Requirements for Signal Delivery Through CD44: Analysis Using CD44-Fas Chimeric Proteins1

Haruko Ishiwatari-Hayasaka,2,3* Takashi Fujimoto,2* Tomoko Osawa,2* Toshiyasu Hirama,4* Noriko Toyama-Sorimachi,† and Masayuki Miyasaka5*

CD44 is a transmembrane glycoprotein involved in various events, including lymphocyte migration, early hemo- poiesis, and tumor metastasis. To examine the requirements of CD44 for signal delivery through the extracellular domain, we constructed a chimeric CD44 fused to the intracellular domain of Fas on its C-terminus. In cells expressing the CD44-Fas fusion protein, could be induced by treatment with certain anti-CD44 mAbs alone, especially those recognizing the epitope group d, which has been previously shown to play a role in ligand binding, indicating that ligation of a specific region of the CD44 extracellular domain results in signal delivery. Of note was that appropriate ligation of the epitope h also resulted in the generation of apoptotic signal, although this region was not thought to be involved in ligand binding. In contrast, the so-called blocking anti-CD44 mAbs (epitope group f) that can abrogate the binding of hyaluronate (HA) failed to induce apoptosis even after further cross-linking with the secondary Ab, indicating that a mere mAb-induced oligomerization of the chimeric is insufficient for signal generation. However, these blocking mAbs were instead capable of inhibiting apoptosis induced by non- blocking mAb (epitope group h). Furthermore, a chimeric protein bearing a mutation in the HA binding domain and hence lacking the ability to recognize HA was incapable of mediating the mAb-induced apoptosis, suggesting that the functional integrity of the HA binding domain is crucial to the signal generation in CD44. The Journal of Immunology, 1999, 163: 1258–1264.

D44 is a versatile adhesion molecule implicated in var- (ECM) components such as collagen or fi- ious cell traffic events, such as migration of activated T bronectin (11, 12). Thus, it is suggested that one of the important C cells into sites of inflammation and distant metastasis of biological functions of CD44 is acting as a matrix receptor that tumor cells (1–3). CD44 has an N-terminal link module involved mediates cell adhesion to ECM (1). in the binding to (HA)6 (4, 5), and multiple lines On the other hand, CD44 can interact with ligands apparently of evidence indicate that recognition of HA is important in various unrelated to ECM, such as serglycin (13, 14), osteopontin (15), and CD44-mediated cellular events. For instance, early lymphopoiesis the MHC class II invariant chain (16). Of interest, serglycin en- is inhibited by anti-CD44 mAbs and also by treatment with hyal- hances CD3-dependent granzyme A release of CTL clones (13), by guest on October 1, 2021. Copyright 1999 Pageant Media Ltd. uronidase (6, 7). CD44-negative cell lines transfected with CD44 suggesting that non-HA ligands can also signal through CD44. cDNA become adherent to high endothelial cells of lymph nodes The CD44-mediated signaling from cell membrane to the nu- or an endothelial cell line, of which binding is sensitive to hyal- cleus was initially shown experimentally using anti-CD44 mAbs. uronidase (8, 9). Soluble CD44-Ig fusion protein binds to high Engagement of CD44 with specific mAbs induces proliferation of endothelial cells of lymph nodes, which is sensitive to hyaluron- T cells when coincubated with anti-CD2 or anti-CD3 mAbs (17– idase and blocked by soluble HA (10). Activation of T cells in- 19). Similarly, stimulation of CD44 with mAbs enhances the cy- creases their binding to HA and enables CD44-mediated lympho- totoxic activity of NK cells or CTL (20, 21) and induces the release cyte rolling (3). In addition to HA, CD44 can interact with other of TNF-␣ and IL-1␤ in monocytes (22). However, various anti-

http://classic.jimmunol.org CD44 mAbs that recognize different epitopes can apparently stim- ulate CD44-positive cells, although it is not clear at present which *Department of Bioregulation, Biomedical Research Center, Osaka University Grad- domain(s) of CD44 should be stimulated and what exactly leads to uate School of Medicine, Suita, Japan; and †Department of Immunology, The Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan signal delivery through this interesting multifunctional molecule. Received for publication April 20, 1998. Accepted for publication May 20, 1999. To further understand the precise requirements for CD44-medi- The costs of publication of this article were defrayed in part by the payment of page ated signaling, we expressed on a lymphoid cell line, the chimeric charges. This article must therefore be hereby marked advertisement in accordance CD44 proteins consisting of the extracellular (EC) domain of Downloaded from with 18 U.S.C. Section 1734 solely to indicate this fact. CD44 and the intracellular (IC) domain of a death-inducing mol- 1 This work was supported by a grant-in-aid from the Ministry of Education, Science, ecule, Fas, which transduce apoptosis signals into cells upon ap- and Culture, Japan, and a grant from the Science and Technology Agency, Japan. propriate cross-linking of the extracellular region. We then inves- 2 H.I.-H., T.F., and T.O. contributed equally to the completion of this work. tigated the important sites for signal delivery in the CD44 EC 3 Current address: Department of Cell Biology, The Tokyo Metropolitan Institute of Medical Science, 3-18-22 Hon-Komagome, Bunkyo, Tokyo 113-8613, Japan. domain using apoptosis as a parameter of CD44-mediated signal- 4 Current address: National Institute of Radiological Sciences, 4-9-1 Anagawa, Inage- ing, subsequent to stimulation with a variety of mAbs against dif- ku, Chiba City 263-8555, Japan. ferent epitopes of the CD44 EC domain. We have previously used 5 Address correspondence and reprint requests to Dr. Masayuki Miyasaka, Depart- a similar strategy to demonstrate that stimulation through the lectin ment of Bioregulation, Biomedical Research Center, Osaka University Graduate domain is crucial to signal generation in a leukocyte adhesion mol- School of Medicine, 2-2 Yamada-oka, Suita 565-0871, Japan. E-mail address: [email protected] ecule, L-selectin (23). The results of the present study indicate that 6 Abbreviations used in this paper: HA, hyaluronic acid; ECM, extracellular matrix; not only the HA binding domain but also another non-HA binding EC, extracellular; IC, intracellular; TM, transmembrane; POD, peroxidase. domain play a critical role in signal generation in CD44.

