Glycobiology vol. 18 no. 1 pp. 104–113, 2008
doi:10.1093/glycob/cwm120
Advance Access publication on October 27, 2007
Expression of the human Sd
aβ-1,4-N-acetylgalactosaminyltransferase II gene is
dependent on the promoter methylation status
Hou-Ren Wang2,3, Chuang-Yi Hsieh3, Yuh-Ching Twu3,
and Lung-Chih Yu1,2,3
2Institute of Biological Chemistry, Academia Sinica, Taipei 115, Taiwan; and 3Institute of Biochemical Sciences, National Taiwan University, Taipei 106, Taiwan
Received on August 22, 2007; revised on October 16, 2007; accepted on October 20, 2007
It has been noted that the expression of Sda, including its
antigenic structure, theβ-1,4-N-acetylgalactosyltransferase II (β4GalNAcT-II) activity responsible for its formation, and the Sdaβ4GalNAcT-II mRNA transcript, is drastically
reduced in oncogenetic processes in gastrointestinal tissues. Markedly reduced metastatic potential has been demon-strated in colon and gastric cancer cells transfected with the Sdaβ4GalNAcT-II gene. In this study, a putative CpG island
encompassing the promoter and exon 1 regions in the human Sdaβ4GalNAcT-II gene was identified, and the investigation
of DNA methylation of the Sda gene promoter region demonstrated a clear association between the methylation status of the CpG island promoter and expression of the Sda
gene in gastrointestinal cancer cell lines. Hypomethylation of the promoter region of the Sda gene was shown in cells
where this gene was expressed. By contrast, there was significant hypermethylation of the Sda gene promoter in
cells that did not express the gene. A specific methylation profile in the Sda gene CpG island was demonstrated in
KATO III gastric cancer cells. In colon cancer cells with the hypermethylated Sdagene promoter, treatment with the
DNA methylation inhibitor, 5-aza-2-deoxycytidine, resulted in demethylation of the promoter region and substantially induced the expression of the Sdagene and the Sdaantigenic
structure. These results strongly suggest that promoter DNA methylation plays a crucial role in the regulation of the Sdaβ4GalNAcT-II gene and Sdaantigen expression.
Keywords: DNA methylation/gastrointestinal cancers/gene expression/N-acetylgalactosaminyltransferase/Sdaantigen Introduction
The Sdaantigen was first reported as a human blood group anti-gen by Macvie et al. and Renton et al. simultaneously in 1967 (Macvie et al. 1967; Renton et al. 1967). The antigen was found in more than 90% of human red blood cells (RBCs), and also in several tissue types, including stomach, colon, and kidney (Morton et al. 1970; Capon et al. 2001; Robbe et al. 2003), 1To whom correspondence should be addressed: Tel: +886-2-3366-4070; Fax:+886-2-2363-5038; e-mail: [email protected]
and oocytes (Dell et al. 2003). Its presence was also demon-strated in various body fluids, such as saliva, milk, serum, and urine, with the greatest concentration occurring in urine (Morton et al. 1970; reviewed in Daniels 2002). The strength of the Sdaantigen expression on Sda-positive RBCs varies greatly.
RBCs with very strong Sda antigen expression, sufficient for
agglutination by Dolichos lectin, have been defined as the Cad phenotype (Sanger et al. 1971). It has been revealed through var-ious studies that the Sda/Cad antigenic epitopes are determined by the terminal trisaccharide structure, GalNAcβ1-4(NeuAcα2-3)Galβ, carried on glycoproteins and glycolipids (Blanchard et al. 1983, 1985; Donald et al. 1983), and that the final step in the formation of the Sda/Cad antigenic epitopes is determined by
the activity of the Sdaβ-1,4-N-acetylgalactosaminyltransferase
II (β4GalNAcT-II), which catalyzes the addition of GalNAc to the Gal of the NeuAcα2-3Galβ terminal structure in the β-1,4 linkage. This enzyme activity was first manifested in the guinea pig kidney (Serafini-Cessi and Dall’Olio 1983), with a subse-quent investigation demonstrating the abundance of this enzyme in the human colon (Malagolini et al. 1989).
Following the first isolation of the Sdaβ4GalNAcT-II cDNA
from murine cytotoxic T lymphocytes (Smith and Lowe 1994), a partial sequence of human Sdaβ4GalNAcT-II cDNA was cloned
from the total RNA fraction of gastric mucosa (Dohi et al. 1996). The full coding sequence of human Sdaβ4GalNAcT-II cDNA was demonstrated by Montiel et al. in 2003 (Montiel et al. 2003). The human Sdagene is located on chromosome 17 and comprises at least 12 exon regions that span about 37 kb of genomic DNA. Notably, two transcript forms, which have different exon 1 (denoted as exon 1Sand exon 1L), but identical
exons 2–11, coding regions, are expressed from the human Sda
gene locus through utilization of alternative promoters. The two transcripts, with respective open reading frames of 1521 and 1701 bp, encode the short- and long-form Sdaβ4GalNAcT-IIs with 506 and 566 amino acid residues, respectively. The two forms of Sdaβ4GalNAcT-II differ with respect to the lengths of their cytoplasmic tails, with the long-form unusually extended at 66 amino acid residues. An independent paper by Lo Presti et al. also reported the cDNA sequence for the long-form Sda
β4GalNAcT-II (Lo Presti et al. 2003).
It has been noted that the expressions of Sda are develop-mentally regulated and increase as a function of age (Macvie et al. 1967; Renton et al. 1967; Dall’Olio et al. 1987, 1990). Increased Sdaantigen expression and Sdaβ4GalNAcT-II
trans-ferase activity has been observed during differentiation of hu-man colon and gastric cancer cell lines (Malagolini et al. 1991; Dohi et al. 1993). Especially, altered expression patterns of Sda
antigen during oncogenetic processes have been noted. It was demonstrated that the Sdaantigen and Sdaβ4GalNAcT-II
activ-ity were markedly diminished (Malagolini et al. 1989; Dohi et al. 1991), and the expression of the Sda gene transcript
DNA methylation regulates Sdagene expression
Fig. 1. Nucleotide sequence of the human Sdaβ4GalNAcT-II gene encompassing the 5promoter, exon 1S, and exon 1Lregions, and the primer locations for bisulfite genomic sequencing and methylation-specific PCR. The nucleotide sequence was derived from the human chromosome 17 genomic contig, deposited in GenBanK as accession number NT_010783. The transcription start nucleotide of exon 1Lis numbered+1. The regions of exon 1Sand exon 1Lare boxed, with asterisks indicating the translation start codons in exon 1Sand exon 1L. Arrows represent the locations and orientations of the primers used in bisulfite genomic sequencing and methylation-specific PCR. It was determined that the−468 position (indicated by a dot) was the G nucleotide in KATO III and MKN45 cells, whereas it was the A nucleotide in the other eight cell lines assessed in this study.
