B

IR

DSK

*VU

BlktspgeOTtsmtTeptlmtpvCfampTapicsh

KH

Gi

GASTROENTEROLOGY 2010;139:1277–1288

ASIC—ALIMENTARY TRACT

nterleukin-11 Reduces TLR4-Induced Colitis in TLR2-Deficient Mice andestores Intestinal STAT3 Signaling

EANNA L. GIBSON,*,‡ MARINIEVE MONTERO,* MARK J. ROPELESKI,§ KIRK S. B. BERGSTROM,* CAIXIA MA,*ANJOY GHOSH,‡,� HELEN MERKENS,¶ JINGTIAN HUANG,* LISA E. MÅNSSON,* HO PAN SHAM,*ELLY M. McNAGNY,¶ and BRUCE A. VALLANCE*

Division of Gastroenterology, BC Children’s Hospital, �The Diabetes Research Program, and ¶The Biomedical Research Centre, University of British Columbia,‡ §

ancouver, British Columbia; Department of Biology, University of British Columbia Okanagan, Kelowna, British Columbia; and Gastrointestinal Diseases Research

nit, Departments of Medicine, Anatomy and Cell Biology, and Pathology and Molecular Medicine, Queen’s University, Kingston, Ontario, Canada

Ivgmrtao(ltgfwrc

tbmNst(dbircctmt

flcTTt

BA

SIC–

ALI

MEN

TARY

TRA

CT

ACKGROUND & AIMS: The roles of intestinal Toll-ike receptors (TLR) in the pathogenesis of colitis are notnown. TLR2 and TLR4 appear to protect against dex-ran sodium sulfate–induced colitis by promoting muco-al integrity, but it is not clear whether this method ofrotection occurs in other models of colitis. We investi-ated the roles of TLR2 and TLR4 and the cell types thatxpress these receptors during infectious colitis. METH-DS: We generated chimeric mice with TLR2�/� or

LR4�/� bone marrow and infected them with the bac-erial pathogen Citrobacter rodentium. We assessed theirusceptibility to colitis and the mechanisms of TLR-

ediated mucosal integrity. RESULTS: TLR2-expressingissue resident cells prevented lethal colitis, whereasLR4-dependent inflammatory responses of hematopoi-tic cells mediated intestinal damage. TLR2 expressionrotected against intestinal damage by maintaining epi-helial barrier function and inducing expression of inter-eukin (IL)-11 from tissue resident cells in the muscularis

ucosae, concurrent with epithelial activation of theranscription factor STAT3. Addition of exogenous IL-11rotected against the lethal colitis in TLR2-deficient miceia STAT3 activation in intestinal epithelial cells. CON-LUSIONS: TLR2-dependent cytoprotective responses

rom tissue resident cells maintain mucosal integritygainst the ultimately lethal TLR4-dependent inflam-atory responses of hematopoietic cells. Whereas TLR2

rotects against various noxious agents, the role ofLR4 during colitis can be either protective or dam-ging, depending on the stimulus. Therefore, thera-eutics that reduce innate immunity (TLR2 signaling

n particular) may not be beneficial to patients witholitis; they could worsen symptoms. Therapies thattimulate cytoprotective responses, like IL-11, couldave benefits for patients with colitis.

eywords: IBD; Inflammatory Bowel Diseases; Intestinalomeostasis; Enteric Bacteria.

astrointestinal health relies on maintaining the deli-cate regulatory balance of proinflammatory and anti-

nflammatory mediators of the mucosal immune system.

ntestinal inflammatory responses must prove sufficientlyigorous to provide host defense against bacterial patho-ens and other noxious stimuli. However, the host responseust be tightly controlled because excessive inflammatory

esponses are damaging whereas the protective mechanismshat promote tissue repair and restoration of homeostasisre required to heal the resulting damage. Failed regulationf these responses can result in inflammatory bowel diseaseIBD), including Crohn’s disease and ulcerative colitis. IBDikely arises from aberrant immune responses triggered byhe normally harmless intestinal microbiota. Althoughermline-based defects in mucosal integrity and barrierunction may facilitate pathologic immune interactionsith these microbes, defects in the ability of the intestine to

epair immune-driven damage may play a central role in thehronic nature of these diseases.

Recent studies have found that the same innate receptorshat recognize and induce inflammatory responses againstacterial products play critical cytoprotective roles by pro-oting mucosal integrity and tissue repair. Rakoff–ahoum et al brought this concept to the forefront by

howing that activation of Toll-like receptors (TLRs) andheir shared adaptor protein, myeloid differentiation factorMyD)88, was unexpectedly protective against dextran so-ium sulfate (DSS)-induced damage in the gut.1 Moreover,locking TLR4 impairs mucosal healing during DSS-

nduced colitis.2 In this model, MyD88-mediated protectioneflects the recruitment of fibroblast-like cells to the base ofolonic crypts.3 TLR2 is also protective during DSS-inducedolitis by specifically regulating tight junction barrier pro-eins.4 What is still not known, however, is which cell types

ediate these specific TLR-dependent roles. Importantly,he epithelium and the underlying hematopoietic cells likely

Abbreviations used in this paper: DSS, dextran sodium sulfate; FITC,uorescein isothiocyanate; H, hematopoietic; IEC, intestinal epithelialell; IFN, interferon; IL, interleukin; MyD, myeloid differentiation factor;LR, Toll-like receptor; TNF, tumor necrosis factor; TR, tissue resident;UNEL, terminal deoxynucleotidyl transferase–mediated deoxyuridineriphosphate nick-end labeling; WT, wild-type.

© 2010 by the AGA Institute0016-5085/$36.00

doi:10.1053/j.gastro.2010.06.057

matwtc�ilcihcajfac

itthhiMttppmtwTeAtabtmriipti

C6(

ic

fs

(wTwcamia

tsavm

frrnat

aeas

f(nfltm

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1278 GIBSON ET AL GASTROENTEROLOGY Vol. 139, No. 4

ediate differential immune responses when their TLRs arectivated. The concept of a cell-specific compartmentaliza-ion of traditional proinflammatory signaling paradigmsas indeed revealed by Nenci et al, showing that constitu-

ive nuclear factor �B activation within intestinal epithelialells (IECs) protects against tumor necrosis factor (TNF)-–induced barrier damage and the spontaneous colitis seen

n mice lacking NEMO in their intestinal epithelium.5 Col-ectively, these studies raise concerns regarding the use ofurrent anti-inflammatory therapies for IBD, because block-ng inflammation may also unpredictably hinder mucosalealing. Although much has been learned about host-mi-robial interactions in defining acute responses to injurynd the mucosal repair response, one caveat is that a ma-ority of studies have solely utilized the DSS model. Tourther define the roles played by innate immunity in colitisnd mucosal protection, exploration of additional modelsould lead to a consensus on TLR function during colitis.

