Metalloproteinase-3)-Deficient Mice Infection in Stromelysin-1 (Matrix Impaired Immunity to Intestinal Bacterial

Home , Metalloproteinase, MMP3, MMP7, Stromelysin 2

The Journal of Immunology

Impaired Immunity to Intestinal Bacterial Infection in Stromelysin-1 (Matrix Metalloproteinase-3)-Deﬁcient Mice1

Chris K. F. Li,* Sylvia L. F. Pender,* Karen M. Pickard,* Victoria Chance,* Judith A. Holloway,* Alan Huett,‡ Nathalie S. Gonc¸alves,* John S. Mudgett,† Gordon Dougan,‡ Gad Frankel,‡ and Thomas T. MacDonald2*

Infection of mice with the intestinal bacterial pathogen Citrobacter rodentium results in colonic mucosal hyperplasia and a local Th1 inﬂammatory response similar to that seen in mouse models of inﬂammatory bowel disease. Matrix metalloproteinases (MMPs) have been shown to mediate matrix remodeling and cell migration during tissue injury and repair in the intestine. We have previously shown enhanced pathology in infected TNFRp55؊/؊, IL-12p40؊/؊, and IFN-␥؊/؊ mice, and here we show that this is associated with an increase in stromelysin-1 (MMP3) transcripts in colonic tissues. We have therefore investigated the role of MMP3 in colonic mucosal hyperplasia and the local Th1 responses using MMP3؊/؊ mice. In MMP3؊/؊ mice, similar mucosal thickening was observed after infection as in wild-type (WT) mice. Colonic tissues from MMP3؊/؊ mice showed a compensatory increase in the expression of other MMP transcripts, such as MMP7 and MMP12. However, MMP3؊/؊ mice showed delayed clearance of bacteria and delayed appearance of CD4؉ T lymphocytes into intestinal lamina propria. CSFE-labeled mesenteric lymph node CD4؉ T lymphocytes from infected WT mice migrated in fewer numbers into the mesenteric lymph nodes and colon ,of MMP3؊/؊ mice than into those of WT mice. These studies show that mucosal remodeling can occur in the absence of MMP3 ,but that MMP3 plays a role in the migration of CD4؉ T lymphocytes to the intestinal mucosa. The Journal of Immunology, 2004 173: 5171Ð5179.

nfection of mice with the intestinal bacterial pathogen (reviewed in Ref. 7). MMPs belong to the matrixin subfamily of Citrobacter rodentium results in colonic mucosal hyperplasia the metzincin superfamily of metalloproteinases. To date, Ͼ23 dif- and a local Th1 inflammatory response similar to that seen in ferent MMPs have been cloned, and additional members continue I 3 mouse models of inflammatory bowel disease (IBD). In suscep- to be identified. Structurally, all MMPs share a similar prodomain tible mouse strains, C. rodentium infection is associated with co- and catalytic domain, which act on a broad spectrum of the extra- lonic crypt hyperplasia, goblet cell depletion, and mucosal erosion cellular matrix components. Most MMPs are secreted as proen- (1, 2). Infection of mice with live C. rodentium or intracolonic zymes and require proteolytic cleavage for activation. Inhibition of inoculation of dead bacteria also induces a large infiltrate of CD4ϩ MMPs is conducted by endogenous MMP inhibitors, tissue inhib- cells into the colonic lamina propria and a highly polarized local itor of metalloproteinases (TIMP), which contain four members Th1 response with increased transcripts for the type I cytokines (TIMP1Ð4). MMPs have been implicated in tissue injury and ma- IL-12, TNF-␣, and IFN-␥ (3, 4). trix remodeling in various chronic inflammatory diseases such as To elucidate the role of proinflammatory cytokines in host de- IBD, asthma, and rheumatoid arthritis. In the intestine, MMPs are fense and mucosal pathology in vivo, mice deficient in TNF-␣, highly expressed at ulcer edges in patients with chronic IBD and IFN-␥, and IL-12 were orally infected with live C. rodentium (5, play a crucial role in the tissue injury that follows T cell activation 6). These animals showed various degrees of impairment in their in explants of human fetal small bowel (8Ð10). In celiac disease, ability to clear infection, but remarkably, mucosal pathology was where a mucosal Th1 response to gluten drives mucosal remodel- greater in these mice than in wild-type (WT) controls. ing and growth, MMP3 is overexpressed in fibroblasts immedi- Matrix metalloproteinases (MMPs) are a family of proteases ately under the epithelium (11). important in the turnover of extracellular matrix and cell migration Taken together, these results suggest that MMPs, especially MMP3, is involved in immune-mediated tissue injury in the intestine. MMPs probably play a role in degrading the matrix to pro- *Division of Infection, Inflammation and Repair, University of Southampton School duce mucosal ulceration, but they may also be important in con- of Medicine, Southampton, United Kingdom; †Merck Research Laboratories, Rah- way, NJ; and ‡Department of Biochemistry, Centre for Molecular Microbiology and trolling matrix turnover and in the mucosal thickening seen in Infection, Imperial College, South Kensington, United Kingdom experimental and clinical intestinal inflammation. Received for publication February 5, 2004. Accepted for publication July 21, 2004. In this work, therefore, we first examined MMP3 in the colon of The costs of publication of this article were defrayed in part by the payment of page WT and TNF-␣, IFN-␥, and IL-12 knockout (KO) mice infected charges. This article must therefore be hereby marked advertisement in accordance with C. rodentium and subsequently investigated mucosal hyper- with 18 U.S.C. Section 1734 solely to indicate this fact. plasia and the local immune responses in MMP3null mice. 1 This work was supported by the Biotechnology and Biological Sciences Research Council. 2 Address correspondence and reprint requests to Prof. Thomas MacDonald, Mail- Materials and Methods point 813, Division of Infection, Inflammation and Repair, Southampton General Mice Hospital, Southampton, SO16 6YD, U.K. E-mail address: [email protected] 3 Abbreviations used in this paper: IBD, inflammatory bowel disease; MMP, matrix Female 6- to 8-wk-old C57BL/6J and B10R111 mice were purchased from metalloproteinase; TIMP, tissue inhibitor of metalloproteinase; DC, dendritic cell; Harlan Olac (Bichester, U.K.) or Bantin and Kingman Universal (Hull, Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ MLN, mesenteric lymph node; MLNC, mesenteric lymph node cell; KO, knockout. U.K.). TNFRp55 / , IL-12p40 / , and IFN-␥ / mice (backcrossed to

