New Insights Into Transcriptional Regulation by RB: One Size No Longer Fits All

The retinoblastoma (Rb) protein is a key regulator of cell proliferation, differentiation, and tumorigenesis. Initial studies of Rb revealed that it binds to, and decreases the activity of, the E2F family of transcription factors. Over the last decade, the mechanisms by which Rb regulates E2F activity have been well-studied. These investigations have lead to a commonly held belief that Rb functions solely as a transcriptional repressor. However, although not as commonly discussed, there are many examples of Rb synergizing with site-specific transcription factors to activate transcription. This Rb-mediated activation appears to be cell type-specific, transcription factor-specific, and even promoter-specific. This chapter details some of the examples of Rb-mediated transcriptional activation and suggests future studies that could provide insight into the mechanisms by which Rb can function to positively regulate transcription. The Classic Model for Rb Function The retinoblastoma (Rb) gene was the first tumor suppressor gene cloned and hundreds of investigators have since studied the biological consequences of deregulation of Rb in human tumors, in mouse models, and in cell culture. The results of these studies have been used to generate a commonly held model for the mechanism by which Rb modulates cell proliferation. In brief, Rb is proposed to block cell cycle progression and to promote differentiation by negatively regulating the transcription of genes whose products are required for the G1/ S phase transition and DNA replication. It has been postulated that Rb-mediated repression is due to its ability to interact with proteins whose biochemical activities favor the creation of an inactive chromatin structure. For example, several groups have found that Rb can interact with a histone deacetylase to repress transcription through the removal of acetyl groups from the lysine tails of histones. The removal of this post-translational modification is associated with a compacted (inactive) nucleosomal structure, which leads to reduced transcriptional competence. Rb can also interact with SUV39H1, a human histone methyltransferase that specifically methylates histone H3 at lysine 9, resulting in the formation of heterochromatin and transcriptional repression. Another activity with which Rb interacts is human SWI/SNF, a protein complex which mediates transcriptional repression via an ATP-dependent remodeling of chromatin. Finally, Rb can cooperate with DNA methyltransferase 1 to repress promoters containing E2F sites. In this case, it is thought that methylation of CpG dinucleotides in the promoter region by DNA methyltransferase 1 leads to the interaction of methyl-CpG binding proteins with the DNA, followed by the recruitment of chromatin remodeling enzymes such as SUV39H1. Fanciulli(Farnham) 11/8/04, 2:24 PM 1 Rb and Tumorigenesis 2 Because Rb lacks a DNA binding domain, it must be localized to promoters through interactions with site-specific DNA binding proteins. The first transcription factor shown to interact with Rb was E2F1.11-13 Subsequently, it has been shown that Rb can bind to several members of the E2F family, with the highest affinity towards E2F1, E2F2, and E2F3 and the lowest affinity towards E2F4. The interaction between Rb and the E2Fs is regulated by cell cycle-dependent phosphorylation of Rb (Fig. 1). During G0 and G1 phases of the cell cycle, hypophosphorylated Rb binds to and inhibits the transcriptional activity of E2F-regulated promoters. As cells progress through G1, cyclin D/cdk4, cyclin D/cdk6, and cyclin E/cdk2 complexes are sequentially activated and then phosphorylate Rb. Hyperphosphorylation of Rb results in its dissociation from, and the resultant activation of, E2F complexes during late G1.14 Since the genes regulated by E2F family members mediate cell cycle progression, DNA repair, DNA replication and DNA recombination, Rb can participate in the control of these processes through its interaction with, and functional inhibition of, the E2F family members. Confounding Facts About Rb Function The model for Rb function described above is entirely consistent with its role as a tumor suppressor; i.e., the presence of Rb normally keeps E2F-regulated genes under tight control, whereas loss of Rb in tumors allows such genes to be transcribed at inappropriately high levels, leading to enhanced