Content uploaded by Yoshimasa Saito
Author content
All content in this area was uploaded by Yoshimasa Saito
Content may be subject to copyright.
Overexpression of a splice variant of DNA
methyltransferase 3b, DNMT3b4, associated
with DNA hypomethylation on pericentromeric
satellite regions during human hepatocarcinogenesis
Yoshimasa Saito*
†
, Yae Kanai*, Michiie Sakamoto*, Hidetsugu Saito
†
, Hiromasa Ishii
†
, and Setsuo Hirohashi*
‡
*Pathology Division, National Cancer Center Research Institute, 5-1-1 Tsukiji, Chuo-ku, Tokyo 104-0045, Japan; and †Department of Internal Medicine,
Keio University School of Medicine, 35 Shinanomachi, Shinjyuku-ku, Tokyo 160-8582, Japan
Edited by Stanley M. Gartler, University of Washington, Seattle, WA, and approved May 31, 2002 (received for review March 1, 2002)
DNA hypomethylation on pericentromeric satellite regions is an early
and frequent event associated with heterochromatin instability dur-
ing human hepatocarcinogenesis. A DNA methyltransferase,
DNMT3b, is required for methylation on pericentromeric satellite
regions during mouse development. To clarify the molecular mech-
anism underlying DNA hypomethylation on pericentromeric satellite
regions during human hepatocarcinogenesis, we examined muta-
tions of the DNMT3b gene and mRNA expression levels of splice
variants of DNMT3b in noncancerous liver tissues showing chronic
hepatitis and cirrhosis, which are considered to be precancerous
conditions, and in hepatocellular carcinomas (HCCs). Mutation of the
DNMT3b gene was not found in HCCs. Overexpression of DNMT3b4,
a splice variant of DNMT3b lacking conserved methyltransferase
motifs IX and X, significantly correlated with DNA hypomethylation
on pericentromeric satellite regions in precancerous conditions and
HCCs (Pⴝ0.0001). In particular, the ratio of expression of DNMT3b4
to that of DNMT3b3, which is the major splice variant in normal liver
tissues and retains conserved methyltransferase motifs I, IV, VI, IX,
and X, showed significant correlation with DNA hypomethylation
(Pⴝ0.009). Transfection of human epithelial 293 cells with DNMT3b4
cDNA induced DNA demethylation on satellite 2 in pericentromeric
heterochromatin DNA. These results suggest that overexpression of
DNMT3b4, which may lack DNA methyltransferase activity and com-
pete with DNMT3b3 for targeting to pericentromeric satellite regions,
results in DNA hypomethylation on these regions, even in precan-
cerous stages, and plays a critical role in human hepatocarcinogenesis
by inducing chromosomal instability.
DNA methylation plays important roles in gene silencing,
chromatin remodeling, and genome stability (1– 4). Aber-
rant DNA methylation is one of the most consistent epigenetic
changes in human cancers (1–4). Generally, the overall level of
DNA methylation is lower in cancer cells than in normal cells (5,
6), although a number of tumor suppressor genes are silenced by
DNA methylation on CpG islands around their promoter regions
in cancer cells (1–4, 7, 8).
We have carefully examined alterations of DNA methylation
status on pericentromeric satellite regions and CpG islands of
specific genes, and expression of DNA methyltransferases and
methyl-CpG-binding proteins, in noncancerous liver tissues show-
ing chronic hepatitis and cirrhosis, which are considered to be
precancerous conditions (9, 10), and in hepatocellular carcinomas
(HCCs) (11–17). Among these alterations, DNA hypomethylation
on pericentromeric satellite regions was detected even in precan-
cerous conditions and appears to be one of the earliest epigenetic
changes during human hepatocarcinogenesis (17). Satellite regions
are located in pericentromeric heterochromatin DNA, and DNA
hypomethylation on these regions is known to result in centromeric
decondensation, enhancing chromosome recombinations (18, 19).
In fact, frequent chromosome 1q copy gain with a pericentromeric
breakpoint was reported in HCCs showing DNA hypomethylation
on satellite 2 (20). DNA hypomethylation on pericentromeric
satellite regions may induce chromosomal instability during hepa-
tocarcinogenesis, even in precancerous conditions.
A newly identified DNA methyltransferase, DNMT3b (21), is
specifically required for DNA methylation on pericentromeric
satellite regions in embryonic stem cells and early mouse embryos
(22). Germ-line mutations of the DNMT3b gene have been re-
ported in patients with immunodeficiency兾centromeric instability兾
facial anomalies syndrome (23, 24), a rare recessive autosomal
disorder characterized by DNA hypomethylation on pericentro-
meric satellite regions (25). These findings encouraged us to
examine genetic alterations of the DNMT3b gene in HCCs.
Down-regulation of DNMT3b is unlikely to underlie DNA
hypomethylation on pericentromeric satellite regions during human
hepatocarcinogenesis, because we and other groups have reported
increased expression of the mRNA for DNMT3b in human cancers,
including HCCs (17, 26, 27). However, four splice variants of human
DNMT3b in the C-terminal catalytic domain are known (ref. 26,
Fig. 1). DNMT3b4 and DNMT3b5, without conser ved methyltrans-
ferase motifs IX and X, probably lack DNA methyltransferase
activity. So far, no studies of the expression of DNMT3b in human
cancers have discriminated between the splice variants.
