Journal of Neuro-Oncology 8: 1-12, 1990. © 1990 Kluwer Academic Publishers. Printed in the Netherlands.

Laboratory Investigation

Platelet Derived Growth Factor (PDGF) autocrine components in human tumor cell lines

Griffith R. Harsh 1, Mark T. Keating 2, Jaime A. Escobedo 3, and Lewis T. Williams 2,3 1 The Brain Tumor Research Center of the Department of Neurological Surgery, the 2 Cardiovascular Research Institute, and the 3Howard Hughes Medical Institute, School of Medicine, University of California, San Francisco, California 94143, USA

Key words: platelet derived growth factor, malignant glioma, autocrine components, oncogenes, messenger RNA Abstract

Tumor cells may stimulate their own proliferation through an autocrine mechanism by simultaneously producing growth factors and growth factor receptors. We now report that numerous human tumor-derived cell lines simultaneously express the genes for platelet-derived growth factor (PDGF) A and B chains and the PDGF receptor (PDGF-R). Measurement of mRNA transcribed from these genes showed that among 16 malignant glioma cell lines tested, 15 expressed the PDGF A gene, 12 expressed the PDGF B gene, and 13 expressed the PDGF-R gene. Of three osteosarcoma lines, three expressed PDGF A, two expressed PDGF B, and three expressed PDGF-R. For eight malignant melanoma lines, seven expressed PDGF A, five expressed PDGF B, and three expressed PDGF-R genes. Thus, 13 of 16 malignant glioma, 3 of 3 osteosarcomas, and 3 of 8 malignant melanoma cell lines expressed the PDGF receptor gene and either or both PDGF genes. Five cell lines were tested for production of biologically active PDGF and PDGF receptor protein. Media conditioned by each of the five cell lines induced tyrosine phosphorylation of a protein identical in size to the PDGF receptor. These five cell lines also produced PDGF receptor protein as measured by Western blot analysis or metabolic labeling and immunoprecipitation using PDGF-R antibodies. The PDGF receptors of these cell lines were activated by human platelet PDGF or by recombinant AA or BB homodimers. Intracellular interaction of these receptors with the growth factor simultaneously produced may provide continuous stimulation to the proliferation of these cells.

Introduction

The abnormal pattern and rapid rate of the growth of tumor cells manifests escape of these cells from the mechanisms that normally regulate cell proliferation. The transformed state may reflect inappropriate expression of genes encoding growth factors, growth factor receptors, and proteins involved in signal transduction. Many transformed cell lines produce polypeptides capable of binding to and stimulating cell division in cells that express

growth factor receptors [1, 2]. These include tumor growth factor a (TGF a) that interacts with the epidermal growth factor (EGF) receptor, TGF that binds to any of a large class of receptors, bombesin, insulin-like growth factors, and plateletderived growth factor-like (PDGF) factors produced by tumor or other transformed cell lines [3-7]. PDGF-Iike factors, whether produced by platelets as a heterodimer of A and B chains or by tumor cell lines as molecules resembling A A or BB homodimers, bind to a cell surface receptor [8],

Fig. 1. PDGF A and B chain mRNA measured by RNAse protection assay. RNA from tumor cell lines, A 172 (lane 1), U-MG 343a (lane 2), SF 188 (lane 3), HOS (lane 4), U 20S (lane 5), WM 115 (lane 6) and Hs 294T (lane 7), was hybridized with 32plabeled RNA probes for PDGF B (510 base protected fragment), PDGF A (195 base protected fragment), and [32microglobulin (144 base protected fragment that served as internal control of quantity and quality of RNA) before RNAse digestion, gel electrophoresis and autoradiography (16 h exposure).

which activates the receptor's tyrosine kinase. The result is autophosphorylation of the receptor and activation of signaling pathways leading to enhanced cell proliferation [9]. Production of P D G F - I i k e mitogenic factors by a transformed cell raises the possibility that the uncontrolled proliferation of the transformed state is self-stimulated. Such P D G F - m e d i a t e d autocrine stimulation depends on expression of the P D G F receptor [10]. Activation of P D G F receptors by the P D G F - l i k e protein produced by simian sarcoma virus (SSV)-transformed cells is the one definitive example of autocrine stimulation of transformed cells [11]. The recent cloning and sequencing of the P D G F receptor gene in our laboratory has permit-

Fig. 2. PDGF-like activity in conditioned media quantitated by its ability to induce phosphorylation of the 185kDa protein. Serum free media conditioned for 12 h by tumor cell lines - A 172, SF 188, U-MG 343a, HOS and U 20S - were dialyzed, lyophilized, reconstituted at 5 × their original concentrations, and used to stimulate 3T3 fibroblasts for 2 h at 4° C. PDGF (10nM) served as a positive control; media conditioned by normal rat kidney cells (NRK) was the negative control. Lysates of target cellswere separated by PAGE, transferred to nitrocellulose and blotted with phosphotyrosine antibody. The activated PDGF receptor appears as a 185 kDa phosphoprotein. ted development of the genetic probes and receptor antibodies necessary to assess the expression of the P D G F receptor gene in tumor-derived cell lines [8, 10, 11]. W e now report expression, at both the m R N A and protein level, of the P D G F receptor in a variety of h u m a n t u m o r cell lines that also express the P D G F A and/or B chain genes.

