459v3 DNAs were mixed and used as templates, almost no products of transcription arising of the linearized template were observed); and (ii) with the circular template, "rolling-circle" transcription is possible, which circumvents kinetically inefficient diffusion-controlled association/ dissociation processes. The type of vector used here is particularly suitable for preparation of uniform RNAs; e.g., for NMR measurements or for crystallization. It is known that DNAs with heterologous 5' or 3' ends hinder crystallization. Similarly, RNAs with heterologous ends are expected to hinder crystallization. When our constructs are used, uniform RNAs with defined 5' and 3' ends can be produced. Indeed, our trimming vector has been proven to be extremely useful for the synthesis of short RNAs (Price et al., 1995). Moreover, it is possible to concatenate entire units in tandem. When 10 units are concatenated, the yield of RNA transcripts increases 10-fold, as compared to the results of transcription from a DNA template with only one unit (of course, in both transcriptions, equimolar amounts of DNA template were used). In addition, different types of RNA sequence can be produced, depending on the kind of insert in the "DNA Template" region indicated in Figure 1 (Ohkawa et al., 1993; Price et al., 1995). Therefore, by concatenating several units, each of which contains a different "DNA Template", we can produce several types of RNA. This methodology is especially useful when ribozymes are to be used against HIV, because HIV is infamous for its high frequency of mutation, which incidentally poses problems for the immune system that is already depressed after HIV-1 infection. This genetic polymorphism not only makes it far more difficult than might be anticipated to find a vaccine for HIV-1 but also poses a challenge to the use of ribozymes as a form of treatment. Although a ribozyme has high sequence-specificity, once the nucleotide sequence in the target RNA chain has been altered, the ribozyme can lose its effectiveness. Nonetheless, although HIV does exhibit such genetic variability, this variability is not limitless. Changes do not occur as frequently in those sections of the RNA chain that code for significant viral functions. Ribozymes that are targeted simultaneously to a number of these highly conserved and less mutable sites should prove to be effective anti-viral agents. Even if one or two of the sites were altered, as long as one or more unaltered sites remained, these sites would be attacked by ribozymes, with resultant inactivation of the functional virus. The probability that every one of the functionally significant sites would undergo simultaneous alteration is extremely low, if not zero. A detailed description of this system of treatment of HIV infection is, however, beyond the scope of this paper. Having characterized the properties of our vector, we can now return to Figure 3 and analyze the rates of transcription and cleavage reactions. Examination of the intensities of the bands from the linearized template (bands 1-6) allows us to conclude that the rates of the transcription and cleavage reactions in vitro are similar. The initial product of transcription (band 1) undergoes self-cleavage to produce the partially digested products (bands 2 and 3). If the rate of transcription were much higher than the rate of cleavage, one would expect an increase in the intensity of band 1. However, if the rate of cleavage were much higher than the rate of transcription, one would expect more intense bands of the completely cleaved products (bands 4-6) with almost no intermediates (bands 2 and 3). Since all six bands are recognizable, the rate of transcription and the rate of cleavage must be similar. Nonetheless, a comparison of bands 2 and 3 reveals that the "3' Processing Ribozyme" appears more active than the "5' Processing Ribozyme", because the intensity of band 2 is much lower than that of band 3. Note here that the material in band 2 was degraded into "Ribozyme for SFL1" (band 4) and "3' Processing Ribozyme" (band 5) by the action of the "3' Processing Ribozyme". However, we now find that the "5' Processing Ribozyme" has a second binding site, and that it forms an inactive complex with respect to the cleavage reaction (see the right hand panel of Fig. 2). It is important to note that, in general, ribozymes tend to form inactive complexes when there exists an alternative binding site. Therefore, care must be taken in choosing the target sequence of a

ribozyme. In fact, the removal of the second binding site (removal of the SacI site) accelerated the cleavage of the intermediate (band 3; data not shown). Therefore, the actual rate of cleavage is higher than the anticipated rate based on the intensity of band 3 in Figure 3. The results described above mean that the cleavage of the cis-acting hammerhead ribozyme occurs more rapidly or at least at a rate similar to the rate of transcription in vitro. Since natural hammerhead ribozymes act in cis during replication by the rolling-circle mechanism (Symons, 1992), there has been no selective advantage to further improvements in the chemical-cleavage step, with respect to its natural function. So it is possible that the active site of the hammerhead ribozyme has not been perfected. In fact, we obtained a chimeric RNA/DNA ribozyme which has much higher activity (under extreme conditions with the cleavage rate constant of 100 min-1; Shimayama et al., 1995) than a natural hammerhead ribozyme (that usually has the cleavage rate constant of about 1 min-1), supporting the possibility to engineer the natural ribozyme or select in vitro of artificial RNA-cleaving agents that are better cleavers of RNAs.

B. Ribozymes are metallo-enzymes: the Mg2+-binding motif of RNA-cleaving agents Here we describe the binding motif of Mg2+ ions. The molecular orbital calculations that we will discuss in this section strongly support a more direct role for the Mg2+ ion as the real catalyst in RNA-cleavage

reactions. Since we are interested in the energetics in RNA-cleavage reactions, we have carried out ab initio molecular orbital calculations using several model compounds (Taira et al., 1990a, 1991a, 1993; Uchimaru et al., 1991, 1992; Storer et al., 1991; Yliniemela et al., 1993; Uebayasi et al., 1994; Zhou and Taira, 1998), and the calculations hint at a positive role for the Mg2+ ion in catalysis (Uebayasi et al., 1991;

Uchimaru et al., 1993; Uebayasi et al., 1994; Zhou and Taira, 1998). Moreover, we analyzed Tetrahymena ribozyme reactions and the quantitative details have been published elsewhere (Uchimaru et al., 1993). In this section we will discuss the role of Mg2+ ion qualitatively.

In Tetrahymena ribozyme reactions, the transesterification reaction is initiated by the attack of the 3'hydroxyl group of the bound guanosine (G) on a phosphodiester linkage to generate the cleaved upstream exon with a 3'-hydroxyl group and an intron with 5'-G (Fig. 4). Divalent magnesium ions are commonly indispensable as cofactors for the self-cleavage of phosphodiester linkages in ribozyme-catalyzed reactions. The self-splicing reactions proceed with inversion of the configuration at the phosphorus center and, thus, an in-line mechanism (SN2(P) process) appears the most likely (McSwiggen et al., 1989; Rajagopal et al., 1989). Consequently, a pentacoordinate oxyphosphorane intermediate/ transition state is postulated for consecutive transesterification reactions of the Tetrahymena rRNA splicing process (structure shown in parentheses in Fig. 4). We have analyzed the electrostatic potential, which represents the energy of interaction between a positive charge and a negative charge of a molecule. Thus, the electrostatic potential should be useful for interpretation of ionic interactions (Uchimaru et al., 1993; Uebayasi et al., 1994). In our model system, we examined the interaction between dianionic trimethoxyphosphorane and the Mg2+ ion (overall, a neutral complex). As shown in parentheses in Figure 4, in the transition state for the transesterification process, most negative charges (up to two charges) are localized on the non-bridging phosphoryl oxygens and this observation appears to support the coordination of a Mg2+ ion between these two non-bridging oxygens.

However, our molecular orbital calculations indicate, instead, that the Mg 2+ coordination occurs preferably in the region between the bridging and non-bridging oxygens, as indicated by the shaded areas in Figure 5A and as also depicted by the circled Mg2+ ions in Figure 4 (Uchimaru et al., 1993). This conclusion is

consistent with the recent findings that the Tetrahymena ribozyme has two metal ions at its catalytic center and that each metal ion interacts with the leaving 3'-oxygen (Piccirilli et al., 1992) and the attacking 3'hydroxyl group of the bound guanosine (G) (Weinstein et al., 1997), respectively. The more symmetrical transition state of the catalytic center of the Tetrahymena ribozymes, as compared to that of the hammerhead ribozyme, makes it easier to carry out expensive molecular orbital calculations. In nature, the symmetrical transition state of the Tetrahymena ribozyme can be used not only to cleave the bonds but also to ligate the bonds. In fact, the Tetrahymena ribozyme has been shown to have polymerase activities (Been and Cech, 1988; Weinstein et al., 1997). Since the Tetrahymena ribozyme does not utilize the 2'-oxygen on the ribose ring of the cleavage site, it can even cleave DNAs (Hershlag and Cech, 1990; Robertson and Joyce, 1990). This kind of property of Tetrahymena-type ribozymes could be advantageous during the development of the DNA world.

A similar electrostatic potential analysis of a hammerhead ribozyme indicates that the Mg2+ coordination occurs preferably in the region between the bridging and non-bridging oxygens, as indicated by the shaded areas in Figure 5B. Further analysis demonstrates that magnesium ion itself is capable of cleaving (or forming from the principle of microscopic reversibility) of a phosphorus-oxygen bond by a direct coordination to the translating oxygen (Uebayasi et al., 1994). In this scenario, the direct coordination of the metal ion with the 2'-oxygen of the attacking nucleotide residue, as shown in Figure 6, polarizes and weakens the 2'-OH bond. As a result, higher concentrations of the active nucleophile, the metal-bound-2'alkoxide of the ribose, becomes available. Therefore, an inverse correlation between the pKa of the metalbound ribose 2'-OH and the ribozyme activity holds. In other words, the lower is the pKa of the metal ion, the higher is the cleavage rate at a given concentration of the metal ion at a fixed pH. Similarly, the direct coordination of the metal ion, that acts as a Lewis acid, with the 5'-oxygen of the leaving nucleotide residue weakens the P-(5'-O) bond. Metal ions with lower pKa values will weaken the phosphorus-(5'oxygen) bond to a greater extent, thereby, activating the ribozyme-mediated cleavage to a greater extent. This kind of metal-ion-binding motif utilized by the Tetrahymena and hammerhead ribozymes, in which the Mg2+ ions coordinate directly with the attacking and leaving oxygens, appears to be conserved even in the protein world. DNA polymerase I from E. coli is a metallo-enzyme and it uses two Mg2+ ions (Beese and Steitz, 1991; Steitz and Steitz, 1993): the coordination sites of these Mg2+ ions are between the

bridging and non-bridging oxygens, in agreement with the results of calculations shown in Figure 5. Next page:

Figure 4. Splicing reactions of the Tetrahymena pre-rRNA. Chemical structures for the splicing reactions are shown at the bottom. The structure of possible transition state structure is depicted in parentheses.

Figure 5. The side view of a three-dimensional representation of the surface of constant electrostatic potential for trimethyloxyphosphorane (A), a model for the transition state structure of the Tetrahymena ribozyme shown in parentheses in Figure 4, and for methylethylene oxyphosphorane (B), a model for the transition state structure of a hammerhead ribozyme shown in Figure 6. The shaded areas represent the regions with most negative electrostatic potential. These regions are the most favorable sites for the coordination of Mg2+ ions. Note that, although dianionic oxyphosphorane concentrates its negative charges on the non-bridging phosphoryl oxygens, the coordination of an Mg2+ ion between these two non-bridging oxygens is unlikely.

III. Conclusion We suggest that the catalytic center of the hammerhead ribozyme has not been perfected for chemicalcleavage reactions because, in its natural role as a cis-acting ribozyme, there has not been strong evolutional pressure towards such perfection since the chemical step appears to be more rapid than the replication processes. Therefore, it seems possible to improve the chemical-cleavage step by, for example, the use of in vitro and in vivo selection procedures or combinatorial chemistry. Molecular orbital calculations predict that a Mg2+ ion does not bifurcate between the most negatively charged non-bridging oxygens in the transition state of the transesterification reactions and, instead, the preferred Mg2+-coordination site is in the space between the bridging and non-bridging oxygens, where Mg2+ ions can act as Lewis acid

catalysts, facilitating the formation and cleavage of phosphorus-oxygen bonds. Finally, RNA-cleaving mechanisms might converge as one unique and universal mechanism, exploited not only by various kinds of ribozyme but also by artificially created metal-ion-dependent DNAzymes and other RNA-cleaving agents that are yet to be identified.

IV. Experimental procedures A. Multitarget-ribozyme expression plasmid (the pGENE8459 series of vectors) and cleavage activities of transcribed ribozymes As described previously, sequences of the various constructed plasmids were confirmed using a DNA Sequencer (model 373A; Perkin Elmer, Applied Biosystems, Foster City, CA, Taira et al., 1991b, 1992). Transcription was carried out in a total volume of 25 mL of solution that contained 2 mL of 5x transcription buffer (1x = 200 mM Tris-HCl, pH 7.5; 30 mM MgCl2; 10 mM spermidine; 0.05% bovine serum albumin); 1.25 mL of 0.2 M DTT; 2.5 mL of NTP mix (500 mM each of UTP, ATP, CTP, and GTP); 1.25 mL of human placental ribonuclease inhibitor (20 units/mL; Toyobo, Tokyo); 0.5 mL of [a-32P] CTP (20 mCi/mL, ~800 Ci/mmol); 2.5 mL of template DNA solution (pGENE8459v3 or pGENE8459v3 with the SacI site removed; 1 mg/mL); and 0.65 mL of T7 RNA polymerase (20 units/ mL; Amersham, Tokyo). Transcription reactions were carried out and kinetics were analyzed at 37째C. The products of transcription and cleavage were analyzed by reference to sequencing ladders of pGENE8459v3 on a 6% polyacrylamide gel that contained 8.3 M urea.

Figure 6. The double-metal-ion mechanism of catalysis for reactions catalyzed by hammerhead ribozymes (Steitz and Steitz, 1993; Uebayasi et al., 1994; Sawata et al., 1995; Pontius et al., 1997; Zhou et al., 1997; Lott et al., 1998; Zhou and Taira, 1998).

B. Molecular orbital calculations GAUSSIAN 88 (Frish et al, 1988) and GAUSSIAN 90 (Frish et al, 1990) program packages were used for geometry optimizations and analyses of Mulliken populations and natural bond orbitals (NBO). SPARTAN 90 (Carpenter et al, 1990) generated the three-dimensional representations of molecular structures and electrostatic potentials. All the calculations in the present work were performed at the Hartree-Fock level. Dianionic trimethoxyphosphorane (Fig. 5A) , which is a model for the reaction center of the Tetrahymena ribozyme, and the locations of Mg2+ ions relative to the trimethoxyphosphorane were geometrically optimized. In addition, electrostatic potential calculations for the trimethoxyphosphorane dianion were performed using self-consistent-field (SCF) densities. For this purpose the 6-31G* optimized structure was used. Further details of the procedures are available elsewhere (Uchimaru et al., 1993). Similarly, dianionic cyclic phosphorane (Fig. 5B) was used as a model compound for the transition state of a hammerhead ribozyme (Uebayasi et al., 1994).

