Introduction

Breast cancer susceptibility genes 1 and 2 (BRCA1 and BRCA2) are tumour suppressor genes, exerting their functions by contributing to DNA repair mechanisms and transcriptional regulation in response to DNA damage.1 Loss-of-function variants in one of these genes trigger the accumulation of genetic alterations and contribute to cancer development. These genes have been characterised as predisposition genes for the development of hereditary breast and ovarian cancers mainly (HBOC syndrome) and also other cancer types such as pancreatic and prostate cancers.2 3 The lifetime risk of developing a cancer for a carrier of a pathogenic BRCA1 variant by the age of 80 is about 72% for breast cancer and 44% for ovarian cancer.4 5 Since loss of function of BRCA1 increases cancer risk, every type of genomic inactivating alterations has been described in this gene, including intragenic duplication.6 However, the identification of an intragenic duplication in the BRCA1 gene with classical molecular techniques, such as short-read sequencing, does not allow the differentiation between: (1) a tandem duplication that would be classified as pathogenic if it induces a premature STOP codon or if it were localised within a functional domain that would be destabilised and (2) an insertion of the extra copy somewhere else in the genome that would be classified of unknown significance. This distinction has a critical impact on the patient’s management and guides the choice for the type of surgery and the type of treatment (PARP inhibitor therapy, chemotherapy, etc). Hence, a technology enabling the identification of tandem intragenic duplication at low-cost and high-speed would be of great interest in clinical routine molecular biology. The interest in nanopore sequencing is growing in the field of molecular pathology, notably because of its capacity to precisely resolve structural variants. Recent work showed that nanopore sequencing can rapidly and accurately provide genetic diagnoses of Mendelian diseases.7 Here, we accurately describe a germline tandem duplication of exons 18–20 of BRCA1 using Oxford Nanopore Sequencing Technology, with adaptive sampling target enrichment. This allowed us to better characterise and classify this structural variant (SV) as pathogenic in a short timeframe.

Patient and methodsPatient and family history

A woman under age 61 years was referred for a genetics consultation because of a triple negative (TN) ductal carcinoma of the breast. Her medical records reported among others tobacco and alcohol consumptions and an ectopic pregnancy. Family was not suggestive of HBOC syndrome. She had two healthy sons under age 25 years. Her two siblings (one brother and one sister, both under age 61 years) were cancer-free. Her father died from cirrhosis before age 60 years; no other information concerning the paternal side of the family was available. The patient’s mother was cancer free after age 61 years. A maternal aunt developed a breast cancer before age 61 years, a maternal uncle died from a lung cancer, and a maternal cousin was diagnosed with a colorectal adenocarcinoma before age 40 years. The patient’s maternal grandfather was diagnosed before age 61 years with a pharyngeal carcinoma.

As the patient was affected with a TN breast cancer before age 61 years, a HBOC multigene panel was prescribed whose results would have a direct and short-term impact on the patient’s management. The patient gave written informed consent for genetic testing and research studies.

Initial molecular analysis

Germline DNA was extracted from blood cells using the QiaSymphony DSP DNA Midikit according to the manufacturer’s instructions. Genic regions of interest were enriched with an Agilent custom SureSelect QXT kit. Short-read next generation sequencing (NGS) was performed on an Illumina NextSeq 500. Bowtie2, VarScan2 and DESeq were used for read mapping on the hg19/GRCh37 reference genome, variant calling and copy number variation (CNV) detection, respectively. Data analysis was restricted to 13 high-penetrance HBOC genes, including BRCA1 (NM_007294.3), according to national guidelines.8 Germline CNV were confirmed by Multiplex Ligation Probe Amplification (MLPA, MRC Holland probe mix P002-D1) on DNA extracted from a buccal swab.

Nanopore sequencing and adaptive sampling

Library preparation was performed in 2 hours using Oxford Nanopore Technologies Ligation Sequencing Kit SQK-LSK110 on 2 µg of genomic germline DNA. Half of the library was then injected into a Minion Flow Cell R9 for 24 hours. After 24 hours of sequencing, the flow cell was washed with a nuclease mix (Oxford Nanopore technology Flow Cell Wash Kit EXP-WSH004), and the second half of the library was injected and sequenced for additional 24 hours. Adaptive sequencing enables target enrichment without additional library preparation steps;9 we targeted regions encompassing the entire loci (including 5’-UTR, 3’-UTR and intronic regions) and 5 kb flanking regions upstream and downstream of 120 genes, including BRCA1, totalling 49 Mb (1.58% of the human genome). Downstream bioinformatic analysis was performed using NanoCliD, a custom bioinformatic pipeline (https://github.com/InstituteCurieClinicalBioinformatics/NanoCliD); NanoCliD relies on guppy for basecalling, minimap210 for alignment, Clair311 for Single Nucleotide Variation (SNV) calling and SVIM,12 Sniffles,13 cuteSV14 and NaNovar15 for SV calling. The data processing toolkit called ‘guppy’ allowed to convert FAST5 files to FASTQ files. Alignments were manually reviewed on Integrative Genomic Viewer.

