transformation of mortal human fibroblasts and activation...

Transformation of Mortal Human Fibroblasts and Activation ofa Growth Inhibitory Pathway by the BovinePapillomavirus E5 Oncoprotein1

Lisa M. Petti2 and F. Andrew RayCenter for Immunology and Microbial Disease, Albany Medical College,Albany, New York 12208

AbstractThe 44-amino acid bovine papillomavirus E5 proteininduces tumorigenic transformation of immortalrodent fibroblasts by binding to and activating theplatelet-derived growth factor b receptor (PDGFbR).Here E5 was expressed in mortal human diploidfibroblasts (HDFs), which lack the accumulatedgenetic changes that are present in immortal rodentcells. E5 induced focus formation and morphologicaltransformation of HDFs without inducing anchorageindependence or immortalization. Similar effectswere observed with the v-sis and neu* oncogenes.E5-PDGFbR complexes were observed in the E5-expressing HDFs, as was constitutive PDGFbRactivation, which was required for the transformingactivity of E5. The E5 HDFs attained a highersaturation density than the control cells, expressingincreased levels of hyperphosphorylatedretinoblastoma protein at subconfluent densities.However, when these cells reached confluence,growth inhibition accompanied by dramatic down-regulation of the PDGFbR, and retinoblastomaprotein was induced apparently by a factor secretedinto the medium. This may represent a novelnegative feedback mechanism controlling PDGFbR-induced proliferation and thereby protecting againstcomplete transformation.

IntroductionThe E5 protein of BPV3 rapidly induces tumorigenic trans-formation of certain immortalized rodent fibroblasts and isthe primary transforming protein of this virus (1–4). The hy-

drophobic 44-amino acid E5 protein forms a disulfide-linkedhomodimer with a subunit size of Mr 7000 and localizesprimarily to intracytoplasmic membranes in BPV-trans-formed cells (5, 6). An important cellular target for the BPV E5protein in fibroblasts is the PDGFbR. In E5-transformed ro-dent fibroblasts, the E5 protein forms a stable complex withPDGFbR and induces constitutive activation of the PDGFbR(7, 8), an event that is required for cell transformation (9, 10).The E5 protein is thought to activate the PDGFbR by bindingto the receptor as a homodimer and facilitating receptordimerization, which then leads to receptor activation (11).

Studies of E5-mediated transformation have been per-formed primarily in mouse C127 cells, a line of immortalizedfibroblasts derived from a mouse mammary carcinoma (12),and the effect of E5 expression alone in mortal and com-pletely untransformed fibroblasts has not been investigated.Analysis of mortal cells has been limited to primary rodent orbovine fibroblasts expressing the entire BPV genome. Forexample, BPV induced morphological transformation of pri-mary hamster embryo fibroblasts at a low frequency, with thetransformed cells being anchorage independent and tumor-igenic (13–15). BPV infection of primary bovine dermal fibro-blasts resulted in morphological transformation, with thetransformed cells possessing activated PDGFbR associatedwith the E5 protein (16). Transfection of primary mouse fi-broblasts with BPV DNA resulted in immortalization afterserial subcloning, but not in focus formation (17, 18). Finally,transfection of rat embryo fibroblasts with BPV mutants re-sulted in immortalization that was dependent on the BPV E6gene (19). From these studies, it was not possible to deter-mine the traits that were directly attributed to E5 expressionbecause of the presence of the entire viral genome and/orthe high spontaneous transformation rates of rodent cells(20).

Here we assessed the effects of E5 expression in mortalhuman fibroblasts to determine the transforming charac-teristics that can be ascribed solely to this small but potentoncogene. Specifically, we used a retroviral vector con-taining the E5 gene to introduce E5 into HDFs. We chosehuman cells because rodent cells are susceptible to spon-taneous immortalization and tumorigenic transformationand because young mortal HDFs lack a background ofaccumulated genetic changes that might modify the ef-fects of E5. We found that the E5 gene can morphologi-cally transform and induce focus formation of these fibro-blasts, which are notoriously difficult to transform. Inagreement with several earlier studies (21–24), we alsofound that v-sis, the viral homologue of the PDGF B gene,can induce a similar transformed phenotype. Unlike theseprevious studies, however, here we show that the trans-formed phenotype of HDFs expressing E5 or v-sis is as-

Received 3/17/00; revised 5/12/00; accepted 5/16/00.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indi-cate this fact.1 Supported by National Cancer Institute Grant 1R29 CA73682-01A1(to L. M. P.) and a Schaffer Fellowship from Albany Medical College(to F. A. R.).2 To whom requests for reprints should be addressed, at Center forImmunology and Microbial Disease, MC-151, Albany Medical College, 47New Scotland Avenue, Albany, NY 12208. Phone: (518) 262-6285; Fax:(518) 262-5748; E-mail: [email protected] The abbreviations used are: BPV, bovine papillomavirus; PDGF, platelet-derived growth factor; PDGFbR; PDGF b receptor; HDF, human diploidfibroblast; ln, natural logarithm; FBS, fetal bovine serum; Rb, retinoblastomaprotein; MAPK, mitogen-activated protein kinase; PD, population doubling.

395Vol. 11, 395–408, July 2000 Cell Growth & Differentiation

sociated with and dependent on activation of thePDGFbR. We also found that neu* (the oncogenic form ofneu which encodes the p185neu receptor tyrosine kinase,activated by a single point mutation resulting in a Val toGlu substitution at position 664 in transmembrane domain)could induce focus formation and morphological transfor-mation of these cells. Although the E5-expressing HDFslost their contact inhibition for growth and continued toproliferate beyond the saturation density of the normalcells, their growth was still impeded once they reachedconfluence. This growth inhibition at confluence was ac-companied by down-regulation of the PDGF receptor, andboth events appeared to be induced by a factor secretedinto the medium. We propose that the E5-expressingHDFs activate a negative feedback mechanism to limit theproliferative effects of sustained PDGFbR activation.

ResultsPhenotypic Effects of Stable E5 Expression in HDFs. E5was stably expressed in HSF4012 cells (NHDF4012; Clonet-ics, San Diego, CA), a human foreskin fibroblast cell strain

with a normal karyotype4 that typically undergoes senes-cence after 55 PDs.5 E5 was introduced into these cells byinfecting low-passage (approximately 15 PDs) HSF4012 cellswith a high titer of an E5-expressing retrovirus containing apuromycin resistance marker. Cells stably expressing E5were established after 1 week of selection in puromycin-containing medium and then verified for expression of the E5protein by immunoblotting (Fig. 2A). Using the analogousretroviral construct without an insert and one containing thev-sis gene, control and v-sis-expressing HDFs, respectively,were established in parallel. The control HDFs were indistin-guishable from normal HSF4012 cells (Fig. 1A). In contrast,the E5-expressing HDFs displayed an altered morphologysimilar to that of murine fibroblast cell lines transformed bythis oncogene (25), appearing more elongated, refractile, anddense than the control cells (compare Fig. 1B with Fig. 1A).Virtually every cell that survived selection acquired the trans-formed morphology within just a few PDs (by 5–6 days after

4 F. A. Ray, unpublished observations.5 E. Okubo and F. A. Ray, unpublished observations.

Fig. 1. HDFs stably expressing theBPV E5 and the v-sis oncogenes aremorphologically transformed. Pho-tomicrographs of normal HDFs har-boring the retroviral vector alone (A)or HDFs stably expressing the BPV E5(B and E) or v-sis (C and F) oncogenesare shown. E5-expressing HDFstreated with AG1296 before conflu-ence are also shown (D). A2D, 347; Eand F, 3283.

396 BPV E5 Transformation of Human Fibroblasts

retroviral infection and just before confluence), suggestingthat morphological transformation was a direct effect of E5expression. Generally, the E5-expressing cells appeared themost refractile and elongated at higher cell densities, i.e.,near and at confluence (Fig. 1E). The E5-expressing cellsalso were multilayered and growing on top of each other (Fig.1B), whereas the control cells appeared as a flat monolayer(Fig. 1A). Moreover, the E5-expressing HDFs resembledHDFs stably expressing v-sis, as shown here (Fig. 1, C and F)and in earlier studies (22, 24). A similar morphology wasobserved previously for HDFs overexpressing c-sis (26) orHDFs treated with PDGF B (21).6 Therefore, E5, like v-sis, caninduce morphological transformation of HDFs.

