E5 oncoprotein mutants activate phosphoinositide 3-kinase independently of platelet-derived growth factor receptor activation.

The E5 oncoprotein of bovine papillomavirus type 1 is a Golgi-resident, 44-amino acid polypeptide that can transform fibroblast cell lines by activating endogenous platelet-derived growth factor receptor beta (PDGF-R). However, the recent discovery of E5 mutants that exhibit strong transforming activity but minimal PDGF-R tyrosine phosphorylation indicates that E5 can potentially use additional signal transduction pathway(s) to transform cells. We now show that two classes of E5 mutants, despite poorly activating the PDGF-R, induce tyrosine phosphorylation and activation of phosphoinositide 3-kinase (PI 3-K) and that this activation is resistant to a selective inhibitor of PDGF-R kinase activity, tyrphostin AG1296. Consistent with this independence from PDGF-R signaling, the E5 mutants fail to induce significant cell proliferation in the absence of PDGF, unlike wild-type E5 or the sis oncoprotein. Despite differences in growth factor requirements, however, both wild-type E5 and mutant E5 cell lines form colonies in agarose. Interestingly, activation of PI 3-K occurs without concomitant activation of the ras-dependent mitogen-activated protein kinase pathway. The known ability of constitutively activated PI 3-K to induce anchorage-independent cell proliferation suggests a mechanism by which the mutant E5 proteins transform cells.

Stable NIH 3T3 cell lines expressing wt BPV-1 E5 and the E5 point mutants Q17G, Q17S, L24A, and L26A were generated as described previously (14,27) by geneticin G418 selection of cells co-transfected with E5 DNA and the neomycin resistance-conferring plasmid, LNCX, at a ratio of 9:1 (E5 DNA:LNCX) using Ca 3 (PO 4 ) 3 -DNA coprecipitation (28). All E5 constructs were tagged at their N terminus with the 6-amino acid AU1 epitope, which is recognized by the AU1 monoclonal antibody (29). The presence of this epitope does not affect the biological activity of E5 (27). Stable NIH 3T3 cell lines expressing the sis oncogene were similarly transfected and selected using a v-sis construct from J. Pierce (National Institutes of Health).
Proliferation Assays-To assess the proliferation of attached cells in defined media, 2 ϫ 10 4 cells were plated in 60-mm tissue culture dishes in 4 ml of DMEM containing 2.5% FBS. After 24 h, 1 dish of each cell line was harvested using trypsin/EDTA treatment and counted (Coulter Electronics) to determine the initial cell density. The remaining cultures were washed with D-PBS, and the medium was replaced with 4 ml of DMEM containing 0.5% FBS (basal medium). To assay PDGF-dependent growth, basal medium was supplemented with recombinant human PDGF BB (Life Technologies, Inc.) at a final concentration of 20 ng/ml. Insulin-dependent growth was assayed in basal medium supplemented with bovine insulin (Life Technologies, Inc.) at a final concentration of 5 g/ml. After 3 days in defined media, the cells were harvested and counted as above to determine the final cell density. Results are expressed as the percent increase in the number of cells, which is 100 ϫ (final density Ϫ initial density)/initial density. To analyze the contribution of PDGF-R signaling to proliferation, the selective PDGF-R kinase inhibitor, tyrphostin AG1296 (30), was added to cultures along with Me 2 SO (final 0.5%) as the cells were shifted into basal medium Ϯ growth factors. A 5 mM stock solution of AG1296 (Calbiochem or Alexis Biochemicals) was prepared in Me 2 SO and stored as aliquots at Ϫ70°C.
To assay anchorage-independent growth, 1 ml of 0.3% agarose containing 1.7 ϫ 10 3 cells was layered over 1 ml of 0.6% agarose in 35-mm dishes. A sterile 3% agarose stock solution was prepared in D-PBS and diluted to the above concentrations by mixing with DMEM (minus phenol red) containing 10% FBS and antibiotics (Life Technologies, Inc.). Cultures initially were overlaid with 0.5 ml of this medium and were given further additions as necessary to prevent desiccation for a period of 3-4 weeks.
For anti-PDGF-R and anti-phosphotyrosine immunoprecipitations, 150-mm tissue culture dishes of 80 -90% confluent cells were placed on ice and washed with 25 ml of D-PBS containing 1 mM Na 3 VO 4 . Cells were scraped into 0.9 ml of SDS lysis buffer (0.4% SDS, 100 mM NaCl, 2 mM EDTA, 50 mM HEPES-NaOH, pH 7.4) at room temperature and were immediately heated for 10 min at 95°C. The lysates were subjected to ultrasonic disruption (2 pulses of 10 s) using a Sonics and Materials, Inc. apparatus fitted with a microprobe. These were then mixed with 0.1 ml of 20% (w/v) Triton X-100 and were clarified by centrifugation for 5 min at 1000 ϫ g. The protein concentration of clarified lysates was determined using the Bio-Rad DC protein assay with bovine IgG as the standard. Aliquots of lysates containing 0.8 mg of protein were mixed with 15 l of protein A-agarose beads (Pierce; ImmunoPure Plus) and 4 g of affinity-purified anti-PDGF-R ␤ polyclonal antibody (Calbiochem) or 4 g of 5H1 anti-phosphotyrosine monoclonal antibody (a gift of A. Burkhardt and J. Bolen) and were rotated end over end for 90 min at 4°C. Subsequently, the beads were washed 3 times with 1 ml of modified radioimmune precipitation assay buffer (150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 1% deoxycholate, 0.1% SDS, 20 mM MOPS-NaOH, pH 7.0) (5 min rotating end-over-end per wash) and 3 times with 1 ml of D-PBS. Bound proteins were eluted by incubating the beads in 40 l of 1ϫ SDS gel sample buffer (3% SDS, 20 mM dithiothreitol, 1.5 mM EDTA, 11% sucrose, 0.008% bromphenol blue, 112 mM Tris-HCl, pH 8.8) for 20 min at 37°C with occasional vortex mixing. Samples were then alkylated by the addition of 80 mM iodoacetamide and incubation for 30 min at 37°C. Electrophoresis was performed in 1.5-mm polyacrylamide Tris-glycine mini-gels (Novex) using alkylated molecular weight markers (Sigma). Gels were electrophoretically transferred to polyvinylidene difluoride membranes and labeled with antibodies as outlined below.
