The role of E2F4 in adipogenesis is independent of its regulatory activity

Rebecca L. Landsberg, Julia E. Sero, Paul S. Danielian, Tina L. Yuan, Eunice Y. Lee, and Jacqueline A. Lees*

Center for Cancer Research, Massachusetts Institute of Technology, Cambridge, MA 02139

Communicated by David E. Housman, Massachusetts Institute of Technology, Cambridge, MA, December 31, 2002 (received for review December 20, 2002) The and pocket families are known to play an impor- are repressed and are then replaced by the activating E2Fs tant role in the regulation of both cellular proliferation and , -2, and -3, as cells reenter the cell cycle (12). Importantly, terminal differentiation. In this study, we have used compound E2F cells that lack either E2F4 and or p107 and p130 have a and pocket protein mutant mouse embryonic fibroblasts to dissect defect in their ability to exit the cell cycle in response to a variety the role of these in adipogenesis. This analysis shows that of growth arrest signals (13–15). In the case of p107Ϫ/Ϫ;p130Ϫ/Ϫ loss of E2F4 allows cells to undergo spontaneous differentiation. cells, this defect correlates with a failure to mediate the normal The ability of E2F4 to prevent adipogenesis seems to be quite repression of certain E2F-responsive genes (15, 16). The repres- distinct from the known properties of E2F. First, it can be separated sive E2F complexes are also required for normal development. from any change in either E2F-responsive expression or cell Mice lacking E2F4 or E2F5 display distinct developmental cycle regulation. Second, it is a specific property of E2F4, and not defects in one or a few tissues that in each case result in neonatal other E2Fs, and it occurs independently of E2F4’s ability to interact lethality (17–19). It is currently unclear whether the develop- with pocket proteins. In addition, E2F4 loss does not override the mental defects reflect a requirement for E2F4 and͞or E2F5 in differentiation defect resulting from pRB loss even though it cell cycle control or whether these proteins play a more direct -completely suppresses the proliferation defect of Rb؊/؊ mouse role in the differentiation process. In contrast, there is consid embryonic fibroblasts. This finding definitively separates the erable evidence that the essential requirement for p107 and p130 known, positive role of pRB in adipogenesis from its cell cycle in long bone development is due to their role in mediating cell function and shows that this pocket protein is required to act cycle exit (20, 21). downstream of E2F4 in the differentiation process. pRB is also an important developmental regulator (as re- viewed in refs. 2 and 22). pRB-deficient embryos die in mid- he E2F transcription factors play a key role in regulating the gestation because of a combination of inappropriate prolifera- Tcell cycle by controlling the transcription of genes encoding tion and apoptosis in a wide variety of tissues. These defects are key components of the cell cycle and DNA replication machinery suppressed by the loss of either E2F1 or , indicating that (reviewed in refs. 1 and 2). The activity of E2F is controlled by they result from the inappropriate release of the activating E2Fs. the that includes the retinoblastoma However, the analysis of Rb;E2f1 and Rb;E2f3 double mutant protein (pRB), a known tumor suppressor, and its relatives p107 embryos reveals a key role for pRB in development that is and p130. The pocket proteins bind to E2F and inhibit the unrelated to the regulation of E2F. In vitro differentiation assays activation of E2F-responsive genes through two mechanisms support a role for pRB in myogenesis, osteogenesis, and adipo- (reviewed in refs. 1 and 2). First, their binding interferes with the genesis, and pRB can function as a coactivator for transcription function of the E2F transactivation domain. Second, the result- factors that promote these differentiation processes (as reviewed ant E2F⅐pocket protein complexes can actively repress transcrip- in ref. 22). tion through recruitment of histone deacetylase and methylase In this study, we investigated the role of the repressive E2F activities. Consequently, the pocket proteins are key regulators complexes in terminal differentiation. We have focused on adipogenesis because this pathway is well defined and it involves of the G1 to S transition. The E2F proteins can be divided into three subgroups based both a proliferation and a differentiation phase. Specifically, on structural and functional homology (reviewed in refs. 1 and confluence-arrested mouse embryonic fibroblasts (MEFs) can 2). E2F6 is the sole member of one subgroup. It does not interact be triggered to undergo differentiation by treatment with the with the pocket proteins and has been shown to play a role in hormone insulin, the glucocorticoid agonist dexameth- vertebrate patterning through interaction with the mammalian asone, and the cAMP phosphodiesterase inhibitor methyl- isobutylxanthine (23). After treatment, the cells express polycomb complex (3, 4). The second E2F subgroup includes ͞ ␤ ͞ ␤ ͞ ␦ E2F1, , and E2F3a. These E2Fs are specifically regulated CCAAT enhancer-binding protein (C EBP ) and C EBP by pRB, and not p107 and p130, in vivo. Combined analysis of and then undergo one to two rounds of mitotic clonal expansion before arresting again and inducing expression of the C͞EBP␣ both overexpressing and deficient cells indicates that they play a ␥ ␥ key role in activating E2F-responsive genes and thereby pro- and peroxisome proliferator-activated receptor (PPAR ) tran- moting S-phase progression (5–7). E2F3b, a recently identified scription factors that are required for lipid accumulation and second product of the E2F3 gene, is most closely related to E2F1, adipocyte induction. The E2F and pocket protein complexes -2, and -3a, but its biological properties remain to be established have already been linked to adipogenesis at several different levels. First, hormonal induction correlates with the dissociation (8–10). ⅐ The final E2F subgroup, comprising E2F4 and E2F5, is the of the repressive E2F4 p130 complex and the expression of the focus of this study. E2F5 binds specifically to p130 in vivo activating E2Fs (24, 25). E2F activity induces clonal expansion PPAR␥ (reviewed in refs. 1 and 2). In contrast, E2F4 associates with and the simultaneous transcriptional activation of ,an E2F-responsive gene (25). Consistent with this scheme, pRB, p107, and p130 and comprises the majority of the Ϫ Ϫ Ϫ Ϫ p107 / ;p130 / MEFs have an increased propensity to undergo E2F⅐pocket protein complexes in vivo (11). These E2Fs are thought to play a key role in the repression of responsive genes through the recruitment of pocket proteins and their associated Abbreviations: pRB, ; MEF, mouse embryonic fibroblast; C͞EBP␤, histone deacetylases. Chromatin immunoprecipitation assays CCAAT͞enhancer-binding protein ␤; PPAR␥, peroxisome proliferator-activated receptor ␥; show that the E2F4͞5⅐p107͞p130 complexes specifically associ- TKO, E2f4Ϫ/Ϫ;p107Ϫ/Ϫ;p130Ϫ/Ϫ; DKO, p107Ϫ/Ϫ;p130Ϫ/Ϫ. ͞ ate with E2F-responsive promoters during G0 G1 when these *To whom correspondence should be addressed. E-mail: [email protected].

2456–2461 ͉ PNAS ͉ March 4, 2003 ͉ vol. 100 ͉ no. 5 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0138064100 Downloaded by guest on October 1, 2021 hormone-induction adipogenesis (26). Second, pRB functions as a coactivator for C͞EBP␤ and thereby plays a positive role in adipogenesis (24, 27, 28). Finally, C͞EBP␣ and PPAR␥ can repress E2F activity and may thereby contribute to the switch from the clonal expansion to the differentiation phase (29–31). Materials and Methods Cell Culture and Analysis. MEFs were generated as described (6, 32). For each genotype, two to four independently derived, passage 4 MEF lines were analyzed. Confluent monolayers were fed every other day with MEF media (DMEM containing 10% FCS, 1% penicillin͞streptomycin, 1% glutamine) in either the absence (cell cycle and spontaneous differentiation assays) or presence (hormone-induced adipogenesis) of 1 ␮M dexameth- asone (Sigma), 0.5 mM 3-isobutyl-1-methylxanthine (Sigma), and 5 ␮g͞ml insulin (Sigma). Lipid accumulation was assessed by quantification of Oil Red O staining (33). Cell cycle progression was determined by Northern blotting (6) or 24-h incubation in media containing 3 mg͞ml BrdUrd and 0.3 mg͞ml 5-fluorode- oxyuridine (FdUrd) to assess DNA synthesis.