proliferation and neoplastic transformation. However, a growing body of evidence suggests that the role of Rb in the cell may be more complex. For example, Rb is not always lost or mutated in tumors. In fact, Rb has been shown to be present at increased levels in colon tumors, as compared to normal colon tissue. One explanation of this seemingly paradoxical finding is that the increased Rb may play a role in the prevention of apoptosis. It has been shown that E2Fs, when overexpressed, can activate apoptosis-inducing genes. Several signal transduction pathways that are often inappropriately activated in cancers have been implicated in controlling the levels and/or activity of the E2F family members. Thus, it is possible that certain colon tumors arise due to initial mutations which activate signaling pathways upstream of E2F, leading to a higher than normal amount of active E2F in the cell. If so, then loss of Rb would allow even higher E2F activity, perhaps resulting in the increased E2F-mediated transcriptional activation of pro-apoptotic genes. Aberrant expression of Rb in a tumor may allow cells to survive long enough to acquire functional mutations in genes that mediate apoptosis. Considered in this light, retaining or increasing levels of Rb may be critical for tumor development, whereas loss of Rb would be detrimental to tumor formation. Accordingly, Yamamoto et al have shown that introduction of antisense mRNA to Rb results in apotosis in colon cancer cells. While these findings support the classical model of Rb functioning to repress E2F-regulated genes, it possible that the increased Rb in colon tumors may play an alternate role. For example, others have shown that expression of a constitutively active Rb in mouse mammary glands results in the development of hyperplastic nodules and adenocarcinomas.18 In this case, Rb appears to be promoting tumor formation when overexpressed in otherwise normal cells (i.e., cells that do not have excessive E2F activity). These results suggest that Rb may play a role in tumor initiation in mammary cells through a mechanism distinct from repressing transcription of E2F target genes. Another observation about Rb that does not fit the commonly held model is that Rb is not always released from the chromatin in S phase. Using a relatively unbiased approach that relies upon a combination of chromatin immunoprecipitation and CpG microarray analysis, Wells et al identified genomic sites bound by Rb in Raji cells. As expected, characterization of these sites identified a subset that showed the cell cycle-dependent changes in protein-DNA occupancy that would be predicted by the classic model. For example, some of the newly identified sites were bound by Rb and E2F in G0 phase, but Rb could not be detected on the site in S phase. On the other hand, certain promoters showed an increase in Rb recruitment in S phase, whereas others showed high level, constitutive binding of Rb throughout the cell Fanciulli(Farnham) 11/8/04, 2:24 PM 2 3 New Insights Into Transcriptional Regulation by RB cycle. These surprising findings raise the question as to whether Rb mediates the same biochemical activities in G0 and in S phase. In other words, is Rb always a transcriptional repressor? It is intriguing that one of the sites that showed an increased amount of bound Rb in S phase was the promoter for the nuclear oncogene c-Myc. The expression of c-Myc correlates with cell proliferation and the c-Myc promoter displays robust activity in S phase. However, too much c-Myc can lead to neoplastic transformation and, therefore, levels of c-Myc must be tightly controlled. One interpretation of the chromatin immunoprecipitation data of Wells et al is that Rb may function as a rheostat. In this model, Rb would keep the Myc gene in a fully off position in quiescent cells but would work in opposition to S-phase specific activators to keep c-Myc transcription at submaximal levels in S phase. However, an alterative explanation could be that Rb is involved in activation of the c-Myc promoter in S phase, in addition to its role as a repressor in G0 phase. In other words, Rb may function as a switch hitter, exchanging corepressors in G0 phase with coactivators in S phase. The concept that Rb may promote tumor formation and/or proliferation by activating transcription of oncogenes such as c-Myc is tantalizing. However, to date, the consequences of removing Rb from the c-Myc promoter in S phase