HeLa cells are phenotypically limiting in cyclin E/CDK2 for efficient human papillomavirus DNA replication.

Human papillomaviral (HPV) origin-containing plasmids replicate efficiently in human 293 cells or cell extracts in the presence of HPV origin-recognition protein E2 and replication initiation protein E1, whereas cervical carcinoma-derived, HPV-18-positive HeLa cells or cell extracts support HPV DNA replication poorly. We recently showed that HPV-11 E1 interacts with cyclin/cyclin-dependent kinase (cdk) complexes through an RXL motif and is a substrate for these kinases. E1 mutations in this motif or in candidate cdk phosphorylation sites are impaired in replication, suggesting a role for cdks in HPV replication. We now demonstrate that one limiting activity in HeLa cells is cyclin E/CDK2. Purified cyclin E/CDK2 or cyclin E/CDK3 complex, but not other cdks, partially complemented HeLa cell extracts. Cyclin E/CDK2 expression vectors also enhanced transient HPV replication in HeLa cells. HeLa cell-derived HPV-18 E1 protein is truncated at the carboxyl terminus but can associate with cyclin E/CDK2. This truncated E1 was replication-incompetent and inhibited cell-free HPV replication. These results indicate that HeLa cells are phenotypically limiting in cyclin E/CDK2 for efficient HPV replication, most likely due to sequestration by the endogenous, defective HPV-18 E1 protein. Further analyses of the regulation of HPV E1 and HPV replication by cyclin E may shed light on the roles of cyclin E/CDK2 in cellular DNA replication.

The initiation of DNA replication in higher eucaryotes requires a multiprotein origin recognition complex, an MCM 1 complex (for minichromosome maintenance) that unwinds the origin, DNA polymerase ␣/primase, and additional proteins (reviewed in Ref. 1). Because of this complexity, the functions of individual proteins in the preinitiation complex and their interactions and regulation have been difficult to investigate.
Small DNA viruses such as SV40/polyomavirus and papillomaviruses each encode one or two viral origin (ori) recognition and initiator proteins and rely heavily on the cellular DNA replication machinery. Because the interactions among viral and cellular proteins mimic those among the cellular counterparts, these viruses have proven to be particularly useful model systems. The SV40 system has aided in the identification and mechanistic description of several important replication components (reviewed in Ref. 2) and more recently has revealed the opposing effects of cyclin E/CDK2 and cyclin A/CDK2 in regulating DNA polymerase ␣/primase via direct phosphorylation (3). The fact that papillomaviruses are highly prevalent and medically important human and animal pathogens also makes them attractive to investigators (reviewed in Ref. 4).
The genome of papillomaviruses is an approximately 8-kilobase pair double-stranded circular DNA. Because the productive life cycle takes place only in keratinocytes undergoing terminal squamous differentiation (4), viral DNA replication is studied either in a cell-free system or in transiently transfected cells (reviewed in Refs. 5 and 6). The ori of all papillomaviruses consists of one binding site for the virus-encoded E1 protein and several highly conserved binding sites for the virus-encoded E2 protein. E2 is a critical origin recognition protein (7)(8)(9)(10)(11)(12)(13)(14)(15), and E1 is the initiator and a DNA helicase (16 -20). They are functionally equivalent to the cellular origin recognition complex and MCM complexes, respectively, during initiation. Replication is dependent on the cellular DNA replication machinery, including DNA polymerase ␣/primase, DNA polymerase ␦, replication protein A (RPA, the single-stranded DNAbinding protein), the proliferating cell nuclear antigen (the processivity clamp for DNA polymerase ␦), replication factor C (the clamp loader), and topoisomerases I and II (8,14). As a consequence of this high degree of conservation in cis sequence elements and trans-acting factors, E1 and E2 proteins from any one virus can replicate all papillomavirus origins (Refs. 21 and 22 and references therein).
E1 protein interacts directly with the p180 and p70 subunits of DNA polymerase ␣/primase (23)(24)(25)(26) and with RPA (27) and is required throughout the initiation and elongation stages, whereas E2 protein is needed only during initiation (28). E2 protein helps recruit E1 protein, preferentially from the homologous HPV genotype, to the origin (21,22,29). The E2 protein of bovine papillomavirus type 1 (BPV-1) interacts with a recombinant 70-kDa subunit of RPA (PRA70) in a far Western blot (30). Our recent studies showed that HPV-18 E2 protein associated with a His-tagged RPA70 in solution and enhanced its single-stranded DNA binding activity and that of RPA. 2 In cell-free replication, high concentrations of E1 can initiate rep-lication even in the absence of an ori. But at low E1 concentrations, E2 protein is absolutely necessary for origin-specific replication and is thought to play a role in the assembly of the preinitiation complex (8,14,28). With few exceptions, E2 is indispensable in vivo (13,21,22), presumably because it is additionally needed to prevent nucleosome condensation of the origin sequences (31) and to help direct E1 protein and ori DNA to the nuclear matrix, where replication occurs in subnuclear foci (32). 3 Additional cellular proteins also interact with the viral replication proteins and play major roles in regulating HPV DNA replication. For example, cellular transcription factor YY1 binds to the E2 protein and inhibits HPV ori replication (33). The human heat shock/chaperone protein Hsp70 and the cochaperone Hsp40 additively stimulate HPV-11 E1 binding to the ori (34). In particular, electron microscopic observations reveal that Hsp40 promotes efficient assembly of di-hexameric HPV-11 E1 complexes on the ori, as would be required for bidirectional replication. Di-hexameric BPV-1 E1 protein complexes have also been observed in the absence of added Hsp40 (20).
