M-CSF and GM-CSF regulation of STAT5 activation and DNA binding in myeloid cell differentiation is disrupted in nonobese diabetic mice.

Abstract

Defects in macrophage colony-stimulating factor (M-CSF) signaling disrupt myeloid cell differentiation in nonobese diabetic (NOD) mice, blocking myeloid maturation into tolerogenic antigen-presenting cells (APCs). In the absence of M-CSF signaling, NOD myeloid cells have abnormally high granulocyte macrophage colony-stimulating factor (GM-CSF) expression, and as a result, persistent activation of signal transducer/activator of transcription 5 (STAT5). Persistent STAT5 phosphorylation found in NOD macrophages is not affected by inhibiting GM-CSF. However, STAT5 phosphorylation in NOD bone marrow cells is diminished if GM-CSF signaling is blocked. Moreover, if M-CSF signaling is inhibited, GM-CSF stimulation in vitro can promote STAT5 phosphorylation in nonautoimmune C57BL/6 mouse bone marrow cultures to levels seen in the NOD. These findings suggest that excessive GM-CSF production in the NOD bone marrow may interfere with the temporal sequence of GM-CSF and M-CSF signaling needed to mediate normal STAT5 function in myeloid cell differentiation gene regulation.

Full text

Page 1

Hindawi Publishing Corporation
Clinical and Developmental Immunology
Volume 2008, Article ID 769795, 8 pages
doi:10.1155/2008/769795
ResearchArticle
M-CSF and GM-CSF Regulation of STAT5 Activation and
DNA Binding in Myeloid Cell Differentiation is Disrupted in
Nonobese Diabetic Mice
B. Rumore-Maton,1J. Elf,2N. Belkin,2B. Stutevoss,2F. Seydel,2E. Garrigan,2and S. A. Litherland2,3
1College of Veterinary Medicine, University of Florida, 2015 SW 16th Avenue, Gainesville, FL 32610, USA
2Department of Pathology, Immunology, and Laboratory Medicine, College of Medicine, University of Florida,
P. O. BOX 100275, 1600 SW Archer Road, Gainesville, FL 32610, USA
3Burnham Institute for Medical Research-Lake Nona, SLSL Buliding M6-1025, Rm 102-9, Kennedy Space Center,
Lake Nona, FL 32899, USA
Correspondence should be addressed to S. A. Litherland, sal@burnham.org
Received 4 May 2008; Revised 13 August 2008; Accepted 23 October 2008
Recommended by Chaim Putterman
Defects in macrophage colony-stimulating factor (M-CSF) signaling disrupt myeloid cell differentiation in nonobese diabetic
(NOD) mice, blocking myeloid maturation into tolerogenic antigen-presenting cells (APCs). In the absence of M-CSF signaling,
NOD myeloid cells have abnormally high granulocyte macrophage colony-stimulating factor (GM-CSF) expression, and as
a result, persistent activation of signal transducer/activator of transcription 5 (STAT5). Persistent STAT5 phosphorylation
found in NOD macrophages is not affected by inhibiting GM-CSF. However, STAT5 phosphorylation in NOD bone marrow
cells is diminished if GM-CSF signaling is blocked. Moreover, if M-CSF signaling is inhibited, GM-CSF stimulation in
vitro can promote STAT5 phosphorylation in nonautoimmune C57BL/6 mouse bone marrow cultures to levels seen in the
NOD. These findings suggest that excessive GM-CSF production in the NOD bone marrow may interfere with the temporal
sequence of GM-CSF and M-CSF signaling needed to mediate normal STAT5 function in myeloid cell differentiation gene
regulation.
Copyright © 2008 B. Rumore-Maton et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
1.INTRODUCTION
Myeloid cell differentiation gives rise to 3 populations of
professional antigen-presenting cells (APC) in the immune
system: monocytes, macrophages, and dendritic cells (DC).
Sequential signaling from interleukin-3 (IL-3), granulocyte-
macrophage-, macrophage-, and granulocyte-colony stim-
ulating factors (GM-, M-, and G-CSF, respectively) sets
the microenvironment necessary for bone marrow myeloid
precursor cells to differentiate and function [1]. The timing
and level of each cytokine in myeloid cell microenvi-
ronmentare tightly controlled in order for myeloid cells
differentiatetofunctionalAPC,andtopreventthepremature
activation of genes involved in macrophage activation, such
as cyclooxygenase/prostaglandin synthase 2 (COX 2/PGS2),
which could promote deleterious inflammation.
The cytokine GM-CSF has a unique dual influence
in myeloid cells, acting first as a differentiation factor in
myeloid hematopoiesis, and then later as an activation
stimulus in mature monocytes and macrophages. Both
overexpression and knockout deletion of GM-CSF in mice
can lead to dysregulation of myeloid differentiation and
autoimmune disease [2], suggesting that GM-CSF influence
exertsdiscretetemporalandquantitativeregulatoryeffectsin
myeloid cell differentiation and mature cell activation. This
switch in GM-CSF function relies heavily on a change in
responsiveness of cells to GM-CSF before and after M-CSF
stimulation [3], and on the ability of GM-CSF to activate
STAT5 phosphorylation at different stages of myeloid cell
maturation [4, 5].
WorkbyPiazzaetal.[4]suggeststhatSTAT5proteinscan
act as key cytokine and growth factor-inducible regulatory

