Human cervical cancer cell lines C33a (HPV-negative), Siha and Caski (HPV-positive) were purchased from the American Type Culture Collection (Manassas, VA). Cells were transduced with CMV-Fluc-IRES-RFP lentiviral particles (GeneChem, Shanghai, China). The cells with stable transfection (RFP⁺) were isolated by FACS. RFP/luciferase-expressing cells were used in living imaging. In cell cycle analysis by flow cytometry, RFP-free cells were used.

The expression vectors for E6 and E7 were constructed as previously described (10). Briefly, the cDNAs coding for E6 and E7 were amplified from total RNA isolated from Caski cell lines using following primers: E6, 5′-GCGGCCGCC ACCATGTTTCAGGACCACAG-3′ (sense) and 5′-CTGCGG CCGCGATTACAGCTGGGTTTTCTCT-3′ (antisense); E7, 5′-GCGGCCGCCACCATGGCATGGCATGGAGATAC ACCT-3′ (sense) and 5′-AGGCGGCCGCGATTATGGTTT CTGAGAACA-3′ (antisense). The cDNAs were inserted into expression vector pcDNA3.1. C33a cells were transfected with plasmid pcDNA3.1-SIX1 (a kind gift from Kong-Ming Wu, Thomas Jefferson University, Philadelphia, PA), pcDNA3.1-E6, pcDNA3.1-E7 or pcDNA3.1 using Lipofectamine 2000 (Invitrogen, Life Technologies). ShSIX1(1) and shSIX1(2) shRNA lentiviral particles (GeneChem) were used to knockdown SIX1 expression, targeting 5′-CCAGCTCAGAAGAGG AATT-3′ and 5′-CACGCCAGGAGCTCAAACT-3′, respectively. Shcon, the shRNA not targeting any known gene, was used as control. Caski cells were transduced with SIX1 shRNA lentiviral particles, or control shRNA lentiviral particles. The cells with stable transfection were selected with G418 (C33a) and puromycin (Caski). The transfected cells were designated as C33a-3.1, C33a-E6, C33a-E7, C33a-SIX1, Caski-shcon, Caski-shSIX1(1) and Caski-shSIX1(2), respectively.

A tissue microarray (TMA) (CIN481) was purchased from Alenabio (Xi-an, China), containing cervical intraepithelial neoplasia (CIN) samples. Tumor samples were acquired by surgery from cancer patients without preoperative treatment, which was approved by the Ethics Committee of the Medical Faculty of Tongji Medical College. Informed consent was obtained from all subjects. The tumor samples were processed into TMA at Outdo Biotech Company (Shanghai, China). TMAs and tissue sections were subjected to immunohistochemical analysis as described previously (11), using the Avidin-Biotin Complex (ABC) Vectastain Kit (Zsgb-Bio, Beijing, China). Rabbit anti-human SIX1 antibody (Sigma-Aldrich) and rabbit anti-human Ki67 antibody (Santa Cruz Biotechnology) were used as primary antibodies. For semi-quantitative evaluation of SIX1 expression in tissue, staining intensity was graded (0, absence; 1, weak; 2, moderate; 3, strong), and assessed using an immunoreactivity-scoring system, HSCORE (12). The HSCORE was calculated using the following formula: HSCORE = ∑Pi(i + 1), where i is the staining intensity of tumor cells and Pi is the percentage of tumor cells at each level of intensity. The percentage of Ki67-positive tumor cells was measured as mean of five hot-spots on one slide from each tumor. Hot-spots were defined as areas with the highest Ki67 intensity. The slides were read by two pathologists and each data point represents the mean of the scores from two pathologists.

Total RNA was extracted from cells with TRIzol reagent (Invitrogen). The relative quantity of mRNA was determined by real-time RT-PCR. Gapdh and EEF1A1 were chosen as reference genes, which were reported as the most reliable combination in cervical cancer (13). The primer sequences are shown in Table I. The expression of genes was quantified using the comparative C_T method. The expression level of each mRNA was normalized to that of GAPDH and EEF1A1 mRNAs, and expressed as n-fold difference relative to the control.

Western blot assay was performed as described previously (14). The antibodies were purchased from Sigma-Aldrich, Biorbyt and Santa Cruz Biotechnology.

The normalized RNA-sequencing data of 116 cases of cervical cancer samples was publicly available from TCGA database (https://tcga-data.nci.nih.gov/tcga/tcgaHome2.jsp). Differential gene expression based on SIX1 expression category was determined using Bioconductor edgeR package (15) as described previously (16). Then, pathway analysis of differentially expressed genes was performed using Onto-Express from Onto-Tools package (17).

