![Integration of ChIP–chip data for P-Stat1, P-Stat3, and P-Stat5 with transcriptome analysis allows identification of novel Stat target genes in LPS-stimulated microglia. a Activation of Stat1, Stat3, and Stat5 in LPS-stimulated primary rat microglial cultures. The immunoblot shows the levels of phosphorylated and total Stats at various times after stimulation with 100 ng/ml LPS; membranes were reprobed with anti-actin antibody to ensure equal loading of proteins. b Genome-wide occupancy of the Stat proteins and transcriptome analyses in primary rat microglial cultures. Chromatin immunoprecipitation and promoter arrays were used to identify the sequences bound by active Stat1, Stat3 and Stat5 proteins in LPS-stimulated. A diagram indicates the number of the genes bound by a given Stat and overlap between these genes (left panel). The ChIP–chip data for Stat binding were analyzed with the expression data from primary rat microglial cultures stimulated for 6 h with LPS (100 ng/ml) [18]. A diagram indicates the number of the genes with P-Stat peaks and reliable expression data (right panel). c Stat-responsive genes (represented in the ChIP–chip experiment) are labeled as significantly up- or downregulated in LPS-stimulated cells compared to control microglia (control vs. LPS t test, p < 0.0001). The cumulative distribution functions of the log2-transformed changes in the expression of the genes with (gray) and without (black) binding peaks for the given P-Stat, the numbers of genes containing peaks, and the p values were calculated from the Kolmogorov–Smirnoff tests. Lower panel Histograms for ranking log2-transformed changes in the expression of the genes containing P-Stat binding peaks among the general set of genes (t test p < 0.0001, FDR < 0.001) ranked in the increasing order on the LPS-induced change in expression. Black horizontal lines show the uniform distributions that would be expected for random binding data](https://www.researchgate.net/profile/Jan-Komorowski-2/publication/257463940/figure/fig1/AS:213426178662404@1427896174303/Integration-of-ChIP-chip-data-for-P-Stat1-P-Stat3-and-P-Stat5-with-transcriptome_Q320.jpg)
Content uploaded by Jan Komorowski
Author content
All content in this area was uploaded by Jan Komorowski on Mar 14, 2014
Content may be subject to copyright.
Available via license: CC BY-NC 2.0
Content may be subject to copyright.
Available via license: CC BY-NC 2.0
Content may be subject to copyright.
ORIGINAL ARTICLE
The signal transducers Stat1 and Stat3 and their novel target
Jmjd3 drive the expression of inflammatory genes in microglia
Piotr Przanowski &Michal Dabrowski &Aleksandra Ellert-Miklaszewska &Michal Kloss &
Jakub Mieczkowski &Beata Kaza &Anna Ronowicz &Feng Hu &Arkadiusz Piotrowski &
Helmut Kettenmann &Jan Komorowski &Bozena Kaminska
Received: 13 May 2013 /Revised: 30 August 2013 /Accepted: 19 September 2013 /Published online: 6 October 2013
#The Author(s) 2013. This article is published with open access at Springerlink.com
Abstract
Most neurological diseases are associated with chronic inflam-
mation initiated by the activation of microglia, which produce
cytotoxic and inflammatory factors. Signal transducers and acti-
vators of transcription (STATs) are potent regulators of gene
expression but contribution of particular STAT to inflammatory
gene expression and STAT-dependent transcriptional networks
underlying brain inflammation need to be identified. In the
present study, we investigated the genomic distribution of Stat
binding sites and the role of Stats in the gene expression in
lipopolysaccharide (LPS)-activated primary microglial cultures.
Integration of chromatin immunoprecipitation-promoter micro-
array data and transcriptome data revealed novel Stat-target genes
including Jmjd3,Ccl5 ,Ezr ,Ifih1 ,Irf7 ,Uba7 ,andPim1.While
knockdown of individual Stat had little effect on the expression
of tested genes, knockdown of both Stat1 and Stat3 inhibited the
expression of Jmjd3 and inflammatory genes. Transcriptional
regulation of Jmjd3 by Stat1 and Stat3 is a novel mechanism
crucial for launching inflammatory responses in microglia. The
effects of Jmjd3 on inflammatory gene expression were indepen-
dent of its H3K27me3 demethylase activity. Forced expression
of constitutively activated Stat1 and Stat3 induced the expression
of Jmjd3 , inflammation-related genes, and the production of pro-
inflammatory cytokines as potently as lipopolysacharide. Gene
set enrichment and gene function analysis revealed categories
linked to the inflammatory response in LPS and Stat1C + Stat3C
groups. We defined upstream pathways that activate STATs in
response to LPS and demonstrated contribution of Tlr4 and Il-6
and interferon-γsignaling. Our findings define novel direct
transcriptional targets of Stat1 and Stat3 and highlight their
contribution to inflammatory gene expression.
Key Message
&Combined analysis of genomic Stat occupancy and
transcriptome revealed novel Stat target genes in LPS-
induced microglia.
&Jmjd3 transcription factor is a novel transcriptional target
of Stat1 and Stat3.
&Stat1 and Stat3 cooperate with Jmjd3 to induce the ex-
pression of pro-inflammatory genes.
&Constitutively active Stat1 and Stat3 fully mimic the LPS-
induced upregulation of inflammatory genes and secretion
of cytokines.
Keywords Inflammation .Signaltransducersandactivators of
transcription .ChIP–chip .Jmjd3 H3K27me3 histone
demethylase .Brain macrophages
Electronic supplementary material The online version of this article
(doi:10.1007/s00109-013-1090-5) contains supplementary material,
which is available to authorized users.
P. Przanowski :M. Dabrowski :A. Ellert-Miklaszewska :
M. Kloss :J. Mieczkowski :B. Kaza
Laboratory of Molecular Neurobiology, Nencki Institute of
Experimental Biology, Warsaw, Poland
A. Ronowicz :A. Piotrowski
Faculty of Pharmacy, Medical University of Gdansk, Gdansk, Poland
F. Hu :H. Kettenmann
Cellular Neurosciences, Max Delbrück Centre for Molecular
Medicine, Berlin, Germany
J. Komorowski
Department of Cell and Molecular Biology, Uppsala University,
Uppsala, Sweden
J. Komorowski
Interdisciplinary Centre for Mathematical and Computational
Modeling, Warsaw University, Warsaw, Poland
B. Kaminska (*)
Neurobiology Center, Nencki Institute of Experimental Biology,
Pasteur 3 str., Warsaw 02-093, Poland
e-mail: bozenakk@nencki.gov.pl
B. Kaminska
Laboratory of Molecular Neurobiology, Nencki Institute of
Experimental Biology, Warsaw, Poland
J Mol Med (2014) 92:239–254
DOI 10.1007/s00109-013-1090-5
Introduction
The initiation of inflammation requires global activation of
gene expression, postranscriptional regulation, epigenetic
modifications, and changes in chromatin structure [1,2].
The transcription factors nuclear factor kappa B (NF-κB)/
Rel, interferon-regulatory factor (IRF), and signal transducer
and activator of transcription (STAT) [3–5] are well-
established regulators of inflammatory gene expression, but
the specific contribution of various factors to the transcrip-
tional circuits underlying inflammation is poorly defined.
STATs are latent transcription factors that are phosphorylated
by activated cytokine receptor associated Janus kinase (JAK),
tyrosine kinase growth factor receptors, and nonreceptor tyro-
sine kinases. Nearly 40 cytokine receptors signal through
STATs, and these signaling cascades can be either pro- or
anti-inflammatory [6–8]. Activated Stats have been implicated
in the regulation of cell growth, differentiation, cell survival,
and cross-talk between cancer and immune cells. Knockout
studies have demonstrated that Stat1 plays an important role in
inflammation, growth arrest, and the promotion of apoptosis;
conversely, Stat3 promotes cell-cycle progression and cellular
transformation and prevents apoptosis [8–10]. On the other
hand, a conditional deletion of Stat3 in macrophages/
neutrophils or in endothelial cells leads to exaggerated inflam-
mation and leukocyte infiltration in multiple organs associated
with increased production of pro-inflammatory cytokines that
suggests the suppressive role of Stat3 [11,12]. Stat1 and Stat3
activated by NF-κB-dependent interleukin-6 (IL-6) ex-
pression or by interferon-γ(IFN-γ) have been implicat-
ed in inflammation [13,14]. Stat5a and Stat5b proteins
alone are weak activators of transcription, and they act
predominantly by cooperating and interacting with other
proteins [15].
