INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The human retinoblastoma tumor suppressor gene (Rb) encodes a nuclear phosphoprotein, which is the prototypical member of a family of pocket proteins that are conserved through evolution (49). Rb-related genes have been identified not only in mammalian cells, but also in Drosophila melanogaster (7), Caenorhabditis elegans (31), and plants (54). Members of the RB family of pocket proteins control proliferation and differentiation by regulating transcription. These pocket proteins are not the classical transcription factors in that they do not possess sequence-specific DNA binding activities. Instead, these pocket proteins are recruited to promoters through interactions with transcription factors. The protein binding sites in RB and related proteins are commonly referred to as pockets. The transcription-regulatory functions are critically dependent on the pockets of these proteins.

Conserved among the RB family of pocket proteins is the A-B domain. The three dimensional structure of the RB A-B domain has been determined (29). The A and B regions each contain a five-helix core motif with a cyclin fold. These two core motifs are packed into a complex domain with additional structural elements to form an extensive A-B interface and a shallow groove that binds to the LXCXE peptide (29). The A-B interface and the LXCXE-binding groove are predicted, based on sequence homology, to be conserved in all the known RB family pocket proteins (29). The LXCXE-motif was first identified in viral oncoproteins that bind to the A-B domain of RB (5, 10, 20, 21, 53). Mutagenesis studies have demonstrated that both the A and B regions are required to form the LXCXE-binding pocket (23). The crystal structure has revealed that the shallow groove for binding to LXCXE is exclusively within the B region (29). The A region does not contain any contact sites for the LXCXE peptide, but it does contribute to the proper folding of the LXCXE-binding pocket. Most importantly, occupancy of the LXCXE-binding pocket does not preclude the binding of an E2F peptide to the A-B domain (29). This observation is consistent with earlier reports showing that the LXCXE binding did not preclude the binding of E2F to RB (11, 22, 39, 46). Thus, the conserved A-B domain contains at least two binding pockets, one for the LXCXE-peptide and the other for a peptide derived from the C-terminal region of E2F-1 (29). Because the A-B domain can simultaneously bind to the LXCXE peptide and an E2F peptide, the A-B domain of the pocket proteins has the ability to assemble protein complexes (49).

A major target of the RB pocket proteins is E2F. The E2F family consists of heterodimeric transcription factors composed of E2F and DP subunits (9, 44). The RB protein itself can interact with several members of the E2F family to repress transcription from E2F-regulated promoters (44). E2F-responsive genes include cyclin E, cyclin A, and other proteins and enzymes required for DNA replication (6, 30, 44). Hence, the repression of E2F-regulated promoters by RB can account for the inhibition of cell proliferation. Previous studies have identified two mechanisms for the transcription repression function of RB. The first mechanism is based on the fact that RB can physically associate with the transactivation domain of E2F-1 and thus block its ability to stimulate transcription (17, 18). Direct evidence supporting this mechanism comes from in vitro transcription studies, in which RB is shown to block the E2F-mediated recruitment of the TFIIA-TFIID transcription preinitiation complex (8, 41). The second mechanism involves the ability of RB to assemble a transcription repression complex at E2F-regulated promoters. In this mechanism, RB simultaneously binds to E2F and a transcription repressor, such as histone deacetylase 1 (HDAC1), to inactivate E2F-regulated promoters by an active process that involves chromatin modification (3, 34, 35). In this model, E2F collaborates with RB to mediate transcription repression. Evidence supporting this model of repression comes from several lines of investigation, including the finding that inhibition of HDAC activity can compromise the transcription repression function of RB (2). Because the mutation of E2F-binding sites in cellular promoters can exert either a negative or a positive effect on transcription, E2F is likely to function both as a transcription activator and a repressor (9, 44). In either case, the interaction with RB pocket proteins will lead only to transcription repression, through the two alternative mechanisms that are not mutually exclusive.

The growth suppression function of RB is important under two types of physiological circumstances: a reversible inhibition of cell cycle progression in response to negative growth signals or an irreversible inhibition of cell proliferation in terminally differentiated cells. RB plays an essential role in G₁ arrest induced by DNA damage (4, 16), by transforming growth factor , or by contact inhibition (56). These negative growth signals activate RB by blocking the phosphorylation of RB, which directly inhibits its pocket functions (25, 26, 27). Indeed, RB mutants that lack a specific combination of phosphorylation sites (PSM.RB mutants) can inhibit cell proliferation even in the presence of growth factors and without any input of negative growth signals (26, 33). The irreversible growth arrest mediate by RB is best illustrated with muscle differentiation. Muscle precursor cells (myoblasts) undergo terminal differentiation when placed under differentiation conditions (high cell density and low serum). Through an RB-dependent mechanism, the differentiated myocytes become refractory to further challenges with serum. In differentiated myocytes, serum does not activate RB phosphorylation, and this is in part mediated by the differentiation-induced upregulation of p21Cip1, which blocks RB phosphorylation (47). Myoblasts lacking a functional RB can initiate the differentiation program, but these RB-deficient myocytes will reactivate DNA synthesis when challenged with serum (14, 15, 42). The p21Cip1 expression is induced in RB-deficient myocytes and its levels are not reduced by serum (48). Thus, p21Cip1 cannot mediate the terminal growth arrest without RB (14, 15).

