Analyses of variant human papillomavirus type-16 E5 proteins for their ability to induce mitogenesis of murine fibroblasts

Background Human papillomavirus type 16 (HPV-16) E5 protein co-operates with epidermal growth factor to stimulate mitogenesis of murine fibroblasts. Currently, little is known about which viral amino acids are involved in this process. Using sequence variants of HPV-16 E5 we have investigated their effects upon E5 transcription, cell-cycling and cell-growth of murine fibroblasts. Results We demonstrate that: (i) introduction of Thr64 into the reference E5 sequence of HPV-16 abrogates mitogenic activity: both were poorly transcribed in NIH-3T3 cells; (ii) substitution of Leu44Val65 or, Thr37Leu44Val65 into the HPV-16 E5 reference backbone resulted in high transcription in NIH-3T3 cells, enhanced cell-cycle progression and high cell-growth; and, (iii) inclusion of Tyr8 into the Leu44Val65 backbone inhibited E5 induced cell-growth and repression of p21 expression, despite high transcription levels. Conclusion The effects of HPV-16 E5 variants upon mitosis help to explain why Leu44Val65 HPV-16 E5 variants are most prevalent in 'wild' pathogenic viral populations in the UK.


Background
A causal association between high-risk human papillomaviruses (HPV) infection -particularly HPV-16 -and cervical cancer has been established. HPV-16 E5 is a minor oncoprotein comprising of 83 amino acids and in silica predictions suggest it comprises of 3 anchor-like α-helices (residues 8-30, 37-52 and 58-76): with only the first being sufficient to span a lipid bilayer (Figure 1). A region within the second helix (residues 41-54) may be the binding site for the pore sub-unit of 16 KDa ATPase [1], though others claim it is located between residues 54-78 [2]. HPV-16 E5 is believed to act in the early stages of the oncogenic process [3][4][5][6] and is membrane-associated, occurring in the Golgi apparatus and endoplasmic reticulum [7].
HPV-16 E5 acts co-operatively with epidermal growth factor (EGF) to stimulate mitosis. Whilst E5 may, or may not, bind directly to the EGF-receptor (EGFr) [8,9], it was initially believed to impair acidification of endosomes via interaction with 16 KDa ATPase [10,11] and thereby promote recycling of EGFr to the cell-surface [10]. Others have suggested: that E5 perturbs EGFr trafficking from early to late endocytic structures rather than influencing acidification [12]; or, that E5 uses 16 KDa ATPase as a chaperone to enter the Golgi [2,13].
Whatever the initial processes, HPV-16 E5 stimulates c-ras, causing c-raf to attach to plasma membranes, activating enzyme cascades through the MEK and MAP kinases, which in turn migrate to the nucleus to phosphorylate cfos transcription factors [14][15][16][17]. Ultimately, this permits assembly of activator protein-1 heterodimers from c-fos and c-jun and stimulation of mitosis [18][19][20]. E5 can interdict this pathway via: (i) the induction of protein-kinase C which activates c-raf [21]; (ii) initiation of c-jun and c-fos and junB transcription [22,23]; and (iii), repression of p21 expression (a cyclin-dependant kinase inhibitor which causes pocket-protein phosphorylation, release of E2F and -via de novo synthesis of cyclins A, B and Emitosis [23]: Figure 2).
We have previously described HPV-16 E5 variants which encode novel E5 protein sequences [24]. Here we conjectured that these natural E5 protein variants may have dif-fering mitogenic properties and that the amino acids involved in this process might be discernable. This proposal was based upon evidence that certain HPV-16 E5 variants are more prevalent than others in wild viral populations in our locality. We detected marked differences in the ability of individual HPV-16 E5 variants to induce transcription in stably transfected long-term NIH-3T3 cell lines, changes in cell-cycle profiles and cell-growth. Some variants were more mitogenic than the reference isolate of HPV-16 E5: others which were poorly mitogenic as a result of either amino acid changes or, low transcriptional efficiencies. Interestingly, those HPV-16 E5 variants most frequently detected in our local population (RFLP pattern 2) -and most commonly associated with cervical lesions -were those which had the greatest mitogenic activity in vitro.  variants: Leu 44 Val 65 (AJ244882); Thr 37 Leu 44 Val 65 (AJ44863); Thr 64 (AJ244840) [24]; Tyr 8 Leu 44 Val 65 , (AJ24481); the HPV-16 reference E5 sequence [25,26] and HPV-6b E5 were amplified and cloned into pcDNA3.1Myc-His. For each HPV-16 E5 variant, a control construct, containing a TAA 'stop' codon (at codon position three) was also cloned into pcDNA3.1Myc-His.

