Article Text

Download PDFPDF

Original research
Homozygous hypomorphic BRCA2 variant in primary ovarian insufficiency without cancer or Fanconi anaemia trait
Free
  1. Sandrine Caburet1,
  2. Abdelkader Heddar2,
  3. Elodie Dardillac3,
  4. Héléne Creux4,
  5. Marie Lambert4,
  6. Sébastien Messiaen5,
  7. Sophie Tourpin5,
  8. Gabriel Livera5,
  9. Bernard S Lopez3,
  10. Micheline Misrahi2
  1. 1 Institut Jacques Monod, Université de Paris, Paris, Île-de-France, France
  2. 2 Faculte de Medecine, Universite Paris Saclay, Hopital Bicêtre APHP, Le Kremlin-Bicetre, France
  3. 3 Institut Cochin, INSERM U1016, UMR 8104 CNRS, Université de Paris, Paris, Île-de-France, France
  4. 4 Service de Gynécologie et Médecine de la Reproduction, CHU de Bordeaux, Bordeaux, Aquitaine, France
  5. 5 UMR Stabilité Génétique, Cellules Souches et Radiations, Université Paris-Saclay, Fontenay aux Roses, Île-de-France, France
  1. Correspondence to Professor Micheline Misrahi, Université Paris Saclay Faculté de Medecine, Hôpital Bicêtre APHP, Le Kremlin-Bicetre, France; micheline.misrahi{at}aphp.fr; Dr Bernard S Lopez; bernard.lopez{at}inserm.fr

Abstract

Background Primary ovarian insufficiency (POI) affects 1% of women under 40 years and is a public health problem. The genetic causes of POI are highly heterogeneous with isolated or syndromic forms. Recently, variants in genes involved in DNA repair have been shown to cause POI. Notably, syndromic POI with Fanconi anaemia (FA) traits related to biallelic BRCA2 truncated variants has been reported. Here, we report a novel phenotype of isolated POI with a BRCA2 variant in a consanguineous Turkish family.

Methods Exome sequencing (ES) was performed in the patient. We also performed functional studies, including a homologous recombination (HR) test, cell proliferation, radiation-induced RAD51 foci formation assays and chromosome breakage studies in primary and lymphoblastoid immortalised cells. The expression of BRCA2 in human foetal ovaries was studied.

Results ES identified a homozygous missense c.8524C>T/p.R2842C-BRCA2 variant. BRCA2 defects induce cancer predisposition and FA. Remarkably, neither the patient nor her family exhibited somatic pathologies. The patient’s cells showed intermediate levels of chromosomal breaks, cell proliferation and radiation-induced RAD51 foci formation compared with controls and FA cells. R2842C-BRCA2 only partially complemented HR efficiency compared with wild type-BRCA2. BRCA2 is expressed in human foetal ovaries in pachytene stage oocytes, when meiotic HR occurs.

Conclusion We describe the functional assessment of a homozygous hypomorphic BRCA2 variant in a patient with POI without cancer or FA trait. Our findings extend the phenotype of BRCA2 biallelic alterations to fully isolated POI. This study has a major impact on the management and genetic counselling of patients with POI.

  • primary ovarian insufficiency
  • BRCA2 exome
  • cancer Fanconi anemia
  • meiosis

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Primary ovarian insufficiency (POI) is a public health issue affecting ~1% of women under 40 years and is clinically heterogeneous with isolated or syndromic forms.1 Most cases are idiopathic, but an increasing number of genetic causes has been recently identified, especially variants in genes involved in DNA repair and recombination.2–4

DNA repair and recombination are essential for genome stability maintenance. Consistently, defects in the pathways ensuring faithful transmission of genetic material result in genome instability associated with developmental abnormalities and cancer predisposition.5 Homologous recombination (HR), an evolutionary conserved process essential to genome stability and cell viability, plays crucial roles in DNA double-strand break (DSB) repair in somatic and meiotic cells.

In mammals, BRCA2 binds to damaged DNA and loads the pivotal HR RAD51 recombinase, which then promotes DNA homology search. Therefore, cells defective in BRCA2 or RAD51 are defective in mitotic HR.6 7 The role of BRCA2 in RAD51 loading in mitotic HR makes it a strong candidate for an involvement in meiotic HR, but this remains to be formally established. Indeed, the severe phenotypes of biallelic inactivation of BRCA2 in human and the early embryonic lethality resulting from germ-line inactivation of this essential gene in animal models hampered meiosis analysis and compromised the study of the putative functions of BRCA2 in gametogenesis.8–10 However, these data underline the strong and pleiotropic consequences of BRCA2 invalidation.

