A cryptic BAP1 splice mutation in a family with uveal and cutaneous melanoma, and paraganglioma

Summary

Inactivating germ line BRCA1-associated protein-1 (BAP1) mutations have recently been reported in families with uveal or cutaneous malignant melanoma (UMM, CMM), mesothelioma, and meningioma. Although apparently predisposing to a wide range of tumors, the exact tumor spectrum associated with germ line BAP1 mutations has yet to be established. Here, we report a novel germ line BAP1 splice mutation, c.1708C>G (p.Leu570fs*40), in a multiple-case Danish UMM family with a spectrum of other tumors. Whole-exome sequencing identified an apparent missense mutation of BAP1 in UMM, CMM, as well as paraganglioma, breast cancer, and suspected mesothelioma cases in the family. Bioinformatic analysis and splicing assays demonstrated that this mutation creates a strong cryptic splice donor, resulting in aberrant splicing and a truncating frameshift of the BAP1 transcript. Somatic loss of the wild-type allele was also confirmed in the UMM and paraganglioma tumors. Our findings further support BAP1 as a melanoma susceptibility gene and extend the potential predisposition spectrum to paraganglioma.

Introduction

The report of a germ line inactivating mutation of the BRCA1-associated protein-1 (BAP1) in a patient with uveal malignant melanoma (UMM) as a part of the initial discovery of frequent somatic BAP1 mutations in UMM suggested a role for BAP1 in cancer predisposition (Harbour et al., 2010). Consistent with this finding, Wiesner et al. (2011) recently identified germ line BAP1 mutations in two families with a syndrome characterized by multiple skin-colored elevated melanocytic tumors, with some individuals also developing UMM or cutaneous malignant melanoma (CMM). At the same time, Testa et al. (2011) reported germ line BAP1 mutations in two families with at least five patients with mesothelioma, as well as individuals diagnosed with other cancers including UMM. More recently, Abdel-Rahman et al. (2011) described a UMM family with cases of CMM, meningioma, and lung adenocarcinoma carrying a BAP1 mutation. Similarly, Njauw et al. (2012) reported two families with BAP1 mutations co-segregating with both UMM and CMM, as well as four other mutations in sporadic UMM and/or CMM cases. All 14 of the germ line BAP1 mutations reported to date are predicted to result in protein truncation via nonsense mutation, frameshift caused by insertions/deletions, or alteration of a canonical dinucleotide splice donor/acceptor sequences. Among them, two families described by Wiesner et al. (2011) report skin-colored melanocytic tumors in all mutation carriers, while carriers from one family and another sporadic case described by Njauw display a similar but distinct nevoid melanoma-like melanocytic proliferations. These observations collectively suggest that BAP1 predisposes to melanocytic tumors as well as a wide spectrum of cancer.

Here, we report a Danish family with multiple UMM and suspected mesothelioma cases, as well as several other cancers including CMM, breast cancer, and paraganglioma, carrying an apparent missense mutation of BAP1 resulting in the creation of a strong cryptic splice donor, aberrant splicing, and a truncating frameshift of the BAP1 transcript. Our findings extend the predisposition spectrum of BAP1 to paraganglioma and highlight the need for rigorous functional and bioinformatic assessment of apparently non-severe coding mutations in BAP1.

We identified a Danish family (Fig. 1a) with three patients with UMM (III-1, III-8, and IV-2), as well as family members with CMM (IV-3), paraganglioma (III-3), breast cancer and unconfirmed peritoneal mesothelioma (II-4), prolactinoma (IV-4), and both malignant fibrous histiocytoma (MFH) and suspected mesothelioma (III-12). Notably, the CMM case displayed an atypical clinical tumor phenotype with a reddish, flat to dome-shaped nodule surrounded by a halo (Fig. 1b, Fig. S1). Halos usually develop around brown-, blue-, or black-pigmented regressing cutaneous metastases and very seldom around regressing primary melanoma. Histology of this reddish primary melanoma does not reveal signs of regression within the tumor. Interestingly, the non-pigmented reddish color and elevated shape of this CMM present similar qualities to the skin lesions described by Wiesner et al. (2011) (Fig. 1b). However, unlike the multiple skin lesions present on these previously reported BAP1 mutation carriers, such lesions were not evident in the present family based on skin examination of five available members (III-1, IV-2, IV-3, IV-5, and III-5, all confirmed to be BAP1 mutation carriers). Additionally, one of the UMM cases (IV-2) is characterized by unusually early age of onset (18 years), a rare (~5% of cases) diffuse multi-centric pathology (Fig. S2) (Font et al., 1968), and lack of metastasis (16 years followup). The histology of the other two UMM cases (III-1 and III-8) was not evaluable, and both patients developed metastases shortly after diagnosis. A clinical description of this family can be found in the Supporting information.

