Platelet-derived Growth Factor

We have prepared radioiodinated purified platelet- derived growth factor ("'I-PDGF) which retains full mitogenic activity. The binding of "'I-PDGF to Swiss 3T3 cells is saturable and highly competed by whole blood serum, purified unlabeled PDGF, and by material from each stage in the purification of PDGF from plate-let-rich plasma. Other purified mitogens and sub- stances tested do not compete. "'I-PDGF binding to fibroblasts, 3T3 cells, and arterial smooth muscle cells shows an apparent dissociation constant of lo-" M, comparable to the range in which PDGF is mitogenic. A clone of Swiss 3T3 cells obtained from a population selected repeatedly against mitogenic response to PDGF shows a greatly reduced mitogenic response to PDGF and binds only 5% as much '251-PDGF/cell. The binding capacity of the different cell types tested ranges from 2,500 binding sites/cell on the poorly re-sponding variant to 390,000 binding sites/cell on one strain of Swiss 3T3 cells. Cell types that do not respond to PDGF do not show specific high affinity binding of Ia6I-PDGF. At 4 "C, '"I-PDGF binding to monolayer cultures is relatively slow. Equilibrium binding of low concentrations of I2'I-PDGF is not achieved during 7 h unless the binding medium is constantly mixed. "'1- PDGF binding at 4 "C shows a broad pH optimum between 6.3 and 8.0. Binding does not seem to require Ca2+ or M&+ but is reduced more than 6-fold if both monovalent and divalent salts are omitted. The initial rate of "'I-PDGF binding is greater at 37 "C than at 4 "C but cell-associated

It has been known for many years that pure polypeptide hormones can profoundly alter the metabolic and mitotic activities of target cells. The mechanism(s) through which this alteration is accomplished remains obscure. Two different approaches have been taken in studying this problem. Some investigators have studied the final physiological effects of hormone addition and attempted to discover, from these effects, what underlying cell process(es) must have been affected. Thus, Robinson et al. (1971) have proposed that changes in cyclic adenosine monophosphate levels mediate the effects of epinephrine on adipose tissue, and Rubin (1975) has proposed that changes in the effective intracellular concentration of Mg2+ underlie the manifold effects of substances mitogenic for animal cells.
A second tactic has been to investigate the initial interaction between hormone and cell with the hope that some characteristic(s) or direct result(s) of this interaction can be shown to be instrumental in effecting cell behavior. This interaction has been especially well studied for insulin (e.g. Cuatrecasas, 1971a) and epidermal growth factor (e.g. Cohen et al., 1975). These studies have demonstrated the existence of cell surface components (receptors) which bind hormone specifically and with high affinity. In this paper, we describe the binding of platelet-derived growth factor, one of the principal mitogens present in whole blood serum (Balk, 1971;Ross et aZ., 1974;Kohler and Lipton, 1974;Westermark and Wasteson, 1975), to cultured animal cells.

Effect of Zodination on the BiologicaZActivity of PDGF2-
We have prepared an iodinated derivative of PDGF using the iodine monochloride method (see "Materials and Methods" for details). This iodinated PDGF contains an average of 0.25 molecule of '251/molecule of PDGF ( Table I, Experiment 3). The iodine monochloride method uses unlabeled lZ7ICl as an oxidizing agent ('271Cl + '"Ic) '"IC1 + 1271-) so that the specific activity of the I2'I is necessarily reduced. As a result, each molecule of PDGF contains an average of 1.7 molecules of lz5I + Iz7I and few molecules of PDGF would escape some extent of iodination.