Heterozygous BRCA2 or PALB2 (FANCN) variants increase susceptibility to breast and ovarian cancers, whereas severe biallelic defects in RAD51 (FANCR), PALB2 or BRCA2 (FANCD1) lead to Fanconi anaemia (FA) syndrome.11 12 In particular, FANCD1 syndrome associates developmental defects, genetic instability, bone marrow failure and cancer predisposition, with cancer developing in the first decade of life, and death before puberty.13 Recently, older patients with XX ovarian dysgenesis were reported to carry compound heterozygous BRCA2 pathogenic or likely pathogenic variants, but the associated clinical phenotypes in the patients and/or relatives, ranging from cancer to the characteristic severe response to mitomycin C (MMC) in chromosomal breakage tests, argue in favour of mild forms of BRCA2-associated FA.14–16

In contrast to those patients, we describe here an adult patient carrying a homozygous missense variant of BRCA2, presenting with only isolated POI, without cancer or signs of FA in the patient or her family. We demonstrate that the mutated R2842C-BRCA2 retains a lower, but significant, residual function when compared with wild-type (WT)-BRCA2. Consistently, the patient’s cells exhibit intermediate levels in chromosomal breaks, cell proliferation and ionising radiation-induced RAD51 foci formation when compared with controls, a patient with FANCD1’s or the heterozygous mother’s cells. This residual HR in somatic cells could explain the absence of developmental defects. Our data extend the phenotype of biallelic variants of BRCA2 to isolated POI. As BRCA2 is a major cancer susceptibility gene, our findings will have a strong impact on the genetic counselling and management of patients with POI and their relatives.

Material and methods

Exome sequencing (ES) and bioinformatics analysis

Genomic DNA was extracted from blood samples by standard protocols. ES, reads quality check and mapping were performed by Beckman Coulter Genomics (Danvers, USA). Exon capture was performed using the hsV5UTR target enrichment kit. Mapping was performed on the GRCh37.p13 reference genome using the Burrows-Wheeler Alignment tool V.0.6.1-r104 with default parameters, and samtools 'Rmdup' was used to remove duplicates. Prior to variant calling, reads were realigned around known or suspected indels by the GATK Realigner commands. The samtools V.2.0 'mpileup' command and the bcftools multiallelic calling model were used for variant calling. Variants were annotated by SnpEff, Variant Effect Predictor and dbNSFP 3.5a. Minor allele frequencies were manually verified using ExAC (http://exac.broadinstitute.org/), Gnomad (https://gnomad.broadinstitute.org/) and Kaviar (http://db.systemsbiology.net/kaviar/) databases.

Chromosome breakage studies

Chromosome breakage studies were performed in Epstein-Barr Virus (EBV)-immortalised lymphoblastoid cell lines derived from the patient, her mother, an FA patient (FANCD1) and a healthy woman as control. They were studied at the Gustave Roussy Institute (Villejuif, France), following a standard in-house protocol. EBV-immortalised cells were cultured under standard conditions for karyotyping. To test for hypersensitivity to DNA cross-linking agents, DNA damage was induced using MMC (Sigma) added for 48 hours. For each sample, three conditions were tested: without MMC to analyse spontaneous damages, and with MMC of 300 and 1000 nM. Chromosome breakages were scored by an experienced cytogeneticist on at least 20 metaphases.

Cell transfection and HR efficiency test

The HR efficiency was assessed in the RG37 cell line, derived from SV40-transformed GM639 human fibroblasts in which we stably integrated the pDR-GFP gene conversion reporter.17 18 RG37 was cultured in DMEM supplemented with 10% foetal calf serum and 2 mM glutamine and was incubated at 37°C with 5% CO2. For the HR efficiency test, the I-SceI meganuclease was expressed by transient transfection of the pCMV-HA-I-SceI expression plasmid19 with Jet-PEI according to the manufacturer’s instructions (Polyplus Transfection), and cells were incubated for 48 hours. Cells were collected in PBS and 50 mM EDTA, pelleted and fixed with 2% paraformaldehyde for 20 min. The percentage of GFP-expressing cells was scored by FACS analysis using a BD Accuri C6 flow cytometer (BD Biosciences).