Copyright © 1999 by The American Association of Immunologists 0022-1767/99/$02.00 The Journal of Immunology 1259

Materials and Methods Table I. Isotypes and epitope specificities of anti-CD44 mAbs used in Construction of CD44-Fas fusion protein this study

Full-length murine CD44H cDNA (24) was inserted into a mammalian Epitope Induction of Effect on HA expression vector pEF-BOS (25) to construct a plasmid encoding wild-type CD44 mAbs Isotypes Specificitiesa Apoptosis Bindinga CD44. A recombinant plasmid, 44/F-4, encoding the EC domain, the trans- membrane (TM) domain, and most of the IC domain of murine CD44 IRAWB14.4 IgG2a d ϩ 11 (nucleotides 1–1092) and the IC domain of human Fas, was constructed as RAWB45.106.2 IgM d ϩ 2 follows. First, we introduced an NheI site in the multicloning site of the KM201 IgG1 f Ϫ 2 pUC119 vector. Then a DNA fragment encoding the full-length CD44 or KM114 IgG1 f Ϫ 2 Fas (26) was inserted into the modified pUC119 to construct pUCcd44 IM7.8.1 IgG2b h ϩb 3 or pUCfas, respectively. The fragment encoding the IC domain of Fas R7 166.7 IgM h ϩ 1 (nucleotides 756-1427) was amplified by PCR by using pUCfas as the RAMBM5.5.8 IgM h ϩ 3 template. The PCR primers used were 5Ј-TGGCTGCAGGTTTGGGT KM703 IgG2a h ϩ (2)c GAAGAGAAAG-3Ј containing a PstI site at the 5Ј terminus and 5Ј- Ј a The data are based on the analysis presented in the article by Zheng et al. (29). TACTTAGCATGCCACTGCATT-3 containing an SphI site. The result- b ant PstI-SphI fragment was ligated with pUC119 (pUCfasPCR). A KpnI- Apoptosis was examined in the presence of a secondary Ab. c This effect depends on types of the cell expressing CD44. PstI fragment containing the nucleotide sequence 1–1092 of CD44 was isolated from pUCcd44 and transferred to pUCfasPCR at KpnI-PstI sites. Then a chimeric fragment consisting of nt 1–1092 of CD44 and 756-1427 of Fas was isolated by digestion with NheI and XbaI, and inserted into pEF-BOS. treated with 20 ␮g/ml of each Ab for 10 min. After pretreatment, cells were The 44/F-4RA plasmid was prepared by modifying 44/F-4. For this incubated for 5 h with 5 ␮g/ml R7 166.7 or overnight with KM703. To purpose, a fragment encoding the EC domain of CD44 (nt 1–827) con- 23 examine HA-induced cell death, cells were incubated for 17 h with 100 taining the Arg Ala mutation was amplified by two-step PCR (amino acid ␮ 23 g/ml human umbilical cord HA (ICN, Costa Mesa, CA) and 5 ng/ml numbers used here are those of mature murine protein; thus, Arg corre- PLUS 41 IRAWB14.4. Cell death was detected by Cell Death Detection ELISA sponds to Arg in man) (27). First, two sets of reaction were performed (Boehringer Mannheim, Mannheim, Germany), which detects mono- and using pUC44/F-4 as a template. The following primers were used (1): oligonucleosomes in cell lysates using mouse mAbs directed against DNA 5Ј-GCGCGGTACCCCGAATTC-3Ј containing a KpnI site at the 5Ј ter- Ј Ј Ј and histones, respectively. Briefly, cells were lysed, and the lysates were minus and 5 -TTTTTACCGCGGATGTCATAG-3 (2), 5 -AAAAATG placed into a streptavidin-coated plate. Subsequently, biotin-labeled anti- GCGCCTACAGTATC-3Ј, and 5Ј-CACTGAGTACCTAGGCTT-3Ј con- Ј histone mAb and peroxidase (POD)-labeled anti-DNA mAb were added taining a BamHI site at the 3 terminus (mutated nucleotides are and incubated for 2 h. After washing the wells to remove unbound Abs, we underlined). In the next reaction, products of the first PCR were used as quantified the amount of nucleosomes derived from apoptotic cells. This primers, and pUC44/F-4 was used as the template. The amplified KpnI- 23 represented the amount of POD-anti-DNA mAb bound to the DNA com- BamHI fragment with the Arg Ala mutation was then exchanged with ponent of the nucleosomes. POD was determined with its substrate ABTS corresponding KpnI-BamHI fragment in 44/F-4. All constructs were veri- (2, 2Ј-azino-di-[3-ethylbenzthiazoline-6-sulfonic acid] by photometric fied by DNA sequencing. analysis measured at 405 nm. Cells Fluorescein-HA binding assay A murine thymoma cell line AKR1 (28) was maintained in DMEM con- Cells were incubated with or without 40 ␮g/ml KM201 for 20 min at 4°C. taining 10% FCS (Dainippon Pharmaceutical, Osaka, Japan), 10 mM After incubation, cells were washed twice with PBS and incubated with 50 by guest on October 1, 2021. Copyright 1999 Pageant Media Ltd. HEPES, 2 mM L-glutamine, 1 mM sodium pyruvate, 1% (v/v) 100ϫ non- ␮g/ml fluorescein-conjugated HA (Seikagaku Kogyo, Tokyo, Japan) dis- essential amino acids, 100 U/ml penicillin, and 100 ␮g/ml streptomycin solved in PBS for 20 min at 4°C. Cells were then washed and analyzed by (complete medium). To prepare clones that express CD44 or CD44-Fas flow cytometry on an EPICS XL flow cytometer (Coulter). fusion protein, the recombinant plasmid together with pSV2bsr carrying blasticidin S resistance gene were introduced into AKR1 cells by electro- T cell proliferation assays poration. Transfectants were selected in complete medium containing 10 Proliferation assays were conducted as previously described (34). Briefly, ␮g/ml blasticidin S (Funakoshi, Tokyo, Japan), and drug-resistant cells Ϫ Ϫ were cloned by limiting dilution. The surface expression of CD44 or surface Ig and Ia T cells were obtained from lymph node of BALB/c mice by immunomagnetic negative selection. The purity of the resulting CD44-Fas fusion protein in each clone was examined by indirect immu- Ͼ ϩ nofluorescence staining with anti-mouse CD44 mAb IRAWB14.4 (9, 29) cell population as assessed by flow cytometry was 97% CD3 . Each ϩ anti-CD44 mAb (5 ␮g/ml) was mixed with the anti-CD3 Ab (2.5 ␮g/ml)

