Epigenetics

ISSN: 1559-2294 (Print) 1559-2308 (Online) Journal homepage: http://www.tandfonline.com/loi/kepi20

5-azacytidine treatment reorganizes genomic

histone modification patterns

Vitalina M. Komashko & Peggy J. Farnham

To cite this article: Vitalina M. Komashko & Peggy J. Farnham (2010) 5-azacytidine treatment
reorganizes genomic histone modification patterns, Epigenetics, 5:3, 229-240, DOI: 10.4161/
epi.5.3.11409

To link to this article: http://dx.doi.org/10.4161/epi.5.3.11409

Copyright © 2010 Landes Bioscience

View supplementary material

Published online: 01 Apr 2010.

Submit your article to this journal

Article views: 331

View related articles

Citing articles: 39 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=kepi20

Download by: [b-on: Biblioteca do conhecimento online UP] Date: 07 June 2016, At: 05:54
 Research paper Research Paper
Epigenetics 5:3, 229-240; April 1, 2010; © 2010 Landes Bioscience

5-azacytidine treatment reorganizes genomic

histone modification patterns
Vitalina M. Komashko and Peggy J. Farnham*
Department of Pharmacology and the Genome Center; University of California-Davis; Davis, CA USA

Key words: DNA methylation, 5-azacytidine, ChIP-chip, cancer cells, H3K27me3, H3K9me3, transcriptional repression

Abbreviations: ChIP-chip, chromatin immunoprecipitation coupled with DNA microarrays; MeDIP, methylated DNA
immunoprecipitation; DE, differentially expressed
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016

Methylation of DNA in combination with histone modifications establishes an epigenetic code that ensures the
proper control of gene expression. Although DNA methyltransferases have been shown to interact with histone
methyltransferases such as EZH2 (which methylates histone H3 on lysine 27) and G9a (which methylates histone H3 on
lysine 9), the relationship between DNA methylation and repressive histone marks has not been fully studied. In cancer
cells, promoters of genes are often aberrantly methylated. Accordingly, 5-azacytidine (a DNA demethylating drug) is
used for treating patients with myelodysplastic syndrome. However, no genome-scale studies of the effects of this drug
have been reported. In this work, we report the effects of 5-azacytidine on global gene expression and analyze ∼24,000
human promoters using ChIP-chip to determine how 5-azacytidine treatment effects H3K27me3 and H3K9me3 levels.
We found that (1) 5-azacytidine treatment results in large changes in gene regulation with distinct functional categories
of genes showing increased (e.g., C2H2 zinc finger transcription factors) and decreased (e.g., genes involved in regulation
of mitochondria and oxidoreductase activity) levels, (2) most genes that show altered expression are not regulated by
promoters that display DNA methylation prior to the treatment, (3) certain gene classes switch their repression mark upon
treatment with 5-azacytidine (from H3K27me3 to H3K9me3 and vice versa), and (4) most changes in gene expression are
not due to relief of repression mediated by DNA or histone methylation.

Introduction myelodysplastic syndrome (MDS). It acts by incorporating into

The most common DNA methylation in human cells is the cova-
lent addition of a methyl group to cytosine bases in CpG nucle-
otides. This reaction is catalyzed by DNA methyltransferases
(DNMTs), using S-adenosyl-methionine as the methyl donor. In
mammals there are three DNMTs: DNMT1, which is involved in
maintenance methylation, and DNMT3a and DNMT3b, which
function as de novo DNA methyltransferases. DNA methylation
is a stable modification that is inherited through cellular divi-
sions. This allows the daughter cells to retain the same expres-
sion pattern as the precursor cells and is important for many
cellular processes including the silencing of repetitive elements,1
X-inactivation and imprinting,2 and repression of developmental
and differentiation-specific genes.3
The development of cancer is characterized by two features:
global genomic DNA demethylation leading to abnormal gene
activation and chromosomal breaks and concurrent gain of
DNA methylation on promoters of tumor suppressing genes.4
This last process has created a strong incentive for the develop-
ment of DNA demethylating drugs. 5-azacytidine is a nucleoside
inhibitor which was the first hypomethylating agent approved
by the U.S. Food and Drug Administration for treatment of the
the genome of proliferating cells during DNA synthesis and trap-
ping DNMTs (which targets them for degradation), resulting in
hypomethylation of the genome. Although therapeutic treatment
with 5-azacytidine is effective and improves overall survival rate
in clinical trials, most patients eventually develop resistance to
the drug.4-6 The mechanisms of this resistance are unknown.
However, treatment with 5-azacytidine results in nonspecific
overall DNA demethylation and this process may affect multiple
regulatory pathways. Also, multi-protein complexes have been
identified which contain various combinations of DNMTs and
histone modifying enzymes.7 For example, the histone methyl-
transferase EZH2, which targets lysine 27 on histone H3, can
recruit DNMT1, DNMT3A and DNMT3B to target genes.8,9
SETDB1, a H3K9 specific methyltransferase, can interact with
DNMT3A and DNMT3B, on promoter regions of cancer cells.10
Also, heterochromatin protein 1 (HP1) can mediate an interac-
tion between G9a (which targets H3K9) and DNMT1, influenc-
ing the silencing of euchromatic genes.11 Taken together, these
previous studies suggest that perhaps destruction of DNA meth-
yltransferases by 5-azacytidine may affect the integrity of histone
methylation complexes and change genomic histone patterns,
which in turn may contribute to the patient responses observed

*Correspondence to: Peggy Farnham; Email: pjfarnham@ucdavis.edu

Submitted: 12/17/09; Accepted: 02/04/10
Previously published online: www.landesbioscience.com/journals/epigenetics/article/11409

www.landesbioscience.com Epigenetics 229

 using an antibody that recognizes 5-methylcytidine
in the control and the treated cells and then perform-
ing PCR to analyze genomic regions known to be
methylated in the control cells. A large difference in
the amount of product of known methylated regions
in the control vs. the treated cells provides evidence
that the 5-azacytidine treatment was successful in
inhibiting DNA methylation.
After confirming that DNA methylation was
reduced in the treated cells, we prepared RNA from
two biological replicates of HEK 293 cells treated
with 50% acetic acid (control), 5 uM 5-azacytdine
for 4 days, and 5 uM 5-azacytidine for 8 days (a total
of 6 samples). We then measured RNA levels from
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016