While previous studies have identified the transcription repression mechanisms of RB and the biological roles of RB in reversible or permanent growth suppression, the precise contribution of the RB pockets to these RB functions has not been elucidated. With the A-B domain structure, it has become possible to dissect the functional contribution of the RB pockets. We focused on the interaction between RB and HDAC1 because HDAC1 contains an LXCXE motif (3, 12, 35) and is likely to bind RB through the shallow groove in the B region of the A-B domain. We reasoned that mutation of the LXCXE-binding site should disrupt the interaction between RB and HDAC1. We also reasoned that as long as the mutation does not compromise the integrity of the A-B domain, it should be possible to maintain the RB-E2F interaction. By replacing asparagine (N) 757 of RB with phenylalanine (F), we have been able to generate an RB mutant that is indeed defective in binding to HDAC1 but retains the ability to inhibit the transactivation function of E2F. The RB(N757F) mutant can suppress cell growth as efficiently as RB(WT) under conditions of RB phosphorylation inhibition, for example, by the Cdk4-inhibitor p16Ink4a. However, the RB(N757F) mutant cannot establish the irreversible growth arrest in myocytes. When challenged with serum, RB(N757F) became phosphosrylated in differentiated myocytes. This is in contrast to RB(WT), which established the permanent growth arrest and blocked its own phosphorylation by serum in differentiated myotubes. These results identify a critical function of the LXCXE-binding pocket of RB in establishing a serum-insensitive state in differentiated myocytes. These results also demonstrate that it is possible to inactivate a specific pocket within the A-B domain of pocket proteins. Such pocket-specific mutants will be useful in dissecting the contribution of each pocket to the biological functions of RB and related pocket proteins.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cell culture and transfection. The RbB-deficient human cervical carcinoma cells C33A and the osteosarcoma Saos-2 cells were obtained from the American Type Culture Collection. Rb^/ and p21^/ fibroblasts were isolated from Rb^/ and p21^/ mouse embryos, respectively. The Rb^/ CC42 myoblasts were provided by Kenneth Walsh (Tufts University). Unless otherwise indicated, these cells were cultured in growth medium (Dulbecco modified Eagle medium [high glucose] supplemented with 15% heat-inactivated fetal bovine serum and antibiotics) at 37°C in 5% CO₂.

/

/

Plasmids. RB(L769D), RB(I768A/L769D), RB(I768D/L769D), RB(N757F) and (PSM.9I)RB(N757F) were constructed by standard in vitro site-directed mutagenesis (detail protocol available upon request) using Pfu DNA polymerase (Stratagene). DNA fragments amplified by PCR were sequenced to confirm the presence of the correct mutations. Wild-type as well as mutant RB cDNAs were cloned into the mammalian expression vector pCMV-neo-Bam (52) to create plasmids pCMV-RB(WT), pCMV-RB(L769D), pCMV-RB(I768A/L769D), pCMV-RB(I768D/L769D), pCMV-RB(N757F), and pCMV(PSM.9I)RB(N757F). The plasmid expressing p16 has been described (26). The plasmid expressing HA-E7 was constructed by PCR using primers 5'-CCGGATCCATGTACCCATACGATGTTCCAGATTACGCTCATGGAATACACC-3' and 5'-CGGAATTCTGGTTTCTGAGAAACGAGTGGGGC-3' to place the hemagglutinin (HA) tag at the N terminus of E7. The PCR product was then cloned into the pCMV-neo-Bam vector to create pCMV-HA-E7. Plasmids expressing HA-E2F-1 and DP-1 have been described (25). The plasmids pCMV-HA-HDAC1, expressing HA-tagged HDAC1, and pCMV-HA-HDAC1IACEE, expressing a HA-tagged HDAC1 with its LXCXE-like motif IACEE deleted [referred to as HDAC1(L) in the text], were kindly provided by Annick Harel-Bellan (12). The MyoD-expressing plasmid pCMV-MyoD has been described (40). Plasmids expressing GST-RB(WT) and GST-RB(N757F) fusion proteins were constructed by PCR amplification, using the plasmids pCMV-RB(WT) and pCMV-RB(N757F) as templates, with primers 5'-TTTGGATCCATGACACAGAGAACACCACG-3' and 5'-GGGTGGATCCTTTCTCTTCCTTGTTTGAGG-3', which cover amino acid residues 353 to 928 of RB. The resulting PCR fragments were cloned into the glutathione S-transferase (GST) fusion protein expression vector pGEX-KG (52). The GST-E7 and GST-E2F-1 fusion protein-expressing plasmids have been described (25). The plasmid expressing GST-HDAC1 was constructed by PCR amplification, using the HA-HDAC1-expressing plasmid as the template, with primers 5'-ATATAGGATCCAATGGCGCAGACTCAGGGC-3' and 5'-ATGCCTCTAGACTCAGGCCAACTTGACCTC-3'. The resulting PCR product was then cloned into the plasmid pGEX-KG.

GST pulldown assays. GST fusion proteins, including GST-E7, GST-E2F-1, GST-RB(WT), GST-RB(N757F), and GST-HDAC1, were expressed in bacteria and purified as described (52). Proteins to be tested for binding to these GST fusion proteins were expressed in C33A cells. Binding assays were carried out by incubating GST fusion proteins with C33A cell lysates in binding buffer (25 mM Tris-HCl [pH 7.5], 250 mM KCl, 0.5% NP-40, 0.1% Triton X-100, 5 mM EDTA, 25 mM NaF, 14 mM 2-mercaptoethanol, and protease inhibitors) as described (52). Bound proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 4 to 20% gradient gels unless otherwise indicated. Immunoblottings were carried with the polyclonal anti-RB antibody 851 as described (50) and the monoclonal anti-HA antibody HA1.1 (Sigma).

