MOLECULAR CARCINOGENESIS 6: 100-1 1 1 (1992)
Regulation of Growth and Gene Expression in Human PapiIlomavirus-Transformed Keratinocytesby Transforming Growth Factor-p: Implications for the Control of Papillomavirus Infection Lundy Braun,' Matthias Diirst, Ryoko Mikurno, Annette Crowley, and Mava Robinson Department of Pathology and Laboratory Medicine (LB, RM, AC, MR), Division of Biology and Medicine, Brown University, Providence, Rhode Island; and lnstitut fur Virusforschung (MD), Deutsches Krebsforschungszentrum, Heidelberg, Germany
Cervical carcinogenesis is a multistep process that appears t o be initiated by infection o f squamous epithelial cells in the cervix w i th one o f a limited number o f human papillomavirus (HPV) types. However, the mechanisms involved in th e evolution o f benign, HPV-induced lesions t o malignancy have not yet been fully elucidated. Transforming growth factor-p (TGF-p), a multifunctional growth factor produced by cells in the skin, inhibits the proliferation o f foreskin and cervical keratinocytes in vitro. We examined the effects of TGF-p on growth and virus early-gene expression in cell lines immortalized by t w o HPV types associated with cervical carcinogenesis as well as th e expression o f TGF-PI mRNA transcripts i n normal and HPV-positive cells in vivo and i n vitro. We found that normal and HPV-positive cells expressed similar levels of TGF-p1 mRNAs and exhibited similar patterns o f responsiveness t o three isoforms o f TGF-p in both monolayer and modified organotypic cultures. Of particular interest is our finding that th e expression of the E6 and E7 early viral transforming regions of both HPV16 and HPVIS was reversibly and rapidly inhibited by TGF-p. In one HPV16-positive cell line examined in detail, inhibition o f HPV expression required protein synthesis and occurred at the level o f transcription. HPVimmortalized cellsselected for resistancet o in vitro differentiation signals remained sensitive [email protected]
growth inhibition. These results, showing th a t both growth and virus gene expression in HPV-transformed cells were responsive to TGF-p, suggest that endogenous growth factors produced by different cell types in squarnous epithelium may play a role in th e progression of cervical neoplasia. D 1992 wi~ey-~iss, Inc. Key words: Viral carcinogenesis, growth factors, cervical cancer, human papillomavirus
INTRODUCTION Human papillomaviruses (HPLS) are epitheliotropic DNA viruses that produce benign hyperproliferative lesions, commonly known as warts, on mucosal and cutaneous surfaces. In the past decade, clinical, molecular, and pathological investigations have demonstratedthat specific HPV types, notably HPV16 and HPVl8, are present in a majority of female genital tract cancers worldwide [ I ] . One of the striking features of papillomavirusinfection is the high frequency with which cutaneous warts undergo spontaneous regression.The cellular events that regulate the disappearance of warts are not understood but probably involve cell-mediated immunity as well as regulation by endogenous growth factors. In recent years, some of the growth factors that are functionally important in the regulation of normal cell growth and differentiation in the skin have been identified, and we have proposed that factors locally expressed by cells in the skin may be involved in the regulation of papillomavirus-host cell interactions 121. Progress in understandingthe pathogenesisof HPV infection has been hampered by the lack of animal or tissueculture model systems that permit virus growth. Several 0 1992 WILEY-LISS, INC.
groups have reported immortalizationof neonatal foreskin [3-51 or adult cervical [6,7] keratinocytes with HPV16 and HPVl8 DNA. Although there have been no reports to date of virus-particle production in any HPV-immortalized cell lines, presumably because cellular factors required for induction of late viral functions are not produced in vitro, culture of these cells on three-dimensional collagen rafts producesepithelium morphologically indistinguishablefrom cervical intraepithelial neoplasia [81. Thus, these cultures represent physiologically relevant model systems for studying the multistep process of papillomavirus-associatedtransformation. Using primary keratinocytes, it has been possible to establish that the E6 and E7 genes transcribed from the early regions of the HPV16 genome are both necessary and sufficient for transformation of skin keratinocytes [9,10]. Evidence that HPVs can interact with and perhaps 'Corresponding author: Department of Pathology and Laboratory Medicine, Division of Biology and Medicine. Brown University, Providence, RI 02912. Abbreviations: DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; GAPDH, glyceraldehyde-3-phosphatedehydrogenase; HPK, HPV16-immortalized keratinocytes; HPV, human papillomavirus; KGM, keratinocyte growth medium; TGF, transforming growth factor.
TGF-p REGULATION OF HPV-TRANSFORMEDKERATINOCYTES
alter the function of cellular proteins is provided by studies showing that HPV E6 and E7 viral oncoproteins bind to the p53 and p l 0 5 retinoblastoma tumor-suppressor proteins [l1-13]. Although it is unclear how binding affects the cellular phenotype, especially since HPV oncoproteins from low-risk H P k also bind the retinoblastoma protein, albeit with lower affinity, these results suggest that modulation of virus activity by cellular genes may be important in regulation of cell growth. The transforming growth factor (TGF)-ps are a family of at least five polypeptide growth factors, of which two, TGF-p1 and TGF-p2, have been extensively characterized 114-1 61. TGF-p1 is a potent growth inhibitor of many epithelial cells, including human and murine keratinocytes[ 151. Despite intensive investigations, the mechanisms by which TGF-p1 exerts its effects in epithelial cells are still poorly understood, but some of the early molecular events associated with growth inhibition of human skin and cervical keratinocytes by TGF-pl include induction of c-fosand c-jun expression [ 2 ] , inhibition of c-myc expression 12,171 and induction of type 1 serinekhreonine protein-phosphatase activity (181. It has been proposed that in some tissues, loss of sensitivity to TGF-p1 growth inhibition may be involved in malignant transformation [ 19,201. Important evidence to support this concept in viral carcinogenesis was provided by the work of Blomhoff et al. , which showed that B lymphocytes infected with Epstein-Barr virus in vitro were resistant to TGF-p1 growth inhibition. In contrast, recent work in our laboratory has shown that the growth of nontumorigenic keratinocytes that express HPVI 6 mRNAs and proteins is inhibited by TGF-p1. Only in HPV-positive cell lines derived from cervical carcinomas did we detect a large decrease in sensitivity to TGF-p1 [2,22,23]. Similar results with HPV16-positive cervical cells have been reported by Woodworth et at. . On the other hand, Pietenpol et al.  have found that HPV-positive keratinocytesare resistant to TGF-P1-mediated growth inhibition. In this study, we confirmed and extended our earlier findings and those of Woodworth et al.  showing that HPVl6-positive keratinocytes were inhibited by TGF-p1 and 4 2 . We also showed that the growth of HPVl8-positive keratinocyte cell lines was regulated by TGF-p1 and that resistance to in vitro differentiation signals was not accompanied by a loss of sensitivity to this family of growth-regulatory factors. MATERIALS AND METHODS Culture of Keratinocytes Keratinocyte cultures were established from neonatal foreskin tissue as previously described . Briefly, the foreskin was incubated overnight in the cold in 0.25% trypsin solution (GIBCO, Grand Island, NY). Epidermal tissue was then separated from the dermis using a fine forceps, minced, and incubated for 1 h in trypsirdethylenediarninetetraacetic acid solution. Keratinocytes were plated on an irradiated feeder layer of NIH 3T3 cells in Dulbecco‘s modified Eagle‘s medium (DMEM; Gibco, Grand Island, NY) supplemented with 5% fetal bovine serum (FBS) and 0.4 pg/mL