http://classic.jimmunol.org and FITC-conjugated goat anti-rat (IgG IgM) (Southern Biotechnology Associates, Birmingham, AL), followed by flow cytometry on an EPICS and immobilized onto 96-well microculture plates. T cells were cultured in ϫ 5 XL flow cytometer (Coulter, Hialeah, FL). complete medium at 1 10 cells/well for 2 days, and then harvested after an 8-h pulse with 0.5 ␮Ci of tritium-labeled thymidine. Monoclonal Abs Results Table I summarizes the anti-CD44 mAbs used in this study and their spec- Construction of CD44-Fas chimeric proteins and their ificities. Rat anti-mouse CD44 mAbs IRAWB14.4, KM201, KM703 (6), expression in AKR1 cell line Downloaded from and IM7.8.1 (30) and anti-mouse MAdCAM-1 mAb MECA-367 (IgG2a) (31) were purified by protein G-Sepharose (Pharmacia Biotech, Uppsala, In the first step we constructed plasmids encoding CD44 and the Sweden). RAMBM5.5.8 and R7 166.7 (32) were purified from hybridoma IC domain of Fas (Fig. 1A). In the 44/F-4 plasmid, the coding culture supernatant using the E-Z-SEP Ab purification kit (Pharmacia Bio- sequence of CD44 including EC, TM, and nearly the entire IC tech). Purified RAWB45.106.2 and KM114 were provided by Dr. Jayne Lesley (The Salk Institute, La Jolla, CA) and Dr. Kensuke Miyake (Saga domains (aa 1–332) was fused to that of the IC domain of Fas. The University Medical School, Saga, Japan), respectively. Anti-CD3 mAb 44/F-4RA plasmid was identical with the 44/F-4 plasmid, except 145-2C11 (33) free of endotoxin was purchased from PharMingen (San that it had a single amino acid substitution of arginine 23 (amino Diego, CA). acid numbers used in this study refer to those of mature protein) to alanine. This substitution corresponds to the mutation at arginine Induction and detection of apoptosis 41 in human CD44, which destroys the B(X7)B motif in the HA For induction of apoptosis, 1 ϫ 104 cells were seeded into each well of a binding domain and thus abrogates the ability of CD44 to recog- 96-well microtiter plate in complete medium containing 5 ␮g/ml anti- nize HA (27). These plasmids or plasmid encoding the wild-type CD44 mAb with or without 5 ␮g/ml goat anti-rat (IgGϩIgM) (Southern Biotechnology Associates) and incubated for 5 h. As a control for Ab CD44 were transfected into CD44-negative murine thymoma treatment, cells were incubated in complete medium containing 5 ␮g/ml AKR1 cells, and multiple clones expressing the corresponding pro- MECA-367. To examine the effect of KM201 or KM114, cells were pre- teins, namely, wild-type CD44 (A-CD44WT), 44/F-4 (A44/F-4), 1260 REQUIREMENTS FOR SIGNAL DELIVERY IN CD44

FIGURE 2. Ligation of CD44-Fas with HA induces apoptosis. A44/F-4 transfectant cells were cultured for 16 h with (ϩHA) or without (ϪHA) 100 ␮g/ml HA in the presence of mAb IRAWB14.4, which enhances HA binding (9, 29). Apoptosis was detected by the amount of nucleosomes in the cell lysates as described in Materials and Methods.

