Mutations in breast cancer susceptibility gene BRCA1 (breast cancer early-onset 1) are associated with increased risk of developing breast and ovarian cancers. BRCA1 is a large protein of 1863 residues with two small structured domains at its termini: a RING domain at the N-terminus and a BRCT (BRCA1 C-terminus domain) repeat domain at the C-terminus. Previously, we quantified the effects of missense mutations on the thermodynamic stability of the BRCT domains, and we showed that many are so destabilizing that the folded functional state is drastically depopulated at physiological temperature. In the present study, we ask whether and how reduced thermodynamic stability of the isolated BRCT mutants translates into loss of function of the full-length protein in the cell. We assessed the effects of missense mutants on different stages of BRCA1-mediated DNA repair by homologous recombination using chicken lymphoblastoid DT40 cells as a model system. We found that all of the mutations, regardless of how profound their destabilizing effects, retained some DNA repair activity and thereby partially rescued the chicken BRCA1 knockout. By contrast, the mutation R1699L, which disrupts the binding of phosphorylated proteins (but which is not destabilizing), was completely inactive. It is likely that both protein context (location of the BRCT domains at the C-terminus of the large BRCA1 protein) and cellular environment (binding partners, molecular chaperones) buffer these destabilizing effects such that at least some mutant protein is able to adopt the folded functional state.
A key component of HR (homologous recombination) is the protein BRCA1 (breast cancer early-onset 1), a master regulator that co-ordinates the response to DNA damage [1,2]. BRCA1 has been found in four complexes: the BRCA1-A complex, which detects and marks sites of DNA damage, the BRCA1-B complex, involved in the S-phase checkpoint, the BRCA1-C complex, which forms part of DNA repair machinery and is critical in the resection of DSBs (double-strand breaks), and the PALB2 (partner and localizer of BRCA2)–BRCA2 complex, involved in Rad51-mediated strand invasion [3–5]. Although the functions of BRCA1 are not yet fully characterized, most studies indicate that HR is critical for the protein's tumour-suppressor activities .
Carriers of germline mutations in BRCA1 have an increased lifetime risk of developing breast and ovarian cancers, and mutations in the BRCA1 gene account for 80% of all familial breast and ovarian cancer cases . BRCA1 encodes a large protein of 1863 residues with only two small structural motifs characterized to date. At the N-terminus, there is a RING finger domain, which, by binding to BARD1 [BRCA1-associated RING domain 1; another RING- and BRCT (BRCA1 C-terminus domain)-domain-containing protein], gives rise to ubiquitin ligase activity. The structural content of the remainder of the protein is currently unknown, although biophysical studies suggest that much of it is intrinsically disordered . At the C-terminus, there is a repeat of two BRCT domains (a phosphoprotein-recognition domain found in numerous proteins involved in DNA repair and maintenance of genomic stability) that bind to proteins key to the cell's HR response upon their phosphorylation, specifically Abraxas, BACH1 (BTB and CNC homology 1) and CtIP [CtBP (C-terminal-binding protein)-interacting protein] .
Genetic screening of cancer patients has led to an unprecedented increase in BRCA1 sequence data. However, unlike nonsense and deletion mutations, where it is usually obvious that they will be disease-causing, predicting clinical significance for missense mutations from the genetic data can be very challenging, as they may be either natural polymorphisms or detrimental to function. It has been predicted for proteins in general that the vast majority of disease-associated missense mutations cause loss of function in an indirect manner by destabilizing the three-dimensional structure of the protein rather than directly by disrupting a binding site or active site [9–11]. It would therefore be useful to have a BRCA1 assay that measures its structural stability so that the effects of mutations can be determined and disease risk thereby assessed. With this aim in mind, we previously established an assay for measuring the thermodynamic stability, i.e. the free energy difference between the unfolded and native state, of the isolated BRCA1 BRCT domains . We analysed the effects on both thermodynamic stability and phosphopeptide binding of 36 missense mutations selected from the BIC database (Breast Cancer Information Core, http://research.nhgri.nih.gov/projects/bic/index.shtml). We found that the mutations ranged from slightly stabilizing to strongly destabilizing. Our data indicated that the majority would result in a significant amount of unfolding of the BRCT domains at physiological temperature. Interestingly, despite the large changes in stability, all of those mutant variants that could be solubly expressed in Escherichia coli (with the exception of those located in the binding site) were able to bind a phosphorylated model peptide with affinities similar to that of WT (wild-type) when experiments were performed under permissive conditions of low temperature (10°C) to favour folding; this finding indicates that the mutations do not induce a misfolded conformation of BRCA1.
