Article Text
PDF
Abstract
Background Genetic analysis of BRCA1 and BRCA2 for the diagnosis of hereditary breast and ovarian cancer (HBOC) is commonly restricted to coding regions and exon-intron boundaries. Although germline pathogenic variants in these regions explain about ~20% of HBOC cases, there is still an important fraction that remains undiagnosed. We have screened BRCA1/2 deep intronic regions to identify potential spliceogenic variants that could explain part of the missing HBOC susceptibility.
Methods We analysed BRCA1/2 deep intronic regions by targeted gene sequencing in 192 high-risk HBOC families testing negative for BRCA1/2 during conventional analysis. Rare variants (MAF <0.005) predicted to create/activate splice sites were selected for further characterisation in patient RNA. The splicing outcome was analysed by RT-PCR and Sanger sequencing, and allelic imbalance was also determined when heterozygous exonic loci were present.
Results A novel transcript was detected in BRCA1 c.4185+4105C>T variant carrier. This variant promotes the inclusion of a pseudoexon in mature mRNA, generating an aberrant transcript predicted to encode for a non-functional protein. Quantitative and allele-specific assays determined haploinsufficiency in the variant carrier, supporting a pathogenic effect for this variant. Genotyping of 1030 HBOC cases and 327 controls did not identify additional carriers in Spanish population.
Conclusion Screening of BRCA1/2 intronic regions has identified the first BRCA1 deep intronic variant associated with HBOC by pseudoexon activation. Although the frequency of deleterious variants in these regions appears to be low, our study highlights the importance of studying non-coding regions and performing comprehensive RNA assays to complement genetic diagnosis.
- Brca1/2
- hereditary breast and ovarian cancer
- deep intronic regions
- splicing
- pseudoexon
Statistics from Altmetric.com
Introduction
Pathogenic germline variants in the tumour suppressor genes BRCA1 (MIM# 113705) and BRCA2 (MIM# 600185) (BRCA1/2) predispose to breast and ovarian cancer (BC/OC). To date, more than ~3500 risk-associated variants in BRCA1/2 have been reported in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/). Carriers of pathogenic variants in BRCA1 result in an increased cumulative risk of developing BC and OC that reaches 66% and 41% at age 70, respectively. Similarly, for BRCA2, the cumulative risks for BC and OC reach 61% and 15%, respectively.1 The identification of pathogenic variants in BRCA1/2 provides accurate clinical management of hereditary breast and ovarian cancer (HBOC) families based on personalised prevention and therapeutic strategies.2–4 However, pathogenic mutations in these genes explain less than 20% of HBOC cases.
In recent years, causative genetic variants in other tumour suppressor genes involved in homologous recombination DNA repair have also been linked to moderate risks of developing BC/OC.5–8 The application of massive sequencing technologies in the clinical setting allows the simultaneous screening of risk-associated HBOC genes, improving the effectiveness of identifying new families at risk.3 9 10 However, there is still an important fraction of HBOC cases for which genetic analysis does not identify causative variants underlying the predisposition to BC/OC.2
Genetic testing commonly identifies variants that generate truncated proteins (nonsense, frameshift, splicing variants) or alter functional domains (missense, in-frame variants). However, deep intronic regions may also contain nucleotide changes that could alter splicing by creating/activating splice sites or by disrupting cis-regulatory splicing elements (intronic enhancers/silencers). These variants have the potential to generate aberrant transcripts by introducing pseudoexons in the mature mRNA.11–13 Such events have been reported in a variety of human genetic syndromes, including cancer,14 15 but their association with HBOC remains largely unexplored. Given that conventional BRCA1/2 analysis in the majority of clinical laboratories is restricted to coding regions and exon-intron boundaries, there is currently a lack of information about the frequency of deleterious spliceogenic variants occurring in deep intronic regions of these genes. Previous works based on RNA analysis of patients with HBOC with uninformative BRCA1/2 results (i.e., no pathogenic variant identified) have also proposed the existence of spliceogenic variants in deep intronic regions that could explain part of the missing HBOC genetic susceptibility.16–18 In the present study, we have analysed BRCA1/2 introns by targeted gene sequencing in a group of Spanish patients with high-risk HBOC testing negative for BRCA1/2. Candidate spliceogenic variants have been selected using in silico approaches and their impact has been characterised in patient RNA. To our knowledge, this is the first study aiming to screen BRCA1/2 deep intronic regions in clinical samples and determine the frequency of spliceogenic variants in these regions in the Spanish population.
