Cell surface expression of HIP, a novel heparin/heparan sulfate binding protein, of human uterine epithelial cells and cell lines.

Previous studies established that uterine epithelial cells and cell lines express cell surface heparin/heparan sulfate (HP/HS)-binding proteins (Wilson, O., Jacobs, A. L., Stewart, S., and Carson, D. D.(1990) J. Cell. Physiol. 143, 60-67; Raboudi, N., Julian, J., Rohde, L. H., and Carson, D. D.(1992) J. Biol. Chem. 267, 11930-11939). The accompanying paper (Liu, S., Smith, S. E., Julian, J., Rohde, L. H., Karin, N. J., and Carson, D. D.(1996) J. Biol. Chem. 271, 11817-11823) describes the cloning of a full-length cDNA corresponding to a candidate cell surface HP/HS interacting protein, HIP, expressed by a variety of human epithelia. A synthetic peptide was synthesized corresponding to an amino acid sequence predicted from the cDNA sequence and used to prepare a rabbit polyclonal antibody. This antibody reacted with a protein with an apparent M of 24,000 by SDS-polyacrylamide gel electrophoresis that was highly enriched in the 100,000 × g particulate fraction of RL95 cells. This molecular weight is similar to that of the protein expressed by 3T3 cells transfected with HIP cDNA. HIP was solubilized from this particulate fraction with NaCl concentrations ≥0.8 M demonstrating a peripheral association consistent with the lack of a membrane spanning domain in the predicted cDNA sequence. HIP was not released by heparinase digestion suggesting that the association is not via membrane-bound HS proteoglycans. NaCl-solubilized HIP bound to heparin-agarose in physiological saline and eluted with NaCl concentrations of 0.75 M and above. Furthermore, incubation of I-HP with transblots of the NaCl-solubilized HIP preparations separated by two-dimensional gel electrophoresis demonstrated direct binding of HP to HIP. Indirect immunofluorescence studies demonstrated that HIP is expressed on the surfaces of intact RL95 cells. Binding of HIP antibodies to RL95 cell surfaces at 4°C was saturable and blocked by preincubation with the peptide antigen. Single cell suspensions of RL95 cells formed large aggregates when incubated with antibodies directed against HIP but not irrelevant antibodies. Finally, indirect immunofluorescence studies demonstrate that HIP is expressed in both lumenal and glandular epithelium of normal human endometrium throughout the menstrual cycle. In addition, HIP expression increases in the predecidual cells of post-ovulatory day 13-15 stroma. Collectively, these data indicate that HIP is a membrane-associated HP-binding protein expressed on the surface of normal human uterine epithelia and uterine epithelial cell lines.


I-HP with transblots of the NaCl-solubilized HIP preparations separated by two-dimensional gel electrophoresis demonstrated direct binding of HP to HIP. Indirect immunofluorescence studies demonstrated that HIP is expressed on the surfaces of intact RL95 cells. Binding of HIP antibodies to RL95 cell surfaces at 4°C was saturable and blocked by preincubation with the peptide antigen. Single cell suspensions of RL95 cells formed large aggregates when incubated with antibodies directed against HIP but not irrelevant antibodies.
Finally, indirect immunofluorescence studies demon-strate that HIP is expressed in both lumenal and glandular epithelium of normal human endometrium throughout the menstrual cycle. In addition, HIP expression increases in the predecidual cells of post-ovulatory day 13-15 stroma. Collectively, these data indicate that HIP is a membrane-associated HP-binding protein expressed on the surface of normal human uterine epithelia and uterine epithelial cell lines.
