Complete Genetic Suppression of Polyp Formation and Reduction of CpG-Island Hypermethylation in Apc^Min/+ Dnmt1-Hypomorphic Mice

Introduction

Promoter CpG island hypermethylation leading to transcriptional silencing of critical genes has been documented widely in many types of human cancer (1, 2, 3, 4) . Three functional DNA methyltransferases have been described in mammals, of which Dnmt1 appears to be the predominant enzyme (5 , 6) . Colorectal cancer is the second leading cause of death by cancer in the United States (7) . The APC tumor-suppressor gene is implicated in the majority of colorectal tumors (8) . Min 3 mice carry a germ-line mutation in the murine Apc ortholog and are predisposed to the development of intestinal neoplasia (9 , 10) . We have shown previously that the combined effects of heterozygosity for a null mutation of the Dnmt1 gene and treatment with the DNA methyltransferase inhibitor 5-aza-2′-deoxycytidine could reduce polyp multiplicity in Apc^Min/+ mice (11) . In this previous study, we could not rule out that the reduction in polyp number by 5-aza-2′-deoxycytidine was because of drug toxicity unrelated to its demethylating capability (12) . Here, we report that hypomorphic alleles of Dnmt1 lead to the complete suppression of intestinal polyp formation. Furthermore, we use the sensitive MethyLight assay to show that CpG island hypermethylation occurs at low frequency in the histologically normal mucosa of Apc^Min/+ mice and is increased in intestinal polyps. Finally, we show that Dnmt1 hypomorphic alleles reduce the frequency of CpG island methylation in the normal mucosa and intestinal polyps.

Materials and Methods

Dnmt1-Hypomorphic Mice.

The Dnmt1^N allele has been described previously (5 , 13) . The Dnmt1^R allele was generated as follows. A 320-bp PvuII-EcoRV fragment from pL1–2LacO (14) was inserted into the unique EcoRV site in the same parent plasmid used for the Dnmt1^N allele (5) to yield plasmid pPWL532. The parent plasmid of this cloning step was later named pBS(10.5) and contains a 10.5-kb fragment spanning exons 2–4 (previously thought to be exon1) cloned into a pSP72 (Promega) vector backbone (5) . The EcoRV site is located in the third intron of the murine Dnmt1 gene, 20 bp upstream of the splice acceptor site of the fourth exon. Subsequently, a PGK-TKNeo cassette was introduced adjacent to the vector backbone by ligation of a 12.1-kb partial XhoI, complete SpeI fragment of pPWL532 with a 3-kb partial SalI and complete XbaI fragment of pPGKTKNeo (15) resulting in an insertion-type targeting vector called pPWL540. This insertion-type vector was digested at the unique HindIII site and transfected by electroporation into J1 murine embryonic stem cells (5) . Gene-targeted clones were subjected subsequently to counterselection of the TKNeo gene by FIAU as described (15) . ES cell clones that had undergone excision of the vector backbone and TKNeo gene but had retained the LacO insertion were selected for injection into blastocysts. Chimeric mice with verified germ-line transmission capabilities were bred with 129/svJae animals to obtain pure 129/svJae mice carrying the LacO insertion. This allele was designated the Dnmt1^R allele for the EcoRV insertion site in the third intron.

Genotype Analysis.

DNA was isolated from tail biopsies as described previously (16) . The genotypes of the Dnmt1 alleles were determined by multiplex-PCR analysis using primers OL106 (5′-GGGAACTTCCTGACTAGGGG-3′), OL168, (5′CCAACAAACCAGTATGTCTCGT-3′), OL173 (5′-CCCAGTTTCCAGAAAGCTACC-3′), and OL369 (5′-CAATTCCACACAACATACGAGC-3′). Reactions were carried out in a 15-μl volume with 20 mm Tris-HCl (pH 8.4), 50 mm KCl, 1.5 mm MgCl₂, 0.2 mm deoxynucleotide triphosphates, 0.3 μm each primer, and 0.35 units of Taq polymerase using a cycling condition of 94°C for 4 min, followed by 35 cycles of 94°C for 50 s, 60°C for 50 s, and 72°C for 1 min 30 s, followed by 72°C for 5 min. OL168 and OL173 produced both a 342-bp wild-type-specific band and a 661-bp Dnmt1^R allele-specific band. OL106 and OL173 produced a 430-bp Dnmt1^N allele-specific band. OL173 and OL369 produced a second 211-bp Dnmt1^R allele-specific band. The Apc genotype was performed by fluorescence-based, allelic discrimination-PCR (TaqMan; Refs. 17 and 18 ). The primer and probe sequences are listed below. Forward primer (5′-GCCCAGCTCTTCTTCCTC AAG-3′), Reverse primer (5′-GATGGTAAGCACTGAGGCCAATAC-3′), Apc^+/+-specific probe (6FAM-TCTCTCTCCAAACTTCTGTCTTTCT-TAMRA), and Apc^Min/+-specific probe (6VIC-TCTCTCTCCTAACTTCTGTCTTTCT-TAMRA). Both probes were synthesized as TurboTaq probes.