Previous studies showed that in murine and bovine fibro-blasts, the E5 protein could complex with and activate bothmature and intracellular metabolic precursor forms of thePDGFbR (7, 9). Therefore, we determined whether thePDGFbR was constitutively tyrosine phosphorylated andcomplexed with the E5 protein in HDFs. To detect E5-PDGFbR complexes in HDFs, coimmunoprecipitation anal-ysis was performed by E5 immunoprecipitation followed byPDGF receptor immunoblotting. Both the slower migratingmature and faster migrating precursor forms of the PDGFbRwere detected in E5 immunoprecipitates from cells express-ing E5, whereas no coimmunoprecipitation of the receptorwas detected in the control cells (Fig. 2B). This indicates thatthe E5 protein can form a stable complex with the PDGFbRin HDFs. Next, the tyrosine phosphorylation status ofPDGFbRs in the E5-expressing HDFs was examined by an-tiphosphotyrosine immunoblotting of PDGF receptor immu-noprecipitates from subconfluent cells. As expected, whencompared with the control cells, tyrosine phosphorylation ofboth mature and precursor forms of the PDGFbR was dra-matically increased in the E5-expressing HDFs (Fig. 2C, topleft panel), indicating that this receptor is constitutively acti-vated in these cells. A similar increase in PDGF receptortyrosine phosphorylation was observed in the v-sis-express-ing HDFs. In both the E5- and v-sis expressing cells, theincrease in tyrosine phosphorylation was not due to a pro-portionate increase in the total amount of receptor expressed(Fig. 2C, bottom left panel). Thus, the E5 protein complexeswith the PDGFbR and induces its constitutive activation inmortal human fibroblasts.

Focus Formation of HDFs by E5. A focus-forming assaywas performed by infecting low-passage (approximately 15PDs) HSF4012 cells with a low concentration of retrovirusexpressing E5 or v-sis. As a control, cells were also infectedwith the analogous retrovirus without an insert. After infec-tion, cells were split and then maintained in the absence ofbiochemical selection and observed for the development offoci. Fig. 3 shows that no foci were induced after infectionwith the control retroviral vector (Fig. 3, Puro). In contrast, E5readily induced the formation of many large, dense foci ofpiled up cells. The foci induced by E5 had a star-shapedmorphology similar to that documented previously for fociinduced by v-sis or overexpression of c-sis (the cellular gene

encoding the PDGF B chain) in HDFs (21, 23, 24, 26). Focusformation induced by E5 was typically observed by 14 daysafter infection, suggesting that it was a direct effect of E5expression rather than a result of mutation during long-termculture. Corroborating the results of previous studies (21, 23,24), we showed that v-sis also could induce the formation ofnumerous star-like foci in HDFs (Fig. 3). In addition, the neu*oncogene also was assessed for focus-forming activity inHDFs. neu* encodes a mutated form of the epidermal growthfactor receptor family member p185neu that is constitutivelyactivated by a Val to Glu substitution at position 664 in thetransmembrane domain (27, 28). We found that retrovirusexpressing the activated p185neu* receptor readily inducedfocus formation of HDFs, whereas the corresponding emptyretroviral vector (LXSN) did not (Fig. 3). Thus, constitutiveactivation of a receptor tyrosine kinase other than the PDGFreceptor can result in focus formation in HDFs.

Inhibition of Transformation by the Kinase InhibitorAG1296. We also examined the effect of treating the E5-and v-sis-expressing HDFs with AG1296, a kinase inhibitor6 Y. Zhang and L. M. Petti, unpublished observations.

Fig. 2. The PDGFbR exists in a complex with the E5 protein and isconstitutively activated in E5-expressing HDFs. Extracts of HDFs stablyexpressing E5, v-sis, neu*, or the empty retroviral vector (C) were preparedas described in “Materials and Methods.” In A, approximately 1000 mg ofprotein were immunoprecipitated with an anti-E5 antiserum (E5 IP) andthen subjected to anti-E5 immunoblotting (E5 blot). In B, approximately400 or 40 mg of protein were immunoprecipitated with anti-E5 (E5 IP) oranti-PR (PR IP) antiserum, respectively, and then subjected to anti-PDGFreceptor immunoblotting (PR blot). In C, subconfluent HDFs stably ex-pressing E5, v-sis, or neu* were either left untreated (2) or treated (1) withAG1296 for 24 h and then lysed. Approximately 900 or 600 mg of extractedprotein were immunoprecipitated with anti-PDGF receptor (PR IP) oranti-p185neu (Neu IP) antibodies, respectively. Ninety percent of eachimmunoprecipitation was subjected to immunoblotting with an antiphos-photyrosine antibody (PY blot). The remaining 10% of each immunopre-cipitation was subjected to anti-PDGF receptor (PR blot) or anti-p185neu

(Neu blot) immunoblotting. The arrows in A and B point to the E5 proteinand the mature (m) and precursor (p) forms of the PDGF receptor (PR),respectively. Numbers to the right in C indicate the position of molecularweight markers in thousands.

397Cell Growth & Differentiation

reported to be specific for the PDGF receptor (29). AG1296treatment before confluence dramatically reduced tyrosinephosphorylation of the PDGF receptor in the E5- and v-sis-expressing cells, indicating that this agent inhibited the ty-rosine kinase activity of the PDGF receptor as expected (Fig.2C, top panel). AG1296 treatment before confluence alsoreversed the morphology change of the E5- (Fig. 1D) andv-sis (data not shown)-expressing HDFs without preventingcontinued growth to confluence. Specifically, the AG1296-treated cells became flatter and less refractile, resemblingthe control cells. It is important to note that the effects ofAG1296 in the E5-expressing cells were reversible; after itsremoval, tyrosine phosphorylation of the PDGF receptor wasrestored within an hour, and the characteristic transformedmorphology of the cells was reestablished within 1–2 days(data not shown). This suggests that E5 expression was notlost in the AG1296-treated cells. AG1296 treatment alsoinhibited focus formation by E5 or v-sis (Fig. 3). This inhibitionwas not simply due to inefficient retroviral infection becausecells treated with AG1296 were taken from the same infec-tions that resulted in focus formation. Thus, these resultsindicate that constitutive activation of the PDGF receptor isrequired for E5- and v-sis-induced morphological transfor-mation and focus formation of HDFs. To ensure that AG1296inhibitor was specific for the PDGF receptor, we assessedwhether it could inhibit morphological transformation andfocus formation induced by constitutive activation of a dif-ferent receptor tyrosine kinase, p185neu*. HDFs stably ex-pressing the activated p185neu* were established by retrovi-ral-mediated gene transfer as described in “Materials andMethods.” These cells appeared to be morphologicallytransformed (data not shown) and expressed abundantamounts of tyrosine-phosphorylated p185neu* as assessedby p185neu immunoprecipitation followed by antiphosphoty-rosine immunoblotting (Fig. 2C, top panel). AG1296 treat-ment before confluence could not inhibit tyrosine phospho-

rylation of p185neu* (Fig. 2C), nor could it reversemorphological transformation by neu* (data not shown).AG1296 was also unable to inhibit focus formation inducedby neu* (Fig. 3) and therefore was not a general inhibitor offocus formation. Taken together, these data indicate that theAG1296 inhibitor is specific for the kinase activity of thePDGF receptor. A structurally similar inhibitor, AG1295, wasshown to be specific for E5-activated PDGF receptors andnot for activated p185neu* in mouse C127 cells (30).

Growth Kinetics of E5-expressing HDFs. HDFs stablyexpressing E5, v-sis, or the empty retroviral vector wereseeded into multiple dishes at the same cell concentrationand then counted at intervals thereafter. As illustrated byrepresentative growth curves (Fig. 4), the E5- and v-sis-expressing HDFs consistently reached a 2–3-fold higher sat-uration density than the control cells. During the initial phaseof exponential growth, the growth rate of the E5-and v-sis-expressing cells was not significantly different than that ofthe control cells. However, the E5- and v-sis-expressing cellscontinued to proliferate exponentially after growth of thecontrol cells had slowed down. Notably, both control andtransformed cells continued to grow, albeit slowly, for manydays after exponential growth. Therefore, although the E5-and v-sis-expressing cells did not display an increasedgrowth rate initially, the exponential growth phase of thesecells was extended, allowing them to attain a higher satura-tion density than the control cells.

Assessment of E5-expressing HDFs for Other Trans-formed Traits. The E5-expressing HDFs were also exam-ined for other characteristics of transformed cells. First,these cells were assessed for their ability to grow under lowserum conditions (Fig. 5). Control or E5-expressing HDFswere seeded into medium containing either 1% or 10% FBSand then counted at various times thereafter. Growth of boththe control and E5-expressing cells in 1% FBS was severelyrestricted compared with growth in 10% FBS. This indicates

Fig. 3. The BPV E5, v-sis, and neu* oncogenes induce focus formation of HDFs. HSF4012 cells were infected with recombinant retroviral vectorsexpressing the BPV E5, v-sis, or neu* oncogenes. For controls, cells were infected with the retroviral vectors without an insert (Puro or LXSN). One day afterinfection, cells were split into new dishes, and 3 days later, the cells were either left untreated (2) or treated (1) with AG1246. Monolayers were maintainedat confluence for 3 weeks thereafter. Crystal violet-stained cells are shown for visualization of foci.


that the E5-expressing cells, like the normal cells, requiredhigher serum concentrations for optimal growth. However,even in 1% FBS, the E5-expressing HDFs were able to attaina higher final cell density than the control cells. Thus, al-though the E5 HDFs were serum dependent for growth, theystill maintained a growth advantage under low serum condi-tions.

