Chromosomal losses are associated with hypomethylation of the gene-control regions in the stomach with a low number of active genes.

Abstract

Transitional-CpG methylation between unmethylated promoters and nearby methylated retroelements plays a role in the establishment of tissue-specific transcription. This study examined whether chromosomal losses reducing the active genes in cancers can change transitional-CpG methylation and the transcription activity in a cancer-type-dependent manner. The transitional-CpG sites at the CpG-island margins of nine genes and the non-island-CpG sites round the transcription start sites of six genes lacking CpG islands were examined by methylation-specific polymerase chain reaction (PCR) analysis. The number of active genes in normal and cancerous tissues of the stomach, colon, breast, and nasopharynx were analyzed using the public data in silico. The CpG-island margins and non-island CpG sites tended to be hypermethylated and hypomethylated in all cancer types, respectively. The CpG-island margins were hypermethylated and a low number of genes were active in the normal stomach compared with other normal tissues. In gastric cancers, the CpG-island margins and non-island-CpG sites were hypomethylated in association with high-level chromosomal losses, and the number of active genes increased. Colon, breast, and nasopharyngeal cancers showed no significant association between the chromosomal losses and methylation changes. These findings suggest that chromosomal losses in gastric cancers are associated with the hypomethylation of the gene-control regions and the increased number of active genes.

Full text

Page 1

INTRODUCTION
The establishment of tissue-specific gene expression pro-
files relies on a variety of transcription factor cascades, which
initiate embryogenesis and development, and on DNA methy-
lation, which prevents the reactivation of inactive genes (1).
The promoter regions of the housekeeping genes always over-
lap with the CpG islands, whereas approximately 60% of the
genes lacking CpG islands are expressed in a tissue-specific
manner (2, 3). However, an inverse correlation between the
methylation state and gene expression is found in only 30%
of the genes randomly selected (4). Therefore, it is believed
that rather than simply acting as an on-off switch in gene reg-
ulation, DNA methylation is essential for maintaining the
well-organized gene expression in a variety of somatic tissues
(5), and that changes in the normal methylation state might
play a role in the genesis and development of cancers (6).
The loss-of-heterozygosity (LOH) event detected by an
analysis of microsatellite sequences represents the unilateral
chromosomal loss, major genetic alterations observed in human
solid tumors (7). Although the LOH event may be a part of
bi-allelic gene inactivation, it is still unclear if a decrease in
the chromosomal dose plays a role in carcinogenesis. Assum-
ing the DNA methylation profile of a given tissue is estab-
lished with respect to the global gene expression profiles, it
is possible that the gene-copy number reduction caused by
LOH can influence the methylation and transcriptional sta-
tus of the genes on the chromosomal copies retained. How-
ever, there are a few reports describing the functional rela-
tionship between the genetic and epigenetic features of can-