whether the cells underwent apoptosis. The isotype and epitope specificity of each mAb are shown in Table I. Preliminary exper- iments indicated that these mAbs bound to both the wild-type CD44 and CD44-Fas fusion proteins equally well (data not shown) and thus were considered to provide a suitable means to stimulate transfectant cells to induce apoptosis. As shown in Fig. 3, six of the eight anti-CD44 mAbs tested induced apoptosis of A44/F-4 cells but not A-CD44WT cells, indicating that the apoptotic effect was specifically mediated by the chimeric CD44-Fas protein. Introduc- tion of a stop codon in the coding sequence of the death domain of Fas abrogated apoptosis (data not shown), supporting the idea that the CD44-Fas chimeric protein transduced apoptosis through Fas upon ligation of the CD44 EC domain. Among the six anti-CD44 mAbs capable of inducing apoptosis, IRAWB14.4 (IgG2a) and R7 166.7 (IgM) have been reported FIGURE 1. Schematic diagram illustrating the generation of CD44-Fas to enhance the HA-binding ability of CD44 (9, 29), but chimeric proteins. A, Construction of recombinant plasmids. Wild-type RAWB45.106.2 (IgM), KM703 (IgG2a), and RAMBM5.5.8 (IgM) mouse CD44 (CD44WT), wild-type human Fas (FasWT), and each of the by guest on October 1, 2021. Copyright 1999 Pageant Media Ltd. chimeric constructs used in this study (44/F-4 and 44/F-4RA) are shown. apparently have no such effect (29) (also see Table I), indicating Numbers above and below each construct refer to the amino acid position of the mouse CD44 and human Fas, respectively. Each construct was cloned into the pEF-BOS vector to generate expression plasmids. B, Ex- pression of CD44-Fas fusion protein in AKR1 transfectants. Expression levels of AKR1 transfectants with wild-type CD44 (A-CD44WT), CD44- Fas chimera (A44/F-4), and CD44-Fas chimera with a point mutation (A44/F-4RA) are shown. Expression of CD44 was examined with anti- CD44 mAb by flow cytometry. http://classic.jimmunol.org

or 44/F-4RA (A44/F-4RA) at high levels were obtained. Repre- sentative results are shown in Fig. 1B.

Induction of apoptosis with anti-CD44 mAbs recognizing various epitopes Downloaded from We have previously reported that a chimeric adhesion molecule, L-selectin-Fas generated as described above, transduces apoptotic signals in cells upon ligation of a functional domain of the EC region with mAb (23). Similarly, as shown in Fig. 2, ligation of the EC domain of CD44-Fas with HA in conjunction with stimulation with a low dose (5 ng/ml) of anti-CD44 mAb IRAWB14.4 that enhances HA binding (9, 29) induced apoptosis in cells, indicating FIGURE 3. Only certain anti-CD44 mAbs can induce apoptosis in that the CD44-Fas protein can also transduce signals upon appro- CD44-Fas-expressing cells. A wild-type CD44 transfectant (A-CD44WT) and a CD44-Fas transfectant (A44/F-4) were incubated overnight with 5 priate stimulation. Therefore, we decided to use this strategy to ␮g/ml of the indicated anti-CD44 mAb or a control Ab (MECA-367) in the investigate the functional sites involved in signal transduction in presence or the absence of 5 ␮g/ml of goat anti-rat (IgG ϩ IgM) as sec- the EC region of CD44. To this end, AKR1 cells expressing the ondary Ab. Apotosis was quantitated as described in Materials and Meth- 44/F-4 chimeric protein (A44/F-4) or the wild-type CD44 (A- ods. The levels of apoptosis are expressed as a percentage of the corre- CD44WT) were treated with various anti-mouse CD44 mAbs with sponding activity obtained with cells treated with mAb IRAWB14.4. The different epitope specificities (29) and then examined to determine data represent the mean Ϯ SD from three separate experiments. The Journal of Immunology 1261

FIGURE 4. Blocking mAb KM201 and KM114 inhibit KM703- or R7 166.7-induced apoptosis. AKR1 transfectants expressing CD44-Fas (A44/ F-4) were pretreated with KM201 or KM114, and apoptosis was subse- quently detected after treatment with 5 ␮g/ml KM703 (overnight) or with R7 166.7 (5 h). The data are presented as described in Fig. 3. They are from a representative experiment, with similar results obtained in two additional experiments. Note that KM201 and KM114 inhibited KM703- or R7 166.7-induced apoptosis.

that the HA binding-enhancing effect was not a major factor for induction of apoptosis in CD44-Fas chimera-expressing cells. With RAWB45.106.2, KM703, and RAMBM5.5.8, apoptosis was enhanced by further ligation with a secondary Ab (goat anti-mouse Ig), suggesting that the efficiency of apoptosis might be determined in part by the extent of cross-linking of CD44-Fas by these mAbs. Interestingly, two groups of mAbs, one recognizing the epitope group d (IRAWB14.4, RAWB45.106.2) and the other recognizing FIGURE 5. A mutant CD44-Fas is unable to bind HA and to transduce the epitope group h (KM703, RAMBM 5.5.8, R7 166.7, IM7.8.1), apoptotic signal. A, Binding of fluorescein (FL)-HA to transfectants ex- were uniformly capable of inducing apoptosis, and all mAbs ex- pressing wild-type CD44-Fas or mutant CD44-Fas. A-CD44WT, A44/F-4, or A44/F-4RA cells preincubated with KM201 (dotted line) or without

by guest on October 1, 2021. Copyright 1999 Pageant Media Ltd. cept IM7.8.1 in these groups induced apoptosis with the primary mAb alone. IM7.8.1 also induced apoptosis once sufficient cross- KM201 (solid line) were incubated with 50 ␮g/ml fluorescein-conjugated linking was established with the secondary Ab. Whereas the HA (FL-HA), and analyzed by flow cytometry. B, A-CD44WT, A44/F-4, ␮ epitope group d is thought to be involved in ligand binding, epitope or A44/F-4RA cells were incubated with 5 g/ml of indicated Abs. The data are presented as described in Fig. 3. The data represent the mean Ϯ SD group h is not involved in HA binding (29). In contrast, both from three separate experiments. blocking mAbs KM201 (IgG1) and KM114 (IgG1) that block HA binding failed to induce apoptosis with the primary mAb alone and induced little apoptosis, if any, with the secondary Ab. Both mAbs main of CD44 is involved in the signal generation. Therefore, we recognize a particular region (epitope group f) of the CD44 link expressed in AKR1 cells a mutant CD44/Fas protein, 44/F-4RA, 23 http://classic.jimmunol.org module and thus block the binding to HA, although neither of them which bears the Arg Ala mutation in the link domain. This mu- directly binds to the B(X7)B motif critical for HA binding (29). tation has been reported previously to completely abrogate HA binding of CD44 (27). As shown in Fig. 5A, 44/F-4RA-expressing Inhibition of mAb-induced apoptosis via CD44-Fas by blocking cells (A44/F-4RA) failed to show HA binding, whereas 44/F-4- mAbs against CD44 expressing cells (A44/F-4) bound HA, albeit at slightly lower lev- Although neither KM201 nor KM114 was capable of inducing els than AKR1 cells that expressed the wild-type CD44 (A-