A robust model of infectious colitis exploits the attach-ng/effacing murine bacterial pathogen Citrobacter roden-ium, which leads to a Th1- and Th17-driven colitis charac-erized by neutrophil and macrophage infiltration, crypt cellyperplasia, mucodepletion, and barrier disruption.6–10 Weave used this model to assess the role of innate receptors in

ntestinal host defense and mucosal homeostasis whereyD88-dependent signaling played a critical role in main-

aining intestinal homeostasis while also defending againsthe pathogen.11 We established that TLR2 was vital forrotecting mucosal integrity,12 whereas TLR4 played aroinflammatory role.13 However, the impact of TLR4 onucosal integrity was not assessed in this model. Moreover,

he mechanisms, cell types, and dynamics of these responsesere not defined. In the current study, we determined thatLR2 activation protects against the development of anxaggerated and ultimately lethal TLR4-driven response.lthough TLR2 mediates barrier function and cytoprotec-

ive responses, including interleukin (IL)-11 and STAT3ctivation from tissue resident cells, TLR4 expressiony hematopoietic cells directs most of the inflamma-ion. Thus, whereas the role of TLR4 in colitis may be

odel dependent, TLR2 promotes mucosal integrity inesponse to diverse noxious agents. Therefore, mimick-ng TLR2-dependent responses by therapeutic admin-stration of cytoprotective factors such as IL-11 mayrove helpful in clinical conditions such as IBD wherehe epithelial barrier is disrupted and mucosal healings a clear objective for any new therapeutic regimen.

Materials and MethodsMiceSix- to 8-week-old C57BL/6J, C57BL/6.129-Tlr2tm1Kir/J,

57BL/6.ly5.1 (Jackson Breeders, Bar Harbor, ME), C57BL/.129-Tlr2,4DKO (double knockout), and C57BL/6 Tlr4�/�

A. Aderem, Institute for Systems Biology) mice were bred (

n-house and maintained under specific pathogen-freeonditions at Child and Family Research Institute.

Bacterial Strains and Infection of MiceMice were orally gavaged with �2.5 � 108 colony-

orming units of C rodentium DBS100 and sacrificed atpecified time points postinfection.

Bone Marrow ReconstitutionBone marrow was isolated from C57BL/6.Ly5.1

WTLy), TLR2�/�, or TLR4�/� mice. Four million cellsere injected intravenously into either wild-type (WT),LR2�/�, or TLR2,4DKO lethally irradiated recipients thatere allowed to recover for 12 weeks. Reconstitution was

onfirmed through fluorescence-activated cell sorternalysis of peripheral blood leukocytes for donor Ly5.1arkers and TLR2 or TLR4 expression. Only mice exhib-

ting greater than 85% peripheral blood chimerism werenalyzed.

Survival and Body Weight MeasurementMice were monitored for mortality and morbidity

hroughout infection and were sacrificed if they showedigns of extreme distress. Body weight data are presenteds the mean percentage of the starting weight, and sur-ival data are presented as the percentage of the initialice still surviving at each time point.

Tissue Collection and Bacterial CountsTissue collection and bacterial counts were per-

ormed as described previously.12 Briefly, mice were sac-ificed and their large intestines excised and imaged mac-oscopically (scale bar � 1 cm) before immersion in 10%eutral buffered formalin (Fisher, Ottawa, Ontario, Can-da) for histologic analyses or processing for tissue pa-hology assays.

Histopathologic ScoringTwo tissue sections from each of 5 mice were

ssessed for submucosal edema, goblet cell depletion,pithelial hyperplasia, epithelial damage, and neutrophilnd mononuclear cell infiltration with a maximum totalcore of 16, as previously described.11,12

Immunofluorescence and TerminalDeoxynucleotidyl Transferase–MediatedDeoxyuridine Triphosphate Nick-EndLabeling Staining of Colonic TissuesStaining of control and infected tissues was per-

ormed as previously described,12 using p-TySTAT3Abcam, San Francisco, CA), IL-11 (Santa Cruz Biotech-ology, Santa Cruz, CA), and smooth muscle actin–uorescein isothiocyanate (FITC; Sigma, Oakville, On-ario, Canada). Terminal deoxynucleotidyl transferase–

ediated deoxyuridine triphosphate nick-end labeling

TUNEL; Roche, Toronto, Ontario, Canada) was per-

fvO

tCkS1lt(GTGGpvgpEac

BgSaac(

hblalwpnoeCwVsu

htat

wNu

Miusggtmwwiltmcmccpf

vdTitwcTcempo

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2010 TLR2 PROTECTS AGAINST TLR4–DRIVEN COLITIS 1279

ormed as described by the manufacturer. Tissues wereisualized using a Zeiss AxioImager 2 (Zeiss, Toronto,ntario, Canada) operated through AxioImager software.

RNA Extraction and Quantitative Real-TimePolymerase Chain ReactionAfter the mice were sacrificed, colonic tissues

ransferred to RNAlater (Qiagen, Mississauga, Ontario,anada) were used to purify RNA using Qiagen RNeasyits (Qiagen). Complementary DNA was synthesized withuperscript II Reverse Transcriptase and Oligo dT 12-6mer (Invitrogen, Burlington, Ontario, Canada) fol-

owed by quantitative real-time polymerase chain reac-ion on an MJ Mini-Opticon Real-Time PCR SystemBio-Rad, Mississauga, Ontario, Canada) using IQ SYBRreen Supermix (Bio-Rad) and IL-17f (forward, TGC-ACTGTTGATGTTGGGAC; reverse, AATGCCCTG-TTTTGGTTGAA) and IL-11 (forward, TGGTGTGCT-ACAAGGCTT; reverse, ACATCAAGACTGTACGGC)rimers from Operon (Eurofins MWG; Operon, Hunts-ille, AL) as well as glyceraldehyde-3-phosphate dehydro-enase, TNF-�, MCP-1, MIP-2, and interferon (IFN)-�rimers.12,13 Quantification was determined using Gene-xMacro OM 3.0 software (Bio-Rad) and is representeds the averaged fold expression relative to the uninfectedontrol.

