Reversible Binding of Platelet-derived Growth Factor-AA, -AB, and -BB Isoforms to a Similar Site on the “Slow” and “Fast” Conformations of ara-Macroglobulin*

The mechanism by which the platelet-derived growth factor (PDGF)-binding protein, a2-macroglob- ulin (azM), modulates PDGF bioactivity is unknown, but could involve reversible PDGF-azM binding. Herein we report that >70% of ‘”I-PDGF-BB or -AB complexed to azM was dissociated by SDS-denatura-tion followed by SDS-polyacrylamide gel electropho- resis, i.e. most of the binding was noncovalent. Reduction of the PDGF*azM complex following denaturation dissociated the cytokine from azM by >go%, suggesting covalent disulfide bond formation. Approximately 50% of the growth factor was dissociated by lowering the pH from 7.5 to 4.0. 12’I-PDGF-BB bound azM in a time-dependent manner (tllz = -1 h), reaching equilibrium after 4 h. The ‘”I-PDGF*BB/a2M complex dis- sociated more slowly (tllz = -2.5 h). “Slow” and “fast” azM bound nearly equal amounts of PDGF-AB or -BB. Trypsin treatment converted PDGF-BB/azM complex to the fast conformation but did not release bound “‘I- PDGF-BB. All PDGF-isoforms (AA, -AB, and

igin. Numerous studies that implicate PDGF as a key mediator in the normal processes of development, tissue maintenance, and wound healing have been reviewed (7). PDGF has also been proposed as a link in the progression of diseases such as atherosclerosis (8) and pulmonary fibrosis (9,10). Two different monomeric chains of PDGF (A and B) give rise to three possible dimers (AA, AB, BB), and these dimeric isoforms recognize dimeric cell-surface receptors composed of a and/or @ chains (11). PDGF isoforms recognize their receptors according to a receptor subunit model, i.e. AA, AB, BB dimers bind to aa receptors; BB, AB dimers to a@; and the BB dimer to p@ receptors (12). The different subtypes of PDGF and receptors could allow for a fine tuning of cell responsiveness, since different cell types can vary greatly in the ratio of isoforms secreted and in the receptor composition which the target cell possesses (13). Cell responsiveness to PDGF in uitro can be further modulated by other growth factors such as TGF-@ (14) and by PDGF-binding proteins (15).
The major PDGF-binding protein is az-macroglobulin (aZM) and PDGF-a2M complexes have been isolated from plasma and from macrophage supernatants (16)(17)(18)(19). This 725-kDa protein apparently serves mutiple functions as a cytokine-binding protein (20), wide spectrum proteinase inhibitor (21)(22)(23)(24)(25), and immune regulator (26,27). azM was first described as a proteinase inhibitor and the mechanism whereby native or electrophoretically "slow" azM covalently entraps proteinases has been extensively studied (see Ref. 25 for review). A proteinase cleaves azM in its "bait region," and this cleavage induces a conformational change in the aZM molecule which entraps the proteinase. The conformational change makes the azM more compact and hence has greater mobility on nondenaturing gel electrophoresis than the native or slow form of azM. The irreversible triggering of the proteinase trap is mimicked by primary amines (28), and the electrophoretically "fast" a2M-proteinase or azM-amine complex is receptor-recognized by fibroblasts (29,30) and macrophages (31)(32)(33)(34). PDGF binds both slow and fast forms of azM (15), and PDGF-stimulated fibroblast proliferation (15) and chemotaxis (35) are inhibited by slow a2M. azM inhibits the binding of PDGF to its cell-surface receptor and thus has been suggested to limit the amount of PDGF that is available to bind to these receptors (16). On the other hand, methylamine-modified, fast, azM synergistically enhances the growth promoting activity of human PDGF purified from platelets (15). Thus, aZM modulates the biological activities of PDGF in uitro. It has been speculated that a2M could serve mutiple functions as a PDGF-binding protein in uiuo (16), including 1) modulation of PDGF biological activity as discussed above, 2) protection of PDGF against proteolytic degradation, and 12837 This is an Open Access article under the CC BY license.

