Inflammation CD44 Suppresses TLR-Mediated

CD44 Suppresses TLR-Mediated Inflammation Hidetada Kawana, Hirokazu Karaki, Morihiro Higashi, Masaru Miyazaki, Frank Hilberg, Motoo Kitagawa and This information is current as Kenichi Harigaya of September 29, 2021. J Immunol 2008; 180:4235-4245; ; doi: 10.4049/jimmunol.180.6.4235 http://www.jimmunol.org/content/180/6/4235 Downloaded from

References This article cites 39 articles, 16 of which you can access for free at: http://www.jimmunol.org/content/180/6/4235.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 29, 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 © 2008 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

CD44 Suppresses TLR-Mediated Inﬂammation1

Hidetada Kawana,2* Hirokazu Karaki,2*† Morihiro Higashi,* Masaru Miyazaki,† Frank Hilberg,‡ Motoo Kitagawa,* and Kenichi Harigaya3*

The cell adhesion molecule CD44, which is the major hyaluronan receptor, has been implicated in the binding, endocytosis, and metabolism of hyaluronan. Previous studies have revealed that CD44 plays crucial roles in a variety of inflammatory diseases. In recent years, TLRs, which are ancient microbial pattern recognition receptors, have been shown to initiate an innate immune response and have been linked to a variety of inflammatory diseases. The present study shows that CD44 negatively regulates in vivo inflammation mediated by TLRs via NF-␬B activation, which leads to proinflammatory cytokine production. Furthermore, our results show that CD44 directly associates with TLR2 when stimulated by the TLR2 ligand zymosan and that the cytoplasmic domain of CD44 is crucial for its regulatory effect on TLR signaling. This study indicates that CD44 plays a protective role in TLR-mediated inflammation and is the first to demonstrate a direct association between CD44 and a TLR. The Journal of Immunology, 2008, 180: 4235–4245. Downloaded from

he adhesion molecule CD44 is a broadly distributed type ings suggest an inflammatory role for CD44. In contrast, in a study I transmembrane glycoprotein receptor for the glycosami- of bleomycin-induced acute lung injury, CD44-deficient mice T noglycan hyaluronan (HA).4 This receptor is known to be showed an enhanced and persistent inflammatory response due to involved in the binding, endocytosis, and metabolism of HA and impaired clearance of apoptotic neutrophils and HA fragments also has additional functions in innate and adaptive immunity (1). from the injury site (8). This study implies an anti-inflammatory http://www.jimmunol.org/ CD44 is constitutively expressed on both hemopoietic and paren- role for CD44. Although CD44 plays a role in inflammation, it is chymal cells. A considerable number of publications have reported becoming increasingly evident that the function of CD44 in in- that CD44 plays crucial roles in a variety of inflammatory diseases flammation is complex and involves multiple cell types, ligands, (2–8), in which CD44 expression is up-regulated on inflammatory and signaling pathways (2). Therefore, it is important to clearly cells. Most evidence for the involvement of CD44 in inflammatory understand the function of CD44 and its mechanisms in inflam- diseases has come from animal studies. Animal studies showed matory processes. that administration of anti-CD44 Abs inhibited inflammation in The current study aimed to investigate the role of CD44 in in- murine models of collagen- and proteoglycan-induced arthritis (3), flammation using a zymosan-induced arthritis (ZIA) model of cutaneous inflammation (4), experimental autoimmune encephalo- CD44-deficient mice developed by our group (9). ZIA is a popular by guest on September 29, 2021 myelitis (5), and IL-2-induced vascular leak syndrome (6). Fur- model of acute arthritis and thought to be mediated by activation thermore, CD44-deficient mice survived better than their wild-type of the alternative pathway of complement and the release of lyso- counterparts, presumably because the lack of CD44 tips physio- somal hydrolases from activated macrophages (10). The recent dis- logical responses toward a favorable outcome. Wang et al. (7) covery of pattern recognition receptors and their roles in innate reported that CD44 in ischemic brain tissue seems to be associated immunity has led to a re-evaluation of our concepts of zymosan- with a selective reduction in inflammatory cytokines. These find- induced inflammation (11). Recently, it became clear in a study on TLR2-deficient mice that TLR signaling pathways have pivotal roles in ZIA (12). *Department of Molecular and Tumor Pathology and †Department of General Sur- TLRs are a family of type 1 transmembrane proteins that are gery, Chiba University Graduate School of Medicine, Inohana, Chuo-ku, Chiba, Japan; and ‡Boehringer Ingelheim Austria, R&D Vienna, Department of Pharmacol- expressed on diverse cell types. Regulation of TLR signaling is a ogy, Dr. Boehringer Gasse, Vienna, Austria key step in acute inflammation, septic shock, and innate/adaptive Received for publication December 27, 2006. Accepted for publication January immunity. Both common-to-all and receptor-specific signals have 7, 2008. been shown to be elicited from individual receptors belonging to The costs of publication of this article were defrayed in part by the payment of page the TLR/IL-1 receptor (TIR) superfamily (13). The ligands for charges. This article must therefore be hereby marked advertisement in accordance TLR2 include lipopeptides, peptidoglycans (14), and zymosan with 18 U.S.C. Section 1734 solely to indicate this fact. (11). 1 This work was supported by a Grant-in-Aid for Scientific Research Priority Area 12215018 from the Ministry of Education, Culture, Sports, Science and Technology, The present study also aimed to examine whether there were any Japan (to K.H.) and a Grant-in-Aid for Scientific Research 13670163 from the Japan differences in the severity of ZIA between CD44-deficient Society for the Promotion of Science (to K.H.). (CD44Ϫ/Ϫ) and wild-type (CD44ϩ/ϩ) mice, and to determine 2 H.K. and H.K. contributed equally to this work. whether there are molecular interactions between CD44 and TLRs 3 Address correspondence and reprint requests to Dr. Kenichi Harigaya, Department in inflammation. of Molecular and Tumor Pathology, Chiba University Graduate School of Medicine, 1–8-1 Inohana, Chuo-ku, Chiba, 260–8670, Japan. E-mail address: harigaya@ faculty.chiba-u.jp Materials and Methods 4 Abbreviations used in this paper: HA, hyaluronan; ZIA, zymosan-induced arthritis; Reagents TIR, TLR/IL-1 receptor; polyIC, poly(I:C); LMW-HA, low molecular mass HA; HMW-HA, high molecular mass HA; B6, C57BL/6J; BM, bone marrow; MEF, Zymosan, a wall product of Saccharomyces cerevisiae, LPS from Salmo- mouse embryonic fibroblasts. nella enteriditis, and TNF-␣ were all purchased from Sigma-Aldrich. Flagellin from Salmonella typhimurium and CpG-DNA (ODN2006) were Copyright © 2008 by The American Association of Immunologists, Inc. 0022-1767/08/$2.00 from Invivogen. R848, an imidazoquinoline-derived antiviral agent, and www.jimmunol.org 4236 CD44 SUPPRESSES TLR-MEDIATED INFLAMMATION dsRNA poly(I:C) (polyIC) were from Amersham Biosciences/GE Cell culture Healthcare. Two types of purified low molecular mass HA (LMW-HA) fragments BM-derived macrophages were prepared by 4 days of culture in 10% FBS- and one type of high molecular mass HA (HMW-HA) from chicken comb containing RPMI 1640 supplemented with 25 ng/ml M-CSF (R&D Sys- (Seikagaku) were free of protein and other glycosaminoglycans and had tems). BM was collected from the femurs of adult mice and resuspended in peak molecular mass of 3, 22, and 940 kDa. These HA fragments contained complete RPMI 1640 containing 25 ng/ml M-CSF. After 2 days in culture, an endotoxin content of Ͻ0.002 ng/mg as determined by Limulus amebo- media and nonadherent cells were aspirated and replaced with fresh media, cyte lysate assays. which was replaced again after 2 days. Assays were performed on day 5 (17). Primary mouse embryonic fibroblasts (MEFs) from CD44Ϫ/Ϫ mice or Animals CD44ϩ/ϩ littermates were isolated from embryos at 12–14 days of ges- The CD44Ϫ/Ϫ mice were described previously (15). These mice and their tation. The embryos were minced and then disaggregated with 0.25% tryp- wild-type counterparts (CD44ϩ/ϩ mice) (hybrids of 129 and C57BL/6 sin and 20 U/ml DNase I in PBS for2hat37°C. The obtained MEFs were strains) were kept and bred in the Animal Unit of Chiba University Grad- maintained in DMEM containing 10% FBS. uate School of Medicine in environmentally controlled and specific patho- Raw 264.7 cells were maintained in DMEM supplemented with gen-free conditions. All data were generated from littermates. CD44Ϫ/Ϫ 10% FBS. mice that had been backcrossed onto the C57BL/6J (B6) background for DNA constructs more than six generations were obtained from The Jackson Laboratory. All experimental procedures were approved by the Institutional An- The expression vectors for the standard and epithelial forms of human imal Care and Use Committee. Typing of CD44 genes was performed CD44 (pCIneo-CD44s and pCIneo-CD44E, respectively), pCMV-MyD88 by PCR of tail DNA using the following primers: CD44 wild-type: (MyD88), pCMV-Mal (Mal), pCMV-IRAK1 (IRAK1), pCMV-TRAF6 5Ј-GGCGACTAGATCCCTCCGTT-3Ј and 5Ј-ACCCAGAGGCATAC (TRAF6), and pFlag-CMV1-TLR2 were described previously (17, 18). De- CAGCTG-3Ј; CD44 knockout: 5Ј-GTTTCATCCAGCACGCCAT-3Ј letion mutants of CD44E and TLR2 were constructed by digestion with Downloaded from and 5Ј-ATTCAGGCTGCGCAACTGT-3Ј. All experiments were per- appropriate restriction endonucleases, filling in by T4 DNA polymerases, formed using 10–14-wk-old mice. ligation to synthetic oligonucleotide linkers containing in-frame termination codons, and cloning into plasmid vectors. The deletion mutant of CD44E Preparation of bone marrow (BM) chimeras (CD44Ed: deletion of aa 292–391) was created by the introduction of a stop codon at alanine 291 (TGC)3Stop (TGA). The deletion mutant of TLR2 CD44Ϫ/Ϫ or CD44ϩ/ϩ mice (10-wk old) were used as irradiated recip- (TLR2cd: deletion of aa 612–784) was created by the introduction of a stop ients of BM cells. The mice were placed in a plastic box and X-irradiated 3