The iodination of a peptide hormone could alter the binding or physiological properties of the hormone, as seems to be the 5162 Specific Binding of PDGF to Cultured Cells

Effect of degree of iodination of PDGF on its binding characteristics and ability to stimulate L3H]thymidine incorporation
For Experiment 1 six identical 1.0-pg samples of PDGF were each preparation to stimulate r3H]thymidine incorporation into pariodinated as described under "Materials and Methods," except that allel cultures was determined using parallel cultures incubated in the amount of total iodine/reaction mix was varied from 0.145 to 3.48 duplicate with 8 concentrations of each PDGF preparation. After an ng (column a). In each case, the ratio of total iodine/IZ5I was main-8-h incubation, [3H]thymidine incorporation was measured and cortained at 7. 05 The values for molecules of Iz5I incorporated/molecule rected for residual "'1-PDGF binding as described under "Materials of PDGF (column b) and for molecules of total iodine + Iz7I) and Methods." The concentration of PDGF needed to produce halfincorporated/molecule of PDGF (column c) were calculated from the maximal stimulation of [3H]thymidine incorporation (column f ) was specific activity of the Iz5I after equilibration with ICI. The binding determined graphically from a plot of r3H]thymidine incorporation characteristics of each preparation were determined using confluent versus initial concentrations of PDGF. The maximum stimulation cultures (3.8 X lo4 cells/2-cm2 well) of 3T3 cells preincubated in 2% produced by each preparation was within 25% of the average maxicalf CMS-I for 48 h. Binding was measured in duplicate at 8 concen-mum stimulation. trations using 1 ml of binding medium/culture incubated for 4 h at For Experiment 2, one 6.0-pg sample was iodinated as described 4 "C with gentle shaking. The values of duplicate determinations under "Materials and Methods" for use in binding experiments, except were within 10% of their mean. Nonspecific binding (2.2% of total that only 0.5 mCi of Na"'I-PDGF was used. For Experiment 3 one binding at concentrations of PDGF giving half-maximal binding) was 6.0-pg sample of PDGF was iodinated as described under "Materials subtracted. The concentration of PDGF needed for half-maximal and Methods" for use in binding experiments. For both experiments, specific binding (column d), and the maximum binding capacity of a parallel 6.0-pg sample was processed, except that no I2'I or IC1 was the cells (column e) were determined graphically from a plot of added. Measurements of binding and stimulation of [3H]thymidine specific binding 'versus the concentration of PDGF remaining in the incorporation were as described for Experiment 1. incubation medium at the end of the binding period. case for the neurohypophysial hormones oxytocin and vasopressin (Cuatrecasas and Hollenberg, 1976). The reaction conditions used to prepare the iodinated derivatives could also damage the hormone (Dedman et al., 1961). We determined the physiological integrity of '251-PDGF by measuring its mitogenic potency in the same assay ([3H]thymidine incorporation into trichloroacetic acid-insoluble material) used to guide the purification of PDGF. lz5I-PDGF is equipotent with unlabeled PDGF in this assay (Table I, Experiments 2 and 3).
T o systematically investigate the possibility that iodination may have affected the biological activity of PDGF, we prepared a series of Iz5I-PDGF derivatives in which the average extent of iodination varied from 0.0-2.61 molecules of iodine/ molecule of PDGF. There was no consistent effect of iodination of mitogenic activity (Table I, Experiment 1). Effect of Iodination on PDGF Binding-To directly compare binding of iodinated uersus native PDGF, we compared binding saturation curves in which the concentration of PDGF was varied using either iodinated or native PDGF. As can be seen in Fig. 5, the curves are superimposable. Had the affinity of '"I-PDGF been lower than that of native PDGF, the curve for native PDGF would have been saturated at a lower concentration than was observed in the curve for Iz5I-FDGF.
In a second set of experiments ( Table I) we examined the effects of increasing iodination of PDGF upon binding capacity and affinity. At the lowest levels of iodination, the average incorporation was less than 0.5 molecule of iodine/molecule of PDGF so that virtually all radioactivity should have been monoiodo-PDGF. As the average incorporation was increased, diiodo-and triiodo-PDGF would have been present in increasing amounts. The cell-binding properties of these preparations showed no consistent changes as the extent of iodination was increased.
