A variety of posttranslational modifications of histones, such as phosphorylation, ADP-ribosylation and methylation, play important roles in the regulation of gene expression through chromatin remodeling. The N-terminal tail of histone H3 is subject to methylation at multiple lysine residues by histone methyltransferase.1 The histone lysine methylation of key histone residues, especially histone H3 lysine 9 (H3K9), is associated with the epigenetic control of heterochromatin assembly and with DNA methylation-induced gene silencing in cancer cells.2, 3, 4, 5, 6
A number of enzymes (EuHMTase1, SETDB1, Suv39H, EZH2 and G9a) in mammalian cells are known to methylate H3K9. Among them, SUV39H1, a mammalian homologue of the Drosophila position effect variegation (PEV) modifier Su(var)3–9 and of the Schizosaccharomyces pombe clr4, was the first histone lysine methyltransferase (HMTase) identified.7 SUV39H1 was initially determined to have an enzymatic activity that specifically methylates H3K9 at the pericentric heterochromatin in vitro and to induce a specialized methylation pattern, which differs from the broad H3K9 methylation present at other chromosomal regions.7, 8, 9, 10, 11 However, recent studies also suggest that SUV39H1 may direct transcriptional repression within the euchromatic promoter.3, 4, 5, 12, 13, 14, 15 SUV39H1 is targeted to the promoters of cell-cycle control genes by retinoblastoma protein (pRb) and also induces the silencing of S-phase genes through H3K9 methylation in differentiating cells.12, 15
The aberrant methylation of CpG islands at the promoter region of tumor suppressor genes is an epigenetic change that has been shown to induce the transcriptional silencing of genes. The methylation status of over 30 genes has been evaluated in cases of colorectal cancer. The aberrant methylation of the tumor suppressor genes p14ARF, p16INK4a and helicase-like transcription factor (HLTF) in particular, has been found in a substantial percentage of colorectal cancer cases.16 Although DNA methylation is recognized as a common and important event in colorectal cancer development, whether SUV39H1 histone methyltransferase affects the euchromatic promoter methylation in colorectal cancer remains unclear.
We analyzed the mRNA levels of SUV39H1 and DNMT1 as well as the CpG island methylation of the p14ARF, p16INK4a and HLTF genes in primary colorectal cancer in an attempt to understand the effect of SUV39H1 on the promoter methylation of genes in the euchromatic region and its relationship to DNMT1.
CI, confidence interval; DNMT1, DNA methyltransferase 1; HLTF, helicase-like transcription factor; HMT, histone methyltransferase; H3K9, histone H3 lysine 9; OR, odds ratio; PCNA, proliferating cell nuclear antigen. The first two authors contributed equally to this project.
Material and methods
This study consisted of a total of 219 colorectal cancer patients who underwent curative surgical resection at the Department of Surgery in the Samsung Medical Center, Seoul, Korea from August 2001 to June 2006. The surgically removed tissue samples were collected from all participants after obtaining written informed consent, which was approved by the Institutional Review Board at the Samsung Medical Center. Paired samples of primary colorectal cancer tissues and their corresponding noncancerous normal tissues were obtained from each patient. The surgically removed tissues were immediately frozen in liquid nitrogen and stored at −80°C until use. A trained interviewer obtained the sociodemographic information on each patient.
The 219 patients consisted of 125 men and 94 women, with a total mean age of 60 ± 12 (mean ± standard deviation) years at diagnosis. Pathological stage was determined by the tumor-node-metastasis (TNM) classification. Eighteen of the 219 patients with colorectal cancers were at Stage 1, 86 were at Stage 2, 93 were at Stage 3 and 22 patients had Stage 4 cancer. The histological subtypes included 193 adenocarcinomas, 16 mucinous adenocarcinomas and 10 other cell types, which included signet-ring cell carcinomas.