By using expression cloning, we previously identified a shortened HPV-18 E1 protein as one of several proteins in HeLa cells that interact with cyclin E/CDK2 (35). HeLa is a cervical carcinoma-derived cell line containing multiple copies of integrated HPV-18 DNA. We then demonstrated that the replication-competent, full-length HPV-11 E1 protein can interact with several different cyclin/cdk complexes, including cyclin E and cyclin A complexes, and these interactions require a conserved RXL motif located in the amino-terminal portion of the E1 protein (35). 4 This motif, first found in the p21 CIP1 class of cdk inhibitors (36), is now recognized as a primary cyclintargeting motif, and in the case of HPV-11 E1, mutation of this motif greatly reduces the efficiency of phosphorylation of E1 by cyclin E/CDK2. The E2 protein does not directly bind cyclin E, but it is also a substrate for phosphorylation when present in a quaternary complex containing the E1 protein. A functional role for at least one cyclin/cdk complex in HPV replication is indicated by the fact that E1 mutations in either the RXL motif or in serine or threonine residues that are consensus cdk substrates dramatically reduce replication efficiency (35). The BPV-1 E1 protein also binds to cyclin E of Xenopus laevis, and the Xenopus cyclin E/CDK2 increases BPV-1 ori replication by E1 in Xenopus egg extracts (37).
Cdks comprise a family of proteins that function at distinct phases of the cell cycle to catalyze cell cycle transitions. Following the activities of cyclin D1/CDK4 and CDK6, S phase entry requires cyclin E-dependent kinases. These kinases inactivate the retinoblastoma susceptibility protein, pRB, which negatively controls the E2F transcription pathway. Overexpression of cyclin E/CDK2 or cyclin E/CDK3 can mobilize quiescent fibroblasts (38 -40). Cyclin E/CDK2 complex also functions downstream of the pRB/E2F pathway (41,42) to activate chromosomal replication. For example, cellular DNA replication in Xenopus egg extracts requires an active CDK2, presumably in complex with cyclin E (43,44). The speculation is that cyclin E/CDK2 may directly regulate the formation or activity of the pre-replication complex, including certain MCM proteins (reviewed in Ref. 45). In contrast, cyclin A/CDK2 and cyclin A/CDC2 are needed for S phase progression and passage through the G 2 /M transition. However, it is currently unclear to what extent cyclin E and cyclin A play redundant or inde-pendent roles in DNA replication, in part because the primary targets of these kinases in the replication process are largely unknown.
We previously showed that HPV ori-containing plasmids replicate transiently but efficiently in a wide range of human and animal cell lines. In contrast, replication is very poor in a number of cervical carcinoma-derived cell lines that contain and express integrated HPV-16 DNA, such as CaSki or SiHa, or HPV-18 DNA, notably HeLa (21). This low activity cannot be entirely attributed to poor transfection, because extracts of HeLa cells also fail to support efficient HPV ori replication, whereas those of human 293 cells (human kidney epithelial cells transformed by adenovirus E1A and E1B) do so efficiently (8). Thus, HeLa cells may contain an inhibitor of HPV ori replication, or they are deficient in certain protein(s) critical for HPV DNA replication. In this study, we demonstrate that HeLa cell extracts are highly competent in supporting efficient SV40 ori replication in the presence of SV40 T antigen and that small amounts of 293 cell extracts complement HeLa cell extracts in initiating HPV ori replication. However, this activity, which is abundant in 293 cell extracts, does not biochemically fractionate in a manner consistent with any known replication proteins and has been most challenging to purify and identify. 5 HeLa cDNA sequences indicate that the amino-terminal portion of the HPV-18 E1 protein, including the putative cyclin E-binding motif RXL, is expressed (35,46,47). We now show that a cyclin E/CDK2 complex partially complements HeLa cells in vivo and in vitro. Conversely, a HeLa cell-derived HPV-18 E1 protein, which is truncated at the carboxyl terminus, inhibits HPV replication in 293 cell extracts and in HeLa cell extracts complemented with cyclin E/CDK2. These observations suggest that sequestration of cyclin E/CDK2 by the endogenous, truncated HPV-18 E1 protein in HeLa cells is at least partly responsible for inefficient HPV replication. By taking advantage of this property of HeLa cell extracts, we demonstrate that cyclin A/CDK2, cyclin A/CDC2, and cyclin B/CDC2 do not possess this complementation activity. The implications of these observations concerning viral and cellular DNA replication are discussed.

