Clinical consequences of BRCA2 hypomorphism

The tumor suppressor FANCD1/BRCA2 is crucial for DNA homologous recombination repair (HRR). BRCA2 biallelic pathogenic variants result in a severe form of Fanconi anemia (FA) syndrome, whereas monoallelic pathogenic variants cause mainly hereditary breast and ovarian cancer predisposition. For decades, the co-occurrence in trans with a clearly pathogenic variant led to assume that the other allele was benign. However, here we show a patient with biallelic BRCA2 (c.1813dup and c.7796 A > G) diagnosed at age 33 with FA after a hypertoxic reaction to chemotherapy during breast cancer treatment. After DNA damage, patient cells displayed intermediate chromosome fragility, reduced survival, cell cycle defects, and significantly decreased RAD51 foci formation. With a newly developed cell-based flow cytometric assay, we measured single BRCA2 allele contributions to HRR, and found that expression of the missense allele in a BRCA2 KO cellular background partially recovered HRR activity. Our data suggest that a hypomorphic BRCA2 allele retaining 37–54% of normal HRR function can prevent FA clinical phenotype, but not the early onset of breast cancer and severe hypersensitivity to chemotherapy.


INTRODUCTION
Fanconi anemia (FA) is a rare genetic DNA repair syndrome characterized by bone marrow failure (BMF) and cancer predisposition particularly to acute myeloid leukemia (AML) and/or solid tumors, mainly head and neck squamous cell carcinoma (HNSCC) 1 . FA cells exhibit chromosomal instability and hypersensitivity to DNA damage especially to DNA interstrand crosslinks (ICLs) 2 . Twenty-two FA-related genes have been identified including tumor suppressors BRCA1 (FANCS) and BRCA2 (FANCD1).
The breast cancer susceptibility 2 (BRCA2) protein is a key component of the DNA homologous recombination repair (HRR) pathway 3 and a downstream component of the FA/BRCA DNA repair pathway 1,2 . Monoallelic germline pathogenic variants in BRCA2 are a common cause of hereditary breast and ovarian cancer (HBOC) 4 , while biallelic pathogenic variants are associated with FA 5 . Individuals with biallelic BRCA2 pathogenic variants usually suffer from a severe form of FA characterized by BMF, multiple congenital defects, and severe cancer predisposition as patients develop leukemia and/or multiple solid tumors (mainly Wilms and brain tumors) in the first decade of life. The overall risk of cancer in these patients is about 97% by the age of 5 years [5][6][7][8][9] .
Here, we describe an unusual clinical case consisting of a female patient who reached adulthood with no clear suspicion of a defined pathology, despite presenting two germline BRCA2 variants: a null frameshift variant c.1813dup, p.(Ile605Asnfs*11) and a missense variant c.7796 A > G, p.(Glu2599Gly). Using a BRCA2 KO HEK293T cell line stably transfected with an HRR GFP marker construct 10 we demonstrated that the missense allele is partially functional providing 37-54% of HRR activity. We propose that p.(Glu2599Gly) is a hypomorphic allele that provides sufficient HRR activity to prevent FA onset, but it is insufficient to avoid tumorigenesis during adulthood and results in acute sensitivity to chemotherapy.