Materials and methods
Patient and control samples acquisition
A total of 192 patients with hereditary BC/OC testing negative for BRCA1/2 were selected according to the following inclusion criteria: personal history of BC before age 36 (n=77), BC with two or more first or second-degree relatives with BC/OC (n=60), personal history of OC before age 60 (n=38). Additionally, we included six patients diagnosed with two BC (bilateral or ipsilateral with or without BC family history); seven patients with BC diagnosed after age 36, with one male BC, OC or pancreatic cancer-affected relative and four patients with BC/OC history affected with colon, endometrium, sarcoma or stomach cancer. All patients were referred for genetic counselling at the High-risk and Prevention Cancer Unit from Vall d’Hebron Hospital, Barcelona and they provided written informed consent for BRCA1/2 testing and research studies. Patients were screened for BRCA1/2 point mutations and large genomic rearrangements by Sanger sequencing and Multiplex ligation-dependent probe amplification (MLPA) (MRC-Holland), respectively.
A total of 327 non-affected control samples were recruited from the Spanish National DNA Bank (Salamanca, Spain). Controls were selected randomly from a population of healthy women above 50 years of age with no personal or family history of cancer. Similarly, control RNAs were obtained from 20 healthy individuals without HBOC history and from four normal breast tissue samples supplied by Biochain (AMSBIO).
Patient DNA was obtained from 10 mL of peripheral blood and isolated using Gentra Puregene Blood Kit (QIAGEN), following manufacturer’s protocol. DNA concentrations were determined using Qubit dsDNA BR Assay kit (ThermoFisher). RNA from variant carriers and controls (n=20) was isolated from 10 mL of peripheral blood samples using Trizol Reagent (ThermoFisher). RNA was cleaned up using RNeasy Minikit (QIAGEN) and treated with RNase-Free DNase Set (QIAGEN) to remove traces of genomic DNA. RNA integrity was determined in E-Gel Precast agarose gels (Invitrogen) and concentrations were measured using a NanoDrop Spectrophotometer (ThermoFisher).
Massively parallel sequencing of BRCA1/2 intronic regions
Agilent SureDesign web-based tool (Agilent Technologies) was used to design a custom Agilent SureSelect bait library of probes targeting whole coding, non-coding and intronic sequences with additional flanking 10 kb genomic sequences of BRCA1 and BRCA2. Captured genomic regions from BRCA1 and BRCA2 spanned chr17: 41,186,312–41,287,500 and chr13: 32,879,617–32,983,809, respectively (see online supplementary figure 1 for BRCA1/2 genomic coverage). Deep sequencing was performed in a MiSeq Instrument (Illumina). DNA library preparation, sequencing protocols and bioinformatics pipeline for sequencing data alignment and variant calling have been extensively described in a previous work from our laboratory.19
Supplemental material
In silico analysis and variant prioritisation
Variants were annotated with ANNOVAR tool using GRCh37/hg19 genome assembly. Variants with a minimum of 10 reads per base, with at least one read in sense (+) and antisense (-) strands, with a variant allele frequency between 75% and 8% and a Minor Allele Frequency (MAF) <0.01 at 1000 Genomes Project database, were included for reannotation using Alamut software v2.10 (Interactive Biosoftware). Reference transcripts NM_007294.3 (BRCA1) and NM_000059.3 (BRCA2) were used for variant reannotation. Population data from the Genome Aggregation Database (gnomAD) (http://gnomad.broadinstitute.org/) and splicing predictions from SpliceSiteFinder-like (SSF), MaxEntScan (MES), HumanSplicingFinder (HSF), GeneSplicer and Splice Site Prediction by Neural Network (NNSPLICE) were incorporated.
Variants located beyond +20/–20 positions from canonical donor (GT) and acceptor (AG) splice sites, respectively, were prioritised for RNA analysis when: (1) MAF in non-Finnish European population (NFE)<0.005; (2) a local splicing effect (i.e., creation of new splice sites or activation of existing cryptic sites) was predicted according to Alamut’s interpretation algorithm , which uses splice site signal scores from MES, NNSPLICE and SSF tools (https://www.interactive-biosoftware.com/alamut-visual/). We applied a final filtering step that consisted of removing variants occurring in >1 patient. A diagram summarising the strategy followed for deep intronic regions screening and variant prioritisation is depicted in figure 1.