Intestinal Polyp Scoring and Size Determination.

The entire intestine was removed immediately after euthanasia, washed with 70% ethanol, and fixed in RNAlater (Ambion, Austin, TX). The entire length of the intestine was measured (cm) and subjected to a careful microscopic screen. Adenomas of at least the size of two villi were included in the counting. The investigator was blind to the genotype while counting. Adenomas and mucosal tissue samples selected for DNA analysis were microdissected from the middle third (cm) of the intestine of male mice. Polyp sizes were determined by measuring the maximum diameter of polyps found in the middle third (cm) of the intestine using calipers with an accuracy of 0.05 mm (MECHANIC Type 6901; Fine Science Tools, Inc., Foster City, CA). Male mice only were used for the size determinations to exclude gender-based differences in polyp size.

DNA Methylation Analysis.

Genomic DNA isolation and sodium bisulfite conversion were performed as described previously (16 , 19) . Agarose beads were incubated for 14 h at 50°C to ensure complete bisulfite conversion. Methylation analysis was performed using the MethyLight assay (20, 21, 22) . Parallel TaqMan PCR reactions were performed with primers specific for the bisulfite-converted methylated sequence for a particular locus and with two reference primers (Lhx1 and Guca2). The ratio between the values obtained using these two reference primers (GENE/Lhx1 and GENE/Guca2) was averaged. The PMR at a specific locus was calculated by dividing the GENE/REFERENCE-averaged ratio of a sample by the GENE/REFERENCE-averaged ratio for SssI-treated (SssI enzyme; New England Biolabs, Beverly, MA) control J1 ES cell DNA and multiplying by 100 (22) . CpG islands were defined as being methylated in a particular sample if the PMR value >1. The MethyLight primer and probe sequences are listed below. In all cases, the first primer listed is the forward PCR primer, the second is the TaqMan probe, and the third is the reverse PCR primer: Apc, GGGCGTAGGTATACGTGATCGA, 6FAM-AATAACACCCCGACAAACTACGCCAATACAA-TAMRA, CCATTTTCGAACCCGACAA; Itga4, CGAGGTGTAGATCGAGGTTTCG, 6FAM-ACAACATCACCGCTTCCCGAAAAACG-TAMRA, CCCGCCTCCTACTCACGTAA; Mgmt, CGACACCCTTACGTCACACACT, 6FAM-AACCACGCCCCGCGTACCAA-TAMRA, TAGTTCGAGGGTGTAAAGCGG; Timp3, GAGAGGCGGTGGGCGTAG, 6FAM-CGATATACGCTACAACGACGTCCCACGA-TAMRA, CGAAAATATAAACTAAACGCGTCCT; Lhx1, AGAGTGTTTGGAAGTTAGGTGAAGGT, 6FAM-CACAATCAACATCCCAAACATATTCACCCA-TAMRA, CACATTCATAAACACAAATTCACACAAC; and Guca2, GGTGTTGTGGTTTAGAAGGTTATGG, 6FAM-TCTCATCATCTTCTACAAACCAAAAC-TAMRA, ACCTTATCCTCAACTTCCAACATACC.

Northern and Southern Blot Analysis.

Total RNA (20 μg) was separated on a 0.03% formaldehyde-denaturing agarose gel buffered with 10 mm sodium phosphate (pH 6.8), blotted, and hybridized with an α-tubulin probe or a Dnmt1 cDNA probe. Genomic DNA (5 μg) isolated from mouse intestinal tissue was digested with the restriction enzymes HpaII and MspI in parallel reactions (New England Biolabs). Southern blot analysis was performed as described previously (11) . Filters were hybridized with a probe containing centromeric minor satellite repeat sequences derived from plasmid pMR150 (23) .