significantly decreased in colon and gastric cancer tissues (Dohi et al. 1996). Although association between the downregulation of Sdaexpression and malignant changes in the gastrointestinal
tract mucosa has been noted, the biological implications of this phenomenon were not manifested until the study by Kawamura et al. (2005). It has been known that the expression of the E-selectin ligands, sialyl Lewisa(sLea) and sialyl Lewisx(sLex), in
gastrointestinal cancer cells is associated with malignant metas-tasis. Kawamura et al. demonstrated that Sda β4GalNAcT-II competes for the acceptor substrates with theα-1,3- and α-1,4-fucosyltransferases responsible for the formation of sLea and sLex carbohydrate structures, and that overexpression of Sda
β4GalNAcT-II eliminates the expression of sLea and sLex in
colon and gastric cancer cells and, consequently, inhibits the metastasis of these cells. Their study suggests a potential for the development of antimetastatic gene therapy for gastrointestinal cancers. The relationship between the expression of Sda and
sLexantigens in normal and cancerous colon tissues has been further elaborated in a recent study (Malagolini et al. 2007).
Although tissue- and cell-type specificities with respect to the distribution of Sda antigen and its altered expression patterns
during oncodevelopmental processes have long been noted, the control mechanism for the Sdagene expression has never been
elucidated. It is well known that the methylation of cytosine residue in the CpG dinucleotide-rich sequence provides one of the regulatory mechanisms for gene expression. The present
study has demonstrated that the expression of the human Sda
β4GalNAcT-II gene is determined by the DNA methylation status of the CpG island in the promoter region.
Results
CpG island in the promoter and exon 1 regions of the Sda β4GalNAcT-II gene
The short- and long-form human Sdaβ4GalNAcT-II transcripts are composed of different exon 1, exon 1S, and exon 1L. Exon 1S
and exon 1Lcontain 38 and 253 bp, respectively, and are
sepa-rated by a 160-bp sequence (Figure 1) (Montiel et al. 2003). The sequence of nucleotides−2000 through +2000 (transcription start of exon 1Las nucleotide+1) of the Sdagene was analyzed
using the CpGPlot utility from the European Bioinformatics In-stitute (http://www.ebi.ac.uk/emboss/cpgplot/) (Figure 2), and showing that the nucleotide region of−537 through +425 ful-filled the criteria for a CpG island defined by Gardiner-Garden and Frommer who first performed a large-scale analysis of CpG islands in vertebrate genomes (Gardiner-Garden and Frommer 1987). They defined a CpG island as a 200-bp (or longer) stretch of DNA with a G+C content greater than 50% and an observed CpG/expected CpG ratio (Obs/Exp CpG) greater than 0.6. Some of the human Alu sequences, which are highly repetitive short interspersed elements with a 280-bp consensus sequence
H R Wang et al.
Fig. 2. CpG island in the Sdaβ4GalNAcT-II gene. The top diagram is a schematic representation of the nucleotide region −2000 through +2000 of the Sdagene (transcription start of exon 1Las+1), with the locations of exon 1S(E1S) and exon 1L(E1L) shown as open boxes. Exon 2 locates about 8 kb downstream to exon 1L, and thus is not shown in the diagram. Open arrows represent the positions and orientations of the Alu repetitive elements. The middle and bottom diagrams present the quantifications of the observed CpG/expected CpG ratio (Obs/Exp CpG) and the percentage of G+C nucleotides in the −2000 through +2000 nucleotide region, respectively. The values for Obs/Exp CpG and G+C% were plotted against the position in the analyzed sequence using the CpGPlot program, with a sliding window of 200 bp and a step increment of 1. The nucleotide region−537 through +425, indicated as gray areas, fulfils the criteria of a CpG island, Obs/Exp CpG greater than 0.6 and G+C content greater than 50%, as defined by Gardiner-Garden and Frommer (Gardiner-Garden and Frommer 1987). (approximately), have relatively high G+C content and Obs/Exp
CpG (Schmid 1998), and thus the nucleotide region from −2000 through +2000 of the Sda gene was analyzed by the
RepeatMasker Web Server (http://repeatmasker.org) to detect the Alu repetitive elements. In total, five Alu elements were lo-cated in the region (Figure 2), but in the regions 5and 3sides of the CpG island, not within the putative CpG island. Takai and Jones have suggested that adoption of a more stringent defini-tion for a CpG island, a DNA region of greater than 500 bp with a G+C content equal to or greater than 55% and Obs/Exp CpG of 0.65, is more likely to distinguish CpG islands associated with the 5regions of genes rather than analogs associated with intragenomic repetitive sequences (Takai and Jones 2002). The nucleotide region−537 through +425 of the Sdaβ4GalNAcT-II
gene has a G+C content of 66.8% and an Obs/Exp CpG of 0.85, fulfilling these more stringent criteria.
Correlation of the methylation status of the CpG promoter with the Sdaβ4GalNAcT-II gene expression
Expression of the Sda gene in different human
gastrointesti-nal cancer cell lines was agastrointesti-nalyzed using reverse transcription-polymerase chain reaction (RT-PCR). The expression profiles were analyzed for the Sdatranscript (with the common region
of the short- and long-form Sdatranscripts PCR amplified), to-gether with the short- (denoted as Sda-S) and long-form Sda
(Sda-L) transcripts in seven colon cancer cell lines, two gastric cancer cell lines, and one pancreatic cancer cell line (Figure 3). Expression of the Sda transcript was demonstrated in Caco-2,
Colo 205, SW1417, SW480, SW948, and MKN45 cells. Similar expression profiles for the short- and long-form Sdatranscripts
were revealed in these cell lines, except that only the short-form Sdatranscript was detected in MKN45 cells. In contrast,
the RT-PCR analysis did not detect any Sda transcript in the
Fig. 3. Expression profiles of the Sdaβ4GalNAcT-II gene in various gastrointestinal cancer cell lines. Expression of the Sdaβ4GalNAcT-II gene in colon (Caco-2, Colo 205, HCT 116, SW1417, SW48, SW480, and SW948), gastric (KATO III and MKN45), and pancreatic (MIA PaCa-2) cancer cell lines was analyzed using RT-PCR. Total RNAs purified from these cells were primed using oligo-dT primer to synthesize the first-strand cDNAs, and PCR amplifications specific for the short-form Sdaβ4GalNAcT-II (denoted as Sda-S), long-form Sdaβ4GalNAcT-II (Sda-L), the common region (exon 4 to exon 6) of the two transcript forms (Sda), andβ-actin cDNA were performed, as described in Materials and methods. The RT-PCR products were analyzed using 1.5% agarose gel electrophoresis. The expected sizes of the products from the Sda, Sda-S, Sda-L, andβ-actin cDNAs were 228, 260, 310, and 626 bp, respectively. The rightmost lanes show the molecular mass standards of the 100-bp ladder.