Strategy followed for the study of BRCA1/2 deep intronic regions in patients with HBOC. The diagram summarises the steps followed from patient selection to RNA analysis of the selected variants. The total number of unique BRCA1/2 variants is denoted in each filtering step. CE, capillary electrophoresis; HBOC, hereditary breast and ovarian cancer; LGR, large genomic rearrangements; MAF, minor allele frequency; MES, MaxEntScan; NFE, non-Finnish European population; NNSPLICE, SpliceSitePrediction by Neural Network; SSF, SpliceSiteFinder-like.
Candidate BRCA1/2 deep intronic variants predicted to alter splicing and characterisation in patient RNA
Characterisation of BRCA1/2 variants in patient RNA
A total of 500 ng of RNA from carriers and controls was retrotranscribed using PrimeScript RT Reagent kit (Takara), combining oligo-dT and random primers. Long PCR fragments (1.5–6 kb) were obtained using Expand Long Range dNTPack (Roche), and short PCR fragments (up to 1.5 kb size) were obtained with EcoTaq (Ecogen). Amplified fragments covered the exons adjacent to intron containing the variant. Primer sequences and PCR conditions are described in online supplementary table 1A. We used 2–5 uL of cDNA to a final PCR reaction of 25–35 uL. Cycling conditions were performed according to manufacturer’s instructions, with elongation times of 2 min for amplicons<1 kb and 7 min for amplicons>1 kb, to allow the amplification of potential long aberrant transcripts present in the samples. RT-PCR products were qualitatively assessed by capillary electrophoresis (CE) in a QIAxcel instrument, using QIAxcel DNA High-resolution kit (QIAGEN). Controls were run in parallel with patient samples and were used as reference to compare RNA patterns. RT-PCR products were purified using ExoSAP-IT PCR Product Cleanup Reagent (ThermoFisher) and sequenced using BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems). Sequencing products were run in a Genetic Analyzer ABI3130xl (Applied Biosystems) and Sanger electropherograms were visualised using SeqScape v2.6 and Sequencing Analysis v2.6 softwares (Applied Biosystems). Polymorphic exonic variants in BRCA1 [c.4308T>C (rs1060915); c.4837A>G (rs1799966)] and BRCA2 [c.-52A>G (rs206118); c.-26G>A (rs1799943); c.865A>C (rs766173); c.1114A>C (rs144848); c.7242A>G (rs1799955)] or other exonic variants present in samples were used to determine potential allelic imbalance derived from frameshift events degraded by the nonsense-mediated decay (NMD). PCR primers and internal sequencing primers used for allelic imbalance assessment are listed in online supplementary table 1B
Supplemental material
Qualitative analysis by capillary electrophoresis of fluorescent amplicons
Analysis by CE of FAM-labelled amplicons was performed to characterise aberrant splicing patterns, providing higher sensitivity and resolution20 21 to rule out the presence of aberrant products not detected with QIAxcel electrophoresis and Sanger sequencing. Amplicons were generated with primers labelled with a FAM molecule at the 5’ end. RT-PCR products were diluted 1:30 and run into a Genetic Analyzer ABI3130xl (Applied Biosystems) under the following electrophoresis conditions: temperature 60°C, 12 s injection at 1.2 kV, 2000 s run at 12 kV. GeneScan ROX500 (Applied Biosystems) was used as internal size standard and peak electropherograms were visualised using GeneMapper Software 5 (Applied Biosystems).
BRCA1 expression analysis by quantitative PCR
Global BRCA1 expression levels were measured in variant carrier and 10 controls in two independent quantitative PCR (qPCR) assays, using Taqman probes targeting exons 5–6 junction (Hs01556196_m1) and exons 23–24 junction (Hs01556193_m1), respectively. A known pathogenic BRCA1 splicing mutation c.302–1G>A22 was used as positive control. Taqman Universal Master Mix II (ThermoFisher) was used for qPCR reactions, and reference genes GAPDH (Hs02758991_g1), B2M (Hs99999907_m1), ACTB (Hs03023880_g1) and HPRT1 (Hs99999909_m1) were used for data normalisation. Experiments were run in a 7900HT Fast Real-Time PCR System (Applied Biosystems) using default cycle conditions. BRCA1 global expression levels were calculated using qBASE +software (Biogazelle), which applies the multiple-gene reference normalisation method.23 All qPCR experiments were performed in triplicate.