Results and Discussion

In previous work, we have shown that Apc^Min/+ Dnmt1^S/+ heterozygous mice treated with 5-azaCdR develop fewer intestinal polyps than untreated, Dnmt1^+/+ Apc^Min/+ mice (11) . This result was substantiated subsequently and expanded on by others (24) . One explanation for this suppression of polyp multiplicity by reduced levels of functional Dnmt1 expression could be that CpG island hypermethylation is an important step in polyp formation and requires sufficient Dnmt1 expression. However, CpG island hypermethylation, though widely studied in human colorectal tumors, has never been documented in mouse intestinal tumors. Therefore, we used the sensitive MethyLight assay (20, 21, 22) to investigate whether CpG island methylation occurs at all in the mouse intestine and whether it is increased in Min polyps. The high sensitivity of the MethyLight assay was crucial for our analysis because most of our samples had limiting amounts of DNA.

We first used a limited number of mucosal and polyp samples to prescreen 10 CpG islands associated with genes known to be methylated in human tumors or likely to influence intestinal tumorigenesis, if silenced (data not shown). Of the 10 CpG islands, only 3 showed a sufficiently frequent methylation in the samples analyzed and were consequently selected for further analysis. These were the CpG islands associated with the genes Itga4, Timp3, and Mgmt. The CpG island associated with Cdkn2a (p16) was hypermethylated in a single polyp sample but was not included for additional study because of its very low frequency of methylation. Subsequently, we screened nine normal mucosal samples and 10 intestinal polyps with each of these three MethyLight reactions.

Fig. 1A ⇓ shows an overview of the frequency of CpG island methylation for each of these genes in the 19 samples analyzed. Fig. 1B ⇓ shows the mean percentage of genes methylated for the mucosal samples versus the polyp samples. The percentage of these three CpG islands that were methylated in the normal mucosal tissue was 41%, whereas the polyp tissues showed an average of 77% methylation for these same genes. We found substantial variation among mice, e.g., mouse 3 was unmethylated, and mice 5 and 7 were methylated at all three genes in both mucosa and polyp. This is consistent with the variation in methylation profiles observed in human colorectal tumors (20) . It is interesting to note that CpG island methylation was detectable in the normal mucosa. This has also been reported for human colorectal tissue and has been shown to increase with age and predispose to subsequent tumorigenesis for some genes (25, 26, 27) . These results indicate that CpG island methylation occurs in normal mucosa and suggest that cells with an increased frequency of CpG island methylation may be predisposed to tumorigenesis.

If a reduction in CpG island methylation is the molecular mechanism by which decreased functional expression of Dnmt1 suppresses polyp formation in Min mice (11 , 24) , then one would expect to see a reduced frequency of CpG island methylation in the normal mucosa or polyp tissue of Min mice with reduced Dnmt1 expression. To avoid the use of drug inhibitors, we developed a hypomorphic allele referred to as the Dnmt1^R allele (see “Materials and Methods”) and combined this allele with the previously described Dnmt1^N allele (5) to generate mice with varying levels of Dnmt1 expression. Fig. 2A ⇓ shows a schematic representation of the three Dnmt1 alleles used in this study. Fig. 2B ⇓ shows the relative expression of these alleles in embryonic stem cells. The Dnmt1^N allele does not show detectable levels of gene expression by Northern blot analysis (Fig. 2B ⇓ and Ref. 13 ) but has been shown to produce low levels of a truncated protein by alternative splicing (13) . Quantitation of the normalized RNA expression level of Dnmt1^R/+ cells versus Dnmt1^+/+ cells indicated that the Dnmt1^R is expressed at ∼60% of that of the wild-type allele (Fig. 2B) ⇓ . This intermediate level of expression is consistent with the observation that the Dnmt1^R allele resulted in reduced viability in conjunction with the null Dnmt1^S allele but was fully viable in combination with the previously described hypomorphic Dnmt1^N allele (data not shown). We tested whether the combination of these two hypomorphic alleles resulted in detectable hypomethylation in vivo. Fig. 2C ⇓ shows that the centromeric minor satellite repeat sequence is significantly hypomethylated in the intestinal mucosa of Dnmt1^N/R mice but not in that of any of the other three allelic combinations. Therefore, we conclude that both Dnmt1^R and Dnmt1^N are hypomorphic alleles with reduced functional expression of Dnmt1 and together result in significant hypomethylation in vivo. Nevertheless, the degree of hypomethylation achieved in these mice is not sufficient to cause an overt phenotype. Dnmt1^N/R mice are comparable in size and weight to their wild-type littermates and are fertile.