codon at phenyalanine 612 (TTC) Stop (TGA). http://www.jimmunol.org/ with 10 Gy using an MBR-1520R machine (Hitachi) at a dose rate of 2 Gy/min. At 24 h after the irradiation, the mice were given an i.v. injection EMSA of other genotype BM cells (2 ϫ 107) through the tail vein. Donor mice (also 10-wk old) were sacrificed by cervical dislocation. BM cells were The details of the EMSA used were described previously (19). BM-derived macrophages were maintained in 10% FBS-containing obtained from the tibia and femur using cold RPMI 1640 (Sigma-Aldrich) ϫ 6 supplemented with heat-inactivated FBS (Vitromex), 100 IU/ml penicillin RPMI 1640. Cells plated in 6-well dishes (1 10 cells/well) were stimulated with 10 ␮ ␣ (Invitrogen Life Technologies), and 100 ␮g/ml streptomycin (Invitrogen g/ml zymosan or 10 ng/ml LPS as samples, 1 ng/ml TNF- Life Technologies). After centrifugation, the BM cells were resuspended in as positive controls, or left untreated as negative controls. At 24 h after the stimulation, the cells were harvested, washed with PBS, and suspended in RPMI 1640 and adjusted to 1 ϫ 108 cells/ml. Next, 200 ␮l of each BM single-cell suspension was injected i.v. into lethally irradiated recipient ice-cold 20 mM HEPES-NaOH (pH 7.9) buffer containing 0.5% Nonidet P-40, 15% glycerol, 300 mM NaCl, 1 mM EDTA, 10 mM NaF, 1 mM mice. At 6 wk after the transplantation, the injected mice were used for by guest on September 29, 2021 experiments. Chimeras prepared by injecting CD44ϩ/ϩ BM cells into ir- DTT, 1 mM sodium orthovanadate, 0.5 mM PMSF, 50 mM calpain inhib- radiated CD44Ϫ/Ϫ mice are herein referred to as CD44ϩ/ϩ BM chimeras itor-1, 1 mg/ml leupeptin, 1 mg/ml pepstatin, and 1 mg/ml aprotinin. After (44ϩ/ϩ344Ϫ/Ϫ). Other chimeras prepared by injecting CD44Ϫ/Ϫ BM 30 min of gentle agitation at 4°C, the supernatants were collected by cen- ϩ ϩ Ϫ Ϫ trifugation as whole-cell extracts. DNA-protein binding reactions (15 ␮l) cells into irradiated CD44 / mice are referred to as CD44 / BM ␮ chimeras (44Ϫ/Ϫ344ϩ/ϩ). were performed by incubating the whole-cell extracts (20 g equivalent of For chimera control groups, CD44ϩ/ϩ BM cells were injected into protein) in a 13 mM HEPES-NaOH (pH 7.9) buffer containing 8% glycerol, 50 mM NaCl, 0.4 mM MgCl , 0.5 mM DTT, 66.6 ␮g/ml poly(dI-dC), irradiated CD44ϩ/ϩ mice (B6 background) and CD44Ϫ/Ϫ BM cells were 2 Ϫ Ϫ and 33.3 ␮g/ml salmon sperm DNA for 15 min on ice, followed by an injected into irradiated CD44 / mice (B6 background), and the resulting 32 chimeras are referred to as CD44ϩ/ϩ mice (44ϩ/ϩ344ϩ/ϩ) and additional 30-min incubation with a P-end-labeled synthetic double CD44Ϫ/Ϫ mice (44Ϫ/Ϫ344Ϫ/Ϫ), respectively. stranded oligonucleotide probe (0.1 ng, 5 nCi) at room temperature. The final concentration of the probe was 0.25 fmol/␮l. Half of the mixture was loaded onto a polyacrylamide gel (5%) in 0.5 ϫ Tris-borate-EDTA buffer Induction and examination of arthritis to separate the DNA-protein complexes, and the separated complexes were Zymosan was prepared and sterilized as a 15 mg/ml suspension in saline. detected by exposing the dried gels to x-ray films. The oligonucleotide sequences for the labeled probe were from Ig ␬ L chain enhancer B as The material was thoroughly agitated before use to ensure that the suspen- follows: 5Ј-GAT CCA GAG GGG ACT TTC CGA GAG-3Ј and 5Ј-GAT sion was homogeneous. Knees were prepared for injection by depilation CCT CTC GGA AAG TCC CCT CTG-3Ј. with a razor to expose the patellar ligament, and the area was cleaned with 70% ethanol. Under diethyl ether anesthesia, the left knee was injected Transcriptional transactivation assays (luciferase assays) with 10 ␮l of the zymosan suspension through a 28-gauge needle fitted to a syringe, and the right knee was injected with saline as a control (9). There Wild-type or CD44Ϫ/Ϫ MEFs were seeded in 12-well dishes (1 ϫ 105 were 14 tibiofemoral joints in seven mice of each genotype and 16 tib- cells/well). Transient transfection was conducted using Lipofectamine iofemoral joints in eight mice of each chimera type. To eliminate undefined 2000 (Invitrogen Life Technologies) and the cells were cotransfected with genetic influences and inflammatory effects, there were 46 additional tib- 0.5 ␮g of p-55Ig␬-luc reporter plasmid for NF-␬B and 10 ng of Renilla iofemoral joints in 46 BM-transplanted mice (44ϩ/ϩ344ϩ/ϩ: n ϭ 22; luciferase internal control plasmid (pRL-CMV; Promega) (20). For CD44 44Ϫ/Ϫ344ϩ/ϩ: n ϭ 16; and 44Ϫ/Ϫ344Ϫ/Ϫ: n ϭ 8) on the B6 back- re-expression experiments, CD44ϩ/ϩ MEFs were cotransfected with 0.5 ground. At 1 wk after the injection, the mice were sacrificed with an over- ␮g of p-55Ig␬-luc reporter plasmid for NF-␬B and 10 ng of Renilla lucif- dose of diethyl ether. Half of the joints were arthritic and the other half erase internal control plasmid (pRL-CMV), while CD44Ϫ/Ϫ MEFs were were saline controls to eliminate injection effects. The knees were re- cotransfected with 0.5 ␮g of p-55Ig␬-luc reporter plasmid for NF-␬B, 10 moved, fixed in 10% neutral-buffered formalin, decalcified in 5% EDTA- ng of Renilla luciferase internal control plasmid (pRL-CMV), and 0.5 ␮g 2Na solution, and embedded in paraffin. Next, the specimens were cut into of pCIneo (Mock), 0.5 ␮g of pCIneoCD44s (for full-length CD44s re- 4-␮m sections and stained with H&E. Sections of ZIA tissues were mi- expression), or 0.5 ␮g of pCIneoCD44E (for full-length CD44E re-expres- croscopically evaluated for the intensities of three parameters of arthritis, sion). At 24 h after the transfection, cells were either left unstimulated (as namely cellular infiltration, synovial cell hyperplasia, and pannus forma- negative controls) or stimulated with 10 ␮g/ml zymosan, 10 ng/ml LPS as tion as described by Ohshima et al. (16) with minor modifications. The samples, or 1 ng/ml TNF-␣ as positive controls for 24 h. Next, the firefly intensity of each parameter was scored from 0 to 3 as follows: 0, within the and Renilla luciferase activities were determined using a Dual Luciferase normal range; 1, mild changes; 2, moderate changes; and 3, severe assay kit (Promega) and a TD20/20 dual luminometer (Turner Designs). changes. The firefly luciferase activity was normalized by the Renilla luciferase The Journal of Immunology 4237