Allele-specific assessment of BRCA1 normal transcript
The ability of the variant allele to generate normal transcripts was investigated with a specific RT-PCR assay. Full-length (FL) transcript specific amplification was performed using a forward primer complementary with the last 21 nucleotides of exon 12 and the first two nucleotides of exon 13 (5’-AGTGACATTTTAACCACTCAGCA-3’) and a reverse primer located in exon 18 (5’-TCCGTTCACACACAAACTCAG-3’) (amplicon size=934 bp). PCR cycling conditions consisted in: denaturing step at 95°C for 5 min; 35 cycles of 95°C for 15 s, 56°C for 5 s and 72°C for 1 min 30 s and a final elongation step of 10 min at 72°C. Products were sequenced by Sanger and heterozygous exonic loci c.4308T>C (rs1060915; exon 13) and c.4837A>G (rs1799966; exon 16) were used to determine biallelic contribution to FL transcript expression.
Levels of BRCA1 FL transcript were estimated using CE data obtained from QIAxcel instrument. Variant carrier and 10 controls were analysed in four RT-PCR experiments spanning exons 11–13 (see primers in online supplementary table 1) and peak areas corresponding to the FL transcript (275 bp) were used to estimate its relative abundance. FL levels in one carrier of the BRCA2 c.6937+594T>G deep intronic variant16 were also measured in three RT-PCR experiments, with a forward primer located in exon 12 (5’-AGGCTTCAAAAAGCACTCCA-3’) and a reverse primer located in exon 14 (5’-TCATCAGAGCCATGTCCATC-3’) (amplicon size=294 bp). FL data were normalised by dividing each FL peak area with the FL mean obtained from the control group. Data were analysed using GraphPad Prism software.
We also measured BRCA1 FL transcripts by qPCR using a Taqman assay targeting exons 12–13 junction (Hs00173233_m1), which is only present in the reference FL transcript but absent in the aberrant transcript. The splicing mutation c.302–1G>A and a large BRCA1 deletion spanning exons 1–13 were used as positive controls. Cycling conditions and data analysis were performed as described in the previous section.
Variant genotyping in BRCA1/2 negative families and controls
We additionally genotyped BRCA1 c.4185+4105C>T in 1030 HBOC Spanish families testing negative for BRCA1/2 and 327 Spanish controls to determine the frequency of this variant in our population. Genotyping of 380 HBOC samples was performed by conventional PCR using intron 12 primers F-AAGCCCCTTGGAGTTGTCAA and R-TTGACAGAGTCCCAAACCCA (amplicon size=184 bp) and posterior Sanger sequencing. The remaining 650 samples and the controls were genotyped using a custom TaqMan SNP assay (ThermoFisher) containing unlabelled PCR primers (F-GTCACCAGTATTCTCCACTTCTTCA, R-GCAAAGAGAGAAAAGGCCTCCTAAA), one VICdye-MGB-labelled probe to detect allele C and one FAMdye-MGB-labelled probe to detect allele T. Taqman Universal Master Mix II (ThermoFisher) was used and 5–20 ng DNA were loaded in each reaction. Allelic discrimination assays were run in a 7900HT Fast Real-Time PCR System (Applied Biosystems) under default cycling conditions. Allelic discrimination plots were obtained using SDS 2.4 software (Applied Biosystems). A positive control (variant carrier) was used in all assays.
Results
RNA analysis of BRCA1/2 deep intronic variants with potential splicing effects
The strategy followed for the study of BRCA1/2 deep intronic regions in 192 high-risk HBOC families identified 30 BRCA1 and 37 BRCA2 candidate splicing variants in 53 patients (28%) (figure 1). Variants are listed in table 1 with detailed information from in silico predictions, gnomAD population frequencies, ClinVar review status and RNA results obtained in this study (splicing analysis +allelic imbalance assessment).
A total of 31 variants were not present in gnomAD, which includes whole-genome data from ≈15 500 unrelated individuals. According to Alamut’s in silico predictions, 29 variants (43.3%) were predicted to create new splice sites (11 donor sites and 18 acceptor sites) and 38 variants (56.7%) were predicted to activate pre-existing splice sites (7 cryptic donor sites and 31 cryptic acceptor sites).