References Altman S. (1989). Ribonuclease P: an enzyme with a catalytic RNA subunit. Adv. Enzymol. 62, 1-36. Been, M. D. and Cech, T. R. (1988). RNA as an RNA polymerase: net elongation of an RNA primer catalyzed by the Tetrahymena ribozyme. Science 239, 1412-1416. Beese, L. S. and Steitz, T. A. (1991). Structural basis for the 3'-5' exonuclease activity of Escherichia coli DNA polymerase: a two metal ion mechanism. EMBO J. 10, 25-33. Bartel, D. P., and Szostak, J. W. (1993). Isolation of new ribozymes from a large pool of random sequences. Science 261, 1411-1418. Breaker, R. R., and Joyce, G. F. (1994) A DNA enzyme that cleaves RNA. Chem. Biol. 1, 223-229. Breaker, R. R., and Joyce, G. F. (1995). A DNA enzyme with Mg2+-dependent RNA phosphoesterase activity. Chem. Biol. 2, 655-660. Carmi, N., Shultz, L. A., and Breaker, R. R. (1996). In vitro selection of self-cleaving DNAs. Chem. Biol. 3, 1039-1046. Carpenter, J. E., Hehre, W. J., and Kahn, S. D. (1990). SPARTAN, version 1.0. Wavefunction Inc. 18401 Von Karman, #370 Irvine, CA 92715. Cech, T. R. (1990). Self-splicing and enzymatic activity of an intervening sequence RNA from Tetrahymena (Nobel Lecture). Angew. Chem. Int. Ed. Engl. 29, 759-768. Cuenoud, B., and Szostak, J. W. (1995). A DNA metalloenzyme with DNA ligase activity. Nature 375, 611-614. Dai, X., Mesmaeker, A. D., and Joyce, G. F. (1995). Cleavage of an amide bond by a ribozyme. Science 267, 237-240. Eckstein, F., and Lilly, D. M. (eds.) (1996). Catalytic RNA, Nucleic Acids and Molecular Biology. vol. 10 (Berlin, Germany, Springer-Verlag). Faulhammer, D., and Famulok, M. (1996). Ca2+ as a cofactor for a novel RNA-cleaving deoxyribozyme. Angew. Chem. Int. Ed. Engl. 35, 2837-2841. Frish, M. J., Head-Gordon, M., Schlegel, H. B., Raghavachari, K., Binkley, J. S., Gonzalez, C., Defrees, D. J., Fox, D. J., Whiteside, R. A., Seeger, R., Melius, C. F., Baker, J., Martin, R., Kahn, L. R., Stewart, J. J. P., Fluder, E. M., Topiol, S., and Pople, J. A. (1988). Gaussian Inc., Pittsburgh, PA. Frish, M. J., Head-Gordon, M., Trucks, G. W., Foresman, J. B., Schlegel, H. B., Raghavachari, K., Robb, M. A., Binkley, J. S., Gonzalez, C., Defrees, D. J., Fox, D. J., Whiteside, R. A., Seeger, R., Melius, C. F., Baker, J., Martin, R. L., Kahn, L. R., Stewart, J. J. P., Topiol, S., and Pople, J. A. (1990). Gaussian Inc., Pittsburgh, PA. Gesteland, R. F., and Atkins, J. F. (eds.) (1993). The RNA World (NY, Cold Spring Harbor Press). Goodchild, J. (1992). Enhancement of ribozyme catalytic activity by a contiguous oligodeoxynucleotide (facilitator) and by 2'-O-methylation. Nucleic Acids. Res. 20, 4607-4612. Haseloff, J., and Gerlach, W. L. (1988). Simple RNA enzymes with new and highly specific endoribonuclease activities. Nature 334, 585-591. Hendry, P., McCall, M. J., Santiago, F. S., and Jennings, P. A. (1992). A ribozyme with DNA in the hybridising arms displays enhanced cleavage ability. Nucleic Acids Res. 20, 5737-5741. Herschlag, D. and Cech, T. R. (1990). DNA cleavage catalysed by the ribozyme from Tetrahymena. Nature 344, 405-409. Kobayashi, S., Amikura, R., and Okada, M. (1993). Presence of mitochondrial large ribosomal RNA outside mitochondria in germ plasm of Drosophila melanogaster. Science 260, 1521-1524. Kobayashi, S., Amikura, R., Nakamura, A., Saito, H., and Okada, M. (1995). Mislocalization of oskar product in the anterior pole results in ectopic localization of mitochondrial large ribosomal RNA in Drosophila embryos. Dev. Biol. 169, 384-386. Lorsch, J. R., and Szostak, J. W. (1994). In vitro evolution of new ribozymes with polynucleotide kinase activity. Nature 371, 31-36.

Lott, W. B., Pontius, B. W., and von Hippel, P. H. (1998). A two-metal ion mechanism operates in the hammerhead ribozyme-mediated cleavage of an RNA substrate. Proc. Natl. Acad. Sci. USA 95, 542-547. McCall, M. J., Hendry, P., and Jennings, P. A. (1992). Minimal sequence requirements for ribozyme activity. Proc. Natl. Acad. Sci. USA 89, 5710-5714. McSwiggen, J. A., and Cech, T. R. (1989). Stereochemistry of RNA cleavage by the Tetrahymena ribozyme and evidence that the chemical step is not rate-limiting. Science 244, 679-683. Nishikawa, S., Nakagawa, K., Kimura, T., Tanaka, M., Yuyama, N., Shimayama, T., Ohkawa, J., Inokuchi, Y., Miyashiro, H., Maeda, H., Furukawa, K., and Taira, K. (1991). In Fundamental Investigations on the Creation of Biofunctional Materials, S., Okamura, T., Tsuruta, Y., Imanishi, and J., Sunamoto, eds. (Kyoto, Kagaku-Dojin), pp 109-123 . Nitta, I., Ueda, T., and Watanabe, K. (1998). Possible involvement of Escherichia coli 23S ribosomal RNA in peptide bond formation. RNA 4, 257-267. Noller, H. F., Hoffarth, V., and Zimniak, L. (1992). Unusual resistance of peptidyl transferase to protein extraction procedures. Science 256, 1416-1419. Ohkawa, J., Yuyama, N., Takebe, Y., Nishikawa, S., and Taira, K. (1993). Importance of independence in ribozyme reactions: kinetic behavior of trimmed and of simply connected multiple ribozymes with potential activity against human immunodeficiency virus. Proc. Natl. Acad. Sci. USA 90, 11302-11306. Paolella, G., Sproat, B. S., and Lamond, A. I. (1992). Nuclease resistant ribozymes with high catalytic activity. EMBO J. 11, 1913-1919. Perreault, J.-P., Wu, T., Cousineau, B., Ogilvie, K. K., and Cedergren, R. (1990). Mixed deoxyribo- and ribooligonucleotides with catalytic activity. Nature 344, 565-567. Perreault, J.-P., Labuda, D., Usman, N., Yang, J.-H., and Cedergren, R. (1991). Relationship between 2'hydroxyls and magnesium binding in the hammerhead RNA domain: a model for ribozyme catalysis. Biochemistry 30, 4020-4025. Piccirilli, J. A., McConnell, T. S., Zaug, A. J., Noller, H. F., and Cech, T. R. (1992). Aminoacyl esterase activity of the Tetrahymena ribozyme. Science 256, 1420-1424. Pieken, W. A., Olsen, D. B., Benseler, F., Aurup, H., and Eckstein, F. (1991). Kinetic characterization of ribonuclease-resistant 2'-modified hammerhead ribozymes. Science 253, 314-317. Price, S. R., Ito, N., Oubridge, C., Avis, J. M., and Nagai, K. (1995). Crystallization of RNA-protein complexes I. Methods for the large-scale preparation of RNA suitable for crystallographic studies. J. Mol. Biol. 249, 398-408. Pontius, B. W., Lott, W. B., and von Hippel, P. H. (1997). Observations on catalysis by hammerhead ribozymes are consistent with a two-divalent-metal-ion mechanism. Proc. Natl. Acad. Sci. USA 94, 2290-2294. Rajagopal, J., Doudna, J. A., and Szostak, J. W. (1989). Stereochemical course of catalysis by the Tetrahymena ribozyme. Science 244, 692-694. Robertson, D. L. and Joyce, G. F. (1990). Selection in vitro of an RNA enzyme that specifically cleaves singlestranded DNA. Nature 344, 467-468. Santoro, S. W., and Joyce, G. F. (1997). A general purpose RNA-cleaving DNA enzyme. Proc. Natl. Acad. Sci. USA 94, 4262-4266. Sarver, N., Cantin, E. M., Chang, P. S., Zaia, J. A., Ladne, P. A., Stephens, D. A., and Rossi, J. J. (1 990). Ribozymes as potential anti-HIV-1 therapeutic agents. Science 247, 1222-1225. Sawa, H. and Shimura, Y. (1992). Association of U6 snRNA with the 5'-splice site region of pre-mRNA in the spliceosome. Genes Dev. 6, 244-254. Sawata, S., Komiyama, M., and Taira, K. (1995). Kinetic evidence based on solvent isotope effects for the nonexistence of a proton-transfer process in reactions catalyzed by a hammerhead ribozyme: implication to the double-metal-ion mechanism of catalysis. J. Am. Chem. Soc. 117, 2357-2358. Scanlon, K. J. (ed.) (1997). Therapeutic Applications of Ribozymes, Methods in Molecular Medicine. vol. 11 (New Jersey, Humana Press). Shimayama, T., Sawata, S., Komiyama, M., Takagi, Y., Tanaka, Y., Wada, A., Sugimoto, N., Rossi, J. J., Nishikawa, F., Nishikawa, S., and Taira, K. (1992). Substitution of non-catalytic stem and loop regions of hammerhead ribozyme with DNA counterparts only increases K M without sacrificing the catalytic step (kcat): a way to improve substrate-specificity. Nucleic Acids. Symp. Ser. 27, 17-18. Shimayama, T., Sawata, S., and Taira, K. (1993). Nuclease-resistant chimeric ribozymes containing deoxyribonucleotides and phosphorothioate linkages. Nucleic Acids Res. 21, 2605-2611. Shimayama, T., Nishikawa, S., and Taira, K. (1995). Extraordinary enhancement of the cleavage activity of a DNAarmed hammerhead ribozyme at elevated concentrations of Mg2+ ions. FEBS Lett. 368, 304-306. Steitz, T. A., and Steitz, J. A. (1993). A general two-metal-ion mechanism for catalytic RNA. Proc. Natl. Acad. Sci. USA 90, 6498-6502. Storer, J. W., Uchimaru, T., Tanabe, K., Uebayasi, M., Nishikawa, S., and Taira, K. (1991). Existence of a marginally stable intermediate during the base-catalyzed methanolysis of ethylene phosphate and ab initio studies of the monohydration of the pentacoordinated oxyphosphorane intermediate. J. Am. Chem. Soc. 113, 5216-5219. Symons, R. H. (1992). Small catalytic RNAs. Annu. Rev. Biochem. 61, 641-671. Taira, K., Uebayasi, M., Maeda, H., and Furukawa, K. (1990a). Energetics of RNA cleavage: implications for the mechanism of action of ribozymes. Protein Eng. 3, 691-701.

Taira, K., Oda, M., Shinshi, H., Maeda, H., and Furukawa, K. (1990b). Construction of a novel artificial-ribozymereleasing plasmid. Protein Eng. 3, 733-737. Taira, K., Uchimaru, T., Tanabe, K., Uebayasi, M., and Nishikawa, S. (1991a). Rate limiting P-O(5') bond cleavage of RNA fragment: ab initio molecular orbital calculations on the base-catalyzed hydrolysis of phosphate. Nucleic Acids Res. 19, 2747-2753. Taira, K., Nakagawa, K., Nishikawa, S., and Furukawa, K. (1991b). Construction of a novel RNA-transcripttrimming plasmid which can be used both in vitro in place of run-off and (G)-free transcriptions and in vivo as multi-sequences transcription vectors. Nucleic Acids Res. 19, 5125-5130. Taira, K., Hirose, J., Hayashida, S., and Furukawa, K. (1992). Analysis of bph operon from the polychlorinated biphenyl-degrading strain of Pseudomonas pseudoalcaligenes KF707. J. Biol. Chem. 267, 4844-4853 Taira, K., and Nishikawa, S. (1992). Construction of several kinds of ribozymes: Their reactivities and utilities. In Gene regulation: Biology of antisense RNA and DNA, R. P. Erickson and J. G. Izant, eds. (New York, Raven Press), pp. 35-54. Taira, K., Uchimaru, T., Storer, J. W., Uebayasi, M., and Tanabe, K. (1993). Properties of dianionic oxyphosphorane intermediates: implication to the reaction profile for base-catalyzed RNA hydrolysis. J. Org. Chem. 58, 3009-3017. Taylor, N. R., Kaplan, B. E., Swiderski, P., Li, H., and Rossi, J. J. (1992). Chimeric DNA-RNA hammerhead ribozymes have enhanced in vitro catalytic efficiency and increased stability in vivo. Nucleic Acids Res. 20, 4559-4565. Turner, P. C. (ed.) (1997). Ribozyme Protocols, Methods in Molecular Biology. vol. 74 (New Jersey, Humana Press). Uchimaru, T., Tanabe, K., Nishikawa, S., and Taira, K. (1991). Ab initio studies of a marginally stable intermediate in the base-catalyzed methanolysis of dimethyl phosphate and nonexistence of the stereoelectronically unfavorable transision state. J. Am. Chem. Soc. 113, 5216-5219. Uchimaru, T., Storer, J. W., Tanabe, K., Uebayasi, M., Nishikawa, S., and Taira, K. (1992). RNA hydrolysis via an oxyphosphorane intermediate. Biochem. Biophys. Res. Commun. 187, 1523-1528. Uchimaru, T., Uebayasi, M., Tanabe, K., and Taira, K. (1993). Theoretical analyses on the role of Mg2+ ions in ribozyme reactions. FASEB J. 7, 137-142. Uebayasi, M., Uchimaru, T., Tanabe, K., Nishikawa, S., and Taira, K. (1991). Preferential chelation of cationic ligands to axial-equatorial oxygens over equatorial-equatorial dianionic oxygens: implication to the mechanism of action of ribozymes. Nucleic Acids Symp. Ser. 25, 107-108. Uebayasi, M., Uchimaru, T., Koguma, T., Sawata, S., Shimayama, T., and Taira, K. (1994). Theoretical and experimental considerations on the hammerhead ribozyme reactions: divalent magnesium ion-mediated cleavage of phosphorus-oxygen bonds. J. Org. Chem. 59, 7414-7420. Uhlenbeck, O. C. (1987). A small catalytic oligoribonucleotide. Nature 328, 596-600. Vaish, N. K., Heaton, P. A., Fedorova, O., and Eckstein, F. (1998). In vitro selection of a purine nucleotide-specific hammerhead like ribozyme. Proc. Natl. Acad. Sci. USA 95, 2158-2162. Warashina, M., Takagi, Y., Sawata, S., Zhou, D,-M., Kuwabara, T., and Taira, K. (1997). Entropically driven enhancement of cleavage activity of a DNA-armed hammerhead ribozyme: mechanism of action of hammerhead ribozymes. J. Org. Chem. 62, 9138-9147. Weinstein, L. B., Jones, B. C., Cosstick, R., and Cech, T. R. (1997). A second catalytic metal ion in group I ribozyme. Nature 388, 805-808. Williams, D. M., Pieken, W. A., and Eckstein, F. (1992). Function of specific 2'-hydroxyl groups of guanosines in a hammerhead ribozyme probed by 2' modifications. Proc. Natl. Acad. Sci. USA 89, 918-921. Wilson, C., and Szostak, J. W. (1995). In vitro evolution of a self-alkylating ribozyme. Nature 374, 777-782. Yang, J.-H., Perreault, J.-P., Labuda, D., Usman, N., and Cedergren, R. (1990). Mixed DNA/RNA polymers are cleaved by the hammerhead ribozyme. Biochemistry 29, 11156-11160. Yang, J.-H., Usman, N., Chartrand, P., and Cedergren, R. (1992). Minimum ribonucleotide requirement for catalysis by the RNA hammerhead domain. Biochemistry 31, 5005-5009. Yliniemela, A., Uchimaru, T., Tanabe, K., and Taira, K. (1993). Do pentacoordinate oxyphosphorane intermediates always exist? J. Am. Chem. Soc. 115, 3032-3033. Yuyama, N., Ohkawa, J., Inokuchi, Y., Shirai, M., Sato, A., Nishikawa, S., and Taira, K. (1992). Construction of a tRNA-embedded-ribozyme trimming plasmid. Biochem. Biophys. Res. Commun. 186, 1271-1279. Zhang, B., and Cech, T. R. (1997). Peptide bond formation by in vitro selected ribozymes. Nature 390, 96-100. Zhou, D. M., Zhang, L. H., and Taira, K. (1997). Explanation by the double-metal-ion mechanism of catalysis for the differential metal ion effects on the cleavage rates of 5'-oxy and 5'-thio substrates by a hammerhead ribozyme. Proc. Natl. Acad. Sci. USA 94, 14343-14348. Zhou, D.-M., and Taira, K. (1998). The hydrolysis of RNA: from theoretical calculations to the hammerhead ribozyme-mediated cleavage of RNA. Chem. Rev. 98, 991-1026.