has not been determined. Therefore, it is not possible to conclude that Rb is an S phase-specific activator of the c-Myc promoter. Although the exact role that Rb played in the regulation of the promoters to which it was bound in S phase was not determined in the studies of Wells et al, there is evidence, as described in the following section, to support the hypothesis that Rb can serve as a transcriptional activator at certain promoters. Figure 1. The classic model of Rb-mediated transcriptional regulation of E2F target genes. Rb serves as a transcriptional repressor of E2F target genes in G0 phase cells. The mechanism by which Rb represses transcription involves the ability of Rb to bring proteins such as histone methyltransferases or histone deacetylases to the basal promoter region. The interaction between Rb and E2F is weakened by the action of cyclin-dependent kinases that phosphorylate Rb in a cell cycle position-dependent manner. Therefore, in late G1 and during S phase, E2F is not bound by Rb and E2F target genes are transcribed. Fanciulli(Farnham) 11/8/04, 2:24 PM 3 Rb and Tumorigenesis 4 Evidence in Support of the Role of Rb As a Transcriptional Activator The first protein shown to cooperate with Rb to stimulate transcription was the site-specific transcription factor Sp1. Robbins et al identified a cis element that mediates transcriptional activation in response to Rb; this element was shown to bind Sp1. Several other genes containing Sp1 sites have also been shown to be activated by Rb, including the fourth promoter of the insulin like growth factor II gene and the hamster dihydrofolate reductase (dhfr) gene. The exact mechanism by which Rb stimulates Sp1-mediated transcription is unclear in many cases and the ability of Rb to function as an activator seems to be cell type-dependent and promoter-specific.26 In some cases, it has been suggested that Rb directly interacts with Sp1 to stimulate Sp1 transcriptional activity, whereas the action of Rb on Sp1 activity has been proposed to be indirect in other cases. For example, it has been proposed that Rb can stimulate transcription by interfering with the interaction between Sp1 and a negative regulator of Sp1 activity.27 Interestingly, Johnson-Pais et al28 showed that Sp1 activity can be inhibited by physical interaction with mdm2 and that expression of Rb results in the release of mdm2 from Sp1, most likely through Rb sequestering the mdm2 protein. Clearly, the degree to which Rb could stimulate Sp1 activity would be dependent on the comparative levels of Sp1 and mdm2 in a particular cell. Rb has also been proposed to function as an activator in other transcription complexes. For example, Thomas et al have shown that Rb physically interacts with the site-specific transcription factor CBFA1 to activate an osteoblast-specific reporter. CBFA1 is a transcriptional regulator that is critical for inducing osteoblast differentation. Thus, loss of Rb may lead to decreased CBFA1 transcriptional activity and a resultant defect in differentiation. The inability to achieve a differentiated state may lead to inappropriate proliferation and, eventually, to tumorigenesis. Interestingly, osteosarcoma is the second most common tumor after retinoblastoma itself among individuals with inherited heterozygous loss of the Rb gene.30 In addition, loss of Rb occurs in up to 60% of sporadic osteosarcomas. These findings support the hypothesis that loss of Rb alters the balance between proliferation and differentiation in osteoblasts. Thomas et al performed chromatin immunoprecipitation assays to demonstrate that Rb is recruited to the osteocalcin and osteopontin promoters, both of which are regulated by CBFA1. However, the mechanism by which Rb stimulates transcriptional activation when bound to these promoters is unknown. For example, does Rb enhance CBFA1-mediated recruitment of basal transcription factors or alternatively does an CBFA-1/Rb complex help recruit another site-specific transcription factor? The Jun family of transcription factors has also been implicated in Rb-mediated transcriptional activation. Xin et al showed that Rb positively regulates expression of the p202 gene, an interferonand differentiation-inducible phosphoprotein. Initial experiments demonstrated that an AP-1 site (to which Jun family members bind in cooperation with Fos family members) is critical for Rb-mediated transcriptional activation of the p202 gene. The authors