FITC-Dextran AssayAssays were preformed as previously described.11

riefly, uninfected and day 6 or 7 postinfection mice wereavaged with 150 �L of 80 mg mL�1 FITC-dextran (FD4;igma) 4 hours before the mice were sacrificed. Mice werenesthetized, blood was collected by cardiac puncturend added to [3%]f acid-citrate dextrose, and plasma wasollected and fluorescence quantified using a VictorX3PerkinElmer, Waltham, MA).

Western BlottingAfter the mice were sacrificed, colonic tissues were

omogenized and centrifuged and the supernatantsoiled in loading dye. Samples (50 �g protein) were

oaded per lane in sodium dodecyl sulfate/polyacryl-mide gel electrophoresis and transferred to nitrocellu-ose membranes using standard procedures. Membranesere incubated with antibodies raised in rabbit (STAT-3,-TySTAT-3 [Abcam], and �-actin [Santa Cruz Biotech-ology]) and secondary goat anti-rabbit horseradish per-xidase– conjugated antibodies and visualized using annhanced chemiluminescence detection kit with thehemiGenius detection system using GeneTools soft-are (Syngene, Frederick, MD) for densitometric analysis.alues were expressed as a ratio to �-actin probed on theame blot as a loading control and denoted in arbitrary

nits. i

In Vivo IL-11 ReconstitutionThe total of 5 ng g�1 of purified recombinant

uman IL-11 (Wyeth Pharmaceuticals, Markham, On-ario, Canada) was injected intraperitoneally for 10 days,nd mice were monitored for morbidity. Tissues wereaken and scored as described previously.

Statistical AnalysisAll the results are expressed as the mean value

ith SEM from 3 to 6 mice per group per experiment.onparametric Mann–Whitney t tests were performedsing GraphPad Prism 4 (GraphPad, La Jolla, CA).

ResultsTLR4 Is Not Required to Protect AgainstC rodentium ColitisWe have previously shown that TLR2�/� and

yD88�/� mice are highly susceptible to C rodentiumnfection, with barrier dysfunction and severe mucosallcerations.12 In contrast, our results infecting micepontaneously lacking TLR4 on diverse genetic back-rounds were conflicting, likely reflecting genetic back-round differences.13,14 To clarify the role of TLR4, ando enable direct comparison with our previous assess-

ent of TLR2 and MyD88 on the C57BL/6 background,e infected TLR4�/� mice on the C57BL/6 backgroundith C rodentium. In stark contrast to our previous find-

ngs where TLR2�/� and MyD88�/� mice succumbed to aethal colitis,11,12 TLR4�/� mice survived infection similaro WT C57BL/6 mice, with no morbidity and few signs of

ucosal injury (Figure 1). This was striking because re-ent studies have shown that both TLR2�/� and TLR4�/�

ice displayed increased susceptibility to DSS-inducedolitis based primarily on their ability to maintain mu-osal integrity.1 Our findings suggest that the roleslayed by TLR4 during C rodentium colitis are distinctrom those elucidated during DSS-induced colitis.

Tissue Resident Cells Are Critical for TLR2-Dependent Protection From Mucosal DamageBecause TLR2 expression is critical for host sur-

ival during C rodentium–induced colitis, we sought toetermine which cell types mediated its protective effect.o differentiate between hematopoietic (H) or tissue res-

dent (TR) cells, we performed reciprocal bone marrowransplantations. Bone marrow from C57BL/6.Ly5.1 miceas injected into lethally irradiated recipient TLR2�/� mice,

reating TLR2�/�TR � WTLyH chimeric mice that lackLR2 expression in tissue resident cells. Similarly, wereated WTTR � TLR2�/�H chimeric mice that lack TLR2xpression in hematopoietic cells. WTTR � WTLyH chi-eric mice were created to control for the reconstitution

rocess. None of the reconstituted mice showed any signsf spontaneous colonic bleeding or inflammation before

nfection (data not shown). Following infection, 100% of

t1sWvrrt

smtWd

dWHinadfaiT

cmpdcrtTcbutr�rcd

mcWdcfg

emTcaaMta

FMdo

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1280 GIBSON ET AL GASTROENTEROLOGY Vol. 139, No. 4

he TLR2�/�TR � WTLyH mice had to be sacrificed by day0 postinfection, a mortality rate that surpassed thateen in infected TLR2�/� mice (Figure 2A). In contrast,

TTR � TLR2�/�H mice infected with C rodentium sur-ived comparably to WTTR � WTLyH and WT mice. Ouresults indicate that tissue resident cells are primarilyesponsible for TLR2-dependent survival during C roden-ium colitis.

Because TLR2�/�TR � WTLyH infected mice could noturvive beyond day 10 postinfection, we sacrificed most

ice by day 8 to ensure we could evaluate the course ofheir pathology (Figure 2C). Colons from TLR2�/�TR �

TLyH mice infected for 8 days had significant midcolon

igure 1. TLR4 is not required to protect against C rodentium colitis.ice were infected 10 days with C rodentium. TLR4�/� infected miceo not display increased (A) morbidity or (B) damage. Histologic image:riginal magnification 200�.

amage and by day 10 had developed massive crypt cell T

estruction (Figure 2B and C) that was not observed inTTR � WTLyH or WTTR � TLR2�/�H chimeric mice.istopathology scoring confirmed that TLR2�/�TR � WTLyH

nfected mice developed worsened colitis, displaying sig-ificantly greater damage compared with WTTR � WTLyH

nd WTTR � TLR2�/�H mice (Figure 2D). We noted someegree of colonic shortening in WTTR � TLR2�/�H in-ected mice with parallel increases in intestinal edemand inflammatory infiltration. Strikingly, the colonic ep-thelial integrity was preserved compared with theLR2�/�TR � WTLyH infected mice (Figure 2B–D).

