Volume 148, Issue 7, Pages 1427-1437.e8 (June 2015) Transforming Growth Factor β Signaling in Colorectal Cancer Cells With Microsatellite Instability Despite Biallelic Mutations in TGFBR2  Noel F.C.C. de Miranda, Maarten van Dinther, Brendy E.W.M. van den Akker, Tom van Wezel, Peter ten Dijke, Hans Morreau  Gastroenterology  Volume 148, Issue 7, Pages 1427-1437.e8 (June 2015) DOI: 10.1053/j.gastro.2015.02.052 Copyright © 2015 Terms and Conditions

Figure 1 p-SMAD2 and SMAD4 immunohistochemical detection in colon cancer tissues. (A–D) p-SMAD2 nuclear accumulation was detected in 78% of MSI-H colon cancers (strong in 44% and weak in 34% of cases). p-SMAD2 nuclear accumulation generally was associated with SMAD4 expression. (E and F) Expression of TMEPAI was associated with the expression of both p-SMAD2 and SMAD4. Scale bars: 50 μm. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Figure 2 Mutations at TGFBR2 and ACVR2A microsatellites. (A) Mutation screening of TGFBR2 and ACVR2A microsatellites was performed by means of fragment analysis on flow-sorted tumor and stromal cellular fractions. Tumor alleles (green) were overlaid with their respective normal samples (blue). Four different allelic states were recognized after fragment analysis. (B) Mutations in TGFBR2 and at the 10th exon of ACVR2A most often were biallelic and concomitant. No significant differences in p-SMAD2 detection were observed between the different genotypes. Polyclonal refers to the observation of more than 2 alleles for TGFBR2. Because chromosomal aberrations are rare in MSI-H CRC and all analyzed tumors were peridiploid, the additional alleles are likely to derive from tumor clones. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Figure 3 Induction of TGFβ pathway activity in a panel of CRC cell lines. (A) TGFβ activation of the (CAGA)12-luciferase reporter was detected in all CRC cell lines except LS411N, the only cell line carrying a biallelic 2-nucleotide deletion in TGFBR2’s microsatellite (Table 1). At least 3 independent experiments were performed. Means and standard deviations are shown (*P < .05, analysis of variance, Bonferroni post-test). (B) SMAD2 phosphorylation on TGFβ stimulation could be suppressed by TGFBR2 knockdown. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; shRNA, short hairpin RNA. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Figure 4 Induction of TGFβ target genes in MSI-H CRC cells: PAI-1 and TMEPAI expression could be induced effectively in SW837 cells. (A and B) This induction was inhibited by TGFBR2 knockdown. (C) Up-regulation of TMEPAI expression also could be observed in LS180 cells when combining the replicates derived from the negative control (empty vector) with the ones from cells expressing nontarget short hairpin RNA (shRNA). (D and E) Increased TMEPAI expression in HCT116 and SW48 cells on TGFβ stimulation was not observed but the basal levels of TMEPAI were significantly lower in cells transduced with the TGFBR2 shRNA. At least 3 independent experiments were performed. Means and standard deviations are shown (*P < .05, **P < .01, Student t test). mRNA, messenger RNA. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Figure 5 Rescue of TGFβ sensitivity in LS411N CRC cells. Constructs containing different TGFBR2 microsatellites were produced to test their ability to rescue TGFβ activity in LS411N cells. (A) Transfection with the TGFBR2 constructs containing A9 or A7 microsatellites rescued TGFβ sensitivity in LS411N cells, whereas the TGFBR2 construct containing 8 adenines (A8) failed to do so. (B) Constructs containing synonymous substitutions at TGFBR2’s A9 microsatellite failed to re-sensitize LS411N cells to TGFβ signaling. (C) The amount of in-frame proteins produced from mutant TGFBR2 constructs with A9 or A8 microsatellites was measured by placing the different TGFBR2 inserts in a NanoLuc expression vector (Promega). To avoid NanoLuc reporter expression from alternative start codons within the TGFBR2 gene, a Myc-tag located at the 5’-end of the TGFBR2 insert allowed for the immunoprecipitation (IP) of the full-length TGFBR2 protein before measurement. Anti-FLAG IP served as a negative control. (D) Expression of TGFBR2 could be detected in 293T cells transfected with TGFBR2’s A10 and A7 constructs, but also with the A9 construct. TGFBR2 expression was undetectable in cells transfected with the A8 constructs and the A9mut construct lacking a microsatellite repeat. At least 3 independent experiments were performed. Means and standard deviations are shown (*P < .05, **P < .01, ***P < .001, Student t test). Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 1 Validation of p-SMAD2 immunohistochemical detection in formalin-fixed, paraffin-embedded colorectal cancer cells. Scale bars: 50 mm. Nuclear and cytoplasmic immunopositivity can be seen on (left) stimulation with TGFβ, which is (right) absent upon further addition of the TGFBR1 kinase inhibitor SB 431542. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 2 Successful knockdown of TGFBR2 expression in SW837 cells by using TGFBR2 short hairpin RNAs (shRNAs). Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 3 TGFBR2 construct sequences were analyzed by Sanger sequencing before any experiment to select for appropriate clones and to exclude contamination. (A10mut–A7mut sequences produced by in situ mutagenesis to substitute the GAA and AAA codons by synonymous GAG and AAG sequences). Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 4 Expression levels of different TGFBR2 constructs transfected into 293T cells were monitored by qPCR. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 5 (Left) Representative example of absent or weak p-SMAD2 expression in normal colon epithelium. (Right) Focal detection of increased p-SMAD2 expression was detected in 3 cases. What seems to be gradient could be observed along the colonic crypt with increasing p-SMAD2 immunopositivity toward the top of the crypts. Scale bars: 100 mm. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 6 Basal levels of p-SMAD2 can be suppressed by using a TGFBR1 kinase inhibitor (SB 431542) and a TGFβ ligand-neutralizing antibody. Ab, antibody; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 7 (A) Proliferation and (B) migration assays were performed in colorectal cancer cell lines in the presence of TGFβ. No differences in proliferation were observed whereas TGFβ-induced migration of SW837 cells could be suppressed effectively by the knockdown of TGFBR2. The basal levels of migration in LS180, without TGFβ stimulation, also appeared reduced (*P < .05, **P < .01). shRNA, short hairpin RNA. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions

Supplementary Figure 8 (A) Detection of basal levels of TGFBR2 in nontransfected colorectal cancer cell lines. (B) Immunoprecipitation of TGFBR1 and TGFBR2 after cross-linking with iodinated TGFβ ligand. IP, immunoprecipitation. Gastroenterology 2015 148, 1427-1437.e8DOI: (10.1053/j.gastro.2015.02.052) Copyright © 2015 Terms and Conditions