NCSSM Broad Street Scientific Research Journal 2011-2012

Broad S treet Scientific

Volume 1 | 2011-2012

The North Carolina School of Science and Mathematics Journal of Student STEM Research

ic Volume 1 | 2011-2012

The North Carolina School of Science and Mathematics Journal of Student STEM Research

Street Broad Scientific

Table of Contents vi 1 2

Broad Street Scientific Staff A Letter from the Chancellor Effect of Hyperthermia and Hypoxia on the Expression of Dysadherin, a Cancer Associated Cell-Membrane Glycoprotein Shown to Increase Metastasis Pranav Haravu, 2012

8 13

Inter-Alpha-Trypsin Inhibitor Heavy Chain 4: Functional Effects on the Inflammatory Response to Lung Injury Greeshma Somashekar, 2012

The RNA-Binding Protein HuR Binds and Stabilizes pre-mRNA in vivo Vipul Vacharajani, 2012

19

The Regulation of Centromere Function and Localization

24

Chomp the Graph

29

Engineering a Robust Gene Network in Pseudo-typed Packaging Cells for in vivo Production of Ecotropic Protein Bound Retroviral Vectors

Alyssa Ferris, 2012 Sam Magura, 2012 Vitchyr Pong, 2012 Elliot Cartee, 2011 Kevin Valakuzhy, 2011

Param Sidhu, 2013

35

Effect of Thiols on the Environmental Fate of Silver Nanoparticles

41

Neutrinos and Their Detection

45

Determination of Activity Coefficients of Wood Smoke Tracer in Artificial and Ambient Organic Semi-Volatile Aerosols

Avi Aggarwal, 2012 Jason Liang, 2013

Suqi Huang, 2012

Street Broad Scientific

Words from the editors Welcome to Broad Street Scientific: The NCSSM journal of student research in the fields of science, technology, engineering, and mathematics. This year marks the founding and creation of the first issue, filed with our students’ incredible research. We created Broad Street Scientific to showcase student research, increase awareness of scientific participation and demonstrate the level of work done by students to readers in and outside of the NCSSM community.

We hope you enjoy reading it.

The editors would like to thank the administration, faculty, and staff of the NCSSM for the opportunity to pursue our research goals in any of the science, technology, engineering or mathematics fields. The educational options that are provided to the students at this school are unparalleled by any high school in the state, and the student body recognizes the significance of such an investment in our, and the state’s, future. We would like to specifically thank our faculty advisor, Dr. Jonathan Bennett, for his advice and guidance through our first edition of the Broad Street Scientific. We would also like thank our Chancellor, Dr. Roberts, for his active support of this publication. Finally, the Broad Street Scientific would like to thank Ms. Lily Pham for providing recommendations and advice regarding the specifics of creating a research publication.

BroadStreetSci Online

broadstreetscientific.ncssm.edu Volume 1 | 2011-2012 | v

Street Broad Scientific

Broad Street Sci Staff Chief Editors

Pranav Maddi, 2012 Hun Wong, 2012

Publication Editors

Vincent Cahill, 2013 Anita Simha, 2013 Rithi Sridhar, 2013 Wey-Wey Su, 2013

Biology Editors

Peter Fan, 2012 William Ge, 2013

Physics Editors

Halston Lim, 2013 Mia De Los Reyes, 2012

Chemistry Editors

Kevin Huang, 2012 Tejas Sundaresan, 2013

Engineering Editor

Christopher Panuski, 2013

Math and Computer Science Editors Webmaster Faculty Advisor

vi | 2011-2012 | Volume 1

Calvin Deng, 2013 Aakash Indurkhya, 2012

Sam Magura, 2012 Dr. Jonathan Bennett

Street Broad Scientific

Letter from the Chancellor

“

”

Somewhere, something incredible is waiting to be known.

-Dr. Carl Sagan

Noted philosopher John Dewey once stated, “scientific principles and laws do not lie on the surface of nature. They are hidden and must be wrested from nature by an active and elaborate technique of inquiry.” This in part defines the essence of scientific research. I am proud to introduce the North Carolina School of Science and Mathematic’s first scientific journal, Broad Street Scientific.

Broad Street Scientific is a showcase of some of the best scientific research being done by students at NCSSM. Providing students with opportunities to apply their learning through research is not only vitally important in preparing and exciting students to pursue STEM degrees and careers after high school, but essential to encouraging innovative thinking that allows students to scientifically address major issues and problems we face in the world today and will face in the future. Opened in 1980, NCSSM was the nations first public residential high school where students study a specialized curriculum emphasizing science and mathematics. Teaching students to do research and providing them with opportunities to conduct high-level research in biology, chemistry, physics, the applied sciences and math is a critical component of NCSSM’s mission to educate academically talented students to become state, national and global leaders in science, technology, engineering and mathematics.

The research showcased in this publication is an example of the significant research that students conduct each year at NCSSM under the direction of the outstanding faculty at our school and in collaboration with researchers at major universities. For twentysix years NCSSM has showcased student research throughour annual Research Symposium each spring and at major research competitions such as the Siemens Competition in Math, Science and Technology, the Intel Science Talent Search, and the International Science and Engineering Fair. The publication of Broad Street Scientific is a great new opportunity to present the outstanding research being conducted by students each year at the North Carolina School of Science and Mathematics.

I would like to thank all of the students and faculty involved in producing Broad Street Scientific, particularly faculty sponsor Dr. Jonathan Bennett and chief editors Hun Wong and Pranav Maddi.

Indeed, “somewhere something incredible is waiting to be known.”

Sincerely, Dr. Todd Roberts, Chancellor North Carolina School of Science and Mathematics

Volume 1 | 2011-2012 | 1

Street Broad Scientific

Research

Effect of Hyperthermia and Hypoxia on the Expression of Dysadherin, a Cancer Associated Cell-Membrane Glycoprotein Shown to Increase Metastasis Pranav Haravu ABSTRACT: Though cancer treatments have significantly progressed, metastasized tumors still pose a significant challenge to modern therapeutic methods, resulting in poor prognoses and lower survival rates for patients with metastasized cancers. In this study we investigated the role of dysadherin in a possible mechanism for both hypoxia- and hyperthermiainduced metastasis. Dysadherin is a cell-membrane protein found highly expressed and glycosylated in cancer cells, but expressed only in a few normal cells. It down-regulates E-cadherin mediated cell-cell adhesion, facilitating metastasis. We treated Panc-1 and PC3 cells with 43°C hyperthermia and .5% O2 hypoxia. Western Blots, with β-actin as a loading control, showed a significant increase in expression of glycosylated dysadherin in Panc-1 cells exposed to hypoxic and hyperthermic conditions. However, unglycosylated dysadherin expression was not affected, leading us to believe unglycosylated dysadherin does not play a significant role in metastasis. IF imaging of PC3 cells also showed a significant increase in dysadherin expression in samples treated with hypoxia or hyperthermia, quantified by the ratio of fluorescence of the secondary, FITC, to that of DAPI. The results suggest hypoxia- and hyperthermia-induced metastasis function by increasing the expression of glycosylated dysadherin, which in turn down-regulates E-cadherin and promotes metastasis. Inhibiting key steps within this mechanism could serve as a potential drug target, greatly reducing the metastatic potential of malignant tumors. To our knowledge, this is the first report indicating that hypoxia- and hyperthermia-induced metastasis may occur by increasing glycosylated dysadherin expression.

Introduction Though cancer treatments have significantly progressed, metastasized tumors still pose a significant challenge to modern therapeutic methods, resulting in poor prognoses and lower survival rates for patients with metastasized cancers. (1) Despite improved therapeutic practices, metastasized tumors remain extremely difficult to treat for two fundamental reasons. One, we are unable to know where additional tumors will grow, and two; we are unable to prevent future tumors from reoccurring. It appears then, that the most promising method of treatment for malignant tumors would include preventing metastasis from occurring, for which we need to identify and understand the mechanism resulting in metastasis. In this report we seek to investigate the role of dysadherin in a possible mechanism for both hypoxia- and hyperthermia-induced metastasis.

Effect of Hypoxia A significant percentage of tumors express hypoxia, a cellular deficiency in available oxygen, due to the sporadic growth of cancerous cells. Normal cells grow at a much slower rate than their cancerous counterparts, allowing for capillaries to form and supply all cells with a steady supply of oxygen. (2) However, in cancerous tumors, high rates of cell growth result in portions of the tumor being distant from tumor vasculature, decreasing the cells’ oxygen supplies due to limitations in the diffusion of oxygen.

(2) Numerous studies have found that hypoxic cells show a much higher metastatic potential, indicating a correla2 | 2011-2012 | Volume 1

tion between hypoxia and metastasis. These in vitro studies have been re-confirmed by in vivo tests, which again show hypoxia-induced metastasis in cancerous tissue. (3) (4)These in vivo and in vitro tests have been even further validated by observing a strong correlation between tumor hypoxia and high levels of metastasis in clinical models.

When used as a prognostic indicator, high levels of tumor hypoxia have generally led to poor prognoses and survival rates. These studies provide grounds for hypothesizing that hypoxia, by some mechanism, leads to metastasis and tumor invasion.

Effect of Hyperthermia Many currently practiced and developing methods of cancer therapy involve treating a tumor with local hyperthermia in conjunction with chemotherapy, for example temperature sensitive liposomes. (5) The current use of localized hyperthermia is primarily to increase vasculature permeability within a tumor and reduce the amount of drug necessary to attain the therapeutic threshold. In addition, cancer cells are more vulnerable to increases in temperature than normal cells, due to a lack of heatshock proteins (HSPs). (6) (7)Normal cells, which undergo slower cell cycles, produce much larger quantities of HSPs than cancerous cells. The function of HSPs is to ubiquitously bind to hydrophobic proteins, preventing the protein from denaturing due to heat. Without HSPs, enzymatic and structural proteins within the cancer cells denature, generally resulting in apoptosis. (6) (7)Though hyperthermia is currently used as a clinical therapy,

Research the full effect of hyperthermia on cancer cell growth is still debated; as it has been shown to increase the rate of metastasis of tumors both in vivo and in vitro. (8) (9) We hypothesize that hyperthermia induced metastasis may be similar to hypoxia induced metastasis, in that both may be due to up-regulation of dysadherin.

Role of Dysadherin In Cell-Cell Adhesion In order for a benign tumor cell to become invasive and malignant, it must overcome cell-cell adhesion and enter a tumor’s vasculature. Cells are normally connected to each other through many junctional structures, such as tight junctions, adherens-type junctions, and desmosomes. (10) One such mechanism for mediating cell-cell adhesion in epithelial cells is E-cadherin, illustrated below (10). Ecadherin molecules expressed on adjacent cells bind to each other in the extracellular domain in a zipperlike fashion. In the intracellular domain, E-cadherin links to β and γ-Catenin, both of which link to α – Catenin, ultimately attaching to the cell’s actin cytoskeleton. (11) The presence of this E-cadherin cell-cell adhesion mechanism has shown to be crucial in preventing metastasis of tumors, indicating a causation rather than correlation. (12) (13) Mutations in the E-cadherin gene and lower expression levels in cancer cells in vivo and in vitro, as well as from clinical data, consistently show increased metastasis and tumor invasion. (14) (15) (16) (17) Hypoxic tumors have been shown to down-regulate E-cadherin expression, decreasing cell-cell adhesion and facilitating metastasis. (18)

Street Broad Scientific main, an extracellular domain, and a cytoplasmic region.

(19) Dysadherin was first thought to be similar to the RIC protein in mice, and was later shown to be FXYD5 (FXYD domain-containing ion transport regulator 5), a member of the FXYD family of proteins. (20) When first discovered, dysadherin was shown to be expressed in the majority of cancers, such as breast, cervical, lung, stomach, colon, and bladder, while only expressed in a few normal cell types, including lymphocytes, endothelial cells, and basal cells. (21) (19) Later however, it was shown to also be expressed in other epithelial tissues, such as kidney, primarily in the cortex, intestine, primarily in the duodenum, and lung. (20) In its original discovery, it was observed that cell lines transfected with increased amounts of dysadherin showed decreased cell-cell adhesion. Upon analysis of E-cadherin and catenin expression, it was found that compared to controls, E-cadherin and α-catenin expression was significantly decreased in cells over-expressing dysadherin. (19) In addition, it was found that dysadherin was over-expressed in regions of loose cell-cell contact; further indicating that dysadherin down regulates E-cadherin expression and promotes metastasis. (19) These initial results were replicated and observed numerous times, including in vivo and patient studies. Over expression of dysadherin leading to the down regulation of E-cadherin was observed in gastric, thyroid, head and neck, cervical, pancreatic, colorectal, and lung carcinomas, as well as a multitude of others.

When high dysadherin levels were observed, metastasis rates increased and survival rates decreased. (23-31) In this report we attempt to show that both hypoxia and hyperthermia may induce an increase in dysadherin expression, partly explaining both hypoxia- and hyperthermia-induced metastasis. This provides the missing link (step A in the schematic below), between hypoxia, hyperthermia, and metastasis.

Figure 1. E-cadherin-mediated cell–cell adhesion. There is an extracellular component that links to other E-cadherin complexes in a zipperlike fashion, and an inner complex that links to the cell’s cytoskeleton.

Discovered in 2002, dysadherin is a cancer associated glycoprotein that also has been shown to down-regulate E-cadherin mediated cell-cell adhesion. Dysadherin is comprised of a 178 amino acid sequence, located on 19q13.12, and consists of two hydrophobic regions, corresponding to a signal peptide and the transmembrane do-

Figure 2. Schematic showing possible major steps from hypoxia and hyperthermia to metastasis. The red arrow, the missing link, has not yet been shown, and is the objective of this paper.

Volume 1 | 2011-2012 | 3

Street Broad Scientific Methods and Materials Cell Culturing PC3 (prostate cancer) and Panc-1 (pancreatic cancer) cell lines were all established in our laboratory, cultured in F12K+ 10% Fetal Bovine Serum + 1% AnitAnti and DMEM + 10% Fetal Bovine Serum + 1% Anit-Anti respectively. For IF imaging, cells were cultured onto BD Biocoat cover slips. All cell culturing and treatment was conducted in a sterile environment.

Cell Treatments Each cell line was split into 4 different 10mL treatment plates (for cells to be harvested for Western blots) or 6-well plates (for cells cultured on cover slips for IF imaging). Hypoxia was induced for 24 hours at .5% O2. Hyperthermia was induced with 2 1 hour water baths, 24 hours apart, at 43°C. Two controls were also used, one set with no treatment (left in incubator), and another was exposed to the same dosage as the hyperthermia, except the water bath was set at 37°C. The water bath control was not used in the IF imaging.

Protein Harvesting Existing media in cell plates was decanted, and each plate was rinsed with PBS. Cells were then scraped into 15mL centrifuge tubes and centrifuged at 4,000 rpm for 10 minutes. Remaining media and PBS was suctioned off, leaving only the cell pellet at the bottom. This pellet was then suspended in lysis buffer (10 mM PBS pH 7.4, 0.5% Triton X-10, 2 mM CaCl2, 10 ug/ml leupeptin, 2 ug/ml pepstatin A, 10 ug/mL aprotinin) (19), transferred to 1.5 mL tubes, and incubated on ice for 30 minutes.

The samples were then centrifuged at 14,500 rpm for 20 minutes, and the supernatant, which contained the Triton X-100 soluble components of the cell (proteins), was collected and stored at -20°C, while the pellet was discarded.

Western Blots As the expected weight of the protein ranged from 40-55 kDa when glycosylated, and 19-20 kDa when unglycosylated, we separated 30μg protein samples in a 12% SDS/PAGE gel. We then transferred the proteins to poly-vinylidene difluoride (PVDF) membranes and blocked overnight at 4°C in a 5% milk blocking buffer. After washing in a 1% milk buffer, the membranes were incubated at room temperature for 2 hours with the primary antibody (1:200). The primary antibody used was Dysadherin (T-14) obtained from Santa Cruz BiotechnologyInc. (sc-30604). This antibody targets the extracellular domain of dysadherin, epitope 50-100.

Membranes were then rinsed and incubated for 2 hours at room temperature with anti-goat secondary antibodies (1:1000), and then developed using an ECL kit.

4 | 2011-2012 | Volume 1

Research Immunofluorescence Microscopy The cover slips onto which cells were cultured were first rinsed with PBS, fixed with 4% PFA, incubated for 2 hours with primary antibody, and then 2 hours with FITC conjugated goat secondary antibody. The slips were then mounted with a photo-bleacher reducer and DAPI nuclear stain. Images were taken for .5 second exposures at wavelengths corresponding to DAPI and FITC. Results Western Blots Initial Western Blots confirmed the expression of dysadherin in both cell lines at a MW of around 50kDa. Following the confirmation of dysadherin expression, we proceeded to expose the cell lines to treatment. The Western Blot for Panc-1 cells exposed to the 4 different treatments is given in figure 3. When we blotted for dysadherin, shown in figure 3, we see two sets of bands in each lane; a strong region of bands around 50 kDa, and a weaker almost background level set of bands between 15 and 20 kDa. The bands with MW ≈ 50 kDa are dysadherin molecules heavily O-glycosylated in the extracellular domain, while the much fainter bands between 15 and 20 kDa are most likely unglycosylated dysadherin. A β- actin loading control enabled us to compare lanes to each other.

Thus, we are able to see that the glycosylated dysadherin bands from cells exposed to hypoxia and hyperthermia (bands A” and D” respectively) are much darker than their controls, indicating that dysadherin expression increased when exposed to hypoxic and hyperthermic conditions. In addition, we notice that the levels of unglycosylated dysadherin expression, bands A’, B’, C’, and D’, are relatively equal between all treatments, indicating that the treatment most likely had no impact on unglycosylated dysadherin expression.

Immunofluorescence Microscopy We used ImageJ to measure the average pixel count for the entire field. The average pixel count of the images taken at the FITC wavelengths, those in Figure 5 (AvgFITC ) gave a ratio of conjugated FITC antibodies to total area, while that of the images taken at DAPI wavelengths, those in Figure 6, (Avg DAPI ) gave the ratio of cell nuclei (cells) to total area. The ratio of AvgFITC : AvgDAPI gives us the average ratio of conjugated FITC antibodies per cell, an indicator of the amount of primary antibody binding and thus the levels of dysadherin expression. These results are given in Figure 6, which shows the AvgFITC : AvgDAPI ratio of each sample, A, B, C, and D. The average AvgFITC : AvgDAPI ratios of all the samples from each set is shown in Figure 7. These results show statistically significant higher levels of FITC fluorescence per cell in the samples treated with hypoxia and hyperthermia than the control, indicating an increase in dysadherin expression.

Research

Street Broad Scientific

Illustrations

Figure 5. Images taken at a wavelength corresponding to FITC. The ordering of the images is identical to that in Figure 5.

Figure 3. Western Blot of Panc-1 cells blotted for dysadherin. Each lane has a different treatment; A has hypoxia, B is the control with no treatment, C is the hyperthermia control (a water bath set at 37°C ), and D is treated with hyperthermia. We see much stronger bands in regions A” and D” than in B” and C”, all of which correspond to glycosylated dysadherin (MW = 47-55 kDa). Regions A’, B’, C’, and D’ show similar band strengths and correlate to unglycosylated dysadherin (MW = 19 kDa). Thus, the expression of glycosylated dysadherin is shown to have increased under hypoxic and hyperthermic conditions.

Figure 6. Graphical representation of the AvgFITC : AvgDAPI ratios (average dysadherin expression per cell in each sample). Each cluster is a different set of samples, and the error bars represent standard error.

Figure 4. Images taken at a wavelength corresponding to DAPI. The first row (1A-1D) consists of the samples treated with hypoxia, the second row (2A-2D) consists of the samples set as a control, and the third row (3A-3D) consists of the samples treated with hyperthermia.

Discussion In the IF microscopy we saw that hypoxia and hyperthermia both significantly increased dysadherin expression levels in PC3 cells. We have observed that in Panc-1 cells, hypoxia and hyperthermia increase the expression of glycosylated dysadherin, as evidenced by the stronger bands at MW = 50 in A” and D” in Figure 4. In addition, we also see that the expression of unglycosylated dysadherin is seemingly not affected by hypoxia or hyperthermia. This has led us to believe that unglycosylated dysadherin would most likely not play a significant role in either hypoxiainduced or hyperthermia-induced metastasis.

Previous studies have found that when expressed in non-cancerous cells, dysadherin is highly unglycosylated and has an observed MW of around 20 kDa. (20) However, when expressed in malignant cancerous cells, dysadherin becomes highly glycosylated, raising its MW Volume 1 | 2011-2012 | 5

Street Broad Scientific

Figure 7. Graphical representation of the average values of the AvgFITC : AvgDAPI ratios of each treatment (average dysadherin expression per cell for each treatment). Here the error bars represent the standard deviation.

to 47- 55kDA, as we observed in the Panc-1 cell line.

(19) Since unglycosylated dysadherin is rarely expressed in cancer cells, we have been led to believe that it does not play a significant role in metastasis. This agrees with our observations that unglycosylated dysadherin expression is unaffected by hypoxia or hyperthermia, two factors that may lead to metastasis. We are then left with the conclusion that glycosylated dysadherin is most likely what plays a role in metastasis.

Studies focusing on topics related to ours have built two strong series of connected causative events; our project suggests the connection between these two halves of the full picture of hypoxia- and hyperthermia-induced metastasis. Within this connection between the two halves is where this project’s novelty lies. The first half is the observation of hypoxic and hyperthermic microenvironments correlating with an increase in the metastatic potential of cancers; the second half is the expression of dysadherin down-regulates E-cadherin mediated cell-cell adhesion, decreasing cell-cell adhesion, and promoting metastasis.

Our project suggests that hypoxia and hyperthermia may cause an increase in glycosylated dysadherin expression as evidenced by the Panc-1 cells, eventually decreasing cellcell adhesion and promoting metastasis.

Conclusions and Future Work Microscopy of PC3 cells showed increased dysadherin expression when exposed to hypoxia and hyperthermia, and western blots of Panc-1 cells exposed to hyperthermia and hypoxia showed significant increases in glycosylated dysadherin expression but no change in the expression of unglycosylated dysadherin. Based on previous literature correlating increased expression of dysadherin with increased metastatic potential, as well as hypoxia and hyperthermia increasing the rate of metastasis, we can conclude the following. This research indicates that hypoxia and hyperthermia may increase the rate of tumor metastasis by increasing the expression of glycosylated dysadherin, and thus down-regulating E-cadherin mediated cell-cell 6 | 2011-2012 | Volume 1

Research adhesion.

Further assays need to be run to determine relative levels of metastatic potential compared to levels of dysadherin expression with and without hypoxia or hyperthermia. We also plan to characterize the relationship between different levels of hypoxia and hyperthermia with dysadherin expression.

The next step after establishing the correlation between hypoxia, hyperthermia, and dysadherin, would be to determine the mechanism for hypoxia and hyperthermia induced up-regulation of dysadherin. Inhibiting key steps within this mechanism could serve as a potential drug target, greatly reducing the metastatic potential of malignant tumors.

It becomes evident that there is much work left before translating this prospect into therapy or even a thorough understanding of the mechanism, but this study opens the door for future work. To our knowledge, this is the first report indicating that hypoxia-induced metastasis and hyperthermia-induced metastasis may occur by increasing glycosylated dysadherin expression.

Street Broad Scientific

Research References 1. Metastasis Mechanism. Geiger, Thomas R and Peeper, Daniel S. 2009, Biochimical et Biophysica Acta, pp. 293-308.

2. Hypoxia modulated gene expression: angiogenesis, metastasis and therapeutic exploitation. Dachs, G.U. and Tozer, G.M. 13, s.l. : European Journal of Cancer, 2000, Vol. 36, pp. 1649-1660.

3. Progression and metastasis in a transgenic mouse breast cancer model: Effects of exposure to in vivo hypoxia. Kalliomaki, Tuula M, et al. 2009, Cancer Letters, pp. 98-108.

4. Tumor Hypoxia Correlates with Metastatic Tumor Growth of Pancreatic Cancer in an Orthotopic Model. Buchler, Peter, et al.

2004, Journal of Surgical Research, pp. 295-303.

5. The development and testing of a new temperature-sensitve drug delivery system for the treatment of solid tumors. Needham, David and Dewhirst, Mark W. s.l. : Advanced Drug Delivery Reviews, 2001, Advanced Drug Delivery Reviews, Vol. 53, pp. 285-305.

6. The cellular and molecular basis of hyperthermia. Hildebrandt, Bert, et al. s.l. : Critical Reviews in Oncology/Hematology, 2002, Vol. 43, pp. 33-56.

7. Effect of hyperthermia on the viability and the fibrinolytic potential of human cancer cell lines. Fukao, Hideharu, et al. 2000, Clinica Chimica Acta, pp. 17-33.