hydrocortisone. Subsequent passages were grown in serumfree keratinocyte growth medium (KGM; Clonetics, San Diego, CAI supplemented with 5 pg/mL insulin, 0.5 pg/mL hydrocortisone, 10 ng/mL epidermal growth factor, and 0.4% bovine pituitary extract (referred to as complete KGM). Transfection of Human Keratinocytes W i t h HPV16and HPV18 Secondary passages of human keratinocyteswere plated at a density of 2-3 x l o 5 cells/35-mm Petri dish and grown to approximately 70% confluency in complete KGM. The plasmids used for transfection were HPV16 cloned into pGEM-3 (obtained from Mark Stoler, Cleveland Clinics) at the BamHl site in the L1 open reading frame of the viral genome and HPV18 cloned into either pBR322  or pGEM-3 a t the EcoRl site in the E l open reading frame .Transfection with lipofectin reagent (Bethesda Research Laboratories, Gaithersburg, MD) was performed in KGM with 5-1 0 p g of linearized pHPV DNA according to the manufacturer‘s instructions. Keratinocytes were then fed fresh KGM and maintained under serum-free conditions with frequent feedings. Cultures were split at a ratio of 1 :2 when confluent and analyzed for viral DNA sequences and RNA expression between the ninth and eighteenth passages. HPV-immortalized cells were selected for their escape from senescence in culture as previously reported . Cell lines established in these experiments were designated P K I 6 or P K I 8. The letters A-H were used to indicate the dish from which the cell line was derived. In the seventeenth passage (PKI 6) or the nineteenth passage (PKI 8), KGM containing 10% FBS was added to both cell lines, and the lines that emerged were maintained in medium containing serum. Cell Culture HPVl 6-immortalized keratinocytes (HPK) were maintained in DMEM:F12 (3: 1) supplemented with 18.2 p,g/mL adenine, 1O-”M cholera toxin, 0.4 pg/mL hydrocortisone, 5 pg/mL insulin, 10 ng/mL epidermal growth factor, and 10% FBS , and were passaged at a ratio of 1 :2 when nearly confluent. PK16 and P K I 8 cells were maintained in complete serum-free KGM. All carcinoma lines were maintained in DMEM containing 10% FBS. To determine the effects of protein-synthesis inhibitors on HPV expression, cycloheximide (Sigma Chemical Co., St. Louis, MO) was added to subconfluent HPK IA cultures at a concentration of 50 pgimL, a concentration that inhibited protein synthesis by approximately95% but was not cytotoxic to HPK IA cells within the time frame of the experiments. Growth Studies Cells were plated at a density of 1-5 x 1O5 cells/35-mm Petri dish and incubated overnight in growth factor-supplemented, serum-free KGM or in serum-containing medium as indicated in the figure legends. Fresh medium containing varying concentrations of TGF-61 or TGF-p2 (R & D Systems, lnc., Minneapolis, MN) or TGF-p3 (M. Sporn, National Cancer Institute) was added for 48 h. Five microcuries of [3Hlthymidine (New England Nuclear, Boston, MA) was
added to each dish for the last 24 h, and cultures were then fixed in methanol and acetic acid and processed for autoradiography by standard techniques. The percentage of labeled nuclei was determined by counting at least 500 cells in four or five low-power fields, and results were expressed either as percentage of control (for cultures grown in the absence of TGF-p) or as percentage labeled nuclei. To study the effects of TGF-p in serum-selected PK16 and PK18 cell lines, cells were maintained as above in KGM containing 10% FBS, but growth studies were performed in serum-free, growth factor-supplemented KGM containing TGF-p. Modified raft cultures were established by the method of Asselineau et al. [291, except that cultures were incubated with TGF-p and labeled with [3Hlthymidinewhile submerged. Northern Analysis Total cellular RNA was extracted from tissue-culture cells or 4-mm biopsies essentially as described by Chirgwin et al. [301 using guanidine thiocyanate. For northern analysis, 15-20 p g of total cellular RNA (as indicated in the figure legends) was fractionated on 1.1% agarose gels containing 0.66 M formaldehyde, transferred to nitrocellulose filters, and hybridizedwith nick-translated, 32P-labeled DNA probes. Hybridization was performed with the following probes: for HPV16 and HPVl8, the full-length genomic viral DNAs [27,3 1I; for glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a 1.2-kb insert of the plasmid pRLC GAP, which contains a full-length cDNA for rat GAPDH [321; for TGF-a, a 0.9-kb fragment of the human TGF-a cDNA; and for TGF-PI, a 1.2-kb Bgll fragment of human TGF-p1 cDNA. Filters were prehybridized for 4-6 h and hybridized at 42°C for 72 h as previously described . To determine if equal amounts of cellular RNA were loaded and transferred, gels were stained with ethidium bromide, and filters were hybridized with the housekeeping gene GAPDH. Nuclear Run-on Assays Isolation of nuclei and nuclear run-on assays were performed essentially as previously described . HPK IA cells were grown to semiconfluence in 150-mm Petri dishes. At the beginning of the experiment, fresh serum-free, growth factor-supplemented KGM containing 10 ng/mL TGF-p1 was added for the times indicated in the figure legends. Nucleifrom approximately 1 x 107cellswere used for each experiment. Five micrograms of denatured plasmid DNA containing 0.4 p g of DNA insert was applied to nitrocellulose filters and hybridized for 3 d at 42°C with 5 x 1O6 cpm trichloroaceticacid-pre~ipitable~~P-labeled RNA. After hybridization, filters were washed five times at 62°C for 20 min in 0.1 x standard saline citrate and 0.1 YOsodium dodecyl sulfate, then exposed to Kodak XAR-2 film. The following double-stranded DNA probes were used: for c-jun, pHJ, which contains the long open reading frame of the human c-jun gene ; for c-myc, a 1.8-kb fragment of human c-myc DNA (Oncor Inc., Gaithersburg, MD); for HPV16, pHPV16, which contains thefull-length HPV16
genome ; for TGF-p1, phTGF-P2, which contains a 2.1-kb human TGF-p1 cDNA; and forfibronectin, a pGEM plasmid containing a 500-bp fibronectin cDNA insert. RESULTS
