Leukemia (2012) 26, 1804 - 1811

& 2012 Macmillan Publishers Limited All rights reserved 0887-6924/12

www.nature.com/leu

ORIGINAL ARTICLE
5-Azacitidine in aggressive myelodysplastic syndromes
regulates chromatin structure at PU.1 gene and cell
differentiation capacity
N Curik1,6, P Burda1,6, K Vargova1,6, V Pospisil1,6, M Belickova2, P Vlckova1, F Savvulidi1, E Necas1, H Hajkova2, C Haskovec2, J Cermak2,
M Krivjanska3, M Trneny1,2,4, P Laslo5,7, A Jonasova1,4,7 and T Stopka1,4,7

Epigenetic 5-azacitidine (AZA) therapy of high-risk myelodysplastic syndromes (MDS) and acute myelogenous leukemia (AML)
represents a promising, albeit not fully understood, approach. Hematopoietic transcription factor PU.1 is dynamically regulated
by upstream regulatory element (URE), whose deletion causes downregulation of PU.1 leading to AML in mouse. In this study a
signicant group of the high-risk MDS patients, as well as MDS cell lines, displayed downregulation of PU.1 expression within
CD34 cells, which was associated with DNA methylation of the URE. AZA treatment in vitro signicantly demethylated URE,
leading to upregulation of PU.1 followed by derepression of its transcriptional targets and onset of myeloid differentiation.
Addition of colony-stimulating factors (CSFs; granulocyte-CSF, granulocyte -- macrophage-CSF and macrophage-CSF) modulated
AZA-mediated effects on reprogramming of histone modications at the URE and cell differentiation outcome. Our data
collectively support the importance of modifying the URE chromatin structure as a regulatory mechanism of AZA-mediated
activation of PU.1 and induction of the myeloid program in MDS.

Leukemia (2012) 26, 1804 -- 1811; doi:10.1038/leu.2012.47

Keywords: PU.1; 5-azacitidine; MDS; AML; cytokines; differentiation

INTRODUCTION microRNA-155, which leads to a decreased protein synthesis of

Myeloid leukemia and myelodysplasia are considered a disorder of
a hematopoietic progenitor that is unable to initiate and complete
myeloid differentiation correctly, a process instructed by the
Ets-transcription factor PU.1.1 - 5 Under normal conditions PU.1
recognizes a purine-rich motif/s and recruits additional factors6
to dose-dependently guide hematopoietic differentiation and
lineage commitment at specic gene promoters.5,7 - 9 PU.1
autostimulates its own expression directly via an upstream
regulatory element (URE).10,11 URE is capable of loading the
RNA polymerase II onto the PU.1 promoter7,12,13 a process
preceded by histone modications (spanning more than 1 kb) at
the URE.7,10 Thus the histone code of the PU.1 gene represents
an important regulatory mechanism, which remains to be fully
understood, to specify different cell fates such as macrophages,
granulocytes and lymphocytes.14
Genetic or epigenetic changes in the PU.1 gene frequently
contribute to the pathogenesis of acute myelogenous leukemia
(AML). First, deletion of URE in mouse resulted in a decrease of
PU.1 to 20% and the onset of AML12 documenting that PU.1 is a
tumor suppressor. Second, mutations of PU.1 are associated with
murine radiation-induced acute leukemia15 or in rare cases of
human AML.16 - 18 Third, integration of a viral DNA of Spleen Focus
Forming Virus near URE in murine progenitors results in a state
where PU.1 is unable to become upregulated to develop myeloid
cells nor can it be silenced to develop red cells.4,19 - 22 Finally, PU.1
expression can be inhibited by a RNA interference mechanism via
PU.1 and a block of terminal myelopoiesis.23 Importantly, restoring
PU.1 levels in acute leukemia as well as other hematologic
malignancies induces cellular differentiation and cell cycle
arrest.4,24
Changes in URE that result in lowered PU.1 expression levels are
not only associated with AML pathology but also with multiple
myeloma where the DNA methylation status of URE correlates also
with increased tumor cell proliferation and shortened overall
patient survival.25 Treatment of multiple myeloma with DNA
demethylating drugs is capable of upregulating PU.1 and induce a
cell cycle arrest and apoptosis in myeloma cells in vitro.25
Interestingly, other epigenetic therapy such as retinoic acid has
been recently documented to upregulate PU.1 levels in acute
promyelocytic leukemia.26,27 Collectively, these observations
provide an important basis for developing epigenetic targets to
manipulate PU.1 expression levels within hematological disorders.
The use of DNA methyl transferase inhibitors (DNMTi) in
intermediate-2 (int-2)/high-risk MDS and some AML patients28 - 30
have doubled the overall survival with signicant improvement in
most of the clinical parameters (Fenaux et al.31,32 and reviewed in
Quintas-Cardama et al.33). AZA affects DNA methylation status of the
derepressed genes34 and may also involve histone modications.35 - 38
Epigenetic therapies may not simply block the tumor cell growth39
but induce cellular differentiation. Recent ndings indicated that
normal murine and human CD34 progenitor cells treated with
the DNMTi decitabine upregulated PU.1 expression by 150% and