Downloaded from apoptosis through the CD44-Fas chimeric protein despite the fact CD44WT). We then examined whether the mutant 44/F-4RA that these mAbs are known to recognize the functional domain of protein could transduce apoptotic signal upon cross-linking with CD44 (29), it is of note that these mAbs were instead capable of mAb. As shown in Fig. 5B, the 44/F-4RA chimeric protein failed inhibiting apoptosis induced by nonblocking anti-CD44 mAbs. As to mediate mAb-induced apoptosis with any of the mAbs tested, shown in Fig. 4, KM201 inhibited KM703-induced apoptosis of even in the presence of a secondary Ab (data not shown), suggest- A44/F-4 cells by ϳ70%, while R7 166.7-induced apoptosis by ing that the functional integrity of the HA binding domain of ϳ80%, although KM201 did not inhibit binding of mAb KM703 CD44-Fas is crucial to the transduction of apoptotic signal into or R7 166.7 to the chimeric CD44-Fas (data not shown). Similar cells. inhibitory effects were observed with another blocking mAb KM114. Stimulation through epitope h of CD44 induces costimulation of T cells Failure of a mutant CD44-Fas protein defective in hyaluronate The above results of the mapping of functional epitopes obtained binding to transduce mAb-induced apoptotic signal by the use of CD44-Fas chimeric proteins indicates that epitope h, Inhibition of apoptosis by a HA-binding blocking mAb KM201 or which has been thought to be uninvolved in ligand binding (29), KM114 observed above (Fig. 4) suggests that the HA binding do- actually plays a role in signal generation. To examine whether this 1262 REQUIREMENTS FOR SIGNAL DELIVERY IN CD44

tor with mAbs that recognize epitope group d or h induced apo- ptosis, whereas mAbs against epitope group f did not, even in the presence of a secondary Ab. By contrast, ligation of wild-type CD44 with any of the mAbs tested did not induce apoptosis. These results suggest that stimulation through a specific site of CD44 is crucial for the generation of a signal and that at least two epitope groups (d and h) in the CD44 EC domain contribute to signal transduction. The epitope group d is recognized by mAbs IRAWB14.4 and RAWB45.106.2 and spans eight amino acid residues (Asp46 to Ser53), among which Lys49 is apparently most important for Ag specificity (29). Because one of the mAbs belonging to this group (IRAWB14.4) has a potent enhancing effect on ligand binding by CD44, this epitope group is thought to be important in regulating the adhesive function of CD44 (29). Indeed, mutation of the lysine residue of human CD44, which is equivalent to Lys49 of the mu- rine counterpart, has been shown to abrogate HA binding (37). Our FIGURE 6. Costimulation of naive T cells via various CD44 epitopes. observation that both IRAWB14.4 and RAWB45.106.2 generated Purified T cells (1 ϫ 105 cells/well) were cultured for 2 days in 96-well apoptotic signal via CD44-Fas further indicates the importance of flat-bottom microculture plates that had been coated with a suboptimal dose of anti-CD3 (2.5 ␮g/well) together with CD28 or anti-CD44 (5 ␮g/ this epitope group. The finding that RAWB45.106.2, which does well) reactive with different epitopes. Tritium-labeled thymidine not enhance HA binding (29), also generated apoptotic signal sug- ([3H]dThd) incorporation is expressed by the mean Ϯ SD of triplicate gests that the ligand binding-enhancing effect of mAbs does not cultures. Data are representative of three similar experiments. necessarily correlate with their ability to generate a signal in CD44. Epitope group h is dependent on two noncontiguous residues, namely Pro108 and Thr113, which are present in two neighboring observation can be extended to the intact CD44 molecule, we regions interrupted by a disulfide bond in the link domain of mu- tested whether stimulation through this epitope leads to T cell rine CD44 (29). Many of the mAbs recognizing this region cross- costimulation. Purified T cells were cultured for 2 days in wells react with human CD44, probably because Pro108 is shared by containing either of the anti-CD44 mAb coimmobilized with a mouse and human CD44 (29). Because most of these mAbs do not suboptimal dose of anti-CD3 mAb in the absence of APC and inhibit hyaluronate recognition by CD44 expressed on the cell sur- examined for cell proliferation. As shown in Fig. 6, three of face, it has been suggested that epitope group h is not directly the four mAbs against epitope group h (IM7, R7 166.7, involved in ligand binding (29). However, in the present study RAMBM5.5.8) induced strong proliferation of TCR-stimulated all four mAbs in this group (R7 166.7, KM703, IM7.8.1, naive T cells, similar to anti-CD28. KM703 recognizing the same by guest on October 1, 2021. Copyright 1999 Pageant Media Ltd. RAMBM5.5) generated apoptotic signal in CD44-Fas-expressing epitope group induced much lower, but significant, T cell costimu- cells in the presence of a secondary Ab, and all but one (IM7.8.1) lation. mAbs KM201 and KM114 against epitope group f also induced apoptosis by themselves. Furthermore, stimulation of na- induced T cell costimulation in agreement with the previous report ive T cells through this epitope induced potent costimulation (Fig. (35). These results are in agreement with those obtained with the 6). These findings indicate that this epitope actually plays a role in chimeric CD44-Fas proteins, in that epitope h plays a role in signal signal generation. In agreement with this idea, IM7.8.1 has actu- delivery. ally been shown to induce shedding of CD44 (38, 39) and inhibit induction of murine arthritis (39). In addition, preliminary results Discussion from our laboratories have shown that at least three of the mAbs in http://classic.jimmunol.org To examine the functional site(s) in the EC domain of CD44, we this group (R7 166.7, KM703, RAMBM5.5.8) can inhibit the bind- constructed the CD44-Fas chimeric protein to detect the CD44- ing of soluble CD44-Ig chimera to hyaluronate (unpublished ob- transduced signal for apoptosis of the chimera-expressing cells. servations), confirming the importance of this epitope group in Fas was fused to almost the very end of CD44 IC domain so as not ligand binding. That IM7.8.1 induced apoptosis only in the pres- to disrupt the association between CD44 and intracellular proteins. ence of a secondary Ab but not by itself could mean that IM7.8.1 We have previously used a similar strategy to study the functional differs in fine specificity from other mAbs of the same group. This

