Ca2+ and BMP-6 Signaling Regulate E2F during Epidermal Keratinocyte Differentiation*

The epidermis consists of a squamous epithelium continuously replenished by committed stem cells, which can either self-renew or differentiate. We demonstrated previously that E2F genes are differentially expressed in developing epidermis (Dagnino, L., Fry, C. J., Bartley, S. M., Farnham, P., Gallie, B. L., and Phillips, R. A. (1997) Cell Growth Differ. 8, 553–563). Thus, we hypothesized that various E2F proteins likely play distinct growth regulatory roles in the undifferentiated stem cells and in terminally differentiated keratinocytes. To further understand the function of E2F genes in epidermal morphogenesis, we have examined the expression, regulation, and protein-protein interactions of E2F factors in undifferentiated cultured murine primary keratinocytes or in cells induced to differentiate with Ca2+ or BMP-6 (bone morphogeneticprotein 6). We find similar patterns of E2F regulation with both differentiating agents and demonstrate a switch in expression from E2F-1, -2, and -3 in undifferentiated, proliferating cells to E2F-5 in terminally differentiated keratinocytes. Inhibition of keratinocyte proliferation by transforming growth factor-β1 did not enhance E2F-5 protein levels, suggesting that this response is specific to differentiation rather than reversible cell cycle withdrawal. E2F-5 up-regulation is also accompanied by formation of heteromeric nuclear complexes containing E2F5, p130, and histone deacetylase (HDAC) 1. Overexpression of E2F5 specifically inhibited DNA synthesis in undifferentiated keratinocytes in an HDAC-dependent manner, suggesting that E2F-5·p130·HDAC1 complexes are likely involved in the permanent withdrawal from the cell cycle of keratinocytes responding to differentiation stimuli.

The skin epithelium, or epidermis, provides a barrier between the internal and external regions of the body, is constantly subjected to physical and chemical stress, and consequently has high renewal rates. The epidermis consists of a stratified squamous epithelium composed mainly of keratinocytes at different stages of differentiation (1,2). Within the epidermis, the basal cell layer is attached to a basement membrane (3)(4)(5), is closest to the dermis, and contains the stem cells. These cells have continuous self-renewal potential and are responsible for renewing and maintaining the epithelium (6). Committed basal cells lose their proliferative capacity, detach from the basement membrane, and initiate terminal differentiation. Differentiating cells migrate upwards and form postmitotic suprabasal skin layers. The signals and intracellular networks that dictate changes in keratinocyte proliferation and differentiation are poorly understood. These networks are extremely important, because they ultimately determine proper skin morphogenesis and homeostasis.
Important cellular networks that regulate proliferation and differentiation in a variety of cell types include cyclins, pRb family proteins and E2F factors. The E2F family of transcription factors consists of six known genes that form heterodimers with DP proteins (reviewed in Refs. 7 and 8). E2F factors participate in a broad spectrum of functions, including control of eukaryotic cell proliferation and patterning during early development (7)(8)(9)(10). Multiple mechanisms regulate E2F activity, such as transcriptional regulation, interactions with the pRb family of proteins (pRb, p107, and p130), acetylation (11)(12)(13), and phosphorylation (14 -17). The importance of E2F activity is underlined by its conservation through evolution from invertebrates (18) to mammals. The multiplicity of mammalian E2F proteins, however, suggests tissue-specific functions. For example, E2F-1 is essential for normal thymocyte apoptosis and selection (19 -21), E2F-4 is indispensable for hematopoietic and intestinal epithelial cell maturation (22,23), and E2F-5 is critical for choroid plexus function (24).
An emerging understanding of the role and regulation of E2F factors during development underlines their importance in this process. For example, expression analysis of E2F during mouse organogenesis has revealed dynamic regulation of these genes in a number of tissues, including neurons and epithelia (25,26). Specifically, in the developing epidermis, E2F-4 transcripts are abundant in the early ectoderm. Following the onset of E2F-4 expression, E2F-2 transcripts are first detected in 14.5 days postcoitus epithelium. High levels of E2F-2 expression are maintained in stratified epidermis and localize to the basal cell layer and to the epithelium surrounding the dermal papilla in the hair follicles. These are the two regions that contain undifferentiated, proliferating cells. In contrast, E2F-5 transcripts are first detected in the suprabasal layers only after the epidermis starts stratification and persist with that distribution during the postnatal period (24,26). Thus, a switch in E2F gene expression appears to occur as epidermal keratinocytes mature and migrate into the differentiated compartment and become postmitotic, terminally differentiated cells.