For silencing experiments, 20 000 cells were seeded 1 day before transfection with siRNAs using INTERFERin, following the manufacturer’s instructions (Polyplus Transfection) with 20 nM of one of the following siRNAs: control (5′-AUGAACGUGAAUUGCUCAA-3′) and BRCA2-3 (5′-GCUUCAGUUGCAUAUCUUA-3′). The BRCA2 siRNA targets the 3′UTR of endogenous BRCA2 messenger RNA (mRNA). All siRNAs were synthesised by Eurofins (France). Forty-eight hours later, the cells were transfected with the pCMV-HA-I-SceI expression plasmid. At least three independent experiments were performed, and HA-I-SceI expression and silencing efficiency were verified by western blot as described further.

Statistical analysis

Statistical analyses were performed using GraphPad Prism V.3.0 (GraphPad Software).

Additional material and methods and additional results are provided in the online supplementary appendix.

Results

Case report

The proposita was born to consanguineous Turkish parents (figure 1A). At the age of 13 years, she had two vaginal bleeds followed by primo-secondary amenorrhoea. She had normal pilosity, breast development and external genitalia. Several hormonal assays confirmed POI, and pelvic ultrasonographical studies showed small ovaries without or with very few follicles (table 1).

Table 1

Clinical, biological and ultrasonographical studies of the proband and relatives

Figure 1

Clinical, biological studies and molecular analysis of the Turkish family. (A) Pedigree of the Turkish family. Double lines indicate consanguineous union. The proband (blue arrow) was analysed by ES. The genotypes for the BRCA2 mutated codon are indicated in red. (B) Position of the variant in BRCA2 gene and protein. The structure of the normal protein for the longest isoform of 3418 residues is shown below the genomic structure with the coding exons as coloured bars (Ensembl, reference transcript ENST00000544455.5). The variant (red arrows) lies at the very end of exon 20, which encodes 48 amino acids (A) encompassing the tower domain at the centre of the OB-fold ssDNA-binding domain (oligonucleotide/oligosaccharide-binding single-strand DNA-binding domain). (C) Partial view of the 3D model of the BRCA2 C-terminal domain (alpha-helices in red, beta-sheets in blue). The mutated residue (purple) is located near the ssDNA (green), at the base of the tower domain, that forms a stem of two long alpha-helices and a helix-turn-helix motif, similar to the DNA-binding domains of recombinases and homeodomain transcription factors. Inserts: difference between the occupancy of the lateral chain of WT (top) and mutated (bottom) residues at this position. ES, exome sequencing; ssDNA; single-stranded DNA; WT, wild type.

Blood counts and liver and thyroid balances were normal. Thyroid autoantibodies were undetectable. Hormonal replacement therapy (HRT) was initiated during her teenage years but was poorly observed, and stopped in early adulthood when she started thinking about getting pregnant. Bone densitometry at the age of 30 showed osteopaenia (T score=−2.3) related to poor adherence to HRT. She is presently 41 years old, displays normal blood assays and has no other clinical signs. The karyotype was 46, XX, and FMR1 premutation screening was negative. After two egg-donation procedures, she had two pregnancies with four healthy children.

The patient and family members have normal stature, normal head circumference and no abnormality in skin pigmentation, skeletal development or dysmorphia (table 1). There was no familial history of infertility or other diseases. The mother married at the age of 15 and had two pregnancies at 18 (a boy) and 22 (the proposita) years. This is in line with a delayed conception followed by secondary amenorrhoea at the age of 33 years, not investigated in Turkey. She is obese with a body mass index of 38. In order to rule out other genetic causes that could explain her subfertility, we performed a targeted next-generation sequencing (see supplementary information). The brother had one healthy son (17 years old) and three healthy daughters aged 15, 13 (both with normal puberty) and 8 years old, respectively. The 71-year-old father, a heavy smoker, had lung cancer at the age of 67 years, treated by radiotherapy and chemotherapy. A paternal uncle developed colorectal cancer at the age of 74 years and died a few months later. Six other paternal and maternal uncles and aunts, aged 61–75 years, had no history of cancer or infertility.

ES identified a homozygous missense variant in the DNA-binding domain of BRCA2

The patient was studied by ES. Familial consanguinity suggested an autosomal recessive inheritance pattern. The variants were therefore filtered on the basis of their homozygosity in the patient, their absence in unrelated fertile in-house controls and a minor allele frequency (MAF) below 0.01 in all available databases. Further filtering on available functional data for a possible role in fertility revealed the missense variant rs80359104, NM_000059.3: c.8524C>T (p.R2842C), located in exon 20 of BRCA2 (figure 1B and online supplementary table s1 and s2). The variant and its familial segregation were confirmed by Sanger sequencing, which also allowed exclusion of mosaicism in the patient’s blood cells (online supplementary figure s1). The variant is very rare and presents only at the heterozygous state in 3 out of 138 342 individuals without known phenotype (MAF 1.10−5) in the non-cancer GnomAD subset. It is absent in the Greater Middle East Variome database dedicated to Middle Eastern populations. It is predicted to be pathogenic by 16 of 17 predictive software (online supplementary table s3 and s4).