cer cells, especially on how the LOH events and methylation
changes collectively influence the global gene expression pro-
files of cancer cells. We previously observed that the methy-
lation-variable CpG sites between unmethylated promoters
and nearby methylated retroelements, including the CpG-
island margins and the non-island-CpG sites the genes lack-
ing CpG islands, are associated with the global as well as
individual gene expression patterns (8-10). These findings
lead us to propose an assumption that a decrease in the num-
ber of active gene copy caused by LOH might facilitate the
methylation changes in the transitional-CpG sites as well as
the changes in the total number of active genes.
1068
Yu-Chae Jung, Seung-Jin Hong,
Young-Ho Kim, Sung-Ja Kim,
Seok-Jin Kang*, Sang-Wook Choi
and Mun-Gan Rhyu
�,
Departments of Microbiology, Clinical Pathology*, and
Internal Medicine
University of Korea, Seoul, Korea
�, College of Medicine, The Catholic
Address for correspondence
Mun-Gan Rhyu, M.D.
Department of Microbiology, The Catholic University of
Korea, 505 Banpo-dong, Seocho-gu, Seoul 135-701,
Korea
Tel : +82.2-590-1215, Fax : +82.2-596-8969
E-mail : rhyumung@catholic.ac.kr
*This work was supported by grants from the Catholic
Institute of Cell therapy Basic Science Programs
Foundation made in the program year of 2006
(2006005041) and the BumSuk Academy Research
Fund (202-82-05650).
J Korean Med Sci 2008; 23: 1068-89
ISSN 1011-8934
DOI: 10.3346/jkms.2008.23.6.1068
Copyright � The Korean Academy
of Medical Sciences
Chromosomal Losses are Associated with Hypomethylation of the
Gene-Control Regions in the Stomach with a Low Number of Active
Genes
Transitional-CpG methylation between unmethylated promoters and nearby methy-
lated retroelements plays a role in the establishment of tissue-specific transcription.
This study examined whether chromosomal losses reducing the active genes in
cancers can change transitional-CpG methylation and the transcription activity in a
cancer-type-dependent manner. The transitional-CpG sites at the CpG-island mar-
gins of nine genes and the non-island-CpG sites round the transcription start sites
of six genes lacking CpG islands were examined by methylation-specific polymerase
chain reaction (PCR) analysis. The number of active genes in normal and cancer-
ous tissues of the stomach, colon, breast, and nasopharynx were analyzed using
the public data in silico. The CpG-island margins and non-island CpG sites tended
to be hypermethylated and hypomethylated in all cancer types, respectively. The
CpG-island margins were hypermethylated and a low number of genes were active
in the normal stomach compared with other normal tissues. In gastric cancers, the
CpG-island margins and non-island-CpG sites were hypomethylated in associa-
tion with high-level chromosomal losses, and the number of active genes increased.
Colon, breast, and nasopharyngeal cancers showed no significant association
between the chromosomal losses and methylation changes. These findings sug-
gest that chromosomal losses in gastric cancers are associated with the hypomethy-
lation of the gene-control regions and the increased number of active genes.
Key Words : Chromosomal Loss; Loss of Heterozygosity; CpG Methylation; Methylation-Specific PCR; Tissue
Expression Profiles
Received : 6 November 2007
Accepted : 1 April 2008

