Kunitz-type protease inhibitor bikunin disrupts phorbol ester-induced oligomerization of CD44 variant isoforms containing epitope v9 and subsequently suppresses expression of urokinase-type plasminogen activator in human chondrosarcoma cells.

We previously found that bikunin (bik), a Kunitz-type protease inhibitor, suppresses phorbol ester (PMA)-stimulated expression of urokinase-type plasminogen activator (uPA). In the present study, we tried to answer this mechanism using human chondrosarcoma HCS-2/8 cells. Our results showed the following novel findings: (a) the standard form of CD44 (CD44s; 85 kDa) is expressed in both unstimulated and PMA-stimulated cells, while CD44v isoforms containing epitope v9 (110 kDa) are strongly up-regulated in response to treatment with PMA; (b) CD44v isoforms containing epitope v9 present on the same cell exclusively form aggregates in stimulated cells; (c) induction of uPA mRNA expression could be achieved by using a second cross-linker antibody to cross-link Fab monomers of anti-CD44; (d) co-treatment of stimulated cells with anti-CD44 mAb alone or anti-CD44v9 mAb alone suppresses PMA-induced clustering of CD44, which results in inhibition of uPA overexpression; (e) bikunin efficiently disrupts PMA-induced clustering of CD44, but does not prevent PMA-induced up-regulation of CD44v isoforms containing epitope v9; and (f) after exposure to bik, approximately 150-kDa band is mainly detected with immunoprecipitation and this band is shown to be a heterodimer composed of the 110-kDa v9-containing CD44v isoforms and a 45-kDa bik receptor (bik-R). In conclusion, we provide, for the first time, evidence that the bik-R can physically interact with the CD44v isoforms containing epitope v9 and function as a repressor to down-regulate PMA-stimulated uPA expression, at least in part, by preventing clustering of CD44v isoforms containing epitope v9.

Bikunin consists of two protease inhibitor domains of the Kunitz type (1). It is found in normal human serum and urine and is constitutively produced and secreted by hepatocytes (1). It is a secreted glycoprotein with a postulated role in protease inhibition. In serum, it occurs mainly in complex with other polypeptides as one of the three chains of inter-␣-inhibitor (I␣I) 1 and as one of the two chains of pre-␣-inhibitor (P␣I). The polypeptides in these proteins are covalently linked via the chondroitin 4-sulfate (C4S) chain of bik molecule. We have shown that bik, but not I␣I, is proposed as a main participant in inhibition of tumor cell invasion and metastasis possibly through both direct inhibition of cell-associated plasmin activity and suppression of uPA expression (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12). We have been studying the function and mechanism of bik on suppression of the invasive capacity of tumor cells.
During the search for novel proteins interacting with bik, we identified at least two types of cell-associated binding protein; the 40-kDa link protein (LP), one of the hyaluronic acid (HA)binding proteins, as a cell-associated bik-binding protein, and a 45-kDa specific receptor for bik (bik-R), which is a membraneassociated unidentified molecule, as a putative bik receptor (bik-R) (13)(14)(15)(16)(17). We postulate that LP is the principal extracellular matrix-binding protein and acts as ligand sink for bik. The nucleotide sequences of human LP cDNA predict a 40 -48-kDa polypeptide. The diverse roles proposed for LP can largely be understood as the manifestation of HA stabilization in the extracellular matrix (18,19).
To more precisely map the regions in the bik that can interact with LP and bik-R, various truncated proteins were tested in the solid-phase binding and ligand blot assays (15,17). We have established that LP can interact with the NH 2 -terminal domain of bik, while bik-R is able to recognize the C4S side chain of bik (15,17). Analysis of binding of native bik and deglycosylated bik to the cells showed that the low affinity binding site is LP and the high affinity site is bik-R (17). Our previous publications also indicated that binding of bik to its binding sites on the cell surface has been implicated in inhibition of protein kinase C (PKC) translocation and activation. More recently we reported that bik markedly suppresses the cell invasion possibly through negative regulation of MEK/ ERK/c-Jun-dependent mechanisms and subsequently suppression of uPA expression and that bik must bind to both of the bik-binding proteins (LP and bik-R) to effectively suppress uPA up-regulation (9,17,20,21). Therefore, bik forms membrane complexes with LP and bik-R, and initiates modulation of signal transduction, which results in bik-mediated suppression of cell invasiveness, suggesting that bik interacts with tumor cells as a negative modulator of the invasive cells.
CD44 is the major cell-surface receptor for HA (22). Since one of the bik-binding proteins is identical to LP, which is apparently held at the cell surface by HA (19), it is reasonable to think that CD44 would be involved in this bik-mediated complex at the cell surface. It has been established that CD44 is a polymorphic integral membrane glycoprotein with a postulated role in matrix adhesion (23), lymphocyte activation (24 -26), lymph node homing (27), and tumor invasion (28,29) and metastasis (30). CD44 plays an important role in the physiology of normal and tumor cells (31,32). CD44s is the basic unit of the CD44 protein. Other isoforms including CD44 variant forms are created by alternative splicing of the mRNA. Upon activation with PMA, it stimulates the dimerization of CD44 (33). Clustering of CD44 on the surface may be important for binding of HA (34) and presumably allows multiple copies of CD44 to interact with a signal molecule of HA. Thus, activation-induced clustering followed by dimerization and oligomerization of CD44 represents an additional signal transduction mechanism for regulating receptor-ligand interactions.
To support the theory that bik would associate with other protein factors in the PMA-stimulated membrane environment, we have examined whether the binding of bik to cells causes the formation of cross-linked CD44 similar to that of cross-linked bik-R. In the present study, we tried to answer the mechanism by which bik efficiently inhibits PMA-induced expression of uPA mRNA and protein using human chondrosarcoma HCS-2/8 cells, which express high levels of CD44 and bik-R on the cell surface. The present results allow us to hypothesize that interaction of bik-R with CD44 proteins is important for structural features that affect bik-dependent signal transduction, that suppression by bik of cell activation induced by PMA requires at least bik-R, and that bik inhibits co-stimulatory signals (i.e. uPA overexpression) delivered by CD44 clustering. It is likely that bik-R is a candidate for functional receptor for bik and may be an accessory receptor for CD44 proteins. Therefore, the present results show that engagement of bik-R by its ligand bik and subsequent coupling of bik-R to CD44 may facilitate inhibition of PMA stimulation by suppression of CD44 activation (that is CD44 clustering), which finally leads to reduction of uPA expression.

