Opposing effects of DNA hypomethylation on intestinal and liver carcinogenesis

Abstract

Genome-wide DNA hypomethylation and concomitant promoter-specific tumor suppressor gene hypermethylation are among the most common molecular alterations in human neoplasia. Consistent with the notion that both promoter hypermethylation and genome-wide hypomethylation are functionally important in tumorigenesis, genetic and/or pharmacologic reduction of DNA methylation levels results in suppression or promotion of tumor incidence, respectively, depending on the tumor cell type. For instance, DNA hypomethylation promotes tumors that rely predominantly on loss of heterozygosity (LOH) or chromosomal instability mechanisms, whereas loss of DNA methylation suppresses tumors that rely on epigenetic silencing. Mutational and epigenetic silencing events in Wnt pathway genes have been identified in human colon tumors. We used Apc(Min/+) mice to investigate the effect of hypomethylation on intestinal and liver tumor formation. Intestinal carcinogenesis in Apc(Min/+) mice occurs in two stages, with the formation of microadenomas leading to the development of macroscopic polyps. Using Dnmt1 hypomorphic alleles to reduce genomic methylation, we observed elevated incidence of microadenomas that were associated with LOH at Apc. In contrast, the incidence and growth of macroscopic intestinal tumors in the same animals was strongly suppressed. In contrast to the overall inhibition of intestinal tumorigenesis in hypomethylated Apc(Min/+) mice, hypomethylation caused development of multifocal liver tumors accompanied by Apc LOH. These findings support the notion of a dual role for DNA hypomethylation in suppressing later stages of intestinal tumorigenesis, but promoting early lesions in the colon and liver through an LOH mechanism.

Fig. 1.

β-Catenin immunostaining of colonic lesions. Note that microadenoma cells as well as tumor cells show accumulation of β-catenin. Accumulation of β-catenin protein is prominent in both nucleus and cytoplasm of microadenoma, whereas the staining is confined at the membrane of adjacent normal crypts. (Bars, 25 μm in microadenoma and 250 μm in tumor.)

Fig. 2.

Opposing effects of DNA hypomethylation on intestinal carcinogenesis in Apc^Min/+ mice. Dnmt1^chip/c; Apc^Min/+ mice develop a significantly higher number of microadenomas than those in other groups (P < 0.03 and P < 0.02, respectively, Mann-Whitney U test) (A). In contrast, DNA hypomethylation decreased the number of intestinal tumors (P < 0.0001 and P < 0.0004, respectively, by Mann-Whitney nonparametric U test) (C) and no tumor developed in the colon of hypomethylated mice (B).

Fig. 3.

Apc LOH analysis of intestinal lesions. Presence of the Apc^Min and Apc⁺ alleles was assayed by PCR followed by cleavage with HindIII (16). The upper band represents the HindIII-resistant Apc^Min PCR product, whereas the lower band represents the wild-type Apc allele cut by HindIII. Lane N, normal-appearing colonic crypts from an Apc^Min/+ mouse; lane T, tumor sample from an Apc^Min/+ mouse showing LOH; lane U, undigested PCR product. Band ratio (Apc⁺/Apc^Min) in each sample was compared with the control lane (N). Note that all tumors and microadenomas show a dominant signal reflecting the Apc^Min allele, suggesting LOH (16).

Fig. 4.

Reduced size and incidence of microadenomas in Apc^Min/+; Dnmt1^chip/c mice. (A) The size of colonic microadenomas and intestinal tumors. Both microadenomas and tumors are smaller in Dnmt1^chip/c mice than other groups. Data represent mean ± SE. (B) The number and size of microadenomas of Dnmt1^chip/c mice at different time points. The number of microadenomas significantly increases with time, whereas the size of microadenomas does not change.

Fig. 5.

Apoptotic and proliferation index in intestinal tumors. (A) TUNEL-positive index of tumors in the small intestine. No significant difference in TUNEL-positive index was observed between tumors in Apc^Min/+; Dnmt1^+/+ and Apc^Min/+; Dnmt1^chip/c mice. The data represent the mean number of TUNEL positive cells per crypt ± SE. (B) BrdUrd-positive cell ratio of tumors in the small intestine. No significant difference was observed in BrdUrd-positive cell ratio in both small and large tumors between Dnmt1^+/+ and Dnmt1^chip/c mice. BrdUrd-positive index correlates with tumor size in both genotypes (P < 0.0001 by Spearman's rank correlation). Representative staining for BrdUrd in small (Upper) and large (Lower) tumors are shown. (Mean ± SE; bars, 500 μm.)

Fig. 6.

Liver tumorigenesis in Apc^Min/+; Dnmt1^chip/c mice. (A) Multiple liver tumors (arrowheads) developed in Apc^Min/+; Dnmt1^chip/c mice. Apc^Min/+; Dnmt1^chip/c mice at 360 days old developed liver tumors, whereas no tumors were found in other genotypes. (B and C) Histological staining of hepatocellular adenomas/carcinomas (arrowheads indicate the border between tumor cells and adjacent hepatocytes) with enlarged hyperchromatic nuclei. (Bars, 250 μmin B and 25 μmin C). (D) Apc LOH analysis at liver tumors in Apc^Min/+; Dnmt1^chip/c mice. Asterisks indicate liver tumors showing a lower band ratio (Apc⁺/Apc^Min) after normalizing to control liver tissue. Imbalances of the band ratio at control samples were observed (35). More than half of the lesions displayed evidence of LOH (n = 6 of 10) as compared to control normal liver from the same animals.

Fig. 7.

β-Catenin accumulation and overexpression of glutamine synthetase in liver tumors. Immunostaining revealed accumulation of β-catenin in liver tumors (arrowheads in A). β-catenin is seen in both cytoplasm and nucleus of tumor cells, whereas adjacent hepatocytes show only membranous staining (B and C). (D-F) Immunostaining for glutamine synthetase (GS), which is a target of β-catenin/tcf pathway. GS is overexpressed in all tumors (D; Dnmt1^chip/c; Apc^Min/+) except one small tumor (n = 23 of 24), whereas no evidence for aberrant GS expression or tumor lesions was observed in any of the age-matched controls (Dnmt1^+/+; Apc^Min/+ in E and Dnmt1^chip/c; Apc^+/+ in F).