then showed that Rb cooperates with JunD to provide an even greater transcriptional activation of the p202 promoter. These studies did not characterize the mechanism by which Rb cooperates with JunD. Therefore, it is not yet known if Rb is bound to the p202 promoter region (as in the case with CBFA1) or if Rb works via removal of an inhibitor (as in certain cases of Sp1-mediated transcription). Others have also shown that Rb can cooperate with Jun family members. Slack et al found that Rb can cooperate with c-Jun to activate the DNA methyltransferase 1 (dnmt1) promoter by facilitating the in vitro binding of c-Jun to a noncanonical AP1 site.34 However, Rb could not be detected in the DNA-bound protein complex. The authors offer two alternative explanations for these results. First, Rb may enhance c-Jun binding to DNA by removal of an inhibitor of c-Jun DNA binding activity (similar to the Sp1 situation). Alternatively, Rb may initially bind to the DNA with c-Jun but be lost from the DNA-bound complex during the electrophoresis of the gel shift assay. Interestingly, although Rb can be detected (along with CBFA1) at the osteopontin promoter using in vivo assays, Thomas et al were not able to detect Rb in the CBFA1-DNA complex in vitro. Results such as these point out the advantages of Fanciulli(Farnham) 11/8/04, 2:24 PM 4 5 New Insights Into Transcriptional Regulation by RB using in vivo methods such as chromatin immunoprecipitation to analyze protein-DNA interactions, rather than relying on in vitro assays. Another family of transcription factors with which Rb cooperates to activate transcription includes the C-EBP proteins. For example, Charles et al revealed that Rb stimulates transcription of the surfactant protein D promoter via a C-EBP binding motif. They also showed that Rb can physically interact with C-EBP alpha, C-EBP beta and C-EBP delta and can be detected in C-EBP/DNA complexes in vitro. Similarly, Gery et al36 have shown that Rb, in the presence of C-EBP beta but not alone, activates promoters of myeloid specific genes such as the granulocyte colony-stimulating factor receptor (G-CSFR) and mim1. In contrast, other studies have shown that Rb can decrease C-EBP beta DNA binding activity, resulting in inhibition of C-EBP beta-mediated transcription in preadipocytes.37 Thus, similar to the effects on Sp1 activity, the direction in which Rb influences transcription mediated by C-EBP family members appears to be cell typeand promoter-specific. It is likely that the presence of other site-specific DNA binding factors bound to a particular promoter will determine the exact role that Rb plays in transcriptional regulation of C-EBP target genes. Several groups have also reported an interaction between Rb and AP-2. For example, Batsche et al showed that Rb can physically interact with AP-2 in vitro and in vivo and that it cooperates with AP-2 to activate the E-cadherin promoter. Similarly, Decary et al have shown that Rb activates the bcl-2 promoter through an AP-2 site. Interestingly, both of these studies showed that the synergy between Rb and AP2 is cell type-dependent, occurring in epithelial cells but not in fibroblasts. Importantly, Decary et al use chromatin immunoprecipitation assays to show that Rb can bind to the bcl-2 and E-cadherin promoters. The cases described above in which Rb stimulates transcription do not constitute a comprehensive list of all known Rb-activated promoters. Rather, they simply provide insight concerning the different types of transcription factors, target genes, and cell types for which Rb may serve a role other than as a transcriptional repressor. There are many other reported instances of Rb-mediated transcriptional activation. For example, Rb has been shown to synergize with transcriptional activators involved in muscle differentiation such as MyoD, Myogenins and Myf-5, ATFa and ATF2, the AH receptor and NF-kb. Rb has also been shown to cooperate with several different steroid receptors to activate transcription. For example, Singh et al46,47 showed that hBrm and Rb (but not the related proteins p107 or p130) cooperated to activate glucocorticoid-receptor mediated transcription. Also, Balasenthil et al showed that Rb positively regulates the cyclin D1 promoter via interactions with the estrogen receptor coactivator PELP1/MNAR. Finally, Hofman et al demonstrate that Rb serves as a coactivator of androgen receptor-mediated transcription. In summary, a