Germline large deletion of BAP1 and decreased expression in non-tumor choroid in uveal melanoma patients with high risk for inherited cancer

Abstract

Uveal melanoma (UM) is the most common phenotype in patients with germline BAP1 mutation. This study aimed to identify selection criteria for BAP1 germline testing and assessed the role of large deletion/duplication and epigenetic inactivation. 172 UM patients with high risk of hereditary cancer were included. Germline variants in BAP1 were assessed by direct sequencing and large deletion/duplication by multiplex ligation-dependent probe amplification. BAP1 expression in unaffected choroid tissue from a patient with UM was assessed by quantitative RT-PCR and methylation by pyrosequencing. Twenty-eight patients had one or more germline sequence variants in BAP1; 7 of these were pathogenic. 140 patients were assessed for large deletion/duplication and in one BAP1 whole gene deletion was detected. In total, eight patients (4.7%) had pathogenic alterations in BAP1 with the highest frequencies of in patients with a personal/family history of ≥ 2 BAP1-related cancers 6/16 (38%), age of onset < 35 years 4/21 (19%) and familial UM 6/34 (18%). One of 19 non-tumor choroid tissues tested showed uncharacteristically low expression as compared to the controls decrease in BAP1 RNA expression but no evidence of constitutional promotor hypermethylation was detected. UM patients with a strong personal or family history of cancers associated with BAP1, early age of onset and familial UM should be assessed for germline variants in BAP1, including large deletions.

INTRODUCTION

Uveal melanoma (UM) is the most common primary intraocular tumor in adults. The tumors originate from melanocytes in the iris, ciliary body, and choroid of the eye. Based on the personal and/or family history of the disease it has been suggested that about 12% of UM patients have a strong hereditary predisposition to cancer.¹

Germline mutation in BRCA1 associated protein 1 (BAP1) is associated with a Tumor Predisposition Syndrome (BAP1-TPDS, OMIM #614327) with at least four main cancers: UM, cutaneous melanoma (CM), mesothelioma (MMe), and renal cell carcinoma (RCC).^2–6 Somatic mutations in BAP1 are observed in many different cancers most commonly in UM.⁷ BAP1 germline mutations have been reported in 1.6–4% of all UM, and in up to 8% of metastatic UM, with the highest frequency, 22%, reported in patients with familial UM.^8–10 The vast majority of germline alterations reported in BAP1 have been single base pair changes and small insertions and deletions (INDELs) that could be readily identified by direct sequencing.¹¹ Large deletions make up a significant percentage of the somatic alterations in BAP1 in UM and other cancers such as MMe, but these have not been investigated as a mechanism of germline inactivation of BAP1 in UM. Large germline deletions are a common mechanism for inactivation of several genes including MSH2, BRCA1, BRCA2, VHL, and RB1.^12–17 In addition, germline epigenetic inactivation has been observed in few genes such as PTEN but has also not been investigated as a mechanism for BAP1 inactivation in UM.^18,19

The aim of this study was to assess the frequency of germline alterations in BAP1 in UM patients at high-risk for hereditary cancer and determine selection criteria for UM patients for germline testing.

MATERIALS AND METHODS

Ethical considerations

Approval for this project was obtained from the Institutional Review Board at The Ohio State University (2006C0045) and Cole Eye Institute. Informed consents were obtained prior to testing. This project was HIPAA compliant and in accordance with the tenets of the Declaration of Helsinki.

Patients

Patients in the study presented for care at either the Ophthalmology clinic at The Ohio State University or Cole Eye Institute, clinical cancer genetics clinic at The Ohio State University or were referred to our cancer genetic program from outside institutions. Our cohort was limited to UM patients with a high risk for hereditary cancer predisposition based on the following criteria: age of diagnosis of UM under 35 years, familial UM, personal and/or family history of cancer including the four main cancers associated with BAP1 TPDS, (Table 1). Family history included all first and second-degree relatives except for UM where patients with any family history of UM were eligible. Patients were accrued over a period of 15 years (2003–2018).

TABLE 1.

