The molecular basis of aberrant hypermethylation of CpG islands observed in a subset of human colorectal tumors is unknown. One potential mechanism is the up-regulation of DNA (cytosine-5)-methyltransferases. Recently, two new mammalian DNA methyltransferase genes have been identified, which are referred to as DNMT3A and DNMT3B. The encoded proteins differ from the predominant mammalian DNA methyltransferase DNMT1 in that they have a substantially higher ratio of de novo to maintenance methyltransferase activity. We have used a highly quantitative 5′ nuclease fluorogenic reverse transcription-PCR method (TaqMan) to analyze the expression of all three DNA methyltransferase genes in 25 individual colorectal adenocarcinoma specimens and matched normal mucosa samples. In addition, we examined the methylation patterns of four CpG islands [APC, ESR1 (estrogen receptor), CDKN2A (p16), and MLH1] to determine whether individual tumors show a positive correlation between the level of DNA methyltransferase expression and the frequency of CpG island hypermethylation. All three methyltransferases appear to be up-regulated in tumors when RNA levels are normalized using either ACTB (β-actin) or POLR2A (RNA pol II large subunit), but not when RNA levels are normalized with proliferation-associated genes, such as H4F2 (histone H4) or PCNA. The frequency or extent of CpG island hypermethylation in individual tumors did not correlate with the expression of any of the three DNA methyltransferases. Our results suggest that deregulation of DNA methyltransferase gene expression does not play a role in establishing tumor-specific abnormal DNA methylation patterns in human colorectal cancer.

Vertebrate cytosine-5 DNA methylation occurs in the context of CpG dinucleotides and is associated with transcriptional inactivity. CpG islands are areas of high CpG density that are normally unmethylated. Aberrant de novo methylation of CpG islands is found in a subset of human colorectal tumors. The abnormal methylation of CpG islands associated with tumor suppressor genes can lead to transcriptional silencing, inactivating the gene through epigenetic rather than genetic means (1). When aberrant CpG island hypermethylation occurs in colorectal tumors, it is frequently not restricted to a single CpG island but affects multiple independent loci (2, 3), reflective of a widespread deregulation of DNA methylation patterns (1). Although the molecular basis of such a methylator phenotype is not known, its global nature suggests a change in trans-acting factors that control the pattern of distribution of methylated residues. It has recently been suggested that the nuclear phosphoprotein FOS transforms cells in part by up-regulating DNMT1 gene expression (4).

DNA methylation results from a methyl transfer reaction performed by trans-acting enzymes known as DNA methyltransferases. Two distinct methyl transfer activities can be distinguished, based on the methylation status of the substrate. Maintenance DNA methyltransferase activity refers to the conversion of hemimethylated substrates to a fully methylated state, whereas de novo methyltransferase activity refers to the new addition of methyl groups at sites that were previously unmethylated. All known DNA (cytosine-5)-methyltransferases are able to perform both reactions. The predominant mammalian DNA (cytosine-5)-methyltransferase, DNMT1, is unusual in that its relative de novo activity is 1–2 orders of magnitude lower than its maintenance activity (5, 6) Recently, two additional mammalian DNA (cytosine-5)-methyltransferase genes have been identified, which are referred to as DNMT3A and DNMT3B. These genes differ from DNMT1 in that the encoded polypeptides DNMT3α and DNMT3β have approximately equal ratios of de novo DNA methyltransferase activity:maintenance DNA methyltransferase activity (7). An additional candidate DNA methyltransferase gene, DNMT2, has been identified, but the encoded protein has not yet been shown to possess methyltransferase activity (8, 9, 10). DNMT3α and DNMT3β are thought to be responsible for the wave of de novo methylation that occurs during embryogenesis (7). Because the abnormal hypermethylation of CpG islands in colorectal tumors involves the new acquisition of DNA methylation (de novo methylation), we have investigated whether transcriptional activation or up-regulation of either DNMT3A or DNMT3B could be responsible for the methylator phenotype observed in colorectal tumors.

Previous studies have analyzed the expression levels of the DNMT1 gene in human tumors and cell lines, with somewhat conflicting results (11, 12, 13, 14, 15). Some reports documented a substantial increase in the expression of DNMT1 or of total DNA methyltransferase enzyme activity in tumor cells, compared to normal counterparts. However, Lee et al.(14) made the important observation that relative determinations of expression levels can be affected by the gene used for the normalization of RNA amounts. Normalization with a gene associated with cell proliferation, such as histone H4 (H4F2) abolished any statistically significantly higher mean expression of DNMT1 in colorectal tumors compared to matched normal mucosa. One additional caveat to these studies is that in all cases, mean expression levels were determined for tumor tissues compared to normal specimens. However, substantial interindividual variability in DNA methyltransferase expression exists. None of these studies analyzed CpG island hypermethylation frequencies in the same samples to correlate these with DNA methyltransferase expression or activity.