PDGF Isoforms Binding az-Macroglobulin
3) clearance of PDGF from the circulation. a2M also binds and modulates the biological activities of several other growth promoting cytokines, including transforming growth factor-@ (TGF-@) (36)(37)(38)(39)(40)(41)(42), tumor necrosis factor-a (TNF-a) (43), basic fibroblast growth factor (bFGF) (44), interleukin-l@ (IL-l@) (45), interleukin-6 (IL-6) (46), nerve growth factor (47) and human growth hormone (48). The biological activity of some of these cytokines is inhibited when bound to a2M, as is the case with TGF-@ (42). Others such as IL-6 (46) and PDGF (15, 18) retain biological activity when complexed to this binding protein. While it is becoming increasingly apparent that a number of different cytokines utilize a2M as a binding protein, the sites on the azM molecule to which these factors bind could differ. Determining whether or not these different cytokines compete for the same binding site(s) on this protein could be important in discerning the in uiuo effects of azM as a potential modulator of cytokine activity. This is likely since many cell types (e.g. macrophages) secrete mixtures of these cytokines and azM (49). Also, it has been suggested that these cytokines bind a2M either covalently, noncovalently, or via a mixture of covalent and noncovalent associations (see Ref. 20 for review).
PDGF has been reported to bind a Z M covalently (17,18). However, other studies suggest that at least part of the PDGF is noncovalently bound, since it can be released by 1 M acetic acid treatment (16,19). Thus, the first objective of the present study was to establish the ratio of covalent/noncovalent binding of PDGF to a2M slow and fast forms. The issue of reversible binding is of paramount importance if one is to elucidate the mechanism(s) by which this binding protein modulates PDGF activity. Furthermore, such information may reconcile the apparent differences that currently exist in the literature regarding covalent and noncovalent associations between PDGF and a2M (16)(17)(18)(19). Second, because only PDGF purified from platelets has been reported to bind a z M , it is not known if one or all of the three different PDGF isoforms (AA, AB, BB) bind to a similar site on the a Z M molecule. Thus, we sought to establish the competition of these isoforms for lZ5I-PDGF-BB binding to a z M . Third, since several other cytokines bind a z M , it is of major importance to determine the possible competitive nature of these factors for PDGF binding to a2M. Herein, we report that the majority of PDGF binding to slow or fast aZM is reversible or noncovalent, and that all PDGF isoforms compete for a similar site on a Z M ,

MATERIALS AND METHODS
Growth Factors and anM-Human PDGF purified from platelets, TGF-01, and TGF-/32 were purchased from R & D Systems (Minneapolis, MN). PDGF-AA, -AB, and -BB isoforms, bFGF, TNF-a, ILlp, and IL-6 were obtained from Upstate Biotechnology (Lake Placid, NY). Human anM was purchased from Calbiochem (San Diego, CA) and bovine plasma aZM was obtained from Boehringer Mannheim.
Conversion of aaM from Slow to Fast-All azM preparations were subjected to dialysis against 100 volumes of distilled water to precipitate fast anM, which was present to some extent in all preparations. The native apM was tested for trypsin binding activity as described below. Slow anM was converted to fast a2M by incubation with 25 mM methylamine (Tris-HC1, 50 mM, pH 8.0) overnight at 25 "C or by incubation with a 4:l molar excess of trypsin or plasmin for 30 min at room temperature. Excess methylamine was removed from tu,M-methylamine complexes by dialysis against 100 volumes of 50 mM Tris-HC1, pH 8.2, at 4 "C. Excess trypsin or plasmin was removed from azM-trypsin or anM-plasmin complexes by gel filtration chromatography (Superose 6 FPLC). Fast azM preparations were stored at 4 "C in 50 mM Tris, pH 8.2, to prevent precipitation. Slow azM was stored at 4 "C in 20 mM sodium citrate buffer, pH 6.5. a2M preparations were tested for PDGF contamination as described previously (15).