Page 2

2Clinical and Developmental Immunology
“switches” for gene expression during myeloid differentia-
tion and activation. In early myeloid differentiation stages,
IL-3 and GM-CSF can induce signaling through both
STAT5A (94-96kD) and B (94-92kD) isoforms which act as
adaptormoleculesforhistoneacetylasesanddeacetylases[4–
7]. Subsequent signaling through M-CSF and G-CSF signals
in matured and activated cells also activates STAT5 [4, 8, 9].
Thus, during cytokine-induced differentiation, STAT5 can
act to promote both gene transcriptional regulation through
epigenetic chromatin dynamics [4, 6, 7].
Incomplete myeloid differentiation is emerging as a
common component of multiple autoimmune pathologies,
both in mice and man. Serreze et al. (1993) [10] found
that myeloid antigen presenting cell (APC) differentiation
is impaired in NOD mouse bone marrow due to a lack
of responsiveness to M-CSF. This nonresponsiveness was
not linked to any defect in M-CSF expression or defect
in its receptor, but to unknown problems with M-CSF
intracellular signaling. Morin et al. (2003) [11] also noted
that high GM-CSF can skew NOD myeloid differentiation
away from macrophage and DC development, and toward
an excess of granulocyte production. Several laboratories
have reported disproportion of immature DC compared
to mature DC in the diabetic NOD, a process depen-
dent on regulation of GM-CSF signaling [10–13]. These
dysfunctional APC phenotypes suggest that NOD myeloid
cell differentiation and activation are defective at a point
or points, where the determination of specific cell lineage
decisions is made based on the cytokine microenvironment
[1, 4]. Similar defective myeloid cell phenotypes have been
seen in human autoimmune diseases such as T1D, SLE,
and autoimmune thyroid diseases (ATD) [10–15]. The
result of these defective myeloid cell properties is a loss
of professional APC that can efficiently promote toler-
ance and regulate appropriate macrophage inflammatory
responses [12, 16, 17]. Loss of these functions sets up an
immune microenvironment conducive to immunopatho-
genesis.
In our studies of APC dysfunction in autoimmunity,
we have found that both NOD and autoimmune human
myeloid cells express abnormally high levels of GM-CSF and
have prolonged STAT5 activation after even brief exposure
to GM-CSF [18–20]. In these studies, we test the potential
connection of GM-CSF signaling and its activation of STAT5
isoformswiththedefectinM-CSFsignalinginNODmyeloid
differentiation in vitro. We find that both GM-CSF overpro-
duction and persistent STAT5 phosphorylation phenotypes
are amplified in autoimmune NOD bone marrow cells, but
can be overridden by in vitro inhibition of endogenous GM-
CSF signaling. Furthermore, the persistent STAT5 phospho-
rylation phenotype seen in NOD myeloid cells can be recre-
ated in non-autoimmune C57BL/6 bone marrow cultures by
blocking M-CSF while stimulating with NOD level GM-CSF.
ThesefindingsinvitrosuggestthatthedefectinNODM-CSF
signaling in vivo may occur during the switch of influence