E2F4 Expression and Immunoblotting. pBabe-E2F4⌬C was gener- ated by digesting pCMV-E2F444 (34) with BamHI and HindIII and subcloning into the pBABE vector. 293 T cells were infected at 60% confluence with 1 ␮g pBabe vector and 1 ␮g of pCL-Eco packaging construct (35) by using FuGENE 6 (Roche Diagnos- tics). The media was replaced 24 h later, and supernatants were harvested at 48 and 72 h, filtered, and used to infect MEFs with 8 ␮g polybrene. Infected cells were selected by culturing in 2 ␮g͞ml puromycin for 48 h. Western blotting assays were per- formed by using 50 ␮g whole cell lysates and the LLF4-1 monoclonal antibody (11). Results It has previously been shown that the loss of E2F4 or p107͞p130 promotes the ability of primary fibroblasts to undergo adipo- genesis in response to hormone induction (25, 26). We wished to determine whether this finding reflects the shared role of these proteins in the formation of repressive E2F complexes and whether it correlates with, or can be separated from, regulation Fig. 1. E2F4 pocket protein repression is required for repression of adipo- of cell cycle exit. To address these questions, we intercrossed the genesis. (A) Cartoon indicating the spectrum of E2F complexes that exist in normal cells and the components (denoted by the Xs) that are affected by the E2f4 (19) and p107;p130 (20) mutant mouse strains to allow the Ϫ Ϫ combined mutation of E2f4, p107, and p130.(B)WT(ϩ͞ϩ), E2f4 / , DKO, and generation of compound E2f4Ϫ/Ϫ;p107Ϫ/Ϫ;p130Ϫ/Ϫ (TKO) mu- TKO MEFs were induced to differentiate for 2 wk, stained with Oil Red O, and CELL BIOLOGY tant MEFs. Fig. 1A depicts how these mutations would affect the quantitated. (C) MEFs of all four genotypes maintained at confluence for 3 wk, various E2F and pocket protein complexes that exist in normal stained with Oil Red O, and quantitated. (D)WT(ϩ͞ϩ), E2f4Ϫ/Ϫ, and TKO MEFs cells. The properties of the TKO cells were directly compared stained with Oil Red O after 6 wk at confluence. with those of the parental p107Ϫ/Ϫ;p130Ϫ/Ϫ (DKO) and E2f4Ϫ/Ϫ mutant MEFs. tained at confluence without hormone induction (Fig. 1 C and E2F4-Deficent MEFs Undergo Spontaneous Adipogenesis. We first D). Importantly, this spontaneous differentiation was observed Ϫ/Ϫ examined the ability of the TKO, DKO, and E2f4Ϫ/Ϫ MEFs to with every preparation of E2f4 MEFs but never with the WT differentiate into adipocytes by using the standard hormone control MEFs (Fig. 1D and data not shown). Given this finding, induction conditions. After 2 wk of induction, the degree of we also tested the ability of DKO and TKO MEFs to accumulate lipids without hormone induction. In this assay, DKO cells differentiation was determined by staining with Oil Red O, a Ϫ/Ϫ lipophillic dye that can be extracted and quantified to assess for behaved differently from E2f4 MEFs in that they consistently induction efficiency. As reported (25, 26), the absence of either failed to undergo spontaneous adipogenesis. This result suggests that E2F4 and p107͞p130 repress adipocyte induction through E2F4 or p107 and p130 led to a 2- to 5-fold increase in distinct mechanisms. This hypothesis is supported by the phe- adipogenesis (Fig. 1B). Strikingly, the combined loss of all three notype of the TKO MEFs. These cells were highly predisposed proteins yielded a significant increase in lipid accumulation Ϫ/Ϫ to undergo spontaneous differentiation, consistently yielding (10-fold). The relative effects of the E2f4 , DKO, and TKO many more Oil Red O-positive cells than the E2f4Ϫ/Ϫ MEFs (Fig. genotypes were observed with numerous preparations of MEFs, 1 C and D). Taken together, these data indicate that E2F4 plays indicating that it is highly reproducible. Thus, E2F4 and the a gate-keeping role to block spontaneous adipocyte differenti- pocket proteins seem to have additive, negative effects on the ation whereas p107͞p130 inhibit induced differentiation in a differentiation process. Clearly, this effect could result from distinct manner. their contribution to either the same or to distinct regulatory processes. E2F4’s Role in Adipogenesis Can Be Separated from Its Role in Cell Ϫ Ϫ During these studies we discovered that E2f4 / MEFs were Cycle Regulation. The induction of adipogenesis in vitro involves predisposed to undergoing adipogenesis when they were main- transitions in cell cycle regulation from confluence arrest, to

Landsberg et al. PNAS ͉ March 4, 2003 ͉ vol. 100 ͉ no. 5 ͉ 2457 Downloaded by guest on October 1, 2021 Fig. 2. Loss of repression leads to defects in cell cycle control and deregulation of E2F target . (A)WT(ϩ͞ϩ), E2f4Ϫ/Ϫ, DKO, and TKO MEFs were grown and analyzed at 0, 4, 6, and 8 days of confluence for expression of E2F responsive genes. Gene expression was quantitated by using the loading control (ARP PO). (Bi)WT(ϩ͞ϩ), (Bii) DKO, (Biii) TKO, and (Biv) E2f4Ϫ/Ϫ MEFs were grown and maintained at confluence for 8 days and then pulsed with BrdUrd for 24 h. The percentage of cycling cells was based on counting 300 DAPI (4Ј,6-diamidino-2-phenylindole)-positive nuclei for each genotype.