To clarify the molecular mechanism underlying DNA hypom-
ethylation on pericentromeric satellite regions during human hepa-
tocarcinogenesis, we examined mutations of the DNMT3b gene and
expression levels of splice variants of DNMT3b in noncancerous
liver tissues showing chronic hepatitis and cirrhosis and in HCCs.
Materials and Methods
Patients and Tissue Specimens. Fifty-nine primary HCCs and the
corresponding noncancerous liver tissues were obtained from sur-
gically resected materials from 49 patients (cases H1 to H49) who
were treated at the National Cancer Center Hospital, Tokyo.
Twelve patients were hepatitis B virus surface antigen (HBs-Ag)-
positive, 29 were anti-hepatitis C virus antibody (anti-HCV)-
positive, one was both HBs-Ag and anti-HCV positive, and seven
were both HBs-Ag and anti-HCV negative. Histological examina-
tion of noncancerous liver tissues from HCC patients revealed no
remarkable findings, findings compatible with chronic hepatitis,
and findings compatible with cirrhosis in 1, 24, and 24 tissues,
respectively. For comparison, normal liver tissues showing no
remarkable histological findings were also obtained from eight
patients (cases C1 to C8) who were both HBs-Ag and anti-HCV
negative and underwent partial hepatectomy for liver metastasis of
primary colon cancer.
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: DNMT3b, DNA methyltransferase 3b; HCC, hepatocellular carcinoma; RT, re-
verse transcription; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
‡To whom reprint requests should be addressed. E-mail: shirohas@ncc.go.jp.
10060–10065
兩
PNAS
兩
July 23, 2002
兩
vol. 99
兩
no. 15 www.pnas.org兾cgi兾doi兾10.1073兾pnas.152121799
PCR–Single-Strand Conformation Polymorphism Analysis and Direct
Sequencing. Genomic DNA was amplified by PCR with 22
rhodamine-labeled intronic oligonucleotide primer sets encom-
passing all coding exons of the DNMT3b gene (GenBank accession
no. AL035071 and Table 1). Primers were designed on the basis of
the exon-intron boundaries as described (28). PCR products were
electrophoresed on 6% polyacrylamide gel. DNA was recovered
from the mobility-shifted bands and sequenced in both directions
with the ABI PRISM BigDye Terminator kit and an A BI PRISM
310 Genetic Analyzer (Applied Biosystems).
Reverse-Transcription (RT)-PCR and Cloning Sequencing. First-strand
cDNA was prepared from total RNA with random
hexadeoxynucleotide primers and SuperScript RNase H
–
reverse
transcriptase (GIBCO兾BRL). cDNA derived from human testis
total RNA (CLONTECH) was used as control. Subsequent PCR
with a primer set of (forward) 5⬘-CCT GCT GAA TTA CTC ACG
CCC C-3⬘and (reverse) 5⬘-GTC TGT GTA GTG CAC AGG AA A
GCC-3⬘(26) amplified all four splice variants in the C-terminal
catalytic domain of the DNMT3b gene. For visual confirmation, the
PCR products were separated electrophoretically on 1.5% agarose
gel. The PCR products were then cloned into the pCRII vector (TA
cloning kit; Invitrogen). The inserts were amplified by colony PCR
and sequenced with both vector primers: M13 forward, 5⬘-GTA
AAA CGA CGG CCA G-3⬘; M13 reverse, 5⬘-CAG GAA ACA
GCT ATG AC-3⬘.
Splice Variant-Specific Quantitative RT-PCR. Oligonucleotide primer
sets specific for DNMT3b3 and DNMT3b4 were designed. Each
primer spans the splice variant-specific exon-exon boundary: 3b3
forward, 5⬘-GAT GAA CAG GAT CT T TGG CTT T-3⬘(exon
20兾23); 3b3 reverse, 5⬘-GCC TGG CTG GAA CTA TTC ACA-3⬘
(exon 23); 3b4 forward, 5⬘-CGG GAT GAA CAG T TA AAG
AAA GTA C-3⬘(exon 20兾22); 3b4 reverse, 5⬘-CCA AAG ATC
CTT TCG AGC TC-3⬘(exon 22兾23). The PCRs were performed
with the SYBR Green PCR Core Reagents kit (Applied Biosys-
tems). Real-time detection of the emission intensity of SYBR
Green bound to double-stranded DNAs was performed with the
ABI PRISM 7700 Sequence Detection System. cDNAs derived
from the HCC cell line Alexander (29) were used as the calibrator
samples. Quantitative PCRs were performed in triplicate for each
sample-primer set, and the mean of the three experiments was used
as the relative quantification value. At the end of 40 PCR cycles, the
reaction products were separated electrophoretically on 3% aga-
rose gel to confirm that no nonspecific product was obtained at each
amplification.
Transfection with DNMT3b4 cDNA. Full-length human DNMT3b4
cDNA was prepared by RT-PCR of total RNA from the Alexander
cell line. The primer set (forward, 5⬘-CGC GGA TCC TGG AA A
GCA TGA AGG GAG ACA C-3⬘; reverse, 5⬘-GCT CTA GAA
CTG TTC ATC CCG GGT AGG TTG-3⬘) was designed such that
after a BamHI兾XbaI double digest the coding sequence of
DNMT3b4 could be ligated in-frame into the ex pression vector with
a myc tag at the C terminus (pcDNA3.1-myc; Invitrogen). To
confirm the fidelity of the PCR, the insert was fully sequenced. The
cloned cDNA was linearized and transfected into 293 cells—human
embryonic kidney cells transformed with adenovirus type 5 DNA
(30)—using FuGENE 6 Transfection Reagent (Roche Diagnos-
tics). After G418 selection of resistant cells, myc-tagged
DNMT3b4-expressing clones were chosen by Western blotting.