MATERIALS AND METHODS
Cells and Culture Conditions-Human chondrosarcoma cell line HCS-2/8 was grown and cultured as previously described (15,35). Briefly, the cells were harvested and aliquoted into tissue culture plates (2 ϫ 10 6 cells/well) in Dulbecco's minimum essential medium with Eagle's salts supplemented with penicillin (100 units/ml), streptomycin (100 g/ml), and 10% heat-inactivated fetal calf serum (Invitrogen, Rockville, MD). The next day, the cells were washed three times with phosphate-buffered saline (PBS) to remove serum, and the medium was replaced with Dulbecco's minimum essential medium supplemented with antibiotics. Serum-free medium plus the test drugs were added and incubation was continued for different time lapses. After culture, medium was aspirated and cells were harvested and washed extensively. In some experiments for flow cytometric analysis, at the end of the incubation, cells were then dissociated with 0.25% trypsin and 0.05 M EDTA solution supplemented with Streptomyces hyaluronidase (10 g/ml; see flow cytometry). Cells were used for measurement of CD44s and CD44v isoforms by flow cytometry, immunoblotting, and immunoprecipitation (36). In addition, medium and cell lysate were used for measurement of uPA (see below). In addition, human ovarian cancer cell line HOC-I (4) was used for further experiments.
Preparation of Antibodies-Rabbit polyclonal antibodies raised against bik (anti-bik pAb), bik-BPs (anti-bik-BP pAb), or LP (anti-LP pAb) were prepared in our laboratory (15). Anti-bik-BP pAb is a mixture of anti-LP pAb and anti-bik-R pAb. Anti-bik-R pAb was purified by LP-coupled-Sepharose 4B. Briefly, purified LP (50 mg) was coupled to CNBr-activated Sepharose 4B (15 g dry weight ϭ 50-ml bed volume; Amersham Biosciences AB, Uppsala, Sweden) according to the manufacturer's recommendations. To remove anti-LP pAb, anti-bik-BP pAb was mixed with LP-Sepharose beads using end to end rocking for 16 h at 4°C. Unbound material was recovered and stored. This step was repeated twice. The amount of protein in the soluble fraction was quantified in a Bradford assay (Bio-Rad) using bovine serum albumin as a standard (37). Anti-bik-R pAb and anti-LP pAb are specific for the respective proteins (see Fig. 1) and abrogate the suppression by bik of the PMA-induced uPA expression (data not shown).
Monoclonal antibodies against CD44 molecules (anti-CD44 mAb and anti-CD44v9 mAb) were obtained from Seikagaku Kogyo Co. Ltd., Tokyo. Anti-CD44 mAb recognizes the epitope involved in HA binding and anti-CD44v9 mAb recognizes the CD44v isoforms containing epitope v9 alone. Therefore, anti-CD44 mAb, but not anti-CD44v9 mAb, inhibits HA binding to the CD44 proteins on their cell surface. Of note that anti-CD44 mAb reacts with both standard form of CD44 (CD44s) and all types of CD44 variant isoforms. Monoclonal antibodies raised against CD44v isoforms containing epitope v3, v4, v5, and v6 were obtained from Novocastra Laboratories Ltd. (Benton Lane, Newcastle, United Kingdom). Rabbit anti-human CD44 polyclonal antibody (pAb) was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).
Cross-linking of Cell Surface Proteins with Disuccinimidyl Suberate (DSS)-The HCS-2/8 cells pretreated with or without 100 nM PMA for 6 h at 37°C were incubated in the presence or absence of bik (1 M) or its truncated proteins (deglycosylated bik and HI-8, carboxyl-terminal domain of bik; each 1 M), cross-linked using 1.5 mM disuccinimidyl suberate (DSS), washed, solubilized in lysis buffer, and subjected to immunoprecipitation with several antibodies. Briefly, cells (5 ϫ 10 7 cells) were pelleted by spinning at 1,000 rpm for 5 min. The cell pellets were washed twice with ice-cold PBS. Cells were treated with 1.5 mM DSS, which was dissolved immediately before use in PBS plus 10 mM Hepes at 23°C for 30 min. The reaction was followed by blocking with excess glycine and simultaneous treatment with 5 mM iodoacetamide for 10 min on ice to alkylate free sulfhydryl groups within the intact cells. The cells were then pelleted and resuspended in 0.5 ml of lysis buffer. The lysate was subjected to centrifugation at 100 ϫ g to pellet unbroken cells and nuclei and to recover the supernatant, which was centrifuged at high speed (40,000 ϫ g for 20 min at 4°C) to prepare the crude membrane fraction. For immunoprecipitation, the supernatant was precleared at 4°C with a 1:50 dilution of agarose-fixed goat antimouse IgG overnight. The solubilized proteins were immunoprecipitated with several antibodies at 1:50 dilution for 6 h and agitated gently for 6 h with a 1:50 dilution of agarose-fixed goat anti-mouse IgG. The beads were washed three times, incubated with a mixture of Streptomyces hyaluronidase and chondroitinase ABC (each 10 g/ml, 1 h, 37°C), resuspended in 2 ϫ sample buffer, heated for 10 min at 95°C, electrophoresed on SDS-polyacrylamide gel under nonreducing conditions, transferred to polyvinilidine difluoride membranes, and immunoblotted using ECL chemiluminescence (Amersham Japan, Tokyo). Western blotting of immunoprecipitates was essentially performed as described previously (39).
Cross-linking of CD44 Proteins on HCS-2/8 Cells-To assess the effects of CD44 cross-linking on the expression of multiple cell surface molecules, cells were incubated with intact antibodies, monomeric Fab antibody, or no first antibody and then washed. They were then incubated in the presence of a second cross-linker antibody (rabbit antimouse Ig or goat anti-rabbit Ig; 10 g/ml) (40,41) and analyzed.
Extraction and Analysis of Extracellular Matrix-associated Macromolecules-Serum-starved HCS-2/8 cells were incubated with or without bik (1 M). At the end of the incubation time, the medium was removed, and the cells were rinsed with PBS. They were harvested with a rubber policeman and then homogenized by hand in a glass homogenizer in PBS, pH 7.4, containing 2% Triton X-100, 10 mM EDTA, 2 mM benzamidine hydrochloride, 1 mM phenylmethylsulfonyl fluoride, and 0.15 mM pepstatin A. The homogenates were centrifuged at 8,000 ϫ g for 1 h. The supernatants were applied to a column (1 ϫ 25 cm) of Sepharose CL-4B previously equilibrated with PBS, pH 7.4, containing 0.5% Triton X-100, 10 mM EDTA, 2 mM benzamidine hydrochloride, and 1 mM phenylmethylsulfonyl fluoride. The excluded fractions from the column were further analyzed using a column (1 ϫ 25 cm) of Sepharose CL-2B. Cesium trifluoroacetate equilibrium centrifugation was performed at 150,000 ϫ g for 120 h at 10°C with a starting density of 1.42 g/ml as previously described (42). HA is known to be extremely large and excluded from a Sepharose CL-2B column and HA would appear at a density of about 1.43 g/ml. Therefore, macromolecules were obtained from the cell extract recovered at a density of 1.41-1.45 g/ml.
