Epidermal terminal differentiation depends on B lymphocyte-induced maturation protein-1 Erna Magnu´sdo´ttir*, Sergey Kalachikov†, Koji Mizukoshi‡, David Savitsky*, Akemi Ishida-Yamamoto‡, Andrey A. Panteleyev§, and Kathryn Calame¶储** *Department of Biological Sciences, Columbia University, New York, NY 10027; †Columbia Genome Center, and Departments of §Dermatology, ¶Microbiology, and 储Biochemistry and Molecular Biophysics, Columbia University College of Physicians and Surgeons, New York, NY 10032; and ‡Department of Dermatology, Asahikawa Medical College, Asahikawa 078-8510, Japan Communicated by Iva S. Greenwald, Columbia University, New York, NY, August 3, 2007 (received for review June 28, 2007)
The cornified layer is a compacted lattice of lipid-embedded corneocytes that provides an organism’s barrier to the external environment. Cornification is the final differentiative step for epidermal keratinocytes and involves dramatic cell condensation before death. Using conditional gene deletion in mice, we identified the transcriptional repressor Blimp-1 (B lymphocyte-induced maturation protein-1) as an important regulator of keratinocyte transition from the granular to the cornified layer. More than 250 genes are misregulated in conditional knockout epidermis, including those encoding transcription factors, signal transduction components, proteinases, and enzymes involved in lipid metabolism. Steady-state mRNA and ChIP analyses of a subset of these genes provide evidence that nfat5, fos, prdm1, and dusp16 are novel direct targets of Blimp-1. Identifying nfat5 as a target of Blimp-1 repression indicates that cornification involves suppression of normal osmotic regulation in granular cells. Consistently, conditional knockout mice have delayed barrier formation as embryos, enlarged granular layer cells and corneocytes, and a morphologically abnormal cornified layer. These studies provide insight into cornification, identifying transcriptional regulatory circuitry and indicating the importance of blocking osmotic homeostasis. cornification 兩 epidermis 兩 prdm1 兩 transcription factors 兩 nfat5
T
he critical barrier between higher organisms and their terrestrial environment is formed by the process of cornification, a complex program of epidermal keratinocyte differentiation, generating a layer of structurally reinforced, dead, anuclear corneocytes embedded into a lipid matrix [reviewed by Candi et al. (1) and Segre (2)]. Although molecular events in cornification are under intense study, its transcriptional regulation is not well understood. We report that B lymphocyte induced maturation protein-1 (Blimp-1) is a critical regulator of maturation of granular layer cells to form corneocytes. Blimp-1 is a transcriptional repressor expressed in multiple embryonic tissues (3) and in a subset of differentiated cells in adult animals. It is required at various stages of early embryonic development in Drosophila melaganoster (4), Xenopus laevis (5), zebrafish (6, 7), and mice (8). In adult mice, it is best characterized in lymphocytes where it is required and sufficient for terminal differentiation of B cells to plasma cells (9). In the T lymphocyte lineage, Blimp-1 is required for peripheral T cell homeostasis and differentiated effector and regulatory function (10, 11). Immunohistochemistry showed abundant Blimp-1 in various epithelial tissues of mice (D. Chang and K.C., unpublished work) (12), and a role for Blimp-1 in sebocyte differentiation was reported recently (13). Here, we explore the role of Blimp-1 in mouse epidermal keratinocytes. Blimp-1 is specifically expressed in granular layer keratinocytes, the most differentiated corneocyte precursors (1). Conditional epidermal deletion of prdm1, the gene encoding Blimp-1, causes not only defective differentiation of sebocytes but also severe defects in terminal differentiation of epidermal keratinocytes, which lead to a delay in the 14988 –14993 兩 PNAS 兩 September 18, 2007 兩 vol. 104 兩 no. 38
formation of the epidermal permeability barrier, hyperkeratinization, and abnormal desquamation. The granular layer of neonatal epidermis is expanded and contains abnormally enlarged keratohyalin granules. More than 250 genes regulated by Blimp-1 in the epidermis were identified, including four direct targets of Blimp-1 repression: nfat5, fos, dusp16, and prdm-1 itself. NFAT5 (nuclear factor of activated T cells 5) is ubiquitously expressed and normally activated during osmotic stress to restore normal water content (14). Our discovery of Blimp-1dependent repression of nfat5 reveals, to our knowledge, previously unknown molecular events involved in the control of water homeostasis during normal cornification. Results Blimp-1 Is Expressed in Terminally Differentiated Keratinocytes.
