Interaction between Human Papillomavirus-Encoded E6 Protein and AurB Induces Cell Immortalization and Proliferation—A Potential Target of Intervention

Simple Summary This study identified that Aurora kinase B (AurB), a cellular protein that is upregulated in human cancers, is a bona fide interacting partner of HPVE6. HPVE6 complexes with AurB at the C-terminus end of E6, upstream of the E6-PBM. The AurB-E6 complex forms predominantly in the nucleus or mitotic cells. The positive correlation between E6 and AurB protein in HPV-positive cancer cells confers an increased cell proliferation and growth, and the eventual tumour formation. This study also underlined that the AurB-E6 complex could be a therapeutic target. However, the commercially available Aurora kinase inhibitors to date might not be selective toward HPV-positive cancer cells. Hence, this lack should be addressed. Abstract The human papillomavirus E6 and E7 oncoproteins interact with a different subset of host proteins, leading to dysregulation of the apoptotic, cell cycle, and signaling pathways. In this study, we identified, for the first time, that Aurora kinase B (AurB) is a bona fide interacting partner of E6. We systematically characterized the AurB-E6 complex formation and its consequences in carcinogenesis using a series of in vitro and cell-based assays. We also assessed the efficacy of Aurora kinase inhibitors in halting HPV-mediated carcinogenesis using in vitro and in vivo models. We showed that AurB activity was elevated in HPV-positive cells, and this correlated positively with the E6 protein level. E6 interacted directly with AurB in the nucleus or mitotic cells. A previously unidentified region of E6, located upstream of C-terminal E6-PBM, was important for AurB-E6 complex formation. AurB-E6 complex led to reduced AurB kinase activity. However, the AurB-E6 complex increased the hTERT protein level and its telomerase activity. On the other hand, AurB inhibition led to the inhibition of telomerase activity, cell proliferation, and tumor formation, even though this may occur in an HPV-independent manner. In summary, this study dissected the molecular mechanism of how E6 recruits AurB to induce cell immortalization and proliferation, leading to the eventual cancer development. Our findings revealed that the treatment of AZD1152 exerted a non-specific anti-tumor effect. Hence, a continuous effort to seek a specific and selective inhibitor that can halt HPV-mediated carcinogenesis should be warranted.


Introduction
Human papillomavirus (HPV) contributes to approximately 5% of human cancers and accounts for 7.5% of cancer deaths in women [1]. Among the over 400 HPV genotypes identified officially, only 13 are classified as "high-risk" types, owing to their ability to cause malignancy. HPV16 and 18 are the most prevalent types contributing to over 70% of cervical cancers [2], as well as other anogenital and head-and-neck cancers. While the incidence of HPV-related cervical cancer is declining, there is an alarming trend of an increase in the incidence of HPV-associated head-and-neck cancers, particularly oropharyngeal cancer [3,4].
Using online Scansite freeware [29], we predicted 4 potential AurB recognition sites within 16 E6: S89, T140, S145, and S150. These amino acids were mutated to Alanine (A), as depicted schematically in Figure S1, using the GeneArt TM site-directed mutagenesis system (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. Primer pairs used to generate these expression constructs are listed in Table S1. The resulting plasmids were all verified by DNA sequencing and transformed into E. coli strain DH5-α.

Cell Lines
HeLa (HPV18 cervical cancer cells), CaSki (HPV16 cervical cancer cells), Human embryonic kidney (HEK) 293 (HPV-null epithelial cells), C33A (HPV-null cervical cancer cells) and U-2 OS (HPV-null human osteosarcoma cells) were purchased from the American Type Culture Collection (ATCC) and maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS, GIBCO) at 37 • C in a humidified incubator with 5% CO 2 . Human primary normal epidermal keratinocytes, adult (HeKa) cell line was purchased from ThermoFisher Scientific (Waltham, MA, USA) and maintained in EpiLife TM medium supplemented with human keratinocyte growth supplement (HKGS). Identities of these human cell lines were validated by short tandem repeat (STR) profiling using the AmpFlSTR Identifiler ® Plus PCR Amplification Kit (ThermoFisher Scientific) with the Applied Biosystems 3500 Series Genetic Analyzer and analyzed by GeneMapper ® Software 5 (Applied Biosystems, Bedford, MA, USA). The STR profile of all cell lines showed >88% concordance with their reference profiles in the ATCC cell line database.

Immunoprecipitation Assay
HEK293 cells (7 × 10 5 ) were seeded onto 10-cm dishes and transfected with 10 µg of hemagglutinin (HA)-tagged HPV18 or HPV16 E6. At 24 h post-transfection, cells were treated with 200 ng/mL nocodazole (Sigma, St. Louis, MO, USA) for 18 h. Cells were collected and lysed by using E1A buffer [50 mM NaCl, 0.1% NP-40, and 50 mM HEPES (pH 7.0)], with gentle syringing, and then placed on ice for 20 min. The cell lysate was then centrifuged at 14,000 rpm for 10 min, and the supernatant was incubated with 20 µL of monoclonal Pierce TM anti-HA agarose beads (ThermoFisher Scientific) at 4 • C for 3 h. Samples were washed thrice with E1A buffer and then subjected to western blot analysis.

Downregulation of HPV Oncoproteins Using siRNA
Downregulation of HPV18 E6 and E6/E7 was performed as described previously [30]. In brief, 2 × 10 5 HeLa cells were seeded into 6-well plates. The cells were transfected with short interference RNA (siRNA) using Lipofectamine LTX (Invitrogen) for 72 h. The total cell lysate was collected and subjected to western blotting.

Fusion Protein Purification and In Vitro Binding Assays
Transformants harboring pGEXT2T:HPVE6s constructs were grown overnight. The culture was then inoculated into a fresh medium and incubated for 1 h. Recombinant protein expression was then induced by the addition of isopropyl-β-D-thiogalactopyranoside (IPTG, Sigma) to a final concentration of 1 mM and incubated for a further 3 h. The bacteria were harvested and lysed with cold 1× PBS containing 1% Triton X-100 and sonicated twice for 30 s (sec). The supernatants were collected and incubated with glutathioneconjugated agarose resin on a rotating wheel overnight at 4 • C. After extensive washing, the amount of immobilized GST fusion proteins were analyzed by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and stained by GelCode TM Blue Stain Reagent (ThermoFisher Scientific).
AurB protein was translated in vitro using the TNT ® Coupled Reticulocyte Lysate System (Promega, Madison, WI, USA), according to the manufacturer's recommendation. Equal amounts of in vitro-translated AurB were added to purified GST fusion proteins and incubated for 1 h at room temperature. After being washed thrice with PBS containing 0.1% Tween 20, the bound proteins were subjected to SDS-PAGE and western blotting for the detection of AurB.

