Hypermethylation and Silencing of the Putative Tumor Suppressor Tazarotene-Induced Gene 1 in Human Cancers Free

Retinoids regulate the growth, differentiation, and apoptosis of normal cells during embryonic development and of premalignant and malignant cells during carcinogenesis. Most of these effects are mediated by nuclear retinoic receptors, which include three retinoid acid receptors (RARs) and three retinoid X receptors, both with designations of α, β, and γ; all of these are ligand-dependent transcription factors. RARs can induce the expression of certain genes in a ligand-dependent manner by binding to the retinoic acid-responsive elements in their promoter regions (1, 2, 3).

In recent years, many investigators have studied ways to exploit retinoids clinically. Tazarotene (AGN-190168), a synthetic retinoid that binds RARβ and RARγ, is used in the treatment of psoriasis, a hyperproliferative skin disorder. One gene induced by tazarotene is tazarotene-induced gene 1 (TIG1), which was cloned in 1996 by Nagpal et al. (4). Tazarotene induces TIG1 in keratinocytes both in vitro and in vivo via a retinoid receptor-dependent mechanism. Previously published information about TIG1 function described its isolation as a gene that is expressed differentially in untreated and retinoid-treated skin in cultures and its increased expression in psoriatic lesions after tazarotene treatment (5, 6). It was also reported that strong TIG1 expression could be induced by RAR-specific but not retinoid-X-receptor-specific agents and that TIG1 resembled CD38, a retinoid-responsive molecule on immune cells (5, 6). Very recently, evidence was reported supporting a tumor suppression role for TIG1 in prostate cancer (7). Jing et al. (7) demonstrated that decreased expression of TIG1 was associated with an increase in malignant characteristics of both prostate cancer cell lines and tissues and that expression of TIG1 is absent in all malignant prostate cells examined.

The mechanism by which TIG1 expression is suppressed in tumor versus normal cells has not been established. One possibility is cytosine methylation of CpG islands, an epigenetic mechanism of gene silencing mediated by modulation of the chromatin structure (8). The discovery that the promoters of numerous tumor suppressor genes are hypermethylated in neoplastic cells has emphasized epigenetic changes as important mechanisms in carcinogenesis (9). This finding led to the development of several techniques for identifying tumor suppressor genes based on aberrant methylation. In an earlier study, we used the methylated CpG amplification-representation difference analysis (MCA-RDA) technique (10) to identify DNA fragments that were differentially methylated in cancer. This technique was also used to identify several genes specifically silenced in cancer cells.

In the present study, we used the same technique and identified TIG1 as a gene whose promoter is methylated and silenced in human cancer. Loss of TIG1 expression might explain the low sensitivity of some neoplastic cells to retinoid-induced growth regulation.

To identify novel putative tumor suppressor genes based on the hypermethylation status of promoters, we applied MCA-RDA to the leukemia cell line K562 (American Type Culture Collection, Rockville, MD), using restriction enzymes with differential sensitivities to 5-methyl-cytosine, followed by PCR amplification, as detailed previously (10). Normal colon mucosa samples were used as a control. RDA was performed as described previously (11). Briefly, two steps of competitive hybridization of the sequences generated from tumor against those from normal cells allowed the selection of exclusively methylated DNA fragments (11).

All of the human cancer cell lines used in this study, except those specified below, were obtained from American Type Culture Collection. The University of Michigan squamous cell carcinoma cell lines UMSCC17A, UMSCC17B, UMSCC19, UMSCC22A, UMSCC22B, UMSCC35, and UMSCC38 were provided by Dr. Thomas Carey (University of Michigan, Ann Arbor, MI). Details about these cell lines, including their karyotyping, have been described previously (12, 13, 14). The cell lines MDA886Ln and MDA1483 were provided by Dr. Peter Sacks (New York University College of Dentistry, New York, NY); the cell line TR146 was provided by Dr. Alfonse Balm (University of Amsterdam, Amsterdam, the Netherlands); and the cell line SqCC/Y1 was provided by Dr. Michael Reiss (Yale University, New Haven, CT). All of the cells were grown at 37°C in a humidified atmosphere composed of 95% air and 5% CO₂ in a monolayer culture consisting of a 1:1 (v/v) mixture of DMEM, 10% regular fetal bovine serum, antibiotics, and either Ham’s F-12 nutrient mixture or RPMI 1640.