Immunoblotting-For immunoblot analysis, cells from 100-mm tis-sue culture dishes at 80 -90% confluency were washed with 10 ml of D-PBS and scraped into 0.3 ml of 2ϫ SDS gel sample buffer. The lysates were heated for 5 min at 95°C and were subjected to ultrasonic disruption, alkylation, and SDS-polyacrylamide gel electrophoresis as above. SDS gels were transferred to Immobilon-P membranes (Millipore) in a Bio-Rad Trans-Blot apparatus in (192 mM glycine, 25 mM Tris-HCl, pH  8.3) at constant current (3200 mA ϫ h for 1.5 mm gels). Membranes were blocked for 30 min at room temperature in WB (0.5% (w/v) Triton X-100, 140 mM NaCl, 10 mM Na 3 PO 4 , pH 7.4) containing 2% bovine serum albumin (ICN 81003) and were labeled for 90 min with primary antibody diluted in WB ϩ 2% bovine serum albumin: anti-PDGF-R ␤ rabbit polyclonal antibody at 2 g/ml, anti-PI 3-K (85 kDa subunit) rabbit polyclonal antibody at 1 g/ml, anti-Erk2 monoclonal antibody at 2 g/ml (Upstate Biotechnology), or anti-active MAPK (Erk1/Erk2) rabbit polyclonal antibody at 25 ng/ml (Promega). Subsequently, membranes were washed 3 times with 20 ml of WB (10 min/wash), labeled for 90 min with alkaline phosphatase-conjugated goat anti-rabbit IgG or goat anti-mouse IgG antibodies (Tropix) in WB ϩ 2% bovine serum albumin (1:5000 dilution), and washed 3 times with WB. Labeled proteins were detected using the CDP-Star chemiluminescent substrate (Tropix) according to the manufacturer's instructions. Anti-phosphotyrosine immunoblots were blocked for 2 h in TBS (150 mM NaCl, 10 mM Tris-HCl, pH 8.0) containing 5% (w/v) bovine serum albumin (Sigma, essentially fatty acid-free) and 1% (w/v) ovalbumin (blocking buffer) and were labeled overnight at 4°C with an antiphosphotyrosine monoclonal antibody (Upstate Biotechnology clone 4G10) at a concentration of 1 g/ml in blocking buffer. The immunoblots were washed 3 times with 20 ml of TBS (10 min/wash), labeled with alkaline phosphatase-conjugated goat anti-mouse IgG (1:5000 dilution in blocking buffer) for 90 min at room temperature, and washed with TBS again. Tyrosine-phosphorylated polypeptides were detected by means of chemiluminescence.
PI 3-Kinase Assay-PI 3-K activity was measured essentially as described by Soltoff et al. (31). Briefly, cells from 150-mm tissue culture dishes at 80 -90% confluency were washed twice with 25 ml of buffer A (137 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 20 mM HEPES-NaOH, pH 7.5) and scraped into 1 ml of lysis buffer (buffer A containing 10% (v/v) glycerol, 1% (v/v) Nonidet P-40, 0.5 mM phenylmethylsulfonyl fluoride, 0.2 mM Na 3 VO 4 , 2 g/ml leupeptin, 2 g/ml pepstatin) at 4°C. After ultrasonic disruption and clarification, tyrosine-phosphorylated proteins were immunoprecipitated from aliquots of lysates containing 0.5 mg of protein as described above. Immunoprecipitates were washed 3 times with lysis buffer and 3 times with kinase assay buffer (100 mM NaCl, 1 mM EDTA, 0.2 mM Na 3 VO 4 , 10 mM HEPES-NaOH, pH 7.0). PI 3-K activity of the immunoprecipitates was assayed by adding 5 l of kinase assay buffer, 10 l of 25 mM MgCl 2 , 250 M ATP, 1 mCi/ml [␥-32 P]ATP, 50 mM HEPES-NaOH, pH 7.0; 20 l of 0.5 mg/ml phosphatidylinositol (Avanti Polar Lipids) in 0.1 mM EGTA, 0.015% (v/v) Nonidet P-40, 10 mM HEPES-NaOH, pH 7.0; and mixing continuously for 10 min at room temperature. Assays were terminated by the addition of 60 l of 2 N HCl and 160 l of chloroform-methanol (1:1) followed by vortex mixing and centrifugation for 30 s in an Eppendorf microcentrifuge. 50 l of the lipid-containing organic (lower) phase was spotted on Silica Gel 60 thin-layer chromatography plates (Merck) that had been prechromatographed in 1% (w/v) potassium oxalate and dried in a 70°C oven. Samples were chromatographed for 30 min using chloroform/methanol/H 2 O/NH 4 OH (60:47:11.3:2) as the solvent. Plates were air-dried, and phosphorylated PI was detected by autoradiography at Ϫ70°C with an intensifying screen. PI and PI phosphate standards (Calbiochem) were chromatographed in parallel with kinase assay samples and were visualized by exposing the chromatography plates to I 2 vapor.