Gene Ther Mol Biol Vol 2, 95-101. August 7, 1998.

Regulation of neuronal differentiation and apoptosis by Brn-3 POU family transcription factors: - potential use in gene therapy (review) David S. Latchman Department of Molecular Pathology, Windeyer Institute of Medical Sciences, University College London Medical School, The Windeyer Building, 46 Cleveland Street, London W1P 6DB, UK ________________________________________________________________________ ___________________________ Correspondence: David S Latchman, Ph.D., Tel: +44-171-380-9343; Fax: +44-171-387-3310; E-mail: d.latchman@ucl.ac.uk Received 26 May 1998; accepted 26 May 1998

Summary The Brn-3 sub group of POU family transcription factors has three members:Brn-3a, Brn-3b and Brn-3c, all of which are expressed in distinct but overlapping groups of neuronal cells in the developing and adult nervous systems. Although these factors are closely related to one another, they have distinct functions. Thus, Brn-3a activates the expression of a number of different genes expressed in neuronal cells whereas Brn-3b represses their expression and inhibits activation by Brn-3a. These stimulatory effects of Brn-3a are paralleled by its ability to stimulate neurite outgrowth by neuronal cells and to protect them from apoptosis/programmed cell death. Thus Brn-3a may be of potential use in a gene therapy approach to human neurological diseases which involve either a failure of neurite outgrowth, such as spinal injury, or excessive neuronal cell death such as Alzheimer's or Parkinson's diseases.

I. Introduction The POU (Pit-Oct-Unc) family of transcription factors was originally identied on the basis of a 150-160 amino acid domain which was common to the mammalian transcription factors Pit-1, Oct-1 and Oct-2 and the nematode regulatory protein Unc-86 (for reviews see Verrijzer and van der Vliet, 1993; Ryan and Rosenfeld, 1997). The POU domain constitutes the DNA binding domain of these factors and is divided into a POU-specic domain and a POU homeodomain which is homologous to that found in the homeobox family of transcription factors. These two sub domains are separated by a short linker region (Figure 1). The four original POU factors all play critical roles in the regulation of gene expression particularly in neuronal cells. Thus, for example, Pit-1 was originally dened on the basis of its being essential for normal development of the pituitary gland and its absence results in congenital dwarsm (Anderson and Rosenfeld 1994) whilst the nematode unc-86 mutation results in the absence of specic neuronal cell types in this organism (Desai et al., 1988). The essential role of these original POU factors led a number of groups to attempt to identify novel members of the POU family which might play similarly critical roles in the regulation of gene transcription in specic cell types. This was done using the approach developed by He et al., (1989) who used degenerate primers derived from conserved regions at each end of the POU domain in a reverse transcriptase/polymerase chain reaction (RT/PCR) using RNA prepared from rat brain. Evidently, RNAs capable of encoding novel POU proteins will contain these conserved regions and will thus be amplied in the RT/PCR reaction.

II. Identication of the Brn-3 POU family transcription factors We utilized this approach to identify the nature of the POU proteins which were expressed in the ND7 cell line which was obtained by the immortalization of dorsal root ganglion (DRG) sensory neurons (Wood et al., 1990). This was of particular interest since the ND7 cells can be induced to cease dividing and differentiate to a non-dividing neuronal phenotype bearing numerous neurite processes by

Figure 1. POU domain sequences of the four founder POU factors. Conserved amino acids are shown in black boxes and the ď&#x192;&#x17E;nal line shows a consensus sequence.

removal of serum from the growth medium or treatment with cyclic AMP (Wood et al., 1990; Suburo et al., 1992). Hence these cells can serve as a model for studies of neuronal differentiation. In this study (Lillycrop et al., 1992) we isolated cDNA clones derived from both Oct-1 and Oct-2 which were known to be expressed in ND7 cells. In addition however, we also isolated cDNA clones derived from Brn-3, a factor initially identied by He et al., (He et al., 1989) as a novel POU protein expressed in the brain. Moreover, we also isolated cDNA clones derived from another POU factor which was much more closely related to Brn-3 than to any of the other POU factors. We therefore renamed the original Brn-3 factor Brn-3a and named our novel factor Brn-3b (Lillycrop et al., 1992). In subsequent studies, others have referred to Brn-3a as Brn-3.0 (Gerrero et al., 1993) and to Brn-3b as Brn-3.2 (Turner et al., 1994). In addition a third member of this family Brn-3c (Ninkina et al., 1993) has been isolated and is also known as Brn-3.1 (Gerrero et al., 1993). The three Brn-3 factors constitute a closely related sub-family of POU factors which are encoded by distinct genes (Theil et al., 1993). In intial studies, all three Brn-3 POU factors were shown to be expressed predominantly in neuronal cells with distinct but overlapping patterns of expression being observed in the developing and adult brain (Lillycrop et al., 1992; Gerrero et al., 1993; Turner et al., 1994; Ninkina et al., 1993; Theil et al., 1993; Fedtsova and Turner, 1996). More recently however, expression of Brn-3a and Brn-3b has also been observed in the cervix (Ndisdang et al., 1998) and the breast and ovary where both Brn-3a and Brn-3b are able to regulate the activity of the oestrogen receptor (Budhram-Mahadeo et al., 1998). Although the importance of the Brn-3 factors in the non neuronal cells where they are expressed is not yet clear, they have been shown to play a critical role in neuronal cell function and differentiation. Thus, Brn-3a expression denes the earliest post-mitotic neurons to form in the central nervous system (Fedtsova and Turner, 1996) and its inactivation in knock out mice results in widespread losses of sensory and motor neurons leading to death of the mice shortly after birth (McEvilly et al., 1996). Similar losses of neurons are also observed in mice lacking functional Brn-3b and Brn-3c, although these are more restricted in nature affecting specically the visual and auditory systems respectively (Erkmann et al., 1996). This critical role for Brn-3c in the auditory system has recently been shown to apply in humans also, with cases of progressive deafness having been shown to result from mutation in the gene encoding Brn-3c (Vahua et al., 1998).

III. Expression of Brn-3a and Brn-3b in ND7 cell differentiation To further investigate the role of Brn-3 factors in neuronal differentiation we used the ND7 cell system and investigated whether the expression of the Brn-3 factors changed during their differentiation. Most interestingly, we observed that the expression of Brn-3a increased from a very low level in the undifferentiated cells to a much higher level in the differentiated cells which were produced following exposure to cyclic AMP or removal of serum. Conversely, Brn-3b was expressed at a high level in the undifferentiated cells and at a much lower level in the differentiated cells whereas the level of Brn-3c remained unchanged upon differentiation (Lillycrop et al., 1992; Budhram-Mahadeo et al., 1994). This observation was of particular interest in view of the results of experiments in which expression vectors containing full length cDNAs encoding Brn-3a or Brn-3b were

transfected with a test promoter containing their binding site upstream of a reporter gene. Thus, in these experiments (Budhram-Mahadeo et al., 1994; Morris et al., 1994) Brn-3a activated the promoter whereas Brn-3b repressed it and interfered with activation by Brn-3a. Similar activation by Brn-3a and repression by Brn-3b has also been observed for a number of promoters derived from genes expressed in neuronal cells such as those encoding the synaptic vesicle protein SNAP-25 (Lakin et al., 1995) the intermediate lament protein -internexin (Budhram-Mahadeo et al., 1995) and the neurolaments (Smith et al., 1997c). Hence, Brn-3a activates the expression of a number of neuronally expressed genes whilst, Brn-3b represses them. Moreover, during the process of ND7 cell differentiation the levels of the activator Brn-3a rise whilst the level of the Brn-3b repressor decreases.

IV. Role of Brn-3a and Brn-3b in ND7 cell differentiation. Both Brn-3a and Brn-3b are transcription factors which evidently act by regulating the expression of other genes encoding specic proteins. Hence, the opposite activities of these factors and their opposite changes in expression pattern during ND7 cell differentiation raises the possibility that the changes in expression of these factors might actually be involved in the differentiation event. Thus, the rise in Brn-3a and fall in Brn-3b levels could result in the activation of a number of different target genes whose protein products were required for either the growth arrest or neurite outgrowth associated with differentiation. Thus, several of the genes which are activated by Brn-3a encode intermedeate lament proteins such as the neurolaments and -internexin whilst SNAP-25 has been shown to be essential for the process of neurite outgrowth (Osen-Sand et al., 1993). To test directly the possibility that the changes in Brn-3a/Brn-3b levels were involved in the differentiation event, we prepared ND7 cell lines which over expressed either Brn-3a, Brn-3b or Brn-3c under the control of a steroid-inducible promoter. In these experiments, the induction of Brn-3a expression by steroid treatment was able to induce the ND7 cells to put out neurite processes even in the absence of the normal stimuli such as serum removal or cyclic AMP treatment which are required to induce this effect (Smith et al., 1997a). This effect was not observed in the cells containing the steriod inducible Brn-3b or Brn-3c constructs or in ND7 cells which had been transfected with plasmid expression vector lacking any insert, all of which showed no response to steroid treatment, paralleling the lack of steroid effect on parental ND7 cells (Smith et al., 1997a). Hence, the over expression of Brn-3a can indeed induce differentiation, as assayed by the outgrowth of neurite processes, in ND7 cells even in the absence of stimuli which normally induce it. This enhancement of neurite outgrowth in the cells expresssing Brn-3a was also accompanied by the enhanced expression of a number of different genes whose promoters had previously been shown to be activated by Brn-3a such as those encoding SNAP-25 (Smith et al., 1997a) and the neurolaments (Smith et al., 1997c). Hence Brn-3a is able to activate the promoters of these genes both in transfection experiments and when the endogenous gene is in its natural chromatin structure in the cell lines. Hence, the rise in Brn-3a which is produced by serum removal or cyclic AMP treatment of ND7 cells plays a direct role in the differentiation process by activating the expression of several different target genes allowing their protein products to then produce neurite

outgrowth. Interestingly, the cell lines over-expressing Brn-3b show a failure of neurite outgrowth even when stimulated by serum removal or treatment with cyclic AMP (Smith et al., 1997b). Hence, the effect of serum removal or cyclic AMP treatment on the outgrowth of neurite processes by ND7 cells is produced both by a rise in the activating transcription factor Brn-3a and by a corresponding fall in the level of the inhibitory transcription factor Brn-3b. This results in the activation of the appropriate target genes and produces neurite outgrowth.

V. Role of the Brn-3a POU domain Interestingly, it has been shown that Brn-3a exists in two different forms which are generated by alternative splicing of its RNA (Theil et al., 1993) (Figure 2). Although both these forms contain the C-terminal POU domain they differ in that the long form of Brn-3a contains an additional 84 amino acids at the N-terminus of the protein which are absent in the short form. The relative proportions of these two forms are regulated during neuronal development and in response to speciď&#x192;&#x17E;c stimuli (Liu et al., 1996). We therefore wished to determine whether both these forms would promote neurite outgrowth in ND7 cells. In fact, ND7 cell lines engineered to express either the long or the short form of Brn-3a were induced to put out neurite processes when Brn-3a expression was induced. Such induction of neurite processes was also observed in ND cell lines expressing only the isolated POU domain of Brn-3a indicating that this region of the protein is sufď&#x192;&#x17E;cient for this effect (Smith et al., 1997a). This parallels the ability of the isolated

Figure 2. Schematic diagram of the long and short forms of Brn-3a showing the genes and processes requiring only the C-terminal POU domain for induction and those which also require the N-terminal domain unique to the long form of the molecule.