There is a striking correlation between Blimp-1 expression and terminal differentiation of mouse and human keratinocytes. Blimp-1 protein was expressed in the epidermal granular layer (Fig. 1 A–C) and the inner root sheath of the hair follicle [supporting information (SI) Fig. 5], consistent with previous reports (3, 13). In addition, we found Blimp-1 in the hair follicle companion layer (data not shown) and mature sebocytes (Fig. 1 A and D). Consistent with expression in other tissues (12), Blimp-1 was exclusively nuclear and was identical between mouse and human. Prdm1 Was Deleted in Keratinocytes. To test the role of Blimp-1 in
epidermis, we crossed prdm1flox/flox mice (15) to mice expressing Cre recombinase under the control of the human keratin-14 promoter (K14-Cre) (16) to achieve epidermis-specific deletion of prdm1. Efficient deletion of the floxed sequence was confirmed by immunohistochemical staining, immunoblotting, and quantitative PCR (SI Fig. 6 A–D). Thus, we refer to the prdm1flox/floxCre⫹ mice as conditional knockout (CKO) mice. Blimp-1 CKO Mice Have Phenotypic Evidence of Abnormal Epidermis.
Prdm1flox/floxCre⫹ mice were born in Mendelian ratios and were indistinguishable from prdm1⫹/⫹Cre⫺, prdm1flox/floxCre⫺, and prdm1flox/⫹Cre⫹ littermates. At postnatal days 4–8 the CKO animals developed abnormally wrinkled and scaly skin (SI Fig. 6 E and F). Hair emerged on the surface of the CKO mouse skin 2 days later than in the wild-type littermates (SI Fig. 6 E and F), but by day 10 the CKOs were indistinguishable from controls Author contributions: E.M., A.A.P., and K.C. designed research; E.M., K.M., D.S., and A.I.-Y. performed research; E.M., S.K., K.M., and A.I.-Y. analyzed data; and E.M. and K.C. wrote the paper. The authors declare no conflict of interest. Abbreviations: Blimp-1, B lymphocyte-induced maturation protein-1; CKO, conditional knockout; NFAT, nuclear factor of activated T cells; E(n), embryonic day. **To whom correspondence should be addressed. E-mail: klc1@columbia.edu This article contains supporting information online at www.pnas.org/cgi/content/full/ 0707323104/DC1. © 2007 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0707323104
with respect to hair growth. By 3 weeks postpartum, CKO mice were scratching incessantly, gradually causing ulceration and hair loss (SI Fig. 6 G–I). All of the CKO mice (n ⫽ 14) that lived to 14 weeks of age developed scarring, which was absent in all WT (n ⫽ 16) littermates. Accompanying scratching was the generation of severe splenomegaly and enlarged lymph nodes (data not shown); the degree of enlargement of the lymphoid organs correlated with the degree of scarring. Multiple Defects Are Present in Epidermis of Blimp-1 CKO Animals.
There was also a delay in the initiation of the first hair cycle. At day 29, in the WT skin the hair follicles were already in advanced stages of anagen, whereas the hair follicles in the CKO skin were either still in telogen, the resting phase of the hair cycle, or had just initiated the anagen phase, thus being 4–5 days behind the normal cycle (SI Fig. 7 C and D). Overall, the epithelial compartment in the Blimp-1 CKO animals showed multiple defects in terminal differentiation, including a abnormal cornified layer and hyperkeratinization of the hair follicle infundibulum.
The cornified layer of day-1 CKO neonates was abnormally compacted, lacking the normal ‘‘basket weave’’ appearance (Fig. 1E). This abnormal compaction persisted into adult life where hyperkeratinization was evident by the presence of a large amount of cornified layer-derived debris attached and adjacent to the epidermal surface of the CKO animals (Fig. 1H). On days 4–8 the CKO epidermis was hyperplastic. Staining for the proliferation marker Ki67 at day 8 showed an increase in the number of proliferating basal cells and also spurious proliferation of suprabasal cells (Fig. 1G). Because Blimp-1 is not expressed in basal keratinocytes, this increase in proliferation is probably secondary to other defects of the CKO epidermis. The hyperplastic phenotype was not observed in day-15 or older CKO animals (data not shown). By day 15 the sebaceous glands of the CKO epidermis were significantly enlarged, consistent with a previous report (13) (data not shown). Furthermore, by day 29, the hair follicle infundibulum was hyperkeratinized (Fig. 1H). The combination of enlarged sebaceous glands and hyperkeratinization of the infundibulum is probably responsible for the accumulation of sebum, generating a cyst-like structure that proved to be filled with lipid when stained with the lipophilic dye Oil Red O (SI Fig. 7 A and B).
