The proto CpG island methylator phenotype of sessile serrated adenomas/polyps

Figures & data

Table 1. Characteristics of the prospectively collected precancerous lesions.

Patient number	Sex	Age	Colon segment involved	Maximum lesion diameter	Paris classification #	Pit pattern classification †	Microscopic appearance	Dysplasia ∆	No. of lesions at study colonoscopy §	BRAF	KRAS
S1	F	88	C	15	0-IIa	IIIs	SSA/P	none	2	c.1799T > A (V600E)	WT
S2	F	68	A	12	0-Is	IIIs, IIIL	SSA/P	none	1	c.1799T > A (V600E)	WT
S3	F	61	A	20	0-IIa	II	SSA/P	dysplasia	4	c.1799T > A (V600E)	WT
S4	F	71	A	15	0-IIa + IIc	IV, V	SSA/P	dysplasia	1	c.1799T > A (V600E)	WT
S5	M	52	C	30	0-IIa	IV	SSA/P	none	1	c.1799T > A (V600E)	WT
S6	F	76	C	25	0-IIa	IV	SSA/P *	dysplasia	2	c.1799T > A (V600E)	WT
S7	F	59	A	14	0-IIa	nr	SSA/P	dysplasia	2	c.1799T > A (V600E)	WT
S8	M	63	A	14	0-Is	II	SSA/P	none	6	c.1799T > A (V600E)	WT
S9	F	69	A	20	0-IIa	II	SSA/P	none	1	c.1799T > A (V600E)	WT
S10	F	45	A	25	0-Is	Is	SSA/P	none	1	c.1799T > A (V600E)	WT
S11	M	57	C	14	0-IIa	IIIs, IIIL	SSA/P	none	2	WT	c.37G > C (G13R)
S12	M	49	A	10	0-IIa	nr	SSA/P	none	3	c.1799T > A (V600E)	WT
S13	M	63	A	15	0-Is	nr	SSA/P	none	3	c.1799T > A (V600E)	WT
S14º	F	65	A	15	0-IIa	IIIs	SSA/P	none	14	c.1799T > A (V600E)	WT
S15	F	36	A	15	0-IIa	nr	SSA/P	none	2	c.1799T > A (V600E)	WT
S16	M	55	C	25	0-IIa + Is	IIIL, IV	SSA/P *	dysplasia	0	WT	c.35G > A (G12D)
S17	F	53	C	15	0-IIa	IIIs	SSA/P	none	1	c.1799T > A (V600E)	WT
A1	M	58	A	25	0-IIa	IV	TA	LGD	1	WT	c.38G > T (G13V)
A2	M	68	A	35	0-Ips	IV	TVA	LGD	1	WT	c.37G > T (G13C)
A3	F	78	A	15	0-Is	IV	TA	LGD	1	WT	WT
A4	F	64	A	30	0-IIa + Is	IV	TA	LGD	1	WT	c.34G > T (G12C)
A5	F	59	A	25	0-IIa	IIIL	TA	LGD	1	WT	WT
A6	F	73	A	10	0-Is	IIIs	TA	LGD	12	WT	WT
A7	F	64	C	20	0-Is	IIIL	TA	LGD	1	WT	WT
A8	M	70	A	10	0-IIa	II, IIIs	TA	LGD	3	WT	WT
A9	F	73	A	30	0-IIa	IIIL	TA	LGD	1	WT	c.35G > T (G12V)
A10	M	77	C	40	0-Is	IIIL, IV	TA	LGD	9	WT	WT
A11	F	84	A	40	0-IIa	IIIL, IV	TA	LGD	5	WT	c.35G > T (G12V)
A12	M	61	A	25	0-Ips	IIIL, IV	TVA	LGD	1	WT	WT
A13	M	83	A	18	0-Is	IIIL	TA	LGD	1	WT	WT
A14	F	78	A	15	0-Is	IIIs	TA	LGD	2	WT	WT
A15	M	75	C	20	0-IIa + Is	IV	TA	LGD	1	WT	c.37G > T (G13C)

Abbreviations: M, male; F, female; C, cecum; A, ascending colon; TA, tubular adenoma; TVA, tubulovillous adenoma; SSA/P, sessile serrated adenoma/polyp; LGD, low-grade dysplasia; nr: not reported.