We asked to what extent does reduced thermodynamic stability of the isolated BRCT domains upon mutation translate into loss of function of the full-length protein in the cell? We assessed the HR activity of a subset of BRCA1 BRCT mutations having a range of destabilizing effects, using chicken DT40 cells as a model system. We found that the majority of the mutations, regardless of how profound their destabilizing effect, had some HR activity and thereby partially rescued the cBRCA1 (chicken BRCA1) knockout. By contrast, the mutation R1699L (Arg1699Leu), which disrupts phosphoprotein binding, but which is not destabilizing, was completely inactive. We conclude that the cell is able to buffer destabilizing mutations such that at least a proportion of the mutant protein molecules can fold into and maintain the correct functional form.
DT40 WT and BRCA1−/− cells stably transfected with a single copy of the recombination repair substrate DR-GFP (direct repeat GFP) plasmid were a gift from K. Hiom . They were grown in RPMI 1640 medium with 10% FBS, 1% chicken serum, 100 units·ml−1 penicillin and 100 μg·ml−1 streptomycin, at 39.5°C under 5% CO2. DT40 cells have been used extensively to study cell cycle and DNA repair .
Missense mutations were introduced by site-directed mutagenesis into pENTRY3C (Invitrogen), containing full-length hBRCA1 (human BRCA1) (amino acids 1–1863) with a C-terminal hexahistidine tag and FLAG tag, a gift from K. Hiom. The tagged BRCA1 and SNVs (single nucleotide variants) were recombined with ptREX-DEST30 (Invitrogen) to give mammalian expression constructs known as hBRCA1 and SNV plasmids.
A total of 5×106 DT40 BRCA1−/− cells at 8×105 cell·ml−1 were centrifuged at 320 g for 5 min and the medium was removed. For the GFP DNA-repair assay and camptothecin assay, the cells were resuspended in 100 μl of solution T from Amaxa Kit T (Lonza) and for the Rad51 IRIF (ionizing-radiation-induced foci) assay, 100 μl of solution R from Amaxa Kit R (Lonza). After addition of vectors, they were electroporated using the Amaxa electroporator program B-023. Cells were then transferred to a 100 mm diameter Petri dish with 14 ml of medium and incubated at 39.5°C for 24 h.
Rad51 IRIF assay
Cells were transfected with 6 μg of hBRCA1 or SNV plasmids (as above), and, 24 h after transfection, they were split into two treatment groups: one half was left in the incubator (0 Gy), whereas the other half was subjected to 5 Gy of ionizing radiation with a Mainance Millennium sample irradiator, before being placed back at 39.5°C.
After 3 h, cells were counted and 106 cells were centrifuged for 5 min at 72 g on to a poly-L-lysine-coated slide in a Shandon Cytopsin. Slides were air-dried and cells were fixed with 4% paraformaldehyde for 10 min and then rinsed in PBS three times. The cells were permeabilized with 0.2% Triton X-100 in PBS for 5 min, and washed with PBS three times. Slides were blocked in 1% BSA in PBS for 30 min at 37°C, before staining with mouse monoclonal anti-BRCA1 antibody (Santa Cruz Biotechnology, sc-6954, clone D9) at 1:25 dilution for 18 h at 4°C and then incubated with a rabbit polyclonal anti-Rad51 antibody (Merck Millipore, PC130, clone Ab-1) at 1:500 dilution for 2 h at 37°C. In some experiments, cells were stained with a mouse monoclonal anti-phospho-histone H2AX (Ser139) antibody (Merck Millipore, 05-636, clone JBW301) at 1:1000 dilution and with a rabbit polyclonal anti-Rad51 antibody (Merck Millipore, PC130, clone Ab-1) at 1:500 dilution for 2 h at 37°C. After three 5 min washes in PBS, slides were incubated with secondary antibodies (Alexa Fluor® 488-conjugated donkey anti-mouse and Alexa Fluor® 594-conjugated donkey anti-rabbit, Jackson ImmunoResearch), both at 1:200 dilution, for 45 min at 37°C. After four 5 min washes in PBS in the dark, slides were mounted with DAPI in DABCO (1,4-diazadicyclo[2.2.2]octane).
Slides were imaged using an Operetta High Content system (PerkinElmer), using the ×60 ND (neutral density) objective, imaging 42 fields per slide. The High Content Harmony software (PerkinElmer) was used to identify the nucleus, the cytoplasm and the cell limit. The fluorescence intensity parameters were calculated for the DAPI, BRCA1 and Rad51 signals, before counting Rad51 foci in cells.