Supplemental material

The variant changes a strictly conserved amino acid (aa) at the base of the tower part of the BRCA2 DNA-binding domain, in close proximity to the groove that binds single-stranded DNA (ssDNA) (figure 1C and online supplementary figure s2). This C-terminal domain is essential for appropriate binding of BRCA2 to ssDNA.20

BRCA2 loads recombinases on ssDNA: RAD51 in mitotic cells, RAD51 and the meiotic DMC1 in germ cells, indicating a crucial role for BRCA2 in mitotic and meiotic HR.21 Cells defective in RAD51 or BRCA2 are defective in mitotic HR,6 7 and in mouse, germ cell-specific BRCA2 deletions lead to meiotic impairment and infertility.22 23 Therefore, we considered R2842C-BRCA2 as a very likely causal variant for isolated POI in our patient, although she does not present FA traits. Hence, we investigated the expression of BRCA2 in human oocytes and the functional impact of the BRCA2 variant on HR in human cells.

BRCA2 is expressed during meiotic prophase I in human foetal ovaries

In order to support a possible role for BRCA2 in female meiotic HR that would explain the patient's infertility, we verified its expression and location in human foetal ovaries. Indeed, Brca2 mRNA expression was reported in murine oocytes,23 and the BRCA2 protein was described to form recombination nodule-like foci along chromosome axes in human spermatocytes,24 but its expression during human female meiosis remained undocumented. Using quantitative reverse transcription PCR on RNA libraries prepared from human ovarian samples at various foetal stages, we detected a predominant expression of BRCA2 mRNA after 11 weeks postfertilisation, when oocytes enter and progress through meiotic prophase I (figure 2A). Immunostaining of human foetal ovarian sections showed that BRCA2 protein was detected mostly in pachytene stage oocytes (figure 2B). BRCA2 staining appeared as thick threads, likely corresponding to meiotic chromosomes. These results show that BRCA2 is indeed present on chromosomes in foetal human oocytes when meiotic DSB repair occurs.

Figure 2

BRCA2 expression in human foetal ovaries. (A) BRCA2 mRNAs were quantified by quantitative reverse transcription PCR from total RNA of pooled human foetal ovaries from various developmental stages (above). Beta-actin was used as a reference, and expression is provided as percentage of the maximum. each RNA library was analysed in triplicate and the bar indicates the mean. Quantification of SYCP3 mRNA is shown at the same stages for comparison (below). (B) BRCA2 immunostaining (purple) in the cortex of a 24-weeks postfertilisation human ovary. Meiotic chromosomes are stained in zygotene and pachytene stages oocytes. No staining is observed when immunohistochemistry is performed in the absence of primary antibody (NEG). Arrows point to germ cells at the indicated stage. Scale bar: 10 µm. NEG, negatif; mRNA, messenger RNA.

R2842C-BRCA2 displays a reduced DSB-induced HR efficiency

Although referenced in ClinVar (VCV000052610) and in the COSMIC database of variants in cancer (COSM23938), R2842C-BRCA2 is considered as a variant of unknown significance for breast cancer predisposition (Variant of Unkown Significance (VUS), International Agency for Cancer Research (IARC) class 3). Previous attempt to classify BRCA2 VUS in a hamster lung fibroblast cell line or in mouse embryonic stem cells showed that this variant displayed a decrease in HR efficiency, at the limit for inferring pathogenicity.25–29 Therefore, the significance of this variant and its classification as a causal variant for human pathology remained unclear. Since human and rodent cells differ in DSB repair efficiency and regulation, we analysed the specific impact of R2842C-BRCA2 on HR in human cells. We used the RG37 cell line,17 a human SV40 immortalised fibroblast line bearing the DR-GFP substrate18 that monitors gene conversion induced by targeted cleavage by the I-SceI meganuclease (figure 3A). Both the expression of WT-BRCA2 and R2842C-BRCA2 stimulated the efficiency of DSB-induced HR (figure 3B). We then silenced the endogenous BRCA2 using a specific siRNA targeting its 3′UTR sequence and complemented these cells with either the WT or mutated BRCA2 (figure 3C). As expected, silencing endogenous BRCA2 decreased HR efficiency, and WT-BRCA2 fully complemented HR efficiency. R2842C-BRCA2, expressed at similar levels than WT-BRCA2, only partially complemented HR efficiency, to 67%±6% compared with WT-BRCA2 (figure 3B).