RNA samples prepared from HCT 116, SW48, KATO III, and MIA Paca-2 cells. Nevertheless, a trace of expression of the short-form Sda, but not the long-form analog, was observed
DNA methylation regulates Sdagene expression
Fig. 4. Methylation status of the Sdaβ4GalNAcT-II gene promoter for different gastrointestinal cancer cell lines. (A) Percentage of methylated CpG sites in the −550 through −1 nucleotide region of the Sdagene. Each vertical line, illustrating the percentage of the methylated CpG site in the analyzed region, represents the result obtained from one individual clone. (B) Evaluation of the methylation status of the Sdapromoter using methylation-specific PCR analysis. Genomic DNA samples from each cell line were bisulfite modified, and PCR amplifications with the primer pairs of USPF+USPR (lanes U) and MSPF+MSPR (lanes M), designed for the unmethylated and methylated status, respectively, were performed. The expected sizes of the PCR products from the USPF+USPR and MSPF+MSPR primer pairs were 494 and 443 bp, respectively. The rightmost lanes show the molecular mass standards of the 100-bp ladder.
in the KATO III cells with more PCR cycles (data not shown).
The methylation status of the Sdagene CpG island in these
cells was analyzed through bisulfite genomic sequencing as described in the Materials and methods section. The region of nucleotide−550 through −1, which contains 47 CpG din-ucleotides (48 in KATO III and MKN45 cells; see Figure 1 legend), was analyzed; 5–8 clones were sequenced per cell line. In the analyzed clones, almost all the cytosine residues not within the CpG dinucleotide were converted to thymidines, with only few exceptions (0–3 per clone), demonstrating that the bisulfite conversions were virtually complete. The percent-age of methylated CpG in the analyzed region obtained from each clone is schematically represented in Figure 4A. The methylation status of the Sda CpG island in each cell line
re-veals a sharp contrast between cells with and without Sdagene
expression. The CpG sites are barely methylated in Caco-2, Colo 205, SW1417, SW480, SW948, and MKN45 cells, which express the Sdagene. Only few sites (0–3 per clone) in the 47
(or 48) CpG dinucleotides were found to be methylated in these cells. The lowest methylation status was observed in SW948 cells as only one methylated CpG was located in the five analyzed clones. In contrast, hypermethylation of the Sdagene CpG island was observed in HCT 116, SW48, and MIA PaCa-2 cells, which do not express the Sdagene. In these three cell lines,
almost all the CpG sites were methylated, with only 0–3 sites per clone unmethylated. MIA PaCa-2 cells showed the highest methylation percentage, with 100% methylation demonstrated for six of the eight analyzed clones. A different feature was observed in KATO III cells. The methylation percentages for the six clones from the KATO III cells were around 60%.
H R Wang et al.
Fig. 5. Methylation profile for individual CpG dinucleotide in the Sdagene CpG island in KATO III cells. The methylation of individual CpG sites in the−550 through+380 nucleotide region of the Sdagene in KATO III cells was analyzed by bisulfite genomic sequencing, with the analyzed region encompassing 85 CpG dinucleotides. The top diagram is a schematic representation of the nucleotide region−600 through +400 of the Sdagene; the locations of exon 1
S(E1S) and exon 1L(E1L) are shown as open boxes. Sequencing results were compiled for 15 clones, prepared from five separate batches of PCR products, and the methylation frequency for each individual CpG site, observed in the 15 analyzed clones, calculated. The methylation percentages (also the number of clones with methylated CpG) for each individual CpG site are illustrated and plotted against the position in the analyzed sequence.
Methylation-specific PCR was established to evaluate the methylation status of the Sda CpG promoter. In this analysis
system, the primer pair designed to demonstrate the methylated status (lanes M, Figure 4B) yielded PCR products only from the bisulfite-treated genomic DNA samples prepared from HCT 116, SW48, KATO III, and MIA PaCa-2 cells, in which the Sda gene promoters were significantly methylated. When the
primer pair designed to demonstrate the unmethylated status was used (lanes U, Figure 4B), only the samples prepared from the cells with the hypomethylated Sdapromoter gave PCR
prod-ucts. However, bisulfite-treated genomic DNA from KATO III cells yielded, in addition to the obvious positive result from the methylated primer pair, a trace of PCR product from the unmethylated primer pair. This outcome may be caused by the semi-methylated status of the Sdapromoter in KATO III cells. The results demonstrated from methylation-specific PCR were consistent with those obtained from bisulfite genomic sequenc-ing, suggesting that this methylation-specific PCR is feasible for rapid analysis of the methylation status of the Sdagene CpG
promoter.
The expression profiles for the Sda gene, revealed by
RT-PCR, and the methylation status of the Sda gene CpG island, demonstrated by bisulfite genomic sequencing and methylation-specific PCR, display a definite inverse relationship between the methylation level in the Sdagene promoter region and gene
expression level in these gastrointestinal cancer cell lines. Methylation pattern of the Sdaβ4GalNAcT-II gene CpG island in KATO III cells
It has been shown in a number of genes that the methylation statuses of the individual CpG sites in the respective CpG is-lands are heterogeneous, and that the overall methylation profile in the CpG islands exhibits a specific pattern (Graff et al. 1997; Kominato et al. 1999; Pao et al. 2001). In the bisulfite genomic sequencing analysis described above, the methylation statuses of the individual CpG sites in the Sda promoter region were
found to be heterogeneous in KATO III cells. Therefore, the
methylation pattern in the KATO III cells was analyzed in detail extending from nucleotides−550 to +380, which nearly encompasses the whole putative CpG island of the Sda gene.