control activity. Values were expressed as the mean relative stimulation analyses of the fluorescence intensities were performed using the same and SD for representative experiments (19). Data were analyzed as the fold software. induction of luciferase activity over unstimulated cells and are representative of the average of three identical experiments. Statistical analysis To investigate the effects of normal or mutant CD44 molecules on TLR Differences between groups were analyzed using the StatView J 4.5 Ϫ Ϫ ϫ 5 signaling, CD44 / MEFs were seeded in 12-well dishes (1 10 cells/ software statistical package (23). Scores for histological grading were well). Transient transfection was conducted using Lipofectamine 2000, and analyzed by the Mann-Whitney U test. Transcriptional transactivation the cells were transfected with p-55Ig␬-luc reporter plasmid (0.5 ␮g), in- ␮ ␮ assays and cytokine production by macrophages were analyzed using ternal control plasmid (0.01 g), and pCIneoCD44E (0.5 g), Student’s t test. pCIneoCD44E (0.1 ␮g) plus pCIneo (0.4 ␮g), pCIneoCD44Ed (deletion of aa 291–361; 0.5 ␮g), pCLneoCD44Ed (deletion of aa 291–361; 0.4 ␮g) plus pCIneo (0.1 ␮g), or pCIneo (Mock; 0.5 ␮g), the volume of which was Results 1.001 ␮g/1 ml. At 24 h after the transfection, the cells were stimulated with CD44 modulates the inflammatory response in ZIA ␮ 10 g/ml zymosan or 100 ng/ml LPS for 24 h. The enzyme activity was To obtain insights into the role of CD44 and the relationships measured as described above. between CD44 and TLRs in inflammatory diseases, we initially Cytokine production from macrophages attempted to develop experimental arthritis in CD44Ϫ/Ϫ and ϩ ϩ Primary cultured BM macrophages from CD44Ϫ/Ϫ mice or CD44ϩ/ϩ CD44 / mice using zymosan, a classic inducer of acute joint littermates were plated in 24-well plates at a density of 2 ϫ 105 cells/well disease. in 0.5 ml of medium. Next, the cells were stimulated with zymosan (10 At 7 days after a zymosan injection, CD44Ϫ/Ϫ mice developed ␮g/ml), LPS (10 ng/ml), polyIC (10 ng/ml), CpG-DNA (ODN2006; 2 markedly more severe lesions in their arthritic joints than ␮M), flagellin (10 ng/ml), or R848 (1 ␮g/ml) for 24 h or left untreated (as CD44ϩ/ϩ mice (Fig. 1, A and B). Histologically, ZIA is charac- Downloaded from negative controls). ELISA kits for mouse IL-6 (R&D Systems or Biosource International) and mouse TNF-␣ (Biosource International) were used to terized by inflammatory cell infiltration, synovial cell hyperplasia, determine cytokine production. and pannus formation in the joint space (24). Mice from each To assess the effect of HA on the LPS signaling pathway, CD44ϩ/ϩ genotype were examined histologically for these three features of Ϫ Ϫ ␮ and CD44 / macrophages were stimulated with 20 g/ml HA of 3, 22, arthritis and graded for their severities to examine whether there or 940 kDa in molecular mass (18). At 2 h after the HA treatment, the cells were further stimulated with LPS (10 ng/ml) or medium. After another were any significant differences. The arthritis assessments of the

24 h, the cytokine production was determined as described above. Exper- individual mice of each group are summarized in Fig. 1C. The http://www.jimmunol.org/ iments were repeated at least twice in triplicate. histological scores of both inflammatory cell infiltration and synovial cell hyperplasia were significantly elevated in CD44Ϫ/Ϫ Overexpression analysis mice compared with CD44ϩ/ϩ mice. Histopathologically, in- To investigate the topological region of the CD44 molecule involved in the creased infiltration of polymorphonuclear cells, mononuclear cells, signaling cascade mediated by TIR-containing receptors, CD44Ϫ/Ϫ and ϩ ϩ ␬ and macrophages, and remarkable hyperplasia of the synovial tis- CD44 / MEFs were transfected with p-55Ig -luc reporter plasmid (0.5 Ϫ Ϫ ␮g), internal control pRL-CMV (0.01 ␮␥), and pCMV-MyD88 (MyD88), sue, were seen in the arthritis in CD44 / mice. Although the pCMV-Mal (Mal), pCMV-IRAK1 (IRAK1), pCMV-TRAF6 (TRAF6), or pannus formation tended to be more severe in CD44Ϫ/Ϫ mice their empty counterpart pCMV vector (Mock) (0.5 ␮g) (17, 21, 22). than in CD44ϩ/ϩ mice, statistical analysis did not show any sig- The MEFs were maintained in DMEM, containing 20% FBS, and sub- nificant differences between the two groups. These results suggest by guest on September 29, 2021 jected to luciferase assays at 48 h after the transfection. that CD44 may have a protective role against the development Western blotting and immunoprecipitation of ZIA. CD44 is present on both hemopoietic cells and parenchymal HEK293 cells were transfected with Lipofectamine 2000. To perform immunoprecipitation experiments, HEK293 cells transfected with various cells, such as epithelial cells and fibroblasts (2). To examine the plasmids were washed with serum-free DMEM and stimulated with zy- cell populations responsible for the severity of the inflammation, mosan in serum-free DMEM. Whole cell lysates were prepared with RIPA we transferred BM cells from CD44Ϫ/Ϫ mice into irradiated buffer (0.1 M Tris-HCl (pH 7.4), 100 mM NaCl, 0.5% Triton X-100, 1% CD44ϩ/ϩ mice (CD44Ϫ/Ϫ BM chimeras, 44Ϫ/Ϫ344ϩ/ϩ) and deoxycholate, and 0.1% SDS) containing 0.5% deoxycholic acid, 2 ϩ ϩ Ϫ Ϫ ϩ ϩ mmol/L sodium vanadate, 50 mmol/L sodium fluoride, 50 mg/ml leupep- CD44 / BM cells into irradiated CD44 / mice (CD44 / tin, 25 mg/ml aprotinin, and 10 mg/ml pepstatin (23). Lysates were incu- BM chimeras, 44ϩ/ϩ344Ϫ/Ϫ). When arthritis was induced by bated with an anti-Flag Ab (M2; Sigma-Aldrich) or an anti-HA Ab zymosan in these mice, CD44Ϫ/Ϫ BM chimeras exhibited more (12CA5) as a control. Immune complexes were precipitated with protein severe changes in inflammatory cell infiltration and synovial cell G-Sepharose (Amersham Biosicences) and analyzed with anti-Flag and hyperplasia than CD44ϩ/ϩ BM chimeras (Fig. 2, A and B). The anti-CD44 (Hermes3) Abs as described previously (19). histological scores of these two parameters were significantly el- Immunofluorescence and confocal laser microscopy evated in CD44Ϫ/Ϫ BM chimeras compared with CD44ϩ/ϩ BM RAW 264.7 cells grown on glass coverslips were washed with serum-free chimeras (Fig. 2C). Although the pannus formation tended to be DMEM and incubated with zymosan in serum-free DMEM for 10 min at more severe in CD44Ϫ/Ϫ BM chimeras than in CD44ϩ/ϩ BM 37°C. After fixation and permeabilization, non-specific sites and Fc recep- chimeras, statistical analysis did not show any significant differ- tors were blocked by incubation in blocking buffer [PBS containing 10% ences between the two groups. Furthermore, the grade of arthritis calf serum and 10 ␮g/ml anti-mouse Fc receptors (2.4G2)]. Double-label- Ϫ Ϫ ing studies were performed using a primary rabbit polyclonal Ab against in CD44 / BM chimeras appeared to be similar to that in CD44 (1: 20; Santa Cruz Biotechnology) and a mouse mAb against TLR2 CD44Ϫ/Ϫ mice, and the grade in CD44ϩ/ϩ BM chimeras ap- (1:20; Abcam). For the double-labeling experiments, the cells were incu- peared to be similar to that in CD44ϩ/ϩ mice. These results in- bated with the primary Abs together overnight. After a 40-min incubation dicate that CD44 on hemopoietic cells could be responsible for the with secondary Abs (FITC-conjugated anti-mouse (1:40) and tetramethyl- severities of inflammatory cell infiltration and synovial cell hyper- rhodamine isomer R [TRITC]-conjugated anti-rabbit (1:40) Abs; both from Dako A/S) in PBS containing 1% BSA, the cells were washed three times plasia in ZIA. in PBS and mounted in 0.1 M Tris-HCl-glycerol containing 5% n-propyl To eliminate other undefined genetic influences and inflam- gallate (WAKO). Control experiments were performed in the same man- matory effects while generating chimeras, we generated ner except that the primary Abs were excluded. Confocal microscopy CD44ϩ/ϩ mice with BM from CD44ϩ/ϩ mice (44ϩ/ϩ3 was performed using an LSM510 META system (Carl Zeiss), equipped ϩ ϩ Ϫ Ϫ Ϫ Ϫ with appropriate dual laser excitation and emission filters for maximum 44 / ), CD44 / mice with BM from CD44 / mice separation of the tetramethylrhodamine isomer R and FITC signals. (44Ϫ/Ϫ344Ϫ/Ϫ), and CD44ϩ/ϩ mice with BM from Stained cells were analyzed with the LSM imaging software. Histogram CD44Ϫ/Ϫ mice (CD44Ϫ/Ϫ BM chimeras, 44Ϫ/Ϫ344ϩ/ϩ), 4238 CD44 SUPPRESSES TLR-MEDIATED INFLAMMATION Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 2. Zymosan-induced arthritis in CD44Ϫ/Ϫ and CD44ϩ/ϩ FIGURE 1. Ϫ Ϫ ϩ ϩ Zymosan-induced arthritis in CD44 / and CD44 / chimeric mice. A and B, Sagittal sections of whole knee joints from a A B ϩ ϩ A mice. and , Sagittal sections of whole knee joints from CD44 / ( ) CD44ϩ/ϩ (A) and CD44Ϫ/Ϫ (B) BM chimeras at 7 days after induction Ϫ Ϫ B and CD44 / ( ) mice at 7 days after induction of arthritis with zymosan. of arthritis with zymosan. Scale bars ϭ 200 ␮m. C, Scores for histological ϭ ␮ C Scale bars 200 m. , Scores for histological grading of inflammatory grading of inflammatory cell infiltration, synovial cell hyperplasia, and cell infiltration, synovial cell hyperplasia, and pannus formation in pannus formation in CD44ϩ/ϩ and CD44Ϫ/Ϫ BM chimeras. The mean ϩ ϩ Ϫ Ϫ CD44 / and CD44 / mice. The mean values are indicated by solid values are indicated by solid black lines. The statistical significance of black lines. The statistical significance of differences was evaluated using differences was evaluated using the Mann-Whitney U test. Each open or U the Mann-Whitney test. Each open or closed circle corresponds to the closed circle corresponds to the score of an individual mouse. score of an individual mouse. all of which had been backcrossed onto a B6 background for hyperplasia than CD44ϩ/ϩ mice (44ϩ/ϩ344ϩ/ϩ) (Fig. 3). Al- more than six generations. Arthritis was then induced by zy- though the pannus formation tended to be more severe in mosan in these mice. CD44Ϫ/Ϫ mice (44Ϫ/Ϫ344Ϫ/Ϫ) and CD44Ϫ/Ϫ mice (44Ϫ/Ϫ344Ϫ/Ϫ) and CD44Ϫ/Ϫ BM chimeras CD44Ϫ/Ϫ BM chimeras (44Ϫ/Ϫ344ϩ/ϩ) exhibited more se- (44Ϫ/Ϫ344ϩ/ϩ) than in CD44ϩ/ϩ mice (44ϩ/ϩ344ϩ/ϩ), vere changes in inflammatory cell infiltration and synovial cell statistical analysis did not show any significant differences among The Journal of Immunology 4239

