Both the amino- and carboxy-termini of BRCA1 are required for BRCA1 regulation of homology directed recombination
We adapted an established assay (15
) for homologous recombination in which two inactive alleles of green fluorescent protein (GFP) are integrated in a single locus in the genome of the cell. One allele contains the 18 bp recognition element for the I-SceI endonuclease. Transfection into the cells of a plasmid for expressing the I-SceI results in a double strand break in one GFP allele, and this break can be repaired by homology directed recombination (HDR) using the second inactive allele of GFP. (The plasmids encoding the recombination substrate and the I-SceI endonuclease were the gift of M. Jasin, Memorial Sloan-Kettering Cancer Center, NY, NY.) This repair results in gene conversion that creates a GFP allele that encodes an active protein, and the recombination can be detected by identifying green fluorescing cells. This plasmid and strategy () has been successfully used in a variety of experiments in which a cell line carrying a mutant version of a protein under study has had this recombination substrate inserted in the cell and the effect of the specific protein was then evaluated (15
). We instead inserted a copy of this recombination substrate into HeLa cells and carefully selected a clone, called HeLa-DR13-9, which has no background GFP fluorescence, but following transfection of the I-SceI expressing plasmid there is a high level of GFP-positive cells. In many repeated experiments, between 10% and 20% of the cells will undergo recombination following I-SceI expression. This cell line is then useful for the analysis of RNAi-depletion of proteins and testing for their effects in HDR.
The Homology-Directed Recombination Assay
HeLa-DR13-9 cells were transfected with either a control siRNA or an siRNA specific to BRCA1. Two days later these cells were re-transfected with the appropriate siRNA plus the plasmid for expression of I-SceI, and three days later the percentage of GFP-positive cells was determined by flow-cytometry. The reduction in GFP-positive cells when BRCA1 is depleted is evident from inspection of the resultant monolayers () and is quantified by flow cytometry (10,000 cells counted per sample). In the experiment for , in the absence of transfected I-SceI expression plasmid, there are no GFP-positive cells (, lane 1). In the presence of I-SceI and a control siRNA 16% of the cells were GFP-positive (lane 2). Depletion of BRCA1 or BRCA2 reduced the number of cells with recombined GFP-alleles by eight-fold or 40-fold, respectively. Similar reductions in HDR were observed with siRNAs targeting other BRCA1 and BRCA2 sequences, indicating that these results are not due to off-target effects of the siRNA (data not shown). Further, these results are consistent with published observations (9
). Importantly, when the siRNA targets the 3’-UTR of the BRCA1, expression of exogenous BRCA1 from a plasmid with a different 3’-UTR results in complete restoration of HDR activity (, lane 4). This result suggests that we have a robust assay for determining the effects of specific BRCA1 mutations in the regulation of the homologous recombination process.
Depletion of BRCA1 protein decreases HDR activity
We next assayed a series of synthetic BRCA1 deletion mutants for function in the HDR assay. In , the four deletion mutants are diagrammed. Using the same time course as in , the endogenous BRCA1 protein was depleted by 3’-UTR-targeting siRNA, and the test BRCA1 protein was expressed from a co-transfected plasmid. Multiple repeat experiments were done, and in each experiment the maximal recombination varies from 10% to 20%, depending on transfection efficiency. Within an experiment the ratio of GFP-positive cells in the BRCA1 depleted sample relative to the control RNAi was consistently 8- to 10-fold reduced. We thus normalized data in each experiment, allowing us to average the results from multiple experiments. Replacing the endogenous BRCA1 with ΔN-BRCA1, which has deleted residues 1–302, results in a 4- to 5-fold reduction in homologous recombination. Expression of the ΔM1-BRCA1, which has deleted residues 305–770, reduced recombination by about 40%. Similarly, expression of the ΔM2-BRCA1, which has deleted residues 775–1292, resulted in a decrease in recombination by about 60%. Expression of the ΔC-BRCA1, which has deleted residues 1527–1863, resulted in a 10-fold loss of recombination. Clearly, the amino-terminus and carboxy-terminus each had a significant role in controlling the recombination reaction. Expression of the two internal domains had intermediate effects. In each case, the concentrations of BRCA1 protein expressed from the transfected plasmid were at or higher levels than the endogenous BRCA1 protein (), indicating that the reduction in HDR activity was not due to failure to express the test protein. Since the expression level of the ΔN, ΔM1, and ΔM2 mutants were significantly higher than the endogenous BRCA1 protein, it is possible that the magnitudes of the deficiencies in HDR activity with these BRCA1 deletion mutants were underestimated. We decided to focus on point mutations in the amino terminus.