In this study, we extend the analysis of DNA methyltransferase expression levels in colorectal tumors in the following ways: (a) we have analyzed the largest data set used thus far for such an analysis (25 pairs of tumor and matched normal mucosa); (b) we have used a highly quantitative RT-PCR3 method (Fig. 1) that is linearly accurate in serial dilutions over 6 orders of magnitude; (c) we have used four different genes [ACTB (β-actin), H4F2 (histone H4), PCNA, or POLR2A (RNA pol II large subunit)] for normalization of RNA levels; (d) we have analyzed the expression levels of two new DNA methyltransferase genes, DNMT3A and DNMT3B, that have not been investigated previously; (e) we investigated whether the individual expression levels of any of the three DNA methyltransferases correlate with the de novo methylation of four commonly hypermethylated CpG islands [APC, ESR1 (estrogen receptor), CDKN2A (p16), and MLH1].

Sample Collection.

A total of 25 paired tumor and normal mucosal tissue samples were obtained from 25 patients with primary colorectal adenocarcinoma. The patients comprised 16 males and 9 females, ranging in age from 39–88 years, with a mean age of 68.8 years. The surgical procedures were performed in Japan, and a preoperative diagnosis was obtained from a biopsy and a histological examination. The mucosal distance from the tumor to the normal specimens was between 10 and 20 cm. Approximately 2 g of the surgically removed tissue were frozen immediately in liquid nitrogen and stored at −80°C until RNA and DNA isolation. The remaining section of the sample was fixed with formalin and used for further histological examination to confirm the diagnosis postoperatively. All histological examinations were performed after staining with H&E.

Nucleic Acid Isolation.

Genomic DNA was isolated by the standard method of proteinase K digestion and phenol-chloroform extraction (16). Total RNA was isolated by the single-step guanidinium isothiocyanate method (17).

Quantitative RT-PCR Analysis.

The quantitation of mRNA levels was carried out using a real-time fluorescence detection method as described previously (18, 19). In brief, after RNA isolation, cDNA was prepared from each sample as described previously (20). The specific cDNA of interest and reference cDNA (ACTB, H4F2, PCNA, or POLR2A) were PCR-amplified separately using an oligonucleotide probe with a 5′ fluorescent reporter dye (6FAM) and a 3′ quencher dye (TAMRA; Ref. 21). The 5′ to 3′ nuclease activity of Taq DNA polymerase cleaved the probe and released the reporter, whose fluorescence could be detected by the laser detector of the ABI Prism 7700 Sequence Detection System (Perkin-Elmer Corp., Foster City, CA). After crossing a fluorescence detection threshold, the PCR amplification results in a fluorescent signal proportional to the amount of PCR product generated. Initial template concentration was derived from the cycle number at which the fluorescent signal crossed a threshold in the exponential phase of the PCR reaction. Relative gene expression was determined based on the threshold cycles of the gene of interest and of the internal reference gene. Use of a reference gene avoids the need to directly quantitate the RNA, which could be a major source of error for analysis. Several reference samples were included on each assay plate to verify plate-to-plate consistency. Plates were normalized to each other using these reference samples, if necessary. Contamination of the RNA samples by genomic DNA was excluded by an analysis of all RNA samples without prior cDNA conversion. The PCR amplification was performed using a 96-well optical tray and caps with a 25-μl final reaction mixture consisting of 600 nm each primer; 200 nm probe; 5 units of Ampli-Taq Gold; 200 μm each of dATP, dCTP, and dGTP; 400 μm dUTP; 5.5 mm MgCl2; 1 unit of AmpErase uracil N-glycosylase; and 1× TaqMan buffer A containing a reference dye at 50°C for 2 min and at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and at 60°C for 1 min. The primer and probe sequences are listed below. In all cases, the first primer is the forward PCR primer, the second primer is the TaqMan probe, and the third primer is the reverse PCR primer.