MATERIALS AND METHODS
DNA Plasmids and Proteins-The HPV-11 ori plasmid, pUC7874-99 (spanning nts 7874 -7933 and 1-99) and the SV40 ori plasmid, pSVori, have been described (8,28). HPV-11 E1 protein, which is tagged at the amino terminus with a glutamate-rich (EE) epitope (48), and native HPV-11 E2 protein were purified from Sf9 cells (8). HPV-11 E1 and E2 were also expressed as fusion proteins in Escherichia coli as follows. To generate pRSET-EE, synthetic oligonucleotides containing, from the 5Ј end, an NdeI linker with an ATG codon, a 27-nt coding sequence for the EE epitope (8), and a BamHI linker at the 3Ј end were cloned between NdeI and BamHI sites of the bacterial expression vector pRSET-A (Invitrogen, Carlsbad, CA) replacing the polyhistidine tag and the enterokinase site. The HPV-11 E1 open reading frame was excised from pBS-EE-11E1 (8) and inserted into pRSET-EE immediately downstream of the epitope to generate pRSET-EE-11E1. pRSET-11E2 was similarly generated except that the E2 protein was tagged at the amino terminus with a 8-amino acid epitope derived from pp65 (a phosphoprotein encoded by the UL83 gene of cytomegalovirus). HeLa-derived HPV-18 E1 cDNA clones (35) were sequenced with the Applied Biosystems Automated DNA Sequencer. An NcoI site at the initiation codon and the RRL 3 ARA mutation were introduced into the HeLa E1 cDNA using the Sculptor mutagenesis kit (Amersham Pharmacia Biotech). The 1.8-kilobase pair NcoI/HindIII fragment, which is long enough to contain the entire E1 open reading frame, was then subcloned into pRSET-EE. Bacterially expressed HPV-11 E1 and E2 and HeLa cellderived HPV-18 E1 fusion proteins were purified by immunoaffinity columns as described previously (8). Hybridomas were gifts of Gernot Walter (EE epitope) and William Britt (pp65 epitope). Bacterially ex- 3  pressed, replication-competent HPV-18 His-tagged E1 and glutathione S-transferase (GST)-E2 fusion proteins were purified as described (33). GST-cyclin E or GST-cyclin A in complex with CDK2, GST-cyclin A or cyclin B in complex with CDC2, and GST-cyclin E in complex with an inactive CDK2 (dn), which contains an Asp-145 3 Ala mutation (40), were purified using glutathione-Sepharose beads from lysates of insect Hi5 cells co-infected with recombinant baculoviruses expressing these proteins (49). SV40 T antigens used were purified from Sf9 cells infected with recombinant baculovirus, some of which was a gift of Dr. Teresa Wang. Protein concentrations were determined by the Bradford assay (Bio-Rad), and the purity was assessed by Coomassie Blue staining after SDS-polyacrylamide gel electrophoresis using known amounts of bovine serum albumin as standards. The HPV-11 E1, the HeLa E1, and the corresponding ARA mutations (Ref. 35 and this study) were also expressed by in vitro translation in the presence of Translabel (ICN Pharmaceuticals, Costa Mesa, CA). Binding of in vitro translated E1 proteins to GST-cyclin E/CDK2 was performed as described (35). The kinase activities were assayed as described (50). Cell Extracts and Cell-free Replication-Suspension cultures of 293 cells were grown in Joklik's modified minimal essential medium with 5% calf serum at 37°C in the absence of CO 2 . HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum at 37°C under 5% CO 2 . Extracts were prepared as described (8). The conditions for cell-free replication in the presence of 2.5 Ci of [␣-32 P]dCTP have been described (8,28). Briefly, the 25-l reaction mixture contained 40 ng of an HPV-11 or SV40 ori plasmid, and 100 or 120 g of cell extract supplemented with purified HPV E1 and E2 proteins or SV40 T antigen, as specified in the figure legends. Additional proteins present in the assays are described for each experiment. DNA replication products were electrophoretically separated in 0.8% agarose gels and documented by autoradiography or with a Phospho-rImager (Molecular Dynamics, Sunnyvale, CA). A Western blot of cell extracts was conducted with monoclonal antibodies against cyclin E and CDK2 (PharMingen, San Diego, CA) simultaneously.