1
First Faculty of Medicine, Institute of pathologic physiology, Charles University, Prague, Czech Republic; 2Institute of Hematology and Blood Transfusion, Prague, Czech Republic;
3
KRD and CEBIOS, Prague, Czech Republic; 4First Medical Department-- Hematology, General Faculty Hospital, Prague, Czech Republic and 5University of Leeds, Leeds, UK.
Correspondence: Dr T Stopka, First Faculty of Medicine, Institute Pathologic Physiology, Charles University, U Nemocnice 5, Prague 12853, Czech Republic.
E-mail: tstopka@lf1.cuni.cz
6
These authors contributed equally to this work.
7
These authors are senior co-authors.
Received 27 May 2011; revised 12 January 2012; accepted 9 February 2012; accepted article preview online 20 February 2012; advance online publication, 13 March 2012

induced morphologic signs and gene expression pattern of
terminal myeloid differentiation.40 DNMTi may cooperate with
colony-stimulating cytokines during induction of cell differentia-
tion. Pre-stimulation of normal progenitors with granulocyte --
colony-stimulating factor (G-CSF) prior the decitabine treatment
in vitro enhanced myeloid differentiation.40 Work acknowledged
in this paragraph paved the way to develop the approach to
increase PU.1 levels in MDS.
The understanding of how DNMTi function in MDS cells and
whether their effect may be further enhanced by growth factors is
incomplete. We herein present evidence that PU.1 expression is
decreased in a signicant proportion of int-2/high-risk MDS
patients. We also show that AZA caused epigenetic reprogram-
ming of the PU.1 gene resulting in the upregulation of PU.1
transcription and induction of cell differentiation. Effect of AZA is
modulated by colony-stimulating cytokines. Understanding the
prodifferentiation effects of AZA in MDS may add to a better
understanding of heterogeneous patient outcome following the
epigenetic therapy.
MATERIALS AND METHODS
Cell lines
Suspension cell lines (OCI-M2, SKM-1 all obtained from DSMZ, Braunschweig,
Germany) were derived from MDS patients following transformation to
AML.41 We initially determined two sublethal concentrations of AZA (1 and
5 mM, freshly prepared every time) and used them as reported elsewhere.34
Experimental design of AZA in vitro treatment was described previously40
and consists of three doses of 1 or 5 mM AZA every 24 h and alternatively
three doses of G-CSF 50 ng/ml (Neupogen, Amgen, Thousand Oaks, CA,
USA) or granulocyte - macrophage (GM)-CSF 50 ng/ml or macrophage
(M)-CSF 50 ng/ml (Sigma-Aldrich, St Louis, MO, USA, GM-CSF-- cat. no.
H5666-5 UG, lot: 041M1362; M-CSF-- cat. no. SRP4237-10 UG, lot: 40737) added
6 h prior AZA.
Patients
Since 2006 we obtained bone marrow (and peripheral blood) samples from
44 patients (in 54 independent samples) with higher-risk MDS (IPSS int-2
and high risk) and AML; 32 males and 12 females, median age 67 years
(range 57 -- 83 years). Diagnostic and prognostic evaluations were
performed according to WHO guidelines (http://www.nccn.com): RAEB 1
(5 patients), RAEB 2 (17 patients), AML (with trilineage dysplasia) with
20 - 30% blasts (12 patients), AML with trilineage dysplasia with 430% blasts
(9 patients) and 1 patient with erythroleukemia. All samples were collected
at the time of diagnosis and prior to any kind of treatment. All patients
signed informed consent in accordance with the Declaration of Helsinki
following institutional guidelines (see Supplementary Patient Material
for clinical data and parameters). The cell samples were separated by
Ficoll gradient (GE Healthcare, Pollards Wood, UK) followed by ow
cytometry and sorting (BD FACSAria IIu, BD Biosciences, San Jose, CA, USA) or
by magnetic sorting (AutoMACS, Miltenyi Biotec, Cologne, Germany) using
antibodies to CD34 (clone 581/CD34 from BD Biosciences) and CD11b
(clone M1/70 from BioLegend, San Diego, CA, USA). Cell culture of the cell
lines was performed according to the manufacturers recommendations;
primary bone marrow cells were immediately cultured in IMDM (Iscoves
Modied Dulbeccos Medium), 10% fetal bovine serum, non-essential amino
acids, antibiotics, stem cell factor 20 ng/ml.
RNA expression
Total RNA was isolated by RiboZol (Amresco, Solon, OH, USA); analyzed by
2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and reverse-
transcribed using High Capacity cDNA Reverse Transcription Kit with
specic primers. Quantitative PCR (9700HT instrument) was run for 40
cycles, 95 1C/15 s-- 60 1C/60 s following the initial denaturation step using
the TaqMan (Roche, Basel, Switzerland) probes. Ct-values of specic (s) and
control (c) amplicons served for calculation using 2(CTcCTs) equation.
Students t-test was also used for statistical analysis.
& 2012 Macmillan Publishers Limited
5-Azacitidine and PU.1 in MDS
N Curik et al
1805
Immunoblotting
Primary cells or cell lines (0.5 to 1  107) were lysed for 12 min in 200 ml
RIPA buffer (50 mM Tris-HCl, pH 8.0; 137 mM NaCl; 1% NP-40; 0.5% sodium
deoxycholate; 0.1% SDS; protease inhibitor cocktail P8340, Sigma) by
vortexing four times for 20 s on ice followed by light sonication (50%
amplitude, 3 cycles of 1,5 s pause in a cold room) on Branson Sonic
Dismembrator (model 500) equipped with a micro tip. Denatured cell
lysates (20 mg protein per lane) were resolved on 4 -- 12% gradient Bis-Tris
gel (NuPage; Life Technologies, Carlsbad, CA, USA). The gels were dry-
blotted by iBlot Gel Transfer System (Invitrogen). Membranes were blocked
by 7.5% nonfat milk in phosphate-buffered saline (0.1% Tween 20). Anti-
PU.1 [T-21] (1:600, sc-352; Santa Cruz Biotechnology, Santa Cruz, CA, USA)
antibody was used. Horseradish peroxidase-conjugated antibody was used
to visualize bands using ECL Plus Western Blotting Detection System (GE
Healthcare) on X-ray lms. Anti-actin horseradish peroxidase-conjugated
antibody (anti-actin [I-19], sc-1616; Santa Cruz) was used to determine
sample loading.
Chromatin immunoprecipitation
Cells (5  106) were processed as described previously.4 The following
antibodies were used: H3K9acetyl (Upstate, Lake Placid, NY, USA, 07-352),
H3K9trimethyl (Abcam, Cambridge, UK, 88-98), H3K4trimethyl (Diagenode,
Liege, Belgium, pAb-003-050), and a control antibody (EMB Biosciences,
San Diego, CA, USA, NI01). DNA from immunoprecipitates was measured
by PCR on the independent amplicons (Supplementary Figure 1) upstream
the PU.1 gene (17.5 to 9.7 kb relative to the PU.1 transcription start site).
These amplicons correspond to murine 14/15 kb URE enhancer (h-17.5
and h-15.9), 12 kb enhancer (h-13.4), 10 kb enhancer (h-11) and 8 kb
enhancer (h-9.7). Standard curves and DNA copy numbers were generated
for all reactions. Protein occupancy on DNA (percentage of input) is
dened as a copy number of a specic DNA fragment in each
immunoprecipitate compared with the copy number of that DNA fragment
within 1/100 input dilution used for immunoprecipitation (1% input DNA).
The control antibody values were subtracted from the values obtained
using the specic antibodies.
DNA methylation analysis
Genomic DNA from the cell lines was subjected to bisulte treatment using
an EpiTect bisulte kit (Qiagen, Venlo, The Netherlands). Bisulte-treated
DNA was then amplied with the primers 50 -GAGAAATGGTTTTTTTGT-
GATTT-30 and 50 -ACAACTACCCCTATTTCCACAT-30 , encoding the 404-bp
long region of PU.1 17-kb 50 upstream regulatory region and covering 19
CpGs of URE as published previously.25 Amplied product was separated
by agarose gel and puried by a QIAquick gel extraction kit (Qiagen). The
PCR fragments were subcloned into a pCR 2.1-TOPO-vector (Invitrogen),
then transfected into the chemically competent Escherichia coli strain
TOP10. In average, 10 recombinant colonies were randomly chosen
and plasmid DNA was isolated using Zyppy plasmid miniprep kit
(ZymoResearch, Irvine-Orange, CA, USA). DNA sequencing was performed
on a DNA sequencer ABI 3500 (Life Technologies) using a BigDye
terminator cycle sequencing chemistry (Life Technologies).
RESULTS
Expression and epigenetic status of PU.1 gene in MDS progenitors
and its response to AZA
Given the importance of PU.1 in AML we set to determine whether
its expression is dysregulated in other myeloid malignancies,
namely MDS (see Supplementary Table with clinical data). We
magnetically puried bone marrow CD34 population and
observed that in comparison with healthy volunteers the PU.1
mRNA level is quite heterogeneously distributed among MDS
patients ( 2 to 2 shown in log2 scale in Figure 1a). Levels of the
normalization factors (G6PD, HPRT1 or GAPDH) did not display
such uctuation. As lower levels of PU.1 transcripts correlate with
the onset of AML, we focused on the MDS patients that displayed
markedly lower PU.1 levels compared with normal controls.
Leukemia (2012) 1804 - 1811
 5-Azacitidine and PU.1 in MDS
N Curik et al