POU domain of Brn-3a but not that of Brn-3b to activate the promoters of the SNAP-25 and neurolament genes (Smith et al., 1997c; Morris et al., 1997b). Hence activation of the genes involved in neurite outgrowth and of neurite outgrowth itself requires only the Cterminal POU domain (Figure 2). These effects of the POU domain of Brn-3a are not observed with the POU domain of Brn-3b which can neither induce neurite outgrowth or activate target genes. These differences in activity must be dependent upon one or more of the seven amino acid differences between the closely related POU domains of Brn-3a and Brn-3b (Lillycrop et al., 1992; Morris et al., 1994). The POU-specic domain is identical in Brn-3a and Brn-3b, the POU homeodomains differ by only a single amino acid and the remaining six differences are contained in the linker region between the two sub-domains which is poorly conserved between different POU proteins. In fact, it is the single difference at position 22 in the POU homeodomain which controls the different activities of the two factors. Thus, alteration of the isoleucine at this position in Brn-3b to the valine found at the equivalent position in Brn-3a, converts Brn-3b into an activator of an artical test promoter (Dawson et al., 1996) and of the SNAP-25 promoter (Morris et al., 1997a) and allows it to stimulate neurite outgrowth (Smith et al., 1997a). Conversely, mutant Brn-3a containing an isoleucine at this position represses the SNAP-25 promoter (Morris et al., 1997a) and inhibits neurite outgrowth (Smith et al., 1997b). Although the POU domain is the DNA binding domain of these factors, these effects are not dependent on any differences in DNA binding ability with wild type Brn-3a, Brn-3b and the mutant forms binding to DNA with equal afnity. Rather, position 22 has been shown to play a critical role in protein-protein interactions of other POU proteins (Lai et al., 1992) and it is likely therefore that the valine/isoleucine difference may control the ability of Brn-3a/Brn-3b to recruit co-activator or co-repressor molecules which are necessary for their effect on transcription.

VI. Role of Brn-3a in the regulation of apoptosis The nding that neurite outgrowth and the expression of a number of Brn-3a target genes can be induced by the isolated POU domain of Brn-3a leads to the question of the function of the remaining regions of the protein and the reason why two different forms of the protein differing at the N-terminus are generated during neuronal development. To investigate this question we attempted to identify genes which were over expressed in the cell lines over expressing the long form of Brn-3a but not in those over expressing the short form of Brn-3a. In these experiments, we were able to show that the Bcl-2 gene was strongly over expressed in the cells overexpressing the long form of Brn-3a but not in those overexpressing the short form. Moreover, in co-transfection experiments the Bcl-2 promoter could be activated by the long form of Brn-3a but not by the short form (Smith et al., 1998). This nding parallels our previous observation that the activation of the internexin promoter also requires the N-terminus of Brn-3a (Budhram-Mahadeo et a., 1995) and suggests that two classes of Brn-3a regulated genes exist with some genes being

activated by the POU domain alone and others requiring the region at the N-terminus unique to the long form of Brn-3a (Figure 2). The regulation of Bcl-2 by Brn-3a is of considerable importance because of the known ability of Bcl-2 to protect both neuronal and other cell types from programmed cell death or apoptosis (White, 1996). We therefore investigated whether the ND7 cells overexpressing the long form of Brn-3a and Bcl-2 would be protected from stimuli which induce apoptosis using a model system in which apoptosis is induced in ND7 cells by removal of serum together with addition of retinoic acid (Howard et al., 1993). In these experiments a clear protective effect was observed in the cells overexpressing the long form of Brn-3a and Bcl-2 compared to the cell death observed in cells overexpressing the short form of Brn-3a, or Brn-3b or Brn-3c neither of which induces enhanced Bcl-2 levels. Similar protection by over-expression of Brn-3a was also observed when we introduced the Brn-3a expression vector into primary cultures of dorsal root ganglion neurons or trigeminal ganglion neurons indicating that these effects are not unique to ND7 cells (Smith et al., 1998) (Figure 3). Hence, the long form of Brn-3a but not the short form is able to up regulate Bcl-2 levels and protect cells from apoptosis (Figure 2).

VII. Potential use of Brn-3a in gene therapy procedures Many human neurological diseases involve excessive cell death which can occur acutely as in stroke or chronically as in Alzheimer's or Parkinson's diseases. The ability of the long form of Brn-3a to prevent apoptotic death suggests that its over-expression may be of use in such diseases. Similarly, the ability of both the long and short forms of Brn-3a to induce neurite outgrowth could be of use in situations such as human spinal injury which involve a failure of neurite outgrowth following injury. Such alterations of Brn-3a expression for therapeutic benet could involve the use of pharmacological procedures to enhance endogenous Brn-3a expression or the use of gene therapy procedures to deliver an exogenous Brn-3a gene. Although many potential viral or non-viral means are available for such in vivo gene delivery we have concentrated on herpes simplex virus (HSV)-based vectors in view of the ability of this virus to establish latent infections specically in neuronal cells (for review see Latchman, 1990). We have now constructed disabled HSV-based vectors which can safely and efciently deliver genes to the central and peripheral nervous systems in vivo (Cofn et al., 1996; Howard et al., 1998). The long form of Brn-3a has recently been introduced into this vector and we have shown that infection with this Brn-3a expressing virus can protect trigeminal ganglion neurons from apoptosis induced by withdrawal of nerve growth factor (Smith et al., 1998) (Figure 3). Hence this virus may respresent an effective means of elevating Brn-3a levels in vivo for therapeutic benet.

Figure 3. Survival of trigeminal ganglion neurons infected with a recombinant herpes simplex virus vector expressing Brn-3a or control vector (c) either in the presence of nerve growth factor (+) or following its removal (-). Note the protective effect of infection with the Brn-3a expressing virus.

VIII. Conclusions The experiments described here have shown that the Brn-3a transcription factor is a bifunctional factor which contains two domains capable of stimulating the expression of specic genes and thereby modifying neuronal phenotype (Figure 2). Thus, the C-terminal POU domain can alone activate the transcription of specic genes and promote neurite outgrowth with this effect being opposed by the Brn-3b transcription factor. In addition however, the N-terminal region of the protein is also able to activate the expression of specic genes such as the Bcl-2 gene and thereby protect neuronal cells against apoptosis whereas Brn-3b has no effect. These ndings suggest therefore that Brn-3a is likely to have a critical role in the correct development of the nervous system by regulating both neuronal differentiation and the rate of apoptosis which plays a critical role in the proper development of the nervous system (Oppenheim, 1991). Moreover, the manipulation of its expression by gene therapy procedures may be of importance in the treatment of neurological disorders involving excessive neuronal death or a failure of neurite outgrowth. Acknowledgements I would like to thank all the members of my laboratory who have contributed to this work. This work has been generously supported by the Cancer Research Campaign, The Medical Research Council, the Sir Jules Thorn Charitable Trust and WellBeing.

References Anderson, B. and Rosenfeld, M.G. (1994) Pit-1 determines cell types during development of the anterior pituitary gland. J Biol Chem 269, 29335-29338,. Budhram-Mahadeo, V., Parker, M. and Latchman, D.S. (1998) The POU domain factors Brn-3a and Brn-3b interact with the estrogen receptor and differentially regulate transcriptional activity via an ERE. Mol Cell Biol 18, 1029-1041. Budhram-Mahadeo, V.S., Morris, P.J., Lakin, N.D., et al. (1995) Activation of the -internexin promoter by the Brn-3a transcription factor is dependent on the N-terminal region of the protein. J Biol Chem 270, 2853-2858. Budhram-Mahadeo, V.S., Theil, T., Morris, P.J., Lillycrop, K.A., Möröy, T. and Latchman, D.S. (1994) The DNA target site for the Brn-3 POU family transcription factors can confer responsiveness to cyclic AMP and removal of serum in neuronal cells. Nucl.Acids Res. 22:3092-3098. Cofn, R.S., Maclean, A.R., Latchman, D.S. and Brown, S.M. (1996) Gene delivery to the central and peripheral nervous systems of mice using HSV1 ICP34.5 deletion mutant vectors. Gene Therapy 3, 886-891. Dawson, S.J., Morris, P.J. and Latchman, D.S. (1996) A single amino acid change converts an inhibiting transcription factor into an activator. J Biol Chem 271, 11631-11633. Desai, C., Garriga, G., McIntire, S.L. and Horvitz, H.R. (1988) A genetic pathway for the development of Caenorhabditis elegans HSN motor neurons. Nature 336, 638-646. Erkman, L., McEvilly, J., Luo, L., et al. (1996) Role of transcription factors Brn-3.1 and Brn-3.2 in auditory and visual system development. Nature 381, 603-606. (Abstract) Fedtsova, N.G. and Turner, E.E. (1996) Brn-3.0 expression identies early pot-mitotic CNS neurons and sensory neural precursors. Mechanisms of Development 53, 291-304. Gerrero, M.R., McEvilly, R.J., Turner, E., et al. (1993) Brn-3.0 A POU domain protein expressed in the sensory, immune and endocrine systems that functions on elements distinct from known octamer motifs. Proc. Natl.Acad. Sci. USA 90, 10841-10845. He, X., Treacy, M.N., Simmons, D.M., Ingraham, H.A., Swanson, L.S. and Rosenfeld, M.G. (1989) Expression of a large family of POU-domain regulatory genes in mammalian brain development. Nature 340:35-42. Howard, M.K., Burke, L.C., Mailhos, C., et al. (1993) Cell cycle arrest of proliferating neuronal cells by serum deprivation can result in either apoptosis or differentiation. J. Neurochemistry 60, 1783-1791. Howard, M.K., Kershaw, T., Gibb, B., et al. (1998) High efciency gene transfer to the central nervous system of rodents and primates using herpes virus vectors lacking functional ICP27 and ICP34.5. Gene Therapy in press. Lai, J-S., Cleary, M.A. and Herr, W. (1992) A single amino acid exhange transfers VP-16 induced positive control from the Oct-1 to the Oct-2 POU homeodomain. Genes Dev 6, 2058-2065. Lakin, N.D., Morris, P.J., Theil, T., et al. (1995) Regulation of neurite outgrowth and SNAP-25 gene expression by the Brn-3a transcription factor. J Biol Chem 270, 15858-15863. Latchman, D.S. (1990) Molecular biology of herpes simplex virus latency. Int J Exp Pathol 71, 133-141. Lillycrop, K.A., Budhram-Mahadeo, V.S., Lakin, N.D., et al. (1992) A novel POU family transcription factor is closely related to Brn-3 but has a distinct expression pattern in neuronal cells. Nucleic Acids Res. 20, 5093-5096. Liu, Y-Z., Dawson, S.J. and Latchman, D.S. (1996) Alternative splicing of the Brn-3a and Brn-3b transcription factor mRNAs is regulated in neuronal cells. J Molec Neurosci 7, 77-85. McEvilly, R.J., Erkman, L., Luo, L., Sawchenko, P.E., Ryan, A.F. and Rosenfeld, M.G. (1996) Requirement for Brn-3.0 in differentiation and survival of sensory and motor neurons. Nature 384, 574-577. Morris, P.J., Dawson, S.J., Wilson, M.C. and Latchman, D.S. (1997a) A single residue within the homeodomain of the Brn-3 POU family transcription factors determines whether they activate or repress the SNAP-25 promoter. NeuroReport 8, 2041-2045. Morris, P.J., Lakin, N.D., Dawson, S.J., et al. (1997b) Differential regulation of genes encoding synaptic proteins by members of the Brn-3 sub family of POU transcription factors. Brain Res 43, 279-285. Morris, P.J., Theil, T., Ring, C.J.A., Lillycrop, K.A., Möröy, T. and Latchman, D.S. (1994) The opposite and antagonistic effects of the closely related POU family transcription factors on the activity of a target promoter are dependent upon differences in the POU domain. Mol Cell Biol 14, 6907-6914. Ndisdang, D., Morris, P.J., Chapman, C., Ho, L., Singer, A. and Latchman, D.S. (1998) The HPV activating cellular transcription factor Brn-3a is over-expressed in women with CIN3 cervical lesions. J Clin Invest 101, 1687-1692. Ninkina, N.N., Stevens, G.E.M., Wood, J.N. and Richardson, W.D. (1993) A novel Brn-3 like POU transcription factor expressed in subsets of rat sensory and spinal cord neurons. Nucleic Acids Res. 21, 3175-3182. Oppenheim, R.W. (1991) Cell death during development of the nervous system. Annu Rev Neurosci 14, 453-501. Osen-Sand, A., Catsicas, M., Staple, J.K., et al. (1993) Inhibition of axonal growth by SNAP-25 antisense

oligonucleotides in vitro and in vivo. Nature 364, 445-448. Ryan, A.K. and Rosenfeld, M.G. (1997) POU domain family values:- exibility, partnerships and developmental codes. Genes Dev 11, 1207-1225. Smith, M.D., Dawson, S.J. and Latchman, D.S. (1997a) The Brn-3a transcription factor induces neuronal process outgrowth and the co-ordinate expression of genes encoding synaptic proteins. Mol Cell Biol 17, 345-354. Smith, M.D., Dawson, S.J. and Latchman, D.S. (1997b) Inhibition of neuronal process outgrowth and neuronal specic gene activation by the Brn-3b transcription factor. J Biol Chem 272, 1382-1388. Smith, M.D., Morris, P.J., Dawson, S.J., Schwartz, M.L., Schlaepfer, W.W. and Latchman, D.S. (1997c) Co-ordinate induction of the three neurolament genes by the Brn-3a transcription factor. J Biol Chem. 272, 21325-21333. Smith, M.D., Ensor, E.A., Cofn, R.S., Boxer, L.M. and Latchman, D.S. (1998) Bcl-2 transcription from the proximal p2 promoter is activated in neuronal cells by the Brn-3a POU family transcription factor. J Biol Chem in press. Suburo, A.M., Wheatley, S.C., Horn, D.A., et al. (1992) Intracellular redistribution of neuronpeptides and secretory proteins during differentiation of neuronal cell lines. Neuroscience 46, 881-889. Theil, T., McLean-Hunter, S., Zornig, M. and Möröy, T. (1993) Mouse Brn-3 family of POU transcription factors: a new amino terminal domain is crucial for the oncogenic activity of Brn-3A. Nucleic Acids Res. 21, 5921-5929. Turner, E.E., Jenne, K.J. and Rosenfeld, M.G. (1994) Brn-3.2: A Brn-3-related transcription factor with distinctive central nervous system expression and regulation by retinoic acid. Neuron 12, 205-218. Vahaua, O., Morell, R., Lynch, E.D. et al. (1998) Mutation in transcription factor POU4F3 associated with progressive hearing loss in humans. Science 279, 1950-1954. Verrijzer, C.P. and van der Vliet, P.C. (1993) POU domain transcription factors. Biochim Biophys Acta 1173, 1-21. White, E. (1996) Life, death and the pursuit or apoptosis. Genes Dev 10, 1-5. Wood, J.N., Bevan, S.J., Coote, P., et al. (1990) Novel cell lines display the properties of nociceptive sensory neurons. Proc Royal Soc London Series B 241, 187-194.