between GM-CSF to M-CSF during myeloid differentiation,
andindicatethatprolongedexpressionofGM-CSFpromotes
thedysregulationofSTAT5,whichmayinterferewithM-CSF
signaling during NOD myeloid cell differentiation.
2.METHODS AND MATERIALS
2.1.Mousestrainsused
Eight to twelve-week old NOD and C57BL/6 female mice
(The Jackson Laboratory, Bar Harbor, Mich, USA) were
used for all studies. These strains were maintained as
breeding stock in our Pathology Department SPF colony,
in microisolator cages with food and water ad libium. All
procedures were conduced to maintain humane treatment
of all animals according to IACUC approved protocols B083
and D754, and humanely euthanized by cervical dislocation
while over-anesthetized. All tissue collections were done
postmortem. A minimum of 3 animals per strain was used
per treatment for each run of each experiment.
2.2.Bonemarrowdifferentiationcultureand
samplepreparation
Long bones from the legs were collected postmortem into
cold RPMI +10%FCS medium (Cellgro, MediaTech, Man-
assas, Va, USA) and the marrow flushed into a sterile tube
by cold medium injected into the bone through a 30-gauge
needle. The marrow samples were washed with cold media
by centrifugation and red blood cells in samples lysed by
incubationinnonisotonic0.84%ammoniumchloridebuffer
(Fisher Scientific, Pittsburgh, Pa, USA). The remaining bone
marrow cells are then plated at 5–10 million cells/well on
tissue culture dishes and fed with fresh sterile medium alone
or with medium containing 1000U/mL of GM-CSF (In
vitrogen-Biosource, Carlsbad, Calif, USA) +2μg/mL anti-
M-CSF blocking antibodies (Pierce-Endogen, Rockford, Ill,
USA), or medium containing 500U/mL M-CSF (In vitro-
gen) + anti-GM-CSF blocking antibodies (Pierce-Endogen).
Cultures were maintained for 24 hours at 37◦C/5%CO2,
then washed, and resupplemented for an additional 24 hours
in culture at 37◦C/5%CO2. Cells were then analyzed for
myeloid phenotypic identification and STAT5 phosphoryla-
tion by flow cytometry and deconvolution microscopy.
2.3.Flowcytometricanddeconvolution
microscopicimageanalysis
Flow cytometric analysis of fluorescently conjugated anti-
bodysurfaceandintracellularbindingwasusedtophenotyp-
ically identify myeloid cells and quantify the level of STAT5
tyrosine phosphorylation (STAT5Ptyr) in cells ex vivo and
after in vitro stimulation. Cells were first labeled with anti-
CD11b antibody FITC or PE conjugates (BD Biosciences,
Calif, USA) to identify monocytes and macrophages in
the sample, then fixed with 2% (v/vCf) formaldehyde
(Sigma-Aldrich, St. Louis, Mo, USA) and permeabilized
using 0.5% saponin (Sigma-Aldrich) in a high protein, iso-
tonic FACS buffer [14]. Intracellular staining of STAT5Ptyr
was done using FITC or APC-conjugated anti-STAT5Ptyr-
specific monoclonal antibodies (Millipore-Upstate mono-
clonal antibody conjugated FITC (Millipore-Upstate) or
with APC using labeling kit, Prozyme, San Leandro, Calif,
USA). Data were analyzed as the percentage of positive
myeloid cells (%STATPtyr+/CD11b+) to assess the number