The breakpoints of the BRCA1 exons 18–20 duplication were subsequently confirmed by Sanger sequencing on genomic DNA, using the forward primer in BRCA1 intron 20 5’-TCGGAAGGCTGAGTTGAGAG-3’ and the reverse primer in BRCA1 intron 17 5’-TCCCAGTGTTTCAAAGGCCC-3’.

Results

The HBOC NGS panel showed the presence of a heterozygous germline duplication encompassing BRCA1 exons 18–20, subsequently confirmed by MLPA on an independent sample. No other variant of interest was detected. Based on the American College of Medical Genetics (ACMG) classification, this variant was located in a functional domain (PM1) and the exons 18–20 duplication had been reported in nine French families, without cosegregation data within the same family (PP1).16 The BRCA1 exonic duplication was therefore classified as a ‘variant of unknown significance’ (class 3), as NGS data analysis could not demonstrate that the reading frame of the BRCA1 transcript was altered (ie, that the event was a tandem duplication). Further cDNA analysis was required to confirm pathogenicity. However, RNA is not routinely available and the technique is very time-consuming (ie, requiring ~2 months for analysis).

In the meantime, we were able to confirm the presence of the duplication and demonstrate that it was indeed a tandem duplication in less than 10 days using nanopore sequencing with adaptive sampling (figure 1). The mean genomic depth of coverage was 2.77× whereas it was about 10 times higher for the targeted regions with a mean coverage of 22.55×. The mean depth of coverage over the BRCA1 locus was 24.63× and eight reads encompassed the breakpoints of the SV. The mean read length was 3.7 Kb. The precise breakpoint coordinates (Chr 17 (NC_000017.10:g.41,208,234_41,216,908dup ; NM_007300.4:c.5138–940_5340+835 dup) collocated with Alu repetitive elements in intron 17 (AluY ; chr17:41,216,845–41,217,145) and intron 20 (AluJb ; chr17:41,208,192–41,208,495) of the BRCA1 gene. This BRCA1 tandem duplication of exons 18–20 alters the reading frame and introduces a premature stop codon. Sanger sequencing (figure 1) subsequently confirmed for clinical purposes the breakpoint, allowing a fast return of results to the patient. The demonstration of the tandem nature of this duplication and the disruption of the reading frame allowing the addition of the PVS1 argument of the ACMG classification,17 the duplication was therefore classified as pathogenic (class 5) within a timeframe compatible with optimal patient management; with this new classification, the patient was now eligible for surgery, and clinical doctors could discuss an annexectomy and complete mastectomy.

Figure 1

Duplication of exons 18–20 of BRCA1 gene. Long-read sequences generated by nanopore sequencing visualised in IGV indicate the presence of the tandem duplication of exons 18–20 of BRCA1 in our patient with hereditary breast and ovarian cancer syndrome. The grey reads represent all the reads generated for the patient’s DNA at this localisation and the coloured reads above represent the reads spanning the duplication. Sanger sequencing chromatogram on the upper panel allows to precisely determine the sequence of the breakpoint: the normal sequence of intron 20 of BRCA1 is framed in blue and the following sequence framed in red is abnormal and represents a sequence of intron 17 of BRCA1 hence confirming the tandem duplication of exons 18–20. IGV, Integrative Genomic Viewer.

Discussion

Genomic SV include deletions, duplications, inversions and triplications of varying sizes and are an important cause of genetic diseases, including SV affecting the BRCA1 gene.18 19 CNV can indeed contribute to human genetic diseases by either influencing the copy number of dosage-sensitive genes or disturbing the gene sequence as a result of intragenic CNV.

The development of microarray technologies, such as Comparative Genomic Hybridization (CGH) and Single Nucleotide Polymorphism (SNP) arrays, and their implementation in the genetic laboratories since the 2000s have progressively supplanted the karyotyping technology with copy number profiling, thus improving the resolution and hence the identification of small CNV of few Kb. More recently, the application of bioinformatic tools to short-read NGS data has emerged as another powerful method to identify SV including fine intragenic CNV that previously remained beyond the resolution limit of conventional microarrays.