Materials and methods Cell culture H u m a n tumor-derived cell lines were obtained f r o m M a r k R o s e n b l u m (SF 126, SF 188, SF 210, SF 215, SF 253, SF 268, SF 456, SF 468, SF 539), Stuart A a r o n s o n (A 172, A 382, A 431, A 1207, A 2781),

C.H. Heldin (U-MG 343a) and the American Type Tissue Culture Association (HOS, SaOS, U 20S, Hs 294T, HT 144, MG 138, MG 251, WM 115, WM 266, SKM 3, SKM5, SKM 24, SKM 28). All cell lines were well established in culture for 30 or more passages [12]. 3T3 murine fibroblasts, Chinese hamster ovary cells, and rat kidney normal cells (NRK) were obtained from the Tissue Culture Facility at the University of California, San Francisco. Cells were propagated in Dulbecco's Modified Eagle Medium (DMEM H21) supplemented with 10% fetal calf serum and antibiotics (penicillin [100 IU/ml] and streptomycin [100 IU/ml]).

Quantification of specific mRNA levels Expression of individual genes at the mRNA level was measured by the RNA solution hybridizationRNAse protection assay that has been described [13]. Twenty/xg of total cellular RNA, prepared by guanidium thiocyanate cell lysis and centrifugation over a cesium chloride cushion [14], was incubated in solution with 32p-labeled single stranded RNA probes produced by in vitro transcription of plasmids containing the following cDNA fragments: a 218 nucleotide Xho II - Sal I fragment of PDGF A (obtained from C.H. Heldin) [15] that yielded transcripts of 223 bases and protected fragments of 195 nucleotides; a 510 nucleotide Ava I - Ava I fragment of c-sis PSM-1 (obtained from F. Wong Staal) [16] that yielded transcripts of 569 bases and protected fragments of 510 nucleotides; a 230 nucleotide Sac I - Sac I fragment of the PDGF receptor [8] that yielded transcripts of 218 bases and a protected fragment of 165 nucleotides; and a 154 nucleotide Pst I - Hinc II fragment of ~2 microglobulin (obtained from W. Lee) [17] that yielded transcripts of 199 bases and protected fragments of 144 nucleotides. After hybridization and incubation with single stranded-specific RNAse, digestion with proteinase K, and phenol-chloroform extraction, the protected species were separated by polyacrylamide gel electrophoresis and analyzed by autoradiography [13].

Assay of conditioned media for PDGF-Iike activity Expression of PDGF genes at the protein level was assessed by measuring the ability of conditioned media to stimulate tyrosine phosphorylation of a 185 kDa protein in mouse fibroblasts. Cells were grown to confluence in T150 flasks, washed 4 times at 5-min intervals with 20 ml of serum free medium, and then incubated for 12 h in media supplemented with bovine serum albumin (BSA) (0.5 mg/ml), insulin (1/xg/ml), transferrin (5/~g/ml), and antibiotics. Collected medium was chilled, centrifuged at low speeds (4000 rpm for 5 min) to remove cellular debris, dialyzed in 3500 Da cutoff tubing against 1N acetic acid, lyophilized, and reconstituted in phosphate-buffered saline and BSA (1 mg/ml) at 1/10 (10x concentration) the original volume. Aliquots of concentrated, conditioned media were diluted with equal aliquots of serum-free media (final conditioned media concentration of 5x) and used to stimulate target 3T3 murine fibroblasts. Growth factor stimulation of phosphorylation of proteins of the size of the PDGF receptor was assayed by Western blotting of target cell extracts with polyclonal phosphotyrosine antibodies that recognize the activated PDGF receptor as a 185 kDa tyrosinephosphorylated protein. Target cells, grown to confluence without a change of media for 4 days, were chilled, washed twice with cold serum-free media and then incubated with test samples for 2 h at 4° C. Cells were then lysed at 4°C with Trisbuffered Triton X-100 (10 mM Tris pH 7.4, 50 mM NaC1, 50mM NaF, 30mM Na pyrophosphate, 100/xM Na orthovanadate, 5raM EDTA, 0.1%, Triton X-100, 1 mg/ml bovine serum albumin, and 1 mM phenylmethylsulfonylfluoride). Lysates were centrifuged (13,000 x g for 5 min) to precipitate a nuclear pellet; the supernatant was suspended in sample buffer to a final concentration of 10% glycerol, 2% SDS, 50 mM DTT, 35 mM Tris pH 6.8 and subjected to 7% polyacrylamide-SDS gel electrophoresis [18]. Proteins were transferred to nitrocellulose electrophoretically. Nitrocellulose filters were sequentially incubated with phosphotyrosine antisera, horseradish peroxidase-labeled goat antirabbit IgG, and horseradish peroxidase color developing reagent [11].