All of the patient samples, both cancerous and normal, were obtained from established tissue banks at The University of Texas M. D. Anderson Cancer Center (Houston, TX) and the Johns Hopkins Hospital (Baltimore, MD). Solid tumor samples were from patients who had undergone surgery, leukemia samples were from bone marrow aspirates and biopsies, and normal samples were from breast, colon, lung, brain, placenta, head, and neck tissues and from bone marrow, fibroblasts, and lymphocytes. Sample collection and distribution were performed under established guidelines of the United States Department of Health and Human Services.

Genomic DNA from cell lines and human tissues was extracted and purified by use of a DNA extraction kit according to the manufacturer’s instructions (Stratagene, La Jolla, CA). Bisulfite treatment was performed as described previously (15). In brief, after genomic DNA isolation, 2 μg of the DNA were treated for 16 h with sodium bisulfite. After the DNA was desulfonated and purified with the Wizard DNA Clean-Up system (Promega, Madison, WI), an aliquot was used as a template for PCR. Methylation of TIG1 was performed by use of combined bisulfite restriction analysis (COBRA), as described previously (15) . The COBRA primers designed to amplify the modified DNA at the three different areas in the promoter of TIG1 (Fig. 1 A) were the first set of primers (COBRA 1) were 5′-GAGAGAATTTAGGGGTTG-3′ (sense) and 5′-AACCAAAAAACAAACAACC-3′ (antisense), which yielded a 221-bp fragment. The second set (COBRA 2) were 5′-GGTAGTTTTAGGATGTTGGGG-3′ (sense) and 5′-TACCCAAATATCACCTCCCAAC-3′ (antisense), which also yielded a 210-bp fragment. The third set of primers (COBRA 3) were 5′-GTTTGGAGAATTTAAGTAG-3′ (sense) and 5′-AATACTTCTAACCCAAACC-3′ (antisense), which yielded a 190-bp fragment. For RARβ isoform 2 (RARβ2) methylation, the PCR primers were 5′-AAGTAGTAGGAAGTGAGTTGTTTAGA-3′ (sense) and 5′-CCAAATTCTCCTTCCAAATAA-3′ (antisense), which produced a 207-bp fragment. TIG1 methylation analysis was performed with 20–40 μl of the amplified products, which were digested with the restriction enzymes HinfI (for PCR products generated by COBRA 1 primers), BssHII (for COBRA 2), and TaqI (for COBRA 3). For RARβ2 methylation analysis, the restriction enzyme TaiI (MBI Fermentas, Hanover, MD) was used to distinguish methylated sequences from unmethylated ones, and the products were subjected to electrophoresis on 3% agarose or 5% acrylamide gels. DNA was visualized by ethidium bromide staining. We calculated the methylation percentage by dividing the density of the methylated band by the summed densities of both methylated and unmethylated bands as determined with a densitometer.

To study TIG1 promoter hypermethylation, we also performed MSP as described previously (16). For the TIG1 promoter methylation study, we designed two sets of primers that could amplify the modified DNA of either the methylated or unmethylated alleles separately. For the methylated allele (118 bp), the sense primer was 5′-GTAGTACGGGCGGGTCGC-3′ and the antisense primer was 5′-GACATCGCCTCCGCAACG-3′. For the unmethylated allele (134 bp), the sense primer was 5′-GAGGTAGTATGGGTGGGTTGT-3′ and the antisense primer was 5′-AATACTAAAATACAACATCACCTCCA-3′. The annealing temperatures for the methylated and unmethylated DNA were 67°C and 61°C, respectively, for 40 s each. Hot-start PCR with a total cycle number of 30 was used in all MSP DNA amplifications. Denaturation and extension cycles were maintained for 30 and 45 s, respectively.

We used the same COBRA primers, sets 1 and 2, as those used for bisulfite-PCR and cloned the DNA fragments into a pCR2.1 TOPO vector (Invitrogen Corporation, Carlsbad, CA) according to the manufacturer’s instructions. Plasmid DNA was extracted and purified with the Qiagen Plasmid Mini Kit (Qiagen, Valencia, CA) and sequenced with an ABI PRISM 377 DNA sequencer (Applied Biosystems, Foster City, CA) at the DNA sequencing core facility at the University of Texas M. D. Anderson Cancer Center.

