A unifying gene signature for adenoid cystic cancer identifies parallel MYB-dependent and MYB-independent therapeutic targets

Abstract

MYB activation is proposed to underlie development of adenoid cystic cancer (ACC), an aggressive salivary gland tumor with no effective systemic treatments. To discover druggable targets for ACC, we performed global mRNA/miRNA analyses of 12 ACC with matched normal tissues, and compared these data with 14 mucoepidermoid carcinomas (MEC) and 11 salivary adenocarcinomas (ADC). We detected a unique ACC gene signature of 1160 mRNAs and 22 miRNAs. MYB was the top-scoring gene (18-fold induction), however we observed the same signature in ACC without detectable MYB gene rearrangements. We also found 4 ACC tumors (1 among our 12 cases and 3 from public databases) with negligible MYB expression that retained the same ACC mRNA signature including over-expression of extracellular matrix (ECM) genes. Integration of this signature with somatic mutational analyses suggests that NOTCH1 and RUNX1 participate with MYB to activate ECM elements including the VCAN/HAPLN1 complex. We observed that forced MYB-NFIB expression in human salivary gland cells alters cell morphology and cell adhesion in vitro and depletion of VCAN blocked tumor cell growth of a short-term ACC tumor culture. In summary, we identified a unique ACC signature with parallel MYB-dependent and independent biomarkers and identified VCAN/HAPLN1 complexes as a potential target.

Figure 1. A unique mRNA gene signature distinguishes salivary gland adenoid cystic cancer (ACC) from matched normal salivary gland tissue

A) Principle component analysis (PCA) for all salivary gland adenoid cystic cancer (ACC) and matched normal samples. Red and green spheres indicate fusion positive and negative tumors respectively. Blue and brown indicate their matched normal tissues respectively. B) Supervised hierarchical-ward clustering analysis using 1160 mRNA probe sets that were significantly expressed in ACC (two sides paired t-test, FDR=0.05; see Supplementary Table 1 for complete gene list). Star depicts tumor sample with negligible MYB expression and matched normal tissue. MYB-NFIB fusion positive samples (P) and fusion negative samples (N). C) Gene localization and molecular enrichment analysis using Fisher's t-test (***, paired t-test, two side p-value<0.0001; **, paired t-test, two side p-value<0.001; ns, paired t-test, two side p-value>=0.05). Gene localization was defined by Ingenuity IPA. D) MYB-related network in ACC. The inner layer: MYB-interaction genes defined by Ingenuity IPA. The outer layer, MYB regulating genes defined by published ChIP-Seq data [33]. E) Average intensities of MYB probe sets for three subtypes of salivary gland tumors and their matched normal samples (***, paired t-test, two side p-value<0.0001). F) MYB average intensity in ACC for each sample. Arrow depicts sample with low MYB. G) qRT-PCR validation of relative MYB expression levels in ACC tumors (***, two sample t-test, two side p-value<0.0001; error bars are S.E.M. of 12 samples).

Figure 2. Exon array analysis is a sensitive tool to identify MYB C-terminal rearrangement status

A) Illustrations of MYB protein structure and MYB-NFIB fusion events in ACC. DBD, DNA binding domain; TAD, transcriptional activation domain; NRD, negative regulation domain. B) MYB exon intensity plot for 12 ACC tumor samples identifies variable expression of MYB C-terminal exons and validates MYB fusion status. Red line, MYB exon intensities in ACC tumor; Blue line, MYB exon intensities in matched normal tissue; Star depicts two false fusion-positive samples validated by FISH.

Figure 3. Integrative analysis identifies both MYB-dependent and MYB-independent genes in ACC

A) Differentially expressed genes with the highest fold change compared to normal in high or low MYB ACC tumors indicate extracellular gene activation. B) The same analysis as panel A was performed on ACC gene expression array data (Affy_HG133) obtained from public GEO database (accession: GSE28996). C) Top scoring genes that are components of extracellular matrix (ECM). MYB binding genes were defined by published ChIP-Seq data. MYB independent targets were defined as genes that were activated in both high and low MYB ACC tumors. D) Integrative analysis of ACC gene signature to recently published ACC mutational sequencing data identifies potential roles of RUNX1 in ACC tumorigenesis. E) Regulation of extracellular gene HAPLN1 by RUNX1.

Figure 4. Comparison of the differential ECM activation in three subtypes of salivary gland tumors and their matched normal samples

***, t-test, two side p-value<0.0001; **, p-value<0.001; *, p-value<0.05; ns, p-value >=0.05. Error bars are S.E.M. of 12 replicates of ACC, 14 replicates of MEC and 11 replicates of ADC.

Figure 5. Distinct miRNA expression in ACC tumorigenesis

A) Supervised hierarchical ward clustering for ACC using 22 mature miRNA seed probe sets that are significantly expressed (two sides paired t-test, FDR=0.05). B) Regulatory miRNA target predictions (TargetScan, Pictar) suggest potential roles of miRNAs in regulation of extracellular genes in ACC tumorigenesis.

Figure 6. Functional analysis of MYB fusion protein

A) Illustration of MYB cDNA clones tested fragments tested and their expression efficiency in human immortalized salivary ductal cells (DC). MYB-NFIB protein showed increased steady-state protein and B) prolonged half-life in human immortalized myoepithelial cells (MC). C) Forced expression of MYB-NFIB enhanced the anchorage independent colony formation of human salivary DC cells. ***, t-test, two side p-value<0.0001; ns, p-value >=0.05. Error bars are S.E.M. of 3 replicates for each group. D) Decreased attachment and increased viability of suspension cells after ectopic MYB-NFIB expression. Error bars are S.E.M. of 3 replicates for each group. E) ACC tumor cell growth inhibition after shRNAi depletion of VCAN, and F) decreased xenograft tumor growth ability in NOD.SCID mice (two sample t-test, error bars are S.E.M. for 4 replicates).