Flow Cytometric Analysis-Culture cells were harvested and washed with washing buffer (PBS supplemented with 2% bovine serum albumin and 0.1% NaN 3 , pH 7.4). Since anti-CD44 mAb does not effectively recognize HA-bound CD44 and the amount of HA present around the cells can greatly influence the binding of this antibody, cells were pretreated with hyaluronidase and used for further experiments. A single cell suspension (10 6 /ml) was incubated with affinity purified antibodies or isotope control antibodies on ice for 1 h. Cells were washed three times with washing buffer, and 2 l of the primary antibody and 3 l of fluorescein isothiocyanate-conjugated second antibody (Dako) were added for 1 h on ice, respectively. Cells were analyzed in a FACScan (Beckton Dickinson). At least 10,000 cells were analyzed per sample in all experiments. All experiments were performed at least twice.
Quantification of Urokinase-type Plasminogen Activator (uPA) by Enzyme-linked Immunosorbent Assay (ELISA)-uPA was quantified using a commercially available ELISA kit (IMUBIND) according to the manufacturer's instructions (American Diagnostica Inc. Greenwich, CT). The uPA ELISA kits and mAbs against uPA A-chain (number 4371) and B-chain (number 3689) were obtained from American Diagnostica Inc. (Dr. R. Hart). All quantifications were done in duplicate.
SDS-PAGE and Western Blot-The culture medium, cell lysate, and purified proteins were dissolved in a sample buffer in the presence or absence of dithiothreitol (DTT). The sample (20 g of protein/lane for cell extracts and 0.05-0.1 g of protein/lane for purified proteins) was processed for electrophoresis, using a 7.5 or 4 -15% gradient SDSpolyacrylamide gel under nonreducing conditions. The resulting gel was electrophoretically blotted onto polyvinylidine difluoride membrane, which was blocked in Tris-buffered saline (TBS) containing 2% bovine serum albumin, and then immunoblotted. The blot was subsequently processed for biotin-avidin-peroxidase method (43). Bands were visualized with the ECL detection system. Briefly, the polyvinylidine difluoride membranes were incubated precisely 20 s in a mixture of 5 ml of each of the ECL detection reagents. The membranes were then placed between two transparencies and exposed to Kodak film. In all experiments, some strips were incubated with non-immune rabbit (or mouse) IgG as a negative control. In the present study, the gel was blotted for bik-BPs with anti-bik-BP pAb, anti-LP pAb, or anti-bik-R pAb, respectively.
Northern Blot Analysis-Total RNA was isolated from cells by lysis in Trizol reagent according to the manufacturer's instructions (Invitrogen); 10 g of RNA were separated in 1.2% agarose gels and blotted onto Hybond N ϩ membranes. uPA mRNA was detected by a radioactively labeled uPA oligonucleotide probe. A 1.0-kilobase EcoRI-PstI fragment of a human uPA cDNA (44) was used as a probe in the hybridization experiments. uPA cDNA was labeled with [ 32 P]dCTP by the random primed DNA labeling technique as described (45). Following hybridization with uPA, blots were stripped and rehybridized with GAPDH as a semiquantitative control by densitometry. After each hybridization, the membranes were washed and exposed on Kodak BioMax MS-1 film at Ϫ70°C.
Statistical Analysis-The data presented are the mean of triplicate determinations in one representative experiment unless stated otherwise. Data are presented as mean Ϯ S.D. All statistical analysis was performed using StatView for Macintosh. The Mann-Whitney U test was used for the comparisons between different groups. p less than 0.05 was considered significant.

Characterization of Polyclonal Antibodies against bik-BPs,
LP, or bik-R-By procedures described previously (13)(14)(15), proteins of the bik-BP family were purified from human HCS-2/8 cell extracts by bik-coupled Sepharose 4B. It has been established that HCS-2/8 cell-derived bik-BPs are mainly composed of two different molecular species, the 45-kDa (bik-BP 45 , which corresponds to bik-R) and the 40-kDa (bik-BP 40 , which is identical to LP) (15). As shown in Fig. 1, anti-bik-BP pAb reacted with not only bik-BP 40 and bik-BP 45 but also purified LP. After immunoabsorption of anti-bik-BP pAb with LP, remaining antibodies (we termed these antibodies anti-bik-R pAb) recognized the bik-BP 45 but not the bik-BP 40 or LP. This shows that the 40-kDa band does not contain more than LP and that anti-bik-R pAb specifically recognize bik-BP 45 . The interaction of anti-LP pAb with bik-BPs was also evaluated by immunoblotting with bik-BP 45 and bik-BP 40 . Anti-LP pAb reacted with bik-BP 40 and purified LP. However, anti-LP pAb failed to react with bik-BP 45 . It is unlikely that the bik-BP 45 has antigenically cross-reactivity with LP. These results indicate that the bikbinding sites purified from HCS-2/8 cells contains at least the 45-kDa bik-R and the 40-kDa LP. These antibodies were used for futher experiments.

Flow Cytometric Analysis of Expression of CD44 Proteins on Unstimulated and Phorbol Ester-stimulated HCS-2/8 Cells-
Binding of mAbs directed against different CD44 variant isoforms was assessed by flow cytometry using the unstimulated and PMA-stimulated HCS-2/8 cells (Fig. 2). Flow cytometric analysis was performed using anti-CD44 mAb that binds to a HA-binding epitope common to all CD44 proteins as well as five anti-CD44v mAbs that selectively recognizes CD44v isoforms containing epitope v3, v4, v5, v6, or v9 on cells cultured in vitro for 6 h with and without 100 nM PMA. Nonimmune mouse IgG was used as the control IgG. Following in vitro culture of cells with PMA, there were 3-and 8-fold increases in cell surface expression of CD44, as defined by anti-CD44 mAb and anti-CD44v9 mAb, compared with unstimulated cells. Furthermore, HCS-2/8 cells were negative for CD44v isoforms containing epitope v3, v4, v5, and v6, irrespective of whether cells were treated with PMA.
