JOURNAL OF CELLULAR PHYSIOLOGY 149235-243 (1991)

Differential Binding, Biological and Biochemical Actions of Recombinant PDGF AA, AB, and BB Molecules on Connective Tissue Cells GARY R. GROTENDORST,* ATSUYUKI IGARASHI, RONALD LARSON, YOSHINAO SOMA, AND MARC CHARETTE Departments of Ophthalmology and Biochemistry & Molecular Biology, College of Medicine, University of South Florida, Tampa, Florida 336 12 (G.R.G., A.I., R.L., Y.S.); Creative BioMolecules, Hopkinton, Massachusetts 0 I748 (M.C.) We have compared the biological and biochemical properties of recombinant PDGF AA, AB, and BB using three types of fibroblastic cells: NIH/3T3, human skin fibroblast, and fetal bovine aortic smooth muscle. PDGF binding, receptor autophosphorylation, phosphatidyl inositol hydrolysis, as well as chemotactic and mitogenic responses of the cells were analyzed. PDCF-AB and PDGF-BB showed similar receptor binding, receptor autophosphorylation, and potent biological activity for all three of the cell types tested. In contrast, PDGF-AA was biologically active only for the NIH/3T3 cells in which binding sites for PDGF-AA were abundant, but was inactive for bovine aortic smooth muscle cells and human skin fibroblasts in which binding sites for PDGF-AA were absent. PDGF-AA could not induce any biochemical changes in the human skin fibroblasts or smooth muscle cells. Western blot studies with anti-Type a and p PDGF receptor antibodies indicate that the NIHi3T3 cells contained both PDGF a and p receptors, whereas the human skin fibroblasts and bovine smooth muscle cells contained only detectable levels of p receptors. These results indicate that cells possessing high levels of PDGF p receptors only are capable of responding equally well to either PDGF AB or BB.

PDGF is a dimeric protein consisting of two related but distinct polypeptide chains, A and B (Johnsson et al., 1982). Three forms of PDGF are known to exist (AA, AB, and BB) and all have been shown to exhibit biological activity. Potential differences in the biological actions of the three dimeric forms of PDGF on a specific cell type are of considerable interest, since only certain isoforms appear to be present in platelets (Hammacher et al., 1988) or to be produced by various cell types, such as endothelial cells (Collins e t al., 1987) or smooth muscle cells (Seifert et al., 1984). Heldin et al. (1988) studied the receptors for the three dimers of PDGF using affinity cross-linking techniques and demonstrated the existence of two different types of PDGF receptors based on molecular size. Results from competition binding assays suggested that the type ci receptor bound all three forms of PDGF and the type p receptor bound PDGF-BB with high affinity and PDGF-AB with low affinity. A cDNA clone which encodes a murine PDGF receptor has been isolated and is related to both the v-kit oncogene and the receptor for macrophage colony stimulating factor (Yarden et al., 1986). A cDNA encoding a human PDGF receptor has also been cloned and it revealed 86% overall amino acid homology with the murine PDGF receptor (Gronwald et al., 1988; Claesson-Welsh et al., 1988). Gronwald et al. (1988) have transfected this cDNA into BHK-570 0 1991 WILEY-LISS, INC.

cells, which do not express a detectable amount of PDGF receptor, and they have reported that the receptor bound PDGF-BB but did not bind PDGF-AB. Claesson-Welsh et al. (1988) have isolated cDNA clones which encode the same protein and reported that this receptor bound PDGF-BB with high affinity and PDGF-AB with lower affinity, suggesting that these cDNAs encode the type p receptor. Two groups have isolated cDNAs encoding what appears to be the type ci receptor (Matsui et al., 1989; Claesson-Welsh e t al., 1989). This receptor appears to bind AB and BB with equal affinity and AA with lower affinity. A number of models have been developed concerning the mechanism of PDGF receptor signal transduction (Heldin and Westermark, 1990; Seifert et al., 1989; Escobedo and Williams, 1988). Some of these models suggest that the receptors must dimerize for signal transduction and that different hetero- and homoreceptor dimers (cia, ap, pp) display different ligand binding properties. Other data indicate that a number Received January 7, 1991; accepted April 19, 1991. *To whom reprint requests/correspondence should be addressed at: Department of Cell Biology and Anatomy, University of Miami School of Medicine, P.O. Box 016960, Miami, Florida 33101.

