Platelet-derived growth factor receptors form a high affinity state in membrane preparations. Kinetics and affinity cross-linking studies.

The specific binding of 125I-PDGF (platelet-derived growth factor) to intact fibroblasts becomes relatively nondissociable during incubation at 37 degrees C. To characterize the interaction of PDGF with its receptors under conditions in which there is no receptor internalization, we have studied the binding of 125I-PDGF to membrane preparations derived from mouse 3T3 cells and rat liver. The binding sites had the affinity and specificity characteristics expected of PDGF receptors. At 37 degrees C (but not at 4 degrees C) the specific binding of 125I-PDGF to membranes gradually became nondissociable as assessed by either dilution or by addition of excess unlabeled PDGF. This tight binding was not due to a covalent interaction since the polyanionic compound suramin readily dissociated specifically bound 125I-PDGF. This property of suramin was used to expose rat liver PDGF receptors which were occupied by endogenous PDGF. Affinity cross-linking studies demonstrated that the formation of the nondissociable state of 125I-PDGF binding was associated with the binding of 125I-PDGF to a 160,000-dalton protein and to a 110,000-dalton species. The cross-linked binding sites could be adsorbed to wheat germ agglutinin and to anion exchange resins. The isoelectric point of both cross-linked species determined by two-dimensional gel electrophoresis was approximately 4.7. These data demonstrate that in membrane preparations, PDGF binds to an anionic 160,000-dalton glycoprotein which is likely to be the receptor. A high affinity state of PDGF binding, which is formed rapidly at 37 degrees C, can be dissociated by suramin.

The specific binding of lZ5I-PDGF (platelet-derived growth factor) to intact fibroblasts becomes relatively nondissociable during incubation at 37 "C. To characterize the interaction of PDGF with its receptors under conditions in which there is no receptor internalization, we have studied the binding of 'T-PDGF to membrane preparations derived from mouse 3T3 cells and rat liver. The binding sites had the affinity and specificity characteristics expected of PDGF receptors. At 37 "C (but not at 4 "C) the specific binding of lZ5I-PDGF to membranes gradually became nondissociable as assessed by either dilution or by addition of excess unlabeled PDGF. This tight binding was not due to a covalent interaction since the polyanionic compound suramin readily dissociated specifically bound '"I-PDGF. This property of suramin was used to expose rat liver PDGF receptors which were occupied by endogenous PDGF.
Affinity cross-linking studies demonstrated that the formation of the nondissociable state of '"I-PDGF binding was associated with the binding of '251-PDGF to a 160,000-dalton protein and to a 110,000-dalton species. The cross-linked binding sites could be adsorbed to wheat germ agglutinin and to anion exchange resins. The isoelectric point of both cross-linked species determined by two-dimensional gel electrophoresis was approximately 4.7. These data demonstrate that in membrane preparations, PDGF binds to an anionic 160,000-dalton glycoprotein which is likely to be the receptor. A high affinity state of PDGF binding, which is formed rapidly at 37 " C , can be dissociated by suramin.
PDGF,' a 32,000-dalton mitogen isolated from human platelets, stimulates the growth in uitro of fibroblasts, smooth muscle cells, and glial cells. The mitogenic response to PDGF is mediated by the interaction of PDGF with specific high * This work was supported by research grants from the National Institutes of Health (HL-29679-01) and the American Heart Association . 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.
$ TO whom all correspondence should be addressed at the University of California. A portion of this work was done during the tenure of an American Heart Association Clinician Scientist Award with funds contributed by the Massachusetts Affiliate.
' The abbreviations used are: PDGF, platelet derived growth factor; DST, disuccinimidyl tartarate; DSS, disuccinimidyl suberate; PMSF, phenylmethylsulfonyl fluoride; SDS, sodium dodecyl sulfate. affinity cell surface receptor sites which in turn activate intracellular events by a process involving tyrosine kinase stimulation (I-3), receptor internalization (4), alteration in ion flux (5), or other as yet unidentified reactions. Recently, receptors for PDGF have been labeled on cultured fibroblasts (6-9), glial cells (6), and vascular smooth muscle cells (8,9) by measuring the binding of "'1-PDGF to intact cells (6-9). These high affinity receptor sites have a well defined specificity and are clearly distinct from receptors for other known growth factors (6-9). Kinetic studies using intact cells have demonstrated that the binding of V P D G F is initially reversible, but rapidly becomes relatively nondissociable (9), Whether this nondissociable state is entirely attributable to internalization of the ligand or is due to the formation of a high affinity state of binding of the ligand cannot be elucidated in intact cell systems. In the current study we have examined the binding of PDGF to its receptor sites in membrane preparations in order to define the kinetic and equilibrium characteristics of the binding reaction in the absence of complicating reactions such as receptor internalization and metabolism. These studies demonstrate that in isolated membranes, as in whole cells, a slowly reversible high affinity state of binding is formed at 37 but not at 4 "C. The finding that the polyanionic compound suramin reversibly dissociates PDGF from its high affinity receptor sites provides a method for unmasking receptors occupied by endogenous PDGF, thus making possible the study of receptors in membranes from intact tissues. Finally, we have used affinity cross-linking techniques to determine the physical characteristics of the high affinity state of the receptor formed in these membrane preparations.