Heparan sulfate proteoglycans (HSPGs) 1 located either on cell surfaces or in extracellular matrices are found in nearly all mammalian tissues (1)(2)(3)(4)(5). Functionally, HSPGs and a variety of HP/HS-binding proteins have been shown to participate in a diverse range of biological processes such as cell attachment, growth factor binding, cell proliferation, migration, morphogenesis, and viral pathogenicity (6 -8). Several lines of evidence indicate that HSPGs play an important role during the initial attachment of the apical plasma membrane of trophectodermal cells of the blastocyst to the apical plasma membrane of the uterine epithelium. In mice, HSPGs are expressed on the cell surfaces of two-cell stage and post-implantation stage embryos (9). Furthermore, blastocyst attachment to laminin, fibronectin, and isolated mouse uterine epithelial cells is inhibited by HP. Embryo attachment also is inhibited by the treatment of embryos with HP/HS lyases or inhibitors of proteoglycan biosynthesis (10,11). Immunological studies of murine embryo implantation sites indicated that the core protein of the basement membrane HSPG, perlecan, and HP/HS chains are located between the apical cell surfaces of trophectodermal cells and uterine epithelial cells during the peri-implantation stage (12). Expression of perlecan on the external trophectodermal surface correlates with acquisition of attachment competence in vitro as well. Externally disposed H/HS-binding sites have been described on the cell surface of mouse uterine epithelial cells (13). Furthermore, using a heterologous cell adhesion assay, we demonstrated that HP/HS-like glycosaminoglycans participate in the initial attachment between two human cell lines, JAR and RL95, used to mimic the initial attachment of the human embryonic trophectoderm to human uterine epithelial cells, respectively (14). As is the case for mouse uterine epithelia, the human uterine epithelial cell line, RL95, has specific, high affinity cell surface HP/HS-binding sites, which are sensitive to mild trypsin digestion of intact cells. Three tryptic peptides that retained HP/HS binding specificity were isolated from such trypsinates and partially aminoterminal sequenced (15). In the accompanying paper (36), the full-length cDNA sequence to one of these proteins, named HIP for HP/HS interacting protein, was obtained and shown to encode a cell surface protein with an M r of 24,000 when expressed in transfected 3T3 cells. HIP is expressed in a cell type-specific fashion by many human cell lines, particularly those of epithelial origin.
In the current study, we have generated and characterized a rabbit antibody to a synthetic peptide designed from a predicted 16-amino acid sequence of HIP. These studies demonstrate that HIP is a peripheral membrane protein that directly binds HP and is expressed on the surfaces of normal human uterine epithelia and many uterine epithelial cell lines.

EXPERIMENTAL PROCEDURES
Materials-Tissue culture media components were obtained from Irvine Scientific (Santa Ana, CA) and Life Technologies, Inc. 125 I-Protein A was from ICN Radiochemicals (Irvine, CA). Tris, glycine, bovine serum albumin, urea, phenylmethylsulfonyl fluoride, polyhema, EDTA, magnesium chloride, heparin, and hemoglobin were purchased from Sigma. Sodium dodecyl sulfate, ␤-mercaptoethanol, acrylamide, bisacrylamide, and Tween 20 were purchased from Bio-Rad. Sodium azide, trichloroacetic acid, acetone, sucrose, paraformaldehyde, ammonium chloride, and calcium chloride were purchased from Fisher. Sodium chloride and methanol were purchased from EM Science (Gibbstown, NJ). Tissue culture plates (100 mm) were purchased from Falcon (Lincoln Park, NJ), and 24-well tissue culture plates were purchased from Corning (Corning, NY). Nitrocellulose membrane (0.45 m) was purchased from Intermountain Scientific Corp. (Bountiful, UT). Dithiothreitol was purchased from Boehringer Mannheim. Ethanol was purchased from AAPER Alcohol and Chemical Co. (Shelbyville, KY). Rabbit anti-Na ϩ /K ϩ -ATPase was purchased from Chemicon International, Inc. (Temecula, CA). Rabbit antibodies to human factor VIII and laminin were purchased from Dakopatt's (Glostrup, Denmark) and Collaborative Research (Bedford, MA), respectively. All chemicals used were reagent grade or better.