C57BL/6 background at least 10 times) were originally purchased from determination of isotype-specific Ab titers, wells were washed with PBS- The Jackson Laboratory (Bar Harbor, ME) and were maintained by ho- Tween 20 before addition of 100 ␮l of an Ig-specific HRP conjugate mozygous matings under contract at Bantin and Kingman Universal. (DAKO, High Wycombe, U.K.) diluted 1/1000 in PBS-Tween 20 contain- MMP3Ϫ/Ϫ mice were obtained from John Mudgett, Merck Research Lab- ing 0.2% (w/v) BSA for2hat37¡C. Finally, after a washing with PBS- oratories (Rahway, NJ) and were maintained by homozygous matings. All Tween 20, bound Ab was detected by addition of o-phenylenediamine mice came from specific pathogen-free colonies. They were kept in specific substrate (Sigma-Aldrich) and the OD490 was measured. Titers were de- pathogen-free environment with free access to sterilized food and water. termined arbitrarily as the reciprocal of the serum dilution corresponding to All experiments used five to six mice per group and were repeated at least an OD of 0.3. twice. Protein extraction Bacterial strains and oral infection of mice Colonic samples (2 cm) were snap-frozen at Ϫ70¡C and homogenized in A nalidixic acid-resistant isolate of C. rodentium (formerly Citrobacter ice-cold extraction buffer (50 mM Tris-HCl (pH 7.4), 10 mM CaCl2, 0.05% freundii biotype 4280) was used in this study. DBS255 (pCVD438) is a C. Brij 35, 0.25% Triton X-100) at maximum speed for 30 s using an IKA rodentium eae (intimin ␣) strain, which expresses biologically active in- tissue homogenizer (Fisher Scientific, Loughborough, U.K.). The homog- timin and is virulent in mice (12). Bacterial inocula were prepared by enates were centrifuged at 13,000 rpm for 10 min at 4¡C, and the super- culturing bacteria overnight at 37¡C in 10 ml of Luria broth containing natants were removed and assayed for protein concentration (Bio-Rad, nalidixic acid (100 ␮g/ml) plus chloramphenicol (50 ␮g/ml). Cultures were Hemel Hempstead, Hertfordshire, U.K.). Homogenates were stored at harvested by centrifugation and resuspended in 0.1 volume of PBS. Mice Ϫ70¡C until required. were orally inoculated with 200 ␮l of the bacterial suspension with a ga- vage needle. The viable count of the inoculum was determined by plating Western blotting on Luria-Bertani agar containing appropriate antibiotics. In all experi- Samples were denatured in reducing treatment buffer and loaded (200 ␮g/ ϫ 9 ments, mice received 1Ð4 10 CFU. lane) onto 10% (v/v) Novex Bis-Tris gels (Invitrogen, Paisley, U.K.). After electrophoresis, the proteins were transferred onto nitrocellulose membrane Measurement of pathogen burden (Bio-Rad) using a Novex blotting apparatus (Invitrogen, Paisley, U.K.) and At selected times postinfection, mice were killed by cervical dislocation. detected using goat anti-mouse MMP3 (1/250) (R&D, Abingdon, U.K.), The terminal 4 cm of the colon were removed, and the colon was weighed HRP-conjugated rabbit anti-goat secondary Ab (DAKO, Cambridge, after removal of fecal pellets. In some experiments, 1-cm samples of distal U.K.), and an ECL Plus kit (Amersham Biosciences, Little Chalfont, U.K.) colon were snap-frozen in liquid nitrogen for subsequent immunohistolog- according to the manufacturer’s instructions. The Ab binding was imaged ical analysis, cytokine RT-PCR, and MMP Western blotting. Spleens and on ECL film (Amersham Biosciences). For control of protein loading, the colons were homogenized mechanically using a Seward 80 stomacher blot was stripped with Tris-HCl buffer (62.5 mM, pH 6.7) containing 10 (Seward Medica, London, U.K.). The number of viable bacteria in organ mM 2-ME for 30 min at 60¡C with gentle shaking. The membrane was then homogenates was determined by viable count on Luria-Bertani agar con- washed with PBS-Tween 20 for 10 min for 3 times at RT. After blocking taining 100 ␮g/ml naladixic acid and 50 ␮g/ml chloramphenicol. The limit with 5% skimmed milk in PBS-Tween 20, the membrane was re-probed ␤ of sensitivity was 10 CFU/organ. with 1/10,000 rabbit anti-mouse -actin (AbCam, Cambridge, U.K.) and 1/1000 HRP-conjugated goat anti-rabbit secondary Ab (DAKO, Cam- Immunohistochemistry and measurement of mucosal thickness bridge, U.K.). The blot was then developed as described as above. A three-step avidin-peroxidase staining was performed on 5-␮m frozen RNA extraction and quantitative RT-PCR sections as described previously (5) using mABs YTS191 (anti-CD4), anti- ␤7 and anti-mouse MAdCAM (both from BD BioSciences, Oxford, U.K.). Total cellular RNA was isolated from frozen colonic tissues by homoge- Biotin-conjugated rabbit anti-rat IgG (1/50) and avidin peroxidase (1/200; nization of the tissue in TRIzol (Invitrogen, Paisley, U.K.) followed by DAKO, Cambridge, U.K.) were diluted in Tris-buffered saline (pH 7.6) chloroform extraction and isopropanol precipitation. Total RNA was mea- containing 4% (v/v) normal mouse serum (Harlan Seralab, Oxon, U.K.). sured at 260 nm by spectrophotometric analysis (Beckman Coulter, High Peroxidase activity was detected with 3,3Ј-diaminobenzidine tetrahydro- Wycombe, U.K.). Cytokine encoding plasmid (pCMQ2), kindly provided chloride (Sigma-Aldrich, Poole, U.K.) in 0.5 mg/ml in Tris-HCl (pH 7.6), by M. F. Kagnoff (Department of Medicine, University of California, San Diego, CA), was used for quantitative competitive PCR for TNF-␣ tran- containing 0.01% H2O2 (Sigma-Aldrich). The density of positive cells in the lamina propria was determined by image analysis (Carl Zeiss, Welwyn scripts as described in Ref. 13. To quantify mouse MMP mRNA levels, we Garden City, Hertfordshire, U.K.). Mucosal thickness was measured by constructed a plasmid that encoded a standard RNA using an approach micrometry on H&E-stained sections, with 10 measurements being taken similar to that described in Ref. 13. Sequences of the oligonucleotide prim- in the distal colons of individual mice (five to six mice/group). Only well- ers used for PCR amplification, and the sizes of the predicted PCR products oriented pieces of tissues were analyzed. from the target and standard RNAs are given in Table I. All primers were synthesized and HPLC purified by Sigma Genosys (Sigma-Aldrich). Prim- Analysis of humoral immune responses ers were designed using Primer 3 (Whitehead Institute for Biomedical Research, Cambridge, MA) from cDNA sequences available in PubMed On day 21 postinfection, the animals were killed, and blood samples were (National Center for Biotechnology Information). The primers span exon- collected by cardiac puncture. The serum was collected and stored at intron boundaries and do not amplify genomic DNA. Sequences are Ϫ20¡C until it was analyzed. For analysis of Ag-specific Ab responses, specific for mouse MMP using blast search (National Center for Biotech- wells of microtiter plates (Maxisorb plates; Nunc, Naperville, IL) were nology Information), and PCR products were checked and sequenced to coated overnight at 4¡C with 100 ␮l of a bicarbonate solution (pH 9.6) verify that the fragment was MMP. The DNA insert containing different containing recombinant EspA (2.5 ␮g/ml). After a washing with PBS with MMP primer sites was cloned into pSP64 poly(A) vector (Promega, Tween 20 (0.05% v/v), wells were blocked by addition of 1.5% (w/v) BSA Southampton, U.K.) to generate poly(A)ϩ transcripts in vitro. To generate in PBS for 1 h. Plates were then washed twice with PBS-Tween 20 before RNA, plasmids were linearized with NotI for pCMQ-2 or EcoRI for mouse sera from individual mice were added and serially diluted in PBS-Tween MMP plasmid and transcribed in vitro with either T7 or SP6 RNA poly- 20 containing 0.2% (w/v) BSA and incubated for2hat37¡C. For the merase under conditions recommended by the supplier (Promega). Serial