The Granular Layer Is Abnormally Expanded in CKO Epidermis. Based on the restricted expression of Blimp-1 in the epidermal granular layer and the apparent defects in formation of the cornified layer, we focused further analysis of the CKO animals on maturation of granular layer cells to form corneocytes. The developmental status of the viable layers of the epidermis was assessed by staining for differentiation markers of suprabasal layers. Keratin-1, a marker for all suprabasal cells, showed no difference between control and CKO epidermis in newborn mouse skin (data not shown). However, analysis of loricrin, a marker of the granular layer, revealed that at day 1, the number of loricrin-positive cell layers in the CKO epidermis was twice that of littermate controls (4–5 versus 2–3 layers) (Fig. 2 A). This difference became more prominent on day 8 (Fig. 2B) and persisted over at least 25 days (Fig. 2C). Expansion of the granular layer, at postnatal day 1 in the absence of suprabasal hyperplasia, is consistent with the idea that the cornification of the CKO epidermis is impeded, thus leading to an accumulation of granular layer cells before their progression to the cornfied layer. To explore this defect further, the ultrastructure of the epidermis of CKO and control neonates was compared by using
Magnu´sdo´ttir et al.
PNAS 兩 September 18, 2007 兩 vol. 104 兩 no. 38 兩 14989
CELL BIOLOGY
Fig. 1. Blimp-1 expression in mouse and human epidermis and analysis of epidermal morphology and proliferation in Blimp-1 CKO epidermis. (A–D) Immunohistochemistry showing Blimp-1 expression in granular layer keratinocytes and mature sebocytes in mouse whisker pad (A), mouse day-1 neonate backskin (B), human granular layer keratinocytes (C), and human mature sebocytes (D). (E) Day-1 CKO neonates have a compacted cornified layer. (F) Day-8 CKO neonates have a hyperplastic epidermis and cornified layer accumulation. (G) Immunohistochemistry for Ki67 on day-8 WT and CKO epidermis showed increased number of basal nuclei staining positive in the CKO (black arrowheads) and suprabasal staining in occasional nuclei (white arrowheads). (H) Day-29 CKO mice have a hyperkeratotic hair follicle infundibulum (white arrows), accumulation of cornified layer debris (black arrows), and enlarged sebaceous glands. GL, granular layer; BM, basement membrane; SG, sebaceous gland; IRS, inner root sheath. (Scale bars: 75 m, A; 30 m, B and H; 50 m, C–G.)
Fig. 2. Differentiation marker and ultrastructural analysis of Blimp-1 CKO epidermis. (A–C) Immunofluorescence for loricrin on day-1 epidermis (A), day-8 epidermis (B), and day-25 epidermis (C) showing an increased number of loricrin-positive layers in the CKO epidermis at all time points. The basement membrane is indicated by white lines. (D) Lower-magnification electron micrograph of the epidermis. Basement membrane is indicated by broken lines. Granular layers (GL) are markedly thickened in CKO compared with WT. (E) The marked rectangular areas in D are shown at higher magnification. Larger keratohyalin granule containing profilaggrin (F) and increased numbers of loricrin (L) granules are noted in the CKO. (F and G) Cornified cell envelopes (CE), prelaminar lipid sheets (arrowheads) derived from lamellar granules cornified layer intercellular lamellae (*) are developed in CKO mice and WT mice. (H) Comparison of the size of coneocytes and granular layer cells between CKO and WT epidermis . Granular layer cell boundaries are marked in red. Both granular layer cells and corneocytes appear larger in the CKO epidermis compared with WT. Nu, nucleus. Osmium tetroxide was used for postfixation in D, E, and H, and ruthenium tetroxide was used in F and G. (Scale bars: 30 m, A–C; 2 m, D and H; 500 nm, E; 50 nm, F and G.)