However, hundreds of genes have been identified by
knockout and biochemical studies as potential Stat3 targets;
only a small fraction has been proven to be direct Stat3
targets in various cells. Recently, molecular insight into the
biology of Stats was gained from a meta-analysis of 29
available ChIP-seq data sets covering genome-wide occu-
pancy of STATs 1, 3, 4, 5A, 5B, and 6 in several cell types
including macrophages, B and T lymphocytes, murine em-
bryonic fibroblasts, embryonic cells, and liver cells [16].
The study revealed that the majority of the classical STAT-
binding sites were located near genes with cell-specific
expression, and each cell type displays a unique STAT binding
pattern.
Most neurological diseases are associated with inflamma-
tion initiated by the activation of microglia—brain resident,
myeloid cells responsible for immune surveillance. Activated
microglia initiate neuroinflammation by producing cytotoxic
and inflammatory factors such as the cytokines IL-1β,tumor
necrosis factor alpha (TNF-α), and IL-6, thereby aggravating
brain damage [17]. Gene expression profiling and biochemi-
cal studies performed on lipopolysaccharide (LPS)-stimulated
microglia in vitro and in an animal model of stroke demon-
strated gross similarity in the transcriptional activation of a
large panel of inflammatory genes [18]. It confirms that events
underlying the inflammatory response in LPS-stimulated mi-
croglia reflect those that contribute to brain inflammation. The
analysis of predominant signaling pathways revealed that Stat
signaling and a subset of potentially Stat-dependent genes are
activated in inflammatory microglia in vitro and in the ische-
mic brain.
To determine the Stat-dependent transcriptional network
and its functions in inflammation, we investigated the geno-
mic distribution of the Stat1, Stat3, and Stat5 binding sites and
the role of Stats in the control of the gene expression in LPS-
stimulated primary microglial cultures. Combined analysis of
gene expression profiles and the genomic Stat occupancy data
led to the identification novel, direct Stat target genes. In
particular, we found that Jmjd3, an H3K27me3 demethylase
and transcription factor, is a novel transcriptional target of
Stat1 and Stat3, and cooperates with Stats to induce the
expression of pro-inflammatory genes. Moreover, our data
demonstrate that constitutively active Stat1 and Stat3 fully
mimic the LPS-induced upregulation of inflammatory genes
and secretion of cytokines.
Materials and methods
Cell culture and treatments
Primary cultures of rat microglia were prepared from 1-day-old
Wistar rat pups as described previously. Briefly, cells were
isolated from cerebral cortices and plated in culture medium
[Dulbecco’s modified Eagle’s medium (DMEM) with
Glutamax and high-glucose formula 4.5 g/L, Gibco] supple-
mented with 10 % fetal bovine serum (FBS) and antibiotics in
poly-L-lysine-coated culture flasks. After 9–10 days, the loose-
ly adherent microglial cells were recovered from confluent glial
cultures by mild shaking and centrifugation. The microglial
cells were suspended and plated at a density of 2–3×10
5
cells/cm
2
in 24-well plates or 60 mm dishes. Nonadherent cells
were removed after 30 min by changing the medium. Adherent
cells (>96 % positive for isolectin B
4
) were incubated for 48 h
to silence the microglial culture. Primary microglial cultures
were prepared from cerebral cortex of newborn C57BL/6 or
TLR4 knockout mice. The TLR KO mice were generated on a
C57Bl/6 background by Dr. Shizuo Akira and colleagues from
the Osaka University, Japan and obtained from Oriental
BioServices Inc., Japan [19]. Cells were stimulated with
100 ng/ml lipopolysaccharide (LPS from Salmonella
enteritidis, Sigma, Germany). In some experiments, 10 μg/ml
cycloheximide (CHX) (Sigma-Aldrich, St. Louis, MO, USA)
240 J Mol Med (2014) 92:239–254
was added together with LPS. Histone methyltransferase inhib-
itors 3-deazaneplanocin and BIX 01294 (Sigma, Germany)
were added for 6 h at a final concentration 3 μM. Primary
mouse microglia cultures were treated by murine recombinant
IL-6 (25 ng/ml) and IFN-γ(10 ng/ml) for 1 h (cytokines were
kindly provided by ImmunoTools, Germany). Murine BV2
immortalized microglial cells (obtained from Dr. Klaus
Reymann) were cultured in DMEM supplemented with 2 %
FBS, 50 U/ml penicillin, and 50 mg/ml streptomycin. The cells
were grown in 12-well plates (for RNA isolation) or 6-well
plates (for protein extraction) at a density of 1× 10
5
or 2.5×10
5
cells per well, respectively, in a humidified atmosphere con-
taining 5 % CO
2
.
Preparation of protein extracts and Western blot analysis
Cell lysates were prepared by scraping the cells in buffer
containing phosphatase and protease inhibitors. The proteins
were then separated by sodium dodecyl sulfate polyacryl-
amide gel electrophoresis and transferred onto nitrocellulose
membranes as described previously [20]. Antibodies used for
Western blot analysis were purchased from Cell Signaling
Technology (Beverly, MA, USA) and included the following:
polyclonal antibodies recognizing phosphorylated and total
Stat1, Stat3, Stat5, Jmjd3, phosphorylated IκB, and horserad-
ish peroxidase-conjugated antirabbit IgG. Protein molecular
weights were estimated with prestained protein markers. The
membranes were stripped and reprobed with horseradish
peroxidase-conjugated anti-β-Actin antibody (Sigma-
Aldrich, St. Louis, MO, USA) to verify that equal amounts
of protein were loaded.
Chromatin immunoprecipitation and hybridization
to microarrays (ChIP–chip)
For ChIP–chip experiments, 1× 10
7
cells were stimulated for
1.5 h with LPS (100 ng/ml). Fixation with 1 % formaldehyde,
sonication, and immunoprecipitation were performed with
components of the ChIP IT kit according to the manufac-
turer’s instructions (Active Motif, Carlsbad, CA, USA).
Each sample was immunoprecipitated with 1 μgofoneof
the following antibodies: anti-P-Stat1 (sc-7988X), anti-P-
Stat3 (sc-7993X), or anti-P-Stat5 (sc-11761X) from Santa
Cruz Biotechnology (Santa Cruz, CA, USA). Normal rabbit
IgG (NI01) from Calbiochem (Darmstadt, Germany) served
as the control immunoprecipitation (IP) antibody. The amount
and quality of the DNA were determined by capillary
electrophoresis with a Bioanalyzer 2100 and a High
Sensitivity DNA LabChip kit from Agilent Technologies
(Santa Clara, CA, USA). For hybridization to microarrays,
the material was amplified with a whole-genome amplifi-
cation WGA3 kit (Sigma-Aldrich, St. Louis, MO, USA)
according to the manufacturer’s protocol. The sequences
of the primers used for ChIP-PCR are presented in the
supplementary Table 1.
The genomic positions of the Stat binding sites were iden-
tified by labeling and hybridizing the immunoprecipitated
chromatin to Rat ChIP 3 ×720K RefSeq promoter oligonucle-
otide chips (Roche NimbleGen, Waldkraiburg, Germany);
two-color competitive hybridization was performed with a
labeled input as a reference. The chromosomal locations of
the binding peaks for P-Stat1, P-Stat3, and P-Stat5 were
identified with the program NimbleScan (version 2.5) using
the default parameters. The chromosomal locations of the
peaks were imported into a relational database and mapped
to the nearest gene with Ensembl (version 60) after providing
the softwarewith the Nimblegen range (−5,000 to 1,000) from
the transcription start site of the gene. Integration of the ChIP-
seq data with gene expression data (Kolmogorov-Smirnov
test) and data visualization were performed in Mathematica
(Wolfram Research).
Transfection with plasmids and siRNA
Microglial BV2 cells were transfected with the Amaxa T kit
(Lonza, Cologne, Germany) and the A-023 program
according to the manufacturer’s instructions. For overexpres-
sion experiments, cells were transfected with 5 μg of DNA per
1×10
6
cells. The following plasmids were used: pEGFP-N1,
Stat1C (kindly provided by Dr. Toru Ouchi, New York
University, New York), Stat3C (kindly provided by Dr.
James Darnell, The Rockefeller University, New York),
Jmjd3, Jmjd3mut (deletion) (kindly provided by Dr. Susana
Sola, University of Lisbon, Lisbon), and Jmjd3mut (catalytic)
(kindly provided by Dr. Amy S. Weinmann, University of
Washington, Seattle). The cells were collected for protein
extracts and RNA isolation 18 and 24 h after transfection,
respectively.