Figure 3

Impact of the R2842C-BRCA2 variant on HR induced by targeted DSB. (A) schematic representation of the DR-GFP substrate for the study of HR. Two inactive GFP (iGFP and SceGFP) genes are organised into direct repeats. The I-SceI meganuclease generates a targeted DSB cleavage into the substrate (red). HR between the two GFP genes generates a functional GFP. The DR-GFP substrate is stably integrated in the SV40-transformed fibroblasts RG37 cell line, and the relative HR efficiency is quantified as the fraction of GFP-positive cells (ie, with a repaired GFP gene after targeted cleavage), scored by FACS. (B) HR efficiency, measured by the fraction of GFP+ cells, in cells expressing either the WT or the mutated BRCA2 protein (normalised to I-SceI transfected cells), and transfected either with a control siRNA (siCT) or a siRNA targeting the 3’UTR of endogenous BRCA2 mRNA (siBRCA2). The values are normalised to the control and represent the average ±SEM (p values from Mann-Whitney test) for at least three independent experiments. (C) Expression of endogenous BRCA2 and exogenous WT-BRCA2 and R2842C-BRCA2. Twenty micrograms of total proteins extracted from cell lines expressing either the WT BRCA2 or the R2842C mutant were electroblotted in the presence of endogenous BRCA2 (siCT) or after specific silencing (siBRCA2). For each condition, the expression of I-SceI and BRCA2 and the efficiency of silencing were measured. We used β-actin as a loading control. Below: relative quantification of BRCA2 versus actin by quantification of bands intensity with ImageJ. * P ≤ 0.05, ** P ≤ 0.01, **** P ≤ 0.0001. DSB, double-strand break; HR, homologous recombination; WT, wild type.

Interestingly, our results are in line with the functional impact on HR recently studied to classify variants of unknown significance or associated with a high risk of cancer using mouse embryonic stem cells.28 In this study, the residual HR activity of the R2842C variant is almost identical to what we obtained (31% and 33%, respectively). Mesman et al also studied other missense variants, classified as benign (class 1), likely benign (class 2) or with unknown significance (class 3, VUS), or associated with a high risk of cancer, such as the R3052W and the G2609D variants. The conclusion is that the residual HR of our variant (R2842C) is intermediate between the pathogenic or likely pathogenic variants (class 4 or 5, such as R3052W or G2609D) with a residual HR of <30%, and those classified as benign, likely benign or VUS with a residual HR of >50%.28

These data show that the R2842C-BRCA2 variant partially affects HR efficiency in human cells. This significant residual activity could account for the absence of developmental defects in the patient.

Increased chromosomal instability in the patient’s cells

Hypersensitivity to the chromosome-breaking effect of cross-linking agents is a reliable marker of FA.30 Therefore, we studied MMC-induced chromosomal breaks in lymphoblastoid cells derived from the patient, her mother, two fertile control women and an FA patient (FANCD1). In the absence of MMC, few spontaneous breaks were observed in cells derived from the proposita and the patient with FANCD1 (figure 4A,B). On exposure to 300 nM MMC, healthy cells are expected to show one to five breaks/cell in no more than 30% of the cells, whereas for a patient with bona fide FA, no undamaged cells are left and most cells show more than 10 breaks/cell.30 At 300 nM MMC, all FANCD1 cells presented breaks, as expected, while the patient's cells exhibited a slight increase of chromosomal breaks, compared with the heterozygous mother’s and the WT control’s cells. Indeed, the number of breaks in the patient’s cells was only 10 times the number in WT cells compared with 41 times in FANCD1’s cells. Furthermore, at high MMC dose (1000 nM), while breaks in FANCD1 cells were too numerous to be quantified, the patient’s cells presented only a modest increase of breaks. Therefore, our data show that the patient’s cells display moderately increased levels of chromosomal breaks compared with WT cells, a cellular phenotype distinct from the typical FANCD1 hypersensitivity.