Gene Ther Mol Biol Vol 2, 103-117. August 7, 1998.

EGR-1 prevents growth arrest by induction of c-myc Scott A. Crist1,2, Sumathi Krishnan1, Seong-Su Han1, Aysegul Nalca1,2, and Vivek M. Rangnekar1,2,3,4,* 1

Department of Surgery, Division of Urology ; Department of Microbiology and Immunology ; Graduate

Center for Toxicology ; and Markey Cancer Center ; University of Kentucky, Lexington, KY 40536. ______________________________________________________________________________________ ________________ * Correspondence: Vivek M. Rangnekar Combs Research Building, Room 303, University of Kentucky, 800 Rose Street, Lexington, KY 40536; Tel. (606) 257-2677; Fax. (606) 257-9608; E-mail: vmrang01@pop.uky.edu

Summary The zinc-finger transcription factor EGR-1 provides protection from G1 phase growth arrest. We present here evidence that this protective effect of EGR-1 is linked to upregulation of c-myc RNA and protein by induction of the c-myc promoter. Growth arrest involves c-myc downregulation and hypophosphorylation of the retinoblastoma susceptibility protein RB, but upregulation of c-myc prevents hypophosphorylation of RB and provides protection from growth arrest. These findings suggest a downstream mechanism for EGR-1 function as an inhibitor of G1 phase growth arrest. Because Egr-1 and c-myc are involved in determining cell fate in response to diverse exogenous signals, the findings of the present study can be extended to model systems for proliferation, cellular differentiation, and programmed cell death.

I. Introduction Regulation of cell cycle progression by exogenous stimuli is associated with a series of complex molecular cascades that are initiated at the plasma membrane and are dependent on inducer- and cell type-specific gene programs. Gene programs in the G1 phase of the cell cycle control entry into, or exit from, the cell cycle and thereby dictate the cell's ultimate responses to exogenous stimuli. Thus, these gene programs couple the biochemical signaling events occurring at the plasma membrane to long-term alterations in cellular phenotypes such as proliferation, growth arrest, differentiation, and programmed cell death. G1 phase genes that encode transcription factors are key regulators of the downstream cascades that determine entry past the G1 phase cell cycle check point into the S-phase. Understanding the role of such transcription factors in the positive and negative regulation of growth will provide important insights into how extracellular stimuli signal specific long-term cellular responses. The transcription factor early growth response-1 (EGR-1; also cited as NGF-IA, TIS8, Zif268, and Krox24), encoded by the Egr-1 gene, is upregulated in response to diverse cellular stimuli including mitogens, membrane depolarization, seizure, synaptic activity, iscemia, nerve injury, and B-cell maturation and differentiation of nerve, bone and myeloid cells (Milbrandt, 1988; Sukhatme, 1988). EGR-1 is a nuclear protein that contains three zinc-fingers of the C2H2 subtype (Cao et al., 1990; Christy and Nathans, 1989; Gashler et al., 1993; Swirnoff and Milbrandt, 1995). Deletion analysis of the Egr-1 cDNA has shown that the amino acids constituting the zinc-finger domain and the flanking regions confer DNAbinding and nuclear localization functions (Gashler et al., 1993; Russo et al.,1993). The amino-terminus contains a serine/threonine-rich region that confers a transcription activation function (Gashler et al., 1993; Russo et al.,1993). EGR-1 is the prototypic member of the Egr family of transcription factors that includes EGR-2/Krox-20, EGR-3, NGFI-C, and the gene product of the Wilms' tumor suppressor gene, WT1 (Call et al., 1990; Crosby et al., 1991; Lau and Nathans, 1987; Lemaire et al., 1988; Lim et al., 1987; Sukhatme et al. 1988; Sukhatme, 1990). The transcription factors of the Egr family have a high degree of homology in the amino acid sequence constituting the zinc-finger region, and they bind to the same GCrich consensus DNA sequence (Christy and Nathans, 1989; Rauscher et al., 1990; Sukhatme, 1990). The direct interaction between the GC-rich consensus DNA elements and the zinc-finger domain of EGR-1 has been confirmed by X-ray crystallography studies (Pavletich and Pabo, 1990). Transient transfection studies using EGR-1 or WT1 expression vectors and reporter genes that contain GC-rich consensus element have shown that these proteins can function either as strong activators or as repressors of transcription, depending upon the cellular context (Drummond et al., 1992; Madden et al., 1991; Maheswaran et al., 1993; Wang et al., 1992). Consistent with the fact that EGR-1 is induced in response to a wide spectrum of mitogenic stimuli, GCrich EGR-1-binding sites have been identified in the promoters of genes such as thymidine kinase, an enzyme integral to DNA biosynthesis (Molnar et al., 1994); in cell cycle regulators such as cyclin D1 (Phillipp et al., 1994) and in the gene encoding the retinoblastoma susceptibility protein RB involved in cellular proliferation (Day et al., 1993). Recent studies, however, suggest that DNA sequences that diverge from the consensus sequence can also bind EGR-1 with high affinity (Molnar et al., 1994; Swirnoff and Milbrandt, 1995). DNA-protein binding studies in vitro have shown that S1-nuclease sensitive regions, which are homopurine/ homopyrimidine-rich [(TCC)n] in nucleotide content, can bind EGR-1 just as avidly as can the GC-rich consensus element (Wang and Deuel, 1992; Wang and Deuel, 1996). Such homopurine/homopyrimidine-rich motifs have been identifed in the promoter regions of several genes encoding growth factors such as platelet-derived growth factor, transforming growth factor (TGF) and basic fibroblast growth factor; growth factor receptors epidermal growth factor-receptor, and the insulin-like growth factor-receptor; and protooncogenes c-Ki-ras and c-myc (Gashler and Sukhatme, 1995; Hu and Levin, 1994). However, the downstream phenotypic consequences of EGR-1-binding to the homopurine/ homopyrimidine-rich motifs in these gene promoters are not known. c-MYC, the product of the c-myc gene is another G1 phase transcription factor that is induced in response to mitogenic stimuli and essential for cell-cycle progression (reviewed in Luscher and Eisenman, 1990). The c-MYC protein contains a basic region-helix-loop-helix domain and a leucine zipper domain at its carboxy-terminus for the DNA-binding function (Blackwell et al., 1993). The target DNA-binding motif for MYC is a 6-bp consensus sequence 5'-CTCGAG-3' referred to as the E-box (Blackwell et al., 1993). The transcription activation function of c-MYC is conferred by the presence at its amino-terminus of a proline/glutamine-rich region and of flanking amino acid residues (Blackwell et al., 1993). Several studies have implicated c-MYC in the regulation of proliferation, programmed cell growth arrest, differentiation, or apoptosis (Alexandrow et al., 1995; Cole, 1986; Evan and Littlewood, 1993; Hanson et al., 1994; Hermeking and Eick, 1994; Hoffman-Liebermann and Liebermann, 1991; Janicke et al., 1994; Janicke et al., 1996; Luscher and Eisenman, 1990; Wagner et al., 1994). Inhibiting c-myc expression by using

antisense oligonucleotides causes growth inhibition of proliferating cells (Evan and Littlewood, 1993). Consistent with this observation, ectopic overexpression of c-myc in quiescent cell cultures is sufficent to induce the cells to re-enter the cell cycle, even in the absence of serum growth factors (Evan and Littlewood, 1993). Moreover, c-myc overexpression can protect cells from the growth arresting action of TGF (Alexandrow et al., 1995). In cells that can be induced to differentiate, enforced expression of c-myc prevents the cells from exiting the cell cycle and thereby inhibits the differentiation pathway (HoffmanLiebermann and Liebermann, 1991). In other studies, cells that are deprived of serum growth factors have been shown to undergo apoptosis when c-myc is ectopically overexpressed (Evan and Littlewood, 1993; Hartwell and Kastan, 1994). Collectively, these studies have firmly established c-myc as a regulator of diverse long-term cellular responses. Although c-MYC plays a central role in the regulation of proliferation, differentiation, and apoptosis, the precise mechanism by which c-MYC regulates these long-term growth responses is not known. Several examples of differential regulation of growth-associated genes by c-MYC have been presented. Enforced expression of c-myc constitutively activates the expression of ornithine decarboxylase, an enzyme integral to polyamine biosynthesis and a mediator of apoptosis after IL-3 withdrawal in murine myeloid cells (Packham and Cleveland, 1994). Constitutive expression of c-myc can transactivate the expression of the cell cycle regulators cyclin D1, D3, and E (Janicke et al., 1996; Janson-Durr et al., 1993) or cyclin A (Janson-Durr et al., 1993). Other studies have shown that in different cell types, constitutive expression of c-myc can transcriptionally regulate the expression of cyclin-dependent kinase 4 (cdk4), a key regulator of cell cycle progression in the G1 phase (Hanson et al., 1994). It is also suggested that c-myc can function as a "molecular matchmaker", a factor that can alter the affinities of two or more interacting factors that mediate cell cycle progression, in a manner independent of transcriptional activation (Hanson et al., 1994). Examples of this activity include the ability of c-MYC to interfere with the interaction between the retinoblastoma susceptibility gene product RB and E2F, and the ability to promote increased complex formation between cyclin A, cyclin dependent kinase 2 and the transcription factor E2F (Marcu et al., 1992; Hanson et al., 1994; Janson-Durr et al., 1993). The formation of these complexes regulates the G1 to S phase transition of the cell cycle (Hartwell and Kastan, 1994; Li et al., 1993; Pietenpol et al., 1990; Qin et al., 1995; Weinberg, 1995). Finally, recent evidence suggests that c-MYC can modulate the activity of cdk4 by transcriptional regulation of cdc25 which encodes a phosphatase that directly controls the activity of cdk4 (Galaktionov et al., 1996). All of these examples represent viable, biologically-relevant mechanisms for c-MYC function; however, the mechanisms are cell type-specific, suggesting multiple mechanisms for cell growth control by c-MYC. Our studies have focused on the role of immediate-early genes in programmed cell growth arrest (Rangnekar et al., 1991; Rangnekar et al., 1992). These studies have used, as an experimental model, human melanoma cells A375-C6 (Endo et al., 1988) that are susceptible to cytokines such as interleukin-1 (IL-1) and tumor necrosis factor- (TNF-) (Rangnekar et al., 1991; Rangnekar et al., 1992). We have demonstrated that in A375-C6 cells, IL-1 induces a G1 phase growth arrest response that is dependent on hypophosphorylation of the retinoblastoma susceptibility protein RB (Muthukkumar et al., 1996). IL-1 causes induction of EGR-1 in A375-C6 cells (Sells et al., 1995), and our initial studies showed that this protein functions to prevent RB hypophosphorylation (Krishnan and Rangnekar, unpublished data) and to protect the cells from growth arrest (Sells et al., 1995). We then sought to determine the mechanism by which EGR-1 may confer this protective effect. We present here evidence that EGR-1 up-regulates c-myc expression by inducing its promoter and thereby protects A375-C6 cells from RB-hypophosphorylation and IL-1-inducible growth arrest.

II. Materials and Methods A. Cell culture and cytokine Growth and maintenance of human melanoma cells A375-C6 have been described previously (Rangnekar

et al., 1991; Rangnekar et al., 1992). Human recombinant IL-1 (specific activity, 1.8 x10 units/mg) was a gift from Craig Reynolds, Biological Response Modifiers Program, National Cancer Institute (Fredrick, MD). IL-1 was used at a concentration of 100 unit/ml as previously described (Rangnekar et al., 1991; Rangnekar et al., 1992).

B. Plasmid constructs Previous studies have described the cytomegalovirus (CMV) promoter-based expression constructs for

mouse-EGR-1 (CMV-mEGR1); the EGR-1-WT1 chimera (CMV-WT1/EGR1) that contains the transactivation domain of WT1 and the three zinc finger domains of EGR-1; and deletion mutants of EGR-1 (from Vikas P. Sukhatme, Harvard Medical School, Boston, MA) (Drummond et al., 1992; Gashler et al., 1993; Madden et al., 1991; Nair et al., 1997). The pAc.myc construct (Hoffman-Liebermann and Liebermann, 1991), which contains a 1.5 kb Sst1-HindIII fragment with 60 bp of untranslated region and all of the coding region (exons 2 and 3) of murine c-myc cDNA cloned into the HindIII site of the actin-promoter construct, pHb APr-1-neo, was kindly provided by Barbara Hoffman (Temple University School of Medicine, Philadelphia, PA). The reporter construct myc(-1141+513)CAT that contains 1141 bp upstream and 513 bp downstream of the P1 promoter of murine c-myc linked to chloramphenicol acetyl transferase (CAT) cDNA was provided by John Cleveland, (Saint Jude Children's Hospital, Memphis, TN); and [myc(E-Box)CAT] that contains the c-MYC response element upstream of CAT cDNA, was provided by Robert Eisenman (Fred Hutchenson Cancer Center, Seattle, WA). The CAT reporter constructs myc(-665+935)CAT and myc(-101+935)CAT, which contain 665 bp and 101 bp upstream, respectively, and 935 bp downstream of the P1 promoter of human c-myc (Wang and Deuel, 1992), were kindly provided by Zhao-Yi Wang (Harvard Medical School, Boston, MA). To generate the reporter constructs p90-CAT, p513-CAT, or p579-CAT, polymerase chain reaction (PCR) was used to amplify various fragments containing the different regions of the human c-myc promoter region. The p579 and p90 fragments used a common downstream primer 5'(CTCGAT)TGCTTTGGGA-3' (which contains 10 nucleotides (underscored) complimentary to nucleotides -96 to -86 upstream of c-myc P1 promoter, and a built-in XhoI site shown in parentheses). For construct p579, a 579 bp fragment containing nucleotides -665 and -86 from the upstream region of human c-myc was synthesized using primer 5'-ATACATGACTCCCCCAACA-3' and the common downstream primer. For construct p90, a 90 bp fragment containing nucleotides -176 and -86 from the upstream region of human c-myc was synthesized by using primer 5'-(GTCGAC)AGAGTGCTCGGC-3' (Sal I site in parentheses) and the common downstream primer. For construct p513, a 513 bp fragment containing nucleotides -685 and -1528 from the upstream region of c-myc (i.e., representing a deletion of the -145 to -115 homopurine/homopyrimidine region) was generated by using the same upstream primer as that used for the 579 bp fragment, and a downstream primer, 5'-ACTCAGCCCGGGCAGCCGAGCACT-3'. Each of the three fragments were subcloned into the SmaI site of pBluescript II SK(-) (Stratagene, La Jolla, CA) and fidelity of the fragments verified by sequencing. Further amplification of the three fragments was accomplished by PCR by using the proof-reading DNA polymerase Pfu (Stratagene), and the primers T3 (5'-AATTAACCCTCACTAAAGGG-3') and T7 (5'-GTAATACGACTCACTATAGGGC-3'). The amplified, blunt-ended fragments were then subcloned into the BglII site of pCAT (Promega Corp., Inc., Madison, WI) after filling-in the 3' overhangs of the BglII site with Klenow fragment. The pCAT vector contains a minimal SV-40 early promoter driving the CAT cDNA. The orientation of the insert and the fidelity of the final constructs were confirmed by sequencing.