236

GROTENDORST ET AL

of auxiliary proteins can associate with the PDGF purchased from Biomedical Technologies, Inc. (Stoughreceptors (Coughlin et al., 1989; Morrison et al., 1989 ton, MA). Purified acidic FGF was purchased from and 1990; Molloy et al., 1989; Kazlauskas and Cooper, Sigma Chemical Company (St. Louis, MO). Anti-PDGF 1989) and it is possible that these proteins could alter IgG was prepared from goat serum immunized with the ligand binding and signal transduction properties purified human PDGF as described previously (Pencev and Grotendorst, 1988). of the PDGF receptor. In a n attempt to further characterize the properties DNA synthesis stimulation assay of PDGF receptors in fibroblastic cells we have studied Cells were grown in 48-well plates to confluence and the effects of recombinant PDGF AA, AB, and BB on several primary cultures of connective tissue cells and used for assay 3 4 days later. Different dimeric forms of fibroblastic cell lines. These proteins exhibit specific PDGF were added to the media and after 18 hours biological activities which are comparable to those of 3H-thymidine (2 kCi/ml) was added. The cells were the natural products. Our results indicate that cells incubated a n additional 2 hours and washed at 4°C possessing only PDGF p receptor respond equally well three times with PBS and five times with 5% TCA. to PDGF AB and BB, whereas cells possessing both TCA-insoluble materials were solubilized in 0.1 N PDGF a and p receptors respond equally well to all NaOH/O.l% SDS, and the amount of incorporated 3Hthree isoforms of PDGF. These data argue against the thymidine was determined with a Beckman liquid need for heterodimeric receptors (i.e., a p ) for cells to scintillation counter. respond to PDGF AB in a t least certain connective Chemotaxis assay tissue cell types. Chemotactic activity was measured using the modiMATERIALS AND METHODS fied Boyden chamber as described previously (GrotenCell culture dorst, 1984, 1987). The test substance was diluted in Human foreskin fibroblasts were grown from ex- DMEM containing 0.2 mg/ml BSA and added to the plants of newborn foreskin. NIH 3T3 cells were ob- lower well of the blind well chamber. The lower well tained from S. Aaronson (NCI). Bovine aortic smooth was covered with a collagen-coated polycarbonate filter muscle cells were grown from explants of fetal bovine (Nucleopore, 8 pm diameter pores). The upper well was aorta as described previously (Grotendorst et al., 1981). then fixed in place and cells (3 x lo5 cells/assay) Cells were cultured in Dulbecco’s modified Eagle’s freshly released from tissue culture flasks were added medium (DMEM) supplemented with 10% fetal bovine to the upper well in DMEM containing 2 mg/ml BSA. serum a t 37°C in a n atmosphere of 10% CO, and 90% After 4 hours incubation at 37°C in 90% air and 10% air. CO, the filters were removed and the cells were fixed and stained using Diff-Quik stain (Harleco). Cells on Growth factors and anti-PDGF IgG the upper surface were removed by scraping with a PDGF AA, PDGF AB, and PDGF BB were obtained rubber policeman and a nuclear stain of the cells on the from Creative Biomolecules, Inc. These peptides are lower surface was extracted with 0.1 N HC1. The produced in Escherichia coli and the denatured mature absorbance a t 600 nm of this extract was measured PDGF A or B chain peptides purified to homogeneity. with Titertek ELISA spectrophotometer (Flow LaboraThe fully reduced peptides are then renatured by tories, McLean, VA). dialysis of the urea-denatured fully reduced peptides in Receptor binding assay the presence of varying ratios of oxidized and reduced glutathione to control the rate of disulfide bond formaConfluent cultues of NIH/3T3 cells, bovine aortic tion. The samples are tested for biological activity and smooth muscle cells, or human foreskin fibroblasts in the dimeric peptides which form are separated from the 6-well plates were used for the assay. Cells were monomers. One surprising finding is that when mix- washed twice with cold binding medium (serum-free tures of PDGF A and B chain are processed the PDGF DMEM, 20 mM HEPES, pH 7.4, 1 mg/ml BSA) and AB heterodimer is the preferred molecule which forms equilibrated a t 4°C. After aspirating the medium, 0.5 (98% of the total dimers present in the purified sample). ml of binding medium was added to each well containThe purity of the PDGF was accessed by several ing different concentrations of PDGF. The plates were methods. Samples of recombinant PDGF run on SDS incubated for 3 hours a t 4°C on a n oscillating platform. gels without reduction and stained with Commassie Then cells were washed twice in PBS for 10 minutes Blue indicated the PDGF was pure and did not contain and bound PDGF was extracted in 1 ml of 1 N acetic any monomers. The electrophoretic mobility of each acid for 10 minutes. Extracts were lyophilized and isoform was also different from all others. Amino acid analyzed by Western blot. Control experiments using sequence analysis indicated greater than 95% purity porcine PDGF, murine EGF, and acidic FGF to compete and a n equal molar amount of A and B chain in the with the recombinant PDGF AB demonstrated specific PDGF AB preparation. Last, we compared the amount receptor binding in this binding assay (data not shown). of immunoreactive A and B chain peptide present with Western blot analysis specific anti A or B chain antisera. When 10 ng of AA or 10 ng of BB was compared to 20 ng of AB we detected Samples were run on 12% SDS polyacrylamide gels. equivalent amounts of A or B chain in each sample. After electrophoresis, peptides on the gel were electroThis confirms the amino acid sequence analysis data blotted onto nitrocellulose filters and the filters were and indicates that the PDGF AB preparation would incubated in Tris-buffered saline (100 mM NaC1, 50 contain no more than 5% BB. Purified murine EGF was mM Tris-HC1, pH 7.4) containing 2.5 mgiml non-fat

RECOMBINANT PDGF AA, AB, BB AND CONNECTIVE TISSUE

powdered milk (TBS-milk) for 4 hours. The filters were incubated overnight in the presence of 15 pg/ml antihuman PDGF IgG diluted in TBS-milk, washed five times in TBS-milk 95 minutes each), and incubated with alkaline phosphatase-conjugated affinity-purified rabbit anti-goat IgG (Organon Teknika Corp., West Chester, PA) at 1: 1,000 dilution in TBS-milk for 90 minutes. The filters were washed five times in TBSmilk and the antigens were detected using a n alkaline phosphatase substrate kit (Kiregaard and Perry Laboratories, Inc., Gaithersburg, MD).