Coimmunoprecipitation and immunoblotting. C33A cells cultured in 100-mm-diameter dishes to 50% confluence were cotransfected with 6 µg of RB-expressing plasmids plus 6 µg of plasmids expressing either HA-E7, HA-E2F1, HA-HDAC1, or HDAC1(L). Forty-eight hours after transfection, cells were harvested and resuspended in an ice-cold lysis buffer (50 mM HEPES [pH 8], 150 mM KCl, 30 mM MgCl₂, 10 mM EDTA, 0.5% NP-40, 10% glycerol, 14 mM 2-mercaptoethanol, 25 mM NaF, and protease inhibitors). Cell resuspensions were rotated at 4°C for 10 min and subsequently cleared by a 10-min centrifugation at 4°C at 14,000 × g to obtain whole-cell lysates. Cell lysates expressing equal amounts of the transfected proteins were incubated at 4°C for 2 to 4 h with either the polyclonal anti-RB antibody 851 (50) or the monoclonal anti-HA antibody HA1.1 (Sigma). A mixture of protein A and G conjugated to agarose was then added to the immunoprecipitations, and the incubation continued for an additional 16 h at 4°C. At the end of incubation, the immunoprecipitates were washed three times with the lysis buffer without protease inhibitors and three times with PBS before being separated by SDS-PAGE in 4 to 15% gradient gels and analyzed by immunoblotting with the indicated antibodies. To determine the phosphorylation patterns of RB in CC42 cells after serum restimulation, the CC42 cells were cotransfected by Superfect (Qiagen) with pBabe-puro, which confers puromycin resistance, plus plasmids expressing RB(WT), RB(N757F), HDAC1, and HDAC1(L). Twenty-four hours after transfection, the cells were cultured in differentiation medium containing puromycin (1.5 µg/ml; Sigma) for 3 days, and this was followed by culturing in growth medium containing 15% fetal bovine serum for 24 h. Whole-cell lysates were then prepared and subjected to SDS-PAGE using a 7% gel (acrylamide-bis acrylamide, 29.6:0.4) followed by immunoblotting with the polyclonal anti-RB antibody 851 (50).

BrdU incorporation, colony formation, and flat-cell formation assays. Saos-2 cells cultured in 100-mm-diameter dishes to 50% confluence were cotransfected with the pH2B-GFP plasmid (24) plus pCMV-neo-Bam, pCMV-RB(WT), or pCMV-RB(N757F). Forty-eight hours after transfection, the cells were divided into three portions for assaying bromodeoxyuridine (BrdU) incorporation, colony formation, and flat-cell formation. For the BrdU incorporation assay, the transfected cells were cultured in growth medium containing 10 µM BrdU for 16 h before immunofluorescence staining was carried out with a rat anti-BrdU antibody and a rhodamine-conjugated donkey anti-rat secondary antibody as described (25). For colony formation and flat-cell formation assays, the transfected cells were cultured in growth medium containing G418 (500 µg/ml) for 14 days and then scored under a light microscope as described previously (51).

Reporter assays. C33A cells and Rb^/ fibroblasts cultured in growth medium in six-well plates to 50% confluence were cotransfected with the pRSV--gal plasmid plus the reporter plasmids in the presence or absence of the plasmids expressing the indicated activators or repressors. The promoters used in the reporters were Gal4/Luc, a minimal thymidine kinase promoter containing five Gal4-binding sites; E2F/Luc, a minimal promoter with only the TATA box and three E2F-binding sites; DHFR/Luc, a wild-type dihydrofolate reductase (DHFR) promoter, and DHFR(-E2F)/Luc, a mutant DHFR promoter without functional E2F-binding sites. When indicated, trichostatin A (TSA) at 100 ng/ml was added to the culture medium at 24 h after transfection. For C33A and Saos-2 cells, 48 h after transfection the transfected cells were harvested and luciferase activities were determined and normalized with the cotransfected -galactosidase activities as described (52). For Rb^/ fibroblasts, the transfected cells were washed once with PBS and cultured in differentiation medium for 24 h after transfection. After 72 h in differentiation medium, the cells were harvested and luciferase activities were determined and normalized with the cotransfected -galactosidase activities.

Myogenic conversion assays. For the determination of myogenic conversion, Rb^/ fibroblasts cultured in six-well plates to 50% confluence were cotransfected with the pH2B-GFP plasmid (24) plus plasmids expressing the indicated effector proteins. Twenty-four hours after transfection, the cells were washed once with PBS and cultured in the differentiation medium for an additional 36 or 72 h as indicated. Immunofluorescence staining was carried out with the monoclonal anti-myosine heavy chain (MHC) antibody MF-20 (1). For the determination of BrdU incorporation after serum stimulation in MyoD-converted RbB^/ fibroblasts, the Rb^/ fibroblasts were cultured and transfected as described above, except that the pH2B-GFP plasmid was omitted. After 72 h in the differentiation medium, the transfected cells were washed once with PBS and cultured in the growth medium containing 15% fetal bovine serum and 10 µM BrdU. Twenty hours later, double immunofluorescence staining was carried out using the rat anti-BrdU antibody with the rhodamine-conjugated donkey anti-rat secondary antibody and a rabbit anti-MHC antibody with a fluorescein isothiocyanate-conjugated goat anti-rabbit secondary antibody.