Immunofluorescence Assay
This assay is carried out as described previously [30]. Approximately 2 × 10 5 U-2 OS cells were plated onto coverslips. The cells were either mock-transfected or transfected with HA-tagged 16 E6 or E7. Cells were fixed with ice-cold absolute methanol 24 h after transfection and incubated with primary antibody against HA (Roche) and Aurora B (Cell Signaling), followed by Alexa Fluor ® 568-conjugated anti-rabbit secondary antibody and Alexa Fluor ® 488-conjugated anti-mouse secondary antibodies (ThermoFisher Scientific), and counterstained with 4 ,6-diamidino-2-phenylindole (DAPI). The subcellular location of HPV16 E6 or E7 and AurB were examined under a fluorescence microscope (Leica, Wetzlar, Germany).

ADP-Glo TM AurB Kinase Assay
HEK293 cells were transfected with FL-AurB. The cell lysate was collected, and the full-length human AurB was then purified from the cells. The kinase reaction was carried out by incubating 10 µL of purified AurB (approximately 2 µg) with equal amounts of purified HPVE6s GST fusion proteins in the presence of 30 µL of kinase reaction buffer containing 25 mM Tris (pH 7.5), 10 mM MgCl 2 , 3.5 mM NaCl and 75 µM ATP for 1 h at 25 • C. The AurB kinase activity was measured using ADP-Glo™ Kinase Assay (Promega), according to the manufacturer's recommendations. Briefly, the kinase reaction was mixed with ADP-Glo™ Reagent at a ratio of 1:1 and incubated in a white base 384-well plate for 1 h. Following this, one volume of Kinase Detection reagent was added and incubated for an additional 1 h at 25 • C in the dark. The amount of luminescence ADP produced was measured using VICTOR Multilabel Plate Reader (PerkinElmer, Waltham, MA, USA).

Telomeric Repeat Amplification Protocol (TRAP) Assay
A PCR-based assay using TRAPeze ® Telomerase Detection Kit (Merck, Rahway, NJ, USA) was employed to detect telomerase activity according to the manufacturer's recommendations. In brief, 5000 U-2 OS (HPV-null osteosarcoma cell lines lack hTERT and telomerase activity) cells were either mock-transfected or with either HA-tagged hTERT, HA-tagged HPV16 E6, Flag-tagged AurB alone, or co-transfected. The cells were collected and lysed. Cell lysate was then subjected to a one-step PCR amplification using master mix provided by the kit. The products were resolved on a nondenaturing polyacrylamide gel. The gel was stained using SYBR Safe DNA Gel Stain (Invitrogen, Carlsbad, CA, USA) and viewed using the ChemiDoc TM Imaging Systems (Bio-Rad, Hercules, CA, USA).

Quantitative Polymerase Chain Reaction (qPCR)
The cells were collected, and total RNA was extracted using RNeasy Mini Kit (QIAGEN) according to the protocol recommended by the manufacturer. The cDNA was synthesized by adding a total of 1 µg RNA for reverse transcription using S SuperScript TM III Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA). Quantitative PCR (qPCR) will be performed in duplicate by adding 10 µM primers and Luna ® Universal qPCR Mastermix (NEB, Ipswich, MA, USA) in the real-time PCR system (Applied Biosystem, Bedford, MA, USA). The primers for AurB, E6, and GAPDH (housekeeping control) are listed below (Table 1):

In Vivo Xenograft Model
Animal ethics and licenses were obtained prior to conducting this experiment. Immortalized primary baby rat kidney epithelial cells (BRK) and HeLa cells (approximately 5 × 10 6 cells) were resuspended with 200 µL of DMEM and injected subcutaneously over the flank of 4-to 5-week-old female athymic nude mice (12 mice for DMSO treated group; 18 for AZD1152 treated group) supplied by LASEC, the Chinese University of Hong Kong. Mice body weight and tumor size were monitored and recorded every other day for 2 weeks. Upon reaching palpable tumors, the littles were given 50 mg/kg of AZD1152 dissolved in sodium carboxymethyl cellulose (Viscosity: 800-1200 mPa.s) via the oral route. In the vehicle control group, the littles were injected with DMSO. Carcasses were disposed of by LASEC. Tumor volume was estimated using the formula (a × b 2 ) × 0.5236, where "a" refers to the largest dimension of the tumor and "b" is the perpendicular diameter [33]. The tumors were paraffin-embedded and sectioned for hematoxylin and eosin (H&E) staining.

Statistical Analysis
Data presented were performed at least in triplicates of independent experiments. Data were expressed as mean ± SEM with statistical analyses being performed using GraphPad Prism 7. The band intensities of immunoblots and tumor burden of HPVpositive and negative-bearing mice were compared with their respective controls using a t-test. The viability of cells upon drug treatments was analyzed via non-linear regression using GraphPad Prism. A p-value of <0.05 was considered statistically significant.

AurB Activity Was Increased in HPV-Positive Cells
Aurora kinases (AurA, B, and C) are activated when they are phosphorylated (residues T288 on AurA, T232 on AurB, and T198 on AurC). Knowing this, we compared the phosphorylated form of Aurora kinases in HPV-positive and HPV-negative cells to reflect their association with HPV oncoproteins. We examined total cell lysates from HeKa (HPVnull primary human keratinocytes), C33A (HPV-null human cervical cancer cells), HeLa (HPV18-positive cervical cancer cells), and CaSki (HPV16-positive cervical cancer cells) by western blotting. As shown in Figure 1A, we observed an increase in the level of AurB phosphorylated at T232 [pAurB (T232) or herein refers as pAurB] in HPV-positive cells (3-fold in HeLa; 1.6-fold in CaSki cells) compared with HPV-null cells. The level of phosphorylated AurA was unchanged, while the phospho-AurC levels were reduced in CaSki cells. We also noted a higher level of total AurB protein in cancer cells (C33A, HeLa, and CaSki) than in normal keratinocytes (HeKa). The level of AurB protein in HeLa (1.5-fold) and CaSki (1.8-fold) was higher than that of C33A. The higher level of pAurB indicates that HPV-positive cancer cells had a higher AurB activity than HPV-negative cells, suggesting that HPV may induce an increased level of AurB activity in HPV-positive cells.