Tissue Resident Cells Mediate TLR2-Dependent Barrier FunctionWe previously found that C rodentium infection

aused excessive epithelial barrier disruption in TLR2�/�

ice before the development of overt mucosal injury,articularly in the midcolon.12 Considering that barrierisruption could play a causal role in the exaggeratedolitis experienced by TLR2�/� mice, and that tissueesident cells are critical for the TLR2-dependent pheno-ype (Figure 2), we examined if barrier dysfunction in theLR2�/� mice was due to loss of TLR2 by tissue residentells. Chimeric mice were infected with C rodentium, andarrier function was measured before the development oflcers (Figure 3A). Early in infection (day 6 postinfec-ion), TLR2�/�TR � WTLyH mice displayed increased se-um FITC-dextran levels, similar to TLR2�/� mice and

3-fold more than WT and WTTR � WTLyH mice. Thisesult reveals that TLR2 signaling from tissue residentells was required to limit epithelial barrier disruptionuring infection.Although C rodentium infection of the WTTR � TLR2�/�H

ice was limited to the apical regions of their colonicrypts, infection was more invasive in the TLR2�/�TR �

TLyH mice. Figure 3B shows C rodentium coming intoirect contact with neutrophils in the underlying submu-osa. These results indicate that TLR2-dependent barrierunction helps to prevent C rodentium from invading theut wall and activating immune cells.

Compensatory TLR4 Responses Injure theIntestinal Mucosa, Leading to Lethal Colitisin the Absence of TLR2Based on these results, it was clear that TLR2-

xpressing stromal cells played a major role in mediatingucosal protection against infectious colitis. AlthoughLR4 signaling was not essential for such protection, itould potentially still contribute to mucosal protections shown in the DSS model by compensating in thebsence of TLR2, because in earlier studies we found thatyD88�/� mice develop far greater and widespread in-

estinal injury during infection than TLR2�/� mice.11 Toddress this issue, we infected mice deficient in both

LR2 and TLR4 (TLR2,4DKO). In contrast to the rapid

wmaCTeWi

TMTtmbi

FmccTttg1TmwflWcsan(smwWsTTt

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2010 TLR2 PROTECTS AGAINST TLR4–DRIVEN COLITIS 1281

eight loss and mortality experienced by MyD88�/�

ice, the TLR2,4DKO mice appeared healthy and actu-lly gained more weight than WT mice (Figure 4A).orrespondingly, no ulcers were identified in infectedLR2,4DKO mice and their tissues showed little damageven in the distal colon, the site of maximal damage in

T mice (Figure 4B and C). Likewise, the exaggerated

igure 2. Tissue resident cellsediate TLR2-dependent mu-

osal integrity. Bone marrowhimeric mice using either WT orLR2�/� as recipients reconsti-uted with bone marrow from ei-her WTLy or TLR2�/� mice wereenerated and infected over0 days with C rodentium. (A)LR2�/�TR � WTLyH infectedice display 100% mortality,hereas WTTR � TLR2�/�H in-

ected mice survive. The midco-ons from the TLR2�/�TR �

TLyH infected mice display ul-eration and bleeding (B) macro-copically (scale bar � 1 cm)nd (C) microscopically (origi-al magnification 200�) with

D) significantly higher damagecores for epithelial damage anducodepletion when comparedith WTTR � TLR2�/�H andTTR � WTLyH mice. *Indicates

ignificance higher score inLR2�/�TR � WyLyH or WTTR �LR2�/�H or TLR2�/� comparedo WT.

nflammatory responses seen during infection of T

LR2�/� mice, including increased IFN-�, MCP-1,IP-2, IL-17f, and TNF-� expression, were lost in the

LR2,4DKO mice (Figure 4D). These results show thathe exaggerated inflammatory response in the TLR2�/�

ice is a result of compensatory TLR4 signaling thatecomes pathologic. Interestingly, the one infection-

nduced pathology seen in both TLR2�/� and

LR2,4DKO mice was exaggerated barrier disruption,

bn((samm

dmstragsoTb4wowpmwmTitn

itToTtcfiib

ptRfaaaieW

FbmbtftwwTmtct

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1282 GIBSON ET AL GASTROENTEROLOGY Vol. 139, No. 4

ecause FITC-dextran serum levels were at levels sig-ificantly higher than in infected WT or TLR4�/� mice

Figure 4E) and similar to those in TLR2�/� miceFigure 3A). These findings indicate that TLR4 expres-ion does not regulate barrier function in this modelnd that the impaired barrier function in TLR2,4DKO

ice is insufficient, in the absence of TLR4, to cause

igure 3. Tissue resident cells are responsible for TLR2-dependentarrier function preventing C rodentium tissue invasion. (A) Chimericice deficient in TLR2 signaling in tissue resident cells but maintained inone marrow cells (TLR2�/�TR � WTLyH) display increased FITC-dex-ran flux across the colonic epithelium after C rodentium 6 days postin-ection compared with chimeric mice maintaining TLR2 signaling withinhe tissue resident cells (WTTR � WTLyH and WTTR � TLR2�/�H). Thisas similar to FITC-dextran flux in infected TLR2�/� mice comparedith WT mice. *Significance � higher barrier dysfunction in TLR2 orLR2�/�TR � WyLyH compared to WT mice. (B) Immunofluorescence ofidcolonic tissues from infected bone marrow chimeric mice. C roden-

ium (Tir; green and arrow), neutrophil (myeloperoxidase; red), and nu-lear staining (4=,6-diamidino-2-phenylindole; blue). Original magnifica-ion: top panel, 200�; bottom panel, 400�.

