Expression of human BRCA1 variants in mouse ES cells allows functional analysis of BRCA1 mutations (original) (raw)
Generating an ES cell–based functional assay for BRCA1. BRCA1 is essential for the viability of mouse ES cells (21, 22). Based on this observation, we used mouse ES cells to generate an assay to study the functional significance of human BRCA1 variants (Figure 1A). We hypothesized that the neutral variants will rescue the lethality of _Brca1_-null (designated as Brca1KO/KO) ES cells and also complement known functions of BRCA1. Conversely, deleterious/hypomorphic variants may either fail to rescue the ES cell lethality or may not be fully functional when tested for known functions of BRCA1 in DNA repair, cell cycle regulation, transcriptional regulation, etc.
Outline of ES cell–based assay. (A) Schematic representation of BRCA1 ES cell–based functional assay. Human BRCA1 BAC DNA with any mutation can be introduced into mouse PL2F8 ES cells containing a conditional allele of Brca1. After Cre-mediated recombination of the conditional allele, a functional copy of the human HPRT1 gene is generated and renders the cells resistant to HAT selection. Depending upon the impact of the mutation on the BRCA1 function, the cells will be viable or not after HAT selection. (B) ES cells lacking functional BRCA1 are not viable. Cre-mediated recombination of the conditional allele of Brca1 in PL2F8 ES cells results in few viable HAT-resistant colonies. Genotyping of these clones shows that they retain a conditional allele of Brca1 (lower panel, upper band).
To establish the functional assay, we engineered mouse ES cells by the gene targeting method. First, we modified one allele of Brca1 as a conditional allele (designated as Brca1CO) by flanking the locus with 2 loxP sites (Supplemental Figures 1 and 2; supplemental material available online with this article; doi:10.1172/JCI39836DS1). We inserted two halves of the human HPRT1 minigene (5′HPRT1 and 3′HPRT1) along with the 2 loxP sites to allow selection of clones that undergo Cre-mediated recombination of the loxP sites in hypoxanthine aminopterin thymidine–containing (HAT-containing) medium (Figure 1A and Supplemental Figure 1). To allow HAT selection, we used Hprt_-deficient AB2.2 ES cells. We also targeted a puromycin resistance gene (Puro) next to the 3′_HPRT1 cassette as an additional selectable marker (Supplemental Figure 1). Next, we disrupted the second allele of Brca1 by deleting exons 3–7 and replacing the region with a blasticidin resistance gene (Blasti) (Supplemental Figure 3).
The engineered ES cell clone used for the functional assay is referred to as PL2F8. Loss of the conditional allele is expected to yield no viable HAT-resistent colonies due to loss of BRCA1. To test this, we transiently expressed Cre recombinase in PL2F8 cells and selected the recombinant clones in the presence of HAT. We obtained 400–600 HAT-resistent colonies per 105_Cre_-electroporated ES cells (Figure 1B). Genotyping of 200 clones by Southern blot analysis revealed all HAT-resistent clones to be Brca1 heterozygous (represented in Figure 1B; designated as Brca1CO/KO). We concluded that none of the HAT-resistent clones were _Brca1_-null (Brca1KO/KO) and all the viable HAT-resistent clones retained a copy of the conditional allele and were Brca1KO/+. The precise mechanism by which the HAT-resistent Brca1CO/KO clones are generated remains unknown. However, we hypothesize that some ES cells may carry trisomy of chromosome 11, giving rise to 2 copies of conditional allele, only one of which recombines to confer HAT resistance. This idea is supported by the karyotypic analysis of HAT-resistent clones, which showed the presence of 3 copies of chromosome 11 (Supplemental Figure 4). In addition, chromosomal anomalies are normally present in some ES cells, and trisomy of chromosome 11 has been reported in mouse ES cells (23).
If, indeed, Brca1KO/KO ES cells fail to survive due to the loss of Brca1, can the lethality of these cells be rescued by the presence of a BRCA1 transgene? To address this, we introduced a BAC clone (clone RPCI11-812-5O) containing a wild-type copy of human BRCA1 into PL2F8 cells. Expression of the BRCA1 transgene was confirmed by Western blot analysis (Figure 2A). This BAC was previously used to rescue the lethality of _Brca1-_deficient mice (24). When we expressed Cre in PL2F8 ES cells expressing the human BRCA1 transgene, we obtained about 1,000 HAT-resistent colonies (Figure 2B), and 44 of 96 (45.8%) were confirmed to be Brca1KO/KO (Figure 2C, WT). The remaining clones were Brca1CO/KO. These results demonstrated the ability of the human BRCA1 to rescue the lethality of Brca1KO/KO ES cells.