each sample using Illumina Human Ref-8 Expression

BeadChip arrays. First, we analyzed each biological
replicate separately to determine the reproducibility of
the experiments. For this step, we identified mRNAs
that were differentially expressed (DE) in the control
Figure 1. Analysis of differentially expressed genes in HEK 293 cells treated with vs. the treated cells using the Bioconductor13 (www.
5-azacytidine. (A) Comparison of differentially expressed genes between individual bioconductor.org/) nudge package.14 This package
biological replicates in cells treated with 5-azacytidine for 4 and 8 days. Overlap of
provides RNA data normalization and identifica-
the differentially expressed genes between two biological replicates for the same
treatment is shown. (B) Comparison of the differentially expressed genes identified tion of DE genes when only a few or no replicates
in 5-azacytidine 4 and 8 days treatments. Overlap between differentially expressed are available. The analysis of the data obtained using
genes in the two treatments is shown. (C) Shown is the number of up and downregu- only one RNA sample imposes statistical limitations
lated genes in the 4 day and 8 day treated cells. and therefore the number of genes obtained in this
initial analysis was rather conservative (a p value cut-
in clinical trials. To investigate a link between histone marks and off of 0.5 was used for selecting DE genes). Using this method,
DNA methylation we analyzed effects of 5-azacytidine on global we identified 154 and 108 DE genes from the two independent
gene expression and on H3K27me3 and H3K9me3 levels at tar- replicates of cells treated with 5-azacytidine for 4 days and 484
get promoters. and 455 DE genes in the two independent replicates obtained
from cells treated with 5-azacytidine for 8 days. The overlap
Results between DE genes in the 4 day treatment datasets was 77 and the
overlap between DE genes in the 8 day treatment datasets was
Analysis of differentially expressed genes after 5-azacytidine 336 (Fig. 1A). The high number of common genes between each
treatment. We began our studies by comparing the levels of of the two separate experiments indicated that the drug treat-
overall DNA methylation in four different human cell lines and ments were fairly consistent between experiments and that they
found that the level of DNA methylation in HEK 293 cells was induced similar changes in gene expression.
very similar to the levels of DNA methylation in breast (MCF7), Having shown that the replicates were comparable, we com-
liver (HepG2) and colon (HCT116) cancer cells (Suppl. Fig. S1). bined biological replicates for each treatment to identify a less
These results are consistent with our previous studies12 in which conservative set of DE genes. We used the Bioconductor beadar-
we showed that 13–17% of low-expressed genes have methylated ray15 package to normalize the data and the limma package to
promoters regardless of whether the cells were derived from nor- perform differential expression analysis.16 The p values were
mal or cancer tissue. Therefore, we chose to analyze the effects adjusted for multiple comparisons using the Benjamini and
of 5-azacytidine on the transcriptome of human embryonic Hochberg method.16,17 The total number of DE genes was 169 in
kidney 293 (HEK 293) cells. We treated the cells with 5 uM the 4 days treated cells and 1,424 in the 8 days treated cells (Fig.
5-azacytidine dissolved in 50% acetic acid for a short (4 days) 1B). Comparison of the differential expressed genes between
and a long (8 days) period. The medium containing 5-azacyti- two treatments showed that a majority of genes with expres-
dine was changed daily to ensure a constant treatment regimen; sion changes in the 4 day treatment were also deregulated in the
the control cells were treated with a 50% acetic acid solution for 8 day treatment; deregulation of only 14 genes was specific to the
the same time periods. Because our preliminary studies revealed 4 day treatment. As shown in panel C, the number of upregulated
that the efficiency of inhibiting DNA methylation could vary genes was 115 and 791 and the number of downregulated genes
from time to time, for all experiments reported in this study was 54 and 633 in the 4 and 8 day treatments, respectively.
we confirmed that DNA methylation was successfully removed Gene ontology analysis of deregulated genes. The next step
using MeDIP-PCR (Suppl. Fig. S2). Briefly (see below for more in characterizing the effects of 5-azacytidine was to perform
details), this involves immunoprecipitating the genomic DNA functional annotation of the deregulated genes. To determine if