1806

(r=0.5)
a 3 b 12

0 6
SKM-1
-1
PU.1Low

-2 OCI-M2

-3 0
Ctrl MDS PU.1high/int. PU.1low

c 100

80

60

40

20

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 CpG

UNTREATED 2xAML1 AML 1PU.1 AML1/PU.1

5M AZA
Figure 1. (a) PU.1 mRNA expression in magnetically sorted CD34 cells. Dots represent TaqMan-based measurements in controls (left)
and int-2/high-risk MDS patients (right, normalized to G6PD, all relative to average of controls). Arrows indicate two MDS cell lines (SKM-1 and
OCI-M2). y axis is shown in log2 scale. MDS patients with low expression are placed within rectangle. (b) Median survival of MDS patients
(months) on AZA in two groups of patients with either intermediate/high (N 8) expression or with low PU.1 expression (N 9). (c) DNA
methylation analysis at PU.1 gene (at the URE) before (empty squares) and after treatment with AZA (5 mM, 72 h indicated by dark squares) in
the MDS OCI-M2 cells. Positions of transcription factor binding sites (for PU.1 and AML1) are indicated by triangles, the positions of CpGs 3 -- 6
and 13 -- 16 are highlighted.

The expression of PU.1 in primary MDS CD34 progenitors
allowed us to categorize the patients into two distinct groups with
either low (below average plus standard error of control CD34
cells) or int-2/high PU.1 expression. Despite limited number of the
patients (N 17) we observed signicant trend toward increased
survival on AZA in patients with PU.1 int-2/high levels (Figure 1b).
To study the mechanism behind downregulated PU.1 expres-
sion in MDS progenitors, we measured the DNA methylation
status at the URE (19 CpGs, approximately 17 kb region, see
Supplementary Figure 2B) within OCI-M2 cells. This cell line
represents an AML transformation of high-risk MDS and expresses
low PU.1 transcripts at levels similar to those measured within our
MDS-PU.1low population (Figure 1a). The URE in OCI-M2 is heavily
methylated with maximum at CpGs 3 -- 5 and 13 -- 16 (Supplemen-
tary Figure 2A). It should be noted that these CpGs neighbor the
PU.1 and AML1 binding sites required for PU.1 autoregulatory
stimulation.
To address whether the DNA methylation can be reduced by
DNMTi, we treated OCI-M2 cells with AZA (see Materials and
Methods section). AZA caused a dose-dependent decrease of DNA
methylation of URE (Figure 1c). It should be noted that within the
AML1 and PU.1 binding regions AZA treatment had little effect
upon DNA demethylation of the CpG 14, while other CpGs
11,13,15,16,19 of these binding regions are signicantly
demethylated.
Our data show that high-risk MDS patients underexpressing
PU.1 displayed shorter survival on AZA therapy and that in a
model cell line OCI-M2 low levels of PU.1 transcripts are associated
with high degree of DNA methylation at URE that can be
signicantly reduced by AZA treatment.
PU.1 mRNA and protein are induced by AZA in MDS cells resulting
in differentiation
To determine whether there is a direct correlation between
URE methylation and regulation of PU.1 transcripts, we studied
the effect of AZA treatment upon PU.1 expression. Indeed,
AZA treatment of OCI-M2 upregulated PU.1 at both mRNA
(Figure 2a) and protein (Figure 2b). Furthermore, AZA treat-
ment induced the transcriptional program of PU.1 target
genes with marked upregulation of CSF3R, EGR2, FOS, GELB,
CEBPA, CSF1R and MPO (Figure 2c). To determine whether
the induction of the PU.1 transcriptional network was suf-
cient to induce cellular differentiation, we used ow cytometry
analysis using antibody to CD11b. Indeed, AZA-induced protein
expression of CD11b (table attached below Figure 2a) was

