JOURNAL OF CELLULAR PHYSIOLOGY 14350-67 (1990)

Expression of Externally-Disposed Heparin/Heparan Sulfate Binding Sites by Uterine Epithelial Cells OSWALD WILSON, ANDREW 1. JACOBS, SHELLEY STEWART, AND DANIEL D. CARSON* Department of Biochemistry and Molecular Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas 77030

A class of high-affinity binding sites that preferentially bind heparidheparan sulfate have been identified on the external surfaces of mouse uterine epithelial cells

cultured in vitro. [3H]Heparinbinding to these surfaces was time-dependent, saturable, and was blocked specifically by the inclusion of unlabeled heparin or endogenous heparan sulfate in the incubation medium. A variety of other glycosaminoglycans did not compete for these binding sites. The presence of sulfate on heparin influenced, but was not essential for, recognition of the polysaccharide by the cell surface binding sites. [‘HI-Heparin bound to the cell surface was displaceable by unlabeled heparin, but not chondroitin sulfate. Treatment of intact cells o n ice with trypsin markedly reduced [3Hlheparinbinding, indicating that a large fraction of the surface binding sites were associated with proteins. Scatchard analyses revealed a class of externally disposed binding sites for heparidheparan sulfate exhibiting an apparent Kd of approximately 50 nM and present at a level of 1.3 x 10’ sites per cell. Approximately 9-14% of the binding sites were detectable at the apical surface of cells cultured under polarized conditions in vitro. Detachment of cells from the substratum with EDTA stimulated [3H]heparin binding to cell surfaces. These observations suggested that most of the binding sites were basally distributed and were not primarily associated with the extracellular matrix. Collectively, these observations indicate that specific interactionswith heparidheparan sulfate containing molecules can take place at both t h e apical and basal cell surfaces of uterine epithelial cells. This may have important consequences with regard to embryo-uterineand epithelial-basal lamina interactions. A number of molecules with diverse functions have been described that share the property of being able to interact more or less specifically with heparin. The list includes certain secreted proteins (Frazier, 1987; Handin and Cohen, 1976), a variety of growth factors (Klagsburn and Shing, 1984; Lobb and Fett, 1984; Sullivan and Klagsburn, 1984), cell surface components, e.g., N-CAM (Cole and Glaser, 1986; Cole et al., 19861, and extracellular matrix components, e.g., fibronectin (Hayashi and Yamada, 1982; Hynes and Yamada, 1982) and laminin (Skubitz et al., 1988; Otto et al., 1982). In the latter cases, the particular heparinbinding domains have been suggested to be involved with some of the adhesive functions of these proteins. In other instances, it appears that heparintheparan sulfate binding or uptake is correlated with the proliferative state of the cell (Herbert and Maffrand, 1989; Ishihara and Conrad, 1989). Consequently, it appears that protein-heparin (heparan sulfate) interactions could play a n important role in a variety of key biological processes. For convenience, heparin usually has been the polysaccharide used to identify and study heparidheparan sulfate binding properties; however, in most cases, including the present studies, it can be @)

1990 WILEY-I,ISS, INC.

shown that these binding sites recognize certain forms of heparan sulfate as well. Consequently, it is believed that heparan sulfate, not heparin, is the biological ligand. A problem with using commercial heparin preparations is that these are inherently heterogeneous with regard to size and molecular composition. Thus, heparin-binding may reflect interactions with only a subset of the heparin molecules presented. As a consequence, estimates of the K,s for heparin-binding may be high owing to dilution with “unrecognized“ heparin forms. Specific, high-affinity cell surface heparin-binding sites have been reported in various systems (Herbert and Maffrand, 1989; Ishihara and Conrad, 1989; Biswas, 1988; Castellot et al., 1985). Such cell surface heparintheparan sulfate binding sites may play important roles in cellular interactions with heparan sulfate-containing matrices or interactions with heparin-binding growth factors.

Received August 2 . 1989; accepted November 29, 1989. “To whom reprint requestdcorrespondence should be addressed.

HEPARIN BINDING TO EPITHELlAL CELL SURFACES

It appears that heparan sulfate proteoglycans of the cell surface of mouse embryos are involved in embryo adhesion to a variety of substrates including fibronectin, laminin, and uterine epithelial cells, their biological substrate (Farach et al., 1987, 1988). Embryo attachment to these matrices can be inhibited specifically by heparin, but not certain forms of heparan sulfate. These observations indicate that structural elements more prevalent in heparin than certain heparan sulfates are involved in embryo attachment to these substrates. In addition, most epithelial cells normally lie on a basal lamina containing heparan sulfate proteoglycans (Farquhar, 1985). Consequently, it is of interest to determine if binding sites for heparan sulfate exist on both apical and basal surfaces of uterine epithelia where they might participate in embryo-uterine and epithelial-basal lamina interactions, respectively. The results of the present studies demonstrate the existence of specific, high-affinity, externally disposed heparidheparan sulfate binding sites on the cell surface of mouse uterine epithelial cells. Our observations suggest that a significant proportion of these binding sites are apically disposed although most are likely to be basally distributed.