Apc^Min/+ Dnmt1 hypomorphs were generated from crosses between C57BL/6 Apc^Min/+ Dnmt1^N/+ males and 129/SvJae Apc^+/+ Dnmt1^R/+ females. Mice were euthanized at 180 days and analyzed for polyp multiplicity and polyp size as described in “Materials and Methods.” Apc^+/+ offspring were euthanized on weaning, because we had shown previously that Dnmt1 mutant mice do not develop intestinal tumors in an Apc^+/+ background (11) . We found a gradual decline in polyp number with diminishing Dnmt1 expression levels with a drop to 0 in the Apc^Min/+ Dnmt1^N/R mice, which have the lowest levels of Dnmt1 expression (Fig. 3A) ⇓ . The number of polyps was reduced significantly in Apc^Min/+ Dnmt1^N/+ mice (P = 0.001) and Apc^Min/+ Dnmt1^N/R mice (P < 0.0001) compared with wild-type Apc^Min/+ mice. We did not observe a single polyp in any of the 14 Apc^Min/+ Dnmt1^N/R mice that we analyzed. This suggests that Dnmt1 hypomorphic alleles act as genetic suppressors of the Apc^Min/+ phenotype. Other studies have reported genetic modifiers of the Min phenotype that reduce the number of polyps (9 , 11 , 28, 29, 30, 31) , but this is the first documentation of complete genetic suppression, to our knowledge.

We also examined the effect of reduced Dnmt1 expression on polyp size. Polyps from Dnmt1^R/+ (P = 0.0021) and Dnmt1^N/+ (P < 0.0001, unpaired t test) were significantly smaller than those from Apc^Min/+ Dnmt^+/+ siblings at 180 days (Fig. 3B) ⇓ , indicating that reduced Dnmt1 expression affects not only polyp formation but also the growth rate of the polyps. This is consistent with the findings of Cormier and Dove (24) , who reported a reduced net growth rate and multiplicity of intestinal adenomas in Dnmt1^N/+ Apc^Min/+ mice.

If CpG island hypermethylation is an essential feature to polyp formation, and if this is the basis for the suppressive effects of reduced levels of Dnmt1 expression, then we would expect to see a lower CpG island methylation frequency in tissues with reduced DNA methylation levels. The results shown in Fig. 4 ⇓ show that this is indeed the case. The level (Fig. 4A) ⇓ and frequency (Fig. 4B) ⇓ of CpG island methylation in the intestinal mucosa is substantially diminished in Dnmt1 hypomorphic mice compared with Dnmt1^+/+ mice. The most striking reduction is seen in Dnmt1^N/R mucosa, where we did not observe a single instance of CpG island methylation >1 PMR threshold (P = 0.0057, unpaired t test). Dnmt1^N/+ mice also showed a statistically significant reduction in the frequency of CpG island methylation in the normal mucosa (P = 0.025) and polyps (P = 0.0002; Fig. 4C ⇓ ). We could not measure the CpG island methylation frequency in Dnmt1^N/R adenomas, because these mice did not develop any polyps.

These results are consistent with a model in which CpG island methylation occurs in the normal intestinal mucosa in a small minority of cells, which then become predisposed to undergo neoplastic transformation. We do not suggest that the three individual genes that we analyzed are necessarily implicated in tumorigenesis but just that they are representative of CpG island hypermethylation occurring at many CpG islands throughout the genome. Presumably, some of these CpG island methylation events occur in growth-controlling genes and contribute to polyp formation and additional growth. This model is supported by the observation that substantially reduced expression of Dnmt1 in the intestinal mucosa results in the complete suppression of detectable polyp formation. One caveat to our results is that we used a binocular stereo microscope to scan the intestine for adenomas, rather than relying on serial sections analyzed with an upright microscope. Therefore, microadenomas or aberrant crypt foci in Dnmt^N/R mice may have been missed. Nevertheless, if such early stage adenomas did arise in these mice, they did not progress further to a stage at which they would be detectable using a dissecting microscope. We found that the RNAlater fixative that we used to allow for subsequent nucleic acid analysis rendered the intestines too fragile for subsequent embedding and analysis at high magnification. With these caveats in mind, we tentatively conclude that CpG island hypermethylation, mediated at least in part by Dnmt1, is an essential and rate-limiting step in intestinal adenoma formation in Apc^Min/+ mice and that Dnmt1 hypomorphic alleles are the first identified true genetic suppressors of the Apc^Min/+ phenotype.