Distinct Capacities of Human 293 and HeLa Cell Extracts to Support HPV-11 and SV40
Cell-free Replication-To examine whether the poor ability of the HeLa cell extracts to support HPV ori replication is due to an inferior quality of HeLa cell extracts relative to 293 cell extracts, we compared their abilities to initiate HPV-11 ori and SV40 ori replication in parallel experiments. In the absence of viral proteins, only a low level of repair synthesis took place, generating Forms I and II DNA (Fig. 1A, lanes 1, 3, 5, and 7). In the presence of SV40 T antigen, HeLa cell extracts supported SV40 ori replication very efficiently, producing slow migrating -form replication intermediates and fast migrating Form I (supercoil) and Form II (relaxed circle) products (lane 8). But these extracts failed to replicate an HPV-11 ori plasmid to any appreciable extent (lane 4). In contrast, 293 cell extracts supported both SV40 and HPV-11 cell-free replication efficiently in the presence of identical amounts of viral proteins as those used in HeLa cell extracts (lane 2 and 6). These experiments demonstrate that both HeLa and 293 cell extracts are fully competent to support ori-specific replication but that there are significant differences in their capacities to meet the requirements of the HPV and SV40 oris.
Inadequate Amounts of Essential Factors in HeLa Cell Extracts to Support HPV-11 Ori Replication-The inability of HeLa cell extracts to support the replication of the HPV-11 ori was not rescued by doubling the amounts of extracts (data not shown). To test why HeLa cell extracts are inefficient in supporting HPV ori replication, we performed a series of mixing experiments. As shown above, replication did not take place in the absence of the HPV replication proteins (Fig. 1B, lane 1). In the presence of HPV-11 E1 and E2 proteins, the ability of 293 cell extracts to support HPV-11 ori replication fell dramatically with decreasing amounts of extracts when the concentrations of essential cellular replication proteins fell below threshold levels (Fig. 1B, lanes 2-7). At 80 g, only weak replication was detected (Fig. 1B, lane 4). 120 g of HeLa cell extracts did not support HPV-11 ori replication (Fig. 1B, lane 8). Decreasing amounts of HeLa cell extracts and increasing amounts of 293 cell extracts were then mixed to a total of 120 g in reaction mixtures (Fig. 1B, lanes 9 -13). The addition of as little as 20 g of 293 cell extracts to 100 g of HeLa cell extracts restored replication to a level higher than that achieved by 60 g of 293 cell extracts alone (compare lanes 9 and 5). Increasing amounts of 293 cell extracts dramatically complemented HeLa cell extracts (lanes 10 -13). The relative activities were quantified in a PhosphorImager and are presented in Fig. 1C. These results indicated that HeLa cell extracts are deficient for certain host factor(s) necessary for efficient HPV ori replication. Con-  -7), 120 g HeLa extracts alone (lane 8), or of a mixture of both to a total of 120 g (lanes 9 -13) were used in the HPV-11 ori replication assay in the presence (lanes [2][3][4][5][6][7][8][9][10][11][12][13] or in the absence (lane 1) of E1 and E2 proteins. The amounts of other reagents are as follows: ori plasmid, 40 ng; 12 ng of E1 protein and 8 ng of E2 protein purified from Sf9 cells, or alternatively, 40 ng of SV40 T antigen, which was also purified from Sf9 cells. C, the total replication products (from B) were quantified by PhosphorImager, and the relative activities of reactions with decreasing amounts of 293 extracts alone (triangles) or mixtures of 293 and HeLa extracts (circles) are shown after the subtraction of the background in lane 1. versely, this missing activity is relatively abundant in 293 cell extracts.
Cyclin E/CDK2 Complements HeLa Cells in Supporting HPV Ori Replication-HeLa cells have been estimated to contain more than 10 copies of integrated HPV-18 DNA. In some of the HPV-18 cDNAs, codons for all but the 5 amino-terminal residues of the E1 protein are spliced out, whereas in others, E1 is truncated within a short span just upstream of nt 2500 toward the carboxyl terminus of the E1 coding region (nt 914 -2885) (46,47). Our previous expression screening of HeLa cDNAs with cyclin E/CDK2 revealed that mRNAs encoding at least a portion of the HPV-18 E1 protein, including the putative cyclin binding RXL motif, is abundantly expressed, as inferred from the very high frequency of HPV-18 E1 clones identified (35). Based on these observations, we speculated that HeLa cell extracts might be phenotypically deficient in one or more cdks for HPV ori replication because they are sequestered by the endogenous truncated HPV-18 E1 protein.
To test this possibility, we added purified GST-cyclin E/CDK2 to the HPV-11 ori replication reaction mixture containing HeLa extracts. As before, at this concentration of E1 protein, only low levels of repair synthesis were observed when unsupplemented HeLa cell extracts were used ( Fig. 2A, lanes 2  and 4 -6, and C, lane 1). Addition of nanogram amounts of GST-cyclin E/CDK2 complex conferred significant replicative activity to HeLa cell extracts, reaching 18 -25% that obtained in 293 cell extracts ( Fig. 2A, lanes 1-3, and C, lanes 1, 2 and 8), but repair synthesis was not stimulated (Fig. 2A, lane 7). In contrast, addition of comparable amounts of GST-cyclin E/CDK2 to 293 cells had no detectable effect on the highly efficient HPV-11 ori replication (data not shown).
