Site-saturation functional screens identify PALB2 missense variants associated with increased breast cancer risk

Abstract

Loss-of-function variants in PALB2 give rise to defects in DNA damage repair by homologous recombination (HR), increasing the risk of breast cancer in female carriers. However, genetic testing frequently reveals missense variants of uncertain significance (VUS) for which the impact on protein function and cancer risk are unclear. Here we assay 84% of all possible missense variants in 11 out of 13 PALB2 exons using site-saturation functional screens with PARP inhibitor sensitivity as a readout for HR. These exons encode the coiled-coil and WD40 domains, which we identify as the minimal regions required for HR. Furthermore, we reveal the functional impact of 6718 missense variants, classifying 3904 variants as functional (58%), 2422 as intermediate (36%), and 392 as damaging (6%). A burden-type analysis shows that damaging missense variants in PALB2 are associated with a significantly increased risk of breast cancer, similar to that observed for truncating variants. These results will be valuable for the classification of PALB2 missense VUS and clinical management of carriers.

Fig. 1. Functional analysis of PALB2 deletion variants.

a Schematic representation of the PALB2 protein in which amino acid numbers are shown to specify the evolutionarily conserved functional domains of PALB2 (top). PALB2 exon numbers 1-13 are specified by cDNA numbers (bottom). In-frame exons, white; out-of-frame exons, grey. b Western blot showing the expression of five PALB2 deletion variants in Trp53^KO/Palb2^KO mES cells, as compared to wild-type (WT) PALB2 and the empty vector (Ev) control. Tubulin is used as a loading control. c DR-GFP HR assay in Trp53^KO/Palb2^KO mES cells expressing the indicated PALB2 deletion variants from (b). HR efficiencies were normalized to the WT PALB2 condition which was set to 100%. Mean ± s.e.m. are shown, n  =  3 biological replicates, dots represent individual data points, one-way ANOVA with two-sided Dunnet’s multiple comparison test. ***P  < 0.001, NS = not significant. d As in c, except for PARPi sensitivity assays. Values indicate the relative resistance to 0.5 μM PARPi treatment with the WT PALB2 condition set to 100%. ****P < 0.0001. Source data are provided as a Source Data file.

Fig. 2. Site-saturation functional screens of PALB2 variants in the CC domain.

a Schematic flow of the site-saturation functional screens performed in this study. NGS next-generation sequencing, AA = amino acid. b Amino acid function map of the CC domain of PALB2 spanning 35 amino acid residues from p.L9 to p.K43 (top). The map shows depletion scores for 638 PALB2 missense and nonsense variants as generated by Enrich2, n = 6 biological replicates. Amino acid characteristics for all variants are indicated (left). Dark red squares represent variants that were depleted in PARPi-treated conditions versus untreated conditions, n = 3 technical replicates. Blue squares represent variants that were enriched. Grey squares represent variants that were either missing in the library or filtered out during the analysis. Grey dots represent the original (wild-type) amino acids. Data for all variants were normalized to WT PALB2, which was set to ‘0’, and to the average of the nonsense variants, which was set to ‘−1’. Source data are provided as a Source Data file.

Fig. 3. Site-saturation functional screens of PALB2 variants in the WD40 domain.

Amino acid function maps that comprise the entire WD40 domain of PALB2 spanning 331 amino acid residues from p.Q856 to p.S1186 (top of each map). Each map corresponds to a distinct WD40 variant library (left and Supplementary Fig. 1). The maps show depletion scores for 6419 PALB2 missense and nonsense variants as generated by Enrich2, n = 3 biological replicates. Amino acid characteristics for all variants are indicated (left of each map). Dark red squares represent variants that were depleted in PARPi-treated conditions versus untreated conditions, n = 3 technical replicates. Blue squares represent variants that were enriched. Grey squares represent variants that were either missing in the libraries or filtered out during the analysis. Grey dots represent the original (wild-type) amino acids. Data for all variants were normalized to WT PALB2, which was set to ‘0’, and to the average of the nonsense variants per library, which was set to ‘−1’. Source data are provided as a Source Data file.

