Germline inactivating mutations in BRCA1 and BRCA2 genes are responsible for Hereditary Breast and Ovarian Cancer Syndrome (HBOCS). Genetic testing of these genes is available, although approximately 15% of tests identify variants of uncertain significance (VUS). Classification of these variants into pathogenic or non-pathogenic type is an important challenge in genetic diagnosis and counseling. The aim of the present study is to functionally assess a set of 7 missense VUS (Q1409L, S1473P, E1586G, R1589H, Y1703S, W1718L and G1770V) located in the C-terminal region of BRCA1 by combining in silico prediction tools and structural analysis with a transcription activation (TA) assay. The in silico prediction programs gave discrepant results making its interpretation difficult. Structural analysis of the three variants located in the BRCT domains (Y1703S, W1718L and G1770V) reveals significant alterations of BRCT structure. The TA assay shows that variants Y1703S, W1718L and G1770V dramatically compromise the transcriptional activity of BRCA1, while variants Q1409L, S1473P, E1586G and R1589H behave like wild-type BRCA1. In conclusion, our results suggest that variants Y1703S, W1718L and G1770V can be classified as likely pathogenic BRCA1 mutations.
Citation: Quiles F, Fernández-Rodríguez J, Mosca R, Feliubadaló L, Tornero E, Brunet J, et al. (2013) Functional and Structural Analysis of C-Terminal BRCA1 Missense Variants. PLoS ONE8(4): e61302. https://doi.org/10.1371/journal.pone.0061302
Editor: Paolo Peterlongo, IFOM, Fondazione Istituto FIRC di Oncologia Molecolare, Italy
Received: January 29, 2013; Accepted: March 7, 2013; Published: April 17, 2013
Copyright: © 2013 Quiles et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by: the Spanish Ministry of Health ISCIII FIS grants PI10/01422, PI09/02483 and PI12/01528, RETICC grants RD06/0020/1051, RD06/0020/1050, RD12/0036/008 and RD12/0036/0031; the AGAUR Catalan Government Agency grants 2009-SGR293 and 2009-SGR290; the Spanish Association Against Cancer (AECC 2010); and the Fundación Eugenio Rodriguez Pascual grant 2012. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Between 10 and 20% of the breast cancer cases appearing in the general population present familial history of the disease . Mutations in BRCA1 and BRCA2 genes confer high lifetime risks of breast and ovarian cancer, among other neoplasias , . Inactivating germline mutations in these genes account for 20–50% of familial cases, depending on the population . Thus, genetic analysis of BRCA1 and BRCA2 is a cornerstone of genetic counseling practice. However, classification of genetic variants as pathogenic is challenging, particularly for missense changes and for silent or intronic variants that cannot be directly associated with increased cancer risk and are classified as variants of uncertain significance (VUS), which are found in 13% of BRCA1 and BRCA2 genetic tests . Diverse multifactorial likelihood algorithms have been developed and applied for both BRCA1 and BRCA2 variants (reviewed in Spurdle et al., 2011) . These models use the combination of a number of independent features (sequence conservation, type of amino-acid change, familial co-segregation, family and personal cancer history, tumor data, co-occurrence with a deleterious mutation, and severity of amino acid change) to establish the likelihood that a given VUS is pathogenic or non-pathogenic. In many cases there is not enough data to classify such mutations for clinical purposes. Therefore, functional analyses that assess specific properties of BRCA1 or BRCA2 may help to classify VUS . In particular, BRCA1 encodes for a protein of 1,863 amino acids and with multiple functional domains . Several functional assays have been conducted to evaluate VUS in BRCA1 at the level of its global and domain-based functions, including ubiquitin ligase activity assays, protease sensitivity assays, phosphopeptide binding assays, small colony phenotype assays, yeast localization phenotype assays, and embryonic stem cell-based functional assays (review in Millot et al., 2012) . In the work presented here, we combine a functional assay - the transcription activation (TA) assay, which is based on the function of the BRCA1 carboxy-terminal region (aa 1396–1863) in transcriptional activation domain when linked to a sequence-specific DNA binding module -  with protein structural analyses  to assess the functional impact of seven BRCA1 C-terminal VUS. Our results indicate that three of them (Y1703S, W1718L, G1770V) have significant functional impact and may represent pathogenic BRCA1 variants while the remaining four do not have a functional impact.