ucosal ulcerations or lethal colitis. d

TLR4-Dependent Signaling From BoneMarrow Cells Worsens Mucosal DamageThe mucosal ulcers seen in infected TLR2�/� mice

id not manifest in the absence of TLR4. Moreover, chi-eric TLR2�/�TR � WTLyH mice developed increased colitis

cores compared with TLR2�/� mice during infection. Weherefore hypothesized that TLR4 and perhaps also TLR2esponses from hematopoietic cells were maladaptive in thebsence of tissue resident TLR2 signaling. To test this, weenerated chimeric mice using TLR2,4DKO recipients recon-tituted with bone marrow from TLR2�/� (TLR4 signaling)r TLR4�/� (TLR2 signaling) mice. We found that infectedLR2,4DKOTR � TLR2�/�H mice displayed increased mor-idity, including greater weight loss, compared with TLR2/DKO or TLR2,4DKOTR � TLR4�/�H mice (Figure 5A). Thereere distinct ulcerations found in the midcolon and cecaf the TLR2,4DKOTR � TLR2�/�H mice that correspondedith increased histologic damage, scoring higher for allarameters compared with TLR2,4DKOTR � TLR4�/�H

ice (Figure 5B–D). Not surprisingly, IFN-� expressionas significantly higher in TLR2,4DKOTR � TLR2�/�H

ice compared with TLR2,4DKO and TLR2,4DKOTR �LR4�/�H mice, whereas TNF-� expression was similarly

ncreased in both chimeric mice (Figure 5E). This revealshat TLR4 from hematopoietic cells induces IFN-� butot TNF-� expression.We previously found that TLR2 was important in lim-

ting colonocyte apoptosis during C rodentium infec-ion.12 Here we found that colons from infectedLR2,4DKOTR � TLR2�/�H mice had increased numbersf TUNEL-positive cells compared with TLR2,4DKOTR �LR4�/�H and TLR2,4DKO mice. These results indicate

hat TLR4 signaling from hematopoietic cells driveolonocyte apoptosis. Curiously, there was no major dif-erence in C rodentium colonization (data not shown),ndicating that despite the critical role of TLR4 in caus-ng inflammation and pathology, it had little impact onacterial loads.

TLR2 Expression Regulates STAT3Activation in IECs Associated With IL-11Expression From Tissue Resident Cells in theMuscularis MucosaeTo determine what other factors could be im-

ortant in mediating TLR2-dependent mucosal protec-ion, we examined various tissue repair responses.ecent studies indicate that STAT3, a nuclear transcription

actor expressed within the intestinal epithelium, playsn important role in mucosal tissue repair.15,16 Tossess whether STAT3 was activated in this model, wenalyzed colonic tissues from uninfected and day 10nfected WT and TLR2�/� mice for phospho-STAT3xpression via Western blotting and immunostaining.

e found that p-STAT3 levels increase in WT mice

uring infection with C rodentium in the distal and

m(lWrtr

tItpT

FrlodT(dsodmciCdtksTtssTfbwl*rc

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2010 TLR2 PROTECTS AGAINST TLR4–DRIVEN COLITIS 1283

id regions of the colon but not in TLR2�/� miceFigure 6A). This corresponded to p-STAT3 nuclearocalization in IECs near the base of colonic crypts in

T mice but not in TLR2�/� mice (Figure 6B). Thisevealed that STAT3 activation and p-STAT3 nuclearranslocation in IECs were TLR2 dependent during C

igure 4. Compensatory TLR4esponses are responsible for theethal colitis in the absencef TLR2. Mice were infected 10ays with C rodentium. (A)LR2,4DKO infected mice do not

B) display increased morbidity oramage in the midcolon macro-copically (arrows indicate ulcers)r (C) score higher for microscopicamage (bottom panel, originalagnification 200�). In A, *indi-

ates significantly less weight lossn TLR2,4DKO compared to WT. In

, *indicates significantly lessamage in TLR2,4DKO comparedo WT. (D) Exaggerated chemo-ine and cytokine gene expres-ion from midcolonic tissues fromLR2�/� infected mice is lost inhe absence of TLR4. *Indicatesignificantly higher gene expres-ion in TLR2�/� compared toLR2,4DKO. (E) TLR2,4DKO in-

ected mice displayed increasedarrier permeability comparedith WT and TLR4�/� mice, simi-

ar to TLR2�/� mice (Figure 3A).Indicates significance higher bar-ier dysfunction in TLR2,4DKO

ompared to WT during infection.

odentium infection. I

There are several factors known to activate or signalhrough STAT3 in IECs, including IL-31, IL-22, IL-24,L-6, and IL-11, a cytokine associated with tissue pro-ection.17 Although IL-11 immunofluorescence ap-eared similar in the colonic epithelium of WT andLR2�/� mice, we identified decreased numbers of

L-11– expressing cells in the muscularis mucosae in

Ttmpshwn

eco

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1284 GIBSON ET AL GASTROENTEROLOGY Vol. 139, No. 4

LR2�/� mice compared with WT mice during infec-ion (Figure 6C). These cells costained for smooth

uscle actin and localized to the muscularis mucosae,utatively identifying them as myofibroblasts. Corre-pondingly, WTTR � TLR2�/�H chimeric infected micead similar IL-11 staining patterns as WT mice,hereas the staining in TLR2�/�TR � WTLyH was sig-

ificantly reduced. These results indicate that IL-11 s

xpression by a population of cells underlying theolonic crypts was induced by C rodentium infectionnly in the presence of TLR2 signaling.

Exogenous IL-11 Therapy Can RescueLethal ColitisBecause IL-11 is known to promote mucosal tis-

Figure 5. TLR4 signaling fromhematopoietic cells is the pri-mary source of inflammation-induced damage. Bone marrowchimeric mice using TLR2,4DKO

as a recipient reconstitutedwith bone marrow from eitherTLR2�/� or TLR4�/� donorswere infected 10 days with C ro-dentium. (A) TLR2,4DKOTR �TLR2�/�H infected mice displayincreased morbidity and dam-age in the midcolons (B) macro-scopically and (C) microsc-opically (original magnification200�) and (D) significantly in-creased damage scores com-pared with TLR2,4DKOTR �TLR4�/�H and TLR2,4DKO mice. InA, * indicates significantly moreweight loss in TLR2,4DKOTR �TLR4�/�H or TLR2,4DKOTR �TLR2�/�H compared to TLR2,4DKO

at specific time points. In D, * Indi-cates significantly more damage inTLR2,4DKOTR � TLR2�/�H com-pared to TLR2,4DKO. (E) IFN-� tran-scripts are significantly higher inTLR2,4DKOTR � TLR2�/�H infectedmice, whereas TNF-� transcriptsare similarly increased when eitherTLR2 or TLR4 is active from hema-topoietic cells compared withTLR2,4DKO mice. *Indicates sig-nificantly higher gene expressionin TLR2,4DKOTR � TLR2�/�H

compared to TLR2,4DKO. (F)TUNEL staining identified in-creased cell death in colons ex-pressing TLR4 from hematopoi-etic cells. *Indicates significanceincreased cell death inTLR2,4DKOTR � TLR2�/�H com-pared to TLR2,4DKO andTLR2�/� compared to WT dur-ing infection.