Effects of TGF-p1 on Growth and Virus mRNA Expression in HPVI 6-Immortalized Keratinocytes We have previously shown that in monolayer culture, the growth of neonatal foreskin keratinocytes immortalized by HPV16 DNA is inhibited by TGF-p1 . To determine whether sensitivity to TGF-PI -mediated growth inhibition is a consistent characteristic of keratinocyteswith transcriptionally active HPVI 6 sequences, we established several cell lines by transfecting early-passage keratinocytes with full-length HPV16 DNA cloned into plasmid vectors as described in Materials and Methods. These cell lines, designated PK16, have been maintained in serum-free KGM supplemented with growth factors for over 30 passages. Within approximately 10-1 2 passages, the cultures were mainly comprised of small cuboidal epithelial cells that retained the ability to differentiate morphologically in the presence of serum. The response of one cell line, PKl6B, to TGF-p1 is shown in Figure 1A. Similar to what we have previously reported in HPK cells I21, the growth of PK16B cells was strongly inhibited by TGF-pl. To determine whether growth inhibition is accompanied by a decrease in virus expression in HPV16-immortalized cells, we examined the effects of TGF-p1 on HPVl6 mRNAs in PK16B cells by northern blot hybridization using fulllength HPV16 DNA as a probe (Figure 1 B). Hybridization with this probe detects multiple HPV16 mRNA species transcribed from the E6 and E7 region of the HPV16 genome that are similar to those detected in cultured cervical carcinoma cells . High levels of HPV16 mRNAs were detected in the PK16B line. By 24 h after treatment with 10 n g h L TGF-p1, a marked decrease in steady-state levels of HPV16 mRNA was observed, indicating that the growth-suppressiveeffects of TGF-p1 were associated with inhibition of virus-specific mRNAs. A summary of the effects of TGF-p1 on HPVl6-positive keratinocytes derived from neonatal foreskin is presented in Table 1. Comparison of t h e Proliferative Response o f Normal and HPV16-Immortalized Keratinocytes to Different lsoforms of TGF-p Since each of the different HPVl6-immortalized cell lines exhibited similar responses to TGF-PI, we selected the best characterized of the HPV16-transformedcell lines, the HPK IA line, for further study. To determine whether HPK cells exhibited a differential response to three closely related members of the TGF-p family, we compared the effects of TGF-p1, TGF-p2, and TGF-p3 on DNA synthesis in HPVl6transformed HPK IA cells in passage 44. This cell line, established by Durst et al. 141 by transfecting full-length HPVl6 into neonatal foreskin keratinocytes, is immortalized but not tumorigenic at early passages. The results of a representative experiment are shown in Figure 2. DNA synthesis was maximally inhibited in HPK
TGF-p REGULATION OF HPV-TRANSFORMEDKERATINOCYTES
Table 1. Summary of Effects of TGF-p1 on HPV-Positive Keratinocytes
Cell line HPV16-positive cells PK16A PK16B PKI 6C PKI 6D HPK IA HPK II HPK Ill HPVI 8-positive cells PK18A PK18B PK18C PK18D PK18H
Percent inhibition bv TGF-B1* DNA HPV mRNA synthesis expression' 60 94 82 69 82 60 77
87 95 84 89
64 76 0 79
29 92 82 77
*Cells treated with 10 ngknL TGF-P1 for 48 h (to examine DNA syn thesis) or 24 h (to examine RNA expression) Data is presented as percent inhibition of TGF-p-treated cells versus untreated cells 'Based on densitometric scanning of the most abundant transcript In some cell lines, several transcripts were scanned
Figure 1. Effects of TGF-01 on PK16 cells. (A) Subconfluent cultures o f PKl6B cells were exposed t o 10 ng/mLTGF-Pl for 48 h in complete KGM. DNA synthesis was measured by labeling cells for the last 24 h with 5 WCi of (3H]thymidine, followed by autoradiography. A t least 500 cells were counted for each point, and the results are expressed as percent labeled nuclei in untreated(-)versustreated(+)cuItures.(B)Fifteen micrograms of total RNA was extracted from control cells ( - ) a n d from cells treated with 10 ng/mLTGF-PI ( + ) f o r 24 h and hybridized with full-length 32P-labeled HPV16 DNA. The bars in the right margin indicate multiple HPV16-encoded transcripts. The arrow in the left margin indicates the position o f 18s ribosomal RNA transcripts.
IA cells by 10 ng/mL TGF-p1, TGF-p2, or TGF-p3, decreasing to 15-19% of the control value (Figure 2A). At lower concentrations, we consistently found that TGF-p3 inhibited HPK IA cells less than either TGF-p1 or TGF-P2. To examine the specificity of TGF-p-mediated inhibitory effects on normal and HPV16-transformed keratinocytes, we compared the response of HPK IA cells to TGF-p1 and TGF-p2 with that of ethionine, which is also a reversible growth inhibitor of keratinocytes . Normal keratinocytesexhibited a similar pattern of growth inhibition in all three agents (Figure 28). In contrast, HPK IA cells were less responsive to ethionine than to either TGF-p1 or TGF-p2. Effects o f TGF-ps on Proliferation o f HPK IA Cells Grown o n Collagen-Fibroblast Lattices Dissociation of cells from their normal tissue microenvironment for growth in culture disrupts cell-cell and cellmatrix interactions, triggering changes in cellular function.
To analyze the effects of TGF-p under more complex culture conditions, we utilized a modification of the raft system developed by Asselineau et al. (291in which epithelial cells are grown on the surface of a collagen-fibroblast lattice. For our experiments, keratinocytes were not raised to the air-liquid interface in order to retain a monolayer that could be uniformly labeled and coated with emulsion for autoradiography. Proliferating HPK IA cells were grown on collagenfibroblast lattices as described in Materials and Methods and exposed to TGF-PI, TGF-p2, and TGF-p3 for 48 h. For the last 24 h, the cells were labeled with [3H]thymidine, after which the entire culture, including the collagenfibroblast lattice, was fixed, dried, and processedfor autoradiography. As shown in Figure 3A, HPK IA cells grown under these conditions actively proliferate with about 65 cells/cm2incorporating label. Treatment of the cultures with 10 n9hLTGF-p led to 94%, 88%, and 71 YO decreases in labeled nuclei for TGF-pl, TGF-p2, and TGF-p3, respectively (Figure 3B-D). These results indicate that the proliferative response of HPVl6-immortalized cells was inhibited by TGF-p, even under relatively physiologicalculture conditions. Regulation of HPV16 Expression by TGF-p Woodworth et al.  have previously reported that growth inhibition of HPVI 6-immortalized cervical epithelial cells by TGF-p1 required protein synthesis and occurred at the level of transcription. However, transformation of foreskin keratinocytes by HPV16 in vivo is rare compared with transformation of cervical epithelium. To determine if the mechanisms of TGF-p action were similar in foreskinand cervix-derived keratinocytes, we examined the time course of TGF-p1 effects on HPVI 6 expression, the effects of protein-synthesis inhibitors on TGF-PI-mediated inhi-
HPK 1A Cells o
TGF-beta1 TGF-bet& TGF-beta3
3 A 0
HPK 1 A
Figure 2. Comparison of the effects of different isoforms of TGF-p and ethionine on the growth of normal (NK) and HPV16positive (HPK IA) keratinocytes. (A) Cells were plated in 35-mrn wells at a density of 1 x lo5 cells/well in complete KGM. One day later. plating medium was removed and replaced with medium containing 1-10 ng/mLTGF-pl, TGF-p2, or TGF-p3. The
cultures were then processed for autoradiography as described in the legend to Figure 1. The results are expressed as percent inhibition of TGF-p-treated cultures relative to untreated controls. (B) Cells were processed as described in (A), except that a single concentration of TGF-p1 and TGF-p2 (10 ng/rnL) was compared with 8 rnM ethionine:
bition of HPVI 6 expression, and the transcriptional effects of TGF-p1 in HPK IA cells.