EXPERIMENTAL PROCEDURES
Material~"Na"~1 was obtained from New England Nuclear; disuccinimidyl suberate, disuccinimidyl tartarate, and Iodogen were obtained from Pierce Chemical Co. BALB/c 3T3 cells (ATCC CCL 163) were obtained from American Type Culture Collection, Rockville, Maryland. These cells were grown according to the supplier's specifications. High molecular weight standards for SDS-polyacrylamide gel electrophoresis were obtained from Bio-Rad. Cellulose acetate filters (EAWP) were obtained from Millipore Corp., Bedford, Massachusetts. PDGF was purified from outdated human platelets as previously described (10,11).
Radioiodination of PDGF-Purified PDGF (5-10 pg) was iodinated by the Iodogen method as previously described (9). The purity of the iodinated product was documented by SDS-polyacrylamide gel electrophoresis.
Membrane Preparations-Balblc 3T3 cells were plated in 10% fetal calf serum a t a density of 4 X lo5 cells/ml and were grown to confluence without a change of media for 5-7 days. The cells were then removed from flasks using phosphate-buffered saline (137 mM sodium chloride, 3 mM potassium chloride, 2 mM Na2HPOI, pH 7.4) containing 0.05% EDTA. The lysates were centrifuged a t 50,000 X g a t 4 "C for 20 min and the pellets were resuspended in phosphate-buffered saline with 2 mM EDTA and 2 mM PMSF using a Potter-Elvehjem homogenizer. After a low speed fraction (300 X g for 10 min) of intact cells and debris was discarded, the supernatant was centrifuged at 50,000 X g at 4 "C for 20 min. The final pellets were resuspended a t a protein concentration of 2-3 mg/ml in 0.25 M sucrose, 30 mM histidine, pH 7.4, and were frozen and stored a t -70 "C. The membrane preparations were devoid of intact cells (assessed microscopically) and contained PDGFstimulated tyrosine kinase activity using [Y-~'P]ATP as a substrate (not shown).
For preparations of rat liver membranes, female Sprague-Dawley CD rats were obtained from Charles River Laboratories. Rats were starved for 20 h prior to being anesthesized with ether. Livers were removed from viable animals, were placed into ice-cold buffer (0.25 M sucrose, 10 mM Tris, pH 7.4, 5 mM EDTA, 25 mM benzamidine, 2 mM leupeptin, 100 kallikrein units/ml of aprotinin, 2 mM PMSF) at a ratio of 1 g of liver/lO ml of buffer, and were homogenized with 3 X 5-s bursts using a polytron homogenizer (setting 11). Suramin (usually 1 mM) was included in the homogenization buffer as indicated in the figure legends. Liver homogenates were spun a t 4 "C in a Sorvall SS-34 rotor for 15 min a t 3,000 rpm. Pellets were discarded and supernatants centrifuged at 4 "C in an SS-34 rotor for 20 min a t 19,000 rpm. The resulting clear pink supernatant was aspirated, and the soft tan-white interface gently resuspended by light vortexing in ice-cold homogenization buffer (1 ml of buffer for every gram of starting liver). These suspensions were centrifuged a t 19,000 rpm for 20 min. The resulting pellets were washed three times a t 4 "C in suramin-free homogenizing buffer, each time resuspending the membrane pellet with 10 up-down strokes on a Potter homogenizer. The final pellet was resuspended in 0.25 M sucrose, 30 mM histidine and stored at -70 "C until used.