Downloaded from role of various domains of L-selectin and found that cross-linking conclusion was recently confirmed in a series of preliminary stud- of the lectin domain in the EC region was essential for signal ies in our laboratory showing that IM7.8.1 and KM703 cross- generation and that mere cross-linking did not lead to signal trans- blocked each other only weakly (unpublished observations). On duction in L-selectin (23). the other hand, it is also possible that IM7.8.1 is unable to induce The link domain of CD44, with a three-dimensional structure a sufficient level of CD44 clustering unless further cross-linked by similar to the lectin domain of selectins (36), possesses at least five a secondary Ab. Fas is known to exert its cytocidal effect by form- different epitope groups (d, e, f, g, and h) identified by the use of ing a homotrimer (40), and it is expected that sufficient oligomer- a panel of anti-CD44 mAbs (29), although the functional signifi- ization or cross-linkage of CD44-Fas is necessary for generation of cance of these groups remains to be fully explored. In this study, the apoptotic signal. Studies by others indicate that receptor oli- we used CD44-Fas chimeric proteins to examine whether any EC gomerization can augment ligand recognition in CD44 (1, 9, 41). domain(s) of CD44 is involved in signal generation. To stimulate Nevertheless, it is evident that a mere Ab-induced receptor oli- a particular site of the CD44 EC domain, we used eight anti-CD44 gomerization is insufficient for signal generation, because mAbs of mAbs recognizing three different epitope groups (d, f, and h). Our epitope f (KM201, KM114) were completely unable to induce apo- results showed that although these mAbs bound to CD44 and ptosis even in the presence of a secondary Ab as discussed below. CD44-Fas equally well, ligation of the CD44-Fas chimeric recep- Clement and Stamenkovic (42) have similarly observed, with The Journal of Immunology 1263