Leukemia (2012) 1804 - 1811 & 2012 Macmillan Publishers Limited

 5-Azacitidine and PU.1 in MDS
N Curik et al

1807
a
b d e

Figure 2. PU.1 mRNA expression (a) in OCI-M2 at 72 h in response to AZA (final concentrations 1, 2 or 5 mM AZA in three doses during 72 h)
relative to average of HPRT1 and GAPDH expression. The % of CD11b-positive cells (in a table bellow) was measured by flow cytometry.
(b) Immunoblotting using antibodies to PU.1 (detecting a specific band of 37 kDa) of either untreated (N) or AZA-treated (5 mM, see Materials
and Methods section) OCI-M2 cells. (c) mRNA expression of PU.1 targets following the treatment of 1 mM AZA (heat map). (d) Cell growth of
OCI-M2 cells in response to AZA (on y axis: cell numbers are expressed as ratio of AZA-treated and untreated cells (1 mM AZA empty; 5 mM AZA
dark) for 3 days. The pie charts indicate simultaneously measured apoptosis in OCI-M2 (shown for 1 and 5 mM AZA). Gray color shows double-
negative cells for annexin V and propidium iodide. (e) Correlation analysis of DNA methylation (at CpG13 -- 16, see also SF2 for detail) and the
expression of PU.1 in int-2/high-risk MDS patients (X axis, dark dots, N 9).

associated with signicant inhibition of cell growth (Figure 2d).
Importantly, the levels of AZA did not induce signicant apoptosis
(Figure 2d).
On the basis of the observations from the OCI-M2 cells, we then
asked if levels of PU.1 transcripts within the CD34 MDS
progenitors correspond to the levels of DNA methylation at URE
of the same patients. Analysis of nine MDS patients (5 PU.1low and
4 PU.1int/high) showed a trend r 0.37 of negative correlation
between PU.1 expression and DNA methylation at URE (Figure 2e).
These data also indicated that OCI-M2 cells represent very unique
model that shows striking similarities with the primary progenitor
cells derived from int2/high-risk MDS patients.
This paragraph shows that PU.1 expression is related to
methylation status of URE and that AZA upregulates PU.1
coincidently with DNA demethylation of URE in MDS cells.
Myeloid colony-stimulating cytokines modulate AZA-mediated
effects on MDS cells via chromatin modications at URE
On the basis of a previous study in which G-CSF potentiated PU.1
upregulation and myeloid differentiation within normal human
CD34 progenitors,40 we asked if this treatment could be utilized
in MDS cells. We used the OCI-M2 cells as our model system and
included M-CSF and GM-CSF to determine whether the cell-stage
specicity of the MDS progenitors could be differentially sensitive
to any cytokine-mediated effects.
Interestingly, all cytokines enhanced signicantly AZA-mediated
DNA demethylation of URE (Supplementary Figure 2A). This was
associated with derepression of PU.1 (and its target EGR2) in the
cells co-treated with GM-CSF and M-CSF (Figure 3a). Notably,
G-CSF mildly reduced the AZA-mediated PU.1 induction coinciding
with a dampering of the EGR2 response. Despite both M-CSF and
GM-CSF augmenting the AZA effect, only in the GM-CSF/AZA-
treated cells was the cell differentiation effectively induced as
demonstrated by expression of CD11b using ow cytometry
(Figure 3b) coincidently with a block of cell proliferation
(Figure 3c). The apoptosis did not signicantly differ between
cells either treated with AZA or AZA in combination with the
cytokines.
In order to understand the consequences that AZA and the
cytokines treatments have on the URE, we have performed
detailed analysis of histone modications at URE in the OCI-M2
(Figure 3d). We have used amplicons at URE and other enhancers;
however, for a concise presentation of the data only three
amplicons are shown (17.5, 16.6 and 15.9 kb upstream the
PU.1 gene start site). Chromatin immunoprecipitation analysis of
an active chromatin mark, histone H3 lysine (K)4 triMethylation
(H3K4Me3), indicated its increase at URE following AZA exposure.
The combination of AZA with GM-CSF or M-CSF did not
signicantly change the H3K4Me3 prole at the URE, while with
G-CSF caused a decrease compared with AZA alone (Figure 3d, left
panel). Next, we investigated another active mark H3K9 acetyla-
tion (H3K9Ac) at PU.1 gene by chromatin immunoprecipitation.
Interestingly, while AZA/G-CSF combination caused increase of
H3K9Ac downstream URE (at 15.9 amplicon), the AZA/GM-CSF
combination appeared to increase H3K9Ac at the URE (16.6 kb)
and this was associated with the most prominent effect on PU.1
level (Figure 4d, right panel).
Thus, use of colony-stimulating cytokines preceding addition of
AZA to MDS cells facilitated changes in DNA demethylation as well
as histone modications at URE resulting in more efcient PU.1
derepression.
Primary MDS progenitors respond to AZA by inducing myeloid
differentiation
Our data implicate AZA as a potent agent in restoring the myeloid
differentiation capacity of MDS CD34 progenitors. To explore
this further, we have studied int-2/high-risk MDS patients in detail
on both the chromatin and expression levels (see Supplementary
Table) and determined in vitro response to AZA in the presence or
absence of colony-stimulating cytokines. G-CSF was used as it was
shown to potentiate expression of PU.1 by DNMTi40 and is
routinely used in clinical hematology. For such studies we have
used ex vivo isolated bone marrow-derived progenitors of the
patient P302, a 67-year-old male with RAEB II (WHO), IPSS-Int2, who
presented with anemia and transfusion dependency of 4 TU/Mo,
neutropenia and 10% blasts in bone marrow (phenotypically
characterized as CD33 CD34 CD117 HLADR CD45 that
were also detectable in peripheral blood by ow cytometry).
Additionally, we also studied patients P301 (76-year-old male, with
RAEB II) and P299 (69-year-old male with RAEB-T). All three