230 Epigenetics Volume 5 Issue 3

 there are any distinct classes of genes that became up or down-
regulated in 4 or 8 days of treatments, we used the DAVID gene
ontology program.18,19 We identified all functional categories of
genes having an enrichment p value of at least 0.01. A bar graph
for each group (up or downregulated, 4 or 8 days) indicating
relative percentage and an enrichment p value for each functional
category are shown in Figure 2. Because of the small size of the
datasets, only few gene categories passed the p value cutoff in
the 4 day up and downregulated sets. However, because most
of the genes identified in the 4 day treatment were also identi-
fied in the 8 treatment (Fig. 1B), most of the gene classes identi-
fied in the 4 day treatment were also found in the corresponding
group (up or downregulated) of the 8 day treatment. For exam-
ple, GAGE, MAGE proteins, tumor antigen and transcription-
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016
related gene classes were found in the upregulated group in the
4 and 8 day treatments (Fig. 2A and C). Similarly, genes with
oxidoreductase activity became downregulated in both 4 and 8
day treatments (Fig. 2B and D). Interestingly, functional catego-
ries in the upregulated group in the 8 day treatment are mostly
represented by genes involved in transcriptional regulation: Zinc-
finger C2H2 type, DNA binding, KRAB box and transcrip-
tional factor activity. On the other hand, gene categories in the
downregulated group include genes involved in the regulation of
mitochondria, oxidoreductase activity, Golgi apparatus and sig-
nal transduction.
DNA methylation analysis of the deregulated genes. Having
identified a set of 169 and 1,424 genes whose expression changed
upon 5-azacytidine treatment, the next step was to investigate
the connection between changes in gene expression and the DNA
methylation status of the promoters of the deregulated genes. To
identify methylated promoters, we performed methylated DNA
immunoprecipitation (MeDIP) with control and treated cells
using an antibody that recognizes methylated cytosine. We con-
firmed the specificity of our MeDIP assays by showing reduction
in DNA methylation on known positive targets in the cells treated
with 5-azacytidine (Suppl. Fig. S2). Then, we proceeded with
amplicon preparation for the methylated DNA from the control
cells. After confirmation that amplicons were representative of
the MeDIP samples, the amplicons were labeled and hybridized
to human hg17 1.5 kb min promoter arrays (see Suppl. Table
1). The enrichment values for each promoter were determined
using the Maxfour algorithm20 and the top ranked set of 2,000
methylated promoters was selected (log2 values were 0.87 to 2.67,
giving fold enrichments of 1.8 to 6.36). We then intersected the
set of top 2,000 methylated promoters with the set of up and
downregulated genes identified in the 4 and 8 day treatment to
determine which of the deregulated genes were regulated by pro-
moters that showed DNA methylation in the control cells (see
Suppl. Table 5 for the lists of up and downregulated genes that
had promoter methylation in the control cells). We found that
promoters of only 7 of the 54 (13.5%) genes that were down-
regulated and 10 of the 115 (9.0%) genes that were upregulated
in the 4 day treatment were methylated in the control cells (Fig.
3A and B). For the 8 day treatment, we found that only 45 of the
633 (8.3%) genes that were downregulated and 76 of the 792
(10.6%) genes that were upregulated had methylated promoters
www.landesbioscience.com Epigenetics 231
in the control cells (Fig. 3C and D). It is possible that these low
numbers were due to selecting a too stringent cut-off for targets
on the MeDIP array (i.e., perhaps there are more than 2,000
methylated promoters in the control cells). Therefore, we also
analyzed the top 5,000 MeDIP-chip targets (log2 values of 0.65
to 2.67, giving fold enrichment of 1.57 to 6.36). Not surpris-
ingly, we identified additional deregulated genes having methy-
lated promoters using this less stringent cut-off (with the values
increasing to 22–30% of each category). However, the methyla-
tion levels of the promoters ranked between 2,000 and 5,000 on
the MeDIP array were quite low and it is likely that most of these
are not true methylation targets. However, whether we used a
more stringent or very lenient cut-off, our results indicate that
gene expression changes are not limited to genes whose promot-
ers are methylated in the control cells; i.e., most gene expression
changes are not a direct consequence of promoter demethylation.
Also, our finding that similar percentages of the up and down-
regulated genes were methylated in the control cells supports our
and others previous studies12,21 suggesting that DNA methylation
might be involved in both activation and repression of transcrip-
tion. We also note that our results showing that methylated pro-
moters can correspond to both active and silent loci in HEK 293
cells are consistent with our previous studies indicating that a
similar percentage of high, medium and low expressed genes have
methylated promoter regions in both normal and cancer cells.12
Effects of 5-azacytidine on histone modifications. In the pre-
vious section we showed that most changes in gene expression
in 5-azacytidine treated cells were from genes whose promoters
were not DNA methylated in the control cells. These results sug-
gest that the 5-azacytidine treatment was altering some other
aspect of transcriptional regulation. The DNA methyltransferase
DNMT1 has been shown to be associated with the histone meth-
yltransferases EZH2,8 and G9a,11 so perhaps 5-azacytidine treat-
ment affected methylation of histones. To test this possibility,
we analyzed levels of H3K27me3 and H3K9me3 in the treated
cells. The largest changes in gene expression were observed in
the long term treatment (8 days); therefore we performed ChIP
assays with control and 8 day treated cells using antibodies to
H3K27me3 and H3K9me3. Each ChIP sample was tested by
PCR using primers for positive and negative controls (see Suppl.
Table 1 for primer sequences). After confirmation that the ChIP
assays were successful, amplicons were prepared and tested using
the same positive and negative primer sets. After demonstrating
that the amplicons were an accurate representation of the start-
ing ChIP samples, they were labeled and hybridized to human
hg17 1.5 kb min promoter arrays (see Suppl. Table 1 for a list
of the ChIP-chip arrays). The enrichment values for each pro-
moter were determined and the top ranked 2,000 promoters for
each antibody in the control and treated cells were identified.
To determine if the 5-azacytidine treatment affected promoters
that were silenced by histone modifications, we first identified
promoters that had high levels of H3K27me3 and H3K9me3 in
the control cells and then determined the levels of these modi-
fied histones on the same promoters in the treated cells. For this
comparison, we selected the set of top 2,000 H3K9me3 (log2
values of 0.95 to 4.55, giving enrichments of 1.9 to 23.4) and the
 EHMT2, SETD2, SUV39H1 and MLL3
in the gene expression array data. We did
not observe any changes in the levels of the
mRNAs encoding these histone methylases.
However, we cannot rule out the possibility of
changes in levels of the modified histones. To
directly test whether modified histones were
decreased, we prepared histone fractions from
control and 5-azacytidine treated cells and
analyzed them by western blot using anti-
bodies against H3K27me3 and H3K9me3
(Fig. 5). We found that levels of H3K9me3
and H3K27me3 in the control HEK293 cells
were very similar to levels in Ntera2 cells.
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016

Surprisingly, the overall level of H3K9me3

greatly increased in 5-azacytidine treated cells.
The level of H3K27me3 also increased in the
treated cells, but not as much as H3K9me3.
These results suggest that although treatment
with 5-azacytidine resulted in removal of
H3K27me3 and H3K9me3 marks from the
promoter regions that were targets in the con-
trol cells, there is an overall increase in these
modified histones. Because these histone
modifications are found only on chromatin-
associated histones, these results suggested
that the modified histones were not simply
lost from chromatin, but rather they must be
Figure 2A and C. Gene ontology analysis of differentially expressed genes. Lists of upregu- targeted to different regions in the control and
lated genes after 4 and 8 days of treatment were analyzed using the DAVID gene ontology 5-azacytidine treated cells. In support of this
program18-19. Shown are the functional categories of upregulated genes after 4 (A) and 8 (C)
days of treatment. The x-axis shows the relative percentage of genes in each category with
hypothesis, we noted that enrichment values
tick marks for each bar. Enrichment p value for each functional category is shown. on the H3K27me3 and H3K9me3 ChIP-chip
arrays from the cells treated with 5-azacyti-
dine for 8 days were as high as the enrichment
top 2,000 H3K27me3 (log2 values of 0.98 to 4.1, giving enrich- values in the control cells (Table 1).
ments of 1.0 to 16.9) promoter targets in control cells (Fig. 4). Gene ontology analysis of H3K9me3 and H3K27me3 targets
We then determined how many of these targets were also in the in control and treated cells. Our ChIP-chip results suggested
set of top 2,000 targets for that same antibody in the treated cells that a different set of genes may have H3K27me3 or H3K9me3
(H3K27me3 log2 values of 1.18 to 4.32, giving fold enrichments marks on their promoters in the control vs. the treated cells. To
of 2.26 to 19.9; H3K9me3 log2 values of 1.0 to 3.31, giving fold determine if different genes became repressed by these two marks
enrichment values of 2.0 to 9.9). We found that only 11.2% of in the treated cells, we selected the top 2,000 H3K27me3 and
the H3K9me3 targets and 8.9% of the H3K27me3 targets from the top 2,000 H3K9me3 promoters in the control and treated
the control cells were in the top 2,000 targets identified using the cells and then performed functional annotation of the targets
same antibody in the treated cells. These results demonstrate that using the DAVID gene ontology program.18,19 We required that
removal of DNA methylation has a profound effect on lysine 9 all functional categories of genes had an enrichment p value of
and lysine 27 trimethylation of histone H3 on promoter regions; at least 0.01 and that each category contains at least ten genes.
i.e., most of the lysine 9 and lysine 27 trimethylation marks on A bar graph showing the relative percentage of genes and an
histone H3 were erased from control target promoter regions enrichment p value for each identified functional category for
upon treatment with 5-azacytidine. We also tested the ZNF333 the control and treated cells are shown in Figure 6. We found
gene in HepG2 cells treated with 5-azacytidine for 8 days and that in the control cells, the most enriched functional catego-
observed H3K9me3 loss at this target after treatment (see Suppl. ries repressed by H3K27me3 are G protein coupled receptors,
Fig. S3). rhodospsin-like GPCR superfamily members, transmembrane
One possible explanation for our observations is that there was proteins, olfactory receptors and glycoproteins (Fig. 6A). These
a global loss of the modified histones in the treated cells, perhaps categories are commonly found for H3K27me3 in adult cells.22
due to downregulation of a histone methylase. We examined Functional categories repressed by H3K9me3 in the control cells
RNA levels for EZH2, SETDB1, DOT1L, SETD7, EHMT1, were enriched in zinc finger proteins, KRAB box, metal-binding