In the above experiments, the fold stimulation achieved by cyclin E/CDK2 addition cannot be determined for lack of activities above the background in the unsupplemented extracts. When a clearly detectable but low level of replication activity was observed in the presence of elevated concentrations of E1 protein or a more active E1 protein preparation, the inclusion of 1 ng of GST-cyclin E/CDK2 enhanced the activity by approximately 4 -5.4-fold, to 44% of that observed in 293 cell extracts (Fig. 2B, compare lanes 1-3, see also Fig. 6A, lanes 5 and 6, and data not shown). In parallel experiments, we added GST-cyclin E in complex with a dominant negative CDK2 (dn), which has an Asp-145 to alanine mutation (40). Up to 2 ng of this complex failed to enhance the replication activity (Fig. 2B, compare lanes 4 -6 to lane 2). Rather, a slight reduction was observed. Identical results were obtained using preparations of HPV-11 E1 protein isolated from Sf9 cells (data not shown). Cdk3 is the closest homolog of CDK2, and both can be activated by A-and E-type cyclins when co-expressed in insect cells, although the relevant cyclin partner and essential function of CDK3 are not known due to its low concentration in cells. Nevertheless, consistent with this close relationship with CDK2, we found that addition of GST-cyclin E/CDK3 also stimulated HPV ori replication in HeLa cell extracts, and the activity was comparable to that observed with GST-cyclin E/CDK2 (Fig. 2C, compare lanes  3-7 to lane 2). These data indicate that a cdk is required for HPV replication in vitro, and this activity is limiting in HeLa extracts.
Other Cyclin-dependent Kinases Do Not Complement HeLa Cell Extracts-Since HPV-11 E1 also binds to cyclins A and B through the same RXL motif and can also be phosphorylated by the associated kinases (35), 4 we examined whether these other cdks can also complement HeLa cell extracts in initiating HPV-11 ori replication. To compare activities, we titrated purified cdk complexes such that approximately equal levels of histone H1 kinase activity were observed over a 125-fold range of concentrations (Fig. 3A). As before, GST-cyclin E/CDK2 activated HPV ori replication at 1 ng, and an activity was detected with as little as 0.02 ng (Fig. 3B, lanes 15-18). In contrast, replication activity was not observed upon addition of GST-cyclin A/CDK2, GST-cyclin A/CDC2, or GST-cyclin B/CDC2 over the entire range of comparable kinase activities; only minimal repair synthesis was observed (Fig. 3B, lanes  3-14). Addition of cyclin A/CDK2 had no effect on the complementation by cyclin E/CDK2 (data not shown). These data indicate that HPV replication specifically requires cyclin E-associated kinases as opposed to a general requirement for a cdk activity. At a minimum, these other cdks are not limiting in HeLa cell extracts.
Co-expression of Cyclin E/CDK2 Increases HPV-11 Ori Replication in HeLa Cells-One prediction from the cell-free replication is that transient replication of an HPV ori-containing plasmid in HeLa cells should be enhanced if the intracellular concentrations of cyclin E/CDK2 are increased. Indeed, cotransfection of vectors expressing cyclin E and CDK2 together with HPV-11 E1 and E2 expression vectors and an HPV-11 ori plasmid enhanced replication by 2.5-fold (Fig. 4, compare lanes  1-4). Co-expression of cyclin E and the CDK2 (dn) mutation abolished replication (lanes 5 and 6). The inactive kinase may have sequestered E1 and E2 proteins into inactive complexes. In parallel with the cell-free results, the replication activity achieved in HeLa cells in the presence of increased cyclin E/CDK2 levels was not as high as is typically achieved with 293 cells (35). It is possible that co-transfection of 5 plasmids is less efficient than that with 3 plasmids. In addition, some other factor may also be limiting in HeLa cells, in agreement with the observation that BPV-1 ori replication requires an activity not required for SV40 ori replication (51). Nevertheless, this experiment clearly demonstrates the importance of cyclin E/CDK2 for HPV replication in vivo.
The Truncated HPV-18 E1 Cloned from HeLa Cells Binds Cyclin E-Restriction mapping and partial sequence analysis of a total of 44 E1 cDNAs derived from the cyclin E expression cloning indicated that all the cDNAs contained a significant E1 coding region spanning the RXL motif. All the published HPV-18 cDNA sequences from HeLa cells contain cellular sequences at the 3Ј end (46,47). Sequence analysis of three of our representative E1 cDNA revealed that they were identical to one another and that the E1 gene diverged from the prototype after nt 2497 and was followed immediately by a termination codon TAA. This message encodes an E1 peptide of 528 residues, as opposed to the prototype E1 of 657 residues. There were 4 amino acid alterations relative to the prototype HPV-18 E1 (T92K, I145N, S177C, and S185C), none of which affected the RXL cyclin-binding motif. There were also several additional third-base changes that did not alter the encoded amino acids. Since there is no consensus splice donor sequence where HeLa E1 diverges from the prototype HPV-18 E1, we suggest that the junction of viral host DNA of the integrated viral DNA template resides between nt 2497 and 2498. These clones are virtually identical to a previously published HeLa cDNA (47), including the downstream host sequence, which had no match to cellular DNA in the GenBank TM . The only exceptions are the above substitution mutations and some minor differences in the host sequences. These results suggest that the chimeric E1 cellular mRNAs are transcribed from the same copy of template out of many integrated HPV-18 DNAs and that a few mutations have occurred during passage of HeLa cells in different laboratories over the years.