Fig. 3. PDGF receptor mRNA measured by RNAse protection assay. Left: RNA from tumor cell fines - A 172 (1), U-MG 343a (2), SF 188 (3), HOS (4), U 20S (5), WM 115 (6), and Hs 294T (7) - was hybridized with a 32p labeled RNA probe for PDGF receptor RNA (165 base protected fragment) before RNAse digestion, gel electrophoresis, and autoradiography (40 h exposure). Right: Parallel RNAse protection assay on RNA from A 431 (N, negative control) and U-MG 343A (2) cell lines for PDGF receptor (above, 96 h exposure) and [3z microglobulin (below, 12 h exposure).

Immunologic Detection of PDGF-Receptor Protein. PDGF receptor protein was identified immunologically by Western blotting of unlabeled cell extracts or by immunoprecipitation of 35S-lab e l e d cell extracts. For both procedures, cells were grown to confluence in six-well plates without a change of media for 4 days. Cells for Western blotting were processed as described above. Alternatively, cells were stimulated for 2 h at 4°C with partially purified PDGF (10 nM), recombinant PDGF A A homodimer (500 nM), or recombinant PDGF BB homodimer (5 nM) [8, 19] before lysis. Nitrocellulose blots of gels containing lysates of these cells were incubated with phosphotyrosine antibodies and developed as described above. Cells for immunoprecipitation underwent metabolic labeling with 35S-methionine (0.25 mCi/ml) in serum and methionine-free medium for 90 min. In pulse-chase experiments, labeled cells were either lysed immediately or washed twice and reincubated in fresh media containing 40 mM L-methionine and 1 mg/ml BSA for 45-180 min before lysis. Lysates were then immunoprecipitated with specific receptor antisera and incubated with protein A sepharose as described [11]. After multiple

washes to remove loosely bound protein, precipitated protein was separated from protein A sepharose by boiling and subjected to 7% polyacrylamide SDS electrophoresis and autoradiography.

Results

PDGF A and B chain mRNA

Cell lines derived from human glioblastomas, osteosarcomas, and malignant melanomas were chosen for these studies because lines from these types of tumors had previously been shown to express PDGF A or B genes [20-23]. The collection of cell lines studied included two groups. First, cell lines known to express either PDGF A or B were included to determine whether one or both PDGF chain genes were expressed and to confirm PDGF gene expression in cell cultures identical to those assayed for PDGF receptor gene expression. Second, cell lines not previously shown to express PDGF-like activity were analyzed for their expression of PDGF autocrine components. Expression of PDGF A and B chain genes by these cell lines

Fig. 4. PDGF receptor protein measured by Western blot. Lysates of tumor cells- A 172 (1), SF 188 (2), U-MG 343a (3), HOS (4), U 20S (5), WM 115 (6), and Hs 294T (7) - and 3T3 fibroblasts (C) - underwent gel electrophoresis, transfer to nitrocellulose, and incubation with PDGF receptor antiserum (A) or preimmune antiserum (B).

were assessed by measuring PDGF A and B mRNA by an RNAse protection assay. From each cell line, 20/.~g of total RNA, quantitated by spectrophotometry, was hybridized with labeled probes for PDGF A, PDGF B, and 62 microglobulin. 62 microglobulin mRNA is produced constitutively at similar levels by most cell types, and its level was used as an internal control for RNA loading [13]. Figure 1 shows an example of an autoradiograph of a polyacrylamide gel from an RNA protection assay for PDGF A and B chains. The 195 base long PDGF A transcript is present in all cell lines. The 510 base long PDGF B transcript is present in six of the seven lines. Of 27 cell lines tested, 25 had detectable A chain mRNA and 19 had detectable B chain mRNA (Table 1). Of the 16 glioblastoma lines tested, 15 expressed PDGF A chain and 12 expressed PDGF B chain. All three osteosarcoma lines expressed PDGF A chain; two out of three expressed PDGF B chain.

All but one of the eight malignant melanoma lines expressed A chain and five expressed B chain. Among those cell lines that expressed both PDGF A and B mRNA, the relative level of A and B mRNA both to each other and to 62 microglobulin mRNA varied with cell type. The most consistent quantitative difference noticed was a relatively high level of PDGF A chain mRNA in malignant melanoma cell lines.