Mock-transfected 293 cells were generated by transfection with
vector alone.
Western Blotting. Total cell extracts were separated by SDS兾PAGE,
transferred electrophoretically onto poly(vinylidene dif luoride) fil-
Fig. 1. mRNA structure and protein structure of splice variants of human
DNMT3b. The four splice variants in the C-terminal catalytic domain (26) and the
numbers of exons and the exon-intron boundaries (28) have been described.
Table 1. The primer sets for PCR–single-strand conformation polymorphism to detect mutation of the DNMT3b gene
Target exons Primer sets Target exons Primer sets
Exon 2 5⬘-TCCCTGCTTCCCTTTCACC-3⬘(sense) Exon 13 5⬘-ACCCCAGGCTTTAGCAGCT-3⬘(sense)
5⬘-TTGTCTGCAGTGACCGCTC-3⬘(antisense) 5⬘-GGAGTTAGAGGAGGCAAGGG-3⬘(antisense)
Exon 3 5⬘-TTAGCAAGGCCGTTCCC-3⬘(sense) Exon 14 5⬘-CCCTTCTCTGGTCTCCGATT-3⬘(sense)
5⬘-CCACGTGATGAAAGCCAAAG-3⬘(antisense) 5⬘-GACTGCAGGAACGTAGGAGC-3⬘(antisense)
Exon 4 5⬘-TGACTTGCTGATACCCTGGG-3⬘(sense) Exon 15 5⬘-CAGGAGACCAGCTCTGACAAAG-3⬘(sense)
5⬘-GAGTGTTGGCCACAAGTGCT-3⬘(antisense) 5⬘-TTTCCAACACCCTGTGCC-3⬘(antisense)
Exon 5 5⬘-AGGCCTCCAGTCACCTAAGG-3⬘(sense) Exon 16 5⬘-GTCTTTGCCCTGTGCCTTC-3⬘(sense)
5⬘-TGCAGTGAGTGAGGCATATCTC-3⬘(antisense) 5⬘-CCTGGCTACCCTGTTGTGAC-3⬘(antisense)
Exon 6 5⬘-TTGCTCTGGCCCAAACTATG-3⬘(sense) Exon 17 5⬘-GAACTGTTCATTTTACCATAGCAGG-3⬘(sense)
5⬘-GGTCCTCAGTCACCCTGG-3⬘(antisense) 5⬘-GGGAAAAAGACAGGAAGAGATG-3⬘(antisense)
Exon 7 5⬘-CCTCTCCTCACTGGGATTTCTT-3⬘(sense) Exon 18 5⬘-TCTAGAAGTGGGTCCAGCTCTC-3⬘(sense)
5⬘-TTCAAAGGGAGGCAGGC-3⬘(antisense) 5⬘-AAGCAAGTGGCTCCTTCTCAG-3⬘(antisense)
Exon 8 5⬘-AGACATGGCACCTGGGACA-3⬘(sense) Exon 19 5⬘-GAACCTGCTGGTCTCAGGGA-3⬘(sense)
5⬘-TCTTGCTTCATCCCTGCTCT-3⬘(antisense) 5⬘-CCTGCCCTGGCTGTCTGT-3⬘(antisense)
Exon 9 5⬘-CACCCCCCCATTCATCA-3⬘(sense) Exon 20 5⬘-CTGCAAAGGTCTGGTTGACAC-3⬘(sense)
5⬘-GACTCTCCCAAGAAGTGGTCC-3⬘(antisense) 5⬘-CCAGGTCTTTCTAGGAAGGCTT-3⬘(antisense)
Exon 10 5⬘-TGGGTGACAGAGCAAGACC-3⬘(sense) Exon 21 5⬘-TGTGCCTAGCAGAGGACCCT-3⬘(sense)
5⬘-CAGAAGAAAGTGCATAGAAAACAGG-3⬘(antisense) 5⬘-AGTCCCCACTTGGAGGTCAC-3⬘(antisense)
Exon 11 5⬘-ACCCAGGCATAGCATGGTCT-3⬘(sense) Exon 22 5⬘-TTGAGCCTTGACTCCCCAG-3⬘(sense)
5⬘-TGATCTGCAGCGTCCTCCT-3⬘(antisense) 5⬘-GCCCATGTCTGCCCATTT-3⬘(antisense)
Exon 12 5⬘-GGAATTGATCTGTACCCGGC-3⬘(sense) Exon 23 5⬘-GGTTGAGGCTGTCAACATCC-3⬘(sense)
5⬘-TGGGTTAAACCCTGACAGGG-3⬘(antisense) 5⬘-ATCACACCTCCTGGGTCCTG-3⬘(antisense)
Saito et al. PNAS
兩
July 23, 2002
兩
vol. 99
兩
no. 15
兩
10061
MEDICAL SCIENCES
ters, and then incubated with anti-myc mAb 9E10 (Santa Cruz
Biotechnology).
