Human inflammatory bowel diseases (IBD) including
Crohn diseases and ulcerative colitis are chronic
immune-mediated disorders whose pathogenesis and
etiology are largely unknown (1). A number of gene-
manipulated murine models of colitis have been devel-
oped. These include IL-10–deficient, T cell receptor-
α–deficient, Gαi2–deficient, Stat4-transgenic, and
T-bet-transgenic mice (2–5). Experimentally induced
models of colitis have also been established by recon-
stituting immunocompromised mice with CD4+
CD45RBhighT cells. All of these models are used in
attempts to elucidate the pathogenesis of IBD. As a
result, it has been proposed that an exaggerated bias
toward Th1 or Th2 polarization in the mucosal immune
system is a key factor in the pathogenesis of colitis (6, 7).
In IL-10–deficient mice, CD4+T cells residing in the
lamina propria region of intestinal tract preferentially
produced IFN-γ (8). Immunocompromised mice trans-
ferred with T cells from IL-10–deficient mice developed
colitis with skewed Th1 responses (9). Furthermore,
treatment with anti–IFN-γ Ab significantly delayed the
onset of the disease and reduced the severity of colitis in
young IL-10–deficient mice (8). Treatment of anti-IL-12
Ab prevented the development of the disease in young
IL-10–deficient, but not in aged IL-10–deficient, mice (7,
10). Thus, the involvement of Th1 cells has become clear
in the pathogenesis of early phase of colitis in IL-10–defi-
cient mice. IL-10 is a well-known regulatory cytokine,
which can provide positive and negative signals on a vari-
ety of immune cells including B cells, T cells, NK cells,
mast cells, and myeloid cells (11). Additionally, IL-10 has
recently been shown to play an important role in the dif-
ferentiation and function of regulatory T cells, which
possess a capacity to inhibit harmful immune disorders
including chronic colitis (12, 13). Although it is now
The Journal of Clinical Investigation|May 2003| Volume 111|Number 9
production of IL-12p40 causes
chronic enterocolitis in myeloid
cell-specific Stat3-deficient mice
Masaya Kobayashi,1Mi-Na Kweon,2Hirotaka Kuwata,1Robert D. Schreiber,3
Hiroshi Kiyono,2,4Kiyoshi Takeda,1,5and Shizuo Akira1,5
1Department of Host Defense, Research Institute for Microbial Diseases, Osaka University, Osaka, Japan
2Department of Mucosal Immunology, Research Institute for Microbial Diseases, Osaka University, Osaka, Japan
3Department of Pathology and Immunology, Center for Immunology, School of Medicine, Washington University,
St. Louis, Missouri, USA
4Division of Mucosal Immunology, Institute of Medical Science, The University of Tokyo, Tokyo, Japan
5Solution Oriented Research for Science and Technology, Japan Science and Technology Corp., Tokyo, Japan
Stat3 plays an essential role in IL-10 signaling pathways. A myeloid cell-specific deletion of Stat3
resulted in inflammatory cytokine production and development of chronic enterocolitis with
enhanced Th1 responses in mice. In this study, we analyzed the mechanism by which a Stat3 defi-
ciency in myeloid cells led to the induction of chronic enterocolitis in vivo. Even in the absence of
Stat1, which is essential for IFN-γ signaling pathways, Stat3 mutant mice developed chronic entero-
colitis. TNF-α/Stat3 double-mutant mice developed severe chronic enterocolitis with enhanced Th1
cell development. IL-12p40/Stat3 double-mutant mice, however, showed normal Th1 responses and
no inflammatory change in the colon. RAG2/Stat3 double-mutant mice did not develop enterocoli-
tis, either. These findings indicate that overproduction of IL-12p40, which induces potent Th1
responses, is essential for the development of chronic enterocolitis in Stat3 mutant mice. Further-
more, enterocolitis was significantly improved and IFN-γ production by T cells was reduced in
TLR4/Stat3 double-mutant mice, indicating that TLR4-mediated recognition of microbial compo-
nents triggers aberrant IL-12p40 production by myeloid cells, leading to the development of entero-
colitis. Thus, this study clearly established a sequential innate and acquired immune mechanism for
the development of Th1-dependent enterocolitis.
