Immunity Requires T Cells and Activation of Innate Humoral

Humoral Immune Response to Flagellin Requires T Cells and Activation of Innate Immunity

This information is current as Catherine J. Sanders, Yimin Yu, Daniel A. Moore III, Ifor R. of September 26, 2021. Williams and Andrew T. Gewirtz J Immunol 2006; 177:2810-2818; ; doi: 10.4049/jimmunol.177.5.2810 http://www.jimmunol.org/content/177/5/2810 Downloaded from

References This article cites 36 articles, 15 of which you can access for free at: http://www.jimmunol.org/content/177/5/2810.full#ref-list-1 http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

by guest on September 26, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2006 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Humoral Immune Response to Flagellin Requires T Cells and Activation of Innate Immunity1

Catherine J. Sanders, Yimin Yu, Daniel A. Moore III, Ifor R. Williams, and Andrew T. Gewirtz2

Bacterial flagellin, the primary structural component of flagella, is a dominant target of humoral immunity upon infection by enteric pathogens and in Crohn’s disease. To better understand how such responses may be regulated, we sought to define, in mice, basic mechanisms that regulate generation of flagellin-specific Igs. We observed that, in response to i.p. injection with flagellin, generation of flagellin-specific Ig required activation of innate immunity in that these responses were ablated in MyD88-deficient mice and that flagellin from Helicobacter pylori, which is known not to activate TLR5, also did not elicit Abs. Mice lacking ␣␤ T cells (TCR␤null) were completely deficient in their ability to make flagellin Abs in various contexts indicating that, in contrast to Downloaded from common belief, generation of flagellin-specific Ig is absolutely T cell dependent. In contrast to Ab responses to whole flagella (H serotyping), responses to flagellin monomers displayed only moderate serospecificity. Whereas neither oral nor rectal administration of flagellin elicited a strong serum Ab response, induction of colitis with dextran sodium sulfate resulted in a MyD88- dependent serum Ab response to endogenous flagellin, suggesting that, in an inflammatory milieu, TLR signaling promotes acquisition of Abs to intestinal flagellin. Thus, acquisition of a humoral immune response to flagellin requires activation of innate

immunity, is T cell dependent, and can originate from ﬂagellin in the intestinal tract in inﬂammatory conditions in the http://www.jimmunol.org/ intestine. The Journal of Immunology, 2006, 177: 2810–2818.

acterial flagella have long been recognized to be one of variety of commensal and pathogenic microbes including E. coli, the major Ags of a variety of flagellated bacteria includ- Salmonella, and Clostridium spp. Purified flagellin has been shown B ing several Salmonella species, pathogenic and commen- to function as a T cell adjuvant (7, 8). Such adjuvanticity can be sal Escherichia coli strains, Pseudomonas, etc. Accordingly, sera envisioned to underlie the elicitation of humoral immune re- from hosts infected/inoculated with such pathogens display robust sponses by flagellin, although this view is at odds with the com- and relatively specific recognition of the flagella of that bacterium, monly held belief that flagella are T independent Ags. thus serving as the basis for H-Ag serotyping used to classify iso- Flagellin-specific Ig generated during Salmonella infection pro- by guest on September 26, 2021 lates of E. coli and Salmonella typhimurium. Flagella isolated from vide protection against reinfection (9); thus, flagellin, which is also such bacteria and monomeric flagellin are efficient elicitors of such a major target of T cells in primary infection (10), has been used humoral immune responses and thus the immunogenicity of flagel- in a variety of vaccine strategies (11, 12). The role of flagellin- lin has been the basis of a variety of vaccine strategies, for example specific Ig in Crohn’s disease is less clear. Flagellin-specific Ig can inserting epitopes from influenza or HIV into the flagellin mole- be envisaged to possibly protect Crohn’s disease patients from cule (1, 2). More recently, it has become appreciated that many infection by flagellated bacteria as such patients are thought to be flagellated bacteria also readily release flagellin monomers into at increased risk of intestinal infection. Conversely, because their milieu (3). Such flagellin monomers are recognized by TLR5, Crohn’s disease is thought by many to be driven by aberrant mu- thus resulting in activating innate immunity, in particular trigger- cosal immune responses against the normally commensal micro- ing a rapid induction of proinflammatory gene expression (3, 4). flora, adaptive immune responses to flagellin may drive the chronic Moreover, flagellin monomers have also been shown to be a target intestinal inflammation that characterizes this disorder. In the latter of the elevated adaptive immune response associated with Crohn’s scenario, Abs to flagellin could be envisioned to play a role in disease, a chronic intestinal inflammatory disorder driven by seem- driving the inflammation associated with Crohn’s disease or, al- ingly inappropriate immune responses to commensal intestinal mi- ternatively, the elevated levels of Abs in Crohn’s disease may sim- croflora (5, 6). Specifically, Crohn’s disease patients exhibit ele- ply reflect increased activation of flagellin-specific T cells that are vated serum IgG and IgA responses to flagellin monomers from a sufficient to drive colitis in susceptible (i.e., immunodeficient) mice (5). In light of the roles of flagellin-specific Ig in mediating outcomes of host-pathogen interactions, we sought to define some Department of Pathology, Emory University, Atlanta, GA 30322 of the basic mechanisms that regulate generation of this aspect of Received for publication January 19, 2006. Accepted for publication June 7, 2006. humoral immunity. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Materials and Methods Mice 1 This work was supported by the National Institutes of Health via Center Grants to Emory University (DK064399) and University of Alabama (DK64400) and by R-01 Various strains of wild-type (WT)3 mice (BALB/c, C57BL/6, and C3H/ Grant DK061417 (to A.T.G.) and via a grant from the Broad Medical Research Pro- OuJ) as well as mice carrying a null point mutation in TLR4 (C3H/HeJ), gram (to A.T.G.). 2 Address correspondence and reprint requests to Dr. Andrew T. Gewirtz, Pathology- Whitehead Research Building 105H, 615 Michael Street, Emory University, Atlanta, 3 Abbreviations used in this paper: WT, wild type; DSS, dextran sodium sulfate; DSP, GA 30322. E-mail address: [email protected] dithiobis (succinimidyl) propionate.