Transforming E5 Mutants Are Defective for PDGF-R Activation and Induction of Growth Factor Independence
To study how BPV-1 E5 oncoprotein mutants transform fibroblast cell lines without activating the PDGF-R, we generated NIH 3T3 cell lines that stably express wt BPV-1 E5 and the E5 point mutants Q17S, L24A, L26A, and Q17G. Q17S E5, in which serine is substituted for glutamine at position 17, transforms NIH 3T3 cells with 39% the efficiency of wt E5 in a focus formation assay (27) but does not induce PDGF-R tyrosine phosphorylation, indicative of receptor activation (17). The L24A and L26A mutants, in which leucine residues at positions 24 and 26 of E5 are replaced with alanine, transform NIH 3T3 cells with 305% and 225% the efficiency of wt E5 but do not induce tyrosine phosphorylation of the PDGF-R (14). As a negative control, we used Q17G E5, in which glycine is substituted for glutamine at position 17. This mutant does not transform NIH 3T3 cells (0% focus formation) (27) and does not activate the PDGF-R (17). As shown in Fig. 1, wt and mutant E5 proteins were expressed at similar levels in the cell lines. The expression of L24A E5 was somewhat higher, and E5 was not detected in normal NIH 3T3 cells. In some experiments, we employed an NIH 3T3 cell line that stably expresses the sis oncogene as a control. Constitutive expression of the PDGF B-like sis gene product in these cells causes transformation due to persistent activation of endogenous PDGF-Rs (32,33). E5 was not detected in the sis-expressing cells (Fig. 1).
PDGF-R Tyrosine Phosphorylation-The ability of E5 mutants to activate the PDGF-R was evaluated in the cell lines used in this study by analyzing levels of tyrosine-phosphorylated PDGF-R in the absence of exogenous growth factors. Initially, tyrosine-phosphorylated proteins were immunoprecipitated from serum-starved cells and then were immunoblotted to detect tyrosine-phosphorylated PDGF-Rs. For this analysis, cells were lysed in the presence of SDS and immediately boiled to avoid potential post-lysis phosphorylation/dephosphorylation or proteolysis. Boiling in SDS also served to disrupt protein-protein interactions, so that the anti-PDGF-R immunoblots reflected tyrosine-phosphorylated PDGF-Rs rather than unphosphorylated PDGF-Rs associated with tyrosinephosphorylated protein(s). In addition, cells were incubated with Na 3 VO 4 for 10 min before lysis to inhibit tyrosine phosphatases and optimize the detection of tyrosine-phosphorylated PDGF-R on the immunoblots. As shown in Fig. 2A, serumstarved, control 3T3 cells exhibited basal levels of tyrosinephosphorylated mature PDGF-R (mPDGF-R), whereas cells stimulated with 10% FBS showed enhanced tyrosine phosphorylation. In contrast, all other cell lines expressing either transformation-competent E5 constructs (wt, L26A, Q17S, or L24A) or the transformation-defective Q17G E5 mutant showed levels of mPDGF-R tyrosine phosphorylation that were comparable with control 3T3 cells. As anticipated, although wt E5 did not increase tyrosine phosphorylation of the mPDGF-R, it did induce significant tyrosine phosphorylation of the immature PDGF-R (iPDGF-R) ( Fig. 2A), consistent with its primary localization in the Golgi apparatus (8). No other E5 mutant induced significant phosphorylation of the iPDGF-R, although L24A E5 did promote a very low level of phosphorylation ( Fig. 2A).