RNA extraction and reverse-transcription PCR
The fresh frozen tissues were sectioned with a microtome, and the serial sections were positioned on slides prior to RNA extraction and stained with hematoxylin-eosin to evaluate the admixture of tumorous and nontumorous tissues. The areas corresponding to the tumor and normal tissues were carefully microdissected. The total RNA was isolated from the fresh frozen tissues using QIAGEN's RNeasy total RNA Isolation kit. The first strand of cDNA was synthesized from 2.0 μg RNA in a reaction mixture with a total volume of 50 μl containing 3 μM random hexamers (Pharmacia, Uppsala, Sweden), 10 mM dithiothreitol, 20 U RNasin Ribonuclease inhibitor (Promega, Madison, WI), 1× RT buffer (500 μM each dNTP, 3 mM MgCl2, 75 mM KCL, 50 mM Tris-HCL, pH 8.3), and 100 U SuperScript II RNase H− reverse transcriptase (GIBCO BRL, Gaithersburg, MD). The amplification reactions were carried out under the following conditions: 5 min at 25°C, 50 min at 42°C, 10 min at 72°C, followed by cooling at 5°C for 5 min. The synthesized cDNA was purified using QIAGEN's Qiaquick columns, eluted with 10 mM Tris-HCl (pH 8.8), and 2 μl of the purified cDNA mixture were used for the subsequent PCR amplification.
Quantitative real time PCR
The ABI PRISM® 7900HT Fast Real-Time PCR System (Applied Biosystems) was used to quantitate the mRNA levels. The TaqMan® primer and probe sequences used to amplify DNMT1 and PCNA were described previously.17 The proliferating cell nuclear antigen (PCNA), rather than actin or 18S rRNA, was used as an internal control to correct for the variations in the amount of RNA since the DNMT1 expression for the maintenance of methylation during S-phase is proliferation-dependent. A premade assay kit (ID: Hs00162471_m1, RefSeq NM_003173.2) from Applied Biosystems was used to analyze the expression of SUV39H1. The PCR reaction consisted of an initial denaturation step at 50°C for 2 min and 95°C for 10 min, which was followed by 50 cycles of 95°C for 15 sec and 60°C for 1 min. The reaction mixture contained 5 μl of TaqMan Universal PCR Master mix (Applied Biosystems), 10–18 μM of each primer, 5–10 μM of TaqMan probe FAM dye and 2 μl of the template cDNA in a final volume of 10 μl.
The Universal Human Reference RNA (Cat Nr:740000, Stratagene, La Jolla, CA) was used to construct a standard curve. Five serial 10-fold dilutions of the reference RNA were amplified to create standard curves of DNMT1, SUV39H1 and PCNA. The relative amounts of the target RNAs were determined using the comparative CT method. The ΔCT value of each sample was derived by subtracting the CT of the target (DNMT1 or SUV39H1) from the CT of the PCNA. The difference between the ΔCT values in the tumor and normal tissue was defined as the ΔΔCT. The target expression level in the tumor relative to the normal tissue was calculated as 2. The real-time PCR reactions were performed in triplicate for each sample-primer set, and the mean of the 3 experiments was used as the relative mRNA level in each sample.
Genomic DNA was extracted from the fresh frozen tissues using a QIAmp tissue kit (Quiagen, Valencia, CA) according to the instructions supplied by the manufacturer. The methylation status of the promoter regions of the p14ARF p16INK4a and HLTF genes was determined by Methylation-Specific PCR (MSP) as previously described by Herman et al.18 (Fig. 1). Two primer sets were designed, 1 specific to unmethylated DNA at the promoter region and the other specific to methylated DNA. The primer sequences and annealing temperatures previously described were used in the MSP procedure.19, 20 The DNA extracted from the peripheral blood lymphocytes of a healthy individual was treated with SssI methyltransferase (New England Biolabs, Beverly, MA) and used as a positive control for the methylated CpG islands. DNA from normal lymphocytes was used as a negative control for the methylated alleles and as a positive control for the unmethylated CpG islands.
The association between the increased expression of SUV39H1 and clinicopathological characteristics was analyzed using a Fisher's exact or Pearson's χ2 test. The Wilcoxon rank-sum test was used to compare the expression levels of SUV39H1 according to age and CEA levels. The degrees of linearity between 2 continuous variables were tested using Spearman's rank correlation coefficients. A multivariate logistic regression analysis was performed to investigate the relationship between the CpG island hypermethylation or DNMT1 elevation and each of the independent variables, after taking the remaining independent variables into account. All of the p values reported were based on two-sided tests.
