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

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

& 2012 Macmillan Publishers Limited 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 11,13,15,16,19 of these binding regions are signicantly
allowed us to categorize the patients into two distinct groups with demethylated.
either low (below average plus standard error of control CD34 Our data show that high-risk MDS patients underexpressing
cells) or int-2/high PU.1 expression. Despite limited number of the PU.1 displayed shorter survival on AZA therapy and that in a
patients (N 17) we observed signicant trend toward increased model cell line OCI-M2 low levels of PU.1 transcripts are associated
survival on AZA in patients with PU.1 int-2/high levels (Figure 1b). with high degree of DNA methylation at URE that can be
To study the mechanism behind downregulated PU.1 expres- signicantly reduced by AZA treatment.
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 PU.1 mRNA and protein are induced by AZA in MDS cells resulting
represents an AML transformation of high-risk MDS and expresses in differentiation
low PU.1 transcripts at levels similar to those measured within our To determine whether there is a direct correlation between
MDS-PU.1low population (Figure 1a). The URE in OCI-M2 is heavily URE methylation and regulation of PU.1 transcripts, we studied
methylated with maximum at CpGs 3 -- 5 and 13 -- 16 (Supplemen- the effect of AZA treatment upon PU.1 expression. Indeed,
tary Figure 2A). It should be noted that these CpGs neighbor the AZA treatment of OCI-M2 upregulated PU.1 at both mRNA
PU.1 and AML1 binding sites required for PU.1 autoregulatory (Figure 2a) and protein (Figure 2b). Furthermore, AZA treat-
stimulation. ment induced the transcriptional program of PU.1 target
To address whether the DNA methylation can be reduced by genes with marked upregulation of CSF3R, EGR2, FOS, GELB,
DNMTi, we treated OCI-M2 cells with AZA (see Materials and CEBPA, CSF1R and MPO (Figure 2c). To determine whether
Methods section). AZA caused a dose-dependent decrease of DNA the induction of the PU.1 transcriptional network was suf-
methylation of URE (Figure 1c). It should be noted that within the cient to induce cellular differentiation, we used ow cytometry
AML1 and PU.1 binding regions AZA treatment had little effect analysis using antibody to CD11b. Indeed, AZA-induced protein
upon DNA demethylation of the CpG 14, while other CpGs 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). detailed analysis of histone modications at URE in the OCI-M2
Importantly, the levels of AZA did not induce signicant apoptosis (Figure 3d). We have used amplicons at URE and other enhancers;
(Figure 2d). however, for a concise presentation of the data only three
On the basis of the observations from the OCI-M2 cells, we then amplicons are shown (17.5, 16.6 and 15.9 kb upstream the
asked if levels of PU.1 transcripts within the CD34 MDS PU.1 gene start site). Chromatin immunoprecipitation analysis of
progenitors correspond to the levels of DNA methylation at URE an active chromatin mark, histone H3 lysine (K)4 triMethylation
of the same patients. Analysis of nine MDS patients (5 PU.1low and (H3K4Me3), indicated its increase at URE following AZA exposure.
4 PU.1int/high) showed a trend r 0.37 of negative correlation The combination of AZA with GM-CSF or M-CSF did not
between PU.1 expression and DNA methylation at URE (Figure 2e). signicantly change the H3K4Me3 prole at the URE, while with
These data also indicated that OCI-M2 cells represent very unique G-CSF caused a decrease compared with AZA alone (Figure 3d, left
model that shows striking similarities with the primary progenitor panel). Next, we investigated another active mark H3K9 acetyla-
cells derived from int2/high-risk MDS patients. tion (H3K9Ac) at PU.1 gene by chromatin immunoprecipitation.
This paragraph shows that PU.1 expression is related to Interestingly, while AZA/G-CSF combination caused increase of
methylation status of URE and that AZA upregulates PU.1 H3K9Ac downstream URE (at 15.9 amplicon), the AZA/GM-CSF
coincidently with DNA demethylation of URE in MDS cells. 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).
Myeloid colony-stimulating cytokines modulate AZA-mediated Thus, use of colony-stimulating cytokines preceding addition of
effects on MDS cells via chromatin modications at URE AZA to MDS cells facilitated changes in DNA demethylation as well
On the basis of a previous study in which G-CSF potentiated PU.1 as histone modications at URE resulting in more efcient PU.1
upregulation and myeloid differentiation within normal human derepression.