Trypsin Binding Assay for apM-Native azM was tested for trypsin binding activity by a modification of a previously described method (50). Increasing concentrations of aZM were added to 96-well microtiter plates to a final volume of 50 pl/well in 25 mM Tris-HC1, 150 mM NaCI, pH 7.4. 3 pl/well of 1 mg/ml trypsin (Sigma) was then added for 10 min to bind available native azM, followed by the addition of 6 pl/well of 1 mg/ml soybean trypsin inhibitor (Sigma), which inhibited all trypsin activity not bound to aZM. After 10 min, 80 pl/well of 0.1 M Tris-HCI, 10 mM CaClZ buffer, pH 8.0, was added, followed immediately by 100 pl/well 3 mM Na-benzoyl-DL-argininep-nitroanilide (BAPNA) hydrochloride (Sigma). The colorimetric reaction was stopped by the addition of 10 pl of glacial acetic acid. The increase in the optical density read at 405 nm is proportional to the quantity of active trypsin (covalently trapped within a2M) available to convert the BAPNA substrate to its product.
Gel Filtration Chromatography-PDGF. a2M complexes were routinely prepared by incubating 1 ng of human 'T-PDGF-AB or human recombinant lZ5I-PDGF-BB with 100 pg of azM fast or slow form for 24 h at 37 "C. These mixtures were isolated by loading onto a gel filtration, molecular weight exclusion column (Superose 6 FPLC, Pharmacia LKB Biotechnology Inc.) equilibrated in phosphate-buffered saline, pH 7.5, operating at a flow rate of 0.5 ml/min. The column was standardized with the following molecular mass markers: aprotinin (6.5 kDa), cytochrome c (12.4 kDa), carbonic anhydrase (29 kDa), bovine serum albumin (66 kDa), &amylase (200 kDa), apoferritin (440 kDa), thyroglobulin (669 kDa), and blue dextran (Vo). Fractions (1 ml) were counted on a y-counter to measure radiolabeled PDGF or assayed for anM by ELISA as described below. Protein was routinely measured by absorbance (280 nm) to ensure that identical amounts of PDGF/a2M were loaded onto the column.
Gel Electrophoresis-Electrophoresis of the PDGF/azM mixtures in a nondenaturing (5% Tris-borate) gel was performed as described previously (51). azM that was incubated with lZ5I-PDGF-AB or -BB as described above for gel filtration chromatography was mixed with Tris-borate buffer (10 pg of azM in 40 pl added to 4 pl of 10 X buffer with 10 p1 of glycerol) and electrophoresed on a native 5% gel. Native gels were either stained with Coomassie Blue and dried for autoradiography or transferred to nitrocellulose, blocked with 5% BSA for 2 h, then shaken with 1:2000 sheep anti-human azM-horseradish peroxidase (Vector Laboratories) in BSA/PBS-Tween for 4 h and developed. For SDS-polyacrylamide gel electrophoresis 80 pl (20 pg) PDGF/a2M samples were mixed with an equal volume of 10% SDS and heated to 37 "C for 1 h, and half of each resulting sample received 10% 2-mercaptoethanol. Samples were run on a 7.5% SDS gel. The gels were stained with Coomassie Blue and dried for autoradiography and y-counting.