transmission electron microscopy. Consistent with the loricrin expression data, the number of keratohyalin containing cells was greatly increased in the CKO (Fig. 2D). Furthermore, both the size and the number of filaggrin-containing granules (Fgranules) and loricrin-containing granules (L-granules) were increased (Fig. 2 D and E). Interestingly, the size of the F-granules appeared similar in lower granular layer cells of the CKO and WT, but in the CKO the granules continued to increase in size in cells positioned closer to the surface (Fig. 2D), indicating a defect in progression of granular cells to become corneocytes. There were no apparent abnormalities in keratin filaments, desmosomes, lamellar granules, cornified cell envelopes, and cornified layer intercellular lamellae in CKO epidermis (Fig. 2 F and G). Also evident from the electron micrographs was the increased size of both granular layer cells and the cells of the lower cornified layer (Fig. 2H). These results provide strong evidence of impeded progression of granular layer cells to become corneocytes in the Blimp-1 CKO epidermis, characterized by an increased number of granular layer cells and enlargement of both granular layer cells and corneocytes. Barrier Formation Is Delayed in Blimp-1 CKO Embryos. The cornified layer, which comprises the epidermal permeability barrier, forms between embryonic days 16.5 and 18.5 (E16.5 and E18.5) (17). Based on the delayed formation of corneocytes in neonates and older CKO animals and the fact that Blimp-1 is expressed in the granular layer of E17 embryos (3), we tested whether barrier formation was delayed during embryonic development of the CKOs. The presence of a permeability barrier was detected by using an X-gal penetration assay at E17.5 (17). In eight of nine controls significant and nearly complete barrier formation was observed. However, in all eight CKO embryos barrier formation was minimal or completely absent (Fig. 3). Barrier formation was complete in CKO neonates tested a few hours postpartum (data not shown). Thus, although the barrier forms in the absence of Blimp-1, formation is delayed, consistent with delayed cornification after birth. 14990 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0707323104
Blimp-1 Regulates Multiple Genes in Epidermis. To elucidate the
molecular mechanism of Blimp-1 action in epidermal keratinocytes, we performed gene expression profiling on epidermis from day-1 neonates. At this time, scratching and inflammation are not yet present, sebaceous glands have not formed, and potential compensatory effects, because of the adjustment of CKO animals to terrestrial environment, were less likely to be induced. More than 250 genes were differentially expressed between the CKO and littermate control epidermis (SI Table 1). Differentially expressed genes with known functions that have potential roles in keratinocytes are shown in SI Table 2. They were grouped into categories of transcription factors, structural genes of the cornified cell envelope, genes involved in lipid metabolism, genes encoding proteinase cascade components, including the genes encoding the proteases Pace4 (1.89-fold) and MT-SP1 (1.43-fold) that are involved the processing of Filaggrin (18, 19), genes involved in signaling, including dusp16 (1.43-fold), metabolic genes, and genes encoding membrane transporters. Thirteen of these genes, selected to cover a range of differential expression ratios, were further validated by quantitative RTPCR. All but one showed differential expression with P ⬍ 0.05 (SI Table 3).
Fig. 3. Lateral (A) and dorsal (B) views of pups from two different litters. Barrier formation is delayed in Blimp-1 CKO epidermis. Epidermal barrier integrity of the WT and Blimp-1 CKO embryos was assayed on E17.5. The acquisition of barrier in the WT is almost complete, whereas in the CKOs the skin stains blue, indicating the absence of barrier.
Magnu´sdo´ttir et al.
Of particular note, several transcription factors, including nfat5 (1.29-fold), prdm1 (6.77-fold), and fos (1.49-fold), were increased in the CKO, consistent with similar observations in B cells (20) and the concept that Blimp-1 regulates transcriptional cascades. In addition, six late cornified envelope (lce) genes, four of which are known to be specifically expressed in epidermis, require Blimp-1 for normal expression. One other lce not normally expressed in epithelia was elevated in the CKO, possibly as a compensatory mechanism. Interestingly, eight solute carrier membrane transproter genes (slc) showed elevated expression in the CKO epidermis, and of those, the one with the greatest elevation, slc6a12 (3.14-fold), is known to be directly activated by NFAT5. These data reveal that epidermal expression of a large number of genes, including many genes thought to be involved in late keratinocyte differentiation, depend on Blimp-1. nfat5, fos, dusp16, and prdm1 Are Direct Targets of Blimp-1 Repression in Differentiated Keratinocytes. The genes that are misregulated in
Discussion This work identifies the transcriptional repressor Blimp-1 as an important and previously unknown regulator of terminal differentiation in epidermal keratinocytes and provides insights into the transcriptional control of granular cell maturation to corneocytes. More than 250 genes were misregulated in CKO epidermis, revealing Blimp-1-dependent regulation of late cornified envelope genes, osmolyte membrane transporters, tranMagnu´sdo´ttir et al.
