ACMG/AMP interpretation of BRCA1 missense variants: Structure-informed scores add evidence strength granularity to the PP3/BP4 computational evidence

Abstract

Classification of missense variants is challenging. Lacking compelling clinical and/or functional data, ACMG/AMP lines of evidence are restricted to PM2 (rarity code applied at supporting level) and PP3/BP4 (computational evidence based mostly on multiple-sequence-alignment conservation tools). Currently, the ClinGen ENIGMA BRCA1/2 Variant Curation Expert Panel uses BayesDel to apply PP3/BP4 to missense variants located in the BRCA1 RING/BRCT domains. The ACMG/AMP framework does not refer explicitly to protein structure as a putative source of pathogenic/benign evidence. Here, we tested the value of incorporating structure-based evidence such as relative solvent accessibility (RSA), folding stability (ΔΔG), and/or AlphaMissense pathogenicity to the classification of BRCA1 missense variants. We used MAVE functional scores as proxies for pathogenicity/benignity. We computed RSA and FoldX5.0 ΔΔG predictions using as alternative input templates for either PDB files or AlphaFold2 models, and we retrieved pre-computed AlphaMissense and BayesDel scores. We calculated likelihood ratios toward pathogenicity/benignity provided by the tools (individually or combined). We performed a clinical validation of major findings using the large-scale BRIDGES case-control dataset. AlphaMissense outperforms ΔΔG and BayesDel, providing similar PP3/BP4 evidence strengths with lower rate of variants in the uninformative score range. AlphaMissense combined with ΔΔG increases evidence strength granularity. AlphaFold2 models perform well as input templates for ΔΔG predictions. Regardless of the tool, BP4 (but not PP3) is highly dependent on RSA, with benignity evidence provided only to variants targeting buried or partially buried residues (RSA ≤ 60%). Stratification by functional domain did not reveal major differences. In brief, structure-based analysis improves PP3/BP4 assessment, uncovering a relevant role for RSA.

Keywords: ACMG/AMP; AlphaFold; AlphaMissense; BRCA1; BayesDel; MAVE; PP3/BP4; RSA; ΔΔG.

Graphical abstract

Figure 1

Summary of relevant structural, functional, and clinical annotations of the 1,867-aa BRCA1 protein The Missense pathogenic variants track displays all BRCA1 missense changes (N = 38) classified as pathogenic by the ClinGen ENIGMA BRCA1/2 VCEP (last consulted August 22, 2024) (scale not preserved). Note that pathogenic missense variants cluster at the RING and BRCT domains, with no pathogenic missense variants reported so far in other regions. The PP3/BP4/BP1 track summarizes ClinGen ENIGMA BRCA1/2 VCEP rules to apply ACMG/AMP predictive evidence to BRCA1 missense changes. Depending on BayesDel-noAF scores, PP3 (≥0.28) or BP4 (≤0.15) computational evidence is applied to missense variants targeting the RING, CC, or BRCT domains. By contrast, PP3/BP4 is not applied to missense variants targeting other regions (mostly, disordered regions). Instead, the BP1_Strong code is applied (regardless of computational predictions). The Domains/Motifs cartoon track represents BRCA1 conserved domains/motifs as defined by the ClinGen ENIGMA BRCA1/2 VCEP, with the RING and BRCT domains defined as clinically important functional domains and the CC (coiled-coil) motif as potentially clinically important (note that the precise boundaries of these domains might be slightly different according to other sources such as UniProt: P38398). The Key partners track shows BRCA1 key interacting proteins BARD1 (interacting with the RING domain), PALB2 (interacting with the CC motif), Abraxas, BRIP1 (also known as BACH1), and CtIP (the latter three interacting with the BRCT domains). The AlphaFold-disorder cartoon track represents BRCA1 disordered regions as deduced from the AlphaFold-2 model AF-P38398-F1 (p.LDDT score <70). The AlphaFold2-models track displays the BRCA1/BARD1 RING heterodimer and BRCT-domain AlphaFold2 models generated for this study. The PDBs track shows ID, descriptive name, and method for experimentally determined 3D structures used in this study.

Figure 2

AlphaMissense, ΔΔG^PDB, ΔΔG^AF, and BayesDel performance at discriminating LoF and FUNC variants at the RING and BRCT domains The figure displays ROC plots and the corresponding auROC value. Overall, AlphaMissense (AM) provides the best discrimination. ΔΔG^AF outperforms ΔΔG^PDB.

Figure 3

PP3/BP4 computational evidence strengths provided by AlphaMissense, ΔΔG^AF, and BayesDel AlphaMissense evidence strengths using ≤0.6 (BP4)//≥0.8 (PP3) (red) or ≤0.65 (BP4)//≥0.75 (PP3) (blue) thresholds (top). ΔΔG^AF evidence strengths using ≤+1 kcal/mol (BP4)//≥+3 kcal/mol (PP3) (red) or ≤+1.5 kcal/mol (BP4)//≥+2.5 kcal/mol (PP3) (blue) thresholds (middle). BayesDel evidence strength with ClinVar ENIGMA BRCA1/2 VCEP recommended thresholds (bottom). Percent of variants falling in the non-informative score range. log₂ LR calculations and plot (including Log2 LR 95% confidential intervals) generated at gwiggins.shinyapps.io/lr_shiny/.

Figure 4

BRIDGES-based breast cancer risk estimates (burden analysis) stratified by computational scores The plot on the right displays breast cancer ORs (and 95% confidential intervals) for variants ≥ (<) the indicated cutoff. The plot on the left shows corresponding distribution of MAVE functional classes. Note that higher ORs correspond to higher proportion of MAVE LoF variants. AM, AlphaMissense; BD, BayesDel; N/A, BRIDGES variants not assessed in MAVE.