Requirement for protein synthesis. We next examined the effect of the protein-synthesis inhibitor cycloheximide on HPV expression (Figure 46). Exponentially growing cultures of HPK cells were pretreated for 15 min with 50 kg/mL cycloheximide. TGF-p1 (10 ng/mL) was then added to the cycloheximide-containingmedium and the cultures incubated for an additional 1 or 4 h, a t which point RNA was extracted. We found that the inhibitory effects of TGF-pl on HPVI 6 mRNA expression were blocked by cycioheximide at both 1 and 4 h after addition of TGF-PI. HPV mRNA expression was 42% of control levels after 1 h of exposure to TGF-p1 and 9% of control after 4 h of continuous treatment (Figure 48). On the other hand, in the presence of both cycloheximide and [email protected]
, HPV16 expression decreased only partially, reaching 72% of control at 1 h and 63% of control a t 4 h. These results indicate that protein synthesis was required for maximal inhibition of HPV16 expression by TGF-PI. In addition, the slight superinduction of HPVl6 mRNA transcripts in the presence of cycloheximide alone suggests that steady-state levels
Timecourse of TGF-pI-mediated inhibition of HPV16. We first examined the kinetics of the decrease in HPV gene expression in response toTGF-pI in HPK IA cells (Figure 4A). By 3 h after the addition of 10 ng/mL TGF-PI, steady-state levels of HPV16 mRNAs were reduced sixfold relative to untreated cultures and remained low for the duration of treatment (48 h). TGF-PI is not cytotoxic to keratinocytes, and this suppression of HPVI 6 mRNA expression did not indicate a generalized impairment of RNA synthesis, since by 24 h after treatment with TGF-PI, a fivefold increase in fibronectin mRNA transcripts was noted (data not shown). No marked changes in the expression of TGF-a transcripts in HPK IA cells were observed in response to TGFpl until 24 h after exposure t o the growth inhibitor, when the steady-state levels of TGF-a increased approximately twofold.
TGF-p REGULATION OF HPV-TRANSFORMEDKERATlNOCYTES
Figure 3. Effects of TGF-p on HPK IA cells grown in modified organotypic culture. HPK IA cells were grown in DMEM/F12 with 10% FBS to semiconfluence on submerged collagen-3T3 rafts. The medium was then aspirated and replaced with fresh medium containing (A) no TGF-p. (B) 10 ng/rnL TGF-pl, (C) 10 nglrnL
TGF-p2, or (D) 10 ng/mL TGF-p3 for 48 h. Labeling with 5 pCi of [3H]thymidinewas performed for the last 24 h. The cultures were then fixed in forrnalin, dried overnight, coated with emulsion, and processed for autoradiography. Results were quantitated by counting labeled nucleilcrn’ in at least four low-power fields.
Figure 4. Regulation of gene expression by TGF-p1 in HPK IA cells. (A) Subconfluent cultures were grown in the absence ( - ) or presence ( + ) of 10 ng/mL TGF-pl for the time periods shown a t the top of the autoradiogram and analyzed by northern blot hybridizationfor the expression of HPV16, TGF-p1, and TGF-a mRNAs. Lane C contained RNA prepared from a confluent culture. (B) Rapidly proliferating cultures were exposed to lOng/mLTGF-pl inthepresence(+)orabsence(-)of cycloheximide (CHX) for 1 or 4 h. RNA was isolated from each culture, and 15 bg was fractionated on agarose/formaldehydegels and
analyzed for HPV16 expression. Control hybridization was performed with the GAPDH probe. ( C ) Nuclear run-on assay of subconfluent cultures of HPK IA cells exposed to 10 ng/mL TGF-p1 for 0, 1, 4, or 24 h using DNA probes specific for c-jun, c-myc, HPV16 (hpv 16). TGF-p1 (tgf-pl), fibronectin (fn), and pBR322 (pBr322) as a control. c-mycwas overexposed in this autoradiogram to detect transcription with TGF-p1 and fibronectin genes. Arrows in the left margins of (A) and (B) indicate the position of 185 ribosomal RNA transcripts.
of HPV mRNAs were under the control of negative regulatory proteins.