BAP1 pathogenic variants in UM patients at high risk for hereditary cancer

	Patients in category	Patients with pathogenic mutations/CNV
Total	172	8 (4.7%)
Sex   - Male   - Female	70 102	2 (2.9%) 6 (5.9%)
Age at diagnosis (average, range)	54.5 [0.3–87]	38.4 [16–67]
UM at young age ( <35) a	21	4 (19%)
Bilateral UM	1	0
Familial UM and additional hx of:   - No other cancers   - CM   - RCC   - MMe   - CM+RCC   - CM+MMe   - Other cancers	34 2 7 3 1 2 1 18	6 (18%) 0 0 1 (33%) 1 (100%) 2 (100%) 1 (100%) 1 (5.6%)
Personal or Family History of BAP1-TPDS cancer other than UMb   - CM   - RCC   -  CM+RCC	65 46 13 6	1 (1.5%) 0 0 1 (17%)
Personal or Family History of ≥ 2 BAP1-TPDS cancers	16	6 (38%)
Personal or Family History of non-BAP1-TPDS cancer	66	0

The four main cancers associated with BAP1-TPDS are UM, uveal melanoma; MMe, malignant mesothelioma; CM, cutaneous melanoma and RCC, renal cell carcinoma.

^aTwo of the patients who were diagnosed at young age had familial UM and one had family history of CM and RCC. The remaining patient was a 16 year girl with whole BAP1 gene deletion.

^bFirst and second-degree relatives only. No family history of MMe was reported in patients other than those with familial UM.

Germline mutation testing

Samples were sequenced for all coding exons of BAP1 and 229 base pairs of the 5’untranslated region (UTR) according to our previously described protocol.²⁰ The following three primer sets were used to sequence the 3’UTR (F1 ACATTCCTTCCATCGTGCCC, R1 TGGGACACCCTACTCCCAAC, F2 AGGTCCTTGTATCATGCCACG, R2: GCAACCCTGTCTCTGCTACC, F3: GTTCTAGGGCTCTTCGCCTTC and R3: AGCAACCACAGGAGGGTTCAT). Sequences were aligned per the reference sequence provided by GenBank accession number NM_004656.2. Sequencing results for coding regions of BAP1 in 138 of these patients as well as family histories of 32/34 of the familial UM patients have been previously reported by our group.^2,8,11,20,21

Assessment of germline large deletion and duplication in BAP1

All patients with no detected germline BAP1 sequence mutations were assessed for deletions and duplications utilizing multiplex ligation-dependent probe amplification (MLPA) analysis. Four patients had insufficient DNA for analysis. A probe mix of 40 probes covering all BAP1 gene exons and several internal control genes was used (SALSA MLPA P417 BAP1 probemix, MRC-Holland) according to the manufacturer’s protocol.

Briefly, after probe hybridization, multiplex PCR of all probes was carried out on GeneAmp 9700 (Applied Biosystems, Waltham, MA). The PCR products were analyzed by capillary electrophoresis on a 3730 DNA Analyzer (Applied Biosystems). Two normalization processes were applied according to the manufacturer’s protocol (www.mlpa.com > MLPA procedure). First, intra-sample normalization was performed by dividing the signal of each target-specific probe by the signal of every single reference probe in that sample. The inter-sample comparison was performed by dividing the normalization constant of each probe in a given sample by the average normalization constant of that probe in all the reference samples. Samples with significant variation in the control probes heights (quality control) were excluded from final analysis. Validation of the copy number variation was carried out using quantitative PCR for copy number assay (probe: Hs00275733_cn, copy number reference assay: 4401631) using the manufacturer’s protocol (Thermofisher Scientific, Waltham, MA).

Assessment of impact of intronic variants on BAP1 gene splicing

For intronic and synonymous variants we assessed the possible effect on splicing by multiple splicing prediction tools (Human Splicing Finder, Fruit Fly splice predictor (NNSPLICE) and NetGene2). The variant c.2057–4G>T within 4 base pairs from exon/intron boundary was further assessed by RT-PCR of RNA extracted from peripheral blood leukocytes and/or tumor using primers spanning the predicted spliced exons (5’-ACCCAAGGAGCTGCTGGC-3’ and 5’-CGTTTCCGCCGGTCAGGCTT −3’).