The BRCA1 N- and C-termini are important in regulation of HDR activity
Identification of BRCA1 point mutations that are critical for homologous recombination activity
We decided to test point mutations in the amino-terminal 71 amino acid residues of BRCA1 that were derived from individuals with a family history of breast cancer and identified from the BIC database. The reasons for using mutations identified in breast or ovarian cancer clinics were two-fold: first, this is the most direct way to identify candidate amino acid residues for critical function without performing saturating mutagenesis and testing a large number of variants; and second, the results of our functional tests might be useful in counseling such individuals who carry these mutations. Sixteen different variants were produced by site-directed mutagenesis of the BRCA1 cDNA-expressing plasmid. One of these variants, C27A, is a synthetic mutation and is not derived from a variant obtained from an individual with a family history with breast cancer. Rather, the C27A variant completes the set of eight different Zn-coordinating residues of BRCA1 in this analysis. The results of multiple experiments are shown in . Consistent with our previous experiments, depletion of BRCA1 by transfection of siRNA targeting the 3’-UTR and co-transfection of the empty plasmid vector, resulted in a 10-fold reduction in GFP-positive cells relative to the control siRNA. Transfection of this BRCA1-specific siRNA along with the wild-type BRCA1 expression plasmid fully restored homology-directed recombination (, lane 3). Strikingly, transfection of each mutant either fully restored recombination to 100% or was fully negative, at the same level of homologous recombination as the vector-transfected control. We note that the deletion mutants used in produced recombination levels that had intermediate results: the ΔM1 and ΔM2 deletions caused partial decrements in homologous recombination. Even deleting the amino terminus, containing all of the residues being tested did not have as severe an effect on the HDR assay as did the point mutants. The expression levels of the point mutants of BRCA1 all were similar to the endogenous protein level (, compare lane 3 to all other lanes) whereas the ΔN deletion, which encompassed all of these point mutants, was significantly overexpressed relative to the endogenous protein and perhaps causing a partial masking of the HDR defect. Alternatively, it is a formal possibility that the point mutations in the BRCA1 protein have both a loss of function phenotype and a dominant negative phenotype.
Identification of BRCA1 missense mutations that affect homologous recombination
Eight of the residues that were tested coordinate zinc ions in the RING domain: C24R, C27A, C39Y, H41R, C44F, C47G, C61G, and C64G. It is anticipated that substitution of any of these amino acid residues would have major structural consequences to the protein. Consistent with that concept, replacement of the endogenous BRCA1 with any of these BRCA1 molecules with mutated zinc-coordinating residues was non-functional in HDR (, lanes 6, 7, 10, 11, 13, 14, 16, 17).
Eight other substitution mutants were tested for function in the homologous recombination pathway. The M18T and T37R variants did not complement the HDR assay. Variants that could complement the HDR activity were I21V, I31M, I42V, L52, D67Y, and R71G.
