Leukemia (2008) 22, 1529–1538
& 2008 Macmillan Publishers Limited All rights reserved 0887-6924/08 $30.00
www.nature.com/leu
ORIGINAL ARTICLE
Genome-wide identification of aberrantly methylated promoter associated CpG islands
in acute lymphocytic leukemia
S-Q Kuang, W-G Tong, H Yang, W Lin, MK Lee, ZH Fang, Y Wei, J Jelinek, J-P Issa and G Garcia-Manero
Department of Leukemia, University of Texas MD Anderson Cancer Center, Houston, TX, USA
We performed a genome-wide analysis of promoter associated
CpG island methylation using methylated CpG island amplification (MCA) coupled to representational differential analysis
(RDA) or a DNA promoter microarray in acute lymphoblastic
leukemia (ALL). We identified 65 potential targets of methylation with the MCA/RDA approach, and 404 with the MCA/array.
Thirty-six (77%) of the genes identified by MCA/RDA were
shared by the MCA/array approach. Chromosomal location of
these genes was evenly distributed in all autosomes. Functionally, 303 of these genes clustered in 18 molecular pathways. Of
the 36 shared genes, 31 were validated and 26 were confirmed
as being hypermethylated in leukemia cell lines. Expression
analysis of eight of these genes was epigenetically modulated
by hypomethylating agents and/or HDAC inhibitors in leukemia
cell lines. Subsequently, DNA methylation of 15 of these genes
(GIPC2, RSPO1, MAGI1, CAST1, ADCY5, HSPA4L, OCLN,
EFNA5, MSX2, GFPT2, GNA14, SALL1, MYO5B, ZNF382 and
MN1) was validated in primary ALL samples. Patients with
methylation of multiple CpG islands had a worse overall
survival. This is the largest published list of potential methylation target genes in human leukemia offering the possibility of
performing rational unbiased methylation studies in ALL.
Leukemia (2008) 22, 1529–1538; doi:10.1038/leu.2008.130;
published online 5 June 2008
Keywords: acute lymphoblastic leukemia; DNA methylation;
microarray; epigenetics; MCA; RDA
Introduction
Acute lymphoblastic leukemia (ALL) is characterized by a block
in differentiation of early lymphoid progenitors, which leads to
an accumulation of immature lymphoblasts.1 Together with
genetic changes, epigenetic alterations such as aberrant DNA
methylation of promoter associated CpG islands (CGIs) have a
role in downregulation of tumor suppressor genes and in the
molecular pathogenesis of ALL.2 Prior data from several
laboratories have indicated that aberrant DNA methylation of
multiple CGIs is common in ALL3,4 and is associated with poor
prognosis.5–7 Furthermore, we have demonstrated that methylation patterns at relapse are stable in a majority of patients with
ALL,8 and that the identification of residual methylation levels is
associated with a higher risk of relapse in patients in
morphological remission (H Yang, submitted). We have also
demonstrated that the reactivation of an epigenetically silenced
gene, such as p57KIP2, results in selective induction of apoptosis
only in cells in which the gene is methylated.2 This body of
Correspondence: Dr G Garcia-Manero, Department of Leukemia,
University of Texas M.D. Anderson Cancer Center, Box 428, 1515
Holcombe Blvd, Houston, TX 77030, USA.
E-mail: [email protected]
Received 21 November 2007; revised 23 April 2008; accepted 24
April 2008; published online 5 June 2008
information indicates that epigenetic silencing of tumor
suppressor genes is important in ALL.
Until recently, one of the limitations of studying DNA
methylation has been the process to select gene(s) to study. It
is possible that several hundred genes could be aberrantly
methylated and silenced in ALL.9 Some of these genes may be
bona fide tumor suppressor genes, whose silencing could be
essential to the pathogenesis of ALL, whereas others may be
silenced as a consequence of global methylation dysregulation
and not directly involved in oncogenesis.10 Therefore, an
unbiased whole genome approach to identify methylation
silenced genes in ALL may reveal novel tumor suppressor genes
and enhance our understanding of the role of aberrant gene
methylation in this disease. To approach this, we have
performed two large-scale genome-wide studies of aberrantly
methylated CpG islands in ALL.
Materials and methods
Cell lines and ALL patient samples
The following human leukemia cell lines were studied: of
lymphoid origin; MOLT4, Jurkat, Peer, T-ALL1, CEM, J-TAG, BJAB, RS4, ALL1, Raji, REH and Ramos. Of myeloid origin: K562
and BV173, HL60, NB4, THP1, U937, ML1, OCI, HEL,
MOLM13 and KBM5R. All were obtained from the American
Type Culture Collection and were cultured in RPMI 1640
(Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal
calf serum (Gemini Bio-Products, West Sacramento, CA, USA)
and penicillin–streptomycin (Invitrogen). Cell suspensions from
bone marrow aspiration specimens from patients with ALL were
obtained prior to any therapy and were stored at established
tissue banks at MD Anderson Cancer Center (MDACC)
following institutional guidelines. All samples were collected
using Ficoll-Paque density centrifugation. DNA was extracted
from leukemia cell lines and ALL samples using standard
phenol-chloroform methods.
Methylated CpG island amplification/representational
difference analysis
The principle underlying methylated CpG island amplification
(MCA) involves amplification of closely spaced methylated SmaI
sites (CCCGGG) to enrich for methylated CGIs.11 About 70% of
CGIs contain two SmaI sites spaced closely enough (200–1000
bases) to be amplified efficiently by PCR. Representational
difference analysis (RDA) is a subtraction hybridization step
used to identify CGIs methylated specifically in tester DNA. This
procedure was performed as described previously.11 MCA
amplicons were generated from a pool of DNA from three
patients with ALL (testers). Of importance, the three ALL tester
patients were selected because they were refractory to induction
DNA Methylation in ALL
S-Q Kuang et al
1530
therapy with hyperCVAD chemotherapy.12 These three patients
were diploid (Mixed Lineage Leukemia (MLL) and Philadelphia
(Ph) chromosome negative). One was a male and two females.