Page 2

Genetic and Epigenetic Changes Related with Transcript Number in Gastric Cancer1069
In this study, we examined the chromosomal losses events,
methylation changes, and global expression profiles in gas-
tric, colonic, mammary and nasopharyngeal cancers in order
to further delineate the interaction of genetic, epigenetic,
and transcriptional alterations using LOH, methylation-spe-
cific polymerase chain reaction (PCR) and Serial Analysis of
Gene Expression (SAGE) data.
MATERIALS AND METHODS
Collection of cancer and normal tissues
Formalin-fixed, paraffin-embedded tissue sections of four
cancer types (stomach, colon, breast, and nasopharynx) were
obtained from 100 patients who had undergone a surgical
resection between March 2004 and July 2006 at St. Paul’s
Hospital and Kangnam St. Mary’s Hospital, The Catholic
University of Korea. The Institutional Review Board approved
this study, and written informed consent was obtained from
each patient before surgery. A single 10-mm-diameter site
containing a homogeneous cell population was selected from
each section that contained a representative of paired normal
and tumor tissues. Seven-μ m-thick hematoxylin-eosin-stained
sections were microdissected under a 40× stereomicroscope
using a surgical scalpel. In the microscopic examination, the
tumor specimens consisted mainly of tumor tissue and the
normal tissues did not show any evidence of tumor cell inva-
sion or significant inflammatory involvement. Approximate-
ly 50 microdissected cells were digested in 1 μ L of a Tween
20-Proteinase K lysis buffer.
PCR-based microsatellite analysis
A pair of normal and cancer DNAs was examined using a
panel of PCR primers that covered 40 microsatellite loci on
eight cancer-associated chromosomes, 3p, 4p, 5q, 8p, 9p, 13q,
17p, and 18q, as reported elsewhere (11-14). The allelic pro-
files of the 40 microsatellite sequences were initially analyzed
for any microsatellite instability (MSI) at the homozygous mark-
ers showing a few stutter bands in a pair of normal and cancer
tissues. If the cancer DNA showed novel bands that were
absent in normal DNA, in >40% of the homozygous mic-
rosatellite alleles, they were interpreted as being a MSI. The
difference in the allelic intensity between the normal and
cancer DNAs was scored as the relative allelic ratio calculat-
ed by dividing the intensity ratio of the cancer by the nor-
mal allelic ratio. A relative allelic ratio >1.5 in the heterozy-
gous case without a MSI was interpreted as a LOH on the
basis of the distribution of relative allelic ratios, as this pro-
vided the best discrimination between wild-type heterozy-
gosity and LOH.
Methylation-specific PCR and sequencing of bisulfite-
modified DNA
Ninety microliters of genomic DNA was denatured with
10 μ L of 3 M NaOH for 15 min at 37℃ and modified with
1,040 μ L of 2.3 M sodium bisulfite and 60 μ L of 10 mM
hydroquinone for 12 hr at 50℃, as described elsewhere (12,
13). The methylation-variable CpG sites reported in previ-
ous study of 11 somatic tissue types were chosen in the tran-
sitional area between promoters and retroelements (9). A total
of 15 CpG sites selected from six CpG-island-negative genes
(MAGEA2,TFF2,DDX53,MSLN,MASPIN,andBGLAP)
and nine CpG-island-positive genes (MUC8, KIAA1752,
CDKN2A, ESR2, PPARG, MLH1, CDH1, VDR, and
RUNX3) were examined using methylation-specific PCR
(MSP) primer sets (Supplementary Table 1 and 2). Supple-
mentary Fig. 1 shows the coverage of CpG islands along with
The mean numbers of active genes and the expressed tags were calculated using two to eight SAGE libraries for each tissue type. Individual genes
were classified according to on the presence or absence of CpG islands and the type of nearby retroelements in the 5′ -end regions.
Genes with CpG islands
Alu and L1 retroelement
Active
genes
Tags per
gene
L1 retroelement
Tissue
Active
genes
Tags per
gene
Alu retroelement
Active
genes
Tags per
gene
L1 retroelement
Active
genes
Tags per
gene
Alu retroelement
Active
genes
Tags per
gene
Genes without CpG islands
Embryo
Placenta
Normal tissues
Stomach
Colon
Breast
Cancer tissues
Stomach
Colon
Breast
513
395
7.1
4.6
3,576
2,885
15.3
7.8
5,817
4,652
16.7
8.1
999
870
5.7
10.3
810
704
6.7
11.9
213
246
332
2.3
3.9
3.5
1,676
2,108
2,386
4.3
5.0
5.6
2,636
3,529
3,988
4.2
5.1
5.9
444
543
672
11.4
7.3
4.6
368
480
543
5.0
3.8
5.2
315
289
334
3.0
2.5
3.1
2,273
2,382
2,474
5.7
4.2
5.4
3,736
3,910
4,122
6.0
4.5
5.9
624
643
641
4.8
4.0
3.9
540
541
552
5.5
3.0
4.0
Table 1. The number of active genes and the number of tags expressed per active gene in embryos and somatic tissues