CD40 fused to Fas or TNF receptor, that only appropriate ligation 4. Goldstein, L. A., D. F. Zhou, L. J. Picker, C. N. Minty, R. F. Bargatze, via a functional site is critical for receptor signaling. J. F. Ding,and E. C. Butcher. 1989. A human lymphocyte homing receptor, the hermes antigen, is related to cartilage core and link proteins. Cell 83 90 91 Epitope group f is dependent on His or Val /Thr (29) and 56:1063. recognized by hyaluronate blocking mAbs KM201 and KM114. 5. Yang, B., B. L. Yang, R. C. Savani, and E. A. Turley. 1994. Identification of a common hyaluronan binding motif in the hyaluronan binding proteins RHAMM, Because this region is distantly located from the B(X7)B motifs CD44 and link protein. EMBO J. 13:286. known to be critical for HA recognition on the secondary structure 6. Miyake, K., K. L. Medina, S. Hayashi, S. Ono, T. Hamaoka, and P. W. Kincade. (36), one possibility is that these blocking mAbs exert their effect 1990. Monoclonal antibodies to Pgp-1/CD44 block lympho-hemopoiesis in long- term bone marrow cultures. J. Exp. Med. 171:477. not by blocking the ligand binding site but by altering conforma- 7. Miyake, K., C. B. Underhill, J. Lesley, and P. W. Kincade. 1990. Hyaluronate can tion of CD44 so as to make hyaluronate recognition difficult, al- function as a cell adhesion molecule and CD44 participates in hyaluronate rec- though further study is clearly required to verify this point. Sub- ognition. J. Exp. Med. 172:69. 83 8. Stamenkovic, I., A. Aruffo, M. Amiot, and B. Seed. 1991. The hematopoietic and stitution of a residue equivalent to His of mouse CD44 to alanine epithelial forms of CD44 are distinct polypeptides with different adhesion poten- in human CD44 has been shown to abrogate HA binding (37). An tials for hyaluronate-bearing cells. EMBO J. 10:343. unexpected observation in our study with mAbs recognizing this 9. Lesley, J., Q. He, K. Miyake, A. Hamann, R. Hyman, and P. W. Kincade. 1992. Requirements for hyaluronic acid binding by CD44: a role for the cytoplasmic epitope group was that KM201 and KM114 blocking mAbs inhib- domain and activation by antibody. J. Exp. Med 175:257. ited mAb-induced apoptosis by mAbs KM703 and R7 166.7 (both 10. Aruffo, A., I. Stamenkovic, M. Melnick, C. B. Underhill, and B. Seed. 1990. belong to group h), while they themselves failed to induce apo- CD44 is the principal for hyaluronate. Cell 61:1303. 11. Jalkanen, S., and M. Jalkanen. 1992. Lymphocyte CD44 binds the COOH-ter- ptosis. Because neither KM201 nor KM114 inhibited binding of minal heparin-binding domain of fibronectin. J. Cell Biol. 116:817. KM703 to any extent (unpublished observations), this suppressive 12. Carter, W. G., and E. A. Wayner. 1988. Characterization of the class III collagen effect by KM201 or KM114 is certainly not attributable to inhibi- receptor, a phosphorylated, transmembrane glycoprotein expressed in nucleated human cells. J. Biol. Chem. 263:4193. tion of mAb binding. Rather, it may be that the ability to recognize 13. Toyama-Sorimachi, N., H. Sorimachi, Y. Tobita, F. Kitamura, H. Yagita, hyaluronate has to be retained by CD44 so as to generate the sig- K. Suzuki, and M. Miyasaka. 1995. A novel ligand for CD44 is serglycin, a nal. This is supported by our finding that the Arg23Ala CD44-Fas hematopoietic cell lineage-specific proteoglycan: possible involvement in lym- phoid cell adherence and activation. J. Biol. Chem. 270:7437. mutant without the ability to bind hyaluronate was completely un- 14. Toyama-Sorimachi, N., F. Kitamura, H. Habuchi, Y. Tobita, K. Kimata, and able to transduce apoptotic signal. Therefore, the functional integ- M. Miyasaka. 1997. Widespread expression of chondroitin sulfate type serglycins rity of the hyaluronate binding region may be critical for signal with CD44 binding ability in hemopoietic cells. J. Biol. Chem. 272:26714. 15. Weber, G. F., S. Ashkar, M. J. Glimcher, and H. Cantor. 1996. Receptor-ligand transduction by CD44. Clement and Stamenkovic (42) also pre- interaction between CD44 and osteopontin (Eta-1). Science 271:509. pared a chimeric human CD44-Fas molecule consisting of CD44 16. Naujokas, M. F., M. Morin, M. S. Anderson, M. Peterson, and J. Miller. 1993. The condroitin sulfate form of invariant chain can enhance stimulation of T cell EC, Fas TM, and Fas IC domains, but were unsuccessful in gen- responses through interaction with CD44. Cell 74:257. erating apoptotic signal by mAb-induced cross-linking, although 17. Shimizu, Y., G. A. Van Seventer, R. Siraganian, L. Wahl, and S. Shaw. 1989. the specificity of the mAbs was not mentioned in their study. Fail- Dual role of the CD44 molecule in T cell adhesion and activation. J. Immunol. 143:2457. ure of generation of the apoptotic signal in their study might have 18. Denning, S. M., P. T. Le, K. H. Singer, and B. F. Haynes. 1990. Antibodies been due to the use of mAb blocking HA recognition. against the CD44 p80, lymphocyte homing receptor molecule augment human Other than the three epitope groups examined in the present peripheral blood T cell activation. J. Immunol. 144:7. 19. Huet, S., H. Groux, B. Caillou, H. Valentin, A. M. Prieur, and A. Bernard. 1989. study, there is also another interesting site in CD44 that would CD44 contributes to T cell activation. J. Immunol. 143:798.

by guest on October 1, 2021. Copyright 1999 Pageant Media Ltd. merit functional analysis if appropriate mAb were available. It is 20. Galandrini, R., R. De Maria, M. Riccoli, L. Frati, and A. Santoni. 1994. CD44 an epitope recognized by mAb Hermes-3 (43) that can inhibit lym- triggering enhances human NK cell cytotoxic functions. J. Immunol. 153:4399. 21. Tan, P. H. S., E. B. Santos, H.-C. Rossbach, and B. M. Sandmaier. 1993. En- phocyte adhesion to high endothelial venules in mucosal lymphoid hancement of natural killer activity by an antibody to CD44. J. Immunol. 150: tissues in the absence of blocking ability to hyaluronate binding 812. (44). However, no anti-mouse mAb against this epitope is avail- 22. Webb, D. S., Y. Shimizu, G. A. Van Seventer, S. Shaw, and T. L. Gerrard. 1990. LFA-3, CD44, and CD45: physiologic triggers of human monocyte TNF and IL-1 able at present, and therefore, the functional importance of this release. Science 249:1295. epitope remains to be determined. 23. Ishiwatari-Hayasaka, H., H. Kawashima, T. Osawa, S. Nagata, and M. Miyasaka. Finally, the present study reiterates the functional importance of 1997. Induction of cell death by chimeric L-selectin-Fas receptors. Int. Immunol. 9:627. the link domain of CD44 in receptor signaling and supports the 24. Zhou, D. F., J. F. Ding, L. J. Picker, R. F. Bargatze, E. C. Butcher, and http://classic.jimmunol.org idea that multiple regions in the link domain must function in D. V. Goeddel. 1989. Molecular cloning and expression of Pgp-1: the mouse concert for an effective ligand binding and signal generation. Our homolog of the human H-CAM (Hermes) lymphocyte homing receptor. J. Im- munol. 143:3390. results also extend those of previous studies by showing that a site 25. Mizushima, S., and S. Nagata. 1990. pEF-BOS, a powerful mammalian expres- apparently uninvolved in ligand binding (epitope group h) is also sion vector. Nucleic Acids Res. 18:5322. important in signal generation. Elucidation of the role of this re- 26. Itoh, N., S. Yonehara, A. Ishii, M. Yonehara, S. Mizushima, M. Sameshima, A. Hase, Y. Seto, and S. Nagata. 1991. The polypeptide encoded by the cDNA gion will enhance our understanding of the complex regulatory for human cell surface antigen Fas can mediate apoptosis. Cell 66:233.

