Hepatocyte Growth Factor Specifically Binds to Sulfoglycolipids”

Hepatocyte growth factor (HGF) is a heparin-binding pleiotropic factor that acts on a variety of epithelial cells. The interaction of human HGF with glycolipids was studied by overlaying them with 12%HGF on thin layer chromatograms and by a solid-phase assay using lipids adsorbed on microtiter plates. Among various glycolipids tested, HGF was found to bind to sulfoglycolip- ids, including galactosylceramide sulfate (SM4), lactosylceramide sulfate (SM3), and gangliotriaosylceramide bis-sulfate. In contrast, HGF failed to bind to gangliosides or neutral glycolipids. HGF binding to SM4 was strongly inhibited by dextran sulfate, heparin, and fu-coidan, whereas neither keratan sulfate nor hyaluronic acid had any inhibitory activity. When glycolipids from a renal cancer cell line, SMKTR3, which overexpresses sulfoglycolipids, were devel- oped on a thin layer chromatogram, method as described previously (3). In short, SMKT-R3 cells were harvested, washed, and reacted on ice for 30 min with a monoclonal antibody to sulfoglycolipid Sulph-I (18) or with a polyclonal antibody to HGF as the first antibody, and subsequently with a fluorescein isothiocyanate-conjugated F(ab'), fragment of rabbit anti- mouse IgG (Dako) or of goat anti-rabbit I g G (Cappel) as the second antibody, respectively. For some experiments, the cells into which gly- colipids had been incorporated as above were preincubated with 0.4 mg/ml HGF on ice for 1 h before the immunofluorescence staining. Fluorescence profiles were determined with a FACScan (Becton Dick-inson).


R3, which overexpresses sulfoglycolipids, were developed on a thin layer chromatogram, SM4 and SM3 were the only glycolipids that bound HGF. We further examined the effect of the incorporation of glycolipids into SMKT-R3 cells on HGF binding to the cells. The incorporation of SM4 into the cells enhanced HGF binding to
SMKT-R3 cells, while that of galactosylceramide, a precursor of SM4, had no effect. These observations indicated that SM4 exogenously incorporated into the cell membranes could react with HGF and suggested that endogenous sulfoglycolipids on SMKT-R3 cells might function as reservoirs for HGF.
Sulfoglycolipids are a class of acidic glycolipids containing one or two sulfate esters on their oligosaccharide chains and are responsible for some of the negative charge on the cell surface. Kidney and brain contain abundant sulfoglycolipids (1). In h u m a n renal cell carcinoma tissue, sulfoglycolipid content is markedly increased as compared with uninvolved tissue (2). Furthermore, the accumulation of sulfated glycolipids was also demonstrated in established cells from human renal cell carcinoma and associated with an elevated activity level of glycolipid sulfotransferase (3).
Although a variety of functions have been suggested for sulfoglycolipids 141, the actual functions in uiuo remain to be clarified. Recently, sulfated glycolipids were found to bind spe-'$This work was supported by the Grant-in-Aid for Scientific Research on Priority Areas No. 05274107 and the Special Grant-in-Aid for Promotion of Education and Science in Hokkaido University Provided from the Ministry of Education, Science and Culture, Japan and the Mizutani Foundation for Glycoscience. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
HGF' is a pleiotropic factor that is produced by mesenchymal cells and acts as a mitogen, motogen, and morphogen for various epithelial cells, including renal cells (11). In our recent study, it w a s d e m o n s t r a t e d that HGF elevated glycolipid sulfotransferase activity in renal cell carcinoma cells.' Since HGF is known to h a v e an affinity for heparin (12), these observations prompted us to examine whether HGF bound to sulfoglycolipids on renal cancer cells.