For each preparation of '251-PDGF, we also determined the percentage of radioactivity that is incapable of high affinity binding to Swiss 3T3 cells ("Materials and Methods," Fig. 4). The values for the percentage of total counts/min which cannot bind to Swiss 3T3 cells ranged from 40% for a PDGF preparation that contained a 38,000-dalton contaminant (visualized on silver-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels and on autoradiograms of the iodinated preparation), to 18% for an lz5I-PDGF preparation showing small amounts of lz5I-gelatin and no other detectable contaminants. These values are comparable to values obtained after iodination of other purified hormones (Cuatrecases and Hollenberg, 1976). The nonbinding material could represent material unrelated to PDGF, PDGF molecules altered during iodination, or incomplete quenching of the iodination reaction with consequent iodination of small amounts of the carrier protein (gelatin) used during dialysis. This type of information concerning the presence of radiolabeled nonbinding components is necessary for accurate Scatchard analysis. Scatchard analysis of binding data requires determination of the concentration of free ligand in solution at the end of the incubation period. Correction must be made for the presence of labeled nonbinding components, since their presence would contribute to the free counts/min without contributing to bound counts/min. This would raise the apparent dissociation constant. After correction for nonbinding radioactivity, all preparations of '251-PDGF used to date have given comparable values of apparent K d for binding to Swiss 3T3 cells.
Determination of Cell-associated '251-PDGF--Before utilizing 1251-PDGF to probe the interaction of PDGF with cultured animal cells, we had to demonstrate that, within our system: 1) the measured interaction is truly a cellular phenom- The mean f range of duplicate determinations is plotted. A, human whole blood serum; A, human platelet-rich plasma; m, PDGF purified through CM-Sephadex; U, PDGF purified through Sephacryl S-200 gel filtration; 0, PDGF purified through heparin-Sepharose chromaenon, i.e. not an artifact of binding to the surface of the culture dish and 2) the interaction is a specific property of PDGFresponsive cells and not just a general adsorptive property of mammalian cell plasma membranes. High affinity, specifically competable binding of hormones to nonreceptor binding sites has been observed in other systems. 1251-Insulin binding to talc shares many of the properties of '251-insulin binding to cell surface receptors (Cuatrecasas and Hollenberg, 1975). In preliminary studies, we found that '"I-PDGF binds to bare culture wells and that this binding can be partially competed by unlabeled PDGF (see "Materials and Methods"). Since, in further studies, we planned to use competition by unlabeled PDGF as being diagnostic of receptor-bound '"I-PDGF, this was unacceptable. To circumvent this problem, we used a solution containing the nonionic detergent Triton X-100 to solubilize cell-associated lZ5I-PDGF without eluting dishbound *251-PDGF (for further details see "Materials and Methods''). A similar procedure was developed independently by Heldin et al. (1981).
Competition for lZ5I-PDGF Binding-As PDGF is increasingly purified from human platelet-rich plasma, smaller concentrations are sufficient to stimulate f'H]thymidine incor-tography; 0, PDGF purified through phenyl-Sepharose hydrophobic chromatography. Note that the preparations of PDGF used here were not derived from the stages in purification of PDGF from a single batch of platelet-rich plasma. Instead, they represent material available from several different preparations. b, competition for '"I-PDGF binding. Parallel cultures were used to determine competition for PDGF binding. Cultures were incubated for 4 h at 4 "C with gentle shaking in binding medium containing 6 PM '"I-PDGF, plus the amount of test competitor indicated on the abscissa. Cell-bound '"I-PDGF was determined as described under "Materials and Methods." Binding has been plotted without correction for nonspecific binding (noncompetable binding) which can be seen from the curve for competition by CMS-I11 PDGF to be 1-2% of total binding. The symbols are the same as those in a. The mean f range of duplicate determinations is plotted. poration ( Fig. la, Table IT, and Raines and Ross, 1982).  Table I1 show that the ability of low concentrations of partially purified preparations of PDGF to compete for "'1-PDGF binding is also increased as the mitogenic activity is purified. We have chosen to use PDGF purified through CM-Sephadex (CMS-111) to determine noncompetable binding in standard assays. At concentrations of 200-400 pg/ml, CMS-111 reduces '251-PDGF binding to 1-2s of binding in the absence of a competitor. Although competition to this extent was not achieved in Fig. l b using the most purified preparation of PDGF, the slope of the competition curve suggests that higher concentrations of this preparation would have been as effective as high concentrations of CMS-111.