azM ELISA-We developed an enzyme immunoassay for human aaM using Immulon 2 Removawell flat-bottomed wells (Dynatech, Chantilly, VA) which were coated 100 pllwell with a 1:lOOO dilution of rabbit anti-azM in PBS (DakoPatts, Santa Barbara, CA) and incubated overnight at 4 "C. The following day, the wells were washed five times with PBS containing 0.05% Tween-20 (PBST) and then 200 pllwell of 1% BSA in PBS was added. After a 5-h incubation at 4 "C, the wells were washed five times with PBST, and 100 pl/well of standard a2M or unknown sample diluted in 1% BSA-PBS were added and incubated overnight at 4 'C. Zero antigen controls and horseradish peroxidase-antibody blanks received only 1% BSA-PBS. The following day, the plate was washed five times with PBST and 100 pllwell of horseradish peroxidase-conjugated, sheep anti-human azM (Serotec, Kidlington, Oxford, United Kingdom) diluted in 1% BSA-PBS (1:5000) was added and incubated at room temperature for 5 h. After washing five times with PBST, the wells were developed for 15-30 min with the diammonium salt of 2,2'-azinobis-3-ethylbenzothiazoline-6-sulfonic acid (Sigma) containing 0.003% hydrogen peroxide. The absorbance (405 nm) was measured on a Titertek Multiskan 96-well plate reader (Flow Laboratories, Mclean, VA). Doubling dilutions of anM (2-1000 ng/ml) were used, and the linear range of the curve was between 5-500 ng/ml.

Isolation of PDGF.azM Complexes by Gel
Filtration Chromatography-To establish the elution profile of the PDGF. a z M complex, human lz5I-PDGF-AB was incubated with human azM in Ham's F-12 buffer with HEPES, CaCL, and 0.25% BSA (binding buffer) at pH 7.4 and the mixture was loaded onto a Superose 6 FPLC column equilibrated in PBS at the same pH. Later experiments were designed to study the formation of this complex on the same column and the subsequent dissociation of human PDGF-AB or recombinant PDGF-BB from human or bovine a2M following a decrease in the pH of the binding buffer. A separate series of experiments addressed the formation of the PDGF.a2M complex on this column in the absence or presence of the three recombinant human PDGF isoforms or several other cytokines (TGF-8, bFGF, IL-18, IL-6, TNF-a) that have been reported to bind to a2M. Human a2M eluted as single peak as measured by ELISA and the major peak of human "'1-PDGF-AB coeluted with this immunoreactive apM (Fig. 1).

Binding of PDGF to Slow and
Fast a2M-Isolation of the ' ' ' 1 -PDGF-AB -a2M complex by nondenaturing gel electrophoresis was performed to demonstrate the purity of slow and fast a2M preparations and to establish the amount of human plasma-derived 12'I-PDGF-AB bound to these forms of a2M. Both purified forms of slow human a2M and fast methylamine-modified a2M bound human "'I-PDGF-AB ( Fig. 2A). A  quantitative analysis of the 12'I-PDGF-AB bound to these a2M forms showed that approximately equivalent amounts of PDGF bound either the fast or the slow form (Fig. 2B).
Preferential Noncoualent Binding of PDGF to azM-To measure the extent of covalent or noncovalent binding of "'I-PDGF-AB to native a2M, the I2'I.PDGF-AB .anM complex was subjected to SDS denaturation or 2-mercaptoethanol reduction prior to SDS-polyacrylamide gel electrophoresis. As expected, SDS treatment dissociated the a2M molecule into its -180-kDa subunits as determined by protein staining and autoradiography of 12'I-a2M-methylamine (Fig. 3). No "' I-PDGF-AB was initially detected by autoradiography at the molecular weight of this a2M subunit (Fig. 3A). Instead, the l2'1-PDGF-AB was detected at the molecular mass of the -30-kDa PDGF dimer (SDS treatment) or the -15-kDa PDGF monomer following