PD GF-like protein Production of PDGF-like protein by five glioma and osteosarcoma cell lines shown to express PDGF A and/or B chain genes was studied by assessing the ability of media conditioned by these cell lines to stimulate tyrosine phosphorylation of the 185 kDa protein of 3T3 murine fibroblasts. Five-fold concentrates of conditioned media from

6

Fig. 5. Metaboliclabeling of PDGF receptor protein. Cells preincubated with 35S-rnethioninefor 15 min (pulse) were allowedto process incorporated 35Sin newly synthesized proteins for various additional intervals (chase) before lysis. Lysates were then subjected to immune precipitation by PDGF antisera before gel electrophoresis and autoradiography. (a) Irnnrnunoprecipitationsof extracts of 3T3 fibroblasts (C), A 172 (1), SF 188 (2), and U-MG 343a (3) cellsfollowinga 15 rain pulse (0) or a 15min pulse and a 90 min chase (90). (b) Immunoprecipitationsof extracts of HOS (above) and U 20S (below)cellsfollowinga 15rain pulse (0) and chase periods of 45, 90, and 180min. The PDGF precursor appears as a 160 kDa protein; the mature form is a 185kDa protein. each of these tumor cell lines induced tyrosine autophosphorylation of this protein (Fig. 2).

PDGF-receptor m R N A Expression of the P D G F receptor gene at the m R N A level was assessed with the R N A s e protection assay using 20 t~g of total cellular R N A and a labeled probe derived from the cytoplasmic portion of the P D G F receptor gene. Similarity in size of protected receptor and 132 microglobulin fragments precluded using both probes in the same hybridization sample. Separate hybridizations with the [~2microglobulin probe showed similar amounts of [~2 microglobulin m R N A for the different cell lines tested (data not shown). Figure 3 shows the radiograph of the polyacrylamide gel obtained with

the R N A s e protection assay for the P D G F receptor in the same seven lines used in Fig. 1. All seven lines have some receptor m R N A , but the amount of receptor m R N A in each line varies greatly. The low levels of m R N A in the U - M G 343a, W M 115, and Hs 294T lines were consistently present on two repeat assays. False positivity was excluded by the absence of signal in an overexposed radiograph of a lane containing R N A from A 431 cells that do not express the receptor gene (lane N, Fig. 3) [10]. Of all the 27 lines tested, 19 had detectable levels of P D G F receptor m R N A . It was present in 13 of the 15 glioblastoma lines, in all the osteosarcoma lines analyzed, and three out of eight malignant melanoma lines (Table 1). The level of expression was greater in glioblastoma and osteosarcoma lines than in the three malignant melanoma lines.

7 PDGF-receptor protein To determine if PDGF receptor protein was expressed in the cell lines expressing PDGF receptor mRNA, PDGF receptor antibodies were used in Western blot and immunopurification experiments. Photographs of Western blots of lysates of the same seven lines shown in Figs. 1 and 3 are shown in Fig. 4a. 3T3 fibroblasts (C) from which the PDGF receptor was originally purified and cloned were used as a positive control (lane 1). The 185 kDa PDGF receptor protein is present in large

amounts in A 172 (lane 2) and U 20S (lane 5) extracts, probably present in SF 188 (lane 2), UMG 343a (lane 3), and HOS (lane 4) extracts, and equivocally present in the extracts from the two melanoma lines (lanes 6 and 7). Figure 4b is a control blot of extracts from parallel cultures using pre-immune sera. No 185 kDa protein is evident. To confirm that PDGF-R protein was expressed, protein from the glioblastoma and osteosarcoma lines was further analyzed by the more sensitive technique of immunoprecipitation of metabolically-labeled proteins. These metabolic-labeling ex-

Table 1. m R N A of P D G F autocrine components in h u m a n t u m o r cell linesa PDGF A

PDGF B

PDGF-R

A/B + R

M alig nant gliomas (n = 16)

15

12

13

13

A 172

+

++

++++

+

A 382

+

++

+

+

A 1207 A 2781

+

++ +++

-

-

M G 138 M G 251

+ +

+ +

+ -

+ -

SF 126

++

+

++

+

SF 188 SF 210

+ ++

+

+++ ++

+ +

SF 215

+

-

+

+

SF 253 SF 268

+ +

+ -

++ +

+ +

SF 456

+

+

+++

+

SF468

+++

+++

++

+

SF 539

+++

-

+

+

U - M G 343a Osteo sarcomas (n = 3)

++ 3

++++ 2

+ 3

+ 3

HOS

+

++

++

+

SaOS U 20S

+ ++

++

+ ++

+ +

M alignant m e l a n o m a s (n = 8)

7

5

3

3

Hs 294T H T 144

+++ +++

+ ++

+ -

+ -

SKM 3 SKM 5

+ ++

+

+

+

SKM 24

+++

-

-

SKM 28

.