& 2012 Macmillan Publishers Limited Leukemia (2012) 1804 - 1811

 5-Azacitidine and PU.1 in MDS
N Curik et al

1808
a
c

Figure 3. (a) mRNA expression of PU.1 and EGR2 in OCI-M2 cell line treated by AZA (5 mM) with 6-h preincubation (before each of the three
additions of AZA) with the colony-stimulating cytokines (G-CSF, GM-CSF, M-CSF, as indicated bellow the x axis, see Materials and Methods
section for detail). The expression was measured at 72 h, data were normalized to average of HPRT1 and GAPDH expression, error bars
indicate s.e.m. of two independent experiments. (b) The numbers in table indicate % of CD11b-positive cells measured by flow cytometry
(at day 6) corresponding to the cell samples analyzed above. (c) Cell growth of OCI-M2 in response to AZA (5 mM) and the indicated cytokines.
Cell numbers on y axis represent AZA G-CSF (empty), AZA GM-CSF (gray) and AZA M-CSF (dark)-treated cells normalized to untreated
cells during the 3 day experiment. The pie charts indicate apoptosis in OCI-M2. Gray color indicates cells double negative for annexin V and
propidium iodide. (d) Chromatin immunoprecipitation analysis using two histone modifications (acetylation of histone H3K9 and trimethy-
lation of H3K4) at the three independent amplicons alongside the URE (B17 kb upstream from the transcription start site, for detail see
Supplementary Figure 1). The X axis shows different treatment of the OCI-M2 cell lines (None, 5 mM AZA, and AZA with the indicated
preincubation with cytokines). Y axis indicates relative occupancy of the histone H3 modificationss.e.m. Nonspecific signal for each cell
culture is shown by empty bars.

patients expressed intermediate levels of PU.1 within CD34
progenitors as compared with controls (data not shown). Despite
expressing intermediate levels of PU.1, the CD34 cells displayed
some marks of DNA methylation at URE: P302 displayed DNA
methylation within the CpGs 3 -- 6 and at three other CpGs
(Figure 4a). Importantly, PU.1 transcript expression (in P302) was
responsive to AZA treatment and this effect was enhanced by
preincubation with myeloid colony-stimulating cytokine G-CSF
(Figure 4b). PU.1 target gene CSF1R was also upregulated by AZA
and AZA/G-CSF (Figure 4c). Importantly, incubation of MDS cells
with AZA and G-CSF signicantly downregulated expression of KIT,
a known stem cell marker (Figure 4c). This phenomenon was also
observed in the OCI-M2 cells (Figure 4d) again demonstrating
similarity of the MDS patients and the OCI-M2 cells. The ow
cytometry experiments indicated that the patient P302 decreased
CD34 positivity and gained CD11b positivity following AZA/G-CSF
(Figure 4e). We have observed enhanced CD11b expression
following in vitro AZA/G-CSF treatment also in the patients P299 and
P301 (data not shown). Chromatin immunoprecipitation indicates
that at the URE the AZA treatment (and this was potentiated by
G-CSF) signicantly changed the histone modication pattern:
it increased H3K9Ac and decreased H3K9Me3 coincidently with
PU.1 gene derepression (Figure 4f).
Activation of PU.1 gene by AZA involves also other regulatory
regions in addition to URE
To determine whether AZA-mediated upregulation of PU.1 is
solely dependent upon the URE, we utilized URE/ mutant
mice.12 As previously described these mice develop AML due to
the dysregulation of PU.1 expression. AML progenitors derived
from bone marrow were magnetically sorted for c-kit positivity as

Leukemia (2012) 1804 - 1811 & 2012 Macmillan Publishers Limited

 5-Azacitidine and PU.1 in MDS
N Curik et al

1809
a b

c d e f

Figure 4. (a) DNA methylation analysis and PU.1 mRNA expression on the CD34 cells derived from MDS patient P302 (see Supplementary
Table for detail) and OCI-M2 cells. (b) Primary ex vivo isolated cells from MDS patient (P302) treated by AZA or AZA G-CSF. PU.1 mRNA
expression was measured in primary bone marrow mononuclear cells (empty bars) and in bone marrow-derived CD34 cells (gray bars).
(c) CSF1R (empty) and KIT (dark) mRNA expression following the treatment with AZA or AZA G-CSF in bone marrow-derived CD34 cells
(P302). (d) KIT mRNA expression in AZA- and AZA/G-CSF-treated OCI-M2 cells. (e) Flow cytometry analyses (with indicated CD34 and CD11b
positivity) in AZA and AZA/G-CSF-treated CD34 cells from the P302 MDS patient. (f ) ChIP analysis of H3K9Ac (empty) and H3K9Me3 (dark)
at URE in the primary (P302) CD11b bone marrow cells.