Transfection efficiency was quantified by stratifying the Alexa Fluor® 488 channel signal (intensity contrast) over the whole cell. The average BRCA1 intensity contrast was calculated for BRCA1−/− cells, on replicate slides, both with and without irradiation. Cells with higher BRCA1 intensity after transient transfection with hBRCA1 and SNV plasmids were considered as expressing BRCA1 and were included in the quantification of Rad51 foci. Cells with four or more Rad51 foci were considered positive. Between 224 and 9277 cells per slide were included in the Rad51 foci count. In some experiments, cells were imaged using an Olympus IX81 epifluorescence microscope.
GFP DNA-repair assay
DT40 BRCA1−/− cells containing the DR-GFP reporter plasmid were transfected with 6 μg of hBRCA1 or SNV plasmids and 6 μg of pCISceI, encoding the I-SceI restriction enzyme that induces DSBs in the reporter GFP gene. Transfection efficiency was assessed by transfecting with 6 μg of hBRCA1 or SNV plasmids and 6 μg of pMAX-GFP (Amaxa). Cells were incubated at 39.5°C for 24 h, and 10 ml cultures were harvested and washed once in 50 μl of PBS. Cells were analysed on a FACSCalibur instrument (Becton Dickinson) for GFP expression and 20000 events were recorded. Flow cytometry data were analysed with FCS Express 3 program (De Novo Software). Data were corrected for transfection efficiency for each mutant as measured by the percentage of cells that expressed GFP after co-transfection with pMax-GFP and the hBRCA1 or SNV plasmids.
BRCA1−/− DT40 cells were transfected with 10 μg of hBRCA1 or SNV plasmids and cells were incubated for 24 h, allowing expression of hBRCA1 and its missense mutations. Cells were treated with 100 nM camptothecin at 39.5°C for 1 h. Cells were harvested and washed three times with PBS and seeded at 105 cells/well in a six-well plate in triplicate. Cell viability was measured using CellTiter-Blue (Promega) after a further 48 h. For each mutant, cell viability after camptothecin treatment was expressed as the percentage of surviving cells compared with untreated cells from the same transfection.
In all assays, data are given as the means±S.E.M. for three experiments.
Choice of BRCA1 missense mutations
Previously we categorized the missense mutations into one of the following groups according to the effect on the thermodynamic stability of the isolated BRCT domains: not destabilizing, mildly destabilizing, moderately destabilizing and strongly destabilizing. We chose a subset comprising 12 of these mutations, covering all four groups (Figure 1). Additionally, R1699L was chosen, as this mutation of a key phosphoprotein-binding-site residue weakens the affinity for a model phosphopeptide by two orders of magnitude, but does not alter BRCT thermodynamic stability .
Details of the hBRCA1 missense mutations used in the present study
Expression of BRCA1 missense mutations
The D9 anti-BRCA1 antibody was raised to residues 1842–1862 of hBRCA1, and it therefore allows us to monitor the BRCT domain in the DT40 cells. This antibody did not react with the endogenous cBRCA1 in WT cells and did not show background staining in BRCA1−/− cells (Figure 2A). There is 33% identity between hBRCA1 and cBRCA1, increasing to 65% in the BRCT domains  (Supplementary Figure S1). Of the mutants evaluated in the present study, three are at non-conserved positions: Met1652, Asp1778 and Met1783 in hBRCA1 correspond to isoleucine, glycine and isoleucine in cBRCA1. The mutation M1652I (Met1652Ile) is a semi-conservative change and a known polymorphism in humans , M1783I (Met1783Ile) is a semi-conserved change, and D1778G (Asp1778Gly) is on the surface of the structure and distant from the phosphopeptide-binding site. The other mutations are of residues that are identical between hBRCA1 and cBRCA1. They are located throughout the structure: six of them are in the BRCT repeats themselves, and four are in, or close to, the interface between the repeats that contains the phosphopeptide-binding site. Overall, we expect hBRCA1 to mimic cBRCA1 functionally in chicken DT40 cells.