In a separate experiment, we assessed whether bik can influence expression of CD44 on their cell surface by flow cytometry (Fig. 2). Cells were incubated with or without PMA (100 nM, 6 h) in the presence or absence of bik (1 M). Bikunin did not inhibit the expression of new CD44 proteins and CD44v isoforms containing epitopes v3, v4, v5, v6, and v9. The 110-, 155-, 195-, and 220-kDa bands and higher mass form (Ͼ250 kDa) were detected by anti-CD44v9 mAb. The 85-kDa band was detected by anti-CD44 mAb, but not by anti-CD44v9 mAb. These data indicate that the 85-kDa band corresponds in size to the standard form of CD44 (CD44s) and the 110-kDa and higher weight forms (155, 195, and 220 kDa) correspond to CD44 proteins containing CD44v isoforms having epitope v9. It is likely that the 170-kDa band detected in Western blot using anti-CD44 mAb corresponds in size to dimers of the 85-kDa CD44s.
To determine whether specific bands representing CD44v isoforms containing epitopes v3, v4, v5, and v6 were present, Western blot of cells cultured in vitro for 6 h and 24 h with 100 nM PMA were performed, using respective antibodies. However, specific bands were not identified in blots using these antibodies (data not shown). In similar experiments in which CD44 proteins from cells were analyzed using nonimmune mouse IgG, no significant signals were observed (data not shown).
It has been established that the presence of DTT abolished the clustering of CD44 proteins (33). To determine whether the protein structure imposed by the disulfide bridge contributes to oligomerization of CD44, we treated PMA-stimulated cells with a reducing agent DTT. Therefore, we examined the effect of reduction on the migration properties of CD44 proteins in SDS-PAGE (Fig. 3, C and D). In the presence of DTT, a pro-portion of CD44v isoforms containing epitope v9 from PMAstimulated cells migrated in Western blot analysis at the 110-kDa form normally detected under reducing conditions (Fig.  3D). In vitro culture for 6 h in PMA significantly induced the 110-kDa CD44v isoforms containing epitope v9. However, treatment with PMA did not significantly change expression of the 85-kDa CD44s (Fig. 3C). The 85-kDa monomeric form of CD44s in PMA-stimulated cells migrates similarly to that in unstimulated cells, independent of whether DTT is used in the experiments. No higher molecular weight forms were observed. Therefore, we consider that two higher mass bands observed at 220 kDa and Ͼ250 kDa under nonreducing conditions are dimers and oligomers of the 110-kDa CD44v isoforms containing epitope v9, respectively. These results suggest that CD44v isoforms containing epitope v9 are able to oligomerize under the PMA stimulation within at least 6 h, and then stimulates clustering of CD44s, a late event in this pathway as shown by the late time course of stimulation (more than 16 h). Of note, PMA markedly enhances the expression of uPA within at least 6 h (data not shown here; see Refs. 15,17,20,and 21). Therefore, we focused our attention on overexpression and dimerization of CD44v isoforms containing epitope v9 on the regulation of uPA expression.
Cross-linking of CD44 Proteins on Cells Induces Expression of uPA-We have already reported that PMA efficiently induce uPA expression and secretion in certain tumor cells containing HCS-2/8 cells (15,17,20,21). Our present data showed that PMA treatment of cells within a 6-h incubation markedly upregulates expression and oligomerization of CD44v isoforms containing epitope v9. Therefore, we speculate that PMA may stimulate the expression of uPA at least in part via dimerization and oligomerization of CD44 proteins. To confirm this theory, we examined whether the level of uPA expression would increase when oligomerization of CD44 proteins is induced with the multivalent antibodies.
We assessed the effects of cross-linking of CD44 proteins on the overexpression of uPA using a specific anti-CD44 mAb and a second cross-linker antibody (e.g. goat anti-mouse Ig) (Fig. 4,  A and B). A specific ELISA for uPA (A) and Northern blot analyses (B) showed that cross-linking of CD44 proteins with subsequent 6-h cultures significantly induced the expression of uPA mRNA (3-fold) and uPA protein (4-fold), respectively, in unstimulated cells in an antibody concentration-dependent manner. This suggests that cross-linking of CD44 proteins significantly stimulates uPA expression at the gene level and at the protein level.
We speculate that, since anti-CD44 pAb recognizes all CD44 proteins including CD44s, anti-CD44 pAb should give the same effect as shown in Fig. 4, A and B. We assessed the effects of cross-linking of CD44 proteins on the overexpression of uPA using a specific anti-CD44 pAb and a second cross-linker antibody (e.g. goat anti-rabbit Ig) (Fig. 4, C and D). As expected, cross-linking using anti-CD44 pAb also significantly induces expression of uPA mRNA and protein. Fig. 4 clearly showed that the addition of anti-CD44 mAb or anti-CD44 pAb and a second cross-linker antibody to the culture induced the expression of uPA mRNA and protein levels without PMA stimulation.
To confirm further the effect on uPA expression of multivalent interactions, cell-bound monomeric Fab was cross-linked by the addition of second cross-linker Ig. For this, papain digests were prepared from the anti-CD44 mAb. Sephacryl G75 gel filtration fractions of anti-CD44 Fab were assayed for cell binding activity and induction of uPA expression in the HCS-2/8 cells. Our experiments demonstrated that Fab monomer requires about 10-fold higher concentration to obtain binding to cells equivalent to intact mAb, while the monovalent Fab fragments alone do not induce uPA expression (Fig. 5A, lane 1). With 10 g/ml second cross-linker Ig, uPA expression was significantly induced (4-fold) by cross-linking of monovalent Fab fragments of anti-CD44 mAb (lane 3). Therefore, in addition to Fab monomer binding, cross-linking is required for uPA expression.