A study led on 184 germline duplications throughout the genome found that the most frequent mechanism is tandem duplication with direct orientation (83% in that study), while others result in the creation of gene fusion at breakpoints, triplications or adjacent duplications, insertional translocations and complex rearrangements.20 21 Intragenic duplications might consequently cause a loss of function if the duplicated exons are organised head-to-tail and either not in-frame, of a large size or in a functional domain of importance in the protein. However, this type of SV can be indistinguishable in short-read NGS data from an insertion of an extra copy somewhere else in the genome. In this case, the SV might not disrupt the gene and might not be pathogenic because the extra copy is not supposed to disrupt the sequence of the gene.

Therefore, interpreting the pathogenic consequences of a duplication ideally requires breakpoint-level analysis as the subsequent clinical management of the patients may be totally different depending on the variant classification.6 18 19 The identification of a duplication in the BRCA1 gene with classic molecular techniques such as short-read sequencing techniques does not differentiate between a tandem duplication that would be classified as pathogenic or a duplication of unknown significance. Usually, in our laboratory, we would have investigated the pathogenicity of such a SV by a cDNA analysis. Yet, such analysis requires RNA, is very time-consuming and the average turn-around time is 2 months.

As an alternative, nanopore whole genome sequencing (WGS) can rapidly and accurately provide genetic diagnoses of mendelian diseases.7 Long reads do not suffer from the same mapping issues as short reads in repetitive regions and permit the detection of SV at a base-pair resolution. However, although WGS provides a more comprehensive genomic analysis, an advantage of targeted sequencing is its greater depth of coverage in regions of interest and lower cost. In the clinical setting, a gene panel approach restricting the analysis to only actionable genes at greater depth of coverage (providing more confidence in the results) is common. Interestingly, a highly flexible targeted approach, called adaptive sampling, has recently been developed for nanopore sequencing.9 22 This strictly computational method targets the sequencing of specific regions of interest with nanopore technology without the need of effortful nucleic acid preparation. During the sequencing steps, fragments of interest are selected in real-time as they migrate through proteic nanopores by converting the electrical current into nucleotide sequence and comparing it to the provided regions of interest. Off-target DNA fragments are ejected from the pores, preserving sequencing capacity for DNA fragments of interest from on-target regions. Hence, by targeting genomic regions of interests, adaptive sequencing can enrich the sequenced data and improve the read depth yield. In our case, we were able to reach a depth of coverage 10 times higher in the targeted genomic regions and to observe a tandem duplication of exons 18–20 of BRCA1 gene. Although the 24× depth obtained by Adaptive sequencing is lower than those possibly obtained by choosing a Crispr enrichment approach, the alteration was unquestionably visible on the long reads. Moreover, the adaptive approach is extremely flexible since the targeted genes can be adapted by a simple change of the bed file, without designing new guides as is the case with the Crispr sequencing. In addition, this method allowed us to precisely identify the SV breakpoints, located in two Alu repetitive elements sharing 74% of identity. This result supports the hypothesis that this SV was mediated by non-allelic homologous recombination, ultimately gleaning further information about a possible mechanism leading to the SV of the BRCA1 gene.

Conventional methods of enrichment typically involve cumbersome and expensive DNA processing steps prior to sequencing such as PCR-based amplification, hybridisation capture or Cas-mediated enrichment. By computationally selecting molecules during the sequencing process, these steps are avoided, making the process simpler, faster and easily adaptable. The possibilities offered by adaptive sampling contribute to the current trend of continuous improvement in applications of molecular biology for human diseases.

Nanopore sequencing coupled with adaptive sampling was demonstrated to be an effective, reliable and fast long-read sequencing technique. In the case reported here, it enabled the accurate resolution of an intragenic duplications of BRCA1 and its classification as a pathogenic variant—ultimately guiding the clinician’s decision and thus improving the clinical management of the patient and her relatives. This result obtained on one case is promising, and protocol optimisations as well as studies on more cases are to be conducted in order to improve yields and enhance the potential of nanopore sequencing.

Ethics statementsPatient consent for publication

Consent obtained directly from patient(s)

Ethics approval

Not applicable.

Acknowledgments

The author would like to thank Trenton Dailey-Chwalibòg for providing writing support and proofreading the manuscript.