AurB Activity Was Increased in HPV-Positive Cells
Aurora kinases (AurA, B, and C) are activated when they are phosphorylated (residues T288 on AurA, T232 on AurB, and T198 on AurC). Knowing this, we compared the phosphorylated form of Aurora kinases in HPV-positive and HPV-negative cells to reflect their association with HPV oncoproteins. We examined total cell lysates from HeKa (HPV-null primary human keratinocytes), C33A (HPV-null human cervical cancer cells), HeLa (HPV18-positive cervical cancer cells), and CaSki (HPV16-positive cervical cancer cells) by western blotting. As shown in Figure 1A, we observed an increase in the level of AurB phosphorylated at T232 [pAurB (T232) or herein refers as pAurB] in HPV-positive cells (3-fold in HeLa; 1.6-fold in CaSki cells) compared with HPV-null cells. The level of phosphorylated AurA was unchanged, while the phospho-AurC levels were reduced in CaSki cells. We also noted a higher level of total AurB protein in cancer cells (C33A, HeLa, and CaSki) than in normal keratinocytes (HeKa). The level of AurB protein in HeLa (1.5fold) and CaSki (1.8-fold) was higher than that of C33A. The higher level of pAurB indicates that HPV-positive cancer cells had a higher AurB activity than HPV-negative cells, suggesting that HPV may induce an increased level of AurB activity in HPV-positive cells.  and Aurora C at amino acid 198 [pAurC (T198)], plus total AurB, were examined via western blotting. Note the increased levels of pAurB (T232) and total Aurora B in HeLa and CaSki. β-actin was included as a loading control. (B). The level of phosphor-AurB (pAurB) upon E6 overexpression. HEK293 cells were transfected with HA-tagged 16E6 or HA-tagged 18E6. The cells were also transfected with LacZ as a transfection efficiency control. The total cell lysate was collected and subjected to western blotting. Representative immunoblots showing the levels of phospho-AurB [pAurB (T232)], total AurB, HA (for E6), plus β-actin as a loading control. Note the increased expression level of pAurB upon overexpression of HA-16 and -18 E6. (C(i)). The level of pAurB and total AurB were examined upon E6 downregulation. HeLa cells were transfected with siRNA against control (siCtrl), HPV18 E6 (si18E6), or HPV18 E6 and E7 (si18E6/E7). After 72 h, the total lysate was collected, and the levels of pAurB and total AurB were ascertained by western blotting. Also shown is the rescued level of p53, indicating the efficient downregulation of E6. β-actin was included as a loading control. (C(ii)). The bar graph shows the quantitation of the levels of pAurB, AurB, and p53 upon HeLa cells transfected with siRNA against control (siCtrl) (black bar), 18E6 (siE6) (green bar) and 18E6/E7 (siE6/E7) (blue bar). Quantitation was performed using ImageJ software, and statistical analysis was performed using Prism. Error bars represent mean ± standard error of the mean (SEM) (n = 3). (ns = not significant, * p < 0.05, ** p < 0.01, *** p < 0.001). (C(iii)). The expression of HPV18E6 and AurB genes upon si18E6 or si18E6/E7 in HeLa cells was quantitated by performing a quantitative polymerase chain reaction (qPCR). The level of the respective mRNA was normalized with GAPDH housekeeping control. Error bars represent mean ± standard error of the mean (SEM) (n = 3). (ns = not significant, * p < 0.05, ** p < 0.01, *** p < 0.001) (Western Blots in Figure S6).

The Level of AurB Protein Correlated Positively with HPVE6
Next, we asked if the increased level of AurB protein has a correlation with either E6and/or E7-encoded by HPV16 and 18. Intriguingly, we observed that the level of AurB protein and its activity in HPV-positive cancer cells correlated positively with the level of HPVE6 protein through the following two approaches: (i) Overexpression of HPV oncoproteins in HPV-null cells Firstly, we overexpressed HA-tagged E6-encoded by HPV16 and HPV18 in HEK293 cells. From our immunoblots ( Figure 1B), we observed an increased level of pAurB, while the expression level of total AurB was unchanged.