8. Hyperthermia increases the metastatic potential of murine melanoma. Oliveira-Filho, R.S., Bevilacqua, R.G. and Chammas, R. 1997, Brazilian Journal of Medical and Biological Research, pp.

941-945.

9. Effects of Total-Body Hyperthermia on Metastases from Experimental Mouse Tumors. Oda, Masayuki, Koga, Shigemasa and Maeta, Michio. 1985, Cancer Research, pp. 1532-1535.

10. The role of the cell-adhesion molecule E-cadherin as a tumoursuppressor gene. Christofori, Gerhard and Semb, Henrik. 1999, Trends in Biochemical Sciences, pp. 73-76.

11. The E-cadherin-catenin complex in tumour metastasis: structure, function and regulation. Beavon, I.R.G. 2000, European Journal of Cancer, pp. 1607-1620.

12. Restoring E-cadherin-mediated cell窶田ell adhesion increases PTEN protein level and stability in human breast carcinoma cells.

Li, Zengxia, et al. 2007, Biochemical and Biophysical Research Communications, pp. 165-170.

13. E-cadherin-mediated Cell-Cell Adhesion Prevents Invasiveness of Human Carcinoma Cells. Frixen, Uwe H, et al. 1991, The Journal of Cell Biology, pp. 173-185.

14. The E-cadherin cell窶田ell adhesion complex and lung cancer invasion, metastasis, and prognosis. Bremnes, Roy M, et al. 2002, Lung Cancer, pp. 115-124.

15. Influence of E-cadherin dysfunction upon local in vasion and metastasis in non-small cell lung cancer. Shibanuma, Hiroyuki, et al. 1998, Lung Cancer, pp. 85-95.

17. E-cadherin complex and its abnormalities in human breast cancer. Jiang, Wen G and Mansel, Robert E. 2000, Surgical Oncology, pp. 151-171.

18. Regulation of E-cadherin: does hypoxia initiate the metastatic cascade? Beavon, I.R.G. s.l. : Journal of Clinical Pathology, April 1999, pp. 179-188.

19. Dysadherin, a cancer-associated cell membrane glycoprotein, down-regulates E-cadherin and promotes metastasis. Ino, Yoshinori, et al. 2002, PNAS, pp. 265-270.

20. Interaction with the Na, K-ATPase and Tissue Distribution of FXYD5 (Related to Ion Channel). Lubarski, Irina, et al. 2005, The Journal of Biological Chemistry, pp. 37717-37724.

21. Dysadherin: a new player in cancer progression. Nam, JeongSeok, Hirohashi, Setsuo and Wakefield, Lalage M. 2007, Cancer Letters, pp. 161-169.

22. Prognostic significance of dysadherin expression in patients with non-small cell lung cancer. Tamura, Masaya, et al. 2005, General Thoracic Surgery, pp. 740-745.

23. Dysadherin expression in gastrointestinal stromal tumors (GISTs). Liang, Jian, et al. 2009, Pathology - Research and Practice, pp. 445-450.

24. Dysadherin: Expression and Clinical Significance in Thyroid Carcinoma. Sato, Haruhiro, et al. 2003, The Journal of Clinical Endocrinology & Metabolism, pp. 4407-4412.

25. Clinical Significance of Dysadherin Expression in Gastric Cancer Patients. Shimada, Yutaka, et al. 2004, Clinical Cancer Research, pp. 2818-2823.

26. Significance of dysadherin and E-cadherin expression in differentiated-type gastric carcinoma with submucosal invasion. Maehata, Yoshitomo, et al. 2011, Human Pathology, pp. 558-567.

27. Prognostic Significance of Dysadherin Expression in Cervical Squamous Cell Carcinoma. Wu, Dan, et al. 2004, Pathology Oncology Research, pp. 212-218.

28. Dysadherin Expression in Head and Neck Squamous Cell Carcinoma; Association With Lymphangiogenesis and Prognostic Significance. Kyzas, Panayiotis A, et al. 2006, American Journal of Surgical Pathology, pp. 185-193.

29. Prognostic significance of dysadherin expression in advanced colorectal carcinoma. Aoki, S, et al. 2003, British Journal of Cancer, pp. 726-732.

30. Dysadherin Overexpression in Pancreatic Ductal Adenocarcinoma Reflects Tumor Aggressiveness: Relationship to E-cadherin Expression. Shimamura, Takeshi, et al. 2003, Journal of Clinical Oncology, pp. 659-667.

31. Dysadherin Expression Facilitates Cell Motilitiy and Metastatic Potential of Human Pancreatic Cancer Cells. Shimamura, Takeshi, et al. 2004, Cancer Research, pp. 6989-6995.

16. Immunohistological Analysis of E-cadherin, a-, B- and yCatenin Expression in Colorectal Cancer: Implications for Cell Adhesion and Signaling. Ghadimi, B.M., et al.

Volume 1 | 2011-2012 | 7

Street Broad Scientific

Research

Inter-Alpha-Trypsin Inhibitor Heavy Chain 4: Functional Effects on the Inflammatory Response to Lung Injury Greeshma Somashekar

Abstract: ITIH4, a serum protein, has been shown to have elevated concentrations in patients with diseases such as early stage ovarian cancer, breast cancer, and chronic obstructive pulmonary disease (COPD). Recent studies have therefore identified ITIH4 as a potential biomarker for these diseases. The goal of this study is to better understand the role of ITIH4 as a biomarker and specifically, how it affects the immune response to lung injury.

Wild type mice (ITIH4+/+) and ITIH4 knockout (ITIH4-/-) mice were exposed to lipopolysaccharide, an endotoxin used to induce inflammation in the lungs. Based on immunohistochemistry, we found that ITIH4 is expressed in the lung in bronchial epithelia and in alveolar macrophages and neutrophils. Tissue injury levels, cytokine and cell counts, and cell migration data were analyzed. These indicate that ITIH4 slows the migration of inflammatory cells to the site of an infection. Results suggest that ITIH4 is a biomarker of containment: it serves to localize the immune response to an infection. This information adds to the growing body of knowledge about ITIH4 and may also allow physicians to learn more about the unique immune response in an individual based on the concentration of ITIH4 present.

Introduction Rationale and Purpose In toxicology, a biomarker is a key molecule that links an environmental exposure to a resulting response in the body. These molecules provide clues about the relationship between certain exposures and the development of human disease. Biomarkers are important as they can potentially be used to determine risk, to screen for disease, to do differential diagnosis, and to monitor disease progression; perhaps their most important function is letting us know that something is out of the ordinary. It is helpful to determine the role of a biomarker in order to link it with a particular process or event in disease. A molecule could be a biomarker for anything from cell migration to increased inflammation. If we knew the exact role of a biomarker, its identification would allow us to learn more about the nature of a disease specific to a patient.

Recent studies based on proteomic analysis have identified inter-alpha-trypsin inhibitor heavy chain 4 (ITIH4) as a candidate biomarker for several diseases (7). Researchers arrived at this conclusion after determining that the concentration of ITIH4 is increased relative to baseline levels in patients with certain diseases such as early stage ovarian cancer, breast cancer, and chronic obstructive pulmonary disease (COPD) (1) (6) (9). The protein has never been researched in the lung and its exact function as a biomarker is unknown. The fact that it persists across several diseases indicates that it may have an important function related to the immune response.

The goal of my project is to better understand the role of ITIH4 as a biomarker of disease. Due to its significance in various diseases, I hypothesized that ITIH4 will have an 8 | 2011-2012 | Volume 1

effect on the immune response to inflammation in the lungs.

A model of inflammation in mice was used to determine, specifically, where ITIH4 is expressed in the lung and how it affects the response to injury, cytokine counts, and immune cell migration and activation. No other research has been recorded on ITIH4’s role in the lung. Determining the role of ITIH4 allows for further investigation into exactly what is affected.

Once we know the aspect of disease that ITIH4 affects and the mechanism behind its actions, we may be able to regulate this process. This would be of significance as it may allow us to, for example, slow the progression of a disease or to inhibit it from attacking the body.

General Background Inter-alpha-trypsin inhibitor heavy chain 4 (ITIH4) is a glycoprotein that was discovered in a high-throughput screening assay a little over a decade ago (3).

ITIH4 is one of five heavy chain proteins belonging to the inter-α-trypsin inhibitor (ITI) family. It is the only heavy chain that is not incorporated into an “inter-α-inhibitor complex”; rather, it circulates unattached in serum, indicating that it may have a unique function unlike that of other proteins in the family.

Human ITIH4 has been sequenced and a homologous area has been mapped on the mouse genome. In this study, the immune response to inflammatory lung injury was analyzed in ITIH4 knockout (ITIH4-/-) mice and compared to that in wild type (ITIH4+/+) mice of the C57Bl/6 strain. Lipopolysaccharide, an endotoxin, was used to induce inflammation in mice (8). Inflammation is the initial response to a foreign pathogen by the innate immune system. Cytokines are soluble proteins sent out by damaged tissue that interact with other # cells and induce a state of inflammation in the infected tissue.

Research KC, MIP-2, IL-6, and TNF-Îą are inflammatory cytokines that we expected to find in our endotoxin-induced model [8. 10]. The infiltration of immune cells such as neutrophils and macrophages into the infected tissue results in pain and swelling. This research aimed to determine how ITIH4 affects different aspects of the immune response.

Results Our first objective was to establish the presence of our protein of interest, ITIH4, in the mouse lung. Immunohistochemical characterization was performed on lung tissue as well as BAL fluid from ITIH4-/- and ITIH4+/+ mice. Lung tissue slides provided clear images of bronchial epithelia, while BAL fluid provided images of inflammatory cells such as macrophages, neutrophils, and lymphocytes typically found in alveolar spaces. As expected, there was no staining for ITIH4 in the ITIH4 deficient (ITIH4-/-) mice, and positive staining was observed in ITIH4+/+ mice.

Protein level is often considered a quantitative indica-

Figure 1. Immunohistochemistry for ITIH4: ITIH4-/mice show no staining for ITIH4 (red) in an airway (A) or on alveolar inflammatory cells (arrowheads; B). Positive staining for ITIH4 was observed in bronchial epithelia (arrows; C) and in macrophages (arrows) and neutrophils (arrows, insert) in ITIH4+/+ mice (D). DAPI was used for nonspecific staining of all cells. 40-80x magnification.

Figure 2. Protein Quantification: A. Protein levels are significantly lower in ITIH4-/- BAL fluid than in ITIH4+/+ BAL fluid.

Street Broad Scientific tion of physical injury. A Bradford assay done previously in the lab had shown that ITIH4-/- BAL fluid contained less protein than ITIH4+/+ BAL fluid (see Figure 2). To analyze tissue injury qualitatively, sectioned lung tissue from the ITIH4-/- and the ITIH4+/+ mice was stained with H&E and imaged using a Confocal microscope at 10x magnification. Vascular leakage around airways and blood vessels was observed in ITIH4+/+ tissue. This response was expected in response to moderate endotoxin exposure. In ITIH4-/- tissue, a decreased level of vascular leakage was observed. Since vascular leakage is indicative of protein level, these data support the results obtained previously and suggests that deficient ITIH4-/- mice sustain less injury post endotoxin exposure than wild type ITIH+/+ mice, indicating that the presence of ITIH4 increases the immune response surrounding an infection.

Figure 3. Qualitative Assessment of Injury: B. H&E stain at 10x magnification of lung injury surrounding airways (1) and blood vessels (2) in ITIH4-/- tissue shows mild vascular leakage (arrowheads). C. ITIH4+/+ tissue at 10x magnification shows an increased amount of leakage (arrows) suggesting greater injury.

Figure 4. Cytokine quantification: The concentrations of KC and MIP-2 are significantly higher in ITIH4+/+ BAL fluid (p < 0.0001). Pro-inflammatory cytokines IL-6 and TNF-Îą, however, had similar concentrations in ITIH4+/+ and ITIH4-/- BAL fluid.

Volume 1 | 2011-2012 | 9

Street Broad Scientific Having analyzed the effect of ITIH4 on injury, the next step was to determine its effect on the inflammatory response. BAL fluid from both groups of mice was assayed for four different cytokines (KC, MIP-2, IL-6, TNF-α) using standard ELISA protocol. Pro-inflammatory cytokines IL-6 and TNF-α were tested initially due to their proven elevation in response to acute inflammation [1, 7].

However, the presence of ITIH4 made little difference in the concentrations of these two cytokines at the acute phase time point of four hours post exposure.

KC and MIP-2 are chemokines that are responsible for the recruitment of inflammatory cells like neutrophils and macrophages to the site of injury. Their concentrations are also known to rise post endotoxin exposure [1].

ELISA testing showed that KC and MIP-2 concentrations are significantly higher in ITIH4+/+ BAL fluid than in ITIH4-/- fluid (p < 0.0001). The abundance of these chemokines suggests that more inflammatory cells are called to the site of inflammation in ITIH4+/+ mice.

This conjecture was tested by counting cells at the same time point.

Figure 5 (above). Cell Differentials: Neutrophil counts are significantly lower in ITIH4+/+ BAL fluid (p<0.0001). However, macrophage and lymphocyte counts are only slightly lower in ITIH4+/+ BAL fluid and were not statistically significant.

Cell counts were performed on inflammatory cells from BAL fluid. Results showed that the number of neutrophils local to the site of injury at four hours post exposure was significantly lower (p < 0.0001) in ITIH4+/+ mice. After finding an increased concentration of neutrophil attractants KC and MIP-2 in ITIH4+/+ mice, this was unexpected. How could there be more signaling molecules calling neutrophils to the site of infection in ITIH4-/- mice, but fewer neutrophils actually present?

The cell counts thus were in apparent contradiction with the chemokines quantification results.

Past research indicates that ITIH4 is known to interact with actin [5, 6]. Actin plays a role in regulating cell motility. Therefore, it was hypothesized that the presence of ITIH4 may have an effect on cell migration. This would explain the decreased numbers of inflammatory 10 | 2011-2012 | Volume 1

Research cells because ITIH4 could be slowing the migration of neutrophils and macrophages to the site of inflammation.

This hypothesis was tested by conducting a neutrophil chemotaxis assay.

Neutrophil migration assay results showed that ITIH4 does, in fact, inhibit cell motility. The greatest level of migration was observed in ITIH4-/- cells through ITIH4-/- serum while the least migration was observed in ITIH4+/+ cells through ITIH4+/+ serum (p < 0.001).

When either the serum or the cells were ITIH4 deficient, intermediate results were observed. The difference was statistically significant even after correction for multiple comparison testing.

Figure 6. (below). Neutrophil in vitro migration assay: Migration is higher in ITIH4-/- serum than in ITIH4+/+ serum. Also, cells deficient of ITIH4 migrate faster than cells with ITIH4. Results were statistically significant (p < 0.05) in both cases.

To test whether our observations were limited to a specific lung injury model or cell type, differential cell counts were also done on ITIH4+/+ and ITIH4-/- BAL fluid three days after bleomycin exposure and three days after naphthalene exposure. Bleomycin is the product of a fungus and exposure is known to induce intratracheal inflammation and fibrosis (scarring) on a time course of 3 – 21 days post exposure. Naphthalene is known to induce temporary airway injury in epithelial cells [7]. Neutrophil levels were significantly higher in ITIH4-/- mice (p < 0.05) post bleomycin exposure while macrophage levels were significantly higher in ITIH4-/- mice (p < 0.01) post naphthalene exposure. These results indicate that the effect of ITIH4 is nonspecific to a particular type of cell or injury.

Conclusion Based on immunohistochemistry, we determined that ITIH4 is expressed in bronchial epithelia and alveolar macrophages and neutrophils. Lung injury, protein levels, and chemokines levels were lower in ITIH4-/mice

Street Broad Scientific

Research

Figure 7. A. Cell differential count three days post bleomycin exposure indicates an increase in neutrophils in ITIH4-/- mice. B. Three days post naphthalene exposure, there is also an increase in macrophages in ITIH4-/- mice. than in ITIH4+/+ indicating that the presence of ITIH4 promotes the inflammatory response at the site of infection.

Additionally, total cell and neutrophil counts are elevated in ITIH4-/- mice relative to the ITIH4+/+ mice. This is explained by migration assay data, which indicates that ITIH4 presence inhibits migration. To summarize: ITIH4 promotes the immune response but slows immune cell migration post endotoxin exposure in the lungs.

We found that ITIH4 is expressed in bronchial epithelia and in alveolar macrophages and neutrophils, which led to questions about its function. Additional results indicate that ITIH4 plays a role in increasing the inflammatory response at the site of infection. It simultaneously slows the migration of inflammatory cells to the affected area. The acute phase immune response is a crucial bodily defense against infection.

In many diseases such as COPD and various cancers, the initial immune response could cause greater damage than the disease itself. Since the concentration of ITIH4 is elevated in these diseases, ITIH4 may serve to prevent the body from overreacting to infections. This information is a significant addition to the current body of knowledge about ITIH4, because we are able to better understand its role as a biomarker.

Based on collected data, we believe that ITIH4 may be a containment protein. The fact that ITIH4 promotes the inflammatory response at the site of infection while slowing the migration of immune cells suggests that its primary purpose is to localize the immune response to a specific region. Often, a large scale immune response is more dangerous than the infection itself. Therefore, ITIH4 may prevent the immune response from proliferating at a rate that would cause more harm than good. At the same time, it contains the response within a reasonable area, preventing the infection from spreading too quickly.

The nature of the interaction between ITIH4 and actin is unknown. Do the two bind directly or are other molecules and proteins involved? Immunoprecipitation can be used to find the binding partners of isolated ITIH4 to determine the physical nature of the bond. All cell migration

assays can be repeated with human neutrophils and macrophages in the future. This is a pending translational study. Human ITIH4 can be purified, ensuring that ITIH4 is the only molecule causing the differences in migration, and can be tested at varying concentrations. The concentration gradient has the potential to suggest a biological range for humans in which the presence of ITIH4 affects the immune response to infection and cell migration. ITIH4 is known to be elevated in sera of patients with many cancers [4] and therefore its effect on cell migration may extend to cancer metastasis. This would be an interesting area to look into.

ITIH4 is known to have three protein domains: the vault domain (VLT), the von Willebrand domain (VWA), and the multi-copper oxidase domain (MCOD). VLT and VWA are homologous to all members of the ITI family, while MCOD is unique to ITIH4. VWA is known to bind to integrins and to promote adhesion, while MCOD plays a role in iron metabolism [3, 4, 5]. These known functions allowed us to make predictions about the roles of these three domains in ITIH4: VWA may affect immune cell migration and MCOD may affect immune cell activation.

Future work will focus on finding a correlation between certain domains and effects on immune cell activation and migration. Focusing our research on specific domains will allow us to determine the mechanisms behind ITIH4â&#x20AC;&#x2122;s role as a biomarker.

To test whether our observations were limited to a specific lung injury model or cell type, differential cell counts were also done on ITIH4+/+ and ITIH4-/- BAL fluid three days after bleomycin exposure and three days after naphthalene exposure. Bleomycin is the product of a fungus and exposure is known to induce intratracheal inflammation and fibrosis (scarring) on a time course of 3 â&#x20AC;&#x201C; 21 days post exposure. Naphthalene is known to induce temporary airway injury in epithelial cells [7]. Neutrophil levels were significantly higher in ITIH4-/- mice (p < 0.05) post bleomycin exposure while macrophage levels Volume 1 | 2011-2012 | 11

Street Broad Scientific were significantly higher in ITIH4-/- mice (p < 0.01) post naphthalene exposure. These results indicate that the effect of ITIH4 is nonspecific to a particular type of cell or injury.

Materials and Methods Animals Right lung tissue and bronchoalveolar lavage (BAL) fluid samples were provided by a contract laboratory. Samples were extracted from ITIH4-/- and ITIH4+/+ mice of the C57Bl/6 strain. All mice had been exposed to endotoxin (5 ml of 1 mg/ml endotoxin solution was added to 200 ml phosphate buffered saline) via a Collision nebulizer for 2.5 hours.

Research standard tissue culture protocol. The solution was centrifuged to pellet cells and resuspended in 1 ml of media for cell counts. The cell concentration was then adjusted to 1 x 106 cells/ml and incubated at 37 °C for 7 days. The chemoattractant fLMP was added to the lower chambers on a 24 transwell plate. Cultured macrophages were added to the upper chambers. A similar template was used as for the neutrophil assay shown above, with fLMP replacing MIP-2 as the attractant. The plate was incubated for 2 hours at 37 °C. Cells were washed, fixed in 2% paraformaldehyde, and stained with DAPI. Transwell filters were removed and placed on slides, which were imaged. Cell counts were performed to quantify macrophage migration.

Cytokine Quantification in Lavage Fluid Commercially prepared ELISA kits (R&D Systems Duoset, Minneapolis, MN) were used to measure the concentrations of murine cytokines (KC, MIP-2, IL-6, and TNF-α) in BAL fluid as per the manufacturer’s provided instructions.

Differential Cell Counts BAL fluid aliquots containing alveolar cells were spun onto slides using Cytospin 3 (Shandon, Pittsburgh, PA).

Neutrophil, macrophage, and lymphocyte counts were determined after Wright-Giemsa staining.

Acknowledgements

Histology and Immunohistochemical Characterization Lungs were inflated with 10% formalin and embedded in paraffin. Five micrometer sections were obtained from the central portion of the lungs. Slides were stained either with hematoxylin and eosin (H&E) or with rabbit polyclonal anti-mouse ITIH4 antibody. Images were produced using a Zeiss 510 NLO Confocal microscope (Zeiss, Petersburg, VA).

[1] Bandow et al, 2008. Improved image analysis workflow for 2-D gels enables large-scale 2-D gel-based proteomics studies – COPD biomarker discovery study. Proteomics 8:1-11.

[2] Bhanumathy et al, 2002. Itih-4, a serine protease inhibitor regulated in interleukin-6-dependent liver formation: role in liver development and regeneration. Developmental Dynamics 223:59-68.

[3] Cai et al, 1998. Identification of mouse itih-4 encoding a glycoprotein with two EF-hand motifs from early embryonic liver. Biochimica et Biophysica Acta 1398:32-37.

[4] Choi-Miura et al, 2000. The novel acute phase protein, IHRP, inhibits actin polymerization and phagocytosis of polymorphonuclear cells.

Inflammation Research 49:305-310.

[5] Kroczynska et al, 2005. BIP co-chaperone MTJI/ERDJI interacts with inter-α-trypsin inhibitor heavy chain 4. Biochemical and Biophysical Research Communications 338:1467-1477.

[6] Mohamed et al, 2008. Lectin-based electrophoretic analysis of the expression of the 35 kDa inter-α-trypsin inhibitor heavy chain H4 fragment in sera of patients with five different malignancies. Electrophoresis 29:2645-2650.

[7] Pineiro et al, 1999. ITIH4 serum concentration increases during acute-phase processes in human patients and is up-regulated by interleukin-6 in hepatocarcinoma hepG2 cells. Biochemical and Biophysical Research Communications 263:224-229.

[8] Rojas et al, 2004. Endotoxin-induced lung injury in mice: structural, functional, and biochemical responses. Am J Physiol Lung Cell Mol Physiol 228:L333-L341.

[9] Tang Y et al, 2008. Progenitor/stem cells give rise to liver cancer due to aberrant TGF-β and IL-6 signaling. PNAS 105: 2445-2450.

[10] Wohlford-lenane et al, 1999. Cytokine gene expression after inhalation of corn dust. The American Physiological Society 99:L736-L743.

Neutrophil Isolation Bone marrow isolated from ITIH4-/- and ITIH4+/+ mice was stored in Hank’s balanced salt solution (HBSS).

Cells were centrifuged over a 52/64/72% Percoll gradient.

Purified neutrophils were removed from the layer between the 52% and 72% bands.

Neutrophil Chemotaxis Assay The chemoattractant MIP-2 was added to the lower chambers on a 24 transwell plate. Isolated neutrophils were added to the upper chambers. The plate was incubated for 1 hour at 37 °C. After incubation, cells in the lower chamber were counted to quantify neutrophil migration.

ITIH4-/- or ITIH4+/+ cells were incubated in ITIH4-/or ITIH4+/+ serum to create a 4x4 template.

Macrophage Migration Assay Cells harvested from extracted bone marrow were placed in media. Media was then removed and adherent cells were washed with prewarmed PBS. Macrophages were trypsinized using 0.25% trypsin-EDTA as per 12 | 2011-2012 | Volume 1

Stavros Garantziotis, Principal Investigator - Matrix Biology Group, NIEHS Vandy Stober, Lab Manager – Matrix Biology Group, NIEHS

Myra Halpin, North Carolina School of Science and Mathematics References

Research

Street Broad Scientific

The RNA-Binding Protein HuR Binds and Stabilizes pre-mRNA in vivo Vipul Vachharajani Abstract: HuR is an RNA-binding protein which is important in normal cell function. Dysregulation of HuR can lead to neoplastic phenotypes and is implicated in carcinogenesis in colon and brain cancer. Because of its ubiquity, HuR is an important target of study both in the context of diseases such as cancer but also in the basic study of RNA regulation.