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 Primary MDS progenitors respond to AZA by inducing myeloid
specicity of the MDS progenitors could be differentially sensitive differentiation
to any cytokine-mediated effects. Our data implicate AZA as a potent agent in restoring the myeloid
Interestingly, all cytokines enhanced signicantly AZA-mediated differentiation capacity of MDS CD34 progenitors. To explore
DNA demethylation of URE (Supplementary Figure 2A). This was this further, we have studied int-2/high-risk MDS patients in detail
associated with derepression of PU.1 (and its target EGR2) in the on both the chromatin and expression levels (see Supplementary
cells co-treated with GM-CSF and M-CSF (Figure 3a). Notably, Table) and determined in vitro response to AZA in the presence or
G-CSF mildly reduced the AZA-mediated PU.1 induction coinciding absence of colony-stimulating cytokines. G-CSF was used as it was
with a dampering of the EGR2 response. Despite both M-CSF and shown to potentiate expression of PU.1 by DNMTi40 and is
GM-CSF augmenting the AZA effect, only in the GM-CSF/AZA- routinely used in clinical hematology. For such studies we have
treated cells was the cell differentiation effectively induced as used ex vivo isolated bone marrow-derived progenitors of the
demonstrated by expression of CD11b using ow cytometry patient P302, a 67-year-old male with RAEB II (WHO), IPSS-Int2, who
(Figure 3b) coincidently with a block of cell proliferation presented with anemia and transfusion dependency of 4 TU/Mo,
(Figure 3c). The apoptosis did not signicantly differ between neutropenia and 10% blasts in bone marrow (phenotypically
cells either treated with AZA or AZA in combination with the characterized as CD33 CD34 CD117 HLADR CD45 that
cytokines. were also detectable in peripheral blood by ow cytometry).
In order to understand the consequences that AZA and the Additionally, we also studied patients P301 (76-year-old male, with
cytokines treatments have on the URE, we have performed 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 (Figure 4e). We have observed enhanced CD11b expression
progenitors as compared with controls (data not shown). Despite following in vitro AZA/G-CSF treatment also in the patients P299 and
expressing intermediate levels of PU.1, the CD34 cells displayed P301 (data not shown). Chromatin immunoprecipitation indicates
some marks of DNA methylation at URE: P302 displayed DNA that at the URE the AZA treatment (and this was potentiated by
methylation within the CpGs 3 -- 6 and at three other CpGs G-CSF) signicantly changed the histone modication pattern:
(Figure 4a). Importantly, PU.1 transcript expression (in P302) was it increased H3K9Ac and decreased H3K9Me3 coincidently with
responsive to AZA treatment and this effect was enhanced by PU.1 gene derepression (Figure 4f).
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 Activation of PU.1 gene by AZA involves also other regulatory
with AZA and G-CSF signicantly downregulated expression of KIT, regions in addition to URE
a known stem cell marker (Figure 4c). This phenomenon was also To determine whether AZA-mediated upregulation of PU.1 is
observed in the OCI-M2 cells (Figure 4d) again demonstrating solely dependent upon the URE, we utilized URE/ mutant
similarity of the MDS patients and the OCI-M2 cells. The ow mice.12 As previously described these mice develop AML due to
cytometry experiments indicated that the patient P302 decreased the dysregulation of PU.1 expression. AML progenitors derived
CD34 positivity and gained CD11b positivity following AZA/G-CSF 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 Thus, effect of AZA treatment is in addition of URE dependent on
by AZA (as outlined above) and mRNA of PU.1 and its targets another enhancer regions or the promoter itself.
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 DISCUSSION
the control c-kit-positive normal bone marrow progenitor cells. In Current therapeutic treatment of hematologic malignancies
the absence of the URE, the PU.1 transcripts were induced in a focuses on the need to induce apoptosis and/or cycle arrest of
dose-dependent manner by AZA treatment in a similar relative the malignant cell population. However, the ability to restore and
fold change as in control progenitors. The levels of PU.1 were induce cellular differentiation represents an alternate strategy,
reected by levels of PU.1 targets Egr2 and G1 in the same and has potential to be very helpful in the treatment of diseases
samples. that are often refractory to standard chemotherapy regimens.
These data indicate that even though URE is required for We herein present new evidence that gene encoding a key
expression of physiological levels of PU.1 it is not exclusively hematopoietic transcription factor PU.1 is a target of chromatin
required for AZA-mediated upregulation of PU.1 levels in AML. 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 262507 and SVV-2012, GAUK 251135 82210, GACR P305/12/1033, UNCE 204021.
MDS patients with higher risk and found that among patients with Collaborators support comes from the Grants NS10632-3/2009, NS9634-4/2008, and
signicantly shorter survival on AZA are patients with low 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
expression of PU.1 (Figure 1). In addition, expression of PU.1
patients participating in this study.