To further analyze the effect of SIX1 expression on DNA replication and cell cycle pathways (gene sets), Gene set enrichment analysis (GSEA) package was used as described previously (18). For this analysis, SIX1 expression was set as continuous label. Heat map was generated by GSEA.

The BrdU colorimetric ELISA kit (Roche) was used to detect 5′-bromo-2′-deoxyuridine (BrdU) incorporated into newly synthesized DNA, according to the manufacturer’s protocol. Briefly, tumor cells were incubated with BrdU solution for 2–16 h at 37°C, then fixed and denatured and incubated with a peroxidase-conjugated antibody against BrdU for 1 h at room temperature. The cells were then incubated with the tetramethyl-benzidine (TMB) substrate solution and absorbance was read at 450 nm using a multiskan spectrum microplate reader (μQuant Bio-Tek Instruments, USA).

Cells were harvested by trypsinization and fixed in 70% ice-cold ethanol overnight at −20°C, and then washed in PBS and incubated in RNase A (10 U/ml) and propidium iodide (20 mg/ml) for 30 min at room temperature. The cells in different phases were detected by flow cytometry on a FACS Calibur system (Becton-Dickinson) and analyzed using FlowJo software. Serum starvation was performed as described previously (19).

Tumor cells were seeded at 2×10³ per well into 96-well plates. Using cell counting kit-8 (CCK8, Boster, China), the proliferation of the cells at each time point was measured with multiskan spectrum microplate reader (μQuant Bio-Tek Instruments, USA) at 450 nm wavelength. Cell proliferation index was expressed as fold change related to day 0.

Cells were suspended in 0.35% agar in DMEM (20% FBS) and plated (1.5×10³ per well) on a layer of 0.5% agar in culture medium in 6-well plates. After two-week culture, colonies were counted and photographed by phase-contrast microscopy (Nikon, Japan). The size of 20 randomly chosen colonies per well was measured. The average size of colony was calculated as (length + width) × 0.5. The colony formation rate was calculated as the number of colonies/1,500 ×100%.

Female NOD-SCID mice (4 weeks old) were purchased from Beijing HFK Bio-Technology Co. Ltd. (Beijing, China). The studies were approved by the Committee on Ethics of Animal Experiments of Tongji Medical College. The mice were maintained in the accredited animal facility of Tongji Medical College. Intramuscular tumors were initiated by injecting 1×10⁶ tumor cells in 50 μl of media into the hind limb of NOD-SCID mice. The sizes of ectopic tumors were monitored by a caliper and calculated by the formula: volume = (width)² × length/2. Orthotopic xenograft model of cervical cancer was performed as previously described (20). Briefly, intramuscular tumors were used as donor tumor fragments for orthotopic transplantation into the cervix of NOD-SCID mice. Tumor growth in cervix was dynamically monitored in living mice by optical imaging of luciferase activity using the IVIS Spectrum system (Caliper, Xenogen, USA). After anaesthetization with 3% pentobarbital sodium, mice were imaged 10 min after intraperitoneal injection of 100 mg/kg D-luciferin. Tumor size was measured by three-dimensional reconstruction of image using Living Image version 4.3.1 software.

Each experiment was repeated at least three times independently. SPSS (version 13.0) software package was used for statistical analysis. Results are expressed as mean value ± standard error of the mean (SEM), and interpreted by one-way ANOVA. Differences were considered to be statistically significant at P<0.05.

To determine the role of SIX1 in cervical cancer, we analyzed the expression of SIX1 in each stage of cancer development. SIX1 expression was negligible in normal cervical epithelium, but was significantly increased in low-grade cervical intraepithelial neoplasia (CIN I, precancerous lesion) (Fig. 1A). Higher expression levels of SIX1 were observed in almost all cases of high-grade cervical intraepithelial neoplasia (CIN II-III). SIX1-expressing cells shared the same distribution area with the extent of intraepithelial neoplasia (Fig. 1A). In three cervical squamous cell carcinoma cell lines, the expression of SIX1 was very low in HPV-negative C33a cells, slightly elevated in Siha cells which contain 1–2 copies of HPV DNA, and much higher in Caski cells which contain 60–600 copies of HPV DNA (Fig. 1B), suggesting that HPV infection increased the expression of SIX1. We then further tested whether SIX1 was induced by E6 or E7 oncoprotein, the driving force of cervical cancer. Overexpression of E7, but not E6, in HPV-negative C33a cells induced SIX1 expression (Fig. 1C). Consistently, when E7 were knocked down in C33a-E7, Siha or Caski cells, the expression of SIX1 was significantly decreased (Fig. 1D). These results suggested that HPV-E7 is one of the factors that are required for inducing the expression of SIX1. Moreover, the correlation of SIX1 expression to the progression of cervical cancer suggested that SIX1 might play an important role in tumorigenesis and progression of cervical cancer.

