|Home | About | Journals | Submit | Contact Us | Français|
S.G.K. conducted all the experiments and wrote the manuscript. P.L. helped with initial concept and generated selection marker cassettes. S.K.S. conceived the idea, generated the conditional ES cell line, supervised the study and wrote the manuscript.
Individuals with mutations in breast cancer susceptibility genes, BRCA1 and BRCA2 have up to 80% risk of developing breast cancer by the age of 70. Sequencing based genetic tests are now available to identify mutation carriers in effort to reduce mortality through prevention and early diagnosis. However, lack of a suitable functional assay hinders risk assessment of more than 1900 BRCA1 and BRCA2 variants in the Breast Cancer Information Core database that do not clearly disrupt the gene product. We have established a simple, versatile and reliable assay to test for functional significance of mutations in BRCA2 using mouse embryonic stem cells (ES-cells) and bacterial artificial chromosomes (BACs) and have used it to classify 17 sequence variants. The assay is based on the ability of human BRCA2 to complement the loss of endogenous Brca2 in mouse ES-cells. This technique may also serve as a paradigm for functional analysis of mutations found in other human disease genes.
Segregation analysis in cancer-afflicted families provides the most reliable information to distinguish between deleterious and neutral alterations identified in BRCA1 or BRCA21. However, there is an enormous need to have a functional assay to classify variants for which such information is not available because most mutations are rare and familial data are often insufficient. A few functional assays have been designed that either utilize a transcriptional regulation activity or rely on functional complementation tests using BRCA1-deficient human tumor cells or BRCA2-deficient hamster cell lines (see reference 2 for review and references). Although these can distinguish between neutral and deleterious variants, such complementation assays have limited application because they rely on using cDNA expression vectors to deliver mutant proteins in cells that are inherently prone to genomic instability.
We have used mouse ES-cells and BACs to design a functional assay to classify BRCA2 variants based on the observation that Brca2 is essential for ES-cell viability3. The ability of BRCA2 variants to rescue the lethality of Brca2-deficient ES-cells is used to evaluate the functional significance of individual variants (Fig. 1a). Sequence variants that fail to rescue Brca2-deficient ES-cells are considered deleterious. The non-lethal variants are screened for a defect in the known function of BRCA2 in DNA repair as described below.
We generated a conditional allele of Brca2 by flanking the entire gene locus with two loxP sites along with two halves of the human HPRT1 mini gene in AB2.2 ES-cells that lack functional Hprt gene and are sensitive to HAT selection (HATs)4. (Supplementary Figures 1 and 2 online) and then disrupted the second copy of Brca2 by targeting a blasticidin resistance gene into exon 11 of the gene (Supplementary Figure 3 online). When Cre was transiently expressed in one of the resulting ES-cell clones (Pl2F7), we obtained only 5–10 HATr colonies (background colonies that retained a wild-type copy after plating 105 cells). Control cells carrying a conditional and a wild-type allele of Brca2 yielded 4,000–6,000 colonies (data not shown).
Next, we rescued the lethality of Brca2-deficient ES-cells by electroporating a human BRCA2-containing BAC retrofitted with a neomycin resistance gene into Pl2F7 cells. We tested three independent Neor clones showing presence of 5′ and 3′ portions of BRCA2 (Fig. 1b) and expressing different levels of human BRCA2 protein (Fig. 1c). All three clones yielded thousands of HATr colonies that lacked endogenous Brca2 (Fig. 1d,e).
To assess the ability to distinguish between a deleterious and a neutral variant using this experimental system, we generated four mutations (279delAC, 6174delT N991D and D2723H) with an indisputable functional effect5–7 in the BAC. First, we examined two known deleterious mutations: 279delAC deletion that results in a stop signal at codon 29 (stop29) and 6174delT, that truncates the protein at codon 20035 (Fig. 1f). As expected, these mutations could not rescue the ES-cell lethality supporting their deleterious nature.