Their median age was 33 years (range 27–66). Therefore, this
represents a group of patients with very poor prognosis not
currently identifiable with current molecular techniques. As
control, we used MCA amplicons generated from a pool of DNA
from two healthy donors (drivers). These controls were one male
and one female, and their ages were 30 and 55 years (median
42.5), respectively (P ¼ 0.48). Briefly, 5 mg of genomic DNA was
digested with 100 units of SmaI (New England Biolabs, Ipswich,
MA, USA) twice and then further digested with 20 units of XmaI
(New England Biolabs). RMCA adaptor was prepared by
annealing of oligonucleotides RMCA 24 and RMCA 12, and
ligated to the digested DNA fragments using T4 DNA ligase
(New England Biolabs). To amplify hypermethylated DNA
fragments, PCR was performed using 3 ml of each of the ligation
mixture as a template in a 100 ml volume containing 100 pmol
RMCA 24 primer, 5 units of Taq DNA polymerase (Life
Technologies, Inc., Carlsbad, CA, USA), 4 mM MgCl2, 16 mM
of NH4 (SO4)2, 10 mg/ml of bovive serum albumin, and 5% v/v
dimethylsulphoxide. The reaction mixture was incubated at
72 1C for 5 min and at 95 1C for 3 min, and then was subjected to
25 cycles of 1 min at 95 1C and 3 min at 77 1C followed by a
final extension of 10 min at 77 1C.
Representational difference analysis was performed on the
test and driver MCA amplicons as described previously.11
Primers used for the first and second rounds of RDA were
JMCA24/JMCA12, and NMCA24/NMCA12. After the second
round of hybridization and selective amplification, the MCA/
RDA products were cloned into a pCR 4.0-TOPO vector
(Invitrogen). To screen for inserts, a total of 800 clones were
sequenced at a core facility (MDACC). Sequence homologies
and chromosomal localization were identified using the Human
Blat program (http://www.genome.ucsc.edu). A list of primers
used in this study is shown in Supplementary Table 1.
MCA/human proximal promoter DNA microarray
The MCA amplicons generated above were labeled with Cy5dUTP (tester) or Cy3-dUTP (driver) according to the manufacturer’s protocol (Invitrogen). Five micrograms of Cy5- and Cy3labeled MCA amplicons were mixed and hybridized to the
Agilent human proximal promoter microarray slides. These
contain approximately 17 000 of the best defined human
transcript associated promoters covering 1.0 kb upstream and
þ 0.3 kb downstream from the transcription start site (Agilent
Technologies, Santa Clara, CA, USA). The hybridization
reaction was carried out in rotisserie hybridization oven at
65 1C, 10 r.p.m. for 40 h and subsequently washed following the
Agilent aCGH array hybridization protocol. Hybridized slides
were scanned using an Agilent G2565AA scanner (Agilent
Technologies). The data from scanned microarray images was
extracted using Agilent feature extraction software. Image
analysis was done using the Agilent G4477AA Chip Analytics
1.2 software package. Genes corresponding to each promoter
DNA spotted on the array were identified and then selected by
the following criteria: a tester’s Cy5 signal intensity of more than
2000 and a Cy5 signal intensity ratio of tester versus driver of
more than 2. Promoter specific CGI methylation was further
determined using an in silico gene-filtering search method.
Genes containing CGI in their promoter regions based on results
from USCS Human Blat analysis (http://www.genome.ucsc.edu)
and having two putative SmaI cutting sites in their CGIs (putative
MCA targets) were selected for further validation.
Leukemia
DNA bisulfite treatment and bisulfite pyrosequencing
Bisulfite treatment of genomic DNA was performed as
described.13 Please refer to Supplementary Information for
further details.
Bisulfite sequencing
Bisulfite sequencing was performed to confirm the pyrosequencing results in selected samples. Please refer to Supplementary
Information for further details.
RNA preparation and real-time PCR
Methods are described in Supplementary information.
5-aza-20 -deoxycytidine and/or trichostatin A treatment
To study the effect of epigenetic modulation, leukemia cell lines
were cultured in media supplemented with 5 mmol/l of 5-aza-20 deoxycytidine (Sigma, St Louis, MO, USA) for 4 days, 5 mmol/l of
5-aza-20 -deoxycitidine for 4 days and then 500 nmol/l trichostatin A (TSA) (ICN Biomedicals, Costa Mesa, CA, USA) for the
last 24 h or 500 nmol/l TSA for 24 h alone. Cells were split 12–
24 h before treatment. Mock-treated cells were cultured
similarly.
Methylation loci mapping and molecular pathway
analysis
The MapChart program (http://www.biometris.wur.nl/UK/Software/MapChart/download) was used to generate a chromosomal map of methylation loci. The ingenuity pathway program
(http://analysis.ingenuity.com) was used to cluster gene function
and signaling pathway networks.
Statistical analysis
Statistical analyses were performed using Prism 4 (GraphPad
Software, Inc.) or Statistica 6 software (Statistica for Windows
version 6.0, StatSoft, Tulsa, OK, USA). The Fisher’s exact test
and t-tests were used to compare gene methylation frequencies
or expression levels in leukemia cell lines or ALL patients and
normal control groups. The Spearman non-parametric test was
used to determine correlations. Effect on survival was analyzed
as previously described.5 All reported P values were two-sided,
and Po0.05 was considered statistically significant.
Results
Global identification of hypermethylated genes in Ph
negative ALL using MCA/RDA and MCA/microarray
techniques
First, we used the MCA/RDA approach to identify aberrantly
methylated CGIs in ALL. We recovered and sequenced 800
clones. Chromosomal locations, gene homology and CGI
information of the sequenced clones were analyzed using the
Human Blat program (http://www.genome.ucsc.edu). Of the
800 clones, 217 (27%) contained CGIs, 74 (9.3%) were
localized within or close to (o200 bp) the promoter region of
identifiable genes and 62 of these 74 clones were matched to
known human gene sequences (Supplementary Table 2). Of
note, five of these clones contained bidirectional promoters.11
These corresponded to either two known genes, or to one
known gene, one mRNA/or spliced EST. In total, therefore, 65
genes were identified.