mitotic clonal expansion, and then terminal cell cycle exit. The WT cells or modulate (either qualitatively or quantitatively) the p107Ϫ/Ϫ;p130Ϫ/Ϫ MEFs have a well documented defect in their gene expression defect that results from the loss of p107͞p130. ability to arrest in response to certain growth conditions (14, 15). In parallel with the gene expression studies, we also monitored It has previously been suggested that this finding could account the confluent monolayers for evidence of inappropriate cell for, or at least contribute to, the altered differentiation proper- cycle entry by culturing them in the presence of BrdUrd for 24 h ties of these cells (26). Given this finding, we wished to establish (Fig. 2B). As expected, there was no evidence of proliferating whether there is any correlation between the differentiation and cells in the WT controls. In contrast, we observed sporadic cells, cell cycle phenotypes of the various mutant genotypes. or pairs of cells, within the E2f4Ϫ/Ϫ monolayer that were Because the in vitro adipogenesis is thought to require con- BrdUrd-positive. Importantly, this cell division specifically oc- fluence arrest (reviewed in ref. 23), we wished to establish the curred in E2f4Ϫ/Ϫ cells that were subjected to confluence and not cell cycle properties of the E2f4Ϫ/Ϫ, DKO, and TKO MEFs under serum starvation arrest (Fig. 2B; refs. 17 and 19). Given this these conditions (Fig. 2). First, we determined how the mutation specificity, and the lack of any detectable gene-expression defect, of the E2F and pocket proteins affected the regulation of we believe that the presence of proliferating cells within the E2F-responsive genes (Fig. 2A). As confluence arrest pro- confluent E2f4Ϫ/Ϫ monolayer is indicative of cells that are ceeded, the levels of known E2F-responsive mRNAs declined undergoing the clonal expansion phase of the adipocyte differ- with identical kinetics in the WT and E2f4Ϫ/Ϫ MEFs (Fig. 2A). entiation process as opposed to evidence of an intrinsic cell cycle This result is entirely consistent with previous conclusions that defect. This hypothesis is supported by our finding that E2f4 there are no defects in the regulation of either E2F target genes mutation had no detectable effect on the magnitude of the cell or cell cycle exit and reentry in E2f4Ϫ/Ϫ MEFs in serum- cycle defect in the DKO MEFs. Even after 8 days of confluence starvation͞restimulation assays (17, 19). In contrast, we found arrest, a similar high proportion of the DKO (60%) and TKO that specific E2F-responsive mRNAs were maintained at high (64%) monolayers underwent inappropriate DNA synthesis levels in the DKO MEFs even 8 days after confluence was during the 24 h labeling window (Fig. 2B). The arrest defects of established (Fig. 2A). This finding correlates with the known the DKO and TKO MEFs were also similar to one another at defect in the down-regulation of a subset of E2F-responsive shorter BrdUrd labeling times (data not shown). We therefore genes in serum-starved DKO MEFs (15). Significantly, the conclude that the DKO and TKO MEFs display indistinguish- pattern of target gene expression in the TKO MEFs was identical able defects in their ability to down-regulate E2F-responsive ͞ to that observed in the DKO MEFs (Fig. 2A). Thus, mutation genes and arrest in G0 G1 in response to confluence arrest. of E2f4 does not alter the regulation of E2F-responsive genes in Importantly, these data strongly suggest the role of E2F4 in

2458 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0138064100 Landsberg et al. Downloaded by guest on October 1, 2021 differentiation can be separated from its role in cell cycle regulation. E2f4Ϫ/Ϫ cells have an increased propensity to un- dergo spontaneous adipogenesis without intrinsic defects in the regulation of E2F-responsive genes and therefore cell cycle entry͞exit. Moreover, E2F4 loss greatly increases the differen- tiation of DKO MEFs without modulating their cell cycle phenotype.

p130 Is the Predominant Mediator of the p107͞p130 Differentiation Effect. We wished to establish whether the enhanced differenti- ation of TKO versus E2f4Ϫ/Ϫ MEFs depends on the combined mutation of p107 and p130 or whether it can be attributed to the mutation of one of these genes. To address this question, we compared the differentiation phenotypes of E2f4Ϫ/Ϫ;p107Ϫ/Ϫ and E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEFs with those of E2f4Ϫ/Ϫ and TKO controls (Fig. 3A and data not shown). p107 loss had no effect on the differentiation of E2f4Ϫ/Ϫ MEFs. In contrast, the E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEFs displayed a wide range of phenotypes in this assay: some were equivalent to E2f4Ϫ/Ϫ MEFs whereas others phenocopied the TKO MEFs (Fig. 3A and data not shown). This degree of variation was