HPV E5 constructs
All constructs and 'stop' controls had the correct DNA sequences after cloning (data not shown). T7 'run-off' transcripts were prepared and translated in cell-free wheat-germ expression assays spiked with 35

S-labelled
Points at which HPV-16 E5 affects the epidermal growth fac-tor signal transduction pathway Figure 2 Points at which HPV-16 E5 affects the epidermal growth factor signal transduction pathway. EGF: epidermal growth factor; EGFr: epidermal growth factor receptor; PKC: protein kinase C; AP-1: activator protein 1.

EGF/EGFr
Putative structure of HPV-16 E5 Figure 1 Putative structure of HPV-16 E5. Cartoon representation of the averaged results of multiple secondary structure predictions of reference sequence of HPV-16 E5 protein [25,26] (e.g. using programmes at http://pref.etfos.hr/split/: data not shown) showing the position of the predicted α-helices, the proposed voltage gating motif, as well as the amino-acid mutations of the natural variants studied.

Voltage gating motif [2]
Hydrophilic residue cysteine (all have 4 cysteines). Autoradiographs of polyacrylamide gel elecprophoresis (PAGE) gels revealed equivalent levels of in vitro translation for all HPV-16 variants, but no evidence of protein products from the equivalent 'stop' controls ( Figure 3A), confirming the fidelity of TAA ('stop') codons. However, different HPV-16 E5 variants were transcribed at different levels in stably-transfected (G418-selected) NIH-3T3 cell-lines, with Tyr 8 Leu 44 Val 65 being expressed at higher level than either Thr 37 Leu 44 Val 65 , or Leu 44 Val 65 , whilst the Thr 64 variant and reference sequence were transcribed at low level (Figure 3B). To assist data interpretation, cell-lines for all 'stop' constructs were pooled prior to subsequent experiments.

E5 variants with increased G 2 M-phase percentages exhibit increased cell-growth
Static analyses of cell-cycle profiles can be difficult to interpret as the percentage values are inter-dependent variables. Thus, different cell populations could have identical cell-cycle profiles, but vastly dissimilar growth rates [27,28]. This problem was addressed by determining cellgrowth at 24

Tyr 8 substitution of HPV-16 E5 is associated with high levels of p21 protein
HPV-16 E5 can repress p21 transcription [29]. We thus determined levels of p21 and cyclin B1 proteins in cells containing the HPV-16 reference E5 sequence as well as a variant with a high-growth rate (Leu 44 Val 65 ) and one with a low-growth rate (Tyr 8 Leu 44 Val 65 ). Cells containing Tyr 8 Leu 44 Val 65 maintained high levels of p21 protein throughout the time course, whereas p21 was much lower in cells containing the reference isolate and, near undetectable for those containing Leu 44 Val 65 (Figure 7). There was also a delay before cyclin B1 could be detected in cells containing Tyr 8 Leu 44 Val 65 as compared to those stably transfected with the other two E5 proteins. These effects were not artefactual as levels of protein loaded were similar as indicated by the β-actin controls.