Table 1
Methylation profile of leukemia cell lines and normal controls
DNA Methylation in ALL
S-Q Kuang et al
Bisulfite pyrosequencing was carried out to determine the methylation status of selected genes in 23 leukemia cell lines and 10 normal peripheral lymphocytic cells (control). Each column represents a
cell line number, and each row indicates a gene identified by both methylated CpG island amplification/representational difference analysis (MCA/RDA) and MCA/DNA promoter microarray. Green box,
methylation density o15%; yellow box, methylation density between 15–29.99%; pink grey box, methylation density 30–59.99%; red box, methylation density 460%. Methylation density 415% was
used as the cutoff for hypermethylation. Methylation frequency was counted as the percentage of positive methylation numbers versus total numbers studied for each gene. P-values were calculated
using a t-test comparing cell line methylatyion versus normal controls.
1531
Leukemia
DNA Methylation in ALL
S-Q Kuang et al
1532
Validation of methylation target genes in leukemia cell
lines
Due to limited yield of this approach, we performed a second
genome-wide DNA methylation search using the MCA/DNA
promoter microarray method. This analysis revealed that 923
(5.4%) of the 17000 promoter specific transcripts featured on the
Agilent chip were expected to be hypermethylated in Ph
negative ALL samples. We further reduced the number of
candidate genes by excluding unknown genes (spliced EST,
mRNA and hypothetical proteins), sex chromosomal genes and
genes whose putative function were not related to oncogenesis,
and focused on 404 autosomal genes (Supplementary Table 3).
We then compared the genes identified by the two techniques:
MCA/RDA and MCA/microarray. Of the 65 known genes
identified by MCA/RDA, only 47 were represented in the array,
36 of them (77%) were found to be hypermethylated by the
MCA/microarray approach. Information on these genes is shown
in Supplementary Tables 2 and 3.
75
50
25
75
50
25
0
0
60
40
20
0
50
25
40
20
25
GNA14
50
25
100
MYO5B
75
50
25
0
100
80
ZNF38
2
60
40
20
0
Controls Patients Cell lines
N=7
N=36
N=23
50
25
Controls Patients Cell lines
N=10 N=41
N=23
Patients Cell lines
N=44
N=23
100
Methylation (%)
N=23
SALL1
75
0
Controls
N=9
Methylation (%)
N=35
25
Controls Patients Cell lines
N=11
N=55 N=23
0
N=9
MSX2
50
N=22
Methylation (%)
N=58
75
Controls Patients Cell lines
N=23
0
N=11
Methylation (%)
Methylation (%)
25
N=35
75
Controls Patients Cell lines
0
Methylation (%)
EFNA5
50
100
50
100
N=13
100
75
Controls Patients Cell lines
N=7
N=41
N=23
75
25
Controls Patients Cell lines
0
0
GFPT
50
0
Methylation (%)
Methylation (%)
Methylation (%)
OCLN
HSPA4L
75
Controls Patients Cell lines
N=8
N=56 N=23
100
60
100
100
0
80
25
Controls Patients Cell lines
N=8
N=45 N=23
ADCY5
75
Controls Patients Cell lines
N=57 N=23
N=11
100
50
0
Methylation (%)
Methylation (%)
Methylation (%)
100
CAST1
80
MAGI
75
Controls Patients Cell lines
N=8
N=46 N=23
Controls Patients Cell lines
N=10 N=31
N=23
100
100
RSPO1
Methylation (%)
100
GIPC2
Methylation (%)
Methylation (%)
100
Several of the genes identified by MCA/RDA and MCA/array in
this study have already been reported in the literature to undergo
promoter hypermethylation and gene silencing in various
human cancers. These include, among several, p73,3 RUNX3,14
RASSF1,15 MGMT,16 EGR3,17 SLC26A418 and TERT.19 This data
validated the techniques used here. To further demonstrate that
other genes identified by these techniques were also aberrantly
methylated in leukemia, we selected 31 genes that had been
identified both by the MCA/RDA and MCA/array. These
included: WNT2B, GIPC2, RSPO1, LHX4, TRIMP58, ECRG4,
MAGI1, CAST1, ADCY5, HSPA4L, ZFP42, OCLN, EFNA5,
MSX2, GFPT2, PNMA2, GNA14, FOXE1, CRTAC1, DPYSL4,
FANK1, DCAMKL1, AGC1, SALL1, SLC13A5, PTPRM, MYO5B,
FBX027, ZNF382, MN1, PCDH11.
MN1
75
50
25
0
Controls Patients Cell lines
N=10 N=46 N=23
Controls Patients Cell lines
N=11
N=53 N=23
Figure 1 Methylation levels of 15 selected genes identified by both MCA-RDA and MCA/DNA promoter microarray in normal controls, primary
acute lymphoblastic leukemia (ALL) patients and leukemia cell lines. DNA methylation was analyzed using bisulfite pyrosequencing. The name of
each gene is shown in the upper left corner of each graph. N represents the number of cases studied for each gene. Each graph represents the
results of an individual gene indicated in the left upper corner of each graph.
Leukemia
DNA Methylation in ALL
S-Q Kuang et al
1533
Table 2
Methylation profile of primary ALL samples and normal controls
Bisulfite pyrosequencing was carried out to determine the methylation status. Each row indicates a gene identified by both methylated CpG island
amplification/representational difference analysis (MCA/RDA) and MCA/DNA promoter microarray. White box, not done; green box, methylation
density o15%; yellow box, methylation density between 15–29.99%; pink grey box, methylation density 30–59.99%; red box, methylation density
460%. Methylation density 415% was used as the cutoff for hypermethylation. Methylation frequency was counted as the percentage of positive
methylation numbers versus total numbers studied for each gene. MCA tester and drivers are the results on the patient samples used for the MCA
experiement. P-values were calculated using a t-test comparing cell line methylatyion versus normal controls.