The ability of the antibody permutations that enhance uPA expression was shown to enhance oligomerization of CD44 proteins in Western blotting (Fig. 5B)

Anti-CD44v9 Antibody Specifically Inhibits PMA-induced
Oligomerization of CD44 Proteins-Oligomerization of CD44 proteins, followed by stimulation of uPA expression was induced with the Fab monomer followed by a second cross-linking antibody. These data raise the following question: what would happen to uPA expression when PMA-stimulated cells were treated with either intact anti-CD44 mAb alone or anti-CD44v9 mAb alone (without a second cross-linker antibody). We examined whether both antibodies can specifically inhibit PMA-stimulated oligomerization of CD44 proteins. Cells were incubated with PMA in the presence of anti-CD44v9 mAb (2, 10, and 50 g/ml, 6 h, 37°C) or nonimmune mouse IgG (50 g/ml, 6 h, 37°C). Western blots of the lysates were probed with biotinylated anti-CD44 mAb, which recognizes all species of the CD44 family. As shown in Fig. 6, left panel, in the presence of nonimmune mouse IgG, a proportion of CD44v isoforms containing epitope v9 from PMA-stimulated cells migrated at the Ͼ250-kDa polydisperse band (oligomerization of CD44v isoforms containing epitope v9), 220-kDa band (dimerization), in addition to the 110-kDa form (monomer) and the 85-kDa form (CD44s monomer). When cells were incubated with PMA plus anti-CD44v9 mAb for 6 h, signal of the 110-kDa band increased but intensity of the Ͼ250-kDa and 220-kDa bands decreased in the antibody concentration-dependent manner. However, anti-CD44v9 mAb did not produce a significant increase in the 85-kDa CD44s. Similar results were obtained with anti-CD44 mAb (data not shown).
As shown in Fig. 6, right panel, in vitro culture for 24 h in PMA markedly induced not only the 220-kDa dimers of v9containing CD44 isoforms but also the 170-kDa dimers of CD44s. In contrast, the addition of anti-CD44v9 mAb to cells significantly decreased the 220-kDa band and higher molecular mass complexes (Ͼ250 kDa), but induced increased expression of monomers of v9-containing CD44 isoforms (110 kDa). Thus, incubation of anti-CD44v9 mAb alone (without a second crosslinker antibody) with cells rather abrogate the ability of PMA to promote dimerization and oligomerization of CD44v isoforms containing epitope v9 in the antibody concentration-dependent manner.
The Effect of Anti-CD44v9 mAb on PMA-stimulated uPA Expression-We examined whether PMA-induced uPA release and its mRNA expression would be inhibited by the addition of anti-CD44v9 mAb (Fig. 7) or anti-CD44 mAb (not shown). The cells exposed to PMA (100 nM, 6 h, 37°C) exhibited about 5-6-fold increase in uPA level in the conditioned medium. As shown in Fig. 7A, when cells were incubated with anti-CD44v9 mAb (10 g/ml), the ability of PMA to stimulate the expression of uPA was inhibited about 50%. Higher concentrations of mAb CD44v9 (50 g/ml) gave similar results on inhibition of PMAdependent stimulation of uPA protein expression. Nonimmune mouse IgG (50 g/ml) did not significantly abrogate PMAinduced uPA expression. Thus, incubation of the cells with anti-CD44v9 mAb had a specifically suppressive effect on the ability of PMA to stimulate uPA expression.
In a parallel experiment, the effect of anti-CD44v9 mAb on the PMA-induced expression of uPA mRNA was studied. RNA was prepared from cells treated with PMA and antibody and hybridized with probes derived from human cDNA clones of uPA and GAPDH (Fig. 7B). We confirmed again that PMA produced a marked increase in uPA mRNA expression at the gene level. The expression of the uPA gene was increased by ϳ4.8-fold at 100 nM PMA for 6 h. This stimulation was abrogated by ϳ40% in cells simultaneously treated with anti-CD44v9 mAb. Furthermore, the level of uPA mRNA gene was decreased by ϳ50% at 1 M bik. These experiments support the hypothesis that oligomerization of CD44v isoforms containing epitope v9 is at least in part involved in a signaling cascade on PMA-dependent uPA expression. Judging from the results in Fig. 6, 50 g/ml anti-CD44v9 mAb almost disrupted the oligomerization. However, Fig. 7 showed only 50% inhibition of the uPA production. It suggests the presence of other signal pathways for uPA in HCS-2/8 cells.

The Effect of bik on PMA-induced CD44 Expression and Oligomerization: bik Inhibits CD44 Oligomerization, but Does Not Prevent PMA-induced up-regulation of CD44
Expression-We next assessed the biological activities of bik on the PMA-induced up-regulation of CD44 protein expression and oligomerization. We previously found that bik must bind to both LP and bik-R on the cell surface to suppress uPA upregulation (15,17,20,21). A deglycosylated form of bik, from which the C4S side chain has been removed, binds only to LP and fails to inhibit PMA-stimulated uPA expression. Furthermore, the carboxyl-terminal domain of bik, HI-8, binds to neither LP nor bik-R and also fails to inhibit PMA-induced uPA expression (20). The effects of native bik on the PMA-induced up-regulation of expression and oligomerization of CD44 proteins were compared with those of the bik and its truncated proteins. Cells were incubated for 6 h with 100 nM PMA supplemented with bik, deglycosylated bik, or HI-8 (each 1 M). Cell lysates prepared in SDS-PAGE sample buffer with DTT were subsequently lysed and immunoblotted with anti-CD44 mAb. As shown in Fig. 8A, lysates of PMA-stimulated cells contained the 85-kDa monomers of CD44s and the 110-kDa monomers of CD44v isoforms containing epitope v9 in the presence of DTT (lane 2). Expression of CD44v isoforms containing epitope v9 was significantly induced by the addition of PMA (lane 2). Both bik (lane 3) and deglycosylated bik (lane 4) failed to prevent PMA-induced up-regulation of CD44s and v9-containing CD44v isoforms at the protein level. HI-8 also failed to prevent PMA-induced up-regulation of CD44 proteins (data not shown).
We could not detect any bik-related complexes by Western blots using anti-CD44 mAb and anti-CD44v9 mAb, if no crosslinker is used in the experiments. Therefore, a cross-linker DSS was used to investigate whether bik influences PMA-induced clustering of CD44 proteins on the cell surface. The cells stimulated with or without PMA were biotinylated and then crosslinked with DSS. CD44 proteins were then immunoprecipitated with anti-CD44v9 mAb. The higher molecular mass immunocomplexes do not enter the gel, unless we include digestion with a mixture of hyaluronidase and chondroitinase ABC. Therefore, the immunocomplexes were preincubated for 1 h with a higher concentration of a mixture of hyaluronidase and chondroitinase ABC (each 10 g/ml) to completely remove HA and its fragments and then analyzed by SDS-PAGE followed by immunoprecipitation (Fig. 8B) followed by immunoblotting (Fig. 8, C-F). Fig. 8B shows that neither monomeric (110 kDa) nor dimeric (220 kDa) CD44v isoforms containing epitope v9 were detected on the unstimulated cells in the absence (lane 1) or presence of bik (lane 2). The 150-kDa band was detected in cells treated with bik. In contrast, significantly more 110-kDa CD44v isoforms was detected on PMA-treated cells (lane 3). In addition, both 220-kDa dimeric and higher molecular mass oligomeric CD44v isoforms containing epitope v9 (Ͼ250 kDa) were detected when the cells were stimulated with PMA (lane 3). The densitometric analysis revealed that high molecular weight v9-containing CD44 isoforms represents approximately twothirds of the total immunoreactive CD44 proteins. In PMAstimulated cells, the addition of bik had a significant effect on the electrophoretic mobility of the CD44 species visualized on immunoprecipitation using anti-CD44v9 mAb, where the ϳ150-kDa band was strongly detected (lane 4). However, monomers, dimers, and oligomers of CD44v isoforms containing epitope v9 were faintly observed (lane 4), demonstrating that bik could effectively block oligomerization of CD44v isoforms containing epitope v9. However, the addition of deglycosylated bik (lane 5) or HI-8 (lane 6) had no significant effect on the electrophoretic mobility of the CD44 species.