Cell Culture-Cells (RL95-2 or HEC-1a) were cultured in Dulbecco's minimal essential medium/Ham's F12, 1:1 supplemented with 100 units/ml penicillin and 10 g/ml streptomycin sulfate and 10% (v/v) heat-inactivated fetal bovine serum at 37°C in a humidified atmosphere of 95% air:5% CO 2 (v/v). For collection of conditioned medium, the same medium was used except that the fetal bovine serum was omitted. Medium was collected after a 24-h incubation. In most experiments, RL95 cells were used; however, in some experiments HEC-1a (purchased from the American Type Culture Collection) or Ishikawa cells (a generous gift of Dr. Erlio Gurpide, Mt. Sinai School of Medicine, New York) were cultured under the same conditions and used for preparation of whole cell extracts or conditioned media. Human endometrial tissue was obtained from routine biopsy specimens.
Peptide Synthesis and Antibody Generation-A synthetic peptide of the following sequence was constructed on a Vega 250 peptide synthesizer using FMOC methodology (16), CRPKAKAKAKAKDQTK. This synthetic peptide was conjugated to the keyhole limpet hemocyanin protein, using the Imject Maleimide Activated Carrier Proteins kit (Pierce) and was used for rabbit immunization following standard protocols (University of Texas M.D. Anderson Cancer Center, Bastrop, TX). Western blot analysis and immunocytochemical studies were conducted using polyclonal antibodies affinity purified with the synthetic peptide linked to maleimide-activated BSA (Pierce) conjugated to cyanogen bromide-activated-Sepharose (Sigma), using the manufacturer's protocol.
SDS-Polyacrylamide Gel Electrophoresis and Western Blotting-Cells or particulate subcellular fractions were initially solubilized and SDS-PAGE and Western blotting performed as described previously (17,18,22) using affinity purified rabbit anti-HIP as primary antibody and 125 I-protein A (30 Ci/g) as the detection system.
Membrane Preparations-RL95 cells were grown to 70% confluency on a 100-mm tissue culture plate. Membranes were isolated by differential centrifugation. Briefly, cells were washed three times with PBS and released from the plate by incubation with 10 mM EDTA in PBS at 37°C for 15-30 min. Cells were pelleted at 1000 ϫ g for 10 min at 4°C and resuspended in homogenizing buffer (0.25 M sucrose, 5 mM Tris-HCl (pH 7.4), 1 mM EDTA, 0.25 mM dithiothreitol, and a mixture of protease inhibitors (10)) and homogenized on ice. The homogenate was centrifuged at 1000 ϫ g for 10 min at 4°C. The 1000 ϫ g supernatant was centrifuged at 10,000 ϫ g for 20 min at 4°C. The 10,000 ϫ g supernatant was centrifuged at 100,000 ϫ g for 1 h at 4°C. Samples of pellets and supernatants were analyzed by SDS-PAGE and Western blotting as described above. RL95 cell surface components were radioiodinated with 1 mCi/ml Na 125 I (carrier-free; Amersham Corp.) in PBS for 30 min on ice by overlaying the cell layers with glass coverslips coated with 10 g of IODO-GEN (Pierce) (19). After this period, the coverslips were removed, and the cell layers were rinsed several times with PBS containing 1 mM NaI. Cells were scraped from the tissue culture dish with a rubber policeman and subsequently homogenized and subjected to the subcellular fractionation scheme described above. Equal protein loads of each fraction was applied to SDS-PAGE, and the location of the 125 I-labeled cell surface components determined by autoradiography of the dried gels.
NaCl Extraction of Membranes-High speed (100,000 ϫ g) membrane fractions were divided into equal parts and extracted either with 0.15, 0.4, 0.8, 1.2, or 1.6 M NaCl in 0.25 M sucrose, 1 mM EDTA, 0.25 mM dithiothreitol, and 5 mM Tris (pH 7.4), incubated overnight at 4°C, and centrifuged the next day at 100,000 ϫ g for 1 h. Supernatants were precipitated overnight at 4°C by the addition of trichloroacetic acid to a final concentration of 10% (w/v). Pellets were dissolved in 0.2 ml of sample extraction buffer and then precipitated and prepared for SDS-PAGE as described above.