Assessment of BAP1 expression in non-tumor choroid tissue

TaqMan 5’ nuclease quantitative (real-time) PCR assays was carried-out using pre-developed assay( #Hs00184962_m1) from Applied Biosystems according to the manufacturer’s protocol using the 9700 ABI real-time PCR machine. The reactions were run in triplicates for each sample in 25 μl with a final dilution of 1x each of Taqman PCR universal master mix and 1x of the pre-developed probes. Only probes amplifying the cDNA and not the genomic DNA were selected. In addition to BAP1 an endogenous control GUSB was tested in separate reactions. The PCR reaction settings were 95°C for 3 min, then 40 to 50 cycles of 95°C for 15 sec, and 60°C for 1 min. The relative expression levels were assessed by the comparative CT method (ddCT) according to our previously published protocol.²²

Germline promotor methylation of BAP1

Screening for promotor methylation of BAP1 was carried out on 19 non-tumor choroids from patients with no detectable germline BAP1 pathogenic mutation and choroids from 6 cadaver eyes. High and low methylated DNAs were used as controls for efficiency of assay. The human BAP1 methylation assay (ADS1756FS1, EpigenDx, Hopkinton MA) was utilized to assess the promoter methylation of the BAP1 gene on 7 hyper methylated regions encompassing the +27 to −23 bases from transcription start site according to the manufacturer’s protocol. Bisulfite treatment of the genomic DNA was carried-out using EZ DNA Methylation-Gold kit (Zymo Research, Irvine, CA). Bisulfite-treated DNA was used as template for the PCR reaction and methylation status was assessed by pyrosequencing of the PCR product.

RESULTS

BAP1 single nucleotide variants and small INDEL detected in the cohort:

Direct sequencing of the 172 patients identified a total of 22 unique variants in 28 patients (Table 2). Based on the consensus recommendation of standards and guidelines for the interpretation of sequence variants of the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP)²³ seven of these variants were null (pathogenic), four were variants of uncertain significance (VUS), and 11 were benign/likely benign (Table 2). Five patients had more than one variant including three with two, one with four and one with six.

TABLE 2.

Summary of germline variants identified in our study

Variant	dbSNP #	Variant type	Allele frequency in UM, (%)	gnomADa	ClinVar Call	Variant type (ACMG)
c.2050C>T, p.Q684*b	-	nonsense	0.0029	0	Pathogenic	Pathogenic
c.1882_1885delTCAC, p. S628Pfs*8 b	-	frameshift	0.0029	0	NRc	Pathogenic
c.1717_1717delC, p.L573Wfs*3 b	-	frameshift	0.0029	0	Pathogenic	Pathogenic
c.1938T>A, p.Y646* b	-	nonsense	0.0029	0	NR	Pathogenic
c.458_459delCT, p.P153Rfs*7b	-	frameshift	0.0029	0	NR	Pathogenic
c. 799C>T, p. Gly267* b	rs387906849	nonsense	0.0029	0	Pathogenic	Pathogenic
c.487G>A, p W5*	-	nonsense	0.0029	0	NR	Pathogenic
Whole gene deletion	-	CNV	0.0029	0	NR	Pathogenic
c.68–28C>T	rs374877579	5’UTR	0.0029	0.0000088	NR	VUSd
c.*646T>C	rs554798397	3’UTR	0.0029	0	NR	VUS
g.5000_5001delTG	-	Intronic	0.0029	0	NR	VUS
c.3412A>G, g.8813A>G	-	3’ UTR	0.0029	0	NR	VUS
c.123–48T>C	rs143273211	Intronic	0.006	0.003151	NR	Likely benign
c.*444C>T	rs123598	3’UTR	0.021	0.04414	Likely benign	Likely benign
c.2057–4G>T	rs149499021	Intronic	0.0015	0	Benign/Likely benign	Likely benign
c.1026C>T, p.Ser342=	rs71651686	coding-synonymous	0.0029	0.003524	Benign/Likely benign	Likely benign
c.2057–22A>C	rs144083199	Intronic	0.006	0.003753	Likely benign	Likely benign
c.*45C>G	rs56898787	3’UTR	0.006	0	Benign	Likely benign
c.1002A>G, p.Leu334=	rs28997577e	coding-synonymous	0.006	0.004198	Benign/Likely benign	Benign
c.931+70A>G	rs141917607e	intronic	0.006	0.004479	NR	Benign
c.931+117_931+118delCC	rs372474338e	intronic	0.006	0	NR	Benign
c.1891–30G>A	rs146661777e	intronic	0.0029	0.004216	NR	Benign
c.660–26T>A	rs139414598	intronic	0.006	0.004209	Likely benign	Benign

^a.GnomAD frequency in non-Finish European

^b.variants previously reported²

^c.NR: non-reported

^d.VUS: variant of uncertain significance

^e.variants identified in patient with truncation mutation.