The primer and probe sequences are as follows: (a) ACTB, TGAGCGCGGCTACAGCTT, 6FAM5′-ACCACCACGGCCGAGCGG-3′ TAMRA, and CCTTAATGTCACACACGATT; (b) H4F2, CTTAGCCTCAGTGCGAATGCT, 6FAM5′-CAGAACCAGAGCACAGCCAAAGCCACTAC-3′ TAMRA, and ACGGTCCCCGGGAGAAT; (c) PCNA, GTGCAAAAGACGGAGTGAAATTT, 6FAM5′-TGTTTCCATTTCCAAGTTCTCCACTTGCAG-3′ TAMRA, and ATCGACATTACTTGTCTGTGACAATTTA; (d) POLR2A, GCCACCCAGATGACCTTGAA, 6FAM5′-CCTTCCACTATGCTGGTGTGTCTGCCA-3′ TAMRA, and GCACACCCAGCGTCACATT; (e) DNMT1, GGTTCTTCCTCCTGGAGAATGTC, 6FAM5′-CCTTCAAGCGCTCCATGGTCCTGAA-3′ TAMRA, and GGGCCACGCCGTACTG; (f) DNMT3A, CAATGACCTCTCCATCGTCAAC, 6FAM5′-AGCCGGCCAGTGCCCTC-GTAG-3′ TAMRA, and CATGCAGGAGGCGGTAGAA; and (g) DNMT3B, CCATGAAGGTTGGCGACAA, 6FAM5′-CACTCCAGGAACCGTGAGA-TGTCCCT-3′ TAMRA, and TGGCATCAATCATCACTGGATT.

DNA Methylation Analysis.

Genomic DNA extracted from tumor and normal samples was treated with sodium bisulfite as described previously (22). After conversion, the DNA was amplified by fluorescence-based, real-time quantitative PCR (as described above, but without the addition of AmpErase). Two sets of primers and probes designed specifically for bisulfite-converted DNA were used: (a) a methylated set for the gene of interest [APC, ESR1, CDKN2A (p16), or MLH1]; and (b) an internal reference set (MYOD1) to control for input DNA. The methylated primers and the probe were designed to contain 1–5 CpG dinucleotides, which amplify only fully methylated molecules. Specificity of the reactions for methylated DNA was confirmed separately using DNAs of known methylation status. The internal reference primers and the probe were designed in a region of the MYOD1 gene that lacks any CpG dinucleotides to allow for unbiased amplification. Parallel TaqMan PCR reactions were performed with primers specific for the bisulfite-converted methylated sequence for a particular locus and with the MYOD1 reference primers. The ratio between the values obtained in these two TaqMan analyses was used as a measure of the degree of methylation at that locus. A ratio greater than or equal to four times the mean ratio for all normal mucosal samples was classified as methylated (Figs. 2 and 3, •), and a ratio less than four was regarded as unmethylated (Figs. 2 and 3, ○). The primer and probe sequences are listed below. In all cases, the first primer listed is the forward PCR primer, the second primer is the TaqMan probe, and the third primer is the reverse PCR primer.

The primer and probe sequences were as follows: (a) APC, TTATATGTCGGTTACGTGCGTTTATAT, 6FAM5′-CCCGTCGAAAACCCGCCGATTA-3′ TAMRA, and GAACCAAAACGCTCCCCAT; (b) CDKN2A, AACAACGTCCGCACCTCCT, 6FAM5′-ACCCGACCCCGAACCGCG-3′ TAMRA, and TGGAATTTTCGGTTGATTGGTT; (c) ESR1, GGCGTTCGTTTTGGGATTG, 6FAM5′-CGATAAAACCGAACGACCCGACGA-3′ TAMRA, and GCCGACACGCGAACTCTAA; (d) MLH1, CTATCGCCGCCTCATCGT, 6FAM5′-CGCGACGTCAAACGCCACTACG-3′ TAMRA, and CGTTATATATCGTTCGTAGTATTCGTGTTT; and (e) MYOD1, CCAACTCCAAATCCCCTCTCTAT, 6FAM5′-TCCCTTCCTATTCCTAAATCCAACCT-AAATACCTCC-3′ TAMRA, and TGATTAATTTAGATTGGGTTTAGAGA-AGGA.

Statistics.

TaqMan analyses performed as described above for either RT-PCR or DNA methylation studies yield values that are expressed as ratios between two absolute measurements (DNMT:normalization gene for RT-PCR and CpG island:MYOD1 for DNA methylation analysis). The ratios for each type of analysis were subsequently normalized such that the mean ratio of the 25 normal samples would equal a value of 1. Consequently, the values for the tumor samples represent a fold increase or decrease relative to a mean normal value of 1. Additional statistical manipulations are described in Tables 1 and 2.