In this report, we explore the roles of E2F proteins in epidermal morphogenesis. Specifically, we address the regulatory pathways and functions of E2F during differentiation in primary cultured murine keratinocytes. These cultures recapitulate differentiation events in vivo, such as up-regulation of suprabasal keratins K1 and K10 (27) and growth arrest of Ͼ95% of cells (28) upon treatment with 1.0 mM extracellular Ca 2ϩ or with BMP-6. We find differential expression and posttranscriptional regulation of a set of E2F proteins in differentiated mouse keratinocytes, which sharply contrast with reported changes in immortalized HaCat human keratinocytes (29). We also report that induction of differentiation in mouse keratinocytes results in formation of E2F-5⅐p130⅐HDAC1 1 complexes, which can trigger keratinocyte entry into quiescence.
Immunoprecipitation Assays-Keratinocyte extracts were prepared from treated or untreated cultures, as indicated in individual experiments. To prepare extracts, cells were lysed in buffer A, as above. 750 g of protein were precleared with protein A-and protein G-Sepharose (Amersham Pharmacia Biotech) and then incubated at 4°C for 16 h with appropriate primary antibodies for immunoprecipitation. Incubation with secondary antibodies (2 h, 22°C) was followed by protein A-or protein G-Sepharose (2 h, 22°C) and extensive washes with lysis buffer A. Sepharose containing immune complexes was dissociated with Laemmli buffer (5 min, 95°) and resolved by SDS-PAGE followed by immunoblotting.
Protein Phosphatase Treatments-Cell lysates were prepared as duplicate samples (50 -100 g), one in buffer A and the other in buffer B (buffer A lacking sodium fluoride and sodium orthovanadate) to prevent inhibition of the phosphatase. Those lysates prepared in buffer B were subsequently incubated with phosphatase (4 units/g of protein; New England Biolabs) for 30 min at 30°C, as suggested by the manufacturer, and analyzed by SDS-PAGE and immunoblotting.
Electrophoretic Mobility Shift Assays-Whole cell extracts were prepared in lysis buffer A. Binding reactions were prepared as described (33) using a 32 P-labeled double-stranded oligonucleotide corresponding to the E2F element in the dihydrofolate reductase promoter (5Ј-CTA GAG CAA TTT CGC GCC AAA CTT GGA TC-3Ј). As a modification of the published protocol (33), binding reactions also contained 8 ng/l of a mutant E2F element oligonucleotide, which contains a C 3 A mutation that abolishes binding to E2F (5Ј-CTA GAG CAA TTT CGA GCC AAA CTT GGA TC-3Ј). We noted that the presence of this mutant oligonucleotide further reduced nonspecific binding without affecting E2F complexes. The binding reactions were allowed to proceed for 45 min on ice and contained 5-15 g of protein. For supershift assays, after the initial 45-min binding reaction, appropriate antibodies were added, and incubation was allowed to proceed for another 45-min period on ice, followed by electrophoresis in 5% nondenaturing polyacrylamide gels (33).
Adenovirus Infections-The E2F-encoding adenoviruses have been described (33) and were generously provided by J. Nevins (Duke University). The infections were conducted at a multiplicity of infection of 50 for 16 h in keratinocyte medium containing 2% serum. At this multiplicity of infection, Ն95% of keratinocytes were infected, as confirmed by 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal) staining of cells infected with a Lac-Z encoding virus. After infection, normal growth medium was added and supplemented, where appropriate, with 1.0 mM CaCl 2 .
[ 3 H]dThd Incorporation into DNA-Keratinocyte cultures were incubated with 1.5 Ci/ml [ 3 H]dThd 5Ј-triphosphate for 2 h, and incorporation into DNA was measured by liquid scintillation counting of trichloroacetic acid-insoluble cell fractions. The 3 H activity was normalized to cell numbers.

