Theilen Feline Sarcoma Virus or Harvey or Kirsten Murine Sarcoma ...

Vol. 48, No. 3

JOURNAL OF VIROLOGY, Dec. 1983, p. 752-764 0022-538X/83/120752-13$02.00/0 Copyright © 1983, American Society for Microbiology

Epidermal Growth Factor Receptor Metabolism and Protein Kinase Activity in Human A431 Cells Infected with SnyderTheilen Feline Sarcoma Virus or Harvey or Kirsten Murine Sarcoma Virus JONATHAN A. COOPER,'* EDWARD M. SCOLNICK,2t BRAD OZANNE,3 AND TONY HUNTER' Molecular Biology and Virology Laboratorv, The Salk Institute, San Diego, California 92138'; Laboratory of Tumor Virus Genetics, National Cancer Instituite, Bethesda, Malyland 202052; and Department of Microbiology, Univ,ersity of Texas Health Science Center, Dallas, Texas 752353

Received 31 May 1983/Accepted 19 August 1983

When human A431 cells, which carry high numbers of epidermal growth factor (EGF) receptors, are exposed to EGF, the total content of phosphotyrosine in cell protein is increased, the EGF receptor becomes phosphorylated at tyrosine, and new phosphotyrosine-containing 36,000- and 81,000-dalton proteins are detected. We examined the properties of A431 cells infected with Snyder-Theilen feline sarcoma virus, whose transforming protein has associated tyrosine protein kinase activity, and Harvey and Kirsten sarcoma viruses, whose transforming proteins do not. In all cases, the infected cells were more rounded and more capable of anchorage-independent growth than the uninfected cells. EGF receptors were assayed functionally by measuring EGF binding and structurally by metabolic labeling and immunoprecipitation. In no case did infection appear to alter the rate of EGF receptor synthesis, but infection reduced EGF receptor stability by about 50% for cloned Harvey sarcoma virus-infected cells and by 80% for cloned feline sarcoma virus-infected cells. The corresponding reductions in EGF binding were 70 and 90%, respectively. The proteins of feline sarcoma virus-infected A431 cells contained an increased amount of phosphotyrosine, and the 36,000- and 81,000dalton phosphoproteins were detected. The EGF receptor was not detectably phosphorylated at tyrosine, however, unless the cells were exposed to EGF. The Harvey and Kirsten sarcoma virus-infected cells did not exhibit elevated levels of phosphotyrosine either in the total cell proteins or in the EGF receptor, nor were the 36,000- and 81,000-dalton proteins detectable. However, these phosphoproteins were found in the infected cells after EGF treatment. Thus, all of the infected A431 cells exhibited reduced EGF binding and increased degradation of EGF receptors, yet their patterns of protein phosphorylation were distinct from those of EGF-treated A431 cells. There are many similarities between the effects of certain cell growth factors and tumor viruses on tissue culture cells. For example, resting monolayers of cells have an increased probability of entering S phase if they are exposed to epidermal growth factor (EGF) (46) or if they are infected with a temperature-sensitive Rous sarcoma virus (RSV) mutant and are shifted from the nonpermissive temperature to the permissive temperature (6, 42). Cell surface ruffling has been observed soon after either manipulation (1, 8, 14), and both EGF and transforming retroviruses cause increased rates of glucose uptake in suitable cells (4, 33). Such t Present address: Merck Sharp & Dohme Research Laboratories, West Point, PA 19486.

752

analogies have prompted the search for further parallels which might provide evidence for common mechanisms of action. Protein kinase activities can be immunoprecipitated together with the transforming proteins of many retroviruses (reviewed in reference 37) and purify with the EGF receptor (9, 15, 16). These protein kinases phosphorylate tyrosine in substrate proteins (37, 55). The activities of such tyrosine protein kinases can be detected in intact cells as an increase in the content of phosphotyrosine in cell proteins (34, 49). In the case of EGF, the use of cells bearing an unusually high number of EGF receptors, human A431 tumor cells (26, 30), has enabled the detection of an 8to 10-fold increase in phosphotyrosine content within 1 min of EGF addition (28, 34, 44).