The mRNA expression levels of SUV39H1 in tumor and their corresponding normal tissue samples were initially quantitated and then displayed as a ratio of the SUV39H1 and the PCNA. The SUV39H1 expression was considered to have been elevated at mRNA level when the relative mRNA expression of SUV39H1 in tumor versus normal tissue was greater than one. The mean mRNA level of SUV39H1 was 1.09 ± 1.76. The elevated mRNA expression of SUV39H1 was found in 54 (25%) of the 219 patients studied. Fourteen of the 54 patients with an elevated mRNA level of SUV39H1 had an mRNA level of 1.0–1.5. Eight patients had mRNA levels of SUV39H1 was between 1.5 and 2.0, while 32 patients were found to have mRNA levels of SUV39H1 over 2.0.
The relationships between the elevated mRNA expression of SUV39H1 and the clinicopathological characteristics in 219 colorectal cancer patients are shown in Table I. The age of the patient with or without elevated SUV39H1 expression was similar (p = 0.65; Wilcoxon rank-sum test). The CEA levels in patients with an elevated mRNA level of SUV39H1 were slightly higher than the levels found in those without an elevated mRNA level of SUV39H1, but this was not found to be statistically significant (p = 0.21; Wilcoxon rank-sum test). The elevated mRNA expression of SUV39H1 was more prevalent in women than men; however, the difference was not statistically significant (28 vs. 22%, respectively; p = 0.37; Pearson's χ2 test). The mRNA levels of SUV39H1 were not associated with tumor location (p = 0.49; Pearson's χ2 test), lymphovascular invasion (p = 0.51; Pearson's χ2 test) or the pathologic stage (p = 0.36; Fisher's exact test). Elevated mRNA expression of SUV39H1 was found in 25% of the adenocarcinomas, 19% of the mucinous adenocarcinomas and 20% of the other cell types; however, this difference was not statistically significant (p = 0.93; Fisher's exact test). Poorly differentiated cells showed a higher prevalence of elevated SUV391 expression than well- and moderately-differentiated cells, and this did not reach a level of statistical significance (p = 0.44; Fisher's exact test).
Table I. Clinicopathological Characteristics (N = 219)
The association of the mRNA levels of SUV39H1 and the CpG island hypermethylation of the p14ARF, p16INK4a and HLTF genes was analyzed. The mean mRNA levels of SUV39H1 in patients with p14ARF methylation and those without were 1.24 and 0.94, respectively, but this difference was not statistically significant (p = 0.48, Wilcoxon rank-sum test). The promoter methylation of the p16INK4a and HLTF genes was not found to be associated with the mRNA levels of SUV39H1 (data not shown). Of the 219 colorectal cancer samples, promoter methylation was detected at a rate of 36% for p14ARF, 51% for p16INK4a and 34% for HLTF (Table II). The promoter methylation of p14ARF and p16INK4a occurred more frequently in patients with an elevated mRNA level of SUV39H1 than in those without, but the difference was not significant (p = 0.62 and p = 0.66, respectively). The prevalence of HLTF methylation also did not differ significantly between patients with an elevated mRNA level of SUV39H1 and those without (28% and 33%, respectively, p = 0.28; Pearson's χ2 test).
Table II. Relationship Between the mRNA Elevation of SUV39H1 and the CpG Island Hypermethylation of the P14ARF, P16INK4a and HLTF Genes (N = 219)
A multivariate logistic regression analysis was conducted to control for the potential confounding effects of variables, such as age and gender, and to calculate the odds ratios. Table III shows the risk of promoter methylation in the p14ARF, p16INK4a and HLTF genes according to the status of an elevated SUV39H1 mRNA level. Patients with an elevated mRNA level of SUV39H1 showed a 1.28 (95% CI = 0.56–2.97) times greater risk of p14ARF methylation than those without, but the difference was not statistically significant (p = 0.41; Wald test). In addition, the mRNA elevation of SUV39H1 was not associated with p16INK4a methylation (OR = 1.15, 95% CI = 0.61–2.18, p = 0.61) or HLTF methylation (OR = 0.63, 95% CI = 0.33–1.19, p = 0.16).