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 REFERENCES
autoregulation mechanism and that it can be reverted by AZA.
1 Laslo P, Spooner CJ, Warmash A, Lancki DW, Lee HJ, Sciammas R et al.
Specically, our data strongly indicate that MDS progenitors with Multilineage transcriptional priming and determination of alternate hematopoie-
heavily repressed URE would require higher dose of AZA and tic cell fates. Cell 2006; 126: 755 - 766.
those that display relatively open URE chromatin may require 2 Dahl R, Simon MC. The importance of PU.1 concentration in hemato-
lower dose of AZA. Importantly, AZA in order to activate PU.1 gene poietic lineage commitment and maturation. Blood Cells Mol Dis 2003; 31:
expression may target additional regulatory regions, other than 229 - 233.
URE (Figure 5). Therefore, more patient data and analyses of the 3 Stopka T, Amanatullah DF, Papetti M, Skoultchi AI. PU.1 inhibits the erythroid
PU.1 upstream regions are however needed to fully comprehend program by binding to GATA-1 on DNA and creating a repressive chromatin
the relationship between the epigenetic status of PU.1 gene and structure. EMBO J 2005; 24: 3712 - 3723.
4 Burda P, Curik N, Kokavec J, Basova P, Mikulenkova D, Skoultchi AI et al. PU.1
clinical responsiveness to AZA.
activation relieves GATA-1-mediated repression of Cebpa and Cbfb during
Alternatively to recommending different AZA levels, we leukemia differentiation. Mol Cancer Res 2009; 7: 1693 - 1703.
searched for mechanism that would enhance AZA-mediated 5 Ross IL, Yue X, Ostrowski MC, Hume DA. Interaction between PU.1 and another Ets
effects on PU.1 gene and focused on colony-stimulating cytokines family transcription factor promotes macrophage-specic Basal transcription
such as G-CSF. It is important to note that additional chromatin initiation. J Biol Chem 1998; 273: 6662 - 6669.
modiers may be required in order to achieve PU.1 gene 6 Yamamoto H, Kihara-Negishi F, Yamada T, Hashimoto Y, Oikawa T. Physical and
reexpression outcome following DNA demethylation in MDS.42 functional interactions between the transcription factor PU.1 and the coactivator
G-CSF was shown to facilitate repression of stem cell genes with or CBP. Oncogene 1999; 18: 1495 - 1501.
without its direct effect to induce PU.1 expression as reported 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
elsewhere.40 Others reported that addition of DNMTi to normal
noncoding transcription by alternate mechanisms at distinct developmental
murine progenitors preceded by G-CSF (or transgenic activation of stages of hematopoiesis. Mol Cell Biol 2007; 27: 7425 - 7438.
PU.1) upregulated PU.1 levels and its program of myeloid 8 Petrovick MS, Hiebert SW, Friedman AD, Hetherington CJ, Tenen DG, Zhang DE.
differentiation.40 Our data in primary MDS progenitors Multiple functional domains of AML1: PU.1 and C/EBPalpha synergize with
(Figure 4c) and OCI-M2 cells (Figure 4d) indicated that indeed G- different regions of AML1. Mol Cell Biol 1998; 18: 3915 - 3925.
CSF inhibited AZA-mediated induction of KIT expression coin- 9 Wontakal SN, Guo X, Will B, Shi M, Raha D, Mahajan MC et al. A large gene
cident with enhanced myeloid differentiation. Our data also network in immature erythroid cells is controlled by the myeloid and B cell
indicate that G-CSF (and also GM-CSF and M-CSF) preincubation transcriptional regulator PU.1. PLoS Genet 2011; 7: e1001392.
prior AZA is capable to inuence both DNA methylation (Figure 1c 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
and Supplementary Figure 2) and chromatin modications at URE
element. Mol Cell Biol 2005; 25: 2832 - 2845.
(Figures 3 and 4), is associated with increased PU.1 mRNA levels 11 Chen H, Ray-Gallet D, Zhang P, Hetherington CJ, Gonzalez DA, Zhang DE et al.
and may be mediated by G-CSF signaling pathway that includes PU.1 (Spi-1) autoregulates its expression in myeloid cells. Oncogene 1995; 11:
PU.1. It should be noted that AZA alone and importantly together 1549 - 1560.
with G-CSF is capable to recapitulate its effects observed in the 12 Rosenbauer F, Wagner K, Kutok JL, Iwasaki H, Le Beau MM, Okuno Y et al. Acute
MDS cell line, also in ex vivo isolated primary high-risk MDS cells. myeloid leukemia induced by graded reduction of a lineage-specic transcription
G-CSF represents a cytokine that is frequently used in MDS and factor, PU.1. Nat Genet 2004; 36: 624 - 630.