Downloaded from mechanisms of ligand binding and signal transduction in CD44. 27. Peach, R. J., D. Hollenbaugh, I. Stamenkovic, and A. Aruffo. 1993. Identification of hyaluronic acid binding sites in the extracellular domain of CD44. J. Cell Biol. 122:257. Acknowledgments 28. Hyman, R., K. Cunningham, and V. Stallings. 1980. Evidence for a genetic basis for the class A Thy-1Ϫ defect. Immunogenetics 10:261. We thank Drs. Jayne Lesley and Robert Hyman (The Salk Institute, La 29. Zheng, Z., S. Katoh, Q. He, K. Oritani, K. Miyake, J. Lesley, R. Hyman, Jolla, CA) and Kensuke Miyake (Saga University Medical School, Saga, A. Hamik, R. M. Parkhouse, A. G. Farr, et al. 1995. Monoclonal antibodies to Japan) for providing anti-CD44 Abs. We also thank Dr. T. Tanaka (Osaka CD44 and their influence on hyaluronan recognition. J. Cell Biol. 130:485. University Graduate School of Medicine, Suita, Japan) for stimulating 30. Trowbridge, I. S., J. Lesley, R. Schulte, R. Hyman, and J. Trotter. 1982. Bio- chemical characterization and cellular distribution of a polymorphic, murine cell- discussions. surface glycoprotein expressed on lymphoid tissues. Immunogenetics 15:299. 31. Streeter, P. R., E. L. Berg, B. T. Rouse, R. F. Bargatze, and E. C. Butcher. 1988. References A tissue-specific endothelial cell molecule involved in lymphocyte homing. Na- ture 331:41. 1. Lesley, J., R. Hyman, and P. W. Kincade. 1993. CD44 and its interaction with 32. Lesley, J., R. Schulte, and R. Hyman. 1990. Binding of hyaluronic acid to lym- extracellular matrix. Adv. Immunol. 54:271. phoid cell lines is inhibited by monoclonal antibodies against Pgp-1. Exp. Cell 2. Naor, D., R. V. Sionov, and D. Ish-Shalom. 1997. CD44: structure, function, and Res 187:224. association with the malignant process. Adv. Cancer Res. 71:241. 33. Leo, O., M. Foo, D. H. Sachs, L. E. Samelson, and J. A. Bluestone. 1987. Iden- 3. DeGrendele, H. C., P. Estess, and M. H. Siegelman. 1997. Requirement for CD44 tification of a monoclonal antibody specific for a murine T3 polypeptide. Proc. in activated T cell extravasation into an inflammatory site. Science 278:672. Natl. Acad. Sci. USA 84:1374. 1264 REQUIREMENTS FOR SIGNAL DELIVERY IN CD44

34. Yashiro, Y., X.-G. Tai, K. Toyo-oka, C.-S. Park, R. Abe, T. Hamaoka, 39. Mikecz, K., F. R. Brennan, J. H. Kim, and T. T. Glant. 1995. Anti-CD44 treat- M. Kobayashi, S. Neben, and H. Fujiwara. 1998. A fundamental difference in the ment abrogates tissue oedema and leukocyte infiltration in murine arthritis. Nat. capacity to induce proliferation of native T cells between CD28 and other co- Med. 1:558. stimulatory molecules. Eur. J. Immunol. 28:926. 40. Takahashi, T., M. Tanaka, J. Ogasawara, T. Suda, H. Murakami, and S. Nagata. 35. Ilangumaran, S., A. Briol, and D. C. Hoessli. 1998. CD44 selectively associates 1996. Swapping between Fas and granulocyte colony-stimulating factor receptor. with active src family protein tyrosine kinases lck and fyn in glycosphingolipid- J. Biol. Chem. 271:17555. rich plasma membrane domains of human peripheral blood lymphocytes. Blood 41. Lesley, J., P. W. Kincade, and R. Hyman. 1993. Antibody-induced activation of 91:3901. the hyaluronan receptor function of CD44 requires multivalent binding by anti- 36. Kohda, D., C. J. Morton, A. A. Parkar, H. Hatanaka, F. M. Inagaki, body. Eur. J. Immunol. 23:1902. I. D. Campbell, and A. J. Day. 1996. Solution structure of the link module: a 42. Clement, M. V., and I. Stamenkovic. 1994. Fas and tumor necrosis factor recep- hyaluronan-binding domain involved in extracellular matrix stability and cell tor-mediated cell death: similarities and distinctions. J. Exp. Med. 180:557. migration. Cell 86:767. 43. Jalkanen, S., R. F. Bargatze, J. De los Toyos, and E. C. Butcher. 1987. Lym- 37. Bajorath, J., B. Greenfield, S. B. Munro, A. J. Day, and A. Aruffo. 1998. Iden- phocyte recognition of high endothelium: antibodies to distinct epitopes of an tification of CD44 residues important for hyaluronan binding and delineation of 85–95-kD glycoprotein antigen differentially inhibit lymphocyte binding to the binding site. J. Biol. Chem. 273:338. lymph node, mucosal, or synovial endothelial cells. J. Cell Biol. 105:983. 38. Camp, R. L., A. Scheynius, C. Johansson, and E. Pure. 1993. CD44 is necessary 44. Culty, M., K. Miyake, P. W. Kincade, E. Sikorski, E. C. Butcher, C. Underhill, for optimal contact allergic responses but is not required for normal leukocyte and E. Silorski. 1990. The hyaluronate receptor is a member of the CD44 (H- extravasation. J. Exp. Med. 178:497. CAM) family of cell surface glycoproteins. J. Cell Biol. 111:2765. by guest on October 1, 2021. Copyright 1999 Pageant Media Ltd. http://classic.jimmunol.org Downloaded from