Materials and Methods
Seven missense BRCA1 C-terminal variants (Q1409L, S1473P, E1586G, R1589H, Y1703S, W1718L and G1770V), identified through genetic testing of patients with suspicion of Hereditary Breast and Ovarian Cancer Syndrome (HBOCS) as part of the Hereditary Cancer Program of the Catalan Institute of Oncology, were included in this study (Figure 1). Written informed consent was obtained from all subjects. The study received the approval of the Ethics Committee of IDIBELL. The pedigrees of these families are depicted in Figure S1. BRCA1 and BRCA2 genetic analyses consisted of screening of point mutations and large genomic rearrangements . Genetic tests were carried out once patients had received appropriate genetic counseling and provided written informed consent.
In Silico Predictions
The BRCA1 sequences containing the VUS were evaluated for their potential pathogenicity using the following algorithms: Align-GVGD , PolyPhen-2 , SIFT , Mutation Assessor , PhD-SNP , SNAP , I-Mutant , PON-P , and Condel . The Align-GVGD algorithm was used at two depths of alignment (human to frog and human to sea urchin). The results of PolyPhen-2 were retrieved from the original webpage (version 2.2.2) but also from version 2.0.22 run by PON-P and version 1 run by Condel, which use them for weighted average scores.
Transcription Activation Assay
The assay was based on the mammalian expression of the firefly luciferase gene under the control of five GAL4 binding sites . In this system, pcDNA3 constructs containing the GAL4 DNA binding domain (DBD) fused to the wild-type or VUS-containing BRCA1 C-terminal region were used. Pathogenic mutation controls (M1775R and Y1853X) were included in the assays. Constructs containing the VUS were generated by site-directed mutagenesis using the QuickChange II Site-Directed Mutagenesis Kit (Agilent Technologies, USA) and the wild-type sequence as a template (U14680). Human embryonic kidney (HEK) 293T cells were co-transfected (X-tremeGENE 9 DNA transfection Reagent, Roche Applied Science, Mannheim, Germany) with the pcDNA3 construct encoding either the wild-type or the VUS sequence, the pG5Luc plasmid that contains the luciferase reporter gene, and the phRG-TK plasmid that contains a Renilla luciferase gene under the control of a constitutive thymidine kinase promoter, which acts as an internal control. Luciferase was quantified using the Dual-Luciferase Reporter Assay System (Promega, Madison, USA).
Protein Structural Rationalizations
To predict the impact of the VUS on both the folding and phosphopeptide binding of the BRCT domains a structural analysis was carried out by mapping the position of the variants on the BRCA1 structures collected from the Protein Data Bank  (PDB, www.pdb.org) and analyzing them manually with Pymol (http://www.pymol.org). The structures were collected by BLAST-querying the PDB using the BRCA1 sequence (UniProt accession code P38398). Only the VUS mapped in the BRCT tandem domains (Y1703S, W1718L and G1770V) were analyzed using this approach, as there are no structural data for the upstream domain or regions.