Southern Blotting. High molecular weight DNA (5
g) was digested
for 24 h with 10 units of either MspIorHpaII, which cut at the
sequence CCGG, per
g of DNA. HpaII does not cut when the
internal cytosine is methylated. The DNA fragments were sepa-
rated by electrophoresis, transferred to nitrocellulose membranes,
and hybridized with
32
P-labeled oligonucleotide probe for satellite
2 (31). Signal intensities were measured with an image analyzer
(model BAS-2500; Fujifilm, Tokyo).
Statistics. Correlations between mRNA levels for DNMT3b splice
variants and DNA methylation status were analyzed with the
Kruskal–Wallis test. Differences in ratios for signal intensities
of Southern blotting were analyzed with the Mann–Whitney U
test. Differences with Pvalues of less than 0.05 were considered
significant.
Results
No Mutation of the
DNMT3b
Gene in HCCs. Fig. 2 shows examples of
PCR–single-strand conformation polymorphism analysis of the
DNMT3b gene in HCC cases. Because all of the shifted bands were
detected in both the noncancerous liver tissue and the HCC of a
particular case, they were considered to be polymorphisms. Se-
quencing revealed that the polymorphisms were located in introns
2 and 13 near the exon-intron boundaries (Table 2). No mutation
of any coding exon of the DNMT3b gene was detected in 59 HCCs.
DNMT3b3 and DNMT3b4 Are Major Splice Variants in HCCs. Fig. 3A
shows examples of RT-PCR products obtained with a primer set
that amplifies all four splice variants of DNMT3b in the C-terminal
catalytic domain. cDNA derived from total RNA of human testis,
in which significant expression of all four splice variants has been
reported (26), was used as control. At the end of 35 cycles of PCR
amplification, PCR products of about 230 bp were visible in almost
all samples (band d in Fig. 3A). PCR products of about 350 bp were
clearly visible in HCCs and Alexander cells, but control testis
contained little of this product (band b in Fig. 3A). RT-PCR
products of sample H35T1 were then cloned into the pCRII vector
and the inserts were amplified by colony PCR (Fig. 3B). Almost all
clones examined possessed inserts of about 230 bp (band d) or 350
bp (band b). Sequencing confirmed that the inserts of about 230 bp
(band d) and 350 bp (band b) corresponded to DNMT3b3, skipping
exons 21 and 22, and DNMT3b4, skipping exon 21, respectively (ref.
26, Figs. 1 and 3C).
Overexpression of DNMT3b4 Significantly Correlates with DNA Hy-
pomethylation on Pericentromeric Satellite Regions in Precancerous
Conditions and HCCs. To accurately quantify levels of mRNA for
DNMT3b3 and DNMT3b4, which were the major splice variants in
HCCs (Fig. 3), splice variant-specific quantitative RT–PCR was
performed in eight normal liver tissues from patients with liver
metastasis of primary colon cancer and in 49 noncancerous liver
tissues and 59 HCCs from 49 HCC patients.
We have previously reported the DNA methylation status on
pericentromeric satellite regions in this cohort (17). In the eight
normal liver tissues, satellites 2 and 3 were heavily methylated (17).
We then categorized the degree of DNA hypomethylation in
Southern blotting as ⫺,⫹,or2⫹:⫺indicates that the HpaII digest
showed the same hybridization pattern as normal liver tissues, ⫹
indicates that smaller fragments were detected in the HpaII digest
Fig. 2. Examples of PCR–single-strand conformation polymorphism analysis
of the DNMT3b gene in HCC cases with the intronic primer set encompassing
exon 13. N, noncancerous liver tissue; T, HCC. The shifted bands of H15N and
H15T (arrows) were sequenced and revealed to be a polymorphism in intron
13 (Table 2).
Table 2. Polymorphisms of the DNMT3b gene
Position Genotype
HCC patients
(n⫽49)
Patients with liver
metastasis of primary
colon cancer (n⫽8)
Intron 2
(22 bp upstream of the 3⬘splice site)
⫺兾⫺38 5
⫺兾c9 3
c兾c2 0
Intron 13
(16–18 bp downstream of the 5⬘splice site)
act兾act 48 8
act兾—10
Fig. 3. Examples of RT-PCR with a primer set that will amplify all four splice
variants in the C-terminal catalytic domain of DNMT3b in HCCs (H) and Alexander
cells. (A) Alexander cells were derived from human HCC (29) and showed DNA
hypomethylation on pericentromeric satellite regions (data not shown). PCR
products of about 420 bp (a), 350 bp (b), 300 bp (c), and 230 bp (d) will correspond
to DNMT3b1, DNMT3b4, DNMT3b5, and DNMT3b3, respectively. (B) RT-PCR
products of H35T1 were cloned into the pCRII vector, and the inserts were
examined by colony PCR. Almost all clones examined possessed inserts of about
350 bp (b) and 230 bp (d). (C) Sequencing confirmed that the insert of about 350
bp (b) corresponded to DNMT3b4, which lacks the conserved methyltransferase
motifs IX and X (Fig. 1) because of the premature stop codon (underlined) arising
from skipping of exon 21.
10062
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.152121799 Saito et al.
compared with normal liver tissues, and 2⫹indicates that the HpaII
digest showed the same hybridization pattern as the MspIdigestof
the same sample and normal liver tissues (17). Nine (18%) of 49
noncancerous liver tissues from HCC patients and 39 (66%) of 59
HCCs showed ⫹or 2⫹DNA hypomethylation (17).