HuR acts in the cytoplasm to prevent degradation of mature mRNA molecules, but HuR’s nuclear function is less well known. Recently, HuR targets which bind only intronic sequences in unspliced RNA have been proposed, which would reveal a HuR function which is distinct from the cytoplasmic one and may be significant in understanding the global cellular function of HuR.t We used RNA immunoprecipitation to confirm HuR binding to intronic only sequences of the proposed targets. We then used siRNA knockdown of HuR to show a functional effect of the observed binding on pre-mRNA transcript abundance.

Our results show a high degree of HuR binding to intronic sequences in the NFATC3 gene, but not to NFATC3 transcripts lacking the putative binding intron. Furthermore, HuR knockdown resulted in a significant decrease in pre-NFATC3 transcript levels. This indicates that HuR participates in functionally-significant, intron-specific binding, and suggests a complex nuclear function of HuR that may include splicing and splicing-concurrent processes.

Introduction HuR is an RNA-binding protein (RBP) which is highly expressed in all human tissues and shuttles between the nucleus and the cytoplasm (Saunders and Barber, 2003; Tenenbaum et al., 2002; Hinman and Lou, 2008). HuR upregulation is implicated in carcinogenesis and sustained growth of glioma and colon cancer (Dixon et al., 2001; Bolognani et al., 2011; Denkert et al., 2006). Furthermore, autoimmune dysfunction of HuR and other neuronallyexpressed Hu-family proteins is linked to neoplastic phenotypes, highlighting the importance of HuR in regulation of gene expression in normal cells (Hinman and Lou, 2008). The ubiquity of and complication associated with deficiency of HuR distinguishes it as a vital protein for all human cells.

Gene expression serves as a vital process in regulation of the cell cycle; specifically, dysfunctional gene regulation is often important in the development of cancer. Therapeutic strategies for cancer that specifically target gene expression patterns, thus, may prove to be extremely effective.

Gene regulation at the messenger RNA (mRNA) level, such as that demonstrated by HuR, has been shown to be a robust and wide-ranging mechanism and thus an interesting target to counteract the drastic changes in expression which occur during cancer (Keene, 2007; Kishore et al., 2010). Factors such as RBPs interact with target mRNA molecules, affecting their processing, stability, and cytoplasmic translation, thus controlling patterns of gene expression (Kishore et al., 2010). Often, these proteins recognize and bind highly-conserved and well-defined sequence motifs in the 3’ and 5’ un-translated regions

(UTRs) and recruit, enhance, or inhibit processing machinery (Kishore et al., 2010).

An emergent property of this sequence-specific targeting is a varied and combinatorial network of RNA-binding interactions, in which a certain mRNA can have several RBPs bound at once in a dynamic time- and sequence-dependent manner. The Post-Transcriptional RNA Operon model posits that these regulatory groups are a basic functional unit of expression in most eukaryotes and enable tight control of mRNA processing from transcription to translation (Keene and Tenenbaum, 2002; Keene, 2007).

This provides not only a powerful means of interpreting knowledge about these RNA-binding proteins but also a strategy for studying them. A necessary step in understanding this tight regulatory system, then, is to determine the function of each RBP both in terms of target binding and in the functional context of an RNA operon.

Unlike other RBPs, HuR lacks a well-defined target sequence motif; however, it is known to bind AU-rich elements (AREs) in the 3’UTR of target mRNA. HuR-ARE binding is known to increase the stability of the targets by blocking access to other destabilizing ARE-binding RBPs (Brennan and Steitz, 2001). However, while the cytoplasmic function of HuR is well-characterized, it is primarily localized in the nucleus, and a nuclear function for HuR remains poorly-defined (Fan and Steitz, 1998). Owing to the ubiquity and high expression of HuR, full characterization of HuR is critical to a complete picture of posttranscriptional gene regulation in cells (Zhu et al., 2006).

The goal of this research was to determine the role of HuR in nuclear pre-mRNA processing.

Volume 1 | 2011-2012 | 13

Street Broad Scientific HuR’s global function has been evaluated in recent studies, which used the novel Photoactivatable Ribonucleoside Cross-Linking and Immunoprecipitation (PARCLIP) technique to determine the binding sites of HuR across the set of transcribed RNA (Hafner et al., 2010).

Surprisingly, using PAR-CLIP, binding sites in intronic sequences were found, suggesting that HuR not only targets mature mRNA through the 3’ UTR but also immature pre-mRNA (Mukherjee et al., 2011; Lebedeva et al., 2011). This project investigated these intronic binding sites further.

In this report, we evaluated the possibility of un-spliced mRNA targeting by the vital RNA-binding protein HuR through RNA immunoprecipitation and quantitative Polymerase Chain Reaction, and used siRNA knockdown of HuR to determine HuR-related functional effects on pre-mRNA. Materials and Methods We used RNA immunoprecipitation to confirm HuR binding to intronic sequences in pre-mRNA and HuR siRNA knockdown to evaluate functional effects of HuR on target transcript abundance. Quantitative PCR was used to determine transcript levels during each of these procedures.

Once functional HuR binding was confirmed, a GFP splicing reporter assay using the pGint plasmid reporter construct was used to more directly look at intronic sequence-dependent effects. Also, a 4-thiouridine incorporation assay was used to investigate HuR-mediated effects on the splicing rates of target pre-mRNAs. Cell Culture Human Embryonic Kidney 293T cells (HEK293, ATCC), the same cell line used by Mukherjee et al, were used for HuR knockdown and RNA extraction (Mukherjee et al., 2011). Cells were cultured in Dulbecco’s Modified Eagle’s Medium (Gibco) at 37 degrees C and 5% CO2.

For knockdowns, cells were cultured in 6-well plates, and for RNA immunoprecipitation (RIP), cells were cultured in 15-cm culture plates. siRNA Knockdown of HuR Anti-HuR siRNA (Applied BioSystems) was transfected into HEK293T cells using Lipofectamine reagent.

Briefly, the siRNA was added to Lipofectamine and the resulting mixture was used to treat 3 wells of a 6-well plate.

The cells were incubated for 72 hours before harvesting to allow full knockdown and degradation of latent HuR present (Invitrogen Life Science, 2006). A negative control group treated in the same manner with random, nonspecific siRNA (Applied Biosystems) was cultured in the remaining 3 wells. Reduced levels of HuR transcript in the knockdown cells, as determined from RT-qPCR, were used as confirmation of successful knockdown of HuR 14 | 2011-2012 | Volume 1

Research protein.

RNA Extraction, Purification, and Reverse Transcription TRIzol, a phenol-based reagent, was used to extract total RNA from the plated cells via a solvent-extraction method (Invitrogen Life Science, 2010). The iScript reverse transcription kit was used to produce cDNA (BioRad Laboratories, 2000). An additional negative control aliquot of RNA was not treated with reverse transcriptase but still underwent the thermal cycling protocol in the same buffer used in the cDNA reaction. The cDNA obtained from both groups was used as a template for the Real Time Quantitative Polymerase Chain Reaction.

RNA Immunoprecipitation and Quantitative PCR (RIP-PCR) RNA Immunoprecipitation (RIP) is a technique in which mRNA bound to an immunoprecipitated RBP is processed via qPCR, microarray, or direct sequencing so as to evaluate the transcripts to which it is bound. The procedure performed here utilized RT-qPCR to measure amounts of bound target mRNA. RIP was performed as previously described (Keene et al., 2006). Briefly, both nuclear and whole-cell lysates were prepared using gentle lysis in Polysome Lysis Buffer at 4 degrees C, so as to preserve RNPs. The lysates were immunoprecipitated using anti-HuR 3A2 antibody on sepharose protein A beads (Sigma-Aldrich). A background control was obtained using normal mouse serum (NMS) instead of 3A2. Western Blotting was used on total, supernatant, and immunoprecipitated fractions of the lysates to determine enrichment for HuR. RNA was extracted from total and immunoprecipitated fractions of the lysates, and RT-qPCR performed on the fractions to determine enrichment of target RNAs. The presence of an enriched RNA species in the immunoprecipitate indicates a HuR target.

PCR Primer Design and RT-qPCR PCR primers for target genes were designed using the online Primer BLAST tool from the National Center for Biotechnology Information (NCBI), as well as the online ExonPrimer tool. Primer BLAST determines oligonucleotide sequences in both the forward and reverse strands that correspond to a unique RNA product, utilizing the NCBI Basic Local Alignment Search Tool (BLAST) to determine specificity (Altschul et al., 1990; Rozen and Skaletsky, 2000). ExonPrimer, originally designed to produce DNA primers which amplify entire exons, produces amplicons which are also appropriate for pre-mRNA amplification. Separate primers were generated to recognize the pre-mRNA and mature mRNA products as follows: The pre-mRNA primers were designed to generate a product that spanned at least one exon-intron junction, so that a controlled product would only be formed by an un-

Street Broad Scientific

Research

spliced RNA template. The mature mRNA primers were designed to create a product spanning at least one exon-exon junction, so that a product of specified size would only be generated by a properly-spliced (i.e. mature) mRNA template.

Product size was kept to a maximum of 250 bp, so as to maintain acceptable qPCR efficiency. The primers are depicted in Table 1:

HuR GAPDH Bactin MDM2 B2M H1A H4B pre-NFATC3 pre-CTCF NFATC3 CTCF

Forward CCTGTTCAGCAGCATTGGTGAAGT AGCCTCCCGCTTCGCTCTCT GGCACCCAGCACAATGAAGATCAA GTACCTACTGATGGTGCTGTAACC AGATGTCTCGCTCCGTGGCCTTA GGAGAAGAACAACAGCCGCAT GGATAACATCCAAGGCATCACC AGCCATGGGAAGGGAAATGTCTGA TTGACTGTCTCTGGACCGCTATCT TATGAAACTGAAGGTAGCCGAGGG AGATGCGCTAGTGGACAGATTGCT

Reverse TTCAGCGTGTTGATCGCTCTCTCT CCAGGCGCCCAATACGACCA ACTCGTCATACTCCTGCTTGCTGA AGCAATGGCTTTGGTCTAAC TGTCGGATGGATGAAACCCAGACA TTGAGCTTGAAGGAACCCGAG CGCCACGAGTCTCCTCATAAAT TTGGAAACCCAAGGTCCAAGGAGA CTGTTGCTGGCAAAGAAGAGCACA TTGGCTTGCAGTAGCGACTGTCTT TTTCGGACTCCTCCACAATGGCTT

Table 1. RT-qPCR Primers used in analysis of putative intronic HuR targets.

RT-qPCR was performed using the Roche LightCycler device on the cDNA with the following protocol: denaturing at 95° degrees C, annealing at 60° C for 7 seconds (s), extension at 72° C for 30 s, and fluorescence measurement at 78° C.

RT-qPCR Data Analysis Changes in transcript levels were quantified through ΔΔCt analysis as follows (Livak and Schmittgen, 2001).

RT-qPCR produces a threshold cycle (Ct) value for each sample, which is the point at which amplicon level is increasing maximally. This value is directly correlated with the amount of starting transcript in the sample.

Because each PCR cycle approximately doubles the amount of amplicon, the threshold cycle can be expressed as Ct ≈ log2(kn), where k is some constant related to replication efficiency and fluorophore fluorescence and n is the amount of starting transcript. Each RNA sample was also probed for a housekeeping gene GAPDH, whose expression is not believed to change between treatments. This gene is used to normalize the transcript numbers to account for cell-to-cell and plate-to-plate variations. Thus, the Ct value for each measured transcript was subtracted from the measured Ct for the GAPDH to yield the logarithm base 2 of a value proportional to the fold difference in expression over GAPDH (LFDGAPDH) as follows: LFDGAPDH = Ct(target) - Ct(GAPDH) = log2(k

)

The LFDGAPDH for each transcript under mock knockdown was then subtracted from that of the knockdown sample to yield the logarithm base 2 of the fold change

in expression of that transcript between mock knockdown and knockdown (LFC). This fold change can then be calculated by evaluating 2LFC. This gives knockdown expression as a fraction of mock knockdown expression.

The fold enrichment of each RIP target was determined using ΔΔCt, controlling 3A2-bound IP by NMS-bound IP. An example calculation is shown in Table 1 for NFATC3.

GapDH

NFATC3 LFDGAPDH

LFC

2-LFC

KD

21.21

23.46

2.25

-1.56

0.33915

Ctrl

18.83

19.52

0.69

Table 2: Example calculation of remaining transcript fraction using the ΔΔCt method. Data shown is for NFATC3 transcript.

Molecular Cloning of Fluorescent Protein Reporter Plasmids A GFP splicing reporter assay was performed using plasmid constructs as follows (Bonano et al., 2007). The pGint and pRint plasmid systems were used to compare splicing efficiency for three putative intronic HuR binding sites. Briefly, intronic sequences spanning the HuR binding sites for NFATC3, CCDC58, and CTCF were amplified via PCR and ligated into the pGEM-T vector system (Promega). The sequence was then further amplified in chemically-competent E. coli DH5-α cells. The inserts were cut from pGEM-T with ApaI and SalI restriction enzymes (New England Biolabs) and blunted with DNA Polymerase I Klenow Fragment (NEB).

The pGint and pRint plasmids, which each contain Green Fluorescent Protein and dsRED coding sequences interrupted by a spliceable multiple cloning site, were also cut Volume 1 | 2011-2012 | 15

Street Broad Scientific with ApaI and blunted. The cut pGint and pRint were dephosphorylated with Calf Intestinal Phosphatase (Promega) to prevent vector-vector ligation. Finally, each intronic HuR-binding sequence was ligated into both pGint and pRint via blunt-ended ligation with T7 DNA ligase (NEB). The resulting final plasmid construct is depicted in Figure 1.

Figure 1. Schematic plasmid map of pGint and pRint. Briefly, DNA sequences flanking the putative HuR binding site in each gene were amplified via qPCR, blunted with DNA Pol I Klenow Fragment, and ligated into the pGint vector. pGint contains a Kanamycin . pGint contains a kanamycin resistance gene, while pRint contains an ampicillin resistance gene (Bonano et al., 2007).

Metabolic RNA Labeling 4-thio-uridine (4sU) was used as to label RNA in a premRNA processing assay, as described previously (Rabani et al., 2011). Briefly, HEK293 cells under knockdown or mock knockdown conditions were exposed to growth medium with 0.2 µm 4sU (Sigma) for a period of 30 minutes.

RNA was extracted with TRIzol reagent and reacted with EZ-link HPDP-Biotin (Thermo Scientific) to bind to the sulfhydryl group present on the 4sU. The RNA was then tumbled for 30 minutes with streptavidin magnetic Dynabeads (Invitrogen). Biotinylated RNA was then magnetically precipitated and the biotin linkage cleaved with 50 mM DTT. The resulting supernatant (non-biotinylated) and precipitate (biotinylated) fractions were analyzed via qPCR and compared to the total RNA (before separation). Because the labeled RNA was necessarily synthesized within the 30 minute labeling period, the ratio of labeled transcript to total transcript was taken as a relative synthesis rate. When applied to the mature pre-mRNA primers, this represents a relative processing rate of the HuR-bound intron.

Results Pre-mRNA Species are Highly Enriched in HuR RIP, but Mature Forms of Intronic-Only Targets are not 16 | 2011-2012 | Volume 1

Research To determine if HuR binds pre-mRNAs in the cell, we performed RIP on HEK293 cells, and measured enrichment of bound targets using RT-qPCR. We used Western Blotting to confirm enriched pulldown of HuR in the IP. As shown in Figure 2, we achieved significant HuR enrichment.

Figure 2. Western Blot on immunoprecipitates (IPs). Lanes in order from left to right: 1) Whole-cell (WC) lysate with Normal Mouse Serum (NMS); 2) WC with 3A2 anti-HuR antibody; 3) Nuclear lysate with NMS; 4) Nuclear lysate with 3A2. The presence of HuR bands (indicated in figure) in the two 3A2 IPs and simultaneous absence from NMS IPs indicates successful enrichment of HuR.

As shown in Figure 3, we found fold enrichments > 300 for the putatively-bound pre-NFATC3 transcript, as compared to <5-fold enrichments for the established non-Hu targets of Histone H1A, Histone H4B, and β2M. The enrichments of pre-NFATC3 are comparable to the established positive control transcripts of β-actin, HuR, and MDM2, demonstrating significant HuR targeting of NFATC3 pre-mRNA. Therefore, the mRNA enrichments we demonstrate are likely due to HuR association. As shown in Figure 4, the mature, spliced form of CTCF and NFATC3 transcripts did not demonstrate enrichment in HuR RIP distinct from the negative controls, indicating that these mature transcripts are not HuR targets. PCR primers for mature transcripts are often designed to span an intron so as to only amplify mature levels; the mature primers used here were more deliberately designed to span putatively HuRbound introns, thus ensuring that the transcript which they amplify do not contain the putative binding site. This suggests that the presence of the bound intron is necessary for HuR binding, and thus that HuR can preferentially target pre-mRNA. HuR Binding has a Functional Effect on Pre-mRNA Abundance HuR binding to the pre-mRNA form of NFATC3 exclusively suggests that HuR affects the transcript in a functional manner. To determine if HuR binding of premRNA had a legitimate functional effect on pre-mRNA abundance, we performed siRNA knockdown of HuR in HEK293 cells and measured NFATC3 and CTCF

Research Figure 3

Figure 5

Fold Enrichment in RIP-qPCR

Remaining Transcript after Knockdown

100%

H

pr eN

B

Transcript

F TC

uR

4B H

1A H

0%

B

H

uR

ac tin M pr D M e2 N FA TC pr e- 3 C TC F B 2M

0

50%

pr eC

200

3

400

TC

600

FA

800

% Transcript remaining

Whole Cell Nuclear

ac tin

1000

Fold Enrichment

Street Broad Scientific

Transcript Figure 5: Differences in un-spliced transcript abundance between HuR

Figure 3. Enrichment in RNA immunoprecipitation for probed mRNA transcripts. Fold enrichment of each target transcript is shown. HuR through MDM2 are positive controls; B2M through H4B are negative controls. PreNFATC3 showed enrichment comparable to the positive controls and so is a HuR target. Pre-CTCF is not highly enriched and so is not a HuR target.

knockdown and mock knockdown.

Fraction of transcript transcript remaining after Figure 5. Differences in un-spliced abundance HuR knockdown is shown. Bars depict mean +/- standard error; decrease in between HuR and mock HuR level servesknockdown as knockdown confirmation whileknockdown.

Bactin is a positiveFracnegatively affected

by HuR knockdown.

Pre-NFATC3 shows tion control of which istranscript remaining after HuR knockdown fold change distinct from 1, so is functionally affected by HuR. Pre-CTCF is shown.

Bars depict mean +/standard error; decrease in has a fold change not distinct from 1, so is not functionally affected by HuR. HuR level serves as knockdown confirmation while Bactin is a positive control which is negatively affected by HuR knockdown. Pre-NFATC3 shows fold change distinct from 1, so is functionally affected by HuR. Pre-CTCF has a fold change not distinct from 1, so is not functionally affected by HuR.

Figure 4

Figure 5

Fold Enrichment in RIP-qPCR (mature mRNA)

Rem ainin g M atur e T r ansc rip t after K n ock dow n Legend Nuclear

Figure 4. Enrichment in RNA immunoprecipitation of mature transcripts. Fold enrichment of each target transcript is shown.

HuR through MDM2 are positive controls; B2M through H4B are negative controls. Neither of mature NFATC3 or mature CTCF showed significant enrichment, and so neither are HuR targets.

TC C

TC FA

F

3

0

N

4B H

1A H

m

B

H

uR

ac tin at ur MD M e 2 N F m A at TC ur 3 e C TC F B 2M

0

tin

200

100

B

400

uR

600

200

H

800

% Tr anscr ipt rem aining

Whole Cell Legend

ac

1000

Tr anscr ipt

Figure 6. Differences spliced transcript Figure 6: Differences in spliced in transcript abundance betweenabundance HuR knockdown and mock knockdown.

Fraction of transcript remaining after HuR knockdown between HuR knockdown and mock knockdown.

Frac- is shown.

Barstranscript depict mean remaining +/- standard error; servesknockdown as knockdown is tion of afterHuR HuR confirmation while Bactin is a positive control. Neither of NFATC3 or CTCF show shown. Bars depict meanlevels. +/- standard error; HuR serves significant decrease in transcript as knockdown confirmation while Bactin is a positive control. Neither of NFATC3 or CTCF show significant decrease in transcript levels.

Volume 1 | 2011-2012 | 17

Street Broad Scientific transcript levels. As shown in Figure 4, HuR was knocked down 70% at the transcript level. This was also accompanied by significant decreases in transcript abundance for the putativeintronic-only targets of pre-NFATC3. Pre-CTCF remained at approximately control levels, while β-actin, the positive control, decreased in abundance. Because the fraction of remaining transcript after knockdown is different from 1 by more than 1 standard error, we can say that these results are legitimate and indicative of a functional effect of HuR binding on the transcript abundance of these target pre-mRNA. As mentioned, HuR regulates many steps of mRNA processing, including binding to mature transcripts. To determine if HuR could also bind mature versions of the pre-mRNA targets, we also measured mature mRNA abundance of putative intronic targets in the knockdown. Unlike the pre-mRNA, the fold change in transcript abundance for these targets was not distinct from 1, as shown in Figure 5. While the sample size used is 3, the data as presented do not allow us to reject the null hypothesis that HuR has no effect on mature transcript levels.

Discussion We demonstrate HuR targeting of exclusively intronic sequence elements in pre-mRNA in human cells. We also show that HuR has a functional effect on stability of targets with exclusively intronic-only binding sites through the analysis of representative transcripts.

The PAR-CLIP study predicted intronic-only binding sites.

A goal of the present study was to investigate these intronic sites; however, until further validation of the relatively-new PAR-CLIP technique is completed, it was difficult to use these data as confirmation of HuR binding to pre-mRNA. As such, it was necessary to confirm that functional intronic binding sites of HuR exist through RIP-PCR. Fold enrichments of pre-NFATC3 in the RIP were comparable to those of β-actin, considered to be a very strong HuR target, thus we can further conclude that HuR can bind NFATC3 pre-mRNA stably. Pre-CTCF a predicted HuR target by the PAR-CLIP, was not enriched in the RIP, confirming the possibility of noise in that target prediction technique and highlighting the importance of confirmation of PAR-CLIP results.

Mature mRNAs lacking the introns for which the only binding sites were predicted were not enriched in the RIP. The fact that the primers used to identify mature species of NFATC3 and CTCF were targeted to span a putatively-bound intron suggests that there is no interaction between the mature transcript and HuR, and furthermore, that the interaction observed with the pre-mRNA is because of sequence elements in the putativelybound intron. This provides strong evidence for HuR pre-mRNA targeting of intronic cis-elements by a process which is independent of mature targeting.

To determine if HuR pre-mRNA binding has functional biological relevance, the effect of HuR knockdown on pre-mRNA levels was further assayed by quantitative PCR. HuR knockdown was accompanied by significant fold changes (different from 1) in transcript abundance of pre-NFATC3, indicating that HuR affects stability at the level of pre-mRNA via exclusive binding to intronic sequences.

18 | 2011-2012 | Volume 1

Research We did not demonstrate a fold change of mature transcript upon knockdown which was distinct from 1, which suggests that HuR has no effect on these transcripts.

However, because pre-mRNA are generally associated with higher turnover in the cell (Zeisel et al., 2011), it is possible that the small fold change in mature levels can be attributed to the relatively lower turnover rates of a reservoir of highly-abundant mature transcripts existing in the cell throughout the knockdown period. We must perform more knockdown procedures to confirm these results. If our results are repeatable, these possibilities can be validated through direct measurement of turnover rates in a 4-thiouridine incorporation assay (see “Future Work” section).

Pre-mRNA abundance is governed by the efficiencies of processes such as transcription, splicing, and degradation. The observed effect on abundance would then be expected to modulate the effective rates of one or more of these processes. While a known function of HuR is cytoplasmic stability through direct inhibition of RNA degradation, it seems unlikely that this established cytoplasmic function extends to the nuclear world as well, given that splicing-independent degradation likely contributes very little to total rates of pre-mRNA depletion (Zeisel et al., 2011). However, it is possible that if HuR plays a major role in stabilizing pre-mRNA independently of splicing, the negligible contribution observed by Zeisel et al. may be because of this stabilization effect, and that HuR depletion causes pre-mRNA degradation to be non-trivial.

Thus, it is possible that HuR has an effect on direct premRNA stabilization.