RESULTS
Biallelic BRCA2-FA disorder diagnosis Patient FA663 was a thirty-three-year-old woman who developed a malignant breast nodule and, after surgery, showed a severe intolerance to chemotherapy. A timeline of patient's progression is shown in Fig. 1A. FA663 was an adopted child and her medical history only indicated slight abnormalities such as low birth weight, short stature, a subtle learning disability, skin hyperpigmentation, macrocytosis, hiatal hernia, hepatic steatosis, and lumbar disc herniation. Due to the severe pancytopenia reaction to chemotherapy, a chromosome fragility test (CFT) 11 was performed in blood T-lymphocytes and the levels of chromosome fragility were compatible with a diagnosis of FA with somatic mosaicism (FAmosaic) as only one-third of her cells presented multiple chromatid-type breaks and radial figures. (Fig. 1B) 11,12 , FA663 case was then included in a cohort of FA patients to be molecularly characterized by Exome Sequencing (ES) 13 , and two variants were identified in the BRCA2/FANCD1 gene. The first, c.1813dup, p. (Ile605Asnfs*11) was a clearly pathogenic variant in BRCA2 exon 10 that produced a frameshift and premature termination of translation (Fig. 1C). This variant had previously been described in FA patients 8 14 and it is reported in ClinVar as pathogenic class 5 classified by the ENIGMA expert panel (https://www.ncbi.nlm.nih.gov/clinvar/variation/37762/). A second missense variant c.7796 A > G, p.(Glu2599Gly) was identified in BRCA2 exon 16 (Fig. 1C) and classified as of "uncertain significance" according to the ACMG guidelines 15 . In ClinVar three submitters also interpreted this variant as of uncertain significance (https:// www.ncbi.nlm.nih.gov/clinvar/variation/483130/). Human Glu2599 is a highly conserved residue among organisms in the BRCA2 DNA binding domain (Fig. 1D) and the detected variant produced a non-conservative change from a negatively charged amino acid (Glu) to a non-polar amino acid (Gly). The p.(Glu2599Gly) is predicted deleterious for BRASS NN (score 0.912 pathogenic), BRASS MLR (score 1.675 pathogenic), AGVGD (Class C65), PolyPhen-2 (score 1 probably damaging) and SIFT (score 0.01 predict non-tolerated) in silico tools. No other potentially pathogenic variants were identified in any other known FA gene.

and in HBOC patients
Patient fibroblasts retain partial DNA damage response To confirm the diagnosis of FA663 as a FA mosaic 11,16 , and to study more in-depth patient's DNA damage response (DDR), FA663 primary fibroblasts were analyzed together with fibroblasts from a wt control and from a patient with a homozygous BRCA2 null variant c.469 A > T; p.(Lys157*). In the CFT, FA663 fibroblasts showed a chromosome fragility level in-between the wt and BRCA2 deficient control and similar to that of blood T-lymphocytes (≈32% aberrant cells; ≈2.88 breaks per aberrant cell) ( Fig. 2A). Similarly, intermediate results were also obtained in ICL induced G 2 /M phase cell cycle arrest (Fig. 2B), RAD51 foci formation (Fig. 2C), and sensitivity to the PARP inhibitors olaparib and veliparib (Fig. 2D, E).
Evidencing the in trans position of BRCA2 c.1813dup and c.7796 A > G variants To assess if c.1813dup and c.7796 A > G variants were in trans, multiplex ligation-dependent probe assay (MLPA) and Sanger sequencing of the BRCA2 gene were undertaken in normal tissue samples and tumor samples from FA663 ( Table 1). The two allelic variants of BRCA2 were found in blood, skin tissue, and in three primary tumor samples: invasive ductal carcinoma (IDC), contralateral ductal carcinoma in situ (DCIS), and colorectal cancer (CRC) identified during a necropsy study (Table 1, Fig. 3A-D and Figure  S1A). However, four metastatic breast cancer samples (hepatic metastasis and supraclavicular, pleural, hilar, and