To further assess the validity of our assay, we tested two missense variants, N991D, a known polymorphism6, and D2723H, a known deleterious missense mutation7. ES-cells expressing the N991D variant survived after the deletion of the endogenous Brca2 and exhibited no difference in sensitivity to various DNA damaging agents compared to control cells (Supplementary Table 1 online). We found D2723H mutant protein to be present exclusively in the cytoplasmic fraction (Fig. 2a) as reported previously7. In our assay, it failed to rescue the ES-cells after Cre expression. Our results confirm that D2723H is deleterious and N991D is a neutral BRCA2 variant demonstrating our ability to successfully distinguish between deleterious and neutral alterations.
We then selected a panel of sequence variants (see Table 1) to further validate the system and to demonstrate its applicability to test variants with different degrees of evolutionary conservation, alterations found in different functional domains, those with conflicting or unclear functional data as well as one splicing variant. In addition we have generated a hypomorphic allele and examined its phenotype in detail using various functional assays.
We first tested 7 BRCA2 variants (see Table 1), five of them (Y42C, D1420Y, V2728I, E2856A, and S2988G) with a limited evolutionary conservation and two (E2663V, G2901D) conserved across the animal kingdom and plants. All less-conserved variants sustained the ES-cell viability after the endogenous Brca2 was deleted and did not show hypersensitivity to any DNA-damaging agents. Four of these mutations, with the exception of S2988G that was found only once, have been frequently recorded in the BIC database8 and epidemiological data (Table 1) strongly suggested that these were neutral variants. Despite these data, Y42C was shown by in vitro studies to disrupt an interaction between the N-terminus of BRCA2 and the replication protein A (RPA)9, a protein essential for DNA repair, which directly contradicts the epidemiological studies10. Lack of any deleterious effect of Y42C in ES cells further validates our experimental system and reveals limitations of experiments in which an isolated portion of the protein is used to draw conclusions about causality.
The pathogenic potential of two evolutionary conserved variants, G2901D and E2663V, was unknown (see Table 1). Our assay shows that G2901D has no effect on BRCA2 function (Supplementary Table 1 online), suggesting that this mutation is neutral. The other variant, E2663V, did not rescue the cells after the Cre expression and, thus, should be classified as deleterious. Interestingly, it has recently been reported as a variant with high probability to be classified as deleterious11.
We have previously demonstrated the reliability of using human BACs in mouse ES-cells to study BRCA1 splicing mutations12. Here, we tested a BRCA2 missense variant T2722R that was recently reported to cause aberrant splicing and results in deletion of exon 18 in the transcript13. Indeed, in mutant ES-cells we found transcripts lacking exon 18 (Δ18, Fig. 3a,b). However, using RT-PCR and Western blot analysis we found that the major product of this allele was the full-length protein (Fig. 3a,c). We believe that unlike in ES-cells used in our studies, in tissue samples from individuals (who are heterozygous), such products may have escaped detection due to the presence of wild-type BRCA2. Nevertheless, the full-length T2722R BRCA2 did not rescue ES-cell viability corroborating the conclusion that the mutation is deleterious.
In an effort to demonstrate the ability to perform various functional assays on Brca2-deficient ES-cells, we generated a known deleterious mutation that results in a protein truncated at codon 3308 (Y3308X)8. We expected this mutation to be a hypomorphic allele based on the phenotype of mice that have a similar truncating mutation in Brca2 (exon 27-deletion)14,15. Evaluation of the Y3308X variant in our assay was also relevant to a similar truncation involving the next codon, E3309X, found in a patient with ovarian cancer (personal communication, Dr. William Foulkes). Based on the deleterious nature of the Y3308X variant, E3309X was expected to be deleterious. However, existence of a known polymorphism a few amino acids downstream, K3326X16, added ambiguity to the interpretation. We, therefore, tested these three truncations in our assay.
We obtained viable ES-cell clones expressing an Y3308X transgene as confirmed by RT-PCR (Fig. 1f) but observed 5–10 fold reduction in the number of HATr colonies. The mutant cells grew significantly slower in vitro as well as in vivo (Supplementary Figure 4 a,b online). These cells had a reduced plating efficiency, and were hypersensitive to crosslinking and methylating agents as well as ionizing radiation (Supplementary Table 1 online and Supplementary Figure 5 online), which is consistent with drug sensitivities known for Brca2-deficient cell lines17–20.