Eighty-five CpG dinucleotides are included in the analyzed region. Fifteen clones, which were prepared from five separate batches of PCR products, were sequenced, and the results compiled to calculate the methylation percentage for each of the 85 CpG sites. The methylation percentages of individual CpG sites in the SdaCpG island in KATO III cells (as
schemat-ically represented in Figure 5) are highly heterogeneous. Many CpG sites were intensively methylated, including 16 sites showing methylated in all of the 15 analyzed clones, whereas several sites exhibited very low methylation frequencies.
Further, a specific methylation pattern in the CpG island was observed in KATO III cells. As shown in Figure 5, the methy-lation density in the region around nucleotides−350 through −200 (the 5region proximal to exon 1
S) is evidently lower than
those in the other regions. The 15 CpG dinucleotides in the 5 150-bp region proximal to exon 1Lhave an average methylation
of 87%, while the 14 CpG sites in the 5150-bp region proxi-mal to exon 1Shave an average methylation of 49%. This lower
methylation density in the region 5to exon 1Smay explain the
trace expression of the short-form Sda transcript in KATO III
cells.
Effect of the DNA methylation inhibitor on the expression of Sdaβ4GalNAcT-II gene and Sdaantigen
The results described above demonstrate a good correla-tion between methylacorrela-tion in the promoter region of the Sda
β4GalNAcT-II gene and restraint of Sda β4GalNAcT-II gene
expression in gastrointestinal cancer cell lines. To discover whether methylation of the promoter region is itself inhibiting the expression of the Sdagene, the effects of the DNA
methyla-tion inhibitor on the expression of the Sdagene and Sdaantigen
in the cells with the hypermethylated Sda promoter were ex-amined. To achieve this, 5-aza-2-deoxycytidine (5-aza-CdR), which is a well-known reagent that specifically inhibits DNA
DNA methylation regulates Sdagene expression
Fig. 6. (A) Induction of Sdaβ4GalNAcT-II gene expression by the DNA methylation inhibitor. Expression of the Sdatranscript in HCT 116 and SW48 cells, with or without 5-aza-CdR treatment (5µM of for 6 days), denoted as CdR and—, respectively, was analyzed using RT-PCR. The primer pair for the amplification of the common region of the short- and long-form Sdatranscripts was used in the PCR. (B) Methylation status of the Sdapromoter in the cells treated with the DNA methylation inhibitor. The methylation statuses of the SdaCpG promoter in HCT 116 and SW48 cells, with or without 5-aza-CdR treatment, was evaluated using methylation-specific PCR. The PCR results for the primer pairs designed for the unmethylated (USPF+USPR) and
methylated statuses (MSPF+MSPR) are shown in lanes U and M, respectively. The molecular mass standards of the 100-bp ladder are shown in the rightmost lanes.
methylation (Momparler 2005), was employed to treat HCT 116 and SW48 cells. The results obtained from RT-PCR analysis showed that the transcription of the Sdagene was significantly
induced by treatment of 5-aza-CdR in these cells (Figure 6A). Demethylation of the Sda promoter regions in HCT 116 and
SW48 cells, after treatment with 5-aza-CdR, was demonstrated through bisulfite genomic sequencing; however, this demethyla-tion was only partial. Methylademethyla-tion-specific PCR yielded positive results using the primer pair adopted for the unmethylated status (lanes U, Figure 6B), showing that the reagent treatment at least partially demethylated the promoter regions of the Sdagenes.
However, the failure of the primer pair for the methylated status to yield a positive result from the 5-aza-CdR-treated HCT 116 sample was unexpected.
The effect of 5-aza-CdR treatment in terms of the expression of Sdaantigen on SW48 cells was investigated. As shown from
flow cytometry analysis and fluorescence microscopy exami-nation (Figure 7), there appeared to be no Sdaepitope on the
surfaces of the untreated SW48 cells; however, there was sub-stantial elicitation of Sdaantigen expression on SW48 cells with the 5-aza-CdR treatment. Fluorescence microscopy revealed that about 20% of the cells became positively bound by
Fig. 7. Induction of Sdaantigen expression by the DNA methylation inhibitor in SW48 cells. Expression of the Sdaantigen on SW48 cells was analyzed using flow cytometry (upper panel; A and B) and fluorescence microscopy (lower panel; C–F). The cell surface Sdaantigen was detected with mAb KM694, with the bound mAbs on the cell surface revealed by FITC-conjugated 2nd Ab, as described in Material and methods. The Sda antigen was not detected on the original SW48 cells (left panel; A and C), whereas, after 5-aza-CdR treatment (5µM of for 6 days), the expression of the Sdaantigen on SW48 cells was clearly demonstrated (right panel) using flow cytometry and fluorescence microscopy (B and D, respectively). Bright field micrographs of B and D are represented in C and D, respectively.
monoclonal antibody for the Sda antigen (mAB KM694)
after treatment with a DNA methylation inhibitor.
These results demonstrate marked induction of Sda β4GalNAcT-II gene, and Sdaantigen expression can be achieved
in cells with the hypermethylated Sdagene promoter through inhibition of DNA methylation, indicating a direct relationship between the methylation status in the Sdapromoter and Sdagene
expression.
Discussion
Tissue- and cell-type specificities have been demonstrated in the expression of the Sdaβ4GalNAcT-II gene, and the alteration of the Sdaβ4GalNAcT-II gene transcription during the
oncoge-netic processes of gastrointestinal cancers has been noted. It is of interest, therefore, to improve understanding of the regula-tory mechanism for Sdagene expression, and of significance to
elucidate the mechanism leading to the cancer-associated down-regulation of the Sda gene transcription. In the present study,
hypermethylation in the CpG island of the Sdagene promoter region was demonstrated in the gastrointestinal cancer cell lines without Sda gene expression, while the promoter region was
H R Wang et al.
hypomethylated in the Sda gene-expressing cells. The results
were consistent for all 10 cell lines assessed in this study. Fur-thermore, the suppression of the Sda gene transcription could
be effectively relieved by the treatment of the DNA methylation inhibitor. These results demonstrate that the DNA methylation status of the CpG island in the promoter region determines the expression of the human Sdaβ4GalNAcT-II gene.
