An oncogenic MYB feedback loop drives alternate cell fates in adenoid cystic carcinoma

Authors

Yotam Drier

Matthew J. Cotton

Kaylyn E. Williamson

Shawn M. Gillespie

Russell J.H. Ryan

Michael J. Kluk

Christopher D. Carey

Scott J. Rodig

Lynette M. Sholl

Amir H. Afrogheh

William C. Faquin

Lurdes Queimado

Jun Qi

Michael J. Wick

Adel K. El-Naggar

James E. Bradner

Christopher A. Moskaluk

Jon C. Aster

Birgit Knoechel

Bradley E. Bernstein

Doi

10.1038/ng.3502

PMID: 26829750 · DOI: 10.1038/ng.3502 · Journal: Nature Genetics (2016)

TL;DR

Drier et al. integrated whole-genome sequencing and chromatin profiling of 20 adenoid cystic carcinomas (ACCs) — 12 published primary tumors, 2 additional primary tumors, and 6 patient-derived primagrafts — and showed that the recurrent MYB rearrangements (15/20 tumors) act not by producing a fusion protein but by juxtaposing super-enhancers from NFIB, TGFBR3, or RAD51B to the MYB locus. MYB protein binds these translocated enhancers, which loop to the MYB promoter, generating a positive feedback loop that sustains oncogenic MYB expression. MYB cooperates with TP63 in myoepithelial cells of low-grade tumors and with Notch in luminal epithelial cells, while grade-3 tumors carry NOTCH1/SPEN-activating events. The BET bromodomain inhibitor JQ1 slowed growth in grade-2 primagrafts but had no effect on Notch-driven grade-3 primagrafts, suggesting BET inhibition for low-grade ACC and Notch inhibition for high-grade ACC.

Cohort & data

  • 20 ACYC tumors analyzed for MYB translocations: 12 primary tumors from EGA EGAS00001000030, 5 additional published primaries, 6 patient-derived xenograft (primagraft) models, plus 2 ACCs sequenced by targeted paired-end sequencing.
  • Chromatin landscapes (H3K4me3, H3K27ac, BRD4 ChIP-seq) mapped in 13 ACCs (5 primary specimens + 8 primagrafts) using chip-seq.
  • MYB and TP63 ChIP-seq performed in 3 ACC primagrafts.
  • Immunohistochemistry for MYB, TP63, ICN1 (active NOTCH1), KIT, and Ki-67 across 19 grade-2 ACCs and 8 grade-3 ACCs (immunohistochemistry).
  • WGS done on Illumina HiSeq 2500 with 100 bp paired-end reads (whole-genome-seq); reads aligned to hg19 with BWA; rearrangements called with dRanger and BreakPointer.
  • Dataset registered as acyc_mgh_2016; new sequencing data deposited at EGA EGAS00001001457.
  • In vivo drug testing in nude (Foxn1nu) mice using 4 ACC primagrafts (2 grade-2, 2 grade-3); BET inhibitor JQ1 at 50 mg/kg daily orally (subcutaneous-xenograft).