Because of the known role of BRCA2 in homologous recombination15, we tested the efficiency of homologous recombination in Y3308X mutant ES-cells using two gene-targeting vectors. One was designed to target a genomic locus located 150 kb upstream of Brca1 (Supplementary Figure 6 a,b online) and contained a negative selection marker (diphtheria toxin gene, DT). The second construct, which targeted the Rosa26 locus, lacked a negative selection marker (Supplementary Figure 6 c,d online). Although the control cells exhibited a targeting efficiency within the range of normal experimental variation, we did not observe a single correctly targeted clone in cells expressing Y3308X BRCA2 (Fig. 2b), suggesting a defect in the homologous recombination process. BRCA2 plays a key role in the recruitment of RAD51 to the sites of DNA repair21, therefore, we examined the effect of the Y3308X mutation on radiation-induced RAD51 foci formation. We observed on average 32 RAD51 foci per cell after irradiation (10 Gy) in control cells whereas none or occasionally only one or two RAD51 foci could be found in the mutant cells (Fig. 2 c–e). These results provide additional evidence demonstrating a defect in RAD51-mediated DNA repair in Y3308X cells.
One of the hallmarks of BRCA2-deficient cells is a marked increase in genomic instability predisposing them to tumorigenesis22. Karyotyping revealed on average 68 chromosomal aberrations such as chromatid gaps and breaks, radial structures and fragmented chromosomes per 100 mutant cells versus 10.5 aberrations per 100 control cells (Fig. 2 f–h). The extent of genomic instability in Y3308X mutant cells strongly suggests that this variant is likely to predispose a cell to tumorigenesis. Interestingly, we found elevated levels of p53 and p21 proteins in unirradiated mutant cells (Supplementary Figure 4c online). This elevated DNA damage response may explain the defects in cell proliferation and plating efficiency of the Y3308X mutant cells.
We also tested the K3326X truncation in our assay and found no effect on BRCA2 function (Supplementary Table 1 online and Supplementary Figure 5 online), which corroborates its classification as a neutral variant. In contrast, E3309X mutant ES-cells exhibited reduced viability and were hypersensitive to various DNA damaging agents (Supplementary Table 1 online and Supplementary Figure 5 online). We, therefore, concluded that, like Y3308X, E3309X was a deleterious variant.
Finally, we tested the impact of different amino acid substitutions at the same position. We examined arginine at position 3052 located at the interface between oligonucleotide/oligosaccharide-binding folds 2 and 3 (OB2 and OB3, respectively) of the human BRCA223, which is found to be changed either to glutamine (R3052Q) or to tryptophan (R3052W) in the human population (see Table 1). A recent study shows R3052Q to have relatively low odds in favor of neutrality (354:1)11. Arginine at this position makes hydrogen bonds with four neighboring amino acid residues, two from each of the OB folds, thus linking them together (Fig. 3d, left panel). Based on the computer simulation, we predicted R3052Q variant to have an activity either similar or partially reduced compared to the wild-type BRCA2 whereas the R3052W alteration would severely disrupt the protein function (Fig 3d, middle and right panels). In our assay, ES cells expressing R3052W variant did not survive (Fig. 3e, right) suggesting it to be deleterious. On the other hand, viable R3052Q-expressing ES cells were obtained and there was no difference in the number of colonies compared to control cells (Fig. 3e, left). When challenged with DNA damaging drugs, R3052Q ES-cells displayed no sensitivity to γ-irradiation and UV but a moderate sensitivity to Mitomycin C, MMS, MNNG and Cisplatin (Fig. 3f and Supplementary Table 1 online), which suggests only a partial loss of BRCA2 function. The moderate defect in BRCA2 function associated with R3052Q variant may potentially contribute to developing cancer, albeit a low risk. Future studies may provide unequivocal proof for the pathogenicity of R3052Q variant. An alternative approach to evaluate the role of such variants in tumorigenesis is to study them in mouse models. We have generated D2723H variant in mice and found it to result in embryonic lethality, which validates the deleterious nature of this variant (Kuznetsov and Sharan, unpublished data). Such deleterious mutations can be studied using conditional knockout allele of Brca2 to evaluate their role in tumor development.