Methylation validation was performed using bisulfite pyrosequencing in 23 leukemia cell lines, 10 normal controls and the 3
Ph negative ALL samples originally used for MCA. Methylation
results are shown in Table 1, Figure 1 and Table 2. Five genes
(TRIMP58, ECRG4, ZFP42, FANK1 and FBXO27) were found
methylated in both leukemia cells lines and normal controls
samples (Table 1), and were considered as non-informative. The
other 26 genes were methylated in at least 2 of the 23 leukemia
cell lines but not or rarely in normal control samples (Table 1).
Spearman coefficient analysis of the 26 genes in leukemia cells
revealed that methylation of these genes was generally
correlated with each other (Supplementary Table 4).
Analysis of gene methylation in primary ALL
We subsequently evaluated the methylation status of 15
randomly selected genes of the 31 genes analyzed in cell lines
in 61 pretreatment samples from patients with ALL. Patient
characteristics are shown in Supplementary Table 4. Results are
shown in Figure 1 and Table 2. Using a methylation cutoff of
Leukemia
1534
Leukemia
DNA Methylation in ALL
S-Q Kuang et al
Figure 2 Gene expression in leukemia cell lines and primary leukemia specimens, and effect of epigenetic modulation on gene expression. Gene expression was analyzed with real-time PCR in
normal bone marrow (BM), peripheral blood specimens (PB) from normal controls, as well as primary acute lymphoblastic leukemia (ALL) samples (ALL). The figure below each sample or cell line
indicates the percent of methylation. Leukemia cells were treated with 5-aza-20 -deoxycytidine (DAC) only, trichostatin A (TSA) only or both (D þ T) as described. Expression of GIPC2 (a), MAGI1 (b),
ADCY5 (c), HSPA4L (d), OCLN (e), EFNA5 (f), GNA14 (g) and MYO5B (h) genes was analyzed by real-time PCR. C, control untreated cells. M % methylation density.
DNA Methylation in ALL
S-Q Kuang et al
1535
15% (as described previously),3 methylation frequencies (percent of patients in which a gene was methylated) ranged from 23
to 100%. Two patients had methylation of 1 gene, 2 of 2 genes,
3 of 3 genes, 14 of 4 genes, 8 of 5 genes, 8 of 6 genes, 11 of 7
genes, 7 of 8 genes, 5 of 9 genes and 3 of 10 genes. No patient
had methylation of more than 11 genes.
To further confirm the results of the pyrosequencing assays,
DNA methylation of RSPO1, ADCY5, HSPA4L and MYO5B was
confirmed in one ALL sample, one normal control and the Raji
cell line using bisulfite sequencing. Sequencing results were in
concordance with the results obtained by pyrosequencing
(Supplementary Figure 1). Spearman coefficient analysis of all
15 genes revealed that methylation of these genes was
significantly correlated with each other (Supplementary Table
4). We also observed an excellent correlation between cell line
methylation results and patients (data not shown).
HSPA4L, OCLN, EFNA5, GNA14 and MYO5B was restored in
methylated leukemia cell lines by either 5-aza-20 -dexocytidine
with or without TSA (Figure 2), a phenomenon associated with
gene demethylation (data not shown). These data indicates that
the DNA methylation and histone deacetylation of the genes
analyzed here are associated with suppressed gene expression in
leukemia cell lines. Furthermore, a relationship was also
observed between methylation and gene expression in primary
ALL samples (Figure 2).
Physical mapping of methylated loci and pathway
analysis of microarray data
The 404 methylated genes identified by MCA/microarray were
analyzed according to their chromosomal location and found to
be distributed evenly in the 22 autosomes. The physical