In order to determine whether competition for IZ5I-PDGF binding is specific for PDGF, we tested other mitogens, including EGF, fibroblast growth factor, and insulin, for their ability to compete for '251-PDGF binding to 3T3 cells (Table  11). Each of these mitogens was tested for ability to compete using concentrations at least 400-fold higher than the concentration required to produce half-maximal stimulation of [3H] thymidine incorporation. Several nonmitogens were also tested, including human thrombospondin and partially puri- Costar 24-well trays and grown to confluence in a 1:l mixture of Dulbecco-Vogt modified Eagle's medium and Ham's F-12 medium + 5% calf serum. The growth medium was then replaced with a 1:l mixture of Dulbecco-Vogt modified Eagle's medium and Ham's F-12 medium + 2% calf CMS-I for 48 h. A , '"I-PDGF binding was determined using triplicate cultures for each concentration by incubating with 1 ml of binding medium containing increasing amounts of '"1-PDGF. The cultures were incubated for 5% h at 4 "C with gentle agitation before measuring cell-bound "'I-PDGF as described under "Materials and Methods." Nonspecific binding was determined as described under "Materials and Methods," and has been subtracted (see Table I11 for values). Cell number was determined as described in the legend to Table 111. The concentrations plotted on the abscissa fied human platelet factor 4, both of which are found in platelets (Lawler et al., 1978;Witte et al., 1978). Platelet factor 4 copurifies with PDGF through the fist purification step (CM-Sephadex cation-exchange chromatography; Raines and Ross, 1982) after defibrinogenation. Trapidil was reported to be a specific inhibitor of mitogenic stimulation of BALB/c 3T3 cells by PDGF (Ohnishi et al., 1981). None of these reduced "'I-PDGF binding to Swiss 3T3 cells in the concentration range tested.
The mannose-binding lectin concanavalin A has been reported to reduce binding of '251-EGF (Carpenter and Cohen, 1977) and '"SI-insulin (Cuatrecasas and Tell, 1973) to their respective receptors. Concanavalin A did not significantly reduce '"I-PDGF binding in our system, suggesting that neither the receptor, nor PDGF, contains available a-mannose residues near regions involved in PDGF binding. Gajdusek et al. (1980) have reported that serum-free medium conditioned by bovine aortic endothelial cells is mitogenic for 3T3 cells. The major mitogenic activity in endothelial cell-conditioned medium seems to be biochemically distinct from PDGF (Gajdusek et al., 1980). Endothelial cell-conditioned medium does show some ability to compete for 12'1-PDGF binding to Swiss 3T3 cells" but, until endothelial cellconditioned medium is resolved into purified components, we will not know how similar this binding competitor activity is to PDGF. Heldin et al. (1980)  incorporation was not determined for human smooth muscle cells or osteosarcoma cells produces a growth factor which may be identical to PDGF and that this factor is equipotent with PDGF as a binding competitor.
Cell Specificity of 12'I-PDGF Binding-In order to demonstrate that lz5I-PDGF binding is not just a genera1 property displayed by cells in culture, as opposed to a specific property related to cell response to PDGF, we measured "'1-PDGF binding to cells which do not respond to PDGF in culture. As an example of a cell type without PDGF receptors, we have used the human cervical carcinoma cell line A431 (Fig. 2). This cell line displayed no measurable specific '*'I-PDGF binding (Fig. 2 A ) and conditioned serum-free medium from this cell had no effect on C3H]thymidine incorporation into trichloroacetic acid-insoluble material and did not block 1251-PDGF binding (data not shown).