Validation of ES cell–based assay by using WT BRCA1 and variants of known clinical implication. (A) Expression analysis of WT BRCA1, M1652I, 2 founder mutations (185delAG and 5382insC), and C61G variants using human BRCA1 BAC in mouse ES cell (2 clones for each). Protein expression was confirmed by immunoprecipitation using a human BRCA1-specific antibody (E1) followed by Western blot analysis using a commercial BRCA1-specific antibody (Ab-1), which gives 2 bands of about 250 kDa (48). Expression of 185delAG was detected by RT-PCR (lower panel). –RT, no RT, negative control. (B) Methylene blue staining of HAT-resistant ES cell colonies with no BAC (PL2F8), WT M1652I, 185delAG, 5382insC, and C61G BRCA1 BAC transgene. (C) Genotyping of HAT-resistant colonies by Southern blot analysis. The upper band representing the conditional allele was deleted in 45.8% of the cells with WT or 41.4% with M1652I BRCA1. In contrast, none of the colonies from PL2F8 cells or those expressing 185delAG, 5382insC, and C61G BRCA1 variants showed loss of the conditional allele. (D) Whole mount of a Brca1KO/KOBACTgC61G embryo at E7.5 showing retarded development, similar to the Brca1KO/KO embryo. Brca1KO/+BACTgC61G control embryo shows normal embryonic development at E7.5. (E) Whole mount of a Brca1KO/KOBACTgC61G embryo at E8.5 shows severe developmental defects compared with a Brca1KO/+BACTgC61G embryo (left).
Analysis of BRCA1 variants with known clinical impact. Based on the ability of wild-type human BRCA1 to rescue the lethality of Brca1KO/KO ES cells, we proceeded to analyze various BRCA1 variants in PL2F8 ES cells (summarized in Table 1). First, we examined 3 variants known to be deleterious (185delAG, 5382insC, C61G) and a neutral variant (M1652I) to test the ability of the ES cell–based system to classify these variants. 185delAG and 5382insC are the 2 well-known founder mutations in BRCA1 that are associated with increased risk of cancer development (25, 26). 185delAG results in a frameshift at codon 23, causing a premature translation stop after codon 38. Similarly 5382insC is known to cause protein truncation at codon 1,829. The C61G amino acid change involves one of the conserved cysteine residues of the RING domain, and the presence of a glycine residue is predicted to disrupt the integrity of this domain (8). In contrast, the M1652I amino acid change is reported to be a neutral variant based on the observation that it is found at a frequency of 4.05% in the population of cancer-free controls of European ancestry (9). We verified the expression of each variant by Western blot analysis with human-specific BRCA1 antibody, with the exception of 185delAG, which is predicted to result in a truncated protein (Figure 2A). Expression of the 185delAG transcript was confirmed by RT-PCR (Figure 2A, lower panel).