EXPERIMENTAL PROCEDURES Materials CF-1 mice were obtained from Harlan/SpragueDawley (Houston). [3HlHgarin (0.3 mCiimg) was from NEN Products (Boston). [ SlMethionine was obtained from ICN Biomedicals, Inc. (Costa Mesa, CA). Tissue culture media and supplements were from Irvine Scientific (Santa Ana, CA). Urea and guanidine hydrochloride were obtained from Schwarz/Mann Biotech (Cleveland). Heparin, bovine kidney heparan sulfate, chondroitin sulfate, hyaluronic acid, trypsin, CHAPS, heparinase, pepstatin, Triton X-100, and octylglucoside were purchased from Sigma Chemical Co. (St. Louis). Millicell HA chambers were purchased from Millipore Continental Water Systems (Bedford, MA). Matrigel was obtained from Collaborative Research (Lexington, MA). Keratan sulfate and heparinase were from Miles Scientific (Naperville, IL). All chemicals used were reagent grade.

E3H1heparin binding to cell su rfaces Primary cultures of uterine epithelial cells were prepared from random cycling mice as described previously (Dutt et al., 1986). In the routine assay, cell layers were rinsed 4 times with ice-cold phosphate buffered saline (PBS)and then incubated with 200 nM [3H]heparin (0.3mCi/mg) in PBS with 2 mM Ca’. and Mg+ for 30 min on ice. At the end of the incubation period the cell layers were rinsed 4 times with ice-cold PBS, and the cell-associated radioactivity was solubilized with 1 ml of a solution containing 4 M guanidine hydrochloride, 1%(w/v) CHAPS, 50 mM Tris-acetate (pH 6.0), and 25 mM EDTA. Radioactivity in these extracts was determined by liquid scintillation counting. To determine non-specific binding, parallel assays were always performed in which 10 pM unlabeled heparin was included in the incubation medium. The median molecular weight of the [3H]heparin used in these studies was 10,000, as determined by molecular exclu+

+

61

sion chromatography (Tang e t al., 19871, and this value was used to estimate the molarity in all cases. Values for L3H]heparin binding observed in the presence of excess unlabeled heparin (non-specific binding) were subtracted from the values determined in the absence of unlabeled heparin (total binding) to determine the specific binding of [3H]heparin. Time dependence of [3H]heparin binding was examined by varying the incubation time of the routine assay from 1 to 30 min. Heparin dependence was determined by varying the concentration of [3H]heparin in the medium from 10 nM t o 1,000 nM. In all cases cell numbers were determined in parallel cultures by dislodging the cells from the substrate with trypsin and performing hemocytometer counting. To determine whether unlabeled heparin was able to displace bound r3H1heparin, cells were incubated with 200 nM C3H]heparin for 30 min on ice. Unbound label was then removed by washing the cells four times with ice-cold PBS. Cells were then incubated on ice for up to 60 min in medium containing either 20 pM chondroitin sulfate or unlabeled heparin. The amount of [3Hlremaining associated with the cells was determined as described above. 3H-heparin binding to polarized cells Polarized cells were grown on Matrigel-coated Millicell HA chambers (0.45 pm pore size 10 mm inner diameter) as described previously (Glasser e t al., 1988). These cultures typically display a number of markers of functional polarity described in detail by Glasser et al. (1988). [3H]Heparin binding assays were performed by adding the routine assay mixture to the apical (upper) compartment of these wells only. To assay I3H1heparin-binding to detached cells, primary cultures of cells were non-enzymatically detached from tissue culture surfaces by incubation at 37°C in culture medium containing 16.5 mM EDTA (pH 7.4). Cell detachment occurred typically within 15 min. The cells then were washed free of the EDTA solution by centrifugation (600gi5 min) and finally were resuspended in assay buffer. Total binding was assayed i n the presence of 10 pM chondroitin sulfate, while nonspecific binding was assayed in the presence of 10 pM heparin. Cell viability exceeded 95%by Trypan Blue exclusion. Cells were incubated in the routine binding assay buffer on ice for 30 min with gentle mechanical agitation every 5 min. At the end of this period, the unbound [3H]heparin was removed by washing three times by centrifugation (600g/5 min) and resuspension in PBS, and the bound radioactivity was determined in the final pellet by liquid scintillation counting. Competition with glycosaminoglycans and enzyme treatments In order to study the specificity of l3H1heparin binding, 0.1 mg/ml (approximately 10 pM) of various polysaccharides were included in the incubation medium. In addition, chemically modified derivatives of heparin, a gift from Dr. T. Irimura (M.D. Anderson Cancer Center) were included in the incubation medium in another set of experiments. The heparin derivatives were prepared (their characteristics were described in detail previously: Irimura et al., 1986) and included: N-desulfated, N-acetylated heparin, N,O desulfated, N-acetylated heparin, 0-desulfated heparin,

62

WILSON ET AL.

and carboxyl-reduced heparin. More than 85% of the sulfate residues were removed from the “fully desulfated” derivatives. Carboxyl-reduced, N-desulfated, and O-desulfated heparin derivates contained approximately 87%, 71%, and 48%of the sulfate of intact heparin, respectively (Irimura et al., 1986). In some cases, cell layers were incubated with trypsin (50 pg/ml) or heparinase (125 mU/ml) for 30 min in PBS prior to performing the routine assay. The enzyme was removed by rinsing the cell layers 4 times with ice-cold PBS prior to the addition of l3H1heparin. In these experiments, cell viability was examined a t the end of the enzyme incubation by Trypan Blue exclusion. Cell viability exceeded 90% in all cases.