Cell lines were treated with 1 μm 5-Aza-dCyd (Sigma Chemical Co., St. Louis, MO). After daily treatments over 5 days, the RNA was extracted and tested for restoration of TIG1 expression. TIG1 induction was also analyzed in the cells treated using 1 μm ATRA for 3 days and 300 nm TSA for 24 h. In a trial to change the ATRA-resistant cell line K562 to ATRA-sensitive, we first treated the cell line K562 for 3 days with 5-Aza-dCyd. The cells were washed and kept with no treatment for 24 h to allow recovery before being treated again with 1 μm ATRA during the last 10 days of the 14-day experiment. The controls were cells kept with either no previous treatment with 5-Aza-dCyd or no more treatment with ATRA.

Total RNA obtained from different cell lines was isolated using Tri Reagent (Molecular Research Center, Inc., Cincinnati, OH). RT-PCR reactions using the Access RT-PCR System (Promega) were performed according to the manufacturer’s instructions, with an annealing temperature of 68°C for 35 cycles. The sense oligonucleotide used for detection of a 134-nucleotide fragment originating from the human TIG1 cDNA was 5′-GAAAAACCCCTTGGAAATAGTCAGC-3′, which corresponded to the region encompassing nucleotides 504–528 in exon 3. The antisense oligonucleotide used for detection of the TIG1 transcript was 5′-AGTGTGACACCTGTGTTGTCATTTCC-3′, which corresponded to positions 638–612 in exon 4, beginning from the translation start site. The GenBank accession number of the cDNA sequence was XM_003103. The sense and antisense oligonucleotide primers for the human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA, used as an internal control, were 5′-CGGAGTCAACGGATTGGTCGTAT-3′ and 5′-AGCCTTCTCCATGGTGGTGAAGAC-3′, respectively. Each sample included reverse transcription-negative controls in which the reverse transcriptase was omitted from the initial reverse transcription step. A normal cDNA panel for different organ sites was purchased from BD Biosciences (Clontech, Palo Alto, CA).

Real-time Q-RT-PCR was performed with the 7700 Sequence Detector (Applied Biosystems; Refs. 17, 18). We used commercially available primers sets with minor groove binder probe for TIG1 and β-actin as an internal control (Applied Biosystems). After RNA isolation, cDNA was synthesized with random hexanucleotide primers and reverse transcriptase. Each sample included reverse transcription-negative controls in which the reverse transcriptase was omitted from the initial reverse transcription step. Reactions for Q-RT-PCR were performed with the TaqMan universal PCR Master Mix kit (Applied Biosystems) in 96-well plates. Each sample was measured in triplicate. Assembled plates were then covered and run using the following conditions: an initial denaturation step of 95°C for 10 min followed by 45 cycles at 95°C for 15 s and 60°C for 1 min. The resulting data were analyzed with ABI Prism 7000 SDS software (Applied Biosystems). The threshold cycles (C_Τ) were determined and the differences in the C_Τ values for β-actin and TIG1 were calculated. Because TIG1 expression was first reported to be lost in prostate cancer cells (7), all Q-RT-PCR data in our present study were normalized to the expression level detected in normal prostate, which was assigned as 100%. Expression results for other tested samples were calculated relative to the expression in normal prostate.

For detecting cell cycle changes and apoptosis, the cells were harvested, fixed with cold 70% ethanol, stained with propidium iodide, and analyzed by flow cytometry as described previously (19). The percentage of dead cells among the treated cells was determined by counting trypan blue-stained cells.

Statistical analyses were performed with the Fisher’s exact test using StatView software (SAS, Cary, NC). Two-sided tests were used to calculate Ps; P < 0.05 was considered statistically significant.

By applying MCA-RDA to the analysis of methylated DNA in the leukemia cell line K562, we identified a hypermethylated DNA fragment corresponding to the TIG1 promoter (Fig. 1,A). Because this was the first evidence of hypermethylation of this gene, we confirmed our results in multiple human cancers by use of bisulfite-PCR. We used the complementary methods COBRA and MSP, as well as DNA sequencing in selected cases. Fig. 1,B shows examples of the cell lines tested with the COBRA analysis. One finding of note was the variation in the degree of methylation among the different cell lines. Specifically, complete methylation was seen in HCT116 and HL-60, whereas partial methylation was seen in Daoy; Hut62 is an example of an unmethylated cell line. Table 1 summarizes the analysis of 53 different cell lines obtained from different organs and sites. In this analysis, we used the COBRA 1 primer set to quantify the relative amounts of methylated DNA in the individual cell lines. As shown in Table 1, among the 53 cancer cell lines tested, 11 had methylation densities >60%, 17 cell lines had methylation densities of 30–60%, and 25 cell lines had methylation densities <30%.