We have provided here a preliminary validation of our ES-cell-based functional assay for evaluating mutations in human BRCA2 by testing 17 variants. Although, the ultimate validation of this assay in a true epidemiological sense is not possible at present, for 13 variants (four truncating variants excluding E3309X, plus six missense variants Y42C, N991D, D1420Y, E2663V, D2723H, V2728H, E2856A, R3052Q and one missense mutation, T2722R, which also affects splicing) the results obtained in our assay completely agreed with predictions based on epidemiological studies and other available functional data6,7,13,24. The functional significance of 4 other mutations (E3309X, G2901D, S2988G, and R3052W) was unknown and reported here for the first time. Future studies will show how the range of phenotypes revealed for different mutations, i.e. cell lethality, degree of sensitivity to DNA damaging agents and, possibly, specific sensitivity to only a subset of toxic compounds correlates with their penetrance and disease presentation in humans.
We have established that this assay is accurate and one of the most comprehensive among those available to date, allowing analysis of virtually any type of mutation and may also serve as a model for investigating other human disease genes that result in a phenotype detectable in ES-cells. We have streamlined the process of mutation analysis. Three to five such sequence variants can be analyzed within 2–3 months (time for each procedure is specified in Fig. 4). This makes our assay a useful, tractable and viable tool for genetic counselors. However, until the assay is fully validated, caution must be exercised when using these data to make clinical decisions.
BAC RP11-777I19, containing full-length human BRCA2 gene25 was used to generate mutations. The mini-lambda-based recombineering system was introduced into BAC containing DH10B cells as described previously26. The loxP site was deleted and loxP511 was replaced with a neomycin resistance gene under the control of Pgk promoter to enable a positive selection of transgenic ES-cells. Mutations were introduced using the oligonucleotide-based “hit and fix” BAC recombineering method as described previously27. All mutations were confirmed by sequencing.
Mouse embryonic stem cells were cultured in M15 media (Knock-out DMEM (Gibco) and supplemented with 15% fetal bovine serum (HyClone), GPS (glutamate-penicillin-streptomycin, Gibco) and 0.1 mM β-mercaptoethanol with SNL feeder cells. 25μg BACμ DNA was electroporated into 107 Pl2F7 ES-cells (for details on generation of these cells from the AB2.2 cell line see Supplementary Figures 1–3 online) in 0.9 ml PBS at 230 V, 500 μF. Transgenic colonies were selected with G418 (Gibco) and analyzed for the presence of BRCA2 gene by Southern blot. Genomic DNA was hybridized with a 581 bp fragment corresponding to intron 1 of BRCA2 as a 5 probe (nucleotides 13,869,785–13,870,365 of NT_024524.13) recognizing a 1,835 bp EcoRI fragment and a 356 bp fragment corresponding to exon 27 of BRCA2 as a 3′ probe (nucleotides 13,953,046–13,953,401 of NT_024524.13) recognizing a 4,275 bp EcoRI fragment (Fig. 1b). Double-positive clones were further tested for expression of BRCA2 by RT-PCR and Western blot analysis. BRCA2 expressing clones were electroporated with Pgk-Cre plasmid4 and 105 cells were seeded in a 100 mm dish with SNL-feeder cells. Recombinant colonies were selected in the HAT media (Gibco) for 5 days followed by 2 days in HT media (Gibco). Viable colonies were analyzed by Southern blot to confirm the loss of the conditional Brca2 allele. A 1.5 kb probe specific to Brca2 exon 11 (nucleotides 5208–6710 of NM 009765) has been used to detect a 2.2 kb EcoRV fragment corresponding to the mutant allele (ko) and a 4.8 kb fragment corresponding to the wild-type or conditional allele (cko) (Fig. 1e and Supplementary Figure 3 online). HATr colonies were also stained with methylene blue (2% methylene blue w/v in 70% ethanol for 15 minutes followed by washing with 70% ethanol) to facilitate their quantification.