VOL. 48, 1983

EGF RECEPTORS OF INFECTED A431 CELLS

Particular proteins which are phosphorylated more extensively at tyrosine after EGF treatment include the EGF receptor itself (34) and two other proteins, one with a molecular weight of 34,000 to 39,000 (36-kDa protein) (25, 34) and one with a molecular weight of 81,000 (81-kDa protein) (34). When A431 cells are infected with RSV, the same 36-kDa protein is constitutively phosphorylated, apparently at the identical tyrosine residue (20). This site is also phosphorylated in vitro when purified 36-kDa protein is incubated with purified RSV-transforming protein (25). However, neither the 81-kDa protein nor the EGF receptor is extensively phosphorylated at tyrosine in RSV-infected A431 cells, suggesting that RSV does not activate the EGF response mechanism of the cells (20). A further link between EGF and transforming retroviruses is the production by some virally transformed cells of EGF-competing factors (21, 23, 24, 41, 51). The secreted factors can themselves act as growth factors, but also have effects not normally attributed to EGF. They can reversibly induce the changes in cell morphology, hexose transport, and anchorage dependence for growth that are usually associated with transformation (21, 38, 41, 51). Hence, these factors and some other non-EGF-competing factors have been termed transforming growth factors (TGFs) (51). These observations suggest a mechanism for cell transformation in which TGF production is induced and the TGFs then stimulate the normal cell response to EGF. EGF-like TGFs further resemble EGF in that TGFs purified from human tumor A673 cells can induce tyrosine phosphorylation in intact A431 cells (44). However, it is not known whether TGFs stimulate tyrosine phosphorylation in the transformed cells in which they are produced. The activation of the EGF receptor tyrosine protein kinase activity could escape detection in TGF-producing rodent cells. The aforementioned stimulation of tyrosine phosphorylation by EGF itself is easier to detect in A431 cells than in most other cell types, probably because the latter have lower numbers of EGF receptors (34). Thus, the failure to detect increased phosphorylation of tyrosine in rodent cells transformed by Harvey sarcoma virus (HaSV) or Kirsten sarcoma virus (KiSV) (49). viruses which do produce TGFs (22, 24, 38, 41), could be due to the insensitivity of the assay. It is also not known whether the production of EGF-like TGFs by virally transformed cells completely accounts for the decreased level of EGF binding by the cells (7, 21, 23, 24, 29, 45, 51-53). We describe here experiments in which A431 cells infected with HaSV or KiSV were compared with A431 cells infected with the SnyderTheilen strain of feline sarcoma virus (ST-