Table I. Primers used for RT-PCR analysis of murine MMP mRNAs

Size of PCR Product (bp)

Target mRNA 5Ј Primer 3Ј Primer Target Standard

MMP3 caggtgtggtgttcctgatg cgtcaaagtgagcatctcca 326 505 MMP7 cggagatgctcactttgaca gaagagggagacaggtgcag 343 505 MMP12 tggaatatgaccccctgttc gcctctgaaccatgcacttt 303 505 TIMP1 actcggacctggtcataagggc aagaagctgcaggcattgat 268 505 The Journal of Immunology 5173

FIGURE 2. MMP3 protein expression in colonic tissue of infected wild- type and MMP3Ϫ/Ϫ mice. Total protein was extracted from colonic tissue FIGURE 1. MMP3 gene expression in colonic tissue of C57BL/6-, TN- of mice orally infected with C. rodentium 14 days previously. Expression FRp55Ϫ/Ϫ-, IL-12p40Ϫ/Ϫ-, and IFN-␥Ϫ/Ϫ-deficient mice. Total RNA was of MMP3 protein was increased in infected B10R111 mice. MMP3 protein Ϫ/Ϫ extracted from colonic tissues of mice orally infected with C. rodentium 14 was absent in infected and uninfected MMP3 mice. ␤-Actin was used days previously. Data depict the mean number of transcripts (ϮSEM) en- as loading control. coding MMP3 in infected and uninfected C57BL/6 and cytokine KO mice. There were significantly more transcripts detected in all groups of infected mice compared with uninfected controls (p Ͻ 0.01 for all 4 groups, Stu- pared as described in Ref. 14 and stained with allophycocyanin-conjugated dent’s t test). Moreover, MMP3 transcripts were significantly more abun- anti-CD4 mAb for MLNC or PE-conjugated anti-CD3 mAb (Caltag Lab- dant in infected IL-12p40Ϫ/Ϫ and IFN-␥Ϫ/Ϫ mice than in infected TN- Ϫ Ϫ oratories). The lymphocytes were then analyzed by FACSCalibur (BD Bio- FRp55 / or WT mice (p Ͻ 0.001). sciences) as described above. The absolute numbers of CFSEϩCD3ϩ or CFSEϩCD4ϩ T cells in the CD3ϩ or CD4ϩ T cells were calculated from the total number of collected cells and the proportion of CFSEϩCD3ϩ or ϩ ϩ 10-fold dilutions of standard RNA (106Ð103 molecules) were co-reverse CFSE CD4 T cells among them. transcribed with total cellular RNA (1 ␮g) at 37¡C for 50 min in a 20-␮l reaction volume containing 50 mM Tris (pH 8.3), 75 mM KCl, 3 mM Statistical analysis ␮ MgCl2, 3 mM DTT, 10 mM dNTP mix, and 0.5 g of oligodeoxythymi- The significance of differences between means was determined using the dylate (Amersham Biosciences), using 100 U of Moloney murine leukemia paired t test and the Mann-Whitney U test. p Ͻ 0.05 was considered virus reverse transcriptase (Invitrogen). PCR amplification was routinely significant. conducted in 50-␮l reaction volume (10 mM Tris (pH 9), 50 mM KCl, 1.5 ␮ Ј Ј mM MgCl2, 200 M each dNTP, and 20 pmol of specific5- and 3 - primers), using1UofTaq polymerase (Amersham Biosciences). The tem- perature profile of the amplification consisted of 35 cycles of 45 s dena- turation at 94¡C, 75 s annealing at 56¡C, and 75 s extension at 72¡C. PCR products were then separated on a 1% agarose gel (Invitrogen), and the image was taken by a UVP gel documentation system (Fisher Scientific). The band intensities were quantified by Quantity One densitometry software (Bio-Rad). The above protocol allows quantification to 103 cytokine mRNA transcripts per ␮g of total RNA. Flow cytometric analysis of dendritic cells (DCs) Mesenteric lymph nodes (MLNs) were removed from infected B10R111 or MMP3Ϫ/Ϫ mice 5 days postinfection and were washed in calcium-magne- sium-free HBSS (Invitrogen). MLNs were then digested with 90 U/ml collagenase 1 and 2.5 U/ml dispase (Sigma-Aldrich) for1hat37¡C. The resulting cell suspension was passed through a 30-␮m pore size cell strainer and washed in RPMI. DC were stained by FITC-conjugated B220, allophycocyanin-conjugated CD11c, and PE-conjugated CD8␣ (Caltag Laboratories, Towcester, U.K.) in PBS with 1% BSA for 30 min at 4¡C. Corresponding fluorochrome-conjugated isotype controls were also used as negative controls as described in the manufacturer’s instructions. After 3 washings in PBS with 1% BSA at 4¡C for 10 min, cells were fixed in 1% paraformaldehyde in PBS. DC were defined as CD11cϩB220Ϫ cells and were further divided into CD8␣ϩ and CD8␣Ϫ populations. DC were then analyzed by FACSCalibur (BD Biosciences, Oxford, U.K.), and the results were analyzed by WinMDI 2.8 software (Scripps Research Institute, La Jolla, CA). The absolute numbers of DCs in the MLNs were calculated from the total number of collected mesenteric lymph node cells (MLNCs) and the proportion of CD11cϩB220Ϫ cells among them. FIGURE 3. Top, Mucosal thickness in infected MMP3Ϫ/Ϫ mice. Data Adoptive transfer of CFSE-labeled cells depict average colonic crypt length (ϮSEM) in uninfected and C. rodentium- Ϫ Ϫ MLNCs were removed from infected MMP3Ϫ/Ϫ mice or B10R111 mice 7 infected WT and MMP3 / mice at different time points of infection. Average days postinfection and were passed through a nylon sieve to release lym- crypt length was significantly greater in infected WT and MMP3Ϫ/Ϫ mice than phocytes. Cells were washed in RPMI 1640 containing 10% FCS, 100 in uninfected controls between days 14 and 35 postinfection (p Ͻ 0.05). How- U/ml penicillin, and 100 ␮g/ml streptomycin (Sigma-Aldrich). The lym- ever, similar increase in crypt length was observed in WT and MMP3Ϫ/Ϫ ␮ phocytes were labeled with 10 M CSFE for 30 min at 37¡C (Molecular mice. Bottom, Pathology of colonic tissue of infected mice. Colonic tissue was Probes, Cambridge, U.K.). Labeling was stopped by adding a 5% final Ϫ Ϫ removed from the distal colon of B10R111 and MMP3 / KO mice 14 days concentration of FCS, and the cells were immediately centrifuged and after oral infection with C. rodentium, and frozen sections were cut and stained washed with ice-cold PBS. The CFSE-labeled cells from infected MMP3Ϫ/Ϫ or B10R111 mice (3 ϫ 107) were injected into naive B10R111 with H&E. The images depict tissues from uninfected B10R111 mice (A), Ϫ/Ϫ or MMP3Ϫ/Ϫ recipient mice via tail vein (three recipients/group). One day infected B10R111 mice (B), uninfected MMP3 (C), and infected Ϫ Ϫ Ϫ Ϫ after the cell transfer, mesenteric lymph nodes and colon were removed MMP3 / (D). Infection caused mucosal thickening in WT and MMP3 / from B10R111 or MMP3Ϫ/Ϫ recipient mice, and lymphocytes were pre- mice. H&E; original magnification, ϫ100. 5174 MMP-3 AND T CELL RESPONSES IN C. rodentium INFECTION