The connection between DNA methylation of the CpG-rich promoter and the transcriptional inactivation of the respective genes has been established (Scarano et al. 2005; Klose and Bird 2006). A recent systematic survey of tissue-specific differen-tially methylated regions of genes has demonstrated a corre-lation between the gene expression and the methycorre-lation status of the respective differentially methylated regions (Song et al. 2005), and the result is highly suggestive of a regulatory role for DNA methylation in gene expression. Mechanisms for methylation-induced gene suppression have been proposed, and binding of the methyl binding proteins to the methylated promoter regions appears to play a central role (Scarano et al. 2005; Klose and Bird 2006). Despite the proposed DNA methylation-triggered gene-silencing mechanisms, some studies have suggested that tissue- or cell-specific methylation may be the result of gene silencing rather than the controlling mechanism, and that the gene silencing occurs prior to DNA methylation, which takes place subsequently to stabilize the silencing (Song et al. 2002; Turker 2002). However, this pro-posed gene silencing prior to promoter methylation might not apply to the human Sdagene, as the suppression of this gene
was effectively relieved, and re-expression of the Sda gene
(Figure 6A), together with the Sdaantigen (Figure 7), was effi-ciently elicited by singular treatment with a DNA methylation inhibitor in the cells with the hypermethylated Sdapromoter. This demonstrates the crucial role of promoter-region DNA methylation in the downregulation of the Sda gene and Sda
antigen expression.
Methylation of individual CpG sites in the Sdagene CpG
is-land exhibits high heterogeneity in KATO III cells. Methylation heterogeneity has been described in the CpG islands of several genes (Kominato et al. 1999; Pao et al. 2001); however, the underlying mechanism leading to this phenomenon remains in-tangible. Except for the methylation heterogeneity, methylation in the SdaCpG island represents a specific pattern in KATO III
cells. Relative to methylation frequencies in the other regions, methylation frequency in the 5 region proximal to exon 1Sis
lower. Differential methylation preference in specific regions of a CpG island has been demonstrated in a number of genes (Graff et al. 1997; Kominato et al. 1999; Pao et al. 2001), and the mechanism for the establishment of a specific methylation pattern in a CpG island has been an attractive subject. It has been demonstrated that a cis-acting sequence in the mouse ade-nine phosphoribosyltransferase (aprt) gene is responsible for the initiation of the conserved methylation pattern. The region was designated the methylation center for the aprt CpG island (Mummaneni et al. 1995). The methylation center sequence is derived from retrotransposons, which contribute to the human Alu repeats and homologous mouse B1 elements (Yates et al. 1999). Furthermore, it has been shown that the Sp1 element at the 5 edge of the aprt CpG island is required for protecting the CpG island from de novo methylation (Brandeis et al. 1994; Macleod et al. 1994; Mummaneni et al. 1998). This intriguing model, which depicts the initiation of methylation spread by
the Alu repeat-containing methylation center and the function-ing of Sp1 elements as boundaries to protect the CpG island from methylation, has been suggested to be applicable to the establishment of methylation patterns of the CpG islands of other genes (Graff et al. 1997). The CpG island of the Sda β4GalNAcT-II gene is flanked by Alu elements (Figure 2), and multiple sites with the consensus sequence for the Sp1 binding are present in the SdaCpG island (not shown). Whether these
elements play any role in the establishment of the methylation pattern in the SdaCpG island awaits further verification.
DNA methylation in CpG-rich promoters of genes is now rec-ognized as a common feature of human neoplasia and cancers (Esteller 2005; Toyota and Issa 2005). Aberrant hypermethyla-tion of gene promoters is known to be a major mechanism asso-ciated with inactivation of tumor-suppressor genes in cancers, and it is involved in almost all the critical steps of oncogenesis. In colorectal cancers, epigenetic changes in selected genes are tightly related to neoplastic transformation (Suzuki et al. 2002; Kondo and Issa 2004), and aberrant DNA methylation appears to arise very early in the colon (initially in mucosa of normal ap-pearance), and may be part of the age-related defect in sporadic colorectal cancers (Ahuja et al. 1998; Toyota et al. 1999). It has been shown that the Sda antigen and Sda β4GalNAcT-II transferase activity are drastically diminished with the malignant changes in gastrointestinal tract mucosa (Malagolini et al. 1989; Dohi et al. 1991), and that the Sda β4GalNAcT-II
mRNA disappears in most cancerous tissue specimens of colon and gastric cancers while it is detected in the majority of the paired tissue specimens from the normal portion (Dohi et al. 1996). Kawamura et al. showed that the forced expression of Sdaβ4GalNAcT-II suppressed the formation of the E-selectin
ligands, sLeaand sLex, in colon and gastric cancer cells and, consequently, reduced the metastasis potential of these cells (Kawamura et al. 2005). Thus, it can be conjectured that the downregulation of Sdagene expression in the oncogenetic
pro-cesses facilitates the formation of the sLeaand sLex antigens
and, therefore, that it promotes metastasis in gastrointestinal cancers. In the present study, the results clearly indicate that DNA methylation in the promoter region plays a decisive role in the suppression of the Sdagene, and that the suppression can
be substantially relieved by treatment with the DNA methylation inhibitor, 5-aza-CdR. 5-aza-CdR is one of the most extensively studied drugs of this type, and therapeutic efficacy has been demonstrated in several kinds of cancers (Momparler 2005). In the present study, however, the results were obtained using gastrointestinal cancer cell lines. It is of importance, therefore, to ascertain whether the downregulation of Sdaβ4GalNAcT-II
gene expression during oncogenetic processes in gastrointesti-nal tissues occurs via this mechanism of DNA methylation in the promoter region, as demonstrated in cell lines. In addition, the possible existence of other control mechanisms, including those relative to positive and negative regulation, for expres-sion of the Sda gene in gastrointestinal tissues should not be neglected, and elucidation of these mechanisms is expected.
Materials and methods
Cell culture
Human colon cancer cell lines Caco-2, Colo 205, HCT 116, SW1417, SW48, SW480, and SW948, gastric cancer cell line
DNA methylation regulates Sdagene expression
KATO III, and pancreatic cancer cell line MIA PaCa-2 were purchased from the American Type Culture Collection (Manas-sas, VA). Human gastric cancer cell line MKN45 was purchased from the Japanese Collection of Research Bioresources (Health Science Research Resources Bank, Osaka, Japan). Colo 205 and KATO III cells were grown in 90% RPMI 1640 medium supple-mented with 10% fetal bovine serum, 50 units/mL of penicillin, and 50µg/mL of streptomycin. The other eight cell lines were grown in 90% Dulbecco’s modified eagle media with identical supplementations.