To silence individual Stats or Jmjd3, the cells BV2 were
transfected by electroporation with 150 pmol of a control small-
interfering RNA (siRNA) (SI03650325) or Stat1
(SI00183547), Stat3 (SI01435287), and Jmjd3 (SI01079631)
specific siRNAs purchased from Qiagen (Germantown, MD,
USA). Transfected cells were seeded onto plates. After 48 h, the
transfected cells were stimulated with 100 ng/ml LPS for 3 or
6hforproteinextractsand9hforRNAcollection.
RNA isolation, reverse transcription, and qPCR analysis
RNA was isolated and complementary DNA (cDNA) was
obtained with a Ambion Cells-to-CT kit (Life Technologies,
Carlsbad, CA, USA). Real-time PCR amplifications were
performed in duplicate with cDNA as the template in a reac-
tion volume of 10 μl; the reaction contained 2x SYBR
GREEN FAST PCR Master Mix (Life Technologies,
Carlsbad, CA, USA) and primer sets for the following genes:
J Mol Med (2014) 92:239–254 241
18S rRNA,iNOS ,Irf7,Uba7,Jmjd3,Ccl5,Il-6,andPim1 .
The primer sequences are given in the Supplementary Table 1.
The expression of all PCR products was normalized to 18S
rRNA and then to untreated controls. Fold changes were
calculated by the 2
−ΔΔ Ct
method with 7500 System SDS
software (Life Technologies). Statistical analyses were
performed with Student’sttest with Statistica software
(StatSoft, Tulsa, OK, USA). The results are expressed as the
mean ± standard deviation (SD). All experiments were
performed with three independently derived microglial cul-
tures or BV2 passages.
Microarray gene expression profiling in BV2 microglial cells
BV2 cells were transfected with Stat1C and Stat3C constructs
or pEGFP-N1 (as a control); 24 h following transfection, the
cells were stimulated with LPS for an additional 9 h or left
untreated. Total RNA was isolated using RNeasy Mini Kit
(Qiagen). The amount and quality of the RNA were deter-
mined by capillary electrophoresis with a Bioanalyzer 2100
and a RNA 6000 LabChip kit from Agilent Technologies.
The microarray experiments were performed with 100 ng
of total RNA as the template. The whole-genome amplifi-
cation procedure was performed with a GeneAtlas WT
Expression Kit according to the manufacturer’sUser
Guide for the GeneAtlas Personal Microarray System
(Affymetrix, Santa Clara, CA, USA). Fragmented and la-
beled cDNA was hybridized to the Affymetrix Mouse Gene
1.1 ST Array Strip (770,317 probes including 28,853
mouse genes). Microarray data were analyzed according
to the previously described RMA preprocessing method
[21]. Briefly, probe set measurements were transformed into
gene-specific measurements using the annotation provided
in the Ensembl database. To remove potential cross-
hybridization effects, all probe sets with annotations for
more than one gene were excluded from further analyses.
Multiple probe sets were annotated for the same gene;
therefore, we choose the best probe set, which was defined
as the probe set with the smallest pvalue obtained from
the Welch’sttest, from each of these sets to obtain gene-
specific measurements. Changes in gene expression were
evaluated with the limma package in the Bioconductor
software [22]. We then computed qvalues for all of the
analyzed genes. The gene set enrichment analysis (GSEA,
describedby[23]) and overrepresentation analysis were
performed with the MSigDB gene collections (http://www.
broad.mit.edu/gsea/msigdb/index.jsp). The preprocessing
methods and all of the statistical computations (except GSEA)
were performed within the R programming environment and
Bioconductor packages. GSEA was performed with the Java
code provided by its authors. Microarray data have been
deposited under the accession number E-MEXP-3659 (The
ArrayExpress Archive, Hixton, UK).
Global H3K27me3 profiling by flow cytometry
To determine global H3K27me3 changes, cells were collected
after stimulation for 9 h with LPS or 24 h post-plasmid
transfection and fixed in 2 % formaldehyde in phosphate-
buffered saline (PBS) for 10 min at room temperature. To stop
the reaction, 125 mM glycine was added, and ice-cold 100 %
ethanol was slowly added to cells at a ratio of 1:1 (ethanol/
glycine). Cells were resuspended in MACS buffer (PBS,
125 mM EDTA, and 0.5 % bovine serum albumin), aliquoted
among two tubes (with or without primary antibody, which
served as a negative control), and permeabilized in MACS
buffer containing 0.1 % Triton X-100. The cells were incu-
bated with anti-trimethyl histone H3 (Lys 27) primary anti-
body [Upstate (Millipore), Billerica, MA, USA] diluted 1:400
in MACS buffer for 1 h at 4 °C and rinsed four times with
MACS buffer. Cells were incubated at room temperature for
1 h with Alexa Fluor 647-conjugated secondary antibody
(diluted 1:2,000, Invitrogen), washed four times in MACS
buffer, and resuspended in 2 % formaldehyde in PBS.
Fluorescence was measured with BD FACSCalibur and
CellQuest software.
Multiplex cytokine flow cytometry assay
Cytokine measurements were performed in conditioned media
using the Procarta Immunoassay Kit (Affymetrix, Santa Clara,
CA,USA)onaLuminex100flowcytometryanalyzer
(Madison, WI, USA) according to the manufacturer’sinstruc-
tions. Conditioned media were collected 24 h after the trans-
fection of BV2 cells with specific plasmids or 24 h after
treatment with LPS. The following pro-inflammatory cyto-
kines were measured: IFN-γ,Il-10/CSIF,Il-1β/ILIF2, Il-6,
RANTES/Ccl5, and TNF-α. Statistical analyses were
performed with Student’sttest with Statistica software
(StatSoft, Tulsa, USA).
Results
Whole-genome identification of P-Stat1, P-Stat3, and P-Stat5
binding sites upon inflammatory stimulation of rat primary
microglial cultures
Western blot analysis demonstrated a rapid increase in the levels
of phosphorylated Stat1, Stat3, and Stat5 from 1 to 6 h, with the
maximum induction occurring between 1 and 1.5 h after stimu-
lation of rat primary microglial cultures with LPS (Fig. 1a). The
direct Stat targets and their role in the transcriptional responses of
microglial cells are unknown. Therefore, we analyzed the distri-
bution of P-Stat1, P-Stat3, and P-Stat5 binding sites in promoter
regions by chromatin immunoprecipitation followed by hybrid-
ization to promoter arrays (ChIP–chip analysis). Microglial cells
242 J Mol Med (2014) 92:239–254
were stimulated with LPS for 1.5 h, when the peak of Stat
phosphorylation was observed (Fig. 1a). Using a false discovery
ratio (FDR) of ≤0.2, we mapped to the genome 3,527 peaks
(corresponding to 3,466 genes) for P-Stat1, 1,074 peaks (corre-
sponding to 1,166 genes) for P-Stat3, and 939 peaks (corre-
sponding to 1,012 genes) for P-Stat5 chromatin immunoprecip-
itation (Fig. 1b).
To identify genes whose transcription is directly regulat-
ed by distinct Stats, we analyzed the current ChIP–chip
data together with global gene expression data from LPS-
stimulated primary rat microglial cultures. First, we evalu-
ated whether the profile of changes in gene expression
following LPS stimulation (microarray data E-MEXP-2466
from [18]) was significantly different in the sets of genes
with and without the P-Stat1, P-Stat3, and P-Stat5 binding
peaks. Using the Kolmogorov–Smirnov test, we compared
the distributions of the log2-transformed fold changes in
the expression of the gene sets with or without P-Stat
binding peaks (Fig. 1c). The comparison was performed
either (1) for all of the genes represented both in the ChIP–
chip and expression dataset or (2) following an optional
filtration step that included only the genes with significant-
ly altered expression. When analyzed for all genes, the two
distributions were not significantly different. However,
among the genes that exhibited significantly altered expres-
sion (ttest p<0.0001, FDR< 0.0003) following LPS treat-
ment, the distributions of the log2-transformed fold changes
in gene expression were clearly different between the genes
with and without P-Stat1 and P-Stat3 binding peaks. We
did not observe this relationship for P-Stat5 binding peaks
(Fig. 1c).
To further explore the effect of P-Stat binding on gene
expression for the set of the genes with significantly
changed expression, we converted the log2 changes to
ranks and used histograms to visualize the distribution of
the ranks of the P-Stat binding genes among all the genes
(Fig. 1c, lower panel). For random binding data (i.e., no
association between the P-Stat binding and the expression
rank), a uniform distribution of ranks (red horizontal
lines in Fig. 1c) of the P-Stat binding genes would be
expected. The genes with P-Stat1 and P-Stat3 binding
sites were overrepresented among the genes most strong-
ly upregulated by LPS treatment (in the top bins of the
respective histograms). For both P-Stat1 and P-Stat3, the
overrepresentation of the P-Stat binding genes in the top
histogram bins was approximately twofold.