J. Clin. Invest.111:1297–1308 (2003). doi:10.1172/JCI200317085.
Received for publication October 8, 2002, and accepted in revised form
January 28, 2003.
Address correspondence to: Shizuo Akira, Department of Host
Defense, Research Institute for Microbial Diseases, Osaka
University, 3-1 Yamada-oka, Suita Osaka 565-0871, Japan.
Phone: 81-6-6879-8302; Fax: 81-6-6879-8305;
Conflict of interest: The authors have declared that no conflict of
Nonstandard abbreviations used: inflammatory bowel diseases
(IBD); phycoerythrin (PE); Toll-like receptor 4 (TLR4).
See the related Commentary beginning on page 1284.
clear that IL-10 plays an essential role in the negative reg-
ulation of inflammatory responses in the intestine, the
mechanism by which IL-10 exerts the anti-inflammato-
ry responses remains unclear.
The STAT family of transcription factors, consisting
of seven members, is involved in cytokine signaling.
The knockout of each member of the STAT family in
mice resulted in impaired responses to corresponding
cytokines, demonstrating that STAT proteins have
essential roles in cytokine-mediated biological respons-
es (14, 15). Unlike the knockout of all other STAT
family members, the knockout of Stat3 resulted in
early death in embryogenesis (16). Therefore, Stat3
has been deleted in a cell- or tissue-specific manner by
the Cre-loxP recombination system in order to exam-
ine biological and functional roles of the transcription
factor in vivo (17–19). For example, mice lacking Stat3
in myeloid cells such as macrophages, neutrophils,
and dendritic cells developed chronic enterocolitis
and showed exaggerated Th1 responses (20). The
immunological and histological analysis revealed very
similar disease development in the intestinal tract of
IL-10–deficient mice. Macrophages from the Stat3
mutant mice did not show any anti-inflammatory
responses mediated by IL-10 and produced increased
levels of inflammatory cytokines in response to bacte-
rial LPS, demonstrating that Stat3 is essential for the
IL-10–mediated signaling pathways in myeloid cells
(20). This study also indicated that myeloid cells,
including macrophages and dendritic cells, are major
targets for IL-10 exhibiting anti-inflammatory
responses in vivo, and the abnormal activation of
myeloid cells is involved in the pathogenesis of chron-
ic colitis in mice. It remains elusive, however, how the
abnormal activation of myeloid cells leads to the
development of colitis in vivo.
In this study, we examined the in vivo mechanism by
which activated myeloid cells in the absence of IL-10 sig-
naling induce chronic colitis using Stat3 mutant mice
with additional deleted alleles, such as Stat1, TNF-α,
IL-12p40, RAG2, and Toll-like receptor 4 (TLR4).
Mice. Stat3 mutant (LysMcre; Stat3flox/flox) mice on a mixed
129 × C57BL/6 genetic background (backcrossed to
C57BL/6 for three generations) were used in this study
(17, 20). Stat1-deficient mice on a C57BL/6 background
were mated with Stat3 mutant mice (21). TNF-α–defi-
cient and IL-12p40–deficient mice on a C57BL/6 back-
ground were kindly provided by K. Sekikawa and J.
Magram, respectively (22, 23). RAG2-deficient mice on a
C57BL/6 background were kindly provided by Fred Alt
(Harvard Medical School and Center for Blood Research,
Boston, Massachusetts, USA). TLR4-deficient mice were
generated as described previously (24). Littermate wild-
type or heterozygous mutant mice were used for experi-
ments as control mice. All mice were housed in a specif-
ic pathogen-free facility at the Research Institute for
Microbial Diseases, Osaka University (Osaka, Japan).