Differential Expression of E2F Proteins in Postnatal Epidermal and Extraepidermal
Tissues-The multiplicity of E2F genes in mammalian cells and their regulatory mechanisms suggests that E2F proteins likely fulfill tissue-and/or developmental stage-specific functions. Indeed, expression analysis of the various E2F forms during murine organogenesis has demonstrated developmental stage-and tissue-specific regulation of E2F mRNA abundance (25,26). To further test this idea, we initially examined whether differential expression of E2F proteins is maintained in postnatal tissues, with special attention to the epidermis. The results of these experiments are summarized in Fig. 1 and in Table I. We were able to detect the presence of E2F-1 through -5 in all tissues examined. However, their abundance relative to each other varied from tissue to tissue. Thus, E2F-1 is present at highest levels in brain and lung, is present at moderate levels in heart, liver, and kidney, and is barely detectable in dermis and epidermis. This distribution is suggestive of a possible neuronal-specific function for E2F-1 in the adult brain. In contrast to E2F-1, E2F-2 levels were barely detectable in brain, relative to other tissues. Maximum E2F-2 expression occurred in lung, heart, and kidney, with intermediate E2F-2 levels present in dermis and liver. Epidermal E2F-2 levels were higher than in brain but lower than in all the other tissues examined.
Expression analysis of both E2F-3 forms (E2F-3a and E2F-3b (34,35) indicates high levels in lung and liver, followed by intermediate levels in epidermis and heart and the lowest levels in dermis and brain ( Fig. 1 and Table I). E2F-4 and E2F-5 were fairly abundant in most tissues examined, with the exception of E2F-4 in kidney and dermis and E2F-5 in dermis and epidermis. We conclude from this analysis that differential regulation of E2F protein expression is maintained from embryonic to adult organs. The relatively low levels of all E2F proteins in epidermis may reflect the high content of structural proteins in this tissue, such as keratins, loricrin, and involucrin.

Regulation of E2F Proteins by Terminal Differentiation or by
Reversible Inhibition of Proliferation-The analysis of E2F expression in epidermis included all keratinocyte types present in this tissue. However, the epidermis contains various cell populations, including undifferentiated stem cells, with capacity for self-renewal or differentiation, as well as terminally differentiated keratinocytes. We had previously determined that, in embryonic murine skin, there is a switch in E2F mRNA expression during keratinocyte maturation characterized by downregulation of E2F-2 and up-regulation of E2F-5 transcripts (26). Therefore, as a first step in understanding the role of E2F factors in keratinocyte maturation, we examined changes in protein levels in undifferentiated, proliferating cultured keratinocytes or in their postmitotic, terminally differentiated counterparts.
In culture, keratinocyte differentiation programs are triggered by increasing the extracellular Ca 2ϩ concentration to Ն0.1 mM (28) or by the presence of BMP-6 (36). Given that the signaling cascades activated by Ca 2ϩ and BMP-6 are different and yet converge on activation of differentiation pathways, we reasoned that using these two agents would allow us to examine the effects of differentiation induction on E2F proteins, irrespective of how the differentiation program was activated.
Initial experiments were designed to confirm that our treatments indeed induced up-regulation of differentiation markers and withdrawal from the cell cycle. Withdrawal from the cell cycle was assessed by measuring incorporation of [ 3 H]dThd into newly synthesized DNA. As shown in Fig. 2A, Ca 2ϩ treatment of these cultures for 24 or 48 h inhibited [ 3 H]dThd incorporation by up to 80% relative to untreated cultures, in agreement with previous observations (28). BMP-6 treatment produced similar decreases in DNA synthesis ( Fig. 2A). To corroborate that these treatments were indeed associated with differentiation, we analyzed cell extracts for the presence of keratin 1 and involucrin, which are, respectively, early and late differentiation markers in keratinocytes. As shown in Fig. 2B, involucrin is up-regulated in the presence of elevated Ca 2ϩ (Ն0.1 mM) or BMP-6. Keratin 1 expression is also increased by 0.1 mM Ca 2ϩ or BMP-6 but not by 1 mM Ca 2ϩ , in agreement with previous reports (27).
Having determined that Ca 2ϩ or BMP-treated keratinocyte cultures withdrew from the cell cycle and activated differentiation genes as predicted, we then examined E2F protein expression profiles in undifferentiated and differentiated cells. This analysis revealed dissimilar regulation of E2F proteins during differentiation. Specifically, E2F-1, -2, and -3 protein levels were markedly down-regulated by Ca 2ϩ and BMP-6 ( Fig.  3). The observed down-regulation of E2F-2 is in agreement with the decrease in E2F-2 mRNA we previously observed in differentiated suprabasal murine epidermal layers (26).