ue repair and its staining pattern is altered in infected

Tacitsaaennntssa

lpstDtduIbar

FICsftSWTa�rm1gmpilWSwTc2mmdWasdamtt(afc

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2010 TLR2 PROTECTS AGAINST TLR4–DRIVEN COLITIS 1285

LR2�/� mice, as compared with WT mice, we nextssessed whether exogenously increasing IL-11 levelsould rescue TLR2�/� mice from their lethal colitis. Dailynjections of IL-11 into infected TLR2�/� mice preventedhe development of mucosal ulcerations. Figure 7A–Chows that exogenous IL-11 kept infected TLR2�/� micelive and protected them from weight loss, ulcerations,nd severe damage to the midcolon. To determine ifxogenous IL-11 mediated its protection via STAT3 sig-aling, we examined tissues from these mice by immu-ofluorescence. We also found p-STAT3 localized to theucleus in cells at the base of colonic crypts in micereated with recombinant IL-11 (Figure 7D). These resultshow that therapeutic use of exogenous IL-11 can rescueusceptible TLR2�/� mice from lethal infectious colitis,

igure 6. STAT3 activation inECs is TLR2 dependent during

rodentium infection and as-ociated with IL-11 expressionrom tissue resident cells inhe muscularis mucosae. (A) p-TAT3 is increased in infectedT mice but not in infected

LR2�/� mice. Western blotnalysis of pSTAT3, STAT3, and-actin from colonic tissues from

epresentative WT and TLR2�/�

ice uninfected or infected for0 days with C rodentium. Barraph shows fold differences ofice infected to uninfected-STAT3/STAT3 ratios. (B) IECs

n transverse sections of midco-onic tissues from infected

T mice stained for nuclear p-TAT3 at day 10 postinfection,hich was decreased in infectedLR2�/� mice (pSTAT3, red; nu-lear stain with 4=,6-diamidino--phenylindole, blue). (C) Nu-erous cells in the muscularisucosa were positive for IL-11uring C rodentium infection inT and WTTR � TLR2�/�H mice,

nd this is impaired in the ab-ence of TLR2 from tissue resi-ent cells shown in both TLR2�/�

nd TLR2�/�TR � LyWTH infectedice (white arrows). The panel to

he right shows with a white arrowhe muscularis mucosa regionstains positive for smooth musclectin) that is magnified (WT unin-ected). IL-11, red; smooth mus-le actin, green.

t least in part through activation of STAT3. t

Conclusions

Recent studies have shown that TLRs, in particu-ar TLR2 and TLR4, in the gastrointestinal tract play arotective role by promoting tissue repair and homeosta-is. Despite this novel and important finding, much ofhe data concerning this role for TLRs has come from theSS model.1,4,12 One exception is TLR5, which was found

o protect against spontaneous colitis in a TLR4-depen-ent fashion.18 The specific cell types and mechanismsnderlying these responses remain poorly characterized.

n this study, we used C rodentium–induced colitis andone marrow chimeric mice to examine the specific rolesnd cell types that mediate TLR2- and TLR4-dependentesponses during infectious colitis. We discovered that

issue resident cells mediated TLR2-dependent protective

rttIdtmt

eaicpdlcaanitee

plhiT

bZetppcwbocc

ttsieiwvmnfdpeit

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1286 GIBSON ET AL GASTROENTEROLOGY Vol. 139, No. 4

esponses in the gut. These responses include the main-enance of barrier function, IL-11 expression by cells inhe muscularis mucosae, and STAT3 activation in IECs.n contrast, hematopoietic cells mediated the TLR4-ependent responses that were ultimately damaging tohe host. Thus, TLR2 cytoprotective responses are critical to

aintain mucosal integrity against an exaggerated and ul-imately lethal TLR4-dependent inflammatory response.

Although the primary function of TLR4 in hematopoi-tic cells is the induction of an inflammatory responseimed at protecting the host from the invading bacteria,ts role in this model actually proved maladaptive,ausing tissue damage and ulceration. This is in contrast toromotion of mucosal integrity by TLR4 during DSS-in-uced colitis.1 During infectious colitis, TLR4 expression

ed to large numbers of infiltrating immune cells and in-reased expression of proinflammatory cytokines. Addition-lly, TLR4-expressing hematopoietic cells drove colonocytepoptosis. Curiously, the exaggerated TLR4 response didot have a significant effect on pathogen load. These find-

ngs reveal that during C rodentium–induced colitis, mostissue damage is due to TLR4-driven responses while TLR2xpression limits the actions of TLR4, thereby preventingxaggerated pathology.

The intestinal barrier integrity between adjacent IECsrovides a physical obstruction to limit the infiltration of

uminal antigens that can activate immune cells. TLR2as been previously shown to maintain barrier function

n the DSS model of colitis.4 We previously found that

LR2�/� mice experience barrier dysfunction, both at p

aseline and during C rodentium infection, by regulatingO-1 and claudin 3 levels.12 Here we show that TLR2xpression by tissue resident cells promotes barrier func-ion and this led to lethal colitis if the mice also ex-ressed TLR4, because TLR2,4DKO infected mice dis-layed overt barrier dysfunction but did not developolitis. This result supports recent findings by Su et al,here pathophysiologically relevant intestinal epithelialarrier dysfunction was insufficient to cause colitis on itswn, but under certain conditions this dysfunction ac-elerated the onset and severity of immune-mediatedolitis.19