Histopathological analysis. Paraffin-embedded colon
samples were sectioned and stained with H&E. Trans-
verse sections are shown in the figures. Severity of coli-
tis was evaluated by the standard scoring system as
described previously (25). Each region of the colon
(cecum; ascending, transverse, and descending colon;
and rectum) was graded semiquantitatively from 0 (no
change) to 5 (most severe change). The grading repre-
sents an increasing incidence and degree of inflamma-
tion, goblet cell loss, crypt abscesses and ulceration,
and fibrosis in the lamina propria. The summation of
the score for each segment of the colon provides a total
disease score per mouse (0–25) where: 0 indicates no
change; 1–5, mild disease; 6–10, mild-moderate; and
11–20, severe. No mice in these studies had a score
above 20 because such severe disease results in death.
The scoring was performed blinded manner by two
independent investigators. The colitis scores were eval-
uated using the program Statview II (SAS Inc., Cary,
North Carolina, USA). P values less than 0.05 were
assumed to be statistically significant.
Cytokine production by macrophages. Mice were intraperi-
toneally injected with 2 ml of 4% thioglycollate medium
(Sigma-Aldrich, St. Louis, Missouri, USA). Three days
later, peritoneal exudate cells were isolated from the
peritoneal cavity by washing with ice-cold HBSS
(Sigma-Aldrich). Cells were cultured for 2 h and washed
with HBSS to remove nonadherent cells. Adherent
monolayer cells were used as peritoneal macrophages.
Peritoneal macrophages were cultured with 10 ng/ml of
LPS (Escherichia coli serotype O55:B5), 1 µM of CpG
DNA, or 50 U/ml of IFN-γ(Genzyme, Cambridge, Mass-
achusetts, USA) plus LPS for 24 h. Concentrations of
TNF-α, IL-6, and IL-12p40 in the culture supernatants
were measured by ELISA according to the manufactur-
er’s instructions (Genzyme; Endogen Inc., Woburn,
Massachusetts, USA). Production of nitric oxide was
measured by the Greiss method using a NO2/NO3Assay
Kit (DOJINDO, Kumamto, Japan).
Cytokine production by dendritic cells isolated from the large
intestine. The large intestine was removed aseptically.
After removal of colonic patches, the large intestine was
washed with RPMI-1640 on ice and stirred with
RPMI-1640 containing 20 mM HEPES, 50 µg/ml of gen-
tamycin, 100 U/ml penicillin, 100 µg/ml streptomycin,
10% FCS, 5 mM EDTA for 60 min at 37°C. The tissues
were then digested with collagenase D (400 U/ml;
Boehringer Mannheim GmbH, Mannheim, Germany) in
RPMI-1640 containing 10% FCS for 30 min at 37°C and
further incubated in the presence of 5 mM EDTA for 5
min. The isolated cells were purified by centrifugation
through a 14.5% Accudenz gradient (Accurate Chemical
& Scientific Corp., Westbury, New York, USA). Cells were
further incubated with mouse CD11c microbeads (Mil-
tenyi Biotech, Bergisch Gladbach, Germany) and select-
ed on autoMACS separation columns (Miltenyi
Biotech). The purified dendritic cell fractions were more
than 95% CD11c+. These cells (105) were seeded on
96-well plate and cultured in the presence or absence of
The Journal of Clinical Investigation| May 2003| Volume 111|Number 9
100 ng/ml LPS for 24 h. Concentrations of IL-6 and
IL-12 in the culture supernatants were analyzed by ELISA.
Cytokine production by splenic and colonic CD4+T cells.