Figure 2shows a binding map of P-Stat1 and/or P-Stat3 to
the promoters of the 56 genes that exhibit P-Stat binding and
are most strongly (FC>2) induced by LPS treatment in
microglial cells. Heatmap representation shows the expression
of 56 genes that exhibit P-Stat binding, and are most strongly
(FC> 2) induced by LPS in microglial cells (Fig. 2a). Despite
the high similarity of the P-Stat1 and P-Stat3 binding motifs, a
majority of the genes with P-Stat binding peaks in their
regulatory regions were found to bind either P-Stat1 or P-
Stat3 but not both. A smaller subset of genes including
Mmp13 and Jmjd3 was found to bind both P-Stat1 and P-
Stat3.
Nine representative candidate genes, based on their
known functions in inflammation, were selected for valida-
tion by PCR with independent biological samples. This
group included Jmjd3 (two independent binding sites),
Ccl5,Ezr,Ifih1 ,Irf7 ,Uba7,Il-6,andPim1. The ChIP-
PCR confirmed the specific binding of Stat1 and/or Stat3
for eight of the nine promoters analyzed (Fig. 2b); the
exception was Tlr7. In particular, we confirmed that P-
Stat1 and P-Stat3 bind to the Jmjd3 gene promoter; one
of the sites originally identified as a P-Stat3 binding site
actually binds both P-Stat1 and P-Stat3. These results dem-
onstrate a subset of genes upregulated in inflammatory
microglial cells, which includes known (Il-6) and novel,
transcriptional targets of Stats (Jmjd3 ,Ccl5,Ezr,Ifih1,
Irf7,Uba7 ,andPim1 ). These data suggest that Stat1 and/or
Stat3 are required to induce expression of the target genes
coding for inflammatory proteins, which are important in
inflammation.
Silencing of Stat1, Stat3, or Jmjd3 affects the expression
of selected targeted genes in LPS-induced microglial cells
Jmjd3, a JmjC family histone demethylase and a transcrip-
tion factor, is induced by the transcription factor NF-κBin
response to microbial stimuli and has been shown to re-
move the histone H3 lysine 27 trimethylation (H3K27me3)
that is associated with transcriptional repression in lineage
determination [24]. More than 70 % of the LPS-inducible
genes were identified as Jmjd3 targets [25].
To determine the functional impact of the absence of Stat1,
Stat3, and Jmjd3 on the transcriptional changes triggered by
LPS stimulation, the expression of these three genes was
silenced with specific siRNAs. Due to the poor transfectability
of primary microglial cultures, genetic manipulation experi-
ments were performed in murine immortalized BV2
microglial cells. These cells are frequently used for mechanis-
tic studies of microglial biology and respond to LPS challenge
in a manner similar to that of primary microglial cultures. We
confirmed that LPS stimulation results in a rapid increase
in the levels of P-Stat1, P-Stat3, and Jmjd3 in these cells,
with kinetics similar to those observed in primary microglial
cultures (Fig. 3a). The regulation of Jmjd3 by NF-κBwas
previously demonstrated in LPS-stimulated bone marrow-
derived macrophages [25]. We found that Jmjd3 is regu-
lated by NF-κB in LPS-stimulated microglial cells, and
Bay 11-7082 (an inhibitor of NF-κB activation) complete-
ly inhibited the LPS-induced Jmjd3 expression in BV2
cells (Fig. 3b).
J Mol Med (2014) 92:239–254 243
Silencing of Stat1 or Stat3 expression alone produced a small
effect on Jmjd3 expression, but silencing both simultaneously
profoundly reduced the level of Jmjd3 expression. This reduc-
tion in Jmjd3 expression was similar to the reduction observed
following addition of Jmjd3-specific siRNA (Fig. 3c).
Furthermore, we investigated the effects of Stat1, Stat3, or
Jmjd3 silencing on the expression of selected genes following
LPS treatment: Il-6 ,iNos ,Ccl5,Pim1 ,Uba7 ,andIrf7 .
Quantification of data from multiple independent experiments
demonstrated that silencing of Stat1 or Stat3 alone had a weak
244 J Mol Med (2014) 92:239–254
effect on the expression of the selected genes, with the exception
of the reduced expression of Il-6 and iNos in Stat3-depleted
cells. Silencing of Stat1 and Stat3 prevented the LPS-induced
upregulation of the messenger RNA (mRNA) levels of the
selected genes. Silencing of Jmjd3 yielded a moderate effect
on the expression of these genes; this silencing effect that was
smaller than the effect of combined silencing of both Stat1 and
Stat3 (Fig. 3d).
Forced expression of constitutively active Stat1 and Stat3
induces Jmjd3 and inflammatory gene expression in BV2
microglial cells
To further investigate the role of Stat1, Stat3, and Jmjd3 in the
response of microglial cells to LPS treatment, we used con-
structs encoding constitutively active forms of Stat1, Stat1C
[26]; constitutively active Stat3, -Stat3C [10] and a wild-type
Jmjd3 [27]. A pEGFP-N1 construct served as the control. In the
absence of other stimuli, forced expression of constitutively
active Stat1 and Stat3 induced Jmjd3 expression in BV2
microglial cells to a level similar to that observed following
LPS treatment. Under these conditions, the NF-κBpathwaywas
Fig. 2 Verification of P-Stat1 and
P-Stat3 binding to selected gene
promoters using ChIP-PCR. a
Heatmap representation of the
selected gene expression in non-
stimulated cells (M, n=6)orLPS-
stimulated microglial cultures (L,
n=4). The correlation of the P-
Stat1 and P-Stat3 binding map for
the genes with P-Stat1 or P-Stat3
binding sites that were most
significantly (ttest p<0.0001)
and at least twofold induced by
LPS. The map indicates the
presence (black ) or absence
(white) of at least one P-Stat
binding peak. b
Immunoprecipitated DNA
samples were isolated from
independently derived microglial
cultures stimulated by LPS for
90 min. Immunoprecipitation
with a neutral IgG served as a
negative control (Neg ); the input
DNA was used as a positive
control. The PCR products were
resolved on 2 % agarose with
ethidium bromide. Similar results
were obtained fromthree
independent microglial cultures
Fig. 1 Integration of ChIP–chip data for P-Stat1, P-Stat3, and P-Stat5
with transcriptome analysis allows identification of novel Stat target
genes in LPS-stimulated microglia. aActivation of Stat1, Stat3, and Stat5
in LPS-stimulated primary rat microglial cultures. The immunoblot
shows the levels of phosphorylated and total Stats at various times after
stimulation with 100 ng/ml LPS; membranes were reprobed with anti-
actin antibody to ensure equal loading of proteins. bGenome-wide
occupancy of the Stat proteins and transcriptome analyses in primary
rat microglial cultures. Chromatin immunoprecipitation and promoter
arrays were used to identify the sequences bound by active Stat1, Stat3
and Stat5 proteins in LPS-stimulated microglia. A diagram indicates the
number of the genes bound by a given Stat and overlap between these
genes (left panel ). The ChIP–chip data for Stat binding were analyzed
with the expression data from primary rat microglial cultures stimulated
for 6 h with LPS (100 ng/ml) [18]. A diagram indicates the number of the
genes with P-Stat peaks and reliable expression data (right panel ). cStat-
responsive genes (represented in the ChIP–chip experiment) significantly
up- or downregulated in LPS-stimulated cells compared to control mi-
croglia (control vs. LPS ttest, p<0.0001). The cumulative distribution
functions of the log2-transformed changes in the expression of the genes
with (red )andwithout(blue ) binding peaks for the given P-Stat, the
numbers of genes containing peaks, and the p values were calculated from
the Kolmogorov–Smirnoff tests. Lower panel P-Stat binding peaks
among the general set of genes (ttest p<0.0001, FDR < 0.001) ranked
in the increasing order on the LPS-induced log2 change in expression.
Red horizontal lines show the uniform distributions that would be
expected for random binding data
J Mol Med (2014) 92:239–254 245
unaffected, as demonstrated by a representative immunoblot
showing no changes in the phospho-IκB level (Fig. 4a).
Forced expression of active Stat1orStat3ledtoasignificant
but moderate increase in the expression of the selected inflam-
matory genes (Il-6 ,iNos ,Ccl5 ,Pim1 ,Uba7,andIrf7)when
compared to controls. Overexpression of Jmjd3 induced only
a small increase in the expression of these selected genes.