ing assays in vitro , the promoter for HPV18 is more active than the HPVIG promoter. To explore the cellular basis of the more aggressive in vivo behavior of HPV18infected lesions, we studied the effects of TGF-p1 on growth and HPVl8 expression in keratinocytes immortalized with HPV18 (Figure 5). Cell lines were established by transfecting secondary passages of neonatal keratinocytes with full-length HPV18 DNA and were maintained in KGM. We found that the growth of five of five PK18 cell lines was strongly inhibited by TGF-PI, with a responsiveness that paralleled that of PK16, HPK, and normal keratinocytes. The expression of HPV18 mRNAs was inhibited by TGF-(31 after 24 h of exposure to this peptide in four of five HPVl 8-transfected cell lines (Figure 5 and Table 1). Similarly, cervical epithelial cells transfected with a plasmid containing only the E6 and E7 open reading frames under the control of the HPV18 promoter were sensitive to TGF-PI (Braun L, Mikumo R, unpublished observations). As we have previously observed in HPVl6-positive lines (data not shown), the magnitude of the inhibitory response of PK18 cells was variable. It is interesting to note that under conditions in which DNA synthesis was suppressed, HPV18 expression in PK18C cells was not inhibited, even
TranscriptionaleffectrofTGF-P7. To determine whether the decrease in steady-state levels of HPV mRNAs by TGF-p1 was due to a decrease in the rate of HPV transcription, we performed nuclear run-on assays using nuclei isolated from TGF-PI-treated HPK cells for 1, 4, and 24 h. As shown in Figure 4C, TGF-p1 reduced HPV16 transcription in rapidly growing HPK IA cells by approximately sevenfold within 1 h after exposure. When shorter autoradiographic exposures were examined, c-myctranscriptionwas inhibited by TGF-PI, whereas c-jun was induced. No change in transcription of TGF-p or fibronectin was observed, although the signals for both of these genes were weak. Similar results were obtained in three separate experiments. Effects of TGF-p1 on HPV18-Positive Cell Lines Although many molecular epidemiological studies do not distinguish between infection with HPVI 6 and HPVI 8, there is some suggestion that cervical lesions infected with HPVI 8 progress more rapidly to malignancy than HPVI 6infected lesions (37,381. In addition, at least in transform-
TGF-/3 REGULATION O f HPV-TRANSFORMED UERATINOCYTES
lopes in vitro in response to either serum or the phorbol ester 12-0-tetradecanoylphorbol-13-acetate. These cells, designated PK16-ss and PK18-ss, have been maintained in serum-supplemented medium for several months. Both P K I 65s and PKI 8-ss cell lines were strongly inhibited by TGF-61, TGF-p2, and TGF-p3 in a dose-dependent manner (Figure 6). No major differences in the patterns of sensitivity to any of the isoforms of TGF-p were observed. Similarly, growth inhibition by TGF-p1 was accompanied by suppression of HPVI 6 and HPVI 8 mRNA transcripts (data not shown). These results indicate that loss of the potential to undergo squamous differentiation in vitro is not associated with altered sensitivity to TGF-p.
TGF-p mRNA Expression in HPV-Positive Cell Lines and Cervical Tumor Tissues To further evaluate the role of the TGF-p family of growth factors in HPV-associatedtransformation, we asked whether
Figure 5. Effects of TGF-pl on HPVl8 rnRNA expression in HPV18-positive cell lines. Subconfluent cultures were grown in the absence ( - ) or presence ( + ) of 10 ng/mLTGF-pl for 24 h. Northern hybridization was performed as described in the legend to Figure 1, except that the RNA was hybridized with fulllength HPVl8 DNA. Arrows indicate the position of the 285 and 185 ribosomal RNA transcripts. In 18A cells. an additional transcript of approximately 1.7 kb was visible after longer exposure of the autoradiographs.
when a short autoradiographic exposure was examined. In general, however, keratinocytes with transcriptionally active HPV18 genomes exhibited a response to this growth inhibitor similar to that of several keratinocyte lines expressing HPV16 (see Table 1). Relationship Between TGF-p Sensitivity and Keratinocyte Differentiation Cervical carcinogenesis is characterized by an altered program of squamous epithelial-cell differentiation. To explore the relationship between differentiation and HPV-associated transformation, we examined the effects of TGF-p1, TGF-p2, and TGF-63 on keratinocytes selected for resistance to serum-induced differentiation in vitro. Differentiationresistant cell lines were established by serial passage and continuous culture of PK16 and P K I 8 cells in growth medium containing 10% serum. Initially, after incubation with serum, most of the cells enlarged, became squamous in appearance, and shed cornified envelopes into the culture medium. However, within 1 mo, small colonies of densely packed, rapidly growing epithelial cells emerged that did not produce cornified enve-
TGF-beta1 TGF-beta2 TGF-beta3
TGF-beta1 TGF-beta2 TGF-beta3
TGF-beta (nglml) Figure 6. TGF-p growth inhibition of differentiation-resistant PK16 and PK18 cells. Subconfluent cultures of serum-selected (5s) lines (A) PK16 (PK16ss) and (B) PK18 (PK18ss) were exposed to varying concentrations of TGF-p1, TGF-p2, and TGF-p3 for 48 h and processed for autoradiography as described in the legend to Figure 2. The results are expressed as percent labeled nuclei in untreated versus TGF-P-treated cultures.
Table 2 . Summarv of HPV-Positive Cell Lines Cell line NK PK16+ PK18' HPK IA* HPK II* HPK Ill* NCE Cask;§ SiHa5 C4-I' c4-115 ME-1805 MS-75Is HeLas TC- 140"
normal foreskin HPV-transfectedforeskin keratinocytes HPV-transfectedforeskin keratinocytes HPV-transfected foreskin keratinocytes HPV-transfected foreskin kerati nocytes HPV-transfectedforeskin keratinocytes normal cervical epithelium cervical carcinoma cervical carcinoma cervical carcinoma cervical carcinoma cervical carcinoma cervical carcinoma cervical carcinoma cervical carcinoma
negative HPV16 HPV18 HPV16 HPV16 HPV16 negative HPVI 6 HPVl6 HPV18 HPV18 unknown HPV18 HPV18 HPV16
Tumorigenic potential* -
+ + + + + + + t
*As assayed in nude mice. -, not tumorigenic; +, tumorigenic. 'Cell lines establisheda5 describedin Materials and Methods. *As described in . §Obtained from American Type Culture Collection. 'ICell line established from a well-differentiated,HPVl6-positivecervical carcinoma (Braun L, Mikumo R, manuscript in preparation).