Figure 1.

A novel BAP1 mutation segregates in a multi-cancer family. (a) Pedigree structure. Solid black circle represents cutaneous malignant melanoma (CMM), three-quarters filled shapes represent uveal malignant melanoma, and half-filled shapes other cancer [Pul, pulmonary adenocarcinoma; Pgn, paraganglioma; Plc, prolactinoma; MFH, malignant fibrous histiocytoma; Brs, breast cancer; (Mes), probable mesothelioma]. c.1708C>G, (c.1708C>G), or WT: confirmed mutant, inferred mutant, or wild-type BAP1 genotype, respectively. (b) A patient with CMM (IV-3) with the germ line BAP1 mutation displays a distinct tumor phenotype. The red papule is about 3 mm in diameter, and the surrounding halo extends to 9 mm.

We performed whole-exome sequencing of three members of the family (III-1, IV-2, and IV-3) using an Illumina TruSeq Exome Enrichment Kit and an Illumina HiSeq2000 Sequencer. We first filtered out changes that were not present in all affected individuals, leaving 5381 protein-changing sequence variants. We subsequently filtered out those variants present in the 1000 Genomes database (2010 November release) and dbSNP (build132), leaving 47 novel variants for follow-up. Among these was a single non-synonymous variant in BAP1 (c.1708C>G, p.Leu570Val), which was also not listed in the sequence derived from approximately 6000 additional human chromosomes annotated by the NHLBI exome project (http://evs.gs.washington.edu/EVS/) or Kaviar (http://db.systemsbiology.net/kaviar/cgi-pub/Kaviar.pl).

While a conservative leucine to valine change is not consistent with previously reported germ line BAP1 mutations, which were all predicted to result in truncated protein (Abdel-Rahman et al., 2011; Njauw et al., 2012; Testa et al., 2011; Wiesner et al., 2011), analysis of the potential effects of the nucleotide change on splicing using the information theory methodology of Nalla and Rogan (2005), Rogan et al. (1998; https://splice.uwo.ca/) suggested that the variant creates a strong cryptic splice donor site. This cryptic donor is predicted to have a higher binding affinity than the natural donor (8.8 Ri compared to 8.1 Ri), which is 21 bases 3’ of the variant. The predicted incorrectly spliced mRNA would be missing 22 bp from the end of exon 13 and would result in a frameshift and truncation 121 amino acids in advance of the normal stop codon (p.Leu570fs*40). We subsequently screened for this mutation in all family members for whom DNA was available, including the three individuals who had been subjected to whole-exome analysis (Fig. 2a), and were able to infer obligate carrier status for others (Fig. 1a, obligate carriers status listed in parentheses). The c.1708C>G mutation was present in all UMM (III-1, IV-2, and III-8) and CMM (IV-3) cases, the patient with paraganglioma (III-3), the case with both breast cancer and unconfirmed mesothelioma (II-4), the individual with both MFH and a suspected mesothelioma diagnosis (III-12), the case with pulmonary adenocarcinoma (II-1), the unaffected members who volunteered for presymptomatic testing (II-6, III-5, and IV-5), but not in the patient with prolactinoma (IV-4).

Figure 2.

An apparent missense mutation results in aberrant splicing of BAP1 transcript. (a) Sequence trace of the region encompassing BAP1 c.1708C>G from a wild-type individual (WT) and a mutant carrier (c.1708C>G). (b) Splice variant analysis. RNA was extracted from whole blood of two patients with uveal malignant melanoma heterozygous for BAP1 c.1708C>G (III-1 and IV-2), and RT-PCR product was separated on a 3% agarose gel. Commercially available normal blood RNA was used as a control. Two different splice variants are depicted on the right. Upper band (124 bp) – wild-type transcript (WT), lower band (102 bp) – aberrantly spliced transcript (Mut), box – exon, solid line – intron, dashed line – splicing event, star – BAP1 c.1708C>G cryptic splice donor site, pink arrows – PCR primer annealing sites.