p<0.02 p<0.01 A Control Zymosan LPS TNF 44+/+→44+/+ p<0.01 p<0.01 44-/-→44+/+ 44-/-→44-/- 3 +/+ -/- +/+ -/- +/+ -/- +/+ -/-

0 Cellular infiltration Synovial cell Pannus formation hyperplasia FIGURE 3. Zymosan-induced arthritis in CD44ϩ/ϩ mice (44ϩ/ϩ3 44ϩ/ϩ), CD44Ϫ/Ϫ mice (44Ϫ/Ϫ344Ϫ/Ϫ), and CD44Ϫ/Ϫ BM chi- Ϫ Ϫ3 ϩ ϩ meras (44 / 44 / ). Scores for histological grading of inflammatory Downloaded from cell infiltration, synovial cell hyperplasia, and pannus formation in CD44ϩ/ϩ mice (44ϩ/ϩ344ϩ/ϩ), CD44Ϫ/Ϫ mice (44Ϫ/Ϫ344Ϫ/Ϫ), and CD44Ϫ/Ϫ BM chimeras (44Ϫ/Ϫ344ϩ/ϩ). The mean values are indicated by solid black lines. The statistical significance of differences was evaluated using the Mann-Whitney U test. Each open or closed circle, or open triangle corresponds to the score of an individual mouse. B LPS http://www.jimmunol.org/

+/+ -/- Ϫ Ϫ Ϫ Ϫ3 these three groups. The results for CD44 / mice (44 / 0 10 20 120 0 10 20 120 ( min ) 44Ϫ/Ϫ), CD44Ϫ/Ϫ BM chimeras (44Ϫ/Ϫ344ϩ/ϩ), and CD44ϩ/ϩ mice (44ϩ/ϩ344ϩ/ϩ) on the B6 background were comparable to those from CD44Ϫ/Ϫ mice, CD44Ϫ/Ϫ BM chimeras, and CD44ϩ/ϩ mice on the hybrid background, respectively. CD44 on both hemopoietic and non-hemopoietic cells has been by guest on September 29, 2021 shown to contribute to the extent of inflammation (2). However, in the case of ZIA, CD44 on hemopoietic cells may have a more important role than CD44 on non-hemopoietic cells, since similar results were observed for CD44Ϫ/Ϫ mice (44Ϫ/Ϫ344Ϫ/Ϫ) and CD44Ϫ/Ϫ BM chimeras (44Ϫ/Ϫ344ϩ/ϩ) in ZIA. These results also suggest that CD44 on hemopoietic cells may have a role in FIGURE 4. NF-␬B activities analyzed by EMSA. A, Increased activities TLR signaling, since TLR signaling plays a pivotal role in ZIA. of NF-␬B in response to zymosan or LPS in CD44Ϫ/Ϫ macrophages. Cells were stimulated with 10 ng/ml zymosan, 10 ng/ml LPS, or 1 ng/ml TNF-␣ CD44 suppresses NF-␬B activation in marrow-derived for 24 h or left untreated (Control). B, Kinetics of NF-␬B activation in macrophages response to LPS in macrophages. Cells were stimulated with 10 ng/ml LPS To assess the effects of CD44-deficency on TLR signaling path- for the indicated periods. Open and closed triangles indicate activated ways, we examined the activation of NF-␬B, one of the principal NF-␬B dimers and free probes, respectively. signal transducers of TLR signaling, in BM macrophages. The activation of NF-␬B was monitored using an EMSA at 24 h after TLR ligand stimulation (13). CD44Ϫ/Ϫ macrophages exhibited elevated NF-␬B activity compared with their CD44ϩ/ϩ counter- ulation (Fig. 4B). These data indicate that CD44 diminished the parts in response to zymosan. Gram-negative bacteria-derived NF-␬B activation induced by the TLR signaling pathway. LPS, a well-characterized ligand for TLR4 (25), also induced el- ␬ evated activation of NF-␬B in CD44Ϫ/Ϫ cells compared with CD44 suppresses NF- B reporter activation after zymosan or CD44ϩ/ϩ cells. In contrast to TLR ligands, TNF-␣ stimulation LPS stimulation provoked a similar robust response in each cell type (Fig. 4A). To evaluate the role of CD44 in the activation of NF-␬B stimula- Similar results were also obtained when primary MEFs from tion by either zymosan or LPS, we examined the transcriptional CD44Ϫ/Ϫ mice were stimulated with either zymosan or LPS (data transactivation activity of NF-␬B by luciferase assays and used not shown). These results indicate that the activation of NF-␬B MEFs from CD44Ϫ/Ϫ and CD44ϩ/ϩ mice. CD44ϩ/ϩ MEFs did was enhanced and prolonged in CD44Ϫ/Ϫ cells compared with not exhibit NF-␬B-dependent transcriptional activation by zymo- CD44ϩ/ϩ cells. In time-course studies, the two genotypes of mac- san, whereas significant NF-␬B-dependent activation was detected rophages displayed similar levels of NF-␬B activation at 10 min in CD44Ϫ/Ϫ MEFs. This CD44-dependent suppression of NF-␬B after LPS stimulation, whereas significantly increased levels of activation was also found in the experiments with LPS (Fig. 5A). NF-␬B activation were observed in CD44Ϫ/Ϫ macrophages com- These results are consistent with our notion obtained by the EMSA pared with CD44ϩ/ϩ controls at 20 and 120 min after the stim- that CD44 suppresses NF-␬B activation. 4240 CD44 SUPPRESSES TLR-MEDIATED INFLAMMATION

A p<0.01 4 A B CD44+/+ ** CD44 -/- 300 CD44+/+ 500 CD44+/+ CD44 -/- ** CD44 -/- p<0.01 250 3 400

200 300

2 ) ( pg/ml

150 α 200 IL-6 ( pg/ml ) ( pg/ml IL-6

100 TNF- Relative Activation 1 100 50

N.D. 0 0 0 Control Zymosan Control Zymosan Control Zymosan LPS C ** B CD44+/+ CD44+/+ CD44 -/- CD44-/- ( Mock ) 10000 CD44-/- ( pCIneoCD44H ) Downloaded from 3 CD44-/- ( pCIneoCD44E ) ** ** ** 8000 4000 **