753

FeSV), which induces both TGF production (45, 52) and tyrosine phosphorylation (2, 5, 45) in transformed mouse cells. By using a monoclonal antibody specific for human EGF receptors, we were able to follow the fate and phosphorylation state of the EGF receptor in these cells. The level of phosphotyrosine in the total cell proteins and the phosphorylation of the 36-kDa and 81kDa proteins were also examined. MATERIALS AND METHODS Cells and virus. NIH/3T3 cells nonproductively transformed by either KiSV or HaSV have been described previously (50), and NIH/3T3 cells nonproductively infected with ST-FeSV were available in the laboratory of E.S. (unpublished data). To produce virus infectious for human cells, each NIH/3T3 cell line was infected with wild mouse amphotropic virus 4070A (32, 43) at a multiplicity of infection of approximately 0.01. The amphotropic virus was grown on NIH/3T3 cells before use in these experiments. NIH/3T3 cells producing the pseudotyped transforming virus were grown to 90% confluence and fed with fresh medium, and then virus complexes were collected 18 h later. Each producer culture yielded virus complexes containing approximately 5 x 105 focusforming units per ml, as assayed on NIH/3T3 cells. Approximately 2 x 105 focus-forming units of either HaSV, KiSV, or ST-FeSV was added to 3 x 104 A431 cells which had been seeded into a 35-mm dish in a medium containing 10% fetal calf serum and a 1:1 mixture of Dulbecco modified Eagle medium (DME) and Ham F12 medium (DME:F12) supplemented with 4 ,ug of Polybrene (Aldrich Chemical Co.) per ml. After morphological transformation of the A431 cells was observed, clones of transformed A431 cells were derived by endpoint dilution in Microtest 11 plates (Falcon Plastics). The mass-infected cultures were also maintained by serial passage in DME:F12 containing 10% fetal calf serum or in DME containing 10% calf serum. Experiments were performed within five passages of infection, during which the microscopic appearance of the cultures was constant. Soft agar colony formation was assayed in DME:F12 containing 10% fetal calf serum and 0.35% Noble agar; 60-mm culture dishes were seeded with either 5 x 104 or 1 x 104 cells, and colonies were counted 10 to 12 days later. EGF binding. Dilutions of 251I-labeled EGF (New England Nuclear Corp.) and unlabeled EGF (a gift from S. Potter and T. Messmer) were added to duplicate 35-mm dish cultures in 0.5 ml of DME containing 0.1% bovine serum albumin. After approximately 1 h at 37°C, the dishes were washed three times with phosphate-buffered saline, and the preparations were solubilized in 1 M NaOH and counted in a gamma scintillation spectrometer. The values for nonspecific binding controls were subtracted (12). Data were analyzed by the method of Scatchard (47). Radioactive labeling. Dishes (35 mm) were seeded with 2 x 105 cells, incubated for 24 to 36 h at 370C, drained, and washed once either with phosphate-free DME containing 4% calf serum or with DME containing 5% of the regular methionine concentration and 4% calf serum. A 1.0-ml portion of the same medium was

754

COOPER ET AL.

added together with either 0.5 mCi of 32P (ICN Radiochemicals) or 0.1 mCi of [33S]methionine (Amersham Searle). After 12 to 20 h at 37°C, EGF was added as needed for the final 1 h of labeling (final concentration, 17 nM), and the cells were prepared for two-dimensional gel analysis (18, 19, 27) or phosphoamino acid analysis (49) or lysed in 0.2 ml of RIPA-EDTA buffer for immunoprecipitation (see below). For the pulse-chase experiment shown in Fig. 7, the cells were placed in methionine-free DME containing 0.1% bovine serum albumin for 15 min, drained, and labeled with 70 ,Ci of [35S]methionine in 0.3 ml of the same medium. After 15 or 30 min at 37°C, the cells were washed once and fed with 2 ml of DME containing 10% calf serum. Immunoprecipitation. Cell lysates made in RIPA buffer (48) containing 2 mM EDTA were clarified by centrifugation at 21,000 x g for 60 min at 4°C, and the supernatants were carefully collected. Samples were removed for acid precipitation (27) and immunoprecipitation as described previously (48). A different protocol was followed when we used the anti-EGF receptor monoclonal antibody, since this antibody does not bind to antigen at 4°C (56). An extract of 4 x 104 cells was mixed with 2 pLg of staphylococcal protein Apurified anti-EGF receptor immunoglobulin G (IgG) (56) and warmed to 25°C for 15 min. The tube was then cooled in ice for 15 min, and 1 ,u1 of goat antiserum to mouse IgG (Miles Laboratories, Inc.) was added. After an additional 30 min at 0°C, the immune complexes were collected on 500 p.g of Staphyvlococcus aureius (Pansorbin; Calbiochem-Behring Corp.) and washed as described previously (48). Samples were analyzed on sodium dodecyl sulfate-polyacrylamide gels (10% acrylamide, 0.13% bisacrylamide, except where stated otherwise) which were impregnated with 2,5-diphenyloxazole (for 35S) or exposed with an intensifying screen (for 32P) (39). The gel bands containint EGF receptor protein were solubilized for counting by the same procedure that was used for the acid precipitates (27), or the radioactivity was estimated from densitometer scans of the autoradiographs.