The incubation was terminated by the addition of 5 ml of cold wash buffer (consisting of 10 mM Tris, pH

'HI-PDGF ADDED fcpm X
FIG. 1. Centrifugation assay of specific "'1-PDGF binding to 3T3 membrane preparations. lZ5I-PDGF (20,000 cpm/ng) was incubated with membranes derived from 3T3 cells (170 pg in 0.6 ml) as described under "Experimental Procedures." The incubations were krminated by centrifugation as described. The concentration of lZ5I-PDGF is indicated on the abscissa and the amount of total, nonspecific and specific binding is plotted on the ordinant. Each value represents the mean of triplicate determinations which varied less than 10% from the mean. 7.4, 0.15 M sodium chloride, 5% fetal calf serum) followed by immediate filtration over 25-mm cellulose acetate filters. The filters were then washed with five successive 5-ml aliquots of cold wash buffer and were counted after drying. This amount of washing was required to overcome the tendency of PDGF to stick to artificial surfaces (12). Nonspecific binding was assessed as the amount of binding which remained in the presence of 10 nM of unlabeled PDGF. This value was chosen because Scatchard analysis of the nonspecific binding yielded a horizontal plot with no evidence of high affinity binding sites. Specific binding was determined by deducting the nonspecific binding from the total amount of bound '"I-PDGF. In general, nonspecific binding constituted 20-35% of the total amount of lZ5I-PDGF binding. Higher levels of nonspecific binding were encountered if the filters were washed less or if platelet poor plasma was omitted. The level of specific binding was not changed by these manipulations.
Affinity Cross-linking-Membranes (0.2-0.3 mg/ml) were incubated in phosphate-buffered saline containing 1 mM MgCI,, and 10% dialyzed platelet poor plasma with IZ5I-PDGF (100,000 cpm) a t 37 "C for 1 h. The incubation was terminated by centrifugation in a Fisher microcentrifuge (model 235A) for 2 min. The membranes were then resuspended briefly in cold phosphate-buffered saline and were centrifuged once again. This washing procedure was repeated one additional time and the membranes were then incubated with crosslinking reagent (DSS) which was present a t 0.005-0.3 mM in phosphate-buffered saline at 15 "C for 15 min. The reaction was then quenched with 25 mM Tris-HCI, pH 7.4, containing 0.5 mM EDTA. The cross-linked membranes were then centrifuged once again in the microcentrifuge and were resuspended in sample buffer prior to SDSpolyacrylamide electrophoresis. Unless indicated otherwise, the electrophoresis was performed under nonreducing conditions. In some experiments cross-linking was blocked by the inclusion of unlabeled PDGF (10 nM unless otherwise indicated) in the incubation prior to the cross-linking reaction.

Specific Binding of '"I-PDGF to 3T3 Cell Membrane Prep-
arations-Two methods were used for assaying the membrane binding sites, each involving extensive washing of the membranes following the incubation of the membranes with 1251-PDGF. Since the amount of specific binding determined by the centrifugation method agreed closely with the level of binding measured by the filtration assay, these methods were used interchangeably. The specific lz5I-PDGF binding sites on 3T3 membranes were saturable (Fig. 1). Scatchard analysis ( Fig. 2A) revealed a maximal density of binding sites of 170 fmol/mg of protein. Half-maximal binding occurred at 0.2-0.3 nM, a value which is only slightly higher than the halfmaximal binding which was observed in intact cells (0.17 nM, Fig. 2 B ) . The recovery of sites in membrane preparations relative to the total number of sites measured in intact cells ranged between 50 and 200% when binding was measured at 37 "C for 1 h. Apparent recoveries greater than 100% in the membrane fractions may be attributed to the presence of receptors in intracellular pools of intact cells or to receptor "down regulation" which occurs in intact cells when the binding is measured at 37 "C (4).
The sites labeled with "'I-PDGF demonstrated the specificity expected of PDGF receptors. Unlabeled PDGF competed for the membrane binding sites, causing half-maximal inhibition of '"I-PDGF binding at a concentration of 0.2 nM (Fig. 3). Reduced PDGF did not inhibit 12'II-PDGF binding at concentrations as high as 1 nM (Fig. 3). The difference in the affinities of native and reduced PDGF for the binding sites correlated with their relative activities as mitogenic agents (9). Insulin, epidermal growth factor, fibroblast growth factor, and nerve growth factor did not compete for the membrane binding sites labeled with '"I-PDGF (not shown). The binding varied less than 20% over the pH range from 5 to 9, but was markedly inhibited by a pH of 4.5 or less (not shown).