Low passage bovine aortic endothelial cells also showed no mitogenic response to addition of PDGF and showed no cornpetable binding. However, since endothelial cells produce a competitor of '""I-PDGF binding, it is possible that these cells do have PDGF receptors but that these receptors are already occupied by an endogenously produced competitor. Such a situation does seem to exist for EGF receptors on Kirsten sarcoma-transformed cells (DeLarco and Todaro, 1980) and for PDGF receptors on one osteosarcoma cell line (Heldin et al., 1980(Heldin et al., , 1981. Because of this type of phenomenon, it is difficult to determine whether a cell type which produces a binding competitor has receptors for the test hormone. by guest on March 18, 2020 http://www.jbc.org/ Downloaded from A second type of evidence for the specificity of '"I-PDGF interaction with responsive cells has been obtained from afflnity-labeling experiments. Using bifunctional cross-linking reagents, Glenn et al. (1982) demonstrated that '251-PDGF bound to several PDGF-responsive cell types can be crosslinked to a single high molecular weight cell surface component. "'I-PDGF binding to this component displays the same properties as the specific '251-PDGF binding measured by Triton X-100 extraction described in this paper. The authors propose that this cell surface component is the PDGF receptor.
PDGF Receptor Affinity-Correlation with Mitogenesis- Fig. 2A shows the concentration dependence of specific 1251-PDGF binding to Swiss 3T3 cells, as well as to low passage monkey and human arterial smooth muscle cells, human foreskin fibroblasts, human lung fibroblasts (WI-38), A431 carcinoma cells, and the Swiss 3T3 variant PF 2 (to be discussed later). '251-PDGF binding is plotted versus the concentration of lz5I-PDGF present in the incubation medium at the time the binding incubation was terminated. For each cell type binding is saturated by 100 PM lZ5I-PDGF. The number of binding sites/cell, among cell types that bind "'1-PDGF, varies over a 130-fold range ( Fig. 2 and Table 111); however, the concentration dependence of binding for each cell type is similar and does not seem to vary along species lines. The binding properties of insulin (Freychet et al., 1971) and EGF  seem similarly highly conserved. The 1251-PDGF concentration at which half-maximal binding is achieved is near 10 PM for each cell type tested (Table 111).
For each cell type, parallel cultures were used to measure the concentration dependence of PDGF stimulation of r3H] thymidine incorporation into trichloroacetic acid-insoluble material (Fig. 2B). These data are plotted uersus the initial concentration of 12'I-PDGF since it is not known in detail how the concentration dependence for mitogenesis may change during progression toward the S phase of the cell cycle. At 37 "C, cultures of 3T3 cells rapidly degrade '251-PDGF (Fig.   3). Consequently, the concentration of PDGF changes greatly during the 18-h incubation prior to determination of [3H] thymidine incorporation. The concentration of PDGF required for half-maximal stimulation of [3H]thymidine incorporation is similar for each cell type tested, about 13 PM (column h of Table 111). Thus, purified PDGF is a potent inhibitor of '"I-PDGF binding in the same concentration range (10-20 PM) in which it is a potent mitogen. The biphasic response of the monkey smooth muscle cells to stimulation by increasing concentrations of PDGF (Fig. 2B) is occasionally seen with this and other cell types at high PDGF concentrations (e.g. see Fig. lA), and has been reported for other mitogens (e.g. Cohen et al., 1975). Its significance is not known.
Radioreceptor Assay for PDGF-Since the apparent Kd for PDGF binding to animal cells is comparable to the concentration range in which PDGF is mitogenic, competition for I-PDGF binding to Swiss 3T3 cells can be used as a sensitive radioreceptor assay to follow the purification of PDGF from human platelet-rich plasma. Material obtained from each successive step in the purification scheme developed by Raines and Ross (1982) is increasingly potent as a binding competitor and as a mitogen (Fig. 1). The overall purification factor calculated from binding data is 4.3 X 105-fold compared to 1.5 X lo5 calculated from r3H]thymidine incorporation data.