To identify the effect of SIX1 on the development of cervical cancer, bioinformatic analysis was performed to predict the effect of SIX1 on gene expression in cervical cancer. The RNA-sequencing data of 116 cervical cancer specimens was obtained from TCGA database. The specimens were divided into SIX1-low expression group (n=58) and SIX1-high expression group (n=58). Bioconductor edgeR package was used to identify the genes which were significantly up-regulated in SIX1-high expression group. Subsequent pathway analysis indicated five pathways which were significantly enriched in the genes potentially upregulated by SIX1 (Fig. 2A). Two of the pathways were correlated with cell proliferation: DNA replication and the cell cycle. Further analysis by using GSEA package confirmed that the enrichment of the upregulated genes related to DNA replication was significantly correlated with the increase of SIX1 gene expression (Fig. 2B, P<0.05, FDR<0.25). The enrichment of the upregulated genes related to cell cycle was correlated, although without statistical significance, with the increase of SIX1 gene expression (Fig. 2C). Importantly, the analysis with both edgeR-Onto and GSEA methods showed that the upregulated genes included many genes related to DNA replication and the cell cycle, and that these genes could be considered as candidate genes which might be up-regulated by SIX1 (Fig. 2D). These results implied that SIX1 may regulate the expression of the genes involved in DNA replication to promote tumor cell proliferation.

Based on the above results, we overexpressed SIX1 in C33a cells (Fig. 3A) to investigate the regulatory effect of SIX1 on DNA replication genes (Fig. 3B). SIX1 expression was knocked down in Caski cells (Fig. 3C) to further confirm the effect of SIX1 (Fig. 3D). The results showed that SIX1 could upregulate the expression of several genes coding for the proteins in five complexes related to DNA replication, including MCM complex (MCM2, MCM3, MCM6), DNA polymerase α-primase complex (POLA1, PRIM1, PRIM2), clamp loader (RFC3, RFC4, RFC5), DNA polymerase δ complex (POLD3) and DNA polymerase ɛ complex (POLE2). These five complexes play critical roles in three important steps of DNA replication initiation: DNA unwinding, synthesis of primer, and switching from DNA polymerase α to polymerase δ or polymerase ɛ (Fig. 4A). Consistently, BrdU incorporation assay indicated that DNA synthesis was more active in SIX1 highly expressed cervical cancer cells (Fig. 4B).

In accordance with the potential of SIX1 to promote DNA replication, overexpression of SIX1 in C33a cells resulted in a decrease of the percentage of cells in G0/G1 phase and a concomitant increase of the percentage of cells in S phase (Fig. 5A), indicating more active cell proliferation. After synchronization by serum starvation and following serum addition, C33a-SIX1 cells progressed into S phase faster than corresponding control cells (Fig. 5B), and knockdown of SIX1 in Caski cells resulted in the opposite effect (Fig. 5C and D). These results demonstrated that SIX1 modulated the cell cycle and promoted G1 to S phase progression.

We next investigated whether SIX1 could promote tumor cell proliferation. Overexpression of SIX1 in C33a cells significantly, although slightly, promoted anchorage-dependent cell growth, while knockdown of SIX1 in Caski cells decreased anchorage-dependent cell growth (Fig. 6A). Intriguingly, in colony-forming assays in soft agar, the tumor cells with highly expressed SIX1 showed remarkably larger size colonies (Fig. 6B) and much higher colony formation rate (Fig. 6C), indicating that SIX1 could more efficiently promote anchorage-independent cell growth.

Expression of the proliferation marker Ki-67, was significantly higher in SIX1 highly expressed cervical cancer specimens (Fig. 7A). Comparing with corresponding control groups, the percentage of Ki67-positive tumor cells was increased in C33a-SIX1 tumors and decreased in Caski-shSIX1 tumors (Fig. 7B), indicating more active cell proliferation in SIX1 highly expressed tumors. In line with this, in both ectopic (Fig. 7C) and orthotopic (Fig. 7D) xenograft models of cervical cancer, overexpression of SIX1 in C33a cells significantly promoted tumor growth, while knockdown of SIX1 in Caski cells decreased tumor growth. Taken together, these results demonstrated that SIX1 could promote not only tumor cell proliferation in vitro, but also tumor growth in vivo.