The E5- and v-sis-expressing HDFs were also assessed foranchorage-independent growth. Equal numbers of normal,empty vector-, E5-, or v-sis-expressing HDFs were mixedwith methylcellulose-containing media, plated, and exam-ined 3 weeks later for the formation of colonies. As a positivecontrol, the HT1080 human fibrosarcoma cell line was alsotested. In three independent experiments, the E5- and v-sis-expressing cells, like the normal cells, formed either no orfew colonies in this semisolid medium (data not shown). Incontrast, a high percentage of the input HT1080 cells formedcolonies in methylcellulose, as was expected (data notshown). Therefore, neither E5 nor v-sis was able to induceanchorage-independent growth of HDFs.

The E5-expressing cells were also tested for the ability tobypass senescence. We found that the E5-expressing cells,like the normal cells, senesced after serial passaging (datanot shown), indicating that E5 was unable to directly induceimmortalization of HDFs. Furthermore, E5 expression did notappear to induce premature senescence of HDFs, as wasreported previously for ectopic expression of activated ras inthese cells (31).

Down-Regulation of the PDGF Receptor in E5-express-ing HDFs. For the PDGFbR and other receptor tyrosinekinases, ligand binding not only incites receptor activationbut also triggers the clustering of receptors in coated pits,followed by endocytosis and lysosomal degradation of re-ceptor-ligand complexes (32, 33), a process referred to asreceptor down-regulation. To determine the effect of E5 onPDGFbR down-regulation in HDFs, we examined PDGF re-

ceptor levels at different cell densities. An equal number ofnormal or E5-expressing cells was seeded into multipledishes, and, at various times during expansion of the cul-tures, monolayers were lysed for immunoblot analysis ortrypsinized in parallel for cell counts. To detect PDGF recep-tor levels in the cells, whole cell lysates were subjected toPDGF receptor immunoblotting. Fig. 6A shows that at anygiven cell density, the level of PDGF receptors in the E5-expressing cells was significantly less than that in the controlcells. This is likely due to receptor down-regulation in re-sponse to E5-induced receptor activation. Strikingly, whenE5-expressing HDFs had just reached confluence (at ap-proximately 2 3 106 cells/60-mm dish; Fig. 6B), PDGF re-ceptor levels were nearly abolished (Fig. 6A, upper panel, E5Lane 3). A similar decrease in the amount of activated, ty-rosine-phosphorylated receptor was observed when thecells reached confluence (data not shown). Thereafter, thegrowth rate of these cells was greatly reduced, and PDGFreceptor levels remained diminished for 3 days (Fig. 6A,upper panel, E5 Lanes 4 and 5). Dramatic receptor down-regulation also was observed when v-sis-expressing cellsreached confluence (data not shown). In contrast, the levelsof receptor in the control cells remained unchanged after thecells attained confluence and exhibited a reduced growthrate (Fig. 6A, control Lane 3). Receptor down-regulation atconfluence in the E5 HDFs was inhibited by AG1296 treat-ment (Fig. 9), suggesting that down-regulation required thekinase activity of the PDGF receptor. Therefore, constitutiveactivation of the PDGFbR by E5 in HDFs appears to lead todramatic down-regulation of the receptor when the cellsreach confluence. Others have demonstrated down-regula-tion of the PDGFbR in response to activation by E5 (34), v-sis(35), or v-fps (36). However, none of these studies relatedPDGFbR down-regulation to cell density.

Fig. 4. Stable expression of E5 and v-sis increases the saturation densityof HDFs. HDFs expressing E5 (f), v-sis (F), or empty vector (control, ❋)were seeded at equal cell densities into multiple dishes and then counted48 h later and every 24 h thereafter. For each time point, cells werecounted in triplicate. Growth curves were generated after plotting the ln ofthe cell number versus the hours after plating. The mean value of the ln ofthe cell number with the SD is shown.

Fig. 5. E5-expressing HDFs display limited growth in low serum. HDFsexpressing E5 or empty vector were seeded at equal cell densities intomedia containing either 10% (Œ, control cells; f, E5 cells) or 1% (❋,control cells; F, E5 cells) FBS and then counted at intervals thereafter. Forthe first four time points, the mean and SD of the ln of the cell number werecalculated from triplicate plates. The fifth and sixth time points werederived by counting two plates and one plate, respectively. The SD forsome points was too small for an error bar to be visualized.


In Fig. 6A (top panel), the conversion of distinct PDGFreceptor bands to faint, smeary bands suggested that thisreceptor was being rapidly degraded. This degradation wasspecific for the PDGF receptor because the levels of actinremained constant at different cell densities, as assessed byan anti-actin immunoblot of a lower portion of the same gel.A pulse-chase experiment confirmed that the half-life of thePDGF receptor was markedly reduced in the E5-expressingHDFs when they reached confluence (Fig. 7). Twelve h beforeconfluence (72 h after plating), a significant amount of re-ceptor was still present after a 1-h chase, and completedegradation occurred after a 4.5-h chase. It is important tonote that at this time, the PDGF receptor form with thelongest half-life is one with an electrophoretic mobility be-tween the mature (Fig. 7, m) and precursor (Fig. 7, p) forms.This may represent a more stable form of the receptor that iseither a specific degradation product of the mature form or adifferent, incompletely processed precursor. In contrast tothe subconfluent cells, in confluent cells (84 h after plating)metabolically labeled PDGF receptor is almost completelydegraded after a 1-h chase. Thus, increased degradationaccounts for down-regulation of the PDGF receptor in con-fluent E5 HDFs.

We also examined the levels of Rb protein in the E5-expressing and control HDFs at different cell densities. Fig.6A (bottom panel) shows that in subconfluent control and E5HDFs (Lanes 1 and 2) both hyper- and hypophosphorylatedforms of Rb (corresponding to the upper and lower bands,respectively) were present. This is to be expected becausethe hyperphosphorylated form should be prevalent in divi-ding cells, allowing for cell cycle progression by releasing thetranscription factor E2F. The amount of Rb in subconfluentE5-expressing cells appeared to be significantly increasedcompared with that in the control cells, particularly when thecells were approximately 50% confluent (Lanes 1). This wasdue primarily to an increase in hyperphosphorylated Rb (in-dicated by a lighter exposure of the Rb blot in Fig. 6A) andcould therefore account for the extended exponential growthof the E5-expressing HDFs compared with the control cells.When both the control and E5 HDFs reached confluence(Lanes 3), a significant reduction in hyperphosphorylated Rbwas observed. In addition, when the E5-expressing HDFsreached confluence, a faster migrating Rb form of approxi-mately Mr 68,000–70,000 appeared that was not evident inthe control cells. The presence of this smaller Rb form mayindicate specific cleavage of Rb, implying that both Rb andthe PDGF receptor are degraded in the E5 HDFs when con-fluence is reached.

We next determined whether the state of confluence wasthe direct cause of PDGF receptor down-regulation in theE5-expressing cells. For example, it is possible that theincreased number of cell-cell contacts in a confluent culturemight trigger a signal for receptor degradation. E5-express-ing HDFs were plated at a high density (2.7 3 106 cells/dish,which is at or slightly higher than confluence) or at a lowdensity (2 3 105 cells/dish). At various times after plating,cells were lysed and examined for PDGF receptor levels byimmunoblotting. Fig. 8 shows that the PDGF receptor wasnot immediately degraded after plating at saturation density.

Fig. 6. Rapid down-regulation of the PDGF receptor and Rb in E5-expressing cells occurs at confluence. An equal number of E5-express-ing or control HDFs were seeded into dishes, and at various timesthereafter, cells were trypsinized and counted or, in parallel, lyseddirectly in Laemmli sample buffer. A, whole cell extracts were electro-phoresed on two 7% polyacrylamide-SDS gels. The upper portion ofeach gel was immunoblotted with either anti-PDGF receptor antibodyor anti-Rb antibody. Anti-actin blotting of a lower portion of each gelindicated that an equal amount of protein was loaded in each lane.Arrows to the left indicate PDGF receptor (PR), Rb, or actin bands.Arrow on the right points to an apparent degradation product of Rb ofapproximately Mr 68,000 –70,000. Numbers on the right indicate theposition of molecular weight markers in thousands. B, counts of E5 (f)and control (F) cells were plotted as the ln of the cell number versushours post seeding. The growth curve in B indicates the densities of thecultures examined by immunoblotting in A, with points 1– 6 in B cor-responding to Lanes 1– 6 in A.


Instead, receptor degradation began some time between 1and 3 days after plating. Thus, in addition to confluence,there appears to be a time requirement for PDGF receptordown-regulation in the E5-expressing cells. In another ex-periment, we measured whether the E5 HDFs could continueto grow after being plated at confluence (approximately2.3 3 106 cells/dish). Indeed, these cells continued to grow,although not at the exponential rate, reaching a final densityof approximately 3 3 106 cells/plate 2 days after plating (datanot shown). This corresponded to the time that a more trans-formed morphology and degradation of the PDGF receptorbecame evident (data not shown). Continued growth afterplating at confluence may be due to the ability of these cellsto grow on top of each other.