Spleens were removed aseptically and single-cell sus-
pensions were prepared. After the lysis of red blood
cells, cells were incubated with CD4 microbeads and
passed through a magnetized column (Miltenyi
Biotech). The purified T cell fractions were more than
95% CD4+and more than 99% viable. Lymphocytes of
the lamina propria of the large intestine were isolated
as previously described (26). In brief, after excision of
colonic patches, the intestinal tissues were treated with
RPMI-1640 medium containing 1 mM EDTA for 20
min to remove the epithelial cells. The tissue was then
digested with 0.5 mg/ml of collagenase (400 U/ml;
Boehringer Mannheim GmbH) for 20 min at 37°C,
and this step was repeated two times. The collected
cells were washed and filtered through a nylon mesh.
CD4+T cells were purified by the magnetic activated
cell-sorter system (Miltenyi Biotech) as described above.
CD4+T cells were seeded on anti-CD3 Ab-coated (10
µg/ml in PBS) 96-well plates and cultured for 24 h.
Concentrations of IFN-γ in the culture supernatants
were analyzed by ELISA, according to the manufactur-
er’s directions (Genzyme).
Immunohistochemical analysis. Freshly obtained large
intestine was washed with PBS and frozen in Tissue-Tec
OCT compound (Sakura, Tokyo, Japan). Cryostat sec-
tions (6 µm) were fixed with cold acetone for 2 min and
dried and rehydrated with PBS. This step was followed
by blocking with PBS containing 20 mM HEPES, 10%
FBS, and 1 µg of Fc-blocking mAb (2.4G2; BD PharMin-
gen, San Diego, California, USA). Sections were stained
with FITC-conjugated anti-CD11c (HL3; BD PharMin-
gen) and phycoerythrin-conjugated (PE-conjugated)
anti-CD3ε (145-2C11; BD PharMingen) or PE-conju-
gated anti-CD11b (M1/70; BD PharMingen) mAbs for
2 h at room temperature. After washing with PBS, the
sections were analyzed by confocal microscopy (Bio-
Rad Laboratories Inc., Hercules, California, USA).
For IL-12 staining, tyramide signal amplification
method was adopted (TSA Biotin System; NEN Life
Science Products Inc., Boston, Massachusetts, USA).
Briefly, fixed cryostat sections were incubated with
biotin-conjugated IL-12 (p40/p70) mAb (C17.8; BD
PharMingen) for 1 h at room temperature, and washed
with washing buffer. Then, sections were stained with
streptavidin-HRP and biotinyl tyramide amplification
reagent. The sections were then incubated with strep-
tavidin-PE (BD PharMingen) and FITC-conjugated
anti-CD11c (HL3; BD PharMingen) or FITC-conjugat-
ed anti-CD11b (M1/70; BD PharMingen).
Northern blot analysis. Peritoneal macrophages were
treated with LPS for 2 h. Total RNA (10 µg) was extract-
ed using the TRIzol reagent (Invitrogen Corp., San
Diego, California, USA), electrophoresed, transferred
to a nylon membrane, and hybridized with cDNA
probe specific for SOCS3, IL-12p40, and β-actin.
Stat1/Stat3 double-mutant mice developed chronic colitis. In
macrophages from myeloid-specific Stat3 mutant
mice, Stat1 is constitutively phosphorylated (20).
Stat1 has been shown to be essential for signaling
mediated by IFN-γ, which is known to be an important
cytokine for the priming of macrophage activity (21).
Therefore, we expected the constitutive activation of
Stat1 to confer abnormal activity of Stat3 mutant
macrophages. To address this issue, we generated mice
lacking both Stat1 and Stat3.
We performed a histopathological analysis on colons
of 8-week-old Stat1/Stat3 double-mutant mice (Figure
1, a–d). Control and Stat1 mutant mice did not show
any sign of pathological changes. On the contrary, the
colonic wall of Stat1/Stat3 double-mutant mice was
thickened, and the gland number was severely reduced,
as is the case in Stat3 mutant mice. These histopatho-
logical changes were observed as early as 5–6 weeks of
age in both Stat3 mutant and Stat1/Stat3 double-
mutant mice. Almost all regions of the colon were
The Journal of Clinical Investigation| May 2003| Volume 111| Number 9
IBDs in Stat1/Stat3 double-deficient mice. (a–d) Histologic examination of the colons of (a) control, (b) Stat1-deficient, (c) Stat3 mutant,
and (d) Stat1/Stat3 double-mutant mice at 8 weeks. H&E staining is shown. ×20. (e)The colitis scores shown for individual mice at 16 weeks
of age were total scores for individual sections as described in Methods. The severity of colitis in Stat1/Stat3 double-mutant mice was sig-
nificantly but only partially improved compared with Stat3 mutant mice (**P < 0.01). *P < 0.001 versus group of control mice.