Figure 5. c-kit-positive cells ex vivo isolated from either murine AML (left) produced in a mouse model lacking URE (PU.1low) as described by
Rosenbauer et al.12 or control mice (right) were subject to reverse transcription -- PCR analyses following AZA (1 and 5 mM) treatment for 72 h in
three doses (Materials and Methods). Expression of PU.1, Egr2 and Gfi1 at 72 h is relative to Gapdh mRNA determined in the same samples.
Fold change relative to untreated cells is indicated on top of the columns.

described previously.12 The progenitor cells were in vitro treated
by AZA (as outlined above) and mRNA of PU.1 and its targets
and control genes were subsequently determined (Figure 5).
Our data conrmed that PU.1 levels in blast-AML of URE/ mice
are approximately vefold downregulated as compared with
the control c-kit-positive normal bone marrow progenitor cells. In
the absence of the URE, the PU.1 transcripts were induced in a
dose-dependent manner by AZA treatment in a similar relative
fold change as in control progenitors. The levels of PU.1 were
reected by levels of PU.1 targets Egr2 and G1 in the same
samples.
These data indicate that even though URE is required for
expression of physiological levels of PU.1 it is not exclusively
required for AZA-mediated upregulation of PU.1 levels in AML.
Thus, effect of AZA treatment is in addition of URE dependent on
another enhancer regions or the promoter itself.
DISCUSSION
Current therapeutic treatment of hematologic malignancies
focuses on the need to induce apoptosis and/or cycle arrest of
the malignant cell population. However, the ability to restore and
induce cellular differentiation represents an alternate strategy,
and has potential to be very helpful in the treatment of diseases
that are often refractory to standard chemotherapy regimens.
We herein present new evidence that gene encoding a key
hematopoietic transcription factor PU.1 is a target of chromatin
modications mediated by AZA in MDS.