Analysis of E2F-4 revealed its relatively high abundance in undifferentiated keratinocytes, mirroring transcript levels in murine epidermis (26), as well as in differentiated cells. In stark contrast, E2F-5 protein levels increase substantially in differentiated cultures relative to actively proliferating cells (Fig. 3). This increase mimics changes in E2F-5 mRNA levels in differentiated suprabasal keratinocytes in vivo (26). The regulation of various E2F family members in primary murine keratinocytes suggests important functional differences between these two factors. Of note are the differences between E2F regulation in this murine model and those reported in immortalized human HaCat keratinocytes induced to differentiate following prolonged culture under serum-free conditions (29). Specifically, E2F-1 protein levels remain constant in differentiated HaCat cells (29), in contrast to their reduction in differentiated primary mouse keratinocytes. However, similar to the mouse model, E2F-1 mRNA is reduced in normal cultured human keratinocytes induced to differentiate with IF-␥ (48 -51), as well as in suprabasal human epidermal layers. 2 The presence of E2F-1 protein, but not mRNA, in human suprabasal layers (29) suggests that the protein may be protected from degradation, perhaps by interaction with pRb (40 -42). Another difference between the mouse and the HaCat model is the marked down-regulation of E2F-5 in differentiated HaCat cells, which contrasts with its up-regulation in differentiated murine keratinocytes. Intriguingly, HaCat cells exhibit changes in E2F proteins in response to serum withdrawal followed by differentiation (43) that resemble those elicited in human keratinocytes by TGF-␤ (44). Because serum withdrawal initially causes entry into quiescence, without immediate activation of differentiation pathways, it is not clear whether the reported changes in E2F factors in differentiated HaCat cells are directly associated with induction of differentiation. Whether the differences we note in E2F regulation in mouse versus human cell types arise from species-specific mechanisms of E2F regulation, pathways for induction of differentiation, or alterations in HaCat growth regulatory networks consequent to immortalization (45,46) has yet to be determined.
Treatment of keratinocytes with TGF-␤ causes reversible cell cycle arrest without activation of differentiation programs (47)(48)(49). Thus, epidermal keratinocytes have the capacity to either withdraw permanently from the cell cycle upon induction of differentiation or reversibly upon treatment with growth inhibitory factors such as TGF-␤. We therefore examined whether the changes in E2F described above were specifically associated with permanent cell cycle withdrawal and differentiation or were simply associated with exit from the cell cycle, irrespective of the initial stimulus. To this end, we compared the changes in E2F proteins in cultures treated TGF-␤1 with those described upon induction of differentiation. Our experiments showed that TGF-␤ treatment decreased DNA synthesis, as expected ( Fig. 2A), and as observed in Ca 2ϩ -or BMP-treated cultures, TGF-␤ also reduced levels of E2F-1, -2, 2 L. Dagnino, unpublished observations.

FIG. 1. Expression of E2F proteins in adult murine tissues.
Extracts from indicated adult mouse tissues (100 g protein/lane) were prepared and resolved on denaturing polyacrylamide gels. After transfer to polyvinylidene difluoride membranes, they were incubated with antibodies against the indicated E2F proteins and detected by enhanced chemiluminescence. The results shown are representative of three experiments. and -3 (Fig. 3). Strikingly, however, the E2F-5 response was completely reversed relative to that triggered by differentiation. That is, treatment with TGF-␤ decreased rather than increased E2F-5 levels.
The described fluctuations in E2F protein levels during the inhibition of keratinocyte proliferation or during induction differentiation are consistent with a model in which E2F-1, E2F-2, and E2F-3 may be needed to maintain the cells in a proliferative state. In contrast, a switch in E2F gene expression

TABLE I Postnatal expression of E2F proteins in murine tissues
Extracts prepared from the indicated tissues (100 g/sample) were resolved by SDS-PAGE and analyzed by immunoblotting with appropriate antibodies. For each E2F protein, ϩ and ϩϩϩϩ, denote, respectively, the tissues with lowest and highest observed levels of a given E2F factor and not relative levels of all E2F proteins within a tissue. The relative levels were estimated from the data shown in Fig. 1.  (n ϭ 12). B, expression of the early and late differentiation markers, keratin 1 and involucrin, respectively, in cultures treated with Ca 2ϩ or BMP-6. Cell lysates were obtained 24 h after initiation of treatment, and total cell protein (50 g/lane) was resolved by SDS-PAGE and analyzed by immunoblots with the indicated antibodies. Note the down-regulation of keratin 1 in cultures treated with 1 mM Ca 2ϩ , which promotes later stages of keratinocyte differentiation, during which keratin 1 but not involucrin is weakly expressed (27).