Summary of Brca1 mutants analyzed in this study
After Cre expression, 185delAG, 5382insC, and C61G resulted in a background level of HAT-resistent colonies (Figure 2B, compare with PL2F8), whereas M1652I yielded colonies similar to those expressing wild-type BRCA1 (Figure 2B). Genotyping of these clones confirmed that 185delAG, 5382insC, and C61G are unable to rescue the lethality of Brca1KO/KO cells and are therefore deleterious (Figure 2C). In contrast, approximately 41% of the HAT-resistent colonies obtained by introducing the M1652I variant were Brca1KO/KO (Figure 2C). Furthermore, when we tested the sensitivity of Brca1KO/KO ES cells rescued by M1652I to various DNA-damaging agents, we observed no differences compared with the control cells (Supplemental Table 1, bottom line), confirming that it is a neutral variant. To further validate our findings and demonstrate the in vivo relevance of our ES cell–based approach to studying BRCA1, we examined the effect of the M1652I and C61G variants in mice. We used the BACs expressing the 2 variants to generate transgenic mice. The expression of mutant BRCA1 in mouse was confirmed by RT-PCR and Western blot analysis (Supplemental Figure 5 and data not shown). The transgenic mice were crossed to Brca1 heterozygous mice. The resulting Brca1KO/+BACTgM1652I and Brca1KO/+BACTgC61G mice were next crossed to Brca1KO/+ mice to test the ability of the two variants to rescue the lethality of Brca1KO/KO, as shown previously for wild-type BRCA1 (24). While we obtained viable Brca1KO/KOBACTgM1652I mice, we failed to obtain Brca1KO/KOBACTgC61G mice. Brca1KO/KO mice rescued by the M1652I variants are phenotypically indistinguishable from control littermates expressing a wild-type copy of Brca1 (data not shown), suggesting that this BRCA1 variant is functionally similar to the wild-type allele. In contrast, the C61G variant failed to rescue the lethality of Brca1KO/KO mice (Table 2). Next, we examined the phenotype of mutant embryos to determine whether the C61G variant is partially functional in mice. At E7.5 and E8.5, the Brca1KO/KOBACTgC61G embryos were phenotypically similar to the Brca1KO/KO embryos (Figure 2, D and E), with a marked defect in embryogenesis compared with the control embryos. The mutant embryos were much smaller in size than control embryos and exhibited no signs of gastrulation or primitive streak formation. In contrast, there was no difference between Brca1KO/KOBACTgM1652I and Brca1+/KOBACTgM1652I embryos at E7.5 (data not shown), providing additional evidence that this variant is neutral. These results suggested that the ES cell–based system reflects the functional impact of BRCA1 variants and is able to recapitulate the in vivo effects of the mutation.
Summary of the number of offspring from the intercross of Brca1 heterozygous mice with either C61G or A1708E mutant BAC transgene (Tg)
Functional analysis of BRCT domain variants. We next generated 3 variants that mapped to the BRCT domain of BRCA1 and are either predicted to be deleterious or are of unknown clinical significance based on BIC database, structural analysis, transcriptional activity assay, and in silico prediction (11, 27–30). The first variant is A1708E, which maps to the first BRCT domain and has been shown to be critical for transcriptional activity. It is considered to be a deleterious, cancer-causing variant. Another variant, V1804D, present in the second BRCT domain, is predicted to be a neutral variant based on the yeast small colony assay and transcriptional activity assay (27, 29). However, the transcriptional activity was estimated to about 50% of the wild-type BRCA1, and structural analysis of the BRCT domain predicted this variant to be deleterious (31). Thus, the predicted role of this variant is controversial. The third variant (R1737X) we tested results in a truncated protein that retains the first BRCT but lacks the second BRCT domain. A mouse model (Brca11700T) expressing such a truncated protein has been generated and predicted to retain some BRCA1 functions based on the observation that the mutant embryos progress a little further in development than the _Brca1_-null embryos (32).
To determine whether these BRCA1 variants were defective in their transcriptional activation function, we examined their ability to bind to the promoters regulated by BRCA1. Failure to bind to the promoter will abrogate the transcriptional activation of its target genes, such as Gadd45a and p21 (33, 34). ChIP results showed that V1804D, A1708E, and R1737X BRCA1 exhibited marked reduction in their binding to the Gadd45a and p21 promoters compared with the wild-type BRCA1, suggesting that the access to these promoters is blocked or reduced (Figure 3A). Therefore, these variants are likely to be defective in their transcriptional activation function. The precise mechanism of the binding defect is unclear. One possibility is that the interaction between these variants and RNA polII transcriptional machinery is disrupted, as previously shown for M1775E BRCA1 (35).
Functional evaluation of BRCT domain variants of BRCA1. (A) A1708E, R1737X, and V1804D BRCA1 variants were analyzed by ChIP to test their binding to Gadd45a and p21 promoters. Wild-type human BRCA1 was used as positive control. Binding of p53 to the p21 promoter was used as an internal control. Quantitation of ChIP-PCR products is shown in the graph. (B) Methylene blue staining of HAT-resistant ES cell colonies with no BAC (PL2F8), WT BRCA1, and A1708E, R1737X, and V1804D variants of BRCA1. (C) Southern blot analysis of HAT-resistant ES cell clones reveals that none of the clones expressing deleterious BRCA1 variants (A1708E, R1737X, V1804D) lost the conditional allele (upper band). Half of clones rescued with WT BRCA1 retained only the mutant band (lower band). (D) Whole mount of Brca1KO/KOBACTgA1708E embryo at E7.5 shows retarded development, similar to the phenotype of Brca1KO/KO embryos at this stage. Brca1+/+ control littermates with or without the A1708E transgene show normal development at E7.5.