We also assessed the effect of fresh media replacementon receptor down-regulation. Briefly, the media of E5 HDFsthat were nearing or had just reached confluence were re-placed with either serum-free media or serum-containingmedia, and at various times thereafter, the cells were lysedand examined for PDGF receptor levels by immunoblotting.Fig. 9 shows that serum deprivation clearly inhibited PDGFreceptor down-regulation, even when serum was removedjust as the cells were starting to degrade their receptors (i.e.,at 84 h). If the media on the cells were replaced with serum-containing media, receptor down-regulation still occurred,although the response was delayed by 1 day and reduced.These results indicate that serum factors are required forreceptor down-regulation and argue against the possibilitythat receptor down-regulation is caused by a depletion of

factors in the medium. If this were the case, down-regulationshould not occur after replacement with fresh serum-containing medium. A more likely possibility is that the cellssecreted a factor that induced receptor down-regulation andthat media replacement removed such a factor. Media re-placement with serum-containing medium would still removesuch a factor, but the presence of serum might allow cells tosecrete more of this factor and eventually induce receptordown-regulation.

To test the hypothesis that a secreted factor inducesPDGF receptor degradation, we determined whether con-ditioned media from confluent E5 HDFs that were degrad-ing their PDGF receptors could induce receptor down-regulation in subconfluent E5 cells that had not yet begunto degrade their receptors (Fig. 10A). Briefly, the media onsubconfluent E5 cells was replaced with media from E5cells that had just reached confluence and were down-regulating their PDGF receptors (S lane in Fig. 10A). Bothtreated and untreated cells were lysed 2.5 and 20 h later,and PDGF receptor levels were examined by immunoblot-ting. Strikingly, a 20-h treatment with conditioned mediafrom confluent cells dramatically down-regulated thePDGF receptors in subconfluent cells (Fig. 10A). Even ashort treatment of 2.5 h induced some receptor degrada-tion. Media taken from confluent control cells not express-ing E5 could not induce receptor down-regulation (Fig.10B). Receptor down-regulation was not a result of thesubconfluent cells becoming confluent and down-regulat-ing their receptors independent of the treatment because

Fig. 7. Degradation of the PDGF receptor in the E5 HDFs is increased at confluence. Control and E5-expressing HDFs were metabolically labeled witha mixture of [35S]methionine (71%) and [35S]cysteine (29%) 12 h before confluence (72 h after plating) or at confluence (84 h after plating). After 45 min,the label was removed, and the cells were either lysed immediately (0 h of chase) or incubated with media containing an excess of cold methionine andcysteine. After 1, 2.5, or 4.5 h of this cold chase, cells were lysed. PDGF receptor was immunoprecipitated from cell lysates and then subjected toSDS-PAGE followed by autoradiography. Arrows on the left point to the mature (m) and precursor (p) forms of the PDGF receptor (PR). The total amountof PDGF receptor present in the E5 HDFs at 72 and 84 h after plating is shown on the left of the PDGF receptor immunoblot in Fig. 9.

Fig. 8. Forced confluence does notdirectly induce PDGF receptordown-regulation in the E5 HDFs. Aculture of subconfluent E5-express-ing HDFs was trypsinized and platedat subconfluent (2 3 105 cells/dish)or confluent (2.7 3 106 cells/dish)densities in 60-mm dishes. Cellswere lysed at various times thereaf-ter, and lysates were subjected toSDS-PAGE followed by PDGF re-ceptor and actin immunoblotting ofthe same gel. PDGF receptor andactin levels were also examined incells immediately before (Lane b)and after (Lane a) trypsinization.Arrows to the left indicate the posi-tion of the mature (m) and precursor(p) forms of the PDGF receptor (PR).


receptor degradation was not evident in the untreatedcells. Moreover, the fact that subconfluent cells coulddegrade their PDGF receptors in response to the condi-tioned media provides further evidence that the confluentstate is not a direct cause of receptor degradation, al-though it may be required for secretion of a factor. Inter-

estingly, a 20-h treatment with this conditioned mediumalso induced down-regulation of hyperphosphorylated Rb(Fig. 10, A and B). Thus, down-regulation of hyperphos-phorylated Rb and degradation of the PDGF receptor areinduced by the same stimulus, presumably a factor se-creted into the medium at confluence.

Fig. 9. PDGF receptor down-regulation in the E5 HDFs requires serum factors as well as the kinase activity of the receptor. E5-expressing HDFs wereplated in 60-mm dishes at 2.5 3 105 cells/dish. Cells were either left untreated or subjected to a media change or AG1296 treatment just before or atconfluence (72 or 84 h after plating, respectively). The media replacement was performed with either serum-free medium or medium containing 10% FBS.AG1296 treatment was performed as described in “Materials and Methods” by adding the inhibitor (1) or DMSO (2) directly to the cells without replacingthe media. Cell lysates were prepared at the times indicated after plating and subjected to SDS-PAGE followed by PDGF receptor or actin immunoblotting.Arrows on the left point to the mature (m) and precursor (p) forms of the PDGF receptor (PR) as well as to actin.

Fig. 10. Conditioned medium from confluent E5 HDFs induces down-regulation of the PDGF receptor and Rb in subconfluent cells. Subconfluent E5 HDFsin 60-mm dishes were either left untreated (2) or had their media replaced (1) with conditioned media from E5 HDFs that had just reached confluence andwere degrading their receptors (Lane S in A). Cells were lysed at the time of treatment (time 0, untreated) or at the indicated times (h) after treatment. Ifunspecified, cells were lysed 20–24 h after treatment. In B, E5 HDFs were treated with conditioned media from confluent E5 HDFs (1E5) or confluent controlHDFs (1C). In C, subconfluent E5 or control HDFs were treated with medium from E5 HDFs. In D, one dish of E5 HDFs was treated with conditioned E5HDF medium that had been boiled for 5 minutes (Lane 1b). In E, E5 HDFs were treated with whole E5 medium (1) or with medium that was fractionatedby centrifugation through a filter device with a molecular weight cutoff of 3000. The retentate or concentrated unfiltered fraction containing high molecularweight solutes was added directly into the existing medium on the cells. The filtrate fraction containing the low molecular weight solutes was added to thecells by media replacement as described in A2D. Cell lysates were subjected to SDS-PAGE followed by PDGF receptor (PR), Rb, or actin immunoblottingas indicated. Arrows on the left point to the mature (m) and precursor (p) forms of the PDGF receptor (PR), hypophosphorylated Rb (pp), hyperphospho-rylated Rb (ppp) and actin.


Treating control fibroblasts with conditioned medium fromconfluent E5 cells resulted in down-regulation of the PDGFreceptor that was delayed with respect to the E5-expressingHDFs (Fig. 10C). After 20 h of treatment, there was minimalreceptor down-regulation, but by 43 h, there was significantreceptor down-regulation in these cells. Conditioned me-dium did not induce receptor activation in the control cells(data not shown). By 43 h, hyperphosphorylated forms of Rbwere also down-regulated in the control cells (data notshown).

We next began to characterize the factor in the condi-tioned medium of confluent E5 HDFs. First, boiling the con-ditioned media before treatment did not inhibit the ability ofthe media to induce down-regulation of the PDGF receptor inthe E5 HDFs (Fig. 10D), suggesting that the down-regulatingactivity in the medium was not enzymatic. To estimate thesize of the factor, we treated subconfluent E5 HDFs withmedium that was fractionated through a centrifugal filterdevice with a molecular weight cutoff of 3000. The majority ofthe down-regulating activity in the medium fractionated tothe filtrate containing low molecular weight solutes (Mr lessthan 3000) and was depleted from the retentate containingthe high molecular weight solutes (Fig. 10E). Taken together,these results suggest that when E5 HDFs reach confluence,they secrete a small, heat-stable soluble factor that inducesdown-regulation of the PDGF receptor and hyperphospho-rylated Rb.

We also determined the effect of this conditioned mediumon DNA synthesis and cell proliferation. Subconfluent E5 orcontrol HDFs were either treated with conditioned mediumfrom confluent E5 HDFs or left untreated. At various timesafter treatment, cells were examined for incorporation of

[3H]thymidine and counted in parallel. Fig. 11 shows that by43 h of treatment, DNA synthesis in the treated E5 HDFs wasdecreased 7-fold compared with the untreated cells. In ad-dition, the treated E5 HDFs could only reach a peak densitythat was half that of the untreated cells, suggesting that cellgrowth also was inhibited. By 65 h after treatment, the E5HDFs showed a 3-fold reduction in cell number comparedwith the untreated cells. Thus, conditioned medium fromconfluent E5 HDFs suppressed DNA synthesis and cell pro-liferation in subconfluent E5 HDFs. It is important to note thatthis effect occurred after the time (20 h) when down-regula-tion of the PDGF receptor and Rb was typically observed. Weexpected to see a decrease in DNA synthesis in the treatedcontrol cells 1 day after receptor down-regulation typicallyoccurred (e.g., as shown in Fig. 10C). In this particular ex-periment, however, these cells probably initiated the normalmechanism of density-dependent growth arrest before fac-tor-dependent growth arrest could occur.

