GSK3β mediates the carcinogenic effect of HPV16 in cervical cancer

Cervical cancer is one of the most prevalent and fatal cancers among women and infection of the human papillomavirus (HPV) is the most important risk factor. This study investigated how HPV16 regulated GSK3β expression and function to promote cervical cancers. The expression of GSK3β was analyzed by quantitative PCR and western blot. The proliferation, invasion, and clonogenic survival of cells with different E6/E7 levels were measured by MTT, transwell invasion assays, and soft agar colony-forming assays, respectively. The levels of GSK3β were correlated with the copy numbers and expression levels of HPV16 E6/E7 genes. HPV16 E6/E7 genes regulated GSK3β transcription through an element located in the promoter 85 and 250 base pairs upstream of the transcription start site. The abilities of cell proliferation, invasion, and clonogenic survival were increased in C33A cells by ectopic HPV16 E6/E7 and decreased in CaSki cells by knocking down HPV16 E6/E7 levels. Meanwhile, LiCl increased GSK3β transcript levels and the proliferation of CaSki cells in a HPV16-dependent manner. These data indicated that GSK3β may participated in HPV16 mediated deregulation of wnt/β-catenin and other signaling pathways promoting the progression and invasion of cervical cancers.

The β -catenin protein has been shown to be abnormally increased or localized in cervical cancer cells [18][19][20] , which has been associated with the progression of cervical cancer 7,19,20 . HPV oncoproteins have been shown to increase β -catenin accumulation by interfering with the β -catenin destruction complex or inhibiting its degradation by proteosome [21][22][23][24] . However, whether GSK3β plays a role in mediating HPV mediated perturbation of wnt/β -catenin in cervical cancer is not known. This study aimed to investigate whether GSK3β is a target of high risk HPV in cervical cancer.

Results
The GSK3β level and HPV copy number were correlated in cervical cancer cells. To check if HPV infection had any effects on the expression level of GSK3β , its mRNA and protein levels in C33A (HPV negative), Siha (1~2 copies of HPV16), and CaSki (~600 copies of HPV16) were analyzed by real-time qPCR and western blot. CaSki cells had the highest level of GSK3β mRNA while C33A cells had significantly lower GSK3β mRNA level compared to Siha cells, which was highly correlated with the expression level of HPV16 E6 (Fig. 1A-C). The Serine 9 phosphorylated and total GSK3β protein levels were also correlated with the HPV16 E6 protein level in these cells (Fig. 1D).
HPV16 oncoproteins E6/E7 upregulated GSK3β mRNA level in cervical cancer cells. Ectopic expression and knockdown of HPV16 E6 and/or E7 were employed to investigate whether HPV16 oncoproteins E6/E7directly regulated the expression of GSK3β . The mRNA level of GSK3β in C33A cells was more than doubled with HPV16 E6 or E7 transfection and increased more than 5-fold with E6 and E7 double transfection ( Fig. 2A). Accordingly, GSK3β protein exhibited a pattern similar to mRNA that was significantly increased by HPV16 E6 and E7 (Fig. 2B). On the other hand, knocking down HPV16 E6 and E7 in CaSki cells with shRNA significantly reduced the mRNA (  HPV16 E6/E7 increased the proliferation of cervical cancer cells. The proliferation rate of C33A cells was increased 14.5%, 17.1%, and 40.2% by exogenous HPV16 E6, E7, and E6/E7, respectively (Fig. 3A). On the other hand, silencing HPV16 E6 and E7 in CaSki cells resulted in a 22% decrease of proliferation (Fig. 3B).

HPV16 E6/E7 increased the metastatic ability of cervical cancer cells. The effects of HPV16
on the invading and colony-forming abilities of cervical cancer cells were evaluated by overexpressing HPV16 E6 and/or E7 in C33A cells or knocking down E6/E7 in CaSki cells. Exogenous HPV16 E6 or E7 significantly increased the numbers of invading cells (Fig. 4A,B) and colonies (Fig. 5A,B). Simultaneously overexpressing E6 and E7 further increased the numbers of invading cells (Fig. 4A,B)  HPV16 E6/E7 regulated GSK3β transcription. Cotransfection of HPV16 E6 and/or E7 with the human GSK3β promoter luciferase reporter gene (pGSK3β 1130) increased GSK3β promoter activity 5 to 9 fold (Fig. 7A). Deleting the promoter from − 1130 bp to − 250 bp did not significantly reduce human GSK3β promoter activity but further deleting the promoter to − 85 bp resulted in the loss of a majority of the promoter activity (Fig. 7B).