distribution of these loci was further mapped using MapChart
program (Figure 3). We also performed pathway and functional
analysis of these genes using the Ingenuity Pathway Analysis
program (http://analysis.ingenuity.com). The initial analysis
divided 303 out of the 404 genes into 29 functional networks,
with the majority of genes clustered into 18 networks (Table 3).
Genes clustered in these networks are involved in a wide variety
Gene expression and effect of epigenetic modulation
To examine the role of the DNA methylation in the control of
gene expression, eight genes were studied.20–27 Leukemia cell
lines were treated with 5-aza-20 -deoxycytidine with or without
TSA. In general, expression of GIPC2, MAGI1, ADCY5,
1
2
29(7%)
31(8%)
TP73
PER3
SPEN
NBL1
EPHB2
E2F2
RUNX3
SESN2
BMP8A
BMP8B
TRIT1
PPT1
RNF11
DHCR24
GIPC2
SOX11
RRM2
ROCK2
ALK
CYP1B1
MTA3
SIX2
MSH6
VRK2
REL
MXD1
TGFA
BIN1
ARNT
EFNA4
MAPBPIP
ARHGEF11
IGSF4B
IGSF8
CREG1
LHX4
HOXD13
NAB1
GLS
TMEFF2
SUMO1
FEV
INHA
IRS1
HRB
NCL
GBX2
HES6
PPP1R7
BOK
HLX1
WNT3A
ARF1
5
6
7
8
9
10
11
25(6%)
15(4%)
25(6%)
24(6%)
28(7%)
21(5%)
19(5%)
DGKQ
TACC3
CRMP1
HTRA3
LGI2
RBPSUH
OGG1
IRAK2
TIMP4
THRB
CTDSPL
TSP50
MAP4
CDC25A
SEMA3F
HYAL2
RASSF1
CACNA2D2
ACY1
ALAS1
BAP1
IL17RB
CAST
PTPRG
MAGI1
IGSF11
GATA2
TRH
SOX14
FAM62C
GMPS
IL12A
MAL
STARD7
MAP4K4
ECRG4
GSTM1
WNT2B
SYCP1
4
26(6%)
3
26(6%)
KDR
IGFBP7
EPHA5
CXCL6
CXCL2
FGF5
TSPAN5
PITX2
TIFA
FGF2
NUDT6
HSPA4L
GAB1
EDNRA
NR3C2
SAP30
HAND2
VEGFC
FAT
ZFP42
TERT
BASP1
AMACR
SKP2
OCLN
MSH3
RASA1
EFNA5
TNFAIP8
HSD17B4
LOX
FBN2
IRF1
SPOCK
MATR3
DND1
HDAC3
PCDH1
KIBRA
NKX2-5
STC2
MSX2
NSD1
GRK6
GFPT2
DUSP22
IRF4
FOXQ1
FOXC1
TXNDC5
BAT2
COL11A2
PACSIN1
PTK7
GTPBP2
HTR1B
ME1
PTPRK
SLC22A3
PDCD2
MAFK
MAD1L1
RBAK
ACTB
DFNA5
HOXA5
PDE1C
CAMK2B
BCL7B
CACNA2D1
AKAP9
CDK6
NPTX2
GNB2
CUTL1
SVH
SLC26A4
NRCAM
CAV1
SMO
MEST
PODXL
NOM1
HLXB9
PTPRN2
MSRA
GATA4
EGR3
TNFRSF10A
LOXL2
EXTL3
NRG1
ZNF703
SFRP1
SDCBP
CA8
NCOA2
E2F5
SDC2
TSPYL5
RPL30
RSPO2
EXT1
NOV
ADCY8
BAI1
GML
SCRIB
PLEC1
PSIP1
MTAP
CDKN2A
FANCG
TPM2
MELK
GNA14
TLE4
UBQLN1
GAS1
CTSL
FANCC
ABCA1
KLF4
TXN
DBC1
GSN
LHX6
HSPA5
AK1
PKN3
BARHL1
TSC1
WDR5
RXRA
LHX3
TRAF2
COBRA1
PFKP
IL15RA
NMT2
VIM
EPC1
ITGB1
NRP1
SLC18A3
NCOA4
UBE2D1
RHOBTB1
HK1
DDIT4
PAPSS2
CRTAC1
BTRC
NEURL
SLC18A2
FANK1
MGMT
DPYSL4
USH1C
PTPRJ
UBE2L6
FADS1
STIP1
RPS6KA4
MEN1
EHD1
SYVN1
RELA
CFL1
PACS1
MTL5
FGF3
SERPINH1
PGR
PPP2R1B
ZBTB16
ST14
12
13
14
15
16
17
18
19*
20
21
22
14(3.5%)
10(2.5%)
9(2.2%)
14(3.5%)
15(4%)
13(3.2%)
8(2%)
22(5.4%)
11(2.7%)
6(15%)
6(15%)
WNK1
EPS8
BCAT1
COL2A1
ACVRL1
CSAD
HOXC10
ERBB3
RASSF3
DYRK2
NTN4
APAF1
CRY1
RASAL1
TNFRSF19
APRIN
DCAMKL1
RCBTB2
RCBTB1
GPC6
CLDN10
DZIP1
LIG4
IRS2
KIAA0323
SNX6
SSTR1
MGAT2
BMP4
PRIMA1
BCL11B
CCNK
YY1
THBS1
MAPK6
ALDH1A2
USP3
MEGF11
TLE3
PKM2
COX5A
IMP3
ARNT2
BNC1
NTRK3
IQGAP1
NR2F2
BAIAP3
TRAF7
PKMYT1
CREBBP
EEF2K
MAPK3
CORO1A
VKORC1
NKD1
SALL1
HERPUD1
CDH5
NFATC3
MARVELD3
APRT
PLD2
SLC13A5
EFNB3
PER1
FLCN
SREBF1
IGFBP4
ADAM11
DLX4DLX4
COL1A1
SFRS1
SUMO2
NPTX1
MRCL3
PTPRM
ROCK1
MIB1
RBBP8
DSC3
PSTPIP2
FVT1
PALM
PRG2
MUM1
GNA11
MATK
ZBTB7A
ICAM1
ICAM5
KEAP1
RAB3A
GDF15
CCNE1
CEBPA
ZNF382
SIX5
HIF3A
SEPW1
ATF5
KLK10
PPP2R1A
KLP1
TRIM28
RASSF2
PAX1
TP53INP2
RBL1
SRC
ADA
NCOA3
CEBPB
PFDN4
GNAS
TNFRSF6B
TPTE
CXADR
GABPA
OLIG2
ETS2
UBE2G2
MN1
NF2
TST
PGEA1
L3MBTL2
TOB2
Figure 3 Chromosomal localization of 404 genes identified by MCA/DNA promoter microarray. Each chromosome is numbered at the top. The
figure below the chromosome number represents the number and percentage of methylated loci identified in each chromosome. Gene names are
indicated beside each methylated locus.