Expression of hBRCA1 WT and mutants in BRCA1−/− cells
The antibody detected the transiently expressed hBRCA1 WT and missense mutations in BRCA1−/− DT40 cells. Importantly, the hBRCA1 protein expressed in chicken BRCA1−/− cells is functional, as shown by the presence of BRCA1 IRIF (Figure 2A). By measuring BRCA1 fluorescence over the cell, we assessed the amount of cells expressing the hBRCA1 mutants. All of the missense mutants but one are expressed at similar levels, with approximately 50% of the cells being successfully transfected with hBRCA1 (Figure 2B). The exception is A1708E (Ala1708Glu), which was only detected in approximately 10% of cells. In the context of the isolated BRCT domain, A1708E was very destabilizing and could not be expressed in a soluble form in E. coli , therefore the lower number of DT40 cells expressing the full-length A1708E mutation is likely to be the result of the low stability causing the protein to be targeted for degradation, or it could be that the antibody cannot detect the mutant if it is predominantly in the unfolded state. In contrast, the other missense mutations previously classified as ‘strongly destabilizing’, S1715C (Ser1715Cys) and S1841N (Ser1841Asn), were expressed in a similar number of cells to the WT.
Foci formation by BRCA1 WT and missense mutations
As described above, a critical aspect of BRCA1 function is to co-ordinate the response to DNA damage, and it has been shown to form nuclear foci after treatment with a number of DNA damaging agents . Figure 3A shows the ability of the transiently expressed hBRCA1 mutants to form nuclear foci after irradiation damage. Some mutants did form IRIF [T1773S (Thr1773Ser), D1778N (Asp1778Asn), D1692N (Asp1692Asn), M1652I and M1783T (Met1783Thr)], whereas others did not [R1699L, G1788D (Gly1788Asp), V1736A (Val1736Ala), A1843P (Ala1843Pro), A1708E, S1715C and S1841N]. The mutants able to form IRIF correspond to those we classified previously as having a mild effect on BRCT stability , with only one classified as moderately destabilizing (M1783T). In contrast, mutants that were not able to form IRIF correspond to those classified previously as having a moderately or strongly destabilizing effect on BRCT stability. Additionally, R1699L, which is located in the phosphoprotein-binding pocket and drastically reduces the affinity for a phosphopeptide, but has no effect on BRCT stability, did not form IRIF. This result confirms the requirement for BRCA1 binding to phosphorylated proteins such as Abraxas, a member of the BRCA1-A complex, involved in the recruitment process of BRCA1 to sites of DNA repair . Those mutants that failed to form IRIF also showed diffuse staining of BRCA1 in the cytoplasm, indicating that the folded state of the BRCT domain is crucial for its correct localization in the cell. Mislocalization has been observed previously in mammalian cells for two mutations, P1749R and M1775R ; these mutations were categorized elsewhere as having either a folding defect (P1749R) or defective phosphopeptide binding (M1775R) .
Effects of hBRCA1 mutants on recruitment of Rad51 to sites of DNA damage
BRCA1 missense mutations cause defective Rad51 IRIF formation
Next, we assayed the short-timescale response of BRCA1 to DNA damage 3 h after irradiation. The DNA damage response is initiated by the phosphorylation of the histone H2AX on Ser139 (γH2AX) and the subsequent recruitment of BRCA1, which in turn recruits the PALB2–BRCA2–Rad51 complex. Therefore we used Rad51 IRIF to monitor downstream recruitment of the DNA repair proteins by BRCA1. Rad51 is essential for the homology search and facilitates the formation of the heteroduplex DNA, before generation of the correctly repaired DNA .
Figure 3B shows staining with antibodies against γH2AX and Rad51 of WT DT40, BRCA1−/− and BRCA1−/−+hBRCA1 cells. WT, BRCA1−/− and BRCA1−/−+hBRCA1 cells show very few (typically less than four per cell) γH2AX or Rad51 IRIF in the absence of DNA damage. WT cells show a normal response to DNA damage with co-localizing γH2AX and Rad51 IRIF present, but Rad51 IRIF are markedly reduced in BRCA1−/− cells, agreeing with the reduced HR efficiency in the absence of cBRCA1 described by Bridge et al. . Rad51 IRIF formation was restored after expression of hBRCA1 in BRCA1−/− cells (Figure 3B and 3D), showing that hBRCA1 is as effective as cBRCA1 at recruiting Rad51 to sites of DNA repair and indicating that, although the sequences are divergent, the regions critical for function across these assays are conserved.
