A Validated Functional Analysis of Partner and Localizer of BRCA2 Missense Variants for Use in Clinical Variant Interpretation

Abstract

Clinical genetic testing readily detects germline genetic variants. Yet, the rarity of individual variants limits the evidence available for variant classification, leading to many variants of uncertain significance (VUS). VUS cannot guide clinical decisions, complicating counseling and management. In hereditary breast cancer gene PALB2, approximately 50% of clinically identified germline variants are VUS and approximately 90% of VUS are missense. Truncating PALB2 variants have homologous recombination (HR) defects and rely on error-prone nonhomologous end-joining for DNA damage repair (DDR). Recent reports show that some missense PALB2 variants may also be damaging, but most functional studies have lacked benchmarking controls required for sufficient predictive power for clinical use. Here, variant-level DDR capacity in hereditary breast cancer genes was assessed using the Traffic Light Reporter (TLR) to quantify cellular HR/nonhomologous end-joining with fluorescent markers. First, using BRCA2 missense variants of known significance as benchmarks, the TLR distinguished between normal/abnormal HR function. The TLR was then validated for PALB2 and used to test 37 PALB2 variants. Based on the TLR's ability to correctly classify PALB2 validation controls, these functional data where applied in subsequent germline variant interpretations at a moderate level of evidence toward a pathogenic interpretation (PS3_moderate) for 8 variants with abnormal DDR, or a supporting level of evidence toward a benign interpretation (BS3_supporting) for 13 variants with normal DDR.

Figure 1

Traffic Light Reporter assay schematic, flow cytometry output, and final assay protocol. A: The Traffic Light Reporter (TLR) assay, adapted from Certo et al, utilizes a cell line with a stably integrated construct encoding mutGFP [a mutant version of the GFP cDNA that does not produce green fluorescent protein (GFP) at baseline due to an engineered I-SceI nuclease target site] connected via the T2A linker to an out-of-frame mCherry cDNA. Position relative to reading frame is indicated in parentheses, where +1 leads to gene expression, whereas +3 is 2 bp out of reading frame. A DNA double-strand break in mutGFP is induced by transfection with a plasmid encoding infrared protein (IFP) and I-SceI nuclease. A plasmid encoding blue fluorescent protein (BFP) and a truncated green fluorescent protein (GFP) construct (truncGFP) provides a homologous donor sequence that can be used by the cell to repair the DNA double-strand break via the homology-directed repair (HR) pathway (left side). However, if the donor sequence is absent, or if HR is not intact (right side), the cell can undergo mutagenic nonhomologous end-joining (NHEJ) thereby restoring mCherry expression if it results in a 2-bp frameshift (about one-third of events). B: Flow cytometry plots show mean fluorescence intensity of live single cells. Expression of IFP (indicating presence of the I-SceI plasmid) versus BFP (indicating presence of the truncGFP donor plasmid) is shown in the upper panels. Expression of GFP (indicating successful HR) versus mCherry (indicating NHEJ) is shown in the lower panels. Untreated 293T/TLR cells were negative for BFP, IFP, GFP, or mCherry. Cells transfected only with truncGFP donor plasmid were reflected by a BFP⁺ population, otherwise negative for IFP, GFP, or mCherry. Cells transfected with I-SceI nuclease are IFP⁺, and some of the population also expressed mCherry, indicating DSB repair by NHEJ due to absence of the homologous donor template. Cells transfected with all components of the TLR assay exhibited presence of both truncGFP and I-SceI (double-positive BFP/IFP population), with distinct populations positive for mCherry or GFP indicating DSB repair by NHEJ or HR, respectively). C: Optimized timing of the TLR assay used for siRNA knockdown and re-expression of BRCA2 and PALB2 variants in 293T/TLR cells. Stably transfected 293T/TLR cells were reverse transfected with siRNA at time 0, followed by transient cotransfection of rescue plasmid (pDEST-FLAG), I-SceI nuclease (coexpressing IFP) and truncGFP donor construct (coexpressing BFP) 24 hours later. Flow cytometry was performed 72 hours after DNA transfection to detect DDR outcomes.