232 Epigenetics Volume 5 Issue 3

Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016

Figure 2B and D. Gene ontology analysis of differentially expressed genes. Lists of downregulated genes after 4 and 8 days of treatment were
analyzed using the DAVID gene ontology program18-19. Shown are the functional categories of downregulated genes after 4 (B) and 8 (D) days of
treatment. The X-axis shows the relative percentage of genes in each category with tick marks for each bar. Enrichment p value for each functional
category is shown.

and genes involved in transcriptional regulation (Fig. 6C). These
categories are also commonly found for genes repressed by
H3K9me3.12 Surprisingly, gene ontology analysis of the treated
cells revealed new patterns of repression. We found that some of
the categories that were repressed by H3K27me3 in the control
cells, such as olfaction, glycoproteins, G protein coupled recep-
tors and rhodospsin like GPCR superfamily of genes, were also
targets in the treated cells. However, we observed that several
functional categories that are usually specific to H3K9me3,
including C2H2-type zinc finger genes and KRAB box domains,
were now enriched in the H3K27me3 target set in the treated
cells (black bars in Fig. 6B). Similarly, gene ontology analysis of
the H3K9me3 targets in the treated cells revealed similarity with
functional categories, such as olfactory receptors, glycoproteins,
G protein coupled receptors and rhodospsin like GPCR super-
family members, which are usually repressed by H3K27me3
(white bars in Fig. 6D). Strikingly, no zinc-finger gene related
categories were found among functional categories repressed by
H3K9me3 in the treated cells. These analyses demonstrate that
upon 5-azacytidine treatment of HEK 293 cells, the two repres-
sive marks H3K27me3 and H3K9me3 switch some of their func-
tional categories of targets.
To determine what percentage of the targets in the control
cells switched to repression by a different mechanism, we per-
formed additional analyses (Fig. 7). We had shown that ∼89%
of the H3K9me3 targets from control cells lost the mark (Fig.
4A); we now determined that of this set, 13% of the targets lost
H3K9me3 and gained H3K27me3 whereas 76% lost H3K9me3
but did not gain H3K27me3. Similarly, we had shown that 91%
of the H3me3K27 targets from control cells lost the mark (Fig.
4B); we now determined that of this set of targets, 22% of the
targets lost H3K27me3 and gained H3K9me3 whereas 69%
lost H3K27me3 but did not gain H3K9me3. To confirm these
ChIP-chip results, we performed PCR analysis of a set of array
targets that showed histone mark switching after the treatment
with 5-azacytidine on the same amplicons that were hybridized
to arrays (Suppl. Fig. S4).
Analysis of histone marks on promoters of differentially
expressed genes. Analysis of the DNA methylation status of the
differentially expressed genes showed that a very small percent-
age of promoters were methylated prior to the treatment with 5-
azacytidine for 4 and 8 days, which means that DNA demethyla-
tion was not a direct cause of most of the changes in gene expres-
sion. However, as indicated above, we found that 5-azacytidine