W M 115 W M 266

++++ +++

++ +

+ -

+ -

Total (n = 27)

25

19

19

19

.

.

.

aTotals for the n u m b e r of cell lines from a given tumor type expressing the P D G F A , B, or R genes and the n u m b e r of cell lines expressing either A or B chain gene and the R gene are presen t e d above the tabulation of the expression of P D G F A , B or R genes by the individual t u m o r cell lines. The m R N A level of each gene was m e a s u r e d by the R N A s e protection assay described in Materials and methods and expressed relative to the level present in the cell line with the highest expression of that gene ( W M 115 for A, U - M G 343a for B, and A 172 for R).

Fig. 6. Stimulation of t-umor PDGF receptor protein by PDGF. Cells from the tumor lines indicated or Chinese hamster ovary (CHO) cells were incubated with serum free media (v, vehicle), human platelet derived PDGF (AB, 10 nM), or yeast derived recombinant PDGF (AA, 500 nM, or BB, 5 nM) for 2 h at 4° C. Cell lysates underwent Western blot with phosphotyrosine antibody. The stimulated, autophosphorylated PDGF receptor appears as a 185 kDa phosphoprotein.

periments allow cells to incorporate large amounts of 35S into proteins with rapid rates of turnover. The protocol was designed to maximize labeling of the 160 kDa PDGF receptor precursor (during a 15 min incubation in 35Smethionine media followed by immediate cell lysis) and of the mature 185 kDa PDGF receptor (after the 15 rain labeling pulse and an additional 15 to 180-rain chase interval of incubation in 35S-free media rich in unlabeled methionine). In Fig. 5a, the glioblastoma cell lines are compared to 3T3 cells. In each line, after the 15 rain pulse, the PDGF receptor antibody recognizes a 160 kDa protein corresponding to the receptor precursor, but the 185 kDa mature receptor protein is not present [11]. After an additional 90 min of incubation, the slightly higher 185 kDa band corresponding to the mature receptor is apparent. In Fig. 5b, a more detailed pulse chase experiment on the osteosarcoma lines is shown. Here, the shift of the protein band recognized by the PDGF receptor antibody clearly represents the processing of the 160 kDa precursors into the mature 185 kDa form.

Thus, in the cell lines tested, biosynthesis and posttranslational processing of the PDGF receptor are similar to that found in untransformed control cells. The functional activity of these receptors was tested by assessing their ability to undergo tyrosine specific autophosphorylation when stimulated by PDGF. Cultures parallel to those used in Fig. 4 were stimulated for 2 h at 4°C with partially purified platelet PDGF (10 nM), recombinant PDGF AA (500nM), or recombinant PDGF BB homodimer (5nM). After PAGE and nitrocellulose transfer, their lysates were blotted with phosphotyrosine antibodies (Fig. 6). For each cell line, the amount of 185 kDa protein labeled with phosphotyrosine antibody was similar to that labeled with PDGF receptor antibody and proportional to the level of PDGF receptor mRNA measured by RNAse protection assay.

Discussion

Cell cycle progression and consequently the rate of cell proliferation is controlled by polypeptide growth factors that bind to specific cell receptors [24]. Uncontrolled cell proliferation is the hallmark of neoplasia. The rapid growth of transformed cells may be mediated by abnormally high concentrations or potencies of growth factors, growth factor receptors, and proteins involved in signal transduction pathways [25]. By producing factors for which they have receptors, some transformed cells may subvert normal regulatory mechanisms and stimulate their own growth [1]. The discovery that the gene encoding the B chain of PDGF is identical to the normal cellular counterpart (c-sis) of the transforming gene (v-sis) of SSV [26-28] provides evidence that such autocrine stimulation may be involved in tumorigenesis. SSV was the first acutely transforming retrovirus to be isolated from a primate. It induces astrocytomas when injected into the primate brain, induces sarcomas when injected into non-neural tissue, and transforms a variety of cell lines to the malignant phenotype [29,