Key findings

  • 15/20 (75%) ACCs harbored MYB rearrangements PMID:26829750. Of the 20 tumors: 6/20 (30%) had canonical NFIB-MYB fusions with loss of MYB 3′UTR; 6/20 (30%) had NFIB-MYB rearrangements that retained the MYB 3′UTR; 2/20 had MYB–TGFBR3 rearrangements; 1/20 had a MYB–RAD51B rearrangement.
  • Several rearrangements occur at the 5′ end of MYB, inconsistent with production of any fusion protein, indicating a regulatory rather than fusion-based mechanism.
  • Translocated regions in NFIB, TGFBR3, and RAD51B all contain super-enhancers active in ACC. Chromosome Conformation Capture (3C) demonstrated that 4/8 H3K27ac elements between 13 kb and 750 kb from MYB physically interact with the MYB promoter in a NFIB-translocated tumor, and 7/9 in a MYB-TGFBR3 tumor.
  • MYB ChIP-seq in 3 primagrafts identified 13,278 high-confidence MYB binding sites (75% distal regulatory, with strong enrichment for the MYB consensus motif CAGTT, p<10⁻⁷⁵⁹). MYB binds the translocated enhancers in NFIB and TGFBR3, supporting a positive feedback loop.
  • Reporter assays in Jurkat cells: 4/5 cloned 250 bp enhancer intervals from NFIB/TGFBR3 strongly induced reporter activity; mutating MYB motifs in 2 of these elements diminished activity.
  • Putative MYB targets in ACC are enriched for development, migration, cell cycle, transcription regulation, and angiogenesis (REACTOME/GO/MSigDB FDR<1%); examples include MYC, BCL2, AURKA, CCND1, MET, FGFR2, IGF1R, MALAT1, CDK6, and GMNN.
  • Second-ranked motif under MYB peaks is TP53/TP63/TP73 (p<10⁻³⁴⁰); 81% of TP63 binding sites in ACC are co-bound by MYB.
  • Putative MYB-driven transcriptional regulators include EN1 (a high-grade ACC biomarker), ARID1A, NOTCH1, the Notch ligands JAG1 and JAG2, and the Notch repressor SPEN.
  • Immunohistochemistry on 19 grade-2 ACCs: TP63 stains the myoepithelial compartment; KIT and ICN1 (active NOTCH1) stain the luminal epithelial compartment; TP63 and ICN1 staining are mutually exclusive.
  • Eight grade-3 ACCs lack TP63 staining and show diffuse ICN1 positivity; activating NOTCH1 mutations or SPEN loss-of-function were present in 7/9 grade-3 tumors but in none of the lower-grade tumors examined. Two additional grade-3 tumors had non-mutational NOTCH1 alterations (a tandem duplication 3′ of NOTCH1 in ACC X11; a partial 5′ NOTCH1 deletion in ACC D1).
  • Differential motif analysis: TP63 motif enriched in grade-2-specific enhancers; RBPJ/Notch motif enriched in grade-3-specific enhancers. Only the oncogenic ΔNp63 isoform is transcribed in the ACC cohort.
  • BET bromodomain inhibitor JQ1 (50 mg/kg daily orally) significantly slowed tumor growth in 2 grade-2 ACC primagrafts (X5M1, X9), with modest decreases in MYB and MYB-target gene expression. Both grade-3 primagrafts (X6, X11) — both Notch-activated — failed to respond to JQ1.

Genes & alterations

  • MYB — recurrent rearrangement target (15/20 ACCs); enhancer hijacking by translocated super-enhancers establishes a positive feedback loop that drives MYB overexpression. MYB binding genome-wide (13,278 high-confidence peaks) cooperates with TP63 and Notch programs.
  • NFIB — partner locus in NFIB-MYB rearrangements (12/20 tumors total: 6 with 3′UTR loss, 6 with retained 3′UTR). Hosts super-enhancers that translocate to MYB.
  • TGFBR3 — novel partner locus in 2/20 ACCs; downstream region contains super-enhancers translocated to MYB.
  • RAD51B — novel partner locus in 1/20 ACC; downstream region contains smaller enhancers translocated to MYB.
  • TP63 — mediator of MYB regulatory program in myoepithelial cells of low-grade ACC; only the ΔNp63 isoform is transcribed.
  • NOTCH1 — activated by point mutations or structural alterations (tandem duplication of 3′ enhancers; 5′ partial deletion) in grade-3 ACC; intracellular NOTCH1 (ICN1) marks luminal epithelial cells.
  • SPEN — loss-of-function mutations contribute to constitutive Notch activation in grade-3 ACC.
  • ARID1A — putative MYB target; mutated in ACC (per Stephens et al. cited reference).
  • BRD4 — BET bromodomain protein occupying super-enhancers at MYB and other MYB targets; functional dependency in grade-2 ACC.
  • JAG1, JAG2 — Notch ligands among top MYB target transcription regulators.
  • RBPJ — RBPJ/Notch motif enriched in grade-3-specific enhancers.
  • EN1 — high-grade ACC biomarker, identified as a putative MYB-driven transcription regulator.
  • KIT — luminal epithelial marker in ACC IHC.
  • MYC, BCL2, CDK6 — among putative MYB target genes in ACC.