Fig. 4Ashows mRNA expression levels for DNMT3b3 normal-
ized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
mRNA. The average levels of mRNA for DNMT3b3, normalized
to GAPDH mRNA, in tissue samples from HCC cases with ⫺,⫹,
and 2⫹DNA hypomethylation were 0.72 ⫾1.20, 0.87 ⫾0.85, and
0.81 ⫾1.01 (mean ⫾SD), respectively. There was no significant
correlation between mRNA levels for DNMT3b3 and DNA meth-
ylation status on pericentromeric satellite regions.
Fig. 4Bshows mRNA expression levels for DNMT3b4 normal-
ized to GAPDH mRNA. The average levels of mRNA for
DNMT3b4, normalized to GAPDH mRNA, in tissue samples from
HCC cases with ⫺,⫹, and 2⫹DNA hypomethylation were 0.31 ⫾
0.23, 0.78 ⫾0.79, and 0.83 ⫾0.94, respectively. There was a
Fig. 4. Splice variant-specific quantitative RT–PCR for
DNMT3b in HCC cases (H) and cases with liver metastasis of
primary colon cancer (C).N, noncancerous liver tissue; T,
HCC. (A) mRNA levels for DNMT3b3 normalized to GAPDH
mRNA. (B) mRNA levels for DNMT3b4 normalized to
GAPDH mRNA. (C) The ratio of DNMT3b4 mRNA to
DNMT3b3 mRNA. We have previously reported DNA meth-
ylation status on pericentromeric satellite regions for these
cases (17). In normal liver tissues (C1 to C8), satellites 2 and
3 were heavily methylated (17). The samples indicated by
white, hatched, and black bars show no (⫺), moderate (⫹),
and strong (2⫹) DNA hypomethylation, respectively, as
defined in ref. 17. mRNA levels for DNMT3b4 normalized to
GAPDH mRNA (B,P⫽0.0001, Kruskal–Wallis test) and the
ratios of DNMT3b4 mRNA to DNMT3b3 mRNA (C,P⫽
0.009, Kruskal–Wallis test) each were significantly corre-
lated with DNA methylation status on pericentromeric sat-
ellite regions.
Saito et al. PNAS
兩
July 23, 2002
兩
vol. 99
兩
no. 15
兩
10063
MEDICAL SCIENCES
significant correlation between mRNA levels for DNMT3b4 and
DNA methylation status on pericentromeric satellite regions (P⫽
0.0001, Kruskal–Wallis test). Fig. 4Cshows the ratio of DNMT3b4
mRNA to DNMT3b3 mRNA. The averages of the ratios of
DNMT3b4 mRNA to DNMT3b3 mRNA in tissue samples from
HCC cases with ⫺,⫹, and 2⫹DNA hypomethylation were 0.71 ⫾
0.52, 1.40 ⫾1.40, and 1.65 ⫾1.70, respectively. There was a
significant correlation between the ratio of DNMT3b4 mRNA to
DNMT3b3 mRNA and DNA methylation status on pericentro-
meric satellite regions (P⫽0.009, Kruskal–Wallis test).
Expression levels of mRNA for DNMT3b1 and DNMT3b5 in all
tissue samples were too low to be detected by this quantitative
RT–PCR system.
Transfection of Human Epithelial 293 Cells with DNMT3b4 cDNA
Induces DNA Demethylation on Satellite 2. Five clones expressing
myc-tagged DNMT3b4, 3b4–1to3b4–5, were obtained. The ex-
pression of myc-tagged DNMT3b4 was weak in 3b4–1, whereas it
was particularly strong in 3b4–5 (Fig. 5A). About 50 days after
transfection, DNA methylation status on satellite 2 was analyzed by
Southern blotting. As shown in Fig. 5B, in 293 parent cells (indi-
vidually cultured parent-1 and parent-2), mock transfectants
(mock-1 to mock-3), and the DNMT3b4-transfected 3b4–1 clone,
which only weakly expresses myc-tagged DNMT3b4, larger DNA
fragments were detected in the HpaII digest compared with the
MspI digest. On the other hand, in clones 3b4–2to3b4–5, which
strongly express myc-tagged DNMT3b4, smaller fragments were
detected in the HpaII digest compared with parent cells, mock
transfectants, and clone 3b4–1. The signal intensity of the 6.6-kbp
HpaII digest was strong in all parent cells, mock transfectants, and
DNMT3b4 transfectants. The ratios of the signal intensity of the
2.0-kbp HpaII digest to that of the 6.6-kbp HpaII digest in clones
3b4–2to3b4–5 were significantly higher than those in parent cells,
mock transfectants, and clone 3b4–1 (Fig. 5B,P⬍0.0001, Mann–
Whitney Utest). Moreover, as shown in Fig. 5, there was a
correlation between expression level of myc-tagged DNMT3b4 and
the ratio of signal intensity of the 2.0-kbp HpaII digest to that of the
6.6-kbp HpaII digest among all DNMT3b4 transfectants. These
results suggest that overexpression of DNMT3b4 induces DNA
demethylation on satellite 2 in human epithelial cells.