Page 3

B. Rumore-Maton et al.3
of cells in the myeloid population with activated STAT5.
After flow analysis, the labeled cells are centrifuged onto
slides and counterstained for chromatin (DAPI, Invitrogen-
Molecular Probes, Carlsbad, Calif, USA). These cells were
used for imaging by deconvolution microscopy as previously
described [19, 20] for 3-dimensional projection rotational
reconstruction image analysis to identify subcellular location
of activated STAT5. Statistical analyses of flow cytometric
data were performed using Prism 5 software (GraphPad,
San Diego, Calif, USA), with pairwise comparisons being
done using Student’s t-test or Mann-Whitney U analyses and
multiple group analyses done using one-way ANOVA.
2.4.Doublechromatinimmunoprecipitation(dbChIP)
ofchromatin-boundphospho-STAT5protein
Five million bone marrow cells were cultured with or
without 100μM Na vanadate (in DMSO, Sigma-Aldrich)
for 30 minutes at 378◦C/5%CO2. Half of the cultures +/−
vanadate were then supplemented with 1000U/mL GM-
CSF, and all were incubated for an additional 90 minutes
at 378◦C/5%CO2. The cells were then fixed in situ, with
1% formaldehyde in 1xPBS (methanol-free, Sigma-Aldrich),
for 10 minutes at 37◦C prior to lysis with high SDS lysis
buffer (Millipore-Upstate). The fixed lysates were sonicated
todisruptmembranesandshearchromatintoapproximately
1000bp fragments then frozen at −80◦C until analyzed.
The sample was split into five 1 × 106cell aliquots for
use in anti-STAT5Ptyr/anti-Histone H3 mediated double
ChIP (dbChIP) protein isolations for Western blot analysis
of STAT5 associated with histone/chromatin complexes.
Aliquots used for dbChIP were precleared with salmon
sperm DNA Protein A or G agarose beads (Millipore-
Upstate), then incubated overnight at 4◦C with antityrosine
phosphorylated STAT5 (STAT5Ptyr) antibodies (Millipore-
Upstate). After incubation, the antibody-bound chromatin
complexes were precipitated using salmon sperm DNA
Protein A agarose beads, and washed extensively with a
series of increasing stringency buffers (low salt, high salt,
LiCl, TE; ChIP reagent kit, Millipore-Upstate). A nonspecific
antibody control ChIP (mouse IgG, Millipore-Upstate) and
a sham IP containing no extract were used as negative
controls.
ChIP extract aliquots dissociated from the beads in
1%SDS, 0.1M bicarbonate buffer (Fisher Scientific). To both
unprecipitated (total) extracts and ChIP samples, NaCl was
added to a final concentration of 500mM and the samples
were incubated for 4 hours at 65◦C to reverse the formalde-
hyde cross-links. The released anti-STAT5Ptyr-selected com-
plexes were neutralized and reprecipitated (“double ChIP”,
dbChIP) overnight at 4◦C using precoupled anti-Histone
H3-Protein G Salmon sperm DNA-coated beads (Millipore-
Upstate). The dbChIP complexes were then washed with
the same sequence of buffers as used in the first ChIP,
and then boiled in 1x Leammili buffer (BioRad, Hercules,
Calif, USA) for Western blot analysis. Western blots were
