RB Family Tumor Suppressor Activity May Not Relate to Active Silencing of E2F Target Genes

Loss of the retinoblastoma suppressor gene, RB1, is a frequent event in human cancer. pRB and its homologs p130 and p107 form the family of pocket proteins and play a key role in cell-cycle regulation. They collectively regulate the family of E2F transcription factors, whose activity is essential for the G₁–S transition. Eleven E2F transcription factors have been identified, classified as “activator E2Fs” (E2F1, E2F2, and E2F3a) and “repressor E2Fs” (E2F3b, E2F4, E2F5, E2F6a, E2F6b, E2F7a, E2F7b, and E2F8). Whereas pRB can interact with E2F1-4, p130 and p107 interact with E2F4 and E2F5. E2F6-8 do not bind pocket proteins (1–3).

Pocket protein–E2F complex formation inhibits transcription of E2F target genes via several mechanisms. First, pocket protein binding blocks E2F's transactivation domain, preventing E2F-mediated transcription. The balance between pocket protein–bound and free E2Fs is regulated by cyclin-dependent kinases that phosphorylate pocket proteins, causing release of E2Fs. This enables activating E2Fs to promote transcription and cell-cycle progression (3). Second, pocket protein “repressor” complexes pRB–E2F3b, pRB–E2F4, p130/p107–E2F4, and p130/p107–E2F5 can recruit chromatin remodeling proteins to E2F target genes by the capacity of pocket proteins to bind proteins containing an LxCxE-like motif (x encoding any amino acid; ref. 4). The LxCxE motif, originally identified in viral oncoproteins, is present in a variety of chromatin remodeling proteins that favor a chromatin state incompatible with transcription (4). These include histone deacetylase 1 (HDAC1), HDAC2, heterochromatin protein 1 (HP1), the histone methyltransferase Suv39h1, and the CtBP-interacting protein (CtIP; refs. 3–7).

Consistent with their role in cell-cycle control, loss of pocket proteins initiates tumorigenesis. In mice, Rb loss induces pituitary and thyroid tumors, and the tumor spectrum is extended by the additional loss of p130 or p107 (8–10). However, it is unclear which of the E2F-regulating activities of pocket proteins is critical for tumor suppression. One view is that regulation of “activator,” rather than “repressor” E2Fs, mediates pRB's tumor suppressor activity (11). However, pocket protein–chromatin remodeling complexes have been implicated in (irreversible) cell-cycle arrest and senescence, which are considered critical for tumor suppression. The recruitment of p130–E2F4 to silenced promoters in G₀–G₁-arrested mouse and human cells suggests active repression (12, 13). Furthermore, pRB and LxCxE-containing proteins have been implicated in repression of E2F target genes upon serum starvation (14, 15), RAS^V12 expression (16), and ectopic expression of p16^INK4A (17).

Although LxCxE-containing proteins have thus been implicated in processes counteracting transformation, the requirement for these proteins to suppress tumorigenesis in vivo remains elusive. To address this issue, we generated mouse embryonic fibroblasts (MEF) and mice expressing a pRB variant in which asparagine at position 750 was replaced for phenylalanine. pRB^N750F protein still inhibited E2F transactivation activity, but was impaired in binding an LxCxE-containing protein and therefore likely unable to recruit chromatin remodelers to E2F-regulated promoters. Because of functional compensation within the pocket protein family (this communication; ref. 8), the essential role of p130-repressor complexes in G₀–G₁ (12, 13), and the ability of p130 to recruit chromatin remodeling proteins via its LxCxE binding site (18, 19), we anticipated the phenotypic consequences of pRB^N750F to be masked by p130 activity. Therefore, we also generated Rb^N750F/N750Fp130^−/− MEFs and Rb^N750F/p130-mutant mice. Our results show that the pRB–LxCxE interaction contributes to cell-cycle arrest upon RAS^V12 expression or γ-irradiation. Remarkably, attenuation of LxCxE-dependent interactions did not promote tumor formation, even in the absence of p130, suggesting that pRB's tumor suppressor activity mainly operates via inhibition of activator E2Fs or an LxCxE-independent repressor mechanism.