We tested whether the mutant BRCA1 proteins used in this study could bind to BARD1 (). We transfected the HA-epitope tagged BRCA1 into HEK293T cells and purified proteins in complex with the variant BRCA1 protein by immunoaffinity purification (IP) using antibody recognizing the HA-tag. Purified proteins were analyzed by immunoblots specific for the HA-epitope to evaluate the effectiveness of expression and purification of the mutant BRCA1 protein and specific for the endogenous BARD1 protein. We found that BRCA1 variant proteins I21V, I31M, I42V, L52F, D67Y, and R71G effectively purified cellular BARD1 protein. These are the same proteins that functioned in homologous recombination. Of note, the BRCA1 variant proteins M18T and H41R had detectable low level of BARD1 binding, but these BRCA1 variants did not complement the HDR assay.
Effect of BRCA1 missense variants on association with BARD1
Correlation of biological function with available clinical data for each BRCA1 missense mutant
We have summarized the available information for each mutant in , including results of HDR function and BARD1 binding (this study), E3 ubiquitin ligase activity and resistance to ionizing radiation (6
), and available clinical information. For many of these variants family history was unavailable, and this limited the analysis we could do. Examples of available family data are supplied in Supplemental Table 1
. We applied an algorithm called VUS Predict (19
) to the 16 variants. VUS Predict calculates the odds of a variant being detrimental based on a variety of characteristics of tumors including estrogen receptor, progesterone receptor, and Her2 status, tumor grade, histopathology, age of onset, and the position of the amino acid residue in a functional domain or evolutionary conserved sequence. If there were no clinical data, then the VUS Predict output was based on the single criterion of evolutionary conservation (asterisked values in ).
Summary of biological functions of BRCA1 proteins in the context of available clinical information for each variant.
Of the 16 BRCA1 variants in this study, four of the eight Zn-coordinating residues have been classified as deleterious when mutant (). One other variant, R71G, has been classified as deleterious (2
), and one, D67Y, had been classified as neutral (4
). One of the substitution mutants, M18T, has a published odds ratio of being deleterious of 31:1 (4
). Using VUS Predict, the M18T mutant has 170.8:1 odds of being deleterious. While these analyses clearly indicate the M18T variant has a trend as a deleterious allele, the magnitude of the ratio did not exceed the threshold of 1000:1 for making a clinical classification. We find that this mutation of BRCA1 causes a complete loss of HDR activity. Though this substitution, when expressed in a truncated BRCA1 peptide in the absence of the BARD1 did have ubiquitin ligase enzymatic activity, it was defective in protection of a cell against ionizing radiation (6
). In another study using the BRCA1/BARD1 heterodimer, the M18T variant was inactive as a ubiquitin ligase (21
). The M18T substitution could affect the BRCA1-BARD1 heterodimerization interface (22
), resulting in reduced BARD1 association with the BRCA1 variant (). Our results in the HDR assay are consistent with this substitution being deleterious.
Genetic analysis of one of these mutants, D67Y, has suggested that it is neutral with regard to cancer predisposition (4
), and consistent with that notion it has the same phenotype as does wild-type BRCA1 in the HDR assay. Though the R71G substitution is considered cancer-promoting, the mutation is thought to affect the splicing of the mRNA (20
), such a feature would be missed when expressing a cDNA as in this experiment. Mutant BRCA1(R71G) protein functioned in the HDR at similar levels as did wild-type BRCA1.
Among the remaining five variants of unknown clinical consequence, the T37R variant was defective in HDR (, lane 9). In new research published while this study was being written has indicated that the similar T37K variant is likely to be deleterious based on clinical and evolutionary analysis (3
). The T37R substitution was also found to defective in providing ionizing radiation resistance (6
), consistent with our HDR results. Taken together, the threonine-37 residue is likely critical to the homologous recombination process, and consequently to radio-resistance. The threonine-37 side chain is solvent exposed, but in a small cavity near the BRCA1-BARD1 heterodimerization interface (22
). Perhaps the bulky lysine or arginine substitutions result in disruption of BRCA1-BARD1 binding, or alternatively another protein important to HDR binds to this pocket.