These results were further evaluated by performing the re-verse experiment under the same experimental conditions, i.e. immunoprecipitation with anti-PDGF-R antibodies and labeling the immunoblot with anti-phosphotyrosine antibodies. As shown in Fig. 2B, a high level of mPDGF-R tyrosine phosphorylation was observed in normal 3T3 cells after the addition of 10% FBS, and an equally elevated level of iPDGF-R tyrosine phosphorylation was seen in the wt E5 cell line. Unstimulated 3T3 cells and mutant E5 cell lines exhibited very low levels of PDGF-R tyrosine phosphorylation. Increased PDGF-R tyrosine phosphorylation in the wt E5 cell line was not due to enhanced expression of the PDGF-R; in fact, these cells contained somewhat lower levels of the receptor (Fig. 2C). Therefore, as judged by tyrosine phosphorylation, activation of the mPDGF-R was induced only by serum, which is compatible with exogenous PDGF binding to cell-surface PDGF-Rs. Only wt E5 was able to induce significant phosphorylation of the iPDGF-R. In agreement with recent studies (14,17), a number of transforming E5 mutants (L26A, Q17S, and L24A) induced only minimal phosphorylation of either the mPDGF-R or iPDGF-R. Proliferation Assays-To employ criteria other than phosphorylation to demonstrate that the L26A, Q17S, and L24A transforming E5 mutants do not significantly activate the PDGF-R, cell lines expressing wt and mutant E5 proteins were tested for their ability to proliferate in 20-fold reduced levels of growth factors. As shown in Fig. 3A, normal NIH 3T3 cells increased in number by only 60% over a period of 3 days in basal medium (DMEM ϩ 0.5% FBS), and cells that expressed the nontransforming Q17G E5 mutant increased by only 160%. When basal medium was supplemented with recombinant PDGF, the number of 3T3 cells and Q17G E5-expressing cells increased by 340% and 440%, respectively. Moreover, the stimulation of cell proliferation by PDGF was sensitive to tyrphostin AG1296, a selective inhibitor of PDGF-R tyrosine kinase activity (30). 20 M tyrphostin AG1296 reduced the proliferation of 3T3 cells and Q17G E5-expressing cells to 120% and 140%, respectively, which were similar to levels of proliferation A, serum-starved NIH 3T3 cells and cell lines expressing E5 constructs were treated with 1 mM Na 3 VO 4 (or 1 mM Na 3 VO 4 ϩ 10% FBS where indicated) for 10 min at 37°C before lysis in the presence of SDS. Tyrosine-phosphorylated (pY) proteins were immunoprecipitated (IP) from equivalent amounts of cell protein, fractionated on 8% SDS gels, transferred to membranes, and reacted with anti-PDGF-R antibodies to detect tyrosine-phosphorylated mature (m) and immature (i) forms of the PDGF-R. B, mature and immature forms of the PDGF-R were immunoprecipitated from equivalent amounts of cell protein and analyzed on immunoblots labeled with anti-phosphotyrosine antibodies. C, anti-PDGF-R immunoblot showing levels of PDGF-R expression in the cell lines. Lanes contain equal amounts of protein.

E5 Oncoprotein Activates PI 3-Kinase
in the absence of PDGF (Fig. 3A). The selectivity of tyrphostin AG1296 for the PDGF-R was demonstrated by its relatively minor effect on insulin-stimulated proliferation in these cell lines. 3T3 cells increased in number by 300%, and Q17G E5expressing cells increased by 350% in basal medium supplemented with insulin. 20 M tyrphostin AG1296 inhibited this growth by only 33 and 26%, respectively (Fig. 3A).
In contrast to the control 3T3 and Q17G E5-expressing cells, wt E5 and sis cell lines grew well in basal medium; cells expressing wt E5 increased by 340%, and cells expressing sis increased by 370% (Fig. 3A). These increases were very similar to those of control 3T3 cells and Q17G E5-expressing cells in the presence of PDGF. Furthermore, the proliferation induced by wt E5 and sis was enhanced only slightly by PDGF (430% increase in cell number for wt E5 and 380% increase for sis) and was sensitive to tyrphostin AG1296 (120% and 110% increase in cell number in the presence of 20 M inhibitor) (Fig. 3A). The fact that tyrphostin AG1296 reduced the proliferation of wt E5-and sis-expressing cells to levels characteristic of control cells in basal medium (i.e. 3T3 and Q17G E5 cells) confirmed that wt E5 and sis enable proliferation in basal medium due to their activation of the PDGF-R.
The proliferation in basal medium of cell lines expressing the transforming L26A, Q17S, and L24A E5 mutants was strikingly similar to that of control cells that do not activate the PDGF-R (Fig. 3A). Cells expressing L26A E5 increased in number by 40%, which was less than the proliferation of normal 3T3 cells. Q17S E5-and L24A E5-expressing cells increased by 140%, which was less than the proliferation of cells that expressed the nontransforming Q17G E5 mutant. The slowed proliferation of these cells in basal medium, however, was not the consequence of a general defect in cell growth, since the L26A E5 and L24A E5 cell lines multiplied at the same rate as normal 3T3 cells, Q17G E5 expressers, and sis-expressing cells in medium containing 10% FBS (Fig. 3B). Cells expressing Q17S E5 multiplied at even a slightly higher rate (Fig. 3B). The results of these proliferation assays are therefore in agreement with the analysis of PDGF-R tyrosine phosphorylation and indicate that among the E5 constructs employed in this study only wt E5 significantly activates the PDGF-R.
The identity of mitogenic signal(s) responsible for the partially enhanced proliferation of Q17G E5-expressing cells relative to normal NIH 3T3 cells in basal medium (additional 100% increase in cell number) is not known. Regardless, this signaling does not induce PDGF-R tyrosine phosphorylation (Fig. 2, A and B) and does not lead to transformation, since the Q17G E5 mutant is nontransforming in focus formation assays.

Transforming E5 Mutants Confer Anchorage Independence
Although the L26A, Q17S, and L24A E5 mutants have previously been shown to induce foci in NIH 3T3 cells (14,27), they do not significantly activate the PDGF-R or support growth factor-independent proliferation. To determine whether these mutants can induce additional characteristics of the transformed phenotype, we evaluated their ability to promote anchorage-independent growth (Fig. 4). In soft agar, cells expressing the nontransforming Q17G E5 mutant generated relatively few small colonies, whereas cells expressing wt E5 formed many colonies, both small and large in size. The L26A and L24A E5 cell lines also formed many small and large colonies. The transforming Q17S E5 mutant triggered an increase predominantly in the number of small colonies compared with the Q17G E5 control (Fig. 4). In summary, transforming E5 mutants that are defective for PDGF-R activation are unable to induce growth factor-independent proliferation but can induce anchorage-independent proliferation.