Heparin Agarose Affinity Chromatography-High speed (100,000 ϫ g) membrane preparations were extracted overnight at 4°C with 0.4 M NaCl in 5 mM Tris (pH 8.0) and centrifuged at 100,000 ϫ g for 1.5 h. The 0.4 M NaCl-insoluble pellet was subsequently extracted with 0.8 M NaCl in 5 mM Tris (pH 8.0) at 4°C for 4 h and centrifuged 1.5 h at 100,000 ϫ g. The protein eluted between 0.4 and 0.8 M NaCl was diluted to 0.15 M NaCl, applied to a 0.5-ml pellet of prerinsed heparin-agarose (Sigma), and incubated overnight batchwise with constant rotary agitation at 4°C. A stepwise elution from heparin-agarose was performed with NaCl extending from 0.15-2.0 M in 5 mM Tris (pH 8.0). All fractions were trichloroacetic acid-precipitated and prepared for SDS-PAGE and Western blotting as described above.
125 I-Heparin Overlay of Two-dimensional Gels-High speed (100,000 ϫ g) particulate preparations were subjected to differential salt extraction with 0.4 M NaCl followed by 0.8 M NaCl as described above. The proteins then were separated by two-dimensional gel electrophoresis (20) and transferred to nitrocellulose as described above for Western blotting. The unblocked nitrocellulose was incubated with 125 I-Bolton-Hunter reagent-derivatized HP (15) in 0.15 M NaCl overnight at 4°C. The blot then was washed 3 times with PBS before drying for autoradiography. The same blot was blocked and then probed with HIP antibody and binding subsequently visualized with a peroxidase ABC system using a diaminobenzidene substrate kit as described by the manufacturer's instructions (Vector Labs, Burlingame, CA). A parallel gel run under exactly the same conditions was silver-stained as described (21) to visualize the migration positions of all proteins on the gel.
Immunocytochemistry-Cells were grown on coverslips for 48 h in Dulbecco's modified Eagle's medium/Ham's F12 with 10% (v/v) fetal bovine serum. After a brief rinse in PBS, the cells were fixed with 2.5% (w/v) paraformaldehyde in PBS for 15 min at room temperature, rinsed twice in PBS, and aldehyde groups blocked by incubation with 50 mM ammonium chloride in PBS for 15 min at room temperature. Incubation with primary and secondary antibodies and mounting of coverslips were as described in Julian et al. (22). For staining of endometrium, human endometrium was rapidly frozen in O.C.T. (Miles; Elkhart, IN) and sections prepared at -20°C on a Reichert Jung cryostat. These sections were fixed in 100% methanol for 10 min at room temperature, rehydrated in PBS for 5 min at room temperature, and immediately used for immunostaining. The stage of the menstrual cycle was identified in all cases by standard histological examination (23) and serum hormone profiles. In all cases, the affinity purified HIP primary antibody was used at a concentration of 25 g/ml and the secondary antibody, fluorescein-conjugated donkey anti-rabbit Ig (Amersham Corp.), at a 1:10 dilution. Rabbit antiserum to human factor VIII was used at a 1:30 dilution. Rabbit antiserum to mouse laminin was used at a 1:50 dilution.