FIGURE 1. Systemic administration of purified flagellin elicits an innate and adaptive immune response independent of LPS. A–C, BALB/c mice were given i.p. 100 ␮l of HBSS containing 0–100 ␮g of purified S. typhimurium flagellin (FliC). Mice were bled at 90 min and 14 days later for measure of IL-6 and flagellin-specific IgG and IgA, respectively. These results are the mean of an ex- periment performed in duplicate and representative of several experiments. Inset in A is a Western blot (for flagellin) of serum collected Downloaded from from a mouse at various times after i.p. injection of flagellin (blot is representative of three experiments). D–F, Groups of WT (C3H/OuJ) or TLR4-deficient (C3H/HeJ) mice (n ϭ 3) were given i.p. HBSS (Ϫ)or ␮

10 g of flagellin. Mice were bled at http://www.jimmunol.org/ 90 min and 14 days for measurement of IL-6 and flagellin-specific IgG and IgA, respectively. Results are mean Ϯ SEM. by guest on September 26, 2021

and mice with targeted deletions in TCR␤ (Tcrbtm1Mom), TCR␦ were pretreated with streptomycin (10 mg/mouse) following procedures (Tcrdtm1Mom), or both (TCR␤␦, Tcr␤dtm1Mom) were purchased from The recently described by Hardt and colleagues (15), so as to allow a more Jackson Laboratory. MyD88null mice were originally generated by Shizuo efficient colonization and yet permit the mice to survive long enough to Akira, were extensively backcrossed onto a C57BL/6 background and generate Abs to a nonattenuated Salmonella strain. Blood was obtained via maintained on standard breeder diets. All experiments were performed on retro-orbital bleeding, and serum was prepared by allowing 30 min to clot female mice 6–12 wk of age. at room temperature followed by 10 min of centrifugation at 5000 ϫ g. Reagents Native flagellin was chromatographically purified from S. typhimurium or Immunoassays E. coli F-18, and purity was verified as previously described (3, 7). S. typhimurium FliC and FljB were purified from previously described iso- Serum IL-6 was measured via a kit assay from R&D systems. To measure genic mutants of SL3201 lacking genes for FliC or FljB, respectively. product-specific (and total) Abs, microtiter plates (from Valeant Pharma- Recombinant his-tagged flagellins FliC and FlaA were prepared as previ- ceuticals) were coated with flagellin (100 ng/well), OVA (2 ␮g/well), or E. ␮ ously described (13). Whole flagella, consisting of FliC, were prepared coli (LPS 10 g/well) applied overnight in 0.1 M NaHCO3 (pH 9.6). After from Sl3201 FljBϪ and subsequently cross-linked with dithiobis (succin- overnight coating, mouse sera was diluted from 1/100 to 1/100,000 in imidyl)propionate (DSP) as previously described (14). Recombinant clos- ELISA wash buffer (HBSS with 0.5% goat serum and 0.1% Tween 20) and tridia-like flagellin, F2, a gift from Charles Elson and Yingzhi Kong (Uni- applied to coated plates. After1hofincubation, product-specific Ig was versity of Alabama, Birmingham, AL) was prepared as previously detected using anti-mouse IgG (or IgM)-HRP (1/1000; Amersham) or a described (5). Carrier-free murine rTNF-␣ was purchased from R&D Sys- two-step method of anti-mouse IgA-biotin (1/5000; Southern Biotechnol- tems. S. typhimurium LPS (ultrapure), OVA (grade VI), and all other ogy) followed by avidin-peroxidase (1/10,000, Jackson ImmunoResearch chemicals were purchased from Sigma-Aldrich except for dextran sodium Laboratories). Peroxidase was then revealed via tetramethylbenzidine substrate sulfate (DSS), which was from Valeant Pharmaceuticals. (Kierkegaard and Perry Laboratories) and, following treatment with H2SO4, OD was read at 450 nm. The Ab titer refers to the reciprocal of the dilution of Inoculations serum equivalent at OD450 to three times background (coated wells and All experiments used three to six mice per condition. In cases where five detection Ab). or more mice were used per condition, up to one mouse per condition was considered an outlier, and subsequently its result excluded, if it exhibited a response of Ͻ1% of all other mice in that condition. Intraperitoneal DSS colitis injections were done with total volumes of 100 ␮l. Rectal administrations were performed with flexible No. 4 French catheters inserted 3–4 cm into DSS colitis was induced as we and others have previously described. the distal colon. Oral gavage was performed in 200-␮l volumes using a Briefly, mice were placed on water containing 2.5 or 3.0% DSS for 7 days 22-gauge round-tipped needle inserted ϳ2 cm into the throat (just beyond before being returned to regular water (16, 17). Induction of colitis was the point of gag reflex). For oral gavaging of live S. typhimurium, mice verified by observing fecal bleeding. 2812 Ab RESPONSE TO FLAGELLIN IS T CELL AND MyD88 DEPENDENT

of flagellin, we observed identical results in C3H/OuJ and C3H/ HeJ mice (Fig. 1, D–E), which carry a point mutation in their TLR4 gene that renders them unresponsive to LPS (19), indicating that our purified flagellin does not contain sufficient levels of LPS (endotoxin) to modulate these responses. The serum Ab response to flagellin was undetectable at 2 days postinjection (OD at 1/100 dilution was not above background, i.e., secondary Ab only) and robust at 14 days postinjection, indicating that it is a primary rather than a memory response (Fig. 2). The level of response from this single injection became maximal ϳ1 mo after flagellin adminis- FIGURE 2. Kinetics of murine Ab response to flagellin. C57BL/6 mice tration and persisted at this level for at least another month, de- were injected i.p. with 50 ␮g of flagellin. Flagellin-specific IgG was mea- clining thereafter but still well above untreated mice 8 mo later sured at indicated times in four individual mice. (latest time measured). A second administration of flagellin 2 wk after initial inoculation resulted in a substantially increased level of flagellin-specific IgG within 3 days of treatment. For example, for Results BALB/c mice given 10 ␮g of flagellin on days 1 and 15 had levels Purified flagellin elicits a MyD88-dependent humoral immune of anti-flagellin IgG OD values: day 15, 2.0 Ϯ 0.3, 1.2 Ϯ 0.4, and response 0.52 Ϯϩ/Ϫ0.3; and day 18, 2.5 Ϯ 0.3, 1.7 Ϯ 0.3, and 1.2 Ϯ 0.2 Downloaded from Flagellin is a major target of adaptive immunity in response to (for serum dilutions of 1/100, 1/1,000, and 1/10,000, respectively). infection with flagellated pathogens and in Crohn’s disease. In Thus, our inoculated mice were indeed capable of making a further light of the dependence of adaptive immunity on innate immunity, Ab response to this purified protein. it seemed logical that the ability of flagellin to elicit adaptive im- Two approaches were utilized to discern whether the Ab re- mune responses was owed to its ability to activate innate immu- sponse to flagellin required the innate response. First, we measured responses to flagellin (FlaA) from H. pylori that we and others