increase in crypt hyperplasia in WT mice (Fig. 3, top). However, compared with infected WT controls, MMP3Ϫ/Ϫ mice showed a similar crypt hyperplasia at various times postinfection (Fig. 3, bottom). There was no difference in the number of C. rodentium in the colon of MMP3Ϫ/Ϫ mice and WT mice on days 7 and 14 of infection but on day 21, when the WT mice had cleared the infection, bacteria were still present in the intestine of infected MMP3Ϫ/Ϫ mice (Fig. 4A). A similar pattern was seen in the spleen on day 21 (Fig. 4B). On day 35, all bacteria were cleared from the colon and spleen of MMP3Ϫ/Ϫ mice. Also on day 21, there was no signiﬁcant difference in the IgG, IgA, or IgM Ab response between infected WT and MMP3Ϫ/Ϫ mice (Fig. 5).

Reduced CD4ϩ cell infiltrates and TNF-␣ transcripts in infected colon of MMP3Ϫ/Ϫ mice Infection with C. rodentium results in heavy infiltrations of CD4ϩ cells into the colon (Fig. 6). However, in MMP3Ϫ/Ϫ mice, the increase in CD4ϩ cell infiltrates in colonic lamina propria was significantly delayed as compared with WT controls (Fig. 6). In WT mice, the numbers of CD4ϩ cells increased steadily from day

FIGURE 4. MMP3Ϫ/Ϫ mice are more susceptible to C. rodentium infection compared with WT mice. Data depict the mean number (ϮSEM) of bacteria recovered from colons (A) and spleens (B) of mice orally infected with 1Ð4 ϫ 109 CFU of DBS255(pCVD438). Bacteria are still present in high numbers in MMP3Ϫ/Ϫ mice on day 21 of infection (p Ͻ 0.05) when WT mice have cleared the infection.

Results Enhanced pathology in infected TNFRp55Ϫ/Ϫ, IL-12p40Ϫ/Ϫ, and IFN-␥Ϫ/Ϫ mice is associated with increased MMP3 transcripts Mice infected with C. rodentium develop colitis and mount a highly polarized intestinal Th1 response. A component of the response in colonic tissue of infected mice includes increased transcripts for IL-12, IFN-␥, and TNF-␣. To determine whether type I cytokines are importance in host defense and mucosal pathology, mice with targeted mutations in the IL-12p40 or IFN-␥ or TNFRp55 gene were orally infected with C. rodentium strain DBS255 (pCVD438) (5, 6). These animals showed various degree of impairment in their ability to clear infection (5, 6). Remarkably, however, mucosal pathology was greater in these mice than in WT controls. The enhanced pathology included increase in crypt length and CD4 cell inﬁltrates. Therefore, we postulated the enhanced pathology in these cytokine KO mice is due to increased mucosal remodeling. Competitive PCR showed that MMP3 transcripts were signiﬁcantly greater in infected colonic tissues of cytokine KO mice than infected WT controls (Fig. 1). In addition, MMP3 transcripts were higher in IL-12p40 or IFN-␥ KO mice, which showed more crypt hyperplasia and enhanced pathology than did TNFRp55 KO mice.

Crypt hyperplasia in MMP3 KO mice infected with C. rodentium To determine whether MMP3 is important in mucosal thickening, Ϫ/Ϫ MMP3 mice were orally infected with C. rodentium. Infection FIGURE 5. Systemic Ab response to C. rodentium. Data depict Day 21 of WT mice showed an increase in MMP3 protein in the mucosa, serum IgG, IgM, and IgA responses to EspA in infected B10R111 and Ϫ/Ϫ and as expected no MMP3 protein was seen in MMP3 mice MMP3Ϫ/Ϫ mice. There was no signiﬁcant difference in EspA-speciﬁc titers (Fig. 2). Oral infection with C. rodentium caused a substantial between infected B10R111 and MMP3Ϫ/Ϫ mice. The Journal of Immunology 5175

FIGURE 7. TNF-␣ transcripts in colonic tissue of infected MMP3Ϫ/Ϫ mice. Total RNA was isolated at weekly intervals from colonic tissue of mice orally infected with C. rodentium previously. Data depict the mean number of transcripts (Ϯ SEM) in infected and uninfected B10R111 and MMP3Ϫ/Ϫ mice. There were signiﬁcantly more transcripts detected in tissues of all infected mice compared with uninfected control at all time points (p Ͻ 0.05). TNF-␣ transcripts were also more abundant in infected WT mice on days 7 and 14 than in infected MMP3Ϫ/Ϫ mice (p Ͻ 0.05). In contrast, although on day 35 TNF-␣ transcripts were virtually absent from WT mice, they were still abundant in MMP3Ϫ/Ϫ mice (p Ͻ 0.001).