the expression of TGF-p1 mRNAs was altered in HPVpositive nontumorigenic and tumorigenic cell lines and in primary cervical tumor tissues relative to normal cultured cells or normal cervical tissue. A summary of the cell lines used in this work, their source, HPV type, and tumorigenic potential is presented in Table 2. Total RNA was extracted from rapidly proliferating cultures of normal and HPVpositive cells, and 15 Fg from each cell line was analyzed by northern blot hybridization using a specific probe for TGF-PI . High levels of TGF-p1 mRNA transcripts were found in cultured keratinocytes isolated from neonatal foreskin or adult cervix (Figure 7A). However, other than in the HPK 111 cell line, no consistent increase or decrease in steadystate levels of TGF-p1 mRNAs in either nontumorigenic or tumorigenic cell lines was observed after correction for the amount of RNA loaded on the gels. Two HPV18-positive, carcinoma-derived cell lines, HeLa and MS-751, expressed low or undetectable levels of TGF-p1 mRNAs; whereas two other HPV18-positive lines expressed amounts of TGF-p1 mRNA transcripts that were equivalent to those produced by normal cervical epithelial cells. Similarly, no consistent pattern of TGF-a mRNA expression was detected in HPVpositive cells. Only a slight increase in the levels of TGF-a mRNA transcripts above the levels expressed by normal keratinocytes was observed in two of five HPV-positive nontumorigenic cell lines (PK18 and HPK Ill). Of the eight cervical carcinoma-derived cell lines examined, sixexpressed lower levels of TGF-a mRNAs than normal cervical cells after normalization to GAPDH message levels. Although most of the carcinoma lines examined had been maintained in culture for many years, one HPVl6positive cell line, TC-140, was recently established in our laboratory from a well-differentiated cervical carcinoma. TC-140 cells expressed similar levels of TGF-pl transcripts
as normal cervical epithelial cells, whereas TGF-a mRNAs were slightly increased above normal (Figure 7A). From these results, we can conclude that, at least in monolayer culture, the presence of transcriptionally active HPV16 or HPV18 does not lead to a consistent change in the mRNA transcripts for either TGF-pl or TGF-a. The expression of cellular genes can differ significantly in vitro and in vivo. To determine whether the expression of either TGF-p1 or TGF-a was altered in cervical carcinomas, we examined the steady-state levels of both these genes in biopsies of four cervical carcinomas and two biopsies of normal cervical tissue (Figure 7B). Total RNA was extracted from biopsies histologically diagnosed as cervical carcinoma and analyzed for TGF-p1 and TGF-a mRNA expression by northern blot hybridization. As we observed in cultured cells, the steady-state levels of TGF-p1 and TGF-a transcripts were variable in normal cervix and in cervical tumor tissue, probably due to variable numbers of different cell types in the tissue samples. DISCUSSION TGF-p1, the best characterizedmember of the TGF-p family of polypeptide growth factors, has been shown to regulate the expression of a wide range of cellular genes, from those coding for proteins important in regulation of the cell cycle to those of extracellular matrix molecules. Moreover, a role for this protein in the regulation of the growth and differentiation of squamous epithelium is supported by recent work showing distinct immunohistochemical staining patterns for TGF-p1 in normal and psoriatic human skin . In this paper, we report that TGF-p1 and TGF-p2 regulated the expression of the €6 and E7 transforming genes of HPVI 6 and HPV18 in neonatal foreskin keratinocytes immortalized by either of these two virus types.
JGF-p REGULATiON OF HPV-TRANSFORMEDKERATINOCMES
Figure 7. Northern blot analysis of TGF-a and TGF-pl expression in (A) cell lines and (B)cervical carcinoma-derived tissues. Fifteen micrograms (for cell lines) or 20 pg (for cervical carcinomaderived tissues) of total cellular RNA was hybridizedsequentially with 32P-labeledcDNA probes for TGF-p1, TGF-a, and GAPDH. NCE, normal cervical epithelial cells; NK, normal keratinocytes;
NCX, normal cervical tissue; T-133, T-140, T-141, T-142. cervical tumor biopsies; T-CASKI, nude mouse tumor obtained by injection of Caski cells into nude mice. All other cell line designations are described in Table 2. Arrows indicate the positions of the major coding transcripts for TGF-p1, TGF-a, and GAPDH.
We found that the growth of HPVl6- or HPVl8-immortalized keratinocytes was consistently inhibited by [email protected]
and TGF-p2, even in cells that have been selected after multiple passages for unresponsiveness to in vitro inducers of squamous differentiation. In most cell lines examined, suppression of HPV mRNA transcripts closely paralleled growth inhibition. In addition, TGF-p1 induced its own message, as has previously been shown in fibroblasts  and cervical cells [241. Although HPVI 6-immortalized cells retained sensitivity to TGF-PI, TGF-p2, and [email protected]
,they acquired partial resistance to ethionineinduced growth inhibition, indicating that immortalization is accompanied by loss of sensitivity to some, but not all, inhibitors of epithelial cell growth. Furthermore, an important finding of this paper is that TGF-p inhibited the growth of HPVl6-positive cells grown on collagen-fibroblast lattices as well as in monolayer culture. Inhibition of HPVI 6 mRNA expression by TGF-p1 was rapid, occurred at the level of transcription, and was partially reversed by cycloheximide. These results imply that a protein factor mediates inhibition, similar to that observed with 5-azacytidine’ a chemical inhibitor of HPV18 transcription (411. Despite the fact that HPVl8 is more active in some biological assays, our work indicates that, at least in relatively early passages, cell lines with transcriptionally active HPVI 8 sequences did not differ dramatically from HPVI 6-immortalized keratinocytes in their response to TGF-p1 . With the exception of P K I 8C cells, growth responsivenessto TGFp-mediated inhibition was always associated with down-
regulation of HPVI 8 transcripts. Whether PK18C cells lose sensitivity to TGF-p1 more rapidly than other HPVI 6- or HPVI 8-positive cell lines is under investigation. However, our findings, together with those of Woodworth et al. , who used HPV-immortalized cervical epithelial cell lines and different plasmid vectors, indicate that the presence of transcriptionally active HPVl6 or HPVl8 sequences alone does not confer resistance to TGF-PI-, TGF-p2-, or TGF-p3mediated growth inhibition. These results differ from a recent study by Pietenpol et al.  in which it was reported that keratinocyte cell lines containing HPV16 or HPV18 were resistant to TGF-pl. The cells used by these authors were clonal lines originally established by transfection of keratinocytes with HPV, followed by plating at low density and growth in high-calcium, serum-supplementedmedium [421. Under these conditions, the majority of the keratinocytes differentiated, thereby eliminating differentiation-competent cells at an early passage. This may have allowed a minor, TGF-P-resistant subpopulation whose growth is normally suppressed by surrounding sensitive cells to proliferate and ultimately dominate the cultures. On the other hand, the [email protected]