Table III. Logistic Regression Analysis1 of the Association Between SUV39H1 Elevation and CPG Island Hypermethylation in 219 Colorectal Carcinomas
The relationship between the mRNA levels of SUV39H1 and DNMT1 was also analyzed. The relative mRNA expression of DNMT1 in tumor versus normal tissue was quantitated in the same manner as SUV39H1. The mean mRNA level of DNMT1 was 1.13 ± 1.26 and the median was 0.76. Ninety-one (42%) of the 219 patients studied had an elevated expression of DNMT1 at the mRNA level. Nineteen patients were found to have mRNA levels of DNMT1 between 1.5 and 2.0, while 34 patients had mRNA levels of DNMT1 over 2.0. The SUV39H1 mRNA levels showed a strong positive correlation with the DNMT1 mRNA levels (Spearman correlation coefficient = 0.5549, p < 0.0001). The mRNA levels of DNMT1 in patients whose mRNA levels of SUV39H1 were between 1.5 and 2.0 and over 2.0 were 2.83 and 1.39, respectively (Fig. 2). When the mRNA levels of SUV39H1 were dichotomized, the mRNA levels of DNMT1 were found to be significantly different between patients with elevated mRNA levels of SUV39H1 and those without (1.62 and 0.91, respectively; p = 0.007, Wilcoxon rank-sum test). Patients with elevated mRNA levels of SUV39H1 showed a higher prevalence of DNMT1 elevation than those without elevated mRNA levels of SUV39H1 (61% vs. 35%, p = 0.0008, Table IV).
Permission to reproduce this table online was not granted by the copyright holder. Readers are kindly asked to refer to the printed version.
A logistic regression analysis was conducted to analyze the association between the mRNA levels of SUV39H1 and the risk of a DNMT1 mRNA elevation, after controlling for confounding factors such as age and gender (Table V). Only 8 patients were found to have mRNA levels of SUV39H1 between 1.5 and 2.0. Therefore, the mRNA levels of SUV39H1 were first divided into three groups; >1.0, 1.0–1.5, and 1.5<. Patients whose mRNA levels of SUV39H1 were between 1.0 and 1.5 had a 5.07 (95% CI = 1.50–17.02, p = 0.009) times greater risk of elevated DNMT1 expression than those without elevated mRNA levels of SUV39H1. In addition, patients with mRNA levels of SUV39H1 greater than 1.5-fold were also found to be associated with a significant risk of elevated DNMT1 expression (OR = 2.21, 95% CI = 1.09–4.48, p = 0.02) in the multivariate analysis. When the data was dichotomized based on the presence or absence of an elevation of SUV39H1, patients with an elevated mRNA level of SUV39H1 were found to have a higher risk of DNMT1 mRNA elevation than those without (OR = 2.71, 95% CI = 1.09–4.48, p = 0.002).
Table V. Multivariate Logistic Regression1 of the Association Between SUV39H1 mRNA Levels and Elevated DNMT1 Expression (N = 219)
We analyzed the mRNA levels of SUV39H1 and DNMT1 and the CpG island hypermethylation of the p14ARF, p16INK4a and HLTF genes in primary colorectal cancer to better understand the relationship between SUV39H1 histone methyltransferase and the hypermethylation of CpG islands at the euchromatic promoter region in human cancer. The results of this study demonstrate that the mRNA levels of SUV39H1 were significantly associated with the mRNA levels of DNMT1 but not with the CpG island hypermethylation of the p14ARF, p16INK4a and HLTF genes in primary colorectal cancers.
The role of SUV39H1 in pericentric heterochromatin has been extensively investigated, but only a few studies have suggested the involvement of SUV39H1 in tumorigenesis. Mice null for SUV39H1 display severely impaired viability and ∼30% of them develop B cell lymphomas after 9–15 months of age.11 Differential expression of SUV39H1 was reported in different cancer types and a prominent increase in the expression of SUV39H1 was observed in preneoplastic nodules and liver tumors induced by methyl deficiency in rats.21, 22 SUV39H1 is also involved in the oncogenic properties of the acute-promyelocytic leukemia (APL)-associated fusion protein PML-retinoic acid receptor (RAR).23 The mRNA levels of SUV39H1 in this study were elevated in 54 (25%) of the 219 colorectal cancer patients studied. These observations suggest that SUV39H1 may play a role in neoplastic transformation. However, the reason for the elevated expression of SUV39H1 in colorectal cancer remains unclear. The clinicopathological variables in the present study were not found to be associated with increased mRNA levels of SUV39H1. Recently, Ozdag et al.21 recently reported that only 2 out of 181 cancer samples had mutations with significant coding-sequence alterations, suggesting that other factors than mutation might be involved in the elevated mRNA levels of SUV39H1.