DiscussionIn this report, we demonstrate that the 44-amino acid BPV E5protein can induce morphological transformation and focusformation and increase the saturation density for growth ofmortal HDFs. Similar effects were observed for expression ofthe v-sis oncogene, as shown previously (21–24). Unlikethese earlier studies, here we provide evidence that consti-tutive activation of the PDGF receptor mediates the trans-forming effects of E5 or v-sis in HDFs. We also found thatconstitutive activation of another receptor tyrosine kinase,p185neu*, imparts similar phenotypic changes in these cells.Therefore, two different receptor tyrosine kinases can appar-

Fig. 11. Conditioned mediumfrom confluent E5 HDFs sup-presses DNA synthesis and cellproliferation. Subconfluent E5 orcontrol HDFs in 24-well plateswere either left untreated (M) ortreated (o) with conditioned me-dium from E5 HDFs that had justreached confluence. At the timeof treatment and various timesthereafter, the cells were as-sayed for DNA synthesis bymeasuring [3H]thymidine incor-poration as described in “Mate-rials and Methods” or trypsinizedand counted in parallel.cpm/100,000 cells (top panels)and cell number (bottom panels)were determined in triplicate.Mean values and SD are shown.


ently use common signaling pathways to elicit transformingphenotypes in human fibroblasts.

One pathway shared by receptor tyrosine kinases is theRas/MAPK pathway (37). However, instead of inducingtransformation, exogenous expression of activated ras inHDFs has been shown to induce premature senescenceassociated with increased p16INK4a and p53 expression (31).In this study, it was proposed that premature senescenceoccurred to protect against an initial mitogenic stimulus byras that could lead to extensive proliferation and tumor-igenesis. Similar results were observed after exogenous ex-pression of constitutively activated forms of Raf or MAPKkinase in HDFs (38, 39), suggesting that the Ras/MAPK path-way is involved in the induction of p16INK4a and p53 expres-sion, which, in turn, promotes senescence. Therefore, acti-vated ras has a very different effect in HDFs than constitutiveactivation of the PDGFbR or p185neu, although Ras is adownstream effector of these receptors. This may be be-cause the phenotypic changes induced by sustained activa-tion of a receptor tyrosine kinase represent the combinedeffect of several interconnected signaling pathways, whichmay be very different from the effect of activating only theRas/MAPK pathway. In the E5-, v-sis-, and neu*-expressingHDFs, perhaps another receptor tyrosine kinase signalingpathway inhibits senescence by suppressing MAPK activa-tion. For example, in certain cell systems, the Raf-MAPKkinase-MAPK pathway is inhibited by activation of the phos-phatidylinositol 39-kinase-protein kinase B pathway throughinhibitory phosphorylation of Raf by protein kinase B (40, 41).We have preliminary evidence suggesting that MAPK activityis only transiently increased in the E5- and v-sis-expressingcells; compared with the control cells, it is only increased atvery low densities and is significantly reduced at higher den-sities (data not shown). A lack of sustained MAPK activitymay explain why these cells do not undergo premature se-nescence. Nonetheless, a different growth inhibitory path-way, which leads to quiescence rather than senescence, isactivated in the E5-expressing HDFs (see below). Impor-tantly, HDFs appear to have multiple growth control path-ways that may serve to protect against complete transfor-mation.

Despite the focus-forming activity of E5, which indicates aloss of contact inhibition, the growth rate of the E5-express-ing HDFs was still was greatly impeded once they reachedconfluence. We also found that the PDGF receptor wasrapidly and dramatically down-regulated in the E5-express-ing cells and not in the control cells when confluence wasreached. We propose that the E5-expressing HDFs activatean alternative growth inhibitory pathway at confluence, re-sulting in density-dependent growth arrest in the absence ofcontact inhibition. Specifically, our data suggest that a sol-uble factor secreted by these cells induces degradation ofboth the PDGFbR and hyperphosphorylated Rb, leading tosubsequent growth inhibition. Because of its insensitivity toheat and its small size, the factor secreted by E5 HDFs is notlikely to be a protease that directly degrades the PDGFbRand Rb. Instead, such a factor is more likely to stimulate anintracellular degradation pathway. This is substantiated bythe fact that only cycling cells and not quiescent cells can

degrade their receptors in response to this factor (data notshown). Thus, we present the possibility that an intracellularproteolysis pathway may be part of a back up density-dependent growth control mechanism in HDFs. Additionalexperiments are required to identify the secreted factor, de-termine the basis for its expression, ascertain the degrada-tion pathway it induces, and determine the role of this path-way in growth inhibition.

The confluent state is probably involved in secretion of thefactor rather than the degradation pathway per se becausedegradation was activated in subconfluent cells in responseto the factor. However, the confluent state is unlikely todirectly cause secretion of the factor because plating cells ata confluent density did not immediately induce PDGF recep-tor down-regulation. Under these conditions, the ability ofthe cells to assume the typical elongated and refractile“transformed” morphology (data not shown) was also de-layed and coincided with PDGF receptor down-regulation.Therefore, it may be the change in cell shape associated withthe confluent state that stimulates the secretion of a factor.

The inability of the control cells to down-regulate the PDGFreceptor at confluence may reflect their inability to secrete afactor rather than respond to one. Indeed, a soluble factorfrom the E5 HDFs could induce down-regulation of the PDGFreceptor and Rb in control cells, albeit more slowly, despitethe fact that the control cells have only a low basal level ofactivated PDGF receptor. Therefore, the down-regulationpathway probably does not require a high level of PDGFreceptor activation, although the increased level of activatedreceptors in the E5 HDFs may expedite this process. Thisseems to be inconsistent with previous studies suggestingthat internalization and degradation of the PDGFbR requirethe receptor to be active and able to phosphorylate one of itssubstrates, phosphatidylinositol 39-kinase (42, 43). It is pos-sible that instead of receptor-mediated endocytosis, the E5HDFs use another pathway such as ubiquitin-mediated pro-teolysis to degrade the PDGF receptor and Rb at confluence.Because conditioned medium from confluent control cellscould not induce receptor down-regulation in the E5 cells(Fig. 10), receptor activation is probably required for secre-tion of the factor. This may explain why AG1296 inhibitsreceptor down-regulation at confluence in the E5 cells (Fig.9). Thus, receptor activation appears to be required for se-cretion of the factor, but receptor activation over a basal levelmay not be required for the down-regulation process itself.

Like the PDGF receptor, Rb is probably down-regulated byproteolytic digestion when the E5 HDFs reach confluence, asevidenced by the disappearance of hyperphosphorylated Rband the appearance of a smaller Mr 68,000–70,000 form.This raises the intriguing possibility that degradation of Rband degradation of the PDGF receptor occur by the samemechanism. This is supported by the fact that down-regula-tion of Rb and the PDGF receptor in subconfluent control andE5 HDFs occurred in response to the same stimulus (i.e., asecreted factor from confluent E5 HDFs). A secreted factorcould stimulate the activation of a specific protease that actson the PDGF receptor and Rb as substrates, suggesting thatthe PDGF receptor and Rb share a common proteolyticrecognition sequence. Similarly, the secreted factor could


activate the specific targeting of these proteins for ubiqui-tinization. Alternatively, a proteolytic product of one proteincould stimulate the degradation of the other. Whatever themechanism, this represents the first report linking posttrans-lational processing of Rb to that of the PDGF receptor.

The Mr 68,000–70,000 Rb form is likely to be a specificCOOH-terminal cleavage product because the antibody thatdetected it was raised against a COOH-terminal peptide.Others have detected a similar Mr 68,000 Rb form in humanleukemia and lung cancer cell lines (44–46). One groupreported that this Rb form results from specific cleavage bya caspase-like activity during apoptosis (44, 45). It is possiblethat degradation of Rb and the PDGF receptor in the E5HDFs is also associated with apoptosis because a drop incell number was observed after treatment of cells with con-ditioned medium from confluent E5 HDFs (Fig. 11). A differ-ent group also detected a COOH-terminal Mr 68,000 Rb formbut found that it was not associated with apoptosis (46).Additional experiments are required to characterize this Mr

68,000–70,000 Rb form and assess its role in growth controlof the E5 HDFs.

We believe that that down-regulation of Rb and the PDGFreceptor plays a role in growth inhibition at confluence be-cause it stands to reason that a dramatic decrease in theamount of activated PDGFbR and hyperphosphorylated Rbmight serve to limit cell growth. Furthermore, when subcon-fluent cells were treated with the secreted factor, down-regulation of both proteins appeared to precede growth in-hibition. However, our data do not rule out the possibility thatdown-regulation of these proteins and growth inhibition re-sult from parallel pathways induced by the same stimulus,and additional experiments are required to definitively dem-onstrate that the two events are causally related.