/

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Targeted mutations of LXCXE-binding pocket in RB. The crystal structure of the RB A-B pocket domain bound to a nine-residue peptide from the viral E7 protein has been determined (29). Importantly, the binding site for the LXCXE peptide in the B region of RB has been defined (29). Based on the crystal structure, we targeted three RB amino acids (N757, I768, L769) for mutation, with the goal of disrupting the LXCXE-binding function yet retaining the structural integrity of RB. We generated four mutants, RB(N757F), RB(L769D), RB(1768A/L769D), and RB(I768D/L769D). Amino acid N757 is conserved among all pocket proteins and is located at the center of the shallow groove that binds to the LXCXE peptide. The side-chain of N757 forms two critical hydrogen bonds with C24 of the E7 LXCXE motif. We mutated N757 to F to disrupt hydrogen bond formation and to provide some steric hindrance with a bulky side chain. Indeed, RB(N757F) was defective in binding the E7 protein (Fig. 1A, compare lanes 5 and 6). Amino acids I768 and L769 of RB interact with L22 of the E7 LXCXE motif (29). Interestingly, RB(L769D) did not abolish binding to E7 (Table 1). The double mutants of L769D and I768A or I768D, however, no longer bound to E7 (Table 1). Thus, the LXCXE-binding site can be disrupted by targeted mutations based on the crystal structure.

View larger version (26K): [in this window] [in a new window]   FIG. 1.   RB(N757F) binds E2F-1 but not E7 or HDAC1. (A) RB(WT) or RB(N757F) was each expressed in C33A cells. GST-E7 or GST-E2F-1 proteins were purified from bacteria and immobilized on glutathione-agarose beads. Whole-cell lysates (WCL) from cells transfected with vector (V) or the indicated RB expression plasmids were used in GST pulldown assays as described in Materials and Methods. RB(WT) and RB(N757F) were expressed to similar levels in C33A cells (lanes 2 and 3). With equal amounts of GST fusion proteins (amido black staining, lanes 1 to 9), GST-E7 bound RB(WT) but not RB(N757F) (lanes 5 and 6), whereas GST-E2F-1 bound to RB(WT) and RB(N757F) (lanes 8 and 9). (B) GST fusion proteins containing the large pocket of RB (A-B and C domains), either WT or N757F, were purified from bacteria. The HA-E2F-1 (lanes 1 and 2), HA-HDAC1 (lanes 3 and 4), and HA-HDAC1(L) (lanes 5 and 6) proteins were each expressed in C33A cells (middle panel, WCL), and tested for binding to GST-RB (bottom panel). The HDAC1(L) protein lacks the LXCXE motif of HDAC1. These HA-tagged proteins were detected by immunoblotting with an anti-HA antibody. (C) GST-HDAC1 was purified from bacteria (top); RB(WT) (lane 1) and RB(N75F) (lane 2) were expressed in C33A cells (middle) and tested for binding to GST-HDAC1. The bound RB protein was detected by immunoblotting with an anti-RB antibody (bottom). (D and E) Coimmunoprecipitation of RB with E7, E2F-1, and HDAC1. C33A cells were cotransfected with the indicated expression plasmids for RB or HA-tagged E7, E2F-1, HDAC1, or HDAC1(L). Whole-cell lysates were immunoprecipitated with either anti-HA (D) or anti-RB (E) and immunoblotted for the coprecipitated proteins. Note that the expression levels of RB(WT) and RB(N757F) were similar in C33A cells. Coimmunoprecipitation with E2F-1 was observed for both RB(WT) and RB(N757F) (lanes 4 and 5). Coimmunoprecipitation with E7 or HDAC1 was observed for RB(WT) but not RB(N757F) (lanes 1, 2, 5, and 6). HDAC1(L) did not coimmunoprecipitate with either RB(WT) or RB(N757F) (lanes 7 and 8).

RB(N757F) inhibits the transactivation function of E2F. A major function of RB is to repress E2F-dependent transcription. This can be achieved by two mechanisms: a direct inhibition of E2F by RB (17, 18) or the recruitment of HDAC1 by RB to E2F-regulated promoters (3, 34, 35). Because RB(N757F) binds to E2F-1 but not HDAC1, it should be crippled in its ability to repress E2F-dependent transcription. We used three different transcription reporters to compare the transcription repression functions of RB(WT) and RB(N757F) (Fig. 2). We also included TSA, which inhibits HDACs, to determine the contribution of HDAC in RB-mediated transcription repression. The first reporter contains five GAL4-DNA binding sites linked to a minimal thymidine kinase promoter and was strongly activated by a hybrid protein of GAL4-DNA binding domain fused to the transactivation domain of E2F-1 (Gal4-E2F-1). In either the absence or the presence of RB, TSA did not affect transcription from this reporter. RB(WT) inhibited the GAl4-E2F-1 activity by about 50%, and RB(N757F) achieved a similar level of inhibition (Fig. 2A). These results were consistent with the ability of RB(N757F) to bind E2F-1 and showed that RB(N757F) could inhibit the transactivation function of E2F-1. The second reporter contains three E2F-binding sites linked to a minimal promoter containing a TATA box and was activated by the coexpression of E2F-1 and DP-1 (Fig. 2B). In the absence of RB, transcription from this reporter was not affected by TSA. Addition of RB(WT) strongly inhibited transcription from this reporter, and TSA partially reversed the negative effect of RB (Fig. 2B). With this reporter, RB(N757F) showed a much reduced repression of transcription, and its effect was not sensitive to TSA (Fig. 2B). Similar results were obtained with a third reporter containing the cellular DHFR promoter, which can be repressed by RB(WT) and RB(N757F) in an E2F-dependent manner (Fig. 2C and D). RB(WT) and RB(N757F) were expressed at comparable levels in these transient-cotransfection experiments, and both proteins associated with the coexpressed E2F (E2F-1-DP-1) as determined by E2F-DNA binding (Fig. 2E). Finally, we found that both RB(WT) and RB(N757F) repressed the wild-type DHFR promoter in a dosage-dependent manner in the absence of ectopically expressed E2F (Fig. 2F). These results confirmed that RB(WT) represses E2F-regulated promoters through HDAC-dependent and -independent mechanisms (2, 8, 41). The RB(N757F) mutant did not bind to HDAC1 and thus was compromised in its ability to repress E2F transcription.