AML and other hematological malignancies to prevent neutro- 13 Steidl U, Steidl C, Ebralidze A, Chapuy B, Han HJ, Will B et al. A distal single
penic complications following chemotherapy.43 Its use in MDS is nucleotide polymorphism alters long-range regulation of the PU.1 gene in acute
myeloid leukemia. J Clin Invest 2007; 117: 2611 - 2620.
not associated with increased frequency of developing AML44,
14 Leddin M, Perrod C, Hoogenkamp M, Ghani S, Assi S, Heinz S et al. Two distinct
thus it represents promising candidate with a potential to auto-regulatory loops operate at the PU.1 locus in B cells and myeloid cells. Blood
augment AZA prodifferentiation effects. 2011; 117: 2827 - 2838.
We observed that some patients whose tumor cells expressed 15 Cook WD, McCaw BJ, Herring C, John DL, Foote SJ, Nutt SL et al. PU.1 is a
higher PU.1 level displayed favorable therapeutic response; suppressor of myeloid leukemia, inactivated in mice by gene deletion and
such observation will require extended follow-up and analyses mutation of its DNA binding domain. Blood 2004; 104: 3437 - 3444.
of additional patients. If this scenario is however correct, 16 Mueller BU, Pabst T, Osato M, Asou N, Johansen LM, Minden MD et al.
further assessment of molecular response to AZA and AZA/ Heterozygous PU.1 mutations are associated with acute myeloid leukemia. Blood
G-CSF will be crucial in predicting responsiveness to AZA, 2002; 100: 998 - 1007.
17 Lamandin C, Sagot C, Roumier C, Lepelley P, De Botton S, Cosson A et al. Are PU.1
potential of AML transformation and other phenotypic features
mutations frequent genetic events in acute myeloid leukemia (AML)? Blood 2002;
of high-risk MDS. Responsiveness of MDS cells to G-CSF 100: 4680 - 4681.
pretreatment prior to AZA in vitro may provide rationale for 18 Vegesna V, Takeuchi S, Hofmann WK, Ikezoe T, Tavor S, Krug U et al. C/EBP-beta,
further evaluating the role of G-CSF in addition to epigenetic C/EBP-delta, PU.1, AML1 genes: mutational analysis in 381 samples of
therapy in MDS going forward. 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
CONFLICT OF INTEREST 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
The authors declare no conict of interest.
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
ACKNOWLEDGEMENTS function and erythroid differentiation by blocking GATA-1 DNA binding. Blood
The major support of this work comes from the grant (investigators TS and MK) of the 2000; 96: 2641 - 2648.
Ministry of Industry and Trade (FR-TI2/509). Other support is from Grants NS10310-3/ 22 Moreau-Gachelin F, Tavitian A, Tambourin P. Spi-1 is a putative oncogene in virally
2009, 2B06077, MSM 0021620806 PRVOUK, MSM 0021620808, LC06044, SVV-2011- induced murine erythroleukaemias. Nature 1988; 331: 277 - 280.

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

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

1811
23 OConnell RM, Rao DS, Chaudhuri AA, Boldin MP, Taganov KD, Nicoll J et al. 34 Komashko VM, Farnham PJ. 5-Azacytidine treatment reorganizes genomic histone
Sustained expression of microRNA-155 in hematopoietic stem cells causes a modication patterns. Epigenetics 2010; 5: 229--240.
myeloproliferative disorder. J Exp Med 2008; 205: 585 - 594. 35 Vire E, Brenner C, Deplus R, Blanchon L, Fraga M, Didelot C et al. The Polycomb
24 Durual S, Rideau A, Ruault-Jungblut S, Cossali D, Beris P, Piguet V et al. Lentiviral group protein EZH2 directly controls DNA methylation. Nature 2006; 439:
PU.1 overexpression restores differentiation in myeloid leukemic blasts. Leukemia 871 - 874.