Discussion
We demonstrate that the reference isolate of HPV-16 E5 can co-operate with EGF to reduce the percentage of cells in G 0 G 1 -phase, increase those in S-phase and increase cellgrowth. HPV-6b E5 exhibited similar though much less marked changes than the HPV-16 E5 reference isolate, in agreement with a previous study [10]. In this report we have examined HPV-16 E5 variant activity under controlled conditions. Indeed, all were transcriptionally expressed under the same (T7) promoter and all had a minimal Kozak sequence [30] inserted around the initial ATG codon to insure equivalent translational efficiency. We have used these constructs to significantly extended previous observations observations on the biologic activity of HPV-16 E5 by analysing: HPV-16 E5 transcription in NIH-3T3 cells; cell-cycle progression; and, cell growth.
Effects of E5 variants assayed with EGF upon cell-cycle pro-files   All HPV-16 E5 constructs -aside from Thr 64 variantcaused a fall in the percentage of cells in G 0 G 1 -phase and an increase in S-phase, indicating that these variant E5 proteins stimulate passage through the G 1 /S cell-cycle checkpoint. Such an increase in the proportion of cells in S-phase may be advantageous to HPV-16, permitting it to increase the number of viral copies per cell prior to division.
The HPV-16 E5 Leu 44 Val 65 and Thr 37 Leu 44 Val 65 variants, had greatest percentages of cells in G 2 M-phase and in greatest cell-growth. This biological activity may help explain why these particular E5 (i.e. RFLP pattern 2) vari-ants are most prevalent (~70%) amongst wild populations of HPV-16 in inner-city London and most strongly associated with the presence of cervical lesions [24]. In contrast, the Tyr 8 Leu 44 Val 65 variant (which induced low cell growth) is an RFLP pattern 5 variant which is detected rarely amongst patients with lesions. Interestingly, the HPV-16 E5 RFLP pattern 2 HPV-16 E5 variants also co-segregate with nucleotide variation in the long control region that results in enhanced viral transcription via co-operation with the human POU transcription factor Brn3A [31].
The Leu 44 Val 65 substitutions may act to improve the structural integrity of E5 protein as both are α-helix stabilisers, whereas isoleucine (present at both sites in the reference isolate) is a helix destabiliser [32]. However, comparison of computer-predicted transmembrane regions [33] of the reference and Leu 44 Val 65 variant did not reveal significant differences (data not shown). Another possibility is that the Iso→Val 65 change may enhance the activity of two putative E5 functional domains: the voltage gating motif x-x-x-His 77 : [34]), present in the HPV-16 E5 reference isolate and all reported mammalian connexins; and, the proposed binding site for 16 kDa ATPase (aa 54-78: [2]). Conversely, another E5 variant with an adjacent change at position 64 (Thr 64 variant) exhibited an impoverished biological activity, confounding E5-induced mitogenesis at the G 0 /G 1 checkpoint.
At least three EGF-independent E5 pathways exist ( Figure  1), most notable being the transcriptional repression of p21 by E5 directly [35] or, indirectly, via E5 induction of  Cell-growth curves Figure 6 Cell-growth curves. Growth curves for cell lines stably transfected with different HPV-16 E5 variants, HPV-6b E5 or HPV-16 'stop'. Dotted lines indicate the initial seeding concentration, the lower graph was added for clarity. Error bars represent the standard error of the mean.  We also observed a reciprocal association between the levels of p21 and cyclin B1, this has been reported by others in several cell-systems and may represent a direct inhibitory effect of p21 upon cyclin-B1 biosynthesis [36,37].   [25,26]; and, from the low cancer-risk virus HPV-6b were investigated.

Construction of recombinant DNA expression vectors
HPV-16 E5 genes were amplified in polymerase chain reactions (PCR) using rTth™ DNA polymerase in two separate reactions, so that E5 open reading frames (ORF) between nucleotides (nt) 3866 and 4077 (encoding E5 amino acids 6-76) were obtained. The first PCR utilised an upstream primer (GGA 3836 GCTAGCTCACCATGGCAAATCTTGATA 3865 ) which included an artificial Nhe-1 cut site (underlined) and a minimal Kozak sequence [30] ( 3847 ACCATGG 3853 ) this motif was incorporated to promote equivalent translational efficiency for all constructs and introduces an artificial alanine residue at codon position two). The second PCR used a 5' primer ( 3836 GGAGCTAG CTCACCATGGCATAACTTGATA 3865 ) that also contained a 'stop' signal (underlined): both PCRs utilised the same downstream primer which encoded an artificial BamH1 site ( 4110 TACAGGATCCTTATG TAATTAAAAAGCGT GCATG 4078 ). E5 PCR products were ligated into the Nhe1 and BamH1 sites of pcDNA3.1Myc-His (Invitrogen Ltd.). The E5 open reading frame (ORF) of HPV-6b E5 was also PCR amplified (using the upstream primer: 4103 TACTATATTGTTGCTAGCCCACCATGGTGCTAA 4135 and downstream, 4366 TACAAATATAAAAAACGGGG ATCCCTAATTCATAT 4332 ) and cloned using the same strategy. Plasmids were transformed into E. Coli JM109 cells and selected by ampicillin resistance. Positive colonies were screened by PCR and then sequenced to confirm the identity of the DNA inserts.

In vitro translation
A cell-free wheat-germ expression assay (Promega Ltd.) was used to determine protein translation of T7 mRNA transcripts with individual reactions supplemented with 5 μl of 35 S-labelled cysteine (1 Ci/l: Amersham International Ltd.). Radiolabelled cysteine was selected since all HPV-16 E5 variants contain 4 cysteine residues. Proteins were subjected to PAGE (below) and radioactivity detected by autoradiography.