cell lines used in our experiments were first propagated by continuous passage at high density either in high-calcium medium containing serum (HPK) (41 or in low-calcium, serum-free medium (PK16 and PK18). Both conditions support the growth of nontumorigenic cells that can differentiate in vitro (PK) or in vivo (HPK) . Thus, the discrepancy
between our results and those of Pietenpol et al. [251 may reflect the selection of HPV-transformed cells with different in vivo differentiation potential due to variation in plating density rather than exposure to serum and high calcium. Since abnormal differentiation is an important phenotype of cervical cancer cells, this data supports the view that selection of cells with partial or complete loss of sensitivity to TGF-p may be important at some stage of HPVassociated carcinogenesis. When the epithelial component of human foreskin is separated from the dermis, TGF-a mRNAs are present at low levels only in epidermal cells, whereas high levels of TGF-p1 mRNAs are found in both the epidermal and dermal cell fractions (data not shown). In this study, we did not observe any consistent change in the patterns of expression of TGF-p1 mRNAs in HPV-positivenontumorigenic cells, in carcinoma-derived cell lines, or in biopsies of cervical carcinoma tissues relative to normal cells. Variation in the levels of TGF-PI and TGF-a in normal and tumor tissue is most likely due to differences in the cellular composition of the biopsied material. Since tumor tissue contains a higher proportion of epithelial cells than normal cervix does, our inability to detect striking differences in expression of either TGF-p1 or TGF-a in the tumors relative to normal tissue suggests that cervical carcinogenesis may not be associated with a change in expression of either TGF-p1 or TGF-a. Further studies are necessary to determine if there are differences in the levels of active TGF-p. In situ hybridization or immunohistochemical analysis of individual cells in tissue sections will be required to establish whether there are, in cervical cancers, localized changes in expression of either of these genes that might reflect the differentiation state of the tumor. Indeed, in the chicken chorioallantoic membrane the response of cells to TGF-p1 appears to be concentration-dependent, with dominant effects on cell migration at low concentrations and antiproliferative effects at higher TGF-p1 concentrations [I51. It is possible that localized changes in the synthesis or activation of TGF-p1, either by epithelial cells or by cells of the immune system, may contribute to spontaneous regression of low-grade intraepithelial neoplasias of the female genital tract. This view is supported by the fact that while HPV genomes are detected in normal epithelium adjacent to papillomas in the respiratory tract and adjacent to wartlike lesions on the cervix, viral RNA expression is uniformly confined to morphologically abnormal tissue (Shah KV, personal communication). As the long latency period between initial papillomavirus infection and the development of genital-tract cancer suggests, other factors in addition to HPV infection are required for tumorigenicity. Infected cells may vary in their sensitivity to TGF-p, depending on whether they have undergone additional virally induced or other genetic changes. In earlier studies, we found that cell lines derived from HPVI 6or HPVI 8-positive cervical tumors were refractory to TGFp l -induced growth inhibition [2,22,23]. Since HPK cells, as well as the PKI 6 and PKI 8 lines described in this report, display patterns of sensitivity to TGF-p1 with respect to growth and cellular gene expression similar to those of normal
keratinocytes, it is likely that these cells are at relatively early stages of transformation. Indeed, HPK cells are quite resistant to malignant transformation with chemical carcinogens (Durst M, unpublished observations) and maintain the potential to differentiate when transplanted into nude mice . The data is consistent with the idea that early in the course of infection with either high-risk HPV type, infected cells are responsive to endogenously produced growth-regulatory proteins. It has recently been reported that human immunodeficiency virus expression is suppressed by TGF-p1 in chronically infected U1 cells . Since the late regions of the HPV16 and HPV18 genomes are not transcribed in vitro, we cannot evaluate the full extent of the effects of TGF-p on papillomavirus infection. Moreover, an additional problem in extrapolating the effects of TGF-p in these culture systems to HPV-induced precursor lesions is that HPV DNA is integrated in the HPV-positive cell lines in culture but is episomal in low-grade intraepithelial neoplasia. However, an HPVl6-positive cell line established from a cervical intraepithelial neoplasia I lesion in which the viral DNA is retained in an episomal form is sensitive to TGF-p1 (Stanley M, personal communication). Thus, these results, together with our previous work I221,suggest that TGF-p1 may mediate regression of preneoplastic HPV-induced lesions in the early stages of infection and that loss of responsiveness to TGF-p, perhaps in conjunction with localized alterations in TGF-p production, may be important in the progression of HPV-associated neoplasia. ACKNOWLEDGMENTS We thank Dr. Michael Sporn for providing purified TGF-p3; Dr. G. Bell for providing the cDNA probes for TGF-a and TGF-p1; Drs. J. Tso, X.-H. Sun, and Ray Wu for their generous gift of the GAPDH probe; Drs. D. Bohman and R. Tjian for the c-jun plasmid; Dr. M . Rojkind for providing the fibronectin probe; and Dr. Mark Stoler for HPVI 6 and HPV18 pGEM plasmids. We also thank Carol White for her help in preparing this manuscript and Mark Fitzgerald for his help with the run-on transcription assays. This work was supported by grants CA-46617 from the National Cancer Institute (to L.B.) and DU 162/1-I from the Deutsche Forschungsgemeinschaft(to M.D.). Received February 21, 1992, revised March 31, 1992, accepted April 1 , 1992
REFERENCES 1 zur Hausen H Human papillomaviruses in the pathogenesis of anogenital cancer Virology 184 9-13, 1991 2 Braun L, Durst M, Mikumo R, Gruoouso P Differential remonse of nontumorigenic and tumorigeic human papillomavirds type 16-positive epithelial cells to transforming growth factor p l . Cancer Res 50:7324-7332. 1990. 3. Pirisi L, Yasumoto S, Felier M, Doniger 1, DiPaolo JA. Transformation of human fibroblasts and keratinocytes with human papillomavirus type 16 DNA. J Virol61: 1061-1 066, 1987. 4. Durst M, Dzarlieva-Petrusevska RT, Boukamp P. Fusenig NE, Gissmann L. Molecular and cyto enetic analysis of immortalized human primary keratinocytes ottained after transfection with human papillomavirus type 16. Oncogene 1 :251-256, 1987. 5. Kaur P. McDougall JK. Characterization of primary human keratinocytes transformed by human papillomavirus type 18. J Viroi 62: 1917-1 924,1988.