probed with anti-STAT5 Ptyr (Millipore-Upstate) antibodies
and cross-linked and uncoupled proteins were visualized
using Amersham ECL plus chemilumenescence reagents (GE
Healthcare/Amersham, Piscataway, NJ, USA). Densitometric
analysis using BioRad Imager and Quantity One Software
was used to estimate molecular weight of cross-linked dimer
complexes and freed monomeric isoforms in STAT5Ptyr
positive bands.
3. RESULTS
3.1. GM-CSFandM-CSFsignalinginterplayto
regulateSTAT5activation
Mouse bone marrow cells can differentiate to mature
different macrophage phenotypes within 4 days culture in
M-CSF or 4–7 days in GM-CSF [3, 21, 22]. Our experiments
attempted to examine myeloid differentiation at stages prior
to complete maturation, at 2 to 48 hours in culture. For
our in vitro myeloid cell differentiation experiments, NOD
and C57BL/6 mouse bone marrow cells were stimulated with
one cytokine (1000U/mL GM-CSF or 500U/mL M-CSF) in
the presence of blocking antibodies for the other cytokine
(anti-GM-CSF or anti-M-CSF, 2μg/mL) for 48 hours at
37◦C/5%CO2.
In earlier studies, we found that blocking GM-CSF
signaling by anti-GM-CSF antibodies or with the Jak2/3
inhibitor AG490 in mature monocytes and macrophages
did not affect the persistence of phosphorylated STAT5
[19]. In contrast, blocking endogenous GM-CSF or M-CSF
stimulation by specific antibodies significantly diminished
STAT5 phosphorylation in NOD bone marrow cells (∗P =
.0240, ANOVA, Figure 1(a)).
Deconvolution microscopic image analysis (see Figure 2)
suggested that we could recreate the NOD STAT5 phos-
phorylation phenotype in non-autoimmune C57BL/6 cells
by blocking M-CSF signaling while increasing the GM-CSF
stimulation to the level seen in the NOD [19, 20]. Flow
cytometric analysis also indicated that blocking endogenous
M-CSF while stimulating with GM-CSF promoted STAT5
phosphorylation in C57BL/6 48-hour bone marrow cultures
(∗p = .0167, Mann-Whitney U test, Figure 1(b)). However,
when we stimulated bone marrow cultures with M-CSF in
vitro in the presence of blocking antibodies to GM-CSF,
STAT5 phosphorylation was also stimulated in C57BL/6
cultures, and remained statistically unchanged in NOD bone
marrow cells (∗p = .0336 and .5556, respectively, Mann-
Whitney U test, Figure 2(c)).
Taken together, these findings suggest that the interplay
M-CSF and GM-CSF signaling is required to properly
regulate STAT5 function, and support the role of temporally
sequenced GM-CSF and M-CSF signaling on controlling
STAT5 activation during myeloid cell differentiation. Fur-
thermore, this interplay is modulated by the timing and
concentration of GM-CSF stimulation early in differentia-
tion, which appears to set the stage for STAT5 activation in
later stages, regardless of further stimulation by GM-CSF
or M-CSF. Because the NOD cells have higher endogenous
GM-CSF production [19], this temporal signaling pattern is
disrupted and leads to aberrant cell responses to GM-CSF.