The E5-expressing cells attained a higher saturation den-sity than the control cells under both high and low serumconditions, indicating that constitutive PDGFbR activationallowed a proliferative advantage. However, because thegrowth rate of the E5-expressing cells was significantly im-paired under low serum conditions, constitutive activation ofthe PDGFbR is not sufficient for optimal growth of HDFs. Thismay explain why exponentially growing E5 HDFs had agrowth rate similar to that of exponentially growing controlcells. We believe that sustained PDGFbR activation providesa proliferative advantage by delaying the onset of a density-dependent growth control mechanism rather than by in-creasing the rate of cell cycle progression. In both the E5 andcontrol HDFs, growth inhibitory mechanisms are activatedonce confluence is reached. However, the E5 HDFs attainconfluence at a higher density than the control cells, therebyraising the threshold density for onset of growth inhibition.Two characteristics of the E5 HDFs may account for this: (a)reduced cell spreading at higher densities, which couldcause each cell to occupy less space; and (b) the ability ofthe cells to grow on top of each other. Both factors may berelated to alterations in cell attachment. The fact that a backup growth control mechanism is activated in the E5 HDFs atconfluence despite a loss of contact inhibition suggests thatconfluence is an important growth-restricting state. Thus,sustained PDGFbR activation may effect changes in cell

attachment, which allow the cells to attain a higher densitybefore reaching a growth-restricting state.

Unlike previous studies that showed that v-sis, c-sis, orPDGF could induce anchorage-independent growth of HDFs(22, 26, 47), we were unable to demonstrate this activity foreither E5 or v-sis. One might argue that the level of E5 or v-sisexpressed in our system was not sufficient to induce detect-able anchorage-independent growth. By preselecting cellsthat expressed high levels of c-sis, Stevens et al. (26) showedthat a threshold level of c-sis expression was required forefficient anchorage-independent growth of HDFs. In our sys-tem, there appeared to be sufficient E5 and v-sis expressionto induce abundant tyrosine phosphorylation of thePDGFbR, yet anchorage independence was not induced.Alternatively, the discrepancy between previous studies andour study may be explained by our more rigorous assess-ment of anchorage independence. First, we used strictercriteria (.100 m) for assigning anchorage-independentgrowth and were therefore less likely to consider false pos-itives significant. Even when the E5- and v-sis-expressingcells displayed a low level of growth in semisolid medium(which was over 100-fold less than that of an anchorage-independent cell line), the same low level was also observedfor the control cells. Second, by using cells that were derivedfrom a transformed colony and expanded, the previous stud-ies may have selected for additional mutations conferringanchorage independence. It is possible that during the ex-pansion and passage of clonal populations, mutations oc-curred that led to increased numbers of anchorage-indepen-dent colonies. In our study, the cells were not cloned andwere used soon after infection. In any event, our data sug-gest that E5 and v-sis are unable to directly induce anchor-age-independent growth. This implies that constitutive acti-vation of the PDGFbR in HDFs is not sufficient to induceanchorage independence. It is possible that the growth in-hibitory pathway activated at confluence also may be acti-vated when these cells are suspended in semisolid medium.

Because E5 was unable to induce anchorage-independentgrowth, immortalization, or complete growth in low serum ofHDFs, we expect that it would be unable to fully transformthese cells to tumorigenicity. These observations are con-sistent with those of several other studies that reported thatv-sis could not impart an extended life span or make cellstumorigenic (22, 23, 24). Thus, although E5 has a potenttransforming effect in HDFs, it provides only some of thesteps necessary for the multistage process of oncogenesis.Recent evidence suggests that one requirement for tumori-genic transformation of human cells is telomere maintenanceand immortalization promoted by expression of the telomer-ase catalytic subunit hTERT (48). Hahn et al. (48) showed thatectopic expression of hTERT, together with activated Rasand the SV40 large T antigen, was sufficient to allow humanepithelial and fibroblast cells to form tumors when injectedinto nude mice. In this study, expression of hTERT with onlyactivated Ras resulted in senescence, as shown previouslyfor expression of activated Ras alone (31). Because E5 didnot appear to induce this premature senescence phenotype,it would be interesting to assess the effect of coexpressingE5 with hTERT and determine what additional oncogene


functions, if any, would be necessary and sufficient to inducetumorigenic transformation. For example, if the Mr 68,000–70,000 form of Rb observed in the E5 HDFs plays a role ingrowth inhibition at confluence, human papillomavirus E7might be able to contribute to complete transformation byinactivating this Rb form.

In conclusion, we provide evidence that constitutiveactivation of the PDGFbR can elicit transforming charac-teristics in mortal human fibroblasts. This stands to reasonbecause the PDGFbR is capable of eliciting signals toinduce cell proliferation and is abundantly expressed infibroblasts. However, it appears that HDFs adopt a strat-egy to abate the proliferative effect of PDGFbR activation,perhaps as a protective mechanism against further trans-formation. We have evidence to suggest that part of thisstrategy may be to activate an intracellular proteolysispathway. Additional experiments are required to identifythis pathway and determine how it may function in normalcells. Thus, the E5-expressing HDFs may allow character-ization of a novel growth regulatory mechanism. Thesecells should also be useful for examining the mechanisticrelationship between loss of contact inhibition and mor-phological transformation apart from other transformingphenotypes.

Materials and MethodsPlasmids. pRVH-E5 and pRVH-v-sis plasmids were obtained fromDaniel DiMaio (Yale University, New Haven, CT). Standard subcloningprocedures were used to subclone the BPV E5 and v-sis oncogenes fromthese plasmids into the pBabepuro retrovirus vector (a gift from JeffSettleman; Harvard University, Cambridge, MA) to generate pBabe-puro-E5 and pBabepuro-sis, respectively. The neu* cDNA was subclonedinto the LXSN retrovirus vector from pSRa-NeuNT (a gift from David Stern;Yale University) using standard methods. High-titer amphotropic retrovi-rus was obtained after transient transfection of the retroviral plasmids intothe Phoenix amphotropic producer cell line (American Type Culture Col-lection with permission from G. Nolan; Stanford University, Palo Alto, CA)as described previously (49).

Stable Expression of E5, v-sis, and neu*. Young HSF4012(NHDF4012) human foreskin fibroblasts were purchased from Clonetics atpassage 1 and maintained in MEM-a-10 [MEM-a (Life Technologies, Inc.)supplemented with 10% FBS, 50 units/ml penicillin, and 50 mg/ml strep-tomycin]. Cell strains stably expressing E5, v-sis, or neu* were establishedusing high-titer recombinant retroviruses expressing these oncogenes toinfect low-passage HSF4012 cells. Briefly, approximately 1 3 106 colony-forming units of retrovirus and 16 mg of Polybrene in 4 ml were added toHSF4012 cells at 80% confluence in a T75 flask. After 1 day, the infectedcells were trypsinized and split into eight to ten 100-mm dishes. Afterallowing the cells to adhere to the dishes, puromycin (Sigma) or G418 (LifeTechnologies, Inc.) was added to the cells at a final concentration of 0.5or 400 mg/ml, respectively. After 1 week of selection, stable cell strainswere established. Control cells were established by infection with thecorresponding retrovirus without an insert.

Focus-forming Assay. Passage 7 HSF4012 cells, which were at 90%confluence in 60-mm dishes, were infected with approximately 1 3 103

colony-forming units of recombinant retrovirus. One day after infection,the cells in each dish were split 1:3. Three days later, when the cells were90% confluent, 7 ml of DMSO or Tyrphostin AG1296 (Calbiochem; 17.5mM, final concentration) were added to the cells. The cells were thenmaintained at confluence in MEM-a-10, with a weekly media changecontaining fresh AG1296 or DMSO. Foci were visible 10–14 days afterinfection, and 3 weeks after infection, monolayers were fixed and stainedwith 1% crystal violet (w/v) in 70% ethanol.

Growth Curves. HDFs expressing E5, v-sis, or the empty retroviralvector were seeded into multiple 60-mm dishes at 1 3 105 cells/dish. Atintervals thereafter, cells in triplicate dishes were trypsinized and counted

using a Coulter counter. The number of cells per dish was converted to theln, and the results from the triplicate dishes were used to determinethe mean and SD at each time point. Cell doublings were calculated usingthe formula

lnF 2 lnIln2

,

where F 5 final cell number, and I 5 the initial cell number.Growth in Low Serum. Cells were diluted into media containing 1.0%

FBS and counted. Cells were then added to media containing either 1%or 10% FBS and plated at a density of 105 cells/60-mm dish. Cells weretrypsinized and counted in triplicate at intervals thereafter. Growth curveswere plotted as described above.

Anchorage Independence Assay. Control, E5- or v-sis-expressingcells (3 3 105) were suspended in 10 ml of 1.5% methylcellulose incomplete media as described previously (50). This suspension was mixedand distributed equally to three gridded 60-mm dishes that had beenprecoated with sterile 1.0% agarose in PBS. HT1080, a cell line derivedfrom a human fibrosarcoma, was used as a positive control and found tobe virtually 100% anchorage independent, based on daily observation ofcolony growth. Dishes were checked for clumps on the day after initiationof the assay, and no significant clumping problems were detected. Afterapproximately 3 weeks, colonies that were greater than 100 mm wereconsidered positive for anchorage-independent growth.