301. There is a large body of evidence that sis-mediated autocrine stimulation underlies transformation by SSV. Cells transformed by SSV secrete v-sis protein, p28 v'is, that is highly homologous to PDGF B and resembles PDGF B antigenically, in binding studies, and by inducing both PDGF receptor autophosphorylation and mitogenesis in cells that express PDGF receptors [28, 31, 32]. The tumorigenicity of SSV-transformed cells correlates with the level of p28 v-s~'secreted [33]. Susceptibility of a cell to transformation by SSV depends on that cell's expression of PDGF receptors [34]. PDGF antibodies inhibit the acute transformation by SSV of human fibroblasts in cell culture [35]. Finally, suramin, which blocks the binding of PDGF with its receptor, reverses down-regulation of the PDGF receptor by p28 v~uand causes reversal of the phenotypic changes that arise from transformation by SSV [36, 37]. Sis- or PDGF B-mediated autocrine stimulation may also occur in tumors not necessarily induced by SSV; many cell lines, whether derived from

tumors (especially gliomas and fibrosarcomas) or transformed by carcinogens other than SSV, express c-sis mRNA and produce a protein antigenically similar to PDGF [20, 38]. In fact, the oncogenicity of the sis gene may lie not in the qualitative differences between v-sis and c-sis encoded proteins, but in the level of their expression. For instance, exogenous PDGF, although not itself acutely transforming, may induce phenotypic changes in cultured cells that resemble those of the transformed state [39]. C-sis DNA cloned from a human glioma, when linked to an appropriate upstream translation initiation site and retroviral LTR promoter, is transforming; cells transformed by these constructs express c-sis mRNA transcripts and secrete protein antigenically related to PDGF B [40]. Glial cells transformed in vitro by chemical carcinogens differ from premalignant glial cells by containing high levels of c-sis mRNA and producing large quantities of PDGF [41]. Although the structural homology and functional identity between PDGF B-B homodimer and p28 v-~ initially focused attention on the sis gene as the basis of autocrine stimulation, subsequent studies have shown that many transformed cell lines express the PDGF A chain gene. Correlation between intracellular A or B mRNA levels and PDGF receptor binding activity of conditioned media has suggested that in many cases A-A homodimers are most commonly secreted [21]. Secretion of one form of PDGF into the medium, however, does not necessarily imply that the particular form is the one responsible for autocrine stimulation. Currently, it is not known whether A - A , B-B, or A-B dimers of PDGF are involved in PDGF autocrine stimulation, or whether the interaction of PDGF and the PDGF receptor occurs intra- or extracellularly. Our findings confirm that many tumor-derived cell lines express PDGF A and B gene chains simultaneously. Gene expression at the mRNA level is accompanied by secretion into the medium of a PDGF-Iike activity capable of stimulating phosphorylation of a protein similar in size to the PDGF receptor. We also found that many of the same cell lines express the PDGF receptor gene at both the mRNA and protein levels. These PDGF receptors can be activated by exogenous

10 application of all three forms of PDGF-AA, BB, and platelet-derived - presumably A-B - dimers, which establishes the possibility of autocrine stimulation in these cells. Although the presence of autocrine components in these cell lines suggests that they were present in the original tumor specimens, it is possible that serial culture of these wellestablished cell lines has selected cells with this property. Experiments to confirm directly the expression of PDGF and PDGF receptor genes in the malignant cells of tumor specimens before culture are currently underway. Whether the autocrine interaction of growth factor and the PDGF receptor occurs intracellularly or extracellularly has been adressed in two other studies. Cell line U-20S clone 6 has been shown to have PDGF receptors and to produce a PDGF A chainlike factor that binds and activates the PDGF receptors of human foreskin fibroblasts [42]. Exogenous PDGF antibodies, however, do not alter the proliferation rate of U-20S clone 6 cells. This suggests either that PDGF-mediated autocrine stimulation is not crucial to the proliferation of these cells, that the antibody raised against rabbit platelet PDGF incompletely neutralizes the form of PDGF expressed, or that the interaction between PDGF-like factors and the PDGF receptor occurs in intracellular compartments inaccessible to PDGF antibodies. Because the PDGF B as well as the A chain gene is expressed in these cells, PDGF B, although not secreted, could be the PDGF-Iike growth factor activity involved in intracellular autocrine stimulation of these cells. This supposition is supported by the finding that autocrine activation of PDGF receptors by the B-B like homodimer of p28 vs~' in SSV transformed NRK cells occurs intracellularly [42]. Immunoprecipitation with phosphotyrosine antibodies showed that both the precursor and mature form are constitutively activated intracellularly and then shunted to degradative pathways different from those followed by the activated receptor in untransformed ceils. This suggests that attempts to interrupt- and to confirm the importance of - autocrine stimulation of tumor cells should be targeted intracellularly. Identification of tumor cell lines that have various combina-

tions of PDGF autocrine components, in conjunction with the development of PDGF antibodies that distinguish between A-A, B-B, and A-B forms with high specificity, should prove useful in dissecting the mechanisms of autocrine-driven tumor cell proliferation.

Acknowledgements Supported by the American Association of Neurological Surgeons Research Foundation, National Institutes of Health Grants 2-RO1-HL32898-03, 1-K08-NS01281-01, and Program Project Grant CA-13525, and the Aaron Silvera Cancer Research Fund. The SF brain tumor cell lines were provided by Mark Rosenblum, M.D.