Figure 4

Increased chromosomal instability and reduced proliferation rate in the patient’s cells. (A) Chromosomal breaks analysis of lymphoblastoid cell lines. Metaphases in the patient’s and FANCD1’s cells in the presence of 300 nM MMC. Chromosomal breaks and radial figures are shown (arrowheads). (B) Quantification of chromosomal breaks in lymphoblastoid cell lines derived from the patient, the mother, an FA patient (FANCD1) and a WT control, in the absence or in the presence of MMC. (C) Reduced proliferation rate of the patient’s primary fibroblasts. Cells from the patient with POI, two WT controls (WT1 and WT2) and an FA patient (FANCD1) were cultured in six-well plates and counted every 2–3 days during 13 days. The value corresponds to the mean±SEM of at least three independent experiments. The statistical significance was calculated from rope comparison of linear regression of growth curves. MMC, mitomycin C; WT, wild type.

Reduced proliferation rate of the patient’s primary fibroblasts

Then we compared the proliferation rate of primary fibroblasts from the patient with POI, WT controls and a patient with FANCD1. As expected, the proliferation of FANCD1 cells was markedly affected when compared with WT cells (figure 4C). Remarkably, the patient’s fibroblasts exhibited a reduced proliferation rate, intermediate between WT and FANCD1 cells.

Altered radiation-induced RAD51 foci formation in the patient’s fibroblasts

The main role for BRCA2 is the loading of the pivotal RAD51 recombinase on damaged DNA, a crucial step for triggering HR. Therefore, we monitored the radiation-induced assembly of RAD51 foci, which are considered sites of HR initiation events. As expected, FANCD1 cells failed to assemble RAD51 foci (figure 5A,B). Consistent with the aforementioned data, the patient’s cells showed an intermediate phenotype: indeed, at 6 Gy, they assemble foci with kinetics comparable with WT cells, but the level of the plateau was about twofold lower than that in WT cells (figure 5B). A dose–response analysis confirmed the complete deficiency in RAD51 foci assembly for FANCD1 cells, at all irradiation doses (figure 5C,D). In the patient’s cells, the number of RAD51 foci increased up to 2 Gy similarly to WT cells but did not further increase at higher doses (figure 5C,D; left panel). Consequently, the patient’s cells showed lower levels of RAD51 foci at high doses (>2 Gy) when compared with WT cells (figure 5C,D; right panel). Together, these data show a dose-dependent sensitivity in the patient’s cells, able to process low levels of DNA damage but failing to assemble RAD51 foci when faced with high levels of DNA damage.

Figure 5

Dose-sensitive alteration of RAD51 foci formation in the patient’s cells. (A) Rad51 nuclear foci assembly in irradiated primary fibroblasts from the patient, two WT controls and an FA patient (FANCD1) (6 Gy, 6 hours). Fixed and permeabilised cells were probed with an anti-RAD51 antibody. (B) Kinetics of Rad51 foci assembly (6 Gy) in irradiated primary fibroblasts. (C,D) Dose–response of Rad51 foci assembly (6-hour postirradiation) in irradiated primary fibroblasts. (C) Complete dose–response of Rad51 foci per nucleus; (D) Rad51 foci at low (2 Gy, left panel) and high (10 Gy, right panel) levels of irradiation, respectively (medians are in blue (n=3)). (B,C) The values correspond to the mean±SEM (n=3); p values (stars) were obtained with the Mann-Whitney test. WT, wild type.

Thus, in accordance with the recommendations of the ENIGMA consortium on functional assays for analysis of variants of uncertain significance of BRCA2,26 our in vitro studies, including RAD51 foci formation and chromosomal breaks studies, confirm previously published data on the hypomorphic nature of the BRCA2-R2842C variant.27 28 Regarding the clinical presentation of the patient, the aforementioned functional evidence, the ACMG-AMP classification and the Sherloc scoring framework31 (online supplementary table s4), this variant can be considered as pathogenic.

No pathogenic variant was found in other FA genes

Biallelic pathological variants in two distinct genes of the FA pathway can be found in a single individual.32 In order to rule out the hypothesis that the cellular phenotypes observed in the patient’s cells could be due to mild biallelic variants in FA genes other than BRCA2, we have analysed the variants found by ES in the 26 other FA genes. The good coverage of each gene, the presence of heterozygous variants (both in coding regions and in UTRs or intronic regions) and unbiased allelic ratios exclude the possibility of having missed a pathogenic variant in another FA gene (online supplementary table s5). This analysis confirms that the cellular phenotypes observed in the patient's cells are not due to variants in FA genes other than BRCA2. In particular, no variants of RAD51 could explain the defects in foci assembly detected in the patient’s cells.