In this study, we found IL-11 expression by cells withinhe intestinal muscularis mucosa, an area known to con-ain myofibroblasts.20 Because myofibroblasts are a majorource of IL-11,21 we hypothesize that these cells aremportant for IL-11 expression in our model. The IL-11xpression coincided with evidence of STAT3 activationn IECs of C rodentium–infected mice. Furthermore, weere able to prevent the normally lethal colitis that de-

elops in infected TLR2�/� mice through exogenous ad-inistration of IL-11. This was associated with increased

uclear localization of p-STAT3 within IECs. We alsoound a modest protective effect in this model using lowoses of IL-6 (not shown). IL-11 has been found torotect against intestinal damage,22 and although thexact mechanisms are unknown, one study found IL-11nduces heat shock proteins that protect against oxida-ive stress,23 while another study found IL-11 activation

Figure 7. Cytoprotective IL-11therapy prevents lethal colitis insusceptible hosts via STAT3 ac-tivation in IECs. TLR2�/� in-fected mice were treated with re-combinant IL-11 for 10 daysduring C rodentium infection anddisplayed (A) survival similar toWT mice, (B) less weight loss,and less damage (C) histologi-cally and (D) quantitatively thanTLR2�/� mock controls (originalmagnification 200�), and (E)this corresponded with nuclearpSTAT3 in IECs (top panel:pSTAT3, red; bottom panel:merged p-STAT3 and nuclearstain with 4=,6-diamidino-2-phe-nylindole, blue). * Indicates sig-nificantly less damage inTLR2�/� � rIL-11 and WT com-pared to TLR2�/�.

rotected against cell death.17 Importantly, clinical trials

op

SNiphsOIcii1sa

NfAtotbihibevwpbttmcaatta

1

1

1

1

1

1

1

1

1

1

2

2

2

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2010 TLR2 PROTECTS AGAINST TLR4–DRIVEN COLITIS 1287

f exogenous IL-11 have shown promise for IBD thera-y.24,25

This is the first report to assess IL-11 expression andTAT3 activation in response to C rodentium infection.otably, STAT3 can be activated by IL-11 and IL-6 and

ts activation in IECs, along with macrophages, has arotective role against IBD.26 –28 Furthermore, studiesave found that IL-11 activates STAT3 in IECs, therebyuppressing experimental colitis29,30 and human IBD.31

ur study supports these findings where exogenousL-11 treatments promote STAT3 activation in IECs, inoncert with providing protection against acute colitis. Its attractive to hypothesize that endogenous IL-11 innfected WT mice activates STAT3 in IECs, because IL-1–positive cells are positioned directly below the IECs,uggesting IL-11 could be secreted to act on these cells,lthough additional mechanisms are also likely.

In the gut, certain molecules such as TNF-� andEMO have proinflammatory roles yet are also required

or protection of the mucosa against noxious stimuli.2,5

lthough anti–TNF-� therapy has revolutionized IBDreatment, it is not efficacious in approximately one-thirdf patients with IBD and its efficacy is not long-lasting inhe majority of responsive patients.32 This may be due tooth the protective and damaging roles that TLR signal-

ng can have. This study points to the potential of addingealing targets to therapies for IBD. We found that TLR2

s protective through maintenance of the epithelialarrier, as well as through the induction of IL-11xpression, with protective effects that appear to in-olve epithelial STAT3 signaling. One must be carefulhen contemplating inhibition or activation of certainroinflammatory molecules in IBD such as STAT3,33

ecause these molecules may be protective in certain cellypes. Although injections of IL-11 or other cytoprotec-ive molecules may prove beneficial in IBD, the develop-

ent of severe thrombosis or cancer would be a seriousoncern.34 It is possible, however, that combination ther-py using both lowered doses of anti-inflammatory ther-pies in addition to cytoprotective factors such as IL-11o induce tissue repair would be more effective in thereatment of IBD than anti-inflammatory treatmentslone.

References

1. Rakoff-Nahoum S, Paglino J, Eslami-Varzaneh F, et al. Recogni-tion of commensal microflora by Toll-like receptors is required forintestinal homeostasis. Cell 2004;118:229–241.

2. Ungaro R, Fukata M, Hsu D, et al. A novel Toll-like receptor 4antagonist antibody ameliorates inflammation but impairs muco-sal healing in murine colitis. Am J Physiol Gastrointest LiverPhysiol 2009;296:G1167–G1179.

3. Brown SL, Riehl TE, Walker MR, et al. Myd88-dependent posi-tioning of Ptgs2-expressing stromal cells maintains colonicepithelial proliferation during injury. J Clin Invest 2007;117:

258–269.

4. Cario E, Gerken G, Podolsky DK. Toll-like receptor 2 controlsmucosal inflammation by regulating epithelial barrier function.Gastroenterology 2007;132:1359–1374.

5. Nenci A, Becker C, Wullaert A, et al. Epithelial NEMO links innateimmunity to chronic intestinal inflammation. Nature 2007;446:557–561.

6. Bergstrom KS, Guttman JA, Rumi M, et al. Modulation ofintestinal goblet cell function during infection by an attachingand effacing bacterial pathogen. Infect Immun 2008;76:796–811.

7. Eckmann L. Animal models of inflammatory bowel disease:lessons from enteric infections. Ann N Y Acad Sci 2006;1072:28–38.

8. Ishigame H, Kakuta S, Nagai T, et al. Differential roles of inter-leukin-17A and -17F in host defense against mucoepithelial bac-terial infection and allergic responses. Immunity 2009;30:108–119.

9. Luperchio SA, Schauer DB. Molecular pathogenesis of Citrobacterrodentium and transmissible murine colonic hyperplasia. MicrobesInfect 2001;3:333–340.

0. Ma C, Wickham ME, Guttman JA, et al. Citrobacter rodentiuminfection causes both mitochondrial dysfunction and intestinalepithelial barrier disruption in vivo: role of mitochondrial as-sociated protein (Map). Cell Microbiol 2006;8:1669–1686.

1. Gibson DL, Ma C, Bergstrom KS, et al. MyD88 signalling plays acritical role in host defence by controlling pathogen burden andpromoting epithelial cell homeostasis during Citrobacter roden-tium-induced colitis. Cell Microbiol 2008;10:618–631.

2. Gibson DL, Ma C, Rosenberger CM, et al. Toll-like receptor 2plays a critical role in maintaining mucosal integrity duringCitrobacter rodentium-induced colitis. Cell Microbiol 2008;10:388–403.