Leukemia
DNA Methylation in ALL
S-Q Kuang et al
1536
Table 3
ID
Genes
1
ADA, AMACR, CXCL2, EPHB2, ETS2, FEV, FGF5, FOXC1, GDF15,
ICAM1, IL12A, IL15RA, IRF1, IRF4, L3MBTL2, MAFK, MAP4K4,
MEN1, MXD1, NCL, NF2, RASA1, REL, RELA, SDC2, SDCBP,
TACC3, TERT, TNFAIP8, TNFRSF19, TNFRSF10A (includes
EG:8797), TNFRSF6B, TRAF2, TXN, ZBTB7A
ACTB, ARF1, BIN1, CCNE1, CDH5, CFL1, CRY1, EGR3, EPS8,
GAB1 (includes EG:2549), GABPA, GATA2, GSN, HLXB9, IGSF8
(includes EG:93185), ITGB1, KDR, KEAP1, MATK, NRG1, NRP1,
PER1, PER3, PITX2, PLD2, PTPRG, PTPRJ, RAB3A, SEMA3F,
SFRS1, SKP2, SRC, SUMO2 (includes EG:6613), TOB2, VEGFC
AK1, CDC25A, CDK6, CDKN2A, CDKN2D, COL11A2, COL1A1,
COL2A1, CREG1, E2F2, E2F5, FGF2, FGF3, IGFBP4, IGFBP7,
LOX, MELK, MGMT, MSX2, NBL1, NOV, OLIG2, PLEC1,
PPP2R1A, PPP2R1B (includes EG:5519), RBL1, ROCK2, RPL30,
RRM2, SFRP1, TGFA, THBS1, TP73, VIM, YY1
ALDH1A2, CEBPA, CEBPB, CREBBP, CSAD, CUTL1, GATA4,
GNA11, HAND2, HOXA5, HOXD13, HSPA5, INHA, IRS2, MAPK3,
ME1, MN1, MTA3, NCOA2, NCOA3, NCOA4 (includes EG:8031),
NFATC3, NR3C2, PCDH1, PGR, RPS6KA4, RUNX3, RXRA,
SCRIB, SREBF1, SUMO1, SYCP1, THRB, TRAF7, TRH
ABCA1, ACVRL1, ADA, APAF1, CCNK, COX5A, CRMP1, CTSL,
DHCR24, DPYSL4, FADS1, GDF15, LOXL2, MAPK6, PFKP,
PKM2, TPM2
CAV1, CXADR, CXCL6, FLCN, GNA14, GNB2, HDAC3, MAD1L1,
MSRA, MTAP, OGG1, PALM, PPP1R7, SAP30, TLE3, ZBTB16,
ZFP42
ARNT, ARNT2, BASP1, BMP4, CLDN10, EEF2K, GAS1, KLF4,
MRCL3, NEURL, NKX2-5, NMT2, SERPINH1, STIP1, UBE2G2,
UBE2L6, WNT3A
ACY1, CYP1B1, DLX4, EFNA4, EFNA5, EFNB3 (includes
EG:1949), EPHA5, ERBB3, GBX2, NAB1, NCL, NTRK3, PKMYT1,
SLC18A3, SMO, SYVN1,
BAP1, BCL11B, DDIT4, GSTM1, MAP4, MIB1, NR2F2, PACSIN1,
PODXL, PSIP1, RNF11, ST14, TRIM28, UBE2D1, UBQLN1
FANCC, FANCG, GDF15, GLS, HES6, HSD17B4, HSPA4L,
MATR3, NSD1, RBBP8, RBPSUH, SALL1, SPEN, SSTR1, TSC1
ABCA1, DSC3, EHD1, EXT1, GFPT2, HIF3A, HLX1, NRP1,
RCBTB2, SFRP1, SLC26A4, SPOCK1, TIMP4, TST
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
APRT, CAST (includes EG:26059), DCAMKL1, EDNRA, EPS8,
GRK6, HRB, MAL, MATK, NPTX2, PTPRK, SNX6, WNK1, ZNF703
GIPC2, KDR, MAGI1, NCOA2, NPTX1, NRCAM (includes
EG:4897), PACS1, PDE1C, PGEA1, PTK7, SIX2, SIX5, SOX11
ARHGEF11, BCAT1, BTRC, DGKQ, FBN2, OCLN, PDCD2,
PFDN4, ROCK1, SESN2, SIAHBP1, TLE4, ZBTB7A
BAI1, BAIAP3 (includes EG:8938), BOK, CACNA2D2, HERPUD1,
HYAL2, ICAM5, IQGAP1 (includes EG:8826), LHX4, LHX6,
PTPRM, RASSF1, TNFRSF10A (includes EG:8797)
AKAP9, ALAS1, ATF5, BCL7B, DYRK2, GABPA, GNAS (includes
EG:2778), HK1, MEST, MTL5, PRIMA1, PTPRN2, SLC18A2
19
20
21
COBRA1, CTDSPL, EPC1, GMPS, LHX3, LIG4, MSH3, MSH6,
PAX1, RBAK, RBBP8, WDR5
ALK, BNC1, CACNA2D1, CAMK2B, CORO1A, DUSP22, GNAS
(includes EG:2778), IL17RB, IRAK2, IRS1, RASSF2, WNT2B
IMP3
BMP8A
FAT
22
23
24
25
26
27
28
29
IGSF4B
PAPSS2
RASAL1
MAPBPIP (includes EG:28956)
HTR1B
PKN3
VKORC1
USH1C
18
Leukemia
Functional class of genes identified by MCA/DNA promoter
Focus
genes
Top functions
35
Cellular Growth and proliferation, cell death,
cancer
35
Cellular growth and proliferation, cellular
movement, cell morphology
35
Cell cycle, cellular growth and proliferation,
cancer
35
Gene expression, viral function, developmental
disorder
17
Inflammatory disease, respiratory disease,
organismal development
17
Cancer, cellular growth and proliferation,
hematological disease
17
Gene expression, connective tissue development
and function, skeletal and muscular system
development and function
Nervous system development and function,
tissue development, cellular development
16
15
15
14
14
13
13
13
13
12
12
1
1
1
1
1
1
1
1
1
1
1
Protein degradation, protein synthesis, posttranslational modification
Genetic disorder, hematological disease, cell
cycle
Cardiovascular system development and
function, embryonic development, tissue
development
Amino-acid metabolism, Post-translational
modification, small molecule biochemistry
Skeletal and muscular system development and
function, tissue development, cancer
Gene expression, cancer, reproductive system
disease
Cell death, renal and Urological disease,
embryonic development
Nervous system development and function,
skeletal and muscular system development and
function, cell death
DNA replication, recombination and repair,
cancer, gene expression
Cancer, cell signaling, post-translational
modification
RNA post-transcriptional modification
Cell morphology, gene expression
Cellular compromise, Nervous system
development
Cancer, cell-to-cell signaling
Developmental disorder, genetic disorder
Cellular development
Cancer, cell death, reproductive system disease
Neurological disease, organismal injury
Cellular assembly and organization
Genetic disorder, hematological disease,
Genetic disorder, neurological disease
DNA Methylation in ALL
S-Q Kuang et al
1537
Figure 4 Survival analysis of patients with acute lymphoblastic
leukemia (ALL) based on number of methylated genes. Patient were
grouped in three different sets: methylation of 0–3 genes, 4–7 genes
and 8–10 genes. These could include any the genes studied.
of biological functions including cell growth and differentiation,
cell death, gene regulation, cell cycle, DNA replication and
repair, signal transduction, transport and metabolism.