Examples of the results of MSP analysis of the cancer cell lines are shown in Fig. 1,C. The RKO and SqCC/Y1 cell lines showed only unmethylated bands, whereas H1944, HCT116, and HL-60 cell lines showed only methylated bands. Individual methylation density percentages of the TIG1 promoter were also confirmed by sequencing of cloned bisulfite-PCR products with three different COBRA primer sets (Fig. 1,A) in the cell lines H1944, MD468, HL-60, K562, MCF-7, RKO, LoVo, SqCC/Y1, HCT116, and H522. Fig. 2 shows examples of methylation densities detected in sequenced cell lines with use of the second COBRA primer set (5–10 clones were sequenced for each cell line tested). Of note, there was a high concordance between the results generated by promoter sequencing and those generated with COBRA (Table 1).

We next studied samples obtained from normal tissues and primary tumors (Fig. 3). The 20 normal tissue samples obtained from organ banks (examples are given in Fig. 3) showed low or no TIG1 promoter methylation, with a mean methylation of 10% (range, 0–15%). Among the 74 total primary malignancies examined, 39 (53%) exhibited relatively high TIG1 promoter methylation density, as measured with the MSP and COBRA methods. Cancer cells with a methylation level higher than corresponding normal tissue cells were considered to be methylated. The overall analysis revealed significantly higher TIG1 promoter methylation values, compared with corresponding normal tissues, in 8 of 14 (57%) acute myeloid leukemia cases, 3 of 7 (42%) chronic myeloid leukemia cases, 4 of 10 (40%) colon cancer cases, 10 of 16 (63%) head and neck cancer cases, 3 of 9 (33%) breast cancer cases, and 11 of 18 (61%) liver cancer cases. Significant concordance (P < 0.0001) was found between the COBRA and MSP results in 20 cell lines and in 25 normal and primary tumor tissues tested with both methods. Similarly, we sequenced the cloned PCR fragments from 10 primary colon tumors with their adjacent normal tissues. Five to 10 clones were sequenced for each pair of samples. All tumor samples tested had higher TIG1 promoter methylation densities than their corresponding individual adjacent normal tissues. Fig. 2 shows an example of the differences in percentage of methylation density as measured with the CpG sequencing method in two pairs of primary colon cancers and their adjacent normal tissues.

To relate TIG1 promoter methylation data with TIG1 expression levels, we first examined constitutive TIG1 expression in a panel of normal tissues by conventional RT-PCR as well as Q-RT-PCR (Fig. 4, A and B). TIG1 expression was detected in total cDNA from most of the normal tissues tested, with some degree of variation. No or minimal promoter methylation was detected in normal tissues that showed diminished expression, such as normal leukocytes.

Regarding cancer cell lines, Table 1 shows the relative individual expression levels in relation to methylation percentage. In this analysis (Table 1), we found that in 11 cell lines with TIG1 methylation densities >60%, all showed complete loss of TIG1 expression. Seventeen cell lines showed a methylation density of 30–60%. Of these, 12 cell lines showed complete loss or markedly diminished TIG1 expression. In the remaining 21 cell lines with TIG1 methylation density <30%, all showed high TIG1 expression, and 4 cell lines were undetermined. A statistical analysis of the data (Table 1) by use of Fisher’s exact test indicated that loss of expression in these cell lines was correlated with hypermethylation of the TIG1 (P < 0.001). As shown in Table 1 and Fig. 4,C, TIG1 gene expression was diminished or lost in 24 of the 47 (51%) cancer cell lines tested by conventional RT-PCR. There was an excellent concordance with the results generated by Q-RT-PCR in 20 cancer cell lines (Fig. 4,D). A few cell lines had relatively high TIG1 methylation values (e.g., MDA-886Ln, Dupro, and MB453) along with various degrees of expression (Table 1). This can probably be explained by the TIG1 promoter being incompletely suppressed through coexistence of unmethylated alleles.