specifically observed with E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEFs and not those derived from any other mutant genotype. Moreover, it only arose between E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEF lines isolated from different embryos and never the same embryo (data not shown). It has recently been reported that the p130 mutant allele used in this study has some residual activity (21) and strain-specific modifiers of p130 have been described (36). It therefore seems likely that this phenotypic range arises because the mixed (C57BL͞6 ϫ 129S͞v) genetic background of the MEFs modulates the residual p130 activity such that it is either above or below a critical threshold level. Importantly, at least in certain settings, our data show that the p130 mutation was sufficient to enhance the E2f4Ϫ/Ϫ differen- tiation phenotype as well as the combined mutation of p107 and p130. Because, p107 loss has no detectable effect in these assays, this strongly suggests that p130 must be the predominant medi- ator of the p107͞p130 differentiation role. We also examined the cell cycle properties of E2f4Ϫ/Ϫ;p107Ϫ/Ϫ and E2f4Ϫ/Ϫ;p130Ϫ/Ϫ confluent monolayers, as judged by both BrdUrd incorpora- tion and E2F-responsive gene expression (data not shown). Regardless of their differentiation phenotype, all preparations of E2f4Ϫ/Ϫ;p107Ϫ/Ϫ and E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEFs had cell cycle phenotypes that were indistinguishable from that of the E2f4Ϫ/Ϫ CELL BIOLOGY MEFs. This suggests that the ability of p130 mutation to enhance the differentiation phenotype of the E2F4-deficient cells occurs Fig. 3. Analysis of E2F4-mediated adipocyte repression. (A) independently of p130’s role in cell cycle control. E2f4Ϫ/Ϫ;p107Ϫ/Ϫ and E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEFs were induced to differentiate in response to hormones for 2 wk, stained with Oil Red O, quantitated, and Pocket Protein Binding Is Not Required for the Repression of Adipo- compared with WT (ϩ͞ϩ), E2f4Ϫ/Ϫ, and TKO MEFs. (B) WT, E2f5Ϫ/Ϫ, E2f4Ϫ/Ϫ, genesis by E2F4. E2F4 and -5 have many overlapping properties and E2f4Ϫ/Ϫ;E2f5Ϫ/Ϫ MEFs were induced to differentiate in response to in vivo including regulation by p130 (reviewed in ref. 2). We hormones for 2 wk and analyzed as in A. (C) Full-length or a C-terminal Ϫ Ϫ therefore tested whether E2F5 inhibits adipogenesis in a manner truncated (E2F4⌬C) E2F4 were expressed in control and E2f4 / MEFs by analogous to E2F4 (Fig. 3B). Significantly, E2F5 loss had little retroviral-mediated infection. (D) Infected cells were induced by hor- mones to differentiate for 2 wk and analyzed as above. (E)WT(ϩ͞ϩ), or no effect on the differentiation phenotype of either wild-type Ϫ/Ϫ Ϫ/Ϫ Ϫ/Ϫ Ϫ/Ϫ Ϫ/Ϫ E2f4 , Rb , and E2f4 ;Rb MEFs were differentiated in response to or E2f4 MEFs. Thus, the inhibition of adipogenesis seems to hormones for 2 wk and analyzed as above. be a unique property of E2F4 and not E2F5. It was important to establish whether the increased differen- tiation of E2f4Ϫ/Ϫ MEFs results for the loss of E2F4 and not a Given these findings, we wanted to determine whether E2F4’s subsequent change in the resulting cell lines. To address this Ϫ Ϫ ability to repress differentiation depends on its ability to bind to issue, we introduced full length E2F4 into wild type and E2f4 / MEFs by using retroviral infection (Fig. 3C). We then compared pocket proteins. To address this question, we generated a mutant the ability of control (pBABE) and E2F4-expressing cells to form of E2F4 that lacks the C-terminal pocket protein binding ⌬ Ϫ/Ϫ undergo adipogenesis (Fig. 3D). Significantly, wild-type E2F4 domain (E2F4 C) and introduced this into wild type or E2f4 reduced the differentiation potential of E2f4Ϫ/Ϫ MEFs to wild MEFs by retroviral infection (Fig. 3C). Although this truncated Ϫ/Ϫ protein is unable to bind pRB-family members (data not shown), type levels. Thus, the differentiation defect of the E2f4 MEFs Ϫ Ϫ is due to the loss of E2F4. it suppressed the differentiation defect of the E2f4 / MEFs Our data indicates that E2F4 and p130 act synergistically to almost as efficiently as the wild-type E2F4 protein. This suggests inhibit adipocyte differentiation. There is also extensive evi- that E2F4 inhibits differentiation independently of pocket dence to indicate that pRB is essential for adipogenesis (27, 28). protein binding.