FIG. 3. Regulation of E2F protein levels during keratinocyte differentiation.
Keratinocytes were cultured for 24 h in the presence of Ca 2ϩ , BMP-6, or TGF-␤ prior to harvesting. Whole cell protein lysates were prepared, resolved by SDS-PAGE (100 g/lane for E2F-1, E2F-3, and DP-2 immunoblots; 50 g/lane for all others), and analyzed with the indicated anti-E2F or anti-DP antibodies. The anti-E2F-3 antibody used recognizes the C terminus of both E2F-3a and E2F-3b, which are indicated. The results show a representative experiment of a total of four assays.
involving E2F-5 up-regulation would be necessary for events associated with terminal differentiation, including irreversible cell cycle arrest.
Post-translational Modifications of E2F Factors in Keratinocytes-In the course of the studies described above, we noted that some E2F proteins exhibit different mobility patterns on denaturing gels, depending on the tissue analyzed ( Figs. 1 and  3). Specifically, in proliferating keratinocytes but not in other tissues examined, E2F-1 is evident as two distinct species, suggestive of post-translational modifications, possibly phosphorylation. To investigate whether the observed presence of various E2F-1 forms is due to differential phosphorylation, we treated keratinocyte extracts with protein phosphatase. We reasoned that differences in mobility caused by phosphorylation would disappear in dephosphorylated proteins, as reported in other systems (11). We observed that, in phosphatasetreated extracts from undifferentiated keratinocytes, E2F-1 bands showed consistent changes in migration properties, indicating that in these keratinocytes, as in other cell types, E2F-1 is a phosphoprotein (Fig. 4). Phosphatase treatment of lysates from differentiated keratinocytes also revealed the presence of phosphorylated E2F-1 species, although the migration of dephosphorylated proteins in these cells differed from that of E2F-1 in undifferentiated keratinocytes (Fig. 4). Thus, other post-translational modifications may occur exclusively in differentiated cells. In contrast to E2F-1, E2F-2 was detected as three different phosphatase-resistant forms specific to keratinocytes and was absent in other tissues analyzed (Figs. 2 and  8). This suggests the presence of tissue-specific modifications in this protein other than phosphorylation. Phosphatase-induced changes in mobility similar to those described for E2F-1 were apparent for E2F-3a, irrespective of whether the cells were exponentially proliferating or differentiated (Fig. 4).
E2F-4 was evident as a series of at least four different forms present in undifferentiated and in differentiated cells, which collapsed into a doublet upon phosphatase treatment. This is consistent with the possible existence of different phosphorylation states, in addition to modifications other than phosphorylation (Fig. 4). Finally, no changes in mobility were apparent in E2F-5, suggesting either that it is not phosphorylated in these cells or that phosphorylation changes may not result in mobility shifts in E2F-5 proteins. To confirm that the phosphatase treatment had been successful in the extracts analyzed for E2F-2 and E2F-5, we subsequently probed those immunoblots for E2F-4, which showed the described collapse of the four species into two in the phosphatase-treated samples (data not shown). We conclude that E2F proteins are post-translationally modified in keratinocytes, although no changes specifically induced by differentiation were apparent. E2F activity is modified by phosphorylation, and at least some E2F forms are substrates of cyclin-dependent kinases (15,17,50,51). Keratinocyte differentiation results in inactivation of cyclin A/cdk2 and cyclin E/cdk2 but not cdk4 (52)(53)(54). The observed phosphorylation of E2F proteins in differentiated keratinocytes is unlikely to involve cdk2 but might be mediated by cdk4. This possibility awaits further investigation Differentiation of Keratinocytes Triggers the Formation of Nuclear E2F-5⅐p130⅐HDAC1 Complexes-A critical factor that determines E2F activity is its ability to bind DNA in multimeric complexes. We hypothesized that the differentiationinduced changes in E2F levels described above would be accompanied by changes in E2F DNA binding activity. Therefore, we conducted band shift assays to determine the nature of the E2F species present in undifferentiated and differentiated keratinocytes. To identify the components of the E2F DNA-binding complexes, we used antibodies specific for members of the pRb family of proteins and various anti-E2F antibodies in supershift assays.