nity. We began to examine this notion by treating mice i.p. with http://www.jimmunol.org/ purified flagellin (without exogenous adjuvant) and measuring se- have shown lacks ability to activate TLR5 (Refs. 13, 20, and 21 rum IL-6 (at 90 min) as a general indicator of the innate response and Fig. 3). In contrast, to recombinant flagellin from S. typhi- and flagellin-specific Abs at 2 wk (and various other times). murium (FliC), administration of purified FlaA did not induce a Flagellin administered in this manner was rapidly maximally (Ͻ15 detectable increase in serum IL-6 or specific Abs. However, when min) detected in the serum (observed by Western blotting; Fig. 1A, FliC and FlaA were administered with CFA, FliC and FlaA were inset), indicating that this is a systemic rather than a local treat- equipotent at eliciting specific Ig, suggesting these molecules have ment. In contrast to healthy humans, the vast majority of whom similar levels of antigenicity but that FliC is more immunogenic display clearly detectable levels of flagellin-specific Abs at dilu- than FlaA due to its ability to activate innate immunity. Next we tions of 1/1000 or greater (6, 18), untreated mice displayed very measured responses to S. typhimurium flagellin in mice deficient in by guest on September 26, 2021 low levels of Abs to E. coli or Salmonella (usually not above the TLR adaptor protein MyD88, which is thought to transduce all background signal at 1/100 dilution), thus making them quite ame- innate immune signaling in response to flagellin. MyD88 mice did nable to this experimental approach. When injected i.p., at doses of not exhibit flagellin-induced serum IL-6 and were severely im- 1–100 ␮g/mouse, flagellin elicited a dose-dependent increase in paired in their ability to make Ab responses to flagellin (Fig. 4). To levels of serum IL-6 and flagellin-specific IgG and IgA (Fig. 1, test whether MyD88null mice could make Abs normally under con- A–C). Importantly, using a submaximal (i.e., nonsaturating) dose ditions expected not to require TLR signaling, we measured the Ab

FIGURE 3. H. pylori ﬂagellin is antigenic but not immunogenic. C3H/HeJ mice (n ϭ 3 per condition) were injected i.p. with 100 ␮l of HBSS containing the indicated dose (or 5 ␮g) of recombinant S. typhimurium or H. pylori ﬂagellin (FliC or FlaA, respectively). Where indicated, 50 ␮l of CFA were added. A and B, Serum IL-6 was measured at 90 min. C, Left, Abs to FliC generated in mice inoculated with FliC, FlaA; or HBSS; right, Abs to FlaA generated in mice inoculated with FliC, FlaA, or HBSS. D, SDS-PAGE immunoblot in which serum from mice receiving FliC or FlaA was used to probe blots of FliC and FlaA to verify that the Abs measured by ELISA were largely directed at the recombinant protein rather than a potential contaminant. E, Generation of FliC and FlaA Abs in mice inoculated in presence of CFA. E is labeled analogously to C. MW, Molecular weight. The Journal of Immunology 2813

FIGURE 4. MyD88 mice do not make innate or adaptive immune response to flagellin but make T-dependent Abs when a cytokine adjuvant is used. A and B, WT (C57BL/6) or MyD88null mice (n ϭ 3) were injected i.p. with HBSS or 50 ␮gofS. typhimurium flagellin (FliC). Serum IL-6 and flagellin-specific IgG were measured at 90 min and 18 days, respectively. C, WT (C57BL/6) or MyD88null mice (n ϭ 8) were injected i.p. with 50 ␮gof OVA Ϯ 1.5 ␮g of recombinant carrier-free TNF-␣. OVA-specific IgG was measured 14 days later.

responses of MyD88 mice and control mice to OVA administered flagellin-specific Igs requires ␣␤ T cells. This conclusion is in Downloaded from with TNF-␣ to serve as a cytokine adjuvant. Such Ab responses accordance with recent observations by Honko et al. (25). were similar between WT and MyD88 mice, suggesting that MyD88 mice are not deficient in ability to make Abs per se but Humoral immune responses to flagellin monomers lack high rather that Ab responses to flagellin, and likely many Ags and/or serospecificity Ag/adjuvant mixtures, require activation of innate immunity Immune responses to flagella are considered highly specific in that