idly (Ͼ4-fold) from day 7, peaked on day 14, and declined from day 21 to a very low number on day 35. As expected, there were no MMP3 transcripts in uninfected and infected colonic tissues of Ϫ/Ϫ FIGURE 6. Top, Cell counts for CD4ϩ cells infiltrating the lamina pro- MMP3 mice. pria in infected mice. Colonic tissue was removed from the distal colon of Transcripts of other MMPs were also abundant in colonic tis- B10R111 and MMP3Ϫ/Ϫ mice weekly after oral infection with C. roden- sues of infected mice. In WT mice, transcripts of MMP7 increased tium. Tissue sections were stained for CD4ϩ T lymphocytes, and cells were significantly on day 14 and reduced in number considerably on day ϩ counted by image analysis. Data depict the mean number of CD4 T cells 21. However, in MMP3Ϫ/Ϫ mice, transcripts of MMP7 increased in infected WT mice compared with uninfected mice. In WT mice, the ϩ remarkably from day 14 and declined steadily at high levels on numbers of CD4 T cells increased steadily from day 7, peaked on day 21, days 21 and 35 (Fig. 8B). Transcripts for macrophage metalloela- Ϫ/Ϫ and reduced on day 35 postinfection. However, in MMP3 mice, the stase (MMP12) were also higher in MMP3Ϫ/Ϫ mice than in WT numbers of CD4ϩ T cells were significantly lower on day 14 than in wild- controls (Fig. 8C). type mice (p Ͻ 0.05, Student’s t test). Bottom, Immunohistochemistry of CD4 cells in colonic tissue from infected WT and KO mice. Colonic tissue The activities of MMPs are also regulated by endogenous in- was removed from the distal colon of B10R111 and MMP3Ϫ/Ϫ mice 14 hibitors TIMPs. TIMP1 transcripts were abundant in colonic tis- days after oral infection with C. rodentium. The images depict tissue from sues infected with C. rodentium (Fig. 8D). The numbers of TIMP1 uninfected B10R111 mice (A), infected B10R111 mice (B), uninfected transcripts in infected colonic tissues increased significantly from MMP3Ϫ/Ϫ mice (C), and infected MMP3Ϫ/Ϫ mice (D). Infection caused day 7, peaked on day 14, and declined to a very low level on days ϩ massive infiltration of CD4 T lymphocytes into the lamina propria in WT 21 and 35. In addition, the number of TIMP1 transcripts increased Ϫ Ϫ ϩ mice. However, in MMP3 / mice, the increase in CD4 T lymphocytes significantly (Ͼ6-fold) in MMP3Ϫ/Ϫ mice compared with WT infiltration is moderate. Immunoperoxidase immunohistochemistry; origi- controls. nal magnification, ϫ100. DCs and CD4 cell migration into the colon Previous studies have shown that MMP3 is needed for the migra- 7, peaked on day 21, and declined on day 35 postinfection. How- tion of Langerhans cells from the skin to the draining lymph nodes ϩ ever, the numbers of CD4 cells were low (Ͻ300 cells/mm2)in to initiate cutaneous sensitivity (15). We therefore reasoned that Ϫ Ϫ MMP3 / mice as compared with WT controls (Ͼ700 cells/mm2) the delayed CD4ϩ cell response in MMP3Ϫ/Ϫ mice infected with Ϫ Ϫ ϩ on day 14. In MMP3 / mice, the numbers of CD4 cells in- C. rodentium might have a similar basis. In both WT and creased from day 14 and peaked on day 21. MMP3Ϫ/Ϫ mice, there was a modest increase in B220ϪCD11cϩ TNF-␣ transcripts were abundant in colonic tissues of infected cells in the MLNs 5 days after infection. There was no significant Ϫ Ϫ WT and MMP3 / mice (Fig. 7). However, the kinetics of the difference in DC number between infected WT and MMP3Ϫ/Ϫ increase in TNF-␣ transcripts was different between infected WT mice (Fig. 9). Ϫ Ϫ and MMP3 / mice, with a delayed increase in the To determine whether CD4 T cells from MMP3Ϫ/Ϫ mice could Ϫ Ϫ MMP3 / mice. migrate into the intestine, we performed transfer experiments. CFSE-labeled CD4ϩ T cells from infected WT or MMP3Ϫ/Ϫ mice Other MMPs expressed in the colon of mice infected with C. migrated in equal number into the normal MLN or colonic lamina rodentium propria, showing no intrinsic defect in CD4ϩ T cell migration in We also conducted a detailed analysis of the kinetics of various MMP3Ϫ/Ϫ mice (Fig. 10, A and C). In contrast, CD4ϩ T cells from MMPs in WT and MMP3Ϫ/Ϫ mice throughout C. rodentium in- WT and MMP3Ϫ/Ϫ mice were impaired in their ability to migrate fection. The number of MMP3 transcripts was low in uninfected into the lamina propria and MLN of MMP3Ϫ/Ϫ mice (Fig. 10, B colon of WT controls (Fig. 8A). MMP3 transcripts increased rap- and D). 5176 MMP-3 AND T CELL RESPONSES IN C. rodentium INFECTION

FIGURE 8. MMP3, -7, and -12, and TIMP1 transcripts in colonic tissues of infected MMP3Ϫ/Ϫ mice. Total RNA was isolated at weekly intervals from colonic tissues of mice orally infected with C. rodentium. A, Mean number of transcripts (ϮSEM) for MMP3 in infected and uninfected B10R111 and MMP3Ϫ/Ϫ mice. In WT mice, there were signiﬁcantly more transcripts detected in infected mice than in uninfected controls on days 7 and 21 (p Ͻ 0.05). MMP3 transcripts were absent in MMP3Ϫ/Ϫ mice. B, Mean number of transcripts (ϮSEM) for MMP7 in infected and uninfected B10R111 and MMP3Ϫ/Ϫ mice. MMP7 transcripts were also more abundant in infected MMP3Ϫ/Ϫ mice than in infected B10R111 mice on days 14, 21, and 35 (p Ͻ 0.05). C, Mean number of transcripts (ϮSEM) for MMP12 in infected and uninfected B10R111 and MMP3Ϫ/Ϫ mice. MMP12 transcripts were also more abundant in infected MMP3Ϫ/Ϫ mice compared with infected B10R111 mice on days 21 and 35 postinfection (p Ͻ 0.05). D, Mean number of transcripts (ϮSEM) for TIMP1 in infected and uninfected B10R111 and MMP3Ϫ/Ϫ mice. TIMP1 transcripts were more abundant in infected MMP3Ϫ/Ϫ mice than in infected B10R111 mice on days 7 and 14 postinfection (p Ͻ 0.05).