Preparation of endogenous heparan sulfate Heparan sulfate was isolated from mouse uterine epithelial cells as described by Tang et al. (1987). Briefly, primary epithelial cell cultures were prepared and harvested. Proteins were precipiated by adding an equal volume of a mixture containing 20% (w/v)trichloroacetic acid and 6% (w/v)phosphotungstic acid. The precipitate was washed twice with 5 ml of 95% (viv) ethanol, and the pellet was digested with chondroitinase ABC followed by mild alkaline hydrolysis (p-elimination) as described previously (Tang et al., 1987). The samples then were fractionated by anion exchange liquid chromatography on a Mono Q column, and by the heparan sulfate-containing fractions, i.e., material eluting between 2.5 and 4 M sodium chloride, were retained for use. Degradation of this material with nitrous acid was performed as described previously (Tang et al., 1987).

‘48

r

10

20 Minutes

30

Fig. 1. Time dependence of [?H]heparin binding to epithelial cell surfaces. Primary cultures of uterine epithelial cells were prepared in 24-well tissue culture plates as described in “Experimental Procedures.” The cell layers (3.1 x lo5 cells per culture) were rinsed several times with ice-cold phosphate-buffered saline (PBS) and then incubated with 200 nM [3Hlheparin (approximately 3 Cilmmol) on ice in the presence ( 0 ) or absence ( 0 )of 10 p M unlabeled heparin. At the indicated times the cell layers were again rinsed with ice-cold PBS 1.0 remove unbound radioactivity, and the cell-associated radioactivity was solubilized as described in “Experimental Procedures.” The dashed line indicates the values calculated for specific binding by subtracting the values obtained in the presence of excess unlabeled heparin (non-specific) from the values obtained in the absence of the competitor (total). The ordinate indicates the pmol of [’Hlheparin bound per culture.

a class of high-affinity binding sites (Fig. 2B)exhibiting an apparent KD of approximately 50 nM. From these data it also was calculated that there were 2-4 x RESULTS lo5 high-affinity binding sites per cell assuming a 1:l ratio of heparin bound per site. However, as discussed i3H]heparin binding to intact cells below additional binding sites were exposed in cells in One approach toward determining if externally dis- suspension indicating that there were actually about posed, heparidheparan sulfate binding sites existed on 1.3 x lo6 sites per cell (see Table 4). Another class of cultured uterine epithelial cells utilized l3H1heparin in binding sites was observed at higher L3H1heparin conligand binding assays. As shown in Figure 1, centrations; however, the apparent affinity of specific [3H]heparin bound to intact cell layers maintained on binding was a t least 2 orders of magnitude lower in this ice in a time-dependent fashion with maximal binding range, and the number of binding sites exceeded that occurring within 20 to 30 min after ligand addition. described above by approximately ten-fold (data not Moreover, binding of radioactive heparin was reduced shown). We chose to focus our attention on the highgreatly if the incubations were performed in the pres- affinity binding sites. Consequently, 200 nM ence of excess unlabeled heparin. In the following stud- [3H]heparin was used in our routine ligand binding ies we defined specific binding as the difference be- assays. We next examined whether l3H1heparin binding to tween the values determined for [3H]heparin binding in the absence (total) and the presence (non-specific)of these sites was reversible. Unlabeled heparin and chondroitin sulfate were compared for their respective excess unlabeled heparin. When the concentration dependence of [3Hlheparin abilities to displace [3Hl-heparinbound to intact mouse binding was examined, it was found that a plateau of epithelial cell layers. As shown in Figure 3 heparin specific binding was reached at approximately 300 nM stimulated a rapid release of the labeled polysaccha(Fig. 2). Overall values for specific binding ranged from ride. In contrast, the bound l3H1heparin was released 40 to 140 pmol per 10’ cells but usually were on the much more slowly in the presence of chondroitin su’lorder of 40 to 80 pmol/lOB cells. Duplicate assays fate alone. Consequently, it appeared that binding to within an experiment never varied by more than 5- the cell layers was reversible and preferentially dis10% (see Table 1). Variations in the total binding that placeable by the appropriate competitor ligand. were observed between experiments may reflect differSpecificity of [3H]heparin binding ences in the hormonal status, i.e., estrus, diestrus, etc., A series of studies was performed to examine the of the mice from which the primary cell cultures were derived or cases where cells were polarized to different specificity of [3H]heparin binding to epithelial cell layextents (see below). The values determined for specific ers. In the first series, a variety of anionic polysacchabinding in several experiments were used for Scatch- rides were examined for their ability to compete for ard analyses. These analyses revealed the existence of [3H]heparin binding. As shown in Table 1, although

63

HEPARIN BINDING TO EPITHELIAL CELL SURFACES

A 0

,010

R‘

lo

B

0

0 +cs (0.1 8 tliep(0.1

tcs (0.1

.008

siteslcell

a a

P

e

.004

m

75

a

Y

* 0

0.2

*

* 0

0.4

.2 .4 .6 .8

[Bound Heparin], nM Fig. 2. Heparin concentration dependence and Scatchard analyses of binding to epithelial cell surfaces. r3HIHeparin binding assays were performed on ice for 30 min as described in “Experimental Procedures.” The concentration of [3Hlheparin was vaned as indicated. The binding of radiolabeled ligand observed in the presence (*) and absence ( 0 ) of 10 pM heparin are presented in A. The dashed line indicates the values calculated for specific binding as described in the text. Each well contained approximately 1.2 X lo6 cells. The ordinate indicates the pmol of t3H]heparin bound per well. The graph shows the results of a representative experiment. B shows the Scatchard analyses of the data obtained from several experiments like the one shown in A using the values calculated for specific binding. From these data it was calculated that the dissociation constant of binding for the highest-affinity class of binding sites was approximately 47 nM and that there were 2-4 x lo5 high-affinity binding sites for heparin per cell. The ordinate in B shows the ratio of the [3Hlheparin bound to the cells (bound) divided by that left in the medium (free). The abscissa indicates the concentration of bound [’Hlheparin in the 200 p1 assays. Approximately 2.6-2.7 x lo5 cells were in each assay.