Kinetics of '"I-PDGF
3. Inhibition of '261-PDGF binding to 3T3 membranes by native and reduced PDGF. "'I-PDGF (50,000 cpm) was incubated with 3T3 membranes as described under "Experimental Procedures." The indicated concentrations of unlabeled purified PDGF were added. The degree of inhibition of binding is plotted as a function of the concentration of added native PDGF (0) or reduced PDGF (0). Nonspecific binding was the amount of binding measured in the presence of 10 nM PDGF. Each value represents the mean of duplicate determinations from three separate experiments. Three separate preparations of native PDGF and reduced PDGF (reduced by 20 mM dithiothreitol as described under "Experimental Procedures") were used in these experiments. attained a steady state (Fig. 4). At 4 "C the binding was considerably slower (Fig. 4) and required 1.5-3 h of incubation to achieve a steady state level of binding, which was 60% of the steady state level achieved at 37 "C (not shown). Previously (9)  '"I-PDGF (15,000 cpm) was added to 3T3 membranes (0.1 mg/ml) at zero time and the amount of specific binding was assessed at the indicated times of incubation at 37 or 4 "C. In some samples (0) an excess (10 nM) of unlabeled PDGF was added after 45 min of incubation at 37 "C. receptor sites on intact cells becomes relatively nondissociable at 37 "C, a phenomenon which we attributed to receptor internalization or to the formation of a state of tight binding to cell surface receptors. Data in Fig. 4 demonstrate that membranes, like whole cells, bound "'1-PDGF in a manner which leads to the formation of a state of binding which is difficult to dissociate by addition of excess unlabeled PDGF. A large dilution of receptor-bound "'I-PDGF was similarly ineffective in dissociating the bound ligand.
Additional kinetic studies (Fig. 5) indicated that this state of tight binding was a time-and temperature-dependent process. Thus, 'T-PDGF binding proceeded at 4 "C until a steady state level of binding was achieved. The membranes were then rapidly warmed to 37 "C and the dissociability of binding was assessed. Immediately following warming, the binding was dissociable (Fig. 5). In three additional experiments (not

High Affinity Membrane
Receptors for PDGF shown) essentially all of the binding was dissociable after rapid warming. However, if the warming period was continued (Fig. 5 ) less of the binding was dissociable (open squares). After 1 h of the warming period, essentially all of the binding was nondissociable (not shown). Effect of Suramin on Tightly Bound '251-PDGF-The state of tight binding of Iz5I-PDGF which occurred at 37 "C did not involve a covalent interaction since the polyanionic compound suramin (13,14) readily dissociated tightly bound PDGF from its membrane binding sites (Fig. 6). This property of suramin was used to expose PDGF receptor sites in rat liver membranes. For these experiments, rat liver membranes were prepared as described under "Experimental Procedures" and a portion of the membranes were exposed to suramin which blocked binding of '2sI-PDGF. After extensive washing, the suramin-treated membranes had significantly more binding sites for 'T-PDGF than the control membranes (Table I). This increment of I2'1-PDGF binding over control binding was presumably due to the unmasking of additional PDGF receptor sites which had been obscured by PDGF released from platelets when the tissue was removed from the rat. The reversibility of suramin inhibition of PDGF binding is demonstrated in Table 11. These experiments were performed on membranes which had previously been exposed to suramin and washed, thereby unmasking all available PDGF receptors. After measuring Iz5I-PDGF binding, the membranes were exposed to suramin for a second time and the loss of binding sites was confirmed by measuring "'I-PDGF binding ( Table  11). The membranes were then washed free of suramin and the binding of lz5I-PDGF was again measured. The results showed that the inhibitory effect of suramin on the binding of lT-PDGF could be reversed by washing the membranes. Dose dependence studies of the suramin effect on PDGF binding (Table 111) revealed that suramin half-maximally inhibited ""I-PDGF binding at a concentration between 0.01 and 0.10 mM.