EXPERIMENTAL PROCEDURES
Materials-NalzsI (17 Ci'mg) was purchased from New England Nuclear. Human recombinant HGF was purified from culture medium of C127 mouse fibroblasts or Chinese hamster ovary cells transfected with plasmid containing human HGF cDNA (13). HGF was radioiodinated by the chloramine-T method as described previously (14). The '"1-HGF had a specific activity of 70-160 pCi/pg protein. Antisera were raised against human recombinant HGF in rabbits (15). Heparin, fucoidan, dextran sulfate, keratan sulfate, and hyaluronic acid were obtained from Sigma. DEAE-Sephadex A-25 was the product of Pharmacia-LKB. Other reagents were of analytical grade.
Cell CuZture"SMKT-R3 cells were established from human renal cell carcinoma and cultured as described previously (16). Cell viability was estimated by the trypan blue exclusion test and was always greater than 95%.
Preparation and Analysis of Glycolipids from Renal Cancer Cells-Glycolipids were prepared from SMKT-R3 cells as described previously ( 3 ) . Briefly, lipids of the cells were extracted with a mixture of chloroforndmethanol/water i60:35:8 and 30:60:8, the ratio ofthe solvent mixture is expressed by volume). The extracts were subjected to mild alkaline hydrolysis t.o destroy ester lipids and then fractionated into neutral and acidic lipid fractions by DEAE-Sephadex A-25 column chromatography. Prepared glycolipids were chromatographed on aluminumbacked silica gel HPTLC plates (Merck! using the solvent system chloroforndmethanol/0.2% CaCl, (60:35:7) and detected with an orcinol reagent (17).
Assay of 'Z,51-HGF Binding to Glycolipids-Binding of lZsII-HGF to glycolipids on HPTLC plates was performed as previously described for '"I-thrombospondin binding (7) except that the incubation with 9 -HGF was done for 2 h at room temperature. The labeled HGF bound to glycolipids was visualized by autoradiography and quantified by densitometry of the autoradiogram. Solid-phase radioassays of HGF binding to glycolipids adsorbed on 96-well microtiter plates (Falcon 3912) were carried out based on the previousiy described method (7). fibroblast growth factor; DMEM, Dulbecco's modified minimal essential The abbreviations used are: HGF, hepatocyte growth factor; FGF, medium; GalCer, galactosylceramide; LacCer, lactosylceramide; Gb,Cer, globotriaosylceramide; Gb,Cer, globotetraosylceramide; SB2, his-sulfogangliotriaosylceramide; SM2, gangliotriaosylceramide 3'-sulfate; SM3, lactosylceramide 3"sulfate; SM4, galactosylceramide 3'-sulfate; HPTLC, high performance thin layer chromatography. Glycolipid Incorporation into Renal Cancer Cells-In order to incorporate glycolipids into SMKT-R3 cells, GalCer or SM4 dissolved in a small amount of dimethyl sulfoxide was added exogenously to the cells. The cells were incubated with glycolipids in serum-free DMEM a t 37 "C for 10 min.

HGF and Sulfoglycolipids
Cytofluorornetric Analysis-Flow cytometry was performed by an indirect immunofluorescence method as described previously (3). In short, SMKT-R3 cells were harvested, washed, and reacted on ice for 30 min with a monoclonal antibody to sulfoglycolipid Sulph-I (18) or with a polyclonal antibody to HGF as the first antibody, and subsequently with a fluorescein isothiocyanate-conjugated F(ab'), fragment of rabbit antimouse IgG (Dako) or of goat anti-rabbit I g G (Cappel) as the second antibody, respectively. For some experiments, the cells into which glycolipids had been incorporated as above were preincubated with 0.4 mg/ml HGF on ice for 1 h before the immunofluorescence staining. Fluorescence profiles were determined with a FACScan (Becton Dickinson).