In Silico Predictions
Several algorithms designed to predict deleterious versus neutral missense changes were used (Table 1). Align-GVGD , PolyPhen-2 , SIFT , Mutation Assessor , PhD-SNP , SNAP  and I-Mutant  differ in the properties taken into account to generate the prediction and in the method and possible training for decision-making (reviewed in Thursberg, et al., 2010) . PON-P  integrates PolyPhen2, SIFT, PhD-SNP, SNAP and I-Mutant using a random forest method, whereas Condel integrates PolyPhen2, SIFT and Mutation Assessor using a weighted average of the normalized scores of the individual methods. Both integration methods claim to outperform individual ones , . Table 1 shows the outcomes of all methods, for comparison with high discrepancy between different programs. The two integrative algorithms indicate that variants Q1409L, S1473P, E1586G and R1589H have a benign effect and that G1770V has a damaging effect, whereas they disagree in their predictions for variants Y1703S and W1718L. Furthermore the results of PON-P and Condel for Q1409L and G1770V strikingly diverge from those of Align-GVGD, which classify Q1409L as most likely to interfere with BRCA1 function and G1770V as less likely using the alignment from human to frog. Also SIFT, Mutation Assessor, PhD-SNP, SNAP and I-Mutant show divergent tendencies.
Transcription Activation Assay
Using the transcriptional assay, three VUS (Y1703S, W1718L and G1770V) showed a significant decrease in reporter expression compared to wild-type BRCA1 (Figure 2). The luciferase activity of these three mutants was very similar to that observed for the known pathogenic mutants used as controls: that is <5%, compared to the wild-type construct. Conversely, the remaining four VUS (Q1409L, S1473P, E1586G and R1589H) displayed similar transcriptional activity to that of the wild-type construct.
Percentage of transcriptional activity was expressed as percentage relative to the wild-type construct. Experiments were performed in triplicate and normalized against the Renilla luciferase internal control. Results are shown in bars: gray for controls and black for variants.
Protein Structural Rationalizations
A structural analysis was performed for those VUS located in the BRCT domains (Y1703S, W1718L and G1770V). Y1703 is located at the BRCT dual-repeat-interaction interface (Figure 3A), directed towards the linker. It forms a hydrogen bond with H1746 (Figure 3B). The change to a serine, by removal of the aromatic ring, is likely to disrupt this bond and the surrounding hydrophobic interactions and to affect the stability of the domain and the dual-repeat interaction. It may also affect the peptide binding through K1702 (Figure 3B), which interacts directly with the peptide’s phosphoserine. W1718 is loosely packed in the core of the BRCT-N domain and is part of a highly conserved motif within the α3-helix (Figure 3C). The substitution of this bulky residue with a leucine likely destabilizes the folding of the domain, analogously to experimentally confirmed observations of substitution by a cysteine , . G1770 is solvent-exposed and located on a tight turn of the loop connecting the β1 strand with the α1 helix in the BRCT-C repeat (Figure 3D). It is highly conserved in the Pfam alignment of the BRCT domain and probably confers to the loop the necessary flexibility for phosphopeptide binding. Substitution by a valine would expose a hydrophobic side chain and confer rigidity to the loop, compromising the binding. In contrast to the other mutations, this change may not affect folding.
Representation of the structure of the BRCT tandem repeat from human BRCA1 in complex with a binding phosphopeptide (with the typical binding motif pSer-X-X-Phe, PDB ID 1t2v): in blue, the N-terminal repeat (BRCT-N); in red, the C-terminal repeat (BRCT-C); in green, the linker region between the two; in cyan, the phosphopeptide with the phosphoserine and the key binding residue Phe(+3); in, magenta the three mutated residues. Y1703 (A) forms a hydrogen bond with H1746 and is adjacent to K1702, which binds directly to the pSer residue in the phosphopeptide. W1718 (B) fills the internal core and stabilizes the BRCT-N domain. G1770 (C) gives the loop connecting the β1 strand with α1-helix in the BRCT-C repeat the necessary flexibility to form a tight turn, bringing the residues N1774 and M1775 close to the phosphopeptide for interaction with the Phe(+3) key residue.