The 6xE2F-luciferase reporter, CMV-HA-E2F1, CMV-HA-E2F4, CMV-Renilla, and pJ3-mRb were provided by Dr. R. Bernards (Division of Molecular Carcinogenesis, The Netherlands Cancer Institute, Amsterdam, The Netherlands); GAL4-TK-luciferase by Dr. R. Medema (Division of Cell Biology, The Netherlands Cancer Institute); and pECE-pRbΔcdk-HA by Dr. J. Lukas (Novo Nordisk Foundation Center for Protein Research, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark). Addition of an N-terminal HA-tag to pRB, mutation of the Rb cDNA and cloning into pcDNA3.1(−) is described in Supplementary Materials and Methods. GAL4-pRB vectors were generated by cloning PCR-amplified Rb fragments (wild-type or mutant) into pM (Clontech; Supplementary Materials and Methods). pSG5-Tag was provided by Dr. J. A. DeCaprio (Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA), pBABE-RAS^V12-puro by Dr. T. Brummelkamp (Division of Biochemistry, The Netherlands Cancer Institute), and pCL-Eco by Dr. D. Peeper (Division of Molecular Oncology, The Netherlands Cancer Institute). The pEYK-TBX2 vector was previously isolated from the pEYK-MCF7 library (20) obtained from Dr. G. Q. Daley (Department of Pediatric Hematology/Oncology, Children's Hospital Boston and Dana-Farber Cancer Institute, Boston, MA).

Rb^−/− 3T9 cells were seeded onto a 12-well plate and transfected the next day with 0.4 μg DNA per well using FuGENE6 Transfection Reagent (Roche) with a DNA-FuGENE ratio of 1 μg to 3 μL. All transfections included either 6xE2F-luciferase or GAL4-TK-luciferase reporter plasmid, Renilla luciferase expression plasmid and pcDNA3.1(−)-HA-mRb or pM-GAL4-pRB plasmids to a total of 0.4 μg DNA/well. For detailed concentrations, see Supplementary Materials and Methods. Twenty-four hours after transfection, cells were washed in PBS, lysed, and measured for Luciferase and Renilla luminescence activity using the Dual-Luciferase Reporter Assay System (Promega E1910). Luciferase luminescence was divided by Renilla luminescence.

Crossings of Rb^N750F/wt animals (21) and Rb^wt/− animals (22) with p130^wt/− mice (22), and intercrossings of the resulting offspring, enabled us to isolate Rb^N750F/N750Fp130^−/−, Rb^N750F/N750Fp130^wt/wt, Rb^−/−p130^wt/wt, Rb^wt/wtp130^−/−, and Rb^wt/wtp130^wt/wt MEFs at embryonic day (E) 14.5 and Rb^−/−p130^−/− at E13.5. Rb^−/−p130^−/−p107^−/− MEFs were previously generated (22).

MEFs were cultured with 100 units/mL penicillin (Invitrogen/Gibco) and 100 μg/mL streptomycin (Invitrogen/Gibco; ref. 22). Cells were plated and treated as follows: Mitogen deprivation: 8 × 10⁵ cells/10-cm dish; 3 hours after plating, cells were washed in PBS and medium without FCS was added. Contact inhibition: 8 × 10⁵ cells/10-cm dish; cells were analyzed 10 days after plating. γ-irradiation: 5 × 10⁵ cells/10-cm dish; cells were irradiated the next day with 5.5, 10, or 15 Gy and analyzed 16 hours after irradiation. All assays: cells were harvested for FACS at the indicated time points and, if applicable, 10 μmol/L bromodeoxyuridine (BrdUrd) was added 1 hour before harvesting. BrdUrd labeling and FACS: See Supplementary Materials and Methods.

Production of retroviruses, retroviral infections, and soft agar plating was performed as described (20). For serial infections, MEFs were cultured in nonvirus-containing media for at least 36 hours between infections and reseeded before infection to obtain optimal cell density.

Cells were seeded in triplicate at a density of 2.5 × 10⁴ cells per well of a 12-well plate (Corning Incorporated; 3512). At the indicated time points (time point 0: 3 hours after seeding), cells were fixed in 4% formaldehyde/PBS, washed in demineralized water, stained with 0.1% crystal violet solution, washed and dried (dark area). The dye was extracted in 1 mL 10% acetic acid per well, and the optical density at 590 nm was measured using a microplate reader (M200 Tecan).