Consistent with previous reports (50,55,56), we detected various E2F complexes in undifferentiated keratinocyte lysates, in which the lower mobility complex A, but not the higher mobility group F, is dissociated in the presence of 0.05% deoxycholate (Fig. 5a). This indicates that the A complex contains other protein components in addition to E2F⅐DP. The lack of effect of deoxycholate on F complexes is consistent with the reported behavior of complexes containing exclusively E2F⅐DP dimers (termed "free" E2F) and demonstrated in various cell types (50,55,56). Further characterization of the A complex using antibody supershift assays revealed that the main components of complex A are pRb and, to a lesser extent, p107 (Fig. 5b).
Treatment of keratinocytes with 1 mM Ca 2ϩ induces substantial changes in low mobility E2F complexes but not in free E2F species. Specifically, the slowly migrating complex A present in exponentially proliferating keratinocytes is substituted by two other complexes in differentiated cells (Fig. 5c, B and C complexes). Differentiation triggers the down-regulation of pRbcontaining complexes, inducing formation of p107-and p130containing species (Fig. 5d, B and C complexes). Thus, in addition to changes in E2F protein levels, differentiation in keratinocytes also triggers changes in protein-protein interactions between E2F and the pRb family. Similar changes oc- FIG. 4. E2F phosphorylation status in proliferating and differentiated keratinocytes. The presence of phosphorylated E2F proteins was assayed by treating cell extracts with protein phosphatase and comparing the mobility of the indicated E2F protein with that observed in untreated extracts. Cultured keratinocytes were harvested 24 h after Ca 2ϩ addition to the culture medium. Replicate extracts were prepared in the presence (control) or absence ( phosphatase-treated samples) of phosphatase inhibitors (sodium fluoride and orthovanadate). Note the greater mobility of phosphatase-treated E2F-1, E2F-3, and E2F-4, indicating their phosphorylation. The arrows marked with E2F-3a and E2F-3b indicate an immunoblot with an antibody raised against the C terminus of E2F-3a and E2F-3b. E2F-3a shows a blot with an antibody raised against the N terminus of E2F-3a only. These blots are representative of experiments conducted three times. curred in cells induced to differentiate with 0.1 mM Ca 2ϩ (data not shown) or with BMP-6 ( Fig. 5f), demonstrating that the up-regulation of p107/E2F and p130/E2F complexes occurs as a consequence of the activation of differentiation programs, irrespective of the initial differentiation signal.
Given that differentiation triggers increased E2F-5 levels and that E2F-5 binds to p130 in a variety of cell types, we next investigated whether E2F-5 is present in p130-containing E2F DNA-binding complexes in differentiated keratinocytes, using supershift assays. We observed substantial changes in complexes B and C, including the appearance of an even lower mobility complex (identified with an asterisk) in the presence of the anti-E2F-5 antibody (Fig. 5e). Thus, E2F-5 proteins that are capable of binding to DNA in differentiated keratinocytes appear to be associated with p130. Gel shift experiments conducted on fractionated extracts indicated that essentially all of the p130⅐E2F complexes were nuclear rather than cytoplasmic, in agreement with the nuclear distribution of p130 (data not shown).
In contrast to E2F-5, less prominent changes were induced by anti-E2F-4 antibodies to the low mobility complexes B and C (Fig. 5e), indicative of a modest contribution of E2F-4 to these complexes. This contrasts with the reported up-regulation of E2F-4⅐p107 complexes in myoblasts induced to differentiate (56) and in immortalized HaCat cells (29), further confirming the existence of cell type-specific mechanisms of E2F regulation and, presumably, function.
To confirm the association between E2F-5 and p130, we also conducted co-immunoprecipitation assays followed by immunoblotting (Fig. 6A). We detected E2F-5 on immunoblots of samples immunoprecipitated with anti-p130 antibodies, and conversely, p130 was present in E2F-5 immunoprecipitates. The availability of E2F-5 may be a limiting factor in the formation of the p130-containing complex, because total cellular levels of p130 in keratinocytes do not appear appreciably altered upon induction of differentiation (Fig. 6B).