DNMT1 Expression in Colorectal Tumors.

Fig. 2 shows the relative expression levels of DNMT1 in 25 individual colorectal adenocarcinoma samples (▪) and 25 matched normal mucosal samples (hatched bars). The expression levels are displayed as ratios between DNMT1 and four reference genes (ACTB, Fig. 2,A; POLR2A, Fig. 2,B; H4F2, Fig. 2,C; and PCNA, Fig. 2,D) to correct for variations in the amounts of RNA. The mean expression levels of DNMT1 in tumors versus normal mucosal samples were calculated using the four normalization genes (Table 1; Fig. 2). DNMT1 appeared to be up-regulated approximately 3.9-fold when ACTB or POLR2A were used for normalization (Table 1; Fig. 2, A and B). However, this apparent up-regulation was absent when the proliferation-associated genes H4F2 or PCNA were used for normalization (Table 1; Fig. 2, C and D). Similar results have been documented previously (14, 23). These results indicate that either DNMT1 is truly up-regulated in tumors and is otherwise normally expressed in a proliferation-independent fashion or that DNMT1 gene expression is proliferation dependent, which accounts for all of its apparent up-regulation in tumors. The latter scenario seems more likely in view of the requirement for maintenance DNA methylation during S-phase and in light of the reported proliferation-dependent gene expression of DNMT1(23, 24, 25).

DNMT3A and DNMT3B Expression in Colorectal Tumors.

Fig. 3 shows the relative expression levels of DNMT3A and DNMT3B in the same samples as described above. The high values seen in some of the samples have been reproduced multiple times and are therefore thought to represent valid ratios. The advantage of the use of multiple normalization genes is apparent from the variable ratios obtained within individual samples. For instance, the apparent high value for DNMT3/PCNA seen in sample 23N is due to an abnormally low expression of PCNA compared to other reference genes in this particular sample and is not due to a high level of DNMT3 gene expression.

The analysis of DNMT3A and DNMT3B expression yielded similar results to the analysis of DNMT1 expression. DNMT3A expression in colorectal tumors appears to be elevated by an average of 2.9- and 2.8-fold when RNA levels are normalized using ACTB and POLR2A, respectively (Table 1; Fig. 3, A and B). Likewise, DNMT3B levels appear to be up-regulated by an average of 3.6- and 4.0-fold in a similar analysis (Table 1; Fig. 3, E and F). However, this up-regulation is absent for both genes when RNA levels are normalized using either H4F2 or PCNA for normalization (Table 1; Fig. 3, C, D, G, and H). The proliferation dependence of DNMT3A and DNMT3B gene expression is not known. These results could indicate true up-regulation of DNMT3A or DNMT3B in tumors, or the situation could be analogous to that for DNMT1. Regardless of which of these two scenarios is correct, if the apparent up-regulation of DNMT3A or DNMT3B is responsible for the methylator phenotype in some human colorectal tumors, then the extent of up-regulation should be greater in tumors with frequent CpG island hypermethylation. We have investigated whether there is a direct link between expression of any of the three DNA methyltransferases and the frequency of CpG island hypermethylation.

CpG Island Hypermethylation and DNA Methyltransferase Expression.

We analyzed the methylation status of four CpG islands known to undergo de novo methylation in human colorectal tumors in all normal and tumor samples. These CpG islands are associated with the genes APC(26), ESR1 (estrogen receptor; Ref. 27), CDKN2A (p16) (2), and MLH1(28, 29). We found that DNMT3A and DNMT3B do not appear to be up-regulated in the two tumors (samples 10 and 17) with the most frequent (three of four) CpG island hypermethylation (Fig. 3). Tumors with relatively high DNMT3A or DNMT3B expression levels tend to have at most one hypermethylated CpG island. There is just one sample (26T) with high DNMT3A and/or DNMT3B expression and two hypermethylated CpG islands. A similar lack of concordance is apparent when CpG island hypermethylation frequencies are compared to DNMT1 expression levels (Fig. 2). ANOVA was used to compare the mean DNMT gene expression between categories of CpG island hypermethylation frequency (Table 2). We performed 12 separate ANOVA calculations to investigate all combinations of the three DNMT genes and the four normalization genes. None of these combinations yielded a statistically significant P value (Table 2), which would have indicated a correlation between the frequency of CpG island hypermethylation and the level of DNMT gene expression. Therefore, regardless of whether or not DNMT3A or DNMT3B gene expression is proliferation dependent, as is DNMT1, the RNA levels of neither of these two genes in individual tumors correlate with CpG island hypermethylation frequency.