IL-6 (pg/ml) IL-6 * 2 2000 http://www.jimmunol.org/

0 LPS Relative Activation ODN R848

1 Control Flagellin Poly (IC) 5000 D CD44+/+ by guest on September 29, 2021 CD44 -/- ** 4000 ** 0 ** Control Zymosan TNF-α 3000 FIGURE 5. NF-␬B activities analyzed by luciferase assays. A, In- creased transcriptional activities of NF-␬B in response to zymosan or LPS (pg/ml) ** Ϫ Ϫ ϩ ϩ Ⅺ Ϫ Ϫ f α in CD44 / MEFs. CD44 / ( ) and CD44 / ( ) MEFs were trans- 2000 fected with a reporter plasmid (0.5 ␮g) and an internal control plasmid TNF- (0.01 ␮g). Cells were left unstimulated (Control) or stimulated with 10 ␮g/ml zymosan or 10 ng/ml LPS. Vertical axes represent the normalized 1000 luciferase activities relative to the mean activity of unstimulated MEFs. * Error bars indicate the SD (n ϭ 3). B, Expression of CD44 reverses the hyperactivation of NF-␬B in CD44Ϫ/Ϫ MEFs. CD44ϩ/ϩ and CD44Ϫ/Ϫ 0 MEFs were transfected with a reporter plasmid (0.5 ␮g), internal control LPS plasmid (0.01 ␮g), and pCIneo (Mock; 0.5 ␮g) pCIneoCD44s (0.5 ␮g), or ODN R848 Control Flagellin pCIneoCD44E (0.5 ␮g). Cells were left unstimulated (Control) or stimu- Poly (IC) lated with 10 ␮g/ml zymosan or 1 ng/ml TNF-␣. Vertical axes represent FIGURE 6. Proinflammatory cytokine production by murine BM mac- the normalized luciferase activities relative to the mean activity of un- rophages. A and B, Increased production of IL-6 (A) and TNF-␣ (B)in stimulated CD44ϩ/ϩ MEFs. Error bars indicate the SD (n ϭ 3). Data are CD44Ϫ/Ϫ BM macrophages stimulated by zymosan (10 ␮g/ml) or left p Ͻ 0.01. untreated (Control). C and D, Increased production of IL-6 (C) and ,ءء .given as means Ϯ SD TNF-␣ (D) in CD44Ϫ/Ϫ macrophages stimulated by various TLR ligands or left untreated (Control). The cells were stimulated with LPS To further examine the effect of CD44 on NF-␬B, CD44 re- (10 ng/ml), polyIC (10 ng/ml), CpG-DNA (ODN; 2 ␮M), flagellin (10 expression experiments were performed. Activation of NF-␬B ng/ml), or R848 (1 ␮g/ml). The cytokine production was determined by by zymosan in CD44Ϫ/Ϫ MEFs was significantly decreased to ELISA. Error bars indicate the SD (n ϭ 3). Experiments were repeated p Ͻ ,ء .the level in wild-type MEFs after transfection of a CD44s ex- at least twice in triplicate, and similar results were obtained .p Ͻ 0.01 ,ءء ;pression vector. Similar effects were found after transfection of 0.05 a CD44E expression vector. These results indicate that both variants of CD44 could suppress NF-␬B reporter activation. In contrast, TNF-␣-dependent NF-␬B activation was not altered in sults indicate that CD44-deficiency is a genetically determined MEFs regardless of the presence or absence of CD44 (Fig. 5B) cause of the NF-␬B hyperactivation induced by zymosan, one to rule out non-specific effects of CD44 transfection. These re- of the TLR ligands. The Journal of Immunology 4241

5 A 6000 p<0.05 CD44+/+ CD44-/- CD44+/+ 4 5000 CD44-/-

p<0.05 3 4000

2 3000 * Relativ Activ e ation IL-6 (pg/ml) 1 2000 * * * 0 Mock M yD 88 M al IR AK1 T R AF6 1000 * FIGURE 8. Effects of CD44 on NF-␬B activation after overexpression N.D. N.D. N.D. N.D. 0 of intracellular mediators in TLR signaling. CD44 has no signiﬁcant effects Cont LPS HA HA HA LPS LPS LPS on the NF-␬B activation induced by intracellular mediators (MyD88, Mal, +HA +HA +HA IRAK1, and TRAF6) of TLR signaling. CD44Ϫ/Ϫ and CD44ϩ/ϩ MEFs MW of HA(Da) - - 3k 22k 940k 3k 22k 940k ␬ ␮ µ were transfected with the p-55Ig -luc reporter plasmid (0.5 g), internal HA( g/ml) - - 20 20 20 20 20 20 ␮ LPS(ng/ml) - 10 - - - 10 10 10 control plasmid (0.01 g) and pCMV-MyD88 (MyD88), pCMV-Mal (Mal), pCMV-IRAK1 (IRAK1), pCMV-TRAF6 (TRAF6), or their empty p<0.01 ␮ ϭ Downloaded from B 5000 counterpart pCMV vector (Mock; 0.5 g). Error bars indicate the SD (n p<0.01 3). CD44+/+ CD44-/- 4000

These results indicate that CD44Ϫ/Ϫ cells are hyperresponsive 3000

to both zymosan and the other TLR ligands examined, suggesting http://www.jimmunol.org/

α (pg/ml) * that CD44 has an inhibitory role in a common signaling pathway

TNF- 2000 of all the TLRs tested. * * * Effects of HA and CD44 on LPS-induced cytokine production 1000 HA is widely distributed under homeostatic conditions as an ex- tracellular and cell surface nonsulfated polysaccharide of high mo-

0 lecular mass, usually several million Da.In contrast, accumulation Cont LPS HA HA HA LPS LPS LPS of low molecular mass forms of HA occurs at sites of inflammation +HA +HA +HA (8). HA interacts with cell surface receptors including CD44, thus by guest on September 29, 2021 MW of HA(Da) - - 3k 22k 940k 3k 22k 940k HA(µg/ml) - - 20 20 20 20 20 20 influencing cell behavior through direct receptor-mediated effects LPS(ng/ml) - 10 - - - 10 10 10 (18, 26). Recently Jiang et al. (1) reported that both TLR2 and FIGURE 7. Proinflammatory cytokine production in response to LPS TLR4 seem to drive lung inflammation in response to the frag- and HA. A and B, Inhibitory effect of CD44 on TLR signaling is not mented form of HA. Currently, both TLR2 and TLR4 are candi- affected by HA. CD44ϩ/ϩ and CD44Ϫ/Ϫ macrophages were stimulated dates for HA-binding receptors. Therefore, we assessed whether by HA with the indicated molecular mass. After HA treatment, the cells HA stimulation of macrophages affected TLR signaling in a CD44- were further stimulated with LPS or medium alone. Production of IL-6 (A) dependent or -independent manner. In these experiments, we used and TNF-␣ (B) was determined by ELISA. Experiments were repeated at three different molecular mass ranges of HA, all of which induced least twice in triplicate with comparable results. ND: Not detected. Error activation of focal adhesion kinase through CD44 in our earlier p Ͻ 0.05 for LPS-stimulated CD44ϩ/ϩ or ,ء .(bars indicate the SD (n ϭ 3 study (18). In our experiments, the sole addition of each different CD44Ϫ/Ϫ macrophages with HA vs those without HA, respectively. molecular mass HA (3, 22, and 940 kDa) did not significantly affect the cytokine production by BM macrophages. In LPS-treated cells, IL-6 and TNF-␣ production levels by both CD44Ϫ/Ϫ and CD44 suppresses proinflammatory cytokine production upon CD44ϩ/ϩ macrophages were significantly reduced by the addi- ligation of TLRs tion of LMW-HA (3 and 22 kDa), indicating that the inhibitory To investigate whether CD44 affects signaling from TLRs, we pre- effect of LMW-HA for LPS-induced NF-␬B activation is not re- pared primary BM macrophages from CD44Ϫ/Ϫ mice or lated to CD44 expression. In contrast, TNF-␣ production by both CD44ϩ/ϩ littermates, treated the cells with zymosan, and exam- CD44 Ϫ/Ϫ and CD44ϩ/ϩ macrophages and IL-6 production by ined the production of the proinflammatory cytokines IL-6 and CD44Ϫ/Ϫ macrophages were not significantly influenced by the TNF-␣. CD44Ϫ/Ϫ BM macrophages stimulated by zymosan pro- addition of HMW-HA (940 kDa) (Fig. 7, A and B), indicating that duced significantly increased amounts of both IL-6 and TNF-␣ HMW-HA may not be involved in the inhibitory effect of CD44 on compared with their CD44ϩ/ϩ counterparts (Fig. 6, A and B). LPS-induced NF-␬〉 activation. As a result, it seems that the in- Furthermore, we assessed the production of both cytokines in re- hibitory effect of CD44 on TLR signaling differs in its mechanism sponse to several other TLR ligands, namely LPS (TLR4 ligand), from that of LMW-HA. polyIC (TLR3 ligand), unmethylated CpG-DNA (ODN2006; ␬ TLR9 ligand), flagellin (TLR5 ligand),and the imidazoquinoline- CD44 does not affect NF- B activation after overexpression of derived antiviral agent R848 (TLR7 ligand). The production of intracellular mediators in TLR signaling both cytokines was significantly increased in CD44Ϫ/Ϫ BM mac- Overexpression of several intracellular components of TLR sig- rophages compared with CD44ϩ/ϩ BM macrophages in response naling has been shown to activate NF-␬B, probably by mimicking to all the TLR ligands tested. the signaling process (17, 22, 27). These components include 4242 CD44 SUPPRESSES TLR-MEDIATED INFLAMMATION