This study combined a functional BRCA1 assay and structural analyses to predict whether specific missense variants in the BRCA1 C-terminal region have pathological significance, contributing to breast cancer risk. The variants selected were VUS identified in a clinical scenario, which are located between amino acid residues 1,396 and 1,863 of the full-length BRCA1 sequence and include part of a coiled-coil domain (aa 1,364–1,437) and the BRCT tandem domains (aa 1,646–1,863) (Figure 1). Notably, six of the VUS (Q1409L, S1473P, E1586G, Y1703S, W1718L and G1770V) had not been recorded in the locus-specific database for mutations in BRCA1 and BRCA2, BIC (Breast Cancer Information Core database, http://research.nhgri.nih.gov/bic/) (Figure 1 and Table 1).
The use of different in silico prediction algorithms gave contradictory results for some of the variants (Table 1) making it difficult to draw any clear conclusion. For example, there is a striking difference between the results for three versions of PolyPhen-2 as run from different sources; W1718L can yield probabilities of pathogenicity of 0.086, 0.497 and 1.000, while the probabilities for Q1409L, S1473P and E1586G range from values below 0.2 to values above 0.95 in the different versions. If we tried to order the variants according to their deleteriousness as predicted from each of the algorithms, they would yield as many orders as algorithms tested. In general terms, the two integrative programs gave quite similar results, predicting a damaging effect for one of the variants (G1770V) and a benign effect for four of them (Q1409L, S1473P, E1586G and R1589H), while variants Y1703S and W1718L were difficult to classify. Therefore, attending to the results obtained in this study, we consider the use of in silico programs not reliable enough to get an approximation of the pathogenicity of a given VUS. In a recent published paper by Li et al., the predictive power of Condel and five individual methods (SIFT and PolyPhen2 among them) in discriminating between pathogenic nsSNVs and other rare nsSNVs, was examined . Their results showed that the predictive power of the combined Condel methodology is not necessarily better than individual methods. Furthermore and in agreement with our data, they show that correlations in the scores between the predictive methods tested are weak or moderate. The observed discrepancies among programs clearly support the need of adding functional assay results in the process of classification of VUS.
The TA assay in mammalian cells was performed as described in previous studies . This assay has been extensively validated, proving to be extremely robust for variants located in the BRCT domain , , , . In a previous study it has been benchmarked using 14 pathogenic and 10 not pathogenic variants, showing high sensitivity and specificity . In this study two constructs containing previously classified pathogenic variants (IARC Class 5), M1775R and Y1853X, were used as negative controls and a construct with the wild-type BRCA1 C-terminal region was used as a positive control. This assay clearly identified a functional defect for three of the variants (Y1703S, W1718L and G1770V), whereas the remaining variants gave results very similar to those of the wild-type controls. It should be noted that residue W1718 appears to be highly prone to mutation, and although the mutation studied here has never been reported, mutations W1718S and W1718C have been described in HBOCS families; in addition, a panel of assays including protease sensitivity, phosphopeptide binding activity and specificity, as well as the TA assay, revealed that both of these variants have strong functional effects . Both constitute large volume changes at a rigid position in a rigid region that is part of the BRCT signature motif WXXXS .
Next, to better interpret and understand the functional data obtained, a structural analysis was performed for those VUS located in the BRCT domains. This analysis was limited to the BRCT domain as its structure has been solved , making it possible to rationalize whether a given variant alters the domain stability or the phosphopeptide-mediated binding to phosphorylated protein partners (Figure 3). The results indicated that the three studied VUS in the BRCT domains seem to damage proper BRCA1 structure. Interestingly, the experimental data obtained in the TA assay support these structural predictions.