However, the expression of both Stat1C and Stat3C resulted
in upregulation of the inflammatory gene expression levels
to levels that were comparable or in some cases higher than
those observed after LPS treatment (Fig. 4b). Furthermore,
overexpression of Jmjd3 with constitutively active Stat3C or
Stat1C yielded an upregulation of inflammatory gene
expression to levels higher than those observed following
LPS treatment.
To determine whether forced expression of Stat1C and/or
Stat3C alone or in combination with Jmjd3 can induce inflam-
matory activation of microglial cells, we measured the secretion
of pro-inflammatory cytokines in conditioned media using a
multiplexed cytokine ELISA (Luminex). Surprisingly,
Fig. 3 Stat1 and Stat3 regulate the expression of Jmjd3 and inflamma-
tion-related genes in LPS-stimulated BV2 microglial cells. aThe levels
of phosphorylated Stat1, Stat3, and Jmjd3 in total protein extracts from
control and LPS-stimulated microglial BV-2 cells. bBay 11-7082 (an
inhibitor of NF-κB) blocks Jmjd3 expression. The immunoblot shows the
reduced level of phosphorylated IκB and the absence of Jmjd3 expression
in LPS-stimulated microglialcells pretreated with Bay 11-7082. cSilenc-
ing of both Stat1 and Stat3 expression prevents LPS-induced Jmjd3
expression as efficiently as pretreatment with siRNA against Jmjd3.
Microglial cells were transfected by an Amaxa electroporation with
control siRNA or siRNAs specific for Stat1, Stat3, or Jmjd3. After
48 h, transfected cells were stimulated with LPS for 6 h. The efficacy of
silencing was verified by Western blot. dSilencing of both Stat1 and
Stat3 expression prevented LPS-induced proinflammatory gene expres-
sion more efficiently than Jmjd3 silencing. Microglial cells were
transfected with control siRNA or siRNAs specific for Stat1, Stat3, or
Jmjd3. After 48 h, the transfected cells were stimulated with LPS for 9 h.
The expression of selected genes in LPS-stimulated BV-2 cells was
determined by quantitative PCR. For a specific experiment, measure-
ments were compared to values for cultures transfected with control
siRNA and stimulated with LPS. The graphs show means and standard
deviations (s.d.) from three independent experiments. Statistical signifi-
cance was calculated by comparison to cells transfected with control
siRNA and stimulated with LPS. *p<0.05; **p< 0.01; ***p<0.001
246 J Mol Med (2014) 92:239–254
overexpression of Stat1C and Stat3C individually or a combi-
nation of either of the constitutive Stats with Jmjd3 increased the
production of pro-inflammatory cytokines to levels similar to
those observed following induction by LPS. The highest levels
of Ccl5, IL-6, IFN-γ,IL-1β,andTNF-αwere detected in the
conditioned media of cells overexpressing Stat1C and
Stat3C or a combination of either of these Stats with
Jmjd3. For IL10 (detected at low levels in all samples),
forced expression of constitutively active Stat1 and Stat3
resulted in a weak induction of cytokine production (Fig. 4c).
Overexpression of Jmjd3 alone caused only a small increase
in cytokine production.
Forced expression of constitutively active Stat1 and Stat3
mimics LPS-induced inflammatory responses in BV2
microglial cells
To assess whether the forced expression of Stat1C and Stat3C
contributes to the global transcriptional network underlying
inflammation, we performed gene expression profiling studies
in BV2 microglial cells overexpressing Stat1C and Stat3C or
stimulated with LPS. Most of the genes whose expres-
sion levels were significantly altered following LPS
treatment exhibited similar altered gene expression in
Stat1C+Stat3C-overexpressing cells. In most cases, overex-
pression of Stat1C and Stat3C resulted in larger changes in
gene expression than those observed in cells induced with
LPS (Fig. 5a).
We compared the percentage of genes whose expression
levels changed in the same direction in the Stat1C+Stat3C
and LPS groups. Depending on the fold-change threshold
used, the majority of genes 75–100 % (p<10
−12
) changed
in the same direction in both groups (Fig. 5b). To deter-
mine whether changes in the expression levels in these
groups correspond to the inflammatory signature, we used
aGSEA[23]. In addition, inflammatory response classes
from MSigDB gene collections were exploited to extract
these genes from the LPS and Stat1C+Stat3C groups. For
a majority of the classes, there was an enrichment of
functional categories linked to the inflammatory response
in both the LPS and Stat1C + Stat3C groups (p<0.25 sig-
nificance threshold as suggested by GSEA authors), as
shown in Fig. 5c. We also identified genes whose expres-
sion was significantly altered in opposite directions follow-
ing LPS induction or Stat1C+ Stat3C overexpression. The
majority of these oppositely modulated genes (95.8 %)
exhibited increased expression in the Stat1C +Stat3C group
and decreased expression in the LPS group. For these
genes, we performed overrepresentation analysis using all
of the classes from the MSigDB gene collection; nine of
the classes appeared to be significantly altered (Fig. 5d). A
majority of these classes are related to inflammatory and
immune responses.
Jmjd3 acts independently of its demethylase activity
in LPS-induced microglial cells
To determine whether histone H3K27me3 demethylation con-
tributes to Jmjd3 activity in LPS-induced inflammatory gene
expression in microglial cells, we used constructs encoding
the following two mutants of Jmjd3: Jmjd3mut (deletion) [27]
and Jmjd3mut (catalytic) [28]. The deletion mutant lacks the
JmjdC domain, while the catalytic mutant contains three point
mutations in the active site of the catalytic domain which
affect H3K27me3 demethylase activity (Fig. 6a). The expres-
sion levels of the wild-type and the two mutant Jmjd3 con-
structs were similar (Fig. 6b). Forced expression of Jmjd3mut
(catalytic) yielded the same changes in the expression levels of
the selected genes as those observed following expression of
wild-type Jmjd3. Deletion of the JmjdC domain abolished the
Jmjd3-dependent transcription of Il-6,iNos ,andCcl5
(Fig. 6b).
We further analyzed H3K27me3 demethylase activity
to determine whether this activity contributes to global
H3K27me3 levels in LPS-stimulated microglial cells. We
determined the relative level of H3K27me3 in LPS-
stimulated BV2 microglial cells by flow cytometry and
demonstrated that neither Jmjd3 silencing nor Jmjd3
overexpression led to global changes in H3K27me3
levels in LPS-stimulated microglial cells (Fig. 6c). As a
positive control, we used microglial cells pretreated with
epigenetic enzyme inhibitors (Fig. 6d). Taken together,
these results demonstrate that the H3K27me3 activity of
Jmjd3 does not contribute to its role in inducing the expres-
sion of inflammatory genes.
Stat1 and Stat3 activation is a secondary response driven
by early cytokine release
To elucidate upstream events leading to LPS-induced Stat
activation, we determined the levels of phosphorylated Stats
in primary microglial cultures from various Tlr knockout
mice. We found that LPS does not induce Stat phosphoryla-
tion (Fig. 7a) and upregulation of inflammatory gene expres-
sion (Fig. 7b) in microglia isolated from Tlr4 knockout mice.
Both Stat phosphorylation and upregulation of inflammatory
genes were not affected in microglial cultures from Tlr1, 2, 7,
and 9 knockouts (not shown). Addition of a medium from
LPS-stimulated microglia from Tlr4wt mice to Tlr4KO
microglial cultures restored Stat phosphorylation and gene
upregulation (Fig. 7b). It suggests that LPS via Tlr4 induces
secreted factors, which in turn activate Stats. This notion is
supported by the results of the experiment with LPS stimula-
tion in the presence of CHX, a protein synthesis inhibitor.
CHX blocks Stat phosphorylation (Fig. 7c) and significantly
reduces the expression of inflammatory genes (Il-6,iNos,
Ccl5,andPim1 ) after the LPS treatment (Fig. 7d). In
J Mol Med (2014) 92:239–254 247
detected as soon as 15 min after LPS stimulation. The second
increase in their mRNA levels was observed 1 and 1.5 h after
the LPS treatment (Fig. 7f).