Apart from pre-mRNA degradation, a significant effect such as the one shown here may be due to HuR-related interactions during the process of splicing, involving either the process of splicing itself or mis-splicing-mediated decay. HuR has been implicated in various alternative exon-inclusion events, in which it either blocks or recruits components of the splicing machinery (Wang et al., 2010; Izquierdo, 2010). Interestingly, the putatively-bound introns surveyed here were constitutively spliced, that is, all known isoforms of the transcripts coded by the gene contained the flanking exons, meaning that alternative inclusion is not an issue. The fact that intronic binding sites in constitutively-spliced introns can be confirmed and functionally proven suggests a broader function of HuR than even alternative splicing (Zhang et al., 1998). Thus, the role of HuR in splicing-related processes remains to be determined. Presumably, since increased levels of HuR result in increased levels of pre-mRNA present, HuR is important for proper splicing of the target, possibly in the role of a splicing factor.

RNA surveillance can provide an explanation for the pre-mRNA abundance response of NFATC3 to HuR. Aberrant splicing can result in degradation pathways linked to the RNA surveillance machinery, which prevents error-containing mRNA from translation (Lareau et al., 2007; Zhang et al., 1998; Milligan et al., 2005). If HuR is in fact a splicing factor, these RNA surveillance

Research pathways may provide a mechanism for the observed decrease in transcript abundance of pre-mRNAs following HuR knockdown. This possibility is currently being evaluated in the in vivo splicing assay and the 4-thiouridine depletion assay.

The Exon-Junction Complex (EJC) provides one potential means by which pre-mRNA interactions can affect mature mRNA interactions. The EJC is a multisubunit complex deposited at exon junctions during splicing which is thought to be involved in downstream processing events such as nuclear export and ultimately translation (Tange et al., 2004). EJC-associated factors have also been shown to induce RNA-surveillance-mediated decay (Wagner and Lykke-Andersen, 2002). Because HuR seems to be bound to mRNAs which have only intronic cis-elements throughout the processing steps, even including mature processing, it is possible that HuR could be involved in EJC deposition or EJCmediated interactions.

In summary, we have demonstrated functional, preferential targeting of unspliced NFATC3 mRNA by the RNA-binding protein HuR. Our results provide insight into the nuclear function of HuR, which is predominantly nuclear, and provide a new avenue of study in Future Work We demonstrate a functional targeting effect of the RNA-binding protein HuR on un-spliced pre-mRNA, which suggests that HuR regulates processing steps in pre-mRNA. Our results corroborate existing literature and provide further insight into the nature of HuR-premRNA interactions.

Functional and Mechanistic Determination While our results demonstrate the importance of HuR during processes before and concurrent with splicing, the exact role of HuR in splicing itself must be confirmed. We propose to accomplish this through a parallel 4-thiouridine incorporation assay and GFP splicing reporter assay.

Future studies would also investigate possible mechanisms for HuR-mediated modulation of splicing.

Binding assays such as co-immunoprecipitation and gel mobility shift would allow for determining interaction of HuR with specific components of the splicing machinery, including the Exon Junction Complex and the spliceosome.

Functional Analysis of Target Transcripts

We have implemented a system for confirming the targets predicted by PAR-CLIP. Once more such intronic targets are confirmed through both RIP and HuR knockdown, comprehensive analysis of these targets genome-wide would provide insight into HuR’s functional role within the context of normal cell function as well. For example, characterization of environmental effects such as oxidative stress on HuR intronic targeting

Street Broad Scientific would allow us to accumulate lists of similarly-regulated intronic target transcripts, revealing functional similarities among the target genes. A comparison of function between intronic and 3’UTR targets could expand the post-transcriptional RNA operon hypothesis to encompass coordinated regulation of pre-mRNA by RBPs; it is possible that splicing targets of HuR define a nuclear RNA operon. Furthermore, intronic targets of HuR may provide additional insight into the role of HuR in disease.

References Altschul, S. F., Gish, W., Miller, W., Myers, E. W., and Lipman, D. J. (1990). Basic local alignment search tool. Journal of molecular biology 215, 403–410. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/2231712 [Accessed August 1, 2011].

Bio-Rad Laboratories (2000). iScript TM cDNA Synthesis Kit. 1–2.

Bolognani, F., Gallani, A.-I., Sokol, L., Baskin, D. S., and MeisnerKober, N. (2011). mRNA stability alterations mediated by HuR are necessary to sustain the fast growth of glioma cells. Journal of neuro-oncology. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/21935689 [Accessed September 25, 2011].

Bonano, V. I., Oltean, S., and Garcia-Blanco, M. a (2007). A protocol for imaging alternative splicing regulation in vivo using fluorescence reporters in transgenic mice. Nature protocols 2, 2166–2181.

Available at: http://www.ncbi.nlm.nih.gov/pubmed/17853873 [Accessed September 6, 2011].

Brennan, C. M., and Steitz, J. a (2001). HuR and mRNA stability.

Cellular and molecular life sciences : CMLS 58, 266–277. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11289308.

Denkert, C., Koch, I., von Keyserlingk, N., Noske, A., Niesporek, S., Dietel, M., and Weichert, W. (2006). Expression of the ELAVlike protein HuR in human colon cancer: association with tumor stage and cyclooxygenase-2. Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc 19, 1261–1269. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/16799479 [Accessed September 5, 2011].

Dixon, D. A., Tolley, N. D., King, P. H., Nabors, L. B., Mcintyre, T. M., Zimmerman, G. A., and Prescott, S. M. (2001). Altered expression of the mRNA stability factor HuR promotes cyclooxygenase-2 expression in colon cancer cells. Cell 108, 1657–1665.

Fan, X. C., and Steitz, J. a (1998). Overexpression of HuR, a nuclear-cytoplasmic shuttling protein, increases the in vivo stability of ARE-containing mRNAs. The EMBO journal 17, 3448–3460.

Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi? artid=1170681&tool=pmcentrez&rendertype=abstract.

Hafner, M., Landthaler, M., Burger, L., Khorshid, M., Hausser, J., Berninger, P., Rothballer, A., Ascano, M., Jungkamp, A.-C., Munschauer, M., et al. (2010). Transcriptome-wide identification of RNA-binding protein and microRNA target sites by PAR-CLIP. Cell 141, 129–141. Available at: http://www.pubmedcentral.nih. gov/articlerender.fcgi?artid=2861495&tool=pmcentrez&rendertyp e=abstract [Accessed June 17, 2011].

Hinman, M., and Lou, H. (2008). Diverse molecular functions of Hu proteins. Cellular and molecular life sciences 65, 3168–3181. Available at: http://www.springerlink.com/ index/54H3W5358HQ14706.pdf [Accessed June 22, 2011].

Volume 1 | 2011-2012 | 19

Street Broad Scientific Invitrogen Life Science (2006). Lipofectamine TM 2000. Available at: http://tools.invitrogen.com/content/sfs/manuals/lipofectamine2000_man. pdf.

Invitrogen Life Science (2010). TRIzol ® Reagent. Available at: http:// tools.invitrogen.com/content/sfs/manuals/trizol_reagent.pdf.

Izquierdo, J. M. (2010). Cell-specific regulation of Fas exon 6 splicing mediated by Hu antigen R. Biochemical and biophysical research communications 402, 324–328. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/20951677 [Accessed June 22, 2011].

Keene, J. D. (2007). RNA regulons: coordination of post-transcriptional events. Nature reviews. Genetics 8, 533–543. Available at: http://www.ncbi. nlm.nih.gov/pubmed/17572691 [Accessed July 15, 2010].

Keene, J. D., Komisarow, J. M., and Friedersdorf, M. B. (2006). RIP-Chip: the isolation and identification of mRNAs, microRNAs and protein components of ribonucleoprotein complexes from cell extracts. Nature protocols 1, 302–307. Available at: http://www.ncbi.nlm.nih.gov/pubmed/17406249 [Accessed July 17, 2011].

Keene, J. D., and Tenenbaum, S. a (2002). Eukaryotic mRNPs may represent posttranscriptional operons. Molecular cell 9, 1161–1167. Available at: http://www.ncbi.nlm.nih.gov/pubmed/12086614.

Kishore, S., Luber, S., and Zavolan, M. (2010). Deciphering the role of RNA-binding proteins in the post-transcriptional control of gene expression. Briefings in functional genomics 9, 391–404. Available at: http://www. pubmedcentral.nih.gov/articlerender.fcgi?artid=3080770&tool=pmcentrez &rendertype=abstract [Accessed June 22, 2011].

Lareau, L., Brooks, A., and Soergel, D. (2007). The coupling of alternative splicing and nonsense-mediated mRNA decay. In Alternative Splicing in the Postgenomic Era, pp. 190–211. Available at: http://citeseerx.ist.psu. edu/viewdoc/download?doi=10.1.1.67.3130&amp;rep=rep1&amp;type=p df [Accessed July 5, 2011].

Lebedeva, S., Jens, M., Theil, K., Schwanhäusser, B., Selbach, M., Landthaler, M., and Rajewsky, N. (2011). Transcriptome-wide Analysis of Regulatory Interactions of the RNA-Binding Protein HuR. Molecular cell, 1–13.

Available at: http://www.ncbi.nlm.nih.gov/pubmed/21723171 [Accessed July 5, 2011].

Livak, K. J., and Schmittgen, T. D. (2001). Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods (San Diego, Calif.) 25, 402–408. Available at: http:// www.ncbi.nlm.nih.gov/pubmed/11846609 [Accessed July 15, 2011].

Milligan, L., Torchet, C., Allmang, C., Shipman, T., and Tollervey, D. (2005).

A nuclear surveillance pathway for mRNAs with defective polyadenylation.

Molecular and cellular biology 25, 9996–10004. Available at: http://mcb. asm.org/cgi/content/abstract/25/22/9996 [Accessed July 15, 2011].

Mukherjee, N., Corcoran, D. L., Nusbaum, J. D., Reid, D. W., Georgiev, S., Hafner, M., Ascano, M., Tuschl, T., Ohler, U., and Keene, J. D. (2011).

Integrative Regulatory Mapping Indicates that the RNA-Binding Protein HuR Couples Pre-mRNA Processing and mRNA Stability. Molecular Cell 43, 1–13. Available at: http://dx.doi.org/10.1016/j.molcel.2011.06.007.

Rabani, M., Levin, J. Z., Fan, L., Adiconis, X., Raychowdhury, R., Garber, M., Gnirke, A., Nusbaum, C., Hacohen, N., Friedman, N., et al. (2011).

Metabolic labeling of RNA uncovers principles of RNA production and degradation dynamics in mammalian cells. Nature biotechnology 29, 436– 442. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?a rtid=3114636&tool=pmcentrez&rendertype=abstract [Accessed July 15, 2011].

Rozen, S., and Skaletsky, H. (2000). Primer3 on the WWW for General Users and for Biologist Programmers. Methods in Molecular Biology 132,

20 | 2011-2012 | Volume 1

Research 365–386.

Saunders, L. R., and Barber, G. N. (2003). The dsRNA binding protein family: critical roles, diverse cellular functions. The FASEB journal : official publication of the Federation of American Societies for Experimental Biology 17, 961–983. Available at: http://www. ncbi.nlm.nih.gov/pubmed/12773480.

Tange, T. Ø., Nott, A., and Moore, M. J. (2004). The ever-increasing complexities of the exon junction complex. Current opinion in cell biology 16, 279–284. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/15145352.

Tenenbaum, S. a, Lager, P. J., Carson, C. C., and Keene, J. D. (2002).

Ribonomics: identifying mRNA subsets in mRNP complexes using antibodies to RNA-binding proteins and genomic arrays. Methods (San Diego, Calif.) 26, 191–198. Available at: http://www.ncbi.nlm. nih.gov/pubmed/12054896.

Wagner, E., and Lykke-Andersen, J. (2002). mRNA surveillance: the perfect persist. Journal of cell science 115, 3033–3038. Available at: http://www.ncbi.nlm.nih.gov/pubmed/12118059.

Wang, H., Molfenter, J., Zhu, H., and Lou, H. (2010). Promotion of exon 6 inclusion in HuD pre-mRNA by Hu protein family members. Nucleic acids research 38, 3760–3770. Available at: http:// www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2887941&to ol=pmcentrez&rendertype=abstract [Accessed June 22, 2011].

Zeisel, A., Köstler, W. J., Molotski, N., Tsai, J. M., Krauthgamer, R., Jacob-Hirsch, J., Rechavi, G., Soen, Y., Jung, S., Yarden, Y., et al. (2011). Coupled pre-mRNA and mRNA dynamics unveil operational strategies underlying transcriptional responses to stimuli.

Molecular Systems Biology 7. Available at: http://www.nature.com/ doifinder/10.1038/msb.2011.62 [Accessed September 14, 2011].

Zhang, J., Sun, X., Qian, Y., and Maquat, L. E. (1998). Intron function in the nonsense-mediated decay of beta-globin mRNA: indications that pre-mRNA splicing in the nucleus can influence mRNA translation in the cytoplasm. RNA (New York, N.Y.) 4, 801–815.

Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi? artid=1369660&tool=pmcentrez&rendertype=abstract.

Zhu, H., Hasman, R. A., Barron, V. A., Luo, G., and Lou, H. (2006). A nuclear function of Hu proteins as neuron-specific alternative RNA processing regulators. Molecular biology of the cell 17, 5105. Available at: http://www.molbiolcell.org/cgi/content/abstract/17/12/5105 [Accessed June 22, 2011].

Review

Street Broad Scientific

The Regulation of Centromere Function and Localization Alyssa Ferris

Summary Centromere function is essential for proper cell division, but the mechanism for determining the formation and location of the centromere is not well understood. Centromeres are a paradox because, while their basic function is very similar among eukaryotes, centromeric DNA sequences are significantly different, even between closely related species. (Lamb and James, 2010). This fact suggests that other forms of regulation are responsible for the location of the centromere, such as epigenetic regulation. This is one possible explanation for how centromere location can stay in the same location on a chromosme and how centromere function is independent of centromeric DNA; however, no such factor has been identified yet.

Background Information A centromere is a visible constriction of DNA located on a condensed chromosome, and where the kinetochore forms during cell division. The kinetochore forms on top of the centromere, and is composed of centromeric proteins (CENP) and other complexes (see figure 1). The most important protein is CENP-A, which serves as a base for the rest of the kinetochore and has been shown to influence where it forms (Buscaino et al., 2010).

When a cell begins to divide, spindle fibers attach to either side of the kinetochore and separate the sister chromatids. Thus, the centromere ensures that the chromosomes are divided correctly among the resulting cells.

Failure results in an uneven division of genetic material which usually results in cell death. The centromeric DNA is typically composed of repeating satellite sequences; each individual sequence is usually between 150 and 200bp (Henikoff et al., 2001).

Figure 1. http://www.edoc.hu-berlin.dae A diagram of a chromosome showing the centromere and the location of the kinetochore with microtubules attached.

In order for a cell to divide correctly each chromosome needs to have one centromere and cellular mechanisms ensure that its location is static. However, in some cases, the centromere forms in a different location on the chromosome for unknown reasons. These neocentromeres have the same general properties as a normal centromere, including a noticeable constriction of the chromosome and functioning kinetochore proteins, which serve as binding sites for microtubules during mitosis and meiosis (Buscaino et al., 2010).

The Role of DNA Sequences Originally, DNA sequence were believed to determine the centromereâ&#x20AC;&#x2122;s location (Karpen and Allshire, 1997). Research in S. cerevisiae showed that the centromere consisted of three distinct satellite sequences which had unique motifs that boundd to centromeric proteins (Karpen and Allshire, 1997). However, research conducted in S. pombe indicated that the centromeric satellite repeats from S. cerevisiae were not conserved (Karpen and Allshire, 1997).

More importantly, satellite sequences could be added or deleted without affecting the stability of the centromere (Karpen and Allshire, 1997). Later studies in Equus showed that evolutionarily new centromeres lacked satellite repeats and also identified locations with the past centromeric identity that still had satellite sequences (Piras et al., 2010). These findings imply that satellite sequences are a part of centromeric identity, but they are not the initial factors that determine centromere identity.

Although, specific sequences in the centromere are not highly conserved, satellite sequences have been found in the DNA sequences of almost all centromeres. For example, human centromeres mainly consist of satellite sequences called alpha satellites, which consist of 178bp repeats (Lee et al., 1997). These repeats constitute approximately 62% of the genetic material surrounding the centromere, and an additional 24% consists of other satelVolume 1 | 2011-2012 | 21

Street Broad Scientific

Review

Figure 2. (Allshire and Karpen, 2008)a) In this model, CENP-A is maintained on the DNA by a chromatin-loading factor. There are two main theories about how this would work, either gaps would remain in the chromatin during replication and CENP-A would be inserted directly into the DNA or the old CENP-A locations would be temporarily replaced by H3 and then restored by a CENP-A-H3 exchange factor later in the cell cycle. b) In this process, centromeric chromatin is modified to include histone H3 which is tagged with specific epigenetic histone modifications, such as dimethylation of lysine 4, and these modifications would specify recruitment of CENP-A. c) In this method, preexisting CENP-A splits during DNA replications leading to the recruitment of additional CENP-A in order to restore chromatin integrity.

lite repeats (Lamb and Birchler, 2010). Studies have also shown that the presence of alpha DNA repeats is positively correlated with centromere stability. Also, if alpha satellites are inserted into ectopic DNA, noncentromeric chromosome sites, an artificial centromere will form.

However, a centromere will not form if neocentromeric DNA is inserted into a chromosome (Lamb and Birchler, 2003). Sequencing data from other species confirms this trend. For example, Arabidopsis centromeres contain satellite sequences and other repeats similar to those found in humans and which are in the same general locations on the chromosome. These satellite sequences are 178bp long and resemble alpha satellites; however, they consist of entirely different DNA sequences. The only commonalities between satellite sequences are their approximate length, between 150bp and 200bp, and the fact that the sequences tend to be rich in adenine and thiamine (Karpen and Allshire, 2003). Additionally, although satellite sequences are often found at the centromere, they can also be present in ectopic DNA (Allshire and Karpen, 2008). Therefore, satellite sequences are neither necessary nor sufficient for centromere formation, and there must be an additional mechanism influencing centromere location.

The Role of Proteins Centromeric proteins are important because, unlike satellite sequences, they are found exclusively at the centromere, and many are only found there during mitosis when the kinetochore forms (Henikoff et al., 2001). CENPA, C and E are especially important because they have been found at active centromeres and neocentromeres but 22 | 2011-2012 | Volume 1

not at old or nonfunctioning centromeres (Henikoff et al., 2001). CENP-A has been shown in previous studies to be the most likely protein to maintain the centromere. Its absence causes most kinetochore proteins to be misplaced, but the absence of other centromere proteins does not affect the deposition of CENP-A. In addition, the overexpression of CENP-A results in its random integration into DNA which causes the formation of ectopic centromeres (Allshire and Karpen, 2008). CENP-A has been found to mark the centromere location by replacing the H3 protein in the histones of a centromeric nucleosome (Henikoff et al., 2001). It has also been found to act as a heritable molecule located on the chromosome during DNA replication, which could allow it to regulate centromere location over multiple generations of cells (Allshire and Karpen, 2008).

Although CENP-A plays an essential part in the formation of the kinetochore, another mechanism must be responsible for its original recruitment and incorporation into centromeric DNA. Allshire and Karpen (2008) proposed three different models for the recruitment of CENP-A. One model is that during DNA replication, a chromatin-loading factor that recognizes CENP-A could deposit it into an identical location in the new DNA strand. This would explain why centromeres are at a single site and why neocentromere locations are maintained once they have been established.

Another possible system proposes that modifications in H3 recruit specific CENP-A assembly proteins, which are responsible for recruiting the proteins that form the kinetochore. This system would also require the presence

Review of an enzyme or binding proteins in the cell which are able to modify the H3. A third model hypothesizes that the intrinsic properties of the CENP-A nucleosomes provide the signals for CENP-A recruitment. For example, the octamer structure could split into two tetramers, and recruitment of CENP-A would be a reconstitution of the tetramers. This model would require proteins to facilitate the division of CENP-A and other proteins to recruit CENP-A when it is reconstituted. Although one of these three models may be responsible for CENP-A deposition in the centromere, none has significant experimental support.

Abnormal Centromeres Some studies have shown that neocentromeres are more likely to form in close proximity to the current centromere, suggesting that either the centromere influences the surrounding DNA or that latent chromosomes maintain some remnants of centromeric identity (Karpen and Allshire, 1997). In Drosophilia, studies have shown that acentric chromosome fragments can gain centromeric function. These neocentromeres are usually formed at the tip of the X chromosome where the chromosome splits, which is approximately 40Mb away from the normal centromere (Karpen and Allshire, 1997). Similar studies in human cells have also identified three locations in the karyotype where neocentromeres are most likely to form, indicating that the location for neocentromere formation is nonrandom (O’Neill and Carone, 2009)(Ketel et al., 2009).

Studies have shown that in some organisms, such as nematodes, centromeres are holocentric, the microtubules attach along whole chromosomes (Karpen and Allshire, 1997). It is possible that in the past most organisms had holocentric centromeres, but they evolved into a single, larger centromere. If the DNA from the holocentric microtubule binding sites had maintained some of its previous identity, neocentromeres would result when factors in the cell, such as CENP-A, had been incorrectly recruited to these sequences. Carone et al. (2008) found that in Macropus eugenii, a retrovirus called KERV was located near transcription sites in the centromere and that it promoted transcription of centromeric DNA. The study also showed that the ncRNA was used to recruit proteins to the centromere to form the kinetochore. Other studies have found that previous centromere locations are located next to KERV sequences (O’Neill and Carone, 2009), and evidence was found that higher order DNA-protein structures suppress alternate centromere locations when a functioning centromere is present (Karpen and Allshire, 1997). It is possible that these retroviral promoters are located by holocentric chromosome locations when they are activated. As a result, the production of a ncRNA could cause recruitment of the structures necessary to form a functioning kinetochore and hence a neocentromere.

Street Broad Scientific There is no one explanation for the stability of centromere location and kinetochore formation. Centromeric DNA sequences are not highly conserved, but satellite sequences that are approximately the same length occur in most eukaryotes. CENP-A has been found to recruit other proteins and complexes to the centromere in order to form the kinetochore; however, no distinct regulation of the production or deposition of CENP-A has been identified. Additionally, neocentromeres form in ectopic DNA, proving that satellite sequences and centromeric DNA is not necessary for centromeric function. However, neocentromeres tend to form at the same locations on chromosomes, indicating that some factor is causing centromere localization.

References Allshire, R. and G. Karpen. 2008. Epigenetic regulation of centromeric chromatin: old dogs, new tricks? Nature Reviews 9:923-937.

Bergmann, H. J., M. G. Rodríguez, N. M. C. Martins, H. Kimura, D. A. Kelly, H. Masumoto, V. Larionov, L. E. T. Jansen, and W. C. Earnshaw.

2011. Epigenetic engineering shows H3K4me2 is required for HJURP targeting and CENP-A assembly on a synthetic human kinetochore. The EMBO Journal 30: 328 – 340.

Buscaino, A., R. Allshire, and A. Pidoux. 2010. Building centromeres: home sweet home or a nomadic existence? Current Opinion in Genetics and Development 20:118-126.

Carone, D., M. Longo, G. Ferreri, H. Hall, M. Harris, N. Shook, K. Bulazel, B. Carone, C. Obergfell, M. O’Neill, and R. O’Neill. 2008. A new class of retroviral and satellite encoded small RNAs emanates from mammalian centromeres. Cromosoma 118:113-125.

Piras F., S. Nergadze, E. Magnani, L. Bertoni, C. Attolini, L. Khoriauli, E. Raimondi, and E. Giulotto. 2010. Uncoupling of satellite DNA and centromeric function in the genus Equus. PLoS Genetics 6:1-10.

Karberg M., R. J. Leavitt, D. A. A. Cabaya, M. E. Van Eden, and X. Y. Jia. 2009. A Method for Quantifying DNA Methylation Percentage Without Chemical Modification. Zymo Research Corporation.

Karpen, G. and R. Allshire. 1997. The case for epigenetic effects on centromere identity and function. Trends in Genetics 13: 489-96.

Henikoff, S., K. Ahmad, and H. Malik. 2001. Centromere paradox: stable inheritance with rapidly evolving DNA. Science 293: 1098-102.

Ketel, C. H. Wang, M. McClellan, K. Bouchonville, A. Selmecki, T. Lahav, M. Gerami-Nejad, and J. Berman. 2009. Neocentromeres form efficiently at multiple possible loci in Candida albicans. PLoS Genetics 5: 1–18.

Lamb, J. and J. Birchler. 2003. The role of DNA sequence in centromere formation. Genome Biology 4:1-4.

Lee, C., R. Weverick, R. Fisher, M. Ferguson-Smith, and C. Lin. 1997.

Human centromeric DNAs. Human Genetics 100: 291-304.

O’Neill. R. and D. Carone. 2009. The role of ncRNA in centromeres: a lesson from marsupials. Progress in Molecular and Subcellular Biology 48: 77-101.

Stimpson, K. M, Sullivan B. A. 2010. Epigenomics of Q1centromere assembly and function. Current Opinion in Cell Biology 22: 1-9.