Page 4

4Clinical and Developmental Immunology
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
STAT5Ptyr +/CD11b+ cells (%)
∗∗P = .0240
ns; P = .4266
ns; P = .4857
ns; P = .0663
∗P = .0441
NOD C57BL/6
Media only
NOD
Anti-GM-CSF
C57BL/6NOD
Anti-M-CSF
C57BL/6
(a)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
STAT5Ptyr +/CD11b+ cells (%)
ns; P = 1
∗P = .0488
ns; P = .5714
∗∗P = .0167
no
α-M
NOD
GM-CSF treated
no
C57BL/6
α-M
(b)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
STAT5Ptyr +/CD11b+ cells (%)
ns; P = .8857
∗∗P = .0159
ns; P = .5556
∗P = .0336
no
α-GM
NOD
M-CSF treated
no
C57BL/6
α-GM
(c)
Figure 1: Flow cytometric analysis of STAT5 phosphorylation in in vitro myeloid differentiation. (a) Bone marrow cultures (1 million
cells/mL) fromNOD and C57BL/6control mice were differentiated in culture with either, (b) 1000U/mL GM-CSFplus 2μg/mL anti-M-CSF
blocking antibodies α-M), or (c) with 500U/mL M-CSF with 2μg/mL anti-GM-CSF blocking antibodies α-GM), for 48 hours at 3◦C/5CO2.
Culturestreatedwithmediumonly(“mediaonly”in(a),“0”in(b)and(c))orwithmediumsupplementedonlywithantibodiestoblockM-
CSF or GM-CSF (ANTI-M-CSF, ANTI-GM-CSF, respectively, in (a)) were run in parallel with the cytokine-stimulated cultures. In addition,
separate aliquots of bone marrow cells were cultured in M-CSF or GM-CSF alone without blocking antibodies (“no” in (b) and (c)). At 48
hours, cells were stained with anti-STAT5Ptyr-FITC and anti-CD11b-PE antibodies for analysis by intracellular immune-histochemical flow
cytometry. Strain and treatment regiments for each pair of cultures are listed on the X-axis. The percentage of STAT5Ptyr+ in CD11b+ cells
is given on the Y-axis. Graphs are representative of 3–9 separate runs of each treatment. The p-values indicated were derived from Mann-
Whitney U test, Student’s t-test, or one-way ANOVA analyses, as appropriate, for the sample group comparisons indicated by brackets.
3.2. NODbonemarrowcellDNAbindingof
STAT5isincreasedbyGM-CSF
To evaluate the possibility that increased STAT5 phosphory-
lation could lead to increased chromatin binding in response
to GM-CSF, we performed double ChIP analyses on NOD
and C57BL/6 bone marrow cells that had been cultured with
GM-CSF for 2 hours, and looked for in situ assembly of
chromatin-associated protein complexes which contain both
STAT5 and Histone H3 (Figure 3). We controlled for the
possibility of the prolonged phosphorylation of cytoplasmic
DNA-unbound STAT5 masking a more transient phos-
phorylation of DNA-bound STAT5 isoforms, by culturing
these cells in the presence and absence of the phosphatase
inhibitor, sodium vanadate. In these dbChIP analyses, we
detected bands containing phosphorylated STAT5 only in
GM-CSF-treated NOD cells (Figure 3). The presence or
absence of the phosphatase inhibitor did not alter the
banding pattern in either NOD or C57BL/6 cell extracts.
Higher molecular weight STAT5-containing complexes were

Page 5

B. Rumore-Maton et al.5
CD11b+/DAPI STAT5Ptyr+Combined
C57BL/6
+GM-CSF
+anti-M-CSF
NOD
+GM-CSF
+anti-M-CSF
C57BL/6
+M-CSF
+anti-GM-CSF
NOD
+M-CSF
+anti-GM-CSF
C57BL/6
+media
only
NOD
+media
only
Isotype
background
Figure 2: GM-CSF/M-CSF cytokine-induced myeloid differentiation in vitro. Bone marrow cultures from NOD and C57BL/6 control
mice were differentiated in culture using regiment of cytokine and anticytokine blocking antibodies. Cells were treated with either
1000U/mL GM-CSF plus 2μg/mL anti-M-CSF blocking antibodies, with 500U/mL M-CSF plus 2μg/mL anti-GM-CSF blocking antibodies,
or with media alone for 48 hours at 3◦C/5CO2. Cells were stained with anti-STAT5Ptyr-FITC (green) and anti-CD11b-PE (red)
antibodies for analysis by flow cytometry (summarized in Figure 1) and deconvolution microscopy (shown here) for the development
of cells with myeloid phenotype (CD11b+) and intracellular expression of tyrosine phosphorylated STAT5. Background staining for
anti-STAT5Ptyr-FITC antibody was determined by parallel sample stained with a nonspecific mouse IgG isotype control (bottom most
panels). The images are composites of 3D projection of 20 optical sections (0.2 micron each) showing both immunohistochemical
labels as well as DAPI (blue) poststaining of chromatin within the cells. Accompanying panels show the STAT5Ptyr-FITC staining alone
(green) or CD11b-PE and DAPI staining (red/blue). Treatment regiment and the cell source strain for each pair of cultures are listed
to the left of the images. Images are representative of 5 random fields observed per sample treatment and 3 separate runs of the
experiment.