RESULTS Infection of A431 cells. In previous studies on the susceptibility of various mammalian cells to infection by murine type C retroviruses, the wild mouse amphotropic virus 4070A has been found to infect many human cells efficiently (32, 40, 43). Therefore, we infected A431 cells with either 4070A virus alone or with viral complexes containing 4070A and HaSV, KiSV, or ST-FeSV in the form of 4070A pseudotypes. A431 cell cultures producing 4070A virus were indistinguishable from uninfected cultures, but within 4 or 5 days morphological alterations in the A431 cultures infected with the transforming viruses were apparent. Most cells in the cultures became much more rounded in shape; as determined by an infectious center assay with NIH/3T3 indicator cells, more than 50% of the HaSV(4070A)-infected A431 cells produced HaSV virus. The efficiency of infection was very high with all three virus stocks. Below we refer

J. VIROL.

to the mass cultures infected with KiSV(4070A) and HaSV(4070A) as KiA431(M) and HaA431(M), respectively. Immunoprecipitation of HaA431 cell extracts showed that HaA431 cells contained HaSV transforming protein p21 (50) (data not shown). However, a biochemical analysis of HaA431(M) and KiA431(M) cultures showed few differences from uninfected A431 cell cultures (see below). That the transforming protein was actually functional was suggested by the observed changes in cell morphology and the ability to grow in soft agar. Three clones of HaA431 cells were derived from one of the HaA431(M) cell cultures. Each of the HaSV-infected clones was morphologically distinguishable from A431. The cells of clones 3 and 4 were much more refractile than A431 cells, with may stellate cells (Fig. 1). Clone 8 cells were also rounded; this clone formed peculiar domelike structures when the cells reached a high density (data not shown). Clones 4 and 8 produced HaSV and 4070A virus, whereas clone 3 was an HaSV-transformed nonproducer. We assessed the ability of A431, HaA431(M), and KiA431(M) cells and of three clones of HaA431 cells to grow in soft agar (Table 1). Photographs of representative fields in the soft agar cultures are shown in Fig. 2. Approximately 70% of the HaA431(M) or KiA431(M) cells seeded in soft agar grew to colonies containing at least 16 cells. About one-third of the macroscopic colonies became very large, containing over 64 cells per colony. Essentially every cell of the three HaA431 clones formed a colony in soft agar. Cells of clones 3 and 4 formed the largest colonies (Fig. 2). Under the growth conditions used and with an observation period of 10 to 12 days, A431 cells remained either as large single cells or as doublets in soft agar (Fig. 2A). Experiments were not performed with the mass cultures infected with ST-FeSV(4070A) (STA431 cells). Two clones of STA431 cells were obtained and studied in detail. Both clone 4 and clone 8 contained many rounded and loosely attached cells (Fig. 1); they produced virus. The biochemical properties of these cells (see below) provided strong evidence that they contained functional ST-FeSV transforming protein P85 (2), and phosphorylated P85 was detected in STA431 cells by immunoprecipitation (data not

shown). EGF binding to infected A431 cells. EGF binding to HaA431(M) and KiA431(M) cells was measured by adding different concentrations of 125I-labeled EGF to duplicate cell cultures. Cell density was important in these experiments, since the number of EGF receptors is increased and their affinity is reduced as A431 cells reach confluence (28). Subconfluent cultures were

VOL. 48, 1983


.

..

-.bs

755

t .N rvjW

-I"''A Al;. -. .

v, .4

Nt. .11 I

'W.-

, -

-4j i /= X> FIG. 1. Morphologies of infected A431 cell clones. Phase-contrast micrographs of subconfluent cultures (magnification, about x200). (A) A431. (B) HaA431 clone 3. (C) HaA431 clone 4. (D) STA431 clone 4. (E) STA431 clone 8.