Landsberg et al. PNAS ͉ March 4, 2003 ͉ vol. 100 ͉ no. 5 ͉ 2459 Downloaded by guest on October 1, 2021 pRB Is Required for Adipogenesis in the Absence of E2F4. Previous (TGF␤)-responsive manner and that the resulting complex studies have shown that RbϪ/Ϫ cells fail to significantly undergo represses the c- gene. Because E2F4’s role in adipogenesis adipogenesis (27, 28). pRB has been shown to serve as a does not require pocket protein binding, it cannot function coactivator for C͞EBP␤, suggesting that pRB loss impairs through this precise mechanism, but one can envisage variations differentiation by reducing the activity of C͞EBP␤. However, it on this theme. Importantly, the finding that E2F4 participates in is also possible that the differentiation defect is a consequence both TGF␤ signaling and adipogenesis indicates that its role of the requirement for pRB in mediating cell cycle arrest. We extends well beyond the regulation of classical E2F-responsive have recently shown that E2F4 loss completely suppresses the genes. It also raises the possibility that E2F4 might play a direct inappropriate proliferation of pRB-deficient cells under growth role in the differentiation of other cell lineages. In this regard, arrest conditions (32). Given this finding, we wished to establish it is interesting to note that the erythroid defect in the E2F4- whether E2F4 loss would modulate the differentiation pheno- deficient mice seems to result from a failure in terminal differ- type of RbϪ/Ϫ MEFs. Consistent with our previous studies, the entiation without any obvious affect on cell cycle regulation E2f4Ϫ/Ϫ MEFs differentiated much better that the wild-type (17, 19). controls in response to hormone induction (Fig. 3E). In con- The second mechanism by which the E2F and pocket proteins trast, the E2f4Ϫ/Ϫ;RbϪ/Ϫ MEFs failed to undergo hormone- contribute to the regulation of adipocyte differentiation involves induced differentiation (Fig. 3E). Moreover, confluence ar- p107 and p130. Consistent with previous reports (26), we find rested E2f4Ϫ/Ϫ;RbϪ/Ϫ MEFs never spontaneously differentiated that DKO MEFs undergo hormone-induced differentiation with (data not shown). We therefore conclude that E2F loss is unable increased efficiency. Our analysis of compound mutant MEFs to override the differentiation defect of the RbϪ/Ϫ MEFs even provides additional insight into their action. First, the loss of though it suppresses the cell cycle defect. This finding supports these proteins cannot trigger the differentiation process in an the prevailing view that pRB promotes differentiation indepen- analogous manner to the absence of E2F4. However, their loss dently of its role in cell cycle regulation. Furthermore, our data significantly enhances adipogenesis that has been induced by indicate that, in adipogenesis, pRB is required to act downstream either classic hormone induction or E2F4 loss. Although our of E2F4. data show that E2F4 can function in a pocket protein- independent manner, it is entirely possible that p107 and p130 Discussion repress adipocyte induction in association with one or more E2F In this study, we have continued to investigate the roles of the proteins, including E2F4. Because E2F4 and p107͞p130 both E2F and pocket proteins in the regulation of adipocyte differ- regulate adipogenesis negatively, we cannot order the action of entiation. It was previously shown that hormone-induced adi- these proteins. However, we were able exploit the cooperative pogenesis is promoted by the loss of either E2F4 or p107 and effects of E2F4 and p107͞p130 deficiency to determine the p130 (25, 26). It seemed highly likely that the shared activity of relative contribution of p107 and p130. Whereas p107 loss had E2F4 and p107͞p130 simply reflects their participation in tran- no detectable effect on the differentiation phenotype of the scriptionally repressive complexes. However, our current anal- E2f4Ϫ/Ϫ MEFs, p130 mutation could increase it, as well as the yses of compound mutant MEFs do not support this hypothesis. combined mutation of p107 and p130. This finding strongly Instead, they suggest that the E2F and pocket proteins contrib- suggests that p130 is the predominant mediator of the p107͞p130 ute to the regulation of adipocyte differentiation through three differentiation function. distinct mechanisms. Moreover, each one of these can be These observations provide important insight into the relative separated from effects on cell cycle control. roles of p107 and p130. To date, the analysis of both development The first mechanism involves the E2F4 . and cell cycle defects in the p107;p130 compound mutant mice We have found that E2F4 loss predisposes MEFs to undergo have largely highlighted the overlapping properties of these adipogenesis. This phenotype includes increasing the proportion proteins (14–16, 20, 21). For example, the p107Ϫ/Ϫ and p130Ϫ/Ϫ of cells that differentiate in response to the standard hormone mice have no obvious developmental defects, but the treatment as well as enabling confluent monolayers to undergo p107Ϫ/Ϫ;p130Ϫ/Ϫ compound mutants are neonatal lethal (20, 38). spontaneous differentiation. We do not understand the precise In contrast, our