Fig. 4. Blimp-1 binds directly to conserved sites in fos, dusp16, nfat5, and prdm1. (A) HaCaT keratinocytes were transduced with a lentiviral vector expressing either GFP alone or GFP-tagged Blimp-1. Analysis for enrichment of selected sequences after ChIP, with an antibody to GFP, was performed with quantitative PCR. The graph shows one representative experiment of three that had similar results. The results are expressed as fold enrichment of signal in Blimp-1-GFP immunoprecipitation over immunoprecipitation of GFP alone. (B) Mouse primary keratinocytes were induced to differentiate for 48 h in 1.2 mM Ca2⫹, and ChIP was performed with an antibody to Blimp-1 or preimmune serum. The graph shows one representative of two similar experiments. The results are expresed as fold enrichment of signal in anti-Blimp-1 immunoprecipitation over that of preimmune serum. *, sites showing significant enrichment.
scription factors, growth factors and cytokines, proteases, and proteins involved in lipid metabolism and signal transduction (SI Table 2). Four direct targets of Blimp-1-dependent repression were identified: nfat5, fos, dusp16, and prdm1 (autoregulation). Horsley et al. (13) demonstrated a role for Blimp-1 in lineage commitment of sebocyte precursors by repressing c-myc. Although the expression studies done by Horsley et al. were mostly consistent with our data (Fig. 1 A–C and SI Fig. 5), they did not report Blimp-1 expression in mature sebocytes (Fig. 1 A and D). Furthermore, the phenotype of the CKO mice reported here is more severe than that reported by Horsley et al. The reasons for these discrepancies are not clear, but may include the usage of different antisera to detect Blimp-1 expression, use of different alleles to obtain epidermis-specific deletion of prdm1, or mouse strain differences, resulting in different genetic backgrounds. A transcript from prdm1, 5⬘ to the deleted region, was highly increased in the CKO in our expression profiles (6.7-fold) (SI Tables 1 and 2 and SI Fig. 8), providing strong evidence that Blimp-1 autorepresses its own expression. There is no evidence that the truncated transcript encodes a protein, because no Blimp-1 protein was detected in sections or immunoblots of CKO epidermis PNAS 兩 September 18, 2007 兩 vol. 104 兩 no. 38 兩 14991
CELL BIOLOGY
the CKO include both direct and indirect targets of Blimp-1 repression. To identify direct targets of Blimp-1 repression, we searched the sequences of all 39 genes listed in SI Table 1 that had elevated expression in the CKO to find matches to the Blimp-1 consensus binding sequence (21), with a preference for sites conserved between mouse and human. Consensus binding sites were found in six genes (dusp16, elf5, fos, nfat5, prdm1, and sprr1a) and analyzed for binding of Blimp-1 in vivo by using ChIP. GFP-tagged Blimp-1 or GFP alone was expressed by lentiviral infection in the human HaCaT keratinocyte cell line, and ChIP was performed with an antibody to GFP. The previously identified Blimp-1 binding site 290 bp 5⬘ of the myc P1 transcriptional start site was a positive control (22); regions lacking any Blimp-1 consensus in the insulin promoter and in the fos gene served as a negative controls. Quantitative PCR analysis of the immunoprecipitated DNA revealed Blimp-1 binding to one site in intron 2 of nfat5 (at position ⫹2120 relative to the transcriptional start site), two conserved sites upstream of fos (⫺3800 and ⫺1895) and one nonconserved in the fos promoter (⫺279), one site in the dusp16 promoter (⫺213) and one in dusp16 intron 1 (⫹28,106), and finally, one site in intron 2 of prdm1 (⫹2274). No binding was observed in the negative controls, in sites in sprr1a and elf5 or in some putative binding sites in the other genes tested (Fig. 4A). To ensure the binding we observed was not an artifact of lentiviral expression, we also performed ChIP for endogenous Blimp-1 on primary mouse keratinocytes induced to differentiate by elevation of calcium levels to 1.2 mM for 48 h. Immunoprecipitations were performed using either an antibody to Blimp-1 (21) or preimmune serum as a negative control. Analysis by quantitative PCR revealed that, consistent with the ChIP performed by overexpression of Blimp-1 in HaCaT cells, Blimp-1 bound to the same conserved sites in intron 2 of nfat5 (⫹3500), the dusp16 promoter (⫺156), and one conserved site upstream of fos (⫺5000). Blimp-1 also bound to one of two conserved consensus binding sites in intron 2 of prdm1 (⫹2870) (Fig. 4B). These data, combined with the microarray and quantitative mRNA analyses, provide stong evidence that nfat5, fos, dusp16, and prdm1 are directly repressed by Blimp-1 in both human and murine keratinocytes.