reduction. However, a longer exposure of the SDS-denatured 12'I-PDGF-AB. a2M complex revealed a minor autoradiographic signal, indicating that some PDGF remained covalently bound to the a2M (Fig. 3B). The same experiments were performed with recombinant 12'I-PDGF-BB. y-Counting the intact "'I-PDGF-BB . a2M complex from the nondenaturing gels and the a2M subunits from denatured or reduced SDS gels (Fig. 4) demonstrated that 20-30% of the "'I-PDGF-BB remained bound to the denatured a2M subunit and less than 10% of the '2sI-PDGF-BB remained bound to the binding protein subunits after reduction (Fig.  4). Similar results were obtained for bovine "'1-PDGF-BB. a2M complex. pH-dependent Release of PDGF from anM-Denaturation of the PDGF.a2M complex by treatment with SDS or 1 M acetic acid was shown to release the majority of PDGF bound to a2M, and these harsh conditions are known to dissociate the anM molecule into its subunits. Thus, we studied the possible release of PDGF from a2M over a range of "physiological" pH (7.5-4.0) where the anM molecule remains intact. Recombinant human lz5I-PDGF-BB incubated with bovine slow a2M at pH 7.5 at 37 "C for 24 h formed a high molecular weight complex that eluted on a Superose 6 FPLC column at the same position as the plasma-derived human lZ5I-PDGF-asM (see Fig. 1). By lowering the pH of solutions containing PDGF. anM complexes with acetic acid for 1 h prior to loading onto the Superose 6 column, a pH-dependent decrease in lZ5I-PDGF-BB bound to bovine anM was demonstrated (Fig. 5). The anM retained its tetrameric (-725 kDa) structure across this pH range, and the amounts of anM that eluted at this high molecular mass were equivalent as determined by protein absorbance (280 nm). The quantity of lz5I-PDGF-BB bound to a2M, i.e. isolated from the PDGF. anM complex on the gel filtration column, decreased -50% as the pH was decreased from 7.5 to 4.0 (Fig. 5). At pH less than 4.0, the a2M was denatured and migrated as subunits with molecular mass >440 kDa.
Time  PDGF-BB (2 ng/ml) were incubated at various time points (30 min to 24 h) at 37 "C in 0.5 ml of binding buffer prior to loading onto a Superose 6 FPLC column. The a2M peak zone (20-28 min) was pooled and bound lZ5I-PDGF-BB quantitated by y-counting as demonstrated in Fig. 1. The association time course showed that PDGF-BB bound a2M with a tIl2 = -1 h, and equilibrium was reached at -4 h (Fig. 6A). Of the total amount of free lZ5I-PDGF-BB added to the binding buffer (1 ng = 25,000-27,000 cpm), approximately half of this radioligand bound to a2M (-12,000-14,000 cpm) after 24 h of incubation. Identical aliquots of lZ5I-PDGF-BB. a2M complex isolated by FPLC after 24 h of incubation (the end point on Fig. 6A) were allowed to further incubate for various time points (0-24 h) at 37 "C to measure the extent of lZ51-PDGF-BB dissociation at pH 7.4. Under these conditions, 50-60% of the bound 12'II-PDGF-BB dissociated from a2M ( t1/2 = -2.5 h) and an equilibrium was reached by -6 h (Fig. 6B).
PDGF Remains Bound to a2M following Proteolytic Conversion to the Fast Form-In order to determine whether or not PDGF would be released from slow a2M during the proteolytic conversion of the a2M from slow to fast by trypsin, recombinant human '251-PDGF-BB complexed to bovine slow a2M at pH 7.5 was treated with a 4 1 molar excess (trypsin/azM) for 20 min prior to nondenaturing gel electrophoresis. Autoradiography of the 1251. PDGF-BB a2M complex before and after trypsin exposure showed that this proteolytic treatment converted the a2M to the electrophoretically fast form, but the 12'II-PDGF-BB remained bound to the trypsin-activated a2M (Fig. 7).