(which is often mediated by MyD88). antisera generated to subspecies of flagellated microbes can be http://www.jimmunol.org/ used to discriminate between such bacteria (the basis of H sero- Generation of flagellin Abs is T cell dependent typing). Conversely, that some flagellin Abs have been observed to Flagella have long been considered thymus-independent Ags display substantial cross-reactivity to various flagellins by Western based largely on pioneering studies by Nossal et al. (22) demon- blotting of flagellin monomers suggests that such specificity may strating the ability of polymerized flagella to directly activate B be largely lost on flagellin monomers which would be consistent lymphocytes in vitro. Yet, studies measuring ability of polymer- with recent studies on the structure of flagellin monomers and their ized flagella to elicit flagella-specific Abs showed a substantial assembly into flagella that showed the majority of the monomer (ϳ75%) diminution in athymic (nude) or thymectomized mice surface is buried in the polymer and only the small portion of the (23), suggesting that although T cell help may not be absolutely molecule that is unique is exposed on the surface of the polymer- by guest on September 26, 2021 necessary for this response, it can play a regulatory role. However, ized flagella (26). To quantitatively test the notion that responses one might expect that such flagellin monomers would lack ability to flagellin monomers might not be highly serospecific, mice were to cross-link BCRs and thus not elicit Abs in the absence of T cells. injected with one of the two flagellin isotypes from S. typhimurium However, given the recent understanding that direct activation of (FliC or FljB) or E. coli FliC (H5 serotype) and collected serum 2 B cells may occur through either cross-linking of BCR or activa- wk later. Such sera was used to probes microtiter plates to measure tion of TLRs and that B cells express mRNA for the flagellin their levels of IgG that could recognize each of these flagellins receptor TLR5 (24), one could imagine that flagellin monomers (Fig. 6). We observed that serum from mice receiving each flagel- might also activate B cells directly. Thus, we decided to examine lin type exhibited a fairly similar ability to recognize the flagellin the role of T cells in generating flagellin-specific Igs. We observed they had received as well as the two types they had not (Fig. 6A). that mice with an engineered deletion of TCR␤, and thus lacking Similarly, serum from mice orally infected with S. typhimurium or all ␣␤ T cells, displayed a complete loss of ability to make flagel- inoculated with flagella displayed similar ability to recognize lin-specific IgG (Fig. 5). Further, in contrast to results previously flagellin monomers from either S. typhimurium or E. coli (Fig. 6, attained with athymic mice, TCR␤null mice also failed to make Ab B and C). If, however, mice were given S. typhimurium flagella responses to flagella or flagella cross-linked with DSP to prevent that were cross-linked with DSP to prevent depolymerization, the its depolymerization. A similar complete inability of TCR␤null resulting serum Abs recognized flagellin monomers from S. typhi- mice to make Ig to flagellin was also observed when assaying for murium but not E. coli (Fig. 6D). That, in the absence of DSP, sera

IgM (OD450 not above background at lowest serum dilution as- from infected and/or flagella-treated mice lacked clear ability to sayed of 1/100). A similar lack of Abs to flagellin was observed in discriminate between Salmonella and E. coli flagellin monomers mice lacking all T cells (i.e., TCR␤␦null), whereas levels of flagel- suggests that in these contexts Abs may be made more to disas- lin-specific IgG were similar in WT mice and mice lacking lacking sembled pieces of flagella (i.e., flagellin monomers) than to the only ␥␦T cells (TCR␦; data not shown). Mice lacking T cells dis- polymerized structure consistent with the recent observation that played early increases in serum IL-6 that were indistinguishable Salmonella organisms contain a specific receptor-mediated mech- from WT mice (Fig. 5C), suggesting that T cells are not involved anism to release flagellin monomers (27). in this aspect of innate recognition of this bacterial product. We Although the apparent antigenic similarity that we observed be- next measured the Ab responses in these TCR␤null mice in re- tween these E. coli and Salmonella flagellins is consistent with sponse to LPS, one of the best defined T-independent Ags. We their high degree of amino acid sequence (ϳ60% identical), some observed that levels of specific Igs elicited by LPS were indistin- flagellins that have been defined as Ags in Crohn’s disease do not guishable between WT and TCR␤null mice, demonstrating that, as share quite such a high degree of similarity. Specifically, use of expected, TCR␤null mice are capable of producing Abs to a T- serologic expression cloning to identify humoral immune targets independent Ag. Thus, in a variety of contexts, generation of of a spontaneously colitic mouse strain identified some flagellins 2814 Ab RESPONSE TO FLAGELLIN IS T CELL AND MyD88 DEPENDENT

FIGURE 5. Ab responses to flagellin are highly T cell dependent. A, Groups (n ϭ 4–5) of WT (C57BL/6; B6) or T cell-deficient (TCR␤Ϫ/Ϫ) mice were inoculated with HBSS or 10 ␮g of isolated flagella, flagella con- verted to monomer by heating, or flagella were cross- linked with DSP to prevent depolymerization to monomer. Serum IgG against purified flagellin monomer were measured 14 days later. Results are shown for individual mice assayed. B, Groups (n ϭ 3) of WT (C57BL/6) or T cell-deficient (TCR␤Ϫ/Ϫ) mice were inoculated with HBSS or 75 ␮g of commercially purchased, highly purified LPS. Levels of serum IgG Downloaded from against such LPS was measured 14 days later by ELISA. Data are the means Ϯ SEM observed at serum dilutions of 1/100 and 1/500. C, WT (C57BL/6) or T cell-deficient (TCR␣␤Ϫ/Ϫ) mice were inoculated with HBSS or the indicated dose of flagellin, and serum IL-6 was measured at 90 min by ELISA. http://www.jimmunol.org/ by guest on September 26, 2021 similar to clostridia flagellin and having ϳ30% amino acid se- gastrointestinal tract would elicit early cytokines or serum Ig. To quence identity to flagellin from E. coli or Salmonella (5). Thus, to address this, mice were given flagellin into the lower esophagus better gauge the degree of antigenic similarity of various flagellins, via a round-tipped needle or into the descending colon by a flexible we measured whether flagellin Abs made to E. coli flagellin would catheter. We observed that mice given even the highest doses of also recognize recombinant clostridia-like flagellin (provided by flagellin that could be delivered by oral gavage or rectal enema did Charles Elson and colleagues at the University of Alabama and not display elevated levels of serum IL-6 (at 90 min or 3 h) or lead referred to by Dr. Elson as F2 flagellin). We observed that serum to detectable levels of flagellin-specific serum Igs (Fig. 7). Repeat from mice inoculated with E. coli flagellin could indeed recognize administration of flagellin did not affect this lack of observed re- such clostridia-like flagellin but did so to a lesser degree than to sponse, nor did we observe detectable response when attempting flagellin from Salmonella consistent with their relative degrees of these experiments in C57BL/6 or C3H mice (in these instances, amino acid sequence homology. However, mice receiving the F2 OD at lowest dilution assayed did not differ from background, i.e., flagellin did not show substantial recognition of E. coli flagellin. secondary Ab only). This lack of response is in striking contrast to Thus, overall these results indicate that consistent with the knowl- the recent work of Strindelius et al. (28), who observed very high edge that some regions of the flagellin molecule are highly con- systemic responses from oral administrations of similar doses of served and some are relatively unique, use of flagellin monomers flagellin. Although we do not know the reason for this difference, to assess immune responses may afford a broad view of levels of our results suggest that generation of high levels of serum anti- exposure to flagellated microbes but may not always be a highly flagellin IgG resulting from exposure to flagellin via the gastroin- specific means of indicating whether a host has been exposed to a testinal tract may require systemic or at least extramucosal expo- specific microbial species. sure to this molecule. The lack of a serum cytokine response to mucosal administra- Induction of colitis, but not simple mucosal exposure to tion of flagellin, together with results that, in both polarized epi- flagellin, leads to robust MyD88-dependent Ab response to thelial cell lines and human colon, TLR5 is not expressed at func- flagellin tional levels on the apical surface (29, 30), suggest that the failure The existence of a potentially large reservoir of flagellin in the of intracolonic flagellin administration to lead to a serum Ab re- intestinal tract suggests that the easily detectable levels of serum sponse might result from a simple failure of such flagellin to cross flagellin-specific Ig displayed by healthy humans and elevated lev- the intestinal mucosa. Thus, we next administered flagellin muco- els displayed by Crohn’s disease patients may be generated in sally in the presence of 50% ethanol, which results in transient response to flagellin originating in the intestinal tract. Thus, we disruption of the epithelial barrier. Flagellin administered to the investigated whether administration of a bolus of flagellin to the colon in this manner indeed resulted in a detectable levels of serum The Journal of Immunology 2815 Downloaded from http://www.jimmunol.org/