We conclude that most cases of frequent hypermethylation of CpG islands in human colorectal tumors do not result from a simple transcriptional up-regulation of any of the three known DNA methyltransferase genes. This leaves open the possibility that one or more of these genes are up-regulated posttranscriptionally. It is also conceivable that other factors regulate the activity of the DNA methyltransferases, either by interacting with the enzymes themselves or by regulating access to the DNA substrate. The molecular basis for the methylator phenotype in human colorectal tumors could be found in the disruption of the control mechanisms preventing access of the DNA methyltransferases to CpG islands rather than in the mere up-regulation of DNA methyltransferase levels in the cell.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

      
1

Supported by NIH/National Cancer Institute Grants R01 CA 71716 (to P. V. D.) and R01 CA 75090 (to P. W. L.).

            
3

The abbreviation used is: RT-PCR, reverse transcription-PCR.

Fig. 1.

Schematic of quantitative RT-PCR (TaqMan) technology. A, in the PCR reaction, an oligonucleotide probe tagged with a 5′ fluorescent reporter and a 3′ quencher is added in addition to the standard PCR components. The probe is complementary to the target sequence of interest and anneals during extension. The close proximity of the quencher to the fluorescent reporter represses fluorescence in the intact probe. As the Taq polymerase synthesizes the new strand, its 5′ to 3′ nuclease activity cleaves the probe, separating the quencher and fluorescent reporter. The fluorescence emitted is proportional to the amount of product accumulated with each cycle. B, the plot represents a sample analysis from the experiments shown in Fig. 3. The plot shows the expression values obtained for three genes (as indicated) in a matched tumor and normal tissue sample. The horizontal bold line indicates the fluorescence level used for the threshold cycle determination in this particular example. ΔRn is defined as the cycle-to-cycle change in the reporter fluorescence signal normalized to a passive reference fluorescence signal.

Fig. 1.

Schematic of quantitative RT-PCR (TaqMan) technology. A, in the PCR reaction, an oligonucleotide probe tagged with a 5′ fluorescent reporter and a 3′ quencher is added in addition to the standard PCR components. The probe is complementary to the target sequence of interest and anneals during extension. The close proximity of the quencher to the fluorescent reporter represses fluorescence in the intact probe. As the Taq polymerase synthesizes the new strand, its 5′ to 3′ nuclease activity cleaves the probe, separating the quencher and fluorescent reporter. The fluorescence emitted is proportional to the amount of product accumulated with each cycle. B, the plot represents a sample analysis from the experiments shown in Fig. 3. The plot shows the expression values obtained for three genes (as indicated) in a matched tumor and normal tissue sample. The horizontal bold line indicates the fluorescence level used for the threshold cycle determination in this particular example. ΔRn is defined as the cycle-to-cycle change in the reporter fluorescence signal normalized to a passive reference fluorescence signal.

Close modal
Fig. 2.

DNMT1 expression in 25 colorectal tumors and matched normal mucosal controls. RNA levels were measured by quantitative, real-time RT-PCR in 25 matched normal () and tumor (▪) colorectal samples. The expression levels are displayed as ratios between DNMT1 and four reference genes (A, ACTB; B, POLR2A: C, H4F2, and D, PCNA) to correct for variations in the amounts of RNA. The ratios for each type of analysis have been normalized such that the mean ratio of the 25 normal samples equals a value of 1. CpG island hypermethylation status for four different genes (APC, ESR1, CDKN2A, and MLH1) is shown below each chart. •, the measured level of DNA methylation at that CpG island was at least 4-fold higher than the mean level in the 25 normal samples; ○, the measured level was below four times the mean level of the normal samples.

Fig. 2.

DNMT1 expression in 25 colorectal tumors and matched normal mucosal controls. RNA levels were measured by quantitative, real-time RT-PCR in 25 matched normal () and tumor (▪) colorectal samples. The expression levels are displayed as ratios between DNMT1 and four reference genes (A, ACTB; B, POLR2A: C, H4F2, and D, PCNA) to correct for variations in the amounts of RNA. The ratios for each type of analysis have been normalized such that the mean ratio of the 25 normal samples equals a value of 1. CpG island hypermethylation status for four different genes (APC, ESR1, CDKN2A, and MLH1) is shown below each chart. •, the measured level of DNA methylation at that CpG island was at least 4-fold higher than the mean level in the 25 normal samples; ○, the measured level was below four times the mean level of the normal samples.