Binding of HGF to Sulfoglycolipids-To examine the ability
of glycolipids on renal cancer cells to serve as receptors for HGF, binding to glycolipids was investigated in two ways (19).
In the first assay technique, glycolipids were separated on a thin layer plate and then overlaid with radiolabeled HGF. As shown in Fig. lA, HGF bound to authentic sulfoglycolipids SM4, SM3, and SB2, but not to either neutral glycolipid or ganglioside standards. Likewise, HGF specifically bound to SM4 and SM3 extracted from a renal cancer cell line, SMKT-R3, which expresses SM4, SM3, and SM2 (3), but not to other acidic or neutral glycolipids from the cells (Fig. 1B ). As shown in Fig. 2 A , HGF binding was dependent on the amount of SM4, detecting as little as 7.5 ng of SM4. When binding of HGF to increasing amounts of authentic SM4 was quantified by densitometric analysis of the autoradiogram, a dose-dependent curve was obtained (Fig. 2B).
In the second assay technique, HGF binding to glycolipids was investigated by a solid-phase binding assay using lipids adsorbed on microtiter plates. Consistent with the data presented above, HGF bound to SM4, SM3, and SB2, but not to SM2, GalCer, or gangliosides (Fig. 3). When SM4 was deacylated, HGF binding to lyso-SM4 was considerably decreased as compared with intact SM4. Dose dependence of HGF binding to  from three separate experiments performed in triplicate and are expressed as percentages of the HGF binding to SM4.
SM4 was also observed in this assay, and binding was abolished completely by excess amounts (100 pgimlt of unlabeled HGF (data not shown). Nonspecific binding of HGF to uncoated wells was less than 1% of the HGF added.
Inhibition of HGF Binding to Sulfoglycolipids-In order to confirm whether the binding observed above was specific to HGF, the effect of anti-HGF antibody on the binding of HGF to SM4 was examined using a solid-phase radioassay. As shown in Fig. 4, the HGF binding was inhibited by coincubation with the antibody in a concentration-dependent manner.
To further define the interaction of HGF with sulfoglycolipids, several anionic polysaccharides were tested for their ability to inhibit HGF binding to SM4 (Fig. 5). A sulfated fucan, fucoidan, was the most potent inhibitor, and dextran sulfate and heparin were also effective. In contrast, keratan sulfate and hyaluronic acid were inactive as inhibitors.
Incorporation of Sulfoglycolipid Enhances the HGF Binding to Renal Cancer Cells-Glycolipids exogenously added to cell culture media are incorporated into plasma membranes (20). This approach was utilized to examine whether the HGF binding to SMKT-R3 cells was increased when sulfoglycolipids were incorporated into the cell membranes. The amount of sulfated glycolipids was analyzed by flow cytometry with a monoclonal antibody directed to a sulfoglycolipid, Sulph-I, which specifically reacts with SM3 and SM4 (18). In our previous study, flow cytometry with Sulph-I was able to detect sulfoglycolipids on the SMKT-R3 cell surface (3). After SM4 was incorporated into SMKT-R3 cells, flow cytometry with Sulph-I was performed. As shown in Fig. 6, the SM4-enriched cells considerably enhanced the reactivity with the antibody. To examine whether the incorporation o f SM4 into the cells increased the binding of HGF to the cells, the cells were pretreated with SM4 or GalCer, a precursor of SM4 that lacks a sulfate group, incubated with HGF, and subjected to flow cytometry using an anti-HGF antibody (Fig. 7). The difference of fluorescence intensity between line 1 (mean fluorescence intensity, 96) and line 2 (mean fluorescence intensity, 546) was interpreted as the endogenous ability to bind HGF on the cells. Pretreatment with SM4 resulted in a n increment of the reactivity with the anti-HGF antibody (mean fluorescence intensity of line 4, 10121, while GalCer had no significant effect (mean fluorescence of line 3, 566). Compared with fine 1, pretreatment with SM4 or GalCer or dimethyl sulfoxide as the vehicle had no influence on the reactivity with an irrelevant antibody (data not shown). These observations indicated that SM4 incorporated into the cell membranes could react with HGF and suggested that exogenous SM4 could function as a n HGF-binding site. It is, there- fore, likely that endogenous sulfoglycolipids on SMKT-R3 cells play a role as HGF receptors.