FIGURE 6. Ab responses measured against flagellin monomers do not display high serospecificty among E. coli (E.c.) and Salmonella (S.t.) flagellins. A, C57BL/6 mice were injected with 10 ␮g of flagellin purified from S. typhimurium (strains expressing only FliC or FljB) or E. coli, and serum was isolated 14 days later. Serum from mice receiving each flagellin was used to probe ELISA plates coated with all three flagellins (data are means Ϯ SEM, n ϭ 4) at indicated serum dilutions. B, C57BL/6 mice were treated with streptomycin and then orally infected with 109 CFU of S. typhimurium (SL3201). Results are shown for three individual mice (ⅷ,AbtoSalmonella flagellin; ⅜,AbtoE. coli flagellin. C and D, Mice were inoculated with 10 ␮g per mouse isolated whole flagella (C) or flagella that were cross-linked with DSP (D). Serum IgG levels were measured 14 days later at the indicated dilutions. Results are means Ϯ SEM, n ϭ 4–5. E, C3H/HeJ mice were injected with either 10 ␮gofE. coli flagellin (FliC) or clostridia-like F2 flagellin. Serum IgG specific for both FliC and F2 was measured at indicated serum dilutions. Results are means Ϯ SEM, n ϭ 4. by guest on September 26, 2021

Abs to flagellin although the magnitude of the response was still 0.13 Ϯ 0.6 and 0.02 Ϯ 0.04; and 50 ␮g of flagellin intracoloni- well below that observed in response to flagellin delivered i.p. cally, 0.16 Ϯ 0.6 and 0.03 Ϯ 0.05), although DSS-treated mice still (Fig. 7C). We next sought to induce the intestinal mucosa into a responded robustly to E. coli flagellin administered i.p. (OD at state resembling that seen in inflammatory bowel disease. Specif- 1/100 and 1/1000 dilutions were 1.8 Ϯ 0.3 and 0.7 Ϯ 0.2, respec- ically, we gave mice drinking water containing DSS, which is tively), suggesting that mimicking the endogenous mucosal re- known to result in a robust colitis thought to be initially driven by sponse by simple administration of exogenous flagellin may not be epithelial barrier disruption. After 7 days of DSS treatment, at a straightforward matter. Nonetheless, together, these results sug- which time mice had bloody stools and other clinical indicators of gest that generating inflammatory conditions in the intestine, and colitis as we and others have described (17, 31), mice were re- perhaps the resulting loss of epithelial barrier function, might be turned to regular drinking water, allowing an apparent healing of sufficient to result in generation of an Ab response to endogenous their colitis. We observed that such treatment with DSS, by itself, bacterial flagellin and that this adaptive response requires TLR resulted in a robust Ab response to clostridia-like F2 flagellin and signaling. a mildly elevated Ab response to E. coli flagellin (Fig. 8). We next investigated whether such DSS-induced acquisition of flagellin- Discussion specific Ig shared the dependency on MyD88 that we observed for Bacterial flagella, and its monomeric subunit flagellin, has been the the Ab response to purified systemically administered flagellin. We focus of a number of immunologic studies because of its utility as reduced the concentration of DSS used to avoid the death of a model system, the important role of motility in bacterial patho- MyD88 mice that can occur in this model (32). Indeed, although genesis, and, consequently, the decision of the immune system to MyD88null mice contract DSS colitis with greater severity than target this molecule. Moreover, there has recently been a new level WT mice (32), they failed to acquire Abs to endogenous flagellin of interest in immune responses to flagellin due to the realization in this colitis model. We next measured whether mucosal admin- that flagellin is a dominant Ag in Crohn’s disease (5). Although the istration of E. coli flagellin following induction of colitis might immunogenicity of flagellin has long been recognized, until re- result in a greater elevation of Abs to this protein. We observed cently the reasons underlying its immunogenicity have largely that neither oral nor intracolonic administration of E. coli flagellin been unknown. One potential reason that has long been offered for given at the height of DSS colitis (after 7 days of DSS treatment) the targeting of flagellin by humoral immunity is that polymerized led to a significant elevation in Abs to this protein as compared flagellin (i.e., flagella) should be able to cross-link BCRs and thus with DSS alone (OD at 1/100 and 1/1000 dilutions were: HBSS, be able to directly elicit a humoral response without the need for 0.12 Ϯ 0.4 and 0.02 Ϯ 0.05; 50 ␮g of flagellin intragastrically, the comparatively complex process of T cell activation. However, 2816 Ab RESPONSE TO FLAGELLIN IS T CELL AND MyD88 DEPENDENT