lymph nodes) infiltrated by the primary breast IDC tumor, and two patientderived tumor xenograft (PDX) models (derived from supraclavicular and hilar metastases) showed loss of heterozygosity (LOH) in BRCA2 (Table 1, Fig. 3E-F and Figure S1B-E). Specifically, all the infiltrated nodes and PDXs samples presented exclusively the c.7796 G allele, suggesting that the two variants were located in trans (Table 1). In addition, the cDNA obtained from blood RNA and cultured fibroblasts was sequenced by Sanger ( Figure S2). We observed a lower signal of the allele carrying the duplication in exon 10 (as a result of degradation by the nonsense-mediated mRNA decay (NMD) pathway) and a higher signal of the G allele of the missense variant c.7796 A > G in exon 16 with respect to that of the A wt allele, which is consistent with the two variants being located in different alleles. Overall, all these results support the position in trans of the two variants. The hypomorphic missense p.(Glu2599Gly) variant displays 37-54% of HRR activity Next, we generated a cellular system to quantify mutation-specific levels of HRR by genetic complementation of a human-derived BRCA2 deficient cell line, stably transfected with a GFP-based construct to detect HRR (BRCA2 KO HEK293T DR-GFP) 10 . The BRCA2 KO cell line was generated in our laboratory by CRISPR/Cas9 technology and it was functionally validated ( Figure S3). Plasmids containing wt BRCA2 cDNA, empty vector (EV), and BRCA2 cDNA bearing either p.(Ile605Asnfs*11) or p.(Glu2599Gly) were transiently transfected in the BRCA2 KO HEK293T-DR-GFP. No expression of p.(Ile605Asnfs*11) BRCA2 protein could be observed by The representative experiment shown, from three independent data sets. c Attenuated RAD51 foci formation in FA663 after IR. Immunofluorescence analysis of wt, BRCA2-mutant, and FA663 primary cell lines 6 h after 5 Gy. The graph represents the mean of three independent experiments with duplicates. Error bars show s.e.m. Statistics are done with the One-Way ANOVA test and Bonferroni correction **P < 0.01, ***P < 0.001. d, e PARPi sensitivity of FA663 primary cells. Data represent the mean of three independent experiments. Error bars represent s.d. Statistics with two-tailed t-Test *P < 0.05, ***P < 0.001, comparing BRCA2-mutant cell line. (wt: human primary skin fibroblasts from a healthy individual; BRCA2: primary skin fibroblasts from a homozygous c.469 A > T, p.Lys157* FA patient; FA663: primary skin fibroblasts established from skin biopsy of patient FA663).
Western blot (WB) analysis, while the missense BRCA2 p. (Glu2599Gly) was expressed at comparable levels with the endogenous wt BRCA2 ( Fig. 4A and Figure S4). BRCA2 mRNA expression was detected in PDX332 (derived from the supraclavicular node) to a similar level of BRCA2-wt PDXs, further evidencing the expression of the missense BRCA2 p.(Glu2599Gly) variant ( Figure S5). The HRR capacity of the two BRCA2 variants was further quantified, using both an ionizing radiation-induced RAD51 foci formation (IR-IF) assay (Fig. 4B) and the DR-GFP reporter test by flow cytometry (Fig. 4C and Figure S6). With both techniques, cells expressing the missense allele had intermediate levels of HRR compared to the BRCA2 wt reconstituted cells that exhibited functional HRR, or the cells expressing the p. (Ile605Asnfs*11) that were HRR deficient (Fig. 4B, C). We established that the residual HRR function in the missense variant p.(Glu2599Gly) was 54%± 11 (s.e.m) (Fig. 4B) and of 37% ± 7 (s.e. m) (Fig. 4C). In agreement with these findings, we observed that the level of baseline RAD51 foci in tumor samples from the metastatic supraclavicular node and the two PDXs were intermediate and significantly different to BRCA2 wt and mutant breast tumors ( Figure S7).