www.landesbioscience.com Epigenetics 233


Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016
Figure 3. The majority of gene promoters that become deregulated
after treatment with 5-azacytidine are not methylated before treat-
ment. MeDIP-chip promoter array data from control cells was matched
to 5-azacytidine 4 and 8 day treatment RNA expression data. Shown are
the percentage of genes that were (A) downregulated after 4 days, (B)
upregulated after 4 days, (C) downregulated after 8 days, (D) upregu-
lated after 8 days of treatment and that were or were not methylated in
the control cells before treatment.
has a profound effect on H3K27me3 and H3K9me3 histone
marks, leading to the loss of the marks on the majority of tar-
gets identified in control cells and switching of some targets to
the other histone mark. It was possible that alterations (either a
loss or a gain) in repressive histone marks could be responsible
for some of the increased or decreased levels of gene expression
observed after treatment with 5-azacytidine. To test this possi-
bility, we determined the percentage of differentially expressed
genes that had any of these modified histone marks before the
treatment; to complete the analysis, we also included DNA meth-
ylation data (Fig. 8). We found that only 20–25% of the differ-
entially expressed genes had any of these three repressive marks
in the control cells.
Discussion
As indicated above, 5-azacytidine is currently in clinical trials
and therefore it is critical to understand the genome-wide effects
that it has on the transcriptome and on chromatin structure.
Therefore, we have treated cells with this drug and examined the
effects of treatment on RNA expression and chromatin patterns.
Our analysis of the effects of 5-azacytidine on HEK 293 cells
has provided the following insights: (1) 5-azacytidine treatment
results in large changes in gene regulation with distinct functional
234 Epigenetics Volume 5 Issue 3
Figure 4. Most promoters lost their repressive mark (H3K9me3 or
H3K27me3) upon treatment with 5-azacytidine for 8 days. The top 2,000
H3K9me3 promoter targets (A) and the top 2,000 H3K27me3 promoter
targets (B) from control cells were identified and then the same chro-
matin modification was analyzed for each promoter in the treated cells.
Shown is the percentage of the promoters that lost or retained the mark
in the treated cells.
categories of genes showing increased (e.g., C2H2 zinc finger
transcription factors) and decreased (e.g., genes involved in regu-
lation of mitochondria and oxidoreductase activity) expression;
(2) most genes that show altered expression are not regulated by
promoters that display DNA methylation prior to the treatment;
(3) treatment with 5-azacytidine results in increased levels of the
repressive histone marks H3K27me3 and H3K9me3 which are
targeted to promoters that are not normally repressed by those
specific marks, with some of the targets switching between
H3K27me3 and H3K9me3 repression (e.g., zinc-finger proteins
switch to repression by H3K27me3 while olfactory receptors,
glycoproteins, G protein coupled receptors and rhodospsin-like
GPCR superfamily members switch to repression by H3K9me3);
(4) most changes in gene expression due to 5-azacytidine treat-
ment are not due to relief of repression (mediated by DNA or
histone methylation).
Treatment with 5-azacytidine leads to changes in gene
expression. We identified a set of 1,424 genes whose expression
was altered by treatment with 5-azacytidine. One novel aspect of
performing a genome-wide analysis is the ability to ask if certain
functional classes of genes are enriched in the set of genes deregu-
lated by the drug treatment. Using this approach, we discovered
that the category of zinc-finger genes was upregulated and the
category of oxidoreductase-related genes was downregulated after
treatment of HEK 293 cells with 5-azacytidine.
Because our analysis of enriched gene categories produced
such striking results, we wanted to compare our studies to other
studies in the field. We searched the publicly available RNA
expression data in the online Gene Expression Omnibus data-
base (GEO; www.ncbi.nlm.nih.gov/geo/) and found only 2
entries for studies using 5-azacytidine. Other studies have used
5-aza-2'-deoxycytidine23-27 (decitabine), but the effects of this
drug are likely to be quite different.5 Decitabine is a deoxyribose,
which allows it to incorporate into DNA only. 5-azacytidine,
on the other hand, has a ribose attached to its ring structure,
which allows it to incorporate not only into DNA, but also into
 RNA. Incorporation into RNA can lead to secondary effects due
to inhibition of protein synthesis and consequently to cell toxic-
ity.28 However, the use of 5-azacytidine in treatments of various
myelodisplastic syndromes has been approved by the FDA and
it is critical to understand its effects on gene expression. We also
searched Pubmed (www.ncbi.nlm.nih.gov/pubmed/) and found
48 entries associated with the term azacytidine; again, most of
these publications present RNA expression data obtained with
cells treated with 5-aza-2'-deoxycytidine and those using 5-aza-
cytidine analyzed only one or several genes.29,30 Therefore, to our
knowledge, this work is the first analysis of whole-genome expres-
sion changes in treatment with 5-azacytidine. We also note that
many of the published studies used short treatment periods (2–4
days).23,24 We have found that a short treatment period (4 days)
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016
does not always result in the successful removal of DNA methyla-
tion (data not shown) and note that many of the previous studies
using short term treatments did not examine the efficiency of
removal of methylation at any target genes.
Most genes that show altered expression are not regulated
by promoters that display DNA methylation prior to the treat-
ment. DNA methylation has been long implicated in transcrip-
tional repression.31 Therefore, it would be reasonable to assume
that treatment of cells with 5-azacytidine would specifically
activate genes that are normally repressed by DNA methylation.
However, when we compared the top 2,000 ranked methylated
promoters from the control cells with RNA expression data in
the treated cells, we found that very few promoters of genes with
altered expression were methylated prior to the treatment with
5-azacytidine. Our choice to focus on the top 2,000 methy-
lated promoters was dictated by the enrichment values from
the array. However, we acknowledge that the top 2,000 cutoff
might be too stringent, especially in comparison to other stud-
ies in which methylated targets were identified as being only 1.1
times enriched over the reference sample.3,32 Therefore, we also
intersected the top 5,000 ranked methylated promoters with the
differentially expressed genes and found only a small increase in
the number of deregulated genes that had methylated promoters.
These results suggest that changes in gene expression were the
result of secondary effects, rather than direct promoter demethy-
lation leading to changes in gene expression. Interestingly, we
note that gene ontology analysis of the upregulated genes identi-
fied transcriptional regulation as a major category. It is possible
that changes in gene expression in the treated cells were achieved
as a two step process: in the beginning, a few transcription fac-
tors became demethylated and as a result were upregulated; these
transcription factors in turn, bound to promoters of other genes,
including other transcription factors, initiating a cascade of tran-
scriptional changes.
Treatment of cells with 5-azacytidine had major effects
on histone methylation. Histone methyltransferases such as
EZH2 (targets H3K27me3) and G9a (targets H3K9me3) have
been found in the same protein complexes as the DNA meth-
yltransferases DNMT1 and DNMT3a/3b. 8,9,11 Although it is
not completely understood how these complexes assemble on
the chromatin, inhibition of one member of the complex may
lead to destruction of the complex and thus affect multiple
www.landesbioscience.com Epigenetics 235
Figure 5. Western blot analysis of H3K9me3 and H3K27me3. HEK 293
cells were treated with acetic acid (control) or treated with 5-azacy-
tidine for 4 and 8 days; extracts from Ntera2 embryonal carcinoma
cells were used for comparison. Blots were probed with antibodies to
H3K9me3 or H3K27me3; Ponceau staining of the histones was used as a
loading control.
Table 1. Fold enrichment for H3K27me3, H3K9me3 and DNA methyla-
tion targets in control and 5-azacytidine treated cells
Median Median
Array Control* 5-azaC*
control^ 5-azaC^
H3K27me3 16.9–1.9 2.38 19.9–2.26 1.44
H3K9me3 23.4–1.9 2.25 9.9–2.0 2.30
5-meC 6.36–1.8 1.99 not done not done
*
shown are the fold enrichments (over total input values for the same
promoter) for the #1 and #2000 ranked promoters on each array.
^
shown are median fold enrichments.
types of repression.33 Therefore, we analyzed the effect of
5-azacytidine on two repressive histone marks, H3K9me3
and H3K27me3. We found that about 90% of the promot-
ers that showed these marks in control HEK 293 cells lost
the marks after treatment with 5-azacytidine. Further analysis
showed that these modifications were not generally lost from
the chromatin. In fact, treatment of cells with 5-azacytidine
or with 5-aza-2'-deoxycytidine (data not shown) resulted in an
increase in levels of histone H3 trimethylated on lysine 9 or
27. We found that different sets of promoters were repressed
by H3K27me3 and H3K9me3 in the treated vs. the control
cells. Interestingly, not only did we find that specific genes
show differences in silencing in the treated cells but we also
observed an unexpected switch in the functional categories
of genes repressed by the two types of modified histones.
For example, zinc-finger genes, which in our previous stud-
ies have been shown to be specifically bound by H3K9me3,
were bound by H3K27me3 in the treated cells. Similarly,
glycoproteins, which are normally repressed by H3K27me3,
were bound by H3K9me3 in the treated cells. We note that
analysis of cells treated for short time periods did not reveal
interchange of H3K27me3 and H3K9me3 targets; therefore
this “epigenetic switch” occurred only after long term and
efficient removal of DNA methylation from the chromatin.
H3K27me3 is known to spread along long stretches of chro-
matin 22,34 and to silence genes involved in development and
differentiation. It has been shown that H3K9me3 is a mark
of pericentrometic chromatin.35 Therefore, loss of these marks
could have severe consequences in cell biology. Future studies
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016