FIGURE 7. Oral and rectal administration of flagellin failed to elicit detectable serum IL-6 or flagellin-specific Ig. A and B, BALB/c mice were given the indicated dose of flagellin i.p. (IP), intragastrically (IG), or intracolonically (IC) as described in Materials and Methods. Serum IL-6 was assayed at

the indicated time and flagellin-specific IgG assayed 15 days later. C, Flagellin was mixed 1:1 with ethanol and administered via the indicated route (100 Downloaded from ␮l containing 50 ␮g of flagellin) and flagellin-specific IgG measured 14 days later. Results are means Ϯ SEM, n ϭ 4.

this explanation would not be able to explain the fact that ﬂagellin sponses to ﬂagellin are dependent on innate immunity, although is also the dominant T cell target upon infection by S. typhimurium we are not aware of another protein that has been shown to pro-

in WT or B cell-deficient mice (10). Moreover, our results herein mote adaptive responses to itself by activating innate immunity. http://www.jimmunol.org/ indicate that in a physiologic infection, the humoral response is Although a recent study has shown that generation of humoral likely being generated to flagellin monomers rather than polymer- responses to T-dependent Ags uses signaling in B cells (24), we ized flagella and that flagellin monomers would be unable to cross- observed herein that T-dependent responses need not always re- link BCR. Recent work by Subramanian et al. (27) indicates that quire TLR signaling in B cells, particularly if the response is bacteria have a regulated host-induced means of secreting flagellin driven by a cytokine such as TNF-␣. Given that murine dendritic monomers; thus, we envisage such monomers, as well as depoly- cells have been observed not to express TLR5 and not directly merized flagella, to be the target of this humoral immune response. respond to flagellin (34), it has been suggested that the dendritic Most importantly, we show that generation of Abs to flagellin cell activation that we and others observed occurs in response to monomers, flagella, or cross-linked flagella is undetectable in mice flagellin in vivo (7) is driven by flagellin-induced cytokines. To- by guest on September 26, 2021 lacking T cells, suggesting that direct activation of B cells by fla- gether, this indicates that the failure of MyD88 mice to make Abs gella/flagellin does not occur to a significant enough extent to re- in response to flagellin is not due entirely to a need for MyD88 sult in a detectable humoral immune response to this bacterial signaling in B cells but at least in part due to a need for MyD88 for product in the absence of T cell help. Although a sense of a po- the cytokine production that may be necessary to activate dendritic tential role for T cells in generating Abs to flagellin can be retro- cells in response to flagellin. Importantly, our results suggest that spectively gleaned from the work that Huchet and Feldman (23) such a role for TLR signaling is true not only in response to pu- performed 30 years ago, it is not entirely clear why they observed rified flagellin but in the acquisition of anti-flagellin Abs associ- a modulatory role for T cells in this process whereas we observed ated with murine colitis. Although one might imagine that such an absolute requirement. Some possible reasons for this difference acquisition of Abs to endogenous flagellin might take place in a are: 1) they used athymic (nude or thymectomized) mice that, in milieu of other microbial products/adjuvants in which the activa- contrast to TCR␤ mice, may still have some extrathymically de- tion of innate immunity by flagellin might be redundant, we have veloped ␣␤ T cells, or perhaps even a so-called second thymus that recently observed that healthy humans carrying a dominant-nega- can allow for some T cell development (33); 2) they used flagella tive TLR5 allele display reduced levels of flagellin-specific Ig (18). that had been dinitrophenylated which can be envisaged to Thus, we speculate that the propensity of many flagellated mi- strengthen its potential ability to directly activate B cells; 3) per- crobes to readily release flagellin monomers much more readily haps their dinitrophenylated flagella contained a product other than than other bacterial-derived adjuvants (e.g., endotoxin, unmethyl- flagellin that had the direct ability to activate B cells, although we ated CpG DNA) may result in the immune system often seeing have observed that flagellin mixed with LPS still failed to elicit flagellin in the absence of a plethora of other bacterial products in Abs to flagellin in TCR␤null mice (data not shown), arguing which context the ability of flagellin to activate innate immunity is against this reasoning. Regardless of whether one or more of these very important. Although recent work has identified a TLR5-in- potential explanations played a role, generation of Abs to flagellin/ dependent and MyD88-independent pathway for recognition of flagella is greatly influenced by, if not absolutely requires, ␣␤ T flagellin (35), our results suggest that this pathway may not pro- cells. vide a sufficient signal 2 to generate an Ab response. Given that, at least in our experimental system, generation of Extrapolating our results to attempt to increase understanding of flagellin-specific Ig is an absolutely T cell-dependent response and how humans naturally acquire Abs to flagellin over their lifetime that flagellin functions as a T cell and MyD88-dependent adjuvant and how such levels might become elevated in Crohn’s disease (7, 8), it seemed likely that this central TLR signaling molecule suggests that the substantial level of serum Abs to E. coli and might be essential for generation of flagellin-specific Ig, and in- Salmonella flagellin that we and others have observed are present deed we observed this was the case. Thus, consistent with one of in healthy humans likely reflect levels of activation of flagellin- the emerging central themes in immunology, adaptive immune re- specific T cells, perhaps resulting from a variety of clinical and The Journal of Immunology 2817