The following variants were identified in the same patient: rs123598, rs149499021 (one patient), rs144083199, rs56898787 (two patients), rs28997577, rs139414598, rs141917607 and rs372474338 (one patient) rs387906849, rs139414598, rs372474338, rs141917607, rs72474338, rs28997577 and rs146661777 (one patient).

For the four VUSs, three were not observed in genomAD while one, rs374877579, had a minor allele frequency (MAF) of 0.0000088. Two of the variants were intronic while two were in the 3’UTR region. For the intronic variants, splicing software suggested no impact on splicing. No tumor tissue was available for further study of the functional impact of the variant in the 3’UTR region on gene expression of the gene.

Five variants (rs139414598, rs372474338, rs141917607, rs28997577, and rs146661777) were detected in a family with a truncating BAP1 mutation p. Gly267*. Four of these (rs28997577, rs139414598, rs141917607, and rs372474338) were also identified in a patient with familial UM but were not detected in the patient’s father who also has UM indicating that they are benign.²³

BAP1 large deletions

Out of the 161 patients assessed for large deletion/duplication in BAP1 140 passed the quality control measures. Of those, one patient showed a large deletion affecting all BAP1 probes (Figure 1). In two patients, MLPA suggested a single exon deletion (exon 16). However, further studies, including RT- PCR with primers spanning the suggested deleted exon as well as quantitative copy number variant PCR, showed no evidence of deletion.

Figure 1.

(A) MLPA demonstrates whole gene deletion in FUM961. All 17 probes of the BAP1 gene showed 50% decrease. (B) Constitutional decrease in BAP1 expression in non-tumor choroid tissue in a patient (7002) marked with red star. Downregulation of BAP1 was observed in 2 additional samples but was not significant compared to the cadaver choroids. Error bars represent standard deviation for technical replicates. (C) No evidence of aberrant promotor methylation in the patient.

Validation of the large deletion was carried out in a clinical laboratory using array comparative genome hybridization (aCGH). A hemizygous deletion with an approximate size of 26 kb was identified. The minimum deletion boundary via aCGH was Chr3:52430497–52456494 (genomic build GRCh37/hg19). The deletion includes at least the two nearby genes DNAH1 and PHF7 in addition to BAP1. DNHA1 and PHF7 are important in spermatogenesis with no known association with cancer so the observed phenotype in the patient is due to BAP1 hemizygous deletion.

Clinical Features of UM patients with germline BAP1 pathogenic alterations

In total, eight patients had pathogenic alterations in BAP1. Six out of these had a family history of UM with personal or family histories of other BAP1-TPDS associated cancers (Tables 1 and 3), and were previously reported by our group.⁸ Of the remaining two patients one was a male with a nonsense mutation, c.15G>A, cDNA.487G>A, p.W5*. The patient developed UM at age 29 years. He had no personal history of other cancers but a family history of a mother with RCC and a paternal uncle with CM. Affected family members were not available for testing. The final patient was a female diagnosed with UM at age 16 years. Germline sequencing did not identify any genetic alteration but MLPA identified 50% decrease of all the 17 probes representing BAP1. Probes from other genes on chromosome 3 were not impacted (Figure 1A). The copy number variation was further validated by quantitative real-time PCR (data not shown). The patient reported a family history of an unspecified liver cancer in her mother at age 26 who died at age 28, a paternal grandmother with breast cancer and a paternal grandfather with pancreatic cancer. Pathology reports were not available on the mother’s tumor. Samples on affected family members were not available for testing. Of the 8 patients with pathogenic BAP1 alterations, only one patient had a confirmed metastasis. However, two other patients died at relatively young age of 43 and 59 years. These two patients had additional primary cancers and the cause of death was not clear from available records. Four patients underwent brachytherapy and three had enucleation. The most common tumor histology was mixed cell type and spindle cell. Of the three tumors with full specimens, none had ciliary body or extrascleral extension.