# Macroscopic appearance of neoplastic lesions was classified according to Paris Endoscopic Classification (Endoscopy 2005; 37:570)

† Morphological analysis of colon crypt patterns according to the Kudo classification (Kudo S et al. Pit pattern in colorectal neoplasia: endoscopic magnifying view. Endoscopy 2001; 33:367).

∆ Low-grade versus high-grade dysplasia as defined by the WHO classification of tumors (Bosman FT et al. WHO classification of tumors of the digestive system: WHO Press, 2010).

§ Numbers include lesions analyzed in our study.

º There is only RNA sequencing data available for this sample.

* Lesions showing histological features of both SSA/Ps and cADNs. In S6, the serrated and adenomatous areas were distinct. S16 was likely a ‘serrated tubulovillous adenoma’, as described by Bettington M et al. (Histopathology 2016, 68:578).

Figure 1. Differentially methylated cytosines (DMCs) in precancerous and cancerous colon lesions. a. MDS plot (Supplementary Experimental Procedures) of DNA methylation levels in SSA/Ps (n = 16); cADNs (n = 15); CIMP(+) cancers (n = 3); CIMP(-) cancers (n = 3); and matched samples of normal mucosa for each lesion. b. Volcano plots showing the magnitude (x axis) and statistical significance (y axis) of the differential methylation observed at DMCs identified in SSA/Ps and cADNs. X-axis: The magnitude of differential methylation was calculated as the M:T ratio (no. methylated reads/total no. reads) for the tumor sample minus M:T ratio for matched normal-tissue control. Black dots: DMCs with absolute methylation differences of <0.1 and P-values > 0.01. Yellow and blue dots: highly significant (P-value < 0.01) DMCs (hypermethylated and hypomethylated, respectively). c. Density plot showing variance at the hypermethylated (top) and hypomethylated (bottom) DMCs (yellow and blue dots of panel B, respectively). d. UpSet plots showing hypermethylated (left) and hypomethylated (right) DMC sets in SSA/Ps and cADNs and their overlaps. Exact numbers of lesion-specific (•) and shared (•-•) DMCs appear above the bars. e. Overlaid density plots showing the distributions of hypo- and hypermethylated DMCs in SSA/Ps and cADNs (left), and CIMP(+) and CIMP(-) cancers (right). X-axis as described in panel B.

Figure 2. Differentially methylated regions (DMRs) in precursor lesions and proto-CIMP in SSA/Ps. a. Volcano plots showing the median differential methylation (x-axis) and its statistical significance (y-axis) for DMRs found in SSA/Ps and cADNs. (See Figure 1B legend for cutoffs and color codes) b. UpSet plots showing the lesion-specificity of hypermethylated and hypomethylated DMRs found in precancerous tumors. Exact numbers of lesion-specific (•) and shared (•-•) DMRs appear above the bars. c. Genomic location of DMRs in SSA/Ps vs cADNs. d. Scatterplots showing the extent of methylation at regions hypermethylated in both SSA/Ps and cADNs, and located within CpG islands (dots in the left panel) or 4-kb peri-TSS windows, i.e., TSS plus 2000-bp upstream and downstream flanking regions (dots in the right panel). In addition, CpG islands or peri-TSS windows with at least one hypermethylated CpG in either SSA/Ps or cADNs are included in the two graphics (i.e., dots along the y or x axis, respectively). e. Venn diagrams showing overlap between sets of genes in SSA/Ps and cADNs whose peri-TSS window contained at least one hypermethylated DMR.