Discussion
The main findings of this study are summarized as follows: (1)
Stat1 and Stat3 binding peaks were associated with the genes
most strongly induced by LPS in microglial cells; (2) chro-
matin immunoprecipitation revealed that Jmjd3 is a direct
target of Stat1 and Stat3; (3) silencing of both Stat1 and
Stat3 strongly affected novel and known Stat target genes in
LPS-stimulated microglial cells; (4) the forced expression of
constitutively active Stat1 and Stat3 was sufficient to induce
the transcription of hundreds of genes, mimicked LPS stimu-
lation and resulted in the production of inflammatory cyto-
kines; (5) Jmjd3 alone weakly induced gene expression, but
Jmjd3 in cooperation with either Stat induced a complete
inflammatory response; (6) the transcriptional effects of
Jmjd3 overexpression appeared to be independent of its
H3K27me3 demethylase activity; and (7) phosphorylation of
Fig. 5 Overexpression of constitutive Stat1C and Stat3C mimics tran-
scriptional changes induced by LPS in BV2 microglial cells. aGlobal
gene expression was studied in untreated pEGFP-N1-transfected BV2
microglial cells, cells overexpressing constitutively active Stat1C and
Stat3C or cells treated for 9 h with LPS, n= 3/group. Heatmap represen-
tation of RMA-normalized microarray expression data for genes whose
expression was significantly altered in LPS-treated cells or cells overex-
pressing constitutively active Stat1C and Stat3C compared to cells
transfected with pEGFP-N1 (q<0.05 and FC >1.5 or <0.66). For a given
gene, an average value in the control samples was computed and
subtracted from each value in an experimental sample. For the heatmap
presentation, values higher than 3 were set to 3. bInfluence of fold-
change threshold on the number of genes whose expression was altered
and on the percentage of genes modulated in the same direction in
microglial cells treated with LPS and cells overexpressing constitutively
active Stat1C and Stat3C. cGene set enrichment analysis (GSEA) of the
LPS and Stat1C+Stat3C data sets using inflammatory response classes
from the MSigDB gene collections. dOverrepresentation analysis using
classes from the MSigDB gene collections on genes whose expression
levels were significantly altered in opposing directions in LPS and
Stat1C+Stat3C data sets
Fig. 4 Overexpression of active Stat1 and Stat3 alone or in combination
with Jmjd3 induces the transcription of inflammatory genes and cytokine
secretion. aEvaluation of the expression levels of Stat1, Stat3, Jmjd3, and P-
IκB proteins in microglial cells overexpressing specific proteins. BV2
microglial cells were transfected by an Amaxa protocol with constructs
encoding constitutively active Stat1, Stat3, Jmjd3, or a pEGFP-N1 control.
Twenty-four hours post-transfection, the cells were left untreated or were
stimulated with LPS for 9 h. bRelative quantification (qPCR) of selected
gene expression in BV2 cells transfected with constructs encoding constitu-
tively active Stat1, Stat3, Jmjd3, pEGFP-N1, or stimulated with LPS. Graphs
show mean ± SD for four independent experiments. In a given experiment,
measurements were compared to the values obtained for pEGFP-N1-
transfected cells (statistical significance was calculated for comparisons of
cells transfected with pEGFP-N1 versus other cells: *p<0.05; **p<0.01;
***p<0.001; cells transfected with a single construct of Stat1 or Stat3 versus
other cells:
#
p<0.05;
##
p<0.01;
###
p<0.001). cHeatmap representation of
the selected cytokine concentrations in conditioned media collected from
BV2 microglial cells transfected with constructs encoding constitutively
active Stat1, Stat3, Jmjd3, pEGFP-N1, or stimulated with LPS (from the
same experiments as described above). Conditioned media were collected
24 h after transfections or 9 h after LPS stimulation and processed for
multiplex cytokine ELISAs using the Luminex system. A color-coded rep-
resentation of cytokine concentrations is presented. The heatmap shows
means of three independent experiments; statistical significance was calculat-
ed by comparing cells transfected with pEGFP-N1 with other cells: *p<0.05;
**p<0.01; ***p<0.001
J Mol Med (2014) 92:239–254 249
Stat1 and Stat3 is induced by an early cytokine release trig-
gered by LPS-induced stimulation of Tlr4 on microglial cells.
Based on this work, we propose a model for the molecular
events triggered by LPS treatment in which microglial cells
launch the inflammatory response (Fig. 8). First, LPS binds to
TLR4 receptor and stimulates early cytokine expression (via
NF-κB or mRNA stabilization). These cytokines when re-
leased induce Stat1 and Stat3 activation. Stat dimers in com-
bination with NF-κB upregulate Jmjd3 expression. Next,
Stat1 and Stat3, in concert with Jmjd3, activate a transcrip-
tional network that results in functional, inflammatory activa-
tion of microglia.
A recent meta-analysis of 29 ChIP-seq data sets covering
genome-wide occupancy of STATs in several cell types, most-
ly under basal conditions [16]revealedthateachcelltype
displays a unique STAT binding pattern, for that reason tran-
scriptional Stat targets could vary in distinct cells. Therefore,
to understand functions of specific Stats, it is crucial to
establish Stat genome-wide occupancy in specific cells and
under given conditions. A microglial cell is both a glial cell of
the central nervous system (CNS) and a mononuclear phago-
cyte. Microglial progenitors colonize the CNS primarily dur-
ing embryonic and fetal periods of development. In the adult
CNS, they act as sentinels of infection and injury, and partic-
ipate in both innate and adaptive immune responses [29]. We
demonstrated for the first time that Stat1, Stat3, and Stat5
binding sites are widely distributed in the microglial genome.
Although many Stat1 and Stat3 binding sites were positively
correlated with genes that are highly upregulated in the LPS-
stimulated microglial cells, we also identified many binding
sites in the promoter regions of genes whose expression was
unaltered in response to treatment. This suggests that the
binding of active Stat1 or Stat3 alone is not sufficient to
modulate the expression of these genes. Most of the tested
genes contained binding peaks for either Stat1 or Stat3. Given
the high similarity of the binding sites for Stat1 and Stat3,
Fig. 6 H3K27me3 demethylase activity of Jmjd3 is not required for the
regulation of transcriptional responses in microglial cells. aAgraphic
depicts the Jmjd3 constructs used in the experiments. Cells were
transfected with constructs encoding wild-type Jmjd3, a deletion mutant
of Jmjd3 (lacking the JmjdC catalytic domain), and a catalytic mutant of
Jmjd3 (containing three point mutations in the catalytic center that
completely abrogate demethylase activity). bQuantification of the ex-
pression of Jmjd3 and three selected inflammatory genes (Il-6 ,iNos,and
Ccl5) in microglial cells transfected with pEGFP-N1 or Jmjd3 constructs.
Total RNA was isolated 24 h after transfection using Cell-to-CT assays,
and gene expression was determined by qPCR. Graphs show means±SD
of three independent experiments (statistical significance for Jmjd3 con-
structs versus the pEGFP-N1-transfected control cells *p<0.05; **p<
0.01; ***p<0.001; or Jmjd3mut(deletion) transfected cells versus other
groups:
#
p<0.05;
##
p<0.01;
###
p<0.001). cOverexpression of Jmjd3
deficient in H3K27me3 demethylase activity does not modulate global
changes in H3K27me3 levels as determined by flow cytometry. dIn the
control experiment global H3K27me3 levels were evaluated in the cells
treated for 6 h with histone methyltransferase inhibitors: 3 μM3-
deazaneplanocin or 3 μM BIX 01294
250 J Mol Med (2014) 92:239–254
these results suggest that Stat1 and Stat3 can each compensate
for the function of the other Stat protein.
We identified a subset of Stat binding genes that were highly
upregulated in inflammatory microglial cells (Fig. 2) including Il-
6, a known target of Stats, and several novel targets: Jmjd3,
Ccl5,Ezr,Ifih1 ,Irf7 ,Uba7 ,andPim1 . These results demon-
strate that Stat1 and Stat3 are required to activate the expression
of genes coding for proteins important in the inflammatory
response. C–C motif ligand 5 (Ccl5/RANTES—regulated upon
activation, normal T cell expressed and secreted) is a chemokine
that facilitates the induction of chemotaxis in immune cells under
conditions of atherosclerosis and cerebral infarction [30]. The
Ifih1 gene encodes IFIH1 (interferon induced with helicase C
domain 1), an interferon-induced helicase known to mediate an
innate immune response [31]. Uba7 (Ube1L) encodes the key
enzyme that activates ISGylation, an interferon-inducible,
ubiquitination-like post-translational protein modification system
[32]. IRF 7 is a transcription factor that mediates interferon-γand
Toll-like receptor signaling. IRF7 expression and activity is
indispensable for appropriate type I IFN production and IFN-
Fig. 7 In LPS-stimulated microglia Stat1 and Stat3 phosphorylation is
driven by Tlr4 activation and early cytokine release. aThe levels of
phosphorylated Stat1 and Stat3 in total protein extracts from primary
microglial cultures isolated from Tlr4 wild-type and knockout mice,
unstimulated or stimulated for 3 h with LPS. Addition of a conditioned
medium from LPS-stimulated microglia from Tlr4wt mice to Tlr4KO
microglial cultures restored Stat phosphorylation and gene upregulation.
bRelative quantification (qPCR) of selected gene expression in primary
microglia cultures isolated from Tlr4 wild-type and knockout mice,
unstimulated or stimulated by LPS for 3 h. A conditioned medium from
LPS-stimulated microglia from Tlr4wt mice restored Stat phosphoryla-
tion and gene upregulation in Tlr4KO microglial cultures. The results are
shown as −ΔΔCT (log
2
RQ). Statistical significance: *p<0.05; **p<
0.01; ***p<0.001. cThe levels of phosphorylated Stat1 and Stat3 in total
protein extracts from primary microglial cultures unstimulated or stimu-
lated for 3 h with LPS in the absence or presence of cycloheximide. d
Quantification (qPCR) of selected gene expression in primary microglia
cultures unstimulated or stimulated for 6 h with LPS in the absence or
presence of cycloheximide. The results are shown as −ΔΔCT (log
2
RQ).