View larger version (38K): [in this window] [in a new window]   FIG. 2.   Partial inhibition of E2F-dependent transcription by RB(N757F). (A) Inhibition of Gal4-E2F-1 by RB. Cotransfection with the indicated reporter and expression plasmids and the quantitation of reporter activity was described in Materials and Methods. Cells were also treated with TSA (100 ng/ml) to inhibit HDAC activity. Note that RB(WT) and RB(N757F) inhibited Gal4-E2F-1 with similar avidity. (B) Repression of a minimal promoter containing E2F-binding sites. Transfection and reporter activity assays were carried out as described in Materials and Methods. Note that RB(WT) repressed this reporter more efficiently than RB(N757F), but the repression activity of RB(N757F) was not sensitive to TSA. (C) Transcriptional repression of wild-type DHFR promoter. RB(N757F) repressed the DHFR promoter with reduced avidity, and this repression was not sensitive to TSA. (D) Transcriptional repression of a mutant DHFR promoter without E2F-binding site. Note that both RB(WT) and RB(N757F) failed to repress this promoter. (E) RB(N757F) interacts with E2F. HA-E2F-1 and DP-1 were coexpressed with RB(WT) or RB(N757F) in C33A cells. Whole-cell lysate was immunoprecipitated with an anti-RB antibody, the associated proteins were released by deoxycholate, and the E2F DNA binding activity was determined by electrophoretic mobility shift assay using a 32P-labeled oligonucleotide containing an E2F-binding site. Where indicated, unlabeled probe (cold probe) was used at a 50-fold molar excess to demonstrate the specificity of the shifted bands. The levels of RB and HA-E2F-1 were determined by immunoblotting of whole-cell lysates. Note that both RB(WT) and RB(N757F) caused an increase in the levels of E2F-1, consistent with previous reports that RB binding stabilizes the E2F-1 protein (19). (F) Dosage-dependent transcriptional repression of wild-type DHFR promoter in the absence of ectopically expressed E2F. Experiments were performed without the cotransfection of E2F-1 and DP-1 expression plasmids as described in Materials and Methods. In panels A to D and F, error bars show standard deviations.

RB(N757F) inhibits cell cycle progression. The E2F transcription factor regulates a number of genes that are important for the progression from G₁ into the S phase of the cell cycle (6, 30, 44). Repression of E2F-mediated transcription by RB, therefore, can cause a cell cycle arrest in G₁. To determine if RB(N757F) can inhibit cell cycle progression, we examined its ability to inhibit BrdU incorporation in two systems (Fig. 3A to C). The first is based on the Rb-negative Saos-2 cells, which become G₁ arrested upon the reintroduction of RB (44). The second is based on Rb-deficient mouse embryo fibroblasts, which become G₁-arrested upon the coexpression of RB and p16Ink4a to inhibit RB phosphorylation (13, 28, 32). A plasmid expressing a green fluorescence protein (GFP) fused to histone 2B (H2B-GFP), which was tightly associated with the cell nucleus (24), was also included to mark the transfected cells (Fig. 3A). In Saos-2 cells, expression of RB(WT) or RB(N757F) inhibited BrdU incorporation to a comparable extent. The inhibition of BrdU was dependent on RB because H2B-GFP-negative cells in the same fields incorporated BrdU similar to vector-transfected cells (Fig. 3B). In Rb-deficient fibroblasts, expression of p16Ink4a alone or RB(WT) alone did not inhibit DNA synthesis. As demonstrated previously, the combined expression of RB(WT) and p16Ink4a inhibited BrdU incorporation (Fig. 3C), because p16Ink4a can block the phosphorylation of RB and thus maintain the pocket functions of RB. Under the condition of coexpression with p16Ink4a, RB(N757F) again inhibited DNA synthesis as efficiently as RB(WT) (Fig. 3C).

View larger version (29K): [in this window] [in a new window]   FIG. 3.   RB(N757F) can inhibit cell proliferation. (A) Inhibition of BrdU incorporation in Saos-2 cells. The RB-negative human Saos-2 cells were cotransfected with a plasmid expressing H2B-GFP plus the indicated expression plasmids. Forty-eight hours after transfection, cells were labeled with BrdU for 16 h and then stained with an anti-BrdU antibody plus Hoechst. (B) Quantitation of BrdU-positive cells. A total of 300 H2B-GFP-positive or H2B-GFP-negative cells from twenty randomly selected fields were counted. The percentage of BrdU-positive cells in vector-transfected culture was set at 100. The values represent data collected from two independent experiments. (C) Inhibition of BrdU incorporation in Rb/ fibroblasts. Fibroblasts derived from Rb/ mouse embryos were cotransfected with the indicated plasmids, and the incorporation of BrdU was measured. Cotransfection with p16, an inhibitor of Cdk4, is to block RB phosphorylation. The percentage of BrdU-positive cells in vector-transfected culture was set at 100. (D) Inhibition of colony formation. Saos-2 cells were transfected with the indicated plasmids and cultured in G418-containing medium for 2 weeks to select for neomycin-resistant colonies. The number of colonies formed in vector-transfected cells was set at 100%. (E) Induction of flat-cell formation. Saos-2 cells were transfected with the indicated plasmids and selected in G418-containing media as in (D). The number of flat cells were counted under phase contrast microscope. Flat cells were defined by single cells with the size of a small colony. The number of flat cells in RB(WT) transfected culture was set at 100%. In panels B to E, error bars show standard deviations.