We next tested their ability to rescue the lethality of Brca1KO/KO ES cells. In all cases, we obtained only background level of colonies, and none were Brca1KO/KO (Figure 3, B and C), suggesting that these variants are all deleterious. These results also demonstrate the essential role of transcriptional control and, possibly, phosphopeptide binding activity of BRCA1. The lethal effect of V1804D mutation in the ES cell–based assay suggests that this variant is deleterious, probably due to the structural disruption of BRCT domain, as previously predicted (29).
To test the in vivo relevance of the results obtained in ES cells, we generated BAC transgenic mice expressing the A1708E variant. As shown in Table 2, the BAC transgenic mice failed to survive on a _Brca1_-null background. Furthermore, the Brca1KO/KOBACTgA1708E embryos exhibited a phenotype similar to that of the Brca1KO/KO embryos at E7.5 (Figure 3D). Thus, the results obtained from ES cell are physiologically relevant and were corroborated by our studies in mice.
Analysis of BRCA1 phosphorylation site variants reveals a role for S1497. To further demonstrate the usefulness of the ES cell-based approach to study BRCA1 function, we examined the physiological significance of various phosphorylation sites identified in BRCA1 by in vitro studies (16–20). We analyzed BRCA1 variants with the changes in serine residues (S308A, S988A, S1497A, S1423/S1524A double mutant, and S1423/S1387/S1524 triple mutant) that are phosphorylated by several kinases including Aurora, ATM, ATR, Chk2, and Cdk2. With the exception of the S1423/S1387/S1524 triple mutant (Supplemental Figure 6A), all other phosphorylation site mutants could rescue the lethality of Brca1KO/KO ES cells. The rescue rate ranged from 10.5% to 50.0% (Supplemental Figure 6B). Although the mRNA of the triple mutant BRCA1 was expressed at levels similar to other variants (data not shown), the protein was barely detectable (Supplemental Figure 6C). We reasoned that the mutant protein might be unstable because of the change of 3 serine residues, explaining its inability to rescue the ES cell lethality. We next performed various assays to determine whether the phosphorylation mutants had any effect on the known functions of BRCA1.
To examine the effect of phosphorylation site mutants on cell cycle regulation, we analyzed the viable Brca1KO/KO ES cells expressing S308A, S988A, S1497A, and S1423/S1524A BRCA1 variants by checking their cell cycle profiles with BrdU/PI double staining. We did not observe any significant differences in phosphorylation site mutants, compared with the Brca1KO/KO ES cells rescued by the wild-type BRCA1 (Supplemental Table 2). These rescued clones also did not reveal any significant change in sensitivity to DNA-damaging agents (Supplemental Table 1). However, when the ES cells rescued by S1497A and S1423/S1524A BRCA1 were γ-irradiated, they exhibited significant hypersensitivity (Figure 4A). The hypersensitivity of S1423/1524 BRCA1 was consistent with previous results indicating that S1423 and S1524 are phosphorylated by the DNA damage–dependent kinases ATM and ATR (16, 18). In contrast, the role of S1497 phosphorylation in DNA repair was previously unknown. The serine residue at position 1,497 is known to be phosphorylated by the cell cycle–dependent Cdk2 kinase (17); therefore, it was predicted to be involved in cell cycle regulation. Surprisingly, this variant had no significant effect on cell cycle progression in the absence of S1497 phosphorylation (Figure 4B and Supplemental Table 2); instead, it resulted in hypersensitivity to irradiation. This hypersensitivity was further confirmed by the clonogenic survival assay (Figure 4, C and D), as well as by the presence of more TUNEL-positive cells after γ-irradiation compared with control cells expressing wild-type BRCA1 (Figure 4E). Interestingly, Cdk2-null MEFs were recently reported to be sensitive to γ-irradiation (36). Although the mechanism remains unknown, a defect in S1497 phosphorylation may account for this phenotype.