Binding of Anti-HIP to RL95 Cell Surfaces-RL95 cells were grown to 90% confluency in 24-well tissue culture plates with Dulbecco's modified Eagle's medium/Ham's F12 containing 10% (v/v) fetal bovine serum. Cells were rinsed three times with Hanks'-buffered saline and preincubated for 15 min at 4°C in 0.5 ml of binding buffer (PBS containing 2 mM CaCl 2 , 2 mM MgCl 2 , 0.1% (w/v) hemoglobin, 1 mM NaI, and 0.02% (w/v) NaN 3 ). The binding buffer was removed, and 0.2 ml of binding buffer containing anti-HIP or nonimmune rabbit IgG was incubated for 45 min at 4°C in duplicate wells. IgG was added at the following concentrations, 0, 10, 50, 100, and 200 g/ml. Cells were rinsed 3 times with binding buffer at 4°C for 5 min and incubated with binding buffer containing 125 I-protein A (1 ϫ 10 6 cpm/well) for 30 min at 4°C. After rinsing 3 times with ice-cold binding buffer, cells were solubilized with 1% (w/v) SDS and 0.5 M NaOH, and the amount of 125 I-protein A bound to RL95 cell surface was determined. To determine nonspecific binding of anti-HIP or nonimmune rabbit IgG, antibody was preincubated for 2 h at 4°C with or without 100 l of peptide affinity matrix and then centrifuged.
Aggregation of RL95 Cells by Anti-HIP-RL95 cells were grown in tissue culture plates to 70% confluency and detached with 10 mM EDTA in PBS without calcium and magnesium. Cells were resuspended in media, (Dulbecco's modified Eagle's medium/Ham's F12 containing 1% (v/v) penicillin/streptomycin and 0.1% (w/v) BSA), at a concentration of 3.5 ϫ 10 6 cells/ml. To prevent adhesion, wells were precoated with 1 mg of polyhema in 100% ethanol at 37°C overnight until dry and then rinsed 3 times with media before use. The following components were added to each well: 1 ml of the media, 100 l of buffer (PBS plus 0.02% (w/v) sodium azide) with or without control IgG antibody, or anti-HIP protein at 25 g/ml and 7 ϫ 10 5 cells (200 l). Plates were incubated for 3 h at 37°C on a rotary shaker at 1700 rpm. Afterward, cellular aggregation was viewed and photographed with a Nikon Diaphot inverted microscope using phase microscopy.

Subcellular
Distribution of HIP-An antibody was generated to a synthetic peptide sequence predicted from the full-length HIP cDNA sequence (36). The sequence was predicted to be hydrophilic and likely to be exposed on the external surface of the intact protein. The antibodies routinely used for the studies described below were affinity purified on a column composed of the BSA-conjugated HIP peptide linked to agarose. As shown below, these antibodies reacted primarily with a protein with the M r of 24,000 as estimated by SDS-PAGE and Western blotting. This molecular weight is similar to that observed for HIP protein expressed by 3T3 cells transfected with full-length HIP cDNA (36). Subcellular fractionation was used as an initial step to partially purify HIP for subsequent analytical studies. Fractionation of RL95 cells and subsequent Western blot analysis determined that HIP was most highly enriched in the 100,000 ϫ g pellet; however, HIP was detected in other particulate fractions as well (Fig. 1). Lower molecular weight components immunologically related to HIP were detected in the 1000 ϫ g/20 min and 10,000 ϫ g/20 min particulate fraction. These components were presumed to be partially degraded forms of HIP. In contrast, HIP appeared to be quantitatively depleted from the 100,000 ϫ g soluble fraction. A similar distribution of HIP was observed in JAR and HEC-1a cells, human trophoblastic and uterine adenocarcinoma cell lines, respectively (data not shown). The high speed particulate fraction was used further as the most convenient source of HIP.
NaCl Solubilization of HIP-The 100,000 ϫ g particulate fraction was subjected to incubations with increasing concentrations of NaCl and then analyzed by Western blot analysis. At a NaCl concentration greater than 0.8 M, HIP was eluted from the membrane fraction into a 100,000 ϫ g soluble fraction (Fig. 2). Analyses of the corresponding supernatants for each salt wash indicated that HIP is partially eluted with 0.4 M NaCl but is completely eluted with NaCl concentrations of 0.8 M or greater. The solubilization of HIP with salt indicates that HIP is likely to be peripherally associated with the particulate fractions of cells. Conditioned media from RL95 cells were centrifuged at 100,000 ϫ g, and the corresponding pellet and supernatant were analyzed by Western blot analysis (Fig. 3). HIP was not detected in secretions from RL95 cells, indicating that this protein is not secreted or released from RL95 cells to a significant extent.