RT-PCR
Expression of the Sdaβ4GalNAcT-II gene in various gastroin-testinal cancer cell lines was evaluated by RT-PCR. Total RNAs of the assessed cancer cell lines were prepared using the QI-Aamp RNA Blood Mini Kit (Qiagen GmbH, Hilden, Germany). The first-strand cDNAs were primed by oligo-dT primer and synthesized by M-MLV reverse transcriptase (Promega Co., Madison, WI).
PCR amplifications specific for the short- and long-form Sda transcripts and the common region of the two transcript forms were performed to determine the Sda gene expression. The synthetic oligonucleotide primers, F41 (ACGGAGTGGAGGT-GATGC, locating exon 4) and R64 (TGGTGTATGTCACGT-GCTG, antisense sequence, locating exon 6), were used to am-plify the common region of the short- and long-form Sda
tran-scripts. The synthesized cDNA sample and 5 pmole of each primer were combined in 12.5 µL of PCR buffer containing 0.2 mM of dNTP and 0.625 U of Expand DNA polymerase (Roche Diagnostics GmbH, Mannheim, Germany). The PCR program included 2 min at 95◦C, followed by 40 cycles of 30 s at 94◦C, 30 s at 62◦C, and 1 min at 72◦C. The forward primers of F3 (GGGATTCGGGATGACTT, locating exon 1S)
and F7 (CTCTGCTTGGAACTCAGA, locating exon 1L) were
used for specific amplification for the short- and long-form Sda cDNAs, respectively, with the reverse primer, R10 (TG-GCTTCACATTTGCACTG, antisense sequence, locating exon 3), common for the two cDNA forms. The PCR conditions were similar to those mentioned above, except for annealing at 60◦C. PCR for theβ-actin cDNA was performed using the forward and reverse primers with sequences of CCTCGCCTTTGCC-GATCC and GGATCTTCATGAGGTAGTCAGTC (antisense sequence), respectively, and 25 cycles of PCR at an annealing temperature of 58◦C were performed. The PCR products were analyzed using 1.5% agarose gel electrophoresis.
Bisulfite genomic sequencing
Genomic DNAs of gastrointestinal cancer cell lines were prepared using the QIAamp DNA Blood Mini Kit (Qiagen). The bisulfite modification method was employed to determine the methylation status of cytosine residue in genomic DNA (Frommer et al. 1992). In this method, sodium bisulfite is used to convert cytosine residue to uracil in single-strand DNA under conditions whereby 5-methylcytosine remains non-reactive. In the following PCR amplification, the uracil and 5-methylcytosine residues in the sequence under investigation are amplified as thymine and cytosine, respectively. Bisulfite modification of the genomic DNA samples was achieved using the MethylEasy High Throughput DNA Bisulphite Modification Kit (Human Genetic Signatures Pty Ltd., Sydney, Australia).
The upper strand of the bisulfite-modified DNA segment encompassing the promoter and exon 1S regions of the Sda
gene was amplified using nested PCR. The first round PCR was performed with primers for the modified sequences, mF1
(TAGGATAGTAGGTGTGGAGTG, nucleotides−601 through
−581, transcription start nucleotide of exon 1L as +1) and
mR2 (AACCAACRCTCCCCATTCCT, complementary of nu-cleotides+56 through +75) (Figure 1). Twenty nanograms of bisulfite-modified genomic DNA and 5 pmole of each primer were combined in 12.5µL of PCR buffer containing 0.2 mM of dNTP and 0.25 U of Taq DNA polymerase (Promega). The PCR program consisted of 5 min at 94◦C followed by 30 cycles of 30 s at 94◦C, 30 s at 55◦C, and 1 min at 72◦C. The second round PCR was performed using the primers for the modified sequences, mF5 (TGAAGTGGAGGGTGGAGGTGGGATT, nucleotides −575 through −551) and mR4 (TCACCTCRCCTCTACAAC-CACCTAT, complementary of nucleotides+1 through +25). The conditions for the second round PCR were similar to those in the first PCR, except that annealing was at 65◦C. The PCR products were cloned into the pGEM-T Easy vector (Promega). DNA sequences were determined using the BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA).
The methylation statuses of individual CpG sites in the Sda gene CpG island in KATO III cells were analyzed in detail. The upper strand of the bisulfite-modified DNA segment en-compassing the majority of the putative CpG island of the Sda gene was PCR amplified. The first round PCR was
per-formed using primers for the modified sequences, mF1 and mR6 (CTTTCAAATATACTTTACCCCRTTA, complementary of nucleotides+446 through +470). The PCR program con-sisted of 5 min at 94◦C followed by 30 cycles of 30 s at 94◦C, 30 s at 58◦C, and 1 min at 72◦C. The second round PCR was performed using the primers for the modified sequences, mF5 and mR8 (ACCACACAACCRCAACTCACTCTACC, comple-mentary of nucleotides+381 through +406), at an annealing temperature of 68◦C. Five separate PCR amplifications were performed, and the PCR products cloned. Fifteen clones from the five separate batches of PCR products were selected and the sequence determined.
Methylation-specific PCR
Methylation-specific PCR analysis was established to evaluate the DNA methylation status in the Sda gene promoter. The primer pair of USPF and USPR for the modified sequences,
GAGTTGTGGGTGGGTTAAGGTTGTG (nucleotides −523
through −499) and ACCACCCACTCCCTAACTCCAACA
(complementary of nucleotides−53 through −30) (Figure 1), respectively, was designed for the unmethylated status. The primer pair of MSPF and MSPR for the modified sequences,
GTTGCGGGTGGGTTAAGGTCGCG (nucleotides −521
through−499) and CTACAAACGCCAAAAAAAACCCGCG
(complementary of nucleotides −103 through −79), respec-tively, was designed for the methylated status. The primer pair of mF1 and mR2 was used for the first round PCR, with the bisulfite-modified genomic DNAs serving as templates. The primer pairs for the unmethylated and methylated status (USPF+USPR and MSPF+MSPR, respectively) were then used separately in the second round PCR. The PCR program in the second round PCR consisted of 5 min at 94◦C followed by 30 cycles of 30 s at 94◦C, 30 s at 65◦C, and 1 min at 72◦C. The products were analyzed using 1.5% agarose gel electrophoresis.
H R Wang et al.