AurB Interacted Directly with HPVE6 at the C-Terminus of HPVE6, Independent of E6-PBM
As we observed an association between AurB and HPVE6, we next wanted to investigate if this association occurs through direct interaction between AurB and E6. We addressed this through co-immunoprecipitation and in vitro binding assays. From both approaches, we observed that AurB interacted directly with HPVE6.
After transfecting HA-tagged E6-and E7-encoded by HPV16 in HEK293 cells, the cells were treated with nocodazole, which is an anti-mitotic agent that can induce an increase in endogenous AurB protein level (Figure 2A,B). We then immunoprecipitated E6 and E7 from cell extracts using HA-conjugated agarose resin and examined the bound AurB via western blotting. We observed that AurB, upon treatment with nocodazole, bound to HA-HPV16E6 but not to HA-HPV16E7 ( Figure 2A). We also detected AurB bound to HA-HPV18E6 in HEK293 cells ( Figure 2B). This indicates that AurB interacts with both E6-encoded HPV16 and 18 in HEK293 cells but not with 16E7.
We next investigated whether the interaction between AurB and E6 is direct. To do this, we performed an in vitro binding assay, where in vitro-translated AurB was incubated with purified GST-tagged E6 fusion proteins encoded by HPV16 or 18, and the bound AurB was detected via western blotting. As shown in Figure 2C, AurB is bound to purified GST-HPV16 and -HPV18E6 fusion proteins.  ii)). The bar graph shows the mean levels of bound FL-AurB quantitated using ImageJ software and normalized with the amount of the relative GST fusion proteins in comparison with WT. (E(i)). Direct binding assay to determine key amino acids of E6 that bind to AurB. By using a similar approach, the indicated purified GST E6 wild type and E6 with its C-terminus PDZ binding motif depleted (E6∆PBM) mutant fusion proteins were incubated with in vitro-translated Flag-tagged AurB (FL-AurB). The upper panel shows immunoblot (IB) detected using an AurB-specific antibody; the lower panel shows the Ponceau stain (PS) of the blot. (E(ii)). The bar graph shows the mean levels of bound FL-AurB quantitated using ImageJ software and normalized with the amount of GST fusion proteins in respective experiments. (F). ADP-Glo TM assay was performed to determine the kinase activity of Aurora kinase B (AurB). The indicated purified GST fusion proteins were incubated with purified AurB. The ability of AurB to convert ATP into luminescent ADP was measured. The bar graph shows the AurB activity in percent (%) normalized with the amount of GST fusion proteins in respective experiments from three independent experiments. Error bars represent the mean ± standard error of the mean (SEM) (n = 3). (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). (Western Blots in Figure S7).
(iii) AurB bound to E6 at the C-terminus of E6, independent of E6-PBM.
As the amino acids S89 to S150 of E6 are located outside, but in very close proximity to the extreme C-terminus end of the PDZ binding motif of E6 (E6-PBM), we wanted to investigate if other amino acids and the integrity of E6-PBM might be important for AurB-E6 complex formation, we, therefore, included E6 mutant in which PBM is depleted (E6∆PBM) and performed a similar in vitro binding assay. Intriguingly, we found that the E6∆PBM mutant bound to in vitro-translated AurB was, in fact, 3-fold stronger than the E6 wild-type ( Figure 2E(i,ii)).
These results indicate that AurB interacts directly with HPV16 and HPV18 E6 in an E6-PBM-independent manner. The amino acid residues at the C-terminus of HPV16E6 (140-150), a region upstream of E6-PBM, are the most critical for the AurB-E6 complex formation.

HPVE6 Formed a Complex with AurB in the Nucleus
Subsequently, we examined the subcellular co-localization of AurB and E6 using an immunofluorescence assay. We ectopically expressed HA-tagged HPV16 E6 or E7 in U-2 OS, an HPV-null human osteosarcoma cell line. This cell line was chosen due to the following reason: (i) the cell is easily transfected with detectable protein level; (ii) morphologically, the cells are flat and have a distinctive nuclear and cytoplasmic region; (iii) the cells lack the expression of hTERT, which is a target of AurB and E6. We found that AurB co-localized with HPV16E6 in the nucleus of these cells. It appeared that E6 co-localized with AurB in the mitotic cells ( Figure S2). We observe a fairly weak or none overlapped expression of E7 with AurB. This result indicates that AurB interacts with E6 directly, and the AurB-E6 complex is formed predominantly in the nucleus and mitotic cells.

E6 Inhibited AurB Kinase Activity
Upon gaining knowledge that AurB expression correlates positively with the expression of HPVE6 through direct interaction between AurB and E6, we were intrigued to know if such an association affects the functions of AurB. As one of the obvious functions of AurB is to act as a mitotic kinase, we performed a sensitive luminescence-based ADP-Glo TM kinase assay. This assay examines the level of consumed ATP (ADP) following kinase activity, with a higher amount of ADP indicating higher kinase activity and vice versa. We incubated purified AurB with purified GST-16E6 fusion proteins. As AurB can undergo autophosphorylation, purified AurB was incubated with purified GST protein alone to serve as a positive control. We found that, in the presence of GST-16E6 fusion proteins, the AurB activity was reduced by more than half ( Figure 2F). Among the E6 mutants, E6 T140A (24.4% ± 4.0%; p < 0.0001) and E6 S145A (38.0% ± 5.8%; p < 0.001) reduced AurB activity more dramatically, followed by wild-type E6 (44.0% ± 9.6%; p < 0.01), E6 S150A (46.9% ± 9.8%; p < 0.01) and E6 S89A (48.7% ± 16.1%; p < 0.05). This indicates that the binding of E6 to AurB, disregarding the affinity of binding of E6 to AurB, can reduce the AurB kinase activity.

E6 Did Not Perturb the Function of AurB in Regulating hTERT Protein
As both AurB and E6 independently perturb hTERT, we then investigated if the AurB-E6 association could affect hTERT protein level and activity. In order to do this, an HPV-null U-2 OS (an osteosarcoma cell line that lacks hTERT expression) was used. HA-tagged hTERT was transfected into U-2 OS cells either alone or together with Flagtagged AurB (FL-AurB) and/or HA-tagged 16E6 (HA-16E6). Cell lysates were collected, and the protein expression level of hTERT was ascertained via western blotting. As shown in Figure 3A, co-expression of hTERT with AurB and E6 (460.5% ± 109.7%, p < 0.05) or with AurB (412.4% ± 161.0%, p < 0.05) led to an increased hTERT expression by at least 4-folds when compared with the cells with hTERT expressed alone. This result indicates that E6 does not perturb the function of AurB in mediating an increase in hTERT expression.