Close modal
Fig. 3.

DNMT3A and DNMT3B expression in 25 colorectal tumors and matched normal mucosal controls. The expression levels are displayed as ratios between DNMT3A (A–D) or DNMT3B (E–H) and four reference genes (A and E, ACTB; B and F, POLR2A: C and G, H4F2;D and H, PCNA) to correct for variations in the amounts of RNA. The ratios for each type of analysis have been normalized such that the mean ratio of the 25 normal samples equals a value of 1. The CpG island hypermethylation status for four different genes (APC, ESR1, CDKN2A, and MLH1) is shown below each chart. •, the measured level of DNA methylation at that CpG island was at least 4-fold higher than the mean level in the 25 normal samples; ○, the measured level was below four times the mean level of the normal samples.

Fig. 3.

DNMT3A and DNMT3B expression in 25 colorectal tumors and matched normal mucosal controls. The expression levels are displayed as ratios between DNMT3A (A–D) or DNMT3B (E–H) and four reference genes (A and E, ACTB; B and F, POLR2A: C and G, H4F2;D and H, PCNA) to correct for variations in the amounts of RNA. The ratios for each type of analysis have been normalized such that the mean ratio of the 25 normal samples equals a value of 1. The CpG island hypermethylation status for four different genes (APC, ESR1, CDKN2A, and MLH1) is shown below each chart. •, the measured level of DNA methylation at that CpG island was at least 4-fold higher than the mean level in the 25 normal samples; ○, the measured level was below four times the mean level of the normal samples.

Close modal
Table 1

Relative mean expression of DNA methyltransferases in tumor versus normal tissue

The mean expression levels are shown of each of the DNMT genes in tumors versus normal mucosal samples. The expression levels were first calculated as ratios between the DNMT gene and one of four reference genes (ACTB, POLR2A, H4F2, and PCNA) as indicated to correct for variations in the amounts of RNA. The ratios were then normalized, such that the mean ratio of the 25 normal samples would equal a value of 1. Subsequently, the ratios for the tumor samples were averaged to give the mean values indicated in the table. Consequently, these values represent an average fold increase or decrease of DNMT gene expression in tumors relative to a mean value of 1 in normal mucosal samples. The P value for each average indicates the probability that the null hypothesis (no difference between tumor and normal) is correct.
Normalization geneDNMT1DNMT3ADNMT3B
Nonproliferation-associated gene    
ACTB 3.9 (P < 0.0001) 2.9 (P = 0.0006) 3.6 (P = 0.0004) 
POLR2A 3.9 (P < 0.0001) 2.8 (P = 0.0015) 4.0 (P = 0.0030) 
Proliferation-associated gene    
 H4F2 0.9 (P = 0.5517) 0.6 (P = 0.0502) 0.8 (P = 0.3823) 
 PCNA 0.8 (P = 0.1920) 1.0 (P = 0.9392) 0.8 (P = 0.7210) 
The mean expression levels are shown of each of the DNMT genes in tumors versus normal mucosal samples. The expression levels were first calculated as ratios between the DNMT gene and one of four reference genes (ACTB, POLR2A, H4F2, and PCNA) as indicated to correct for variations in the amounts of RNA. The ratios were then normalized, such that the mean ratio of the 25 normal samples would equal a value of 1. Subsequently, the ratios for the tumor samples were averaged to give the mean values indicated in the table. Consequently, these values represent an average fold increase or decrease of DNMT gene expression in tumors relative to a mean value of 1 in normal mucosal samples. The P value for each average indicates the probability that the null hypothesis (no difference between tumor and normal) is correct.
Normalization geneDNMT1DNMT3ADNMT3B
Nonproliferation-associated gene    
ACTB 3.9 (P < 0.0001) 2.9 (P = 0.0006) 3.6 (P = 0.0004) 
POLR2A 3.9 (P < 0.0001) 2.8 (P = 0.0015) 4.0 (P = 0.0030) 
Proliferation-associated gene    
 H4F2 0.9 (P = 0.5517) 0.6 (P = 0.0502) 0.8 (P = 0.3823) 
 PCNA 0.8 (P = 0.1920) 1.0 (P = 0.9392) 0.8 (P = 0.7210) 
Table 2

Analysis of the relationship between CpG island hypermethylation and DNA methyltransferase expression in individual tumors (ANOVA)