TRAF6 (a RING ﬁnger protein), IRAK1 (a serine/threonine kinase), and MyD88 and Mal/TIRAP (adaptor molecules) (13, 28, 29). We cotransfected CD44ϩ/ϩ and CD44Ϫ/Ϫ MEFs with TRAF6, IRAK1, MyD88 or Mal/TIRAP, and the NF-␬B-dependent luciferase reporter. Overexpression of TRAF6, IRAK1, MyD88, or Mal/TIRAP resulted in activation of NF-␬B in both CD44ϩ/ϩ and CD44Ϫ/Ϫ MEFs to similar degrees (Fig. 8). These results suggest that CD44 can restrain TLR signaling upstream of TRAF6, IRAK1, MyD88, and Mal/TIRAP. These results imply that CD44 may inhibit signaling from TLRs proximally to receptors on the cell membrane.

Zymosan stimulation induces an association between CD44 and TLR2 To obtain insights into the molecular mechanisms for how CD44 inhibits TLR signaling, we ﬁrst examined the subcellular localizations of CD44 and TLR2 using double-labeling studies involving immunoﬂuorescence and confocal laser microscopy. In agreement with previous examinations (11), TLR2 was enriched on phago- Downloaded from somal membranes containing zymosan particles (Fig. 9A, particles 1 and 2) in a murine macrophage cell line (RAW 264.7). CD44 was also found in the areas around zymosan particles (Fig. 9B, particles 1 and 2), which are comparable to phagosomal membranes. Simultaneous detection of TLR2 and CD44 revealed that

the two molecules resided in the areas around zymosan particles http://www.jimmunol.org/ (Fig. 9C, particles 1 and 2), indicating that TLR2 and CD44 were colocalized around zymosan particles. Histograms of the intensities of TLR2 and CD44 on a dashed white line (Fig. 9C) revealed that the intensity pattern of TLR2 showed a similar distribution to that of CD44 (Fig. 9D), conﬁrming that the two molecules were preferentially colocalized around zymosan particles in endogenous cells. Next, we addressed the direct association between CD44 and TLR2 using immunoprecipitation assays. As shown in Fig. 9G, by guest on September 29, 2021 CD44 and TLR2 were coimmunoprecipitated from an extract of HEK293 cells stimulated by zymosan, but were not coimmunoprecipitated from an extract without stimulation. To assess the roles of the cytoplasmic domains of TLR2 and CD44 in the direct association, we constructed a truncated CD44 mutant (CD44Ed) pCIneoCD44Ed291–361 that lacked the carboxyl-terminal 71 amino acids and almost all the cytoplasmic domain (Fig. 9E). We also constructed a truncated TLR2 mutant pFlag-CMV1-TLR2cd that lacked the cytoplasmic carboxyl-terminal 173 amino acids including the TIR domain (Fig. 9F). CD44Ed and full-length TLR2

the white dashed line in C. The x-axis represents the point-to-point distance on the white dashed line in C in mm, and the y-axis shows the ﬂuorescence intensities of TLR2 (green) and CD44 (red) on an arbitrary scale. Similar patterns of intensity distribution are observed for TLR2 and CD44. E, Schematic diagrams of CD44E and CD44Ed (CD44E deletion mutant: deletion of aa 291–361). The deletion mutant lacks most of the cytoplasmic section (71 amino acids) of CD44E, which includes two functional domains (ezrin-radixin-moesin-binding domain and ankyrin-binding domain). F, Schematic diagrams of TLR2 and TLR2cd (cytoplasmic deletion mutant of TLR2). The deletion mutant lacks most of the cytoplasmic sec- FIGURE 9. Double-staining and immunoprecipitation of TLR2 and tion (173 amino acids; deletion of aa 612–784) of TLR2, which includes CD44. A–D, Double-staining confocal microscopy study with anti-TLR2 the TIR domain. G–I, CD44 is coimmunoprecipitated with TLR2. HEK293 (A) and anti-CD44 (B) Abs in RAW 264.7 cells. A, TLR2 staining (green) cells were transfected in the presence (ϩ) or absence (Ϫ) of expression is detected around internal zymosan (particles 1 and 2) in a RAW 264.7 vectors for CD44 (150 kDa), cytoplasmic domain-deleted CD44 (CD44cd; cell. B, CD44 staining (red) is also detected around internal zymosan par- 150 kDa), TLR2 (90 kDa) (G and H), and cytoplasmic domain-deleted ticles (1 and 2) in a RAW 264.7 cell. C, A merged confocal image of TLR2 TLR2 (TLR2cd; 75 kDa) (I) as indicated. After 10 min of incubation with (green) and CD44 (red) indicates that TLR2 and CD44 colocalize around or without zymosan, cells were lysed and immunoprecipitated with an anti- the zymosan particles (1 and 2). N: Nucleus of the RAW 264.7 cell. D, Flag Ab for TLR2 or TLR2cd, or an anti-HA Ab as a control. TLR2 and Histogram of the intensities of TLR2 (green) and CD44 (red) staining on CD44 in the precipitate were detected by Western blotting. The Journal of Immunology 4243

A the cytoplasmic domain of CD44 has a regulatory role in TIR signaling and leads to NF-␬B activation by zymosan or LPS.

2 p < 0.05 Discussion Eleven TLR paralogues encoded by mammalian genomes sense p < 0.05 macromolecules specific to microbes and initiate the production of cytokines that elicit an inflammatory response (13, 28, 29). In their intracellular domains, TLRs share structural homology with the 1 receptor for a proinflammatory cytokine, IL-1. Both common-to-

Relative Activation Relative all and receptor-specific signals have been shown to be elicited from individual receptors belonging to the TIR superfamily (13). One of the major pathways activated by TLRs culminates in the activation of the transcription factor NF-␬B, which acts as a master 0 Zym( µ g/ml) 10 10 10 10 10 10 switch for inflammation. Throughout this study of ZIA in mice, the mock( µg) 0.5 0.4 - 0.5 0.4 - absence of CD44 molecules in BM cells aggravated inflammation CD44E( µg) - 0.1 0.5 - - - and enhanced TLR2-mediated NF-␬B activation. These results in- CD44Ed( µg) - - - - 0.1 0.5 dicated that CD44 has a regulatory role in the inflammatory responses induced by TLR ligands. Recruitment of inflammatory B p < 0.05 3 cells is one of the earliest events in the acute inflammatory re- Downloaded from sponse and CD44 plays a significant role in the localization of cells in an inflammatory lesion (2). This role of CD44 may affect the kinetics of the response in ZIA models. We histologically con- p < 0.05 2 firmed the maximal severity of ZIA at day 7, when the most prom- inent features of ZIA should be observed (24, 33). The increased p = 0.08