TABLE I Unmasking of PDGF receptors on rat h e r membranes by suramin
Rat liver membranes were prepared as described under "Experimental Procedures" using either control buffer without suramin (0.25 M sucrose, 5 mM Tris, pH 7.4, 5 mM MgCI,, 1 mM PMSF, and 100 kallikrein units/ml of aprotinin) or the same buffer plus 1 mM suramin during the homogenization and centrifugation steps. After washing membranes three times in suramin-free buffer, PDGF binding was measured. Each value is the mean of three determinations which varies less than 10% from the mean. This experiment is reuresentative of two similar experiments. The 12'II-PDGF binding sites exposed by suramin treatment of liver membranes had a high affinity for lz5I-PDGF as indicated by the low concentration (0.08 nM) of PDGF required for half-maximal binding (not shown) in agreement with the binding data for 3T3 membranes (Figs. 1, 2, and 3). The specificity of binding in liver membranes was appropriate for PDGF receptors in that reduced PDGF did not compete for the receptor sites at a concentration of 1 nM. In addition, the size of the receptor in liver membranes appeared to be the same as that in 3T3 membranes (see below). Binding sites with similar characteristics were exposed by suramin treatment of human placenta membranes (not shown).
Affinity Cross-linking Studies-The membrane binding sites in 3T3 membrane preparations were further characterized by affinity cross-linking techniques. In these experiments '"I-PDGF which was specifically bound to membranes was covalently linked to its binding sites by using either DSS or

Reversibility of suramin inhibition of PDCF binding
Rat liver membranes were prepared using 1 mM suramin to expose binding sites and were washed four times as described under "Experimental Procedures." These membranes ("control") were assayed for specific 12'II-PDGF binding and were then treated (a second time) with 1 mM suramin for 20 min at 4 "C in assay buffer and 12'I-PDGF bindingwas measured in the presence of suramin (B). The membranes were then washed three times to remove the suramin and 12'II-PDGF was again determined (C). Negative binding (B) indicates that the total binding of '"I-PDGF in the absence of unlabeled PDGF was less than the binding in the presence of 1 nM unlabeled PDGF. This experiment is representative of two experiments, each done in triplicate.

Concentration dependence of suramin inhibition of 12sI-PDCF
binding Rat liver membranes were incubated with T -P D G F in the presence of the indicated concentrations of suramin and specific binding was measured as described under "Experimental Procedures." The extent of inhibition of specific bindingwas determined in two separate experiments. 100% inhibition refers to total inhibition of specific "' 1-PDGF binding (5,000-13,000 cpm/mg of protein in two experiments).
~ DST. Analysis by SDS-polyacrylamide gel electrophoresis revealed two macromolecular species representing the products of the cross-linking reaction (Fig. 7). The molecular weight of these species was 190,000 and 140,000. Assuming that these weights include the weight of PDGF then the corresponding native molecules prior to cross-linking would have M, = 160,000 and 110,000. It should be noted that nonreducing conditions were used for electrophoresis to preserve the disulfide linkages between the polypeptide chains of PDGF. The weights of the cross-linked species determined on 7.5% acrylamide gels were confirmed by 5% polyacrylamide gels. Both the 160,000-and 110,000-dalton cross-linked species had a high affinity for purified PDGF since unlabeled PDGF at concentrations as low as 0.1 nM significantly blocked binding and cross-linking of '2'I-PDGF to both species (Fig.  7). This inhibition of cross-linked '*'I-PDGF by unlabeled PDGF is consistent with the inhibition of specific membrane binding assessed by the filtration or centrifugation assays (Fig. 3). The addition of protease inhibitors such as PMSF (2 mM), aprotinin (100 kallikrein units), pepstatin (2 pM), and leupeptin (100 p~) during membrane preparation or during the binding reaction did not influence the distribution of the two products of the cross-linking reaction. When 1 mM EDTA was included during the cell lysis and membrane preparation, more of the '*'I-PDGF was cross-linked to the 160,000-dalton protein and less to the 110,000-dalton protein (Fig. 8).
The proportion of cross-linked "'I-PDGF in the high and low molecular weight species depended on the concentration of cross-linker. A t low concentrations of cross-linker (0.01-0.03 mM) the predominant species was the high molecular weight form. At higher concentrations of cross-linker a greater proportion of cross-linked '"I-PDGF appeared in the low molecular weight form and less in the high molecular weight form (not shown). This phenomenon may represent intramolecular cross-linking of the receptor which could result in an underestimate of the molecular weight of the receptor PDGF complex.
Membranes from other tissues such as rat liver and human skin fibroblasts have '*'I-PDGF binding sites which demonstrate a cross-linking pattern similar to that of mouse 3T3 cells (not shown). By contrast, membranes prepared from tissues such as osteosarcoma cells, which do not respond to PDGF, do not have binding sites identifiable by these crosslinking techniques (not shown).