Discussion
The GSK3β expression including Ser9 phosphorylation was correlated with the HPV16 copy number and the expression of oncoprotein E6/E7. HPV16 E6/E7 regulated GSK3β transcription through a region between 85 bp and 250 bp upstream of the transcription initiation site of human GSK3β gene. Knocking down HPV16 E6/E7 reduced GSK3β expression in CaSki cells while overexpressing HPV16 E6 and/or E7 increased GSK3β expression in C33A cells. The proliferation, invading, and colony-forming abilities of cervical cancer cells were associated with HPV16 E6/E7 and GSK3β levels. A common GSK3β inhibitor LiCl increased GSK3β expression levels in CaSki cells, which was abolished by silencin HPV16 E6/E7 with shRNAs.
GSK3β serves as a master molecule for transducing various signals including Wnt and Notch to regulate cell cycle progression, cell differentiation, proliferation, and cell death 25 . In autochthonous transgenic prostate cancer TRAMP mice, inhibition of GSK3β resulted in reduction of tumor number and size through the derepression of C/EBPα and inhibition of E2F expression 16   resulted in reduced cell growth and colony-formation 15 . However, whether the inhibition of GSK3β could sensitize pancreatic cancer to gemciatbine treatment was inconclusive 15,26,27 .
This is the first study to investigate the relationship between high-risk HPV and GSK3β in cervical cancer. In cervical cancer cells, HPV16 E6/E7 promoted the expression of GSK3β at least in part through transcription. HPV16 E6 augmented the transcription activities of β -catenin responsive genes in HEK293 cells and cyclin D1 protein levels were correlated with HPV copy number in cervical cancer cells (C33A, SiHa, CaSki, and HeLa) 23 . In COS7 cells and primary human foreskin keratinocytes, HPV16 was shown to interact with DVL2 and β -catenin and enhanced the expression of c-myc and other β -catenin target genes 22 . HPV16 E6/E7 increased nuclear accumulation of β -catenin in oropharyngeal cancer cells by inhibiting the mediator of ubiquintination, Siah-1 21 . These results demonstrated that oncoprotein HPV E6/E7 regulated β -catenin protein level and nuclear accumulation at multiple levels and with different pathways. Our data presented a complicated picture regarding the effects of E6 and E7 oncoproteins of high-risk HPV on the regulation of GSK3β (Fig. 8), which in turn could modulate cellular β -catenin levels. HPV16 E6/E7 increased GSK3β transcripts by up-regulating GSK3β gene transcription, which resulted in the increase of steady-state GSK3β protein and Ser9 phosphorylated GSK3β protein levels.
Lithium chloride (LiCl) has been widely used to stimulate Wnt/β -catenin signal or regulate other pathways by deactivating the kinase activity of GSK3β [28][29][30][31] . LiCl treatment resulted in a significant increase of both mRNA and protein levels of GSK3β in CaSki cells, which was blunted by a knockdown of HPV16 E6/E7 with shRNA. The effect of LiCl on GSK3β was HPV16 E6/E7 dependent as LiCl was failed to increase GSK3β mRNA level in C33A cells. Moreover, as the increased level of GSK3β protein by HPV16 E6/E7 was in Ser9 phosphorylated state, the net effect would be the activation of wnt/β -catenin pathway 28 or inflammation pathway 31 . As conventional wisdom believed that LiCl could only inhibit the kinase activity of GSK3β 28-31 , it was surprising that LiCl increased GSK3β mRNA level in HPV16 positive cervical cancer cells, which might have resulted from increased E6/E7 transcriptional activity by LiCl or GSK3β inactivation causing the activation of other signal pathway which cooperated with E6/ E7. We are currently investigating the exact mechanisms underlying HPV16 E6/E7-dependent GSK3β mRNA level increase caused by LiCl.