TABLE 1. Competition by various glycosaminoglycans for [3Hlheparin binding to epithelial cell surfaces’ Treatment (competing elvcosaminoelvcan) None Heoarin Heparan sulfate (uterine epithelial cell) Heparan sulfate (bovine kidney) Chondroitin sulfate Hyaluronate Keratan sulfate

pmol [3Hlhegarin bound uer 10 cells 38 2 3.5 4 3 1.5 7 2 0.6 38 35 34 46

2 2 2 ?

2.5 1.5 1.5 0.5

8 of control bindine 100 11 18 100 92 89 121

‘Primary cultures of uterine epithelial cells (4.0 x lo5 cells per culture) were prepared and [3H]heparin-bindingassays performed using 200 nM 13Hlheparin at 4°C for 20 min as described in “Experimental Procedures.”The incubation medium used in the binding assays contained 0.1 mgiml (approximately10 kM1 of the indicated polysaccharide.

heparin itself and heparan sulfate endogenous to uterine epithelial cells effectively blocked ligand binding, other polysaccharides tested failed to inhibit this process significantly at 0.1 mg/ml (approximately 10 pM). In addition, higher concentrations (1 mg/ml) of chondroitin sulfate, hyaluronate, keratan sulfate, and bovine kidney heparan sulfate failed to compete for [3H]heparin-binding (data not shown). Experiments also were done to assess the relative ability of heparan sulfate endogenous to the uterine epithelial cells (Tang et al., 1987) to compete for binding of L3H1heparin. As shown in Figure 4, endogenous heparan sulfate reduced r3H]heparinbinding in a concentration-dependent manner. Levels of [3H]heparinbinding were reduced to the

Minutes

Fig. 3. Displacement of L3H]heparin from epithelial cells by addition of heparin. Cells were allowed to bind r3HIheparin (200 nM) for 30 min on ice, as described in “Experimental Procedures.” Cultures were then incubated in PBS containing 0.1 mg/ml (10 r M ) of chondroitin sulfate in the presence (*) or absence ( 0 )of 0.1 mglml(10 FM) heparin on ice. The amount of label associated with the cells at the indicated times was determined as described in “Materials and Methods” and are expressed as a percentage of the zero time values. The zero time value for binding was 62 t 3 pmolil0’ cells.

background levels observed using commercial heparin as competitor. Digestion of the heparan sulfate preparations with nitrous acid greatly reduced their inhibitory activity indicating that it was the heparan sulfate components of these fractions that inhibited binding. Near-maximal competition was observed when 19 pg of endogenous heparan sulfate was added to the 200 ~1 assays. Most of the endogenous heparan sulfate chains had Mr of approximately 10,000 (Tang et al., 1987). Consequently, the endogenous heparan sulfate at approximately 8.5 pM competed nearly as effectively as 10 pM heparin. The dose response obtained using commercial heparin as competitor was similar to that obtained using endogenous heparan sulfate (data not shown). Interestingly, although endogenous heparan sulfate inhibited L3H1heparinbinding as efficiently as heparin, bovine kidney heparan sulfate failed to do so. Bovine kidney heparan sulfates differ compositionally from heparin and heparan sulfates from a variety of other cellular sources, e.g., 0-sulfate content, iduronate content (Gallagher and Walker, 1985). Thus, binding appeared to be dependent on sequences found in commercial heparin and certain forms of heparan sulfate. To extend these observations, several chemically modified heparin derivatives were compared for their ability to act as competitors for the binding of [3H]heparin to intact cells (Table 2). Removal of either N- or 0-sulfate groups markedly reduced the ability of heparin to compete for cell surface binding sites; however, removal of both N- and 0-sulfates consistently restored this activity. Removal of N-sulfates without subsequent reacetylation resulted in a derivative that failed to compete for binding in these assays (data not shown). These obser-

64

WILSON ET AL 0.3

TABLE 2. Ability of chemically modified heparins to compete for ['Hlheparin binding to mouse uterine epithelial cell surfaces

0.2

Heparin derivative' None Heparin N-desulfated, N-acetylated 0-desulfated N, 0-desulfated, N-acetylated Carboxyl reduced

U 3

m"

f

.-C L Cl

Tr

k

Q of control

bindine 100 10 104 81 9 56 -

'L'HlHeparin-binding assays were performed as described in "Experimental Procedures." Each well contained approximately 3.5 x lo5 cells, and the incubation medium contained 0.1 mg/ni1 (approximately 10 p M ) of the specified heparin derivatives. Derivatives were prepared as described (Irimura e t a]., 1986).

m

0

pmoles c3H1heparin bound Der 10' cells 74.5 i 7.8 7.4 i 0.5 77.0 2 0.1 60.3 '-+ 0.2 6.8 i 0.5 41.7 i 0.2

0.1

TABLE 3. ['Hlheparin binding following digestion of epithelial cell surfaces wlth trvpsin or heuarinase

0

10

20

pg Heparan Sulfate

Enzyme used' None Trypsin Heparinase

pmoles bound per 10' cells 52 ? 2.0 26 -t 0.6 59 ? 3.0

% of control

binding 100 50 113

Fig. 4. Endogenous heparan sulfate inhibits binding of L3H1heparin to epithelial cell surface. Heparan sulfate was prepared from primary cultures of uterine epithelial cells as described in "Experimental Procedures." For [3Hl-heparin binding assays, primary cell cultures were prepared in 24-well tissue culture plates (approximately 4 x lo5 cells/well). Cell layers were rinsed several times with PBS and r3H]heparin-binding assays were performed in 0.2 ml volumes for 30 min as described in Figure 1. The points indicate the averages and ranges of duplicate determinations in each case. The ordinate indicates the pmol of L3H]heparin bound per well. Symbols: 0,heparan sulfate; 0 , nitrous acid-digested heparan sulfate; ---,binding observed in the presence of 10 FM heparin.