We next measured the ability of the hBRCA1 missense mutations to recruit Rad51 to sites of DNA damage. Figure 3C shows the Rad51 IRIF counted by the Harmony software in WT cells before and after DNA damage. In non-irradiated cells, all of the missense mutants behave as the WT with less than 5% of cells showing more than four IRIF per cell (Figure 3D). However, after 5 Gy of irradiation, there are notable differences between the missense mutants in their ability to recruit Rad51 to IRIF. The absence of cBRCA1 in the BRCA1−/− cells caused a very significant reduction in the recruitment of Rad51 after DNA damage (P≤0.001), but this defect was rescued by the expression of WT hBRCA1. A number of mutations showed a statistically significant reduction in Rad51 foci formation compared with cells transfected with WT hBRCA1: R1699L, D1778N, M1783T, G1788D, V1736A and A1708E. The majority of those found to be defective in forming BRCA1 IRIF also failed to recruit Rad51, as would be expected from the dependence of one on the other.
In those mutant proteins that show a significant reduction in Rad51 recruitment, the mutated residue is often situated in the inter-domain region between the BRCT domains. These mutations were also found in our previous study to be moderately destabilizing (loss of free energy of unfolding of 2.2–5 kcal·mol−1; 1 kcal=4.184 kJ). Given that phosphorylated proteins bind in a cavity formed between the two domains, it is likely that the structural integrity of this region is particularly sensitive to the destabilizing effect of mutations.
As R1699L is a mutation in the phosphoprotein-binding site that disrupts the affinity for ligands, it is unable to interact with CtIP, a protein necessary for end resection of the DNA before the coating of single-stranded DNA by Rad51 [13,22,23], and indeed this mutation showed a level of Rad51 IRIF similar to that of the BRCA1−/− cells.
The low number of Rad51 foci observed for A1708E is consistent with its low transfection, arising from low stability as discussed above. It was only found expressed in approximately 10% of the cells, and the cells that did express this mutant formed very few Rad51 foci, comparable with the levels seen in BRCA1−/− cells.
Surprisingly, the three missense mutations with the lowest thermodynamic stabilities, S1715C, S1841N and A1843P, were all able to recruit Rad51 to sites of DSBs at a level similar to the WT protein. These mutations are not located on the phosphoprotein-binding surface itself, but they form part of the scaffold beneath it. Our results therefore suggest that, by forming a complex with other proteins, BRCA1 is stabilized and this can compensate for the destabilizing effects of the mutations.
Effects of BRCA1 missense mutations on GFP DNA repair
The Rad51 IRIF assay examines the ability of the cells to recruit the DNA-repair machinery, and this second assay examines whether the DNA lesion can be repaired. A defective GFP gene, missing an internal fragment, was stably inserted into DT40 cells . This GFP gene is cut by the transiently transfected restriction enzyme, I-Sce1, which produces a DSB. The successful repair by HR of the DNA break using the missing internal GFP fragment located downstream results in a correct GFP gene; subsequent GFP expression can then be quantified by flow cytometry . Supplementary Figure S2 shows flow cytometry data for WT and BRCA1−/− cells in this assay. After I-Sce1 transfection, the proportion of WT cells performing HR increases from 0.1% to 11.8%, whereas only 1.4% of BRCA1−/− cells show a GFP signal, a 9-fold decrease in HR efficiency in the absence of cBRCA1. The expression of human WT BRCA1 increases HR efficiency to 5.7%, but does not restore it to WT levels (Figure 4). Only three of the mutations, D1778N, M1783T and S1841N, were able to repair GFP to the same extent as hBRCA1. All of the others showed a statistically significant reduction in HR. As in the previous assay, R1699L showed the same level of GFP expression as the BRCA1−/− cells consistent with phosphoprotein binding to BRCA1 being required for the recruitment of the HR machinery to DSBs.
Effects of hBRCA1 mutants on GFP DNA repair
The results of this assay when compared with those obtained in the Rad51 IRIF assay suggest that, although HR proteins may be successfully recruited by some of the hBRCA1 missense mutations, that alone does not guarantee that HR can be carried out successfully. The failure at different stages of the HR process may be a reflection of the involvement of BRCA1 in multiple complexes with distinct roles in the detection of DSBs, DNA resection and recruitment of repair proteins.
Effects of BRCA1 missense mutations on sensitivity to camptothecin
This whole-cell assay looks at the effect of the BRCA1 missense mutations after a 72 h period. The cells expressing mutant hBRCA1 were challenged with the anti-cancer compound camptothecin. Camptothecin is a specific inhibitor of topoisomerase I that acts by stabilizing the topoisomerase I-cleavage complex, thus preventing further replication and inducing DSBs in one of the sister chromatids . BRCA1, together with CtIP, is required to eliminate the chemical modification caused by camptothecin to allow HR repair to commence . Therefore any detrimental mutation that impairs BRCA1 function will decrease cell survival when challenged with camptothecin.