Discussion

We report here a BRCA2 homozygous hypomorphic variant in a patient with POI, and, remarkably, without cancer or FA trait in the patient and her family. The patient clearly presents an isolated POI because she has no characteristic clinical signs of FA, such as congenital malformations or skin pigmentation abnormalities.33 She also has a normal head circumference, and her size, similar to that of her heterozygous mother, is within normal ranges (according to Turkish data34). The fact that she does not present any bone marrow failure or malignancy, together with the lack of FA case in all her family, also excludes the possibility of a ‘Fanconi-like’ syndrome, such as the Estren-Dameshek syndrome, a mild form of FA. In this syndrome, very few or no congenital anomalies could be found in patients, but the diagnosis of a classical FA was made by characteristic cytogenetic analysis in the patient and by typical phenotypes found in the same sibship.33 At the cytogenetic level, the most specific criteria for FA is the hypersensitivity of patients’ somatic cells to cross-linking agents. On exposure to MMC, our patient's cells exhibited only a slight increase of chromosomal breaks compared with WT control cells. These cellular defects are largely below the instability observed in typical FA cells,30 and somatic mosaicism was excluded because of the homogeneity of the variant detection provided by ES. Therefore, this patient cannot be considered clinically as an FA case.

The filtering of variants obtained by ES in our patient yielded the BRCA2 R2842C variant as the only candidate to be involved in the POI of our patient. A highly likely role for BRCA2 in germ cell HR stems from several lines of evidence. Meiotic recombination is a complex and highly regulated process occurring in meiotic prophase I that involves specific meiotic factors such as MEIOB and known BRCA2 partners such as DMC1 and MEILB2.35–38 BRCA2 has also been shown to be essential for meiotic DNA repair in various animal models, such as zebrafish, Caenorhabditis elegans and Drosophila.14 39–41 In mouse, the rescue of the invalidated murine gene with a human transgene led to low levels of functional protein and to a complete impairment of male meiosis and a 10-fold loss of the follicular pool, with numerous meiotic abnormalities in the remaining oocytes.23 Very recently, it was shown in mouse that an oocyte-specific Brca2 invalidation leads to meiotic impairment, germ cell depletion and infertility.11 22

While the role and clinical impact of homozygous and heterozygous BRCA2 variants in mitotic cells are widely studied, the impact of such defects in human germ cells remains less understood, mainly because almost all patient with FANCD1 die before puberty. Here, we show that BRCA2 mRNA and protein are indeed expressed in human foetal ovaries in pachytene stage oocytes, when meiotic HR occurs. The main role of BRCA2 is the loading of the pivotal recombinase RAD51 on damaged DNA to allow the repair of DSBs by HR. The location of the variant in the ssDNA-binding domain of BRCA2 suggests it can affect the protein function. Indeed, using an HR reporter assay, we showed that the R2842C-BRCA2 mutant exhibited a reduced DSB-induced HR efficiency. Consistently, the patient’s cells display a reduced proliferation rate and a dose-sensitive alteration of radiation-induced assembly of RAD51 foci. Furthermore, we verified that these somatic cellular defects could not be attributed to variants in other genes implicated in the FA pathway. Taken together, our data strongly support a crucial role of BRCA2 in human meiotic HR, and therefore a likely impact of R2842C-BRCA2 reduced activity in this process in our patient with POI.

Our observation describes the first adult patient with fully isolated POI due to a biallelic variant in BRCA2, without cancer or FA traits despite a detailed clinical and cytogenetic description. Indeed, patients carrying biallelic damaging variants in BRCA2 usually present with the severe FANCD1 syndrome and die before puberty. The possibility of a heterogeneous clinical presentation in carriers of biallelic BRCA2 variants is currently emerging from the report of older patients. As shown in online supplementary table s6, we can conclude that in all previous patients with POI, there was either evidence of FA or not enough information to fully exclude it, especially microcephaly and short stature (frequently reported in patients with FA related to FANCD1 mutations42) and absence of cytogenetic studies, critical for the diagnosis of FA.