Role of S1497 residue of BRCA1 in irradiation-induced DNA damage repair. (A) Relative survival after γ-irradiation (IR) of 2 independent clones expressing S1497A compared with ES cells expressing no transgene (PL2F8), WT BRCA1, and S1423/1524A double mutant. *P < 0.006 for WT and S1497A-1. Error bars represent ± SEM. *P < 0.006 (B) Western blot analysis of 3 clones expressing S1497A and 1 with WT BRCA1 using phospho-S1497 antibody (upper panel). The lower panel shows expression of BRCA1. Rabbit IgG was used as negative control (NC); HeLa cells were used as positive control (PC). (C) Clonogenic survival assay to examine hypersensitivity of S1497A ES cells in response to 0.2 and 0.5 Gy of IR. (D) Quantitation of the number of colonies in C. (E) Quantitation of TUNEL-positive ES cells expressing S1497A BRCA1 after IR (2 Gy). ES cells expressing WT BRCA1 were used as a control. Con, control, before IR. Error bars represent ± SEM. (F) Phosphorylation of the S1497 residue affects the hyperphosphorylation of BRCA1 after IR. Wild-type or S1497A cells were left untreated or pretreated with an inhibitor of Cdk2 kinases. After IR, BRCA1 phosphorylation was analyzed by IP–Western blotting with a BRCA1 antibody and a phospho-S1497–specific antibody. Densitometric quantitation of the S1497 residue is indicated below the left panel. HEK293 cells were used as a control to show the difference in mobility of phosphorylated and unphosphorylated BRCA1. (G) Quantitation of chromosomal abnormalities in response to IR in ES cells (n = 100) expressing WT or S1497A BRCA1. (H) Representative picture of metaphase spreads of WT and S1497A cells after IR. Inset shows magnified view of 3 chromosomal aberrations. The scale bar on the left applies to the insets; the scale bar on the right applies to the larger images.
To determine the role of S1497 in radiation-induced DNA repair, we examined the phosphorylation of the 1,497 serine residue in response of γ-irradiation using a phospho-specific antibody. As shown in Figure 4F, after irradiation, we observed a 9.1-fold increase in phosphorylation of wild-type BRCA1, with a specific 3.5-fold increase in S1497 phosphorylation. However, the S1497A variant displayed virtually no (1.3-fold) hyperphosphorylation after irradiation, even though the amount of BRCA1 was increased (Figure 4F). Therefore, the hypersensitivity of S1497A mutant ES cells upon IR treatment is due not to the protein stability or other defects in BRCA1 function, but rather to the abrogated phosphorylation of BRCA1, which is required for the proper recruitment of the protein to the site of damaged DNA. More interestingly, the hyperphosphorylation of BRCA1 was inhibited by pretreatment with 2(_bis_-(hydroxyethyl)amino)-6-(4-methoxybenzylamino)-9-isopropyl-purine, a Cdk2 inhibitor, suggesting a role of Cdk2 in radiation-induced phosphorylation of BRCA1 (Figure 4F).
To determine the functional consequence of impaired phosphorylation of S1497 on overall genomic integrity in response to DNA damage, we examined the karyotype of ES cells expressing wild-type and S1497A BRCA1 in response to 1 Gy of γ-irradiation. We examined 100 nuclei of each genotype for the presence of various chromosomal abnormalities and found a marked increase in the number of gaps, radials, and breaks in the S1497A ES cells (Figure 4, G and H). In untreated ES cells of both genotypes, 0%–2% of cells showed these abnormalities. However, after irradiation, while only 5.5% of ES cells expressing wild-type BRCA1 showed gaps, radials, and breaks, we found that 15.5% of S1497A ES cells exhibited such abnormalities. These results clearly demonstrate that the S1497 residue plays an important role in maintaining the genomic stability in response to IR-induced DNA damage. We next tested whether the S1497A variant had any effect on homologous recombination, a process known to involve BRCA1. We measured the gene targeting efficiency at the Rosa26 locus in ES cells as described previously (15). We observed no difference in the homologous recombination activity of S1497A variant compared with wild-type BRCA1 (Supplemental Table 3). These results suggest that although S1497A mutant ES cells exhibited an increase in genomic instability in response to IR, the homologous recombination function of the ES cells was not compromised.