differentiation studies provide clear evidence for mechanism by which E2F4 mediates this apparent gatekeeper differential properties of p130 and p107 that is commensurate function, but our analysis suggests that it is quite distinct from the with the specificity of their expression in arrested (p130) vs. known properties of E2F. First, the increased differentiation of cycling (p107) cells. Given these findings, it will be interesting the E2F4-deficient cells can be separated from any gross changes to compare the phenotypes of the E2f4Ϫ/Ϫ;p130Ϫ/Ϫ and in either E2F-responsive gene expression or cell cycle regulation. E2f4Ϫ/Ϫ;p107Ϫ/Ϫ mutant mice to determine whether the differ- Second, this differentiation phenotype is neither shared nor ential roles of p130 and p107 can be observed in vivo and whether augmented by the loss of E2F5, with which it is known to they are specific for adipocyte differentiation or extend to other cooperate in the control of cell cycle exit, or any other E2F family tissues. member (ref. 25; Fig. 3; unpublished findings), indicating that it The analysis of the E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEFs also raises some is a specific activity of E2F4. Finally, E2F4’s ability to suppress question of the role of cell cycle effects. The DKO MEFs have differentiation can occur in the absence of its C terminus a major defect in their ability to exit the cell cycle, and this was sequences. This deleted region includes the region of the protein thought to account for, or at least contribute to, the propensity that mediates both transcriptional activation and pocket protein of these cells to undergo adipogenesis (14, 15, 26). However, the binding functions. Thus, E2F4’s ability to inhibit differentiation E2f4Ϫ/Ϫ;p130Ϫ/Ϫ MEFs do not have an obvious cell cycle defect must occur independently of its ability to either directly activate even though they display a differentiation defect as profound as target genes, or to mediate their repression through recruitment TKO MEFs. Thus, at least in the absence of E2F4, the effect of of the pocket proteins and their associated histone deactylates p130 mutation seems largely cell cycle independent. and͞or methylases. The final mechanism by which the E2F and pocket proteins Our observations do not rule out the possibility that E2F4 contribute to the regulation of adipocyte differentiation involves contributes to the regulation of adipogenesis by influencing, pRB. Numerous studies have shown that pRB is required for the either positively or negatively, the activity of other transcription differentiation of MEFS in vitro. This finding is believed to be factors. Indeed, there is some precedence for this mode of action. due to pRB’s role as a coactivator for C͞EBP␤, but it could also Chen et al. (37) recently showed that the E2F4͞p107 complex result from the profound cell cycle defect of the RbϪ/Ϫ cells (24, binds to the Smad proteins in a transforming growth factor ␤ 27, 28). Indeed, Fajas and coworkers (39) recently reported that

2460 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0138064100 Landsberg et al. Downloaded by guest on October 1, 2021 pRB also negatively regulates adipogenesis via inhibition of PPAR␥ Our data clearly show that the three distinct mechanisms by and suggested that pRB’s dominant positive role might primarily which the E2F and pocket proteins regulate adipogenesis can be depend on its key role in cell cycle exit. Because E2F4 loss separated from one another and, in certain situations, from cell suppresses the inappropriate proliferation of pRB-deficient cells cycle control. However, in vivo, it seems highly likely that there (32), we have also examined the differentiation phenotype of the Ϫ/Ϫ Ϫ/Ϫ will be extensive cross-talk between these pathways to ensure E2f4 ;Rb MEFs. Significantly, these cells are completely ordered progression through sequential stages of the differen- impaired in their ability to undergo either spontaneous or hor- tiation process and link each one to the appropriate cell cycle mone-induced differentiation. This finding represents evidence that the requirement for pRB in adipogenesis is independent of any cell condition. Thus, the future challenge will be to establish the cycle effect. Moroever, it shows that pRB acts downstream of E2F4 underlying basis of each mechanism and determine how they are in the regulation of adipogenesis. Because C͞EBP and E2F bind to coordinated in vivo. the same region of pRB, we initially suspected that E2F4 would inhibit adipogenesis by preventing the pRB-C͞EBP␤ interaction. We thank Tyler Jacks for kindly providing the p107 and p130 mutant However, our data show that E2F4’s inhibitory activity does not mouse strains, and members of the Lees laboratory for helpful discus- require the C-terminal, pocket protein-binding domain. Thus, sions during this study and the preparation of this manuscript. R.L.L. was whereas this competitive mechanism may operate in some situa- supported by a fellowship from the Anna Fuller Fund. This work was tions, it cannot fully account for the adipogenic properties of the supported by National Institutes of Health Grant RO1-GM53204 (to E2F4-deficient cells. J.A.L.).