using a mAb for Blimp-1 that would have detected such a product (SI Fig. 6). A prdm1 probe located directly within the deleted region gave only a background signal in the CKO, consistent with efficient deletion (SI Figs. 6 and 8 and SI Tables 1 and 2). Our data suggest that Blimp-1 autorepression occurs by both direct and indirect mechanisms. Binding of Blimp-1 to a site in intron 2 of the gene (Fig. 4) provides evidence of direct repression, whereas repression of fos is predicted to decrease the levels of AP-1, an activator of prdm1 (23). Although the biological significance of autorepression is not clear, Blimp-1 can be toxic (24), and we have subsequently also observed autoregulation in the T cell lineage (K.C. and L. Cimmino, unpublished work), suggesting a generalized mechanism. Dusp16 appears to be a direct target of Blimp-1 repression (SI Table 2 and Fig. 4). Dusp genes are negative regulators of MAPK signaling (25), and MAPK affects many aspects of keratinocyte differentiation and survival (26). Furthermore, Ancelin et al. (27) found that Blimp-1 bound to dusp2, suggesting Blimp-1 may enhance MAPK signaling by repressing various dusp genes. Blimp-1 represses fos, a component of the AP-1 transcription factor complex (SI Table 2 and Fig. 4). Although fos⫺/⫺ mice have no epidermal phenotype, presumably because of redundancy of Fos-related proteins, mice expressing a keratinocytespecific v-fos transgene show that repression of fos in the epidermis is important (28). v-fos transgenic mice and Blimp-1 CKO mice both have elevated fos levels and have strikingly similar phenotypes (SI Table 2, SI Fig. 6, and Figs. 1 and 2), including hyperplasia, hyperkeratosis, alopecia in areas of mechanical stress, and development of a very prominent granular layer (28). Thus, a part of the Blimp-1 CKO phenotype probably results from a failure to repress fos transcription in the granular layer. Importantly, our work identified transcription factor NFAT5 as a target of Blimp-1 repression during keratinocyte maturation (SI Table 2 and Fig. 4). NFAT5 is ubiquitously expressed and is the only NFAT family member not activated by calcium. Furthermore, it is the only mammalian transcription factor known to be induced in response to hypertonicity (14). NFAT5 activates transcription of osmocompensatory genes, including a number of membrane transporters. These transporters facilitate the flow of compatible osmolytes across the plasma membrane, thereby causing water uptake into the cell, restoring normal osmolarity, and promoting cell survival. This osmotic flux of water concurrently changes cell volume (14). Consistent with lack of Blimp1-dependent repression of nfat5, electron micrographs of the CKO epidermis show that both the granular layer cells and corneocytes have an increased cell size compared with controls, suggesting that normal cell volume regulation is defective (Fig. 2H). Furthermore, mRNA levels for several osmolyte transporters were increased in the CKO epidermis in our expression study, with a previously identified direct target of NFAT5, slc6a12, showing the greatest increase (3.14-fold) (SI Tables 1 and 2). Measurements of water concentration in human epidermis have revealed a dramatic reduction in water content at the granular–cornified layer interface, compared with that of the spinous layer (29, 30), effectively exposing the granular layer cells to constant hyperosmotic stress. Based on our results, we suggest a model in which Blimp-1 would inhibit the normal cellular response to osmotic stress by repressing nfat5, thereby preventing an NFAT5-dependent increase in intracellular osmolytes and the subsequent water uptake. Therefore, differentiating Blimp-1⫺/⫺ keratinocytes would not lose water normally, resulting in the observed cell volume increase (Fig. 2H). Blimp1-mediated transcriptional repression of nfat5 may facilitate osmotic stress-driven death of granular cells and their concomitant cornification as a final step of epidermal keratinocyte differentiation. Indeed, early experiments studying the terminal differentiation of cultured human epidermal cells showed that 14992 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0707323104