Patient RNA could be obtained to assess the effect of 27 BRCA1 and 27 BRCA2 variants. All variants were characterised by RT-PCR assays comparing splicing profiles with healthy controls and posterior Sanger sequencing. Visual inspection of RT-PCR products in QIAxcel and Sanger electropherograms did not detect splicing alterations in any variant carrier (see table 1 and online supplementary figures 2 and 3), with the exception of BRCA1 c.4185+4105C>T carrier which generated an extra transcript absent in control samples (figure 2A). BRCA1/2 allelic imbalance was ruled out in 28 patients (table 1) by inspection of Sanger electropherograms at heterozygous exonic loci (see online supplementary figure 4), but an allelic imbalance was detected in BRCA1 c.4185+4105C>T carrier (figure 2B).
BRCA1 c.4185+4105C>T characterisation in patient RNA. (A): QIAxcel electrophoresis of RT-PCR assay covering exons 11–13 (275 bp). An extra band of ~400 bp was detected in variant carrier (ins114nt), not present in controls. (B): Sanger electropherogram showing allelic imbalance at polymorphisms c.4308T>C and c.4837A>G. (C): Sanger sequencing confirmed the insertion of 114 nucleotides (nt) between exons 12 and 13, generating a new transcript that we annotated as ▼12A (r.4185_4186ins4185+3990_4185+4103). This transcript is predicted to encode for a truncated BRCA1 protein (p.Gln1395_Gln1396insSerLysSerLeu*).
A total of 16 variants occurring in >1 patient (6 BRCA1 and 10 BRCA2) were also identified but not prioritised for RNA analysis (online supplementary table 2). Among these, two individuals carried the BRCA2 c.6937+594T>G variant previously detected in the French population and reported as the first BRCA2 deep intronic variant generating an aberrant transcript by activation of a cryptic splice site.16
Family origin and clinical features from BRCA1 c.4185+4105C>T carrier
The family is originally from Lleida (western Catalonia). The proband was diagnosed with a high-grade ovarian carcinoma with papillary serous histology at age 58. The proband’s father was diagnosed with prostate cancer at age 70, and a paternal female cousin was diagnosed with BC at age 40. After 2 years of follow-up, the patient was diagnosed with a grade 1 infiltrating ductal breast carcinoma, with positive hormonal receptors (ER+,PR+) and negative HER2 receptors (see family pedigree in online supplementary figure 5).
In silico splicing analysis of BRCA1 c.4185+4105C>T
BRCA1 c.4185+4105C>T variant was detected in co-occurrence with BRCA1 c.80+909T>C (MAF: ALL=0.11%, AFR=0.40%) (rs186169069). In silico analysis of BRCA1 c.80+909T>C using Alamut visual v2.10 (including MES, NNSPLICE, GeneSplicer, Human Splicing Finder and SSF tools) predicted the activation of a pre-existing donor site. For BRCA1 c.4185+4105C>T, only SSF-like predicted the activation of a pre-existing atypical GC donor site (wild-type=75.9 vs variant=78.5), and the remaining tools predicted the creation of a de novo donor site, probably due to their inability to detect GC sites (online supplementary figure 6). The Alamut’s algorithm that uses SSF-like, NNSPLICE and MES to predict a local splice effect, defined this variant as creating a new donor splice site (table 1). To our knowledge, BRCA1 c.4185+4105C>T variant is not present in genetic databases Leiden Open Variation Database (LOVD), Breast Cancer Information Core (BIC), BRCA Share, Human Gene Mutation Database (HGMD) and ClinVar, as of April 2018. Moreover, it is not present in gnomAD and it has not been reported before in the literature.
Characterisation of BRCA1 c.4185+4105C>T splicing effect in patient RNA
RT-PCR experiments were performed covering exons 11–13 (275 bp) in variant carrier and 10 control samples. Experiments were performed in duplicate with mRNA from variant carrier drawn at two different time-points. Products visualised in QIAxcel instrument revealed an extra band in patient sample at ~400 bp (figure 2A). Sanger sequencing confirmed the insertion of a pseudoexon, consisting of 114 nucleotides from intron 12 (figure 2C). To determine whether this transcript could be a minor alternative BRCA1 isoform, we performed high-sensitivity CE of fluorescent amplicons in variant carrier, 20 blood and 4 breast tissue samples from healthy controls. The novel transcript was only present in variant carrier and was detected in the two mRNA extractions (figure 3).
Capillary Electrophoresis (CE) assays in carrier and controls. CE of FAM-labelled amplicons from variant carrier, five blood controls and four normal breast tissues. Results from two different RNA extractions in variant carrier are shown. Full-length transcript and the minor alternative isoform ∆13 p were detected in all samples, whereas aberrant transcript ▼12A was only detected in variant carrier. FL, full-length.