Page 3

1070 Y.-C. Jung, S.-J. Hong, Y.-H. Kim, et al.
the distribution of retroelements in the 5′ -end regions of
the 15 genes examined.
For semiquantitative MSP analysis, a minimum number
of PCR rounds to reach sub-plateau DNA amplification were
performed using a radioisotope. The bisulfite-modified DNA
was amplified and labeled by a hot-start PCR containing α -32P
dTTP (PerkinElmer, Boston, MA, U.S.A.) and dNTP mix-
ture through 32 PCR cycles. The PCR products were load-
ed onto a nondenaturing polyacrylamide gel and visualized
by repeated autoradiography using a radioluminograph scan-
ner (BAS 2500, Fuji Photo Film, Kanakawa, Japan) and ana-
lyzed with TINA image software (Raytest Isotopenmeβ ger-
ate, Straubenhardt, Germany). The specificity of each MSP
primer set was validated using a standard curve for the uni-
versal methylated and unmethylated DNAs, as described
elsewhere (9).
The relative proportion of the methylated CpG band to
the total intensity of the methylated and unmethylated CpG
bands was calculated from the MSP bands amplified by the
MSP primer set. The proportion of methylated CpGs was
divided into 5 levels; level 1 (0-20% methylation), level 2
(21-40% methylation), level 3 (41-60% methylation), level
4 (61-80% methylation), and level 5 (81-100% methyla-
tion). The results of the LOH and MSP analyses on the four
cancer types are listed in Supplementary Table 3.
The methylation-variable CpGs of the TFF2, MSLN,
BGLAP, CDKN2A, and MLH1 genes were analyzed by
cloning and sequencing of the common PCR DNA (Sup-
plementary Fig. 2). The common PCR primer sets were
designed to span both the unmethylated and methylated
CpGs in the CpG amplicons. The PCR product of each com-
mon primer set was cloned into the T&A cloning vector
(Real Biotech, Taipei, Taiwan). The DNA sequencing was
performed using a BigDye Terminator Kit (PE Biosystems,
Genes with CpG islands
L1 element
EmbryoPlacenta
Alu and L1
EmbryoPlacenta
Alu element
Embryo Placenta
L1 element
Embryo Placenta
Alu element
EmbryoPlacenta
Genes without CpG islands
Normal tissues
Stomach1
2
3
4
1
2
1
2
3
4
5
6
0.48
0.39
0.57
0.47
0.34
0.62
0.08
0.15
0.11
0.17
0.11
0.15
0.16
0.17
0.27
0.42
0.10
0.07
0.32
0.19
0.36
0.34
0.30
0.24
0.33
0.31
0.48
0.50
0.20
0.24
0.69
0.55
0.57
0.61
0.61
0.64
0.59
0.59
0.72
0.73
0.43
0.48
0.59
0.44
0.46
0.58
0.45
0.51
0.44
0.51
0.63
0.47
0.20
0.27
0.68
0.63
0.70
0.61
0.70
0.75
0.45
0.54
0.81
0.54
0.32
0.35
0.75
0.70
0.70
0.66
0.67
0.56
0.05
0.04
0.09
0.10
0.33
0.22
0.73
0.61
0.51
0.57
0.39
0.64
0.01
0.02
0.03
0.27
0.14
0.06
0.20
0.17
0.19
0.21
0.10
0.21
0.20
0.21
0.17
0.20
0.02
0.07
0.05
0.06
0.25
0.19
0.31
0.24
0.10
0.07
0.11
0.12
0.05
0.13
0.01
0.01
0.39
0.90
0.18
0.08
Colon
Breast
Cancer tissues
Stomach1
2
3
4
5
6
1
2
1
2
3
4
5
6
7
8
0.26
0.28
0.21
0.25
0.33
0.31
0.45
0.29
0.34
0.19
0.14
0.34
0.17
0.33
0.18
0.40
0.17
0.35
0.21
0.26
0.30
0.37
0.12
0.13
0.50
0.37
0.13
0.20
0.28
0.42
0.35
0.18
0.68
0.66
0.65
0.59
0.63
0.66
0.29
0.38
0.65
0.60
0.70
0.69
0.70
0.64
0.61
0.70
0.78
0.57
0.58
0.57
0.79
0.83
0.50
0.35
0.78
0.77
0.51
0.72
0.63
0.76
0.72
0.73
0.53
0.53
0.50
0.68
0.54
0.72
0.37
0.24
0.46
0.56
0.63
0.72
0.57
0.47
0.40
0.58
0.66
0.70
0.64
0.50
0.60
0.73
0.54
0.46
0.55
0.53
0.45
0.67
0.73
0.54
0.44
0.67
0.68
0.55
0.41
0.64
0.66
0.73
0.55
0.44
0.54
0.56
0.56
0.59
0.39
0.46
0.62
0.54
0.15
0.21
0.14
0.12
0.22
0.18
0.29
0.14
0.17
0.18
0.08
0.48
0.08
0.13
0.18
0.13
0.29
0.17
0.20
0.12
0.38
0.32
0.09
0.35
0.21
0.28
0.30
0.09
0.31
0.18
0.20
0.32
0.19
0.79
0.75
0.25
0.53
0.78
0.02
0.28
0.51
0.58
0.07
0.01
0.18
0.56
0.31
0.82
Colon
Breast
Table 2. Statistical analysis (R values) of the number of expressed tags for gene expression similarities between the embryo and somat-
ic tissues
Using the SAGE data in silico, the number of expressed tags of individual genes was compared between the embryo (embryonic cell line of passage
16 and the first trimester placenta) and somatic tissues. Total genes were classified according to the presence or absence of CpG islands and the
type of the nearby retroelements. Pearson’s correlation coefficients of p<0.05 are marked in bold. R values ≥0.5 are indicated by italics.