Phosphatidylinositol hydrolysis Confluent monolayers of NIH/3T3, human foreskin fibroblasts, or bovine smooth muscle cells in 6-well culture dishes were labeled for 24 hours with 4 pCi/ml 3H-inositol (Dupont-New England Nuclear) in DMEM containing 10% FCS. Cells were then rinsed with and incubated for 30 minutes a t 37"C, i n a buffer containing 20 mM LiC1, 110 mM NaC1, 5 mM KC1, 1 mM CaCl,, and 20 mM Hepes pH 7.2. Following this incubation, PDGF AA, AB, or BB was added to cells, and reaction was terminated by replacement of the Hepes buffer with 1 ml ice-cold 10% perchloric acid. Cells were then incubated for 30 minutes a t 4°C to extract inositol phosphates; acid-soluble material was neutralized with 5 N KOH in the presence of 20 mm Hepes pH 7.4, and extracts were applied to columns containing Dowex Ag 1 x 8 formate anion exchange resin (BioRad). Inositol phosphates were eluted from columns by increasing ammonium formate ion concentration, and 3H-inosito1 phosphates were detected by liquid scintillation counting. Receptor auto-phosphorylation studies and type (Y and p PDGF distribution studies For studies in intact cells confluent cultures of cells in 6-well plates were incubated for 15 minutes at 37°C in 1ml of DMEM containing 0.1 mg/ml BSA and 5 to 20 ng/ml of PDGF-AA, AB, or BB. After washing twice with ice-cold PBS, cells were scraped off the plates with a rubber policeman in PBS. Cells were pelleted by centrifugation for 5 minutes at 2,OOOg and resuspended in 1% Triton X-100, 10 mM Tris pH 7.4, 3 mM phenylmethysulfonyl fluoride. After centrifugation a t 20,OOOg (10 min) the supernatant was collected and further processed. For total receptor tyrosine phosphorylation this sample was lyophilized and solubilized in reducing SDS-sample buffer prior to electrophoresis on SDS polyacrylamide gels. Samples were r u n on 7.5% SDS polyacrylamide gel and electroblotted to nitrocellulose. Non-specific binding was blocked by incubation in TBS containing 2% bovine serum albumin for 2 hours. Tyrosine phosphorylated peptides were detected by incubation with anti-phosphotyrosine antibody PY20 (ICN Biochemicals, Cleveland, OH) a t 1:1,000 dilution overnight a t 25°C. For the second antibody, alkaline phosphataseconjugated affinity-purified goat anti-mouse IgG (Organon Teknika Corp., West Chester, PA) was used a t 1 5 0 0 dilution. For analysis of Type (Y and p PDGF receptors, membranes were extracted from the cells and analyzed as above, except that rabbit anti-Type (Y or Type p antisera was used a s primary antibody a t

237

1 5 0 0 dilution and alkaline phosphatase-conjugated goat anti-rabbit IgG (Organon Teknika) was used a s a second antibody. Membrane preparations were made from confluent cultures of NIH/3T3 cells. Cell layers were washed twice with ice-cold isotonic buffer (140 mM NaC1, 10 mM Hepes, 1 mM CaCl,, 2 mM MgSO,) and scraped into isotonic buffer. After centrifugation for 5 minutes at 2,00Og, cells were resuspended in ice-cold sucrose buffer (0.125 M sucrose, 10 mM Hepes pH 7.4) and incubated for 10 minutes. The cells were lysed by 30 strokes of Dounce type homogenizer. The homogenate was checked microscopically for cell lysis and centrifuged for 5 minutes at 4,OOOg to eliminate nuclei, followed by a n additional 30 minute centrifugation a t 15,OOOg to pellet membranes. The pellet was resuspended with receptor solubilization buffer containing 20 mM Hepes pH 7.4, 3 mM MgCl,, 0.1 mM Na3V04, 0.5 mM EGTA, 5 mM NaF, 0.5% Triton X-100, 5% glycerol, and 0.5 mM PMSF. PDGF-stimulated phosphorylation experiments were performed as follows. The membrane preparation (30 pl) was incubated with the indicated amount of the PDGF isoform for 15 minutes at 4"C, followed by the addition of 5 pCi of (y-"P)ATP, and was incubated an additional 15 minutes at 4°C. The reactions were terminated by adding 1 pl of 20 mM cold ATP. Tyrosine phosphorylation was measured by immunoprecipitation with specific antisera. Anti-phosphotyrosine antibody (1 pl) was added and incubated for 1 hour a t 4°C followed by 30 pl of ultrapure protein A-agarose (Genzyme, Boston, MA) and a n additional hour of incubation. After centrifugation (12,OOOg, 5 min) the pellet was washed twice with Tris-buffered saline pH 7.4 containing 0.1% Triton X-100 and the specifically bound peptides eluted with 50 pl of 10 mM phosphotyrosine containing 1 mM EDTA. Following incubation for 20 minutes at 4"C, the sample was centrifuged and the supernatant was lyophilized for SDS gel electrophoresis and autoradiography. For ligand-dependent receptor precipitation the radiolabeled membrane preparation was incubated with 100 p1 of anti-PDGF IgG affinity matrix for 2 hours at 4°C. Bound receptor ligand complexes were recovered by centrifugation (12,OOOg, 5 min), the matrix was washed 3 x with TBS containing 0.1% Triton X-100 and the bound peptides were eluted by resuspension of the pellet in 20 pl of 1 N acetic acid. Following incubation for 20 minutes a t 4"C, the sample was prepared and analyzed as described above. RESULTS Differential biological activity of PDGFs on target cells We examined the biological activities of recombinant PDGF isoforms on two primary connective tissue cell types: skin fibroblasts, vascular smooth muscle cells, and the NIH/3T3 cell line. PDGF is known to act a s both a chemoattractant and a mitogen for cells which express the cognate receptors for this peptide. The data in Figure 1 summarize the results of these studies. All three cell types responded to PDGF AB and BB molecules as both a chemoattractant and a mitogen. The dose-response curves for these proteins were not differ-

GROTENDORST ET AL.