Sullivan, B. A. and G. H. Karpen. 2004. Centromeric chromatin exhibits a histone modification pattern that is distinct from both euchromatin and heterochromatin. Nature Structural & Molecular Biology 11: 1076 – 1083.

Conclusion Volume 1 | 2011-2012 | 23

Street Broad Scientific

Research

Chomp the Graph Sam Magura, Vitchyr Pong, Elliot Cartee, Kevin Valakuzhy Introduction Chomp the Graph is a two player game played on a finite graph G=(V,E). On a player’s turn, they remove part of the graph, either removing a vertex and all edges incident to it or just a single edge; such a removal is called a move. The player who cannot make any more moves (i.e. is faced with an empty graph) loses.

Figure 1. An example progression of Chomp the Graph. The initial graph is the leftmost. The player that makes a move on the rightmost graph, which contains a single vertex, wins the game.

We say that a graph is an N-position the player who plays next will win if using optimal strategy and a P-position otherwise. Our goal is to determine whether a given graph is an N-position or P-position. From these definitions, we see that, from an N-position, there must be a move that will send the game to a P-position; otherwise, the first position would not be an N-position. Likewise, from a P-position, there is no move that will send the game to another P-position.

Background Chomp the Graph is an impartial game; that is, the available moves depend only on the state of the game, the same moves are available for each player, and the payoffs for winning the game are the same for each player. In other words, the only difference between the two players is that one goes first while the other goes second. Additionally, Chomp the Graph is played under the normal play convention; the player who cannot move loses. Finally, Chomp the Graph is terminating because there are no infinite lines of play. Indeed, for each move, the total number of edges and vertices must always decrease by at least one, making an infinite line of moves impossible.

We can now relate the Chomp the Graph game to another game, Nim. In the game Nim, there are several heaps of “stones”. On a player’s turn, they choose a heap and remove one or more stones from it.

Figure 2. A progression of a game of Nim with one heap.

Any game of Nim, along with any other impartial game, has an associated nimber, that describe whether the po24 | 2011-2012 | Volume 1

sition in N or P. By the Sprague-Grundy theorem, any impartial game that is terminating and played under the normal play convention is equivalent to a Nim heap of a certain size. Therefore, any possible game of Chomp the Graph is equivalent to some Nim heap. Nimbers are defined recursively. Let G={G0, G1, G2, … } be a Nimequivalent game with options G0, G1, G2, … .That is, a single move can change G to G0 or G1 or G2, and so on.

Additionally, let *H denote the nimber of a game H. A Nim heap of size *H can sent to any other Nim heap with a size that is less than *H. Therefore,

*G= mex({*G0, *G1, *G2, … }) [1] Here mex(S) denotes the least nonnegative integer not contained in the set S. The mex of an empty set is 0, so a game with no available moves has a nimber of 0. Nimbers are useful because they provide an easy way to predict the outcome of games and sums of game (e.g. a game of Nim with two heaps, each with a known nimber). Nimbers can be added, though not in the same way as integers. Instead, finite nimbers are added with a nim sum as if they were normal integers using the bitwise exclusive-or operator.

EV Parity In this section, we prove a relationship between the nimber of a game of Chomp the Graph and the number of vertices and edges in the graph. The nimber of the empty graph is 0, since there are no moves that can be made on it. Now, we can recursively determine the nimbers of more complicated graphs using Equation 1.

Through casework, we discovered the following relationship that holds in bipartite graphs between the number of vertices and edges of a graph and its nimber.

A graph with an even number of edges and an even number of vertexes has a nimber of 0.

A graph with an even number of edges and an odd number of vertexes has a nimber of 1.

A graph with an odd number of edges and an even number of vertexes has a nimber of 2.

A graph with an odd number of edges and an even number of vertexes has a nimber of 3.

We call this system of using the parity of the edges and vertices to predict the nimber of a graph EV parity. The EV parity of a graph can be thought of either as a twodigit binary number (ev)2 , with e and v each representing single digits, or an ordered pair (e, v). The value of e is 1 if the graph as an odd number of edges, or 0 if it has an even number of edges. The value of v is 1 if the graph as an odd number of vertices, or 0 if it has an even number

Research of vertices. The binary representation of EV parity is the nimber of the graph.

Proof for Bipartite Graphs In this section, we prove that a bipartite graph with EV parity (e, v) is equivalent to a nimber ev2. We do this by establishing that, for a graph G with nimber *G, there exist moves to change the graph’s nimber to *G - 1, *G - 2, …

1, 0, but that there is no move that will leave the graph’s nimber at *G. Propostion 1: For any graph, there is no move that will not change the graph’s EV parity.

Proof. The player has two options: Remove a vertex, and all edges connected to it. This changes the vertex parity, and may or may not change the edge parity.

Remove an edge. This changes the edge parity.

Lemma 1: If a graph has an odd number of vertices --- that is, a vertex parity v of 1 --- then the graph contains at least one vertex of even degree.

Proof. Proceed by contradiction; suppose a graph with an odd number of vertices has only vertices of odd degree.

However, because there are an odd number of vertices, all of which have odd degree, then the sum of the degrees would be odd. This is impossible since the sum of degrees is the number of edges times two, and therefore must be even. Thus, at least one vertex has even degree.

Proposition 2: If presented with a graph of (0, 1) EV parity, there exists a single move that will change the graph’s parity to (0, 0).

Proof. To reset this position to (0, 0), we must remove a vertex of even degree. By Lemma 1, this graph --- which has a vertex parity of 1 --- has such a vertex.

Proposition 3: If presented with a graph of (1, 0) EV parity, there exists a single move that will change the graph’s parity to (0, 1) and a single move that will change the graph’s parity to (0, 0).

Proof. If the graph’s EV parity is (1, 0), there is at least one edge. Remove this edge to change the graph’s parity to (0, 0). To change the parity to (0, 1), a vertex of odd degree must be removed. This is always possible. Proceed by contradiction; suppose that in a graph of (1, 0) EV parity, all vertices are of even degree. Then the sum of degrees of all vertices in the graph is divisible by four since there are an even number of vertices in the graph each with an even degree. Because twice the number of edges is the sum of the degrees, the number of edges in this graph must also by even. This is a contradiction, so there is at least one vertex of odd degree in a graph with (1, 0) parity, and the parity can be changed to (0, 1).

Proposition 4: If presented with a bipartite graph of (1, 1) EV parity, there exist moves to change the graph to (1, 0), (0, 1), or (0, 0) parity.

Proof. To change the parity to (1, 0), a vertex of odd degree must be removed. By Lemma 1, since this graph

Street Broad Scientific has an odd number of vertices, there is at least one vertex of even degree. To change the parity to (0, 1), remove any edge. Since the edge parity of the graph is 1, there is at least one edge. To reset the parity to (0, 0), a vertex of odd degree must be removed. This is always possible. Proceed by contradiction; suppose the EV parity of a graph is (1,1) and every vertex has even degree. Define such a graph to be special. However, there are no graphs that are both bipartite and special cases, as proven below.

Lemma 3: It is impossible for a graph to be both bipartite and a special case.

Proof. Let G be bipartite and have an EV parity of (1,1).

Let its vertices be divided into sets A and B. Starting from just these two groups of vertices, we will attempt to add an odd number of edges in a way such that: Each edge is incident on one vertex in A and one vertex in B. Each vertex has even degree.

Let σ(A) be the sum of the degrees of vertices in A. Let σ(B) be the sum of the degrees of vertices in B. Before any edges are added σ(A) = 0 and σ(B) = 0. Each time an edge is added, σ(A) and σ(B) each increase by one, because a vertex of a bipartite graph can only be adjacent to vertices in the other set. So when m edges have been added, σ(A) = m and σ(B) = m.

Because the EV parity of the final graph is (1, 1), the final value of m is odd. Therefore, both the sum of the degrees of every vertex in A and the sum of the degrees of every vertex in B are odd. Because the sum of a set of even numbers cannot be odd, at least one vertex in each set must have odd degree.

Theorem 1: A bipartite graph with EV parity (e, v) has a nimber of (ev)2.

Proof. First, let us impose an order on EV parities such that if the EV parity of graph G is greater than the EV parity of graph H, then the proposed nimber of G is also greater than the proposed equivalent nimber for H. Thus, the EV parities, from least to greatest, are: (0, 0), (0, 1), (1, 0), and (1, 1).

The nimber of a game is defined as the smallest nimber not present in the set of the game’s options. Our theorem is true if and only if, for all bipartite graphs, the graph’s EV parity is the smallest EV parity not contained in the set of its options. We prove this restatement of the theorem by considering it for each of the four EV parities.

A non-empty graph of EV parity (0, 0) cannot have another graph of EV parity (0, 0) in its set of options, by Proposition 1, which states that there is no move that will leave a graph’s EV parity unchanged. Thus the smallest EV parity not contained in the set of options is (0, 0), and the theorem holds for this case.

The empty graph, which also has EV parity (0, 0), has no options. The smallest EV parity that is not in the empty set is the smallest EV parity, (0, 0). The theorem holds for this case.

Volume 1 | 2011-2012 | 25

Street Broad Scientific A graph of EV parity (0, 1) has a graph with EV parity (0, 0) in its set of options, by Proposition 2. It does not a graph of parity (0, 1) in its options by Proposition 1. Thus, the smallest EV parity not contained by its set of options is (0, 1).

A graph with EV parity (1, 0) has graphs of EV parity (0, 1) and (0, 0) in its set of options by Proposition 3.

It does not have a graph of EV parity (1, 0) in its set of options by Proposition 1, thus the smallest EV parity not contained by its set of options is (1, 0).

A bipartite graph with EV parity (1, 1) has graphs of EV parity (1, 0), (0, 1), and (0, 0) in its set of options by Proposition 4. It does not have a graph of EV parity (1, 1) in its set of options by Proposition 1, thus the smallest EV parity not contained by its set of options is (1, 1).

Thus the theorem holds for all bipartite graphs.

This theorem allows us to predict whether a given bipartite graph is an N-position or a P-position, since: If the bipartite graph has non-(0, 0) EV parity, it has a nonzero nimber. A Nim heap with a nonzero number of stones is a N-position, so bipartite graphs with non-(0, 0) EV parity are N-positions.

If the bipartite graph has (0, 0) EV parity, the graph has a nimber of 0. A Nim heap with 0 stones is a P-position, so graphs with (0, 0) EV parity are P-positions.

Odd Cycles and EV Parity Most graphs can be solved using this idea of EV parity.

However, an example of a graph that cannot be solved using EV parity is a triangle.

Research an EV parity of (1, 0). To win, the first player removes any edge from the cycle. We also investigated single odd cycle graphs where a central odd cycle has trees branching off from some or all of its vertexes. For example:

We classified these graphs based on their EV parity and the nature of the odd cycle’s vertexes.

O: Each of the cycle’s vertexes is incident to an odd number of edges E: Each of the cycle’s vertexes is incident to an even number of edges ME: One of the cycle’s vertexes is incident to an odd number of edges. The rest are incident to an even number of edges.

MO: One of the cycle’s vertexes is incident to an even number of edges. The rest are incident to an odd number of edges.

MM: The graph does not fall into any of the previously listed cases.

Through casework, we investigated these categories of graphs and created the following table. Blank spots could be either winning positions or losing positions, or are unknown. N represents a N position, and P represents a P position.

O In this case, the EV parity is (1, 1). However, this graph cannot be sent to a (0, 0) position because there is no vertex of odd degree. (Note: this does not contradict our previous application of EV parity because it is not a bipartite graph.) Even cycles are bipartite graphs, so their positions are easy to calculate. In general, all odd cycles display this property: they have EV parities of (1,1) and have no vertices of odd degree. Still, some graphs containing odd cycles still have predictable positions.

Predictable Odd Cycle Graphs Some graphs containing odd cycles can be won through a technique called “cycle killing.” We focused on cycle killing for graphs that contain a single odd-cycle. In this technique, the player attempts to make a move that both follows EV parity and removes the odd cycle. By doing this, the player changes the graph into a (0, 0) bipartite graph, a losing position for their opponent. The cycle killing strategy works for any single odd cycle graph that has 26 | 2011-2012 | Volume 1

(1,1) (1,0) (0,1) (0,0)

N N

E

MO

ME

MM

N N

N N N

N N N

N N N L

Non-Planar Graphs It is important to note that the argument for EV parity only requires the graph to be bipartite; the graph can be non-planar. Though planarity could potentially play a role in complex odd cycle graphs, it has no effect on graphs that can be predicted by EV parity.

Identical Subgraph Theorum Consider a graph G that can be categorized into three distinct subgraphs H1, H2, and H3 where H1 and H3 are isomorphic and do not share an edge on G, and H3 is equivalently connected to H1 and H2, which means that for every edge between a vertex v3 on H3 and vertex v1 on

Research H1, there is an edge between v3 on H3 and vertex v2 on H2, in which v2 is the isomorphic counterpart to v1. If a graph satisfies these criteria, it is said to be reducible.

Figure 3. A graph that is reducible. H1 and H2 are isomorphic and do not share and edge. H3 is equivalently connected to both H1 and H2.

In the example above, H1 and H2 are isomorphic: they are both have a center vertex vc that is connected to two other vertices. Also, H1 and H2 and do not share an edge.

H3 is equivalently connected to both H1 and H2. In both cases, the edge is connected to v* and vc. So, this graph is reducible.

Street Broad Scientific and that it is Player 1â&#x20AC;&#x2122;s turn. No matter what Player 1 does, Player 1 cannot avoid playing a game that is equivalent to H3. Any moves involving H1 or H2 are stalling moves that do not affect H3.

Player 1 has three stalling moves: If Player 1 removes an edge connecting H1 and H3, Player 2 can remove the equivalent edge connecting H2 and H3. This end position is also reducible, and H3 is maintained.

If Player 1 removes a vertex in H1, Player 2 can remove the equivalent vertex in H2. This end position is also reducible, and H3 is maintained.

If Player 1 removes an edge within H1, Player 2 can remove the equivalent edge within H2. This end position is also reducible, and H3 is maintained.

In each of these cases, H3 is maintained and the graph remains reducible. Since H1 and H2 are finite, there are a finite number of stalling moves Player 1 can make, so H1 and H2 will eventually be removed from the overall graph without affecting H3. If Player 1 ever plays on H3, then Player 2 plays on H3 as if H1 and H2 are not part of the game. To show why this does not affect H1 or H2, examine the moves on H3 that Player 1 can make.

If Player 1 removes an edge in H3, this will not affect H1 or H2.

If Player 1 removes a vertex v* in H3, this will not uniquely affect H1 or H2. Since this graph is reducible, v* must be connected to the same vertices in H1 and H2, so H1 and H2 are identically affected (e.g. both vc on H1 and H2 lose a degree).

Since moves involving H1 and H2 are only stalling moves, and moves involving H3 identically affect H1 and H2, the overall graph has the same position as H3. The IST is especially useful when the position of a graph is not easily predictable using EV parity or a cycling-killing strategy.

Figure 4. A graph that is not reducible given its current categorizations.

In the example above, H3 is not equivalently connected to H1 and H2. Between H1 and H3, there is an edge between v* and vc, whereas between H2 and H3, there is an edge between v* and v2, H3.

The Identical Subgraph Theorem (IST) states that reducible graphs has the same position (N or P) as the subgraph H3.

Proof: Suppose a graph G, with subgraphs H1, H2, and H3 is reducible.

Without loss of generality, suppose H3 is N-position

Figure 5. An example use of the Identical Subgraph Theorem to predict a graph.

In this example, the IST is used to simplify a graph containing two odd cycles. Since the graph on the right is a bipartite graph with an EV parity of (0, 0), we know that the original graph had a P-position.

Volume 1 | 2011-2012 | 27

Street Broad Scientific Conclusion Chomp the Graph is a terminating impartial game that adheres to normal play convention. By the SpragueGrundy theorem, Chomp must have a nimber, which determines if a position can be won if played optimally.

In bipartite graphs, casework proves that this nimber is equal to (ev)2 , where e is the number of edges and v is the number of vertices in the graph. This approach is limited by the odd cycles. A graph with only one odd cycle can be predicted by eliminating the cycle and keeping the edge and vertex parity in mind. When a graph goes beyond the scope of cycle-killing, the Identical Subgraph Theorem (IST) provides a method for simplifying the graphs that adhere to a more general standard.

28 | 2011-2012 | Volume 1

Research

Street Broad Scientific

Review

Engineering a Robust Gene Network in Pseudotyped Packaging Cells for in vivo Production of Ecotropic Protein Bound Retroviral Vectors: Prospectives and future directions from a comprehensive review of existing literature Param Sidhu ABSTRACT: An emerging question in the fields of both synthetic biology and gene therapy is the potential for use of a retroviral vector in cell targeted protein and drug delivery. The vector should be able to hold DNA for transduction, produce proteins or substances for delivery and bind those substances to its surface membrane. The goal of this review was, therefore, to determine the feasibility of introducing a robust, multifaceted gene network into pseudotyped HEK-293 packaging cells in order to produce retroviral vectors. The vectors would be assembled using packaging cell enzymatic machinery in vivo. The applications of the work include selective tumor necrosis and targeted drug delivery.

I. Overview of Retroviral Vectors Introduction to Gene Therapy Gene therapy refers to disease treatment involving the modification, insertion, or deletion of genes. Alterations to the genetic library of an organism results in modified protein production, which ultimately has physiological implications for the organism. There are four major ways in which gene therapy seeks to address diseases: insertion of a functional gene, replacement of a dysfunctional gene with a normal gene, repair of abnormal genes through selective reverse mutation, and regulation of protein production by modifying DNA (Anson).

In gene therapy, the modification or insertion of genetic information is conducted using carrier molecules known as vectors. This transport agent is typically a benign virus or bacteria that has been genetically altered to carry normal human DNA to the organism’s cells. The vector will infect or enter the cell, introducing its genetic plasmid into the genome of the cell to induce perceptible changes in protein production.

There are currently four major vectors that are used in gene therapy: herpes simplex viruses, which are cold sore causing viruses that very specifically target neurons; adeno-associated viruses, which attach to a specific binding site on chromosome 19; adenoviruses, which are a class of viruses with double stranded DNA that cause a number of respiratory and eye infections; and retroviruses, which are a group of viruses that can create double-stranded DNA copies of their RNA genomes for integration into host cells. The scope of this article will focus on retroviral vectors.

Overview and Structur of Retroviral Vectors Retroviruses were first discovered as cell free oncogenic factors in chickens, but have subsequently been determined to be present in a large variety of animal species. Retroviral virions are typically spherical particles 80-100 nanometers in length. At the heart of the virion are two identical cop-

ies of RNA genome molecules, which are tRNA primers responsible for reverse transcription. The capsid contains a variety of proteins. The gag gene codes for proteins responsible for virion maturation, proteins generated by the pol gene are responsible for synthesis of viral DNA, and proteins created by the env gene are responsible for entry of the virion into the cell. The plasma membrane itself is derived from the host cell’s lipid bilayer membrane and contains the envelope proteins.

Retroviruses also elicit a wide range of pathologic conditions in their host; retroviruses range in action from completely vestigial, such as the spumaviruses, to being somewhat aggressive in their progression, such as Human Immunodeficiency Virus. The large number of benign types of retroviruses makes it an excellent candidate for use in gene therapy as there is typically not as significant an immune response (Templeton). Retroviral vectors themselves can be developed using a variety of packaging methods, can have all DNA removed, and can store up to eight kilobase of exogenous DNA. These viruses also have pre-coded genetic sequences responsible for regulation and replication of viral DNA. These predefined functions give retroviruses a pertinent use in gene therapy.

Components of Retroviral Vectors A simple backbone of a vector system is composed of three components: the lipid encapsulated retroviral protein capsid, the viral proteins, and the vector genome. There are currently two types of vector construction: MMLV vectors and retroviral DNA vectors.

The Moloney murine leukemia virus (MMLV) vector is the most commonly used kind of vector for gene therapy.

This kind of vector is separated into two components: the vector and the packaging cell line. The vector is composed of a lipid encapsulated retroviral protein layer, which contains the vector genome with the viral DNA “knockeddown.” This means that only proteins coding for reverse transcription, integration, and packaging are left in Volume 1 | 2011-2012 | 29

Street Broad Scientific

Review

Figure 1. “A diagram depicting the current mechanism of action in a Moloney murine leukemia virus (MMLV) vector.” Source: Verma, et al

the viral genome. The desired genes replace the proteins ordinarily coding for viral replication and virion maturation (Verma et al).The genes to be delivered by the vector are inserted or cloned into the genome construct and its expression is promoted by the 5’ long terminal repeat. The second component of the vector is a packaging cell line that delivers all the viral proteins coded for by the gag, pol, and env genes to the vector in trans. Proteins responsible for insertion of the vector genome into a host cell (transduction) are contained within the packaging construct and become active upon binding with the cell surface (Templeton).

The second type of retroviral vector, which is significantly less common, is the retroviral DNA vector. These are typically composed of vector plasmids inserted between the long terminal repeat promoters of the original virus genome. This allows for simple plasmid manipulation. The use of this vector however, is beyond the scope of this article.

Efficacy and mechanisms of action The most attractive features of the retroviral vector are its unique life cycle and expedient design of the retroviral genome. A retrovirus binds to a host cell through interaction of the Env glycoprotein with receptors on the cell surface. This results in fusion of the cell lipid bilayer with the viral sheath. Upon binding of the retrovirus to 30 | 2011-2012 | Volume 1

host cell, the virus converts its RNA genome into double stranded DNA, which is then efficiently integrated into the cell genome. This robust process occurs as the reverse transcriptase enzyme converts viral RNA into double stranded DNA, which is then randomly inserted into the host cell genome by the enzyme integrase. This enables efficient protein synthesis as the provirus (the integrated DNA) is transcribed as normal cellular genetic material.

Thus, the viral proteins and the viral RNA genome can be transcribed without the need for de novo protein synthesis – the virus itself does not need to produce infectious proteins or particles, which minimizes an immune response to the vector (Anson).

The genetic structure of the retrovirus also makes it an attractive vector. The DNA based proviral form allows for simple molecular manipulation. We can easily create a replication defective (non-reproducing) form by manipulating cis and trans elements of the genome. The cis elements of the retroviral DNA (biologically active nucleic acids) and the trans elements (proteins coding sequences) are non-overlapping. Thus, the two can easily be separated in the manufacture of replication defective vectors: the nucleic acid elements can be introduced on a transfer vector construct and protein elements can be expressed using standard recombinant plasmid expression systems. By constructing helper cell lines that produce the trans-acting viral gene products, we can propagate the cis components in manufacture of a fully replication defective cell line. In

Review this way, the resulting provirus will be free of viral DNA responsible for viral replication and cell lysis. The resulting vector is replication dead and therefore nonmalignant and suitable for protein introduction (Markwoitz et al). The diagram below explains in detail the mechanism for production of retroviral vectors and the action of said vectors on target cells. suitable for protein introduction (Markwoitz et al). The diagram below explains in detail the mechanism for production of retroviral vectors and the action of said vectors on target cells.

II. Advantages of Retroviral Vectors Defective Vector System A replication defective vector system refers to a retroviral vector without genes responsible for viral reproduction.

This defect occurs naturally in certain mouse retroviruses because part of the normal viral genome has been replaced with a cDNA copy of a cellular oncogene. This type of vector is beneficial in gene therapy because its viral proteins do not have to be introduced to the host cell; rather, proteins can simply be provided in trans by the producer cell. No de novo protein synthesis is required in maintenance of the provirus, thus, immune response commonly associated with viral protein transfer and production is minimized. The repression of replication and reduced immune response associated with replication defective retroviruses makes them an attractive agent for protein and drug transfer (Hindmarsh et al).

Well Documented Integration One of the primary advantages of retroviral vectors over others is that they are able to integrate their genetic information into the host cell efficiently, and the mechanism for this has been well documented. The process begins when the virus RNA genome is reverse transcribed into linear DNA before being converted into double stranded DNA. The ends of the long terminal repeats found at the termini of the linear viral DNA are recognized by integrase. The next step is formation of preintegration complexes, large protein structures composed of linear viral DNA; several viral proteins including matrix reverse transcriptase, nucleocapsid, and viral integrase; and at least two cellular proteins, high-mobility-group [HMG-I(Y)] and barrier to autointegration factor (BAF). This complex enters the nucleus via the nuclear pores or after the disintegration of the nuclear membrane in cell division.

Once the complex enters the nucleus and associates with the chromosomes, viral integrase is released. This enzyme catalyzes the insertion of viral DNA into the host genome by bringing the linear viral DNA together with the host DNA. Lastly, a two base pair sequence is lost from each end of the viral DNA, four to seven base pairs are duplicated on the ends of the host DNA, and integrase binds the host and viral DNA. Cellular proteins mediate

Street Broad Scientific repair of damage to the newly generated provirus from the binding process. The excellent documentation and efficient insertion of desired genes into target cells make retroviruses ideal vectors for DNA transduction (Anson).

