Deep sequencing of uveal melanoma identifies a recurrent mutation in PLCB4

Authors

Peter Johansson

Lauren G. Aoude

Karin Wadt

William J. Glasson

Sunil K. Warrier

Alex W. Hewitt

Jens Folke Kiilgaard

Steffen Heegaard

Tim Isaacs

Maria Franchina

Christian Ingvar

Tersia Vermeulen

Kevin J. Whitehead

Christopher W. Schmidt

Jane M. Palmer

Judith Symmons

Anne-Marie Gerdes

Göran Jönsson

Nicholas K. Hayward

Doi

10.18632/oncotarget.6614

PMID: 26683228 · DOI: 10.18632/oncotarget.6614 · Journal: Oncotarget (2016)

TL;DR

Johansson et al. performed whole-genome (WGS, n=14) or whole-exome (WES, n=14, including 9 tumors and 5 primary cell lines) sequencing of 28 uveal melanoma (UM) samples to search for novel driver mutations beyond the canonical GNAQ, GNA11, EIF1AX, SF3B1, and BAP1. They identified a recurrent hotspot mutation in PLCB4 (c.G1888T, p.D630Y) in 2 of 28 samples, mutually exclusive with GNAQ/GNA11 mutations, validated by Sanger sequencing and corroborated in 1 of 56 published UM tumors. Because PLCB4 is the canonical downstream effector of GNAQ/GNA11 in the Gαq signaling pathway, the authors propose PLCB4 p.D630Y is a gain-of-function driver activating the same pathway. The cohort had a low mutation burden (mean 10.6 protein-changing mutations/sample, 0.50/Mb), no UV signature, and a predominant BRCA (signature 3) mutation signature.

Cohort & data

  • 28 untreated UM samples with matched germline (blood or saliva): 14 primary tumors profiled by WGS, 9 tumors plus 5 primary tumor-derived cell lines profiled by WES.
  • Cancer type: uveal melanoma (UM).
  • Datasets: um_qimr_2016; external validation drawn from prior UM WGS/WES (n=56) and the TCGA UM cohort (uvm_tcga, 80 samples released 2015-11-14 while paper was under review).
  • Reference genome: GRCh37 (hg19); alignment with BWA, GATK indel realignment / base quality recalibration, variant calling with SAMtools/bcftools, annotation with ANNOVAR.
  • Structural variants called with Janda v0.8.1; CNV/LOH segmentation with binocular v0.2 (Circular Binary Segmentation variant).
  • Mutational signature decomposition adapted from Alexandrov et al. (30 COSMIC signatures).