HIP Binds Heparin-The 0.8 M NaCl eluate from the 100,000 ϫ g particulate fraction was diluted to 0.15 M NaCl and incubated with heparin-agarose (Fig. 4). Elution of the heparinagarose with increasing concentrations of NaCl demonstrated that HIP bound to heparin and was released at NaCl concentrations greater than 0.75 M. Staining with Coomassie Blue (data not shown) indicated that multiple proteins were present in the heparin-binding fractions. Therefore, a HP overlay assay was employed to demonstrate the ability of HIP to bind HP directly (15). Samples were subjected to two-dimensional SDS-PAGE, transferred to nitrocellulose, and sequentially probed with 125 I-HP and anti-HIP. Profiles of total proteins were visualized in a parallel sample by silver staining. As shown in  Cell Surface Localization of HIP-Anti-HIP was used to determine if this protein was expressed on the external surface of intact cells. Initially, concentration dependence and saturability of anti-HIP binding was examined. Fig. 6A shows that binding of anti-HIP to intact RL95 cells was both specific and saturable as compared with binding of nonimmune rabbit IgG. Furthermore, when anti-HIP protein was pre-absorbed with peptide affinity matrix, its binding was reduced to the level observed with nonimmune rabbit IgG (Fig. 6B). Next, anti-HIP was used to examine the distribution of this protein on HEC-1a cell surfaces. As shown in Fig. 7, immunostaining of methanolpermeabilized, paraformaldehyde-fixed HEC-1a cells with anti-cytokeratins demonstrated a strong positive signal (panel B). In contrast, fixed, nonpermeabilized cells displayed only background staining (panel A) comparable with that observed when primary antibody was omitted (panel D). Staining of fixed, nonpermeabilized cells with anti-HIP was uniformly distrib-uted on the surfaces of all cells in these cultures including points of cell-cell contact (panel C). Similar results were obtained with RL95 cells (data not shown). Collectively, these data indicated that reactivity with anti-HIP was reflective of cell surface staining and not due to permeabilization in human uterine epithelial cell lines.
It was further reasoned that if HIP was on RL95 cell surfaces then non-fixed, single cell suspensions of living RL95 cells could be aggregated by anti-HIP. As shown in Fig. 8, incubation of single cell suspensions of RL95 cells with anti-HIP greatly enhanced cell aggregation. Parallel controls, including PBS, PBS containing 0.02% sodium azide and an antibody to the cytoplasmic tail of the mucin, MUC1 (24), did not enhance RL95 cell-cell aggregation. Collectively, these data strongly indicate that HIP is located on the extracellular surface of the plasma membrane of human uterine epithelial cell lines.
Experiments also were performed to determine if HIP is expressed by other human uterine epithelial cell lines as well as normal human uterine epithelium in situ. As shown in Fig.  9, Western blots of several human uterine epithelial cell lines as well as human endometrium displayed a prominent band corresponding to the molecular weight of HIP. A 1.3-kilobase transcript is detected in all three cell lines by Northern analyses using HIP cDNA as a probe (36).