C. Northern analysis Total RNA preparation and Northern blot analysis was performed as described (Muthukkumar et al., 1995; Sells et al., 1995). The EcoR1/HindIII fragment of pMV-7/mycER which contains exons 2 and 3 of human c-myc was used as a probe for c-myc. The cDNA probe for human gro- has been described (Rangnekar et al., 1991). To verify equal loading of RNA on the gel, the blots were probed with cDNA for glyceraldahyde-3phosphate dehydrogenase (GAPDH). The hybridization signals were scanned with a densitometer to measure the amount of mRNA, and the amount of fold induction or reduction was calculated after normalizing the hybridization signal with respect to GAPDH.

D. Assay for 3[H]thymidine incorporation These studies were performed in 96-well plates and percent growth inhibition was calculated as described previously (Rangnekar et al., 1991; Rangnekar et al., 1992).

E. Effect of IL-1 on spheroid growth Cells were plated in plates coated with semi-solid growth medium (RPMI 1640 growth medium containing 4% gelatin) with a top layer of the liquid growth medium. When cells began to clump and form spheroids, individual spheroids of similar size and shape were transfered to a 24-well plate. The spheroids were then either exposed or left unexposed to IL-1 . The dimensions of each spheroid were determined

daily by microscopy with a scaled grid and the volume (width2 X length) was calculated.

F. DNA transfections and stable transfectant cell line Transfections were performed by using calcium phosphate coprecipitation method described previously (Sells et al., 1995). Stable transfectant clones were selected by culture in 300 !g/ml G418 sulfate (BRL/ Life Technologies, Inc.). Pools of G418-resistant clones were maintained as cell lines.

G. CAT assays Transient transfections and CAT assays were performed as previously described (Sells et al., 1995). The values for percent conversion of [14C]chloramphenicol to acetylated forms in different samples from a given experiment were normalized with respect to the corresponding protein concentration and were expressed as relative CAT activity.

H. Western (immunoblot) analysis Whole-cell protein extracts were subjected to Western blot analysis with the indicated antibody (1 mg/ml) and 125I-protein A, as described previously (Ahmed et al., 1996; Muthukkumar et al., 1995). The retinoblastoma susceptibility gene product RB was detected by using the anti-RB antibody C-15 (from Santa Cruz Biotechnology, Santa Cruz, CA) as previously described (Muthukkumar et al., 1996). The anti--actin monoclonal antibody was purchased from Sigma Chemical Company (St. Louis, MO). Blots that were first probed with the monoclonal antibodies were subsequently probed with a rabbit anti-mouse antibody (from Southern Biotechnology, Birmingham, AL) before incubation with 125I-protein A. The

rabbit polyclonal antibody for MYC, 50-39 was a generous gift from Steve Hann (Vanderbilt University, Nashville, TN). Western blot analysis for the detection of MYC was carried out as described above, except that after incubation with the MYC antibody, the Enhanced ChemoLuminescence (ECL; Amersham, Arlington Heights, IL) system was used for detection.

III. Results A. Ectopic expression of EGR-1 protects cells from growth arrest We have previously reported that EGR-1 mRNA and protein are induced in A375-C6 cells after 3 to 4 hours of exposure to IL-1, and that inhibition of EGR-1 expression or function enhances the growthinhibitory effect of IL-1 (Sells et al., 1995). We sought to examine whether ectopic overexpression of EGR-1 can protect A375-C6 cells from IL-1-inducible growth arrest. These experiments used constructs that encoded either full-length EGR-1 or functional variants of EGR-1 such as those that lacked the transactivation region (TA) or DNA-binding region (ZF). A375-C6 cells were stably transfected with the different constructs and pools of stably transfected clones were maintained as cell lines. The transfected cell lines were exposed to IL-1 and growth inhibition was determined by [3H]thymidine incorporation assays.

As shown in Fig. 1 (A & B), ectopic expression of full-length EGR-1 from CMV-mEGR1 caused a significant decrease (P<0.0001 by the Students t test) in growth inhibition relative to cells containing the empty vector after exposure to IL-1. By contrast, cells expressing either the EGR-1-mutant ZF that was deficient in DNA-binding or TA that was deficient in transcriptional activity were similar to those containing the empty-vector in terms of their susceptibility to IL-1-inducible growth arrest (Fig. 1B). These findings suggest that EGR-1 confers protection from a growth arrest signal and that the protective effect is dependent on the ability of EGR-1 to transcriptionally regulate downstream genes.

B. EGR-1 induces the c-myc promoter In the course of our studies aimed at identifying downstream targets of EGR-1, we tested c-myc because its promoter (P1) contains an EGR-1-binding element with a homopurine/homopyrimidine-rich sequence (Wang and

Figure 1. Effect of ectopic overexpression of EGR-1 or deletion mutants on IL-1-inducible growth arrest. (A) Pools (L1 or L2) of stably transfected clones expressing empty vector or EGR-1 were exposed to IL-1 or vehicle for 24, 48, or 72 hours. (B) Pools (L1) of stably transfected clones expressing empty vector, EGR-1, EGR-1transactivation deletion mutant TA, or EGR-1-zinc finger deletion mutant ZF were exposed to IL-1 or vehicle for 72 hours. The cells were subjected to [3 H]thymidine incorporation studies and percent growth inhibition was calculated. Each value point is a mean of 12 observations from 3 different experiments. Error bars indicate Âąstandard deviations.

Deuel, 1992) [(TCC)n located -142 to -115 bp; depicted in Fig. 2]. This element has been previously shown to bind, in electrophoretic mobility shift assays, to purified EGR-1 (Wang and Deuel, 1992). However, the ability of this element to induce EGR-1-dependent transactivation of the c-myc promoter has not been tested. Secondly, significant overlap in EGR-1 and c-myc functions has been identified: (i) EGR-1 and c-MYC are both associated with induction of proliferation in most cell types (Evan and Littlewood, 1993; Gashler and Sukhatme, 1995); (ii) forced expression of Egr-1 or c-myc can block or restrict differentiation along a specific cellular lineage (Hoffman-Liebermann and Liebermann, 1991; Nguyen et al., 1993); and (iii) Egr-1 or c-myc expression sensitizes cells to apoptotic stimuli (Ahmed et al., 1996; Hermeking and Eick, 1994; Janicke et al., 1994; Sells et al., 1995; Wagner et al., 1994). To determine the ability of EGR-1 to transcriptionally activate the promoter of cmyc, we used CAT-reporter constructs containing various deletions of the P1 promoter of c-myc. Cells were cotransfected with CMV-mEGR1, or as controls with an empty vector or CMV-WT1/EGR1, and the CAT-reporter constructs. The CMV-WT1/ EGR1 construct was expected to transrepress the promoter constructs that were transactivated by CMV-mEGR1. When a constant amount of the CAT-reporter construct [myc(-1141+513)CAT] which contains 1141 bp upstream or 513 bp downstream of the c-myc promoter (Fig. 2A) was cotransfected with increasing amounts of CMV-mEGR1, a 5 to 6 fold increase in CAT activity occurred (Fig 2B). On the other hand, cotransfection with CMV-WT1/EGR1 caused a decrease in CAT activity from the [myc(-1141+513)CAT] construct (Fig. 2B). These findings suggest that ectopic expression of EGR-1 leads to transcriptional upregulation of the c-myc promoter.

Figure 2. Transactivation of the c-myc promoter deletion constructs by EGR-1. (A) Schematic of myc(-1141+513)CAT, myc(-665+935)CAT, myc(-101+935)CAT reporter constructs used for transfections. The cap site and the (TCC)n region are indicated. A375-C6 cells were cotransfected (B) reporter plasmid myc(-1141+513)CAT (4 !g) and with various amounts of CMV-mEGR1 or CMV-WT1/EGR1; with (C) myc(-1141+513)CAT (4 !g) and 5 !g of vector, CMV-mEGR1, CMV-mEGR1. TA, or CMV-mEGR1. ZF; or with (D) myc(-665+935)CAT or myc(-101+935)CAT reporter constructs and various amounts of CMV-mEGR1. The total amount of plasmid DNA used in each transfection was brought up to 34 !g by using vector DNA. Cell extracts were assayed for CAT activity and relative CAT activity with each effector construct is expressed relative to that with vector 30 !g of empty vector.

We then examined whether induction of the c-myc promoter required the DNA-binding and transactivation functions of EGR-1. Cells were transiently cotransfected with vector, CMV-mEGR1, CMV-mEGR1.ZF, or CMV-mEGR1.TA and the reporter plasmid myc(-1141+513)CAT, and cell lysates were examined for CAT activity. As seen in Fig. 2C, CMV-mEGR1 induced the c-myc promoter. On the other hand, the EGR-1 mutants CMV-mEGR1.ZF or CMV-mEGR1.TA did not induce the promoter (Fig. 2C). These findings suggest that the DNA-binding and transactivation functions of EGR-1 are both required for induction of the c-myc promoter. To further localiz e the EGR-1 respon sive site(s) within the cmyc promo ter, transie nt transfe ctions were perfor med by using the reporte r constr uct myc(-6 65+93 5)CAT , which is a truncat ed variant of myc(-1 141+5 13)CA T and contai ns 665 bp of sequen ce upstrea m of P1 promo

ter (Fig. 2A). When cotrans fected with CMVmEGR 1, myc(-6 65+93 5)CAT showe da2 to 3 fold increas e in CAT activit y (Fig. 2D). Cotran sfectio n of CMVWT1/ EGR1 caused a dosedepend ent decreas e in CAT activit y from this same constr uct (data not shown ). These results sugges t that the EGR-1 respon sive elemen t is contai

ned within 665 bp upstrea m and 995 bp downs tream of the c-myc promo ter. When reporter construct myc(-101+935)CAT containing 101 bp upstream and 935 bp downstream of the promoter [i.e., lacking the -142 to -115 homopurine/ homopyrimidine-rich region and upstream sequence] was cotransfected with CMV-mEGR1 (Fig. 2D) or CMV-WT1/EGR1 (data not shown) neither an increase nor a decrease in CAT activity was detected. The lack of response of this reporter construct to EGR-1 or the WT1/EGR-1 chimera suggests that the putative EGR-1 response element is located in the region between -665 and -101 bp of the c-myc promoter.

C. EGR-1 regulates c-myc expression via the (TCC)n EGR-1binding motif in the c-myc promoter We next tested the (TCC)n motif located between 142 to 115 upstream of P1 promoter for inducibility with EGR-1. Cells were cotransfected with CMV-mEGR1 and the following reporter constructs (Fig. 3A): p90-CAT which contains the (TCC)n region; construct p513-CAT in which the (TCC)n region is absent; or p579-CAT which contains the region -665 to -86 and was expected to be an EGR-1-responsive control (based on results from Fig. 2D), and CAT assays were then performed. CMV-mEGR1 caused a 3 to 3.5 fold induction in CAT activity from p90-CAT or p579-CAT but did not alter the basal CAT activity from p513-CAT, (Fig. 3B). These findings suggest that EGR-1 causes induction of the c-myc promoter via the (TCC)n motif.

Figure 3. Transactivation of c-myc promoter via the (TCC)n sequence by EGR-1. (A) Schematics of pSV40CAT, p579-CAT, p90-CAT and p513-CAT. These constructs used the SV-40 promoter and various fragments corresponding to the c-myc promoter region. The (TCC)n region and the cap site are indicated. Note that p513CAT lacks the (TCC)n motif. (B) The reporter plasmids (10 !g) were cotransfected with 0 or 10 !g of vector or CMV-mEGR1 and cell extracts were assayed for CAT activity. Relative CAT activity for each reporter construct is a ratio of the CAT activity with 10 !g of CMV-mEGR1 and 10 !g of empty vector.

Figure 4. EGR-1 induces c-myc expression at the mRNA and protein level. Total RNA isolated from C6/ EGR1.L1 or C6/vector.L1 cells was subjected to Northern blot analysis for c-myc or GAPDH. The increase in induction of c-myc RNA normalized to GAPDH expression levels is indicated. (B) Whole-cell extracts from C6/ vector.L1, C6/EGR1.L1, C6/myc.L1 or C6/myc.L3 cells were prepared and subjected to Western blot analysis by u antibody, 50-39. For a loading control, the same blot was probed with anti- actin antibody. The arrow indicates the position of MYC protein (about 68 kd).

D. Overexpression of EGR-1 causes increased expression of c-myc RNA and protein Our results suggested that the c-myc promoter represented one of the targets of EGR-1. We then ascertained that this effect was not restricted to the promoter and that it resulted in an increase in c-myc RNA and protein. Total RNA was isolated from cell lines that were stably transfected with CMV-mEGR1 or vector and was subjected to Northern blot analysis for c-myc. Data representative of four different pools of transfected clones maintained as cell lines are shown in Fig. 4. As compared to C6/vector cells, C6/ EGR1 cells showed a 5 fold increase in c-myc RNA (Fig. 4A). To determine whether the protein levels of c-MYC were also higher in cells overexpressing EGR-1, we prepared whole-cell protein extracts from the transfectants and subjected them to Western blot analysis for c-MYC protein. As seen in Fig. 4B, C6/ EGR1.L1 cells express a 4 fold higher level of c-MYC protein than do C6/vector.L1 cells. These results suggest that EGR-1 can upregulate endogenous c-myc, leading to increased levels of c-MYC protein.