We next tested the effect of bik and its truncated proteins on the migration properties of CD44 proteins by immunoprecipitation followed by Western blot under nonreducing conditions (Fig. 8C). The cells were incubated with 100 nM PMA and 1 M bik (or deglycosylated bik or HI-8) for 6 h, biotin labeled, and then treated with the cross-linking reagent DSS. The cells were subsequently lysed, incubated again with a mixture of hyaluronidase and chondroitinase, and immunoprecipitated with each antibody. CD44v isoforms containing epitope v9 were not co-precipitated by anti-bik pAb (lane 1), and bik was not coprecipitated by the anti-CD44v9 mAb (lane 5), indicating no direct interaction between CD44v isoforms containing epitope v9 and bik. Also, v9-containing CD44v isoforms were not coprecipitated by anti-LP pAb and vice versa (lanes 2 and 6). Interestingly, a distinct heterodimer band between CD44v isoforms containing epitope v9 and bik-R could be detected in PMA-stimulated cells co-incubated with bik (lanes 4 and 8). It is therefore likely that the 150-kDa band is a heterodimer between CD44v isoforms containing epitope v9 (110 kDa) and bik-R (45 kDa). We showed that in the absence of bik, there is no association of bik-R with CD44v isoforms containing epitope v9 (lanes 3 and 7), demonstrating that bik-R requires bik for an association with CD44v isoforms containing v9 epitope. These results allow us to conclude that bik-R is able to physically interact with CD44 proteins only in the presence of bik on the cell surface.
The effect of bik and its truncated proteins on the migration properties of CD44 proteins was tested in immunoprecipitation followed by Western blot under reducing conditions (Fig. 8E). CD44v isoforms containing epitope v9 were co-precipitated by anti-bik-R pAb, and vice versa (lanes 4 and 8). Since distinct heterodimer bands between bik-R (45 kDa; lane 4) and CD44v isoforms containing epitope v9 (110 kDa; lane 8) could dissociate under reducing conditions, it is most likely a covalently linked heterodimer.
In a parallel experiment, we examined whether CD44v isoforms can interact with bik-R in the presence of deglycosylated bik or HI-8. We tested the effect of truncated biks on the migration properties of CD44v isoforms by immunoprecipitation followed by Western blot under nonreducing (Fig. 8D) and reducing (Fig. 8F) conditions. Unlike native bik, however, bik-R could not interact with CD44v isoforms containing epitope v9 in the presence of deglycosylated bik (Fig. 8, D and F, lane 1), Another important experiment is to establish if bik can dissociate the oligomers of CD44v isoforms after their form (that is 1 M bik is added 2 h after PMA stimulation). Bik has no or little capacity to dissociate the oligomers of CD44 proteins (Fig.  8B, lane 5). This result is in agreement with our previously published data in which no significant decrease of uPA release is observed when 1 M bik is added to the medium after stimulation by PMA (21). Therefore, bik has no ability to inhibit the pre-existing clusters of CD44 proteins.
The Effect of bik on uPA Expression in Other Type Cells-We verified whether our observations made with HCS-2/8 are applicable to human ovarian cancer cell line HOC-I. Two-color FACS analysis with a phycoerythrin-conjugated anti-uPA mAb and fluorescein isothiocyanate-conjugated anti-CD44 mAb revealed that HOC-I expressed uPA and CD44 proteins (data not shown). Like HCS-2/8, in vitro culture of HOC-I with PMA resulted in overexpression of CD44 and uPA (data not shown). We determined whether PMA stimulates the dimerization of FIG. 7. Effect of anti-CD44v9 mAb on PMA-induced HCS-2/8 cell-dependent uPA expression. A, we examined the ability of anti-CD44v9 mAb (2, 10, and 50 g/ml) and nonimmune mouse IgG (50 g/ml) to reduce PMA (100 nM)stimulated expression of uPA protein. uPA protein level in the conditioned medium was assayed by ELISA. Experiments were performed twice with similar results. Data are mean Ϯ S.D. based on two experiments. B, induction of uPA mRNA by PMA in the presence of anti-CD44v9 mAb (50 g/ml), nonimmune mouse IgG (NI-IgG; 50 g/ml), or bik (1 M). Northern blot analysis of total RNA (10 g) from cells for uPA mRNA levels. The bar graph was derived from the ratio of uPA and GAPDH densitometric measurements for each condition. Experiments were performed twice with similar results.
CD44 on HOC-I and bik suppresses PMA-stimulated dimerization of CD44. HOC-I were cultured in medium with or without PMA. After culture for 6 h, cells were surface biotinylated, immunoprecipitated, and Western blotted with avidin peroxidase. As shown in Fig. 9, under nonreducing conditions, CD44 proteins precipitated from unstimulated cells migrated as a broad band with a molecular mass of about 90 -110 kDa. In contrast, CD44 proteins precipitated from stimulated cells had two bands, a 90 -110-kDa band and a higher molecular mass 200 -220-kDa band. The 200 -220-kDa protein is detected only under nonreducing conditions, but not under reducing conditions. Therefore, it is likely that the 200 -220-kDa protein represents the disulfide-linked dimers of the 90 -110-kDa CD44 proteins. The addition of bik to PMA-stimulated cells had a significant effect on the electrophoretic mobility of the CD44 species visualized by immunoprecipitation using anti-CD44 mAb, where the 150 -160-kDa band was detected. However, monomers and dimers of CD44 proteins were weakly observed, demonstrating that bik could effectively block oligomerization of CD44 proteins.