DISCUSSION
In the present study, we have demonstrated that HGF bound to sulfoglycolipids on plastic plates, on thin layer chromatograms, and on renal cancer cells. The properties of HGF binding to sulfoglycolipids were similar to those of thrombospondin (71, antistasin (21), and properdin (22) in regard to half-maximum binding and the potency of sulfated glycoconjugates to inhibit binding among sulfatide-binding proteins (23). HGF bound to SM4 and SM3, but not to GalCer or LacCer, which are the respective precursor glycolipids. Moreover, additional Gal-NAc at the nonreducing terminal of SM3 resulted in the inability of HGF to bind to SM2. On the other hand, HGF could bind to SB2, which i s sulfated at the nonreducing terminal residue of SM2. These observations suggest that a sulfate ester on the nonreducing terminal residue is critical for HGF binding to

FIG. 7. Effect of incorporation of sulfoglycolipid into renal cancer cells on HGF binding to the cells.
After SMKT-R3 cells were preincubated a t 37 "C for 10 min in serum-free DMEM containing 100 pu SM4 or GalCer or 0.1% dimethyl sulfoxide as vehicle (none) and washed free of excess glycolipids, they were allowed to react with HGF on ice for 1 h followed by flow cytometry with anti-HGF antibody or control antibody.
sulfoglycolipid. HGF also appeared to recognize the fatty acid moiety to some extent since there was a considerable decrease in the binding of HGF to lyso-SM4 compared with SM4. Alternatively, it is possible that the free amino group of lyso-SM4 affects the binding.
There are two classes of HGF-binding sites on the surfaces of target cells, one with high affinity called c-Met, and one with low affinity (24)(25)(26)(27)(28)(29). We also observed the presence of both binding sites on SMKT-R3 cells.3 Our present observations suggest that sulfoglycolipids on renal cancer cells can act as low affinity binding sites. Since HGF was eluted from the low affinity sites by excess heparin, it was suggested that the binding site most likely corresponded t o matrixor cell-associated heparan sulfate proteoglycans (25,26). HGF has an affinity for heparin (12), and the N-terminal hairpin domain and the second kringle domain are responsible for heparin binding.3 Heparin-binding sites are involved in sulfoglycolipid binding to laminin (30), thrombospondin (31), and antistasin (32). Thus, the N-terminal hairpin structure of HGF is possibly associated with sulfoglycolipid binding. It is necessary to further explore the role of sulfoglycolipids in the low affinity binding sites on renal cancer cells.
It is clear from previous studies that the high affinity receptor (c-Met) is needed for the biological response of target cells (29). Although there has been no evidence for the involvement of the low affinity receptor in HGF signal transduction, it is unknown a t present whether cellular responses to HGF require :' T. Nakamura, H. Tajima, and K. Matsumoto, unpublished data. only c-Met or its association with other molecules. Recent studies have demonstrated that basic FGF in the absence of cell surface heparan sulfate proteoglycans does not bind to its high affinity receptor and is not active, suggesting that basic FGF binding to cell surface heparan sulfate proteoglycans is a prerequisite for its high affinity binding (33). Furthermore, a deletion-mutant protein of HGF that lacked the N-terminal hairpin structure did not bind to heparin and lost its biological activities (34). It is, therefore, conceivable that low affinity binding sites such as heparan sulfate proteoglycans and sulfoglycolipids are required for the high affinity binding or exertion of biological activities of HGF. Moreover, sulfoglycolipids on renal cancer cells may play a role as binders and reservoirs for HGF to accumulate it on the cell surface, to protect it from degradation, or to transfer it to the high affinity receptors that initiate the cellular response. It is of importance to characterize these low affinity receptors for the elucidation of the HGF signal pathway.
Although there are a number of heparin-binding growth factors (35), including HGF, acidic and basic FGF, and interleukin-3, the presence of a heparin-binding site is not always sufficient for binding to sulfoglycolipids (30,36). To our knowledge, the present study is the first report that sulfoglycolipids bind a growth factor. It is essential to further study the binding properties of HGF to sulfoglycolipids and to examine the possibility of binding of other growth factors to sulfoglycolipids.