Protein isolation, separation on 3% to 8% Tris-Acetate and 4% to 12% Bis-Tris NuPage gradient gels (Invitrogen), and immunoblotting were performed according to standard protocols. Cells were lysed in 150 mmol/L NaCl, 50 mmol/L Hepes pH 7.5, 5 mmol/L EDTA, 0.1% NP-40, and 1 tablet complete protease inhibitor cocktail (Roche) per 50 mL. Protein concentrations were determined using Bio-Rad protein assay (Bio-Rad).

For immunoprecipitation experiments, C33A cells cotransfected with pSG5-TAg plus pcDNA3.1(−)-HA-mRb (wild-type, mutant, or empty) were used. Protein isolation was performed in the presence of 5 mmol/L NaF, 0.5 mmol/L vanadate, 20 mmol/L β-glycerolphosphate, and 1 mmol/L phenylmethylsulfonylfluoride. Lysates were incubated with anti-HA antibody and 50% protein A/G beads (Santa Cruz) in lysis buffer, separated on 12% polyacrylamide gels, and immunoblotted. For antibodies, see Supplementary Materials and Methods.

Rb^N750F/N750Fp130^wt/−, Rb^N750F/wtp130^−/−, and Rb^N750F/N750Fp107^wt/− mice plus appropriate control genotypes were inspected twice a week and sacrificed when ill. Mice were examined histologically; blood was collected into heparinized tubes and diluted 2× in PBS. Red- and white blood cell counts were determined using a Coulter counter.

Cell populations from spleen, bone marrow, blood (leukocytes), and thymus were stained in antibody cocktails and analyzed by FACS as described in Supplementary Materials and Methods.

Binding of pRB to LxCxE-containing proteins involves four conserved residues in the B domain: Asn757, Tyr756, Lys713, and Tyr709 (23). Substituting asparagine at position 757 for phenylalanine disrupted LxCxE binding, whereas binding to E2F1 remained intact (24, 25). We introduced the corresponding N750F substitution in murine pRB and found this to fully abrogate interaction with the LxCxE-containing protein SV40 large T antigen (TAg; Fig. 1A). As controls, we used pRB^C699F (human pRB^C706F) and pRB^R654W (human pRB^R661W). pRB^C706F is incapable of binding E2Fs (26) and LxCxE proteins (27), impaired in transcriptional repression (28) and unable to inhibit proliferation of SAOS2 cells (26). pRB^R661W is a low-penetrance pRB mutant defective in binding E2Fs and LxCxE-containing proteins (29, 30).

To test the ability of pRB^N750F to reduce transcription, we introduced a luciferase gene preceded by six E2F binding sites in mouse Rb^−/− 3T9 cells. Luciferase activity induced by ectopic expression of E2F1 (Fig. 1B, lanes 1 and 2) was efficiently suppressed by coexpression of wild-type pRB (pRB^wt) or pRB^N750F (Fig. 1B, compare lanes 3 to 5 with lane 6 to 8), but not by pRB^C699F (Fig. 1B, lanes 9 to 11). Also, without ectopically expressed E2F1, pRB^wt and pRB^N750F repressed the luciferase reporter, whereas pRB^C699F did not (Fig. 1C). These experiments demonstrate pRB^N750F to effectively block E2F transactivation activity.

E2F4 generally acts as a transcriptional repressor (12, 13), likely by forming E2F4-pocket protein–chromatin remodeling complexes. However, E2F4 can also function as a transcriptional activator (Fig. 1D, compare lanes 1 and 2; ref. 31). pRB^N750F only partly repressed the E2F-luciferase reporter induced by E2F4 (Fig. 1D, compare lanes 2, 3, and 4). We propose that pRB^wt suppressed E2F4-mediated transcription by simultaneously inhibiting transactivation and forming E2F4–pRB–chromatin remodeling complexes and that pRB^N750F was impaired in the latter. Consistently, when pRB was fused to a GAL4-DNA binding motif, targeting pRB to a cotransfected GAL4-TK-luciferase reporter pRB^N750F was severely, but perhaps not completely, impaired in repressing transcription (Fig. 1E).

We conclude that pRB^N750F is able to bind E2F transcription factors but unable to establish complete transcriptional repression, likely because it is impaired in recruiting chromatin remodeling complexes.