'In these experiments epithelial cell cultures (6.5 x lo5 cells per culture1 were preincubated with either 50 pgiml of trypsin or 125 mU/ml of heparinase for 30 min at 4°C prior to initiation of the binding assays. The data presented in this case reflect the values determined in duplicate determinations for specific binding, i.e., the difference in the binding observed in the absence and presence of 10 p M unlabeled heparin.

vations indicated that sulfate residues were not required for binding but appeared to influence expression or conformation of the recognized structural elements. It also was found that reduction of the carboxyl groups of heparin led to a marked decrease in inhibitory activity. It was concluded that both sulfation and carboxylation patterns strongly influenced recognition of heparidheparan sulfate by the cell surface binding sites.

'Polarized uterine epithelial cell cultures were prepared as described by Glasser et al. (19881 using mature, randomly-cycling mice. ['HIHeparin binding to the apical surfaces of these cells was measured using the routine assay conditions. Cells were non-enzymatically detached from tissue culture plates using 16.5 mM EDTA as described in "Experimental Procedures." Values for non-specific binding were subtracted from each determination. The range of values for attached, non-polarized cells were taken from the values reported in the other figures and tables in this manuscript.

TABLE 4. P'Hlheparin binding to surfaces of polarized or non-enzymatically detached cells ~