Competition of PDGF Isoforms and Other Cytokines for 1251-PDGF-BB Binding to a2M-The three different isoforms of PDGF were tested for competitive binding to a2M to evaluate the capacity of this binding protein to potentially modulate the activity of all of the PDGF dimers. Human recombinant "'I-PDGF-BB (2 ng/ml) and bovine a2M (200 pg/ml) were incubated in the absence or presence of an excess of nonradioactive PDGF-AA, -AB, or -BB (8 pg/ml) for 24 h at 37 "C. All three nonradioactive isoforms inhibited the majority of l2'1-PDGF-BB binding as determined by a decrease in the radioactivity associated with the slow bovine a2M peak on the Superose 6 FPLC column (Fig. 8). The same experiment was performed using fast, methylamine-modified, a2M. Importantly, several other cytokines that have been reported to bind a2M (i.e. bFGF, IL-10, IL-6, TNF-a, TGF-01, and TGF-p2) were tested for their ability to compete with the I2'I-PDGF-BB for binding to both slow and fast a2M. The column fractions containing the PDGF.a2M complex (10-30 min elution time) were pooled and counted for radioactivity. Nonradioactive PDGF-AA, -AB, and -BB all inhibited complex formation between '251-PDGF-BB and a2M or 1251-PDGF-BB and a2M-methylamine by 65-80% compared to control treatments that received no excess isoform (Fig. 9). IL-10, IL-6, bFGF, TNF-a, TGF-01, and TGF-02 each at an excess concentration (8 pg/ml) did not inhibit the binding of 12'I-PDGF-BB to slow a2M. TGF-01 inhibited complex formation between '251-PDGF-BB and a2M-methylamine by as much as 50%, while TGF-(32 inhibited this interaction by only 10%. IL-6 and bFGF inhibited '"I-PDGF-BB binding to a2Mmethylamine 30-35%, while IL-10 and TNF-a had negligible inhibitory effects on this complex formation. Bovine a2M and  Fig. 4) was pooled for each column run. None of the cytokines tested, other than the PDGF isoforms, competed for slow a2M, while bFGF, IL-6, and TGF-01 prevented complex formation between PDGF-BBIazM 30-50%. Inactivation of the proteinase binding capacity of the slow a2M by methylamine was measured by the trypsin binding assay described under "Materials and Methods" (inset); closed circles show that slow arM possesses trypsin binding activity that is proportional to an increase in absorbance, while methylamine-activated a2M (open circles) does not bind trypsin.

UX-S
a2M-methylamine bound approximately equivalent amounts of '251-PDGF-BB on the gel filtration column, a result similar to that obtained with human "'I-PDGF-AB binding to human a2M and a2M-methylamine on a nondenaturing gel (Fig. 2). Bovine a2M and a2M-methylamine were not run on a nondenaturing gel to test for differences in electrophoretic migration because, unlike human a2M-methylamine, bovine a2Mmethylamine does not migrate as a fast form in a nondena-turing gel (27). For this reason, the complete conversion of bovine a2M to a2M-methylamine was determined by the inability of the methylamine-treated azM to bind and thus inactivate trypsin (Fig. 9). Trypsin that is trapped within the a2M molecule is still able to react with small molecules such as the BAPNA reagent (50), producing a yellow product that increases in proportion to the amount of slow a2M in the reaction mixture (Fig. 9). These data demonstrated that methylamine treatment inhibited the trypsin binding capacity of bovine slow a2M.

DISCUSSION
Earlier studies on plasma-derived PDGF introduced the concept that this growth factor forms covalent bonds with its binding protein, a2M (17,18). In the process of isolating and purifying PDGF from plasma and macrophage supernatants, we and others observed that it was necessary to acidify these biological fluids in order to separate the PDGF from its higher molecular weight-binding proteins before the growth factor could be detected by immunoassay or receptor assay (16,19). The principal PDGF-binding proteins in these fluids were identified as a-macroglobulins and they were found to inhibit the binding of PDGF to either its cell-surface receptor or anti-PDGF antibodies, presumably by masking the receptor and antibody recognition site on the growth factor (19). The observation that PDGF could be detected after acidification suggested that at least a portion of the PDGF complexed to a2M was bound noncovalently. The issue of covalent versus noncovalent binding of PDGF to a2M is key to understanding the biological role(s) that this binding protein could serve in affecting the growth promoting activity of this cytokine. For example, aZM has been proposed as a clearance protein for PDGF released into the circulation following platelet degranulation and the PDGF.a2M complex could be cleared in the liver via a2M receptors on hepatocytes (52). In extravascular tissues, macrophages, among other cell types, produce PDGFlike molecules and azM (19,53,54), and the proliferative response of fibroblasts to PDGF may be inhibited or enhanced by a2M, depending on whether it is in the slow form or the fast receptor-recognized conformation (15). Thus, azM could serve as a clearance pathway for PDGF in the circulation and in extravascular tissues, but also as a positive or negative regulator of growth factor activity. We postulated that the control of PDGF-stimulated growth by a2M likely involves the release of the growth factor from this binding protein, allowing PDGF to bind to its own receptor and trigger a mitogenic response (15). For this reason, the observation that the majority of the PDGF is bound to azM noncovalently (Figs. 3 and 4) and the demonstration that approximately 50% this growth factor can be released from its binding protein by lowering the pH of the incubation medium from 7.5 and 4.0 (Fig. 5) or in a time-dependent manner at pH 7.4 ( Fig. 6) are consistent with our hypothesis that PDGF binding to azM is reversible. Such information is basic to our understanding of the mechanism(s) by which the binding protein influences cytokine activity.