To study E2F target gene expression in a physiologic setting, we generated Rb^N750F/N750F primary embryonic fibroblasts (MEFs) carrying the N750F substitution at the endogenous Rb gene (Supplementary Fig. S1A and ref. 21). As the consequences of abrogating pRB's LxCxE-binding capacity may be masked by p130 and/or p107 (8, 22) and p130 is the main pocket protein recruited to repressed promoters in G₀–G₁, we also generated Rb^N750F/N750Fp130^−/− MEFs and appropriate controls (Rb^−/−, p130^−/−, and Rb^−/−p130^−/−). Unfortunately, we have not obtained Rb^N750F/N750Fp107^−/− MEFs as in our mouse background p107 deficiency alone already affected embryonic development.

pRB-N750F protein level was similar to wild-type pRB (Supplementary Fig. S1B and S1C). Rb^N750F/N750F MEFs displayed elevated expression of the E2F target genes Cyclin E and p107 during asynchronous proliferation (Supplementary Fig. S1B, lanes 1 and 3), upon expression of RAS^V12 (Supplementary Fig. S1B, lanes 4 and 6) and after serum starvation (Supplementary Fig. S1B, lanes 7 and 9), indicative for similar derepression as in Rb^−/− cells. The effect on Cyclin A seemed less prominent. Strikingly, derepression was not enhanced by the absence of p130 (Supplementary Fig. S1C). Thus, the pRB–LxCxE interaction contributes to repression of E2F target genes in a physiologic setting.

We tested the ability of Rb^N750F/N750F and Rb^N750F/N750Fp130^−/− MEFs to arrest in response to antiproliferative signals.

Bypass of cell-cycle arrest in response to mitogen deprivation or cell–cell contact only occurred upon the combined loss of pRB and p130 (Table 1). Rb^N750F/N750Fp130^−/− MEFs arrested as efficiently as wild-type and p130^−/− MEFs (Table 1), showing that the LxCxE interaction of pRB did not contribute to cell-cycle arrest.

G₁ arrest in response to γ-irradiation relies on pRB (32). Indeed, Rb^−/− MEFs were severely impaired in arresting the cell cycle in response to γ-irradiation: at 5.5 Gy, wild-type and p130^−/− MEFs displayed a 45% to 50% reduction in S phase cells, opposed to only 25% in Rb^−/− MEFs (Fig. 2A, lanes 1–6). This effect was not aggravated upon concomitant ablation of p130 (Fig. 2A, lanes 7–8). Also the fold increase in (G₁ + G₂):S ratio of irradiated versus untreated cells revealed an impaired arrest in Rb^−/− MEFs, irrespective of p130 (Fig. 2B, lanes 1–4). Similarly irradiated Rb^N750F/N750F MEFs displayed a reduction in S phase cells of 35% (Fig. 2A, lanes 9 and 10) and the fold increase in (G₁ + G₂):S ratio was similar as in Rb^−/− MEFs (Fig. 2B, lanes 2 and 5). Again, this effect was not modulated by p130 (Fig. 2A, lanes 9–12; Fig. 2B, lanes 5 and 6).

At higher doses of irradiation, Rb^N750F/N750F MEFs displayed an intermediate phenotype: at 10 or 15 Gy, Rb^N750F/N750F MEFs arrested less efficiently than wild-type MEFs, but more efficiently than Rb^−/− MEFs (Supplementary Fig. S2A, lanes 4–6 and 7–9). Moreover, irradiation with 15 Gy induced a shift from G₁ toward G₂ arrest in both Rb^N750F/N750F MEFs and Rb^−/− MEFs (Supplementary Fig. S3).

To measure the stringency of radiation-induced cell-cycle arrest, serum-starved Rb^N750F/N750F, Rb^−/−, and Rb^wt/wt MEFs were restimulated with serum and irradiated 7 hours later. Rb^−/− MEFs entered S phase earlier than Rb^wt/wt MEFs, with Rb^N750F/N750F MEFs showing an intermediate phenotype (Supplementary Fig. S2B). Consistent with earlier findings (33), these results indicate that the pRB–LxCxE interaction contributes to DNA-damage–induced G₁ arrest.