The kinetic characteristics of the receptor identified by cross-linking (Fig. 9) were identical with the characteristics of specific binding measured by filtration or centrifugation of membrane preparations (Figs. 5 and 6). For example, when 12'"IPDGF was bound to membranes at 37 "C, neither dilution nor addition of excess PDGF prior to cross-linking altered the appearance of the cross-linked bands seen on electrophoresis. However, when suramin ww added to membranes containing '2'I-PDGF bound to its receptors, the bound "'I-  ( I and J ) . The band at the bottom of the gel represents '*'I-PDGF that was bound to the membranes but was not cross-linked and thus was dissociated from the membranes when exposed to the SDS in sample buffer. FIG. 8 (right). Effect of EDTA treatment on the distribution of cross-linked "'I-PDGF in the high and low molecular weight binding sites. 31'3 cells were removed from flasks as described under "Experimental Procedures" and were kept in the presence (lane 2 ) or absence (lane 1 ) of 1 mM EDTA throughout the membrane preparation. In both cases, binding and affinity crosslinking were performed as described under "Experimental Procedures" except that all reactions occurred in the presence of 1 mM EDTA for the sample in lane 2. An autoradiogram of a 7.5% polyacrylamide gel is shown. This experiment is representative of two separate experiments.
were present upon subsequent addition of DSS (Fig. 9). When suramin was added after cross-linking occurred, there was no effect on the appearance of the high and low molecular weight species (Fig. 9).
Since the kinetic characteristics of binding were different at 4 and at 37 "C, we compared the cross-linking pattern a t these temperatures and found that the same distribution between the 160,000-dalton binding protein and the 110,000dalton protein was seen at 4 "C and was observed a t 37 "C (not shown). Thus, formation of the slowly reversible state of binding at 37 "C was not associated with a change in molecular weight of the binding sites since the same apparent sizes of cross-linked species were seen a t a temperature a t which the slowly reversible state did not occur.
Specificity studies revealed that insulin, epidermal growth factor, histone B, and fibroblast growth factor did not block cross-linking to either the high or low molecular weight species when added to the incubation at a concentration of 10 nM prior to the addition of ""I-PDGF. In contrast, a much lower concentration of native PDGF (1 nM) totally blocked cross-linking of '*'I-PDGF to these proteins (Fig. 7). PDGF which had been reduced by dithiothreitol (20 mM) and alkylated with iodoacetamide did not block cross-linking of ""I-PDGF to the high and low molecular weight species when added a t a concentration of 1 nM (not shown). This observation is consistent with the inability of reduced PDGF to bind to receptors measured by the filtration and centrifugation assays (Fig. 3).
T o determine the efficiency of cross-linking, membranes were incubated with "'1-PDGF and then washed a t 4 "C under conditions which did not permit dissociation of specific binding. Specifically bound '''I-PDGF was then cross-linked and the membranes were solubilized in SDS and applied to a Sephacryl S-200 gel filtration column which separated "'1-PDGF cross-linked to its receptor from free '2'I-PDGF (identified electrophoretically) which had been dissociated from the receptor when detergent was added. The results indicated that 15% of specifically bound "'I-PDGF was cross-linked to its receptor under the conditions employed in these experiments.
FIG. 9. Dissociation of specifically bound '251-PDCF assessed by cross-linking. 12sII-PDGF was incubated with 3T3 membranes for 1 h a t 37 'C as described under "Experimental Procedures." After steady state level of binding had been achieved the following manipulations were performed lanes A and B (control), the incubation was continued for 20 min prior to cross-linking; lanes C and D, a n excess of unlabeled PDGF (10 nM) was added for 20 min prior to cross-linking; lanes E and F, the samples were diluted 100-fold for 20 min prior to cross-linking; lanes G and H, suramin was added for 20 min prior to cross-linking; lanes I and J, 1 mM suramin was added after cross-linking. Samples in lunes B, D, F, H, and J contained 10 nM PDGF during the initial incubation of "'1-PDGF with membranes. This experiment is representative of two experiments.  10. Two-dimensional electrophoresis of '251-PDGF cross-linked to its receptor. Affinity cross-linking of Iz5I-PDGF to 3T3 membranes was performed as described under "Experimental Procedures" except that the samples were solubilized in 9.0 M urea, 2% Nonidet P-40, and 5% ampholine. Two-dimensional isoelectric focusing/SDS-polyacrylamide gel electrophoresis was performed by the methods of O'Farrell (25) except that in the isoelectric focusing, the tube gels were prepared using 1:l mixture of ampholines (LKB Instruments Inc.) in the ranges of 3.5-5 and 5-8 and the electrode solutions were 0.1 M HnPOI and 0.1 M NaOH. Isoelectric focusing was performed a t 400 V for 19 h a t 10 'C. The gel was embedded in 1% agarose on the top of a stacking gel (4% polyacrylamide) and SDS-polyacrylamide gel electrophoresis was performed using a 7.5% running gel. The autoradiogram shown is representative of three separate experiments. The arrows indicate the position of the proteins cross-linked to lZsI-PDGF. Bovine serum albumin focused at a position corresponding to an isoelectric point of 5.3.