Figure 3. Bisulfite pyrosequencing verification of 6 hypermethylated DMRs. Methylation levels at each CpG site of (a) 3 SSA/P-specific DMRs and (b) 3 DMRs shared between SSA/Ps and cADNs. Verification assay results in 61 FFPE colon tissue samples (left) are compared with those of genome-wide bisulfite sequencing of the fresh tissue series (right). DMR-containing loci (red box) and CpG islands (blue horizontal bars) are schematically represented below graph pairs. c. Hierarchical clustering heatmap of the 61 FFPE tissue samples based on the mean methylation level across all CpGs in each DMR. The cADN cluster contains two subgroups, probably reflecting their different frequencies of KRAS mutations. The SSA/P cluster also contains a subgroup of lesions with higher levels of methylation at all 6 markers, presumably reflecting these patients’ significantly older age when the lesions were discovered (mean: 66.3 years versus 54.5 years in the subgroup with lower methylation levels; P = 0.01). Significant differences between subgroups regarding other clinical variables (Supplementary Table 3) were not observed.

Figure 4. Analysis of the SSA/P and cADN transcriptomes in light of the lesions’ methylome profiles. (a) MDS plot and (b) hierarchical clustering heatmap of the 17 SSA/Ps and 15 cADNs and their normal mucosal samples. Asterisks indicate the two samples with intermediate profiles (for details, see Table 1). Hierarchical clustering in B was based on expression levels of the 10,000 genes with most highly variable expression. c. Venn diagrams showing the precursor-lesion specificity of the most dysregulated genes (P-value < 10⁻¹⁰ and log₂ fold change > 1). d. Scatter plots showing the variable relation between the magnitude of differential DNA methylation (x axis) and the expression (y axis) for SSA/Ps (vs matched normal mucosa samples). Genes with at least one DMR (P-value < 0.05) in their peri-TSS windows and dysregulated expression (P-value < 0.05) are shown. Red dots: subsets of genes with a median methylation difference of ≥ 0.1 and log₂ fold change in expression of >1. ZIC2, ZIC5, and HUNK are highlighted to illustrate the heterogeneous relation between DNA methylation and gene expression. e. Methylation levels (top) and gene expression (bottom) data for the ZIC2, ZIC5, and HUNK loci in SSA/Ps (pink), cADNs (blue), and corresponding samples of normal mucosa (light-pink and light blue, respectively). Areas outlined in black are those displaying differential methylation (see details in Supplementary Figures 5 and 6).

Figure 5. In situ hybridization analysis of ZIC2, HUNK, and ANXA10 mRNAs in SSA/Ps and cADNs. ZIC2 expression (red punctate labeling) is (a) present in SSA/Ps and (b) absent in cADNs. HUNK (brown punctate labeling) is (c) almost absent in SSA/Ps (very low-level expression was present at the bottom of serrated crypts, see inset) and (d) expressed in cADNs. Control staining for ANXA10, a known SSA/P-specific marker[25,73–75] was (e) strongly positive in SSA/Ps (dense red punctate labeling) and (f) absent in cADNs.

Sakamoto N, Feng Y, Stolfi C, et al. BRAF(V600E) cooperates with CDX2 inactivation to promote serrated colorectal tumorigenesis. Elife. 2017 Jan 10;6. doi: 10.7554/eLife.20331. PubMed PMID: 28072391; PubMed Central PMCID: PMCPMC5268782.

 Google Scholar

Kim JH, Kim KJ, Rhee YY, et al. Gastric-type expression signature in serrated pathway-associated colorectal tumors. Hum Pathol. 2015 May;46(5):643–656. PubMed PMID: 25704805.

 PubMed Web of Science ®Google Scholar

Sajanti SA, Vayrynen JP, Sirnio P, et al. Annexin A10 is a marker for the serrated pathway of colorectal carcinoma. Virchows Arch. 2015 Jan;466(1):5–12. PubMed PMID: 25395067.

 PubMed Web of Science ®Google Scholar