Consistent with this sequence information, in vitro translation of HeLa-derived E1 cDNAs led to the production of a 68-kDa protein (Fig. 5, lane 7), a result that was confirmed upon expression and purification of an EE epitope-tagged 68-kDa E1 protein from bacteria (data not shown). For comparison, the wild type HPV-11 or HPV-18 E1 proteins are predicted to be 70 kDa, but when expressed in mammalian cells, in insect cells, in bacteria, or by in vitro translation, each migrates at an apparent size over 80 kDa (Fig. 5, lane 1) (8,21,33). Collectively, these cDNA data indicate that HeLa cells contain mRNAs with a capacity to encode a carboxyl terminus truncated HPV-18 E1 protein. This truncation occurs in the ATPase domain which also helps determine the specificity for E2 protein association (Ref. 22 and references therein) and would therefore not be expected to function as a replication initiator. Indeed, in the presence of HPV-18 E2, HeLa cell-derived E1 protein failed to support HPV ori replication in vitro or in vivo (data not shown). However, this truncated E1 protein retained the ability to interact with cyclin E/CDK2 in an RXL-dependent manner (Fig. 5, lanes 7-12), similar to the replication-competent HPV-11 E1 protein (Fig. 5, lanes 1-6) (35). These data are consistent with the proposal that the defective HPV-18 E1 protein may have sequestered cyclin E/CDK2, making HeLa cells phenotypically deficient and unable to replicate HPV origin containing DNA efficiently.
Exogenous Truncated HeLa HPV-18 E1 Protein Inhibits Cellfree HPV Ori Replication-Our hypothesis would predict that addition of the replication-incompetent HPV-18 E1 protein cloned from HeLa cells should function as a dominant negative inhibitor for HPV ori replication regardless of the HPV types from which the E1 and E2 proteins are derived. Recombinant HeLa cell-derived HPV-18 E1 protein was expressed in and purified from E. coli and added to HPV-11 ori replication reactions in HeLa cell extracts complemented with GST-cyclin E/CDK2 in the presence of matched pairs of HPV-11 or HPV-18 E1 and E2 proteins. We used two protein pairs to demonstrate the generality of the inhibition by the HeLa cell-encoded E1 protein. The truncated E1 inhibited replication effectively in a dose-dependent manner, reducing or eliminating the activities gained upon addition of cyclin E/CDK2 (Fig. 6A, compare lanes  1-4, lanes 5-8). A similar inhibition was also observed when the replication was conducted with 293 cell extracts (Fig. 6A,  compare lanes 9 -11 and 12-14). The small differences in the extent of inhibition in the presence of HPV-18 or HPV-11 E1 and E2 proteins might reflect the relative affinity of the two E1 proteins for cyclin E. DISCUSSION We have previously shown that HPV-11 E1 protein binds to several cyclins and that phosphorylation by one or more cyclin/ CDK complexes is essential for efficient HPV ori replication in transient and cell-free replication. However, the identity of the cyclin/CDK that functions during the initiation of replication was not known (35). In this study, by taking advantage of the observation that HeLa cells and cell extracts support HPV ori replication poorly (8) due to a deficiency in one or more novel factors (Fig. 1) and acting on an educated guess, we conducted experiments demonstrating that a functional cyclin E/CDK2 or cyclin E/CDK3 complex, but not other cyclin-dependent kinases, comprises a significant portion of the limiting components (Figs. 2-4). Both CDK2 and CDK3 complement a temperature-sensitive mutation in the CDC28 kinase in budding yeast and are implicated in the G 1 /S transition in mammalian cells, but the available evidence indicates that they are not functionally redundant (Refs. 38, 40, and 52 and references therein). These kinases inactivate the pRb growth control pathway and can also function downstream of this pathway to activate chromosomal DNA replication in the absence of pRb phosphorylation (39,42,49). One proposed role for the cyclin E/CDK2 complex is the phosphorylation of MCM proteins, but precisely how it regulates MCMs and perhaps additional targets during initiation of replication is unknown (45). Furthermore, it is not known whether cyclin A/CDK2 can substitute for cyclin E/CDK2 in these phosphorylation activities. However, the tight correlation between cyclin E and S phase entry would seem to suggest that the cyclin E/CDK2 complex, rather than cyclin A/CDK2, functions during initiation of DNA replication. The present study, together with our previous analyses of E1 mutations in potential phosphorylation sites and in the cyclinbinding site (35), demonstrates for the first time that the phosphorylation activity of cyclin E/CDK2 or cyclin E/CDK3 plays a unique and critical role in the initiation of DNA replication. Most importantly, this role cannot be fulfilled by an excess of other cdks.