Human lZ5I-PDGF-AB and recombinant human lZ5I-PDGF-BB both bound human or bovine slow azM, as well as fast a2M that was prepared by reaction with either methylamine, plasmin, or trypsin. All three recombinant PDGF isoforms were observed to compete for lZ5I-PDGF-BB binding to either a2M or a2M-methylamine, suggesting that these isoforms all bind to a similar site on the azM molecule and that conversion from the slow to fast conformation by methylamine does not alter PDGF isoform binding. Indeed, azM and azM-methylamine that had been incubated with radiolabeled PDGF and then isolated by nondenaturing gel electrophoresis were found to contain approximately equivalent amounts of human lZ5I-PDGF-AB or human recombinant lZ5I-PDGF-BB. It will be of interest to determine whether or not the biological activities of PDGF-AA, -AB, and -BB are modulated by azM in a similar manner. Interestingly, the lZ5I-PDGF-BB remained bound to bovine a2M after the lZ5I-PDGF-BB. aZM complex was treated with an excess of trypsin to convert the a2M to fast form (Fig. 7). Thus, proteinases apparently do not displace PDGF bound to azM, and these data suggest that a2M could serve to protect PDGF from proteolytic degradation. Such a role for azM has been suggested for IL-6, which is inactivated by trypsin, but retains IL-6-like activity in the presence of trypsin when complexed to aZM (46). These observations also suggest that PDGF and IL-6 interact with azM by a mechanism different from that of proteinases and primary amines.
The  6). A calculation of the rate constants for association ( kl) and dissociation ( k2), which must assume an excess of radioligand over receptor (or binding protein), were not performed on these kinetic data due to the low molarity of lZ5I-PDGF-BB (6.7 X lo-" M) relative to azM (3 X

M);
i.e. there was likely an excess of binding sites for PDGF. Further studies using saturation binding and kinetic studies are in progress to address the relative affinities of PDGF for azM versus the PDGF cell-surface receptor. The time course data shown in Fig. 6 for association of lZ5I-PDGF-BB with a2M are closely similar to the time course of lZ5I-bFGF binding aZM (44); i.e. low concentrations of both bFGF and PDGF binding to 200 pg/ml aZM reached an equilibrium state at about 4 h. lZ5I-PDGF-BB was released rapidly from the 1251-PDGF-BB.a2M complex by lowering the pH over a physiological range of 7.4-4.0 (Fig. 5). Such low pH could be encountered within lysozomes following internalization of the PDGF.azM complex via the azM receptor. In this case, it would be of interest to learn whether or not released PDGF (which is acid stable) could then be recycled to the cell-surface and remain bioactive. Thus, it is conceivable that PDGF could be dissociated from the PDGF. azM complex by either depletion of unbound PDGF via internalization by PDGF cellsurface receptors (which would favor the release of azM-bound PDGF to establish a new extracellular equilibrium) or PDGF could be released under conditions where pH is reduced.