The advantage of the radioreceptor assay is that it is more selective for PDGF. The [3H]thymidine incorporation assay determines the overall mitogen (and, to some extent, cofactor) content of the test preparation, while the binding assay detects only substances that utilize the PDGF receptor or which acid-soluble 1251 present in the incubation medium. The latter values were corrected for trichloroacetic acid-soluble "' I present in unused binding medium (1.5% of total '"I). Binding medium incubated at 37 "C without cells for 9 h was still 98% trichloroacetic acid-precipitable. The mean f S.E. of triplicate determinations is plotted: 0, cellassociated "'1; 0, cell-associated '"1 plus acid-soluble I2'I in the medium. Additional triplicate cultures were incubated 8% h with '"I-PDGF, then for ?h h in fresh PDGF-containing medium before cellassociated 12'1 (0) was measured. themselves bind PDGF (e.g. carrier proteins). Thus, the high potency of human serum in the r3H]thymidine assay relative to its potency as a binding competitor (Table 11) probably results in part from the presence in serum of other substances active in the [3H]thymidine incorporation assay. Human platelet-rich plasma, and all further purifications of it, are enriched in PDGF and show much lower ratios of EDs0 in the binding assay to ED, in [3H]thymidine incorporation assays (Table 11). The disadvantage of the radioreceptor assay, as it is currently employed, is its somewhat lower sensitivity-it is easier to detect a several-fold increase in r3H]thymidine incorporation than a less than 25% decrease in Iz5I-PDGF binding.
In most determinations reported here (Tables I-111), the concentration of PDGF needed for half-maximal binding competition has been near the concentration needed for half-maximal stimulation of [3H]thymidine incorporation. The relationship between receptor occupancy and physiological response has been extensively studied in other hormone systems. The ED% for binding competition by EGF has been reported to be 2-fold  to 10-fold (Aharonov et al., 1978b) higher than the ED50 for mitogenesis. Reported values for the quantitative relationship between 1251-insulin binding and stimulation of glucose oxidation have ranged from reports of an ED50 for binding competition 200-fold higher than the ED50 for metabolic stimulation of liver cells (Freychet et al., 1971) to an ED% for binding competition 40-fold higher than the ED% for metabolic stimulation of fat cells (Kono and Barham, 1971) to a report in which the two processes show essentially the same ED50 (Cuatrecasas, 1971a). The discrepancies between these reports seem to result largely from differing estimations of binding affinity. Findings such as these have led to the concept of "spare receptors" which postulates that a maximal physiological signal can be generated by occupancy of only a fraction of all receptors. At present there are two major reasons why we cannot determine from our data whether spare PDGF receptors exist. 1) All determinations of binding affinity reported here were made at 4 "C while all those of mitogenic capacity were made at 37 "C. The Kd for insulin binding to cells has been shown to be very dependent on temperature (Cuatrecasas, 1971b). 2) During the 18-h 37 "C incubation used in mitogenesis studies the concentration of intact PDGF in the medium continuously decreases as it is bound and degraded (Fig. 3). By the end of the assay period, the majority of the original lZ5I-PDGF has been degraded (Fig. 3). It is not clear how to quantitatively express the relationship between mitogenesis and PDGF concentration in a system in which the concentration of mitogen is rapidly decreasing, and in which the importance of the mitogen to the cell may be changing. For this reason, all mitogenesis data are plotted uersus the initial concentration of PDGF. If plotted against the concentration at 4 h, the EDso would be significantly reduced. It is thus clear that the precise quantitative relationship between concentrations needed to inhibit binding by 50% and to stimulate 50% of maximal [3H]thymidine incorporation is dependent on choice of assay system, and on the form in which the data are expressed.