E5 Does Not Constitutively Activate the Erk/MAPK Signal Transduction Pathway
Numerous growth and differentiation signals initiate a protein phosphorylation cascade in which activated Ras triggers the sequential activation of Raf, MAPK kinase (MEK), and the 44 kDa and 42 kDa MAPKs, Erk1, and Erk2 (34 -36). Since constitutively active forms of Ras, Raf, and MEK are oncogenic (34,35,(37)(38)(39), we asked whether transforming E5 mutants might directly or indirectly activate component(s) of the Erk/ MAPK signal transduction pathway, leading to the constitutive activation of Erk1 and Erk2.
The activation of Erk1 and Erk2 involves their phosphorylation (by MEK) on proximal N-terminal threonine and tyrosine residues (40). Levels of active Erk1/Erk2 can be quantified on immunoblots using antibodies raised against a doubly phosphorylated peptide corresponding to the Erk phosphorylation site (41). As shown in Fig. 5A, active Erk1/Erk2 were not detectable on immunoblots of serum-starved NIH 3T3 cells but were readily detected 10 min after 10% FBS was added to the cells (compare the first and second lanes). There was no evidence of Erk activation in serum-starved cells expressing wt E5 (third lane), demonstrating that wt E5 does not constitutively activate these kinases to levels associated with acute seruminduced signaling. Active Erk1/Erk2 also were not detected in exponentially growing 3T3 cells in the continuous presence of 10% FBS (fourth lane). Immunoblots labeled with an antibody that recognizes Erk2 irrespective of its phosphorylation state showed that these same samples contained equal amounts of Erk2 protein (lower panel of Fig. 5A).
To detect lower levels of Erk activation in exponentially growing and serum-starved cells, it was necessary to increase the amount of phosphorylated Erk1 and Erk2 on the immunoblots. Treatment of serum-starved 3T3 cells with Na 3 VO 4 (to inhibit tyrosine dephosphorylation) and okadaic acid (to inhibit serine/threonine dephosphorylation) for 10 min before lysis made possible the detection of active Erk1/Erk2 ( Fig. 5A; compare the second and sixth lanes). As expected, cells grown in the continuous presence of 10% FBS and treated with these same inhibitors for 10 min showed even higher levels of Erk activation (fifth lane). The ability to detect active Erk1/Erk2 in serum-starved 3T3 cells now permitted an analysis of the effects of the various E5 mutants on Erk phosphorylation (upper panel of Fig. 5B). Treatment of serum-starved cells expressing wt or mutant E5 constructs with Na 3 VO 4 and okadaic acid demonstrated that Erk activation was identical in all cell lines tested. Thus, none of the transforming E5 constructs (including wt E5) constitutively activate the Erk/MAPK signaling pathway as determined by the phosphorylation state of Erk1 and Erk2.

Transforming E5 Mutants Activate PI 3-K
Heterodimeric PI 3-K is a critical component of signal transduction pathways that are involved in the mitogenic response to various growth factors and in oncogenic transformation (18,19). In rodent fibroblasts, expression of constitutively active PI 3-K in the presence of serum is sufficient to induce characteristics of cellular transformation, including anchorage-independent growth (25). Given our finding that transforming E5 mutants do not activate Erk/MAPK signaling, studies were undertaken to ascertain whether or not these oncoproteins activate PI 3-K signaling.
PI 3-K Tyrosine Phosphorylation-The activation of heterodimeric PI 3-K often involves phosphorylation of the 85-kDa regulatory subunit (p85) on tyrosine (23,24,42,43). Therefore, we measured levels of tyrosine-phosphorylated p85 in serumstarved 3T3 cell lines expressing wt and mutant E5 proteins to screen for possible effects of E5 on PI 3-K activity. Tyrosinephosphorylated proteins were immunoprecipitated from cell lysates, and the immunoprecipitates were probed for p85 on immunoblots. As with our analysis of PDGF-R phosphorylation, lysates were boiled in SDS to disrupt protein-protein interactions before immunoprecipitation so that anti-p85 immunoblots would not contain unphosphorylated p85 that was associated with another tyrosine-phosphorylated protein(s). This technique proved to be quantitative within a defined range of protein concentrations, since the amount of tyrosinephosphorylated p85 that was detected increased linearly with increasing quantities of lysate (Fig. 6A).

E5 Oncoprotein Activates PI 3-Kinase
transforming Q17G E5 mutant (third lane) did not increase p85 tyrosine phosphorylation in serum-starved cells (12% of the FBS-stimulated level); however, wt E5 (fourth lane) induced tyrosine phosphorylation of p85 to nearly the same level as FBS (84% of the FBS-stimulated level). Most importantly, we found that transforming E5 mutants that were defective for PDGF-R activation nevertheless induced tyrosine phosphorylation of p85 in the absence of growth factors (sixth to eighth lanes). Serum-starved cell lines expressing L26A E5, Q17S E5, or L24A E5 contained tyrosine-phosphorylated p85 at 51, 22, or 130% of the FBS-stimulated level, respectively. These values represent a 2-13-fold increase in p85 tyrosine phosphorylation relative to control 3T3 cells and Q17G E5-expressing cells. The total amount of p85 present in each of the cell lines varied by at most 16% (Fig. 6C) and therefore could not account for the much greater variation in levels of tyrosine phosphorylated p85.