A possible explanation for the failure of cells to migrate into the probably due to the delayed appearance of CD4ϩ T cells into the intestines of MMP3Ϫ/Ϫ mice could be differences in MAdCAM lamina propria and associated reduction of TNF-␣ expression in ␤ ϩ expression. We therefore analyzed MAdCAM and 7 integrin ex- the colon. The reduced migration of CD4 T cells from the donor pression in uninfected and infected mice (representative images WT or MMP3Ϫ/Ϫ mice into the lamina propria of recipient ϩ shown in Fig. 11). Only a few ␤7 cells were seen in uninfected MMP3Ϫ/Ϫ mice strongly suggests that MMP3 is important in the Ϫ/Ϫ colon from WT and MMP3 mice. In contrast, on day 14 of tissue environment for migration of CD4ϩ T lymphocytes into the ␤ ϩ infection, the colon of WT mice was infiltrated with 7 cells. In lamina propria. Ϫ/Ϫ MMP3 mice, however, there was also an increase but much The nature of the protective immune response needed for ster- less so than in WT mice, consistent with the reduced infiltrate of ilizing immunity to this pathogen is still unclear although immu- CD4 cells. MAdCAM was also expressed on vessels of uninfected nity requires T and B cells (16). We and others have previously WT and MMP3Ϫ/Ϫ mice. On day 14 of infection, there was mas- shown that mice deficient in B cells or CD4 cells are highly sus- sive mucosal thickening and the striking appearance of long ϩ ceptible to infection (17, 18). However, we were unable to clear MAdCAM vessels in both groups of mice. bacterial infection from the intestine of B cell-deficient mice with Discussion immune sera, but recently Bry and Brenner (17) achieved immunity by transferring immune sera into infected CD4-deficient mice. C. rodentium infection of mice produces features of T cell re- Taken together, these results clearly indicated that immunity needs sponse and pathology in the colon, which are virtually identical with that seen in mouse models of IBD (3). It also represents an T cell-dependent Ab production. Nonetheless, we have also shown important animal model to study immune defenses against patho- that cytokine-deficient mice are more susceptible to Citrobacter Ϫ/Ϫ gens that colonize via attaching and effacing lesions (6). In the infection (5, 6). IL-12 mice, for example, have very delayed present study, we have shown for the first time that MMP3 plays clearance of the bacteria and severe pathology; some mice die; a critical role in controlling pathogenic bacterial populations in the however, serum Ab responses are the same as those in WT mice. colon and the mobilization of T lymphocytes in the lamina propria. Likewise, in this study we also could see no difference in Ab re- Ϫ Ϫ Moreover, mucosal remodeling can occur in the absence of sponses between WT and MMP3 / mice. In the context of this Ϫ Ϫ MMP3, suggesting the redundancy of MMPs in tissue remodeling study, our observation that MMP3 / mice had delayed appear- in vivo. In infected MMP3Ϫ/Ϫ mice, there is a deficiency of im- ance of CD4ϩ T cells into the mucosa and reduced TNF-␣ tran- mune response to clear pathogenic bacteria in the colon. This is scripts again suggests a role of CD4ϩ T cells, perhaps not in The Journal of Immunology 5177

model of colitis and transgenic mice defective in TGF-␤ signaling (20). Also in our study, elevated expression of MMP3 transcripts in colonic tissues is strongly associated with enhanced pathology in infected TNFRp55Ϫ/Ϫ, IL-12p40Ϫ/Ϫ, and IFN-␥Ϫ/Ϫ mice. However, crypt hyperplasia is induced in infected MMP3Ϫ/Ϫ mice to an extent similar to that in infected WT mice. The occurrence of crypt hyperplasia in the absence of MMP3 strongly suggests the redundancy of MMPs in tissue remodeling in vivo. The crypt hyperplasia in infected MMP3Ϫ/Ϫ mice is probably due to the compensatory increase of matrilysin (MMP7) and macrophage metalloelastase (MMP12) transcripts in the infected colon. Due to the FIGURE 9. DCs in the MLNs of C. rodentium-infected WT and similar and broad substrate specificity of different MMPs, matrix MMP3Ϫ/Ϫ mice. DCs were isolated from MLNs by collagenase digestion; degradation could be conducted by other MMPs in tissues where stained with Abs against mouse B220 and CD11c, and analyzed by flow they coexpress and coexert together. Specifically, MMP7 is cytometry; 20,000 events were collected. Mouse DCs were defined as strongly implicated in bacterial infections and cancers. It is highly Ϫ ϩ B220 CD11c fraction in lower right region of each dot plot. A, Repre- expressed in epithelial cells and macrophages and can degrade sentative dot plot of DC isolated from infected B10R111 mice. B, Repre- proteoglycans, gelatins, and elastin (21). MMP12 is macrophage Ϫ/Ϫ sentative dot plot of DC isolated from infected MMP3 mice. There was specific and preferentially degrades elastin as well as other matrix no significant difference in DC frequency between infected B10R111 and Ϫ Ϫ components. It is produced predominantly by infiltrating macro- MMP3 / mice. APC, Allophycocyanin phages and appears essential for macrophage migration through the extracellular matrix. In disease, MMP12 has been strongly im- clearing the bacteria from the colon but in controlling bacterial plicated in pathogenesis in emphysema (22). In addition, the ele- Ϫ Ϫ number early in the infection. vated level of TIMP-1 expression on day 14 in infected MMP3 / The balance between MMPs and their inhibitors has been im- mice suggests the homeostatic mechanism to control the excessive plicated in the tissue remodeling and ulceration in the intestines. In activity of MMPs in matrix remodeling. Therefore, it would be Ϫ Ϫ Ϫ Ϫ the intestine, stromelysin-1 (MMP3) has been shown to expressed interesting to study the phenotypes of MMP7 / and MMP12 / at high levels at the ulcerated areas of clinical IBD and affected mice in C. rodentium infection. areas of celiac diseases (9, 11). In fetal intestinal explant model, The deficiency in controlling the bacteria in the colon of in- MMP3 is the most critical molecule in tissue injury after T cell fected MMP3Ϫ/Ϫ mice suggests that MMP3 is important in bac- activation by PWM (19). Furthermore, in animal models, MMP3 is teria defense. MMPs are strongly induced in epithelial cells during also expressed at high level in colitic samples of the SCID transfer bacterial infections. MMP7 is strongly induced in epithelial cells

FIGURE 10. Migration of CFSE-labeled T cells into MLN and lamina propria of WT or MMP3Ϫ/Ϫ mice. B10R111 or MMP3Ϫ/Ϫ donor mice were orally infected with C. rodentium. Seven days later, lymphocytes were isolated from MLNs of donor mice and labeled with CFSE. Thirty ϫ 106 MLN cells were then injected into B10R111 or MMP3Ϫ/Ϫ recipient mice through the tail vein. One day after cell transfer, MLN and colons were removed and lymphocytes in MLNs and lamina propria were isolated by collagenase digestion. Cells isolated from MLNs and lamina propria mononuclear cells were stained with Abs against CD4-PE for MLNC or CD3-PE for lamina propria mononuclear cells were analyzed by ﬂow cytometry. The percentage of positive cells was used to calculate labeled cells recovered. A and C show that lower numbers of CFSEϩ cells were recovered from the MLNs and lamina propria of recipient p Ͻ 0.05). In contrast, B and D show there is no difference in the number of CFSEϩ cells recovered ,ء) MMP3Ϫ/Ϫ mice than from recipient B10R111 mice from the MLN or lamina propria of WT B10R111 mice when donor cells were taken from B10R111 or MMP3Ϫ/Ϫ mice. 5178 MMP-3 AND T CELL RESPONSES IN C. rodentium INFECTION