Flexible Component Organization and Gene Expression Although in a rudimentary retroviral vector system the 5â&#x20AC;&#x2122; long terminating repeats acts as a promoter for the ordinarily coding for viral replication and virion maturation (Verma et al).The genes to be delivered by the vector are inserted or cloned into the genome construct and its expression is promoted by the 5â&#x20AC;&#x2122; long terminal repeat. The second component of the vector is a packaging cell line that delivers all the viral proteins coded for by the gag, pol, and env genes to the vector in trans. Proteins responsible for insertion of the vector genome into a host cell (transduction) are contained within the packaging construct and become active upon binding with the cell surface (Templeton).

III. Goals in Developing Retroviral Vectors Target Specificity One of the foremost goals in vector design is creating a vector that can target specific cells. Creating feature of a vector would grant it a significantly greater clinical relevance because it would transduce only in targeted cells.

Development of this technology would allow in vivo delivery of a vector to an afflicted cell and would allow for treatment of specific cells using gene therapy. Proteins on the viral membrane surface mediate the mechanism by which a virus is guided towards its target cell in nature. The interaction of viral surface proteins with receptors on the cell surface detery mines entry of the virus into the cell. Thus, one step toward creating a target specific vector involves tagging the vector surface with the appropriate proteins to interact with the surface of the target cell. Upon binding of a vector to a cell surface receptor, the receptor will either cause a change in the physiological protein structure of the virus to grant it entry or it will cause acidification of the viral sheath to induce structural changes.

This type of targeting is known as vector pseudotyping.

Regulated Gene Expression Maintaining an appropriate level of gene expression in cells with transduced vector DNA has been problematic in retroviral vectors in the past. Integration of a vector genome into the provirus in the host cell is an advantage of current retroviral vectors, however, regulation of the gene expression of the vector DNA has not been effectively documented. This problem is true of all types of vectors; it is not limited to retroviruses in particular. There are two main barriers to effective gene regulation: interaction of the cell with vector promoter sequences and interaction of the host immune system with vector generated proteins.

Although a promoter in the vector genome may drive gene Volume 1 | 2011-2012 | 31

Street Broad Scientific expression in vitro, the in vivo action of the promoter sequence is unpredictable – interaction of the sequence with existing promoters or override of the promoter by cellular promoters minimizes control over the vector gene expression. The immune system also has the ability to recognize foreign promoters that have been integrated into the host genome and inactivate them. Suppression of the long terminal repeat promoter has been studied and will be further reviewed in Section V. The second factor that limits gene regulation is interaction of the immune system with vector produced proteins. Immune identification of endogenous foreign proteins, even if produced by the host cell’s machinery, causes mediated cell death in vectortransduced cells. Untreated, the loss of vector viability is inevitable even if promoter sequences operate correctly (Verma et al). One way that scientists have circumvented this problem is by using proteins that limit T-Cell response. This was exemplified in the study regarding Y-interferon transfer into tumor cells, conducted by Dr. Gansbacher at Sloan-Kettering Cancer Center. In the work, roviral vectors were used to introduce the γ-interferon (IFN-γ) gene into CMS-5 cells.” T-Cell Activation was repressed and this allowed for a long term, stable expression of the endogenous proteins. Future study should seek to minimize immune response to vector proteins and allow for the efficient regulation of gene expression.

Effective Integration The inherent risk of mutagenesis following an ineffective integration has been a point of study since 2002, when a group of ten monkeys exposed to myeloblatic irradiation and subsequently transplanted with hematopoetic stem cells treated with a viral vector (Templeton). The introduction of the replication competent retroviral vector resulted in mutagenesis of the target cells. It is therefore

Figure 1. “Components and functionality of a gene regulatory network.”

Source: U.S. Department of Energy Genome Programs 32 | 2011-2012 | Volume 1

Review pertinent for us to ensure comprehensive error checks after integration of vector genome.

IV. Overview of Regulatory Gene Networks Introduction to Gene Regulatory Networks A gene regulatory network is a collection of DNA, which has its expression regulated by the interaction of secondary products of gene expression (including proteins and mRNA). A regulatory network typically involves translation of modular DNA into mRNA sequences that code for specific proteins, which can include structural proteins, enzymatic proteins, and transcription factors.

Every module of DNA evaluates and responds to a number of inputs, using activators and repressors to gain transcriptional control over epigenetic expression. Networks are capable of completing a variety of tasks – they are quite versatile and robust. For example, depending on the protein coded for by genes in the network, endogenous protein concentration can be upregulated, biologically catalyzed reactions can be induced, and transcription factor binding to promoter sequences can activate genes.

A regulatory gene network can essentially be thought of as a network of computational units in the sense that the network functions on logic connectivity and ordinary stochastic processes. This means that most of the functionality of the network is provided by using simple logic operators like “or” functions, “and” functions” and “switch” functions”. Although the basis of the network is simple, complex arrangement of the nodes (DNA elements) of the network can cause stochastic behavior. The caustic nature of these networks is also significant to study – they arrangement of DNA sequences is intentional and caustic – it provides temporal regulation of intergene connectivity. The potential for engineering synthetic genetic regulatory networks is an emerging point of study in the field of synthetic biology that may have lucrative applications in retroviral vector gene therapy (Lu et al).

Types of Gene Circuity

The most significant types of genetic circuitry involved with the creation of a robust gene network are the gene oscillator, cell-cell communicator, and genetic toggle switch controller. The gene oscillator is capable of manipulating a network architecture based in fluid positive feedback loops to cause an oscillation in the expression of a gene. The most famous cell-cell communicator developed thus far was that developed by Thomas Bulter. His circuit used “a threshold concentration of acetate to induce gene expression by acetate kinase and part of the nitrogenregulation two-component system.” His communicator essentially made an artificial quorum sensor in E. Coli that allowed the bacteria to respond to changes in its environment. J.J. Collitns characterized the genetic toggle switch controller effectively at the turn of the century in his

Review

Street Broad Scientific

renowned study “Construction of a genetic toggle switch in Escherichia coli.” The controller is composed of two repressible promoter sequences and is flipped between two stable states to switch between expressions of two genes or to turn the expression of a single gene on and off.

Other genetic circuits that have been developed include digital logic evaluators, filters, and sensors, but those topics are out of the scope of this review. The diagram in Figure 2 represents the typical functionality of a regulatory gene network.

Challenges of Engineering Regulatory Networks Although significant strides have been made in the field of synthetic biology since the pioneering of the inaugural devices by J.J. Collins and associates, current gene networks lack robustness that would allow them to demonstrate predictable behaviors. An insufficient library of modular component parts prevents effective computational modeling in silico – it is therefore our lack of characterized interoperable parts that prevents the construction rather than a flaw in the construction method itself.

Therefore, a significant time in any synthetic biology project will error checks after integration of vector genome.

Retroviral vectors were used to introduce the γ-interferon (IFN-γ) gene into CMS-5 cells.”T-Cell Activation was repressed and this allowed for a long term, stable expression of the endogenous proteins. Future study should seek to minimize immune response to vector proteins and allow for the efficient regulation of gene expression.

Volume 1 | 2011-2012 | 33

Street Broad Scientific References

Anson, Donald S. “Retroviral Vectors.” Genetic Vaccines and Therapy. Biomed Central, 13 Aug. 2004. Web. 17 Feb. 2012. <http:// www.gvt-journal.com/content/2/1/9>.

Anson, Donald S. “The Use of Retroviral Vectors for Gene Therapy-what Are the Risks? A Review of Retroviral Pathogenesis and Its Relevance to Retroviral Vector-mediated Gene Delivery.” Pub Med Central - NCBI. National Institutes of Health. Web. 17 Feb. 2012. <http://www.ncbi.nlm.nih.gov/pmc/articles/PMC515179/>.

Butler, Thomas, Sun-Gu Lee, Wilson W. Wong, Eileen Fung, Michael R. Connon, and James C. Liao. “Design of Artificial Cell–cell Communication Using Gene and Metabolic Networks.” PNAS. org. Proceedings of the National Academies of Sciences of the US. Web.

Cole, Caroline, Jian Qiao, Timothy Kottke, Rosa M. Diaz, Atique Ahmed, Luis Sanchez-perez, Gregory Brunn, Jill Thompson, John Chester, and Richard G. Vile. “Tumor-Targeted, Systemic Delivery of Therapeutic Viral Vectors.” MedScape Today Scientific News. Web. 17 Feb. 2012. <http://www.medscape.com/viewarticle/514625_4>.

Cronin, Michelle, Ali R. Akin, Sara A. Collins, Jeff Meganck, JaeBeom Kim, Chwanrow Baban, Susan A. Joyce, Gooitzen M. Van Dam, Ning Zhang, Douwe Van Sinderen, Gerald C. O’Sullivan, Noriyuki Kasahara, Cormac C. Gahan, Kevin P. Francis, and Mark Tangney. “PLoS ONE: High Resolution In Vivo Bioluminescent Imaging for the Study of Bacterial Tumour Targeting.” PLoS ONE : Accelerating the Publication of Peer-reviewed Science. Web. 17 Feb. 2012. <http://www.plosone.org/article/info:doi/10.1371/journal.pone.0030940>.

“Engineered Bacteria Effectively Target Tumors, Enabling Tumor Imaging Potential in Mice.” Science Daily: News & Articles in Science, Health, Environment & Technology. Web. 17 Feb. 2012. <http://www.sciencedaily.com/releases/2012/01/120125172319. htm>.

Gansbacher, Bernd, Rajat Bannerji, Brian Daniels, Karen Zier, Kathy Cronin, and Eli Gilboa. “Retroviral Vector-mediated γ-Interferon Gene Transfer into Tumor Cells Generates Potent and Long Lasting Antitumor Immunity.” Cancer Research. AACR Journals, 15 Dec. 1990. Web. 17 Feb. 2012. <http://cancerres. aacrjournals.org/content/50/24/7820.short>.

Ghani, Karim, Sylvine Cottin, Pedro Otavio De Campos-Lima, Marie-Christine Caron, and Manuel Caruso. “Characterization of an Alternative Packaging System Derived from the Cat RD114 Retrovirus for Gene Delivery.” The Journal of Gene Medicine. Wiley Online Library, 8 June 2009. Web. 17 Feb. 2012. <http://onlinelibrary.wiley.com/doi/10.1002/jgm.1351/full>.

Hindmarsh, Patrick, and Jonathan Leis. “Retroviral DNA Integration.” Pubmed Central - NCBi. National Institutes of Health, 1999. Web. 17 Feb. 2012. <http://www.ncbi.nlm.nih.gov/pmc/articles/PMC98978/>.

Hu, Wei-Shau, and Vinay K. Pathak. “Design of Retroviral Vectors and Helper Cells for Gene Therapy.” Pharmocological Reviews. Aspet Journals, 1 Dec. 2000. Web. 17 Feb. 2012. <http://pharmrev. aspetjournals.org/content/52/4/493.full#title12>.

Hughes, S. H., and H. E. Varmus. “Principles of Retroviral Vector

34 | 2011-2012 | Volume 1

Review Design.” NCBI Bookshelf. Ed. J. M. Coffin. National Institutes of Health / Cold Spring Harbor Laboratory Press, 1997. Web. 17 Feb. 2012. <http://www.ncbi.nlm.nih.gov/books/NBK19423/>.

Janes, Sam. “Mesenchymal Stem Cell Delivery of TRAIL Can Eliminate Metastatic Cancer.” Cancer Research. AACR Journals. Web. 17 Feb. 2012. <http://cancerres.aacrjournals.org/content/69/10/4134. full>.

Markowitz, D., S. Goff, and A. Bank. “A Safe Packaging Line for Gene Transfer: Separating Viral Genes on Two Different Plasmids.”

Journal of Virology. American Society for Microbiology. Web. 17 Feb. 2012. <http://jvi.asm.org/content/62/4/1120.abstract>.

Sorge, Joe, Dowain Wright, Valerie D. Erdman, and Ann E. Cutting. “Amphotropic Retrovirus Vector System for Human Cell Gene Transfer.” Pubmed Central - NCBI. National Institutes of Health.

Web. 17 Feb. 2012. <http://www.ncbi.nlm.nih.gov/pmc/articles/ PMC368980/pdf/molcellb00151-0068.pdf>.

Stricker, Jessee, Scott Cookson, Matthew R. Bennett, William H. Mather, Lev A. Tsimring, and Jeff Hasty. “A Fast, Robust and Tunable Synthetic Gene Oscillator.” Access : Nature. Nature: International Weekly Journal of Science, 29 Oct. 2008. Web. 17 Feb. 2012. <http://www.nature.com/nature/journal/v456/n7221/full/nature07389.html>.

Sugimoto, Y., A. Aksentijevich, M. M. Gottesman, and I. Patsman.

“Efficient Expression of Drug-selectable Genes in Retroviral Vectors under Control of an Internal Ribosome Entry Site.” PubMed.org NCBI. National Institutes of Health, July 1994. Web. 17 Feb. 2012. <http://www.ncbi.nlm.nih.gov/pubmed/7764914>.

Templeton, Nancy Smyth. Gene and Cell Therapy: Therapeutic Mechanisms and Strategies. New York: Marcel Dekker, 2004. Print.

“The V-ras Oncogene Inhibits the Expression of Differentiation Markers and Facilitates Expression of Cytokeratins 8 and 18 in Mouse Keratinocytes.” Wiley Online Library. Web. 17 Feb. 2012. <http:// onlinelibrary.wiley.com/doi/10.1002/mc.2940030608/abstract>.

Verma, Inder M., and Nikunj Somia. “Gene Therapy – Promises, Problems and Prospects.” Western Washington University. Web. 17 Feb. 2012. <http://fire.biol.wwu.edu/trent/trent/genetherapy2.pdf>.

Yu, S. F., T. Von Ruden, P. W. Kantoff, C. Garber, M. Seiberg, U. Ruther, W. F. Anderson, E. F. Wagner, and E. Gilboa. “Self-inactivating Retroviral Vectors Designed for Transfer of Whole Genes into Mammalian Cells.” Abstract - PNAS.org. Proceedings of the National Academies of Sciences of the US, 1 May 1986. Web. 17 Feb. 2012. <http://www.pnas.org/content/83/10/3194.short>.

Acknowledgements Dr. Myra Halpin The North Carolina School of Science and Mathematics Research in Chemistry Program Duke University Pratt School of Engineering 2012 Duke iGEM Mentors Dr. Jingdong Tian Dr. Nicolas Buchler Dr. Charles Gersbach Mr. Aakash Indurkhya

Street Broad Scientific

Research

Effect of Thiols on the Environmental Fate of Silver Nanoparticles Avi Aggarwal Abstract: Silver nanoparticles (AgNPs) are used in consumer goods for their antimicrobial properties, yet little is known about their environmental impact. When they are released into aquatic systems, their fate is influenced by organic ligands that may adsorb to particle surfaces and modify reactivity, causing particles to aggregate or dissolve into potentially bioavailable forms of silver. The thiol (S-H) ligand is likely to play a key role because of its affinity for silver. We assessed the effect of thiols on citrate-coated AgNPs using glutathione (GSH), a low molecular weight peptide produced by organisms during oxidative stress and exposure to toxic metals. Aggregation of AgNPs was measured through time-resolved dynamic light scattering (DLS). Silver dissolution over time was determined by filtering batch suspensions to separate dissolved from particulate silver and measured with inductively coupled plasma mass spectroscopy (ICP-MS). Also, GSH concentrations measured with high pressure liquid chromatography (HPLC) and zeta potential measured with DLS were used to monitor surface modifications on the AgNPs. Results show that glutathione sorbs on the surface of AgNPs, reduces growth rate, and improves AgNP stability in solution, and moreover, that these processes occur simultaneously. This has implications for persistence of silver in aquatic systems and less bioavailability to organisms.

Introduction Silver nanoparticles (AgNPs) have unique antimicrobial properties and are widely used in a variety of applications including food storage, wound dressings, and fabrics.

Applications for the NPs are increasing, but little is known about their environmental implications and risks associated with their use. The silver ion has cytotoxic effects on organisms through oxidative damage1. While much of the recent work assessing risks of AgNPs has focused on pathways of toxicity1, 15, it does not consider environmentally relevant means of exposure. An important first step in researching the risks posed to ecological systems and organisms is to determine means of exposure to AgNPs, which will influence transport, transformation, bioavailability, and toxicity of the nanomaterials2.

Silver nanoparticles are likely to reach aquatic environments through runoff and consumer waste, and their fate is affected by dissolved organic matter (DOM) and ligands found within. Organic ligands can modify particle surfaces and alter the stability of the nanoparticles in water, causing aggregation into nanoparticle conglomerates and dissolution into other potentially bioavailable forms of silver.

This potential for multiple transformations is what makes nanomaterials unique contaminants in the environment, unlike other emerging organic pollutants.

The thiol group (SH) has been shown to stabilize HgS and ZnS nanoparticles by adsorbing on their surface 3, 4, 5, and it is likely to play a key role in the fate of AgNPs because of its high affinity for silver. Silver nanoparticles are known to form highly insoluble sulfides or chlorides in freshwaters 6, 7, wastewater treatment plants 8, and seawater 9, but when thiol containing organic compounds are present in significant concentrations, they may compete with sulfide for binding to silver and increase dissolved

silver concentration10. In this work we used GSH, a low molecular weight thiol-containing peptide produced by organisms in response to oxidative stress, to model the fate and transport of AgNP in thiol rich environments through the central processes of dissolution, aggregation, and ligand surface sorption. Examples of thiol-rich environments include the anaerobic pore water of sediments11, contaminated marine waters containing phytochelatins (product of plants response to increased metal concentrations12), inside of cells (where concentrations can reach up to hundreds of millimolars13), and blood14.

Nanoparticle suspensions are typically manufactured with a stabilizing coating to prevent flocculation of particles. This coating can be steric, which involves polymers adsorbed to the surface to prevent contact, or electrostatic, such as citrate, which creates a barrier of counter ions around the particle surface and causes electrostatic repulsion between particles. However, in natural environments, particles come in contact with organic ligands that can adsorb to particle surfaces and replace the coating. This sorption, along with changes in ionic strength in natural waters, can alter colloidal stability and lead to more aggregation, which in turn could decrease exposed NP surface area and slow dissolution. Conversely, because of their high affinities, the ligands may attach to and remove ions at the NP surface, inducing surface-modified dissolution (See Figure 1.) These inter-related processes occur simultaneously. The goal of this research was to determine the mechanisms of the processes taking place and how they are related.

Volume 1 | 2011-2012 | 35

Street Broad Scientific

Figure 1. Surface adsorption by ligands will interfere with NP solubility and bioavailability by influencing aggregation, dissolution kinetics, and metal speciation.

(Diagram created by and used with persmission from mentor.)

Methods and Materials Materials All chemicals used in this work were ACS reagent grade and purchased from Sigma-Aldrich unless otherwise stated. Stock solutions were prepared using filtered (<0.2 μm) ultrapure water (Barnstead Nanopure, >17.8 MΩ-cm). Trace-metal grade acids were used to adjust the pH of solutions. Ultrahigh purity nitrogen was utilized for purging oxygen from aqueous samples. All borosilicate glass containers were acid-washed through soaking overnight in 1 N HCl followed by three rinses with ultrapure water. Stock solution of 3.25 mM Lglutathione was prepared with degassed water, stored at 4 °C, and utilized within 2 weeks of preparation. Synthesis of silver nanoparticles with citrate coating (Ag-CIT) followed previously published methods15. Particle monomer size and shape were characterized with Transmission Electron Microscopy (TEM). Synthesis and characterization of NPs was done by research group associates.

Ag-CIT consisted of mostly spherical with some oval shaped particles and an average geometric diameter of 19.1 ± 12.7 nm (average of 188 particles measured from TEM images). From the number-based size distribution of the two stock suspensions, surface area per mass was calculated: 16.1 m2/g. Stock solution of Ag-CIT was stored at 4 °C. Experimental solutions were prepared of 1 ppm AgCIT (9.26 µM) in a pH 7.5 buffer solution with 7 mM NaHCO3, 10 mM NaNO3, and varying glutathione concentrations of 1 µM, 10 µM, and 100 µM GSH. Bicarbonate buffer solution was prepared fresh daily.

Methods Dissolved silver was measured by preparing replicates 36 | 2011-2012 | Volume 1

Research of solutions containing 1 ppm AgNP-CIT (9.26 µM), 7 mM NaHCO3 buffer solution, 10 mM NaNO3, and varying GSH concentrations of 1 µM, 10 µM, and 100 µM and filtering at individual time points over the time range of zero to forty eight hours. Solutions were filtered with 0.025 μm membranes (VSWP Millipore) to separate particles from suspension. Membranes were rinsed with 7 mM NaHCO3 buffer solution before used for the suspensions. The duration of filtering ranged from eight to ten minutes. Silver concentration in the filtrate was measured after digesting samples with 2% HNO3 and 1% HCl by inductively coupled plasma mass spectroscopy (Agilent Technologies).The silver concentration of the acidified samples was kept below 50 ppb to avoid precipitation of silver chloride. The ICP-MS measurements were taken by a member of the research group presiding over my work.

Glutathione measurements were also taken from filtrates; GSH measurements were quantified by reverse phase high performance liquid chromatography (Varian ProStar)16, 17. Samples were diluted in 0.5 M sodium acetate buffer dissolved in water and adjusted to pH 6 and 2, 2’-dithiobis(5-nitropyridine) (DTNP) dissolved in in acetonitrile to derivatize the thiol. HPLC-grade solvents were utilized for all reagents. HPLC involved a C18 column and UV-Vis detector. These measurements were also done by the same presiding member of the research group.

DLS (Zetasizer Nano Series, Malvern Instruments) was used to monitor particle growth by measuring the intensity-weighted average hydrodynamic diameter of particles precipitating in test solutions over a range of 96 hours using incident light (λ = 633 nm) scattered at 173°. Individual measurements were taken at 0, 3, 24, and 48 hours. Immediately after addition of AgNP to the test solution, a 1 mL aliquot was dispensed into a polycarbonate disposable 1 cm cuvette cleansed with ultrapure water and placed in the instrument. The remaining solution was then preserved for the next measurements. The average hydrodynamic diameter of particles was estimated approximately every 3 min by averaging 16 individual measurements runs. All particle growth measurements were conducted at 25° C. Zeta (ζ) potential of Ag NPs, which is a measure of surface potential and stability, was calculated from electrophoretic mobility measurements of the particle suspensions (Zetasizer Nano Series, Malvern Instruments). Mobility was calculated by applying electric potential through a capillary cell with electrodes at either end of the sample holder. The velocity of the moving charged particles was measured and used to infer the magnitude of their surface charge and zeta potential. Zeta potential was measured 4 times for each time point after the initiation of AgNP aggregation at 25 °C. Additionally, a pH replicate for each test solution was measured at each of the time points.

Results and Discussion Previous data on silver dissolution, size, zeta potential, and pH collected from control solution with no added organic ligands (1 ppm AgNP-CIT (9.26 µM), 7 mM NaHCO3, 10 mM NaNO3) by collaborating researcher

Street Broad Scientific

Research was added to measurements. All data presented are either individual measurements or averaged measurements with standard deviation error bars.

Silver Dissolution – (Figure 2) Solutions containing GSH displayed elevated dissolved Ag concentrations compared to solutions with no ligand. 10 μM and 100 μM GSH solutions generally showed greater dissolution than 1 μM GSH solution. Dissolution increased rapidly within the initial measurements (0, 3, 6hrs) and continued to dissolve over the remaining period. Solution of 10 μM GSH appears to have the highest dissolved Ag but shows highly variable concentrations within replicate measurements, so a trend between dissolution and GSH concentration cannot be characterized.

FIGURE 3

Figure 3. GSH detected after filtering 1 ppm Ag-CIT in 7 mM NaHCO3, 10 mM NaNO3 and GSH concentrations of 1 μM GSH, 10 μM GSH, and 100 μM GSH. GSH was found to be stable over time in concentrations corresponding to original values. Bottom- Percent of GSH measured in filtrate of original GSH in solutions.

Figure 2. Dissolution of 9.3 μM Ag-CIT in the presence of 1 μM GSH, 10 μM GSH, 100 μM GSH, and no ligand. Solution pH was 7.5-8.5 buffered with 7 mM NaHCO3. Ionic strength was controlled with 10 mM NaNO3. Measured total silver. GSH significantly increases dissolution compared to solution with no ligand; however a definite trend with respect to concentration cannot be determined.