Discussion
We have reported DNA hypomethylation on pericentromeric
satellite regions, DNA hypermethylation on CpG islands of
genes such as p16,E-cadherin, and HIC-1 (hypermethylated-in-
cancer), overexpression of DNA methyltransferases, and re-
duced expression of methyl-CpG-binding proteins during hepa-
tocarcinogenesis (12–17). Among these findings, DNA
hypomethylation on pericentromeric satellite regions was ob-
served even in precancerous conditions and appears to be one of
the earliest events during hepatocarcinogenesis. To prevent the
development of HCCs in hepatitis virus carriers suffering from
chronic hepatitis or cirrhosis, clarification of the molecular
mechanisms underlying early events during hepatocarcinogen-
esis is needed. We have focused on the mechanism of DNA
hypomethylation on pericentromeric satellite regions.
DNMT3b is required for de novo DNA methylation on pericen-
tromeric satellite regions in embryonic stem cells and during mouse
development (22). In adult cells, the role of DNMT3b remains
unclear and it is unknown how DNA methylation on pericentro-
meric satellite regions is maintained. DNMT1 has so far been
recognized as the ‘‘maintenance’’ DNA methyltransferase. Once
established by DNMT3b during development, DNA methylation on
pericentromeric satellite regions is likely to be maintained by
DNMT1 in adult cells. In cancer cells, however, targeting of
DNMT1 to substrate DNA may be disrupted by mechanisms such
as dysfunction of p21
WAF1
(32), which competes with DNMT1 for
binding to proliferating cell nuclear antigen (33). In fact, in human
cancers, overexpression of DNMT1 is significantly correlated with
CpG island methylation phenotype, which is defined by frequent
DNA hypermethylation on CpG islands that are not methylated in
normal cells (27). If DNMT1 does not maintain DNA methylation
on pericentromeric satellite regions because of disturbance of its
targeting in cancer cells, DNMT3b must rescue DNA methylation
on these regions. Moreover, some recent studies have proposed that
all active DNA methyltransferases, DNMT1 and members of the
DNMT3 family, probably possess both de novo and maintenance
DNA methyltransferase activity in vivo, regardless of their prefer-
ence for hemimethylated or unmethylated substrates in vitro (4,
34–36). Therefore, it is likely that DNMT3b is required to maintain
DNA methylation on pericentromeric satellite regions even in
somatic cells, including cancer cells. Therefore, we focused on
alterations of DNMT3b during hepatocarcinogenesis.
When analyzed previously by quantitative RT-PCR w ith a primer
set that cannot discriminate between splice variants of DNMT3b,
expression of total mRNA for DNMT3b did not correlate with
DNA hypomethylation on pericentromeric satellite regions in HCC
cases (17). Thus, it is unlikely that reduced expression of DNMT3b
simply causes DNA hypomethylation on these regions during
hepatocarcinogenesis. Although germ-line mutations of the
DNMT3b gene have been reported in patients with immunodefi-
ciency兾centromeric instability兾facial anomalies syndrome (23, 24),
in HCCs we detected no somatic mutations of the DNMT3b gene
in any coding exons.
There have been a small number of studies on expression of the
four splice variants in the C-terminal catalytic domain of DNMT3b
in vivo. DNMT3b3 possesses the N-terminal region and conserved
methyltransferase motifs I, IV, VI, IX, and X, but lacks 63 aa
residues between motifs VI and IX. Because mouse DNMT3b3 did
not methylate an applied unmethylated substrate in an in vitro study
(37), DNA methyltransferase activity of human DNMT3b3 should
also be carefully discussed. However, it has been reported that
Fig. 5. Transfection of human epithelial 293 cells with DNMT3b4 cDNA. (A)
Western blotting with monoclonal anti-myc antibody. Band a, myc-tagged
DNMT3b4; band b, endogenous c-myc. Among DNMT3b4 transfectants, expres-
sion of myc-tagged DNMT3b4 was weak in 3b4–1, whereas it was particularly
strong in 3b4–5. About 50 days after transfection, DNA methylation status on
satellite 2 was analyzed by Southern blotting (B). In parent cells (individually
cultured parent-1 and parent-2), mock transfectants, and 3b4–1, larger DNA
fragments were detected in the HpaII (H) digest compared with the MspI (M)
digest. In clones 3b4 –2to3b4–5, smaller fragments were detected in the H digest
compared with parent-1 to parent-2, mock-1 to mock-3, and 3b4 –1.The ratios of
signal intensity of the 2.0-kbp H digest (arrow) to that of the 6.6-kbp H digest
(arrowhead), whose intensity was strong in all parent cells, mock transfectants,
and DNMT3b4 transfectants, are shown at the bottom (P⬍0.0001, Mann–
Whitney Utest).
10064
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.152121799 Saito et al.
DNMT3b3 is ubiquitously expressed in normal human tissues (26).
Our data also indicate that the major variant in normal liver tissues
is DNMT3b3. Therefore, we cannot rule out the possibility that
DNMT3b3 carries the DNMT3b activity at least in vivo, for example
in human liver tissue. DNMT3b4 probably does not show methyl-
transferase activity because it lacks the conserved methyltrans-
ferase motifs IX and X, although it retains the N-terminal domain
required for targeting to heterochromatin sites through binding to
RP58 (38, 39). In vivo, it has been reported that normal human
tissues, with the exception of testis, do not express significant levels
of DNMT3b4 (26). Here, we confirmed the trace level of
DNMT3b4 expression in normal liver tissues.
Overexpression of DNMT3b4 and elevation of the ratio of
DNMT3b4 mRNA to DNMT3b3 mRNA were both significantly
correlated with the degree of DNA hypomethylation on pericen-
tromeric satellite regions in precancerous conditions and HCCs.