Treatment of cells with a DNA methylation inhibitor
HCT116 and SW48 colon cancer cells were split into six-well culture plates at a density of 1× 105cells/mL, and 5-aza-CdR
(Sigma-Aldrich Co., St. Louis, MO), a potent and specific in-hibitor of DNA methylation, was added to a final concentration of 5µM. Cells were incubated for 6 days before harvesting total RNAs and genomic DNAs and examining cell surface Sda antigen expression.
Analysis of cell surface Sdaantigen expression
The cell surface Sdaantigen was detected using the monoclonal
antibody (mAB) KM694, a generous gift from the Pharmaceu-tical Research Center, Kyowa Hakko Kogyo Co. Ltd. (Tokyo, Japan). The mAB KM694 was initially established as a mAB against GM2 ganglioside; however, greater binding affinity for the Sda structure was subsequently demonstrated (Kawamura
et al. 2005). To examine the expression of the Sdaantigen on the surfaces of the SW48 cells, the cells were incubated with 0.25µg/mL of mAB KM694 after blocking with 5% bovine serum albumin in phosphate-buffered saline. The bound mABs were detected by incubation with 0.5 µg/mL of fluorescein isothiocyanate (FITC)-conjugated goat antimouse IgM (Santa Cruz Biotechnology Inc., Santa Cruz, CA), and then the cells were subjected to flow cytometry. If the FITC signal bound on the cell surfaces was to be examined by fluorescence microscopy, the cells were incubated with 5µg/mL of mAB KM694.
Funding
National Science Council, Taiwan (NSC 96-2320-B-002-074-MY3).
Conflict of interest statement
None declared.
Abbreviations
Aprt, adenine phosphoribosyltransferase; 5-aza-CdR, 5-aza-2 -deoxycytidine; β4GalNAcT, β-1,4-N-acetylgalactosaminyl-transferas; FITC, fluorescein isothiocyanate; RBCs, red blood cells; RT-PCR, reverse transcription-polymerase chain reaction; sLea, sialyl Lewisa; sLex, sialyl Lewisx.
References
Ahuja N, Li Q, Mohan AL, Baylin SB, Issa J-PJ. 1998. Aging and DNA methylation in colorectal mucosa and cancer. Cancer Res. 58:5489–5494. Blanchard D, Cartron J-P, Fournet B, Montreuil J, van Halbeek H,
Vliegenthart JFG. 1983. Primary structure of the oligosaccharide deter-minant of blood group Cad specificity. J Biol Chem. 258:7691–7695. Blanchard D, Piller F, Gillard B, Marcus D, Cartron J-P. 1985. Identification
of a novel ganglioside on erythrocytes with blood group Cad specificity. J Biol Chem. 260:7813–7816.
Brandeis M, Frank D, Keshet I, Seigfried Z, Mendelsohn M, Nemes A, Temper V, Razin A, Cedar H. 1994. Sp1 elements protect a CpG island from de novo methylation. Nature. 371:435–438.
Capon C, Maes E, Michalski J-C, Leffler H, Kim YS. 2001. Sda-antigen-like structures carried on core 3 are prominent features of glycans from the mucin of normal human descending colon. Biochem J. 358:657–664.
Dall’Olio F, Malagolini N, Di Stefano G, Ciambella M, Serafini-Cessi F. 1990. Postnatal development of rat colon epithelial cells is associated with changes in the expression of theβ1,4-N-acetylgalactosaminyltransferase involved in the synthesis of Sdaantigen and ofα2,6-sialyltransferase activity towards
N-acetyl-lactosamine. Biochem J. 270:519–524.
Dall’Olio F, Malagolini N, Serafini-Cessi F. 1987. Tissue distribution and age-dependent expression of β-4-N-acetylgalactosaminyltransferase in guinea-pig. Biosci Rep. 7:925–932.
Daniels G. 2002. Human Blood Groups. Oxford: Blackwell.
Dell A, Chalabi S, Easton RL, Haslam SM, Sutton-Smith M, Patankar MS, Lattanzio F, Panico M, Morris HR, Clark GF. 2003. Murine and human zona pellucida 3 derived from mouse eggs express identical O-glycans. Proc Natl Acad Sci USA. 100:15631–15636.
Dohi T, Hanai N, Yamaguchi K, Oshima M. 1991. Localization of UDP-GalNAc: NeuAcα2,3Gal-R β1,4(GalNAc to Gal)N-acetylgalacto-saminyltransferase in human stomach. J Biol Chem. 266:24038–24043. Dohi T, Nakasuji M, Oshima M. 1993. Induction of the fundic mucosa-specific
glycolipid with dimethylformamide in gastric-cancer cell lines. Int J Cancer. 53:137–140.
Dohi T, Yuyama Y, Natori Y, Smith PL, Lowe JB, Oshima M. 1996. Detection of N-acetylgalactosaminyltransferase mRNA which determines expression of Sdablood group carbohydrate structure in human gastrointestinal mucosa and cancer. Int J Cancer. 67:626–631.
Donald ASR, Yates AD, Soh CPC, Morgan WTJ, Watkins WM. 1983. A blood group Sda-active pentasaccharide isolated from Tamm-Horsfall urinary glycoprotein. Biochem Biophys Res Commun. 115:625–631.
Esteller M. 2005. Aberrant DNA methylation as a cancer-inducing mechanism. Annu Rev Pharmacol Toxicol. 45:629–656.
Frommer M, McDonald LE, Millar DS, Collis CM, Watt F, Grigg GW, Molly PL, Paul CL. 1992. A genomic sequencing protocol that yields a positive display of 5-methylcytosine residues in individual DNA strands. Proc Natl Acad Sci USA. 89:1827–1831.
Gardiner-Garden M, Frommer M. 1987. CpG islands in vertebrate genomics. J Mol Biol. 196:261–282.
Graff JR, Herman JG, My¨oh¨anen S, Baylin SB, Vertino PM. 1997. Mapping patterns of CpG island methylation in normal and neoplastic cells implicates both upstream and downstream regions in de novo methylation. J Biol Chem. 272:22322–22329.
Kawamura YI, Kawashima R, Fukunaga R, Hirai K, Toyama-Sorimachi N, Tokuhara M, Dohi T. 2005. Introduction of Sda carbohydrate antigen in gastrointestinal cancer cells eliminates selectin ligands and inhibits metastasis. Cancer Res. 65:6220–6227.
Klose RJ, Bird AP. 2006. Genomic DNA methylation: The mark and its mediators. Trends Biochem Sci. 31:89–97.