Glutathione in filtrate - (Figure 3) GSH concentrations were measured from the same filtrates to quantify adsorption of GSH to the particle surface, disregarding the potential confounding factor of oxidation of GSH. GSH measured for each solution was relatively stable across all time points measured, indicating that ligand-NP adsorption occurs soon after NPs are exposed to ligands and does not fluctuate much after the initial exchange. This suggests that GSH stabilizes AgNP for at least 48 hours. Right: Concentrations from A were converted to percentage of original GSH in solutions. Solution with 1 μM GSH solution had significantly lower percentage than solutions with 10 μM and 100 μM GSH, which could be attributed to limiting behavior of low GSH or inconsistency in measuring low concentrations.

Zeta Potential – (Figure 4) Zeta potential, which is an indicator of particle surface charge and stability, was measured over time. All solutions containing GSH displayed increasingly negative zeta potential values over time, suggesting increased repulsion between particles and more stability. Both 10 μM and 100 μM GSH solutions, which contained GSH in excess of Ag-CIT, stabilized at similar values; 1 μM GSH solution showed less negative values.

This data shows that increased GSH causes more surface modification among AgNPs.

Volume 1 | 2011-2012 | 37

Street Broad Scientific Figure 4. (Previous page) Zeta potential of 9.3 μM AgCIT in the presence of 1 μM GSH, 10 μM GSH, 100 μM GSH, and no ligand. Solution pH was 7.5-8.5 buffered with 7 mM NaHCO3. Ionic strength was controlled with 10 mM NaNO3. The presence of GSH appears to decrease zeta potential, implying stability.

Aggregation – (Figure 5) Aggregation leads to large clusters that settle out of solution faster than small particles and is an important process in the behavior of nanoparticles. It also influences surface reactions by reducing the available surface area and slowing diffusion of chemicals.

The presence of GSH at low concentration was found to induce NP aggregation compared to a control solution with no ligand, but increasing GSH concentrations resulted in less aggregation. This suggests that GSH significantly alters NP stability even at low concentrations, but when present at high concentrations in excess of the AgNP, such as in the 10 μM and 100 μM solutions, it can alternately promote stability and prevent interaction of particles.

Research on the growth, dissolution, and surface modification of citrate-coated silver nanoparticles can be summarized as follows:

Figure 7. (above) Initial Ag-CIT growth rates were calculated by fitting a linear trend to the first 0-3 hours of hydrodynamic diameter measurements (see Fig. 5) for each solution and are plotted on a logarithmic scale. Additional measurements in this time range are necessary to characterize growth; values are imprecise and displayed with standard error. GSH was found to reduce growth rate.

Figure 5. (above) Aggregation of 9.3 μM Ag-CIT in the presence of 1 μM GSH, 10 mM GSH, 100 μM GSH, and no ligand in 7 mM NaHCO3 solution pH buffered at 7.58.5. Ionic strength was controlled with 10 mM NaNO3.

GSH promotes aggregation at low concentrations, but at high concentrations it reduces growth.

Figure 6. (below) The pH of all solutions was also measured over time with replicates to measure changes in solution behavior and give context for zeta potential values. All solutions displayed an increase in pH value over the time range.

Figure 8. Rates of dissolution over time range calculated from fitting a linear trend to the Ag dissolution measurements (see Fig. 2) normalized to initial total surface area (μmol/m2•h) and plotted over a logarithmic scale for each replicate tested. Definite trend cannot be determined, but shows that intermediate GSH concentrations increase dissolution rates.

From the above data, the effects of GSH concentration 38 | 2011-2012 | Volume 1

Research

Figure 9. Zeta potential for each GSH concentration (see Fig. 4) tested at two selected time points, 3 hours and 48 hours, plotted over a logarithmic scale. Shows that over multiple time ranges, GSH decreases zeta potential of AgNPs; this implies increased charge, surface modification, and stability.

Although other research confirms that the fate of nanomaterials is determined by several processes including dissolution, aggregation, and surface reactions18, few studies investigate these processes together. This research is unique in that it studies several inter-linked mechanisms simultaneously affecting the environmental fate and bioavailability of AgNPs in natural settings (See Figure 10.)

Figure 10. Processes affecting the fate of AgNP in the environment. (Created by and used with permission from collaborating member of research group.)

Measurements of zeta potential and of glutathione concentrations in filtrates quantify the surface modification of the nanoparticles through measuring surface charge and GSH adsorption. Size measurements of the hydrodynamic radius of nanoparticles over time quantify aggregation and speciation. Dissolved silver measurements quantify the dissolution of Ag+ from the NPs and speciation. Together, these results elucidate the role of the thiol ligand in the fate of AgNPs under various conditions and confirm that when present in significant concentrations, thiols can alter the behavior of silver nanomaterials in the environment and should be considered in assessing their fate along with other key inorganic ligands such as sulfides and chlorides.

Street Broad Scientific Environmental Implications Environmental implications of this work include the persistence of AgNPs in aquatic systems. Glutathione appeared to stabilize the AgNPs at high concentrations through reduced aggregation, reduced Ag+ dissolution, and greater surface modification, which could prevent phase separation or sedimentation of the NPs out of water. In thiol-rich environments, this may cause particles to persist in solution and remain subject to modification by other inorganic ligands, natural organic matter, and uptake by organisms.

However, the stabilization of Ag NPs through thiol-containing molecules such as glutathione can be environmentally beneficial in wastewater treatment plants. Concerns have arisen about the negative effects of Ag NP on bacteria in activated sludge9, but these concerns may not be valid if Ag NPs encounter bacterially secreted thiol-containing compounds before they reach the bacteria.

Biological Implications The behavior of Ag NP with GSH also has biological implications for the means of heavy metal toxicity in organisms. Glutathione and other thiol-containing antioxidants and enzymes are produced in response to oxidative stress and metal toxicity, but Ag NPs may sorb to these molecules and deplete their availability.

Additionally, this research will affect models of Ag NP drug delivery. Ag NPs are being considered for use in drug delivery applications19, but in the body they are likely to come into contact with high concentrations of GSH and other thiols, which will alter their availability and solubility.

Conclusions and Future Work The environmental fate and bioavailability of silver nanoparticles is likely to be affected by the thiol ligand, which, due to its high affinity for silver metal, can adsorb to particle surfaces and alter their stability in waters. We studied the nature of thiol-Ag NP interactions through measuring the aggregation, dissolution, and zeta potential of citrate-coated silver nanoparticles in the presence of varying concentrations of glutathione. Results show that GSH reduces aggregation rates, increases dissolution of Ag+ from nanoparticles, and decreases zeta potential, implying increased nanoparticle stability as a result of thiol sorption. Thiols play a critical role in the behavior and uptake of Ag NPs and should be included with other key ligands in assessing the risks of silver nanomaterials. Moreover, results stress the need for a multiple-focus holistic approach looking at all of the interacting processes that are occurring together and affecting one another.

Further experimentation is necessary to determine if GSH is adsorbing to NP surfaces as inferred; measurements of GSH could be due to GSH oxidation as well as AgNP sorption; reliable methods to measure radical oxidative species or silver oxides are required. Future work should also include studies of additional biogenic thiols besides glutathione to confirm the role of the thiol ligand in the interactions.

Volume 1 | 2011-2012 | 39

Street Broad Scientific Also, the role of thiols is limited by their presence relative to competing ligands such as sulfide and chloride6, 7, 8, 9, and the behavior of AgNP in environments with multiple influential ligands should be examined. Furthermore, we only conducted experiments at a single ionic strength, controlled at 10mM NaNO3, but ionic strength varies between freshwater and saline environments, and a gradient of ionic strengths should be considered. Likewise, changes in GSH content significantly affected Ag NP behavior in experiments, and additional GSH contents should be tested to characterize silver nanoparticle behavior for a full spectrum of possible thiol concentrations in various environments.

References Ercal, N.; Gurer-Orhan, H; Aykin-Burns, N. Toxic metals and oxidative stress part I: Mechanisms involved in metal-induced oxidative damage. Curr. Top. Med. Chem. 2001, 1, 529-539.

Aiken, G., et. al. Influence of Dissolved Organic Matter on the Environmental Fate of Metals, Nanoparticles, and Colloids. Environ. Sci. Technol., 2011, 45(8), 3196-3201.

Gondikas, A. P.; Jang, E. K.; Hsu-Kim, H., Influence of amino acids cysteine and serine on aggregation kinetics of zinc and mercury sulfide colloids. Journal of Colloid and Interface Science 2010, 347 (2), 167-171.

Lau, B. L. T.; Hsu-Kim, H., Precipitation and growth of zinc sulfide nanoparticles in the presence of thiol-containing natural organic ligands. Environmental Science & Technology 2008, 42 (19), 7236-7241.

Deonarine, A.; Hsu-Kim, H., Precipitation of Mercuric Sulfide Nanoparticles in NOM-Containing Water: Implications for the Natural Environment. Environmental Science & Technology 2009, 43 (7), 2368-2373.

Adams, N. W. H.; Kramer, J. R., Silver speciation in wastewater effluent, surface waters, and pore waters. Environ Toxicol Chem 1999, 18 (12), 2667-2673.

Kramer, J. R.; Adams, N. W. H.; Manolopoulos, H.; Collins, P. V., Silver at an old mining camp, Cobalt, Ontario, Canada. Environ Toxicol Chem 1999, 18 (1), 23-29.

Kim, B.; Park, C. S.; Murayama, M.; Hochella, M. F., Discovery and Characterization of Silver Sulfide Nanoparticles in Final Sewage Sludge Products. Environmental Science & Technology 2010, 44 (19), 7509-7514.

Miller, L. A.; Bruland, K. W., Organic Speciation of Silver in Marine Waters. Environmental Science & Technology 1995, 29 (10), 2616-2621.

Adams, N. W. H.; Kramer, J. R., Reactivity of Ag+ ion with thiol ligands in the presence of iron sulfide. Environ Toxicol Chem 1998, 17 (4), 625-629.

Shea, D.; Maccrehan, W. A., Determination of Hydrophilic Thiols in Sediment Porewater Using Ion-Pair Liquid-Chromatography Coupled to Electrochemical Detection. Analytical Chemistry 1988, 60 (14), 1449-1454.

Ahner, B. A.; Morel, F. M. M., Phytochelatin Production in Marine-Algae .2. Induction by Various Metals. Limnology and Oceanography 1995, 40 (4), 658-665.

40 | 2011-2012 | Volume 1

Research Fahey, R. C.; Brown, W. C.; Adams, W. B.; Worsham, M. B., Occurrence of glutathione in bacteria. J Bacteriol 1978, 133 (3), 1126-9.

Meyer, J. N.; Lord, C. A.; Yang, X. Y. Y.; Turner, E. A.; Badireddy, A. R.; Marinakos, S. M.; Chilkoti, A.; Wiesner, M. R.; Auffan, M., Intracellular uptake and associated toxicity of silver nanoparticles in Caenorhabditis elegans. Aquatic Toxicology 2010, 100 (2), 140150.

Hsu-Kim, H., Stability of metal - glutathione complexes during oxidation by hydrogen peroxide and Cu(II) catalysis. Environmental Science & Technology 2007, 41, 2338-2342.

Vairavamurthy, A.; Mopper, K., Field Method for Determination of Traces of Thiols in Natural-Waters. Analytica Chimica Acta 1990, 236 (2), 363-370 Wiesner, M.R., Lowry, G.V., Alvarez, P., Dionysiou, D., Biswas, P., 2006. Assessing the risks of manufactured nanomaterials. Environ.

Sci. Technol. 40, 4336-4345.

Langer, R.; LaVan, D. A.; McGuire, T., Small-scale systems for in vivo drug delivery. Nat Biotechnol 2003, 21 (10), 1184-1191.

Street Broad Scientific

Review

Neutrinos and Their Detection Jason Liang

Introduction Neutrinos are among the most mysterious and ethereal particles in the universe. They are so weakly interacting that they were not found until 1956; even today huge detectorshave to be built to produce only a few measurable interactions per day. However, many physicists today think that this peculiar property of neutrinos can be exploited togain insights into other phenomena. Since neutrinos only react rarely and weakly, they can provide scientists with information about events that other methods of remote sensing, such as photons, cannot, such as the earliest ages of the universe and the conditions inside the supernovae.

In this literature review, I will briefly give an overview of neutrinos and the physics associated with them and then explain the different neutrino detection methods.

Neutrinos Prediction and Discovery In 1930 Wolfgang Pauli was the first to hypothesize the existence of neutrinos in order to explain missing momentum and energy in certain nuclear reactions, specifically the missing energy of the electron in beta decay. His predicted particle had no charge and spin 1/2 . In 1933, Enrico Fermi used neutrinos in his theory of beta decay. They were also soon used to explain the decay of particles produced by cosmic rays (Kaneyuki).

However, it was not until more than 20 years later in 1956 that neutrinos were detected by Clyde Cowan and Frederick Reines at the Savannah River nuclear reactor.

This is because neutrinos are leptons and are not affected by the strong nuclear force. Also, they are not affected by the electromagnetic force and have too little mass to be considerably affected by gravity. As a result only the weak nuclear force measurably affects them so they are extremely hard to detect.

Neutrinos come in three flavors: electron, muon, and tau, corresponding to the three lepton flavors. When electron neutrinos react with an atomic nucleus, an electron is produced. Similarly, muons are produced when a muon neutrino interacts, and tau particles are produced in the case of a tau neutrino.

The Solar Neutrino Problem The first solar neutrino detector was built by Raymond Davis at the Homestake Mine in South Dakota in the late

1960s. After many years of collecting data, Davis found that the measured neutrino flux at Homestead was only less than a third of the flux predicted from the standard solar model. General consensus was that his experimental methods were sound, and similar experiments confirmed this, but no one knew where the missing neutrinos were.

This prompted a deluge of revisions to the standard solar model of the Sun’s interior to correct for the predicted neutrino flux (Bachall).

The Atmospheric Neutrino Problem Another paradox was discovered during the 1980s and 1990s when several detectors measured the number of cosmic ray neutrinos produced by cosmic ray showers in the Earth’s atmosphere. Theory predicted that cosmic ray showers would produce two muon neutrinos for every electron neutrino, with those particles resulting from the decay sequence of a pion:

π + → µ+ ν µ → ν e + e + + ν µ + ν µ

However, at SuperKamiokande, only 1.3 muon neutrinos were detected for every electron neutrino (Kajita).

This contradiction with theory was called the atmospheric neutrino problem. In particular, the ratio of muon to electron neutrinos was less for neutrinos going upward (through the bulk of the Earth) than for neutrinos coming downward. The only explanation for this that did not result in major contradictions in other areas of physics was that neutrinos could oscillate, or change flavors.

Most of the detectors then in use, including SuperKamiokande, could not detect tau neutrinos because to be detectable in those detectors, tau neutrinos must interact with a nucleus via the weak interaction and produce tau particles, which requires an immense amount of energy that the neutrino usually does not have. Thus, the phenomenon of muon neutrinos changing into tau neutrinos would explain the discrepancy in the ratio of electron neutrinos to muon neutrinos. Other more complicated three-flavor oscillations also explain the experimental results (Kajita).

Neutrino Oscillation and Mass Oscillation implies that neutrinos have mass, which contradicts the results obtained from most forms of the Standard Model. Even so, in the past 20 years it has Volume 1 | 2011-2012 | 41

Street Broad Scientific become accepted that neutrinos actually have an extremely small but nonzero mass. Neutrino mass seems to be the most likely explanation for the atmospheric neutrino paradox.

Quantum mechanics can provide the explanation for neutrino oscillation because all particles with mass have wave properties. In the case of neutrinos, the wavefunction determines the probability of the neutrino being a muon neutrino instead of being an electron neutrino.

Neutrinos of different masses have dierent frequencies for their wavefunctions, and as the wavefunctions propagate through space the probability oscillates (Figure 1).

Figure 1. Oscillation of neutrino wavefunctions (Source: College of William and Mary)

The frequency of the oscillation depends on the several factors, including the energy of the neutrino and the difference of the squares of the masses (Kaneyuki). A slow oscillation corresponds to a small difference in mass of the neutrinos, while a fast oscillation indicates a large dierence.

This has allowed scientists to place limits on the masses of the different flavors of neutrinos. Thus far, the upper limit of the mass is 2.2 eV for the electron neutrino, 0.17 MeV for the muon neutrino, and 15.5 MeV for the tau neutrino (Meszaros).

Review do not corroborate very well (Figure 2). Also, since the original solar neutrino experiments, the Sudbury Neutrino Detector has confirmed that the total flux of all three types of neutrinos from the Sun is equal to the theoretically predicted value, showing that the detection methods of SuperKamiokande and other detectors could not have been the sources of the deficiency (Freedman).

Neutrino Detection Methods Introduction Neutrinos do not respond to the strong or electromagnetic forces, so detection methods cannot detect them directly but rather rely on detecting the products of their reaction with other particles. There are two main methods of detecting neutrinos: via scattering or capture.

Detecting neutrinos with the scattering method involves imaging the Cherenkov radiation that particles emit after their collision with a neutrino. Neutrino capture (or charged-current) detection methods record the energies of electrons or antielectrons that are emitted when a neutrino collides with a proton and the proton undergoes beta decay or inverse beta decay. As a result, this method can only detect electron neutrinos of electron flavor.

Radiochemical Detectors The first detector specically designed to find solar neutrinos was a chlorine detector built by Raymond Davis of Brookhaven National Laboratory in the 1960s.

His detector consisted of 100000 gallons of perchloroethylene (C2Cl4) in an underground tank at the Homestake Gold Mine in Lead, South Dakota (Freedman). When a neutrino struck the nucleus of a chlorine atom, one of the neutrons was converted into a proton, turning the chlorine into a radioactive atom of 37Ar (Bahcall).

This reaction,

νe +37 Cl → e− +37 Ar

Figure 2. Comparison of models incorporating neutrino oscillation to models which do not (Source: Kajita)

As for proof that neutrino mass can provide the explanation for the atmospheric neutrino problem, models of the number of neutrino detection events versus the angle of the incoming neutrino which incorporate neutrino oscillation agree well with the data that were collected, while regular models which do not take oscillation into account, 42 | 2011-2012 | Volume 1

could capture neutrinos with a threshold energy of 0.814 MeV at the rate of 10-35 neutrinos per atom per second, which could detect all the major sources of neutrinos except the common pp neutrinos. To determine the eciency of the collection method, samples of 36Ar and 38Ar were added before the radioactive argon was collected.

The argon was separated chemically fromthe perchloroethylene every few months by circulating helium gas through the detector, which collected the argon onto a charcoal trap where it was absorbed (Bahcall). The number of atoms was then counted by measuring the number of electrons produced by the decay of 37Ar, which has a halflife of 35 days. This process removed the argon with an efficiency of 95%. This experiment provided the first evidence of neutrino oscillation, since the measured neutrino flux was 2.1±.9 SNU (where 1 SNU = 10-36

Review captures per target atom per second) while the predicted flux was 7.9 Âą1.33 SNU (Bachall), indicating a deficit in the flux of electron neutrinos.

Scattering Detectors The main type of scattering detector is the water Cherenkov detector. When particles in a medium move faster than the speed of light in that medium a shock wave of photons is created, similar to the creation of a sonic boom by an object moving faster than the speed of sound, which results in a cone of blue light in the direction of the particle (Scholberg). (This does not violate special relativity because this theory states that nothing can ever travel faster than the speed of light in a vacuum, not just the speed of light in a particular medium.) Photomultiplier tubes lining the inside of the detector then amplify the photons into measurable electrical pulses and record them.

When an energetic neutrino enters the detector and collides with protons, positrons and other particles are produced. These particles recoil, and many of them recoil faster than the speed of light in water, producing Cherenkov radiation. The number of photons produced by a charged particle moving through the detector is proportional to the amount of energy lost by the particle. This is the principle upon which water Cherenkov detectors operate. A drawback of these detectors is that any particle that can scatter electrons is detected, not just neutrinos, so there have to be strict lters on which events are analyzed (Scholberg).

One of the oldest continuously operating detectors is Kamiokande (currently SuperKamiokande), located in Kamioka, Japan. Kamiokande was originally built in the 1980s as a proton decay detector. To screen out other particles, such as muons from cosmic rays, which could produce similar ashes of light, Kamiokande is located one kilometer under Mount Ikenoyama.

It is a cylinder which contains 50 kilotons of ultrapure water to help light maintain its intensity over a long distance and more than 10000 inward-facing photomultiplier tubes each half a meter in diameter to pick up interactions inside the cylinder. Muons produce events which are similar to the ones which neutrinos cause, so almost 2000 photomultiplier tubes face outward to detect light from charged particles, which allows muons and neutrinos to be differentiated because muons have charge while neutrinos do not (Kajita).

The main producers of Cherenkov radiation in water Cherenkov detectors are electrons. The threshold energy for these types of detectors is 0.8 MeV because the electron must have a velocity greater than c/n (where n is the index of refraction of water), about 225000 km/s, to produce Cherenkov photons.

By analyzing the properties of the light cones, such as their size, shape, and intensity, scientists can infer many properties of the incoming neutrinos. For example, they can also tell what kind of neutrino produced the light

Street Broad Scientific cone (Figure 3). The rings of light produced by muon neutrinos are relatively sharp, while electron neutrinos result in a more diuse ring because electron neutrinos produce electrons, which then scatter photons more effectively than the muons produced by muon neutrinos (Kaneyuki).

Also, the direction of the light cone is the direction of the particle produced by the neutrino, which is extremely close to the incident direction of the neutrino and thus points to the direction of the neutrino source. Lastly, the number of photons produced lets scientists know the energy of the particle and allows the energy of the neutrino to be inferred.

The results of SuperKamiokande provided significant evidence for neutrino oscillation and led to its widespread acceptance by the scientific community as the solution to the solar neutrino problem and atmospheric neutrino problem. As noted in Figure 2, the data collected by SuperKamiokande fit models incorporating oscillation but correlatedpoorly with models that did not.

Another type of scattering detector is the organic scintillation detector. Scintillation detectors are composed of hydrocarbons, specically alkanes (Scholberg). Photomultiplier tubes observe photons produced by the de-excitation of molecular orbitals. Just like in water Cherenkov detectors, the energy loss is proportional to the number of photons collected. However, light emission from de-excitation is uniform in all directions, so unlike water Cherenkov detectors scintillation detectors cannot determine the direction of the incoming neutrino. An example of this kind of detector is Borexino in Italy.

Volume 1 | 2011-2012 | 43

Street Broad Scientific

Review

Heavy Water Detection There are several detectors that can detect neutrinos using both the scattering and capture methods. One example is a heavy water detector in which each hydrogen atom in water is replaced by a deuterium atom. This is the detection method that the Sudbury Neutrino Observatory (SNO) in Canada uses (Freedman). SNO can easily detect all three types of neutrinos, while detectors utilizing electron capture can only detect one flavor, and detectors using neutrino-electron scattering are far more sensitive to one type of neutrino than to other types.

When a neutrino of any flavor passes through the detector, it can eject the neutron out of the deuterium nucleus in a neutral-current reaction. The ejected neutron is soon captured by another nucleus, producing a gamma ray which can be detected by photomultiplier tubes. The SNO can also detect neutrinos via scattering and absorption using charged currents, which are the methods that other detectors such as Super-Kamiokande use and which are limited to mainly detecting electron neutrinos. The neutral-current reactions are: νe + d → p + p + e −

ν+d→ν+p+n

The main result of this experiment was the confirmation of the theory of neutrino oscillation (Freedman). SNO could detect all three flavors of neutrinos unlike previous detectors, and when the total neutrino flux was added up the result matched the solar neutrino flux predicted by the standard solar model.

Liquid Argon Detectors The last major type of neutrino detector is the liquid argon detector, which uses a time-projection chamber (Figure 4). The time-projection chamber forms a threedimensional image of the recoil electrons produced by the charged-current reaction by using a constant electric field to drift electrons into a recording plane, with the time of drift and final position of the electron determining the original position (Bahcall). The reaction is

νe +40 Ar → e− +40 K ∗

with a threshold energy of 5.885 MeV. The excited state of potassium emits gamma rays almost immediately, which allows the time of the event to be recorded. This is the detection method used by the ICARUS detector in the Gran Sasso lab.

44 | 2011-2012 | Volume 1

Figure 4. A diagram of an argon detector.

(Source: Symmetry Magazine)

Conclusions In the past 20 years, there have been many groundbreaking studies in the field of neutrino physics. This is exemplified by the discovery of neutrino oscillations and the realization that neutrinos have mass at SuperKamiokande in Japan. In addition to research being done at SuperKamiokande, many new detectors are in the first stages of planning and construction, including a new large-scale liquid argon detector in the United States.

These potential “neutrino telescopes” will allow scientists access to new and innovative methods of neutrino detection and analysis, which may result in groundbreaking discoveries relating to the most high-energy events in the universe, such as the Big Bang and supernovae.

References Cited Bahcall, John H. Neutrino Astrophysics. Cambridge: Cambridge University Press, 1989. Print.

Freedman, Roger A., Robert M. Geller, and William J. Kaufman III. Universe. 9th ed. New York: W. H. Freeman and Company, 2011. Print.