Statistical significance: *p<0.05; **p<0.01; ***p<0.001. eThe levels
of phosphorylated Stat1 and Stat3 in total protein extracts from murine
primary microglial cultures untreated or exposed for 3 h with LPS or
recombinant murine cytokines Il-6 or Ifn-γ.fRelative quantification
(qPCR) of the levels of Ifn-γ,Il-6,Ifn-β,andTnf-αmRNAs in primary
microglial cultures at various times after the treatment with LPS
J Mol Med (2014) 92:239–254 251
mediated physiological functions [33]. Pim1 belongs to a family
of serine/threonine protein kinases that are involved in the control
of cell growth, differentiation and apoptosis. Pim1 can be in-
duced by several external stimuli, including a number of cyto-
kines relevant to the immune response [34]. Our results provide
new evidence for direct interactions between Stats and the Pim1,
Jmjd3,Ccl5 ,Ezr ,Ifih1 ,Irf7 ,andUba7 gene promoters.
We focused on the role of Stats in the regulation of the Jmjd3
gene, which we identified as a new transcriptional Stat target.
The Jmjd3 gene has both Stat1 and Stat3 binding sites within its
promoter region. Jmjd3 is a JmjC family member and a histone
demethylase implicated in H3K27me3 demethylation, which
erases the transcriptional repression involved in lineage deter-
mination. Jmjd3 is an important regulator of inflammatory gene
transcription in peripheral macrophages [25,35]. The relevance
of the demethylase activity of JMJ enzymes in regulating cellu-
lar responses is unclear. De Santa and colleagues [25] reported
that Jmjd3, induced by LPS and interferon-γstimulation, asso-
ciates with inflammation-related genes in peripheral macro-
phages. However, most Jmjd3 target genes were not associated
with detectable levels of H3K27me3, and the expression of
Jmjd3 binding genes was unaffected or only moderately im-
paired by a mutant Jmjd3 with impaired demethylase activity. It
suggests that Jmjd3 regulates the transcriptional response in a
H3K27 demethylation-independent manner [25]. However, a
recent study showed that a small-molecule inhibitor of the
H3K27me3-specific JMJ subfamily reduces LPS-induced in-
flammatory cytokine production by human primary macro-
phages [25]. We found that inhibition of the demethylase activ-
ity of Jmjd3 did not alter the expression of the inflammatory
Jmjd3-dependent genes and had no effect on global H3K27me3
levels in LPS-activated microglial cells (Fig. 5) that support a
notion that Jmjd3 regulates the transcriptional, inflammatory
response in a H3K27 demethylation-independent manner.
This is a first report that provides evidence for Stat-
dependent regulation of Jmjd3 expression during LPS-
induced inflammatory responses. (Fig. 3b); however, NF-κB
also participates in the transcriptional regulation of Jmjd3. The
effects of Stat1 and Stat3 silencing or the ectopic expression of
persistently activated Stat1 and Stat3 on inflammation-related
genes are much stronger than the changes induced by Jmjd3
silencing or overexpression. This suggests that Stat1 and Stat3
contribute to the induction of gene expression not only
through Jmjd3 activation but also by regulating transcriptional
changes themselves (Fig. 3). Forced Jmjd3 expression in
microglial cells resulted in minor effects on the mRNA levels
of inflammatory genes and a modest effect on inflammatory
cytokine production (Fig. 4) compared to LPS or Stat overex-
pression. However, Jmjd3 cooperated with the constitutively
active Stat1C and Stat3C to upregulate the mRNA levels of
Fig. 8 Model depicting the role
of the Stat proteins in the
regulation of inflammatory
transcriptional responses in
microglia. LPS activates TLR4
receptor that results in NF-κB
activation and early cytokine
production. Released cytokines
leads to phosphorylation of Stat1
and Stat3. Activation of NF-κB,
Stat1, and Stat3 leads to the
expression of Jmjd3, which,
together with Stat1 and Stat3,
drives the expression of pro-
inflammatory genes. The dashed
line represents relations which
have not been studied in this work
252 J Mol Med (2014) 92:239–254
inflammatory genes and cytokine production to levels com-
parable to those observed following LPS stimulation (Fig. 4).
These results suggest that Jmjd3 cooperates with Stat1 and
Stat3 to additionally alter the expression level. Although data
regarding Jmjd3 DNA-binding sites in LPS-stimulated rat
microglial cells are not available, computational comparison
of our Stat binding data and Jmjd3 binding data from LPS-
stimulated mouse intraperitoneal macrophages [25]revealsa
significant correlation (p<10
−12
). This suggests that similar
target genes are regulated by Stat1, Stat3, and Jmjd3, which
cooperate to fully activate transcriptional responses during the
inflammatory response.
Interestingly, forced expression of Stat1C and Stat3C leads to
cytokine production (Fig. 4c); therefore, other signaling events
are initiated in addition to the activation of transcriptional ma-
chinery. Stat1C and Stat3C overexpression was more effective
and produced larger changes than LPS treatment, with the
exceptionofIL10production.IL10isananti-inflammatory
cytokine and participates in the resolution of the inflammatory
response. Therefore, while activation of Stat1 and Stat3 seems to
be sufficient to activate inflammation, it cannot support a com-
plex physiological response that involves termination of
inflammation.
The analysis of upstream signaling events leading to Stat
activation shows that Tlr4 knockout impaired LPS-induced Stat
phosphorylation and pro-inflammatory gene expression in pri-
mary microglial cultures. However, the effect of Tlr4 signaling
on Stat activation in microglial cells is indirect and likely caused
by early cytokine release (Fig. 7). Similar mechanism was
shownearlierinacaseofTLR4–NF-κB–IL-6–Stat3 signaling
in an experimental model of LPS/TLR4-mediated septic shock
[13]. It suggests that a loop consisting of early released cytokines
(IFN-γand IL-6) and activated Stats is responsible for most of
changes occurring in inflammatory microglia.
Altogether, our results demonstrate that Stat1 and Stat3 are
essential, and sufficient for initiating an appropriate inflamma-
tory response. This challenges the emerging idea that blocking
Stat3 may inhibit antitumor immune responses. While Stat1 has
been repeatedly linked to the initiation of classical inflammatory
responses, Stat3 is thought to act as an oncogene in cancer cells,
and as an inducer of immunosuppression when activated in
immune cells within the tumor microenvironment. There have
been many attempts to block Stat3 expression or activity in
immune cells located within the tumor microenvironment [6].
Our results show that both Stat1 and Stat3 contribute to activat-
ing the expression of inflammatory genes. It raises concerns that
blocking Stat3 may not be a good strategy for restoring the
expression of the mediator proteins necessary to activate the
immune response against tumor cells.
Acknowledgments We would like to thank Dr. Marta Maleszewska for
assistance in determining the H3K27me3 levels by flow cytometry (ex-
periments were performed in the Laboratory of Cytometry of the Nencki
Institute of Experimental Biology) and Anna Siuda for assistance in
determining cytokine levels with the Luminex system. We would like to
thank Prof. Dr. Seija Lehnardt, Karen Rosenberger, and Dr. Katja Derkow
from Charite Medical University, Berlin (who organized breeding and
transport of all TlrKO mice). We would like to thank ImmunoTools for
providing us with recombinant cytokines (IT-Box-Cy55M ImmunoTools
Award 2013). Studies were supported by grants (N301 239536) from the
Ministry of Science and Higher Education (JK) and (2011/03/N/NZ1/
03143) from the Polish National Science Centre (PP), and the Foundation
for Polish Science, the International PhD Projects program (PP). The
funders had no role in the study design, data collection, and analysis,
decision to publish, or preparation of the manuscript.