affected in Stat3 mutant mice, but not in Stat1/Stat3
double-mutant mice at 24 weeks of age. In addition,
occurrence of ulcer and abscess was reduced in
Stat1/Stat3 double-mutant mice compared with Stat3
single-mutant mice. Thus, Stat1/Stat3 double-mutant
mice developed chronic enterocolitis, although the
severity of the colitis was diminished compared with
Stat3 mutant mice (Figure 1e).
We next analyzed production of IFN-γ by CD4+T
cells isolated from spleens of Stat1/Stat3 double-
mutant and Stat3 mutant mice. Splenic T cells were
stimulated with PMA and ionomycin, and production
of IFN-γ and IL-4 by CD4+cells was analyzed by intra-
cellular staining (Figure 2a). IFN-γ–producing CD4+T
cells were significantly elevated in Stat3 single-mutant
mice. IFN-γ–producing cells were also increased in
Stat1/Stat3 double-mutant mice compared with that
of control mice, albeit the reduced numbers were in
comparison to Stat3 mutant T cells. We further puri-
fied CD4+T cells from the intestinal lamina propria of
these mice and analyzed for IFN-γproduction by ELISA
(Figure 2b). Intestinal CD4+T cells from Stat1/Stat3
double-mutant mice produced increased level of IFN-γ
compared with control T cells. Thus, these findings
demonstrated that double mutation of Stat1 and Stat3
resulted in the increase of IFN-γ production by intes-
tinal CD4+T cells in addition to splenic CD4+T cells,
indicating an enhanced Th1 response.
The Journal of Clinical Investigation| May 2003| Volume 111| Number 9
Enhanced IFN-γ production in Stat1/Stat3 double-defi-
cient mice. (a) Spleen cells were stimulated with PMA
and ionomycin for 6 h, then stained for CD4, fixed, and
finally stained for IFN-γ and IL-4. Cells were analyzed on
FACS by gating on CD4+population. The frequency of
cytokine-producing CD4+cells is indicated as a per-
centage. (b) CD4+T cells were purified from lamina
propria and stimulated with plate-bound anti-CD3 Ab.
The concentration of IFN-γ was measured by ELISA. (c)
Mice were intraperitoneally injected with thioglycollate,
and 3 days later peritoneal macrophages were isolated.
Macrophages were stimulated with 10 ng/ml LPS for
24 h. Concentrations of TNF-α, IL-6, and IL-12p40 in
the culture supernatants were measured.
IBDs in TNF-α/Stat3 double-deficient mice. (a–d) Histologic examination of the colons of (a) control, (b) TNF-α–deficient, (c) Stat3
mutant, and (d) TNF-α/Stat3 double-mutant mice at 16 weeks. H&E staining is shown. ×20. (e) The colitis score of TNF-α/Stat3 dou-
ble-mutant mice at 16 weeks of age. TNF-α/Stat3 double-mutant mice developed severe chronic colitis, similar to Stat3 mutant mice.
*P < 0.001 versus group of control mice. (f) Mice were intraperitoneally injected with thioglycollate, and 3 days later peritoneal
macrophages were isolated. Macrophages were stimulated with 10 ng/ml LPS for 24 h. Concentrations of TNF-α, IL-6, and IL-12p40 in
the culture supernatants were measured. *Not detected.