Since several other cytokines bind to a2M we sought to establish if these growth factors compete for PDGF binding to azM. Excess concentrations of TGF-Dl, TGF-82, TNF-a, bFGF, IL-1P, and IL-6 were tested for their inhibitory potency in preventing complex formation between lZ5I-PDGF-BB and aZM or azM-methylamine. None of these cytokines inhibited the binding of Iz5I-PDGF-BB to the slow form of azM (Fig.  9). These data suggest that PDGF isoforms could bind to slow azM unhindered in the presence of cytokine mixtures i n vivo.
In contrast, some of these cytokines competed for PDGF binding to fast azM. TGF-Dl, but not TGF-P2, inhibited complex formation between Iz5I-PDGF-BB and a2M-methylamine by -50%. While both TGF-Pl and TGF-P2 bind to fast a2M (42), it is conceivable that these two factors have differing affinities for a2M or bind to different sites on the a2M molecule. Thus, the TGF-81-binding site on the fast a2M could overlap or allosterically modulate the PDGF-binding site. Similarly, IL-6 and bFGF inhibited complex formation between 1261-PDGF-BB and a2M-methylamine by -20-30%, which could suggest some overlap or allosteric hinderance by PDGF and these cytokines binding simultaneously to fast a2M. TGF-81 has been reported to inhibit the binding of 1251-bFGF binding to a2M, while PDGF does not compete for this interaction (44). It is conceivable that certain azM-binding cytokines, such as TGF-81 and bFGF, could mediate the release of PDGF from the fast form of a2M, but not the slow form. Other cytokines apparently do not interfere with PDGF binding to a2M. For example, TNF-a binds fast a2M-methylamine (43) but did not inhibit the formation of the PDGF-BB-a2M-methylamine complex (Fig. 9). Such interactions between these cytokines and the two forms of a2M are likely to be complex and require further study.
A variety of growth-promoting cytokines, including PDGF, are likely to be involved in the processes of normal tissue maintenance and repair, and their abberant expression may well be linked to pathogenic disorders such as pulmonary fibrosis and atherogenesis that are characterized by an increase in cell proliferation and extracellular matrix production (7). A growing number of studies suggest that a2M could play a role in modulating the biological activity, clearance, and degradation of these cytokines in either the circulation or in extravascular tissues (36-48). Because a2M also serves as a proteinase inhibitor and since proteinases irreversibly convert a2M to a fast or receptor-recognized form, this adds another layer of complexity to the problem of understanding the mechanisms by which a number of cytokines interact with a common binding protein. Proteinases apparently are capable of "turning on" the modulatory effects that a2M has for some cytokines in at least two different ways: 1) cytokines such as TNF-a (43) and IL-18 (45) preferentially bind fast a2M (42) and thus the activity of these cytokines would not be expected to be directly regulated by slow form a2M, and 2) PDGF binds both native and proteinase-reacted forms and a2M converted to the fast form by methylamine synergistically enhances the growth promoting activity of PDGF, while the slow form inhibits PDGF-stimulated growth (15). It is presently unclear whether or not a2M-proteinase complexes potentiate or inhibit the growth promoting activity of the different PDGF isoforms, and this is the subject of ongoing studies. Understanding the interactions between the network of proteinases, a2M, and growth factors which bind a2M will be fundamental to our knowledge of cytokine function in vivo.
In summary, the majority of the PDGF-a2M association is noncovalent and all three PDGF isoforms (-AA, -AB, and -BB) bind similarly to a2M. PDGF binds to slow and fast a2M and the slow to fast conversion by trypsin does not dissociate bound PDGF. PDGF-BB dissociates slowly from isolated PDGF. a2M complex ( tlI2 = 2-3 h) at pH 7.4 and reaches an equilibrium state after 6 h. Furthermore, a decrease in pH from 7.4 to 4.0 causes a rapid, progressive release of PDGF from aZM. Because PDGF binding to a2M is reversible, a2M could release PDGF near the cell surface in close proximity to its own receptor.