The Association of AurB and E6 Led to Increased Telomerase Activity
Next, we assessed if the telomerase activity in hTERT-expressing cells was perturbed by AurB-E6 by performing a PCR-based TRAPeze telomerase detection assay. U-2 OS cells were transfected with FL-AurB and HA-16E6, or with HA-hTERT and FL-AurB, or HA-hTERT, FL-AurB, and HA-16E6. As shown in Figure 3A(iii) and S3A, at least a 2-fold increment of telomerase activity was detected in cells with AurB and E6 (265.4% ± 45.8%, p < 0.05), hTERT, and AurB (282.0% ± 65.9%, p < 0.05), or hTERT was co-expressed with AurB and E6 (251.3% ± 30.4%, p < 0.01) when compared with the basal telomerase activity in the mock-transfected U-2 OS cells. This result demonstrates that E6 does not perturb the function of AurB in regulating telomerase activity. More strikingly, we also observed that AurB and E6 overexpression were able to increase the basal level of telomerase activity.  (ii)). The bar graph shows the mean levels of hTERT protein quantitated using ImageJ software and normalized with loading control in respective experiments. (A(iii)). The bar graph shows the quantification of telomerase activity by measuring the signal of all telomerase ladder bands using ImageJ software and by dividing by the intensity of the bands from the internal controls. (B(i)). C33A (HPV-null cervical cancer cells) and CaSki (HPV16-positive cervical cancer cells) were treated with a potent Aurora kinase B inhibitor (AZD1152) or pan-Aurora kinase inhibitors (Reversine, AMG900, and Tozasertib) at concentrations indicated for 24 h. Simultaneously, the cells were also treated with DMSO, which served as vehicle control. The total cell lysate was collected, and the expression levels of phosphor-Aurora kinase B (pAurB), total Aurora kinase B (AurB), HPV16E7, and p53 were ascertained via western blotting. Beta-actin (βactin) was included as a loading control. (B) The bar graphs show the mean levels of (ii) pAurB, (iii) total AurB, (iv) HPV16E7, and (v) p53, quantitated using ImageJ software and normalized with LacZ was included as a transfection efficiency control. The total cell lysate was collected, and the levels of hTERT, phosphor-AurB [pAurB (T232)], AurB, and 16E6 were ascertained via western blotting. Note the increased expression of hTERT upon co-expression with either AurB and/or 16E6. (A(ii)). The bar graph shows the mean levels of hTERT protein quantitated using ImageJ software and normalized with loading control in respective experiments. (A(iii)). The bar graph shows the quantification of telomerase activity by measuring the signal of all telomerase ladder bands using ImageJ software and by dividing by the intensity of the bands from the internal controls. (B(i)). C33A (HPV-null cervical cancer cells) and CaSki (HPV16-positive cervical cancer cells) were treated with a potent Aurora kinase B inhibitor (AZD1152) or pan-Aurora kinase inhibitors (Reversine, AMG900, and Tozasertib) at concentrations indicated for 24 h. Simultaneously, the cells were also treated with DMSO, which served as vehicle control. The total cell lysate was collected, and the expression levels of phosphor-Aurora kinase B (pAurB), total Aurora kinase B (AurB), HPV16E7, and p53 were ascertained via western blotting. Beta-actin (β-actin) was included as a loading control. (B) The bar graphs show the mean levels of (ii) pAurB, (iii) total AurB, (iv) HPV16E7, and (v) p53, quantitated using ImageJ software and normalized with loading control in respective experiments. Error bars represent the mean ± standard error of the mean (SEM) (n = 3). (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). (C) The bar graph shows the quantification of telomerase activity by measuring the signal of all telomerase ladder bands using ImageJ software and by dividing by the intensity of the bands from the internal controls. (Western Blots in Figure S8).
Next, we were intrigued to understand the interplay between hTERT and Ras/MEK/ERK pathway in the presence of AurB and E6. This is because hTERT can inhibit the Ras/MEK/ERK pathway, which is one of the key oncogenic pathways [34]. We found that hTERT, regardless of when co-expressing with AurB or E6, can reduce the levels of Ras/MEK/ERK pathway ( Figure  S3B). When hTERT co-expressed with AurB and E6, the levels of Ras, MEK1/2, and ERK1/2 were similar to that of in the mock-transfected cells, indicating that the AurB-E6 complex may override hTERT inhibition on Ras/MEK/ERK axis, thereby allowing cell survival.
Altogether, these results indicate that the AurB-E6 association can maintain a high hTERT protein level. More importantly, AurB and E6 can accelerate telomerase activity synergistically in HPV-positive cancer cells, which is critical for E6 to induce immortalization during cancer progression.

d. Apoptosis activation
From the above results, we observed that 300 nM of AZD1152 treatment reduced the protein levels of phosphor-and total AurB and HPV16E7, as well as increased the level of p53 protein), and knowing that AZD1152 is an anti-tumor agent, we next wanted to study if 300 nM of AZD1152 could activate the apoptotic pathway. We treated HPV-positive (HeLa and Caski) and -negative (C33A) cancer cells with 300 nM of AZD1152 for 3 days. We collected total cell lysate on each day of the treatment and analyzed the expression levels of caspase-8 and p53 via western blotting. We observed an increase in caspase-8 and p53 proteins in both HeLa and CaSki cells when these cells were treated with 300 nM of AZD1152 ( Figure S4A). Interestingly, we also observed increased levels of these proteins in a treatment day-dependent manner, in which the lowest expression of caspase-8 and p53 was observed on Day 0 and the highest expression on Day 3. This increment was not obvious in C33A cells treated with 300 nM of AZD1152. On the contrary, treatment of C33A with 300 nM of AZD1152 reduced the expression of caspase-8 from Day 0 to Day 3. These results indicate that HPV-positive cells are more susceptible to cell death induced by treatment with 300 nM AZD1152 compared with HPV-null cells.
We then wanted to scrutinize the effect of AZD1152 treatment on telomerase activity using similar experimental settings as described in Figure 3A. We also included HeLa cells, which are known to carry a high endogenous telomerase activity, mock-transfected with a pcDNA3.1 vector. To further demonstrate the inhibitory effect of Aurora kinase inhibitors on telomerase activity, HA-hTERT was overexpressed in HeLa cells. After 24 h, the cells were treated with 300 nM of AZD1152.
Under the treatment of 300 nM of AZD1152, we observed a higher level of telomerase activity in cells with hTERT, AurB, and E6 co-expressed (120.4% ± 5.2%, p < 0.05) when compared with the basal telomerase activity in the mock-transfected U-2 OS cells ( Figure 3C and Figure S3C). Whereas a similar telomerase activity was detected in cells with AurB and E6 (118.9% ± 16.8%, p = 0.32) or AurB and hTERT (109.2% ± 9.6%, p = 0.39) expressed when compared with the mock-transfected U-2 OS cells. Even though a higher telomerase activity was detected in mock-transfected HeLa cells (155.5% ± 16.8%, p < 0.05) and hTERT-transfected HeLa cells (208.5% ± 35.3%, p < 0.05) compared to mock-transfected U-2 OS cells, there was no significant difference in the telomerase activity between mockor hTERT-transfected HeLa cells (p = 0.25). These results show that AZD1152 treatment inhibits telomerase activity when AurB and E6 are co-expressed in U-2 OS cells. However, the treatment is insufficient to suppress telomerase activity orchestrated by hTERT-AurB-E6 in HPV-positive cancer cells.