Attached, non-polarized Polarized cells, apical side Detached cells

~~~~~

prnol boundil0" cellsL 34 - 67 25 -t 6 217 -t 12

marily due to binding to HS endogenous to these cells Enzymatic susceptibility of "-heparin (Tang et al., 1987). Collectively, these data suggested binding sites that a large fraction of the cell surface binding sites for In another series of experiments, intact cell layers heparinheparan sulfate were associated with proteins. were predigested with trypsin or heparinase prior to the ligand binding assays to determine the effects on Cell surface distribution of [3H]heparin binding to the cell surface. As shown in 3H-heparin-binding sites Table 3, trypsin treatment reduced [3Hlheparin bindTwo types of experiments were performed to detering by 50% indicating that a large fraction of the binding sites were associated with proteins with accessible mine if the heparin-binding sites were likely to be astryptic sites. If a battery of protease inhibitors (Dutt et sociated with apical versus basolateral cell surfaces. In al., 1986) were included in the trypsinization mixture, the first experiment, cells were grown under conditions no reduction in binding was observed (data not shown). in which they expressed a polarized phenotype, i.e., This indicated that it was the proteolytic action of the distinct apical versus basolateral plasma membrane trypsin that was responsible for the decrease in cell domains, and functions (Glasser et al., 1988). In this surface binding sites. Cells also recovered their full case, extracellular matrix components also are excomplement of [3H]heparin-binding sites within a few pected to be deposited on the basal aspect of these cells. hours following trypsinization (data not shown). Hep- [3H]Heparin binding to the apical surfaces of the pmoarinase had no effect on expression of cell surface larized cells was found to be comparable to the lower [3H]heparin-binding sites. Approximately 50% of the range of values observed for L3H1heparin binding to cell surface 35S04-labeled material releasable by cells grown on plastic (Table 4). In the second experitrypsin was releasable by heparinase (data by shown). ment, epithelial cells were separated from underlying Therefore, it did not appear that [3H]heparin was pri- extracellular matrix by detaching the cells with EDTA.

HEPARIN BINDING TO EPITHELIAL CELL SURFACES

The cells were washed and used in heparin-binding assays. Approximately an eight-fold increase in the number of heparin-binding sites was observed in the cells in suspension as were found on the apical surface of polarized cells. The average number of total binding sites found on detached non-polarized cells (217 pmoll 10' cells) was similar to the number found for detached polarized cells (196 pmoYlO' cells). Collectively, these observations suggested that the heparin-binding sites were associated with firmly attached, cell surface components, and not extracellular matrix components deposited by the cells. Furthermore, most of these binding sites appeared to be basally distributed.

DISCUSSION An intriguing example of cell adhesion is embryo implantation into the uterine wall. In this case, the development of the embryo must be coordinated with the hormonal regulation of uterine receptivity (Finn and Martin, 1974; Psychoyos, 1973). In situations in which this coordination is disturbed, implantation occurs with a very low frequency (Smith and Biggers, 1968). Furthermore, it appears that regulation of the initial phase of embryo attachment takes place at the apical surface of the epithelial cells that line the uterine wall (Salomon and Sherman, 1975; Nilsson, 1967). Consequently, it is critical to identify the components of the epithelial cell surface that take part in embryo interactions if we are to understand how implantation is controlled at a molecular level. Studies utilizing mouse blastocyst stage embryos indicated that heparidheparan sulfate proteoglycans of the embryo cell surface are involved in adhesive interactions with a variety of substrata including primary cultures of uterine epithelial cells (Farach et al., 1987, 1988). Therefore, we undertook the present studies to determine if the cell surface of mouse uterine epithelial cells displayed heparinlheparan sulfate binding sites. We found that uterine epithelial cells expressed at their external surface a class of specific, high-affinity binding sites recognizing either heparin or certain forms of heparan sulfate. The heparin-binding sites displayed an apparent KD similar to those of other heparin-binding molecules such as laminin (Skubitz et al., 1988) and fibronectin (Tarsi0 et al., 1987). Interestingly, the heparidheparan sulfate-binding sites displayed a similar specificity as mouse embryo attachment t o uterine epithelial cells with regard to the ability to interact with heparin, but not certain preparations of heparan sulfate (Farach et al., 1987). These observations not only underscore the specificity of both interactions, but also indicate that both interactions involve structural elements prevalent in heparin. In this regard, previous studies have indicated that heparan sulfate of mouse embryos has heparin-like characteristics (Farach et al., 1987). Heparan sulfate from different cellular origins has been shown to vary in the content of O-sulfate groups and iduronic acid (Gallagher and Walker, 1985; Zimina et al., 1987). For example, bovine kidney heparan sulfate exhibits substantially less O-sulfate groups than heparan sulfate obtained from other cell types or heparin (Gallagher and Walker, 1985).Bovine kidney heparan sulfate also does not compete for either embryo attachment (Farach et al., 1987) or [3Hl-heparin binding to uterine epithe-

65

lial cell surfaces. A large fraction of the cell surface I3H]heparin-binding sites were sensitive to trypsin, but not heparinase. Thus, the binding appeared to be associated with cell surface proteins and not due to heparan sulfate-heparan sulfate interactions (Fransson et al., 1980). Consistent with these observations, we have found that uterine epithelial cells express proteins at their cell surface with externally disposed heparinbinding domains (Wilson, Stewart, and Carson, manuscript submitted). Furthermore, Lankes et al. (1988) recently isolated a 78,000 Mr protein from bovine uteri that also binds only certain forms of heparan sulfate. However, it is not clear if any of these proteins are related to the cell surface heparan sulfate-binding sites studied here. We found that removal of either O-sulfate or N-sulfate groups markedly reduced the ability of heparin to compete for L3H]heparin-bindingsites on the cell surface. Remarkably, removal of both N- and O-sulfate groups restored the capacity of heparin to compete for these binding sites. These observations suggest that sulfate is not recognized per se, but