Figure 6A and B. Demethylation of DNA switches histone marks. Lists of the top ranked 2,000 targets on H3K27me3 promoter arrays from the control
and 5-azacytidine treated cells were analyzed using the DAVID gene ontology program18-19. Shown are the functional categories identified in lists
of H3K27me3 targets in control cells (A) and in cells treated with 5-azacytidine for 8 days (B). The x-axis shows relative percentage of genes in each
category with tick marks for each bar. Enrichment p value for each functional category is shown. The white bars indicate the gene categories bound
by H3K27me3 from control cells (panel A) that switch to H3K9me3 in treated cells (panel D); the black bars indicate the gene categories bound by
H3K9me3 in control cells (panel C) that switch to H3K27me3 in treated cells (panel B); the grey bars indicate the gene categories bound by H3K27me3
in control cells (panel A) that remain the same in treated cells (panel B).

are required to determine the significance of this disruption of
normal chromatin patterns on cellular differentiation and chro-
mosome integrity.
Most changes in gene expression due to 5-azacytidine treat-
ment are not due to relief of repression. Although we demon-
strated that 8 days of treatment with 5-azacytidine resulted in
efficient removal of methylated DNA from promoters and caused
major changes in the binding patterns of modified histones, we
also note that most genes that showed increased or decreased
transcript levels were not located in silenced chromatin in the
control cells. This suggests that the 5-azacytidine treatment is
mostly affecting genes located in active chromatin domains.
Future studies are required to determine if the changes in tran-
script levels are due to changes in transcription rate or to changes
in RNA stability caused by intercalation of the 5-azacytidine into
the transcripts.

236 Epigenetics Volume 5 Issue 3

Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016

Figure 6C and D. Demethylation of DNA switches histone marks. Lists of the top ranked 2,000 targets on H3K9me3 promoter arrays from the control
and 5-azacytidine treated cells were analyzed using the DAVID gene ontology program18-19. Shown are the functional categories identified in lists
of H3K9me3 targets in control cells (C) and in cells treated with 5-azacytidine for 8 days (D). The x-axis shows relative percentage of genes in each
category with tick marks for each bar. Enrichment p value for each functional category is shown. The white bars indicate the gene categories bound
by H3K27me3 from control cells (panel A) that switch to H3K9me3 in treated cells (panel D); the black bars indicate the gene categories bound by
H3K9me3 in control cells (panel C) that switch to H3K27me3 in treated cells (panel B).

Conclusion Materials and Methods

In summary, we found that 5-azacytidine treatment can cause
both loss of DNA methylation and alterations in target genes
bound by modified histones. However, most of the changes in
gene expression caused by the treatment with 5-azacytidine are
from genes whose promoter regions are not silenced by DNA
methylation or repressive histones. Rather, most changes (both
increases and decreases in expression) are from genes that reside
in active chromatin.
Cell culture. The human HEK 293 embryonal kidney cells were
obtained from Dr. David Segal’s laboratory at the Genome and
Biomedical Science Facility, UC Davis (ATCC cat# CRL-1573)
and grown in DMEM, 10% FBS, 2 mM L-glutamine and 1%
penicillin/streptomycin. The human Ntera2 embryonal car-
cinoma cells were obtained from ATCC (cat# CRL-1973) and
grown in Dulbecco’s Modified Eagle Medium supplemented
with 10% FBS, 2 mM L-glutamine and 1% penicillin/strep-
tomycin. The human HepG2 hepatocellular carcinoma cells

www.landesbioscience.com Epigenetics 237

 (DNMT1+/+) colorectal cell line was obtained
from the Cell Center of the Genetic Resources
Core Facility at John Hopkins University and
grown in McCoy’s 5A medium supplemented
with 10% FBS.
Treatment with 5-azacytidine. 5-azacytidine
was purchased from Sigma (cat # 2385) and dis-
solved in 50% acetic acid. HEK 293 cells were
treated with 5 uM 5-azacytidine dissolved in
medium for 4 or 8 days; medium was changed
daily. The control cells were treated with 50%
acetic acid solution for the same time periods. The
Figure 7. Global analysis of histone switching. The top 2000 H3K9me3 (A) and H3K27me3 amount of acetic acid was the same by volume as
(B) targets in control cells were identified and then both sets were analyzed for H3K9me3 used for 5-azacytidine solution.
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016

and H3K7me3 marks in the treated cells. Shown in (A) are the percentages of promoters
that either did not lose the H3K9me3 mark or that lost H3me3K9 and either did or did not
Analysis of overall DNA methylation lev-
gain H3K27me3. Shown in (B) are the percentages of promoters that either did not lose els. Global levels of DNA methylation were
the H3K27me3 mark or that lost H3K27me3 and either did or did not gain H3K9me3. analyzed using the Imprint Methylated DNA
Quantification kit (Sigma, cat # MDQ1) follow-
ing manufacturer’s instruction.
RNA expression arrays. Total RNA was
prepared from 5 x 106 cells using RNAeasy
Kit (QIAGEN) following the manufacturer’s
instructions. RNA quality was ensured using
the Agilent Systems Bioanalyzer. Biotinylated
RNA was prepared with the Illumina TotalPrep
RNA Amplification kit (Ambion) and hybrid-
ized to Illumina Human Ref-8 v3 Expression
BeadChips, which contain 24,500 well annotated
RefSeq transcripts. Hybridization signals were
detected with Illumina BeadArray Reader. RNA
labeling, amplification and array hybridization
was performed by the Expression Analysis Core
at UC Davis (http://genomecenter.ucdavis.edu/
expression_analysis/).
ChIP and MeDIP assays and amplicon
preparation. ChIP assays with 5 x 106 cells were
performed following the previously described pro-
tocol.36 The primary antibodies were as follows:
rabbit polyclonal H3K9me3 IgG (Abcam cat #
ab8898; 3 ug per ChIP assay) and rabbit poly-
clonal H3K27me3 IgG (Millipore cat # 07-449;
Figure 8. Analysis of repressive marks on the differentially regulated genes. The promot- 3 ug per ChIP assay). For MeDIP assays, genomic
ers of the genes that were differentially expressed after 4 or 8 days of treatment with
5-azacytidine were examined for DNA methylation and for H3K27me3 and H3K9me3.
DNA was extracted by shaking cells in digestion
Shown are the percentages of upregulated genes in 4 day treatment (A), downregulated buffer (100 mM NaCl, 10 mM TrisCl, pH 8, 25
genes in 4 day treatment (C), upregulated genes 8 day treatment (B) and downregulated mM EDTA, pH 8, 0.5% SDS) and 0.1 mg/ml
genes in 8 day treatment (D) whose promoters are bound by one of the repression marks Proteinase K for 12–18 hours at 50°C and purified
or by any combinations of the marks; also shown is the percentage of promoters that are using phenol-chlorophorm extraction method.
not bound by any of the repression marks.
Extracted DNA was sonicated to an average size
of 500 bp, denatured at 95°C for 10 min and
were obtained from ATCC (cat # CRL-10741) and grown in quickly chilled on ice. Immunoprecipitation of methylated DNA
Dulbecco’s Modified Eagle Medium supplemented with 10% was performed with 4 ug mouse monoclonal 5-Methylcytidine
FBS, 2 mM L-glutamine and 1% penicillin/streptomycin. The (Eurogentec cat# BI-MECY-0100) antibody per 4 ug of sonicated
human MCF7 breast adenocarcinoma cells were obtained from genomic DNA using the same buffers as for the ChIP assays.
ATCC (cat# HTB-22) and grown in Dulbecco’s Modified Eagle The secondary rabbit anti-mouse IgG was purchased from MP
Medium supplemented with 10% FBS, 2 mM L-glutamine and Biomedicals (cat # 55436). Incubation with primary antibody
1% penicillin/streptomycin. The human HCT116 parental was performed for 2 hours and for 1 hour with the secondary