Fig. 4. Classification and validation of PALB2 missense variants from site-saturation functional screens.

a Histogram showing the distribution of 7106 depletion scores of nonsense, synonymous and missense PALB2 variants from Figs. 2b and 3. Dashed lines indicate functional classification thresholds determined by mixture modelling. Dashed line on the right separates functional (right) from intermediate (middle) variants at a depletion score of −0.0759. Dashed line on the left separates intermediate (middle) from damaging (left) variants at a depletion score of −0.7037. b ROC curve showing 98% sensitivity and specificity for classification of variants from ‘a’ was generated by fitting a two-component mixture model to depletion scores of 339 nonsense and 49 synonymous variants. c Histogram showing the distribution of 6,718 depletion scores of PALB2 CC and WD40 missense variants. Dashed lines indicate functional classification thresholds as in b. Number and fraction (%) of missense variants classified as functional (green), intermediate (orange) or damaging (red) are indicated. d Correlation analysis between outcomes of PARPi sensitivity assays and depletion scores from site-saturation functional screens of missense variants in PALB2. n = 140 PALB2 variants (43 CC and 97 WD40 variants). PARPi assays were performed twice with similar results (Supplementary Data 1 and Source Data Fig. 4d). Depletion scores are from Figs. 2b and 3. Dashed lines indicate functional thresholds as in (a). r = Pearson correlation coefficient. Two-sided P < 0.0001. e As in d, except for DR-GFP assays. f SGE was used to introduce SNVs across a region of exon 10 of PALB2 encoding p.G1000 to p.I1037 of the WD40 domain. A gRNA/Cas9 construct was transfected with a plasmid library containing SNVs within ~100 bp of genomic sequence, homology arms, and synonymous variants within the CRISPR target site to prevent re-cutting. Cells were collected 9 days after transfection and targeted sequencing was performed to quantify SNV abundances and calculate SGE scores. g Correlation analysis between SGE scores and depletion scores from site-saturation functional screens of 179 PALB2 variants in the WD40 region spanning p.G1000 to p.I1037 (Supplementary Data 1). SGE experiments were performed twice with similar results (Supplementary Data 1 and Source Data). Depletion scores are from Fig. 3. Dashed lines indicate the functional classification thresholds as in (b). r = Pearson correlation coefficient. Two-sided P < 0.0001. Source data are provided as a Source Data file.

Fig. 5. Molecular validation of the impact of PALB2 missense variants in site-saturation functional screens.

a YPF/GFP pulldowns of the indicated PALB2 CC variant proteins following transient expression in U2OS cells. PALB2 CC variants are indicated in three colours reflecting their functional outcome in the site-saturation screens in ‘Fig. 2b’; green is functional, orange is intermediate, red is damaging. GFP-NLS and YFP-PALB2-L35P served as negative controls. Western blot analysis was performed using antibodies against GFP and BRCA1. Representative blots of two independent experiments with similar results are shown. b Western blot analysis of the expression of human PALB2 WD40 variants in Trp53^KO/Palb2^KO mES cells using an antibody directed against PALB2. The empty vector (Ev) served as a negative control on each blot. Tubulin was used as a loading control. *indicates a non-specific band. PALB2 variants are indicated in colour as in (a). Expression levels indicated below the blots are normalized to Tubulin and relative to the normalized WT PALB2 expression level, which was set to 1. Blots of one experiment are shown. c Fluorescence microscopy analysis (left) and quantification of the cellular distribution (right) of human YFP-PALB2 WD40 variants following transient expression in U2OS cells. Coloring of variants is as indicated in (a). Data represent the mean ±s.e.m. n = 3 biological replicates, dots represent individual data points, two-way ANOVA with two-sided Dunnet’s multiple comparison test. *P < 0.05, **P  <  0.01, ***P  <  0.001, NS = not significant. Source data are provided as a Source Data file.

Fig. 6. Classification of PALB2 missense variants from ClinVar by impact in site-saturation functional screens.

a Radial bar chart of PALB2 SNVs reported in ClinVar (as of June, 2025) (left), pie chart showing the distribution PALB2 missense VUS from ClinVar which were examined in site-saturation functional screens and are located in the CC and WD40 domains. Variants in the pre-CC (start codon to first codon of the CC domain) and middle regions were not examined in these screens. Histogram showing the frequency distribution of depletion scores of PALB2 CC and WD40 VUS from ClinVar (right). Dashed lines indicate thresholds from ‘Fig. 4a’, which were used to classify PALB2 CC and WD40 VUS from ClinVar as either functional (green), intermediate (orange) or damaging (red). b Schematic representation of the PALB2 protein in which amino acid numbers are shown to specify the evolutionarily conserved functional domains of PALB2 (bottom). PALB2 variants are color-coded as indicated. c DR-GFP HR assay in Trp53^KO/Palb2^KO mES cells expressing the indicated PALB2 variants from (b). HR efficiencies were normalized to the WT PALB2 condition which was set to 100%. Ev is the empty vector control. d PARPi sensitivity assay using Trp53^KO/Palb2^KO mES cells expressing the indicated PALB2 variants variants from (b) and (c). Values indicate the relative resistance to 0.5 μM PARPi treatment with the WT PALB2 condition set to 100%. In c and d, mean is shown, n  =  2 biological replicates, dots represent individual data points. Source data are provided as a Source Data file.