Clinical implications

  • The unifying mechanism in ACC is enhancer hijacking, not a fusion protein — diagnostic FISH/PCR strategies should look beyond MYB-NFIB fusion transcripts to detect the broader spectrum of regulatory rearrangements (including MYB-TGFBR3, MYB-RAD51B, and NFIB-MYB events that retain the MYB 3′UTR).
  • Low-grade (grade 2) ACCs are dependent on the MYB-driven enhancer circuit and are sensitive to BET bromodomain inhibition with JQ1 in primagraft models, suggesting BET inhibitors as a therapeutic candidate for non-resectable cribriform/low-grade ACC.
  • High-grade (grade 3) ACCs harbor activating Notch pathway events (NOTCH1 mutations/structural alterations or SPEN loss in 7/9 grade-3 tumors) and are insensitive to JQ1; the authors note recent evidence that Notch-mutant ACCs respond to Notch inhibitors, supporting Notch inhibition as the appropriate strategy for grade-3 disease.
  • TP63 and ICN1 IHC staining patterns can stratify ACC into low-grade (mixed cribriform with TP63-positive myoepithelial and ICN1-positive luminal compartments) versus high-grade (TP63-negative, diffusely ICN1-positive solid) tumors, and may help predict therapeutic vulnerability.

Limitations & open questions

  • ACC primagrafts and primary tumors largely substitute for the absence of faithful in vitro ACC models — the authors explicitly state that further validation of putative MYB targets “will require the development of faithful in vitro models for ACC.”
  • The drug testing cohort is small (4 primagrafts; 4–9 mice per group), limiting statistical inference about JQ1 response heterogeneity.
  • WGS lacked matched normal controls; rearrangement calls were filtered against a panel of 100 unmatched normals and DGV germline variants — residual germline contamination cannot be fully excluded.
  • The mechanism by which the translocated enhancer engages the MYB promoter genomically (looping in cis vs. trans, role of TAD disruption) is demonstrated by 3C but the chromosomal architecture was not mapped at higher resolution (e.g., Hi-C).
  • IHC differences in MYB intensity between grade-2 and grade-3 tumors may partly reflect technical variation; whether lower MYB protein levels causally underlie the grade-3 BET-resistance phenotype was not directly tested.
  • No matched longitudinal samples — whether the grade-2 → grade-3 transition is accompanied by acquired Notch alterations within a single patient remains unresolved.

Citations from this paper used in the wiki

  • “We identified MYB translocations as the main recurrent event (13 out of 18 ACCs) in these tumors… This yielded a total of 15 (out of 20) ACCs with MYB rearrangements.” (Results, p. 3)
  • “We identified canonical NFIB-MYB fusions with loss of the MYB 3′ UTR in 6 of the 20 tumors (30%)… An additional 6 tumors (30%) harbor an NFIB-MYB rearrangement but retain the MYB 3′UTR.” (Results, p. 3)
  • “MYB binding patterns… share a statistically significant overlap: 62% overlap with MYB-bound promoters in MCF7 (p<10⁻⁶); 60% overlap with MYB targets in mouse myeloid progenitors (p<10⁻⁵¹).” (Results, p. 4–5)
  • “we found that 81% of TP63 binding sites in ACC are co-bound by MYB” (Results, p. 5)
  • “These mutations are present in 7 out of 9 grade 3 tumors, but none of the lower grade tumors examined” (re: NOTCH1/SPEN in grade-3 ACC, Results, p. 6)
  • “BET inhibition significantly slowed tumor growth in the grade 2 primagrafts… In contrast, the grade 3 tumors did not respond to BET inhibition.” (Results, p. 7)
  • “All new data has been deposited at the European Genome-phenome Archive (EGA)… under accession number EGAS00001001457.” (Methods, p. 9)

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