exposing the cells to high concentrantions of NaCl effectively promoted cell envelope formation (31). No epidermal abnormalities were reported in nfat5 knockout mice (32), which is consistent with our model, where nfat5 expression would need to be repressed to block a normal hypertonic response at the granular-cornified layer interface. Taken together, our data demonstrate that Blimp-1 plays critical roles in terminal differentiation of epidermal keratinocytes and identify genes that depend on Blimp-1 for their appropriate expression in epidermis. Identification of osmoregulator nfat5 as a target of Blimp-1 repression and elevated expression of both nfat5 and NFAT5 target slc6a12 in CKO epidermis provide a coherent molecular mechanism to explain, at least in part, the delayed barrier formation, enlarged granular layer cells, and abnormal cornified layer observed in CKO mice. These observations suggest the transcriptional control of cellular water content as prerequisite for maturation of granular keratinocytes to corneocytes. The data also provide a basis for future molecular studies into mechanisms of normal skin cornification and skin cornification disorders. Materials and Methods Mice. Prdm1f lox/f lox mice (15) were crossed with K14-Cre mice (16) to generate Prdm1flox/floxK14-Cre⫹ and Prdm1floxc/flox K14Cre⫺ mice. Mice were housed in the Columbia University barrier facility under standard conditions. Human Tissue. Human skin samples were obtained from discarded
tissue. Histology. Human and mouse skin specimens were fixed over-
night in 10% neutral buffered formaldehyde at pH 7.4 before paraffin embedding sectioning and staining in H&E at the histology core facility at Columbia University. Specimens from three animals of each genotype were analyzed at each age. Immunostaining. Details for immunohistocheimisty and immunofluorescence staining are described in SI Text. Electron Microscopy. Small pieces of skin samples were fixed in half-strength Karnovsky fixative, followed by fixation in 1% osmium tetroxide in distilled water. After en bloc staining with uranyl acetate, specimens were dehydrated in ethanol and embedded in Epon812 (Taab, Berkshire, U.K.). Ultrathin sections were stained with uranyl acetate and lead citrate. For visualization of membrane structures in the cornified layer, ruthenium tetroxide postfixation was used as described (33). Microarray Analysis. The mRNA expression profiles of four WT
(prdm1flox/floxCre⫺) and four CKO (prdm1flox/floxCre⫹) neonates were compared by using the Illumina Sentrix MouseRef-8 Expression BeadChip. See SI Text for details. Dye Penetration Assay. Dye penetration assay was performed on embryos or neonates as described (17). Cells and Constructs. Lentivirus was generated by transfecting 293T fibroblasts with packaging vector pCMV-deltaR8.9 (34), pMD.G encoding the vesicular stomatitis virus G protein, for pseudotyping (35), and either FUGW (34) expressing GFP alone or FUW-PRDI-BFI-GFP (36). HaCaT keratinocytes were subsequently infected with the lentivirus and harvested for ChIP 48 h postinfection. Primary mouse keratinocytes were harvested from neonates essentially as described (37) and cultured in Keratinocyte SFM (Invitrogen, Carlsbad, CA). At confluence CaCl2 was added to the media to a final concentration of 1.2 mM for 48 h. Magnu´sdo´ttir et al.
ChIP. ChIP experiments were performed essentially as described
We thank Dr. E. Fuchs (The Rockefeller University, New York) for K14-Cre mice; Dr. A. Christiano and members of her laboratory at Columbia University, H. Bazzi, and H. Kim for advice, discussions, and
reagents; and members of K.C.’s laboratory, members of Dr. R. Morris’s laboratory at Columbia University, Dr. B. Reizis, Jose Galan, and Dr. A. Engelhard for discussions and advice. This work was supported by National Institutes of Health Grants RO1AI50659 and RO1AI43576 (to K.C.), National Institutes of Health Grant K01AR02204 (to A.A.P.), and National Institutes of Health Grant P30AR044535 (to Columbia University Skin Disease Research Center).