In vitro results were concordant with in silico predictions, indicating that BRCA1 c.4185+4105C>T variant converts a pre-existing GC site into a strong GT donor site that, together with an upstream cryptic acceptor site (online supplementary figure 6), promotes the inclusion of a pseudoexon between exons 12 and 13. We annotated this new transcript as ▼12A (r.4185_4186ins4185+3990_4185+4103), which is predicted to introduce four new amino acids and a stop codon, generating a truncated BRCA1 protein (p.Gln1395_Gln1396insSerLysSerLeu*) (figure 2C).
BRCA1 global expression was measured in two independent qPCR assays using probes located in exons 5–6 and exons 23–24, respectively. Results showed a notable reduction (>2-fold) of BRCA1 expression levels in carrier compared with controls (figure 4). This reduction was similar to a known BRCA1 pathogenic splicing variant c.302–1G>A, used as positive control.
BRCA1 expression in carrier and controls by qPCR. BRCA1 global expression was measured in two independent qPCR assays, using two probes targeting exons 5–6 and 23–24 junctions, respectively. Both assays show a reduction of BRCA1 levels in c.4185+4105C>T carrier compared with controls (C1–C10). Grid lines represent BRCA1 expression levels in controls (mean=1.2). Variant carrier and a pathogenic splicing variant c.302–1G>A show similar BRCA1 levels. Mean±95% CI are shown.
BRCA1 reference FL transcript specific assessment was determined by qualitative, semiquantitative and quantitative approaches. Qualitative allele-specific analysis was performed by FL specific amplification and posterior Sanger sequencing of two exonic polymorphisms (rs1060915, rs1799966) known to be heterozygous at DNA level in variant carrier. Visual examination of RT-PCR products by CE detected less amplification of FL transcript in variant carrier, and visual inspection of Sanger peaks at polymorphic positions showed main contribution from only one allele (figure 5A). Semiquantitative measurement of FL transcript using QIAxcel CE data showed a 2-fold reduction of FL levels in carrier sample compared with controls (figure 5B), suggesting that variant allele does not produce normal transcript. Accordingly, specific amplification of FL transcript by qPCR using a probe targeting exons 12–13 junction also showed a significant reduction of FL levels (figure 5C). Consistent with data obtained from qualitative and semiquantitative experiments, these data indicate that the variant allele is not generating normal transcript.
Specific assessment of BRCA1 full-length transcript. (A) Specific amplification of FL transcript by RT-PCR shows less amplification in carrier compared with controls. Sanger inspection shows main contribution to FL from only one allele. (B) FL levels measured by semiquantitative CE experiments show a >2-fold reduction in variant carrier. Grid line represents mean of normalised FL levels in control group (y=1). Mean±SEM are shown. (C) FL transcript measurement by qPCR using a Taqman probe targeting exons 12–13 junction. FL levels show a >2-fold reduction in variant carrier. BRCA1 pathogenic variants c.302–1G>A and exon1-13 deletion were used as positive controls (in red). Mean±95% CI are shown. (D) RT-PCR evaluation of BRCA1 and BRCA2 deep intronic variants and semiquantitative measurement by QIAxcel electrophoresis of FL transcript. Variant carriers were analysed in parallel with non-carrier controls (10 for BRCA1 and 20 for BRCA2). Full-length levels were not drastically reduced in the BRCA2 carrier, compared with BRCA1. Grid line represents mean of normalised FL levels in control group (y=1). Mean ±SEM are shown. B, blank; FL, full-length transcript.
Furthermore, we compared FL levels between BRCA1 c.4185+4105C>T and the BRCA2 c.6937+594T>G deep intronic variant reported to alter splicing by Anczuków and colleagues.16 Semiquantitative CE data from QIAxcel showed lower FL levels in BRCA1 c.4185+4105C>T carrier (mean=0.27) compared with BRCA2 c.6937+594T>G carrier (mean=0.74) (figure 5D).
Variant genotyping in Spanish HBOC families and controls
A total of 1030 index cases from Spanish HBOC families testing negative for BRCA1/2 and 327 Spanish controls were genotyped at BRCA1 c.4185+4105C>T position. The variant was not identified in any additional family or control, suggesting that this variant is a very rare event (online supplementary figure 7). Moreover, this variant is not reported in 1000 genomes database which includes a set of 165 Spanish controls (77 females and 88 males) and has not been reported in gnomAD which includes whole-genome data from ≈15 500 unrelated individuals.