Based on the result obtained by the analysis of CD4+
T cells, our next experiment was to characterize
macrophages. Although the analysis of the peritoneal
macrophages does not directly reflect involvement of
macrophages in intestinal inflammation, the inflam-
matory response can be estimated in cells of other ori-
gins, because the enhanced IFN-γ production was
observed not only in the intestinal T cells but also in
splenic T cells (Figure 2, a and b). Peritoneal
macrophages isolated from Stat1/Stat3 double-mutant
mice produced increased levels of TNF-αand IL-6 com-
pared with control or Stat1 mutant macrophages,
albeit lower compared with Stat3 mutant macrophages
(Figure 2c). IL-12p40 production from Stat1/Stat3
double-mutant macrophages was comparable to that
of Stat3 single-mutant macrophages. From these find-
ings we speculated that inflammatory cytokine syn-
thesis contributes to the development of chronic ente-
rocolitis in Stat1/Stat3 double-mutant mice.
TNF-α/Stat3 double-deficient mice suffered from severe chron-
ic enterocolitis. TNF-αis involved in a variety of inflamma-
tory processes, including the development of IBD (27).
Indeed, our previous and present studies demonstrated
that the level of TNF-αproduction was increased in Stat3
mutant mice (Figure 2c). Therefore, we next generated
mice lacking both Stat3 and TNF-α. Surprisingly, inflam-
matory changes including loss of gland, infiltration by
inflammatory cells, crypt abscess, and ulcers were
observed in the colon of TNF-α/Stat3 double-mutant
mice. The severity of disease was similar to that of Stat3
mutant mice (Figure 3, a–e). All of the TNF-α/Stat3
double-mutant mice analyzed (six mice) developed severe
chronic enterocolitis, as did Stat3 mutant mice (Figure
3e). We next analyzed the LPS-induced inflammatory
cytokine production by peritoneal macrophages.
TNF-α/Stat3 double-mutant macrophages produced
increased levels of IL-6 and IL-12 compared with
macrophages isolated from control mice (Figure 3f). In
particular, an extremely high level of IL-12p40 was pro-
duced by TNF-α/Stat3 double-mutant macrophages.
In the next set of experiments, production of IFN-γby
CD4+T cells isolated from spleen or intestinal lamina
propria of TNF-α/Stat3 double-mutant mice with coli-
tis was analyzed (Figure 4). CD4+T cells isolated from
both spleen and intestinal lamina propria of TNF-α/
Stat3 double-mutant mice as well as Stat3 single-
mutant mice produced increased levels of IFN-γ. Thus,
TNF-α/Stat3 double-mutant mice also showed skewed
Th1 responses and developed chronic enterocolitis.
The Journal of Clinical Investigation| May 2003| Volume 111| Number 9
IFN-γ production by CD4+T cells in spleen and lamina propria of
TNF-α/Stat3 double-mutant mice. CD4+T cells were purified from
spleen and lamina propria (LPL), and stimulated with plate-bound
anti-CD3 Ab. The concentration of IFN-γ was measured by ELISA.
CD4+T cells from TNF-α/Stat3 double-mutant mice produced sig-
nificantly increased levels of IFN-γ, similar to Stat3 mutant mice.
IL-12/Stat3 double-mutant mice showed no inflammatory change in the colon. (a) Peritoneal macrophages were stimulated with 10 ng/ml
LPS for 24 h. Concentrations of TNF-α, IL-6, and IL-12p40 in the culture supernatants were measured. *Not detected. (b–d) Histopathol-
ogy of the colons of (b) control, (c) IL-12–deficient, and (d) IL-12/Stat3 double-mutant mice at 16 weeks. H&E staining is shown. ×20. (e)
The colitis score of IL-12/Stat3 double-mutant mice at 16 weeks of age. None of the IL-12/Stat3 double-mutant mice showed any inflam-
matory change in the colon. *P < 0.001 between Stat3 mutant mice and control or IL-12/Stat3 double-mutant mice.