To further clarify the role of PDGF-R activation in E5-induced PI 3-K activation, serum-starved cell lines expressing wt E5 or L26A E5 were treated with 0.3-3 M tyrphostin AG1296 for 60 min before lysis and analysis of tyrosine phosphorylated p85. Tyrphostin AG1296 previously has been shown to completely inhibit PDGF-R kinase activity in intact Swiss 3T3 cells and in isolated cell membranes at these concentrations (30). As indicated in Fig. 6D, 3 M tyrphostin AG1296 decreased p85 tyrosine phosphorylation by 70% in wt E5-expressing cells, suggesting that a component of PI 3-K activation in these cells is the consequence of PDGF-R activation. However, 3 M tyrphostin AG1296 had no effect on the tyrosine phosphorylation of p85 in cells expressing L26A E5. Thus, E5 mutants that cannot significantly activate the PDGF-R, such as L26A E5, apparently activate PI 3-K by a separate, tyrphostin-insensitive mechanism. It is also likely that the residual (30%) tyrphostin-resistant activity of wt E5 indicates utilization of the same alternative pathway to PI 3-K activation.
PI 3-K Activity-In vitro lipid kinase assays were performed to determine whether E5-induced tyrosine phosphorylation of the PI 3-K p85 regulatory subunit indeed reflected increased PI 3-K activity. Tyrosine-phosphorylated proteins were immunoprecipitated from serum-starved cells and tested for their ability to phosphorylate exogenous PI in the presence of [␥-32 P]ATP. As expected, essentially no PI kinase activity was detected in control NIH 3T3 cells or in cells expressing the nontransforming Q17G E5 mutant (Fig. 7A, second and third  lanes). When PDGF-R-signaling pathways were activated by adding FBS to the medium (first lane) or by constitutive expression of wt E5 (fourth lane) or the sis oncogene (fifth lane), 32 P-labeled PI phosphate was increased. However, since Ras activates PI 3-K independently of p85 phosphorylation (22), it is not surprising that the addition of serum, which rapidly activates Ras, elevated PI 3-K activity to a greater extent than p85 tyrosine phosphorylation. An elevated level of PI 3-K activity also was present in cells expressing the L26A, Q17S, and L24A E5 mutants that do not activate the PDGF-R (sixth through eighth lanes). These results show that a number of transforming E5 mutants induce phosphorylation and activation of heterodimeric PI 3-K by means of a pathway that does not require activation of the PDGF-R.
The independence of PI 3-K activation from PDGF-R signaling in L26A E5-expressing cells was further evidenced by its insensitivity to tyrphostin AG1296. PI 3-K activity in these cells was not affected by treatment with 0.3-3 M tyrphostin AG1296 for 60 min before lysis, whereas identical tyrphostin AG1296 treatment of wt E5-expressing cells partially lowered levels of PI 3-K activity (Fig. 7B). These results are in agreement with the effects of tyrphostin AG1296 on p85 tyrosine phosphorylation induced by wt and L26A E5 (Fig. 6D) and support the view that wt E5 utilizes both PDGF-R-dependent and PDGF-R-independent mechanisms to activate PI 3-K.
FIG. 6. Transforming E5 mutants induce tyrosine phosphorylation of the PI 3-K p85 regulatory subunit. A, serum-starved wt E5-expressing cells were treated with 1 mM Na 3 VO 4 for 10 min at 37°C before lysis in the presence of SDS. Tyrosine-phosphorylated proteins were immunoprecipitated (IP) from 0.2-1.6 mg of cell protein, fractionated on a 10% SDS gel, transferred to polyvinylidene difluoride, and labeled with an antibody recognizing the p85 regulatory subunit of heterodimeric PI 3-K. The ϩ control lane contains a p85 positive immunoblotting control (Upstate Biotechnology). A molecular mass marker (in kDa) is shown on the left. The relative amount of tyrosinephosphorylated p85 in each lane, as determined by laser densitometry of the film, is plotted versus the corresponding quantity of protein used for immunoprecipitation. B, serum-starved NIH 3T3 cells and cell lines expressing E5 constructs or the sis oncogene were treated with 1 mM Na 3 VO 4 (or 1 mM Na 3 VO 4 ϩ 10% FBS, where indicated) for 10 min at 37°C before lysis and analysis of tyrosine-phosphorylated PI 3-K p85 subunit as above.

DISCUSSION
Two new classes of E5 oncoprotein mutants have been identified that are hypertransforming but show only minimal PDGF-R activation (14,17). To investigate the mechanism of cell transformation by these mutants, we sought to identify other mitogenic signal transduction pathways that are constitutively activated in cell lines.
Transforming E5 Mutants Defective for PDGF-R Activation-wt BPV-1 E5 induces trans-phosphorylation and activation of the PDGF-R by effectively cross-linking endogenous PDGF-R monomers via transmembrane interactions (10, 14 -16). Nevertheless, E5 mutants that induce little or no receptor phosphorylation because they are defective for PDGF-R binding (Q17S, L21A, and L24A) or homodimerization (L7A, L25A, and L26A) can efficiently transform fibroblast cell lines (14,17). In the current study, we show that members of both classes of transforming E5 mutants (Q17S, L24A, and L26A) not only fail to efficiently stimulate PDGF-R phosphorylation in NIH 3T3 cells (Fig. 2) but also fail to functionally activate the PDGF-R in these same cells. Thus, the proliferation of cell lines that express Q17S E5, L24A E5, or L26A E5 requires PDGF and/or other growth factors, whereas cells that express wt E5 or the sis oncogene are relatively growth factor-independent (Fig. 3).