TABLE 3.

Clinical characteristics of patients with pathogenic BAP1 alterations included in the study

Patient	Proband UMa (age/sex)	BAP1 variant	Proband (other cancer)	Family history			Treatment	Tumor size in mm	Tumor histology	Metastasis/ follow up
				UM	BAP1 cancers	Other cancers
FUM064c	41/F	p.Gln684*	Yes	Yes	Yes	Yes	brachytherapy	9.9 × 1.7	NA	Unknown/ Dead 26 ms
FUM104a,c	67/F	p. S628Pfs*8	Yes	Yes	Yes	Yes	NA	NA	NA	Unknown/ Dead 96 ms
FUM152c	18/F	p.L573fs*3, rs71651686	None	Yes	Yes	No	brachytherapy	3.5 × 6.5	NA	Yes/ Dead 108 ms
FUM327 c	62/M	p. Y646*	None	Yes	No	Yes	brachytherapy	15 × 2.96	NA	No/ Alive 69 ms
FUM340 c	22/F	p.P153Rfs*7	None	Yes	Yes	Yes	enucleation	8 × 1	Mixed cell type (spindle, epithelioid)	No/ Alive 126 ms
FUM036 c	52/F	p. Gly267*, rs139414598, rs372474338, rs141917607, rs72474338, rs28997577, rs146661777	Yes	Yes	Yes	Yes	brachytherapy	NA	NA	Unknown/ Dead 40 ms
FUM969	29/M	p W5*	None	No	Yes	Yes	enucleation	8.99 × 2.91	Mixed cell, predom spindle	No/ Alive 26 ms
FUM961	16/F	WGDb	None	No	No	Yes	enucleation	15 × 6	Spindle cell melanoma	No/ Alive 19 ms

^aConsented posthumously, NA: not available

^bWhole gene deletion

^cReported previously²

BAP1 associated cancers are uveal melanoma, cutaneous melanomas, renal cell carcinoma and mesotehliomas.

All tumors were choroidal except FUM340 which was a ring melanoma of the iris.

BAP1 germline epigenetic inactivation

RNA from non-tumor choroids was available on 24 patients with no detectable pathogenic BAP1 alteration. The expression levels of BAP1 were compared to the average expression in choroids from 6 cadaver eyes by qRT-PCR. BAP1 expression levels in the non-tumor choroids ranged from 0.32 to 2.88 while expression levels in the cadaver choroids ranged from 0.6 to 1.35, both compared to the average expression in the choroids. Using the ddCT mean + 2.5 X standard deviation the only outlier was the non-tumor choroid 7002, indicating uncharacteristically low expression of BAP1 as compared to the controls (Figure 1B). This patient was a 54-year-old female with family history of lung, breast, and unspecified skin cancers. No germline variant was detected in the sequenced coding and non-coding regions of BAP1.

We investigated BAP1 promotor methylation as a potential mechanism for epigenetic inactivation in 19 subjects with sufficient tumor and non-tumor tissues including the proband 7002 with downregulation of BAP1 in non-tumor choroid. We did not identify any significant promotor methylation in any of the samples (Figure 1C).

DISCUSSION

Selection of the UM patients for germline testing is an important decision for clinicians managing them. The reported frequencies of germline pathogenic and likely pathogenic variants in BAP1 vary between 1.6–3% in the unselected UM patient population.^7,10 Our study was designed to identify patients who should be prioritized for germline BAP1 mutation testing. This study focused on patients at high-risk for germline BAP1 mutations, including those with early onset of UM (<35 yrs) and those with strong personal and/or family history of cancer. To our surprise, even for these high-risk patients only 4.7% (8/172) had detectable pathogenic alterations in BAP1. The highest frequency of germline BAP1 pathogenic alterations was identified in UM patients with ≥2 personal or family history of cancers associated with BAP1-TPDS (36%, 6/16) followed by those with young age of onset of their tumors (19%, 4/21) and those with familial UM (18%, 6/34).