Whereas pRB alone is the central player in the cellular response to ionizing radiation, bypass of RAS^V12-induced growth inhibition required concomitant loss of pRB and p130, pRB and p107, or all three proteins (8, 34). We therefore studied the effect of RAS^V12 expression in a p130-defective background. Similar to Rb^−/−p130^−/− MEFs, Rb^N750F/N750Fp130^−/− MEFs maintained a high proliferation rate upon RAS^V12 expression, whereas Rb^wt/wtp130^−/− MEFs decelerated proliferation (Fig. 3). This result indicates that the pRB–LxCxE interaction is critical for RAS^V12-induced inhibition of proliferation.

We previously showed that Rb^−/−p130^−/− MEFs, but not Rb^−/− or p130^−/− MEFs, could proliferate anchorage independently upon expression of RAS^V12 and TBX2 (20). We therefore studied the transforming capacity of RAS^V12 in Rb^N750F/N750Fp130^−/− MEFs rather than in Rb^N750F/N750F MEFs. Surprisingly, Rb^N750F/N750Fp130^−/− MEFs did not grow anchorage independently upon expression of RAS^V12 and TBX2 (Fig. 4). Thus, ablation of the pRB–LxCxE interaction alleviated RAS^V12-induced growth inhibition (in the absence of p130) but was not sufficient to support RAS^V12/TBX2-induced transformation.

To investigate whether the pRB–LxCxE interaction is critical for pRB's role in embryonic development, we intercrossed Rb^N750F/wt mice. Although full ablation of Rb was embryonic lethal (35), Rb^N750F/N750F animals were born according to Mendelian ratio and survived into adulthood.

In contrast, in crosses with p130^+/− animals, life-born Rb^N750F/N750Fp130^−/− animals were not obtained (Supplementary Table S1). Rb^N750F/N750Fp130^−/− embryos were detected at embryonic day 18 but died prenatally. We could not test the effect of p107 disruption on the viability of Rb^N750F/N750F mice as p107^−/− mice died in utero. However, combination of three mutant alleles, Rb^N750F/N750Fp130^+/−, Rb^N750F/wtp130^−/−, and Rb^N750F/N750Fp107^+/−, allowed normal embryonic development and survival into adulthood.

Ablation of the pRB LxCxE-interacting domain did not affect lifespan or tumorigenesis (Fig. 5A). The absence of tumor development in Rb^N750F/N750F mice sharply contrasted with the high incidence of pituitary and thyroid tumors in Rb^+/− and chimeric Rb^−/− mice (9, 10). We previously showed that the tumor spectrum induced by Rb loss was broadened by concomitant loss of p130 or p107, and included retinoblastoma, osteosarcoma, lymphosarcoma, pheochromocytoma, and coecal adenocarcinoma (8, 36). We therefore wondered whether tumor development in Rb^N750F/N750F mice was suppressed by compensatory activities of p130 or p107. Because Rb^N750F/N750F animals were not viable in the absence of p130 or p107 (Supplementary Table S1), we studied tumor incidence in mice with three mutant alleles, i.e., Rb^N750F/N750Fp130^+/−, Rb^N750F/wtp130^−/−, or Rb^N750F/N750Fp107^+/−, anticipating that similar to Rb^+/−p130^−/− or Rb^+/−p107^−/− chimeric mice, tumor development may ensue from loss of the respective wild-type alleles (10). Although hemizygosity for p107, and perhaps also for p130, reduced the survival of Rb^N750F/N750F mice (Fig. 5B and D, respectively), Rb^N750F/N750Fp130^+/− and Rb^N750F/N750Fp107^+/− animals were not predisposed to tumorigenesis (Supplementary Table S2).

Also, Rb^N750F/wtp130^−/− mice were not tumor prone, although they showed a severe reduction in life span (Fig. 5C), their median survival being reduced from >70 weeks in control littermates to 32 weeks. Rb^N750F/wtp130^−/− animals were smaller and skinnier than control littermates and had an irregular fur. Of 37 animals, 24 had to be sacrificed due to sudden illness and 8 suddenly died before 60 weeks of age. Surprisingly, microscopic examination of pituitary (n = 10), thyroid (n = 16), brain (n = 24), eyes (n = 23), digestive system (stomach, intestines, liver, pancreas; n = 22), testes (n = 7), ovary (n = 15), heart (n = 15), lungs (n = 22), adrenals (n = 16), spleen (n = 22), thymus (n = 19), and bone marrow (sternum, hind leg, head; n = 16) did not reveal abnormalities, except in the spleen (see below). Only three animals, older than 50 weeks, were diagnosed with tumor(s): one alveolar-bronchiolar adenoma, two adenomas in the pituitary gland, and one lymphoma (Supplementary Table S2). For comparison, Rb^+/− animals developed pituitary tumors between 32 and 36 weeks (10), and in Rb^+/−p130^−/− chimeric mice, 5 of 15 animals developed a tumor (8). Taken together, we conclude that Rb^N750F/wtp130^−/− animals had a reduced life span but were not predisposed to tumorigenesis.