238

MITOGENIC ACTIVITY

CHEMOTACTlC ACTlVlM

400

-

300

-

/QHbE

:/ /y

zl2200 -

BB

‘z 1 5 0 t

X

AA

flpg-a

-

Z 100 W

z

g

0

-

100

4.AB

r T

50.

I I

n

W

01

400

-

300

-

,

0

100 -

.---dAB BB

’A/

z X

y-’

75.

1

8

100

a

/:y4

0

z

v

-

AB BB

4

o/--o-O-o

-

25I

’ 0

nI

’ tO// -o-o--o

AA

rn

0 .

0

BB AB

BOO

/ ’ A’

600 0

8

2 *O

50

I

eo

BB

n I

!X 2

80

40

8

200

0

0.5

o / o

o/o-o-

20

O

-0-

I

M

I 10

20

50

10.0

200

PDGF (ng/ml)

0.5

10

20

50

100

20.0

POGF (ng/rnl)

Fig. 1. Biological activity of the recombinant PDGF AA, AB, and BB molecules in both chemotactic and mitogenic assays for NIHi3T3 cells, human skin fibroblasts, and bovine aortic smooth muscle cells. Cells were cultured and the assays were performed as described under “Materials and Methods.” Each of the cell types was tested for both chemotactic and mitogenic responses to the various isoforms of recombinant PDGF. Only NIH/3T3 cells responded to all forms of PDGF a t the concentrations examined. Both human skin fibroblast

and bovine aortic smooth muscle cells were unresponsive to PDGF AA as a mitogen, although the fibroblast exhibited a detectable response to PDGF AA as a chemoattractant. The dose-response curves parallel the PDGF binding data and are comparable to native PDGF specific activity in these assays. These are the average of duplicate determinations and are representative of six separate experiments. In all panels symbols are AA (o),AB (e), and BB ( A ) .

ent between the cell types. However, a different pattern of response to PDGF AA was seen. NIH 3T3 cells were the only cell type to exhibit both a chemotactic and a mitogenic response to PDGF AA at concentrations which were in the 1-20 ng/ml range. Neither the human skin fibroblasts nor fetal bovine aortic smooth

muscle cells responded to PDGF AA as either a chemoattractant or a mitogen.

Differential binding of PDGFs to target cells We then compared the binding of the recombinant PDGF AA, AB, and BB peptides to NIH/3T3 cells,

RECOMBINANT PDGF AA, AB, BB AND CONNECTIVE TISSUE

(B) HSF

(A) 3T3 1

2

3

4

s

1

2

3

239

(CISMC 4

5

1

2

3

4

5

AA

AB

BB Fig. 2. Binding assay of PDGF-AA, AB, and BB. Cells were incubated with different concentrations of PDGF for 3 hours at 4°C. After washing, bound PDGF was extracted and analyzed by Western blot. A NIHi3T3 cells. 0 ngiml (lane l),5 ngiml (lane 2), 20 ngiml (lane 3), 50 ngiml (lane 4), 100 ngiml (lane 5). B. Human skin fibroblasts. 2 ngiml

(lane l), 5 ng/ml (lane 2), 10 ngiml (lane 3), 20 ngiml (lane 4), 50 ngiml (lane 5). C: Bovine aortic smooth muscle cells. 0 ngiml (lane 11, 10 ngiml (lane 2), 20 ngiml (lane 3), 50 ngiml (lane 4), and 10 ng of standard PDGF (lane 5). These results are representative of three separate experiments.

primary human skin fibroblasts, and fetal bevine aortic smooth muscle cells. The data from the Western blot assays of the receptor-bound PDGF are shown in Figure 2. PDGF AB and BB bound equally well to all three of the cell types examined. The amount of bound peptide increased in a dose-dependent manner. In marked contrast, only the NIH/3T3 cells exhibited a significant level of PDGF-AA binding. No binding was detected with either the human skin fibroblasts or fetal bovine aortic smooth muscle cells. PDGF AB heterodimers are binding to the fibroblasts and the smooth muscle cells, as we can detect PDGF A chain bound with a n A-chain specific antibody. Because AA does not bind to these cells this binding could only arise from AB heterodimer binding (data not shown). These results parallel the differences observed in our biological activity studies of the different recombinant PDGFs and suggest that NIH/3T3 cells possess a PDGF receptor class which is distinct from that found on bovine aortic smooth muscle cells. This receptor appears to bind PDGF AA and is present in non-functional amounts on both the human skin fibroblast and the bovine aortic smooth muscle cell.