None of the 53 UM patients with personal or family history of CM without personal or family history RCC and MMe had pathogenic variants in BAP1. This suggests that the UM-CM families may constitute a unique phenotype associated with candidate genes other than BAP1. Of note, we have previously tested germline mutations in CDK4 and CDKN2A two genes associated with familial melanoma and did not identify any pathogenic variants in a cohort of 53 high-risk UM patients.⁶ Also, none of the 66 patients with personal or family histories of cancers other than the four main cancers associated with BAP1 had pathogenic mutations in BAP1, which again suggests the possibility of additional candidate genes/phenotypes.

One of the important findings in our study is that 1/8 (13%) of the germline pathogenic variants detected in BAP1 was due to a whole gene deletion. The deletion was not detected by direct sequencing. Although large deletions have been reported as a common somatic alterations in BAP1 in UM²⁴ and other cancers,^25,26 to our knowledge this is only the second report of germline whole-gene deletion of BAP1 in a UM patient.² This indicates that whole gene deletion should be included in the assessment of germline BAP1 alterations. It is worth noting that in this study direct sequencing had only 87.5% sensitivity for detection of germline alterations in BAP1 in patients with UM. Inclusion of BAP1 copy number variation assay will be crucial for proper assessment of germline variation in BAP1. Several algorithms have been developed for analysis of data generated from deep sequencing platforms to allow assessment of large copy number variations. Proper counseling of patients and physicians for the sensitivity of the assay used for detection of alterations in BAP1 highlighting the limitation of direct sequencing will be crucial. The deletion in the patient included two genes in the vicinity of BAP1, DNAH1 and PHF7. PHF7 and DNAH1 are both important in spermatogenesis with no association with cancer so the phenotype is caused mostly by deletion of BAP1. The patient with whole gene deletion was diagnosed at age 16 years which has been reported as the youngest age of onset of UM in patients with pathogenic mutations in BAP1²⁷.

Another important finding in this study is the occasional downregulation of BAP1 in the germline non-tumor tissue of UM patients. A potential explanation is germline epigenetic inactivation of BAP1 in these subjects. Pyrosequencing showed no significant BAP1 promotor methylation in any of the 19 studied subjects including the subject with significant constitutional downregulation of BAP1 in the non-tumor tissue. This suggests other potential mechanisms for germline epigenetic regulation of BAP1 such as alterations in miRNA and/or an enhancer element.

In addition to the pathogenic variants identified in BAP1 we detected several polymorphisms in 13% of UM patients. Lin et al²⁸ identified several common germline genetic variations in BAP1 that were associated with renal cell carcinoma and lung cancer. None of these variants were identified in our cohort. A recent study showed that variants in the G-quadruplexes in the BAP1 promoter region strongly regulate its expression. This promotor region is GC rich and cannot easily be sequenced by direct/Sanger sequencing.²⁹ In our study we only sequenced 229 bases of the 5’UTR region so the presence of regulatory variants in our patient cannot be totally excluded. Further investigation using next generation sequencing strategies could identify variants in the 5’UTR of BAP1.

Identification of high frequency of benign germline variants and VUS in BAP1 in UM patients highlights the importance of carrying out germline genetic testing through clinical geneticists/genetic counselors, as these professionals are more capable of counseling patients and families with benign variants and VUS. Also, our findings highlight the importance of establishing a registry for properly assessing these variants.

It has been suggested that germline BAP1 mutations are preferentially associated with metastatic UM.³⁰ In a larger follow up study the same group did not confirm such association¹⁰ but suggested that tumors from germline BAP1 mutations carriers present with relatively larger tumors, average largest tumor diameter of 15.9 mm, and a high frequency, 75%, of ciliary body involvement, two criteria that have been linked to aggressive diseases. Our results do not support or dispute the association of germline BAP1 mutations with aggressive disease and further larger studies will be needed. However, the strong evidence of association of somatic biallelic inactivation of BAP1 with aggressive UM⁷ warrants managing these patients as high-risk for metastasis.

In conclusion, our results suggest that in UM patients, germline BAP1 mutation testing should be prioritized to 1) those with familial UM; 2) those with early age of onset of their tumors (< 35 years old); and 3) those with personal or family history of at least one cancer associated with BAP1-TPDS such as MMe or RCC. Assessment for large deletions should be included in testing for germline alterations in BAP1. Finally, epigenetic germline inactivation of BAP1 and alterations in additional candidate genes could explain the hereditary predisposition of a subset of UM patients to cancer.