Heterozygous BRCA2 variants increase susceptibility to familial breast and ovarian cancers.43 Such increased susceptibility has not been observed in our patient's family. Furthermore, the fact that neither the patient nor her relatives have yet developed other pathologies implies that the R2842C-BRCA2 variant is hypomorphic and retains a residual HR activity, as supported by our functional studies. Interestingly, we showed that RAD51 foci assembly was not affected at low irradiation doses in the patient’s somatic cells. In addition, the patient’s cells displayed a growth rate and levels of chromosomal breaks intermediate between those of WT and FANCD1 cells. The residual HR activity of the R2842C-BRCA2 mutant could account for the absence of somatic pathologies in vivo. Somatic cells are rarely spontaneously confronted with high levels of simultaneous DSBs, while by contrast, meiotic recombination in germ cells is initiated by hundreds of DSBs requiring HR to form crossing-overs mandatory to produce functional haploid gametes.38 In our functional test, RAD51 foci assembly by the mutated BRCA2 was significantly decreased at high doses (>5 Gy), generating a great number of DSBs (with an estimation of 30–40 DSB/Gy/mammalian genome44). Therefore, the R2842C-BRCA2 variant could be expected to affect the processing of such a high number of simultaneous meiotic DSBs.

The study of DNA repair-deficient knock-out mouse models and in vitro assays suggest that oocytes are equipped to execute the repair of a wide variety of DNA lesions.45 DNA repair genes such as MLH1, PMS2, MSH6 and ERCC1 are also expressed in the ovary45 and play a major role in ensuring female fertility. Major alteration of these genes are involved in many non-reproductive cancers such as Lynch syndrome or skin cancers.45 Thus, in the same way, we can speculate that hypomorphic variants of many of these DNA repair genes will be identified in isolated POI cases without evident somatic pathology.

The hypomorphic nature of a BRCA2 variant does not exclude a risk of cancer. Indeed, in a recent case–control study on breast cancer using genotyping data from almost 50 000 cases (48.159 cases and 48.231 controls), it was shown that the hypomorphic missense variants of BRCA2 are associated with a moderate risk of breast cancer and therefore require appropriate long-term monitoring.46 In particular, variants G2508S and W3035S with a residual HR activity of 50% according to Mesman and colleagues28 are associated with a moderate risk of breast cancer in Asian and in the European populations, respectively.46

It is therefore reasonable to propose that clinical management of patients with POI harbouring hypormophic variants of BRCA2 (residual HR ≤ 50% as described by Mesman et al, demonstrated by functional assays), as in our patient, should include intensified screening programmes, and presymptomatic genetic testing for other family members should be discussed. Regarding the possibility of undergoing prophylactic surgery, there is no point in maintaining a risk of ovarian cancer, and an oophorectomy should therefore be discussed in these patients, since patients with POI with biallelic BRCA2 mutations have small ovaries without follicle. In addition, according to the current literature, we suggest favouring a breast monitoring approach rather than a mastectomy.

In conclusion, we describe and functionally characterise here a homozygous hypomorphic variant of BRCA2 in a patient with isolated POI. As BRCA2 is a major susceptibility gene for breast and ovarian cancer, this observation should change the genetic counselling and pretest information for patients with POI. Indeed, the genetic counselling should now be addressed while keeping in mind a possible defect in a major DNA repair gene such as BRCA2. More generally, this study also has a wide impact for the understanding of the processes controlling genome plasticity and the consequences of their defects in somatic and germ cells.

Acknowledgments

We thank Baptiste Fouquet for his help in some experiments, Jean Soulier and the Cellulothèque des hémopathies de l’Hôpital Saint-Louis for the gift of FANCD1 primary fibroblasts and Alexandra Benachi for foetal ovaries.

References

Footnotes

  • SC, AH and ED are joint first authors.

  • Twitter @SCaburet

  • Contributors MM, BSL and GL designed research studies. ED, AH, ST and SM conducted experiments. SC, ED, AH, GL, BSL and MM acquired and analysed data. AH, HC and ML collected clinical data and biological samples of the patient and their parents. MM, BSL, SC, AH and GL wrote the manuscript.

  • Funding This study was supported by Université Paris Diderot (SC), Université Paris Sud-Paris Saclay (ED, AH and MM), by the Agence Nationale de Biomédecine (AH and MM) and by Institut Universitaire de France (GL). BSL was supported by the Ligue Nationale contre le cancer 'Equipe labellisée 2017', Agence Nationale de la Recherche (ANR-16-CE12-0011-02 and ANR-16-CE18-0012-02), AFM-Téléthon and Institut National du Cancer (INCa-PLBIO18-232).

  • Competing interests None declared.

  • Patient consent for publication Obtained.

  • Ethics approval The study was approved by all the institutions involved and by the Agence de Biomedecine (reference number PFS12-002).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available upon reasonable request. Data supporting the claims made in this paper are available upon request after its publication.