Discussion
The aim of this study was to identify novel germline BRCA1/2 variants in deep intronic regions that could explain hereditary predisposition to BC/OC in high-risk families with uninformative BRCA1/2 test results. The analysis of 192 high-risk HBOC families by targeted gene sequencing identified 28% of patients carrying rare (MAF <0.005) BRCA1/2 deep intronic variants located beyond positions +20/–20 from canonical splice sites, with indicative in silico predictions of altering splicing. Overall, our results ruled out the presence of predominant splicing alterations occurring in variant carriers, indicating a low specificity for in silico tools used in this study. Only BRCA1 c.4185+4105C>T variant was correctly predicted and its effect was confirmed in patient RNA, producing a novel frameshift transcript ▼12A due to the activation of a cryptic donor site (figure 2A–C). High-resolution CE did not identify this transcript in control samples (blood and normal breast tissue) (figure 3), and it has not been reported in previous RNA studies.24 25 The presence of this transcript was confirmed in a second RNA extraction, indicating that ▼12A is a true splicing event and it is not a product of illegitimate splicing occurring due to technical artefacts, such as blood processing or blood ageing.26 27 This transcript introduces a premature stop codon (PTC) 99nt upstream of the next exon-exon junction (pseudoexon-exon13 junction), making the transcript highly likely to be degraded by the NMD system. This mechanism by which PTC transcripts are detected and degraded within cells has been well described in eukaryotes,28–30 and it is considered a surveillance mechanism to target aberrant mRNAs that would lead to the synthesis of proteins with deleterious effects for the organism. The activity of NMD has been well documented for BRCA1 PTC transcripts, showing that NMD is triggered by the majority of BRCA1 PTC mutations, resulting in a 1.5-fold to 5-fold reduction in mRNA abundance.31 We quantified global BRCA1 expression in variant carrier by quantitative PCR, obtaining an indirect measure of NMD activity. We used two different Taqman assays and both showed a >2-fold decrease of BRCA1 mRNA levels in carrier compared with controls (figure 4), indicating the degradation of transcripts produced by the variant allele and leading to a state of haploinsufficiency. Another approach indicating transcript degradation was the examination of heterozygous BRCA1 exonic positions in variant carrier, which showed differential allelic expression (figure 2B).
Nucleotide conservation analysis across 100 vertebrate species using PyloP, PhastCons and Multiz Alignment tools, showed less conservation in the pseudoexon from BRCA1 ▼12A transcript compared with exons 12 and 13 (online supplementary figure 8), suggesting a non-functional role for this region. Furthermore, nucleotide conservation comparison between the pseudoexon included in the aberrant transcript BRCA1 ▼12A and the pseudoexon included in the alternative isoform BRCA1 ▼13A, indicated a higher conservation in the alternative isoform (online supplementary figure 9). The alternative transcript ▼13A has been detected in human control blood tissue and breast tissue,24 32 whereas ▼12A has not been described previously.
BRCA1 c.4185+4105C>T variant was identified in co-occurrence with BRCA1 c.80+909T>C, located in intron 2. A recent study analysing the functional impact of non-coding BRCA1 variants remarked the importance of this intron because it contains regulatory regions that may affect BRCA1 promoter activity.33 In our study, we could not formally exclude that this variant contributes to the allelic imbalance observed in the carrier. However, this variant is located outside the non-coding regulatory sequences from intron 2 (CNS-1 and CNS-2), known to alter transcriptional activity when mutated,34 supporting a non-functional role for this variant.
A systematic BRCA1/2 RNA analysis in patients with HBOC in the French population identified the first BRCA2 deep intronic variant (c.6937+594T>G) causing the inclusion of a pseudoexon by activation of a cryptic donor site.16 Authors identified this variant in eight additional HBOC families and indicated a pathogenic role for the variant. However, a recent study based on case-control analysis did not observe an association between BRCA2 c.6937+594T>G and BC risk,35 and the variant has been classified as benign by the Evidence-based Network for the Interpretation of Germline Mutant Alleles (ENIGMA) expert panel (https://enigmaconsortium.org/). The differences observed in FL levels from BRCA1 c.4185+4105C>T and BRCA2 c.6937+594T>G carriers measured by semiquantitative CE, suggested that BRCA2 c.6937+594G allele still generates FL transcripts (figure 5D), in agreement with its clinical classification as benign. Although these results need to be confirmed with quantitative approaches and allele-specific assays, semiquantitative measurement of FL transcripts could serve as indicative of variant pathogenicity. The frequency of BRCA1 c.4185+4105C>T variant in Spanish HBOC cases and control population was also investigated to collect more clues about its pathogenicity, and genotyping analysis did not identify additional carriers in either group. Although variant absence in a larger control group (>1000 individuals) is required to consider evidences of moderate pathogenicity,36 there is a general assumption that high-penetrance disease-causative variants occur at very low population frequencies. This, combined with the clinical phenotype from variant carrier (BC+OC), supports pathogenicity for BRCA1 c.4185+4105C>T.