References 1. Newton, S. M., T. M. Joys, S. A. Anderson, R. C. Kennedy, M. E. Hovi, and B. A. Stocker. 1995. Expression and immunogenicity of an 18-residue epitope of HIV1 gp41 inserted in the flagellar protein of a Salmonella live vaccine. Res. Microbiol. 146: 203–216. 2. Jeon, S. H., T. Ben-Yedidia, and R. Arnon. 2002. Intranasal immunization with synthetic recombinant vaccine containing multiple epitopes of influenza virus. Vaccine 20: 2772–2780. 3. Gewirtz, A. T., P. O. Simon, Jr., C. K. Schmitt, L. J. Taylor, C. H. Hagedorn, A. D. O’Brien, A. S. Neish, and J. L. Madara. 2001. Salmonella typhimurium translocates flagellin across intestinal epithelia, inducing a proinflammatory response. J. Clin. Invest. 107: 99–109. 4. Hayashi, F., K. D. Smith, A. Ozinsky, T. R. Hawn, E. C. Yi, D. R. Goodlett, J. K. Eng, S. Akira, D. M. Underhill, and A. Aderem. 2001. The innate immune response to bacterial flagellin is mediated by Toll- like receptor 5. Nature 410: 1099–1103. 5. Lodes, M. J., Y. Cong, C. O. Elson, R. Mohamath, C. J. Landers, S. R. Targan, M. Fort, and R. M. Hershberg. 2004. Bacterial flagellin is a dominant antigen in FIGURE 8. Induction of colitis with DSS results in a MyD88-dependent Crohn disease. J. Clin. Invest. 113: 1296–1306. acquisition of Ab response to endogenous flagellin. A, C57BL/6 mice were 6. Sitaraman, S. V., J. M. Klapproth, D. A. Moore, III, C. Landers, S. Targan, bled (day 0) and then given drinking water containing 3% DSS for 7 days I. R. Williams, and A. T. Gewirtz. 2005. Elevated flagellin-specific immuno- globulins in Crohn’s disease. Am. J. Physiol. Gastrointest. Liver Physiol. 288: and serum assayed for IgG to E. coli or clostridia-like flagellin (F2) 14 days G403–G406. after their removal from DSS. B, C57BL/6 or MyD88 mice were given 7. McSorley, S. J., B. D. Ehst, Y. Yu, and A. T. Gewirtz. 2002. Bacterial flagellin ϩ 2.5% DSS for 7 days, returned to regular water, and assayed for serum IgG is an effective adjuvant for CD4 T cells in vivo. J. Immunol. 169: 3914–3919. Downloaded from to clostridia-like flagellin (F2) 15 days later. 8. Didierlaurent, A., I. Ferrero, L. A. Otten, B. Dubois, M. Reinhardt, H. Carlsen, R. Blomhoff, S. Akira, J. P. Kraehenbuhl, and J. C. Sirard. 2004. Flagellin promotes myeloid differentiation factor 88-dependent development of Th2-type response. J. Immunol. 172: 6922–6930. 9. McSorley, S. J., and M. K. Jenkins. 2000. Antibody is required for protection against virulent but not attenuated Salmonella enterica serovar typhimurium. In- subclinical infectious and/or transient ulcerations. Whereas the fect. Immun. 68: 3344–3348. 10. McSorley, S. J., B. T. Cookson, and M. K. Jenkins. 2000. Characterization of first-described T cell epitope of Salmonella flagellin is somewhat http://www.jimmunol.org/ CD4ϩ T cell responses during natural infection with Salmonella typhimurium. specific for Salmonella (i.e., present in many Salmonella serovars J. Immunol. 164: 986–993. but not E. coli) (10), the T cell epitopes of Salmonella flagellin 11. Harada, H., F. Nishikawa, N. Higashi, and E. Kita. 2002. Development of a targeted during primary infection have recently been defined to be mucosal complex vaccine against oral Salmonella infection in mice. Microbiol. Immunol. 46: 891–905. in highly conserved regions of the protein common to E. coli and 12. Kirkpatrick, B. D., K. M. Tenney, C. J. Larsson, J. P. O’Neill, C. Ventrone, other common flagellins (36). This suggests that the not highly M. Bentley, A. Upton, Z. Hindle, C. Fidler, D. Kutzko, et al. 2005. The novel oral typhoid vaccine M01ZH09 is well tolerated and highly immunogenic in 2 vaccine serospecific responses observed herein is likely polyclonal rather presentations. J. Infect. Dis. 192: 360–366. than directed toward one specific conserved epitope of flagellin, 13. Gewirtz, A. T., Y. Yu, U. S. Krishna, D. A. Israel, S. L. Lyons, and R. M. Peek, with the highly conserved portion of the flagellin monomer that Jr. 2004. Helicobacter pylori flagellin evades Toll-like receptor 5-mediated innate immunity. J. Infect. Dis. 189: 1914–1920. by guest on September 26, 2021 mediates polymerization and is targeted by TLR5 (14) being one 14. Smith, K. D., E. Andersen-Nissen, F. Hayashi, K. Strobe, M. A. Bergman, structural feature that might likely be so targeted. However, inter- S. L. Barrett, B. T. Cookson, and A. Aderem. 2003. Toll-like receptor 5 recog- estingly, when assessing Ab cross-reactivity between these typical nizes a conserved site on flagellin required for protofilament formation and bacterial motility. Nat. Immunol. 4: 1247–1253. flagellins and a more distantly related clostridia-like flagellin that 15. Barthel, M., S. Hapfelmeier, L. Quintanilla-Martinez, M. Kremer, M. Rohde, was serologically cloned as a dominant Ag in murine colitis, we M. Hogardt, K. Pfeffer, H. Russmann, and W. D. Hardt. 2003. Pretreatment of mice with streptomycin provides a Salmonella enterica serovar Typhimurium observed somewhat of a one-way specificity in that Abs made in colitis model that allows analysis of both pathogen and host. Infect. Immun. 71: response to E. coli flagellin reacted moderately with a clostridia- 2839–2858. like flagellin but not vice versa whether the response was made to 16. Gewirtz, A. T., and J. L. Madara. 2000. Lipoxin stable analogs: potential thera- peutic down-regulators of intestinal inflammation. In Eicosanoids and Inflamma- purified protein or acquired from native flora upon induction of tion. D. Perez, ed. Springer-Verlag, Berlin. colitis. Although the mechanisms governing such cross-reactivity 17. Castaneda, F. E., B. Walia, M. Vijay-Kumar, N. R. Patel, S. Roser, remain to be determined, the lack of significant levels of Abs to E. V. L. Kolachala, M. Rojas, L. Wang, G. Oprea, P. Garg, et al. 2005. Targeted deletion of metalloproteinase 9 attenuates experimental colitis in mice: central coli flagellin in healthy laboratory mice and their highly preferen- role of epithelial-derived MMP. Gastroenterology 129: 1991–2008. tial induction of responses to clostridia-like flagellin upon induc- 18. Gewirtz, A. T., M. Vijay-Kumar, S. R. Brant, R. H. Duerr, D. L. Nicolae, and J. H. Cho. 2006. Dominant-negative TLR5 polymorphism reduces adaptive im- tion of DSS colitis suggest that flagellated E. coli may not be a mune response to flagellin and negatively associates with Crohn’s disease. major component of our mouse microflora. However, given that Am. J. Physiol. Gastrointest. Liver Physiol. 290: G1157—G1163. mice did not respond well to a bolus of intracolonically adminis- 19. Poltorak, A., X. He, I. Smirnova, M. Y. Liu, C. V. Huffel, X. Du, D. Birdwell, E. Alejos, M. Silva, C. Galanos, et al. 1998. Defective LPS signaling in C3H/HeJ tered E. coli flagellin under a variety of conditions but could re- and C57BL/10ScCr mice: mutations in Tlr4 gene. Science 282: 2085–2088. spond well to its endogenous flagellin upon induction of colitis, it 20. Lee, S. K., A. Stack, E. Katzowitsch, S. I. Aizawa, S. Suerbaum, and is suggested that this surface may be normally be quite strongly set C. Josenhans. 2003. Helicobacter pylori flagellins have very low intrinsic activity to stimulate human gastric epithelial cells via TLR5. Microbes Infect. 5: 1345– to be hyporesponsive to such products except when delivered un- 1356. der appropriate danger signals delivered with Ags in a coordinated 21. Andersen-Nissen, E., K. D. Smith, K. L. Strobe, S. L. Barrett, B. T. Cookson, S. M. Logan, and A. Aderem. 2005. Evasion of Toll-like receptor 5 by flagellated manner. Because mucosal administration of flagellin to other sur- bacteria. Proc. Natl. Acad. Sci. USA 102: 9247–9252. faces results in robust responses (2, 37), intestinal hypresponsive- 22. Nossal, G. J., G. L. Ada, and C. M. Austin. 1964. Antigens in immunity. ness to flagellin may be restricted to surfaces with large commen- II. Immunogenic properties of flagella, polymerized flagellin and flagellin in the primary response. Aust. J. Exp. Biol. Med. Sci. 42: 283–294. sal bacterial loads such as the gastrointestinal tract. Such intestinal 23. Huchet, R., and M. Feldmann. 1974. Thymus-independent IgM and IgG re- hyporesponse to flagellin may in part be mediated by both its sponses in mice to flagellar antigens. Int. Arch. Allergy Appl. Immunol. 46: highly selective barrier function and/or the intestinal population of 600–611. 24. Pasare, C., and R. Medzhitov. 2005. Control of B-cell responses by Toll-like regulatory T cells. receptors. Nature 438: 364–368. 25. Honko, A. N., N. Sriranganathan, C. J. Lees, and S. B. Mizel. 2006. Flagellin is an effective adjuvant for immunization against lethal respiratory challenge with Disclosures Yersinia pestis. Infect. Immun. 74: 1113–1120. The authors have no financial conflict of interest. 26. Yonekura, K., S. Maki-Yonekura, and K. Namba. 2003. Complete atomic model 2818 Ab RESPONSE TO FLAGELLIN IS T CELL AND MyD88 DEPENDENT