PDGF receptor binding assays and the biological action of the PDGF isoforms on these cells and indicate that PDGF AB and BB are equally effective in acting through a receptor which is distinct from that for PDGF AA.

Phosphatidylinositol hydrolysis induced by the PDGF isoforms Because PDGF is a potent stimulator of phosphatidylinositol hydrolysis i t is thought that PI turnover may mediate some of the biological responses to PDGF. To compare the effects of the different isoforms of PDGF on PI turnover, cell layers were labeled overnight with 3H-inositol and treated with 10 nglml of PDGF AA, AB, or BB. After incubation for the indicated period of time, the cells were extracted and the hydrolysis of PIP,, was determined. As seen in Figure 3A, all PDGF isoforms were equally effective in stimulating the hydrolysis of PIP,, in the NIH/3T3 cell line. However, only PDGF AB and BB were active in human skin fibroblasts and the bovine smooth muscle cells. These results are consistent with the differences we measured in both the

Differential tyrosine kinase activity of the PDGF receptors The two PDGF receptors t h a t have been identified to date exhibit auto-tyrosine phosphorylating activity when exposed to the appropriate ligand. To study these receptors in connective tissue cells, we have compared the auto-tyrosine phosphorylating activity of the PDGF receptors present in intact NIHi3T3 cells, human skin fibroblasts, and bovine aortic smooth muscle cells. Initially, we compared the ability of the three PDGF isoforms to stimulate tyrosine phosphorylation of PDGF receptors in each of the three cell types. Confluent monolayers of cells were treated with 20 ng/ml of each of the PDGF isoforms for 15 minutes a t 37°C. The membrane fraction was isolated and the tyrosine phosphorylation was measured by Western blot with a n anti-phosphotyrosine monoclonal antibody (Fig. 4). We found that all three isoforms could stimulate tyrosine phosphorylation in the NIHi3T3 cells. However, only the B-chain containing isoforms were active in the fibroblasts and the smooth muscle cells. These differences in target cell specificity are identical with those in our previous experiments which analyzed biological activity, ligand binding, and inositol turnover and indicate that the NIH/3T3 cells possess receptors which can be activated by PDGF AA whereas the other cell types do not. We then wanted to compare the relative specific activity of three PDGF isoforms on receptor phosphorylation in intact NIHi3T3 cells. Confluent monolayers of NIHi3T3 cells were exposed to increasing concentrations of the recombinant PDGF isoforms for 15 minutes at 37°C and the PDGF receptors were extracted as described under “Materials and Methods.” Tyrosine

240

GROTENDORST ET AL.

phosphorylation of the PDGF receptors was measured by Western blot analysis of the cell extracts with a n 4ooo1 anti-phosphotyrosine monoclonal antibody. The data in Figure 5A indicate that while 5 and 10 ngiml of PDGF AB and BB are capable of stimulating a high level of 3ooo receptor tyrosine phosphorylation, the PDGF AA isoform was not a s effective at these concentrations. Importantly, all three of these isoforms were equally n effective in stimulating PI turnover and DNA synthesis in these cells a t concentrations of 2-10 ng/ml. There“= fore, it appears that the PDGF AA isoform is not as lo00 effective a s PDGF AB or BB in stimulating tyrosine c kinase activity of the total PDGF receptors present in the NIHi3T3 cells. Alternatively, there may be fewer 01 1 0 10 20 30 receptors for PDGF AA or the removal of the phosphate from the AA-receptors may be more rapid than for the W (NINUIES) other receptors types. B) HSF We next analyzed membrane preparations of the NIH/3T3 cells in a n in vitro receptor phosphorylation assay. In our initial experiments, samples were immunoprecipitated with anti-phosphotyrosine IgG and the precipitated phosphorylated peptides were analyzed by SDS polyacrylamide gel electrophoresis and autoradiography. The membrane preparation was incubated with the indicated amount of the PDGF isoform for 15 minutes a t 4“C, followed by the addition of 5 pCi of gamma 32P-ATP,and then incubated for an additional 3 15 minutes at 4°C. The reactions were terminated by addition of excess ATP and precipitation buffer, as described under “Materials and Methods.” The results of the anti-phosphotyrosine precipitation in Figure 5B indicate that both PDGF AB and BB stimulate a 5-10 fold increase in tyrosine phosphorylation of a PDGF receptor. In contrast, equivalent concentrations of PDGF AA were much less effective and stimulated only a 2-3 fold increase. We then performed ligand dependent receptor precipitation using a n anti-PDGF IgG affinity matrix to absorb the ligand-phosphorylated receptor complex. Here we wanted to examine whether the ligand added to the reaction mixture could be co-precipitated with the phosphorylated receptor. Phosphorylated membrane preparations as prepared in Figure 5B were precipitated using anti-PDGF IgG coupled to Affi-gel 10 agarose beads. As seen in Figure 5C, striking differences were apparent between the AA and the AB/BB isoforms. Both PDGF AB and BB co-precipiRYE ( M I N m ) tated with a 180 kD phosphorylated peptide, whereas the PDGF AA did not co-precipitate with a peptide Fig. 3. Phosphatidylinositol hydrolysis induced by PDGF AA, AB, which incorporated any 32P-label. Even at amounts up and BB isoforms in NIHi3T3 cells, human skin fibroblasts, and bovine to 80 ng of PDGF AA we could not precipitate a aortic smooth muscle cells. Confluent monolayers of each of the cell cultures were labeled for 24 hours with 4 kCi/ml of 3H-inositol. The phosphorylated receptor by this method. These data cultures were then treated with 10 ngiml of the indicated form of support the existence of a PDGF AA receptor which is PDGF and the hydrolysis of PIP, was determined as described under distinct from the one through which PDGF AB or BB Materials and Methods. The data in A indicate that all isoforms of acts. PDGF were equally effective in the stimulation of PIP, hydrolysis in

A) NIH/3T3

E B

~

NIH/3T3 cells. The data in B and C show that PDGF AB and BB are active as inducers of PIP, hydrolysis in human skin fibroblast and bovine smooth muscle cell cultures, but PDGF AA had no activity at the tested concentration. These data are representative of the results of three separate experiments.