238 Epigenetics Volume 5 Issue 3

 antibody. Antibody-DNA complexes were captured using Staph
A cells, then the DNA was washed, eluted and incubated with
Proteinase K. The non-specific rabbit IgG used as a negative con-
trol in the ChIP and MeDIP assays was purchased from Alpha
Diagnostics (cat # 20009-5). PCR analysis of ChIP samples and
amplicons preparation was performed as described in Komashko
et al.12 Primers used for ChIP-PCR and amplicons confirmations
are indicated in Supplementary Table 1.
Histone extraction and western blot. Histone extracts were
prepared following the Abcam extraction protocol. 2 ug of
histone extracts were separated using 15% Tris-glycine SDS-
polyacrylamide gel. The primary antibodies were used as follows:
H3K9me3 (Abcam, cat # ab8898; 1 ug per blot), H3K27me3
(Millipore, cat # 07-449; 0.5 ug per blot).
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016
ChIP-chip assays. Amplicons were applied to human UCSC
hg17 1.5 kb promoter array (a single array design containing
24,275 promoters). The labeling and hybridization of DNA
samples for ChIP-chip analysis was performed at UC Davis as
described in Komashko et al.12 Information about promoter
arrays can be found in Supplementary Table 1.
Data analysis. RNA hybridization signals were analyzed
using BeadStudio Gene Expression Module v.3.2.3 (Illumina).
To normalize gene expression data and determine differentially
expressed genes we used the R Project for Statistical Computing
v.2.7.0,37 (http://www.r-project.org/) and Bioconductor13 pack-
ages (http://www.bioconductor.org/) beadarray,15 nudge14,16 and
limma.16 Quantile method was used for normalization (beadar-
ray package). Raw and quantile normalized RNA expression
data can be found in Supplementary Table 2. Information about
differentially expressed genes can be found in Supplementary
Table 3.
Fluorescence intensity raw data were obtained from scanned
images of the oligonucleotide tiling arrays using NimbleScan
References
1. Robertson KD, Wolffe AP. DNA methylation in health
and disease. Nat Rev Genet 2000; 1:29-30.
2. Bird A. DNA methylation patterns and epigenetic
memory. Genes Dev 2002; 16:6-21.
3. Weber M, Hellmann I, Stadler MB, Ramos L, Paabo S,
Rebhan M, et al. Distribution, silencing potential and
evolutionary impact of promoter DNA methylation in
the human genome. Nat Genet 2007; 39:457-66.
4. Issa J-P. DNA methylation as a therapeutic target in
cancer. Clin Cancer Res 2007; 13:1634-7.
5. Issa JP, Kantarjian HM. Targeting DNA methylation.
Clin Cancer Res 2009; 15:3938-46.
6. Tuma RS. Epigenetic therapies move into new territory,
but how exactly do they work? J Natl Cancer Inst 2009;
101:1300-1.
7. Cedar H, Bergman Y. Linking DNA methylation and
histone modification: patterns and paradigms. Nat Rev
Genet 2009; 10:295-304.
8. Vire E, Brenner C, Deplus R, Blanchon L, Fraga MF,
Didelot C, et al. The polycomb group protein EZH2
directly controls DNA methylation. Nature 2006;
439:871-4.
9. Hernandez-Munoz I, Taghavi P, Kuijl C, Neefjes J,
van Lohuizen M. Association of BMI1 with polycomb
bodies is dynamic and requires PRC2/EZH2 and the
maintenance DNA methyltransferase DNMT1. Mol
Cell Biol 2005; 25:11047-58.
www.landesbioscience.com Epigenetics 239
2.2 extraction software (NimbleGen Systems). For each spot on
the array, log2 ratios of the Cy5-labeled test sample versus the
Cy3-labeled reference sample were calculated. Then, the biweight
mean of this log2 ratio was subtracted from each point; this pro-
cedure similar to mean-normalization of each channel. To sum-
marize binding regions and call peaks for each factor we used
the Maxfour peak calling method.20 Briefly, a value was assigned
based on the highest mean of four consecutive probes in each
promoter. Log2 enrichment values for each factor for control and
5-azacytidine 8 day treated cells can be found in Supplementary
Table 4.
RNA expression arrays and promoter arrays were matched
based on unique Entrez Gene IDs.
Functional annotations were performed using the pro-
gram Database for Annotation, Visualization and Integrated
Discovery (DAVID) 2007,18,19 (http://niaid.abcc.ncifcrf.gov/).
The same parameters were used for all analyses presented in this
study. These parameters were Gene Ontology (GO) Molecular
Function term, level 2; InterPro name is the Protein Domains
section; and SP_PIR_Keywords in the Functional Category sec-
tion. The EASE Score Threshold was set at 0.01.
Venn Diagrams were plotted using R package Vennerable.
Acknowledgements
We thank Dr. David Segal (Genome Center; UC Davis) for pro-
viding HEK 293 cells and members of the Farnham lab for help-
ful discussions; this work was supported in part by Public Health
Service grants CA45250, R21CA128471 and DK067889.
Note
Supplementary materials can be found at:
www.landesbioscience.com/supplement/
KomashkoEPI5-3-Sup.pdf
10. Li H, Rauch T, Chen Z-X, Szabo PE, Riggs AD, Pfeifer 17. Benjamini YaH Y. Controlling the false discovery rate:
GP. The histone methyltransferase SETDB1 and the a practical and powerful approach to multiple test-
DNA methyltransferase DNMT3A interact directly ing. Journal of the Royal Statistical Society Series B