Immunoprecipitation and Immunoblotting. For the experimentshown in Fig. 2C, HSF4012 cells stably expressing E5, v-sis, or neu* in60-mm dishes at 80% confluence were treated with 10 ml of DMSO or 10ml of 10 mM AG1296 (final concentration, 25 mM). Twenty-four h later, thesecells as well as the corresponding untreated control cells (expressing theempty retroviral vectors pBabepuro or LXSN) were lysed in cold EBCbuffer as described previously (7, 8, 15). For the experiment shown in Fig.9, cells were treated with AG1296 in the same manner but lysed 2 daysafter treatment in protein sample buffer as described below. For Fig. 2, Aand B, cells were lysed in cold 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid extraction buffer as described previously (29).PDGF receptor was immunoprecipitated as described previously (8, 9, 15)by adding approximately 1 ml of anti-PR-C3a antibody (a gift from DanielDiMaio; this antibody recognizes the COOH-terminal 13 amino acids ofthe human PDGFbR) per 100–200 mg of protein extract. p185neu* wasimmunoprecipitated from EBC cell extracts by adding 10 ml of anti-p185neu antibody Ab-4 (Calbiochem) to approximately 600 mg of proteinextract. Immunoprecipitation of the E5 protein and its associated proteinswas performed as described previously (8, 15) by adding approximately 1ml of anti-E5 antibody (a gift from Daniel DiMaio; this antibody recognizesthe COOH-terminal 16 amino acids of the E5 protein) per 100–200 mg ofprotein extract. Washed immunoprecipitates were resuspended in 23Laemmli sample buffer. To prepare whole cell extracts for the experimentsshown in Figs. 6–10, the monolayers were washed twice with PBS andthen lysed by adding 200–250 ml of hot 23 Laemmli sample buffer. Wholecell extracts or immunoprecipitates were boiled; electrophoresed on aSDS-7%, 7.5%, or 15% (for E5 immunoblotting) polyacrylamide gel; andtransferred to nitrocellulose or polyvinylidene difluoride (for E5 immuno-blotting) at 100 V for 2 hours. Immunoblotting was performed as describedpreviously (8, 15) using a 1:1000 dilution of antiphosphotyrosine mono-clonal antibody 4G10 (Upstate Biotechnology, Inc.), a 1:1000 dilution ofanti-PDGF receptor antiserum (PR-C3a), a 1:100 dilution of anti-p185neu

monoclonal antibody (Ab-3, Calbiochem), a 1:500 dilution of anti-E5 an-tiserum, a 1:750 dilution of anti-Rb antibody (C-15; Santa Cruz Biotech-nology), or a 1: 500 dilution of anti-actin antiserum (Sigma). Proteins weredetected by enhanced chemiluminescence using a protein A- or anti-mouse IgG (for p185neu immunoblot)-horseradish peroxidase conjugate(Pierce).

Pulse-Chase Experiment. Monolayers of E5 or control HDFs thatwere either 80% (72 h after plating) or 100% (84 h after plating) confluentin 60-mm dishes were washed twice with PBS and preincubated forapproximately 30 min in serum-free, methionine-free, and cysteine-freeDMEM. Cells were then incubated in 1 ml of fresh serum-free, methionine-free, and cysteine-free medium containing approximately 418 mCi ofExpre35S35S label (New England Nuclear; 71% [35S]methionine and 29%[35S]cysteine) for 45 min. Cells were then washed twice with PBS andeither lysed immediately in EBC buffer or incubated in fresh mediumcontaining 150 mg/ml L-methionine and 120 mg/ml L-cysteine for various


times before lysis. After preclearing the extracts by incubation with proteinA-Sepharose beads, PDGF receptor was immunoprecipitated and sub-jected to SDS-PAGE, followed by fluorography and autoradiography.

Fractionation of Conditioned Medium. Conditioned medium fromE5 HDFs that had just reached confluence was fractionated by centrifu-gation at 1600 3 g for 3 h through a Centriplus (Millipore) centrifugal filterdevice with a molecular weight cutoff of 3000. The retentate, which wasconcentrated down to a smaller volume and contained solutes with mo-lecular weights of .3000, was added directly to the existing medium ofsubconfluent E5 HDFs to restore its original concentration. The filtrate,which contained solutes with molecular weights of ,3000, was added tosubconfluent E5 HDFs after removal of the existing medium.

DNA Synthesis Assay. For the experiment shown in Fig. 11, E5 orcontrol HDFs were seeded at 1–2 3 104 cells/well in 24-well plates, and2 days later, when the cells were still subconfluent, they were either leftuntreated or treated with conditioned medium from E5 HDFs that had justreached confluence. At the time of treatment and various times thereafter,the cells were assayed for [3H]thymidine incorporation or trypsinized forcell counts. For [3H]thymidine incorporation, cells were incubated for 2 hwith 100 ml/well of 1.5 mCi/ml [3H]thymidine (New England Nuclear).Triplicate wells were assayed for cold trichloroacetic acid-precipitable,hot perchloric acid-soluble radioactivity as described previously (30). Cellcounts were performed in parallel from triplicate wells using a hemocy-tometer.

AcknowledgmentsWe gratefully acknowledge Daniel DiMaio, David Stern, and Jeff Settle-

man for essential reagents and Julia Schaefer and Daniela Drummond-Barbosa for subcloning neu* and E5, respectively, into the appropriateretroviral vectors. We also thank Daniel DiMaio, Judith Laffin, TomFriedrich, John Lehman, Ying Zhang, and Jeff Banas for invaluable dis-cussions and advice.

References1. DiMaio, D., Guralski, D., and Schiller, J. T. Translation of open readingframe E5 of bovine papillomavirus is required for its transforming activity.Proc. Natl. Acad. Sci. USA, 83: 1797–1801, 1986.

2. Schiller, J. T., Vass, W. C., Vousden, K. H., and Lowy, D. R. E5 openreading frame of bovine papillomavirus type 1 encodes a transforminggene. J. Virol., 57: 1–6, 1986.

3. Burkhardt, A., DiMaio, D., and Schlegel, R. Genetic and biochemicaldefinition of the bovine papillomavirus E5 transforming protein. EMBO J.,6: 2381–2385, 1987.

4. Bergman, P., Ustav, M., Sedman, J., Moreno-Lopez, J., Vennstrom, B.,and Pettersson, U. The E5 gene of bovine papillomavirus type 1 is suffi-cient for complete oncogenic transformation of mouse fibroblasts. Onco-gene, 2: 453–459, 1988.

5. Schlegel, R., Wade-Glass, M., Rabson, M. S., and Yang, Y-C. The E5transforming gene of bovine papillomavirus encodes a small hydrophobicprotein. Science (Washington DC), 233: 464–467, 1986.

6. Burkhardt, A., Willingham, M., Gay, C., Jeang, K-T., and Schlegel, R.The E5 oncoprotein of bovine papillomavirus is oriented asymmetrically inGolgi and plasma membranes. Virology, 170: 334–339, 1989.

7. Petti, L., Nilson, L. A., and DiMaio, D. Activation of the platelet-derivedgrowth factor receptor by the bovine papillomavirus E5 transformingprotein. EMBO J., 10: 845–855, 1991.

8. Petti, L., and DiMaio, D. Stable association between the bovine pap-illomavirus E5 transforming protein and activated platelet-derived growthfactor receptor in transformed mouse cells. Proc. Natl. Acad. Sci. USA,89: 6736–6740, 1992.

9. Nilson, L. A., and DiMaio, D. Platelet-derived growth factor receptorcan mediate tumorigenic transformation by the bovine papillomavirus E5protein. Mol. Cell. Biol., 13: 4137–4145, 1993.

10. Drummond-Barbosa, D., Vaillancourt, R. R., Kazlauskas, A., andDiMaio, D. Ligand-independent activation of the platelet-derived growthfactor b receptor: requirements for bovine papillomavirus E5-inducedmitogenic signaling. Mol. Cell. Biol., 5: 2570–2581, 1995.

11. Lai, C-C., Henningson, C., and DiMaio, D. Bovine papillomavirus E5protein induces oligomerization and trans-phosphorylation of the platelet-derived growth factor b receptor. Proc. Natl. Acad. Sci. USA, 95: 15241–15246, 1998.

12. Lowy, D. R., Rands, E., and Scolnick, E. M. Helper-independenttransformation by unintegrated Harvey sarcoma virus DNA. J. Virol., 26:291–298, 1979.

13. Amtmann, E., and Sauer, G. Bovine papillomavirus transcription:polyadenylated RNA species and assessment of the direction of transcrip-tion. J. Virol., 43: 59–66, 1982.

14. Geraldes, A. Malignant transformation of hamster cells by cell-freeextracts of bovine papillomas (in vitro). Nature (Lond.), 222: 1283–1285,1969.

15. Zhang, Y. L., Lewis, A., Jr., Wade-Glass, M., and Schlegel, R. Levelsof bovine papillomavirus RNA and protein expression correlate with vari-ations in the tumorigenic phenotype of hamster cells. J. Virol., 61: 2924–2928, 1987.

16. Petti, L., and DiMaio, D. Specific interaction between the bovinepapillomavirus E5 transforming protein and the b receptor for platelet-derived growth factor in stably transformed and acutely transformed cells.J. Virol., 68: 3582–3592, 1994.

17. Mantyjarvi, R., Sarkkinen, H., Parkkinen, S., Laatikainen, A., Ryhanen,A., and Syrjanen, K. Tumorigenicity of primary mouse fibroblasts trans-formed by bovine papillomavirus type 1. Intervirology, 29: 311–319, 1988.