2007; 21: 1050 - 1059. 36 Hernandez-Munoz I, Taghavi P, Kuijl C, Neefjes J, van Lohuizen M. Association
25 Tatetsu H, Ueno S, Hata H, Yamada Y, Takeya M, Mitsuya H et al. Down-regulation of BMI1 with polycomb bodies is dynamic and requires PRC2/EZH2 and the
of PU.1 by methylation of distal regulatory elements and the promoter is required maintenance DNA methyltransferase DNMT1. Mol Cell Biol 2005; 25: 11047 - 11058.
for myeloma cell growth. Cancer Res 2007; 67: 5328 - 5336. 37 Li H, Rauch T, Chen ZX, Szabo PE, Riggs AD, Pfeifer GP. The histone
26 Mueller BU, Pabst T, Fos J, Petkovic V, Fey MF, Asou N et al. ATRA resolves the methyltransferase SETDB1 and the DNA methyltransferase DNMT3A interact
differentiation block in t(15;17) acute myeloid leukemia by restoring PU.1 directly and localize to promoters silenced in cancer cells. J Biol Chem 2006; 281:
expression. Blood 2006; 107: 3330 - 3338. 19489 - 19500.
27 Gu ZM, Liu CX, Wu SF, Zhao M, Xu HZ, Liu W et al. PU.1 directly regulates retinoic 38 Smallwood A, Esteve PO, Pradhan S, Carey M. Functional cooperation between
acid-induced expression of RIG-G in leukemia cells. FEBS Lett 2010; 585: 375 - 380. HP1 and DNMT1 mediates gene silencing. Genes Dev 2007; 21: 1169 - 1178.
28 Silverman LR, McKenzie DR, Peterson BL, Holland JF, Backstrom JT, Beach CL et al. 39 Jiemjit A, Fandy TE, Carraway H, Bailey KA, Baylin S, Herman JG et al. p21(WAF1/
Further analysis of trials with azacitidine in patients with myelodysplastic CIP1) induction by 5-azacytosine nucleosides requires DNA damage. Oncogene
syndrome: studies 8421, 8921, and 9221 by the Cancer and Leukemia Group B. 2008; 27: 3615 - 3623.
J Clin Oncol 2006; 24: 3895 - 3903. 40 Hu Z, Negrotto S, Gu X, Mahfouz R, Ng KP, Ebrahem Q et al. Decitabine maintains
29 Fenaux P, Mufti GJ, Hellstrom-Lindberg E, Santini V, Finelli C, Giagounidis A et al. hematopoietic precursor self-renewal by preventing repression of stem cell genes
Efcacy of azacitidine compared with that of conventional care regimens in the by a differentiation-inducing stimulus. Mol Cancer Ther 2010; 9: 1536 - 1543.
treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, 41 Drexler HG, Dirks WG, Macleod RA. Many are called MDS cell lines: one is chosen.
phase III study. Lancet Oncol 2009; 10: 223 - 232. Leuk Res 2009; 33: 1011 - 1016.
30 Silverman LR, Demakos EP, Peterson BL, Kornblith AB, Holland JC, Odchimar- 42 Si J, Boumber YA, Shu J, Qin T, Ahmed S, He R et al. Chromatin remodeling is
Reissig R et al. Randomized controlled trial of azacitidine in patients with the required for gene reactivation after decitabine-mediated DNA hypomethylation.
myelodysplastic syndrome: a study of the cancer and leukemia group B. J Clin Cancer Res 2010; 70: 6968 - 6977.
Oncol 2002; 20: 2429 - 2440. 43 Dotti G, Stella CC, Mangoni L, Cottafavi L, Caramatti C, Almici C et al. Granulocyte
31 Fenaux P, Gattermann N, Seymour JF, Hellstrom-Lindberg E, Mufti GJ, Duehrsen U colony-stimulating factor (G-CSF) prevents dose-limiting neutropenia in lympho-
et al. Prolonged survival with improved tolerability in higher-risk myelodysplastic ma patients receiving standard dose chemotherapy. Haematologica 1995; 80:
syndromes: azacitidine compared with low dose ara-C. Br J Haematol 2010; 149: 142 - 145.
244 - 249. 44 Estey E, Thall P, Andreeff M, Beran M, Kantarjian H, OBrien S et al. Use of
32 Fenaux P, Bowen D, Gattermann N, Hellstrom-Lindberg E, Hofmann WK, granulocyte colony-stimulating factor before, during, and after udarabine
Pfeilstocker M et al. Practical use of azacitidine in higher-risk myelodysplastic plus cytarabine induction therapy of newly diagnosed acute myelogenous
syndromes: an expert panel opinion. Leuk Res 2010; 34: 1410 - 1416. leukemia or myelodysplastic syndromes: comparison with udarabine plus
33 Quintas-Cardama A, Santos FP, Garcia-Manero G. Therapy with azanucleosides for cytarabine without granulocyte colony-stimulating factor. J Clin Oncol 1994; 12:
myelodysplastic syndromes. Nat Rev Clin Oncol 2010; 7: 433 - 444. 671 - 678.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

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