A likely role for E2F-5⅐p130 complexes may be transcriptional regulation, such as transcriptional inhibition of genes that promote cell growth. Transcriptional regulation occurs via multiple mechanisms, including modification of histone acetylation status (57)(58)(59)(60). Specifically, transcriptional repression can arise from histone deacetylation. To investigate whether the differentiation-induced E2F-5⅐p130 complexes contained histone deacetylases, we extended the co-immunoprecipitation assays to test for HDAC1 and confirmed its presence in both E2F-5 and p130 immunoprecipitates (Fig. 6A). We propose that HDAC1 activity may mediate inhibitory effects of p130⅐E2F-5 complexes on transcription of various genes, including those associated with proliferation.
Biological Function of E2F Proteins in Keratinocytes-As a next step in understanding the role of individual E2F genes in keratinocyte growth and differentiation, we examined the ability of various E2F factors to modulate proliferative responses, by inducing ectopic expression in keratinocytes. Based on the observations described above, we hypothesized that E2F-1, -2, and/or 3 may be involved in maintaining proliferation, whereas E2F-5 may mediate entry into quiescence. To test this theory, we overexpressed various E2F proteins by adenovirus-mediated gene transfer and measured the ability of infected cultures to incorporate [ 3 H]dThd into newly synthesized DNA. Consistent with our hypothesis, undifferentiated keratinocytes infected with E2F-1 or E2F-2 adenovirus exhibited increased DNA synthesis (Fig. 7A). Because these cultures were maintained under conditions that stimulated proliferation, we speculate that the increased DNA synthesis observed in these cells may arise from a larger proportion of cells entering S phase, possibly because of cell cycle shortening. In analogous experiments on Ca 2ϩ -differentiated keratinocytes, we observed a similar ability of E2F-1 and E2F-2 to induce abnormal DNA synthesis (Fig. 7B). This confirms a possible role for E2F-1 and -2 in maintenance of the proliferative state and the need for down-regulation of these proteins during differentiation. We observed massive death in keratinocytes infected with adenoviruses encoding E2F-3 and were unable to obtain measurements of [ 3 H]dThd incorporation.
In a complementary set of experiments, we infected keratinocytes with E2F-5-encoding adenovirus. We postulated that the changes in E2F-5 levels and formation of E2F-5⅐p130 and/or E2F-5⅐p130⅐HDAC1 might result in an impaired capac- Electrophoresis mobility shift assays were conducted using undifferentiated keratinocyte lysates in the absence or in the presence of 0.05% deoxycholate, which dissociates higher order E2F complexes (A) containing proteins in addition of E2F⅐DP dimers. b, presence of pRb family proteins in E2F DNA-binding complexes in undifferentiated keratinocyte cultures. DNA binding was allowed to proceed for 45 min prior to addition of the indicated antibodies. After 1 h of incubation in the presence of antibody, samples were resolved on nondenaturing polyacrylamide gels. Note the presence of supershifted complexes in the presence of the pRb and p107 antibodies, indicating the presence of those proteins in complex A. c, changes in DNA-binding E2F complexes upon keratinocyte differentiation. Extracts from keratinocytes cultured in the presence of 1 mM Ca 2ϩ for the indicated times were prepared and subjected to binding to a 32 P-labeled oligonucleotide containing the E2F-binding element present in the dihydrofolate reductase promoter. Complexes were resolved on nondenaturing polyacrylamide gels. Note the presence of complex A in undifferentiated cells and its replacement by complexes B and C in differentiated keratinocytes. d-f, E2F DNAbinding complexes in keratinocytes cultured with 1 mM Ca 2ϩ or BMP-6, as indicated, for 24 h. DNA binding was allowed to proceed for 45 min prior to addition of the indicated antibody. After 1 h of incubation in the presence of antibody, samples were resolved on nondenaturing polyacrylamide gels. Each binding reaction contained 15 g of protein extract for Ca 2ϩ -treated cultures, and 25 g of protein for BMP-6treated cultures. Note the presence of the B and C complexes containing p130 and p107 but not pRb in cultures treated with either Ca 2ϩ or BMP-6. The results shown are representative of at least four experiments.