The BRCA1−/− cells showed reduced cell survival compared with WT DT40 (by approximately 65%) and cell survival could be restored by transient transfection of WT hBRCA1 (Figure 5). All of the missense mutations, with the exception of S1715C, showed an increased sensitivity to camptothecin compared with hBRCA1, and they reduced cell survival by between 25% and 60%. These changes were statistically significant at P≤0.01, with the exception of T1773S and D1778N which were statistically significant at P≤0.05. Notably, the founder mutation, A1708E, reduced cell survival by ~50%, and we can compare this effect with those of other mutations, whose clinical significance is not known. M1652I is generally considered to be a polymorphism; however, in our assay, it impaired cell survival to the same extent as A1708E. R1699L, a mutation of a critical residue in the phosphoprotein (CtIP)-binding site that decreased binding in in vitro assays by two orders of magnitude, also reduced cell survival to an extent similar to that of the BRCA1−/− cells, consistent with the role of CtIP in camptothecin response.
Effects of hBRCA1 mutants on cell survival after camptothecin treatment
Phenotypic rescue of the knockout by BRCA1 missense mutations
In order to assess how the missense mutations performed in HR compared with the WT hBRCA1, and whether they were able to rescue the phenotypes seen in the absence of cBRCA1 in BRCA1−/− cells (i.e. drastic reductions in Rad51 IRIF formation, GFP DNA-repair efficiency and cell survival), the activity of each of the mutants in the three assays was compared with that of the transiently expressed WT hBRCA1. We set the rescue provided by the WT hBRCA1 to 100%, and calculated the percentage of rescue provided by each of the mutants in the three assays. The mutants were then placed into three categories on the basis of their ability to perform HR: >50%, 30–50% and <30% of WT activity. Mutants showing >50% of WT activity in two assays were classified as having no effect (T1773S, D1778N, M1783T, S1715C and S1841N). Mutants showing 30–50% of WT activity in two assays and mutants having one assay in each of the three categories were classified as having a moderate effect (D1692N, M1652I, G1788D and A1843P). Lastly, mutants showing activity <30% of WT in two or more assays were classified as having a strongly deleterious effect (R1699L, V1736A and A1708E). These results are summarized in Figure 6A.
Relative activity of the BRCA1 missense mutations compared with WT hBRCA1 in Rad51 IRIF, GFP DNA repair and cell survival assays
The rare mutation found in the Icelandic population associated with breast cancer, D1692N, was found to have a moderately detrimental effect on HR, whereas the founder mutation in the Sephardic Jews, A1708E, was found to be strongly detrimental (Figure 6B). The R1699L mutation, which has a disrupted phosphoprotein-binding site, was strongly detrimental, having the same effect as the cBRCA1 knockout and underlining the importance of phosphoprotein binding to the BRCT domains for efficient protein interaction and downstream signalling. The three moderately destabilizing mutations, G1788D, V1736A and A1843P, were also detrimental in the cell.
Relationship between loss of structural stability of BRCT domains and loss of BRCA1 cellular function
A number of studies have characterized missense mutations in the BRCA1 BRCT domains in vitro using assays such as phosphopeptide binding, protease sensitivity and transcriptional activation [19,27–29]; however, fewer have looked at the effects on the full-length protein in the cell [30–32]. Moreover, many of the cell-based studies have focused on mutations that disrupt the phosphoprotein-binding site, whereas the effects of loss of thermodynamic stability on BRCA1 function in the cell have not been investigated systematically to date. In the present study, we examined the effects on a number of different stages of BRCA1-mediated HR of missense BRCT mutations having a range of different stabilities. The first assay looked at the initial response to DNA damage, namely BRCA1 localization to sites of damage and its subsequent recruitment of HR factor Rad51; the second measured the ability of the BRCA1 missense mutations to repair a GFP gene ; the third looked at cell survival after DNA damage by camptothecin, a topoisomerase I poison which causes DNA lesions repaired by HR .
Overall, there is not a strong correlation between loss of stability of the isolated BRCT domains and loss of function of the full-length protein in the cellular context (Figure 6B), with many of the mutants displaying different effects in the three assays of BRCA1 function. For example, S1841N was detrimental in the cell-survival assays, but not deleterious in the Rad51 IRIF assay or the HR assay. This variation may reflect different effects of the mutations on the different binding partners in the multiple protein complexes of which BRCA1 is a member, and/or the different types of DNA damage introduced in the IRIF/HR assays compared with the camptothecin assay leading to recruitment of BRCA1 into different complexes . One mutation, the BRCA1 founder mutation A1708E, behaved consistently in the three assays and has been found to be deleterious in a number of previously published assays [19,29,34]. It was detrimental in all three assays and was the only mutant to show reduced transfection indicative of loss of stability. Ala1708 is located in the interface between the two BRCT domains, and it is likely to disrupt the hydrophobic packing between the repeats and prevent phosphoprotein binding.