and localize to promoters silenced in cancer cells. J Biol (Methodological) 1995; 289-300.
Chem 2006; 281:19489-500. 18. Dennis GJ, Sherman BT, Hosack Da, Yang J, Gao W,
11. Smallwood A, Esteve PO, Pradhan S, Carey M. Lane HC, et al. DAVID: Database for annotation,
Functional cooperation between HP1 and DNMT1 visualization and integrated discovery. Genome Biol
mediates gene silencing. Genes Dev 2007; 21:1169- 2003; 4:3.
78. 19. Huang da W, Sherman BT, Lempicki RA. Systematic
12. Komashko VM, Acevedo LG, Squazzo SL, Iyengar and integrative analysis of large gene lists using DAVID
SS, Rabinovich A, O’Geen H, et al. Using ChIP-chip bioinformatics resources. Nat Protoc 2009; 4:44-57.
technology to reveal common principles of transcrip- 20. Krig SR, Jin VX, Bieda MC, O’Geen H, Yaswen P,
tional repression in normal and cancer cells. Genome Green R, et al. Identification of genes directly regulated
Research 2008; 18:521-32. by the oncogene ZNF217 using chromatin immuno-
13. Gentleman RC, Carey VJ, Bates DM, Bolstad B, precipitation (ChIP)-chip assays. J Biol Chem 2007;
Dettling M, Dudoit S, et al. Bioconductor: open 282:9703-12.
software development for computational biol- 21. Yasui DH, Peddada S, Bieda MC, Vallero RO, Hogart
ogy and bioinformatics. Genome Biol 2004; 5: A, Nagarajan RP, et al. Integrated epigenomic analyses
80. of neuronal MeCP2 reveal a role for long-range interac-
14. Dean N, Raftery AE. Normal uniform mixture differ- tion with active genes. Proc Natl Acad Sci USA 2007;
ential gene expression detection for cDNA microarrays. 104:19416-21.
BMC Bioinformatics 2005; 6:173. 22. Squazzo SL, Komashko VM, O’Geen H, Krig S, Jin
15. Dunning MJ, Smith ML, Ritchie ME, Tavare S. VX, Jang S-W, et al. Suz12 silences large regions of
beadarray: R classes and methods for Illumina bead- the genome in a cell type-specific manner. Genome
based data. Bioinformatics 2007; 23:2183-4. Research 2006; 16:890-900.
16. Smyth GK. Linear models and empirical bayes meth- 23. Mueller W, Nutt CL, Ehrich M, Riemenschneider
ods for assessing differential expression in microar- MJ, von Deimling A, van den Boom D, et al.
ray experiments. Stat Appl Genet Mol Biol 2004; Downregulation of RUNX3 and TES by hypermethy-
3. lation in glioblastoma. Oncogene 2007; 26:583-93.
 24. Dannenberg LO, Edenberg HJ. Epigenetics of gene
expression in human hepatoma cells: expression profil-
ing the response to inhibition of DNA methylation and
histone deacetylation. BMC Genomics 2006; 7:181.
25. Kim H, Lapointe J, Kaygusuz G, Ong DE, Li C,
van de Rijn M, et al. The retinoic acid synthesis gene
ALDH1a2 is a candidate tumor suppressor in prostate
cancer. Cancer Res 2005; 65:8118-24.
26. Fandy TE, Herman JG, Kerns P, Jiemjit A, Sugar
EA, Choi SH, et al. Early epigenetic changes and
DNA damage do not predict clinical response in an
overlapping schedule of 5-azacytidine and entinostat
in patients with myeloid malignancies. Blood 2009;
114:2764-73.
27. Arai M, Yokosuka O, Hirasawa Y, Fukai K, Chiba T,
Imazeki F, et al. Sequential gene expression changes in
cancer cell lines after treatment with the demethylation
agent 5-aza-2'-deoxycytidine. Cancer 2006; 106:2514-
25.
28. Kihslinger JE, Godley LA. The use of hypomethylating
Downloaded by [b-on: Biblioteca do conhecimento online UP] at 05:54 07 June 2016
agents in the treatment of hematologic malignancies.
Leuk Lymphoma 2007; 48:1676-95.
29. Jahroudi N, Foster R, Price-Haughey J, Beitel G,
Gedamu L. Cell-type specific and differential regula-
tion of the human metallothionein genes. Correlation
with DNA methylation and chromatin structure. J Biol
Chem 1990; 265:6506-11.
240 Epigenetics Volume 5 Issue 3
30. Majid S, Dar AA, Ahmad AE, Hirata H, Kawakami
K, Shahryari V, et al. BTG3 tumor suppressor gene
promoter demethylation, histone modification and cell
cycle arrest by genistein in renal cancer. Carcinogenesis
2009; 30:662-70.
31. Miranda TB, Jones PA. DNA methylation: the nuts
and bolts of repression. J Cell Physiol 2007; 213:384-
90.
32. Fouse SD, Shen Y, Pellegrini M, Cole S, Meissner
A, Van Neste L, et al. Promoter CpG methylation
contributes to ES cell gene regulation in parallel with
Oct4/Nanog, PcG complex and histone H3 K4/K27
trimethylation. Cell Stem Cell 2008; 2:160-9.
33. Jin B, Tao Q, Peng J, Soo HM, Wu W, Ying J, et al.
DNA methyltransferase 3B (DNMT3B) mutations in
ICF syndrome lead to altered epigenetic modifications
and aberrant expression of genes regulating develop-
ment, neurogenesis and immune function. Hum Mol
Genet 2008; 17:690-709.
34. Bracken AP, Dietrich N, Pasini D, Hansen KH, Helin
K. Genome-wide mapping of Polycomb target genes
unravels their roles in cell fate transitions. Genes Dev
2006; 20:1123-36.
35. Rice JC, Briggs SD, Ueberheide B, Barber CM,
Shabanowitz J, Hunt DF, et al. Histone methyltrans-
ferases direct different degrees of methylation to define
distinct chromatin domains. Mol Cell 2003; 12:1591-
8.
36. Acevedo LG, Iniguez AL, Holster HL, Zhang X, Green
R, Farnham PJ. Genome-scale ChIP-chip analysis using
10,000 human cells. Biotechniques 2007; 43:791-7.
37. Team RDC. R: A Language and Environment for
Statistical Computing. Vienna, Austria: R Foundation
for Statistical Computing 2006.