""I-PDGF covalently bound to its receptor was used to determine some of the physical characteristics of the receptor. Receptor bound '*"I-PDGF could be adsorbed to diethylaminoethyl cellulose and eluted with 0.4 M NaCl, suggesting that the receptor is an acidic molecule. Two-dimensional polyacrylamide gel electrophoresis confirmed that both the high and low molecular weight cross-linked species were acidic with an isoelectric point of approximately 4.7 (Fig. 10). This observation suggests that the high and low molecular weight species are similar. The actual isoelectric point of the receptor must be somewhat lower than the value of the cross-linked complex since the basic nature of PDGF would be expected to increase the isoelectric point of the receptor-hormone complex. Since most cell surface receptors are glycoproteins, we examined the ability of cross-linked PDGF receptors to bind to lectins. For these experiments membranes were incubated with "'I-PDGF, cross-linked with DSS as described above, and solubilized with 1% Triton. The soluble fraction was adsorbed to a column of wheat germ agglutinin linked to Sepharose 4B and was specifically eluted from matrix with N-acetylglucosamine. Analysis of the eluates showed that the cross-linked 160,000-and 120,000-dalton proteins could be adsorbed to and specifically eluted from the resin using 0.4 M N-acetylglucosamine (not shown).

DISCUSSION
Recently radioligand binding techniques have been used to label PDGF receptor sites in intact cell preparations using 3T3 cells (6-8) and arterial smooth muscle cells (9). Although these studies can provide qualitative information about the receptor sites, quantitative measurements of affinity and kinetic characteristics require the use of membrane preparations or soluble preparations of receptors in order to avoid the complications of receptor internalization and metabolism. In addition, a first step in receptor purification is to isolate binding sites in membrane fractions rather than intact cells especially if a solid tissue such as liver or placenta is used as a source of PDGF receptors. An additional rationale for using subcellular preparations rather than intact cells for binding studies is that the effects of possible intracellular regulatory agents (such as ions and nucleotides) and of other reagents such as suramin can be directly tested on the binding sites. For these reasons we have developed methods for measuring membrane bound recepbor sites for PDGF. The principal findings of this work include the following. 1) High affinity PDGF binding sites which have the characteristics of PDGF receptors can be identified in membrane preparations derived from cells known to respond to PDGF. The affinity of these sites for PDGF appears to be similar to the affinity determined in intact cell studies (9). Membranes derived from cells which do not respond to PDGF, do not have binding sites.
2) At 37 "C the binding of I2'1-PDGF to its membrane receptors becomes relatively nondissociable. Since these studies were performed in membrane preparations, the nondissociable state is not due to internalization as was postulated previously (9). The formation of the state of tight binding is a temperature-dependent process which does not occur a t 4 "C.
3) Tightly bound PDGF can be dissociated with the heterocyclic anionic compound suramin, indicating that the tight binding of PDGF is not due to covalent bond formation. The suramin effect on membranes is fully reversible and can be used to unmask PDGF receptors which are obscured by the presence of endogenous PDGF released by blood clotting during removal of the tissue from the animal.