Twenty-four percent (n = 21) of Rb^N750F/wtp130^−/− animals showed a reduced number of hematopoietic cells in the red pulp of the spleen (“reduced cellularity”), a phenomenon not seen in Rb^N750F/wtp130^+/− and Rb^wt/wtp130^−/− animals (both n = 6). Of the five cases, only two showed reduced hematopoiesis in the bone marrow (not shown). Furthermore, red and white blood cell counts were sometimes reduced in ill Rb^N750F/wtp130^−/− animals (data not shown), but this did not correlate with the reduced cellularity in the spleen. Therefore, the cause and significance of this observation remain elusive.

FACS analysis revealed normal percentages of Lin⁻Sca1⁺cKit⁺ cells, long-term hematopoietic stem cells, short-term hematopoietic stem cells, myeloid progenitors, lymphoid progenitors, erythroid progenitors, monocytes, and granulocytes in the bone marrow of Rb^N750F/wtp130^−/− mice (not shown). However, 2 of 4 Rb^N750F/wtp130^−/− animals had a decreased population of cells with intermediate expression of CD19 and B220, possibly reflecting a defective pre–B-cell population (mouse a and c in Supplementary Table S3; 0.75% and 1.7%, respectively, vs. on average 6.6% in Rb^N750F/wtp130^+/− and Rb^wt/wtp130^−/− control animals). However, mature B cells, characterized by high CD19/B220 expression, were only mildly reduced (1.7% and 2.1%, vs. on average 3.3% in control animals, Supplementary Table S3) and B- and T-cell numbers were normal in spleen, peripheral blood, and thymus (not shown), arguing against a B-cell defect in Rb^N750F/wtp130^−/− mice. In conclusion, although pocket proteins have been implicated in proliferation and mobilization of hematopoietic stem cells in the bone marrow (37, 38), both microscopic and FACS analyses argue against a major bone marrow defect in Rb^N750F/wtp130^−/− animals.

A defect that could relate to early death of Rb^N750/wtp130^−/− mice was suggested by their skinny appearance. Indeed, at the time they became ill, Rb^N750/wtp130^−/− mice showed strongly reduced or even absent subcutaneous adipose depots (Supplementary Fig. S4A) that lacked unilocular lipid droplet-containing adipocytes (Fig. 6A). Fat deposits, if present, looked like brown rather than white fat. The back skin of Rb^N750/wtp130^−/− animals was thinner and more transparent than in control mice (Supplementary Fig. S4A): in particular, the adipose tissue layer in the lower dermis was thin and in some Rb^N750/wtp130^−/− animals totally absent (Fig. 6B). Four to seven weeks old Rb^N750/wtp130^−/− animals already showed reduced thickness of the dermal adipose layer containing fewer adipocytes and smaller lipid droplets (Supplementary Fig. S4B). Anterior and posterior subcutaneous fat deposits were present, but again the size of the lipid droplets was smaller than in control littermates (Supplementary Fig. S4C). These observations indicate a shift of white to brown fat in Rb^N750/wtp130^−/− animals, which may impact on thermoregulation and viability of the animals.

Given the proposed role of the pRB–LxCxE interaction in silencing of E2F target genes (4, 14), we asked whether this interaction is essential for pRB's role in cell-cycle regulation, development, and suppression of tumorigenesis. To this aim, we abrogated the LxCxE binding site of mouse pRB by substituting asparagine at position 750 for phenylalanine. pRB^N750F retained the ability to inhibit E2F1 transcriptional activity, but was unable to bind an LxCxE-containing protein and to effectively mediate E2F4-dependent transcriptional repression. We introduced the N750F substitution at the endogenous Rb gene and studied its consequences in the absence of p130.

We found the pRB–LxCxE interaction to contribute to growth inhibition in response to γ-irradiation and expression of RAS^V12, but not after growth factor deprivation or contact inhibition.