Kajita, Kakaaki, Edward Kearns and Yoji Totsuka.

“Detecting Massive Neutrinos.” Scientic American August 1999: 48-55. Print.

Kaneyuki, Kenji and Kate Scholberg. “Neutrino Oscillations.”

American Scientist May-June 1999: 222-231 Print.

Scholberg, Kate. “Supernova Neutrino Detection.” Annual Review of Nuclear and Particle Science 2012. Print.

Street Broad Scientific

Research

Determination of Activity Coefficients of Wood Smoke Tracer in Artificial and Ambient Organic Semi-Volatile Aerosols Suqi Huang Abstract: Aerosols are known to affect the climate and have been linked to adverse health effects. An effective way of reducing air pollution is to determine and to control the sources of the harmful particles. The chemical mass balance (CMB) receptor model is a method of source apportionment that uses tracer compounds, which should be distinct to certain sources and conserved in the atmosphere. In this experiment, the Integrated Volume Method developed by Saleh and Khlystov (2009) was used to predict the volatility of a wood smoke tracer, levoglucosan, and the interaction of the tracer with artificial and ambient semi-volatile organic aerosols. Artificial mixtures were made with monocarboxylic acids, and ambient particles were collected using a filter and extracted in ethanol. Aerosols were generated from mixtures with different mole fractions of levoglucosan and sent through a heated thermodenuder. The change in volume was calculated by comparing measurements before and after heating and was graphed versus the mole fraction. Aerosol generated from levoglucosan exhibited a change in volume, which suggests that it is semi-volatile. Mixtures of levoglucosan and monocarboxylic acids do not form a solution. Mixtures of levoglucosan and ambient extracts showed interaction and did not display ideal behavior, and the activity coefficients of levoglucosan were determined.

Introduction Organic Semi-Volatile Aerosols: An aerosol is a suspension of small particles in a gas.

These particles may be solid, liquid, or a mixture, and a large majority of them are emitted naturally by sources such as volcanoes, forest fires, and plants (Voiland). Those that are produced by human activities are known as anthropogenic and are emitted from sources including biomass burning, food cooking, motor vehicle driving, and cigarette smoking (Voiland). Aerosols affect the climate through the absorption and scattering of solar radiation, which causes temperature changes, and influence cloud formation (Chung and Seinfeld 2002, Voiland). Atmospheric particles have also been associated with increased mortality and adverse health effects, including illnesses, cancer, and other harmful effects when they are inhaled (Lewtas 2007). Ambient aerosols contain organic compounds, inorganic compounds, and water; of the organic compounds, a large portion are semi-volatile (Saleh and Khlystov 2009). Semi-volatile organic aerosols display gas-particle partitioning, which means that the particles change back and forth between the gas phase and the particulate phase (Seinfeld and Pankow 2003, Saleh and Khlystov 2009). The behavior of semi-volatile organic aerosols in the atmosphere is extremely difficult to model because of the large variety of possible chemical reactions and products, all of which affect the volatility of the aerosol (Kroll and Seinfeld 2008). This difficulty comes from the constant partitioning of the particles between the gas phase and the particulate phase. The particles can undergo many potential different reactions in either phase, which makes modeling and predicting organic aerosol very complex and challenging to accomplish accurately.

Modeling with Activity Coefficients: Semi-volatile aerosols are usually modeled under the assumption that the organic particles form an ideal solution, allowing calculations and properties to be described directly using concentrations, partial pressures, and mole fractions (Pankow 1994). However, for nonideal mixtures, the activity coefficient can be included to adjust the model to better fit the behavior of the aerosols. This value can range from 0.3 to 198 and reflects the interaction of the chemical components within the aerosol and reveals any deviation from ideality (Bowman and Melton 2004). If the activity coefficient is greater than one, then a chemical component is dissimilar to the overall aerosol mixture (Bowman and Melton 2004). Models that incorporate activity coefficients agree more with experimental data than do models that assume ideal behavior (Bowman and Melton 2004). If the mixture is ideal and Raoultâ&#x20AC;&#x2122;s Law is applied, then the activity coefficient equals 1 at any mole fraction. Setting the activity coefficient as 1 for aerosols that display non-ideal behavior causes significant errors in calculations, so it is essential to determine its value in order to achieve more accurate models (Bowman and Melton 2004).

Problem Source Apportionment: Air pollution is a challenging problem to resolve because it is difficult to filter the atmosphere of harmful particles, while having every person wear a gas mask is impractical and inconvenient. The best approach to reduce the amount of harmful particles is to find out which sources are emitting particles, to estimate how much these sources are Volume 1 | 2011-2012 | 45

Street Broad Scientific contributing to the ambient particles, and to limit or to decrease the amount of particle emission from sources that contribute significant amounts.

Different methods of source apportionment are used to identify source contributors, which can include motor vehicle engine exhaust, road dust, food cooking, wood combustion, and others (Schauer et al. 1996). The time and money spent on air pollution control could be used more effectively if there are more data about the many different contributors to airborne particles and how much each source contributes, which can help determine the relative importance of the sources (Cass 1998).

Tracer Compounds One well-known method of source apportionment is the chemical mass balance (CMB) receptor model. This method originally used chemical elements as tracer compounds, such as nickel, lead, and aluminum, but air pollution control has now largely eliminated these from emissions (Cass 1998). Instead, organic compounds that are distinctive to certain source classes are being used to help â&#x20AC;&#x153;traceâ&#x20AC;? the particles back to their original sources (Baek et al. 2005, Shrivastava et al. 2007). The CMB technique solves a system of equations in which the concentration of chemical compounds in an ambient sample is set equal to a linear combination of the relative chemical compositions of contributing sources (Schauer et al. 1996). This assumes that the chemical properties of the atmospheric particles are a direct result of the linear accumulation of the emitted particles and requires accurate data concerning the concentration of organic compounds in source emissions and in ambient samples (Schauer et al. 1996). However, this method produces uncertain and possibly inaccurate results if measured concentrations and source profiles are based on limited data (Shrivastava et al. 2007, Baek et al. 2005).

In addition, all sources known to contribute the particular tracer compound used must be included in the calculations (Cass 1998). Levoglucosan as a Wood Smoke Tracer A reliable tracer compound must be emitted from specific sources and be stable and conserved from the point of emission to the receptor point where data is collected, i.e. not be changed drastically by volatilization or chemical reactions while in the atmosphere (Schauer et al. 1996).

This means that compounds such as polycyclic aromatic hydrocarbons would not make good tracer compounds because they are emitted by almost all combustion sources, react quickly in the atmosphere, and evaporate to the gas phase (Cass 1998). Levoglucosan (1,6-Anhydro-β-Dglucopyranose) is a compound commonly used as an organic tracer for biomass burning and wood smoke (Shrivastava et al. 2007). Jordan et al. (2006) analyzed emission samples from the smoke of woodstoves and compared

46 | 2011-2012 | Volume 1

Research the results to ambient samples collected in Launceston, Australia. Levoglucosan was found at expected mass fractions in the ambient samples, while the other compounds tested had much lower mass fractions ( Jordan et al. 2006).

However, other studies have suggested that levoglucosan is semi-volatile and experiences changes in volume through evaporation to the gas phase (Oja and Suuberg 1999, Huffman et al. 2009). If levoglucosan is unstable and interacts with other particles in the atmosphere, then its use as a tracer compound becomes questionable. Using the activity coefficient of levoglucosan to correct for its nonideal behavior will make calculations for the apportionment model more accurate.

Research Goal The goal of this research was to investigate the volatility of levoglucosan and the interaction of levoglucosan with artificial and ambient organic semi-volatile aerosols.

The Integrated Volume Method developed by Saleh and Khlystov (2009) was used to calculate the change in aerosol volume, which is a measure of the aerosol volatility, of mixtures with different amounts of levoglucosan. The value of the activity coefficient of levoglucosan within a semivolatile mixture provides information about the interaction of the tracer compound with the other particles and determines whether those particles have any effect on the volatility of the tracer.

Integrated Volume Method Saleh and Khlystov (2009) developed the Integrated Volume Method (IVM) to estimate the activity coefficient of semi-volatile organic aerosols in binary solutions. Figure 1 shows their experimental setup. Compounds based on mole fractions are dissolved in a solvent, usually water or ethanol, and aerosol is generated by spraying the solution with an atomizer. The aerosol then enters a large mixing chamber for dilution and drying. For solutions dissolved in ethanol, the aerosol is drawn through an activated carbon denuder to make sure that the aerosol is dry.

After exiting, the aerosol is split into two lines. For one line, reference measurements are taken using the upstream Scanning Mobility Particle sizer (SMPS), while the other passes through a thermodenuder, a stainless steel tube which can be heated up. After the semi-volatile particles evaporate and reestablish equilibrium at a higher temperature in the thermodenuder, the volume is measured again via the downstream SMPS. From these data, the change in volume before and after the aerosol is heated can be calculated by comparing the measurements from the upstream SMPS with those from the downstream SMPS.

Street Broad Scientific

Research

The experiment is repeated multiple times with different mole fractions. To check the ideality of the aerosol, the change in volume is graphed along the y-axis with the mole fraction along the xaxis. If the resulting graph does not display a straight line, the conclusion is that the mixture is not ideal and that the evaporation does not follow Raoult’s Law. If the aerosol had shown ideal behavior, the change in volumewould be linearly proportional with the mole fraction and the activity coefficient would equal 1.Determination of Activity Coefficients: The method of determining the value of the activity coefficient involves repeating an algorithm until convergence is reached, which is described in Saleh and Khlystov (2009). First, the change in volume of component i in an aerosol with more than one component is represented by: Δvp,i = (Mi)/(ρρ,iRT0) * (xi, yi,1Psat,i,0) where T0 is the initial equilibrium temperature, xi is the mole fraction of the component, yi is the activity coefficient, Psat,i is the saturation pressure of the pure component, and the subscript 1 indicates the conditions at the new higher temperature. Therefore, the total change in volume of the aerosol is the sum of the change in volumes of the individual components: Δvp,total = ∑ Δvp,i Assuming that the aerosol is binary and only has two components A and B, the following set of equations apply: C A,0 + γA, xA,0Csat, A, 0 = C A,1 + γA, xA,1Csat, A, 1 CB,0 + γB( 1 - xA,0 ) Csat, B,0 = CB,1 + γB( 1 - xA,1 ) Csat, B,1 Where C is the molarity, Csat is the saturation molar concentration, and the subscripts 0 and 1 indicate

conditions at the initial temperature T0 and the at the higher temperature, respectively. Using the following equation, one can solve for CB :

xA =

CA CA + CB

CB = 1−XXAA CA Using this value of CB and combining the previous equations forms a quadratic equation in terms of mole fractions:

[gγBCsat , B , 1 − γg ACsat , A , 1]x A2 ,1 +[CA , 0 + CB , 0 + γg A , xA , 0Csat , A , 0 + γgB(1 − xB , 0)Csat , B , O + gγ ACsat , A , 1 − gγBCsat , B , 1]xA , 1 + [−CA , 0 − gγ A , xA , 0Csat , A , 0] =0 The Van Laar equation (Smith and Van Ness 1987), shown below, is substituted into the formula used to calculate the total change in volume of the aerosol:

lnγg1 =

A (1+ BAxx21 )2

lnγg 2 =

B (1+ BAxx21 )2

where A and B are constants determined by graphing the total volume change of the aerosol against the mole fraction x. The activity coefficients are calculated using the Van Laar equation above. Because the mole fractions change depending on the volatility of the components, the quadratic equation must be solved again for the mole fractions using the values of the activity coefficients. The process is then repeated until an agreement is reached between the mole fractions and the activity coefficients.

Volume 1 | 2011-2012 | 47

Street Broad Scientific

Research

Procedure For artificial aerosols, mixtures of six monocarboxylic acids were made in ethanol. The monocarboxylic acid mixtures included pentadecanoic acid, palmitic acid, heptadecanoic acid, stearic acid, nonadecanoic acid, and eicosanoic acid, which are C15-C20 monocarboxylic acids. The tracer compound tested was levoglucosan, which is a wood smoke tracer. Solutions of pure monocarboxylic acids, pure levoglucosan, and several different ratios of levoglucosan to acid mixtures were made at least a day in advance before running the experiments. Changing the ratios subsequently changed the mole fractions. For ambient aerosols, samples were collected on glass-fiber filters using a HighVolume air sampler for three weeks on the rooftop of one of the university buildings. The filters were extracted in ethanol and then filtered with a syringe to separate out particles smaller than 0.2 micrometers. Aerosol was generated from a solution made from the filter extract and the mass of the ambient aerosol was estimated from the SMPS measurements. Using the estimated mass, it was possible to calculate the amount of levoglucosan needed to create mixtures with ratios of 0.5:1, 1:1, 1:2, and 1:3.

After the aerosol was generated and stabilized at equilibrium, the thermodenuder was heated up to 60°C for five measurements. It was then unplugged and allowed to cool back down to room temperature, or about 25°C. The change in volume, Δv, was calculated by comparing the reference SMPS measurements with the heated SMPS measurements. By using interpolation, Δv was determined at a specific temperature for each mixture. Levoglucosan was defined as component A, while the monocarboxylic acids or ambient extract were treated as component B. Therefore, the mole fractions were calculated by dividing the number of moles of levoglucosan by the total number of moles in solution.

Results

Change in Aerosol Volume of LevoglucosanMonocarboxylic Acid Mixtures at 35°C

∆ν

�

µm3 cm3

�

Change in Aerosol Volume of LevoglucosanMonocarboxylic Acid Mixtures at 40 °C

∆ν

Levoglucosan Mole Fraction Figure 1

48 | 2011-2012 | Volume 1

Figures 1 and 2 show the change in aerosol volume of levoglucosan-monocarboxylic acid mixtures at 35°C and 40°C for different mole fractions of levoglucosan. The mole fraction of 0 corresponds with an equimolar mixture of the six monocarboxylic acids listed previously in he procedure, and the mole fraction of 1 corresponds with a pure levoglucosan mixture. If the levoglucosan and monocarboxylic acids had no interaction with each other, then the total change in volume for the aerosol would be equal to the sum of the changes in volume of each of the components in the mixture, i.e., the sum of the change in volume of pure levoglucosan and the change in volume of pure monocarboxylic acids. With no interaction among the particles, the change in volume is not dependent on the mole fraction of the individual components and does not need to be corrected with activity coefficients. This can be seen in the graph because the change in volume for three different mole fractions is almost the same. The data collected at 35°C supports the conclusion that there is no interaction among the components in the aerosol.

The change in volume of the pure levoglucosan mixture was measured at 17.6µm³/cm³, and the change in volume of the pure monocarboxylic acid mixture was measured at 40.5µm³/cm³, making the sum, or total change in volume of the aerosol, approximately 58.1µm³/cm³ with no interaction. The total change in volume of the aerosol for all three mole fractions is around 55µm³/cm³. These values are close enough to conclude that the components did not interact with each other and did not affect each other’s volatilities within the aerosol at 35°C. At 40°C, the changes in volume are greater than they were at 35°C, which is expected because the temperature is higher and more particles evaporated to the gas phase. If there was no interaction among the particles, which was the conclusion reached from the previous data set, then the total change in aerosol volume would be the sum of the change in volume of pure levoglucosan, 42.6µm³/cm³, and the

�

µm3 cm3

�

Levoglucosan Mole Fraction Figure 2

Street Broad Scientific

Research Change in Aerosol Volume of LevoglucosanAmbient Extract Mixtures at 35 °C

∆ν

�

µm3 cm3

�

Activity Coefficients of Levoglucosan in Mixture with Ambient Particles at 35 °C

γ

Levoglucosan Mole Fraction

Levoglucosan Mole Fraction Figure 3

change in volume of the pure monocarboxylic acid mixture, 92.6µm³/cm³. This sum is equal to approximately 135.2µm³/cm³if there was no interaction.

Because the data at 35°C shows that there is no interaction between levoglucosan and monocarboxylic acids, it can be said that these compounds do not form a solution.

Although both levoglucosan and monocarboxylic acids are polar solutes and ethanol is a polar solvent, it is expected that these compounds would dissolve. Further work could be done to examine the interaction between levoglucosan and other polar compounds, such as dicarboxylic acids.

Figure 3 shows the change in aerosol volume of levoglucosan-ambient extract mixtures at 35°C for different mole fractions of levoglucosan. Unlike the previous graph of levoglucosan-monocarboxylic acid mixture, these components do form a solution and interact with each other because the change in volume is less than the sum of the changes in volume of the individual compounds. In this aerosol, the sum of the change in volume of the pure levoglucosan mixture, 17.6µm³/cm³ , and the change in volume of the pure ambient particle mixture, 13.8µm³/cm³ , is approximately 31.4µm³/cm³ . All of the data points for the mole fractions lie below 20µm³/cm³ , so the levoglucosan and ambient particles must interact with each other.

According to the data, when the majority of the mixture is ambient particles, the total change in volume drastically decreases. When the majority of the mixture is levoglucosan, the total change in volume nearly matches the change in volume of the pure levoglucosan mixture. The behavior of these mixtures is non-ideal and does not follow Raoult’s law. The data points clearly cannot be fitted to a line. If the behavior had been ideal, then the change in volume would be linearly proportional to the mole fraction, and the data points would lie in a line that connects the points that represent change in volume for the pure mixtures. Since this is not the case, activity coefficients for levoglucosan need to be determined and included in models to correct for the non-ideal behavior. Figure 4 shows the activity coefficients for levoglucosan when in a mixture with par-

Figure 4

ticles from ambient extracts as a function of levoglucosan mole fraction. As established earlier, the behavior of these mixtures is non-ideal, and the activity coefficients are not equal to 1.

Discussion Levoglucosan-Monocarboxylic Acid Mixtures Two sets of data were graphed for the total change in aerosol volume fotr levoglucosanmonocarboxylic acid mixtures, one at 35°C and one at 40°C. It was clear from the data collected at 35°C that levoglucosan and monocarboxylic acids do not interact with each other because they do not form a solution. The aerosol generated from mixtures with different mole fractions of levoglucosan displayed volume changes approximately equal to the sum of the volume changes of levoglucosan and monocarboxylic acids, indicating that there is no interaction among these particles. All individual components partitioned to the gas phase as they normally would in a pure mixture. There was no need to calculate activity coefficients because the volatilities of the compounds were not affected. The mixture of levoglucosan and monocarboxylic acids is an example of a combination of a tracer compound and a group of semi-volatile organic aerosols that do not interact with each other. This suggests the possibility of other tracer compounds and semi-volatile organic aerosol 14 mixtures that also do not affect each other. When completing models and calculations for source apportionment, one should be aware ofone should be aware of the organic aerosols in the atmosphere surrounding the tracer compound and account for their relationships, i.e., whether or not they form a solution and incorporating activity coefficients when applicable. The fact that monocarboxylic acids do not change the volatility of levoglucosan could help support and encourage the use of levoglucosan as a tracer compound because this specific group of organic aerosols does not affect its partitioning.

Levoglucosan-Ambient Extract Mixtures The mixtures of levoglucosan and particles from ambient extracts were discovered to be non-ideal, meaning that the change in aerosol volume could not be graphed as a linear function of the levoglucosan mole fraction. The values of the activity coefficients Volume 1 | 2011-2012 | 49

Street Broad Scientific of levoglucosan determined with these experimental methods and algorithm were not equal to 1, which is expected for nonideal mixtures. The compounds in the ambient extracts do affect the volatility of levoglucosan, so the activity coefficients should be included in source apportionment calculations to account for the effect of the ambient particles on the volume change of thelevoglucosan and to more accurately determine the amount of particles emitted from sources.

Other Studies The data showed that levoglucosan is not completely nonvolatile, since it displayed differences in volume between the reference measurements and heated measurements, which indicate that a portion of it evaporated atthe higher temperature.

At 35°C, the change in aerosol volume of the pure levoglucosan mixture was 17.6µm³/cm³ , and the change in volume increased to 42.6µm³/cm³ at 40°C. This conclusion is consistent with previous studies that suggest that levoglucosan is semi-volatile (Oja and Suuberg 1999, Huffman et al. 2009).

Conclusions and Future Work The goal of this research project was to examine the volatility of a wood smoke tracer, levoglucosan, and the interaction of levoglucosan with artificial and ambient organic semivolatile aerosols with the Integrated Volume Method developed by Saleh and Khlystov (2009). The results showed that levoglucosan partitioned to the gas phase and evaporated at 35°C and 40°C, suggesting the possibility that it is semi-volatile, which has also been suggested by previous work. The data collected at 40°C for levoglucosan-monocarboxylic mixtures is inconclusive, and further experiments are needed to confirm or reject the conclusion drawn from the data collected at 35°C that levoglucosan and monocarboxylic acids do not form a mixture and do not interact with each other. Future work could also include finding the interaction of levoglucosan with other semi-volatile aerosols, such as dicarboxylic acids. Mixtures of levoglucosan and ambient particles did form a solution, and the generated aerosol displayed nonideal behavior. Activity coefficients corresponding with the mole fractions of levoglucosan were calculated. As expected for non-ideal mixtures, none of the activity coefficients were equal to 1. Levoglucosan is only one of many tracer compounds. The Integrated Volume Method can be applied to many other possible mixtures to test the volatility of the tracer compounds and the interaction between components in a variety of mixtures. This project only looked at binary mixtures, so the relationship between levoglucosan and multiple different components is still unknown.

References Baek, J., Park, S. K., Hu, Y., Russell, A. G., 2005: Source apportionment of fine organic aerosol using CMAQ tracers. 2005.

Bowman, F. M., Melton, J. A., 2004: Effect of activity coefficient models on predictions of secondary organic aerosol partitioning. Aerosol Science, 35, 1415-1438.

Cass, G.R., 1998: Organic molecular tracers for particulate air pollution sources. Trends in Analytical Chemistry, 17, 356-366.

Chung, S. H., Seinfeld, J. H., 2002: Global distribution and climate forcing of carbonaceous aerosols. J. Geophys. Research, 107, 4407.

50 | 2011-2012 | Volume 1

Research Hennigan, C. J., Sullivan, A. P., Collett, J. L., Robinson, A. L., 2010: Levoglucosan stability in biomass burning particles exposed to hydroxyl radicals. Geophysical Research Letters, 37, L09806.

Hoffmann, D., Tilgner, A., Iinuma, Y., Herrmann, H., 2009: Atmospheric stability of levoglucosan: a detailed laboratory and modeling study. Environmental Science and Technology, 44(2), 694-699.

Huffman, J. A., Docherty, K. S., Mohr, C., Ulbrich, I. M., Ziemann, P. J., Onasch, T.B., and Jimenez, J. L., 2009: Chemicallyresolved volatility measurements of organic aerosol from different sources. Environmental Science and Technology, 43(1), 5351–5357.

Lewtas, J., 2007: Air pollution combustion emissions: Characterization of causative agents and mechanisms associated with cancer, reproductive, and cardiovascular effects. Mutation Research, 636, 95-133.

Oja, V., Suuberg, E. M., 1999: Vapor pressures and enthalpies of sublimation of D-glucose, Dxylose, cellobiose, and levoglucosan. Journal of Chemical and Engineering Data, 44, 26–29.

Pankow, J. F., 1994: An absorption model of gas/particle partitioning of organic compounds in the atmosphere. Atmospheric Environment, 28, 185-188.

Saleh, R., Khlystov, A., 2009: Determination of activity coefficients of semi-volatile organic aerosols using the integrated volume method. Aerosol Science and Technology, 43:8, 838-846.

Schauer, J. J., Rogge, W. F., Hildemann, L. M., Mazurek, M. A., Cass, G. R., 1996: Source apportionment of airborne particulate matter using organic compounds as tracers. Atmospheric Environment, 30, 3837-3855. 17 Seinfeld, J. H., Pankow, J. F., 2003: Organic atmospheric particulate material. Annual Review of Physical Chemistry, 54, 121-140.

Simoneit, B. R. T., 2002: Biomass burning – a review of organic tracers for smoke from incomplete combustion. Applied Geochemistry, 17(3), 129-162.

Smith, J. M., Van Ness, H. C. (1987). Introduction to Chemical Engineering Thermodynamics, Chemical Engineering Series. New York: McGraw-Hill.

Shrivastava, M. K., Subramanian, R., Rogge, W. F., Robinson, A. L., 2007: Sources of organic aerosol: Positive matrix factorization of molecular marker data and comparison of results from different source apportionment models. Atmospheric Environment, 41, 9353-9369.

Voiland, Adam. “Aerosols: Tiny Particles, Big Impact.” NASA Earth Observatory. NASA, 02 Nov. 2010. Web. 21 June 2011. <http://earthobservatory.nasa.gov/Features/Aerosols/page1.php>.

Street Broad Scientific

Street Broad Scientific

52 | 2011-2012 | Volume 1

Street Broad Scientific

Questions?

Comments?

Submissions? broadstreetscientific@gmail.com

Volume 1 | 2011-2012 | 53