Treatment with AZD1152 Suppressed the Cell Proliferation and Tumour Formation in an HPV-Independent Manner
We performed a cell counting kit-8 (CCK8) assay to assess the ability of viable cells to digest tetrazolium salt into orange formazan after the cells were treated with AZD1152. We found that 206 nM ( Figure S4B(i)), 151 nM ( Figure S4B(ii)), and 179 nM ( Figure S4B(iii)) of AZD1152 achieved 50% inhibition of the viability of HeLa, CaSki, and C33A cells, respectively. These results show that Aurora kinase B inhibition, following treatment with 300 nM of AZD1152, can inhibit the proliferation of both HPV-null and -positive cells. The inhibitory effect exerted by AZD1152 is not specific toward HPV-positive cancer cells.
We then wanted to recapitulate the efficacy of AZD1152 treatment in inhibiting tumor formation using biologically relevant 3-dimensional organotypic culture and animal models. Therefore, we addressed this by performing the following assays: (i) Three-dimensional organotypic raft culture We grew C33A, HeLa, and Caski cells in an organotypic raft culture, where these cells were seeded onto an epidermal equivalent with a collagen layer containing fibroblast cells and supplemented with growth and differentiating factors. Culturing cervical cancer cells under this condition can mimic the microenvironment of cervical carcinoma to a large extent [35]. After 14 days of culture, the cells were treated with 300 nM of AZD1152 continuously for an additional 7 days. From the tissue sections (Figure 4), C33A, HeLa, and CaSki were able to differentiate under DMSO treatment, even though the thickness of CaSki cells was relatively thinner than that of C33A and HeLa cells. Treatment with 300 nM of AZD1152 reduced the thickness of all three cell lines. We observed approximately 40-50% of C33A, 10-20% of HeLa, and 20-30% of CaSki cells attached to the epidermal equivalent. It is worth noting that the majority of the HeLa cells treated with 300 nM of AZD1152 appeared to dislodge from the epidermal equivalent. Our results reveal that 300 nM of AZD1152 is sufficient to inhibit differentiation and survival of both HPV-null and -positive cancer cells.
(ii) HPV-athymic nude mice As depicted in Figure 5A, we injected HeLa or immortalized primary baby rat kidney (BRK) cells into the dorsal flank of 4-5 weeks old female athymic nude mice. The BRK cells were included as a negative control. This cell line is HPV-null and is able to maintain tumor formation in athymic nude mice, as described in our previous study [36]. After 2 weeks, when a palpable tumor was achieved, the mice were fed with 50 mg/kg of AZD1152 via oral route once every 2 days, for a total of 6 courses within 12 days. The body weight and tumor formation were monitored closely along the course of the experiment ( Figure S5). The littles were sacrificed, and tumors were collected and weighed. We observed that the tumor volume of nude mice injected with HeLa cells was reduced dramatically when treated with AZD1152 (34.7% ± 20.1%, p < 0.05) compared with DMSO-treated mice ( Figure 5B). While tumor volume of mice injected with BRK control cells also reduced significantly upon AZD1152 (47.9% ± 10.0%, p < 0.001) treatment ( Figure 5C). Taken together, both the results from 3D organotypic culture and animal study consistently show that treatment using AZD1152 inhibits tumor formation of cancer cells, independent of HPV-status of the cells. (ii) HPV-athymic nude mice As depicted in Figure 5A, we injected HeLa or immortalized primary baby rat kidney (BRK) cells into the dorsal flank of 4-5 weeks old female athymic nude mice. The BRK cells were included as a negative control. This cell line is HPV-null and is able to maintain tumor formation in athymic nude mice, as described in our previous study [36]. After 2 weeks, when a palpable tumor was achieved, the mice were fed with 50 mg/kg of AZD1152 via oral route once every 2 days, for a total of 6 courses within 12 days. The body weight and tumor formation were monitored closely along the course of the weighed. We observed that the tumor volume of nude mice injected with HeLa cells was reduced dramatically when treated with AZD1152 (34.7% ± 20.1%, p < 0.05) compared with DMSO-treated mice ( Figure 5B). While tumor volume of mice injected with BRK control cells also reduced significantly upon AZD1152 (47.9% ± 10.0%, p < 0.001) treatment ( Figure 5C). Taken together, both the results from 3D organotypic culture and animal study consistently show that treatment using AZD1152 inhibits tumor formation of cancer cells, independent of HPV-status of the cells.