18. Mantyjarvi, R., Sarkkinen, H., Parkkinen, S., Ryhanen, A., Karjalainen,H., Syrjanen, K., and Syrjanen, S. Phenotypic transformation of primarymouse fibroblasts by BPV1 DNA. Arch. Virol., 100: 17–25, 1988.

19. Cerni, C., Binetruy, B., Schiller, J. T., Lowy, D. R., Meneguzzi, G., andCuzin, F. Successive steps in the process of immortalization identified bytransfer of separate bovine papillomavirus genes into rat fibroblasts. Proc.Natl. Acad. Sci. USA., 86: 3266–3270, 1989.

20. Kraemer, P. M., Travis, G. L., Ray, F. A., and Cram, L. S. Spontaneousneoplastic evolution of Chinese hamster cells in culture: multistep pro-gression of phenotype. Cancer Res., 43: 4822–4827, 1983.

21. Johnsson, A., Betsholtz, C., Heldin, C-H., and Westermark, B. Anti-bodies against platelet-derived growth factor inhibit acute transformationby simian sarcoma virus. Nature (Lond.), 317: 438–440, 1985.

22. Johnsson, A., Betsholtz, C., Heldin, C-H., and Westermark, B. Thephenotypic characteristics of simian sarcoma virus-transformed humanfibroblasts suggest that the v-sis gene product acts solely as a PDGFreceptor agonist in cell transformation. EMBO J., 5: 1535–1541, 1986.

23. Fry, D. G., Milam, L. D., Maher, V. M., and McCormick, J. J. Trans-formation of diploid human fibroblasts by DNA transfection with the v-sisoncogene. J. Cell. Physiol., 128: 313–321, 1986.

24. Fry, D. G., Hurlin, P. J., Maher, V. M., and McCormick, J. J. Trans-formation of diploid human fibroblasts by transfection with the v-sis,PDGF2/c-sis, or T24 H-ras genes. Mutat. Res., 199: 341–351, 1988.

25. Settleman, J., Fazeli, A., Malicki, J., Horwitz, B. H., and DiMaio, D.Genetic evidence that acute morphologic transformation, induction ofcellular DNA synthesis, and focus formation are mediated by a singleactivity of the bovine papillomavirus E5 protein. Mol. Cell. Biol., 9: 5563–5572, 1989.

26. Stevens, C. W., Brondyk, W. H., Burgess, J. A., Manoharan, T. H.,Hane, B. G., and Fahl, W. E. Partially transformed, anchorage-indepen-dent human diploid fibroblasts result from overexpression of the c-sisoncogene: mitogenic activity of an apparent monomeric platelet-derivedgrowth factor 2 species. Mol. Cell. Biol., 8: 2089–2096, 1988.

27. Bargmann, C. I., Hung, M-C., and Weinberg, R. A. Multiple independ-ent activations of the neu oncogene by a point mutation altering thetransmembrane domain of p185. Cell, 45: 649–657, 1986.

28. Bargmann, C. I., Hung, M-C., and Weinberg, R. A. The neu oncogeneencodes an epidermal growth factor receptor-related protein. Nature(Lond.), 319: 226–230, 1986.

29. Kovalenko, M., Gazit, A., Bohmer, A., Rorsmann, C., Ronnstrand, L.,Heldin, C-H., Waltenberger, J., Bohmer, F-D., and Levitzki, A. Selectiveplatelet-derived growth factor receptor kinase blockers reverse sis-trans-formation. Cancer Res., 54: 6106–6114, 1994.


30. Klein, O., Polack, G. W., Surti, T., Kegler-Ebo, D., Smith, S. O., andDiMaio, D. Role of glutamine 17 of the bovine papillomavirus E5 protein inplatelet-derived growth factor b receptor activation and cell transforma-tion. J. Virol., 72: 8921–8932, 1998.

31. Serrano, M., Lin, A. W., McCurrach, M. E., Beach, D., and Lowe, S. W.Oncogenic ras provokes premature cell senescence associated with ac-cumulation of p53 and p16INK4a. Cell, 88: 593–602, 1997.

32. Nilsson J., Thyberg, J., Heldin, C. H., Westermark, B., and Wasteson,A. Surface binding and internalization of platelet-derived growth factor inhuman fibroblasts. Proc. Natl. Acad. Sci. USA, 80: 5592–5596, 1983.

33. Goldstein, J. L., Brown, M. S., Anderson, R. G., Russell, D. W., andSchneider, W. J. Receptor-mediated endocytosis: concepts emergingfrom the LDL receptor system. Annu. Rev. Cell Biol., 1: 1–39, 1985.

34. Nilson, L. A., Gottlieb, R. L., Polack, G. W., and DiMaio, D. Mutationalanalysis of the interaction between the bovine papillomavirus E5 trans-forming protein and the endogenous b receptor for platelet-derivedgrowth factor in mouse C127 cells. J. Virol., 69: 5869–5874, 1995.

35. Garrett, J. S., Coughlin, S. R., Niman, H. L., Tremble, P. M., Giels,G. M., and Williams, L. T. Blockade of autocrine stimulation in simiansarcoma virus-transformed cells reverses down-regulation of platelet-derived growth factor receptors. Proc. Natl. Acad. Sci. USA, 81: 7466–7470, 1984.

36. Anderson, D. H., and Ismail, P. M. v-fps causes transformation byinducing tyrosine phosphorylation and activation of the PDGF b receptor.Oncogene, 16: 2321–2331, 1998.

37. Mulcahy, L. S., Smith, M. R., and Stacey, D. W. Requirement for rasproto-oncogene function during serum-stimulated growth of NIH 3T3cells. Nature (Lond.), 313: 241–243, 1985.

38. Zhu, J., Woods, D., McMahon, M., and Bishop, J. M. Senescence ofhuman fibroblasts induced by oncogenic Raf. Genes Dev., 12: 2997–3007,1998.

39. Lin, A. W., Barradas, M., Stone, J. C., van Aelst, L., Serrano, M., andLowe, S. W. Premature senescence involving p53 and p16 is activated inresponse to constitutive MEK/MAPK mitogenic signaling. Genes Dev., 12:3008–3019, 1998.

40. Rommel, C., Clarke, B. A., Zimmermann, S., Nunez, L., Rossman, R.,Reid, K., Moelling, K., Yancopoulos, G. D., and Glass, D. J. Differentiation

stage-specific inhibition of Raf-MEK-ERK pathway by Akt. Science(Washington DC), 286: 1738–1741, 1999.

41. Zimmermann, S., and Moelling, K. Phosphorylation and regulation ofRaf by Akt (protein kinase B). Science (Washington DC), 286: 1741–1744,1999.

42. Joly, M., Kazlauskas, A., Fay, F. S., and Corvera, S. Disruption ofPDGF receptor trafficking by mutation of its PI-3 kinase binding sites.Science (Washington DC), 263: 684–687, 1994.

43. Shpetner, H., Joly, M., Hartley, D., and Corvera, S. Potential sites ofPI-3 kinase function in the endocytic pathway revealed by the PI-3 kinaseinhibitor, wortmannin. J. Cell Biol., 132: 595–605, 1996.

44. Bing, A., and Dou, Q. P. Cleavage of retinoblastoma protein duringapoptosis: an interleukin 1b-converting enzyme-like protease as candi-date. Cancer Res., 56: 438–442, 1996.

45. Fattman, C. L., Bing, A., and Dou, Q. P. Characterization of interiorcleavage of retinoblastoma protein in apoptosis. J. Cell. Biochem., 67:399–408, 1997.

46. Chen, W., Otterson, G. A., Lipkowitz, S., Khleif, S. N., Coxon, A. B.,and Kaye F. J. Apoptosis is associated with cleavage of a 5 kDa fragmentfrom RB which mimics dephosphorylation and modulates E2F binding.Oncogene, 14: 1243–1248, 1997.

47. Palmer, H., Maher, V. M., and McCormick, J. J. Platelet-derivedgrowth factor or basic fibroblast growth factor induce anchorage-inde-pendent growth of human fibroblasts. J. Cell. Physiol., 137: 588–592,1988.

48. Hahn, W. C., Counter, C. M., Lundberg, A. S., Beijersbergen, R. L.,Brooks, M. W., and Weinberg, R. A. Creation of human tumour cells withdefined genetic elements. Nature (Lond.), 400: 464–468, 1999.

49. Pear, W. S., Scott, M. L., and Nolan, G. P. Generation of high titre,helper-free retroviruses by transient transfection. In: P. Robbins (ed.),Methods in Molecular Medicine: Gene Therapy Protocols, pp. 41–57.Totowa, NJ: Humana Press, 1996.

50. Ray, F. A., Peabody, D. S., Cooper, J. L., Cram, L. S., and Kraemer,P. M. SV40 T antigen alone drives karyotype instability which precedesneoplastic transformation of human diploid fibroblasts. J. Cell. Biochem.,42: 13–31, 1990.


transformation of mortal human fibroblasts and activation...

Documents