PDGF (Y and p receptor levels in skin fibroblasts, vascular smooth muscle cells, and NIH/3T3 cells Because of the differences in biological and biochemical activities of the PDGF isoforms in the three cell types, we wanted to determine in a n independent fashion the relative levels of the Type a and Type p

RECOMBINANT PDGF AA, AB, BB AND CONNECTIVE TISSUE

m C AAABBB

HSF

C AAABBB

24 1

SMC

C AA A 8 BB

Fig. 4. Receptor phosphorylation by PDGF isoforms. Confluent monolayers of the indicated cell types were exposed to PDGF AA (40 ngiml), AB (10 ngiml), or BB (10 ngiml) for 15 minutes a t 37°C. The tyrosine phosphorylated receptors were detected with a n anti-phosphotyrosine monoclonal antibody a s described under Materials and Methods. PDGF AB and BH were active in all three cell types. PDGF AA exhibited activity in the NIHi3T3 cells only.

receptors on these cell types. Specific anti-sera for cell types exhibited similar dose-response curves to either the Type (Y or Type p PDGF receptor was PDGF AB and BB suggesting that the NIH/3T3 cells generously provided by S. Aaronson (NCI, Bethesda) have unique receptors that are capable of binding and Western blot analysis was performed on membrane PDGF AA which is not present on the fibroblast or extracts of the three cell types. The results of these smooth muscle cells tested here. The differences seen in studies demonstrated that NIH/3T3 cells possess both the biological response of the various cells to the Type (Y and Type p PDGF receptors, whereas in our different PDGF isoforms are consistent with the results human skin fibroblasts and bovine smooth muscle cells of our biochemical studies. These studies demonstrated only the Type p receptor was detected (Fig. 6). These that NIH/3T3 cells contained binding sites for PDGF results suggest that the presence of the PDGF a AA, whereas no PDGF AA binding sites were detected receptor is essential for cells to respond to PDGF AA on either fibroblast or smooth muscle cells. Yet all three cell types expressed nearly equivalent levels of but not AB or BB. binding sites for PDGF AB and BB. Second, our studies DISCUSSION of PIP, hydrolysis and receptor auto-phosphorylation At the current time, two distinct PDGF receptors, indicate that while all three isoforms of PDGF are designated Type alpha and beta (or Type a and p), have active in stimulating these biochemical changes NIH/ been identified (Yarden e t al., 1986; Claesson-Welsh 3T3 cells, only B-chain containing isoforms were active et al., 1988; Gronwald et al., 1988; Matsui et al., 1989). in the fibroblasts and smooth muscle cells, with PDGF Data from several laboratories indicate that the Type p AA exhibiting no activity. PDGF AA appears to be receptor preferentially binds PDGF-BB and is not much less effective a t stimulating total PDGF receptor activated or is only poorly activated by PDGF AB or AA auto-phosphorylation and does not co-precipitate with a (Claesson-Welsh et al., 1988; Gronwald et al., 1988; 180 kD phosphorylated protein as do PDGF AB or BB Matsui et al., 1989). In recombinant cells, the Type (Y when anti-PDGF IgG is used a s the precipitating receptor exhibits a different pattern of binding activity antibody. Analysis of the presence of Type (Y and p in that it appears to be activated equally well by PDGF PDGF receptors in these three cell types indicated that BB and AB and can be activated by PDGF AA, but only only NIHi3T3 cells contained (Y receptors (as few as when very high concentrations are used ( > l o 0 ng/ml) 500-1,000 receptors per cell could be detected) whereas (Matsui e t al., 1989). all three cells exhibited p receptors. Our results indicate that NIH/3T3 cells are capable of Taken together our data on the presence of the Type responding to low concentrations of PDGF AA (1-20 a and Type @ PDGF receptors and the biological and ng/ml) as a mitogen, whereas human dermal fibro- biochemical activities of the three PDGF isoforms on blasts and aortic smooth muscle cells do not. All three these cell types indicate somewhat different binding

242

GROTENDORST ET AL.

properties for the Type (Y or Type @ PDGF receptors than those previously reported. Also, our results that PDGF AB and BB are equally active on cells possessing only Type @ PDGF receptors suggest that there may be other mechanisms in addition to heterodimeric receptors (Type or-@ dimer) for transduction of biological signals from PDGF AB molecules (Seifert et al., 1989). There are several possible mechanisms to explain these data. First, the forms of the PDGF @-receptors expressed in certain connective tissue cells could be different from those which have been cloned and expressed in recombinant cell types. These differences

a receptor

p receptor

O L L O

%$$) 18Ok6

180kd-

A AA

0

AB

BB

1020

5 1 0 2 0 5

5 x )