Preparation of stable NIH-3T3 cells expressing HPV E5 mRNA
NIH-3T3 cells (ATCC Ltd.) were maintained in Dulbecco's minimum essential medium supplemented with 40 mM L-glutamine, 2 × 10 6 U/l benzyl-penicillin, 2 g/l streptomycin sulphate (DMEM) and 10% (v/v) fetal calf serum (DMEM/FCS), in a humidified atmosphere of 5% (v/v) carbon dioxide in air. Cells were transfected with E5 containing plasmids using Lipofectin™ and grown in DMEM/ FCS containing 1 g/l of G418 for 3 weeks (a dose cytotoxic for non-transfected NIH-3T3 cells within one week). Indi-vidual colonies of G418-selected cells were isolated and expanded in DMEM/FCS containing G418 to produce cell-lines for each variant. Individual cell-lines for each 'stop' construct were pooled. Variant cell-lines and the 'stop pool' were tested intermittently for mycoplasma infections: all were negative. Cell lines were tested for E5 mRNA expression using an adaption of the method described by Biswas et al. [5]. Briefly reverse transcriptase (RT) reactions were primed using the HPV-16 E5 downstream primer ( 4110 TACAGGATCCTTATGTA ATTAAAAAGCGTGCAT 4078 ) with Moloney murine leukemia virus reverse transcriptase then samples were subjected to a nested PCR using internal primers located within the E5 ORF [5]. After a further wash in PBS cells were re-suspended in, and stained with, 1 ml of propidium iodide/RNAse solution (50 g propidium iodide and 200 g RNAse/l PBS) for 15 min at rt before analysis on a Becton-Dickenson flow cytometer (FACSCalibur™). Data from 10,000 events (i.e. stained cells) were analysed (for a minimum of four times) for each reading after separation of single intact nuclei from debris and cell-clumps by gating on an FL2 area/width plot. Cells were then characterised on the basis of detection of green (FITC) and red (propidium iodide) fluorescence.

Cell-growth
Cells were grown in 2% (v/v) in DMEM containing 2% (v/ v) fetal calf serum, supplemented with EGF (20 μg/l), after seeding at a density of 0.1 × 10 6 cells in 20 cm 3 Petri dishes. Cells were fed daily with fresh media containing EGF and cultures harvested at 24, 48 and 72 h by detachment from Petri dishes by exposure to Versene (5 ml for 2 min at 37°C). One millilitre of DMEM/FCS was added to inactivate the trypsin then cells were pelleted by centrifugation (200 g for 15 min at rt). Cell pellets were re-suspended in DMEM/FCS and viable cell numbers determined immediately by trypan-blue exclusion staining [38].

Western blots
NIH-3T3 cells (0.1 × 10 6 cells in 5 ml of DMEM/FCS containing no G418) into 20 cm 3 Petri dishes and left overnight. Cells were serum-starved for 24 h in DMEM, grown in DMEM supplemented with EGF (as above), then harvested over an 18 h period. Cells were lysed by suspension in RIPA buffer and cellular debris removed by centrifugation at 10,000 g for 30 sec at rt. Protein concentrations were determined using a commercial assay (BioRad Ltd.) and adjusted to enable 5 μg of total protein to be added to each well of a PAGE gel. Cellular proteins were subjected to PAGE under reducing conditions through a 4-12% Tris/glycine gradient gel (Novex Ltd.) and blotted onto Hybond-P™ membranes (Amersham International Ltd.). Each blot was next cut into strips containing the proteins of interest: cyclin B1 (60 kDa), β-actin (42 kDa) and p21 (21 kDa). Strips were blocked overnight at rt using 5% (w/ v) dried milk powder (Marvel™, Premier Brands UK Ltd.) in Tris-buffered saline/Tween™ (20 mM Tris [hydroxymethyl]-amino methane; 0.2 M sodium chloride and 0.1% [v/v] Tween-20™, pH 7.5: TBS-T).
Blot strips were washed three times with TBS-T then incubated for 1 h at rt with rabbit anti-β-actin (1/1000 dilution), mouse anti-p21 (2.5 mg/l) or, mouse anti-cyclin B1 (2 mg/l: Pharminogen Ltd.) antibodies. After two washes in TBS-T strips were immersed in the appropriate horseradish peroxidase-labelled secondary antibody (sheep anti-mouse immunoglobulin or donkey anti-rabbit immunoglobulin: Amersham International Ltd. each at a dilution of 1/250) for 1 h at rt. Bound antibody was detected using an ECL-plus™ chemiluminescence kit and ECL Hyperfilm™ (Amersham International Ltd.).

Statistical analyses
Students' t-tests were used to assist the interpretation of data.