growing body of evidence supports the hypothesis that Rb can function to mediate either activation or repression of transcription, most likely by serving as a platform for the recruitment of both coactivators and corepressors. As noted above, the mechanisms by which Rb serves as a repressor of E2F-activated transcription have been fairly well-studied (Fig. 2A). For example, Rb can interact with SUV39H1, a methylase specific for lysine 9 of histone H3. Nicolas et al have shown, using chromatin immunoprecipitation and synchronized cell populations, that Rb-repressed promoters such as dhfr show higher levels of histone H3 methylated on lysine 9 in G0 phase (when Rb is bound) than in S phase (when Rb is released). They also showed that levels of histone H3 acetylated on lysine 9 increase at the dhfr promoter as cells progress from G0 to S phase. Others have shown that HDAC1 is associated with the dhfr promoter in G0 phase, but not in S phase. Thus, a model can be developed for G0 phase-specific Rb-mediated repression of the dhfr promoter that entails Rb-mediated recruitment of an HDAC (to deacetylate H3), followed by or concomitant with Rb-mediated recruitment of SUV39H1 for methylation of lysine 9 of histone H3. For many promoters, the phosphorylation of Rb in S phase would result in disruption of Rb/E2F interactions, loss of Rb (and thus loss of HDACs and histone methyltransferases) from the promoter, and activation of transcription. However, it must be recalled that many promoters, such as the c-Myc promoter, are transcriptionally Fanciulli(Farnham) 11/8/04, 2:24 PM 5 Rb and Tumorigenesis 6 active and bound by Rb in S phase. Recent studies suggest a mechanism by which Rb might change roles and serve as a transcriptional activator in S phase (Fig. 2B). Fanciulli et al showed that a protein called Che-1 can reverse the Rb-mediated repression of the dhfr promoter in Saos-2 cells, but has no effect on basal transcription or E2F1-mediated activation of the dhfr promoter in the absence of Rb. Further studies revealed that Che-1 can bind directly to and displace HDAC1 from Rb.53 The binding of Che-1 to E2F target promoters is cell cycle regulated, showing the highest binding in S phase. Interestingly, Che-1 can also interact with the RPB11 subunit of RNAP II. Thus, the cell cycle-dependent interaction of Che-1 with Rb could turn Rb from a repressor to an activator, by displacing HDACs and serving as an adaptor between Rb and the basal transcriptional machinery.52 Because Rb has been shown to bind more than 100 proteins, many of which are transcriptional regulators, it would not be surprising to find that the ability to repress E2F activity is an important, but not sole, function of Rb, especially in differentiated cell types. Unfortunately, very few studies have been performed using chromatin IP to analyze markers of active vs inactive chromatin on non E2F-regulated promoters that are activated by Rb. However, one fairly recent study did use chromatin IP to show a correlation between transcriptional activation, binding of Rb and AP2, and the presence of acetylated histone H4 (a mark of active chromatin) on the bcl2 promoter.39 This suggests that perhaps Rb can recruit a histone acetyltransferase (Fig. 2b). As described above, it is also possible that Rb mediates transcripFigure 2. Mechanisms by which Rb can influence transcription. Depiction of four different mechanisms by which Rb may influence transcription. In panel A, Rb represses transcription of E2F target genes by serving as a platform for the binding of histone methyltransferases (HMTases), histone deacetylases (HDACs), DNA methyltransferases (DNMTs), or other corepressors. Similarly, in panel B, Rb activates transcription of CBFA1 target genes by serving as a platform for the binding of histone acetyltransferases (HATs) or other coactivators such as Che-1. In panel C, Rb stimulates transcription by serving as a bridge between two DNA binding factors and enhances their binding to the promoter DNA. Finally, in panel D, Rb sequesters a factor which can inhibit either the DNA binding or activity of a transcription factor (such as Sp1). Fanciulli(Farnham) 11/8/04, 2:24 PM 6 7 New Insights Into Transcriptional Regulation by RB tional activation via mechanisms that are independent of changes in chromatin structure, such as assisting in the recruitment of other site-specific factors (Fig. 2C) or the sequestration of inhibitory proteins (Fig. 2D).