A Role of bik-R in Regulating uPA Expression-uPA expression is observed in HCS-2/8 cells after treatment with phorbol ester (Fig. 7). Induction of uPA expression requires hours and involves increased CD44 dimerization on the cell surface. Here, we have investigated some parameters of specific antibodyinduced inhibition of uPA expression function of CD44 and bik-R. Requirement for heterologous interaction of bik-R with CD44 is addressed.
We investigated a role of bik-R/bik in regulating uPA expression through suppression of CD44 oligomerization. For this, monovalent Fab fragments were prepared from two antibodies, anti-CD44 pAb and anti-bik-R pAb. Of note that Fab monomers require about 10 -20-fold higher concentrations to obtain binding to cells equivalent to intact antibodies. As shown in Fig. 10, neither each antibody alone nor second cross-linker Ig alone does not affect uPA expression. As expected, the formation of homologous aggregates by Fab monovalent preparations of anti-CD44 pAb followed by second cross-linker Ig induced detectable increases in uPA expression at concentrations of 10 g/ml Fab monomers, but the magnitude of the suppression was conditions. The cells were incubated with 100 nM PMA supplemented with 1 M bik, deglycosylated bik, or HI-8, and subsequently biotin-labeled and treated with DSS. Then, CD44 proteins were immunoprecipitated with anti-CD44v9 mAb and stained with avidin peroxidase. In a separate experiment, 1 M bik was added 2 h after PMA stimulation (lane 5). C and E, the CD44 proteins were immunoprecipitated with anti-CD44v9 mAb and this immunocomplex was stained with specific antibodies to bik, LP, and bik-R, and vice versa under nonreducing (C) and reducing (E) conditions. D and F, the cells were incubated with 100 nM PMA supplemented with deglycosylated bik (1 M), HI-8 (1 M), or PBS, and subsequently biotin-labeled and treated with DSS. Then, CD44 proteins were precipitated with anti-CD44v9 mAb and the immunocomplex was stained with pAb bik-R under nonreducing (D) and reducing (F) conditions. A representative result of three independent experiments is shown. IP, immunoprecipitation; and IB, immunoblotting. a, the 85-kDa monomers of CD44s; b, the 110-kDa monomers of CD44v isoforms containing epitope v9; c, the 150-kDa complex between v9-containing CD44 isoforms and bik-R; d, the 220-kDa dimers of CD44v isoforms containing epitope v9; and e, macroaggregates and oligomerization of CD44v isoforms containing epitope v9. much lower than that induced by 100 nM PMA alone. The uPA expression induced by the homodimerization of CD44 could be suppressed under the formation of bik-R/CD44 heterodimer. In the presence of 100 nM PMA, the formation of homologous aggregates by Fab monovalents of anti-CD44 followed by second cross-linker Ig failed to induce significant increases in uPA expression, although the uPA release into the medium was slightly higher than that induced by PMA alone. On the other hand, suppression of PMA-induced uPA expression could be achieved by using goat anti-rabbit Ig to cross-link Fab monomers of anti-CD44 and Fab monomers of anti-bik-R (e.g. formation of CD44 homodimers, bik-R homodimers, and CD44bik-R heterodimers. The molecular ratio among CD44 homodimers, bik-R homodimers, and CD44-bik-R heterodimers at the end of the incubation was estimated to be ϳ1:1:2.) Thus, heterologous binding between CD44 and bik-R was required for the inhibition of PMA-induced signal transduction through suppression of CD44 oligomerization. We conclude, therefore, that bik-R is inhibiting unstimulated or PMA-induced uPA expression by direct influencing the microdistribution of CD44 (e.g. CD44 oligomerization formation) on the cell surface and that bik-R can also function as a membrane-associated binding protein for CD44 molecules on HCS-2/8 cells.
Analysis of Macroaggregates in the HCS-2/8 Cell Extracts-Hyaluronic acid-rich matrix was extracted from cells treated with or without bik. The aquisition of aggregates composed of HA was investigated using conventional SDS-PAGE. Fig. 11 shows that, when cells treated with bik were lysed, most of immunoreactivities of HA, bik, bik-R, CD44v isoforms containing epitope v9 and LP was recovered as high molecular mass complexes that did not enter the gel (A). In contrast, when the macromolecules were treated with Streptomyces hyaluronidase and DTT, bik, bik-R, v9-containing CD44v isoforms, and LP were dissociated from the high molecular weight complexes and then migrated within the gel (B). These results suggest the presence of the macroaggregates composed of the CD44-HA-LP-bik-bik-R in the HCS-2/8 cell extracts. On the other hand, the macromolecules consisting of CD44, HA, and LP were detected in the extract of HCS-2/8 cells incubated without bik (C and D). DISCUSSION A growing body of evidence has accumulated on the biological function of bik. We previously found that bik forms membrane complexes with bik-binding protein/receptor and initiates modulation of signal transduction, which results in bik-mediated suppression of cell invasiveness (9,15,17,20,21). We have clearly demonstrated in recent studies (15,17,20,21) that PMA efficiently induces uPA expression and secretion in HCS-2/8 cells (17), that co-treatment with PMA and bik induces a strong reduction of uPA expression by the cells that is dependent on bik dose, demonstrating that bik can suppress PMAstimulated up-regulation of uPA, and that exogenous bik forms aggregates (bik⅐LP⅐bik-R complex) on the surface of the PMAstimulated cells. However, the nature of the interaction between these complexes and CD44 proteins or between CD44 multimerization and uPA expression remains unclear. Furthermore, nothing is known about the mechanism by which bik-R would affect the function of CD44 proteins. In the present study, we tried to answer the mechanism, by which bik inhibits PMA-induced uPA expression at the gene level and the protein level. To elucidate the ligand-receptor interactions that mediate the function, we examined and compared the mechanisms of suppression by bik of PMA-induced up-regulation of uPA expression with bik-dependent suppression of PMA-stimulated or specific antibody-induced multimerization of CD44 proteins using human chondrosarcoma HCS-2/8 cells and ovarian cancer HOC-I cells.