References 1. Sporn MB, Roberts AB: Autocrine growth factors and cancer. Nature 331: 745-747, 1985 2. Anzano MA, Roberts AB, Smith JM, Sporn MD, Delarco JE: Sarcoma growth factor from conditioned medium of viraUytransformed cells is composed of both types ct and [3 transforming growth factors. Proc Natl Acad Sci USA 80: 626445268, 1983 3. Sporn MB, Roberts AB, Wakefield LM, Assonian RK: Transforming growth factor-[3: biological function and chemical structure. Science 233: 532-534, 1986 4. Reeve AE, Eccles MR, Wilkins ILl, Bell IG, Millon LJ: Expression of insulin-like growth factor-II transcripts in Wilm's tumour. Nature 317: 258-260, 1985 5. Bowen-Pope D, Vogel A, Ross R: Production of PDGFlike molecules and reduced expression of PDGF receptors accompany transformation by a wide spectrum of agents. Proc Natl Acad Sci USA 81: 2396-2400, 1984 6. Moody TW, Bertness V, Carney DN: Bombesinqike peptides and receptors in human tumor cell lines. Peptides 4: 683-686, 1983 7. Marquardt H, Todaro GT: Human transforming growth factor. J Biol Chem 257: 5220--5225, 1982 8. Escobedo JA, Navankasatussas S, Cousens LS, Coughlin SR, Bell GI, Williams LT: A common PDGF receptor is activated by homodimeric A and B forms of PDGF. Science 240: 1532-1534, 1988 9. Ross R, Raines EW, Bowen-Pope DF: The biology of PDGF. Cell 46: 155-169, 1986 10. Yarden Y, Escobedo JA, Kuang W-J, Yang-Feng TL, Daniel TO, Tremble PT, Chen EY, Ando ME, Harkins RW,

11

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

Francke U, Fried VA, Ullrich A, Williams LT: Structure of the receptor for platelet derived growth factor helps define a family of closely related growth factor receptors. Nature 323: 226-232, 1986 Keating MT, Williams LT: Processing of the platelet-derived growth factor receptor. J Biol Chem 262: 7932-7937, 1987 Rosenblum ML, Gerosa MA, Wilson CB, Barger GA, Pertuiset BF, DeTribolet N, Dougherty DA: Stem cell studies of human malignant brain tumors. Part I: Development of the stem cell assay and its potential. J Neurosurg 58: 170-176, 1983 Daniel TO, Gibbs VC, Millay DF, Garovoy MR, Williams LT: Thrombin stimulates c-sis gene expression in microvascular endothelial cells. J Biol Chem 261: 9579-9582, 1986 Chirgwin JM, Przybyla AE, McDonald RJ, Rutter WJ: Isolation of biologically active RNA from sources rich in ribonuclease. Biochemistry 18: 5294-5299, 1979 Starksen NF, Harsh GR, Gibbs VC, Williams LT: Regulated expression of the PDGF A chain gene in microvascular endothelial cells. J Biol Chem 262: 14381-14384, 1987 Josephs SF, Ratner L, Clarke MF, Westin EH, Reitz MS, Wong-Staal F: Transforming potential of human c-sis nucleotide sequences encoding PDGF. Science 225: 636-639, 1984 Suggs SV, Wallace B, Hirose T, Kawashima EA, Itakura K: Use of synthetic oligonucleotides as hybridization probes: isolation of clonal cDNA sequences for human 132 microglobulin. Proc Natl Acad Sci USA 78: 6613-6617, 1981 Laemmeli UK: Cleavage of structural proteins during the assembly of the head of bacteriophage. Nature 227: 680685, 1970 Williams LT, Tremple PT, Antonaides HN: PDGF binds to specific receptors on vascular smooth muscle cells and becomes non-dissociable. Proc Natl Acad Sci USA 79: 50675070, 1982 Eva A, Robbins KC, Anderson PR, Srinivasan A, Tronick SR, Reddy EP, Ellmore NW, Galen AT, Lautenberger JA, Papas TS, Westin EH, Wong-Staal F, Gallo RC, Aaronson SA: Cellular genes analogous to retroviral onc genes are transcribed in human tumour cells. Nature 295: 116--119, 1982 Betsholtz C, Johnsson A, Heldin CH, Westermark B, Lind P, Urden MS, Eddy R, Shows TB, Philpott K, Mellor AL, Knott RJ, Scott J: cDNA sequence and chromosomal localization of human PDGF A chain and its expression in tumour cell lines. Nature 320: 695-699, 1986 Westermark B, Johnsson A, Paulsson Y, Betshultz C, Heldin CH, Herlyn M, Rodeck U, Koprowski H: Human melanoma cell lines of primary and metastatic origin express the genes encoding the chains of PDGF and produce a PDGF-Iike growth factor. Proc Natl Acad Sci USA 83: 7197-7200, 1986