The deficiency of cyclin E/CDK2 activity in HeLa cell extracts as far as HPV replication is concerned cannot be attributed to a low concentration of cyclin E, as a Western blot revealed similar amounts of cyclin E and CDK2 in HeLa and 293 cell extracts (Fig. 7). Rather, we suggest that HeLa cells are phenotypically limiting in cyclin E/CDK2 activity for HPV ori replication attributable to sequestration by the endogenous, replication-incompetent HPV-18 E1 protein. At present, the truncated HPV-18 E1/cyclin E/CDK2 complex in HeLa cells cannot be demonstrated directly for lack of an antibody to the HPV-18 E1 protein. Nevertheless, several observations support this interpretation. First, HeLa-derived HPV-18 E1 protein binds to cyclin E/CDK2 in vitro, as does the full-length HPV-11 E1 (Fig. 5), which also forms a complex with cyclin E/CDK2 in vivo in insect cells and mammalian cells (35). Second, the observation that a low level of replication was detected in unsupplemented HeLa cell extracts in the presence of higher concentrations or more active preparations of E1 (Fig. 2, data   FIG. 5. HeLa-derived HPV-18 E1 is truncated but retains the ability to associate with cyclin E/CDK2 in an RXL motif-dependent manner. 35 S-Labeled, in vitro translated HPV-11 and HeLa-derived HPV-18 E1 (with or without mutation of the RRL cyclin interaction motif to ARA) were incubated with immobilized GST-cyclin E/Cdk2 or GST as a control (1 h, 4°C), and washed beads were subjected to SDS-polyacrylamide gel electrophoresis and autoradiography. The amount of in vitro translated E1 proteins in control lanes (lanes 1, 4, 7, and 10) was 25% of the amount used in the binding experiment. Arrows point to HPV-11 E1 and HeLa cell-derived HPV-18 E1 proteins. Faster migrating bands are premature termination products generated during in vitro translation. not shown) is consistent with a competition between the exogenous E1 and the endogenous E1 for a limiting factor. Third, HPV ori replication in 293 or complemented HeLa cell extracts is inhibited by exogenous HeLa-derived E1 protein (Fig. 6A). However, we cannot exclude the possibility of additional mechanisms of inhibition by the HeLa E1 protein, such as a titration of host DNA replication proteins or the formation of non-functional, mixed oligomers with the full-length E1 protein, as the addition of the replication-competent, full-length HPV-18 E1 reduced replication by HPV-11 E1 and E2 proteins somewhat (data not shown). Be that as it may, none of these other modes of repression can account for the unique ability of cyclin E/CDK2 or cyclin E/CDK3, among the various cdks, to complement HeLa cell extracts.
One might ask why cellular and SV40 DNA replication were not affected by the endogenous HPV-18 E1 protein. Phosphorylation by cdks of threonine 124 of the T antigen of SV40 or polyomavirus is known to regulate its replication function. Although a role of cyclin E/CDK2 has not been examined directly in a replication reaction, this threonine residue is phosphorylated in vitro efficiently by cyclin A and cyclin B-dependent kinases but poorly by cyclin E/CDK2 (53-56). Thus, it is possible that T antigen is not regulated by cyclin E/CDK2 and hence insensitive to cyclin E/CDK2 sequestration by the defective HPV-18 E1 protein. Yet, the possibilities remain that cyclin E/CDK2 is also required for some aspect of SV40 replication, as is cellular DNA replication, but T antigen and cellular substrates are capable of effectively competing with the endogenous HPV-18 E1 in HeLa cells. We note that both 293 and HeLa cells harbor elevated cyclin E protein or mRNA relative to normal fibroblasts devoid of viral oncogenes (57,58) due to a transcriptional up-regulation by, respectively, the adenovirus E1A and HPV E7 protein. Both oncoproteins target the pRB/ E2F pathway (Ref. 4 and references therein). Alternatively, the truncated HPV-18 E1/cyclin E/cdk complex might be capable of phosphorylating substrates relevant for cellular and SV40 replication. Indeed, we have previously shown that the E1/cyclin E/CDK2 complex retains activity toward the exogenous substrate histone H1 and the associated substrate E2 (35). Thus, in principle, cyclin E may be capable of performing at least some of its functions even when associated with E1. Consistent with some of the hypotheses, addition of the HeLa-derived HPV-18 E1 protein also partially inhibited cell-free SV40 ori FIG. 6. The truncated E1 protein derived from a HeLa cDNA library is a dominant negative inhibitor of HPV ori replication and can also inhibit SV40 ori replication. A, effects of HeLa cell-derived HPV-18 E1 on HPV ori replication. Four sets of replication assays were conducted with HPV-11 ori, two sets each with 100 g of HeLa cell extracts (lanes 1-8) or 293 cell extracts (lanes 9 -14), and two sets each