Although serum starvation was found to induce pRB- and HDAC1-dependent deacetylation and silencing of the Cyclin E promoter (15) and a mutation in the LxCxE binding site of pRB increased Cyclin E expression during serum starvation (our experiments; ref. 14), Rb^N750F/N750Fp130^−/− MEFs arrested normally in response to serum starvation, whereas Rb^−/−p130^−/− MEFs did not. Therefore, cell-cycle arrest after serum withdrawal does not rely on pRB's LxCxE binding site but is effectuated by other pocket protein functions, possibly inhibition of E2F transactivation or another activity of pRB.

In contrast, pRB's critical role in cell-cycle arrest after γ-irradiation partially relied on its interaction with LxCxE proteins. Similar observations were made in response to UV irradiation and cisplatin exposure (33, 39). Cycle arrest upon γ-irradiation coincided with reduced CDK2 kinase activity, indicative for pRB–HDAC-mediated repression of the CDK2 regulator Cyclin E (32). However, other LxCxE proteins may be involved as well, such as the LxCxE-containing protein RF-Cp145, which enhanced survival after DNA damage in the presence of pRB (39); LxCxE members of the p200 family, which may stabilize the hypo-phosphorylated form of pRB upon DNA damage (40); the APC–Cdh1 complex that binds pRB in an LxCxE-dependent manner causing downregulation of Skp2 and stabilization of p27^KIP1 (41).

RAS^V12-induced growth inhibition also required pRB's LxCxE-interacting domain. Consistently, expression of the DNA binding domain of E2F1 (E2F-DB) could bypass RAS^V12-induced senescence (42). Because E2F-DB cannot activate transcription but only displace endogenous E2Fs from promoter regions, these results implied that relieve of E2F-mediated transcriptional repression can overcome RAS^V12-induced senescence. In addition, Narita and colleagues (16) detected recruitment of HP1γ and the presence of methylated H3K9 to promoter regions in RAS^V12-induced senescent cells, but not in confluent quiescent cells.

Surprisingly, although loss of pRB's LxCxE-binding activity in Rb^N750F/N750Fp130^−/− MEFs allowed proliferation in the presence of RAS^V12, it did not allow transformation by RAS^V12 and TBX2. Possibly, derepression of Cyclin E in Rb^N750F/N750Fp130^−/− MEFs induced CDK2 activity to a level sufficient to bypass RAS^V12-induced senescence, but insufficient to counteract the dramatic drop in kinase activities that occurs upon loss of anchorage (20). It is also possible that anchorless growth is prohibited by inhibition of E2F transactivation activity or another function of pRB that was not affected by the N750F mutation. Of note, others reported that ablation of the pRB–LxCxE interaction promoted anchorless growth upon expression of dominant-negative p53 and RAS^V12 (43) or RAS^V12 alone (44). However, in the first case, the effect was modest (2-fold) and no comparison was made with Rb^−/− cells; in the second, cells were plated at low density before removal of anchorage, which may have caused selection for (rare) cells that had escaped cell-cycle arrest. Importantly, our in vitro and in vivo results are consistent: ablation of pRB–LxCxE interactions did not promote anchorless growth in vitro or tumorigenesis in vivo (see below).

Our in vitro experiments show that ablation of the pRB–LxCxE interaction attenuated cell-cycle arrest upon γ-irradiation and expression of RAS^V12, but did not alleviate cell-cycle arrest upon serum starvation or cell–cell contact and was not sufficient to support anchorage-independent growth. It was therefore of interest to determine whether recruitment of LxCxE-containing proteins by pRB is critical for mouse development and survival and for tumor suppression, in particular in the absence of p130 or p107.

The perinatal death of Rb^N750F/N750Fp130^−/− animals as opposed Rb^N750F/N750F mice suggests that recruitment of chromatin remodeling complexes via LxCxE-mediated interactions is essential for completion of embryonic development. This is consistent with embryonic failure of p130^−/−p107^−/− and E2F4^−/−E2F5^−/− mice (45, 46).