4) The membrane sites for tight binding of Iz5I-PDGF appear to consist predominantly of a 160,000-170,000-dalton prot,ein which can be identified by affinity cross-linking techniques using membrane preparations. At least one additional lower molecular weight cross-linked species can also be identified. This lower molecular weight species may be related to the 160,000-dalton receptor. This cross-linking methodology has made it possible to estimate the isoelectric point of the receptor and to develop methods to solubilize receptor binding activity and to partially purify active receptor sites using lectin affinity chromatography and ion exchange chromatography.' Whether formation of the state of tight binding is related to the activity of the receptor is not known. Similar high affinity states of binding have been described for other polypeptide hormones including nerve growth factor (16), epidermal growth factor (17), and insulin (18,19). The high affinity state of binding defined by our data is different from the covalent binding of epidermal growth factor to its receptors which occurs when the chloramine-T method of iodination is used for preparation of '"I-epidermal growth factor (20, 21). Our data show that the state of tight binding of "'I-PDGF is clearly not due to covalent bond formation since the polyanionic compound suramin readily dissociates tightly bound "'I-PDGF. A similar dissociation of ""I-PDGF binding is seen in the presence of the detergent SDS (data not shown). However, unlike detergent, suramin dissociates bound PDGF and leaves the unoccupied membrane-bound receptors in a state in which they can rebind PDGF following the removal of suramin. We exploited this property of suramin to expose receptors which were occupied by endogenous PDGF which was presumably released from platelets (during preparation L. T. Williams and P. Tremhle, unpublished data. of the tissue) and was tightly bound to the receptor sites. This use of suramin made the study of PDGF receptors in membranes derived from intact organs feasible.
The affinity cross-linking studies in these membrane preparations demonstrated that '251-PDGF can be cross-linked to a 160,000-dalton protein at very low concentrations of the cross-linker disuccinimidyl suberate and the cross-linker with a shorter chain link, disuccinimidyl tartarate. Since the 160,000-dalton protein was the only cross-linked species seen at extremely low concentrations of cross-linker (for both DSS and DST), it is likely that this protein represents the main binding site for PDGF. These concentrations of cross-linking agents used for PDGF are less than 1/10 concentration of cross-linker required for other hormone systems (15,22). This special susceptibility of PDGF to cross-linking may be due to the presence of numerous lysine groups (23) which are potential cross-linking sites. At higher concentrations of crosslinking agent, a t least one additional lower molecular weight cross-linked species ( M r = 110,000) was observed. This low molecular weight species might be an artifact due to the formation of intramolecular cross-linked bonds in the PDGF receptor. Alternatively, the low molecular weight species could represent a membrane macromolecule which neighbors the PDGF receptor or is associated with the PDGF receptor. It is also possible that the presence of the low molecular weight species is due to proteolysis of the PDGF receptor. The presence of a variety of protease inhibitors did not alter the distribution between high and low molecular weight crosslinked species, suggesting the proteolysis is not related to the low molecular weight species or to the formation of the low molecular weight species. However, the observation that the presence of EDTA during the preparation of membranes alters the distribution between high and low molecular weight forms of the receptor (Fig. 81, raised the question that a calcium-sensitive protease could be involved in the formation of the lower molecular weight species, a process which occurs in the case of the epidermal growth factor receptor (24). If the 110,000-dalton species is a proteolytic fragment of the 170,000-dalton species, it probably does not contain a site for tyrosine phosphorylation since in other studies we have not found a 110,000-dalton phosphorylated species in experiments designed to detect autophosphorylation of the PDGF receptor in U~U O .~ Finally, another explanation for the existence of two cross-linked species is that the low molecular weight form is a synthetic precursor of the high molecular weight form. Additional biosynthetic studies will be required to elucidate these issues.
Bowen-Pope and Ross (8) have reported cross-linking studies of '"I-PDGF bound to intact 3T3 cells. As in their study, we had difficulty resolving the high molecular weight regions of the gels when intact cells were used for cross-linking. This problem was presumably due to large aggregates of cell surface proteins which become cross-linked and incompletely penetrated the gels (8). The use of membrane preparations for binding and cross-linking greatly facilitated the studies reported here, especially when the regions greater than the 170,000-daltons were of interest. However, despite this difficulty it is possible to demonstrate cross-linking to the 160,000dalton species in intact cells (8). Whether the lower molecular weight component of binding (110,000 daltons) is present in intact cell experiments is not entirely clear due to the high background on the autoradiograms.
In summary, these studies have identified a tightly bound state of I2'1-PDGF in membrane preparations derived from both cultured cells and intact tissues. Tightly bound PDGF

High Affinity Membrane
Receptors for PDGF must be dissociated (by suramin treatment) from some tissues in order to measure the total pool of binding sites. The use of suramin for dissociation PDGF from its receptors may also provide a method for interfering with PDGF action in uiuo.
The cross-linking studies reported here demonstrated that the state of tight PDGF binding is seen when PDGF binds to a 160,000-dalton species and perhaps to a lower molecular weight species. These binding sites are anionic and adsorb to lectin affinity resins. These results should be useful in developing strategies for purification of PDGF receptors and for studies on the mechanism by which PDGF activates its receptor.