HIP Expression in Human Endometrium-Expression and localization of HIP was examined in methanol-fixed frozen sections of human endometrium taken at various stages throughout the menstrual cycle. In all cases, strong reactivity of lumenal and glandular epithelia was detected. Through the proliferative and until post-ovulatory day 7 of the cycle, HIP reactivity was not detected in underlying stroma cells (Fig. 10). Nonimmune IgG failed to react with these tissues (data not shown). Furthermore, the epithelial identity of the HIP-positive cells was confirmed by demonstration of reactivity with antisera to cytokeratins and Muc-1 in serial sections (data not shown). Strong reactivity was detected at both the apical and basal aspects of these cells. Some variation in the intensity of signal between these glandular structures was noted. It is unclear if this variation reflects differences between glands or regional differences in HIP expression of individual glands that normally extend from the uterine lumen (functionalis) to deep within the endometrium (basalis). By post-ovulatory day 13, additional staining for HIP was detected within the underlying stroma (Fig. 11). As expected, the underlying stroma extracellular matrix also displayed strong expression of the decidual marker, laminin (27), at this time. In contrast, laminin expression was confined to basal lamina in stromal tissue of late proliferative stage uteri. The heparan sulfate proteoglycan, perlecan, also has been reported to be expressed by decidualizing stroma cells (26); however, stromal staining for perlecan was much less intense than that of basal lamina (data not shown). As mentioned above, HIP was not detected in stromal cells through the entire proliferative phase of the cycle. These data demonstrated that HIP is a protein normally expressed by uterine epithelia.
FIG. 6. Binding of anti-HIP to intact RL95 cells is saturable and specific. A, monolayers of RL95 cells were grown in a 24-well tissue culture plate to 90% confluency. Cells were incubated at 4°C for 45 min with anti-HIP (q) or nonimmune rabbit IgG (Ⅺ) as described under "Experimental Procedures." The data represent the average Ϯ S.E. of duplicate determinations. The triangles represent the average Ϯ S.E. obtained for specific binding (anti-HIP binding) minus the average binding obtained with nonimmune rabbit IgG (nonspecific). Binding is both specific and saturable between 5 and 10 g of anti-HIP/ml. B, in a similar experiment 25 g of anti-HIP or nonimmune rabbit IgG was preincubated without (striped boxes) or with (open boxes) 100 l of peptide affinity matrix for 2 h before incubation with RL95 cells. The data are the averages Ϯ S.E. for duplicate determinations in each case.  (24) and represents another negative control. Panel D shows cells incubated with anti-HIP. In panels B and C, some cell-cell aggregation is noted after 3 h of incubation; however, panel D indicates that this aggregation is greatly enhanced by the presence of anti-HIP. Samples were photographed with a Nikon Diaphot inverted microscope using inverted phase microscopy. DISCUSSION A number of studies described above have demonstrated that HSPGs are expressed on the surfaces of mouse blastocysts and human trophoblastic cell lines where they function in cell adhesion events. In these studies, it was further demonstrated that adhesive activity resides in the constituent HS chains of the HSPGs. Consistent with these observations, specific HP/ HS-binding sites were identified on the surfaces of both mouse uterine epithelial cells and human uterine epithelial cell lines (13,15). HP/HS-binding sites have been described on the surfaces of a number of cell lines (28 -31); however, identification of these proteins has been elusive. N-CAM represents one well described cell surface HP/HS-binding protein (32) but is not expressed in the uterus. Recently, heparin-binding epidermal growth factor-like growth factor was identified at mouse implantation sites (33) and is one potential ligand for embryonic HSPGs. Several other candidate proteins have been described that display HP/HS-binding activity (34, 35) but have not been well characterized. In previous studies, we were able to obtain a partial amino-terminal sequence of several tryptic peptides derived from RL95 cell surfaces that retained HP/HS-binding activity. This sequence was used to obtain a full-length cDNA and predicted amino acid sequence of one of these proteins (36). This protein is referred to as HIP. Inspection of the predicted amino acid sequence of HIP using several protein structurepredicting algorithms indicated regions likely to be antigenic and exposed on the exterior surface of the protein. One of these sequences was chosen for preparation of antibodies, and these antisera have been used in the present study.