Exploratory analysis of effect of DNA methylation in
survival
Prior studies have indicated a relationship between hypermethylation of multiple genes and worse survival.5,6 To explore this
effect, we analyzed the impact of methylation of gene(s) on
overall survival. To do this analysis, patients were divided into
three groups: those with methylation (X15%) of 0–3 genes, 4–7
genes and 8–10 genes. The median survival of patients in the
first group of patients had not been reached, whereas it was 173
weeks for the second group and 68 weeks for the third group
(P ¼ 0.012) (Figure 4).
Discussion
Using MCA/RDA and an MCA/DNA promoter microarray, we
identified a large cohort of genes that are potential targets of
aberrant DNA methylation in ALL. Using MCA/RDA we
identified 65 targets of aberrant DNA methylation, and with
the MCA/array 404. Thirty-one of the genes (77%) were
identified by both techniques. Of these 31 genes, 26 genes
were found to be informative in cell lines (not methylated in
normal controls), and 15 genes were confirmed to be
methylated in primary ALL samples. None of the genes that
were validated as informative in cell lines was discarded by the
analysis of primary ALL samples. Furthermore, a number of
these genes had been already shown to be hypermethylated in
cancer. This evidence further demonstrates the validity of the
assays used here.
In the current study, and as a proof of principle, we validated
the methylation of 31 genes not previously reported to be
hypermethylated in ALL. Methylation analysis excluded
TRIMP58, ECRG4, ZFP42, FANK, DHX32 and FBXO27 as they
were found to be methylated in normal controls. The other 26
genes were found to be specifically hypermethylated in
leukemia cell lines, and 15 of them were further confirmed to
be hypermethylated in a panel of 61 ALL patients. Methylation
of these genes was highly correlated both in cell lines and
primary samples. Thus, confirming the presence of a hypermethylator phenotype in ALL, as described previously.3
To demonstrate that the level of methylation identified here
was associated with gene expression inactivation, we studied
the effect of treating cell lines with a hypomethylating agent
with or without a histone deacetylase inhibitor. As expected,
treatment of cell lines with these epigenetic modulators resulted,
in general, in gene reactivation, an effect that was mediated by
induction of DNA hypomethylation. An inverse relationship
between methylation and gene expression was also observed in
primary ALL samples.
Targets of aberrant DNA methylation were evenly distributed
in all autosomes with no evidence of particular chromosomal
hot spots for methylation. The function of these genes clustered
in 29 different networks involved in multiple cellular functions.
Finally, and as described in other previous studies,5,6
methylation of multiple genes was associated with a worse
outcome. Those patients with methylation of more than four
genes appeared to have a significant worse outcome when
treated with hyperCVAD based chemotherapy.
The implications of this work are very significant. First, we
have made public the largest list of potential targets of aberrant
DNA methylation in ALL. This provides a significant source to
perform further studies of DNA methylation in ALL, and
potentially other leukemias. We provide information that these
genes cluster together in specific functional networks and
chromosomal locations. This provides the opportunity to
investigate specific functions of genes located in specific
chromosomal locations. It should be noted that this is not the
first large-scale analysis of methylation in ALL. Recently, Taylor
et al.28 reported on 262 methylated genes using a different CGI
microarray method. We compared the gene lists identified by
our method and their CGI array and found that only nine genes
(NASP, FEV, HRB, EFNA5, GTPBP2, RPIB9, CAV1, NOV and
SMC2L1) were concordant. This probably relates to the different
methods used, in particular the regions analyzed, and the data is
probably complementary and it argues that potentially other
targets of methylation exist that have not been identified by
these two studies combined. It also should be noted that we
have identified a number of genes methylated in normal
samples. The significance of this observation is not known at
the present time.
In summary, we have identified a group of aberrantly
methylated genes that can be used as potential epigenetic
biomarkers of ALL. We have demonstrated that ALL exhibits
widespread epigenetic alterations. Although probably not all
methylation events identified here are relevant, clinically or
functionally, in ALL directly, the study of this group of genes
may provide further insights into the pathogenesis of ALL.
Acknowledgements
This work was supported by a Leukemia SPORE (P50CA100632)
Career Development Award to S-Q K, and NCI grants CA100067
and CA105771, the Leukemia & Lymphoma Society of America
and a MD Anderson Cancer Center’s Physician-Scientist Award
from the Commonwealth Cancer Foundation for Research (all to
G G-M).
References
1 Pui CH, Relling MV, Downing JR. Acute lymphoblastic leukemia.
N Engl J Med 2004; 350: 1535–1548.
2 Kuang SQ, Ling X, Sanchez-Gonzalez B, Yang H, Andreeff M,
Garcia-Manero G. Differential tumor suppressor properties and
transforming growth factor-beta responsiveness of p57KIP2 in
Leukemia
DNA Methylation in ALL
S-Q Kuang et al
1538
3
4
5
6
7
8
9
10
11
12
13
14
15
16
leukemia cells with aberrant p57KIP2 promoter DNA methylation.
Oncogene 2007; 26: 1439–1448.
Garcia-Manero G, Daniel J, Smith TL, Kornblau SM, Lee MS,
Kantarjian HM et al. DNA methylation of Multiple Promoterassociated CpG Islands in Adult Acute Lymphocytic Leukemia.
Clin Cancer Res 2002; 8: 2217–2224.
Gutierrez MI, Siraj AK, Bhargava M, Ozbek U, Banavali S,
Chaudhary MA et al. Concurrent methylation of multiple genes in
childhood ALL: correlation with phenotype and molecular
subgroup. Leukemia 2003; 17: 1845–1850.
Shen L, Toyota M, Kondo Y, Obata T, Daniel S, Pierce S et al.
Aberrant DNA methylation of p57KIP2 identifies a cell-cycle
regulatory pathway with prognostic impact in adult acute
lymphocytic leukemia. Blood 2003; 101: 4131–4136.