Conflict of interest All authors declare no conflict of interest.
Open Access This article is distributed under the terms of the Creative
Commons Attribution License which permits any use, distribution, and
reproduction in any medium, provided the original author(s) and the
source are credited.
References
1. Nicodeme E, Jeffrey KL, Schaefer U, Beinke S, Dewell S, Chung
CW, Chandwani R, Marazzi I, Wilson P, Coste H et al (2010)
Suppression of inflammation by a synthetic histone mimic. Nature
468:1119–1123
2. O’Shea JJ, Plenge R (2012) JAK and STAT signaling molecules in
immunoregulation and immune-mediated disease. Immunity 36:542–
550
3. Ivashkiv LB, Hu X (2004) Signaling by STATs. Arthritis Res Ther 6:
159–168
4. Hayden MS, West AP, Ghosh S (2006) NF-kappaB and the immune
response. Oncogene 25:6758–6780
5. Honda K, Taniguchi T (2006) IRFs: master regulators of signalling
by Toll-like receptors and cytosolic pattern-recognition receptors. Nat
Rev Immunol 6:644–658
6. Yu H, Kortylewski M, Pardoll D (2007) Crosstalk between cancer
and immune cells: role of STAT3 in the tumour microenvironment.
Nat Rev Immunol 7:41–51
7. Kaminska B, Swiatek-Machado K (2008) Targeting signaling path-
ways with small molecules to treat autoimmune disorders. Expert
Rev Clin Immunol 4:93–112
8. Akira S (1999) Functional roles of STAT family proteins: lessons
from knockout mice. Stem Cells 17:138–146
9. Meraz MA, White JM, Sheehan KC, Bach EA, Rodig SJ, Dighe AS,
Kaplan DH, Riley JK, Greenlund AC, Campbell D et al (1996)
Targeted disruption of the Stat1 gene in mice reveals unexpected
physiologic specificity in the JAK-STAT signaling pathway. Cell 84:
431–442
10. Bromberg JF, Wrzeszczynska MH, Devgan G, Zhao Y, Pestell RG,
Albanese C, Darnell JE (1999) Stat3 as an oncogene. Cell 98:295–303
11. Takeda K, Clausen BE, Kaisho T, Tsujimura T, Terada N, Förster I,
Akira S (1999) Enhanced Th1 activity and development of chronic
enterocolitis in mice devoid of Stat3 in macrophages and neutrophils.
Immunity 10:39–49
12. Kano A, Wolfgang MJ, Gao Q, Jacoby J, Chai GX, Hansen W,
Iwamoto Y, Pober JS, Flavell RA, Fu XY (2003) Endothelial cells
require STAT3 for protection against endotoxin-induced inflamma-
tion. J Exp Med 198:1517–1525
13. Greenhill CJ, Rose-John S, Lissilaa R, Ferlin W, Ernst M, Hertzog PJ,
Mansell A, Jenkins BJ (2011) IL-6 trans-signaling modulates TLR4-
dependent inflammatory responses via STAT3. J Immunol 186:
1199–1208
J Mol Med (2014) 92:239–254 253
14. Sikorski K, Czerwoniec A, Bujnicki JM, Wesoly J, Bluyssen HA
(2011) STAT1 as a novel therapeutical target in pro-atherogenic
signal integration of IFNγ, TLR4 and IL-6 in vascular disease.
Cytokine Growth Factor Rev 22:211–219
15. Kornfeld JW, Grebien F, Kerenyi MA, Friedbichler K, Kovacic B,
Zankl B, Hoelbl A, Nivarti H, Beug H, Sexl V et al (2008) The
different functions of Stat5 and chromatin alteration through Stat5
proteins. Front Biosci 13:6237–6254
16. Kang K, Robinson GW, Hennighausen L (2013) Comprehensive meta-
analysis of signal transducers and activators of transcription (STAT)
genomic binding patterns discerns cell-specific cis-regulatory modules.
BMC Genomics 14:4
17. Glass CK, Saijo K, Winner B, Marchetto MC, Gage FH (2010)
Mechanisms underlying inflammation in neurodegeneration. Cell
140:918–934
18. Zawadzka M, Dabrowski M, Gozdz A, Szadujkis B, Sliwa M, Lipko
M, Kaminska B (2012) Early steps of microglial activation are
directly affected by neuroprotectant FK506 in both in vitro inflam-
mation and in rat model of stroke. J Mol Med (Berl) 90:1459–1471
19. Hoshino K, Takeuchi O, Kawai T, Sanjo H, Ogawa T, Takeda Y,
Takeda K, Akira S (1999) Cutting edge: Toll-like receptor 4 (TLR4)-
deficient mice are hyporesponsive to lipopolysaccharide: evidence
for TLR4 as the Lps gene product. J Immunol 162:3749–3752
20. Sliwa M, Markovic D, Gabrusiewicz K, Synowitz M, Glass R,
Zawadzka M, Wesolowska A, Kettenmann H, Kaminska B (2007)
The invasion promoting effect of microglia on glioblastoma cells is
inhibited by cyclosporin A. Brain 130:476–489
21. Mieczkowski J, Tyburczy ME, Dabrowski M, Pokarowski P (2010)
Probe set filtering increases correlation between Affymetrix GeneChip
and qRT-PCR expression measurements. BMC Bioinforma 11:104
22. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S,
Ellis B, Gautier L, Ge Y, Gentry J et al (2004) Bioconductor: open
software development for computational biology and bioinformatics.
Genome Biol 5:R80
23. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL,
Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES et al
(2005) Gene set enrichment analysis: a knowledge-based approach
for interpreting genome-wide expression profiles. Proc Natl Acad Sci
U S A 102:15545–15550
24. De Santa F, Totaro MG, Prosperini E, Notarbartolo S, Testa G, Natoli
G (2007) The histone H3 lysine-27 demethylase Jmjd3 links inflam-
mation to inhibition of polycomb-mediated gene silencing. Cell 130:
1083–1094
25. De Santa F, Narang V, Yap ZH, Tusi BK, Burgold T, Austenaa L,
Bucci G, Caganova M, Notarbartolo S, Casola S et al (2009) Jmjd3
contributes to the control of gene expression in LPS-activated mac-
rophages. EMBO J 28:3341–3352
26. Sironi JJ, Ouchi T (2004) STAT1-induced apoptosis is mediated by
caspases 2, 3, and 7. J Biol Chem 279:4066–4074
27. Solá S, Xavier JM, Santos DM, Aranha MM, Morgado AL, Jepsen
K, Rodrigues CM (2011) p53 interaction with JMJD3 results in its
nuclear distribution during mouse neural stem cell differentiation.
PLoS One 6:e18421
28. Miller SA, Mohn SE, Weinmann AS (2010) Jmjd3 and UTX play a
demethylase-independent role in chromatin remodeling to regulate T-
box family member-dependent gene expression. Mol Cell 40:594–
605
29. Saijo K, Glass CK (2011) Microglial cell origin and phenotypes in
health and disease. Nat Rev Immunol 11:775–787
30. Mirabelli-Badenier M, Braunersreuther V, Viviani GL, Dallegri F,
Quercioli A, Veneselli E, Mach F, Montecucco F (2011) CC and
CXC chemokines are pivotal mediators of cerebral injury in ischae-
mic stroke. Thromb Haemost 105:409–420
31. Yoneyama M, Kikuchi M, Matsumoto K, Imaizumi T, Miyagishi
M, Taira K, Foy E, Loo YM, Gale M, Akira S et al (2005)
Shared and unique functions of the DExD/H-box helicases RIG-I,
MDA5, and LGP2 in antiviral innate immunity. J Immunol 175:
2851–2858
32. Dao CT, Zhang DE (2005) ISG15: a ubiquitin-like enigma. Front
Biosci 10:2701–2722
33. Ning S, Pagano JS, Barber GN (2011) IRF7: activation, regulation,
modification and function. Genes Immun 12:399–414
34. Bachmann M, Möröy T (2005) The serine/threonine kinase Pim-1.
Int J Biochem Cell Biol 37:726–730
35. Kruidenier L, Chung CW, Cheng Z, Liddle J, Che K, Joberty
G, Bantscheff M, Bountra C, Bridges A, Diallo H et al (2012)
A selective jumonji H3K27 demethylase inhibitor modulates the
proinflammatory macrophage response. Nature 488:404–408
254 J Mol Med (2014) 92:239–254