E. IL-1 mediated growth arrest is dependent on downregulation of c-myc RNA in A375-C6 cells To study the biological relevance of EGR-1-inducible expression of c-myc, we used A375-C6 cells that undergo a time- and dose-dependent growth arrest in the G1 phase of the cell cycle when exposed to IL-1. Because the levels of c-myc expression correlate with the growth status of various cell lines, we began these studies by examining the effect of IL-1 on c-myc expression in A375-C6 cells. Total RNA was isolated from subconfluent monolayers of A375-C6 cells before or after exposure to IL-1 and was examined for c-myc expression by Northern analysis. As seen in Fig. 5A, within 1 hour of exposure to IL-1, c-myc levels decreased by 8 fold relative to the basal levels; thereafter, sustained low levels of cmyc were maintained upto 48 h of exposure to IL-1. When the same Northern blot was probed with grocDNA, a rapid and sustained induction of the gro- gene was seen indicating responsiveness to IL-1 as expected from our previous studies (Joshi-Barve et al., 1993; Rangnekar et al., 1991; Rangnekar et al., 1992). We also ascertained that the c-myc gene in A375-C6 cells is positively or negatively inducible by growthstimulatory or growth-inhibitory signals, respectively. The cells were grown in serum-containing medium (unstarved cells) or serum-free medium (serum-starved cells) for 48 h and were then exposed to 10% serum for various intervals of time. Total RNA was prepared and subjected to Northern blot analysis for c-myc. As seen in Fig. 5B, serum-starvation caused a rapid decrease in the steady state levels of c-myc, and serumstimulation caused a strong induction of c-myc within 2 to 3 hours. These findings suggest that the expression levels of c-myc in these cells are positively regulated by growth-stimulatory signals and are negatively regulated by growth arresting signals.

Figure 5. IL-1-inducible growth arrest is associated with downregulation of c-myc mRNA. (A) A375-C6 cells were treated with vehicle (untreated with IL-1 [UT]) or were treated with IL-1 for various intervals of time, as indicated. Total RNA was then prepared and subjected to Northern analysis for c-myc. The same blot was sequentially probed for gro-, and finally for GAPDH expression. (B) A375-C6 cells were grown in medium containing 10% serum (unstarved [US]), or were serum-starved for 48 hours (0 h of serum-stimulation) and were then exposured to 10% serum. Total RNA was isolated at various time points after serum restimulation and was subjected to Northern blot analysis for c-myc and GAPDH.

F. Ectopic overexpression of c-myc protects A375-C6 cells from IL-1inducible growth inhibition. To determine whether downregulation of c-myc is required for the growth-inhibitory action of IL-1 in A375-C6 cells, we studied whether ectopic overexpression of c-myc driven by the -actin promoter (which was expected to be unresponsive to IL-1) could alter the growth-inhibitory response. A375-C6 cells were stably transfected with pAc.myc, an eukaryotic expression vector in which exons 2 and 3 of c-myc are under the transcriptional control of the -actin promoter. Pools of stably transfected clones were maintained as cell lines and were subjected to Northern blot analysis to verify expression of the 1.8 kb RNA from the transgene and to Western blot analysis to verify increased c-MYC protein levels. As seen in Fig 6A, several G418-resistant clones (L1 and L3, but not L2) expressed high levels of the 1.8 kb transgenic c-myc RNA species and of the 2.4 kb RNA species representing endogenous c-myc. Western blot analysis indicated that myc.L1 and myc.L3 cells that expressed the transgenic c-myc RNA showed an higher amount of c-MYC protein than did cells transfected with vector alone (Fig. 4B). A [3H]thymidine incorporation assay performed to study the growth rate of the different cell lines indicated that the doubling times for all of the C6/myc and C6/vector transfected cell lines examined over a 72 h period were similar (i.e., about 24 h; data not shown). Finally, to ascertain whether the transgenic c-MYC protein was functional, we used transient transfection of the myc.L1 or vector.L1 cultures with a CAT construct containing the MYCresponsive E-box sequence. As seen in Fig. 6B, myc.L1 cells showed a 3 to 4 fold higher CAT activity than did the vector.L1 cells. Together, the findings of these studies indicated that transgenic c-myc is both expressed and functional in A375-C6 transfectants.

Figure 6. Expression of c-myc in stably transfected clones. (A) Total RNA was prepared from A375-C6 cells that were stably transfected with vector or c-myc expression plasmid and Northern blot analysis was performed by using c-myc cDNA as a probe. Note that C6/myc.L1 and C6/myc.L3 expresses both the endogenous c-myc 2.4 kb

RNA and the transgenic c-myc 1.8 kb RNA; whereas C6/vector.L1 and C6/myc.L2 express only endogenous c-myc RNA. (B) Transfected cell lines C6/vector.L1 or C6/myc.L1 were transiently transfected with MYC(E-box)-CAT reporter plasmid which contained the MYC-response element upstream of the CAT cDNA or with min-CAT plasmid, which contained a minimal promoter but lacked the E-box. The cell extracts were then assayed for CAT activity.

To determine whether ectopic expression of c-myc could rescue the cells from growth arrest by IL-1, C6/vector or C6/myc cell lines were left unexposed or exposed to IL-1 for 24, 48, or 72 h, and the effect on growth was examined by [3H]thymidine incorporation studies. The vector-transfected cell lines showed approximately 25, 40, or 70% growth inhibition in response to IL-1 at 24, 48, or 72 h, respectively (Fig. 7A). By contrast, the C6/myc cell lines showed a significant decrease (P<0.0001 by the Student t test) in susceptibility to IL-1, with a maximum of approximately 50% growth inhibition after 72 hours exposure to IL-1 (Fig. 7A). We also examined whether ectopic expression of c-myc rescued cells grown as spheroids from the growth arresting action of IL-1. Stable transfectants expressing c-myc or vector were grown at maximum density on a non-adherent culture surface to obtain spheroids. Individual speroids were transferred to a 24-well culture plate and then either left unexposed or exposed to IL-1, and the increase in spheroid volume was determined over a 11 day period. The findings from these experiments (Fig. 7B) indicated that IL-1 caused a 30 to 45% growth inhibition of spheroids from vector-transfected cells, and a 10 to 15% growth inhibition of spheroids from myc-transfected cells. These findings are consistent with those from the [3H]thymidine incorporation studies and suggest that ectopic expression of c-myc protects A375-C6 cells

from the growth-inhibitory action of IL-1.

G. Ectopic overexpression of c-myc prevents hypophosphorylation of RB The retinoblastoma susceptibility gene product RB is a key regulator of the G1 phase growth arrest action of IL-1 in A375-C6 cells (Muthukkumar et al., 1996). Exposure to IL-1 causes hypophosphorylation of RB protein that is functionally required for growth arrest (Muthukkumar et al., 1996). Because ectopic expression of c-myc protects the cells from IL-1-inducible growth arrest, we sought to determine whether this protective mechanism was linked to the phosphorylation status of RB. The C6/vector.L1 or C6/ myc.L1 cell lines were left unexposed or exposed to IL-1 for 48 h and protein extracts were subjected to Western blot analysis. As seen in Fig. 7C, untreated C6/vector. L1 cells contained about 10% of total RB in the hypophosphorylated form, and treatment with IL-1 caused an accumulation of the faster-migrating, hypophosphorylated form of RB with a concomitant decrease (about 50%) in the slower-migrating, hyperphosphorylated form of RB. By contrast, there was only a minimal change in the phosphorylation status of RB in the C6/myc.L1 cells exposed to IL-1, with <15% hypophosphorylated RB accumulating after 48 h (Fig. 7C). These findings indicate that ectopic expression of c-myc abrogates the IL-1-inducible events that lead to hypophosphorylation of RB.

Figure 7. The effect of c-myc overexpression on IL-1-inducible growth inhibition and hypophosphorylation of RB. (A) C6/myc.L1 or vector.L1 cultures were exposed to IL-1 or vehicle for 24, 48, and 72 h, and then subjected to [3 H]thymidine incorporation studies. Each value point is a mean of 12 observations from 3 different experiments. Error bars indicate Âąstandard deviations. (B) Spheroids produced from C6/myc.L1 or C6/vector.L1 cultures were exposed to IL-1 or vehicle, and 3-dimensional growth was determined at the various time points indicated. Each value point is a mean of 24 observations from 3 separate experiments. (C) C6/myc.L1 or C6/ vector.L1 transfected cells were unexposed (-) or exposed (+) to IL-1 for 48 hours, and then whole-cell protein extracts were prepared from the cells and subjected to Western blot analysis using the anti-RB antibody. The slow-migrating differentially phosphorylated forms of RB (pRB) and the fast-migrating hypophosphorylated form of RB (RB) are indicated.

IV. Discussion The present study used an in vitro growth arrest system to determine the effect of EGR-1 expression on a G1 phase growth arrest pathway. We determined that EGR-1 functions to protect A375-C6 cells from the growth arresting action of IL-1. Furthermore, this study demonstrated that EGR-1 can regulate the expression of the cmyc gene via a (TCC)n EGR-1-binding element. EGR-1 has been previously shown to directly bind to this element (Wang and Deuel, 1992), and our present studies indicate an interaction between EGR-1 and the c-myc promoter leading to upregulation of the c-myc gene. Consistent with these observations, ectopic expression of functionally active c-MYC was sufficient to protect the A375-C6 cells from the growth arresting action of IL-1. Our previous studies (Muthukkumar et al., 1996) have defined RB hypophosphorylation as a key requirement for IL-1-inducible G1 phase cell cycle growth arrest. The findings of the present study indicated that rescue of the cells from IL-1-inducible growth arrest by ectopic c-MYC protein is linked to maintenance of RB in the hyperphosphorylated form. Thus, this study has identified c-myc as a downstream target of EGR-1 that counteracts growth arrest by preventing RB hypophosphorylation. Our previous studies have shown that EGR-1 is induced by IL-1 in A375-C6 cells (Sells et al., 1995). Inhibition of either EGR-1 expression by using an antisense oligomer or EGR-1 function by using a dominant-negative mutant enhances the G1 phase growth arrest response to IL-1 (Sells et al., 1995). Consistent with these observations, ectopic expression of EGR-1 results in abrogation of IL-1-inducible growth arrest. A number of previous studies have provided a circumstantial link between EGR-1 induction and a mitogenic response in diverse cell types (Gashler and Sukhatme, 1995). Moreover, EGR-1 null female mice are infertile, suggesting that EGR-1 is a positive effector in the reproduction process (Lee et al., 1996). In general, these findings suggest a role for EGR-1 in positive modulation of cell growth. The mechanism by which EGR-1 abrogates growth arrest and thereby positively modulates growth is dependent on the ability of the protein to function as a transcription factor. Although consensus EGR-1binding sites had been identified in the promoter regions of several growth-associated genes, the transregulation of these gene promoters by EGR-1 had neither been demonstrated nor shown to have a biological significance. The present study used transfection assays to demonstrate that a novel (TCC)n motif that is found in the promoter regions of many growth-related genes, can confer EGR-1responsiveness on the c-myc promoter. Thus, a role for EGR-1 can be envisioned in the upregulation of other genes, such as those encoding growth factors, growth factor receptors, or protooncogenes, that contain the EGR-1- responsive (TCC)n motif. Previous studies of diverse cell types have shown that modulation of c-myc expression can directly alter cell growth (Evan and Littlewood, 1993). The demonstration that EGR-1 can transactivate the c-myc promoter and upregulate the expression of c-myc at the RNA and protein levels suggests a novel mechanism for cell growth regulation by EGR-1. Moreover, the findings suggest that by upregulating the expression of c-myc, EGR-1 may regulate the expression of c-myc-responsive genes, and thereby expand the number of potential downstream target genes for enhanced signal transduction. Moreover, because there is an overlap in the phenotypic responses to EGR-1 and c-MYC, we hypothesize, on the basis of the findings of this study, that the overlapping functions are a consequence of a linear regulatory pathway, in which EGR-1 upregulates the expression of c-myc. Future studies may test the validity of this hypothesis.

The A375-C6 cells served as an excellent model system for studying the relevance of c-myc expression because in response to IL-1 or serum-starvation these cells show a downregulation of c-myc expression. Ectopic overexpression of c-MYC abrogated the growth arrest response to IL-1, suggesting that downregulation of c-MYC is functionally required for growth arrest. These findings are consistent with those reported for another growth-inhibitory cytokine TGF, which causes c-MYC downregulation as part of a growth arrest response in other tumor cells (Alexandrow et al., 1995; Pietenpol et al., 1990). Most importantly, IL-1 shows pleiotropic effects on cell growth: it inhibits the growth of certain tumor cells but stimulates the growth of other tumor cells (cited in Rangnekar et al., 1992). We and others have shown that in fibroblast cells in which it serves a growth-stimulatory signal, IL-1 induces the expression of c-myc (Joshi-Barve et al., 1993; Kessler et al., 1992; Rangnekar et al., 1991). The present findings about the functional requirement of c-myc downregulation as part of a growth arrest response to IL-1, suggest that the pleiotropic responses to IL-1 may be dependent upon whether IL-1 induces or down-regulates c-myc expression. An analysis of c-myc expression in a broad panel of IL-1-responsive cell lines whose growth is positively or negatively regulated by IL-1 will help evaluate this hypothesis. IL-1-inducible growth arrest of A375-C6 cells is associated with an accumulation of hypophosphorylated RB (Muthukkumar et al., 1995), and ectopic overexpression of an EGR-1 target gene product, c-MYC, prevents growth arrest by maintaining RB in a hyperphosphorylated state. This finding is in agreement with those of other studies that have shown that c-MYC can modulate the phosphorylation status of RB (Galaktionov et al., 1996; Marcu et al., 1992). Because c-MYC can modulate the activity of cdk4 by transcriptionally regulating the expression of cdc25, a phosphatase that directly controls the activity of cdk4, and because kinase-active cdk4 is required for maintenence of hyperphosphorylated RB (Galaktionov et al., 1996), this pathway should be further investigated to identify the potential mechanism by which cMYC prevents RB hypophosphorylation in response to IL-1.

V. Concluding remarks This study has identified a novel mechanism by which EGR-1 counteracts negative growth signals and thereby acts as a positive modulator of growth. The identification of c-myc, a key regulator of positive or negative growth responses, as a functional downstream gene target of EGR-1 suggests an important role for EGR-1 in growth control. Thus, by using c-MYC as a downstream target, EGR-1 may expand the spectrum of its potential target genes and phenotypic endpoints. Because EGR-1 and c-MYC show overlapping biological functions (such as rescue from growth arrest and the stimulation of proliferation, differentiation, or apoptosis), the findings of this study can be extended to determine the effect of the crosstalk on these processes in diverse experimental models.