TGF-p REGULATION OF HPV-TRANSFORMEDKERATINOCYTES 6 Woodworth CD. Bowden PE, Doniger J, et a1 Characterization of normal human exocewical epithelial cells immortalized in vitro by papillomavirus types 16 and 18 DNA Cancer Res 48 4620-4628, 1988. 7 Pecoraro G, Morqan D, Defend1V. Differential effects of human papillomavirus type 6, 16, and 18 DNAs on immortalization and transformation of human cervical epithelial cells. Proc Natl Acad Sci USA 86:563-567, 1989. 8. McCance DJ, Kopan R. Fuchs E, Laimins LA. Human papillomavirus type 16 alters human epithelial cell differentiation in vitro. Proc Natl AcadSci USA85:7169-7173,1988. 9. Hawley-Nelson P, Vousden KH, Hubbert NL, Lowy DR, Schiller JT. HPVl6 E6 and E7 proteins cooperate to immortalize human foreskin keratinocytes. EMBO J 813905-3910. 1989. 10 Munger K, Phelps WC, Bubb V, Howley PM, Schlegel R. The E6 and E7 genes of the human papillomavirus type 16 together are necessary and sufficient for transformation of primary human keratinocytes.J Virol63:4417-4421, 1989 11. Munger K, Werness BA, Dyson N, Phelps WC, Howley PM. Complex formation of human papillomavirus E7 proteins with the retinoblastomatumor suppressor gene product. EMBO J 8:40994105, 1989. 12. Werness BA. Levine AJ. Howlev PM. Association of human papillomavirus types 16 and 18 E6 proteins with p53 Science 248 76-79,1990 13 Dvson N. Howlev PM, Munqer K, Harlow E The human papilloma virus-16 E j oncoproteh is able to bind to the retinoblastoma gene product. Science 243:934-937, 1989. 14. Massague J. The transforming growth factor-p family. Annu Rev Cell Biol 6:597-641, 1990. 15. Moses HL, Yang EY, Pietenpol JA. TGF-P stimulation and inhibition of cell proliferation: New mechanistic insights. Cell 63:243247,1990. 16. Sporn MB, Roberts AB. The multifunctional nature of peptide growth factors. In: Sporn MB, Roberts AB, (eds), Handbook of Experimental Pharmacology: Peptide Growth Factors and Their Receptors, Vo195/1. Springer-Verlag,Heidelberg, 1990, pp. 3-1 5. 17. Pietenpol JA, Holt JT, Stein RW, Moses HL. Transforming growth factor B1 SuDoression of c-mvcaene transcriDtion: Role in inhibition of keratinocyte prolifehon. Proc Nati Acad Sci USA 87. 3758-3762, 1990. 18 Gruppuso PA, Mikumo R, Brautiqan DL, Braun L Growth arrest induced by transforming growth factor-beta 1 is accompanied by protein phosphatase activation in human keratinocytes J Biol Chem 266 344-3448.1991 19 Goustin AS, Leof EB, Shipley GD, Moses HL Growth factors and cancer Cancer Res 46 1015-1 029, 1986 20 Roberts AB, Thompson NL, Heine U, Flanders C, Sporn MB Transforming growth factor-p Possible roles in carcinogenesis Br J Cancer 57 594-600, 1988 21 Blomhoff HK Smeland E Mustafa AS Godal T Ohlsson R EosteinBarr virus mediates a switch in responsiveness to transforming growth factor type beta in cells of the B cell lineage. Eur J lmmunol 17:299-301. 1987. 22. Braun L, Lauchlan S, Mikumo R. Gomez M. Tumorigenic transformation of cervical epithelial cells is associated with loss of growth inhibition by TGF-beta. J Cell Biochem Suppl 13C:181, 1989. 23. Braun L, Lauchlan 5, Mikumo R, Gomez M. Inhibition of seruminduced DNA synthesis in normal but not tumorigenic cervical epithelial cells by transforming growth factor-pl. in: Howley PM, Broker TR (eds), Papillomaviruses, Voi 124, UCLA Symposium on Molecular and Cellular Biology. Wiley-Liss, New York, 1990, pp. 157-167. 24. Woodworth CD, Notario V, DiPaolo JA. Transforming growth factors beta 1 and 2 transcriptionally regulate human papillomavirus (HPV) type 16 early gene expression in HPV-immortalizedhuman genital epithelial cells. J Virol 64:4767-4775, 1990. 25. Pietenpol JA, Stein RW, Moran E, et al. TGF-p1 inhibition of c-myc
transcription and growth in keratinocytes is abrogated by viral transforming proteinswith pRB binding domains. Cell 12:2-10. 1990. 26. Hawley-NelsonP, Sullivan JE, Kuny M. Hennings H, Yuspa SH. Optimized conditions for the growth of human epidermal cells in culture. J Invest Dermatol75: 176-182. 1980. 27. Boshart M, Gissman L. lkenberg H. Kleinheinz A, Scheurlen W, zur Hausen H. A new type of papillomavirus DNA, its presence in genital cancer biopsies and in cell lines derived from cervical cancer. EM80 J 3: 1 1 51 -1 1 57, 1984. 28. Stoler MH, Broker TR. In situ hybridization detection of human papillomavirus DNA and messenger RNA in genital condylomas and a cervical carcinoma. Hum Pathol 17: 1250-1 258, 1986. 29. Asselineau D, Bernard B, Bailly C, Darmon M, Prunieras M. Human epidermis reconstructedby culture: Is it "normal." J Invest Dermatol 86.181-1 86, 1986. 30. Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ. Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 18:5294-5299, 1979. 31. Durst M, Gissmann L, lkenberg H, zur Hausen H. A papillomavirus DNA from a cervical carcinoma and its presence in cancer biopsy samples from different geographic regions. Proc Natl Acad Sci USA80:3812-3815, 1983. 32. Tso J, Sun K-H, KaoT-H, Reece KS, Wu R. Isolation and characterization of rat and human glyceraldehyde-3-phosphate dehydrogenase cDNAs: Genomic complexity and molecular evolution of thegene. Nucleic Acids Res 13:2485-2501. 1986 33. Matrisian LM, Glaichenhaus N, Gesnel M-C, Breathnach R. Epidermal growth factor and oncogenes induce the same cellular mRNA in rat fibroblasts. EMBO 14: 1435-1440, 1985. 34. Bohmann D, Bos JT, Admon A, Nishimura T, Vogt KT, Tjian R. Human protooncogene c-jun encodes a DNA binding protein with structural and functional properties of transcription factor AP-1. Science238:1386-1392. 1987 35. Yee C, Krishnan-Hewlett I, Baker CC. Schlegel R, Howley PM. Presence and expression of human papillomavirussequences in human cervical carcinoma cell lines. Am J Pathol 119:361-366, 1985. 36. Wilke MS, Edens M, Scott RE. Ability of normal human keratinocytes that grow in culture in serum-free medium to be derived from suprabasal cells. J Natl Cancer lnst 80: 1299-1304, 1988. 37. Shah KV, Howley PM. Papillomaviruses.In: Fields 8N. Knipe DM (eds), Virology, 2nd ed. Raven Press, New York, 1990, pp. 1651-1676. 38. Romancyuk H, Villa LL, Schlegel R, Howley PM. The viral transcriptional regulatory region upstream of the E6 and E7 gene is a major determinant of the differential immortalization activities of human papillomavirus types 16 and 18. J Virol65:2739-2744, 1991. 39. Kane CJM, Knapp AM, Mansbridge JN, Hanawalt PC. Transforming growth factor-p1 localization in normal and psoriatic epidermal keratinocytes in situ. J Cell Physiol 144: 144-1 50, 1990. 40. van Obberghen-Schilling E, Roche NS, Flanders KC, Sporn MB, Roberts AB. Transforming growth factor p l positively regulates its own expression in normal and transformed cells J BiorChem 263 7741-7746.1988 41 Rosl F, Durst M, zur Hausen H Selective suppression of human papillomavirus transcription in non-tumorigenic cells by 5-azacytidine EMBOJ7 1321-1328, 1988 42 Schlegel R, Phelps WC, Zhanq Y-L, Barbosa M Quantitative keratinocyte assay detects two biological activities of human papillomavirus DNA and identifiesviral types associated with cervicalcarcinorna. EMBOJ7:3181-3187, 1988. 43. Durst M. Bosch FX.Glitz D, Schneider A, zur Hausen H. Inverse relationship between human papillomavirus (HPV) type 16 early gene expression and cell differentiation in nude mouse epithelial cysts and tumors induced by HPV-positive cell lines. J Virol 65:796-804. 1991 44. Poli B, Kinter Al, Justement IS, Bressler P, Kehrl JH, Fauci AS. Transforming growth factor p suppresses human immunodeficiency virus expression and replication in infected cells of the monocytei macrophage lineage. J Exp Med 173.589-597, 1991