may influence expression of recognized sequences. Reduction of carboxyl groups also lead to a loss of inhibitory activity. Consequently, it appears that both sulfation and disposition of carboxyl groups influence heparin-binding in a complex fashion. Casu et al. (1988) have suggested that certain conformations of iduronate-containing polysaccharides are recognized by their respective binding proteins. Their data indicates that particular conformations of iduronate are strongly influenced by the positioning of nearby sulfate groups, although several different types of substitutions can produce these conformations. It appears that the heparidheparan sulfate receptors of the uterine epithelial cells also display a specificity more subtle than the simple presence or absence of particular charged moieties, perhaps involving polysaccharide conformation. Detailed examination of these structural requirements is currently underway. The maximum number of heparin-binding sites was observed with intact cells in suspension and was considered to represent the total number of cell surface binding sites. From these data it aptears that there is a total of approximately 1.3 x 10 heparan sulfatebinding sites per cell. Only 16-31% of the total cell surface binding sites were accessible in cells attached to tissue culture plastic. Fewer binding sites (9-14% of the total) were accessible at the apical cell surface of polarized cell cultures. Thus, it appears that more than just the apical complement of binding sites are exposed by cells on tissue culture plastic. There may be partial access to basolaterally disposed binding sites on subconfluent cells cultured on plastic. In addition, the cell surfaces of these cells may be incompletely segregated into apical and basolateral domains owing to their culture under non-polarizing conditions. Therefore, greater variability observed in the number of binding sites on cells cultured on tissue culture plastic may reflect variations in the basal accessibility and/or degree of polarity attained in different cell preparations. It is unlikely that the heparinlheparan sulfate binding sites are primarily associated with extracellular matrix or peripherally associated components. In the polarized cells, most extracellular matrix components

66

WILSON ET AL.

Uterine Epithelium

z

Basal

Lamina

Fig. 5. Model of heparan sulfate-dependent interactions a t the cell surface of uterine epithelial cells. The model depicts uterine epithelial cells with apically disposed heparan sulfate binding sites ( y ) interacting with heparan sulfate proteoglycans (HSPG;? ) at the cell surface of trophectodermal cells of the blastocyst. At the basal aspect of the uterine epithelium are additional heparan sulfate binding sites which interact with heparan sulfate proteoglycans in the underlying basal lamina. For further details see text.

would be expected to be deposited basally (Farquhar, 1985). In the cells in suspension, most of the extracellular matrix or peripheral components would be expected to be left either on the tissue culture substratum or lost during the washing procedure. The observation that additional binding sites were exposed in the cells in suspension indicates that these sites are not associated with extracellular matrix components. Since the matrix upon which the cells are cultured contains a large amount of heparin-binding components itself, e.g., laminin, we were unable to directly measure basally disposed heparin-binding sites. Nonetheless, i t seems likely that most of the additional sites exposed on the cells in suspension are basally distributed where they may interact with the heparan-sulfate proteoglycans produced by these cells and found in their basal lamina (Tang et al., 1987). It must also be considered that EDTA treatment may release endogenous heparan sulfate bound to cell surface sites and expose “blocked” receptors. In fact, endogenous heparan sulfate proved to be a n effective competitor for L3Hlheparin binding. Pretreatment of cells with heparinase failed to stimulate cell surface binding of 13H1heparin; however, a substantial fraction of cell surface (trypsinreleasable) heparan sulfate was not released by heparinase. This latter pool of cell surface heparin sulfate may be bound to proteins that protect these polysaccharides from heparinase action. As shown in Figure 5, the apically-disposed heparin1 heparan sulfate binding sites may function, in part, to promote attachment of the trophectodermal cells of the embryo. Basally distributed binding sites may promote adhesion to the underlying basal lamina. It is clear that the ability of the epithelium to support embryo attachment is strictly controlled by steroid hormones (Psychoyos, 1973; Finn and Martin, 1974). It is intersting to note that a s the uterine epithelium becomes receptive for embryo attachment these cells loosen their contact with the basal lamina (Sherman and Wudl, 1976). Consequently, it is possible that steroid

hormones induce a loss or rearrangement of basally disposed binding sites to the apical surface to produce a receptive interface with the embryo. Consistent with this are Parr’s (1980, 1982) observations of increased basal-to-apical membrane trafficking during the receptive uterine phase. It also appears that some of the heparan sulfate proteoglycans of uterine epithelia are expressed by the apical surface (Tang et al., 1987; Mor- ris et al., 1988a; Carson et al., 1988) and that the turnover of these molecules is stimulated by estrogen (Morris, et al., 198813). Thus, metabolic loss of apically disposed proteoglycan may make additional heparan sulfate-binding sites available a t this cell surface and promote embryo attachment. Further studies must examine the precise structural requirements for heparinlheparan sulfate binding to cell surface binding sites. It will be of interest to determine if the ability of heparan sulfate structures to compete for [3Hlheparinbinding sites in uterine epithelial cells is paralleled by their ability to inhibit embryo adhesion to epithelial cell layers. It will be necessary to isolate and develop specific antibodies to cell surface heparan sulfate receptors in order to definitvely study their expression and function. As a result, we should be able to determine how expression of such proteins is regulated within uteri and other tissues.