There is a correlation between the stability of the isolated BRCT domain and the formation of BRCA1 foci after DNA damage, but not between stability and the ability to form Rad51 IRIF after DNA damage. For example, the very destabilizing mutants S1841N, S1715C and A1843P did not form BRCA1 foci, but were able to recruit Rad51 to DSBs (Figure 3A). All of these mutations are located in the core of the protein. S1841N had previously been shown to have detrimental effect on phosphopeptide binding, protease sensitivity and transcription activation, whereas S1715C was classified as “uncertain” . That these profoundly destabilized mutant BRCA1 proteins were still able to recruit Rad51 could potentially be explained by the fact that the PALB2–BRCA2–Rad51 complex does not interact with BRCA1 via the BRCT domains and therefore its binding may be unaffected by the folding status of the BRCT domains. Alternatively, it has been shown that proteins other than BRCA1 are important for the sustained localization of the PALB2–BRCA2–Rad51 complex at sites of DNA damage . The lack of BRCA1 IRIF seen for several mutations may be due to a change in the interaction with the BRCA1-A complex [Abraxas–RAP80 (receptor-associated protein 80)] involved in localizing BRCA1 to DSBs; these changes could lead to altered kinetics of IRIF formation/dispersion for these mutants such that foci could still have been formed, but may have been missed if they were late-forming or had already dispersed, as we only looked at one time point. The mutation G1788D was found to be even less effective than the BRCA1 knockout in recruiting Rad51 after irradiation. This temporal delay or inhibition of Rad51 IRIF may be a result of an interference of G1788D with BRCA1-independent Rad51 recruitment.
Among the missense mutations that we have classified here as having a moderate effect on HR function, two were thought previously to be benign and functional: M1652I (a polymorphism; cBRCA1 has an isoleucine residue at this position) and D1692N (a rare mutation found in the Icelandic population). For M1652I, the results obtained may be cell-type- and assay-specific, as in mouse embryonic stem cells, no hypersensitivity to a range of DNA-damaging agents was seen; however, camptothecin was not tested in that drug screen, making it difficult to compare those results with ours . Other studies showed that D1692N did not impair the transcription activity of BRCA1 [19,36], but had not investigated HR.
The mutation V1736A is classified differently in various previously published studies. Val1736 is conserved in 18 different vertebrate species; it is distant from the phosphoprotein-binding pocket and does not make direct contact with the phosphoprotein ligand . V1736A was predicted by computational analysis to be deleterious [38,39] and was classified as deleterious in a transcription assay , but it was classified as class 1 (intermediate overall) by IARC (http://brca.iarc.fr/LOVD) and as “uncertain” in a structural/functional analysis of the isolated BRCT domain . Our finding that it has a very deleterious effect on DNA repair (especially in recruiting Rad51 and in carrying out HR) agrees with studies showing that it reduces the localization of BRCA1 at sites of DSBs  and disrupts the interaction with RAP80 . BRCA1 interacts with RAP80 via Abraxas [40–42], and all three proteins are part of the BRCA1-A complex involved in the initial stages of DSB repair and in targeting BRCA1 to DSBs . Studies have indicated that the formation of the BRCA1-A complex prevents access of other BRCA1-containing complexes to the repair site, thereby actually inhibiting BRCA1-driven HR [43,44]. We showed previously that V1736A caused a moderate destabilization of the BRCT domains and a 5-fold decrease in phosphopeptide affinity . If V1736A disrupted BRCA1-A complex formation without affecting the formation of other complexes, one might expect it to enhance HR [43,44]. Therefore our finding that V1736A inhibits HR suggests instead that the formation of other HR-promoting BRCA1 complexes is also reduced.
In summary, it is clear that BRCA1 functions are regulated by dynamic interactions with multiple complexes, and it is possible that different BRCT domain mutations might differentially disrupt the formation of some of these complexes and thereby lead to the use of a different DNA repair pathway.