The predicted pI of HIP, Ͼ10, is consistent with its behavior on isoelectric focusing gels. Alternatively, HIP may be posttranslationally modified. No consensus sites for glycosylation are indicated by the predicted sequence; however, other modifications are possible. Subcellular fractionation studies indicate that HIP is most highly enriched in the high speed particulate fraction and is quantitatively depleted from the high speed supernatant, i.e. cytosolic fraction. We have detected various plasma membrane markers in this fraction including Na ϩ /K ϩ -ATPase and radioiodinated cell surface components 2 ; however, rearrangement of peripheral membrane components like HIP may occur during such fractionation making interpretation of subcellular locale by this approach problematic. The ability of NaCl to release HIP from the particulate fraction is consistent with the lack of a potential membrane spanning domain in the predicted sequence of HIP and demonstrates that HIP is a peripheral membrane protein. Digestion of membranes with a mixture of HP/HS lyases did not release HIP into the 100,000 ϫ g soluble fraction. This suggests that HIP is not retained by membrane-bound HSPGs. Therefore, it is possible that other membrane components bind and retain HIP. Alternatively, it is possible that HIP binds to a region of HS close to the protein core and protects HS from enzymatic digestion. Characterization of the HIP-binding sites is necessary to define the nature of the HIP-membrane interaction.
Several lines of evidence indicate that HIP is displayed on cell surfaces. Antibodies to this protein bind specifically and in a saturable manner to intact RL95 cells under conditions where endocytosis should not occur. Assuming a 1:1 stoichiometry of IgG binding to HIP and protein A to antibody, it can be calculated that there is an average of approximately 1.5 ϫ 10 4 molecules of HIP displayed on the surface of each RL95 cell. If each IgG binds to two HIP molecules and each protein A tetramer binds four IgG molecules then this estimate may be as high as 1.2 ϫ 10 5 HIP molecules per cell surface. In either case, these numbers are well below the number of [ 3 H]HP-binding sites (9 ϫ 10 6 ) previously determined for RL95 cells (15). Consequently, even given potential inaccuracies in both estimates, it seems that HIP can only be one of multiple cell surface HP/HS-binding proteins displayed on RL95 cell surfaces. It is possible that many HIP molecules are occupied by HS at the cell surface and masked from antibody binding. HS lyase pretreatment of cells did not expose additional anti-HIP-binding sites 3 ; however, if, as discussed above, HIP binding "protects" HS chains from digestion then HS lyases might not be expected to expose more HIP.
Antibodies to HIP also display staining patterns on intact RL95 cells that are consistent with those of cell surface components, e.g. enrichment at cell peripheries and regions of cell-cell contact. Similar patterns of immunoreactivity with anti-HIP are detected on human trophoblastic and breast cancer cell lines. 4 Furthermore, these same antibodies specifically aggregate RL95 cells in suspension, a property expected for antibodies reacting with epitopes displayed on the cell surface. Experiments with an impermeant chemical cross-linking reagents destroyed antibody reactivity with HIP, but larger cell associated bands were not observed. 2 Thus, while in one sense these experiments suggest a cell surface disposition of the protein, the apparent destruction of the epitope confuses interpretation. Collectively, these data strongly argue that at least a fraction of the population of HIP is displayed on RL95 cell surfaces where these proteins may directly participate in HP/HS binding.
HIP is detected in several human uterine epithelial cell lines and in human endometrium by Western blotting of total protein extracts. Moreover, anti-HIP strongly reacts with uterine epithelial cells in sections of human endometrium through post-ovulatory day 7 of the cycle. By post-ovulatory day 13, HIP is also detected in the predecidual cells of the uterine stroma. The HSPG, perlecan, is expressed by human decidual cells (26).
It is possible that HS chains of perlecan also serve as ligands for HIP in basal lamina and in the decidual extracellular matrix. In any event, these observations indicate that HIP is expressed by normal human endometrium. Potential functions could involve binding to basal lamina or intercellular HSPGs expressed by uterine epithelia or HSPGs expressed by blastocysts during implantation. The antibody described in the present studies does not react with mouse uterine components either by immunostaining or Western blotting. Current efforts are being placed toward generating probes to the mouse homologue so that the physiological role of this protein in the uterus can be more rigorously examined by molecular genetic approaches.