The pedigree structure of patients with these missense mutations was, in general, not very informative for co-segregation analysis (see Figure S1), either due to the lack of relatives with breast (or ovarian) cancer or because no DNA samples for other relatives were available. Nevertheless, it is worth noting that variant Q1409L, which appears to have no effect on TA activity in the present study, was found in a patient with a pathogenic mutation in BRCA2 (c.5720_5723delCTCT), and although a small number of carriers of two pathogenic mutations - one in BRCA1 and one in BRCA2 - have been described worldwide this event has a very low probability. In spite of this, and due to the fact that mutation Q1409L is mapped in the BRCA1/PALB2 interaction region, it would be worth to test if mutant L1409 is hampering the BRCA1/PALB2 interaction . Moreover, mutation G1770V, which shows a clear impairment of TA activity, has been found in two families of Moroccan origin presumed to be independent. In both cases, patients suffer from breast or ovarian cancer at a very early age of onset (27 and 30 years of age, respectively).
In summary, in a clinical context, extensive worldwide analysis of BRCA1 has identified a large number of VUS, most of them unique, which need to be classified for the purposes of diagnosis and genetic counseling. By combining functional analysis and structural rationalizations, this study shows that three BRCA1 VUS (Y1703S, W1718L and G1770V) have profound functional impact and may be pathogenic variants. By contrast, the other four variants did not alter the function of BRCA1 in the TA assay and should remain classified as VUS. The implementation of combined approaches, such as the one presented here, is therefore of considerable clinical relevance, especially in those cases where co-segregation studies are not possible.
We thank Montserrat Soler from the Institute for Research in Biomedicine (IRB) for the fruitful discussion and suggestions. We would also like to thank the Spanish Association Against Cancer (AECC) for recognizing our group with one of its awards.
Conceived and designed the experiments: GC MÀP PA AM CL. Performed the experiments: FQ JFR LF ET RM. Analyzed the data: FQ JFR LF RM JB IB CL. Contributed reagents/materials/analysis tools: RM MÀP PA AM. Wrote the paper: FQ RM MÀP GA AL CL.
- 1. Balmain A, Gray J, Ponder B (2003) The genetics and genomics of cancer. nature genetics 33: 238–244.
- 2. Futreal PA, Liu Q, Shattuck-Eidens D, Cochran C, Harshman K, et al. (1994) BRCA1 mutations in primary breast and ovarian carcinomas. Science 266: 120–122.
- 3. Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman K, et al. (1994) A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science 266: 66–71.
- 4. Kurian AW (2010) BRCA1 and BRCA2 mutations across race and ethnicity: distribution and clinical implications. Curr Opin Obstet Gynecol 22: 72–78.
- 5. Frank TS, Deffenbaugh AM, Reid JE, Hulick M, Ward BE, et al. (2002) Clinical characteristics of individuals with germline mutations in BRCA1 and BRCA2: analysis of 10,000 individuals. J Clin Oncol 20: 1480–1490.
- 6. Spurdle AB, Healey S, Devereau A, Hogervorst FB, Monteiro AN, et al. (2011) ENIGMA–evidence-based network for the interpretation of germline mutant alleles: an international initiative to evaluate risk and clinical significance associated with sequence variation in BRCA1 and BRCA2 genes. Hum Mutat 33: 2–7.
- 7. Carvalho MA, Couch FJ, Monteiro AN (2007) Functional assays for BRCA1 and BRCA2. Int J Biochem Cell Biol 39: 298–310.
- 8. Clark SL, Rodriguez AM, Snyder RR, Hankins GD, Boehning D (2012) Structure-Function Of The Tumor Suppressor BRCA1. Comput Struct Biotechnol J 1: e201204005.
- 9. Millot GA, Carvalho MA, Caputo SM, Vreeswijk MP, Brown MA, et al. (2012) A guide for functional analysis of BRCA1 variants of uncertain significance. Hum Mutat 33: 1526–1537.
- 10. Phelan CM, Dapic V, Tice B, Favis R, Kwan E, et al. (2005) Classification of BRCA1 missense variants of unknown clinical significance. J Med Genet 42: 138–146.
- 11. Williams RS, Green R, Glover JN (2001) Crystal structure of the BRCT repeat region from the breast cancer-associated protein BRCA1. Nat Struct Biol 8: 838–842.