Regarding primary tumor samples, TIG1 expression was decreased in one and was lost in two of the six acute myeloid leukemia samples (Fig. 4,E) and decreased or lost in three of the seven colon cancer samples obtained from different patients. Fig. 4,F shows the relative differences in TIG1 expression as measured by Q-RT-PCR in two primary colon cancer with their adjacent normal mucosa tissues. Fig. 5 shows the correlation between loss of TIG1 expression in two other colon tumor cases with their corresponding adjacent normal tissues, and TIG1 methylation density.

We next tested the ability of the DNA methylation inhibitor 5-Aza-dCyd to restore TIG1 expression in a number of cell lines that showed diminished expression (Fig. 6,A). All of the tested cell lines (K562, H460, H522, MCF-7, SW48, HCT116, HL-60, and UMSCC38) showed reactivation and restoration of TIG1 expression after 5-Aza-dCyd treatment, albeit to different degrees. These data indicated that repression of TIG1 was mediated at least in part by TIG1 promoter hypermethylation. On the other hand, the histone deacetylase inhibitor TSA failed to restore or increase TIG1 expression, even in the unmethylated cell lines. We did, however, observe augmentation of TIG1 expression when TSA was combined with 5-Aza-dCyd (Fig. 6 B). Because no commercial antibody was available at the time, we restricted our analysis to mRNA levels in this study.

To determine whether TIG1 promoter methylation was associated with methylation of the promoter of RARβ2, another retinoic-induced putative tumor suppressor gene, we examined methylation of the RARβ2 promoter by COBRA in a panel of 29 human cancer cell lines. Thirteen (45%) of the cell lines showed reversed states of methylation for the two promoters, and there was no correlation between TIG1 and RARβ2 promoter methylation (P = 0.78). Table 2 shows examples of cancer cell lines discordant for RARβ2 and TIG1 methylation.

TIG1 expression is normally induced by retinoic acid treatment. We therefore tested the effect of methylation on this induction. Table 2 shows a correlation between the level of constitutive TIG1 expression and its methylation status as well as its inducibility by treatment with 1 μm ATRA for 3 days. Specifically, all four cell lines with low methylation densities (≤20% in the cell lines mentioned in Table 2) of TIG1 promoter showed constitutive expression of TIG1 and inducibility by ATRA. TIG1 expression was not induced in most of the cell lines with a higher promoter methylation density (>60% in the cell lines mentioned in Table 2) after the same ATRA treatment. However, 5-Aza-dCyd restored or increased expression in all cell lines tested, including the cell lines not restored by ATRA (HCT116, MCF-7, H522, and H460; Table 2 and Fig. 6,A). The effect of ATRA on TIG1 induction is shown in Fig. 6 C. Cancer cell lines with lower methylation densities (H1792 and RKO) had adequate induction of TIG1 after ATRA treatment. In contrast, the cell line with higher methylation density (H460) showed neither expression nor induction of TIG1 after the same treatment with ATRA.

To determine the potential role of TIG1 in retinoic acid responsiveness independently of RARβ2, we studied the cell line K562. This cell line showed dense methylation of the TIG1 promoter and reduced TIG1 expression, which was reversible by 5-Aza-dCyd treatment (Fig. 6,A), but no methylation of the RARβ2 promoter (data not shown). Cells treated with ATRA alone for 10 days did not show cell growth suppression or an increased percentage of apoptosis (Fig. 7,C), indicating ATRA resistance. However, as shown in Fig. 7,D, ATRA treatment for 10 days after an initial 3-day treatment with 5-Aza-dCyd induced significant apoptosis (sub-G₁, 50%). Compared with control cells (Fig. 7,A), 5-Aza-dCyd-treated cells without ATRA showed no change in the percentage of apoptosis measured at the end of the experiment (Fig. 7 B). The sub-G₁ percentage representing the apoptosis population was confirmed by a comparable percentage of cell death, as measured by trypan blue staining (data not shown). Similar results were obtained when we studied other cell lines, such as the breast cancer cell lines CAMA-1 and BT-474, using 1 μm ATRA after 0.1 μm 5-Aza-dCyd for 3 days. However, the lung cancer cell line H460 did not show significant differences between the different groups (data not shown). Next we measured TIG1 expression using the relative ratio of the RT-PCR products (TIG1:GAPDH) to test whether the induction of apoptosis in K562 cells (which do not express TIG1 at basal levels) is related to restoration of TIG1 expression. Whereas ATRA induced TIG1 in K562 (20% in arbitrary units), combined treatment with ATRA after 5-Aza-dCyd had a synergistic effect on TIG1 induction (70% in arbitrary units), as measured on day 10 of the experiment. However, some other retinoid-induced gene(s) might be involved in the apoptotic response to such a combined treatment.