Page 4

Genetic and Epigenetic Changes Related with Transcript Number in Gastric Cancer 1071
Foster City, CA, U.S.A.) and an ABI automated DNA se-
quencer (PE Biosystems, Warrington, U.K.). The intensity
of the MSP bands was compared with the distribution of
methylated CpGs, which were determined by cloning and
sequencing of the common PCR amplicons.
Methylation analysis of normal somatic tissues
The methylation status of the transitional CpGs examined
in 11 normal somatic tissues was previously reported (9).
This study extended the previous data by adding six transi-
tional sites (TFF2, MSLN, BGLAP, MUC8, KIAA1752,
and PPARG) and three somatic tissues (fat, ovary, and testis).
A total of eleven tissue types were collected from 51 indi-
viduals. The bone marrow and fat stromal cells were obtained
from three individuals (10) and each of the remaining nine
tissue types were obtained from five individuals. The stom-
ach and colon was divided into proximal and distal portions
according to the anatomic landmarks, gastric body and the
splenic flexure of the colon.
Analysis of in-silico data
A total of 32 SAGE libraries for the embryonic stem cells,
placenta, stomach, colon, and breast tissues were obtained
from a public database (http://cgap.nci.nih.gov/SAGE/). Six
expressed sequence tag (EST) libraries for the nasopharyn-
geal tissues were collected from a public database (http://cgap.
nci.nih.gov/Tissues) because of no SAGE data available. The
SAGE and EST libraries analyzed are listed in Supplemen-
tary Table 4. The SAGE tags and EST tags were assigned
using UniGene cluster ID by accumulating the total expressed
tags to the matched genes at each tissue library. A total of
434,325 expressed tags in the 32 SAGE libraries correspond-
ed to the 15,770 gene symbols through tag-to-gene matching.
The genomic location of the NCBI RefSeq cDNA sequ-
A
17p 18q 9p 13q 5q
Individual chromosomal loss
Gastric cancer
4p 3p 8p
Gastric cancer
Fig. 1. Chromosomal losses detected in the gastric, colonic, mammary, and nasopharyngeal cancers (25 cases for each cancer type).
(A) Individual chromosomal losses and (B) the number of chromosomal losses were evaluated by PCR-based analysis using 40 microsatel-
lite markers on chromosomes 3p, 4p, 5q, 8p, 9p, 13q, 17p, and 18q.
No. of cases
20
16
12
8
4
0
17p 18q 8p 13q 5q
Individual chromosomal loss
Colonic cancer
9p 3p 4p
Colonic cancer
8p 13q 17p 3p 4p
Individual chromosomal loss
Mammary cancer
5q 9p 18q
Mammary cancer
9p
Individual chromosomal loss
Nasopharyngeal cancer
8p 13q 3p 5q 17p 18q 4p
Nasopharyngeal cancer
B
0
No. of chromosomal losses
12345678
No. of cases
8
6
4
2
0
0
No. of chromosomal losses
123456780
No. of chromosomal losses
123456780
No. of chromosomal losses
12345678
The methylation of the CpG-island margins were divided into four levels (+, ++, +++, and ++++) according to the genome-wide DNA methylation of
the embryo (27, 28) and the placenta (29) and the CpG-island margin methylation estimated in somatic tissues (Fig. 4). The number of active genes
and the transcript number per active gene were divided into four or five levels according to the ranks observed in the embryo and somatic tissues
(Fig. 5). Hypomethylation changes in cancer tissues are detailed in Fig. 3. NE, not examined.
Nasopharynx BreastColonStomachPlacenta Embryo
CpG-island margin methylation
Association between high-level LOH
and hypomethylation
Genes with CpG islands
No. of active genes
Transcript number per gene
Genes without CpG islands
No. of active genes
Transcript number per gene
L1-close transcript proportion
+
NE
++
NE
++++
Yes
++++
No
+++
No
+++
No
+++++
+++++
++++
++++
+
+
++
++
+++
+++
NE
NE
+++++
++
+
++++
++++
+++
+ ++
++
+++
+++
+
++
NE
NE
++
+++
++++
Table 3. The methylation status of CpG-island margins, the number of active genes, and the transcript number per active gene in
embryo and somatic tissues