Kominato Y, Hata Y, Takizawa H, Tsuchiya T, Tsukada J, Yamamoto F. 1999. Expression of human histo-blood group ABO gene is dependent upon DNA methylation of the promoter region. J Biol Chem. 274:37240–37250. Kondo Y, Issa J-PJ. 2004. Epigenetic changes in colorectal cancer. Cancer
Metastasis Rev. 23:29–34.
Lo Presti L, Cabuy E, Chiricolo M, Dall’Olio F. 2003. Molecular cloning of the humanβ1,4 N-acetylgalactosaminyltransferase responsible for the biosynthesis of the Sdahisto-blood group antigen: The sequence predicts a very long cytoplasmic domain. J Biochem. 134:675–682. Macleod D, Charlton J, Mullins J, Bird AP. 1994. Sp1 sites in the mouse aprt
gene are required to prevent methylation of the CpG island. Genes Dev. 8:2282–2292.
Macvie SI, Morton JA, Pickles MM. 1967. The reactions and inheritance of a new blood group antigen, Sda. Vox Sang. 13:485–492.
Malagolini N, Dall’Olio F, Di Stefano G, Minni F, Marrano D, Serafini-Cessi F. 1989. Expression of UDP-GalNAc: Neuα2,3Galβ-R β1,4(GalNAc to Gal)
N-acetylgalactosaminyltransferase involved in the synthesis of Sdaantigen in human large intestine and colorectal carcinomas. Cancer Res. 49:6466– 6470.
Malagolini N, Dall’Olio F, Serafini-Cessi F. 1991. UDP-GalNAc: NeuAcα2,3Galβ-R (GalNAc to Gal) β1,4-N-acetylgalactosaminyl-transferase responsible for the Sdaspecificity in human colon carcinoma CaCo-2 cell line. Biochem Biophys Res Commun. 180:681–686. Malagolini N, Santini D, Chiricolo M, Dall’Olio F. 2007. Biosynthesis and
expression of the Sda and sialyl Lewisx antigens in normal and cancer colon. Glycobiology. 17:688–697.
Momparler RL. 2005. Epigenetic therapy of cancer with 5-aza-2-deoxycytidine (decitabine). Semin Oncol. 32:443–451.
Montiel M-D, Krzewinski-Recchi M-A, Delannoy P, Harduin-Lepers A. 2003. Molecular cloning, gene organization and expression of the human
DNA methylation regulates Sdagene expression
UDP-GalNAc: Neu5Acα2-3Galβ-R β1,4-N-acetylgalactosaminyltransfe-rase responsible for the biosynthesis of the blood group Sda/Cad antigen: Evidence for an unusual extended cytoplasmic domain. Biochem J. 373:369– 379.
Morton JA, Pickles MM, Terry AM. 1970. The Sdablood group antigen in tissues and body fluids. Vox Sang. 19:472–482.
Mummaneni P, Walker KA, Bishop PL, Turker MS. 1995. Epigenetic gene inactivation induced by a cis-acting methylation center. J Biol Chem. 270:788–792.
Mummaneni P, Yates P, Simpson J, Rose J, Turker MS. 1998. The primary function of a redundant Sp1 binding site in the mouse aprt gene pro-moter is to block epigenetic gene inactivation. Nucleic Acids Res. 26: 5163–5169.
Pao MM, Tsutsumi M, Liang G, Uzvolgyi E, Gonzales FA, Jones PA. 2001. The endothelin receptor B (EDNRB) promoter displays heterogenous, site specific methylation patterns in normal and tumor cells. Hum Mol Genet. 10:903–910.
Renton PH, Howell P, Ikin EW, Giles CM, Goldsmith KLG. 1967. Anti-Sda, a new blood group antibody. Vox Sang. 13:493–501.
Robbe C, Capon C, Maes E, Rousset M, Zweibaum A, Zanetta J-P, Michalski J-C. 2003. Evidence of regio-specific glycosylation in human intestinal mucins. J Biol Chem. 278:46337–46348.
Sanger R, Gavin J, Tippett P, Teesdale P. 1971. Plant agglutinin for another human blood-group. Lancet. i:1130.
Scarano MI, Strazzullo M, Matarazzo MR, D’Esposito M. 2005. DNA methylation 40 years later: Its role in human health and disease. J Cell Physiol. 204:21–35.
Schmid CW. 1998. Does SINE evolution preclude Alu function? Nucleic Acids Res. 26:4541–4550.
Serafini-Cessi F, Dall’Olio F. 1983. Guinea-pig kidney β-N-acetylgalacto-saminyltransferase towards Tamm-Horsfall glycoprotein. Biochem J. 215:483–489.
Smith PL, Lowe JB. 1994. Molecular cloning of a murine N-acetylgalacto-samine transferase cDNA that determines expression of the T lymphocyte-specific CT oligosaccharide differentiation antigen. J Biol Chem. 269:15162–15171.
Song F, Smith JF, Kimura MT, Morrow AD, Matsuyama T, Nagase H, Held WA. 2005. Association of tissue-specific differentially methylated region (TDMs) with differential gene expression. Proc Natl Acad Sci USA. 102:3336–3341. Song JZ, Stirzaker C, Harrison J, Melki JR, Clark SJ. 2002. Hypermethylation trigger of the glutathione-S-transferase gene (GSTP1) in prostate cancer cells. Oncogene. 21:1048–1061.
Suzuki H, Gabrielson E, Chen W, Anbazhagan R, van Engeland M, Weijenberg MP, Herman JG, Baylin SB. 2002. A genetic screen for genes upregulated by demethylation and histone deacetylase inhibition in human colorectal cancer. Nat Genet. 31:141–149.
Takai D, Jones PA. 2002. Comprehensive analysis of CpG islands in human chromosomes 21 and 22. Proc Natl Acad Sci USA. 99:3740–3745. Toyota M, Ahuja N, Ohe-Toyoto M, Herman JG, Baylin SB, Issa J-PJ. 1999.
CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci USA. 96:8681–8686.
Toyota M, Issa J-PJ. 2005. Epigenetic changes in solid and hematopoietic tumors. Semin Oncol. 32:521–531.
Turker MS. 2002. Gene silencing in mammalian cells and the spread of DNA methylation. Oncogene. 21:5388–5393.
Yates PA, Burman RW, Mummaneni P, Krussel S, Turker MS. 1999. Tandem B1 elements located in a mouse methylation center provide a target for de novo DNA methylation. J Biol Chem. 274:36357–36361.