Discussion
Both E6 and E7 viral oncoproteins are small in molecular mass. However, they possess a robust ability to interact with key cellular proteins and instigate their degradation in a proteasome-dependent manner: E6 degrades p53 and PDZ proteins (hDlg, hScribble, and MAGIs); E7 degrades pRB and its related pocket proteins. These are the most well-known targets of E6 and E7, but the list of their cellular targets is expanding. In this report, we identify for the first time that E6 proteins from cancer-causing HPV types (HPV16 and HPV18) interact with a cell cycle modulator, Aurora kinase B (AurB). We employed a systematic strategy to investigate the oncogenic association between E6 and AurB. Firstly, we investigated the direct interaction between E6 and AurB using biochemical approaches. Secondly, we provided molecular evidence which delineated the consequence of AurB-E6 association to the functions of AurB and their effect in orchestrating oncogenic events of HPV-positive cancer cells. To provide molecular and biochemical evidence on the association between AurB and E6, we utilized transient overexpression using immortalized cancer cell lines with high transfection efficiency. These approaches allow us to study the cellular functional consequences when AurB and E6 were expressed at a 1:1 ratio. Similarly, we were able to study the levels of pAurB and AurB upon E6 depletion at ≥50%. In addition, we focused on delineating the association between E6-encoded by HPV16 and AurB, as HPV16 contributes to the majority of the cancer cases. Thirdly, using biologically relevant preclinical models, we study the specificity and efficacy of commercially available Aurora kinase inhibitors in halting HPV-mediated hallmarks of cancer phenotype.
Of the Aurora kinase family members, our and previous studies have focused on the association between AurA and HPV [22,27,28], while little is known about the role of AurB in HPV-associated cancers. Although AurB is known to be upregulated and even suggested to be a diagnostic marker in several human cancers, including head-and-neck and cervical cancers [37], the available studies were conducted in disregard of HPV oncoproteins. Our findings provide essential molecular evidence that clearly shows the upregulation of AurB in HPV-immortalized cancer cells correlates directly with the E6 protein level. In this study, we used nocodazole treatment to increase the levels of endogenous AurB. Nocodazole is a microtubule poison and can activate the DNA damage pathway. Upon receiving DNA damage stimuli, due to drug treatment, or upon HPV infection [38], Chk1 is activated. Chk1 then phosphorylates AurB, resulting in its activation and increased expression [39,40] ( Figure 6A). We found that the AurB protein corresponds positively with the E6 protein level in HPV-positive cancer cells. During HPV-induced cancer progression, the increased expression of E6 could lead to increased AurB stability and its activity. Of note, the level of AurB is unlikely to be affected by p53, owing to the fact that AurB acts upstream of p53, where AurB phosphorylates p53, leading to the degradation of p53 protein [41,42]. Previous studies have shown that phosphorylation of E6 at E6-PBM upon DNA damage response induction resulted in the inactivation of p53 transcriptional activity [43]. At the same time, the upregulation of AurB mediated by E6 in HPV-containing cervical cancer cells may potentially lead to further inactivation of p53.
Through the binding assays, we identified that 10 amino acid residues (140 to 150) located immediately upstream of the E6-PBM mediate a previously unidentified oncogenic function of E6. This region is important to form a complex with AurB and AurA [28]. The binding of E6 to AurB may change the conformation of AurB and reduce its ATP binding affinity to AurB ( Figure 6B). Even though E6 mutants showed a reduced binding ability to AurB, the weak association between AurB and E6 might be sufficient to perturb the kinase activity of AurB. However, in HPV-positive cells, the association between E6 and AurB increased phosphorylation of AurB, particularly at T232 of AurB. As AurA and AurB share almost 70% homology, they play different roles in cell cycles. AurA plays a central role during interphase and mitotic entry, while AurB is involved in the mitotic phase and cytokinesis. From our previous and current studies, despite binding to the same region of E6, an association of E6 with AurA and AurB appears to have different consequences. We observed that the binding of E6 to AurB reduced AurB kinase activity but did not affect AurA [28]. Nonetheless, the AurA-E6 association could regulate G1/S and mitotic phases of the cell cycle through cyclin E and histone H3 [28]. While AurB-E6 association affects human telomerase, reverse transcriptase (hTERT), and Ras/MEK/ERK signaling axis. Therefore, it is tempting to speculate that when AurA and AurB proteins are aberrantly upregulated in cancer cells, E6 could bind to them simultaneously to abrogate cell cycle checkpoints, allowing cells to survive and proliferate. Figure 6. A schematical diagram summarizes the molecular consequences of the association between Aurora kinase B and HPVE6 to cellular events, and treatment of AZD1152 can disrupt the oncogenic event mediated by the AurB-HPVE6 complex. (A). Aurora kinase B (AurB) is a kinase with its ATP acceptor site located at amino acid residue 120. Phosphorylation of AurB at amino acid residue threonine (T) 232 and/or T331 increases its kinase activity. AurB is also phosphorylated by Chk1 at T331. HPVE6 and AurB, independently, favor phosphorylation histone H3 at position serine (S) 28. AurB activates transcription factors specific to protein 1 (Sp1) and C-myc, which in turn leads to increased expression (indicated by the solid red arrow) of human telomerase reverse transcriptase (hTERT). E6 also has a role in increasing the hTERT protein level. While increased hTERT protein resulted in increased cell proliferation, hTERT also inhibited the Ras/MEK/ERK axis. (B). HPVE6 binds to AurB at the C-terminal end of E6 (residues 140 to 150). This interaction may result in altered protein conformation of Aur. This, in turn, results in the distortion of the ATP binding site within AurB. However, the phospho-status of AurB at T232 remains unaffected. In addition, the association between AurB and HPVE6 did not disrupt the ability of AurB to increase hTERT protein level and its telomerase activity, as well as the post-translational modification of histone H3: acetylation of histone H3 at residue lysine (K) 9 and phosphorylation of histone H3 at residues S10 and S28. In addition, the hindrance effect of hTERT on Ras/MEK/ERK expression is also overridden by the AurB-E6 complex, leading to increased expression of this key oncogenic signaling pathway. As the lesion progresses, the level of E6 protein increases. This will also increase the AurB protein level and activity, leading to cell immortalization, proliferation, and survival, with dysregulated cell division and formation of the eventual carcinoma. (C). Treatment with AZD1152, a potent and specific Aurora kinase B inhibitor, suppresses AurB activity and activates the apoptotic pathway. The treatment also can reduce hTERT protein level and its telomerase activity, halting HPV-mediated carcinogenesis, albeit the inhibitory effect occurs in a non-HPV-specific manner (indicated by dashed red arrows).
Through the binding assays, we identified that 10 amino acid residues (140 to 150) located immediately upstream of the E6-PBM mediate a previously unidentified oncogenic function of E6. This region is important to form a complex with AurB and AurA [28]. The binding of E6 to AurB may change the conformation of AurB and reduce its ATP binding affinity to AurB ( Figure 6B). Even though E6 mutants showed a reduced binding ability to AurB, the weak association between AurB and E6 might be sufficient to perturb the kinase activity of AurB. However, in HPV-positive cells, the association between E6 and AurB increased phosphorylation of AurB, particularly at T232 of AurB. As AurA and AurB share almost 70% homology, they play different roles in cell cycles. AurA plays a central role during interphase and mitotic entry, while AurB is involved in the mitotic phase and cytokinesis. From our previous and current studies, despite binding to the same Figure 6. A schematical diagram summarizes the molecular consequences of the association between Aurora kinase B and HPVE6 to cellular events, and treatment of AZD1152 can disrupt the oncogenic event mediated by the AurB-HPVE6 complex. (A). Aurora kinase B (AurB) is a kinase with its ATP acceptor site located at amino acid residue 120. Phosphorylation of AurB at amino acid residue threonine (T) 232 and/or T331 increases its kinase activity. AurB is also phosphorylated by Chk1 at T331. HPVE6 and AurB, independently, favor phosphorylation histone H3 at position serine (S) 28. AurB activates transcription factors specific to protein 1 (Sp1) and C-myc, which in turn leads to increased expression (indicated by the solid red arrow) of human telomerase reverse transcriptase (hTERT). E6 also has a role in increasing the hTERT protein level. While increased hTERT protein resulted in increased cell proliferation, hTERT also inhibited the Ras/MEK/ERK axis. (B). HPVE6 binds to AurB at the C-terminal end of E6 (residues 140 to 150). This interaction may result in altered protein conformation of Aur. This, in turn, results in the distortion of the ATP binding site within AurB. However, the phospho-status of AurB at T232 remains unaffected. In addition, the association between AurB and HPVE6 did not disrupt the ability of AurB to increase hTERT protein level and its telomerase activity, as well as the post-translational modification of histone H3: acetylation of histone H3 at residue lysine (K) 9 and phosphorylation of histone H3 at residues S10 and S28. In addition, the hindrance effect of hTERT on Ras/MEK/ERK expression is also overridden by the AurB-E6 complex, leading to increased expression of this key oncogenic signaling pathway. As the lesion progresses, the level of E6 protein increases. This will also increase the AurB protein level and activity, leading to cell immortalization, proliferation, and survival, with dysregulated cell division and formation of the eventual carcinoma. (C). Treatment with AZD1152, a potent and specific Aurora kinase B inhibitor, suppresses AurB activity and activates the apoptotic pathway. The treatment also can reduce hTERT protein level and its telomerase activity, halting HPV-mediated carcinogenesis, albeit the inhibitory effect occurs in a non-HPV-specific manner (indicated by dashed red arrows).
AurB is involved in regulating the function of hTERT and Ras/MEK/ERK signaling cascade. We showed that AurB increased hTERT expression, and this is in line with previous findings [25]. AurB, on its own, activates specific protein 1 (Sp1) and c-Myc transcription factors, leading to increased expression of hTERT [25], as depicted in Figure 6A. The upregulation of hTERT could potentially be due to activation of other signaling pathways, post-transcriptional or -translational modification mediated by AurB. This is unclear and worth exploring in the future. In this study, we observed that E6-AurB was essential to maintain a high expression of hTERT protein and its telomerase activity ( Figure 6B). Previous studies have shown that the Ras/MEK/ERK axis regulates the expression of both hTERT [44] and AurB [45], while HPVE6 activates the Ras/MEK/ERK pathway [46]. Nonetheless, the expression of hTERT represses the Ras/MEK/ERK pathway [34]. Similarly, we also found that, regardless of the presence of E6 or AurB alone, hTERT protein represses the levels of Ras/MEK/ERK proteins ( Figure 6A). Strikingly, a different phenomenon was observed when AurB-E6-hTERT was co-expressed in cervical cancer cells. AurB-E6 association could override the hTERT-mediated suppression on Ras/MEK/ERK pathway ( Figure 6B). This could be a mechanism of how E6 exploit AurB to perturb this key oncogenic signaling pathway to enhance cell proliferation and survival. Hence, leading to the increased thickness of the suprabasal layer of HPV-bearing tumor during HPV-induced carcinogenesis.
Upon gaining knowledge about the oncogenic role of AurB in mediating E6-induced oncogenesis, we compared the effect of Aurora kinase inhibition on HPV oncoprotein expression, apoptotic activation, cell proliferation, and tumor formation of HPV-containing and HPV-null cervical cancer cell lines. To date, there is no inhibitor that disrupts the association between AurB and E6. Therefore, we treated these cells with commercially available AZD1152 or Barasertib, a potent and specific AurB inhibitor, which is an ATPcompetitive inhibitor currently used in clinical trials for solid tumors. The efficacy and selectivity of AZD1152 on HPV-associated cancers have not been clearly elucidated. Here, we provided preclinical evidence, which unleashes that AZD1152 exhibited a greater efficacy and selectivity in activating apoptotic pathways and inhibiting HPV oncoprotein expression in HPV-containing cervical cancer cells compared to HPV-null cancer cells. The inhibitor also suppressed telomerase activity augmented by AurB and/or E6 ( Figure 6C). However, the inhibition is insufficient to suppress telomerase activity in HPV-positive cancer cells, in which hTERT, AurB, and E6 proteins are often produced in high abundance. Consistently, treatment with AZD1152 inhibits proliferation and tumor formation of cancer cells. These effects were observed by disregarding HPV status. Most of the anti-tumor therapeutics available to date exert a broad-spectrum killing effect and often pose adverse after-treatment effects to the patients. As HPV-containing cervical cancer cells might be more susceptible to apoptotic induction, a continuous effort to explore a more target-specific therapeutic option for HPV-associated cancers should be warranted. In addition, it is also worth mentioning that the use of primary keratinocytes to recapitulate the association between AurB and E6 and treatment efficacy should be performed. Our data revealed that the AurB-E6 complex is a drug-able target that could offer a new horizon for the generation of next-generation therapies for HPV-associated cervical cancer and potentially for other HPV-associated cancers.