Buffering of destabilizing mutations due to protein and cellular contexts: stabilization due to protein–protein interactions
A number of studies have examined the effects of destabilizing mutations on protein evolution [45–50]. They showed that mutations that reduce the structural stability of a protein constrain its capacity to evolve function. The importance of stability effects underlined by these studies resonates with computational analysis of disease-associated missense mutations, which indicate that the vast majority cause loss of function indirectly by destabilizing the structure rather than directly by disrupting functional sites (e.g. ). Both phenomena follow from the fact that functional sites need a globally stable scaffold to underpin them and the number of residues required to maintain the scaffold greatly exceeds the number of residues required, for example, for an active site. These studies indicate that proteins can accommodate a loss of stability upon mutation of 1–2.5 kcal·mol−1 without any detrimental cellular effect [9,46,49,51–54]. Any greater loss and the protein will not be stable enough to resist misfolding, aggregation and degradation, all processes that reduce the levels of functional protein in the cell; the background mutational landscape must also be taken into account when considering the cell's ability to buffer destabilizing mutations [55,56]. Previous studies have suggested that only 10% of the protein molecules needs to be folded for there to be no detrimental impact on the cell [57,58]; in our previous analysis of the isolated BRCA1 BRCT domain, we showed that many of the mutations are so destabilizing that much less than 10% of the molecules are folded at 37°C, so how does one explain why some, at least, of these BRCA1 missense mutations are able to retain function in the cell? It is important to note that many studies of stability effects have used small single-domain enzymes as models. BRCA1, however, is representative of a very different class of protein; consequently, additional factors may be at play when it comes to translating stability in the test tube into function in the cell. BRCA1 is a giant scaffold protein, possessing both structured and unstructured regions that are involved in multiple protein–protein interactions. Our results suggest that proteins such as BRCA1 can compensate for the destabilizing effects of mutations both intermolecularly via the stability gained from interactions with their natural binding partners and intramolecularly via interactions between its domains, as discussed in .
The cellular stabilization of BRCA1 that we observe may also be achieved via the action of molecular chaperones, as has been found for other proteins . Interestingly, in this regard, it has been shown that the molecular chaperone Hsp90 (heat-shock protein 90) can enhance the stability of a BRCA1 mutant truncated in the BRCT region; as a result of this stabilization, the protein was able to recruit the PALB2–BRCA2–Rad51 complex [61,62]. The potential for natural binding partners to buffer BRCA1 against the destabilizing effects of mutations is consistent with our previous in vitro analysis showing that binding of a phosphopeptide to the BRCT domains results in restoration of a destabilizing mutation to the WT stability . Indeed, it has been proposed that binding proteins can act as evolutionary capacitors , as do molecular chaperones . In future, estimation of these cellular effects will need techniques that can measure protein thermodynamic stability directly in the cell [63–67]. Another example of the cell's buffering capacity is provided by a study of phosphoglycerate kinase , which was found to be more resistant to thermal unfolding in the cell than in the test tube. If BRCA1 mutants were similarly stabilized by the cellular context, then there could be sufficient folded protein available to maintain DNA repair. Clearly, we will need a much better understanding of these factors before we can predict the effects of mutations on protein stability, and more importantly protein activity, in the cell; a major challenge today in the fields of biology and medicine.
breast cancer early-onset 1
BRCA1 C-terminus domain
CtBP (C-terminal-binding protein)-interacting protein
direct repeat GFP
partner and localizer of BRCA2
receptor-associated protein 80
single nucleotide variant
David Gaboriau and Pamela Rowling contributed equally to the design, execution and analysis of the experiments. Project advice was given by Ciaran Morrison and Laura Itzhaki. The project was conceived by Laura Itzhaki. David Gaboriau, Pamela Rowling and Laura Itzhaki wrote the paper. All authors critically revised and approved the final paper.
We thank Kevin Hiom (University of Dundee, Dundee, U.K.) for the DT40 cells with the DR-GFP construct (Maria Jasin, Memorial Sloan Kettering Cancer Center, NY, U.S.A.), Ko Sato (MRC Cancer Cell Unit, Cambridge, U.K.) for help with DT40 work, and members of the Itzhaki and Morrison laboratories for useful discussion
This work was supported by the Medical Research Council (MRC) [including grant number G1002329] and the European Commission [BOOSTER project (BiO-dOSimetric Tools for triagE to Responders), project number 242361, call number SEC-2009-4.3-02], a Medical Research Foundation Senior Fellowship to L.S.I. and a Science Foundation Ireland Principal Investigator award [grant number 10/IN.1/B2972 (to C.G.M.)].
These authors contributed equally to this work.