severity of ZIA observed for CD44Ϫ/Ϫ BM cells at the peak pe- http://www.jimmunol.org/ riod supports the notion that CD44 regulates the magnitude of 1 TLR-mediated inflammation. Relative Activation Relative CD44 plays a role in inflammation, and CD44 on both hemopoietic and non-hemopoietic cells has been known to contribute to the extent of inflammation (2). However, our findings for ZIA in 0 LPS (ng/ml) 100 100 100 100 100 100 chimeric mice imply that CD44 expression on hemopoietic cells is mock( µg) 0.5 0.4 - 0.5 0.4 - more important for the extent of ZIA than CD44 expression on CD44E ( µg) - 0.1 0.5 - - - CD44Ed ( µg) - - - - 0.1 0.5 non-hemopoietic cells. ZIA is thought to be mediated by the activation of the alternative pathway of complement from activated by guest on September 29, 2021 FIGURE 10. Roles of the cytoplasmic domain of CD44. A and B, macrophages (9, 10). Recently, zymosan has become accepted as NF-␬B activities analyzed by luciferase assays. Expression of full-length CD44E reduces zymosan-induced (A) and LPS-induced (B) NF-␬B acti- one of the ligands for TLRs, and a study of the ZIA model of vation in CD44Ϫ/Ϫ MEFs, whereas expression of CD44Ed (deletion of aa TLR2-deficient mice revealed that TLR signaling contributes to 291–361) appears to enhance LPS-induced NF-␬B activation in CD44Ϫ/Ϫ the extent of ZIA (12). Thus, our findings indicate that CD44 may MEFs (B). contribute to the extent of ZIA in TLR signaling. When we examined the cytokine production in response to TLR ligands for TLR2, TLR3, TLR4, TLR6, TLR8, and TLR9, en- (Fig. 9, G and H) or full-length CD44 and TLR2cd (Fig. 9I) were hanced cytokine production was always found in CD44Ϫ/Ϫ BM not coimmunoprecipitated in the presence or absence of zymosan macrophages compared with CD44ϩ/ϩ cells. These results sug- stimulation. gest that CD44 negatively regulates TLR-mediated cytokine pro- These results suggest that zymosan stimulation leads to an as- duction, possibly through its common-to-all signaling. sociation between CD44 and TLR2. Furthermore, it seems that the TLR signaling is negatively regulated by several molecules, TLR2 cytoplasmic domain including the TIR region and the CD44 such as SOCS-1 (21, 22), IRAK-M (34), single immunoglobulin cytoplasmic domain are necessary for the association between IL-1R-related molecule (35), and ST2 (27). Among these, these two molecules. SOCS-1, IRAK-M, and ST2 were barely expressed at steady-state and strongly induced after LPS stimulation in macrophages. These The CD44 cytoplasmic domain may have a regulatory role in three factors are required to establish LPS tolerance (21, 22, 27, TLR signaling 34, 35), a protection mechanism against LPS-induced tissue dam- Previous studies have shown that the cytoplasmic domain of CD44 age. Thus, we examined the ability of CD44Ϫ/Ϫ macrophages to has intercellular signaling motifs and protein domains necessary develop LPS tolerance. Contrary to our expectations, CD44Ϫ/Ϫ for interactions with cytoskeletal components (30). To assess the cells did not differ from CD44ϩ/ϩ cells in their LPS tolerance role of the cytoplasmic domain of CD44 in TLR signaling, we (data not shown). It seems likely that CD44 is a factor that defines examined the transcriptional activation of NF-␬B in MEFs trans- the overall magnitude of the LPS response but not LPS tolerance, fected with a truncated CD44 mutant pCIneoCD44Ed (deletion of in contrast to other negative regulators. Although CD44-deficiency aa 291–361), which behaves as a dominant-negative molecule in enhanced the LPS-mediated response as effectively as zymosan in the functioning of CD44 (31, 32). Transfection of the CD44Ed vitro, the in vivo role of LPS remains unclear. Thus, we generated (deletion of aa 291–361) mutant into CD44ϩ/ϩ MEFs enhanced LPS-sublethal shock models of CD44Ϫ/Ϫ and CD44ϩ/ϩ mice. the activation of NF-␬B while transfection of full-length CD44E Surprisingly, CD44Ϫ/Ϫ mice displayed dramatic hypersensitivity into CD44Ϫ/Ϫ MEFs reduced both the zymosan and LPS-induced to LPS compared with CD44ϩ/ϩ mice (data not shown), consis- activations of NF-␬B (Fig. 10, A and B). These results indicate that tent with data in a recent report (36). From this finding, it is clear 4244 CD44 SUPPRESSES TLR-MEDIATED INFLAMMATION that CD44 plays an indispensable role in the regulation of TLR- diological Science, Chiba, Japan) for the BM transplantation; mediated responses in vivo. Further analyses could lead to the N. Suzuki and S. Suzuki for useful comments and helpful discussions; and development of new treatments for human LPS-induced disease. T. Saito (RIKEN Research Center for Allergy and Immunology, Kana- CD44 is a well-known HA-binding receptor protein, and now gawa, Japan) for critical reading of the manuscript. both TLR2 and TLR4 are also candidates for HA-binding receptor proteins. Thus, we assessed the effects of HA on TLR-mediated Disclosures cytokine production by BM macrophages. We showed that The authors have no financial conflict of interest. LMW-HA (3 and 22 kDa) had inhibitory effects on TLR signaling, but these effects were independent of CD44 expression. As a re- References sult, it seems that the inhibitory effect of CD44 on TLR signaling 1. Jiang, D., J. Liang, J. Fan, S. Yu, S. Chen, Y. Luo, G. D. Prestwich, M. M. Mascarenhas, H. G. Garg, D. A. Quinn, et al. 2005. Regulation of lung differs in its mechanism from that of LMW-HA. Furthermore, it is injury and repair by toll-like receptors and hyaluronan. Nat. Med. 11: 1173–1179. known that synthesis of HA increases under inflammatory circum- 2. Pure, E., and C. A. Cuff. 2001. A crucial role for CD44 in inflammation. Trends stances, and that TNF-␣ is an important stimulator of HA produc- Mol. Med. 7: 213–221. ␣ 3. Zeidler, A., R. Brauer, K. Thoss, J. Bahnsen, V. Heinrichs, tion (37). In our experiments, TNF- stimulation resulted in sim- D. Jablonski-Westrich, M. Wroblewski, S. Rebstock, and A. Hamann. 1995. ilar levels of NF-␬B activation in both CD44ϩ/ϩ and CD44Ϫ/Ϫ Therapeutic effects of antibodies against adhesion molecules in murine collagen genotypes of BM macrophages and MEFs. These results support type II-induced arthritis. Autoimmunity 21: 245–252. 4. Camp, R. L., A. Scheynius, C. Johansson, and E. Pure. 1993. CD44 is necessary the conclusion that the inhibitory effect of CD44 on TLR signaling for optimal contact allergic responses but is not required for normal leukocyte is independent of HA. extravasation. J. Exp. Med. 178: 497–507.