In contrast to our results with NIH 3T3 cells, Klein et al. (44) found that Q17S E5 increases PDGF-R phosphorylation 7-fold in a different murine cell line, C127. This discrepancy may derive from the use of different cell lines, since Q17S E5 has a 7.5-fold higher transformation efficiency in C127 cells than in NIH 3T3 cells (27). A similar biological disparity exists in murine hematopoietic cells. Although another Q17S E5 construct appears to activate the PDGF-R and induce interleukin-3 (IL-3)-independent proliferation when co-expressed with exogenous PDGF-Rs in Ba/F3 murine hematopoietic cells (44), our Q17S E5 mutant fails to induce IL-3-independent prolifer-ation in 32D murine hematopoietic cells when co-expressed with the PDGF-R (17). It remains to be determined whether these opposite activities derive from different levels of E5 expression, from epitope-tagging of the protein, or from the use of different cell lines. Regardless of the results with Q17S E5, however, we have employed two additional E5 mutants that only minimally activate the PDGF-R yet exhibit a 2-3-fold increase in transforming activity relative to wt E5.
E5 Activates Heterodimeric PI 3-K Independently of PDGF-R Activation-We have shown that PI kinase activity is present in anti-phosphotyrosine immunoprecipitates of serum-starved cells that express wt E5 or the Q17S-, L24A-, or L26A-transforming E5 mutants. In contrast, PI kinase activity is barely detectable in similar immunoprecipitates of serum-starved normal NIH 3T3 cells and control cells expressing the nontransforming Q17G E5 mutant. Therefore, three different transforming E5 mutants that are defective for PDGF-R activation are still able to activate a PI-kinase(s). PI-kinase activation appears to be constitutive since it occurs in serumstarved cells.
There is compelling evidence that the activated PI-kinase detected in our study is heterodimeric (p85/p110) PI 3-K, a key component of mitogenic signaling pathways (18,19,22,24). Heterodimeric PI 3-K can phosphorylate PI (Fig. 7) as well as PI 4-phosphate and PI 4,5-bisphosphate and is present in antiphosphotyrosine immunoprecipitates (Fig. 6), unlike PI 4-kinases and other classes of PI 3-Ks (18 -20, 24, 45). Moreover, the activation of heterodimeric PI 3-K by the PDGF-R involves tyrosine phosphorylation of the p85 regulatory subunit (23,24,42,43). We show that constitutive activation of the PDGF-R in cell lines that express wt E5 or the sis oncogene leads to increased PI kinase activity in anti-phosphotyrosine immunoprecipitates and to increased tyrosine phosphorylation of the PI 3-K p85 subunit on immunoblots. Activation of receptor tyrosine kinases (including the PDGF-R) by 10% FBS also elicits a concomitant increase in immunoprecipitable PI kinase activity and p85 tyrosine phosphorylation. Finally, increased tyrosine phosphorylation of the PI 3-K p85 subunit is correlated with the increased PI kinase activity induced by Q17S E5, L24A E5, and L26A E5, which do not significantly activate the PDGF-R.
Because heterodimeric PI 3-K can be directly activated by the PDGF-R, it appears that wt E5 induces PI 3-K activation via two signaling pathways: one in which PDGF-R activation is an intermediate step and another, which does not require PDGF-R activation. Transforming E5 mutants that are defective for PDGF-R activation have to activate PI 3-K via the latter pathway. This model is supported by our observation that the highly selective PDGF-R kinase inhibitor, tyrphostin AG1296, partially inhibits PI 3-K activation and tyrosine phosphorylation of the PI 3-K p85 subunit induced by wt E5 but does not inhibit PI 3-K activation and p85 tyrosine phosphorylation induced by L26A E5, which elicits little or no PDGF-R activation.
Mechanism of PI 3-K Activation by E5 Mutants-It has been reported that expression of the BPV-1 E5 oncoprotein in NIH 3T3 cells constitutively activates Ras and elevates levels of PI 3-K activity present in anti-epidermal growth factor receptor (EGF-R) immunoprecipitates (26). Although Ras can activate heterodimeric PI 3-K (22), it seems unlikely that Ras is involved in the activation of PI 3-K by the transforming E5 mutants that we have studied. Typically, Ras activates the Erk/MAPK signaling pathway, which includes sequential activation of the Raf, MEK, and Erk protein kinases (34 -36). We have not observed activation of Erk1 and Erk 2 in serumstarved NIH 3T3 cells that express mutant or wt E5 proteins. Since E5 is predominantly localized in membranes of the Golgi FIG. 7. PI kinase assays. A, immunoprecipitation of PI 3-K activity from serum-starved NIH 3T3 cells and cell lines expressing E5 constructs or the sis oncogene. Thin-layer chromatography was used to detect the phosphorylation of PI by anti-phosphotyrosine immunoprecipitates in the presence of [␥-32 P]ATP. The cells were treated with 1 mM Na 3 VO 4 (or 1 mM Na 3 VO 4 ϩ 10% FBS where indicated) for 10 min at 37°C before lysis, and equivalent amounts of cell protein were immunoprecipitated for each lipid kinase assay. B, PI 3-K activity in antiphosphotyrosine immunoprecipitates of serum-starved wt E5 and L26A E5 cell lines after treatment with 0.3-3 M tyrphostin AG1296 for 60 min at 37°C. The basal level of PI 3-K activity associated with an anti-phosphotyrosine immunoprecipitate from serum-starved normal 3T3 cells is shown for comparison. apparatus (8) and Ras-mediated activation of Raf occurs at the plasma membrane (37,46), cellular compartmentalization may prevent the downstream activation of MEK and Erk. More importantly, Ras activation of PI 3-K involves the GTP-dependent interaction of Ras with the PI 3-K p110 catalytic subunit and does not result in tyrosine phosphorylation of the PI 3-K p85 regulatory subunit (22). Our finding that PI 3-K activation induced by wt E5 and the Q17S, L24A, and L26A E5 mutants is associated with increased tyrosine phosphorylation of the PI 3-K p85 subunit argues against Ras-mediated activation. Since E5 has no intrinsic protein kinase activity, this finding implies that the Q17S, L24A, and L26A E5 mutants activate a proteintyrosine kinase other than the PDGF-R, which in turn activates PI 3-K by phosphorylating p85.