1. Dyson, N. (1998) Genes Dev. 12, 2245–2262. Harlow, E., Beach, D., Weinberg, R. A. & Jacks, T. (1996) Genes Dev. 10, 2. Trimarchi, J. M. & Lees, J. A. (2002) Nat. Rev. Mol. Cell Biol. 3, 11–20. 1633–1644. 3. Trimarchi, J. M., Fairchild, B., Wen, J. & Lees, J. A. (2001) Proc. Natl. Acad. 21. Rossi, F., MacLean, H. E., Yuan, W., Francis, R. O., Semenova, E., Lin, C. S., Sci. USA 98, 1519–1524. Kronenberg, H. M. & Cobrinik, D. (2002) Dev. Biol. 247, 271–285. 4. Storre, J., Elsasser, H. P., Fuchs, M., Ullmann, D., Livingston, D. M. & 22. Lipinski, M. M. & Jacks, T. (1999) Oncogene 18, 7873–7882. Gaubatz, S. (2002) EMBO Rep. 3, 695–700. 23. Rosen, E. D. & Spiegelman, B. M. (2000) Annu. Rev. Cell Dev. Biol. 16, 145–171. 5. Helin, K. (1998) Curr. Opin. Genet. Dev. 8, 28–35. 24. Richon, V. M., Lyle, R. E. & McGehee, R. E., Jr. (1997) J. Biol. Chem. 272, 6. Humbert, P. O., Verona, R., Trimarchi, J. M., Rogers, C., Dandapani, S. & 10117–10124. Lees, J. A. (2000) Genes Dev. 14, 690–703. 25. Fajas, L., Landsberg, R. L., Huss-Garcia, Y., Sardet, C., Lees, J. A. & Auwerx, 7. Wu, L., Timmers, C., Maiti, B., Saavedra, H. I., Sang, L., Chong, G. T., J. (2002) Dev. Cell 3, 39–49. Nuckolls, F., Giangrande, P., Wright, F. A., Field, S. J., et al. (2001) Nature 414, 26. Classon, M., Kennedy, B. K., Mulloy, R. & Harlow, E. (2000) Proc. Natl. Acad. 457–462. Sci. USA 97, 10826–10831. 8. He, Y., Armanious, M. K., Thomas, M. J. & Cress, W. D. (2000) Oncogene 19, 27. Chen, P. L., Riley, D. J., Chen, Y. & Lee, W. H. (1996) Genes Dev. 10, 3422–3433. 2794–2804. 9. Adams, M. R., Sears, R., Nuckolls, F., Leone, G. & Nevins, J. R. (2000) Mol. 28. Charles, A., Tang, X., Crouch, E., Brody, J. S. & Xiao, Z. X. (2001) J. Cell. Cell. Biol. 20, 3633–3639. Biochem. 83, 414–425. 10. Leone, G., Nuckolls, F., Ishida, S., Adams, M., Sears, R., Jakoi, L., Miron, A. 29. Slomiany, B. A., D’Arigo, K. L., Kelly, M. M. & Kurtz, D. T. (2000) Mol. Cell. & Nevins, J. R. (2000) Mol. Cell. Biol. 20, 3626–3632. Biol. 20, 5986–5997. 11. Moberg, K., Starz, M. A. & Lees, J. A. (1996) Mol. Cell. Biol. 16, 1436–1449. 30. Porse, B. T., Pedersen, T. A., Xu, X., Lindberg, B., Wewer, U. M., Friis-Hansen, 12. Takahashi, Y., Rayman, J. B. & Dynlacht, B. D. (2000) Genes Dev. 14, 804–816. L. & Nerlov, C. (2001) Cell 107, 247–258. 13. Gaubatz, S., Lindeman, G. J., Ishida, S., Jakoi, L., Nevins, J. R., Livingston, 31. Altiok, S., Xu, M. & Spiegelman, B. M. (1997) Genes Dev. 11, 1987–1998. D. M. & Rempel, R. E. (2000) Mol. Cell 6, 729–735. 32. Lee, E. Y., Cam, H., Ziebold, U., Rayman, J. B., Lees, J. A. & Dynlacht, B. 14. Bruce, J. L., Hurford, R. K., Jr., Classon, M., Koh, J. & Dyson, N. (2000) Mol. (2002) Cancer Cell 2, 463–472. Cell 6, 737–742. 33. Ramirez-Zacarias, J. L., Castro-Munozledo, F. & Kuri-Harcuch, W. (1992) 15. Hurford, R. K., Jr., Cobrinik, D., Lee, M. H. & Dyson, N. (1997) Genes Dev. Histochemistry 97, 493–497. 11, 1447–1463. 34. Verona, R., Moberg, K., Estes, S., Starz, M., Vernon, J. P. & Lees, J. A. (1997)

16. Mulligan, G. J., Wong, J. & Jacks, T. (1998) Mol. Cell. Biol. 18, 206–220. Mol. Cell. Biol. 17, 7268–7282. CELL BIOLOGY 17. Rempel, R. E., Saenz-Robles, M. T., Storms, R., Morham, S., Ishida, S., Engel, 35. Naviaux, R. K., Costanzi, E., Haas, M. & Verma, I. M. (1996) J. Virol. 70, A., Jakoi, L., Melhem, M. F., Pipas, J. M., Smith, C. & Nevins, J. R. (2000) Mol. 5701–5705. Cell 6, 293–306. 36. LeCouter, J. E., Kablar, B., Whyte, P. F., Ying, C. & Rudnicki, M. A. (1998) 18. Lindeman, G. J., Dagnino, L., Gaubatz, S., Xu, Y., Bronson, R. T., Warren, Development (Cambridge, U.K.) 125, 4669–4679. H. B. & Livingston, D. M. (1998) Genes Dev. 12, 1092–1098. 37. Chen, C. R., Kang, Y., Siegel, P. M. & Massague, J. (2002) Cell 110, 19–32. 19. Humbert, P. O., Rogers, C., Ganiatsas, S., Landsberg, R. L., Trimarchi, J. M., 38. Lee, M. H., Williams, B. O., Mulligan, G., Mukai, S., Bronson, R. T., Dyson, Dandapani, S., Brugnara, C., Erdman, S., Schrenzel, M., Bronson, R. T. & Lees, N., Harlow, E. & Jacks, T. (1996) Genes Dev. 10, 1621–1632. J. A. (2000) Mol. Cell 6, 281–291. 39. Fajas, L., Egler, V., Reiter, R., Hansen, J., Kristiansen, K., Debril, M.-B., 20. Cobrinik, D., Lee, M. H., Hannon, G., Mulligan, G., Bronson, R. T., Dyson, N., Mirad, S. & Auwerx, J. (2002) Dev. Cell 3, 903–910.

Landsberg et al. PNAS ͉ March 4, 2003 ͉ vol. 100 ͉ no. 5 ͉ 2461 Downloaded by guest on October 1, 2021