DISCUSSION
The role of BRCA2 in DDR is relevant not only for tumor suppression, but also for organisms' viability. Preceding studies in mice demonstrated that Brca2 is essential for normal embryonic development 17 and indirect evidence indicates that most of the time this is also true for humans. Pathogenic variants such as c.5946delT (also known as 6174delT) with a carrier frequency of around 1-1.5% in Ashkenazi Jewish population have not been ever detected in the homozygous state as well as many common BRCA2 pathogenic variants that are located in exon 11 9 . Nevertheless, it has been shown that biallelic deleterious BRCA2 variants in humans are compatible with life, despite being associated to severe forms of FA and childhood cancer predisposition [6][7][8][9] . In line with these formerly reported cases 6 , FA patients with biallelic BRCA2 pathogenic variants previously identified in our laboratory presented elevated chromosomal fragility, severe congenital aberrations and/or high cancer incidence by age 3 or younger (Table S1). On the contrary, patient FA663 had a tumor-free childhood and mild Fanconi Anemia-associated clinical features, which remained practically unnoticed to clinicians. Our work focuses on the detailed molecular characterization of a patient being a carrier of a pathogenic variant and a hypomorphic variant in BRCA2 that survived to adulthood with mild Fanconi Anemiaassociated clinical features, but nonetheless developed youngonset breast cancer and had fatal evolution. Our functional complementation studies showed that p.(Glu2599Gly) missense allele retains between 37 and 54% of normal HRR activity levels, likely sufficient to suppress BMF, leukemia, and solid tumors during childhood, but not breast or colon cancer at adult age. Our accurate molecular screening of DNA from normal and tumor tissues demonstrated that the c.1813dup and c.7796 A > G variants were positioned in trans. The predicted frameshift caused by the FA663 variant c.1813dup produces a premature stop codon, which implies a degradation of the resulting transcript by NMD 18 . Our results of Sanger sequencing of blood and fibroblasts cDNA show a low signal of one allele supporting the NMD hypothesis ( Figure S2).
Most of the BRCA-associated tumors undergo the loss of the BRCA-wt allele during tumor evolution, which turns the tumor into an HRR deficient (HRD) state and results in tumor sensitivity to DNA damaging chemotherapy. Intriguingly, while the primary tumors from this patient did not undergo LOH (  Figure S1B-E). This could be an evolutionary advantage acquired by the metastatic tumors, rendering them capable of partly repairing the DNA damage induced by chemotherapy and resulting in chemoresistance. This hypothesis underscores an active role of the missense variant in the tumorigenic phenotype.
Remarkably, at least six FA cases of biallelic BRCA2 patients due to the combination of one clearly deleterious allele and one missense allele have been previously reported 6 . Three of the variants, p.(Lys2729Asn), p.(Asn372His), and p.(Ile2490Thr) showed functional HRR 19,20 and they appear classified in ClinVar as "Benign" (by expert review), implying that were probably not the second causal variants in these FA-D1 patient 6 . The variants, p. (Leu2510Pro) and p.(Trp2626Cys) were able to partially rescue the lethality of Brca2-null (Brca2 KO/KO ) ES cells, but with a 70-80% reduction in the number of colonies and an HRR severely impaired 19,20 . Cells expressing these variants had residual HRR activity, below 5%, and a 20-fold increase in ICLs induced chromosome fragility when compared to the wt BRCA2 expressing cell lines 19,20 . Similarly, the p.(Arg2336His) variant, due to the c.7007 G > A substitution in DNA, also behaves as a BRCA2 null allele when expressed in ES cells 19,20 . The c.7007 G > A change is a splicing alteration that generates two frameshift mRNA transcripts (skipping exon 13 and exons 12-13, respectively), with the appearance of an early stop codon in exon 14 19,20 . Since functional studies demonstrated a deep impairment of HRR 20 , these last three variants cannot be considered hypomorphic, in agreement with the acute FA phenotype and early cancer predisposition shown by these patients 6 . In contrast, p. (Glu2599Gly) found in our patient exhibited substantial levels of HRR in all functional assays performed in our study but well below the wt levels (Fig. 2, Fig. 4 and Figure S5), in accordance to recent data 21 . Moreover, we observed the same intermediate chromosome fragility in both primary lymphocytes and fibroblasts from FA663 patients. This is indicative of a hypomorphic effect of the p. (Glu2599Gly) variant and not of FA somatic mosaicism. Indeed, a diagnosis of FA somatic mosaicism is usually consistent with the presence of higher induced chromosome fragility in patient's fibroblasts than patient's lymphocytes 12,22,23 . The p.(Glu2599Gly) variant presented by our patient is located in the helical domain (2479 to 2668aa) of the DNA binding region of BRCA2. This domain interacts with DSS1 24 , which promotes the exchange of BRCA2-RPA1 to BRCA2-RAD51 and thus the ssDNA interaction 25 . The BRCA2 DNA binding domain is also involved in the stabilization of BRCA2 and in the regulation of R-loopassociated DNA damage, a particular source of damage resolved by FA proteins 26,27 . In addition, it has been shown that hypomorphic missense variants in the DNA binding domain of BRCA2 with partial effects on protein function may confer moderate risks of breast cancer 28 , being the hypomorphic variants associated with this domain the only missense variants in BRCA2 increasing cancer risk. Although we show that BRCA2 p. (Glu2599Gly) is hypomorphic in terms of DNA repair function, this study cannot provide a quantification of its estimated reduced cancer penetrance, due to the fact that it is presented together with another clearly deleterious variant and the lack of a significant number of patients harboring this missense variant.
Our findings together with two family study cases that reported biallelic BRCA2 patients with a less severe FA phenotype and early onset breast and/or colon cancer susceptibility 29,30 are relevant for the interpretation of variants with unknown significance (VUS) in BRCA2. In the evaluation of the potential pathogenicity of a variant in BRCA2, it is considered that its co-occurrence in trans with a clearly pathogenic variant indicates its benign character 31 . However, our results show that this criterion cannot be applied without careful exploration of the clinical phenotype.
We conclude that the BRCA2 p.(Glu2599Gly) is a truly hypomorphic BRCA2 allele that retains sufficient HRR activity for normal organism development and to minimize FA-related phenotypes, but not enough to avoid early onset of tumorigenesis and severe chemotoxicity. Furthermore, we propose a straightforward methodology that produces conclusive results in a reasonable turnaround time to classify BRCA2 variants of uncertain significance (VUS) in cancer patients.  Figure S1. *indicates the variant location.