- 12. Feliubadaló L, Lopez-Doriga A, Castellsagué E, del Valle J, Menéndez M, et al.. (2012) Next-generation sequencing meets genetic diagnostics: development of a comprehensive workflow for the analysis of BRCA1 and BRCA2 genes. European Journal of Human Genetics In press.
- 13. Tavtigian SV, Deffenbaugh AM, Yin L, Judkins T, Scholl T, et al. (2006) Comprehensive statistical study of 452 BRCA1 missense substitutions with classification of eight recurrent substitutions as neutral. J Med Genet 43: 295–305.
- 14. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, et al. (2010) A method and server for predicting damaging missense mutations. Nat Methods 7: 248–249.
- 15. Kumar P, Henikoff S, Ng PC (2009) Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc 4: 1073–1081.
- 16. Reva B, Antipin Y, Sander C (2011) Predicting the functional impact of protein mutations: application to cancer genomics. Nucleic Acids Res 39: e118.
- 17. Capriotti E, Calabrese R, Casadio R (2006) Predicting the insurgence of human genetic diseases associated to single point protein mutations with support vector machines and evolutionary information. Bioinformatics 22: 2729–2734.
- 18. Bromberg Y, Rost B (2007) SNAP: predict effect of non-synonymous polymorphisms on function. Nucleic Acids Res 35: 3823–3835.
- 19. Capriotti E, Fariselli P, Rossi I, Casadio R (2008) A three-state prediction of single point mutations on protein stability changes. BMC Bioinformatics 9: S6.
- 20. Olatubosun A, Valiaho J, Harkonen J, Thusberg J, Vihinen M (2012) PON-P: integrated predictor for pathogenicity of missense variants. Hum Mutat 33: 1166–1174.
- 21. Gonzalez-Perez A, Lopez-Bigas N (2011) Improving the assessment of the outcome of nonsynonymous SNVs with a consensus deleteriousness score, Condel. Am J Hum Genet 88: 440–449.
- 22. Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, et al. (2000) The Protein Data Bank. Nucleic Acids Res 28: 235–242.
- 23. Thusberg J, Olatubosun A, Vihinen M (2011) Performance of mutation pathogenicity prediction methods on missense variants. Hum Mutat 32: 358–368.
- 24. Williams RS, Chasman DI, Hau DD, Hui B, Lau AY, et al. (2003) Detection of protein folding defects caused by BRCA1-BRCT truncation and missense mutations. J Biol Chem 278: 53007–53016.
- 25. Glover J (2006) Insights into the Molecular Basis of Human Hereditary Breast Cancer from Studies of the BRCA1 BRCT Domain. Fam Cancer: 89–93.
- 26. Li M-X, Swan JSH, Bao S-YB, Yang W, Ho S-L, et al. (2013) Predicting Mendelian Disease-Causing Non-Synonymous Single Nucleotide Variants in Exome Sequencing Studies. PLOS Genetics 9(1): e1003143 .
- 27. Carvalho MA, Marsillac SM, Karchin R, Manoukian S, Grist S, et al. (2007) Determination of cancer risk associated with germ line BRCA1 missense variants by functional analysis. Cancer Res 67: 1494–1501.
- 28. Lee MS, Green R, Marsillac SM, Coquelle N, Williams RS, et al. (2010) Comprehensive analysis of missense variations in the BRCT domain of BRCA1 by structural and functional assays. Cancer Res 70: 4880–4890.
- 29. Mirkovic N, Marti-Renom MA, Weber BL, Sali A, Monteiro AN (2004) Structure-based assessment of missense mutations in human BRCA1: implications for breast and ovarian cancer predisposition. Cancer Res 64: 3790–3797.
- 30. Zhang F, Fan Q, Ren K, Andreassen PR (2009) PALB2 functionally connects the breast cancer susceptibility proteins BRCA1 and BRCA2. Mol Cancer Res 7: 1110–1118.