Despite many observations suggesting that SUV39H1 plays a role in euchromatic gene repression, the present study suggests that there is a complex relationship between SUV39H1 expression and euchromatic promoter methylation in colorectal cancer. The reason for the lack of relationship between SUV39H1 expression and DNA methylation is unclear; however, several factors may partially explain this complex relationship. SUV39H1 is primarily known to direct H3K9 trimethylation specifically at the pericentric heterochromatin7, 8, 23, 24 and G9a is known to be responsible for all detectable H3K9 dimethylation and a significant amount of H3K9 monomethylation within silent euchromatin.25 Therefore, SUV39H1 may play a minor role in euchromatic promoter methylation.
Another possibility is that SUV39H1 may be involved in methylation-independent euchromatic gene repression, considering the fact that there are no reports of DNA methylation in most studies demonstrating the involvement of SUV39H1 in euchromatin repression. Several proteins, such as Rb and HP1, may affect the effect of SUV39H1 on DNA methylation in euchromatic loci. Rb is known to associate with SUV39H1 and HP1 in vivo via its pocket domain, which is used to recruit SUV39H1 to HP1 and to target HP1 to the promoter.12 HP1 is also required for DNA methylation,26 and the IY165/168EE mutation in the chromo shadow domain of mHP1 α abrogated a direct binding to SUV39H1.27 Finally, another component of SUV39H1 action may affect the relationship, since H3K9 methylation alone is able to suppress transcription but is insufficient for HP1 recruitment, suggesting the need of chromatin-associated factors to read the histone code.28
In the present study, DNMT1 was also not found to be associated with the CpG island hypermethylation of the p14ARF, p16INK4a HLTF genes, which was consistent with the previous finding.17 However, human DNMT1 and SUV39H1 are known to interact both in vitro and in vivo, and both of these proteins have been implicated in the transcriptional repression of genes and the formation of heterochromatin.29, 30 The patients in this study with an elevated mRNA level of SUV39H1 showed a higher prevalence of a DNMT1 elevation than those without elevated mRNA levels of SUV39H1 (61% vs. 35%, p = 0.0008), suggesting that the two proteins may act in concert. Therefore, we reanalyzed the possible requirement of both SUV39H1 and DNMT1 for the promoter methylation of the p14ARF, p16INK4a and HLTF genes. However, the simultaneous elevation in the mRNA levels of SUV39H1 and DNMT1 were not associated with promoter methylation in any of the 3 genes (Fig. 3). Esteve et al.31 recently reported that G9a histone methyltransferase coordinates DNA methylation by directly interacting with DNMT1 during DNA replication. Taken together, these observations suggest that SUV39H1 works in concert with DNMT1 for transcriptional repression but they are not sufficient for the hypermethylation of the CpG islands in euchromatic loci.
This study was severely limited by several factors. First of all, the methylation-independent repression of SUV39H1 on genes in the euchromatic region cannot be evaluated because of the lack of expression data on the p14ARF, p16INK4a and HLTF genes. In addition, 21 (39%) of the 54 patients with an elevated mRNA level of SUV39H1 did not show an elevated mRNA level of DNMT1, which suggests that other factors are required for DNMT1 expression. The lack of expression data on Rb and HP1 made it difficult to analyze a clear relationship between SUV39H1 expression and DNA methylation. Also, whether SUV39H1 affects DNMT1 expression is unclear and the mechanism responsible for the elevated SUV39H1 expression in colorectal cancer also remains unclear. In conclusion, the present data suggest that the transcriptional upregulation of SUV39H1 is positively associated with the DNMT1 mRNA level; however, it is not sufficient for euchromatic promoter methylation in colorectal cancer.
The authors wish to thank Hee-Jin Jo and Hyang-Suk Jung for the data collection and management.