20

200 116 97

66

B

AB

AA 0

5

2

0

BB M

2 0 5

5

-200 -116 97

Fig. 6. Western blot analysis of PDGF a and p receptors in NIHi3T3 cells, human skin fibroblasts, and bovine aortic smooth muscle cells. Extracts of cells were prepared, and electrophoresis and immunoblots were performed as described under Materials and Methods. PDGF a receptors were detected only in NIHi3T3 cells. PDGF p receptors were detected in all three types.

could give rise to different ligand binding properties of these receptor isoforms. Second, there could be nonligand binding subunits which associate with the receptor and alter its binding properties in the natural cell types and which are not present in the recombinant cells. Alternatively, a third yet to be identified PDGF receptor which binds PDGF AB andlor BB would account for these data.

~

-66

43

C

AA 0

40

80

AB 10

BB 20

10

20

-116 -

97

-

66

43

Fig. 5. PDGF receptor tyrosine autophosphorylation in whole NIH/ 3T3 cells and membrane preparations of NIHi3T3 cells. A Confluent monolayer cultures of NIHi3T3 cells were treated with the indicated amount of PDGF AA, AB, or BB for 15 minutes at 37°C. The PDGF receptors were then extracted and the amount of tyrosine phosphorylation was determined by immunoblots using a monoclconal antiphosphotyrosine antibody as described under Materials and Methods. All three forms of PDGF at high concentrations (20 ngiml) induced receptor tyrosine phosphorylation, whereas only PDGF AB and BB were active at low concentrations (5 and 10 ngiml). B,C: Immunoprecipitation of phosphorylated PDGF receptors with anti-phosphotyrosine (B) and anti-PDGF IgG (C). Membrane preparations were isolated from NIHi3T3 cells and used in the cell free receptor auto-phosphorylation assay as described under Materials and Methods. Membrane extracts were treated with the indicated amount of PDGF for 15 minutes a t 4°C and then incubated for an additional 15 minutes at 4°C in the presence of 5 FCi of gamma "P-ATP. The reaction was terminated by the addition of excess cold ATP and dilution into immunoprecipitation buffer. Labeled peptides were then immunoprecipitated with either anti-phosphotyrosine or anti-PDGF IgG and analyzed with 7.5%)acrylamide SDS gels and autoradiography. Gels were exposed for 4 days at -70°C with a n intensifying screen prior to the development of the X-OMAT R film. As seen in B, PDGF AA was much less effective in stimulating tyrosine auto-phosphorylation of PDGF receptors than either PDGF AB or BB. When the ligand receptor complexes were precipitated with anti-PDGF, only PDGF AB and BB co-precipitated with phosphorylated receptors, whereas PDGF AA did not (C).All of these experiments were repeated at least three times with similar findings.

RECOMBINANT PDGF AA, AB, BB AND CONNECTIVE TISSUE

Our finding that fibroblasts and smooth muscle cells lack sufficient numbers of PDGF receptors which are capable of binding PDGF AA may help to explain some previous reports which indicated that PDGF AA lacked mitogenic activity for these cells (Nister e t al., 1988; Kazlauskas et al., 1988). The lack of responsiveness of both human dermal fibroblast and bovine aortic smooth muscle cells to PDGF AA raises questions concerning the effectiveness of autocrine produced PDGF AA as a mitogen for these non-transformed connective tissue cells (Seifert et al., 1984; Paulsson et al., 1987; Soma and Grotendorst, 1989). The differences observed in the cellular distribution of the PDGF receptors and the responsiveness of the various cell types to the different PDGF dimers increases the specificity of the in vivo regulatory functions of the members of this peptide growth factor family.

ACKNOWLEDGMENTS The authors thank Dr. Stuart Aaronson for his generous gift of specific PDGF a and p receptor antisera, and Linda Gibeaut and Sean Snyder for assistance in the typing of this manuscript. This work was supported by grants from the NIH (GM37223) and the Florida High Technology Council. LITERATURE CITED Claesson-Welsh, L., Eriksson, A,, Moren, A,, Server-insson, L., Ek, B., Ostoman, A., Betsholtz, C., and Helkin, C.-H. (1988) cDNA cloning and expression of human platelet-derived growth factor (PDGF) receptor specific for B-chain-containing PDGF molecules. Mol. Cell. Biol., 8t3476-3486. Claesson-Welsh, L., Hammacher, A., Westermark, B., Heldin, C.-H., and Nister, M. (1989) Identification and structural analysis of the type 01 receptor for platelet-derived growth factor. J. Biol. Chem., 264t1742-1747. Collins, T., Pober, J.S., Gimbrone, M.A., Hammacher, A., Betsholtz, C., Westermark, B., and Heldin, C.-H. (1987) Cultured human endothelial cells express platelet derived growth factor A chain. Am. J. Pathol., 126:7-12. Coughlin, S.R., Escobedo, J.A. and Williams, L.T. (1989) Role of Phosphatidylinositol kinase in PDGF receptor signal transduction. Science 243:1191-1194. Escobedo, J.A., and Williams, L.T. (1988) A PDGF receptor domain essential for mitogenesis but not for many other responses to PDGF. Nature 335:85-87. Gronwald, R.G.K., Grant, F.J., Haldeman, B.A., Hart, C.E., OHara, P.J., Hagen, F.S., Ross, R., Bowen-Pope, D.F., and Murray, M.J. (1988) Cloning and expression of a cDNA coding for the human platelet-derived growth factor receptor: Evidence for more than one receptor class. Proc. Natl. Acad. Sci. U.S.A., 85t3435-3439. Grotendorst, G.R. (1984) Alteration of the chemotactic response of NIHi3T3 cells to PDGF by growth factors, transformation, and tumor promoters. Cell, 36279-285. Grotendorst, G.R. (1987) A rapid spectrophotometric assay for the quantitation of cell migration in the Boyden chamber chemotaxis assay. In: Methods of Enzymology Peptide Growth Factors. Barnes and Sirbasku, eds. Academic Press, New York, Vol. 147, pp. 144-152.