For Western blot analysis proteins were extracted in RIPA buffer (50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 150 mM NaCl, 0.1% SDS, 1% Triton X-100, 0.25% sodium deoxycholate, 1 mM sodium fluoride, 1 mM orthovanadate). NE-PER buffer system (Pierce) was used to isolate nuclear and cytoplasmic protein fractions. Proteins were then separated on a precast NuPAGE 4–12% Bis-Tris gel (Invitrogen) in MOPS SDS running buffer (Invitrogen). The following antibodies have been used: BRCA2 (Ab-2, Calbiochem, 1:500), actin (Ab-5, NeoMarkers, 1:1,000), p53 (CM5p, Novocastra), p21 (F-5, Santa Cruz). Secondary antibodies were goat anti-rabbit IgG-HRP (sc-2004, Santa Cruz) at 1:5,000 dilution and goat anti-mouse IgG-HRP (sc-2005, Santa Cruz) at 1:2,000 dilution. ECL Plus Western Blotting Detection system (Amersham) was used for chemiluminescent detection.
8,000 ES-cells per well (16,000 cells per well for Y3308X and E3309X mutants to compensate for lower seeding and growth efficiency) were seeded in 96-well plates in 2–4 replicas. Drug treatment started 18 hours later (concentrations indicated in Supplementary Figure 5 online). For UV sensitivity testing 32,000 ES-cells (64,000 cells for Y3308X and E3309X) per well were seeded in triplicates in 24-well plates. Next day cells (without media) were irradiated at indicated doses in a UV-crosslinker (Stratagene) and then cultured in fresh media. For γ-irradiation plates were exposed to a 137Cs source at 146.3 rad/minute without media change. 72 hours later the relative number of living cells was measured using XTT assay28. Values were adjusted for “no cells” background.
40,000 cells per well were plated in gelatinized SonicSeal plastic slides (NUNC). 48 hours later slides were irradiated with 10 Gy. Six hours later cells were fixed with 4% paraformaldehyde for 5 minutes, washed twice with PBS and permeabilized in PBS-buffered 0.1% Triton X-100 for 10 minutes. After two additional washes with PBS, cells were blocked in a blocking solution (1% BSA, 0.05% Triton X100, 10% donkey serum in PBS). The antibody staining and imaging was performed as described previously29.
Exponentially growing ES-cells were electroporated with each of the targeting vectors described in Supplementary Figure 6 online. Clones were selected for resistance to hygromycin or puromycin. 96 colonies for each cell line were analyzed for homologous recombination by Southern blot analysis as shown in Supplementary Figure 6 online.
ES-cells were treated with colcemid (Invitrogen) for 1.5 hours to arrest at metaphase. Cells were trypsinized, washed and resuspended in hypotonic solution at 37°C (0.075M KCL) for 15 minutes and fixed in a methanol/acetic acid mixture (3:1 v/v). Air-dried preparations were stained in Giemsa solution (10% Sorensen’s buffer and 2% Giemsa, J.T. Baker). Two hundred well-spread metaphases containing at least 40 chromosomes were examined blindly from each genotype for structural aberrations.
Crystal structure 1MIU for a C-terminal portion of mouse BRCA2 was retrieved from the protein data bank at http://www.rcsb.org/ and analyzed using Sirius software available at http://sirius.sdsc.edu and DS ViewerPro from Accelrys Software Inc.
Results obtained from testing mutant ES-cells for sensitivity to genotoxins were further evaluated for statistical significance using ANOVA test. Because Y3308X and E3309X cells showed identical phenotype, they were later tested as one group (n=6) against the group of controls (two WT clones and K3326X polymorphism, n=9) using two-tailed t-test with unequal variance.
We thank Drs. J. Acharya, K. Biswas, S. Chang, I. Daar, G. Merlino, A. Nussenzweig, S. Philip and L. Tessarollo for help discussions and critical review of the manuscript. We also thank Dr. Amie Deffenbaugh (Myriad Genetics Inc.) for providing epidemiological data, B. Martin, S. Burkett, L. North and S. Stauffer for technical assistance, L. Cleveland for help with DNA sequencing, R. Frederickson and A. Kane for illustrations, Dr. D. Du for Rosa26 genomic construct, Dr. K. Biswas for Brca1 targeting construct, Dr. Stephen West (Cancer Research UK, South Mimms) for RAD51 antibody, Dr. M. Lewandoski for helpful discussions. We thank Dr. Carol Ware (University of Washington, Seattle) for providing human ES cell pellet and Dr. Jack Collins (Advanced Biomedical Computing Center, SAIC, NCI-Frederick) for help with Computer modeling. Research was sponsored by the Center for Cancer Research, National Cancer Institute, National Institutes of Health.