FIGURE 11. Representative images of ␤7ϩ cells and MAdCAMϩ vessels in the colon of WT and MMP3Ϫ/Ϫ mice on day 14 of C. rodentium infection. Top, ␤7ϩ cells, which are few in uninfected mice. Infection results in a marked increase in ␤7ϩ cells in WT mice, but there are fewer positive cells in MMP3Ϫ/Ϫ mice, consistent with the smaller numbers of CD4 cells. Bottom, MAdCAMϩ vessels in uninfected and infected WT and MMP3Ϫ/Ϫ mice. Pos- itive vessels are obvious in uninfected animals; however, in both groups of infected mice, long positive vessels can be seen throughout the mucosa. Immu- noperoxidase immunohistochemistry; original magniﬁcation, ϫ200.

of inflamed mucosa of the stomach in Helicobacter pylori infec- support this notion, it has been shown that treatment of mice with tion (23). MMP9 is strongly induced in human gingival epithelial MMP inhibitors results in an accumulation of lymphocytes on the cells infected with Porphyromonas gingivalis (24). MMP7 has lymph node endothelium and reduced diapedesis (28). Whether the been shown to activate antimicrobial peptides, ␣-defensins (crypt- MMP inhibitors were acting via lymphocyte-derived MMPs or en- dins), in Paneth cells of mouse small intestine (25). MMP7Ϫ/Ϫ dothelial-derived MMPs was not established. We also emphasize mice are more susceptible to infection with Salmonella typhi- that in the experiments shown here, recipient animals were not murium. The role of MMP3 in bacteria defense is emphasized in infected with C. rodentium and that endothelial changes induced that proforms of MMP7 and MMP9 could be proteolytically acti- by infection might add a further layer of complexity to the system. vated by MMP3. In this regard, it was striking that although there were fewer CD4 The reduction of CD4ϩ lymphocytes in the lamina propria in cells in the colonic lamina propria of infected MMP3Ϫ/Ϫ mice on infected MMP3Ϫ/Ϫ mice suggests that MMP3 is important in mi- day 14, on days 21 and 35, the numbers were the same. This would gration of CD4ϩ lymphocytes into the lamina propria. There are suggest that inflammation overrides this early block in lymphocyte two possibilities by which MMP3 affect the CD4ϩ T lymphocyte extravasation. response. MMP3 is required for the migration of Langerhans cells Another explanation for the failure of cells to migrate into the from the skin to the draining lymph nodes to initiate cutaneous intestine could be the absence of the intestine-homing integrin ␣ ␤ sensitivity (15). The first possibility is therefore that mobilization 4 7 on lymphocytes or endothelial MAdCAM expression in of DCs is impaired in infected MMP3Ϫ/Ϫ mice, but the similar MMP3Ϫ/Ϫ mice. MAdCAM was expressed basally in MMP3Ϫ/Ϫ increase in the number of DCs recovered from draining lymph mice, and long, positively staining vessels were seen in the hyper- nodes of infected WT and MMP3Ϫ/Ϫ mice suggests that there are plastic colonic mucosa of infected MMP3Ϫ/Ϫ mice, making the no intrinsic defects of MMP3Ϫ/Ϫ mice to mobilize DCs from the latter possibility unlikely. Cells infiltrating the colon of infected Ϫ/Ϫ ␤ ϩ intestine into draining lymph nodes. The second possibility of the WT and MMP3 mice were also 7 integrin . immune defect of MMP3Ϫ/Ϫ mice is the inability of CD4ϩ lym- The mechanisms by which the MMP controls T cell migration phocytes to home to the intestine. It is evident in the CSFE cell and bacterial defense remain virtually unexplored. The data pre- transfer experiment there is a reduced migration of T lymphocytes sented in this study, using C. rodentium as a model system, add to taken from both WT and MMP3Ϫ/Ϫ mice into the mesenteric our understanding of the implications of MMPs in tissue remod- lymph nodes and lamina propria of MMP3Ϫ/Ϫ mice. Our data eling in the intestine as well as mucosal immunity in general. The therefore suggest that the tissue environment is important in mi- identification of other factors in controlling tissue remodeling in gration of T lymphocytes. Endothelial cells can make MMP3 (26, vivo and control of T cell migration by MMP are the focus of our 27), and there is a possibility that when an intestine-homing CD4ϩ current studies. cell binds to a MAdCAMϩ vessel in the colon, it signals to the endothelial cells to release MMP3 into the pericellular space to Acknowledgments open up tight junctions, and the lymphocytes migrate through. To We thank Mike P. Broome for expert assistance in tail vein injection. The Journal of Immunology 5179