Receptor tyrosine kinases other than the PDGF-R are known to activate PI 3-K, including the colony-stimulating factor-1 receptor, EGF-R, and insulin receptor (19,24), and could potentially explain the ability of the PDGF-R-independent E5 mutants to transform cells. However, a number of results argue against a role for these kinases in PI 3-K activation by E5. 1. Goldstein et al. (15) show that the co-expression of wt BPV-1 E5 and the PDGF-R causes IL-3-independent proliferation in 32D murine hematopoietic cells that normally require IL-3 for growth. In contrast, IL-3-independent proliferation was not observed when the colony-stimulating factor-1 receptor or EGF-R were co-expressed with wt E5, indicating that E5 does not activate these receptors. NIH 3T3 cells have approximately 10% as many EGF-Rs as PDGF-Rs (47,48), and the addition of EGF to serum-starved NIH 3T3 cells increases PI 3-K activity in anti-EGF-R immunoprecipitates by only 45% (26). In our study, E5 mutants that are defective for PDGF-R activation increase PI 3-K tyrosine phosphorylation up to 13-fold. 3. PI 3-K activation by the insulin receptor does not involve tyrosine phosphorylation of the PI 3-K p85 subunit but rather is a direct result of p85 binding to distinct, tyrosine-phosphorylated insulin-receptor substrate proteins (19,49). E5 mutants that are deficient in PDGF-R activation do not support growth factorindependent proliferation in basal medium. Growth factor-independent proliferation does occur in cells where the PDGF-R is constitutively activated due to the expression of wt E5 or the sis oncogene. If alternative growth factor receptors were activated by E5 mutants, one might expect reduced growth factor requirements in these cells. However, despite these reservations, additional experiments will be required to definitively rule out the participation of alternative receptor tyrosine kinases. PI 3-K has also been detected in complexes with numerous nonreceptor tyrosine kinases, including focal adhesion kinase, Src, Fyn, Lyn, Abl, and Lck (50 -55). These interactions can lead to the activation of PI 3-K (50, 52, 54), although tyrosine phosphorylation of the PI 3-K p85 subunit is not obligatory (50,52).
Interestingly, the transforming activity of wt E5 and the PDGF-R-independent E5 mutants correlates with their ability to alkalinize the lumen of the Golgi apparatus in NIH 3T3 cells. Ratio-imaging experiments using a pH-sensitive fluorescent probe selectively targeted to the Golgi apparatus indicate that E5 inhibits acidification of the Golgi lumen by the vacuolar H ϩ -ATPase. 2 Inactivation of the vacuolar H ϩ -ATPase presumably results from binding interactions between its 16-kDa poreforming subunit and E5 (12). Further investigation will be necessary to determine whether or not an alkalinized Golgi compartment, pH 7.0, can lead to the activation of a protein kinase that in turn phosphorylates the p85 subunit of PI 3-K.

PI 3-K Activation Is Correlated with Anchorage-independent
Growth-We have shown that wt E5, which constitutively activates the PDGF-R and PI 3-K, causes the proliferation of NIH 3T3 cells to become independent of growth factors and attachment, whereas transforming E5 mutants that constitutively activate PI 3-K, but not the PDGF-R, confer anchorage independence only. This finding is consistent with the hypothesis that the Q17S, L24A, and L26A E5 mutants activate PI 3-K without activating any receptor tyrosine kinases, since inducible expression of a constitutively active PI 3-K mutant in rat embryo fibroblasts promotes anchorage-independent growth but does not eliminate the growth factor requirement for proliferation (25). PI 3-K activation is sufficient to cause serumstarved cells to enter S phase of the cell cycle, but progression through the entire cell cycle additionally requires growth factor receptor signaling (25). It may be that PI 3-K activation abrogates anchorage requirements for proliferation because anchorage normally is required for the activation of PI 3-K in response to growth factors (57).
Constitutive activation of the focal adhesion kinase and Abl tyrosine kinases, and Rho and Cdc42 small G-proteins, leads to cell proliferation that is anchorage-independent but growth factor-dependent (58). It is likely that some of these signaling proteins may confer anchorage independence by activating PI 3-K, whereas others may be downstream targets of PI 3-K that are important for enabling anchorage-independent growth (18,19,50,54). Since anchorage-independent growth is closely correlated with tumorigenicity in animal models (59) and transient expression of constitutively active PI 3-K promotes carcinoma invasion (56), dissecting the effect of the E5 oncoprotein on PI 3-K-dependent signaling pathways may provide valuable insights into normal cell proliferation and neoplasia.