& 2012 Macmillan Publishers Limited Leukemia (2012) 1804 - 1811

 5-Azacitidine and PU.1 in MDS
N Curik et al
1810
We initially determined expression of PU.1 in CD34 cells of
MDS patients with higher risk and found that among patients with
signicantly shorter survival on AZA are patients with low
expression of PU.1 (Figure 1). In addition, expression of PU.1
negatively correlated with DNA methylation at URE. Presence of a
relatively stable repressive mark such as DNA methylation and
H3K9Me3 at URE in the higher risk MDS progenitors indicates that
repression of URE may represent an important barrier of PU.1
autoregulation mechanism and that it can be reverted by AZA.
Specically, our data strongly indicate that MDS progenitors with
heavily repressed URE would require higher dose of AZA and
those that display relatively open URE chromatin may require
lower dose of AZA. Importantly, AZA in order to activate PU.1 gene
expression may target additional regulatory regions, other than
URE (Figure 5). Therefore, more patient data and analyses of the
PU.1 upstream regions are however needed to fully comprehend
the relationship between the epigenetic status of PU.1 gene and
clinical responsiveness to AZA.
Alternatively to recommending different AZA levels, we
searched for mechanism that would enhance AZA-mediated
effects on PU.1 gene and focused on colony-stimulating cytokines
such as G-CSF. It is important to note that additional chromatin
modiers may be required in order to achieve PU.1 gene
reexpression outcome following DNA demethylation in MDS.42
G-CSF was shown to facilitate repression of stem cell genes with or
without its direct effect to induce PU.1 expression as reported
elsewhere.40 Others reported that addition of DNMTi to normal
murine progenitors preceded by G-CSF (or transgenic activation of
PU.1) upregulated PU.1 levels and its program of myeloid
differentiation.40 Our data in primary MDS progenitors
(Figure 4c) and OCI-M2 cells (Figure 4d) indicated that indeed G-
CSF inhibited AZA-mediated induction of KIT expression coin-
cident with enhanced myeloid differentiation. Our data also
indicate that G-CSF (and also GM-CSF and M-CSF) preincubation
prior AZA is capable to inuence both DNA methylation (Figure 1c
and Supplementary Figure 2) and chromatin modications at URE
(Figures 3 and 4), is associated with increased PU.1 mRNA levels
and may be mediated by G-CSF signaling pathway that includes
PU.1. It should be noted that AZA alone and importantly together
with G-CSF is capable to recapitulate its effects observed in the
MDS cell line, also in ex vivo isolated primary high-risk MDS cells.
G-CSF represents a cytokine that is frequently used in MDS and
AML and other hematological malignancies to prevent neutro-
penic complications following chemotherapy.43 Its use in MDS is
not associated with increased frequency of developing AML44,
thus it represents promising candidate with a potential to
augment AZA prodifferentiation effects.
We observed that some patients whose tumor cells expressed
higher PU.1 level displayed favorable therapeutic response;
such observation will require extended follow-up and analyses
of additional patients. If this scenario is however correct,
further assessment of molecular response to AZA and AZA/
G-CSF will be crucial in predicting responsiveness to AZA,
potential of AML transformation and other phenotypic features
of high-risk MDS. Responsiveness of MDS cells to G-CSF
pretreatment prior to AZA in vitro may provide rationale for
further evaluating the role of G-CSF in addition to epigenetic
therapy in MDS going forward.
CONFLICT OF INTEREST
The authors declare no conict of interest.
ACKNOWLEDGEMENTS
The major support of this work comes from the grant (investigators TS and MK) of the
Ministry of Industry and Trade (FR-TI2/509). Other support is from Grants NS10310-3/
2009, 2B06077, MSM 0021620806 PRVOUK, MSM 0021620808, LC06044, SVV-2011-
Leukemia (2012) 1804 - 1811
262507 and SVV-2012, GAUK 251135 82210, GACR P305/12/1033, UNCE 204021.
Collaborators support comes from the Grants NS10632-3/2009, NS9634-4/2008, and
Yorkshire Cancer Research (PL). We thank Drs Ulrich Steidl and Daniel Tenen for
providing us with the mouse model of PU.1low AML. We are very thankful to all MDS
patients participating in this study.
REFERENCES
1 Laslo P, Spooner CJ, Warmash A, Lancki DW, Lee HJ, Sciammas R et al.
Multilineage transcriptional priming and determination of alternate hematopoie-
tic cell fates. Cell 2006; 126: 755 - 766.
2 Dahl R, Simon MC. The importance of PU.1 concentration in hemato-
poietic lineage commitment and maturation. Blood Cells Mol Dis 2003; 31:
229 - 233.
3 Stopka T, Amanatullah DF, Papetti M, Skoultchi AI. PU.1 inhibits the erythroid
program by binding to GATA-1 on DNA and creating a repressive chromatin
structure. EMBO J 2005; 24: 3712 - 3723.
4 Burda P, Curik N, Kokavec J, Basova P, Mikulenkova D, Skoultchi AI et al. PU.1
activation relieves GATA-1-mediated repression of Cebpa and Cbfb during
leukemia differentiation. Mol Cancer Res 2009; 7: 1693 - 1703.
5 Ross IL, Yue X, Ostrowski MC, Hume DA. Interaction between PU.1 and another Ets
family transcription factor promotes macrophage-specic Basal transcription
initiation. J Biol Chem 1998; 273: 6662 - 6669.
6 Yamamoto H, Kihara-Negishi F, Yamada T, Hashimoto Y, Oikawa T. Physical and
functional interactions between the transcription factor PU.1 and the coactivator
CBP. Oncogene 1999; 18: 1495 - 1501.
7 Hoogenkamp M, Krysinska H, Ingram R, Huang G, Barlow R, Clarke D et al. The
Pu.1 locus is differentially regulated at the level of chromatin structure and
noncoding transcription by alternate mechanisms at distinct developmental
stages of hematopoiesis. Mol Cell Biol 2007; 27: 7425 - 7438.
8 Petrovick MS, Hiebert SW, Friedman AD, Hetherington CJ, Tenen DG, Zhang DE.
Multiple functional domains of AML1: PU.1 and C/EBPalpha synergize with
different regions of AML1. Mol Cell Biol 1998; 18: 3915 - 3925.
9 Wontakal SN, Guo X, Will B, Shi M, Raha D, Mahajan MC et al. A large gene
network in immature erythroid cells is controlled by the myeloid and B cell
transcriptional regulator PU.1. PLoS Genet 2011; 7: e1001392.
10 Okuno Y, Huang G, Rosenbauer F, Evans EK, Radomska HS, Iwasaki H et al.
Potential autoregulation of transcription factor PU.1 by an upstream regulatory
element. Mol Cell Biol 2005; 25: 2832 - 2845.
11 Chen H, Ray-Gallet D, Zhang P, Hetherington CJ, Gonzalez DA, Zhang DE et al.
PU.1 (Spi-1) autoregulates its expression in myeloid cells. Oncogene 1995; 11:
1549 - 1560.
12 Rosenbauer F, Wagner K, Kutok JL, Iwasaki H, Le Beau MM, Okuno Y et al. Acute
myeloid leukemia induced by graded reduction of a lineage-specic transcription
factor, PU.1. Nat Genet 2004; 36: 624 - 630.
13 Steidl U, Steidl C, Ebralidze A, Chapuy B, Han HJ, Will B et al. A distal single
nucleotide polymorphism alters long-range regulation of the PU.1 gene in acute
myeloid leukemia. J Clin Invest 2007; 117: 2611 - 2620.
14 Leddin M, Perrod C, Hoogenkamp M, Ghani S, Assi S, Heinz S et al. Two distinct
auto-regulatory loops operate at the PU.1 locus in B cells and myeloid cells. Blood
2011; 117: 2827 - 2838.
15 Cook WD, McCaw BJ, Herring C, John DL, Foote SJ, Nutt SL et al. PU.1 is a
suppressor of myeloid leukemia, inactivated in mice by gene deletion and
mutation of its DNA binding domain. Blood 2004; 104: 3437 - 3444.
16 Mueller BU, Pabst T, Osato M, Asou N, Johansen LM, Minden MD et al.
Heterozygous PU.1 mutations are associated with acute myeloid leukemia. Blood
2002; 100: 998 - 1007.
17 Lamandin C, Sagot C, Roumier C, Lepelley P, De Botton S, Cosson A et al. Are PU.1
mutations frequent genetic events in acute myeloid leukemia (AML)? Blood 2002;
100: 4680 - 4681.
18 Vegesna V, Takeuchi S, Hofmann WK, Ikezoe T, Tavor S, Krug U et al. C/EBP-beta,
C/EBP-delta, PU.1, AML1 genes: mutational analysis in 381 samples of
hematopoietic and solid malignancies. Leuk Res 2002; 26: 451 - 457.
19 Rekhtman N, Radparvar F, Evans T, Skoultchi AI. Direct interaction of
hematopoietic transcription factors PU.1 and GATA-1: functional antagonism in
erythroid cells. Genes Dev 1999; 13: 1398 - 1411.
20 Nerlov C, Querfurth E, Kulessa H, Graf T. GATA-1 interacts with the myeloid PU.1
transcription factor and represses PU.1-dependent transcription. Blood 2000; 95:
2543 - 2551.
21 Zhang P, Zhang X, Iwama A, Yu C, Smith KA, Mueller BU et al. PU.1 inhibits GATA-1
function and erythroid differentiation by blocking GATA-1 DNA binding. Blood
2000; 96: 2641 - 2648.
22 Moreau-Gachelin F, Tavitian A, Tambourin P. Spi-1 is a putative oncogene in virally
induced murine erythroleukaemias. Nature 1988; 331: 277 - 280.
& 2012 Macmillan Publishers Limited