We next performed RT-PCR analysis to evaluate whether aberrant splicing of the BAP1 transcript occurs precisely as predicted in carriers of the c.1708C>G mutation. We amplified BAP1 from RNA isolated from whole blood drawn from two family members with UMM (III-1 and IV-2) using primers encompassing the exon 13–14 junction. Product from control individuals with only the wild-type (WT) BAP1 gene results in a single 124-bp band (Fig. 2b). As predicted, blood from individuals harboring the c.1708C>G variant yields both the 124-bp product as well as a second band 22 bp shorter, suggesting that the cryptic splice donor causes aberrant splicing. Sequencing of this shorter product confirmed that splicing utilized the cryptic donor and natural acceptor (Fig. S3b). Sequencing of the correctly spliced 124-bp product from c.1708C>G heterozygous individuals indicated no detectable trace of the c.1708C>G allele, suggesting that the majority of transcript from the mutant allele is improperly spliced (Fig. S3a).

To investigate whether this inactivating mutation is accompanied by somatic loss of the WT copy during tumor progression, we analyzed the BAP1 sequence of available cancer tissues. Loss of the WT allele was indeed observed in one UMM (IV-2) and the paraganglioma (III-3) tissue (Fig. S4), but not in pulmonary metastasis of MFH tissue (III-12). Another UMM case (III-1) displayed monosomy of chromosome 3, but the DNA was of insufficient quantity to genotype c.1708C>G. BAP1 immunohistochemistry was performed on available archival tumor tissues from carriers of this germ line BAP1 mutation but was not successful.

Our data are consistent with and further support the assertions made by recent studies implicating BAP1 as a melanoma and mesothelioma predisposition gene. Of the eight reported multi-cancer families harboring germ line inactivating BAP1 mutations, including the family reported here, three have both UMM and mesothelioma (or suspected mesothelioma). Our findings further extend the spectrum of the BAP1-related syndromes to include novel phenotypes. Notably, we identified a patient with paraganglioma carrying a germ line BAP1 mutation and showed that the WT BAP1 allele was lost somatically. While there were other patients with cancer in this family, specifically MFH and prolactinoma cases, neither patient displayed a germ line mutation accompanied by loss of the WT allele, consistent with a link between BAP1 and a specific set of cancer phenotypes. We suggest that BAP1 is truly a broad-spectrum tumor suppressor gene and that multi-cancer families and cases of apparently sporadic paragangliomas warrant screening for germ line BAP1 mutations.

In addition to providing additional support for and extending the BAP1 cancer predisposition spectrum, our data also highlight the need for careful informatic annotation of clinical sequencing data, including whole-genome and whole-exome studies. The mutation we observed initially appeared to be a novel conservative amino acid substitution using widely used informatic tools designed for annotation of ‘next-generation’ sequencing data. Specifically, this exonic mutation is located 21bp away from the natural splice donor, and thus would not be flagged as a potential deleterious splicing mutation by most annotation tools that focus on natural splice donor/acceptor positions. More careful analysis of the relatively tolerable mutation carried by the family reported here revealed the creation of a cryptic splice donor resulting in an aberrantly spliced BAP1 transcript. Our study suggests that thorough annotation should be employed for high-throughput clinical sequencing studies.

Supplementary Material

Supplemental methods figures

Data S1. Methods.

Figure S1. A CMM patient with the BAP1 cryptic splice mutation displays a distinct clinical phenotype.

Figure S2. Micrographs demonstrating the multi-centric localization of the uveal melanoma from patient IV-2 in hematoxylin-eosin staining.

Figure S3. BAP1 c.1708C>G creates a strong cryptic splice donor site and results in an aberrantly spliced transcript.

Figure S4. The wild-type BAP1 allele is lost in cancer tissues.

Significance.

Germ line BAP1 mutations have recently been identified in a few families with melanoma and mesothelioma. We report an apparent missense mutation of BAP1 resulting in aberrant splicing and a truncating frameshift in a family with melanoma, suspected mesothelioma, and paraganglioma. Our study suggests that BAP1 is truly a broad-spectrum tumor suppressor gene and also supports this gene as a target for mutation screening in multi-cancer families and paraganglioma cases. Further, these data highlight the need for rigorous bioinformatic assessment of apparently non-deleterious, novel coding mutations in whole-genome and whole-exome sequencing studies.