DNMT3b4 may compete with the major variant, DNMT3b3, for
targeting to pericentromeric satellite regions. This may be the
reason overexpression of DNMT3b4, especially in relation to the
expression of DNMT3b3, results in DNA hypomethylation on
pericentromeric satellite regions in precancerous conditions and
HCCs. To confirm this possibility, we introduced DNMT3b4 into
human epithelial 293 cells, which express a significant level of
DNMT3b3 mRNA and a trace of endogenous DNMT3b4 mRNA
(data not shown). DNA demethylation on satellite 2 was observed
in DNMT3b4 transfectants, depending on the expression level of
myc-tagged DNMT3b4. Although the overexpression levels in
DNMT3b4 transfectants may exceed the levels of DNMT3b4 in
tissue samples from HCC patients, the results of the transfection
studies support the possibility that overexpression of DNMT3b4
induces DNA demethylation on pericentromeric satellite regions in
human epithelial cells. On the other hand, a few exceptional tissue
samples from HCC patients without overexpression of DNMT3b4
or an elevated ratio of DNMT3b4 mRNA to DNMT3b3 mRNA
also showed DNA hypomethylation on pericentromeric satellite
regions. Therefore, overexpression of DNMT3b4 should not be
considered as the only mechanism for DNA hypomethylation on
pericentromeric satellite regions.
We and other groups have shown that chromosomal instability,
e.g., allelic imbalance on chromosomes 1 and 16, accumulates even
in noncancerous liver tissues showing chronic hepatitis and cirrho-
sis, and is more frequent in HCCs (15, 16, 40). Satellite 2 is the major
sequence of the heterochromatin region adjacent to the centro-
meres of chromosomes 1 and 16. In fact, frequent 1q copy gain with
a breakpoint in heterochromatin DNA was reported in HCCs with
DNA hypomethylation on satellite 2 (20). Overexpression of
DNMT3b4 may lead to chromosomal instability through induction
of DNA hypomethylation on pericentromeric satellite regions, even
in the precancerous stages of HCCs. On the other hand, it was
recently shown that CpG islands and genes are relatively commonly
located in heterochromatin regions (41). The growth rate of
DNMT3b4 transfectants became about double that of mock trans-
fectants soon after introduction of DNMT3b4, when allelic imbal-
ance may not yet have accumulated (unpublished data). This
growth change may be caused by alterations in the expression of
genes in regions in which DNA methylation status is affected by
DNMT3b activity. In fact, the expression levels of several genes that
potentially participate in signal transduction pathways and兾or cell
growth were altered in the DNMT3b4 transfectants (unpublished
data).
We conclude, therefore, that overexpression of DNMT3b4 in-
duces DNA demethylation on pericentromeric satellite regions
even in precancerous stages and may play critical roles in the
development of HCC through chromosomal instability and aber-
rant expression of cancer-related genes. It would be of further
interest to investigate whether aberrant expression of DNMT3b
splice variants is of general significance during carcinogenesis, even
in organs other than the liver.
This study was supported by a Grant-in-Aid for the Second Term
Comprehensive 10-Year Strategy for Cancer Control and a Grant-in-Aid
for Cancer Research from the Ministry of Health, Labor, and Welfare
of Japan. Y.S. is a recipient of a Research Resident Fellowship from the
Foundation for Promotion of Cancer Research in Japan.
1. Jones, P. A. & Laird, P. W. (1999) Nat. Genet. 21, 163–167.
2. Jones, P. A. & Takai, D. (2001) Science 293, 1068–1070.
3. Baylin, S. B., Esteller, M., Rountree, M. R., Bachman, K. E., Schuebel, K. & Herman, J. G.
(2001) Hum. Mol. Genet. 10, 687–692.
4. Robertson, K. D. (2001) Oncogene 20, 3139–3155.
5. Gama-Sosa, M. A., Slagel, V. A., Trewyn, R. W., Oxenhandler, R., Kuo, K. C., Gehrke, C. W.
& Ehrlich, M. (1983) Nucleic Acids Res. 11, 6883–6894.
6. Vachtenheim, J., Horakova, I. & Novotna, H. (1994) Cancer Res. 54, 1145–1148.
7. Yoshiura, K., Kanai, Y., Ochiai, A., Shimoyama, Y., Sugimura, T. & Hirohashi, S. (1995) Proc.
Natl. Acad. Sci. USA 92, 7416 –7419.
8. Eguchi, K., Kanai, Y., Kobayashi, K. & Hirohashi, S. (1997) Cancer Res. 57, 4913–4915.
9. Tsuda, H., Hirohashi, S., Shimosato, Y., Terada, M. & Hasegawa, H. (1988) Gastroenterology 95,
1664–1666.
10. Yasui, H., Hino, O., Ohtake, K., Machinami, R. & Kitagawa, T. (1992) Cancer Res. 52,
6810–6814.
11. Kanai, Y., Ushijima, S., Tsuda, H., Sakamoto, M., Sugimura, T. & Hirohashi, S. (1996) Jpn. J.
Cancer Res. 87, 1210–1217.
12. Kanai, Y., Ushijima, S., Hui, A. M., Ochiai, A., Tsuda, H., Sakamoto, M. & Hirohashi, S. (1997)
Int. J. Cancer 71, 355–359.