The role of S308 residue in differentiation of ES cells. The S308 residue is known to be phosphorylated by Aurora-A kinase, and its phosphorylation is suggested to play a role in G2/M phase progression and centrosome microtubule nucleation (20, 37). Although the rescue of Brca1KO/KO ES cells by S308A BRCA1 was reduced (10% compared with 40%–50% by M1652I or wild-type; Figure 2C and Supplemental Figure 6B), cell cycle analysis revealed no significant defects (Supplemental Table 2). Surprisingly, when we cultured the S308A mutant ES cells without feeder cells, we observed a marked reduction in cell growth by day 4 (Figure 5A). We reasoned that this reduction in proliferation rate might be associated with cell differentiation because we observed clear changes of the S308A mutant ES cell morphology (Supplemental Figure 7) in the absence of feeder. Based on these results, we further analyzed the differentiation potential of S308A mutant cells by generating embryoid bodies. As shown in Figure 5B, the number of cells in embryoid bodies gradually increased and then became constant when control ES cells were used. In contrast, S308A ES cells exhibited a significant reduction in cell number in the embryoid bodies after day 7. The decrease in cell number was due to a relative increase in apoptotic cells, as marked by an increase in TUNEL-positive cells in the center of the embryoid body (Figure 5C). In addition, the embryoid bodies generated from control and S308A ES cells displayed a clear difference in their morphology based on histological analysis (Figure 5D). Wild-type embryoid bodies at day 7 showed the presence of well-defined layers of primitive ectoderm and mesoderm, whereas the embryoid bodies derived from S308A ES cells exhibited irregular structures, with many apoptotic cells in the center. From these observations we concluded that the S308A variant results in a differentiation defect of the ES cells.
Role of S308 residue in ES cell differentiation. (A) Growth analysis of S308A ES cells compared with WT BRCA1 cells cultured without feeders. Error bars represent ± SEM. (B) Growth of embryoid bodies generated from ES cells expressing WT human BRCA1 and S308A, measured by trypsinizing the embryoid bodies and counting the total number of cells at each indicated time points. (C) TUNEL staining of embryoid bodies generated from S308A and WT ES cells after 7 days in culture. (D) H&E staining of embryoid bodies generated from S308A and PL2F8 ES cells after 14 days in culture. (E) Growth of teratomas generated from PL2F8 (n = 4) and S308A (n = 5) ES cells. A single ES cell clone of PL2F8 and S308A was analyzed for each genotype. (F) Histological analysis of teratomas generated from WT and S308A ES cells. The WT teratomas show numerous neural rosettes (arrows) that were greatly reduced in S308A teratomas.
To test the differentiation potential of S308A mutant ES cells in vivo, we injected one S308A ES cell clone subcutaneously into 5 nude mice and monitored the formation of teratomas that represent a differentiated mass of cells. While all the teratomas from control cells (PL2F8, n = 4) grew rapidly within a few weeks, only 2 teratomas from S308A cells (n = 5) grew at a rate comparable to that of the control teratomas. Interestingly, the other 3 teratomas from S308A cells grew extremely slowly, suggesting a proliferation or differentiation defect (Figure 5E). Furthermore, histological analysis of the two S308A teratomas that grew rapidly revealed the presence of more differentiated neural cells, compared with the immature, fast-growing cells present as a neural rosette structure in the control teratomas (Figure 5F). These results suggested a role for the S308 residue in stem cell differentiation, in addition to its known role in the G2/M phase transition by Aurora-A kinase–mediated phosphorylation.
Analysis of the RXRXH mutant revealed a marginal effect on cell cycle progression. The 358LXCXE motif was initially predicted to be one of the RB-binding domains in BRCA1 (38). Subsequently, it was reported that the disruption of the 358LXCXE motif by changing it to 358RXRXH resulted in functional alteration of BRCA1, including misregulation of RB family proteins (39). Based on this finding, we tested the effect of 358RXRXH BRCA1 in our functional assay. The Brca1KO/KO ES cells expressing this variant were viable (Supplemental Figure 6). To find out the functional defect related to this mutation, we first examined the cell cycle profiles of these cells. As shown in Supplemental Figure 8, we found an elevated percentage of S phase cells and a correspondingly decreased percentage of G2/M phase cells in 2 mutant ES clones. The differences are not dramatic but are statistically significant and reproducible. It is notable that the changes in cell cycle profile observed in the mutant ES cells are similar to the changes in the prostate cancer cell line DU-145 (39), suggesting that the mouse ES cells can phenocopy the defects observed in human cancer cell lines. We also observed increased RB level and slower growth rate of the mutant ES cells, but the differences were marginal or not statistically significant (data not shown).