RB(N757F) cooperates with MyoD. The requirement of RB for the proper differentiation of muscle has been well documented (37, 42, 55). We therefore examined the ability of the RB(N757F) mutant to stimulate MyoD-dependent transcription. We used two different reporters, one with the promoter of muscle creatine kinase (MCK/Luc) and the other with four E box binding sites (4RE/Luc), to measure the effect of MyoD and RB in Rb-deficient fibroblasts (Fig. 4A). In the absence of RB, MyoD stimulated these reporters by about fivefold. Coexpression with RB(WT) caused a two- to threefold further increase in the reporter activity. Coexpression with RB(N757F) also increased the reporter activity, but it was less active than RB(WT) in cooperating with MyoD (Fig. 4A). The expression of RB(WT) or RB(N757F) did not affect the expression of the cotransfected MyoD in these experiments (not shown). We also examined the activation of the endogenous MHC gene (Fig. 4B and C). Similar to the reporter assays, RB(WT) enhanced the MyoD-dependent activation of MHC by two- to threefold (Fig. 4C). Moreover, the MHC-positive cells derived from the coexpression of MyoD and RB(WT) were elongated with more than one nucleus, indicative of a more advanced differentiation (Fig. 4B). Coexpression of RB(N757F) with MyoD also stimulated MHC-positive cells and the differentiated morphology (Fig. 4B). We noticed that the appearance of MHC-positive myoctyes was delayed with RB(N757F) relative to RB(WT) (Fig. 4C). These results showed that RB(N757F) could collaborate with MyoD in stimulating muscle differentiation, albeit with reduced avidity than RB(WT).

View larger version (28K): [in this window] [in a new window]   FIG. 4.   RB(N757F) cooperates with MyoD to stimulate myogenic differentiation. (A) Stimulation of MyoD-dependent transcription. Rb/ fibroblasts were cotransfected with the indicated expression plasmids and two different MyoD-dependent repeaters. The 4RE/Luc reporter contains four E box binding sites; the MCK/Luc contains the promoter of the muscle creatine kinase gene. Transfected cells were transferred into differentiation medium for 48 h, and the luciferase activity was quantitated as described in Materials and Methods. Values represent means and standard deviations (error bars) of three independent transfections. (B) Stimulation of differentiated phenotype. Rb/ fibroblasts were cotransfected with a plasmid expressing H2B-GFP plus the indicated expression plasmids. Twenty-four hours after transfection, cells were cultured in differentiation medium (low serum) for an additional 72 h before immunofluorescence staining with a monoclonal anti-MHC antibody was performed (Materials and Methods). (C) Quantitation of MHC-positive cells. Transfected cells were collected at 36 or 72 h after the transfer into differentiation medium. None of the H2B-GFP-positive cells in vector-cotransfected cultures became MHC positive. With MyoD alone, between 10 and 15% of the transfected cells became MHC positive, and this value was set as 1. The stimulation of MHC-positive cells by RB was determined in two independent transfection experiments. Error bars, standard deviations.

RB(N757F) does not establish permanent growth arrest in myocytes. Previous reports have established that RB is required for muscle cells to permanently withdraw form the cell cycle (37, 42, 55). Stimulation of differentiated muscle cells with serum does not activate DNA synthesis and cannot induce the phosphorylation of RB. However, serum stimulation will reactivate DNA synthesis in Rb-deficient muscle cells. To test whether RB(N757F) could establish a permanent growth arrest during muscle differentiation, we restimulated the MyoD-converted myocytes with serum and measured the incorporation of BrdU into MHC-positive cells (Fig. 5). In vector-transfected cultures of Rb-deficient fibroblasts, no MHC-positive cells were observed, and these fibroblasts incorporated BrdU upon serum stimulation (Fig.5). MHC-positive cells converted by MyoD alone also incorporated BrdU. The percentage of BrdU-positive cells was similar to the percentage of MHC-negative cells in the same culture (Fig. 5). Coexpression of MyoD with RB(WT), as expected, blocked the incorporation of BrdU in MHC-positive cells (Fig. 5). However, MHC-positive cells converted by the coexpression of MyoD with RB(N757F) incorporated BrdU when they were restimulated with serum (Fig. 5). Despite its ability to cooperate with MyoD in the stimulation of differentiation, RB(N757F) was unable to prevent the reactivation of DNA synthesis in the differentiated cells.

View larger version (41K): [in this window] [in a new window]   FIG. 5.   RB(N757F) cannot establish permanent growth arrest in myocytes. (A) Incorporation of BrdU in differentiated myocytes. Rb/ fibroblasts were converted to myocytes by the ectopic expression of MyoD followed by culturing in differentiation medium (low serum) for 72 h as described in Materials and Methods. The transfected cultures were then stimulated with 15% serum and incubated with BrdU for 24 h. Cells were then fixed and stained with Hoechst and anti-MHC and anti-BrdU antibodies as indicated in the figure. Arrows denote BrdU-positive nuclei in MHC-positive cells, whereas arrowheads denote BrdU-negative nuclei in MHC-positive cells. (B) Quantitation of BrdU-positive cells from three independent experiments. The percentage of BrdU-positive cells in vector-transfected culture was set as 100. MHC-positive cells converted by MyoD alone responded to serum with an efficiency indistinguishable from MHC-negative cells in the same culture. MHC-positive cells converted by MyoD and RB(WT) did not respond to serum. MHC-positive cells converted by MyoD and RB(N757F) responded to serum and incorporated BrdU. Thus, RB(N757F) was defective in establishing permanent cell cycle withdrawal in myocytes. Error bars, standard deviations.