5. Brocke, S., C. Piercy, L. Steinman, I. L. Weissman, and T. Veromaa. 1999. Downloaded from An emerging concept in signal transduction is that cell adhesion Antibodies to CD44 and integrin ␣4, but not L-selectin, prevent central nervous molecules can function as coreceptors. CD44 is known to serve as system inflammation and experimental encephalomyelitis by blocking secondary a coreceptor for growth factors and associate with their receptors, leukocyte recruitment. Proc. Natl. Acad. Sci. USA 96: 6896–6901. 6. Rafi-Janajreh, A. Q., D. Chen, R. Schmits, T. W. Mak, R. L. Grayson, such as tyrosine kinase receptor ErbB or c-Met (38, 39). These D. P. Sponenberg, M. Nagarkatti, and P. S. Nagarkatti. 1999. Evidence for the findings prompted us to examine the subcellular localizations of involvement of CD44 in endothelial cell injury and induction of vascular leak CD44 and TLRs. As expected, CD44 was found in the area around syndrome by IL-2. J. Immunol. 163: 1619–1627. 7. Wang, X., L. Xu, H. Wang, Y. Zhan, E. Pure, and G. Z. Feuerstein. 2002. CD44 phagosomes containing zymosan where TLR2 can also exist. We deficiency in mice protects brain from cerebral ischemia injury. J. Neurochem. http://www.jimmunol.org/ also found that CD44 directly associated with TLR2 by using 83: 1172–1179. HEK293 cells after endocytosis of zymosan. In addition, no sig- 8. Teder, P., R. W. Vandivier, D. Jiang, J. Liang, L. Cohn, E. Pure, P. M. Henson, and P. W. Noble. 2002. Resolution of lung inflammation by CD44. Science 296: nificant differences were found between CD44Ϫ/Ϫ and CD44ϩ/ϩ 155–158. BM macrophages regarding alterations in their expressions of 9. Keystone, E. C., H. U. Schorlemmer, C. Pope, and A. C. Allison. 1977. Zymosan- induced arthritis: a model of chronic proliferative arthritis following activation of TLR2 and TLR4, LPS-binding abilities, and endocytic activities the alternative pathway of complement. Arthritis Rheum. 20: 1396–1401. for zymosan (data not shown). The study involving overexpression 10. Schorlemmer, H. U., D. Bitter-Suermann, and A. C. Allison. 1977. Complement of intracellular signaling molecules for TLRs suggested that CD44 activation by the alternative pathway and macrophage enzyme secretion in the pathogenesis of chronic inflammation. Immunology 32: 929–940. may inhibit signaling from TLR2 and TLR4 proximally to recep- 11. Underhill, D. M., A. Ozinsky, A. M. Hajjar, A. Stevens, C. B. Wilson, tors on the cell membrane. Transfection of a CD44 cytoplasmic M. Bassetti, and A. Aderem. 1999. The toll-like receptor 2 is recruited to mac- by guest on September 29, 2021 deletion mutant did not reproduce the inhibitory effect on TLR rophage phagosomes and discriminates between pathogens. Nature 401: 811–815. signaling, and the mutant did not associate with TLR2 after stim- 12. Frasnelli, M. E., D. Tarussio, V. Chobaz-Peclat, N. Busso, and A. So. 2005. ulation with zymosan. Therefore, the cytoplasmic domain of CD44 TLR2 modulates inflammation in zymosan-induced arthritis in mice. Arthritis may be critical for the regulation of TLR signaling, possibly by Res. Ther. 7: R370–R379. 13. Akira, S., and K. Takeda. 2004. Toll-like receptor signalling. Nat. Rev. Immunol. interacting with the TIR domain. Indeed, a truncated TLR2 mutant 4: 499–511. that lacked the TIR domain could not associate with CD44 under 14. Schwandner, R., R. Dziarski, H. Wesche, M. Rothe, and C. J. Kirschning. 1999. Peptidoglycan- and lipoteichoic acid-induced cell activation is mediated by toll- zymosan stimulation. From these results, it is tempting to speculate like receptor 2. J. Biol. Chem. 274: 17406–17409. that CD44 functions as an inhibitory molecule that rapidly moves 15. Protin, U., T. Schweighoffer, W. Jochum, and F. Hilberg. 1999. CD44-deficient to phagosomes and physically associates with TLR2. However, it mice develop normally with changes in subpopulations and recirculation of lym- phocyte subsets. J. Immunol. 163: 4917–4923. remains possible that the inhibitory effect of CD44 is modulated by 16. Ohshima, S., Y. Saeki, T. Mima, M. Sasai, K. Nishioka, S. Nomura, M. Kopf, endogenous ligands for CD44 other than HA. Y. Katada, T. Tanaka, M. Suemura, and T. Kishimoto. 1998. Interleukin 6 plays This report is the first to demonstrate that CD44 directly asso- a key role in the development of antigen-induced arthritis. Proc. Natl. Acad. Sci. USA 95: 8222–8226. ciates with TLR2 and negatively regulates TLR-mediated re- 17. Suzuki, N., S. Suzuki, G. S. Duncan, D. G. Millar, T. Wada, C. Mirtsos, sponses. Accordingly, the ultimate function of CD44 may be to H. Takada, A. Wakeham, A. Itie, S. Li, et al. 2002. Severe impairment of interleukin-1 and toll-like receptor signalling in mice lacking IRAK-4. Nature 416: protect tissues or organs against damage due to a severe inflam- 750–756. matory response in vivo. 18. Fujita, Y., M. Kitagawa, S. Nakamura, K. Azuma, G. Ishii, M. Higashi, H. Kishi, It is clear that CD44 is a potentially important molecule that T. Hiwasa, K. Koda, N. Nakajima, and K. Harigaya. 2002. CD44 signaling through focal adhesion kinase and its anti-apoptotic effect. FEBS Lett. 528: regulates inflammatory responses. Therefore, we believe that elu- 101–108. cidating the molecular mechanisms of CD44 in TLR signaling 19. Kitagawa, M., T. Oyama, T. Kawashima, B. Yedvobnick, A. Kumar, K. Matsuno, pathways will help toward achieving an understanding of the and K. Harigaya. 2001. A human protein with sequence similarity to Drosophila mastermind coordinates the nuclear form of notch and a CSL protein to build a mechanisms of inflammatory diseases and thereby provide novel transcriptional activator complex on target promoters. Mol. Cell. Biol. 21: insights into new therapies against inflammatory disorders. 4337–4346. 20. Fujita, T., G. P. Nolan, H. C. Liou, M. L. Scott, and D. Baltimore. 1993. The candidate proto-oncogene bcl-3 encodes a transcriptional coactivator that acti- Acknowledgments vates through NF-␬ B p50 homodimers. Genes Dev. 7: 1354–1363. 21. Nakagawa, R., T. Naka, H. Tsutsui, M. Fujimoto, A. Kimura, T. Abe, E. Seki, We thank B. Seed (Harvard Medical School, Boston, MA), T. Fujita (The S. Sato, O. Takeuchi, K. Takeda, et al. 2002. SOCS-1 participates in negative Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan), regulation of LPS responses. Immunity 17: 677–687. N. Watanabe, H. Shimada, A. Asari (Seikagaku, Tokyo, Japan), S. Akira, 22. Kinjyo, I., T. Hanada, K. Inagaki-Ohara, H. Mori, D. Aki, M. Ohishi, H. Yoshida, S. Sato (Osaka University, Osaka, Japan), and N. Suzuki and S. Suzuki M. Kubo, and A. Yoshimura. 2002. SOCS1/JAB is a negative regulator of LPS- induced macrophage activation. Immunity 17: 583–591. (RIKEN Research Center for Allergy and Immunology, Kanagawa, Japan) 23. Kawana, H., J. Tamaru, T. Tanaka, A. Hirai, Y. Saito, M. Kitagawa, A. Mikata, for the reagents used in this study; K. Yoshida (National Institute of Ra- K. Harigaya, and T. Kuriyama. 1998. Role of p27Kip1 and cyclin-dependent The Journal of Immunology 4245

kinase 2 in the proliferation of non-small cell lung cancer. Am. J. Pathol. 153: 32. Lesley, J., Q. He, K. Miyake, A. Hamann, R. Hyman, and P. W. Kincade. 1992. 505–513. Requirements for hyaluronic acid binding by CD44: a role for the cytoplasmic 24. Van Meurs, J. B., P. L. van Lent, A. E. Holthuysen, I. I. Singer, E. K. Bayne, and domain and activation by antibody. J. Exp. Med. 175: 257–266. W. B. van den Berg. 1999. Kinetics of aggrecanase- and metalloproteinase-in- 33. Van de Loo, F. A., S. Kuiper, F. H. van Enckevort, O. J. Arntz, and duced neoepitopes in various stages of cartilage destruction in murine arthritis. W. B. van den Berg. 1997. Interleukin-6 reduces cartilage destruction during Arthritis Rheum. 42: 1128–1139. experimental arthritis: a study in interleukin-6-deficient mice. Am. J. Pathol. 151: 177–191. 25. Poltorak, A., X. He, I. Smirnova, M. Y. Liu, C. Van Huffel, X. Du, D. Birdwell, 34. Kobayashi, K., L. D. Hernandez, J. E. Galan, C. A. Janeway, Jr., R. Medzhitov, E. Alejos, M. Silva, C. Galanos, et al. 1998. Defective LPS signaling in C3H/HeJ and R. A. Flavell. 2002. IRAK-M is a negative regulator of toll-like receptor and C57BL/10ScCr mice: mutations in Tlr4 gene. Science 282: 2085–2088. signaling. Cell 110: 191–202. 26. Ponta, H., L. Sherman, and P. A. Herrlich. 2003. CD44: from adhesion molecules 35. Wald, D., J. Qin, Z. Zhao, Y. Qian, M. Naramura, L. Tian, J. Towne, J. E. Sims, to signalling regulators. Nat. Rev. Mol. Cell Biol. 4: 33–45. G. R. Stark, and X. Li. 2003. SIGIRR, a negative regulator of toll-like receptor- 27. Brint, E. K., D. Xu, H. Liu, A. Dunne, A. N. McKenzie, L. A. O’Neill, and interleukin 1 receptor signaling. Nat. Immunol. 4: 920–927. F. Y. Liew. 2004. ST2 is an inhibitor of interleukin 1 receptor and Toll-like 36. Liang, J., D. Jiang, J. Griffith, S. Yu, J. Fan, X. Zhao, R. Bucala, and P. W. Noble. receptor 4 signaling and maintains endotoxin tolerance. Nat. Immunol. 5: 2007. CD44 is a negative regulator of acute pulmonary inflammation and lipo- 373–379. polysaccharide-TLR signaling in mouse macrophages. J. Immunol. 178: 28. Beutler, B. 2004. Inferences, questions and possibilities in toll-like receptor sig- 2469–2475. nalling. Nature 430: 257–263. 37. Wilkinson, T. S., S. Potter-Perigo, C. Tsoi, L. C. Altman, and T. N. Wight. 2004. Pro- and anti-inflammatory factors cooperate to control hyaluronan synthesis in 29. Barton, G. M., and R. Medzhitov. 2003. Toll-like receptor signaling pathways. lung fibroblasts. Am. J. Respir. Cell Mol. Biol. 31: 92–99. Science 300: 1524–1525. 38. Yu, W. H., J. F. Woessner, Jr., J. D. McNeish, and I. Stamenkovic. 2002. CD44 30. Entwistle, J., C. L. Hall, and E. A. Turley. 1996. HA receptors: regulators of anchors the assembly of matrilysin/MMP-7 with heparin-binding epidermal signalling to the cytoskeleton. J. Cell. Biochem. 61: 569–577. growth factor precursor and ErbB4 and regulates female reproductive organ re- 31. Jiang, H., R. S. Peterson, W. Wang, E. Bartnik, C. B. Knudson, and W. Knudson. modeling. Genes Dev. 16: 307–323. 2002. A requirement for the CD44 cytoplasmic domain for hyaluronan binding, 39. Orian-Rousseau, V., L. Chen, J. P. Sleeman, P. Herrlich, and H. Ponta. 2002. pericellular matrix assembly, and receptor-mediated endocytosis in COS-7 cells. CD44 is required for two consecutive steps in HGF/c-Met signaling. Genes Dev. Downloaded from J. Biol. Chem. 277: 10531–10538. 16: 3074–3086. http://www.jimmunol.org/ by guest on September 29, 2021