Page 5

1072Y.-C. Jung, S.-J. Hong, Y.-H. Kim, et al.
ences was obtained from the genome web site (UCSC Gold-
en Path May, 2004 assembly, http://genome.ucsc.edu/). A
3-kb sized non-overlapping window was used to analyze a
DNA segment upstream and downstream of the transcrip-
tion start site. The coverage of CpG islands and the distri-
bution of retroelements compiled from searches of the genome
database were evaluated by inputting the sequence data that
was delimited from the 5′ -end regions into a local program.
The annotation of retroelements was made using the Repeat-
Masker program (http://ftp.genome.washington.edu/RM/
RepeatMasker.html). A total of 15,770 active genes show-
ing the tag-and-gene match were demarcated by the pres-
ence or absence of CpG islands at the transcription sites as
well as the types of nearby retroelements existing in a 3-kb
window. For the fidelity of the genome-wide expression data,
we compared SAGE and Affymetrix GeneChip (http://sy-
matlas.gnf.org/SymAtlas), and there was strong agreement
in major mRNA content of the analyzed tissue types (data
not shown) as previous reports (15, 16). However, the SAGE
data was found to reliably reflect the wide-range of transcrip-
tion level by counting the sequence-based ‘digital’ tags, while
the microarray data based on the fluorescence signal was not
suitable for defining the active or inactive transcription sta-
tus as well as the estimation of strong transcription activi-
ties due to the limit of probe hybridization method.
Statistical analysis
A chi-square test was used to compare the methylation
changes between gastric, colonic, mammary, and nasopha-
ryngeal cancers. The Pearson’s correlation coefficients of the
expressed tag numbers in different tissues were calculated to
determine the similarities of the individual gene expression.
A two-sided pvalue <0.05 was considered significant.
RESULTS
The level of chromosomal losses estimated in four cancer
types
The LOH events in each cancer were determined using 40
MAGEA2
Fig. 2. Methylation changes in the transitional-CpG sites of the 15 selected genes examined in four cancer types. The methylation status
of the transitional-CpG sites was estimated using a semiquantitative methylation-specific PCR method. The methylation changes were
scored based on the difference in the level of methylation between the normal and tumor tissues. The frequency of methylation changes
in each transitional area is indicated as a percentage in 25 cancer cases. The level of chromosomal losses was evaluated by PCR-based
loss-of-heterozygosity analysis. The cancer tissues were grouped into high-level chromosomal losses (H) four or more chromosomes and
low-level chromosomal losses (L) involving less than four chromosomes.
HL H LH L H L
Stomach Colon BreastNaso-
pharynx
HL H L H LH L
Stomach Colon BreastNaso-
pharynx
HLH L H LH L
Stomach Colon BreastNaso-
pharynx
HL H L H LH L
Stomach Colon BreastNaso-
pharynx
%
100
80
60
40
20
0
DDX53 MASPINTFF2
MSLN
%
100
80
60
40
20
0
BGLAP MUC8ESR2
CDH1
%
Frequency of hypo- (
) and hyper- ( ) methylation
100
80
60
40
20
0
KIAA1752PPARG VDR
CDKN2A
%
100
80
60
40
20
0
MLH1 RUNX3