243

Grotendorst, G.R., Seppa, H.J., Kleinman, H.K., and Martin, G.R. (1981) Attachment of smooth muscle cells to collagen and their migration toward platelet-derived growth factor. Proc. Natl. Acad. Sci. U.S.A. 78:3669-3672. Hammacher, A., Hellman, U., Johnsson, A,, Ostman, A,, Gunnarsson, K., Westermark, B., Wasteson, A., and Heldin, C.-H. (1988) A major part of platelet-derived growth factor purified from human platelets is a heterodimer of one A and one B chain. J. Biol. Chem., 263t16493-16498. Heldin, C.-H., Backstrom, G., Ostman, A., Hamrnacher, A., Ronnstrand, L., Rubin, K., Nister, M., and Westermark, B. (1988) Binding of different dimeric forms of PDGF to human fibroblasts: evidence for two separate receptor types. EMBO J., 7:1387-1393. Heldin, C.-H., and Westermark, B. (1990) Signal transduction by receptors for platelet-derived growth factor. J . Cell Sci. 96t193-196. Johnsson, A,, Heldin, C.-H., Westermark, B., and Wasteson, A. (1982) Platelet-derived growth factor: identification of constituent polypeptide chains. Biochem. Biophys. Res. Commun., 104r6674. Kazlauskas, A., Bowen-Pope, D., Seifert, R., Hart, C.E., and Cooper, J.A. (1988) Different effects of homo- and heterodimers of plateletderived growth factor A and B chains on human and mouse fibroblasts. EMBO J., 7t3727-3735. Kazlauskas, A. and Cooper, J.A. (1989) Autophosphorylation of the PDGF receptor in the kinase insert region regulates interactions with cell proteins. Cell 58:1121-1133. Matsui, T., Heidaran, M., Miki, T., Popescu, N., La Rochelle, W., Kraus, M., Pierce, J., and Aaronson, S. (1989) Isolation of a novel receptor cDNA establishes the existence of two PDGF receptor genes. Science, 243:800-804. Molloy, C.J., Bottaro, D.P., Fleming, T.P., Marshall, M.S., Gibbs, J.B., and Aaronson, S.A. (1989) PDGF induction of tyrosine phosphorylation of GTPase activating protein. Nature 342t711-714. Morrison, D.K., Kaplan, D.R., Escobedo, J.A., Rapp, U.R., Roberts, T.M., and Williams, L.T. (1989) Direct Activation of the serinei threonine kinase activity of Raf-1 through tyrosine phosphorylation by the PDGF beta-receptor. Cell 58:649-657. Morrison, D.K., Kaplan, D.R., Rhee, S.G., and Williams, L.T. (1990) Platelet-derived growth factor iPDGF)-dependent association of phospholipase C-gamma with the PDGF receptor signaling complex. Mol. and Cell. Biol. 10:2359-2366. Nister, M., Hammacher, A., Mellstrom, K., Siegbahn, A,, Ronnstrand, L., Westermark, B., and Heldin, C.-H. (1988) A glioma-derived PDGF A chain homodimer has different functional activities from a PDGF AB heterodimer purified from human platelets. Cell, 52t791799. Paulsson, Y., Hammacher, A,, Heldin, C.-H., and Westermark, B. (1987) Possible positive autocrine feedback in the prereplicative phase of human fibroblasts. Nature, 328t715-717. Pencev, D., and Grotendorst, G.R. (1988) Human peripheral blood monocytes secrete a unique form of PDGF. Oncogene Res., 3r333342. Seifert, R.A., Schwartz, S.M., and Bowen-Pope, D.F. (1984) Developmentally regulated production of platelet-derived growth factor-like molecules. Nature, 311:669-671. Seifert, R.A., Hart, C.E., Phillips, P.E., Forstrom, J.W., Ross, R., Murray, M.J., and Bowen-Pope, D.F. (1989)Two different subunits associate to create isoform-specific platelet-derived growth factor receptors. J. Biol. Chem., 264t8771-8778. Soma, Y., and Grotendorst, G.R. (1989) TGF-p stimulates primary human skin fibrobast DNA synthesis via a n autocrine production of PDGF-related peptides. J . Cell. Physiol., 140:246-253. Yarden, Y., Escobedo, J.A., Kuang, W.-J., Yang-Feng, T.L., Daniel, T.O., Tremble, P.M., Chen, E.Y., Ando, M.E., Hardins, R.N., Francke, U., Fried, V.A., Ullrich, A,, and Williams, L.T. (1986) Structure of receptor for platelet-derived growth factor helps define a family of closely related growth factor receptors. Nature, 3 2 3 t 2 2 6 232.