The level of GSK3β expression correlated with copy numbers and expression levels of oncogenes of high-risk HPV in cervical cancer cell lines. The GSK3β level could be increased by ectopically expressing HPV16 E6/E7 in HPV negative C33A cells or reduced by knocking down E6/E7 level in HPV16   For LiCl treatment, C33A and CaSki cells were seeded into 6-well plates and cultured overnight before being treated with or without 10 mM LiCl for 24 hr. In the case of E6/E7 knockdown, CaSki cells were seeded into 6-well plates and infected with pLL3.7-shE6/pLL3.7-E7 or pLL3.7-shcontrol viruses for 36 hrs before treated with 10 mM LiCl for another 24 hrs. At the end of 24 hr LiCl treatment, cells were collected for MTT assay or RNA and protein works. CCTTCCT as the 3′ primer from 293T genomic DNA and cloned into KpnI and BglII (TaKaRa Bio, Shiga, Japan) digested pGL3-basic vector (Promega, Madison, WI).
The HPV16 E6 and E7 coding sequences were amplified from the cDNA generated from CaSki total RNA with primers GAAGATCTATGTTTCAGGACCCACAGG and CGGAATTCTTAC AGCTGGGTTTCTCTAC for E6 and GAAGATCTATGCATGGAGATACACCTAC and CGGAATTCTT ATGGTTTCTGAGAACAG for E7. The PCR products were digested with BglII and EcoRI and inserted into pMSCV-puro, which was digested with the same enzymes.
The short hairpin sequences were designed to target HPV16 E6 and/or E7 expression based on previously published sequences 32 . Oligonucleotides TGCACAGAGCTGCAAACAACTCAAGAGAGTTG TTTGCAGCTCTGTGCTTTTTTC and TCGAGAAAAAAGCACAGAGCTGCAAACAACTCTCTTGA GTTGTTTGCAGCTCTGTGCA for E6 and TGACAGAGCCCATTACAATATCAAGAGATATTGT AATGGGCTCTGTCTTTTTTC and TCGAGAAAAAAGACAGAGCCCATTACAATATCTCTTGAT ATTGTAATGGGCTCTGTCA for E7 were synthesized by Life Technologies (Shanghai, China), annealed, and inserted into HpaI/XhoI digested pLL3.7 (MIT, Cambridge, MA). All constructs were verified by DNA sequencing (Life Technologies).
The retroviruses and lentiviruses were produced and titered by Biowit Technologies (Shengzheng, China).
MTT assay. The cells were seeded in 96-well plates a day before being infected with retroviruses or lentiviruses. 48 hr after the virus infection, the cell culture medium was replaced with 100 μ l fresh growth medium containing 10 μ l of 12 mM 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, Life Technologies) and continued incubation at 37 °C for 4 hr, and then the medium was replaced with 150 μ l of DMSO. The plate was protected against light and incubated at room temperature for 10 min with shaking before being read at 490 nm on a plate reader (Molecular Devices, Sunnyvale, CA).
Cell invasion assay. The cells were trypsinized and adjusted to a concentration of 5 × 10 4 cells/ml with serum-free medium. 200 μ l cells were seeded in the upper chambers of 12-well Transwell plates (BD BioSciences, San Jose, CA) and 600 μ l of complete growth medium containing 10% FBS was put into the lower chambers. After culturing for 24 hr at 37 °C, the cells were fixed with methanol for 15 min at room temperature and stained with Giemsa solution for 15 min before rinsing with PBS and being wiped off with Matrigel (BD BioSciences) on the upper surface. The number of migrated cells was counted from 5 random fields (200x).

Colony formation assay.
A single cell suspension was diluted to a concentration of 10 3 cells/ml with complete growth medium and 100 μ l of the cell suspension was mixed with 0.3% agar Noble agar (DIFCO, Detroit, MI) and was seeded onto a layer of solidified agar which was previously prepared in a 24-well cell culture plate (Corning, Tewksbury, MA). The plate was incubated at 37 °C until clones were formed. Statistical analysis. All experiments were independently performed in triplicates. The data were expressed as mean ± standard deviation. SPSS 16.0 statistical software was used for data analysis of variance (ANOVA) or independent sample t-test. A P value less than 0.05 was considered statistically significant.