Triple Rb^N750F/N750Fp130^+/−- and Rb^N750F/wtp130^−/−-mutant mice developed to term and survived into adulthood, although the lifespan of the latter was severely reduced. This may be related to strongly reduced dermal and subcutaneous fat deposits, with white adipose tissue having acquired a morphology resembling brown fat. Previous reports have shown the involvement of pocket proteins in the control of white versus brown fat cell differentiation, the latter predominating in the absence of Rb or p107 (47, 48). Perhaps Rb^N750F/wtp130^−/− mice suffered from excessive heat loss due to the reduced presence of dermal white adipocytes and thin skin. Heat loss may be compensated for by “browning” of residual subcutaneous white fat depots through recruitment of thermogenic adipocytes in normally white adipose depots. This may result in excessive energy consumption in adipose tissue, possibly depleting energy-rich substrates required for maintaining normal thermoregulation. However, further experiments are required to support this scenario.

Our previous studies demonstrated that Rb^+/−p107^−/− and Rb^+/−p130^−/− chimeric mice developed a wide spectrum of tumors following loss of the wild-type Rb allele (8). Surprisingly, not only Rb^N750F/N750F mice but also Rb^N750F/N750Fp130^+/−, Rb^N750F/wtp130^−/−, and Rb^N750F/N750Fp107^+/− mice were not tumor prone. Although one may argue that the wasting syndrome in Rb^N750F/wtp130^−/− mice may have precluded tumor development, this cannot explain the absence of tumor development in Rb^N750F/N750Fp130^+/− and Rb^N750F/N750Fp107^+/− mice that did not show this syndrome. Therefore, the absence of tumor susceptibility in these triple allele mutant mice implies that Rb^N750F/N750Fp130^−/− and Rb^N750F/N750Fp107^−/− cells, which are expected to frequently arise by spontaneous loss of heterozygosity, do not readily undergo oncogenic transformation. Our results therefore argue against a major role for LxCxE-containing proteins in tumor suppression.

It remains possible that loss of the pRB–LxCxE interaction stimulates tumor formation only under specific conditions; for example, mice with an ablated pRB–LxCxE interaction were prone to genotoxin-induced liver tumorigenesis, consistent with the role of pRB–LxCxE interactions in DNA damage response (33). Also, abrogation of the pRB–LxCxE interaction enhanced tumorigenesis in combination with mutant or inactive p53 (43, 49), an effect that may be related to the involvement of pRB's LxCxE-binding activity in chromosome condensation and cohesion (49, 50).

The absence of tumor development in Rb^N750F/p130- and Rb^N750F/p107-mutant mice indicates that disturbance of transcriptional regulation by loss of pRB–LxCxE interactions is insufficient to promote spontaneous tumor development. Thus, pRB's tumor suppressor activity may be effectuated through regulation of activator E2Fs rather than by active silencing of E2F target genes. However, in some tissues, activator E2F activity was dispensable for proliferation (51–53). It may therefore be possible that pRB–LxCxE and pRB-activator E2F interactions are each dispensable for tumor suppression but not both together. Alternatively, a different function of pRB is critical for tumor suppression, such as LxCxE-independent transcriptional repression or a function of pRB in S-phase during DNA replication. Irrespective of the mechanism, our results suggest that the in vitro cellular response to mitogen deprivation or cell–cell contact is a better predictor of tumorigenesis than RAS^V12-induced growth inhibition.

No potential conflicts of interest were disclosed.

Conception and design: T.L. Vormer, K. Wojciechowicz, J.B. Hansen, H. te Riele

Development of methodology: J.B. Hansen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T.L. Vormer, K. Wojciechowicz, S. de Vries, A. van der Wal, E. Delzenne-Goette, S.H. Naik, J.-H. Dannenberg, J.B. Hansen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T.L. Vormer, K. Wojciechowicz, J.-Y. Song, J.-H. Dannenberg, J.B. Hansen, H. te Riele

Writing, review, and/or revision of the manuscript: T.L. Vormer, J.-H. Dannenberg, J.B. Hansen, H. te Riele

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Dekker

Study supervision: J.B. Hansen, H. te Riele

The authors thank their colleagues Marieke Aarts, Sietske Bakker, Rob Dekker, Tanja van Harn, Camiel Wielders, and Eva Wielders for helpful discussions and comments on the article, and members of their animal and animal pathology facilities for technical assistance.

This work was supported by the Dutch Cancer Society (grant numbers NKI 2000-2232 and NKI 2002-2634) and a fellowship from the Federation of European Biochemical Societies (J.B. Hansen).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.