Figure 2

Validation of the Traffic Light Reporter (TLR) assay using BRCA2 gold standard controls. A: Raw flow cytometry data for %HR and homology-directed repair/nonhomologous end-joining (HR/NHEJ) ratio are shown for BRCA2 wild-type (open blue circles), empty vector (open black squares), and gold standard benign (closed blue circles) and pathogenic (closed black squares) validation controls (normalized against the internal mock-treated control for each plate). B: Batch effects due to plate-to-plate experimental variability were addressed by normalization based on a linear mixed model (LMM) with random effects, utilizing the assay results for wild-type and empty vector controls included in each plate, demonstrating improvement in separation between the different classes of variants. C: TLR assay results for the BRCA2 experiments are summarized, demonstrating HR function after siBRCA2 knockdown and rescue with plasmids containing BRCA2 variants or controls. HR function is rescued by BRCA2 wild-type cDNA and benign validation controls but is impaired with empty vector and pathogenic validation controls. Each rescue condition is depicted across the x axis. The y axis represents the LMM normalized HR/NHEJ ratio. Thresholds for normal and abnormal functional impact are indicated by gray dashed lines. Data are expressed as mean ± SEM. n ≥ 3 independent experiments; n = 21 wild-type and empty vector controls.

Figure 3

Representative PALB2 Traffic Light Reporter (TL)R assay flow cytometry data. Typical flow plots for one run of the TLR assay. Row 1 (left to right): controls for siRNA/DNA transfection and batch effects. Row 2 (left to right): transient rescue with PALB2 variants L35P (potentially pathogenic missense variant), N497Mfs∗64 (known pathogenic), and A712V (known benign). GFP, green fluorescent protein; HR, homology-directed repair; NHEJ, nonhomologous end-joining; siP2, siRNA against PALB2.

Figure 4

Summary of PALB2 TLR assay readout for all variants tested. A: Traffic Light Reporter (TLR) assay results for the PALB2 experiments are summarized, demonstrating homology-directed repair (HR) function after siPALB2 knockdown and rescue with plasmids containing PALB2 variants or controls. HR function was rescued by PALB2 wild-type cDNA and benign validation controls but was impaired with empty vector and pathogenic validation controls. Each rescue condition is depicted across the x axis. The y axis represents the linear mixed model (LMM) normalized homology-directed repair/nonhomologous end-joining (HR/NHEJ) ratio. Thresholds for normal and abnormal functional impact are indicated by gray dashed lines. B: Lollipop diagram of PALB2 protein variants tested in the TLR assay, colored by functional impact. Benign validation controls are shown in blue, pathogenic validation controls are shown in black. Variants of uncertain significance (VUS) are coded by functional impact, based on the thresholds set for normal (yellow) or abnormal (red) assay readout. VUS with intermediate assay readout are labeled in pink. Coiled-coil (CC), chromatin association motif (ChAM), and WD40 protein domains (WD1-7, seven blades of the WD beta propeller) are indicated by colored boxes. Data are expressed as mean ± SEM. n ≥ 3 independent experiments; n = 40 wild-type and empty vector controls.

Figure 5

Traffic Light Reporter (TLR) assay HR/NHEJ values correlate with published homology-directed repair (HR) functional data. Top row (left to right): linear regression of normalized mean HR/nonhomologous end-joining (NHEJ) versus %HR efficiency (Boonen et al²⁴), or HDR fold change (Wiltshire et al²²). Bottom row (left to right): linear regression as above, but using normalized mean %HR instead of HR/NHEJ ratio. Benign/likely benign variant controls are shown as blue circles, variants of uncertain significance (VUS) as magenta triangles, and truncating variants as black squares. n = 15 (Boonen et al²⁴); n = 16 (Wiltshire et al²²).