with bacterially expressed, replication competent, bacterially expressed HPV-11 E1 (15 ng) and E2 (8 ng) (lanes 1-4 and 9 -11) or replication competent HPV-18 E1 (30 ng) and E2 (10 ng) (lanes 5-8 and 12-14), as indicated (ϩ). The highest amount of GST-cyclin E/CDK2 (ϩ in lanes 2-4 and 6 -8) as that used in the experiments described in Fig. 3, 30 or 60 ng of the EE-tagged, truncated HeLa cell-derived HPV-18 E1 protein (lanes 3, 4, 7, 8, 10, 11, 13, and 14) were added to some reactions. Relative activities were determined by using a PhosphorImager. The highest activities in the each set of experiments achieved in the absence of the HeLa-derived E1 protein were set as 100% (lanes 2, 6, 9, and 12). B, effects of HeLa-derived E1 on SV40 ori replication. Each reaction mixture contained 40 ng of SV40 ori DNA, 200 ng of SV40 T antigen, and 100 g of HeLa cells extracts. To lanes 2-8 were added 30 or 60 ng of HeLa E1 protein as indicated. The highest amount of GST-cyclin E/CDK2, -cyclin A/CDK2, -cyclin A/CDC2, -cyclin B/CDC2, or -cyclin E/CDK2 (dn) as those used in Fig. 3 was added to lanes 4 -8. The SV40 T antigen used in this experiment was a different preparation from those used in Fig. 1, and 200 ng were required to elicit efficient replication. The relative activities were determined by a PhosphorImager .   FIG. 7. HeLa and 293 cell extracts contain comparable amounts of cyclin E and CDK2. A Western blot was conducted to reveal cyclin E (top arrowhead) and CDK2 (bottom arrowhead) simultaneously. Lanes 1 and 3 contained 10 or 8 g of 293 cell extracts. A 20% difference in cyclin E is clearly discernible. Lane 2 contained 10 g of HeLa cell extracts. There is no detectable difference in cyclin E to that in 10 g of 293 cell extracts shown in lane 1. replication in the presence of T antigen, and the activity was partially restored by the addition of cyclin E/CDK2, but not cyclin E/cdk (dn) (Fig. 6B, compare lanes 1-4 and lane 8). Interestingly, cyclin E/CDK2 and cyclin A/CDK2 had comparable activities (lanes 4 and 5). In contrast, neither cyclin A/CDC2 nor cyclin B/CDC2 had any effect (lanes 6 and 7). When replication was conducted in the presence of a much higher concentration of T antigen, no inhibition was observed (data not shown). These results indicate that the HeLa E1 the SV40 T antigen compete for various cdks and that not only cyclin A/CDK2 but also cyclin E/CDK2 may function in SV40 ori replication. Although it is possible that the exogenous cdks titrated HeLa E1 from the SV40 T antigen, this interpretation cannot explain the inactivity of cyclin E/cdk (dn), cyclin A/CDC2, or cyclin B/CDC2 that also bind to HPV-11 E1, and presumably HeLa E1, with similar affinity (35).
We note that several cyclin/cdk substrates are present in the preinitiation complex, including E1, E2 (35), and the p180 and p70 subunits of the DNA polymerase ␣/primase (3). In vitro RPA is a substrate of cyclin A/cdk, but not cyclin E/CDK2 because it binds to cyclin E poorly (59,60). However, since E1 appears to bind to cyclin E tightly (35,37) and these other cellular replication proteins interact with E1 or E2 proteins, the E1/E2 complex may serve as a docking platform for the kinase to phosphorylate both viral and host proteins, thereby promoting initiation of replication. A number of possible consequences of phosphorylation of these viral and host proteins remain to be tested. These include the activation of the E1 helicase, the transition from the preinitiation complex to a post-initiation complex by the release of the E2 protein (25,26,28), and the up-regulation of DNA polymerase ␣/primase activity as previously demonstrated during the SV40 cell-free replication (3). Our data do not address the basis for the unique activity of cyclin E/CDK2 among the cdks. This distinction conceivably may reside in a difference in the phosphorylation sites or the ability of these other kinases to phosphorylate the viral and host proteins when they are assembled into the preinitiation complex.
In summary, we have shown in this work that, among the cyclin-dependent kinases, cyclin E/CDK2 or cyclin E/CDK3 can uniquely activate HPV ori replication in HeLa cells that are phenotypically limited in cyclin E/CDK2 and do not support efficient HPV ori replication. Our data are consistent with an interpretation that this deficiency is most likely caused by the sequestration of these kinases by the endogenous replicationincompetent, truncated HPV-18 E1 protein that competes for cyclin E efficiently against exogenous E1 protein. We propose that the simplicity of the HPV replication system makes it an excellent experimental model for higher eucaryotic systems in deciphering, at the molecular level, the role of cyclin E/CDK2 in the initiation of DNA replication, linking this kinase uniquely to the G 1 /S transition.