ANOVA was used to compare mean DNMT gene expression between categories of CpG island hypermethylation frequency. Twelve separate ANOVA calculations were performed to investigate all combinations of the three DNMT genes and the four normalization genes. The P value indicates the probability that the null hypothesis (no difference in the mean DNMT gene expression levels in the different categories of CpG island hypermethylation frequency) is correct.
Normalization geneDNMT1DNMT3ADNMT3B
Nonproliferation-associated gene    
ACTB P = 0.0659 P = 0.5288 P = 0.2216 
POLR2A P = 0.2233 P = 0.6959 P = 0.3947 
Proliferation-associated gene    
H4F2 P = 0.3603 P = 0.4338 P = 0.4500 
PCNA P = 0.7098 P = 0.7209 P = 0.8879 
ANOVA was used to compare mean DNMT gene expression between categories of CpG island hypermethylation frequency. Twelve separate ANOVA calculations were performed to investigate all combinations of the three DNMT genes and the four normalization genes. The P value indicates the probability that the null hypothesis (no difference in the mean DNMT gene expression levels in the different categories of CpG island hypermethylation frequency) is correct.
Normalization geneDNMT1DNMT3ADNMT3B
Nonproliferation-associated gene    
ACTB P = 0.0659 P = 0.5288 P = 0.2216 
POLR2A P = 0.2233 P = 0.6959 P = 0.3947 
Proliferation-associated gene    
H4F2 P = 0.3603 P = 0.4338 P = 0.4500 
PCNA P = 0.7098 P = 0.7209 P = 0.8879 

We thank Dennis Salonga and Ji Min Park for help in generating cDNAs and in designing TaqMan oligonucleotides.