Key findings

  • Low mutation burden, no UV signature. Mean 10.6 protein-changing mutations per sample (range 0–53); mean 0.50 mutations/Mb across the genome (range 0.22–0.66/Mb) and 0.53/Mb in coding regions (range 0.06–2.52/Mb). No sample showed COSMIC signature 7 (UV).
  • Predominant BRCA mutation signature. 79% of samples carried COSMIC signature 3 (defective double-strand break repair, “BRCA” signature); signatures 1 and 5 were ubiquitous; signatures 12 or 16 contributed in almost every sample. No significant association between signature 3 and BAP1 mutation status.
  • Canonical UM drivers recapitulated. 11 BAP1 mutations (7 frameshifting indels, 2 splice, 1 nonsense, 1 missense); 14 GNA11 p.Q209P; 7 GNAQ p.Q209P and 1 GNAQ p.G48L; 4 EIF1AX mutations (p.P2L, p.G6V, p.G8R, splice) mutually exclusive with BAP1; 3 SF3B1 mutations (p.R625C, p.R625H, p.K666T). GNA11 and GNAQ mutations were mutually exclusive.
  • Recurrent PLCB4 p.D630Y hotspot. PLCB4 c.G1888T p.D630Y (chr20:9389753, NM_000933) was the only additional gene with a recurrent mutation (2 of 28 samples). Sanger-validated. Predicted deleterious by SIFT, probably damaging by PolyPhen2, GERP++ score 5.69, located in the Y-domain of the catalytic core. The identical variant was found in 1 of 56 previously published UM samples.
  • Mutual exclusivity with GNAQ/GNA11. Both PLCB4 p.D630Y samples lacked GNAQ and GNA11 mutations, and the same exclusivity held in published cohorts — consistent with PLCB4 acting at the same node of the Gαq pathway.
  • TCGA UM (released during review) corroborates the hotspot. 2/80 TCGA UM samples had codon p.D630 PLCB4 mutations: TCGA-VD-A8KD carried adjacent c.G1888T + c.A1889T variants (yielding p.D630Y + p.D630V if biallelic, or p.D630F in cis); TCGA-YZ-A985 carried c.G1888A (p.D630N).
  • PLCB3 candidate. 1 of 28 samples carried PLCB3 c.G2694C p.K898N (NM_001184883) in the CTD linker, a region implicated in GNAQ activation. The PLCB3 protein also binds GNAQ/GNA11 in STRING interactome analysis.
  • Other recurrent genes. Beyond the established drivers and PLCB4, only 4 genes were mutated in more than one sample (each with 2 missense mutations): MUC3A, TCHH, TTN, and LLGL1. LLGL1 is the most plausible additional candidate — a known tumor suppressor in glioblastoma and esophageal squamous cell carcinoma.
  • Copy-number landscape. Monosomy 3 or copy-neutral LOH of chromosome 3 was the most common event (8 of 14 WGS samples, 57%); all 6 BAP1-mutant samples were hemizygous for chromosome 3. Additional recurrent aberrations: 8q gain (n=6), 8p loss (n=3), 16q loss (n=2), X loss (n=3). Sample 554 was an outlier with near-genome-wide gains (triploid 4, 5, 6, 11, 12, 13, 17q, 21; tetraploid 7, 8, 18, 20, X). Diploid-3 samples had fewer large aberrations.
  • 8q gain is a late event. Bimodal VAF distributions across the six 8q-gained samples imply duplication occurred after most mutations accrued; on average 93% (range 87–97%) of mutations on gained 8q arose before duplication.
  • Structural variant burden is low. Total of 297 SVs across 14 WGS samples (mean 273 reported — likely typo for ~21; range 9–42 per sample). Genes hit by SVs in more than one sample: CDH13, FAM135B, GFRAL, LRRC16A, MOK, SEMA3E, VPS13B. Only two in-frame fusions: sample 553 GSPT1–HSD17B3 and sample 550 FAM135B–PDSS2 (interchromosomal between commonly aberrant chromosomes 6 and 8). Sample 539 had a 154 bp intronic BAP1 deletion alongside copy-neutral chromosome 3 LOH.

Genes & alterations

  • PLCB4 — recurrent hotspot p.D630Y (c.G1888T) in 2/28 UM samples, mutually exclusive with GNAQ/GNA11; proposed gain-of-function driver in UM. Contrasts with cutaneous melanoma where PLCB4 is mutated in 21–28% of cases along the length of the gene without recurrence at p.D630, suggestive of loss-of-function role in CMM.
  • GNAQ — 7 samples with p.Q209P and 1 with p.G48L; mutually exclusive with GNA11.
  • GNA11 — 14 samples with p.Q209P (most frequent driver in this cohort).
  • BAP1 — 11 truncating/splice/missense mutations; all 6 BAP1-mutant WGS samples were chromosome-3 hemizygous; one intronic 154 bp deletion identified via SV calling.
  • SF3B1 — 3 mutations: p.R625C, p.R625H, p.K666T.
  • EIF1AX — 4 mutations (p.P2L, p.G6V, p.G8R, splice); mutually exclusive with BAP1.
  • PLCB3 — single p.K898N (c.G2694C) in CTD linker; candidate for involvement in the GNAQ/GNA11/PLCB axis.
  • LLGL1 — 2 missense mutations; flagged as a plausible additional candidate driver based on tumor-suppressor roles in other cancers.
  • MUC3A, TCHH, TTN — each with 2 missense mutations; flagged as recurrent but without functional support.
  • GSPT1–HSD17B3 — in-frame fusion in sample 553.
  • FAM135B–PDSS2 — in-frame interchromosomal fusion in sample 550 (chromosomes 6 and 8).