23. Harsh GR, Rosenblum ML, Williams LT: Oncogene related growth factors and growth factor receptors in human brain tumor cell lines. J Neurooncol, in press, 1988 24. Rozengurt E: Early signals in the mitogenic response. Science 234: 161-166, 1986 25. Weinberg RA: The action of oncogenes in the cytoplasm and nucleus. Science 230: 770-776, 1985 26. Chiu IM, Reddy EP, Grivol D, Robbins KC, Tronick SR, Aaronson SA: Nucleotide sequence analysis identifies the human c-sis protooncogene as a structural gene for PDGF. Cell 37: 123-129, 1984 27. Johnsson A, Heldin CH, Wasteson A, Westermark B, Deuel TF, Huang JS, Seeburg PH, Gray A, Ullrich A, Scrace G, Stroobant P, Waterfield MD: The c-sis gene encodes a precursor of the B chain of PDGF. Embryo J 3: 921-928, 1984 28. Doolittle FJ, Hunkapiller MW, Hood LF, Devare SG, Robbins KC, Aaronson SA, Antonaides HN: Simian sarcoma virus onc gene, v-sis, is derived from the gene (or genes) encoding PDGF. Science 221: 275-276, 1983 29. Deinhardt F: Biology of primate retroviruses. In: Klein G (ed). Viral Oncology, pp 357-398. New York: Raven Press, 1980 30. Leal F, Williams LT, Robbins KC, Aaronson SA: Evidence that the v-sis gene product transforms by interaction with the receptor for PDGF. Science 230: 327-329, 1985 31. Deuel TF, Huang JS, Stroobant P, Waterfield MD: Expression of a PDGF-Iike protein in simian sarcoma virus transformed cells. Science 221: 1348-1350, 1983 32. Pantazis P, Pelicci PG, Dalla-Favera R, Antoniades HN: Synthesis and secretion of proteins resembling plateletderived growth factor by human glioblastoma and fibrosarcoma cells in culture. Proc Natl Acad Sci USA 82: 24042408, 1985 33. Huang JS, Huang SS, Deuel TF: Transforming protein of simian sarcoma virus stimulates autocrine growth of SSVtransformed cells through PDGF cell surface receptors. Cell 39: 79-87, 1984 34. Leal F, Igarishi H, Gazit A, Williams LT, Notario V, Tronick SR, Robbins KC, Aaronson SA: Mechanism of transformation by an oncogene coding for a normal growth factor. Biochem Mol Epidem Cancer 1: 155-165, 1986 35. Johnsson A, Betsholtz C, Heldin CH, Westermark B : Antibodies against platelet-derived growth factor inhibit acute transformation by simian sarcoma virus. Nature 317: 438440, 1985 36. Betsholtz C, Johnsson A, Heldin CH, Westermark B: Efficient reversion of simian sarcoma virus transformation and inhibition of growth factor-induced mitogenesis by surarain. Proc Natl Acad Sci USA 83: 6440-6444, 1986 37. Garrett JS, Coughlin SR, Niman HL, Tremble PM, Giels GM, Williams LT: Blockade of autocrine stimulation in SSV-transformed cells reverses down regulation of PDGF receptors. Proc Natl Acad Sci USA 81: 7466-7470, 1984 38. Bowen-Pope D, Vogel A, Ross R: Production of platelet-

12 derived growth factor-like molecules and reduced expression of platelet-derived growth factor receptors accompany transformation by a wide spectrum of agents. Proc Natl Acad Sci USA 81: 2396--2400, 1984 39. Anzano MA, Roberts AB, Sporn MB: Anchorage-independent growth of primary rat embryo cells is induced by platelet-derived growth factor and inhibited by type B transforming growth factor. J Cell Physiol 126: 312-318, 1986 40. Gazit A, Igarishi H, Ciu IM, Srinivasan A, Yaniv A, Tronick SR, Robbins KC, Aaronson SA: Expression of the normal human sis/PDGF-2 coding sequence induces cellular transformation. Cell 39: 89--97,1984

41. Lens P, Altena B, Nusse R: Expression of c-sisand plateletderived growth factor in in vitro transformed glioma cells from rat brain tissue transplacentally treated with ethyl nitrosourea. Mol Cell Biol 6: 3537-3540, 1986 42. Keating MT, Williams LT: Autocrine activation of intracellular PDGF receptors in v-sis transformed cells. Science 239: 914-916, 1988

Address for offprints: G. Harsh, Department of Neurological Surgery, c/o The Editorial Office, 1360 9th Avenue, Suite 210, San Francisco, CA 94122, USA