Candi E, Schmidt R, Melino G (2005) Nat Rev Mol Cell Biol 6:328–340. Segre J (2003) Curr Opin Cell Biol 15:776–782. Chang DH, Cattoretti G, Calame KL (2002) Mech Dev 117:305–309. Ng T, Yu F, Roy S (2006) Dev Genes Evol 216:243–252. de Souza FS, Gawantka V, Gomez AP, Delius H, Ang SL, Niehrs C (1999) EMBO J 18:6062–6072. Baxendale S, Davison C, Muxworthy C, Wolff C, Ingham PW, Roy S (2004) Nat Genet 36:88–93. Wilm TP, Solnica-Krezel L (2005) Development (Cambridge, UK) 132:393–404. Vincent SD, Dunn NR, Sciammas R, Shapiro-Shalef M, Davis MM, Calame K, Bikoff EK, Robertson EJ (2005) Development (Cambridge, UK) 132:1315–1325. Shapiro-Shelef M, Calame K (2005) Nat Rev Immunol 5:230–242. Kallies A, Hawkins ED, Belz GT, Metcalf D, Hommel M, Corcoran LM, Hodgkin PD, Nutt SL (2006) Nat Immunol 7:466–474. Martins GA, Cimmino L, Shapiro-Shelef M, Szabolcs M, Herron A, Magnusdottir E, Calame K (2006) Nat Immunol 7:457–465. Angelin-Duclos C, Cattoretti G, Lin KI, Calame K (2000) J Immunol 165:5462– 5471. Horsley V, O’Carroll D, Tooze R, Ohinata Y, Saitou M, Obukhanych T, Nussenzweig M, Tarakhovsky A, Fuchs E (2006) Cell 126:597–609. Ho SN (2003) Arch Biochem Biophys 413:151–157. Shapiro-Shelef M, Lin KI, McHeyzer-Williams LJ, Liao J, McHeyzer-Williams MG, Calame K (2003) Immunity 19:607–620. Vasioukhin V, Degenstein L, Wise B, Fuchs E (1999) Proc Natl Acad Sci USA 96:8551–8556. Hardman MJ, Sisi P, Banbury DN, Byrne C (1998) Development (Cambridge, UK) 125:1541–1552. Pearton DJ, Nirunsuksiri W, Rehemtulla A, Lewis SP, Presland RB, Dale BA (2001) Exp Dermatol 10:193–203. List K, Szabo R, Wertz PW, Segre J, Haudenschild CC, Kim SY, Bugge TH (2003) J Cell Biol 163:901–910.
20. Shaffer AL, Lin KI, Kuo TC, Yu X, Hurt EM, Rosenwald A, Giltnane JM, Yang L, Zhao H, Calame K, Staudt LM (2002) Immunity 17:51–62. 21. Kuo TC, Calame KL (2004) J Immunol 173:5556–5563. 22. Lin Y, Wong K, Calame K (1997) Science 276:596–599. 23. Ohkubo Y, Arima M, Arguni E, Okada S, Yamashita K, Asari S, Obata S, Sakamoto A, Hatano M, O-Wang J, et al. (2005) J Immunol 174:7703–7710. 24. Messika EJ, Lu PS, Sung YJ, Yao T, Chi JT, Chien YH, Davis MM (1998) J Exp Med 188:515–525. 25. Liu Y, Shepherd EG, Nelin LD (2007) Nat Rev Immunol 7:202–212. 26. Eckert RL, Efimova T, Dashti SR, Balasubramanian S, Deucher A, Crish JF, Sturniolo M, Bone F (2002) J Invest Dermatol Symp Proc 7:36–40. 27. Ancelin K, Lange UC, Hajkova P, Schneider R, Bannister AJ, Kouzarides T, Surani MA (2006) Nat Cell Biol 8:623–630. 28. Greenhalgh DA, Rothnagel JA, Wang XJ, Quintanilla MI, Orengo CC, Gagne TA, Bundman DS, Longley MA, Fisher C, Roop DR (1993) Oncogene 8:2145–2157. 29. Warner RR, Myers MC, Taylor DA (1988) J Invest Dermatol 90:218–224. 30. Caspers PJ, Lucassen GW, Carter EA, Bruining HA, Puppels GJ (2001) J Invest Dermatol 116:434–442. 31. Rice RH, Green H (1979) Cell 18:681–694. 32. Lopez-Rodriguez C, Antos CL, Shelton JM, Richardson JA, Lin F, Novobrantseva TI, Bronson RT, Igarashi P, Rao A, Olson EN (2004) Proc Natl Acad Sci USA 101:2392–2397. 33. Hou SY, Mitra AK, White SH, Menon GK, Ghadially R, Elias PM (1991) J Invest Dermatol 96:215–223. 34. Lois C, Hong EJ, Pease S, Brown EJ, Baltimore D (2002) Science 295:868–872. 35. Naldini L, Blomer U, Gallay P, Ory D, Mulligan R, Gage FH, Verma IM, Trono D (1996) Science 272:263–267. 36. Kuo TC-C (2005) PhD thesis (Columbia University, New York). 37. Pirrone A, Hager B, Fleckman P (2005) Methods Mol Biol 289:3–14.
(4). See SI Text and SI Tables 4 and 5 for details.
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.
CELL BIOLOGY
19.
Magnu´sdo´ttir et al.
PNAS 兩 September 18, 2007 兩 vol. 104 兩 no. 38 兩 14993