of the bacterial flagellar filament by electron cryomicroscopy. Nature 424: 643– 32. Rakoff-Nahoum, S., J. Paglino, F. Eslami-Varzaneh, S. Edberg, and 650. R. Medzhitov. 2004. Recognition of commensal microflora by Toll-like receptors 27. Subramanian, N., and A. Qadri. 2006. Lysophospholipid sensing triggers secre- is required for intestinal homeostasis. Cell 118: 229–241. tion of flagellin from pathogenic Salmonella. Nat. Immunol. 7:583–589. 33. Terszowski, G., S. M. Muller, C. C. Bleul, C. Blum, R. Schirmbeck, J. Reimann, 28. Strindelius, L., M. Filler, and I. Sjoholm. 2004. Mucosal immunization with L. D. Pasquier, T. Amagai, T. Boehm, and H. R. Rodewald. 2006. Evidence for purified flagellin from Salmonella induces systemic and mucosal immune re- a functional second thymus in mice. Science 312: 284–287. sponses in C3H/HeJ mice. Vaccine 22: 3797–3808. 34. Means, T. K., F. Hayashi, K. D. Smith, A. Aderem, and A. D. Luster. 2003. The 29. Gewirtz, A. T., T. A. Navas, S. Lyons, P. J. Godowski, and J. L. Madara. 2001. Toll-like receptor 5 stimulus bacterial flagellin induces maturation and chemo- Cutting edge: bacterial flagellin activates basolaterally expressed tlr5 to induce kine production in human dendritic cells. J. Immunol. 170: 5165–5175. epithelial proinflammatory gene expression. J. Immunol. 167: 1882–1885. 35. Molofsky, A. B., B. G. Byrne, N. N. Whitfield, C. A. Madigan, E. T. Fuse, 30. Rhee, S. H., E. Im, M. Riegler, E. Kokkotou, M. O’Brien, and C. Pothoulakis. K. Tateda, and M. S. Swanson. 2006. Cytosolic recognition of flagellin by mouse 2005. Pathophysiological role of Toll-like receptor 5 engagement by bacterial macrophages restricts Legionella pneumophila infection. J. Exp. Med. 203: flagellin in colonic inflammation. Proc. Natl. Acad. Sci. USA 102: 1093–1104. 13610–13615. 36. Bergman, M. A., L. A. Cummings, R. C. Alaniz, L. Mayeda, I. Fellnerova, and 31. Gewirtz, A. T., L. S. Collier-Hyams, A. N. Young, T. Kucharzik, W. J. Guilford, B. T. Cookson. 2005. CD4ϩ-T-cell responses generated during murine Salmo- J. F. Parkinson, I. R. Williams, A. S. Neish, and J. L. Madara. 2002. Lipoxin a4 nella enterica serovar Typhimurium infection are directed towards multiple analogs attenuate induction of intestinal epithelial proinflammatory gene expres- epitopes within the natural antigen FliC. Infect. Immun. 73: 7226–7235. sion and reduce the severity of dextran sodium sulfate-induced colitis. J. Immu- 37. Ben-Yedidia, T., and R. Arnon. 1998. Effect of pre-existing carrier immunity on nol. 168: 5260–5267. the efficacy of synthetic influenza vaccine. Immunol. Lett. 64: 9–15. Downloaded from http://www.jimmunol.org/ by guest on September 26, 2021