Clinical implications

  • Reinforces that 83% of UM acquire GNAQ or GNA11 mutations (well-established) and now extends the Gαq pathway driver spectrum to include PLCB4 p.D630Y, suggesting that PLCB4-mutant UM tumors should be considered functionally equivalent to GNAQ/GNA11-mutant tumors for pathway-targeted therapeutic strategies.
  • BAP1 mutation co-occurrence with chromosome 3 hemizygosity is consistent with prior reports that BAP1 loss requires two hits and associates with metastatic risk; the present cohort does not directly assess prognosis.
  • The BRCA-like mutational signature (signature 3) in 79% of samples, with no significant link to BAP1 status, raises but does not test the hypothesis of DNA double-strand break repair vulnerability in UM.

Limitations & open questions

  • Small discovery cohort (28 samples; 14 WGS) limits power for rare driver detection.
  • Functional validation of PLCB4 p.D630Y as gain-of-function is not performed — authors call out that functional studies are warranted.
  • The mechanistic basis of the BRCA-like signature in UM is unexplained; no BRCA1/BRCA2 mutations are mentioned and the signature does not segregate with BAP1.
  • LLGL1, MUC3A, TCHH, TTN flagged as recurrent but uncharacterized; LLGL1 is plausibly oncogenic but requires further study.
  • The reported “mean 273 SVs per sample” alongside “range 9 to 42” appears to be a typographical inconsistency in the manuscript; the per-sample range suggests the mean is closer to ~21.
  • Cross-cancer context: PLCB4 hotspot at p.D630 is specific to UM and is absent from 159 reported non-synonymous PLCB4 mutations in CMM, supporting context-specific oncogene vs. tumor-suppressor roles — but this remains a speculation grounded only in mutation distribution.

Citations from this paper used in the wiki

  • “We performed WGS or WES on 28 untreated UM samples and identified a total of 297 non-synonymous mutations (mean 10.6, range 0 to 53).”
  • “PLCB4, phospholipase C, beta 4, was the only other gene that had a recurrent mutation (c.G1888T, p.D630Y, chr20:9389753, NM_000933), which occurred in 2 of 28 samples.”
  • “A search of mutations in other UM WGS/WES data sets identified the same PLCB4 mutation in 1 of 56 samples, which also occurred mutually exclusive to GNAQ and GNA11 mutations.”
  • “The two samples we identified with PLCB4 mutations did not have mutations in either GNAQ or GNA11.”
  • “PLCB4 is highly mutated (21% to 28%) in CMM but in contrast to UM, these mutations occur along the length of the gene, suggestive of them being loss-of-function.”
  • “The most common event was monosomy 3 or copy neutral LOH of the entire chromosome 3, which was seen in 8 of 14 samples (57%). Notably, all samples with BAP1 mutation (n=6) were hemizygous for chromosome 3.”
  • “The majority (79%) of samples also had signature 3, which is associated with defects in DNA double-strand break-repair (the ‘BRCA’ signature) but there is no significant association between BAP1 mutation and this signature in the samples analyzed here.”
  • “On average 93% (range 87-97%) of the mutations were estimated to occur before the duplication showing that gained chromosome 8q is a late event.”
  • “Sample TCGA-VD-A8KD carries two mutations at adjacent bases in the same codon, c.G1888T and c.A1889T… Sample TCGA-YZ-A985 carries a c.G1888A mutation resulting in amino acid change p.D630N.”

This page was processed by crosslinker on 2026-05-14.