The present study showed that (a) the HCS-2/8 cells stimulated with 100 nM PMA for 6 h markedly enhance the expression of splice variants of CD44 (CD44v isoforms containing epitope v9), but not standard form of CD44 (CD44s), on their cell surface; (b) within a 6-h incubation, PMA up-regulates the clustering of CD44v isoforms containing epitope v9 (dimerization of CD44 after PMA treatment has been shown before in other cell types (33)), while CD44s protein does not form homodimer and oligomeric clusters in the plasma membrane of the cells stimulated with PMA; (c) cross-linking of CD44 proteins on cells by either anti-CD44 mAb or anti-CD44 pAb followed by a second crosslinker antibody induces expression of uPA. The cells exposed to PMA showed about a 4.8-fold increase in uPA mRNA level, while the cross-linking of CD44 proteins exhibited about a 3-fold increase. These data indicated that PMA stimulates the expression of uPA, at least in part, via oligomerization of CD44 proteins; (d) addition of anti-CD44 mAb or anti-CD44v9 mAb alone (without a second cross-linker antibody) suppresses PMA-induced oligomerization of CD44 proteins; (e) PMA-induced uPA production was also inhibited by the addition of anti-CD44 mAb or anti-CD44v9 mAb; (f) bik does not inhibit the synthesis of new CD44 proteins; (g) bik can disrupt homodimerization of v9-containing CD44v isoforms possibly through coupling between CD44v isoforms containing epitope v9 and bik-R on the plasma membrane (we showed that deglycosylated bik or HI-8 does not have the same effect as the native bik); (h) bik-R requires bik for an association with CD44 molecules; and (i) treatment of cells with reducing agents disrupts heterodimerization between CD44v isoforms containing epitope v9 and bik-R: this fact suggests that disulfide mediated coupling (heterodimer formation) of CD44 proteins and bik-R that is sensitive to the effect of DTT is important for bik-mediated suppression of PMA-induced uPA expression. The present results clearly show that PMA stimulation is associated, at least in part, with dimerization of CD44 proteins, that PMA-induced clustering of CD44 proteins, but not simple upregulation of CD44 proteins, is necessary to allow up-regulation of uPA expression, and that suppression of clustering of CD44 proteins is required for bik-mediated down-regulation of uPA expression.
Therefore, we propose the following theory regarding the role of bik in suppression of PMA-stimulated signal transduction (Fig. 12). Clustering of CD44v isoforms containing epitope v9 stimulated by PMA may be able to function in the membrane environment. This might, in turn, stimulate the expression of uPA mRNA and protein. Exogenous bik could bind to bik-R with high affinity, free bik, or possibly the bik⅐bik-R complex also binds to LP (17), the LP⅐bik⅐bik-R complex formation might bring an easy access of bik-R to CD44 proteins, the subsequent complex suppresses PMA-stimulated dimerization/oligomerization of CD44 proteins possibly by steric hindrance or allosteric change, which would inhibit signal transduction involved in uPA expression. That is bik may promote the disaggregation of CD44 protein clustering through the formation of a putative multimeric structure containing LP and bik-R. We propose to refer to the bik-R as a "functional bik receptor" or a "CD44 accessory protein." These results allow us to speculate that the effect of LP is to enhance the recruitment of bik into a bik⅐bik-R binary complex with CD44 proteins, and bik-R functions by the inhibitory mechanism of PMA-induced clustering of CD44v isoforms containing epitope v9. To validate this mechanism, the effect of deglycosylated bik and HI-8 was studied: deglycosylated bik can bind to LP but not to bik-R. As expected, bik can disrupt the clustering of CD44 proteins, while deglycosylated bik or HI-8 does not, indicating that the C4S side chain of bik is important to display its function.
One may estimate that disruption of the oligomers with bik and simultaneous decrease in the uPA expression is not evidence for the involvement of CD44v oligomers in uPA regulation. To resolve this problem, we tried to induce or prevent oligomerization of CD44 proteins by other methods which are independent of bik: intact antibody or anti-CD44 Fab monomer and a second cross-linker antibody was used for induction of oligomerization of CD44 proteins. In addition, anti-CD44 mAb or anti-CD44v9 mAb alone without a second cross-linker antibody was used for suppression of oligomerization. Our observations revealed that the connection between dimerization of CD44 proteins and uPA expression are promising. It is unlikely therefore that PMA may well induce both processes separately, without any interplay between them. However, other possibilities are that upon bik treatment the bik-R may initiate a signal that prevents the PMA-stimulated uPA expression and that the affinity of CD44 proteins to recognize bik-R on the cell surface may increase when CD44 proteins are activated by PMA. The model in Fig. 12 shows a direct interaction of bik-R with CD44 molecules. The data suggesting an interaction rely on immunoprecipitaions which do not rule out the possibility that other intervening protein partners are mediating the interaction. Taken together, possible mechanisms of bik-R action include: (a) a change in the microdistribution of CD44 molecules in relationship to each other; (b) a change in distribution in relationship to other molecules including bik-R on the cell surface; (c) a conformational change in a certain domain of CD44; or (d) a combination of distribution and conformational effects. The ability of recombinant CD44s or CD44v9 to bind to recombinant bik-R would need to be assayed to assess direct interactions. These processes are not clear at the moment, because we have not characterized "bik-R." Cell contact with the extracellular matrix component HA plays an important role in many physiological and pathological processes including tumor cell invasion and metastasis (28 -30). It has also been reported that CD44 proteins form molecular aggregates in the plasma membrane, and that the ability of CD44 proteins to oligomerize correlates well with their affinity for HA (32,46,47). Therefore, PMA-induced microaggregation or clustering of CD44 proteins may be important for both CD44-dependent signal transduction and HA binding. Our preliminary data show that CD44 aggregation mediated by PMA results in enhanced binding of soluble HA to its receptor and that the ability of HA binding is reduced by bik, 2 although bik fails to inhibit the expression of CD44 proteins. This fact also strongly supports our hypothesis that bik efficiently suppresses the clustering of the CD44 proteins and/or induces conformational change in a certain domain of CD44.
Since the discovery of the binding proteins or receptors for bik, the pharmacological and biochemical profiles of this glycoprotein have been better defined; however, there are many aspects of biology of the binding proteins or receptors for bik that still remain poorly understood. A major problem that still remains is that bik-R has not yet been identified. Bik-R, from our previous data (17), can only be said plasma membraneassociated bik-binding protein(s)/receptor(s) other than LP. Although we have clearly demonstrated that CD44v isoforms containing epitope v9 aggregates on the surface of the PMAstimulated cells, which results in the up-regulation of uPA expression, the molecular nature of the interaction between CD44 protein multimerization and uPA overexpression remains unclear. In addition, the precise receptors (bik-R and CD44v)-intracellular effector interactions present in the cells may have not been adequately replicated. Also, nothing is known about the molecular mechanism by which bik-R coupling to CD44 proteins. Notwithstanding these limitations, this is the first report demonstrating the suppression by bik of PMA-induced expression of uPA mRNA and protein possibly through inhibition of dimerization and oligomerization of CD44 proteins on the plasma membrane of tumor cells.