used since this allowed maximal phosphorylation of tyrosine residues when 8 nM EGF was added (34). We found that the apparent numbers of EGF-binding sites on HaA431(M) and KiA431(M) cells were 25 to 35% less than the numbers on the control cells (Fig. 3A and Table 2). Binding of 125I-labeled EGF to two of the clones of HaA431 cells was also measured. In each case, the number of binding sites was reduced by about 70%, whereas the dissociation constant was relatively unaltered (Fig. 3B and Table 2). The lower reduction in EGF binding in the uncloned cultures may have reflected the presence of cells which did not contain a transforming viral genome. EGF binding to STA431 clone 4 and clone 8 cells was more dramatically reduced. Less than 10% of the control levels of

EGF-binding sites were detected on each of these clones (Fig. 3B and Table 2). Phosphoproteins of control and EGF-treated HaA431 and KiA431 cells. Cultures of A431, HaA431(M), and KiA431(M) cells that were 50% confluent were labeled for 12 h with 32p to approach isotopic equilibrium, harvested, and analyzed by two-dimensional gel electrophoresis. The gels were incubated in alkali to facilitate detection of proteins containing phosphotyrosine and phosphothreonine (13, 18). The phosphoproteins of A431 cells were not significantly altered by mass infection with either virus (Fig. 4). We did not identify p21 on our two-dimensional gels, although this protein was readily detected by immunoprecipitation (data not shown). When the cells were incubated with 17

756

COOPER ET AL.

J. VIROL.

TABLE 1. Colony formation by HaSV- and KiSVinfected A431 cells in soft agar" % Of colonies containing:

Cells tested

1-2

> 16

>64

Cells

Cells

Cells

A431 >90 0 0 A431producing 4070A >90 0 0 A431 producing KiSV(4070A) 20-30 40-50 10-20 A431 producing HaSV(4070A) 20-30 40-50 10-20 HaA431 clone 3 -10 60-70 20-30 HaA431 clone 4 -10 60-70 5-10 HaA431 clone 8 -10 60-70 20-30 " The cells were seeded into duplicate 60-mm dishes containing either 5 x 104 or 1 x 104 cells in 3 ml of soft agar (see text). Plates were incubated at 370C with 10% CO2 and fed with 2.5 to 3.0 ml of fresh medium containing 0.35% agar 5 days later. A total of 300 colonies of each cell type were counted on each dish. The percentages were calculated relative to the total number scored. Photographs of representative fields are shown in Fig. 2.

nM EGF for the final 1 h of labeling, phosphorylation of the 81-kDa and 36-kDa proteins was detected in both normal and infected cells (Fig. 4). Substantially the same results were obtained

with all three clones of HaA431 cells (data not shown). The total proteins of similarly labeled cultures of A431, HaA431 clone 3, and HaA431 clone 4 cells were partially acid hydrolyzed, and their phosphoamino acids were quantitated (Table 3). The phosphotyrosine contents were not significantly affected by infection. In each case the phosphotyrosine level was increased substantially by EGF. In this respect, the infected cultures differed from uninfected A431 cells bearing subnormal numbers of EGF receptors (10). In such cells, the tyrosine phosphorylation after EGF treatment is proportional to the EGF receptor number (10). Phosphoproteins of control and EGF-treated STA431 cells. Cultures of A431, STA431 clone 4, and STA431 clone 8 cells were labeled for 17 h with 32p. EGF was added to a final concentration of 17 nM for the final 1 h of labeling. An analysis of the labeled cultures by two-dimensional gel electrophoresis revealed the presence of the 36-kDa and 81-kDa phosphoproteins in untreated STA431 cells (Fig. 5). Although present, ST-FeSV transforming protein P85 was not detected by two-dimensional gel electrophoresis. Incubation of the cells with EGF did not noticeably alter the phosphorylation of the pro-

A

B

C

D *

9

p

FIG. 2. Colonies in soft agar. Photographs of representative fields were taken 10 days after seeding 5 x 104 cells in 60-mm dishes. (A) A431. (B) HaA431 clone 4. (C) HaA431 clone 8. (D) HaA431 clone 3.

VOL. 48, 1983


0*

.1

0.8

0

0.

757

0.4 a