View larger version (51K): [in this window] [in a new window]   FIG. 6.   RB(N757F) is hyperphosphorylated upon serum restimulation in CC42 myocytes. (A) Incorporation of BrdU in differentiated CC42 myocytes. The Rb/ CC42 myoblasts were cotransfected with a plasmid expressing a puromycin-resistant gene plus the indicated expression plasmids. Twenty-four hours later, the transfected cells were cultured in differentiation medium (low serum) containing puromycin (1.5 µg/ml) for 72 h as described in Materials and Methods. The cells were subsequently stimulated with 15% serum and incubated with BrdU for 24 h before being fixed and stained with Hoechst and anti-MHC and anti-BrdU antibodies. BrdU-positive cells from three independent experiments were then quantitated. The percentage of BrdU-positive cells in vector-transfected culture was set as 100. (B) Hyperphosphorylation of RB(N757F) in CC42 myocytes after serum restimulation. CC42 myoblasts were cotransfected, selected for puromycin resistance in differentiation medium, and stimulated with serum as described for panel A. Whole-cell lysates from the transfected cells were then subjected to immunoblotting with an anti-RB antibody to detect hyperphosphorylated RB (Hyper-RB) and hypophosphorylated RB (Hypo-RB). (C) Expression of p21 in differentiated CC42 myocytes. CC42 myoblasts were cotransfected with a plasmid expressing a puromycin-resistant gene plus an empty vector (lanes 2 to 4), RB(WT) (lanes 5 to 7), RB(N757F) (lanes 8 to 10), or (PSM.9I)RB(N757F) (lanes 11 to 13). Transfected cells were selected by growth in medium containing puromycin (1.5 µg/ml) 24 h after transfection. Whole-cell lysates from transfected cells cultured in growth medium (lanes 2, 5, 8, and 11), in differentiation medium for 3 days (lanes 3, 6, 9, and 12), or in differentiation medium for 3 days and then in growth medium for 1 day (lanes 4, 7, 10, and 13) were subjected to SDS-PAGE and immunoblotting using a polyclonal anti-p21 antibody. A monoclonal anti-tubulin antibody (Sigma) was used to ensure equal loading. Note that p21 was absent in the whole-cell lysate prepared from p21/ fibroblasts (lane 1).

Serum-induced phosphorylation of RB(N757F) in differentiated myocytes. Because the hypophosphorylayted form of RB(N757F) can establish a permanent growth arrest in Saos-2 cells (Fig. 3), we reasoned that serum restimulation of myocytes might have caused the phosphorylation and the inactivation of RB(N757F). To examine this possibility, we expressed RB(WT) or RB(N757F) in the Rb-deficient CC42 myocytes, allowed them to differentiate, restimulated differentiated myocytes with serum, and then examined the phosphorylation status of the RB protein. No RB protein was detected in vector-transfected CC42 cells (Fig. 6B, lane 1). The RB(WT) and RB(N757F)proteins were expressed to similar levels in CC42 myocytes (Fig. 6B, compare lanes 2 to 7 with lanes 8, 10, and 12). Only a trace of RB(WT) was phosphorylated upon serum stimulation of differentiated myotubes (Fig. 6B, lanes 3, 5, and 7), consistent with the inhibition of BrdU incorporation (Fig. 6A). By contrast, a majority of RB(N757F) became hyperphosphorylated in differentiated myocytes following serum stimulation (Fig. 6B, lanes 9, 11, and 13). Thus, the failure of RB(N757F) to establish a permanent growth arrest in differentiated myotubes can be attributed to its susceptibility to serum-induced phosphorylation. To further demonstrate this point, we introduced the N757F mutation into a phosphorylation site-mutated (PSM.9I)RB that cannot be inactivated by phosphorylation (25, 26). (PSM.9I)RB(N757F) was not phosphorylated in the CC42 myocytes, either before or after stimulation (Fig. 6B, lanes 14 to 19). Consistent with the growth suppression of Saos-2 cells, (PSM.9I)RB(N757F) rescued the defect of CC42 myocytes and established a growth arrest that cannot be reversed by serum stimulation (Fig. 6A). These results showed that the LXCXE-binding function of RB is dispensable for growth arrest per se, but this pocket is essential for RB to establish the inhibition of its own phosphorylation in differentiated myocytes.

Phosphorylation of RB(N757F) occurred despite the expression of p21Cip1. Previous studies have identified p21Cip1 as the inhibitor of RB phosphorylation during myogenic differentiation (13, 14). The induction of p21Cip1 is mediated by the myogenic transcription factor MyoD and can occur in Rb-deficient myocytes (15, 48). To examine whether p21Cip1 expression might account for the phosphorylation of RB(N757F), we examined the levels of p21Cip1 in CC42 myocytes transfected with RB(WT), RB(N757F), or (PSM.9I)RB(N757F) (Fig. 6C). The basal level of p21Cip1 was low in CC42 cells prior to the induction of differentiation, irrespective of the expression of RB (Fig. 6C, lanes 2, 5, 8, and 11). A detectable increase in p21Cip1 was observed with CC42 cells transfected with (PSM.9I)RB(N757F) prior to differentiation (Fig. 6 C, lane 11), possibly because this RB mutant could block the mitogenic effect of serum and hence partially activated differentiation in growth medium. Following differentiation, the p21Cip1 levels increased in CC42 myocytes, irrespective of the expression of RB (Fig. 6C, lanes 3, 6, 9, and 12). When differentiated myocytes were stimulated with serum, we found that p21Cip1 levels were maintained and this was again irrespective of which form of RB was expressed (Fig. 6C, lanes 4, 7, 10, 13). The anti-p21 antibody did not detect any signal in lysates prepared from p21Cip1-deficient fibroblasts (Fig. 6C, lane 1). Taken together, these results show that RB(N757F) did not interfere with the expression of p21Cip1 in differentiated myocytes and suggested that p21Cip1 alone was not sufficient to block serum-induced phosphorylation of RB in myocytes.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