The Number of PDGF Receptors-The number of PDGF binding sites/cell varied from 3.9 X 105/cell on a high binding strain of Swiss 3T3 cells to 3.6 X 104/cell on human arterial smooth muscle cells and 2.5 X 103/cell on the variant Swiss 3T3 clone PF 2 (Table 111). With the exception of 3T3 clone PF 2, there is no consistent relationship between the number of receptors for PDGF and the magnitude of the mitogenic response to PDGF (Fig. 2). Aharonov et al. (1978a) have reported that three lines of mouse cells which have different numbers of EGF receptors do not show pardel differences in mitogenic responsiveness. It is possible that the dissociation of receptor number from the magnitude of the mitogenic response to PDGF reflects the presence of spare receptors on most cell types. Only when the number of receptors is less than the minimum number necessary to initiate a full mitogenic response wouid responsiveness be proportional to receptor number.
We have found that a clone of Swiss 3T3 cells (3T3-PF 2) which was derived from a population selected by E3H]thymidine suicide against mitogenic response to PDGF (see "Materials and Methods") showed a 20-fold reduction in lZ5I-PDGF binding and a 7-fold reduction in mitogenic response to PDGF. We cannot be certain that the reduced binding was the cause of the reduced mitogenesis, but this seems very probable. The defect in this clone is not a general defect in mitogenic responsiveness, since its responsiveness to EGF was not reduced-1 ng/ml of EGF plus 1 pg/ml of insulin stimulated [3H]thymidine incorporation 26-fold in the experiment (Fig. 2) in which PDGF stimulated only 3-fold. To the extent that the reduced binding determines reduced responsiveness, the phenotype of this clone supports a causal relationship between the measured high affinity lZ5I-PDGF binding and mitogenesis.
Scatchard Analysis of Binding Data-For more rigorous analysis of the binding data, we have plotted the data according to the method of Scatchard (1949). Fig. 6 shows Scatchard plots for three cell types with greatly differing numbers of receptors. It can be seen that the plots are not straight lines, but rather show a hook at low values of bound lZ5I-PDGF (low lZ5I-PDGF concentrations). When the apparent K d is determined from the slope of the curve at a half-maximal ratio of bound/free, the values (12.2, 8.6, and 17 PM, respectively) agree reasonably well with values determined graphically from saturation curves (7.5, 6.5, and 10.5 PM, respectively; Table  111). It is possible that the curvature of the Scatchard plots reflects a degree of positive cooperativity in the binding of lZ5I-PDGF to its receptor. On the other hand, the curvature may also reflect an artifact in the measurement of binding. It is possible that very low concentrations of PDGF are particularly vulnerable to degradation or inactivation. Another possible source of error in the hook region of the Scatchard plots is incomplete equilibration between bound and free lZ5I-PDGF. As will be discussed below, complete equilibrium binding of very low lz5I-PDGF concentrations may not be achieved during the incubation period. Consequently, the true a f h i t y for low concentrations of lZ5I-PDGF could be underestimated. This could account for some of the reduction in ratios of bound/free Iz5I-PDGF at low concentrations. A similar observation has been made in study of lZ5I-EGF binding to fibroblasts . Since very long incubation of some cell types (e.g. Swiss 3T3) at 4 "C causes some morphological changes in the cells, we have not attempted to achieve complete equilibrium in this system, preferring to make these measurements with membrane preparations or solubilized receptor^.^ The value (lo-" M) which we have obtained for the apparent Kd for lZ5I-PDGF binding to fibroblasts is considerably lower than the value (lo-' M) reported by Heldin et al. (1981).
We have replicated their experimental protocol and obtained a value (2.5 X 10"' M, data not shown) close to the value which Heldin et al. (1981) reported. It is likely that some of the difference between the two estimates of K d result from two differences in binding conditions. 1) The total concentration of receptors in their assay is much higher than the concentration of receptors employed in the present study (2 X 10"' M uersus an average of 2 X 10"2 M). Cuatrecasas and Hollenberg (1976) have pointed out that binding competition curves accurately reflect the affinity of the hormone only if the concentrations of labeled hormone and binding sites are substantially less than the dissociation constant for the labeled hormone. 2) As will be discussed below, equilibrium binding of low concentrations of Iz5I-PDGF is approached very slowly in the absence of efficient mixing. Binding periods of 3 h at 4 "C as employed by Heldin et al. (1981) do not permit approach to equilibrium at low concentrations unless the binding medium is kept well mixed. If equilibrium binding is not approached at low lZ5I-PDGF concentrations, the true K d will be overestimated. The binding of high concentrations of lZ5I-PDGF is relatively insensitive to the kinetic problems discussed; therefore, the protocols used by both Heldin et al. (1981)  Kinetics of PDGF Binding a t 4 "C-At 4 "C the rate of binding of low concentrations of lZ5I-PDGF to monolayer cultures is relatively slow (Fig. 7A). Continuous gentle agitation of the cultures during incubation greatly increases the rate of binding (Fig. 7B), but even with gentle shaking, complete equilibrium binding is achieved by 4 h only at high (160 PM) and very high (300 pg/ml of CMS-I11 PDGF) concentrations of PDGF.