Roman-Gomez J, Jimenez-Velasco A, Castillejo JA, Agirre X,
Barrios M, Navarro G et al. Promoter hypermethylation of cancerrelated genes is a strong independent prognostic factor in acute
lymphoblastic leukemia. Blood 2004; 104: 2492–2498.
Bueso-Ramos C, Xu Y, McDonnell TJ, Brisbay S, Pierce S,
Kantarjian H et al. Protein expression of a triad of frequently
methylated genes, p73, p57Kip2, and p15, has prognostic value in
adult acute lymphocytic leukemia independently of its methylation
status. J Clin Oncol 2005; 23: 3932–3939.
Garcia-Manero G, Bueso-Ramos C, Daniel J, Williamson J,
Kantarjian HM, Issa JP. DNA methylation patterns at relapse in adult
acute lymphocytic leukemia. Clin Cancer Res 2002; 8: 1897–1903.
Herman JG, Baylin SB. Gene silencing in cancer in association with
promoter hypermethylation. N Engl J Med 2003; 349: 2042–2054.
Rauch T, Li H, Wu X, Pfeifer GP. MIRA-assisted microarray
analysis, a new technology for the determination of DNA
methylation patterns, identifies frequent methylation of homeodomain-containing genes in lung cancer cells. Cancer Res 2006;
66: 7939–7947.
Toyota M, Ho C, Ahuja N, Jair KW, Li Q, Ohe-Toyota M et al.
Identification of differentially methylated sequences in colorectal
cancer by methylated CpG island amplification. Cancer Res 1999;
59: 2307–2312.
Garcia-Manero G, Kantarjian HM. The hyper-CVAD regimen in
adult acute lymphocytic leukemia. Hematol Oncol Clin North Am
2000; 14: 1381–1396, x–xi.
Garcia-Manero G, Kantarjian HM, Sanchez-Gonzalez B, Yang H,
Rosner G, Verstovsek S et al. Phase 1/2 study of the combination of
5-aza-20 -deoxycytidine with valproic acid in patients with
leukemia. Blood 2006; 108: 3271–3279.
Li QL, Ito K, Sakakura C, Fukamachi H, Inoue K, Chi XZ et al.
Causal relationship between the loss of RUNX3 expression and
gastric cancer. Cell 2002; 109: 113–124.
Dammann R, Li C, Yoon JH, Chin PL, Bates S, Pfeifer GP.
Epigenetic inactivation of a RAS association domain family protein
from the lung tumour suppressor locus 3p21.3. Nat Genet 2000;
25: 315–319.
Esteller M, Hamilton SR, Burger PC, Baylin SB, Herman JG.
Inactivation of the DNA repair gene O6-methylguanine-DNA
17
18
19
20
21
22
23
24
25
26
27
28
29
methyltransferase by promoter hypermethylation is a common
event in primary human neoplasia. Cancer Res 1999; 59: 793–
797.
Matsubayashi H, Canto M, Sato N, Klein A, Abe T, Yamashita K
et al. DNA methylation alterations in the pancreatic juice of
patients with suspected pancreatic disease. Cancer Res 2006; 66:
1208–1217.
Xing M, Tokumaru Y, Wu G, Westra WB, Ladenson PW, Sidransky
D. Hypermethylation of the Pendred syndrome gene SLC26A4 is
an early event in thyroid tumorigenesis. Cancer Res 2003; 63:
2312–2315.
Liu L, Saldanha SN, Pate MS, Andrews LG, Tollefsbol TO.
Epigenetic regulation of human telomerase reverse transcriptase
promoter activity during cellular differentiation. Genes Chromosomes Cancer 2004; 41: 26–37.
Kirikoshi H, Katoh M. Molecular cloning and characterization of
human GIPC2, a novel gene homologous to human GIPC1 and
Xenopus Kermit. Int J Oncol 2002; 20: 571–576.
Kotelevets L, van Hengel J, Bruyneel E, Mareel M, van Roy F,
Chastre E. Implication of the MAGI-1b/PTEN signalosome in
stabilization of adherens junctions and suppression of invasiveness. FASEB J 2005; 19: 115–117.
Ludwig MG, Seuwen K. Characterization of the human adenylyl
cyclase gene family: cDNA, gene structure, and tissue distribution
of the nine isoforms. J Recept Signal Transduct Res 2002; 22:
79–110.
Takahashi H, Furukawa T, Yano T, Sato N, Takizawa J,
Kurasaki T et al. Identification of an overexpressed gene,
HSPA4 L, the product of which can provoke prevalent humoral
immune responses in leukemia patients. Exp Hematol 2007; 35:
1091–1099.
Ichimiya S, Kojima T. Cellular networks of human thymic
medullary stromas coordinated by p53-related transcription
factors. J Histochem Cytochem 2006; 54: 1277–1289.
Shi H, Guo J, Duff DJ, Rahmatpanah F, Chitima-Matsiga R,
Al-Kuhlani M et al. Discovery of novel epigenetic markers
in non-Hodgkin’s lymphoma. Carcinogenesis 2007; 28:
60–70.
Liu AM, Wong YH. Activation of nuclear factor {kappa}B by
somatostatin type 2 receptor in pancreatic acinar AR42J cells
involves G{alpha}14 and multiple signaling components: a
mechanism requiring protein kinase C, calmodulin-dependent
kinase II, ERK, and c-Src. J Biol Chem 2005; 280: 34617–34625.
McGurk L, Tzolovsky G, Spears N, Bownes M. The temporal and
spatial expression pattern of myosin Va, Vb and VI in the mouse
ovary. Gene Expr Patterns 2006; 6: 900–907.
Taylor KH, Kramer RS, Davis JW, Guo J, Duff DJ, Xu D et al.
Ultradeep bisulfite sequencing analysis of DNA methylation
patterns in multiple gene promoters by 454 sequencing. Cancer
Res 2007; 67: 8511–8518.
Shu J, Jelinek J, Chang H, Shen L, Qin T, Chung W et al. Silencing
of bidirectional promoters by DNA methylation in tumorigenesis.
Cancer Res 2006; 66: 5077–5084.
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
Leukemia