ity to proliferate and/or synthesize DNA. As shown in Fig. 7A and in contrast to the observed response to E2F-1 and -2, ectopic E2F-5 expression in proliferating cells reduced DNA synthesis. Because of the presence of HDAC1 in E2F-5 complexes described above, we also examined whether HDAC activity was involved in this inhibition of DNA synthesis by E2F-5. To this end, we incubated E2F-5 infected cultures with trichostatin A, a histone deacetylase inhibitor. We found that the ability of E2F-5 to inhibit DNA synthesis was abrogated in the presence of trichostatin A (Fig. 7), indicating its dependence on histone deacetylase activity and consistent with a proposed functional role for E2F-5⅐p130⅐HDAC1 complexes in modulation of cell cycle progression in keratinocytes. Differentiated keratinocytes infected with E2F-5 adenovirus were incapable of initiating DNA synthesis (Fig. 7B), further contrasting the biological properties of E2F-1 and -2 with those of E2F-5. The up-regulation of E2F-5⅐p130 and/or E2F-5⅐p130⅐HDAC1 complexes in keratinocytes treated with Ca 2ϩ or BMP-6 and the HDAC-dependent ability of E2F-5 to suppress DNA synthesis in keratinocytes are consistent with a role for E2F-5⅐p130⅐HDAC1 complexes in terminal differentiation of murine keratinocytes. Histone acetylation status controls multiple gene transcription events, and consequently the cellular effects of histone deacetylases and their inhibitors are complex. For example, HDAC inactivation by trichostatin A can give rise to cell cycle re-entry, apoptosis, or differentiation, depending on the cellular context (57,61). With regards to E2F-regulated transcription, the association of HDAC1 with pRb and p130 mediates repression of growth-related E2F-responsive promoters (62,63). This effect is consistent with the presence of HDAC1⅐p130⅐E2F-5 complexes in differentiated keratinocytes and with the abrogation by ectopic E2F-5 of DNA synthesis in proliferating keratinocytes (Fig. 7). Taken together, our observations are consistent with a model in which keratinocytes that receive a differentiation signal up-regulate E2F-5 synthesis, possibly by mechanisms involving increased transcription and translation. Elevated levels of E2F-5 allow for the formation of HDAC1⅐p130⅐E2F-5 complexes, which then repress transcription of genes involved in DNA synthesis and cell cycle progres-FIG. 6. Formation of E2F-5⅐p130⅐HDAC1 complexes in undifferentiated and in differentiated keratinocytes. A, replicate samples of whole cell lysates (750 g/lane) from untreated or Ca 2ϩ -treated (24 h) keratinocyte cultures were immunoprecipitated with the antibodies indicated under the IP column. The immunoprecipitates were dissociated in Laemmli buffer, resolved by SDS-PAGE, and transferred to polyvinylidene difluoride membranes for analysis of co-immunoprecipitating proteins. The membranes were then probed with the indicated antibodies (Immunoblot column). B, Analysis of p130 levels in undifferentiated and in differentiating keratinocytes. Extracts from cultured keratinocytes were prepared at the indicated times after the addition of Ca 2ϩ to the culture medium. Total cellular protein extracts were resolved on denaturing polyacrylamide gels (100 g protein/lane) and transferred to polyvinylidene difluoride membranes. The membranes were incubated with an anti-p130 antibody. Note the presence of p130 as a doublet, corresponding to hyper-and hypophosphorylated forms of this protein. After detection of p130 by enhanced chemiluminescence, we probed the membrane with an antibody against ␤-tubulin to normalize protein loading. Lysates prepared from 293 human embryonic kidney cell extracts were used as positive controls, for comparison. sion, which may include those of the "proliferative" E2F forms, E2F-1, -2, and -3. Indeed, E2F⅐p130 complexes occupy E2F sites in the E2F-1, -2, and -3a promoters in quiescent cells (64 -68), mediating their repression. An intriguing possibility would be that the E2F-5 complexes occupy these promoters, thus causing the observed down-regulation in E2F-1, -2, and -3 genes. Thus, although the events that trigger and regulate epidermal stem cell commitment and differentiation are still poorly defined, it appears that these processes involve tight regulation of expression and activity of the E2F family of transcription factors.