References 16. Vallance, B. A., W. Deng, K. Jacobson, and B. B. Finlay. 2003. Host suscepti- bility to the attaching and effacing bacterial pathogen Citrobacter rodentium. 1. Johnson, E., and S. W. Barthold. 1979. The ultrastructure of transmissible murine Infect. Immun. 71:3443. colonic hyperplasia. Am. J. Pathol. 97:291. 17. Bry, L., and M. B. Brenner. 2004. Critical role of T cell-dependent serum anti- 2. Barthold, S. W. 1980. The microbiology of transmissible murine colonic hyper- body, but not the gut-associated lymphoid tissue, for surviving acute mucosal plasia. Lab. Anim. Sci. 30:167. infection with Citrobacter rodentium, an attaching and effacing pathogen. J. Im- 3. Higgins, L. M., G. Frankel, G. Douce, G. Dougan, and T. T. MacDonald. 1999. munol. 172:433. Citrobacter rodentium infection in mice elicits a mucosal Th1 cytokine response and lesions similar to those in murine inflammatory bowel disease. Infect. Immun. 18. Simmons, C. P., S. Clare, M. Ghaem-Maghami, T. K. Uren, J. Rankin, A. Huett, R. Goldin, D. J. Lewis, T. T. MacDonald, R. A. Strugnell, G. Frankel, and 67:3031. ϩ 4. Higgins, L. M., G. Frankel, I. Connerton, N. S. Goncalves, G. Dougan, and G. Dougan. 2003. Central role for B lymphocytes and CD4 T cells in immunity T. T. MacDonald. 1999. Role of bacterial intimin in colonic hyperplasia and to infection by the attaching and effacing pathogen Citrobacter rodentium. Infect. inflammation. Science 285:588. Immun. 71:5077. 5. Goncalves, N. S., M. Ghaem-Maghami, G. Monteleone, G. Frankel, G. Dougan, 19. Pender, S. L., S. P. Tickle, A. J. Docherty, D. Howie, N. C. Wathen, and D. J. Lewis, C. P. Simmons, and T. T. MacDonald. 2001. Critical role for tumor T. T. MacDonald. 1997. A major role for matrix metalloproteinases in T cell necrosis factor ␣ in controlling the number of luminal pathogenic bacteria and injury in the gut. J. Immunol. 158:1582. immunopathology in infectious colitis. Infect. Immun. 69:6651. 20. Hahm, K. B., Y. H. Im, T. W. Parks, S. H. Park, S. Markowitz, H. Y. Jung, 6. Simmons, C. P., N. S. Goncalves, M. Ghaem-Maghami, M. Bajaj-Elliott, J. Green, and S. J. Kim. 2001. Loss of transforming growth factor ␤ signalling in S. Clare, B. Neves, G. Frankel, G. Dougan, and T. T. MacDonald. 2002. Impaired the intestine contributes to tissue injury in inflammatory bowel disease. Gut 49: resistance and enhanced pathology during infection with a noninvasive, attach- 190. ing-effacing enteric bacterial pathogen, Citrobacter rodentium, in mice lacking 21. Murphy, G., M. I. Cockett, R. V. Ward, and A. J. Docherty. 1991. Matrix met- IL-12 or IFN-␥. J. Immunol. 168:1804. alloproteinase degradation of elastin, type IV collagen and proteoglycan: a quan- 7. Sternlicht, M. D., and Z. Werb. 2001. How matrix metalloproteinases regulate titative comparison of the activities of 95 kDa and 72 kDa gelatinases, strome- cell behavior. Annu. Rev. Cell Dev. Biol. 17:463. lysins-1 and -2 and punctuated metalloproteinase (PUMP). Biochem. J. 277 (Pt 8. Saarialho-Kere, U. K., M. Vaalamo, P. Puolakkainen, K. Airola, W. C. Parks, and 1):277. M. L. Karjalainen-Lindsberg. 1996. Enhanced expression of matrilysin, collage- 22. Morris, D. G., X. Huang, N. Kaminski, Y. Wang, S. D. Shapiro, G. Dolganov, nase, and stromelysin-1 in gastrointestinal ulcers. Am. J. Pathol. 148:519. ␣ ␤ ␤ A. Glick, and D. Sheppard. 2003. Loss of integrin v 6-mediated TGF- acti- 9. Vaalamo, M., M. L. Karjalainen-Lindsberg, P. Puolakkainen, J. Kere, and vation causes MMP12-dependent emphysema. Nature 422:169. U. Saarialho-Kere. 1998. Distinct expression profiles of stromelysin-2 (MMP- 23. Bebb, J. R., D. P. Letley, R. J. Thomas, F. Aviles, H. M. Collins, S. A. Watson, 10), collagenase-3 (MMP-13), macrophage metalloelastase (MMP-12), and tissue N. M. Hand, A. Zaitoun, and J. C. Atherton. 2003. Helicobacter pylori upregu- inhibitor of metalloproteinases-3 (TIMP-3) in intestinal ulcerations. lates matrilysin (MMP-7) in epithelial cells in vivo and in vitro in a Cag depen- Am. J. Pathol. 152:1005. dent manner. Gut 52:1408. 10. Pender, S. L., J. M. Fell, S. M. Chamow, A. Ashkenazi, and T. T. MacDonald. 24. Fravalo, P., C. Menard, and M. Bonnaure-Mallet. 1996. Effect of Porphyromonas 1998. A p55 TNF receptor immunoadhesin prevents T cell-mediated intestinal gingivalis on epithelial cell MMP-9 type IV collagenase production. Infect. Im- injury by inhibiting matrix metalloproteinase production. J. Immunol. 160:4098. mun. 64:4940. 11. Daum, S., U. Bauer, H. D. Foss, D. Schuppan, H. Stein, E. O. Riecken, and R. Ullrich. 1999. Increased expression of mRNA for matrix metalloproteinases-1 25. Wilson, C. L., A. J. Ouellette, D. P. Satchell, T. Ayabe, Y. S. Lopez-Boado, J. L. Stratman, S. J. Hultgren, L. M. Matrisian, and W. C. Parks. 1999. Regulation and -3 and tissue inhibitor of metalloproteinases-1 in intestinal biopsy specimens ␣ from patients with coeliac disease. Gut 44:17. of intestinal -defensin activation by the metalloproteinase matrilysin in innate 12. Frankel, G., A. D. Phillips, M. Novakova, H. Field, D. C. Candy, D. B. Schauer, host defense. Science 286:113. G. Douce, and G. Dougan. 1996. Intimin from enteropathogenic Escherichia coli 26. Mach, F., U. Schonbeck, R. P. Fabunmi, C. Murphy, E. Atkinson, restores murine virulence to a Citrobacter rodentium eaeA mutant: induction of J. Y. Bonnefoy, P. Graber, and P. Libby. 1999. T lymphocytes induce endothelial an immunoglobulin A response to intimin and EspB. Infect. Immun. 64:5315. cell matrix metalloproteinase expression by a CD40L-dependent mechanism: im- 13. Eckmann, L., J. Fierer, and M. F. Kagnoff. 1996. Genetically resistant (Ityr) and plications for tubule formation. Am. J. Pathol. 154:229. susceptible (Itys) congenic mouse strains show similar cytokine responses fol- 27. Pintucci, G., P. J. Yu, R. Sharony, F. G. Baumann, F. Saponara, A. Frasca, lowing infection with Salmonella dublin. J. Immunol. 156:2894. A. C. Galloway, D. Moscatelli, and P. Mignatti. 2003. Induction of stromelysin-1 Ϫ Ϫ 14. Viney, J. L., and T. T. MacDonald. 1992. Lymphokine secretion and proliferation (MMP-3) by fibroblast growth factor-2 (FGF-2) in FGF-2 / microvascular en- of intraepithelial lymphocytes from murine small intestine. Immunology 77:19. dothelial cells requires prolonged activation of extracellular signal-regulated ki- 15. Wang, M., X. Qin, J. S. Mudgett, T. A. Ferguson, R. M. Senior, and nases-1 and -2 (ERK-1/2). J. Cell Biochem. 90:1015. H. G. Welgus. 1999. Matrix metalloproteinase deficiencies affect contact hyper- 28. Faveeuw, C., G. Preece, A. Ager. 2001. Transendothelial migration of lympho- sensitivity: stromelysin-1 deficiency prevents the response and gelatinase B de- cytes across high endothelial venules into lymph nodes is affected by metallo- ficiency prolongs the response. Proc. Natl. Acad. Sci. USA 96:6885. proteinases. Blood 98:688.