Activin signaling in microsatellite stable colon cancers is disrupted by a combination of genetic and epigenetic mechanisms

Abstract

Background: Activin receptor 2 (ACVR2) is commonly mutated in microsatellite unstable (MSI) colon cancers, leading to protein loss, signaling disruption, and larger tumors. Here, we examined activin signaling disruption in microsatellite stable (MSS) colon cancers.

Methods: Fifty-one population-based MSS colon cancers were assessed for ACVR1, ACVR2 and pSMAD2 protein. Consensus mutation-prone portions of ACVR2 were sequenced in primary cancers and all exons in colon cancer cell lines. Loss of heterozygosity (LOH) was evaluated for ACVR2 and ACVR1, and ACVR2 promoter methylation by methylation-specific PCR and bisulfite sequencing and chromosomal instability (CIN) phenotype via fluorescent LOH analysis of 3 duplicate markers. ACVR2 promoter methylation and ACVR2 expression were assessed in colon cancer cell lines via qPCR and IP-Western blots. Re-expression of ACVR2 after demethylation with 5-aza-2'-deoxycytidine (5-Aza) was determined. An additional 26 MSS colon cancers were assessed for ACVR2 loss and its mechanism, and ACVR2 loss in all tested cancers correlated with clinicopathological criteria.

Results: Of 51 MSS colon tumors, 7 (14%) lost ACVR2, 2 (4%) ACVR1, and 5 (10%) pSMAD2 expression. No somatic ACVR2 mutations were detected. Loss of ACVR2 expression was associated with LOH at ACVR2 (p<0.001) and ACVR2 promoter hypermethylation (p<0.05). ACVR2 LOH, but not promoter hypermethylation, correlated with CIN status. In colon cancer cell lines with fully methylated ACVR2 promoter, loss of ACVR2 mRNA and protein expression was restored with 5-Aza treatment. Loss of ACVR2 was associated with an increase in primary colon cancer volume (p<0.05).

Conclusions: Only a small percentage of MSS colon cancers lose expression of activin signaling members. ACVR2 loss occurs through LOH and ACVR2 promoter hypermethylation, revealing distinct mechanisms for ACVR2 inactivation in both MSI and MSS subtypes of colon cancer.

Figure 1. Loss of expression of three components of activin signaling in primary MSS colon cancers.

Immunohistochemical analysis of paraffin-embedded primary colon cancers assessed expression of target protein (brown) as compared to adjacent normal tissue. A and B (Top row): ACVR1 expression is lost in a subset of MSS tumors. Example of tumor with expression in both normal and colon tumor tissue (left panel) and selective loss in a subset of tumors, but not adjacent normal colonic tissue (right panel). Overall, ACVR1 loss was observed in 2/51 or 4% of all MSS colon cancers analyzed. C and D (Middle row): ACVR2 expression is lost in a subset of MSS tumors. Two examples of selective loss of ACVR2 in colon tumor, but not normal colonic tissue (left and right panel) are shown. Overall, ACVR2 loss was observed in 7/51 or 14% of all MSS colon cancers analyzed. E and F (Bottom row): pSMAD2 expression is lost in a subset of MSS tumors. Example of expression in both normal and colon tumor tissue (left panel) and selective loss in a subset of tumors, but not normal colonic tissue (right panel). Overall, pSMAD2 loss was observed in 5/51 or 10% of all colon cancers analyzed.

Figure 2. Summary of ACVR2 loss and its mechanisms in primary MSS colon cancer specimens.

Of the 51 MSS colon cancers from the NCCCS cohort tested for ACVR2 and ACVR1 loss, 8 revealed loss of either receptor (7 lost ACVR2 and 2 lost ACVR1 with one tumor losing both, see Table 1). Of those 8, 1 lost pSMAD2. Of the 7 tumors with ACVR2 loss, 3 revealed LOH and 3 had selective ACVR2 promoter hypermethylation, while no mutations were found in any of the three kinase domain hotspots or the coding polyadenine tract of exon 10, commonly mutated in MSI colon tumors. Neither of the 2 tumors with ACVR1 loss revealed LOH at the ACVR1 locus. Conversely, of the 43 tumors expressing both ACVR2 and ACVR1, 4 or 9% lost pSMAD2 expression, while maintaining total SMAD2 and TGFBR2 expression, underscoring loss of SMAD2 phosphorylation capability as an additional primary event to disrupt activin signaling.

Figure 3. ACVR2 promoter hypermethylation and LOH in colon cancer specimens and the MSS HT29 cell line and correlation of ACVR2 promoter hypermethylation and loss of ACVR2 expression.

A) ACVR2 promoter with map of positioning of MSP primers B) Using a CpG islands search program, we identified the CpG islands within the ACVR2 promoter based upon following stringent criteria: ∼CG percentage>55%; observed CpG/expected CpG >0.65; length >500 bp. All CpG dinucleotide sequences are represented by vertical bars across the horizontal line depicting the promoter sequence. Each circle in the bottom panel illustrates an individual CpG site corresponding to the vertical bars depicted in the upper panel. Based upon bisulfite sequencing, each CpG site was scored quantitatively as 100% methylated (dark circles), >50% methylated (dark grey circles), <50% methylated (light grey circles) or unmethylated (white circles). As indicated, 3 of 7 ACVR2-negative CRCs demonstrated complete methylation of the critical region of ACVR2 promoter, while none of the ACVR2-expressing tumors showed any evidence for high degree/complete methylation of ACVR2 promoter. A methylation pattern similar to that seen in ACVR2 negative colon cancers was observed in the MSS colon cancer cell line HT29 with genomic DNA from HT-29 allowing for sequencing of a slightly larger region of the ACVR2 promoter. C) Six colon cancer cell lines with different microsatellite instability backgrounds (see Table 3) were analyzed for ACVR2 protein expression using immunoprecipitation techniques. Two cell lines, HCT116 (an MSI colon cancer cell line with biallelic frameshift mutations in ACVR2) and the MSS colon cancer cell line HT29 (with ACVR2 promoter hypermethylation), revealed loss of ACVR2 protein expression. D) Loss of ACVR2 protein expression correlated with decrease in ACVR2 mRNA transcription via quantitative PCR in HT29 cells when compared to the ACVR2 expressing cell line FET using GAPDH for standardization. This experiment was performed three times in triplicates, and the bar graph represents one experiment with * indicating a statistically significant difference with a p<0.001. E) ACVR2 protein expression was re-established following demethylation treatment with 5′Aza.