3. Khan MA, Ma C, Knodler LA, et al. Toll-like receptor 4 contrib-utes to colitis development but not to host defense duringCitrobacter rodentium infection in mice. Infect Immun 2006;74:2522–2536.

4. Vallance BA, Deng W, Jacobson K, et al. Host susceptibility to theattaching and effacing bacterial pathogen Citrobacter rodentium.Infect Immun 2003;71:3443–3453.

5. Pickert G, Neufert C, Leppkes M, et al. STAT3 links IL-22 signal-ing in intestinal epithelial cells to mucosal wound healing. J ExpMed 2009;206:1465–1472.

6. Jiang H, Patel PH, Kohlmaier A, et al. Cytokine/Jak/Stat signalingmediates regeneration and homeostasis in the Drosophila mid-gut. Cell 2009;137:1343–1355.

7. Wang RJ, Peng RY, Gao YB, et al. Jak/STAT signaling pathway ofIL-11 in the protection of intestinal epithelial cells from neutronradiation [in Chinese]. Xi Bao Yu Fen Zi Mian Yi Xue Za Zhi2009;25:27–30.

8. Vijay-Kumar M, Sanders CJ, Taylor RT, et al. Deletion of TLR5results in spontaneous colitis in mice. J Clin Invest 2007;117:3909–3921.

9. Su L, Shen L, Clayburgh DR, et al. Targeted epithelial tightjunction dysfunction causes immune activation and contributesto development of experimental colitis. Gastroenterology 2009;136:551–563.

0. Grassl GA, Valdez Y, Bergstrom KS, et al. Chronic enteric salmo-nella infection in mice leads to severe and persistent intestinalfibrosis. Gastroenterology 2008;134:768–780.

1. Hoang B, Trinh A, Birnbaumer L, et al. Decreased MAPK- andPGE2-dependent IL-11 production in Gialpha2�/� colonic myo-fibroblasts. Am J Physiol Gastrointest Liver Physiol 2007;292:G1511–G1519.

2. Boerma M, Wang J, Burnett AF, et al. Local administration ofinterleukin-11 ameliorates intestinal radiation injury in rats. Can-

cer Res 2007;67:9501–9506.

2

2

2

2

2

2

2

3

3

3

3

3

R

RVb

A

T

C

F

RBSCCIB

ASIC

–A

LIMEN

TARY

TRA

CT

1288 GIBSON ET AL GASTROENTEROLOGY Vol. 139, No. 4

3. Ropeleski MJ, Tang J, Walsh-Reitz MM, et al. Interleukin-11-induced heat shock protein 25 confers intestinal epithelial-specific cytoprotection from oxidant stress. Gastroenterology2003;124:1358–1368.

4. Herrlinger KR, Witthoeft T, Raedler A, et al. Randomized, doubleblind controlled trial of subcutaneous recombinant human inter-leukin-11 versus prednisolone in active Crohn’s disease. Am JGastroenterol 2006;101:793–797.

5. Sands BE, Winston BD, Salzberg B, et al. Randomized, con-trolled trial of recombinant human interleukin-11 in patientswith active Crohn’s disease. Aliment Pharmacol Ther 2002;16:399–406.

6. Sugimoto K. Role of STAT3 in inflammatory bowel disease. WorldJ Gastroenterol 2008;14:5110–5114.

7. Takeda K, Noguchi K, Shi W, et al. Targeted disruption of themouse Stat3 gene leads to early embryonic lethality. Proc NatlAcad Sci U S A 1997;94:3801–3804.

8. Alonzi T, Newton IP, Bryce PJ, et al. Induced somatic inactivationof STAT3 in mice triggers the development of a fulminant form ofenterocolitis. Cytokine 2004;26:45–56.

9. Kiessling S, Muller-Newen G, Leeb SN, et al. Functional expres-sion of the interleukin-11 receptor alpha-chain and evidence ofantiapoptotic effects in human colonic epithelial cells. J BiolChem 2004;279:10304–10315.

0. Peterson RL, Wang L, Albert L, et al. Molecular effects of recom-binant human interleukin-11 in the HLA-B27 rat model of inflam-matory bowel disease. Lab Invest 1998;78:1503–1512.

1. Sands BE, Bank S, Sninsky CA, et al. Preliminary evaluation ofsafety and activity of recombinant human interleukin 11 in pa-tients with active Crohn’s disease. Gastroenterology 1999;117:58–64.

2. Cosnes J, Seksik P. Inflammatory bowel diseases: from sul-fasalazine to biologics [in French]. Gastroenterol Clin Biol 2009;

33:692–701. H

3. Mitsuyama K, Matsumoto S, Masuda J, et al. Therapeutic strat-egies for targeting the IL-6/STAT3 cytokine signaling pathway ininflammatory bowel disease. Anticancer Res 2007;27:3749–3756.

4. Bollrath J, Phesse TJ, von Burstin VA, et al. gp130-mediatedStat3 activation in enterocytes regulates cell survival and cell-cycle progression during colitis-associated tumorigenesis. Can-cer Cell 2009;15:91–102.

Received October 8, 2009. Accepted June 10, 2010.

eprint requestsAddress requests for reprints to: Bruce A. Vallance, PhD, ACB,

oom K4-188, 4480 Oak Street, BC’s Children’s Hospital,ancouver, British Columbia, Canada V6H 3V4. e-mail:vallance@cw.bc.ca; fax: (604) 875-3244.

cknowledgmentsM.M. and M.J.R. made similar contributions to this study.TLR4�/� and TLR2,4DKO mice were provided by Dr Alan Aderem.

he STAT3 antibodies were a gift from Drs Hui Xiao and Xiaoxia Li.

onflicts of interestThe authors disclose no conflicts.

undingSupported by grants from the Canadian Institutes of Health

esearch and the Crohn’s and Colitis Foundation of Canada (to.A.V.). K.M.M. is a Michael Smith Foundation for Health Researchenior Scholar. B.A.V. is a Canada Research Chair (Tier 2) and theHILD Foundation Research Scholar. D.L.G. is supported by theanadian Association of Gastroenterology/AstraZeneca/Canadian

nstitutes of Health Research and the Michael Smith Foundation for

ealth Research.