Chromosome fragility assays and G2/M cell cycle arrest
For the Chromosome fragility test, peripheral blood T-lymphocytes and primary fibroblasts were treated with diepoxybutane (DEB) at the indicated concentrations and processed as in 11,32 . For the G2/M cell cycle arrest analysis, primary fibroblasts were treated with Mitomycin-C at the indicated concentrations and processed as in 32 . 300000 HEK293T cells were seeded on coverslips (previously coated with Poly-L-Lysine) in 6 cm Petri dishes and the same procedure was followed.

Plasmid transfection
Transfections were carried out using 10 μL Lipofectamine 2000 (Life Technologies) and 250 μL Optimem (Life Technologies) plus 10 μg of each pCDNA3 plasmid construct the day after plating 600000 HEK293T cells/well in 6-well plates with complete medium.

HRR assay
HRR assay was performed as in 36 with some modifications: 200000 cells HEK293T expressing pDR-GFP construct in a stable manner were seeded in 12-well plates. pCDNA3 plasmids were co-expressed with an I-SceI expressing plasmid by transient transfection. HR-dependent repair of I-SceI induced DNA double-strand breaks was quantified by fluorescenceactivated cell sorting (FACS) of GFP + cells after 48 h.

Generation of PDX models
Biopsy specimens from metastatic lesions were immediately implanted into the lower flank of 5-week-old female NOD SCID GAMMA mice (NSG, Charles River) and then expanded in NMRI-Foxn1nu/nu mice (NMRI, Janvier Labs). This procedure was conducted following the European Union's animal care directive (2010/63/EU) and was approved by the Ethical Committee of Animal Experimentation of the Vall d'Hebron Research Institute.

BRCA2 mRNA expression
RNA was extracted from PDX samples (15-30 mg) by using the PerfectPure RNA Tissue kit (five Prime). cDNA was obtained using the PrimeScript RT Reagent kit (Takara). Quantitative RT-qPCR was performed in a 7900HT Fast Real-Time PCR System (Applied Biosystems) using TaqMan Universal Master Mix II (Applied Biosystems) and predesigned human-specific primers and TaqMan probes (Hs99999908 for GUSB, Hs99999903 for ACTB, and Hs01037414 for BRCA2). The comparative CT method was used for data analysis, in which geNorm algorithms were applied to select the most stably expressed housekeeping genes (GUSB and ACTB) and geometric means were calculated to obtain normalized CT values 37 .
RNA extraction, reverse transcription and cDNA amplification RNA was isolated from patient's peripheral blood samples and cultured fibroblasts using Trizol Isolation Reagent (ThermoFisher Scientific). RNA was cleaned up using Rneasy Minikit (QIAGEN) and digested with RNase-Free DNase Set (QIAGEN) to remove genomic DNA. A total of 500 ng were reverse transcribed with PrimeScript RT reagent kit (Takara) that supplies random and oligo-dT primers. BRCA2 cDNA was amplified in two PCR fragments spanning exons 10 to 14 (forward 5'-tatgtccaaatttaattgataat-3' and reverse 5'-ttggtctgcctgtagtaatc-3') and 11 to 18 (forward 5'ttctgatgttcctgtgaaaacaa-3' and reverse 5'-attctggggcttcaagaggt-3'), respectively, using Expand™ Long Template PCR System (Roche). Sequencing was performed with BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems) in a Genetic Analyzer ABI 3130xl (Applied Biosystems). Sanger electropherograms were visualized using SeqScape v2.6 and Sequencing Analysis v2.6 softwares (Applied Biosystems). DNA copy number analysis by multiplex ligation-dependent probe assay (MLPA) Allelic copy number changes in BRCA2 were performed by MLPA kit #P090 (MRC-Holland), and resulting electrophoresis on an ABI 3130xl capillary sequencer (Applied Biosystems). Analysis was done using Coffalyser.Net software (MRC-Holland, Amsterdam, The Netherlands). Blood samples from healthy donors were included for sample normalization. Animal experiments were conducted following the European Union's animal care directive (2010/63/EU) and the protocols were approved by the Ethical Committee of Animal Experimentation of the Vall d'Hebron Research Institute and the appropriate governmental agency and carried out in accordance with the approved guidelines.