Inactivation of tumor suppressor genes can be accomplished in several ways, including deletion, mutation, and epigenetic silencing of gene expression by methylation or changes in histone acetylation. MCA-RDA has been useful in the identification of putative tumor suppressor genes based on hypermethylation status of promoters (10). Using this method, we identified a DNA fragment corresponding to the TIG1 promoter-associated CpG islands in the leukemia cell line K562. We confirmed that the TIG1 promoter was hypermethylated in approximately half of the cases tested in both cancer cell lines and primary tumors samples. We also found that TIG1 expression was correlated with the methylation status of its promoter and was restored in selected cell lines after 5-Aza-dCyd treatment. These findings suggest that TIG1 expression is silenced primarily by hypermethylation.

Treatment of several cell lines with the histone deacetylase inhibitor TSA failed to restore TIG1 expression. This failure suggests that the acetylation status of the TIG1 gene was not dominant in explaining silencing. In several earlier studies using leukemia cell lines, TSA was shown to inhibit histone deacetylation but not to activate transcription as a single agent (20, 21). In these studies, ATRA was required in addition to TSA to induce engagement of histone acetyltransferase, thereby acetylating adjacent histones and promoting chromatin relaxation and binding of transcription factors to response elements in DNA to initiate transcription.

Interestingly, we found that TIG1 expression was decreased or lost in all of the cell lines that had a methylated TIG1 promoter and could not be restored in those cell lines after 3 days of treatment with 1 μm ATRA. On the other hand, similar treatment of cell lines that exhibited basal expression of TIG1 resulted in a significant induction of expression. In a previous study (22), we observed the same pattern in the inducibility of RARβ2 expression, which occurred only in cell lines that had at least some basal endogenous expression of RARβ2. These data suggest that individual cell lines with a minimum endogenous level of TIG1 expression exhibit an increase in expression after retinoic acid treatment. These results imply that silencing of TIG1 by hypermethylation contributes to the loss of retinoic acid responsiveness in some neoplastic cells.

Approximately half of the 29 cell lines in which the TIG1 and RARβ2 promoter status was determined showed divergence in the methylation status of the two genes, although both genes are located in the same region on chromosome 3p. Furthermore, induction of TIG1 expression was independent of RARβ2 promoter methylation status. The retinoic acid-induced apoptosis of K562 cells, which have TIG1 but not RARβ2 promoter methylation, was also in line with the proposed independent role of TIG1 in retinoic acid responsiveness. The induction of apoptosis after combined treatment with 5-Aza-dCyd and ATRA is possibly explained by retinoic acid acting on the demethylated cell population generated after 5-Aza-dCyd treatment. The K562 cell lines was previously reported to be resistant to differentiation induction by ATRA (23, 24). Similar results were obtained when we studied other cell lines, such as the breast cancer cell lines CAMA-1 and BT-474, but using 1 μm of ATRA after only 0.1 μm of 5-Aza-dCyd instead of 1 μm as in the cell line K562. Our results suggest that ATRA combined with 5-Aza-dCyd restores the responsiveness of ATRA independent of RARβ2 reactivation, pointing the way to human clinical trials for combining ATRA or tazarotene with demethylating agents such as 5-Aza-dCyd to restore lost TIG1 expression in some human cancers.

Jing et al. (7) recently demonstrated tumor suppression properties for the TIG1 gene in prostate cancer. They showed that transfection of TIG1 into prostate cancer cells decreased invasiveness in vitro and tumorigenicity in vivo in nude mice. Our study shows that methylation may be one mechanism by which the tumor suppression characteristics of this gene are inactivated by silencing. Interestingly, we included in our present study four prostate cancer cell lines that were used in the study by Jing et al. (7). We found that three of these four cell lines had diminished TIG1 expression as a result of hypermethylation of TIG1 promoter. These results can complement and support the potential suppressor role of TIG1 in prostate cancer.

In conclusion, our results demonstrate that hypermethylation-associated silencing of the TIG1 gene occurs in a variety of human cancers and that 5-Aza-dCyd can reverse TIG1 silencing, indicating that TIG1 repression is mediated in part by promoter hypermethylation. These data also reinforce the potential importance of retinoids in cancer therapy (25, 26).