Additionally, the analysis by targeted sequencing of other BC/OC susceptibility genes in BRCA1 c.4185+4105C>T carrier, including ATM, BRIP1, CHEK2, EPCAM, MLH1, MSH2, MSH6, PALB2, PMS2, PTEN, RAD51C, RAD51D, STK11 and TP53, did not identify any deleterious variant that could explain the family phenotype,19 supporting also a pathogenic role for the BRCA1 deep intronic variant.
According to the ENIGMA guidelines for BRCA1/2 variant classification, when the variant allele is assessed by specific transcriptional assays and reveals only the expression of aberrant transcripts, the variant should be classified as pathogenic. In our case, although semiquantitative and quantitative analyses supported a monoallelic contribution to FL transcript expression (figure 5B–C), a residual signal (≈5%–10%) from variant allele was observed at heterozygous sites after the specific amplification of FL transcript (figure 5A). However, whether this is a true contribution from variant allele or a PCR artefact caused by unspecific primer hybridisation cannot be concluded from our results. In any case, the production of functional BRCA1 protein from the variant allele would be lower than the rescue threshold (≈30%) proposed in de la Hoya work,37 supporting a likely pathogenic role for our variant. Taking this into account and from an analytical point of view, the variant should be classified as likely pathogenic (Class-4). Further evidence from segregation analysis within family relatives and loss of heterozygosity analysis in tumour samples could help to unequivocally define a pathogenic classification, but unfortunately no samples were available.
In clinical laboratories, deep intronic regions of HBOC genes are generally not screened, and the frequency of pathogenic mutations in these regions could account for a proportion of HBOC cases by either affecting splicing, transcriptional activation or mRNA stability.38 Pseudoexon insertion events directly related to cancer pathologies caused by the creation of new splicing donor or acceptor sites represent the more frequent occurrence of this type of mutational events.15 The identification of pseudoexons is particularly interesting for the design of novel therapeutic molecules based on RNA biology.39 The use of therapies based on antisense oligonucleotides has shown to prevent the inclusion of cryptic exons by blocking the binding of trans-splicing regulatory factors to mutant 5’ or 3’ splice sites and restore the correct reading frame.16 40 41
Here, we report the first deep intronic mutation occurring at BRCA1 locus that promotes the inclusion of a pseudoexon in mature mRNA and is associated with HBOC risk. Although the frequency of pseudoexon events caused by spliceogenic variants in BRCA1/2 deep intronic regions appears to be low in our population, our findings highlight the relevance of integrating massive sequencing of whole genomic regions of HBOC genes and RNA analysis to complement genetic diagnosis of familial breast and ovarian cancers.
Acknowledgments
The authors acknowledge the Cellex Foundation for providing research facilities and equipment. We also thank Leo Judkins for English language editing.
References
Footnotes
Contributors GM, OD and SG-E designed the study. GM conducted RNA experiments and drafted and edited the manuscript. GM, SB, AT, AG-B and VB performed experiments and procedures. AM-F performed bioinformatics analysis. EC, AL-F, NS and JB provided samples and patient data. OD and SG-E supervised experiments. All authors read and reviewed the manuscript.
Funding This work was supported by Spanish Instituto de Salud Carlos III (ISCIII) funding, an initiative of the Spanish Ministry of Economy and Innovation partially supported by European Regional Development FEDER Funds: FIS PI12/02585 and PI15/00355 (to O Diez), PI13/01711 and PI16/01218 (to SG-E). SG-E and SB are supported by the Miguel Servet Program (CP16/00034) and Asociación Española Contra el Cáncer (AECC) contract, respectively.
Competing interests None declared.
Patient consent Not required.
Ethics approval Clinical Research Ethics Committee (CEIC), Vall d’Hebron Research Institute (VHIR).
Provenance and peer review Not commissioned; internally peer reviewed.