Conclusions
In summary, our study shows, for the first time, AurB is a genuine and previously unrecognized target of E6 encoded by cancer-causing HPVs. We also provided mechanistic evidence on how E6-mediates cancer progression through AurB by increasing the hTERT protein level and its telomerase activity to induce cell immortalization, proliferation, and survival. More importantly, we revealed that treatment with AZD1152, a potent and specific AurB inhibitor, inhibited tumor formation of HPV-positive cells by activating the apoptotic pathway. Nonetheless, the inhibitory effect exerted by AZD1152 is independent of HPV status in the cancer cells.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cancers15092465/s1, Figure S1: Schematic diagram showing the basic structure of cancer-causing HPV16E6 and potential amino acids recognised by Aurora B. E6 contains two major functional domains: two zinc (Zn) finger binding domains, the regions where E6 forms complex with E6AP and p53; and the unique dual-functional PDZ binding motif (PBM) at its extreme C-terminal end, the region where E6 binds to PDZ proteins (Dlg-1, MAGI and Scribble), and which is also a target for PKA/AKT phosphorylation, which switches binding affinity to the 14-3-3 protein family. The lower panel shows the Aurora B recognition motif [R/K]1-3-X-[S/T] and the key amino acids within HPV16 and 18E6 predicted to be recognised by Aurora B: HPV16E6 at amino acids S89, T140, S145 and S150 of HPV16E6; HPV18E6 at amino acid S59. Figure S2. Subcellular colocalization of E6 with AurB predominantly in the nucleus. Immunofluorescence staining was