ACKNOWLEDGMENTS We thank Drs. A. Dutt, M.C. Farach-Carson, N. Raboudi, Ms. J.-P. Tang, and Ms. J. Laidlaw for their critical reading of this manuscript and many helpful discussions. We also are indebted to Dr. T. Irimura (M.D. Anderson Cancer Center) for providing the various heparin derivaties. We acknowledge the excellent preparation of this manuscript by Ms. Ellen Madson and Mrs. Harriette Young. This work was supported by a n American Cancer Society grant (BC-503) and a n NIH grant (HD 25235) awarded to D.D.C. O.W. was supported by funds from the National Institutes of Health training grant (HD 07325). LITERATURE CITED Biswas, C. (1988)Heparin and heparan sulfate binding sites on B-16 melanoma cells. J. Cell. Physiol., 136t147-153. Carson, D.D., Tang, J.-P., Julian, J . and Glasser, S.R. (1988) Vectorial secretion of proteoglycans by polarized rat uterine epithelial cells. J. Cell Biol. 107:2425-2435. Castellot, J.J., Jr., Wong, K., Herman, B., Hoover, R.L., Albertini, D.F., Wright, T.C., Caleb, B.L.,andKarnovsky, M.J. (1985)Binding and internalization of heparin by vascular smooth muscle cells. J. Cell. Physiol., 124t13-20. Casu, B., Petitou, M., Prouasoli, M., and Sinay, P. (1988) Conformational flexibility: A new concept for explaining binding and biological properties of iduronic acid-containing glycosaminoglycans. TIBS, 13:221-225. Cole, G.J., and Glaser, L. (1986) A heparin-binding domain from NCAM is involved in neural cell-substration adhesion. J. Cell Biol., 102:403-412. Cole, G.J., Loewy, A., and Glaser, L. (1986) Neuronal cell-cell adhesion depends on interactions of N-CAM with heparin-like molecules. Nature, 320:445-447. Dutt, A., Tang, J.-P., Welply, J.K., and Carson, D.D. (1986) Regulation of N-linked glycoprotein assembly in uteri by steroid hormones. Endocrinology 118:661-673.0 Farach, M.C., Tang, J.-P.,Decker, G.,and Carson, D.D. (1987)Heparin/ heparan sulfate is involved in attachment and spreading of mouse embryos in vitro. Dev. Biol., 123:401-410. Farach, M.C., Tang, J.-P., Decker, G., and Carson, D.D. (1988) Differential effects of p-nitrophenyl-D-Xylosideson mouse blastocysts and uterine epithelial cells. Biol. Reprod., 39:443-455.

. ,

HEPARIN BINDING TO EPITHELIAL CELL SURFACES Farquhar, M.G. (1985) The glomerular basement membrane: A selective macromolecular filter. In: Cell Biology of the Extracellular Matrix. E.D. Hay, ed. Plenum Press, New York, pp. 335-378. Finn, C.A., and Martin, L. (1974) The control of implantation. J. Reprod. Fertil., 39:195-206. Fransson, L.A., Havsmark, B., Niedzynski, LA., and Huckerby, T.N. (1980) Interaction between heparan sulfate chains. 11. Structural characterization of iduronate- and glucuronate-containing sequences in aggregating chains. Bioehim. Biophys. Acta, 633.95-104. Frazier, W.A. (1987) Thrombospondin: A modular adhesive glycoprotein of platelets and nucleated cells. J . Cell Biol., 105t625-632. Gallagher, J.T., and Walker, A. (1985) Molecular distinctions between heparan sulfate and heparin. Biochem. J., 2 3 0 5 6 5 4 7 4 . Glasser, S.R., Julian, J., Decker, G.L., Tang, J.-P., and Carson, D.D. (1988)Development of morphological and functional polarity in primary cultures of immature rat uterine epithelial cells. J. Cell Biol., 107:2409-2423. Handin, R.I., and Cohen, J.H. (1976) Purification and binding properties of human Dlatelet factor four. J. Biol. Chem.. 251t4273-4282. Hayashi, M., and Yarnada, K.M. (1982) Divalent cation and modulation of fibronectin binding to heparin and to DNA. J. Biol. Chem., 257:5263-5267. Herbert, J.-M., and Maffrand, J.-P. (1989) Heparin interactions with cultured human vascular endotheial and smooth muscle cells: Incidence on vascular smooth muscle cell proliferation. J . Cell. Physiol., 138:424-432. Hynes, R.O., and Yamada, K.M. (1982) Fibronectins: Multifunctional modular glycoproteins. J . Cell Biol., 95:369-377. Irimura, T., Nakajima, M., and Nicholson, G.L. (1986) Chemically modified heparins as inhibitors of heparan sulfate specific endoP-glucuronidase (heparanase) of metastatic melanoma cells. Biochemistry, 255322-5328. Ishihara, M., and Conrad, H.E. (1989) Correlations between heparan sulfate metabolism and hepatoma growth. J . Cell. Physiol., 138: 467-476. Klagsburn, M., and Shing, Y. (1984) Heparin affinity of anionic and cationic capillary endothelial cell growth factors: Analysis of hypothalamus-derived growth factors and fibroblast growth factors. Proc. Natl. Acad. Sci. U.S.A., 82305-809. Lankes, W., Griesmacher, A,, Grunwald, J., Schwartz, A., Albiez, R., and Keller, R. (1988) A heparin-binding proteins involved in inhibition of smooth muscle cell proliferation. Biochem. J., 251 :831842. Lobb, R.R., and Fett, J.W. (1984) Purification of two distinct growth factors from bovine neural tissue by heparin-affinity chromatography. Biochemistry, 2333295-6299.

67

Morris, J.E., Potter, S.W., and Gaza-Bulesco, G. (1988a) Estradiolstimulated turnover of heparan sulfate proteoglycan in mouse uterine epithelium. J. Biol. Chem., 263:4712-4718. Morris, J.E., Potter, S.W., and Gaza-Bulesco, G. (1988b) Estradiol induces an accumulation of free heparan sulfate glycosaminoglycans in uterine epithelium. Endocrinology, 122:242-253. Nilsson, 0. (1967)Attachment of rat and mouse blastocysts onto uterine epithelium. Int. J. Fertil., 125-13. Otto, U., Odermatt, E., Engel, J . , Furthmayr, H., andTimpl, R. (1982) Protease resistance and conformation of laminin. Eur. J. Biochem., 123.53-72. Parr, M.B. (1980) Endocytosis at the basal and lateral membranes of rat uterine epithelial cells during early pregnancy. J. Reprod. Fertil., 60:95-99. Parr, M. (1982) Apical vesicles in the rat uterine epithelium during early pregnancy: A morphometric study. Biol. Reprod., 26:915-924. Psychoyos, A. (1973) Hormonal control of ovoimplantation. Vitam. Horm., 31.201-256. Salomon, D.S., and Sherman, M.I. (1975) Implantation and invasiveness of mouse blastocysts on uterine monolayers. Exp. Cell Res., 90r26 1-268. Sherman, M.I., and Wudl, L.R. (1976) The implanting mouse blastocyst. In: The Cell Surface in Animal Embryogenesis and Development. G. Poste and G.L. Nicolson, eds. ElsevieriNorth Holland Biomedical, Amsterdam, pp. 81-125. Skubitz, A.D.N., McCarthy, B.J., Charonis, A.S., and Furcht, L.T. (1988) Localization of three distinct heparin-binding domains of laminin by monoclonal antibodies. J . Biol. Chem., 263:4861-4868. Smith, D.M., and Biggers, J.D.(1968) The estrogen requirement for implantation and the effect of its dose on the implantation response in the mouse. J. Endocrinol., 41:l-9. Sullivan, R., and Klagsburn, M. (1984) Purification of cartilage-derived growth factor by heparin afinity chromatography. J. Biol. Chem., 260.2399-2403. Tang. J.-P., Julian, J., Glasser, S.R., and Carson, D.D. (1987) Heparan sulfate proteoglycan synthesis and metabolism by mouse uterine epithelial cells cultured in vitro. J. Biol. Chem., 262: 12832-1 2842. Tarsio, J.F., Reger, A.L., and Furcht, L.T. (1987) Decreased interaction of fibronectin, type IV collagen and heparin due to nonenzymatic glycation. Implications for diabetes mellitus. Biochemistry, 26r1014-1020. Zimina, N.P., Dmitriev, LP., and Rykova, V.I. (1987) Composition and degree of sulfation of glycosaminoglycans from the tissues of animals of different species: Heterogeneity and tissue specificity of heparan sulfates. Biokhimiia, 52:848-854.