23 OConnell RM, Rao DS, Chaudhuri AA, Boldin MP, Taganov KD, Nicoll J et al.
Sustained expression of microRNA-155 in hematopoietic stem cells causes a
myeloproliferative disorder. J Exp Med 2008; 205: 585 - 594.
24 Durual S, Rideau A, Ruault-Jungblut S, Cossali D, Beris P, Piguet V et al. Lentiviral
PU.1 overexpression restores differentiation in myeloid leukemic blasts. Leukemia
2007; 21: 1050 - 1059.
25 Tatetsu H, Ueno S, Hata H, Yamada Y, Takeya M, Mitsuya H et al. Down-regulation
of PU.1 by methylation of distal regulatory elements and the promoter is required
for myeloma cell growth. Cancer Res 2007; 67: 5328 - 5336.
26 Mueller BU, Pabst T, Fos J, Petkovic V, Fey MF, Asou N et al. ATRA resolves the
differentiation block in t(15;17) acute myeloid leukemia by restoring PU.1
expression. Blood 2006; 107: 3330 - 3338.
27 Gu ZM, Liu CX, Wu SF, Zhao M, Xu HZ, Liu W et al. PU.1 directly regulates retinoic
acid-induced expression of RIG-G in leukemia cells. FEBS Lett 2010; 585: 375 - 380.
28 Silverman LR, McKenzie DR, Peterson BL, Holland JF, Backstrom JT, Beach CL et al.
Further analysis of trials with azacitidine in patients with myelodysplastic
syndrome: studies 8421, 8921, and 9221 by the Cancer and Leukemia Group B.
J Clin Oncol 2006; 24: 3895 - 3903.
29 Fenaux P, Mufti GJ, Hellstrom-Lindberg E, Santini V, Finelli C, Giagounidis A et al.
Efcacy of azacitidine compared with that of conventional care regimens in the
treatment of higher-risk myelodysplastic syndromes: a randomised, open-label,
phase III study. Lancet Oncol 2009; 10: 223 - 232.
30 Silverman LR, Demakos EP, Peterson BL, Kornblith AB, Holland JC, Odchimar-
Reissig R et al. Randomized controlled trial of azacitidine in patients with the
myelodysplastic syndrome: a study of the cancer and leukemia group B. J Clin
Oncol 2002; 20: 2429 - 2440.
31 Fenaux P, Gattermann N, Seymour JF, Hellstrom-Lindberg E, Mufti GJ, Duehrsen U
et al. Prolonged survival with improved tolerability in higher-risk myelodysplastic
syndromes: azacitidine compared with low dose ara-C. Br J Haematol 2010; 149:
244 - 249.
32 Fenaux P, Bowen D, Gattermann N, Hellstrom-Lindberg E, Hofmann WK,
Pfeilstocker M et al. Practical use of azacitidine in higher-risk myelodysplastic
syndromes: an expert panel opinion. Leuk Res 2010; 34: 1410 - 1416.
33 Quintas-Cardama A, Santos FP, Garcia-Manero G. Therapy with azanucleosides for
myelodysplastic syndromes. Nat Rev Clin Oncol 2010; 7: 433 - 444.
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
& 2012 Macmillan Publishers Limited
5-Azacitidine and PU.1 in MDS
N Curik et al
1811
34 Komashko VM, Farnham PJ. 5-Azacytidine treatment reorganizes genomic histone
modication patterns. Epigenetics 2010; 5: 229--240.
35 Vire E, Brenner C, Deplus R, Blanchon L, Fraga M, Didelot C et al. The Polycomb
group protein EZH2 directly controls DNA methylation. Nature 2006; 439:
871 - 874.
36 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 - 11058.
37 Li H, Rauch T, Chen ZX, Szabo PE, Riggs AD, Pfeifer GP. The histone
methyltransferase SETDB1 and the DNA methyltransferase DNMT3A interact
directly and localize to promoters silenced in cancer cells. J Biol Chem 2006; 281:
19489 - 19500.
38 Smallwood A, Esteve PO, Pradhan S, Carey M. Functional cooperation between
HP1 and DNMT1 mediates gene silencing. Genes Dev 2007; 21: 1169 - 1178.
39 Jiemjit A, Fandy TE, Carraway H, Bailey KA, Baylin S, Herman JG et al. p21(WAF1/
CIP1) induction by 5-azacytosine nucleosides requires DNA damage. Oncogene
2008; 27: 3615 - 3623.
40 Hu Z, Negrotto S, Gu X, Mahfouz R, Ng KP, Ebrahem Q et al. Decitabine maintains
hematopoietic precursor self-renewal by preventing repression of stem cell genes
by a differentiation-inducing stimulus. Mol Cancer Ther 2010; 9: 1536 - 1543.
41 Drexler HG, Dirks WG, Macleod RA. Many are called MDS cell lines: one is chosen.
Leuk Res 2009; 33: 1011 - 1016.
42 Si J, Boumber YA, Shu J, Qin T, Ahmed S, He R et al. Chromatin remodeling is
required for gene reactivation after decitabine-mediated DNA hypomethylation.
Cancer Res 2010; 70: 6968 - 6977.
43 Dotti G, Stella CC, Mangoni L, Cottafavi L, Caramatti C, Almici C et al. Granulocyte
colony-stimulating factor (G-CSF) prevents dose-limiting neutropenia in lympho-
ma patients receiving standard dose chemotherapy. Haematologica 1995; 80:
142 - 145.
44 Estey E, Thall P, Andreeff M, Beran M, Kantarjian H, OBrien S et al. Use of
granulocyte colony-stimulating factor before, during, and after udarabine
plus cytarabine induction therapy of newly diagnosed acute myelogenous
leukemia or myelodysplastic syndromes: comparison with udarabine plus
cytarabine without granulocyte colony-stimulating factor. J Clin Oncol 1994; 12:
671 - 678.
Leukemia (2012) 1804 - 1811