13. Sun, L., Hui, A. M., Kanai, Y., Sakamoto, M. & Hirohashi, S. (1997) Jpn. J. Cancer Res. 88,
1165–1170.
14. Kanai, Y., Hui, A. M., Sun, L., Ushijima, S., Sakamoto, M., Tsuda, H. & Hirohashi, S. (1999)
Hepatology 29, 703–709.
15. Kondo,Y., Kanai, Y., Sak amoto, M., Mizokami, M., Ueda, R. & Hirohashi, S. (2000) Hepatology
32, 970–979.
16. Kanai,Y., Ushijima, S., Tsuda, H., Sak amoto, M. & Hirohashi, S. (2000) Cancer Lett. 148, 73–80.
17. Saito, Y., Kanai, Y., Sakamoto, M., Saito, H., Ishii, H. & Hirohashi, S. (2001) Hepatology 33,
561–568.
18. Kokalj-Vokac, N., Almeida, A., Viegas-Pequignot, E., Jeanpierre, M., Malfoy, B. & Dutrillaux,
B. (1993) Cytogenet. Cell Genet. 63, 11–15.
19. Jeanpierre, M., Turleau, C., Aurias, A., Prieur, M., Ledeist, F., Fischer, A. & Viegas-Pequignot,
E. (1993) Hum. Mol. Genet. 2, 731–735.
20. Wong, N., Lam, W. C., Lai, P. B., Pang, E., Lau, W. Y. & Johnson, P. J. (2001) Am. J. Pathol.
159, 465–471.
21. Okano, M., Xie, S. & Li, E. (1998) Nat. Genet. 19, 219–220.
22. Okano, M., Bell, D. W., Haber, D. A. & Li, E. (1999) Cell 99, 247–257.
23. Hansen, R. S., Wijmenga, C., Luo, P., Stanek, A. M., Canfield, T. K., Weemaes, C. M. & Gartler,
S. M. (1999) Proc. Natl. Acad. Sci. USA 96, 14412–14417.
24. Xu, G. L., Bestor, T. H., Bourc’his,D., Hsieh, C. L., Tommer up, N., Bugge, M., Hulten, M., Qu,
X., Russo, J. J. & Viegas-Pequignot, E. (1999) Nature (London) 402, 187–191.
25. Maraschio, P., Zuffardi, O., Dalla Fior, T. & Tiepolo, L. (1988) J. Med. Genet. 25, 173–180.
26. Robertson, K. D., Uzvolgyi, E., Liang, G., Talmadge, C., Sumegi, J., Gonzales, F. A. & Jones,
P. A. (1999) Nucleic Acids Res. 27, 2291–2298.
27. Kanai, Y., Ushijima, S., Kondo, Y., Nak anishi, Y. & Hirohashi, S. (2001) Int. J. Cancer 91,
205–212.
28. Xie, S., Wang, Z., Okano, M., Nogami, M., Li, Y., He, W. W., Okumura, K. & Li, E. (1999) Gene
236, 87–95.
29. Alexander, J. J., Bey, E. M., Geddes, E. W. & Lecatsas, G. (1976) S. Afr. Med. J. 50, 2124–2128.
30. Graham, F. L., Smiley, J., Russell, W. C. & Nairn, R. (1977) J. Gen. Virol. 36, 59–74.
31. Tagarro, I., Fernandez-Peralta, A. M. & Gonzalez-Aguilera, J. J. (1994) Hum. Genet. 93,
383–388.
32. Baylin, S. B. (1997) Science 277, 1948–1949.
33. Chuang,L. S., Ian, H. I., Koh, T. W., Ng, H. H., Xu, G. & Li, B. F. (1997) Science 277, 1996 –2000.
34. Vertino, P. M., Yen, R. W., Gao, J. & Baylin, S. B. (1996) Mol. Cell. Biol. 16, 4555–4565.
35. Rhee, I., Jair, K. W., Yen, R. W., Lengauer, C., Herman, J. G., Kinzler, K. W., Vogelstein, B.,
Baylin, S. B. & Schuebel, K. E. (2000) Nature (London) 404, 1003–1007.
36. Liang, G., Chan, M. F., Tomigahara, Y., Tsai, Y. C., Gonzales,F. A., Li, E., L aird, P. W. & Jones,
P. A. (2002) Mol. Cell. Biol. 22, 480 –491.
37. Aoki, A., Suetake, I., Miyagawa, J., Fujio, T., Chijiwa, T., Sasaki, H. & Tajima, S. (2001) Nucleic
Acids Res. 29, 3506–3512.
38. Bachman, K. E., Rountree, M. R. & Baylin, S. B. (2001) J. Biol. Chem. 276, 32282–32287.
39. Fuks, F., Burgers, W. A., Godin, N., Kasai, M. & Kouzarides, T. (2001) EMBO J. 20, 2536–2544.
40. Kishimoto, Y., Shiota, G., Wada, K., Kitano, M., Nakamoto, K., Kamisaki, Y., Suou, T., Itoh,
T. & Kawasaki, H. (1996) J. Cancer Res. Clin. Oncol. 122, 585–589.
41. Barry, A. E., Howman, E. V., Cancilla, M. R., Saffery, R. & Choo, K. H. (1999) Hum. Mol. Genet .
8, 217–227.
Saito et al. PNAS
兩
July 23, 2002
兩
vol. 99
兩
no. 15
兩
10065
MEDICAL SCIENCES