1
Jones P. A., Laird P. W. Cancer epigenetics comes of age.
Nat. Genet.
,
21
:
163
-167,  
1999
.
2
Ahuja N., Mohan A. L., Li Q., Stolker J. M., Herman J. G., Hamilton S. R., Baylin S. B., Issa J. P. Association between CpG island methylation and microsatellite instability in colorectal cancer.
Cancer Res.
,
57
:
3370
-3374,  
1997
.
3
Liang G., Salem C. E., Yu M. C., Nguyen H. D., Gonzales F. A., Nguyen T. T., Nichols P. W., Jones P. A. DNA methylation differences associated with tumor tissues identified by genome scanning analysis.
Genomics
,
53
:
260
-268,  
1998
.
4
Bakin A. V., Curran T. Role of DNA 5-methylcytosine transferase in cell transformation by fos.
Science (Washington DC)
,
283
:
387
-390,  
1999
.
5
Bestor T. H. Activation of mammalian DNA methyltransferase by cleavage of a Zn binding regulatory domain.
EMBO J.
,
11
:
2611
-2617,  
1992
.
6
Smith S. S. Biological implications of the mechanism of action of human DNA (cytosine-5)methyltransferase.
Prog. Nucleic Acid Res. Mol. Biol.
,
49
:
65
-111,  
1994
.
7
Okano M., Xie S., Li E. Cloning and characterization of a family of novel mammalian DNA (cytosine-5)methyltransferases.
Nat. Genet.
,
19
:
219
-220,  
1998
.
8
Okano M., Xie S., Li E. Dnmt2 is not required for de novo and maintenance methylation of viral DNA in embryonic stem cells.
Nucleic Acids Res.
,
26
:
2536
-2540,  
1998
.
9
Van den Wyngaert I., Sprengel J., Kass S. U., Luyten W. H. Cloning and analysis of a novel human putative DNA methyltransferase.
FEBS Lett.
,
426
:
283
-289,  
1998
.
10
Yoder J. A., Bestor T. H. A candidate mammalian DNA methyltransferase related to pmt1p of fission yeast.
Hum. Mol. Genet.
,
7
:
279
-284,  
1998
.
11
El-Deiry W. S., Nelkin B. D., Celano P., Yen R. W., Falco J. P., Hamilton S. R., Baylin S. B. High expression of the DNA methyltransferase gene characterizes human neoplastic cells and progression stages of colon cancer.
Proc. Natl. Acad. Sci. USA
,
88
:
3470
-3474,  
1991
.
12
Issa J. P., Vertino P. M., Wu J., Sazawal S., Celano P., Nelkin B. D., Hamilton S. R., Baylin S. B. Increased cytosine DNA-methyltransferase activity during colon cancer progression.
J. Natl. Cancer Inst.
,
85
:
1235
-1240,  
1993
.
13
Schmutte C., Yang A. S., Nguyen T. T., Beart R. W., Jones P. A. Mechanisms for the involvement of DNA methylation in colon carcinogenesis.
Cancer Res.
,
56
:
2375
-2381,  
1996
.
14
Lee P. J., Washer L. L., Law D. J., Boland C. R., Horon I. L., Feinberg A. P. Limited up-regulation of DNA methyltransferase in human colon cancer reflecting increased cell proliferation.
Proc. Natl. Acad. Sci. USA
,
93
:
10366
-10370,  
1996
.
15
Kautiainen T. L., Jones P. A. DNA methyltransferase levels in tumorigenic and nontumorigenic cells in culture.
J. Biol. Chem.
,
261
:
1594
-1598,  
1986
.
16
Wolff R. K., Frazer K. A., Jackler R. K., Lanser M. J., Pitts L. H., Cox D. R. Analysis of chromosome 22 deletions in neurofibromatosis type 2-related tumors.
Am. J. Hum. Genet.
,
51
:
478
-485,  
1992
.
17
Chomczynski P., Sacchi N. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction.
Anal. Biochem.
,
162
:
156
-159,  
1987
.
18
Gibson U. E., Heid C. A., Williams P. M. A novel method for real time quantitative RT-PCR.
Genome Res.
,
6
:
995
-1001,  
1996
.
19
Heid C. A., Stevens J., Livak K. J., Williams P. M. Real time quantitative PCR.
Genome Res.
,
6
:
986
-994,  
1996
.
20
Bender C. M., Pao M. M., Jones P. A. Inhibition of DNA methylation by 5-aza-2′-deoxycytidine suppresses the growth of human tumor cell lines.
Cancer Res.
,
58
:
95
-101,  
1998
.
21
Livak K. J., Flood S. J., Marmaro J., Giusti W., Deetz K. Oligonucleotides with fluorescent dyes at opposite ends provide a quenched probe system useful for detecting PCR product and nucleic acid hybridization.
PCR Methods Applications
,
4
:
357
-362,  
1995
.
22
Xiong Z., Laird P. W. COBRA: a sensitive and quantitative DNA methylation assay.
Nucleic Acids Res.
,
25
:
2532
-2534,  
1997
.
23
Teubner B., Schulz W. A. Regulation of DNA methyltransferase during differentiation of F9 mouse embryonal carcinoma cells.
J. Cell. Physiol.
,
165
:
284
-290,  
1995
.
24
Papadopoulos T., Pfeifer G. P., Hoppe F., Drahovsky D., Muller-Hermelink H. K. Proliferation-associated expression of DNA methyltransferase in human embryonic lung cells.
Virchows Archiv. B Cell. Pathol. Incl. Mol. Pathol.
,
56
:
371
-375,  
1989
.
25
Szyf M., Bozovic V., Tanigawa G. Growth regulation of mouse DNA methyltransferase gene expression.
J. Biol. Chem.
,
266
:
10027
-10030,  
1991
.
26
Hiltunen M. O., Alhonen L., Koistinaho J., Myohanen S., Paakkonen M., Marin S., Kosma V. M., Janne J. Hypermethylation of the APC (adenomatous polyposis coli) gene promoter region in human colorectal carcinoma.
Int. J. Cancer
,
70
:
644
-648,  
1997
.
27
Issa J. P., Ottaviano Y. L., Celano P., Hamilton S. R., Davidson N. E., Baylin S. B. Methylation of the oestrogen receptor CpG island links ageing and neoplasia in human colon.
Nat Genet.
,
7
:
536
-540,  
1994
.
28
Herman J. G., Umar A., Polyak K., Graff J. R., Ahuja N., Issa J. P., Markowitz S., Willson J. K., Hamilton S. R., Kinzler K. W., Kane M. F., Kolodner R. D., Vogelstein B., Kunkel T. A., Baylin S. B. Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma.
Proc. Natl. Acad. Sci. USA
,
95
:
6870
-6875,  
1998
.
29
Veigl M. L., Kasturi L., Olechnowicz J., Ma A. H., Lutterbaugh J. D., Periyasamy S., Li G. M., Drummond J., Modrich P. L., Sedwick W. D., Markowitz S. D. Biallelic inactivation of hMLH1 by epigenetic gene silencing, a novel mechanism causing human MSI cancers.
Proc. Natl. Acad. Sci. USA
,
95
:
8698
-8702,  
1998
.