RB and RB-related pocket proteins are conserved through evolution, and they play important roles in the regulation of proliferation and differentiation. These pocket proteins contain a conserved A-B domain that binds to a number of transcription regulators. The crystal structure of the A-B domain has defined at least one highly conserved pocket that binds to the LXCXE peptide motif (29). Using the crystal structure as a guide, we have been able to create a mutant, RB(N757F), that is specifically defective in binding the LXCXE motif. We have shown that RB(N757F) does not bind to the viral E7 protein or the cellular HDAC1 but retains the ability to bind E2F-1 (Fig. 1; Table 1). The properties of RB(N757F) demonstrate that the binding sites for E2F and LXCXE motif in RB are separate and distinct structures. Because the LXCXE-binding pocket can be inactivated by the substitution of N757, which is conserved in all of the known A-B domains, this mutation may also be used to inactivate the LXCXE-binding pocket of other RB-related proteins.

The growth suppression function of the RB-pocket proteins is associated with their ability to repress E2F-dependent transcription. This is because E2F regulates genes that are required for DNA synthesis (6, 30, 44). Previous work has established two mechanisms of repression by these pocket proteins, a direct inhibition of E2F transactivation function and the recruitment of transcription corepressors, such as HDAC1 to E2F-regulated promoters (2, 44). With RB(N757F), which can inhibit the transactivating function of E2F-1 (Fig. 2; Table 1) but cannot recruit HDAC1 (Fig, 1 and 2; Table 1), we were able to investigate the relative contribution of these two mechanisms in growth suppression. With several well-established assays, we found RB(N757F) and RB(WT) to have comparable activity in suppressing cell growth (Fig. 3; Table 1). These observations suggest that the LXCXE-binding pocket is dispensable for growth suppression under conditions of RB phosphorylation inhibition (e.g., in Saos-2 cells and cotransfection with p16Ink4a). Mutation of N757 does not abolish the binding of RB to BRG or CTIP (Doug Dean, personal communication), which are two other transcription repressors recruited by RB to E2F-regulated promoters (36, 46). However, our results strongly suggest that HDAC1-mediated repression through RB is dispensable for growth arrest.

The RB(N757F) mutant also contains an intact C-terminal region that is not a part of the A-B domain (29). The A-B domain and the C-terminal region of RB are both required to suppress cell growth. The crystal structure has demonstrated that the A-B domain can fold into pockets without the C region (29). However, neither the A-B domain itself nor the coexpression of A-B domain with the C-region fragments can suppress cell growth (50). Thus, the multiple pockets in the A-B domain and the C region must cooperate to achieve growth suppression (49, 51, 52). Previously, we and others have shown that an RB mutant (R661W), which was unable to bind either E2F or the LXCXE motif, can still suppress cell growth albeit with reduced activity (43, 52). The growth suppression function of RB(R661W) can be inactivated by specific mutations in the C region (49, 52). Taken together, the accumulated data suggest that the growth suppression function of RB can be mediated by multiple mechanisms through the different pockets in RB.

We have identified an important defect with RB(N757F); it cannot establish the irreversible growth arrest associated with terminal differentiation (Fig.5 and 6; Table 1). This defect is not intrinsic, because RB(N757F) can establish long-term growth arrest when its phosphorylation is blocked (Fig. 3) and (PSM.9I)RB(N757F) can allow myocytes to withdraw from the cell cycle permanently (Fig. 6C). The defect of RB(N757F) lies with its inability to establish a serum-refractory state in differentiated myocytes (Fig. 6). Serum restimulation of myocytes did not activate the phosphorylation of RB(WT), but it did induce the phosphorylation of RB(N757F) (Fig. 6B). Taken together, the results have identified a critical function of the LXCXE-binding pocket of RB in blocking serum-induced phosphorylation of RB in differentiated cells. The inhibition of RB phosphorylation in differentiated myocytes has been attributed to the upregulation of p21Cip1 (14, 15, 47). The upregulation of p21Cip1 is mediated by MyoD (15, 38), and this occurs without RB (48). Interestingly, RB(WT) or RB(N757F) did not affect the levels of p21Cip1 in differentiated CC42 myocytes with or without serum stimulation (Fig. 6C). Despite the similar levels of p21Cip1, myocytes with RB(N757F) allowed RB phosphorylation following serum stimulation (Fig. 6C). These results suggest that the upregulation of p21Cip1 may inhibit RB phosphorylation during the initial phase of differentiation, but it is not sufficient to maintain RB in a dephosphorylated state.

At present, we do not know the precise mechanism by which RB(WT) establishes the serum-refractory state in differentiated myocytes. It is of interest to note that the differentiated myotubes of newt can activate RB phosphorylation and reenter DNA synthesis in response to serum (45). The myotubes of newt, which can regenerate its limbs, thus resemble mammalian myocytes expressing the RB(N757F) mutant. With the mammalian RB, the LXCXE-binding function has evolved to prevent serum from inducing RB phosphorylation in terminally differentiated myocytes. Because the upregulation of p21Cip1 is not dependent on RB and because the levels of p21Cip1 do not correlate with the phosphorylation of RB (Fig. 6C), this Cdk inhibitor is not likely to be the target of regulation by RB(WT) in myocytes. With the RB(N757F) mutant, it should be possible to identify the critical RB targets that are responsible for the resistance of myocytes to the mitogenic stimulation of RB phosphorylation.