Acknowledgments This work was supported by NIH Grant R01 CA60872 and by Council for Tobacco Research-USA Grant 3490 (to VMR). Received 27 May 1998; accepted 10 June, 1998

References Ahmed, M. M., K. Venkatasubbarao, S. M. Fruitwala, S. Muthukkumar, D. P. Wood, Jr., S. F. Sells, M. Mohiuddin, and V. M. Rangnekar (1996) EGR-1 induction is required for maximal radiosensitivity in A375-C6 melanoma cells. J. Biol. Chem. 271, 29231-29237. Alexandrow, M. G., M. Kawabata, M. Aakre, and H. L. Moses (1995). Overexpression of the c-Myc oncoprotein blocks the growth-inhibitory response but is required for the mitogenic effects of transforming growth factor beta 1. Proc. Natl. Acad. Sci. USA 92, 3239-3243. Blackwell, T. K., J. Huang, A. Ma, L. Kretzner, F. W. Alt, R. N. Eisenman, and H. Weintraub (1993) Binding of MYC proteins to canonical and noncanonical DNA sequences. Mol. Cell. Biol. 13, 5216-5224. Blackwood, E. M., L. Kretzner, and R. N. Eisenman (1992) Myc and max function as a nucleoprotein complex. Curr. Opin. Genet. Dev. 2, 227-235. Call, K. M., T. Glaser, C. Y. Ito, A. J. Buckler, J. Pelletier, D. A. Haber, E. A. Rose, A. Kral, H. Yeger, W. H. Lewis, C. Jones, and D. E. Housman (1990) Isolation and characterization of a zinc finger polypeptide gene at the human chromosome 11 Wilms' tumor locus. Cell 60, 509-520. Cao, X., R. A. Koski, A. Gashler, M. McKiernan, C. F. Morris, R. Gaffney, R. V. Hay, and V. P. Sukhatme (1990) Identification and characterization of the EGR-1 gene product, a DNA-binding zinc finger protein Induced by differentiation and growth signals. Mol. Cell. Biol. 10, 1931-1939. Christy, B. A., and D. Nathans (1989) DNA binding site of the growth factor-inducible protein Zif268. Proc. Natl.

Acad. Sci. USA 86, 8737-8741. Cole, M. D. (1986) The myc oncogene: its role in transformation and differentiation. Annu. Rev. Genet 20, 361-385. Crosby, S. D., J. J. Puetz, K. S. Simburger, T. J. Fahrner, and J. Milbrandt (1991) The early-response gene NGFI-C encodes a zinc finger transcriptional activator and is a member of the GCGGGGGCG (GSG) element-binding protein family. Mol. Cell. Biol. 11, 3835-3841. Day, M. L., S. Wu, and J. W. Basler (1993) Prostatic nerve growth factor inducible A gene binds a novel element in the retinoblastoma gene promoter. Cancer Res. 53, 5597-5599. DesJardins, E., and N. Hay (1993) Repeated CT elements bound by zinc finger proteins control the absolute and relative activities of the two principal human c-myc promoters. Mol. Cell. Biol. 13, 5710-5724. Drummond, I. A., S. L. Madden, P. Rohwer-Nutter, G. I. Bell, V. P. Sukhatme, and F. J. Rauscher III (1992) Repression of the insulin-like growth factor II gene by the Wilms' tumor suppressor WT1. Science 257, 674-678. Endo, Y., Matsushima, K., Onozaki, K., and J. J. Oppenhiem (1988) Role of ornithine decarboxylase in the regulation of cell growth by interleukin-1 and tumor necrosis factor. J. Immunol 141, 2342-2348. Evan, G., and T. D. Littlewood. (1993) The role of c-myc in cell growth. Curr. Opin. Genet. Dev. 3, 44-49 Galaktionov, K., X. Chen, and D. Beach (1996) Cdc25 cell-cycle phosphatase as a target of c-myc. Nature 382, 511-517 Gashler, A. L., S. Swaminathan, and V. P. Sukhatme (1993) A novel repression module, an extensive activation domain, and a bipartite nuclear localization signal defined in the immediate-early transcription factor EGR-1. Mol. Cell. Biol. 13, 4556-4571. Gashler, A., and V. P. Sukhatme (1995) Early growth response protein 1 (Egr-1): Prototype of a zinc finger family of transcription factors. Prog. Nucl. Acids Res. Mol. Biol. 50, 191-224. Goodrich, D. W., and W. H. Lee (1992) Abrogation by c-myc of G1 phase arrest induced by RB protein but not by p53. Nature 360, 177-179. Hanson, K. D., M. Shichiri, M. R. Follansbee, and J. M. Sedivy (1994) Effects of c-Myc expression on cell cycle progression. Mol. Cell. Biol. 14, 5748-5755. Hartwell, L. H., and M. B. Kastan (1994) Cell cycle control and cancer. Science 266, 1821-1827. Hermeking, H., and D. Eick (1994) Mediation of c-myc-induced apoptosis by p53. Science 265, 2091-2093. Hoffman-Liebermann, B., and D. A. Liebermann (1991) Interleukin-6 and leukemia inhibitory factorinduced terminal differentiation of myeloid leukemia cells is blocked at an intermediate stage by constitutive cmyc. Mol. Cell. Biol. 11, 2375-2381. Hu, R. M., and E. R. Levin (1994) Astrocyte growth is regulated by neuropeptides through Tis 8 and basic fibroblast growth factor. J. Clin. Invest. 193, 1820-1827. Janicke, R. U., Lee, H. H. F., and A. G. Porter (1994) Nuclear c-myc plays an important role in the cytotoxicity of tumor necrosis factor alpha in tumor cells. Mol. Cell. Biol. 14, 5661-5670. Janicke, R. U., X. Y. Lin, F. H. Lee, and A. G. Porter (1996) Cyclin D3 sensitizes tumor cells to tumor necrosis factor-induced, c-Myc-dependent apoptosis. Mol. Cell. Biol. 16, 5245-5253. Janson-Durr, P., A. Meichle, P. Steiner, M. Pagano, K. Finke, J. Botz, J. Wessbecher, G. Draetta, and M. Eilers (1993) Differential modulation of cyclin gene expression by MYC. Proc. Natl. Acad. Sci. USA 90, 3685-3689. Joshi-Barve, S. S., V. V. Rangnekar, S. F. Sells, and V. M. Rangnekar (1993) Interleukin-1-inducible expression of gro- via NF-B activation is dependant on tyrosine kinase signaling. J. Biol. Chem. 268, 18018-18029. Kessler, D. J., M. P. Duyao, D. B. Spicer and G. E. Sonenshein (1992) NF-B-like factors mediate interleukin 1 induction of c-myc gene transcription in fibroblasts. J. Exp. Med. 176, 787-792. Krishnan, S., and V. M. Rangnekar. Unpublished data. Lau, L. F., and D. Nathans (1987) Expression of a set of growth-related immediate early genes in BALB/c 3T3 cells: coordinate regulation with c-fos or c-myc. Proc. Natl. Acad. Sci. USA 84, 1182-1186. Lee, S. L., Y. Sadovsky, A. H. Swirnoff, J. A. Polish, P. Goda, G. Gavrilina, and J. Milbrandt (1996) Luteinizing hormone deficiency and female infertility in mice lacking the transcription factor NGFI-A (Egr-1). Science 273, 1219-1221. Lemaire, P., O. Relevant, R. Bravo, and P. Charnay (1988) Two mouse genes encoding potential transcription factors with identical DNA-binding domains are activated by growth factors in cultured cells. Proc. Natl. Acad. Sci. USA 85, 4691-4695. Li, L-J., G. S. Naeve, and A. S. Lee (1993) Temporal Regulation of cyclin A-p107 and p33cdk2 Complexes Binding to the Human Thymidine Kinase Promoter Element Important for G1-S Phase Transcriptional Regulation. Proc. Natl. Acad. Sci. USA 90, 3554-3558. Lim, R. W., B. C. Varnum, and H. R. Herschman (1987) Cloning of tetradecanoyl phorbol esterinduced "primary response" sequences and their expression in density-arrested Swiss 3T3 cells and a TPA nonproliferative variant. Oncogene 1, 263-270. Luscher, B., and R. N. Eisenman (1990) New light on Myc and Myb. Genes Dev. 4, 2025-2035. Madden, S. L., D. M. Cook. J. F. Morris, A. Gashler, V. P. Sukhatme, and F. J. Rauscher III. (1991) Transcriptional repression mediated by the WT1 Wilms' tumor gene product. Science 253, 1550-1553. Maheswaran, S., S. Park, A. Bernard, J. F. Morris, F. J. Rauscher III, D. E. Hill, and D. A. Haber (1993) Physical and

functional interaction between WT1 and p53 proteins. Proc. Natl. Acad. Sci. USA 90, 5100-5104. Marcu, K. B., S. A. Bossone, and A. J. Patel (1992) myc function and regulation. Annu. Rev. Biochem. 61, 809-860. Milbrandt, J (1988) A nerve growth factor-induced gene encodes a possible transcriptional regulatory factor. Science 238, 797-799. Molnar, G. , A. Crozat, and A. B. Pardee (1994) The immediate-early gene Egr-1 regulates the activity of the thymidine kinase promoter at the Go -to-G1 transition of the cell cycle. Mol. Cell. Biol. 14, 5242-5248. Murray, A. W. (1992) Creative blocks: cell-cycle checkpoints and feedback controls. Nature 359, 599-604. Muthukkumar, S., P. Nair, S. F., Sells, N. G. Maddiwar, R. J., Jacob, and V. M. Rangnekar (1995) Role of EGR-1 in thapsigargin-inducible apoptosis in the cell line A375-C6. Mol. Cell. Biol. 15, 6262-6272. Muthukkumar, S., S. F. Sells, S. A. Crist, and V. M. Rangnekar (1996) Interleukin-1 induces growth arrest by hypophosphorylation of the retinoblastoma susceptibility gene product RB. J. Biol. Chem 271, 5733-5740. Nair, P., S. F. Sells, S. Muthukkumar, V. P. Sukhatme, and V. M. Rangnekar (1997) EGR- 1 dependent apoptosis is mediated by p53. J. Biol. Chem 272, 20131-20138 Nguyen, H., B. Hoffman-Liebermann, and D. Liebermann (1993) The zinc finger transcription factor Egr-1 is essential for and restricts differentiation along the macrophage lineage. Cell 72, 197-209. Packham, G., and J. L. Cleveland (1994) Ornithine decarboxylase is a mediator of c-MYC-induced apoptosis. Mol. Cell. Biol. 14, 5741-5747. Pavletich, N. P., and C. O. Pabo (1991) Zinc finger-DNA recognition: crystal structure of a Zif268-DNA complex at 2.1 A. Science 252, 809-817. Phillipp, A., A. Schneider, I. Vasrik, K. Finke, Y. Xiong, D. Beach, K. Alitalo, and M. Eilers (1994) Repression of cyclin D1: a novel function of MYC. Mol. Cell. Biol. 14, 4032-4043. Pietenpol, J. A., J. T. Holt, R. W. Stein, and H. L. Moses (1990) Transforming growth factor 1 suppression of c-myc gene transcription: role in Inhibition of keratinocyte proliferation. Proc. Natl. Acad. Sci. USA 87, 3758-3762. Qin, X. Q., D. M. Livingston, M. Ewen, W. R. Sellers, Z. Arany, W. G. Jr. Kaelin (1995) The transcription factor E2F-1 is a downstream target of RB action. Mol. Cell. Biol. 15, 742-755. Rangnekar, V. V., S. Waheed, and V. M. Rangnekar (1992) Interleukin-1-inducible tumor growth arrest is characterized by activation of cell type-specific early gene expression programs. J. Biol. Chem. 267, 6240-6248. Rangnekar, V. V., S. Waheed, T. J. Davies, F. G. Toback, and V. M. Rangnekar (1991) Antimitogenic and mitogenic actions of interleukin-1 in diverse cell types are associated with induction of gro gene expression. J. Biol. Chem. 266, 2415-2422. Rauscher, F. J., J. F. Morris, O. E. Tournay, D. M. Cook, and T. Curran (1990) Binding of Wilms' tumor locus zinc finger protein to the EGR-1 consensus sequence. Science 250, 1259-1262. Russo, M. W., C. Matheny, and J. Milbrandt (1993) Transcriptional activity of the zinc finger protein NGFI-A is influenced by its interaction with a cellular factor. Mol. Cell. Biol. 13, 6858-6865. Sells, S. F., S. Muthukkumar, V.P. Sukatme, S.A. Crist, and V. M. Rangnekar (1995) The zinc-finger transcription factor EGR-1 impedes interleukin-1-inducible growth arrest. Mol. Cell. Biol. 15, 682-692. Sukhatme, V. P. (1990) in cell growth: The Egr Family. J. Am. Soc. Nephrol. 1, 859-866. Sukhatme, V. P., X. Cao, L. C. Chang, C.-H Tsai- Morris, D. Stemenkovich, P. C. P. Ferreira, R. Cohen, S. A. Edwards, T. B Shows, T. Curran, M. M. Le Beau, E. D. Adamson (1988) A zinc finger-encoding gene coregulated with c-fos during growth and differentiation, and after cellular depolarization. Cell 53, 37-43. Swirnoff, A. H., and J. Milbrandt (1995) DNA-binding specificity of NGFI-A and related zinc finger transcription factors. Mol. Cell. Biol. 15, 2275-2287. Wagner, A. J., J. M. Kokontis, and N. Hay (1994) Myc-mediated apoptosis requires wild-type p53 in a manner independent of cell cycle arrest and the ability of p53 to induce p21/waf1/cip1. Genes Dev. 8, 2817-2830. Wang, Z. Y. , S. L. Madden, T. F. Deuel, F. J. Rauscher III (1992) The Wilms' tumor gene product, WT1, represses transcription of the platelet-derived growth factor A-chain gene. J. Biol. Chem. 267, 21999-22002.

Wang, Z Y., and T. F. Deuel. (1992) An S1 nucleasesensitive homopurine/ homopyrimidine domain in the PDGF A-chain promoter contains a novel binding site for the growth factorinducible protein EGR-1. Biochem. Biophys. Res. Commun. 188, 433-439. Wang, Z. Y. and T. F. Deuel (1996) S1-nuclease-sensitive DNA structures contribute to transcriptional regulation of the human PDGF A-chain. Prog. Nucl. Acid Res. Mol. Biol. 55, 227-244. Weinberg, R (1995) The retinoblastoma protein and cell cycle control. Cell 81, 323-330.