There are several factors which may contribute to the apparently longer times needed to approach equilibrium binding in our system than have been reported for EGF (e.g. Carpenter et al., 1975), insulin (e.g. Cuatrecasas, 1971b), or PDGF (Heldin et al., 1981). Most determinations of the time needed to approach equilibrium are performed with relatively high concentrations of '251-hormone to maximize counts/min bound at early time points. Use of high concentrations of lZ5I-PDGF would underestimate the time required for approach to equilibrium binding at lower concentrations. The very high affinity of PDGF-cultured cells for lZ5I-PDGF probably also contributes to the problems of achieving equilibrium binding of very low concentrations of '251-PDGF. At very low concentrations of 1251-PDGF, the concentration of lz5I-PDGF is comparable to, or greater than, the concentration of receptors (Table 111) and as much as 50% of the lZ5I-PDGF in the binding medium is cell-bound after 4 h of incubation with gentle shaking at 4 "C. Thus, for equilibrium binding to be approached, the cells must have effective access to the entire volume of binding medium. Since the cells occupy a monolayer of 10 pm in a fluid depth of 500 pm, it is possible that, without adequate mixing, a local zone of depleted medium is established above the cell monolayer. Further binding would depend upon the rate at which this zone was replenished by diffusion from the medium. Gentle shaking may help to prevent the formation of this layer, Kinetics of PDGF Binding and Degradation at 37 "C-The initial rate of 1251-PDGF binding was greater at 37 "C than at 4 "C ( Figs. 3 and 7). At 37 "C, however, cell-associated 1 did not increase to a stable plateau value, but instead, reached a maximum value and began to decline soon thereafter ( Fig. 3 and Heldin et al., 1981). Coincident with this decline, trichloroacetic acid-soluble ' ' ' I appeared in the incubation medium. After 9 h of incubation with 40 PM 1251-PDGF, the amount of cell-associated lZ5I was reduced to less than 25% of the initial value (Fig. 3A). Part of the decline was due to depletion of lZ51-PDGF in the binding medium (90% by 9 h). However, most of the decline reflected a true decrease in the binding capacity of the cells since an additional 30-min incubation in fresh 'Z51-PDGF-containing medium did not restore lZ5I-PDGF binding to the value seen during the fwst 30 min of incubation. Concomitant with the decrease in cell-associated I there was an increase in trichloroacetic acid-soluble ' ' ' I (degraded lZ5I-PDGF) in the medium. At 320 PM '251-PDGF similar results were obtained (Fig. 3B) except that maximal binding was achieved earlier (30 min), medium depletion was less severe (18% at 9 h), and the rate of appearance of trichloroacetic acid-soluble Iz5I was somewhat greater.
In experiments in which lZ5I-PDGF was bound at 4 "C, followed by rinsing with saline and reincubation at 4 or 37 "C in the absence of lZ5I-PDGF in the medium, greater than 90% of the radioactivity remained cell-associated at 4 "C for at least 4 h, while greater than 80% of the radioactivity was released in trichloroacetic acid-soluble form within 1 h at 37 "C (data not shown).
The reduction in binding capacity following incubation with PDGF cannot be due to masking of PDGF receptors during the incubation period, since the PDGF used for incubation