The Hybrid Capture^® 2 system (hc2, Digene Corp., USA) was purchased by the Cancer Hospital affiliated to Xinjiang Medical University. The second-generation hybrid capture 2 (hc2) assay is the only method currently approved for clinical hrHPV DNA testing by the US Food and Drug Administration (FDA) and the China Food and Drug Administration (CFDA). A total of 37 HPV genotypes including 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68, 6, 11, 42, 43, 44, 81, 53, 26, 34, 40, 54, 55, 57, 61, 67, 69, 70, 71, 72, 73, 82, 83 and 84, and multiple infections could be accurately detected with both a sensitivity and specificity > 95%. Especially, based on the guidelines used for clinical diagnosis, the method can detect 13 HR-HPV types (-16, -18, -31, -33, -35, -39, -45, -51, -52, -56, -58, -59 and -68) or 5 LR-types (-6, -11, -42, -43 and -44), as well as the HPV DNA load. The assay was a non-radioactive signal-amplification method based on the hybridization of the target HPV-DNA to labeled RNA probes in solution, utilizing signal amplification of antibody capture and detection of chemiluminescence signals in 96-well plates.

Two groups (HSIL, CSCC) of HPV16/18-positive patients with cervical lesions (a total of 31 cases) and the cervical tissue specimens without HPV16/18 lesions or other HPV subtypes (NC) as the controls were selected from 60 specimens previously collected. RT² profiler PCR array gene expression analysis was used for differential analysis of TLR signaling pathways (QIAGEN, Cat. no. PAHS- 018Z), as well as innate and adaptive immune response pathways (84 genes) (QIAGEN, Cat. no. PAHS-052Z) ( Tables S2A, B ) by using the human tissue sample described above. PCR arrays were used to analyze the differential expression of receptors, virus-specific response molecules, pathway regulatory molecules, upstream and downstream effector molecules in innate and adaptive immune response, Th1, Th2, Th17, Treg marker molecules, inflammatory response and anti-viral immune response molecules. The operational steps were performed according to Qiagen’s operating procedures. The protocol can be briefly described as follows: Total RNA was extracted from tissue using a RNeasy Microarray Tissue Mini Kit and purified (RNeasy@ MinElut™, Qiagen) to ensure RNA quality. Reverse transcriptase MLV-RT was used to synthesize the cDNA using 1.0 µg of total RNA as a template. 2 x SYBR Green Fluor qPCR Master Mix was prepared according to the requirements of the assay. Subsequently, sample cDNA and MasterMix were added to 96 wells, respectively in which 84 genes, 5 housekeeping genes and 7 controls were detected. Online data analysis software was used to analyze the original CT values obtained by real-time quantitative PCR to form a scatter map, and the RNasy microarray tissue mini-kit map was grouped.

HFF-1, and Caski cells were purchased from the Cell Bank of Chinese Academy of Sciences. H8 and SiHa cells were kindly provided by Prof. Abulizi Abudula. SiHa, Caski and H8 cells were cultured in RPIM1640 media (Gibco, US) and HFF-1 cultured in DMEM (HyClone, US) supplemented with 10% fatal bovine serum (Biological Industries, Israel), and 100 U/ml penicillin/streptomycin (Biological Industries, Israel). Cells were cultured in a 5% CO₂ incubator at 37°C. When cultures became confluent, the cells were treated with 0.25% trypsin (Gibco, US) containing 0.02% EDTA for 2 min at room temperature.

CRISPR/Cas9 single guide sequences specifically targeting HPV16 infection marker E6/E7 (sg E6/E7) were designed in Vector Builder (www.vectorbuilder.com) and produced by the Cyagen Company. The E6/E7 gRNA guide sequence used were 5’-CAACAGTTACTGCGACGTG-3’ and 5’-TCCGGTTCTGCTTGTCC AGC-3’. SiHa cell lines were transduced twice with first sgRNAE6/E7 lentiviruses, EGFP lentiviruses as the positive control and transfection reagent as the negative control. Puromycin (Merck, Germany) was added on day 4 at a minimal toxic dose to select transduced cells. Then after 3 days of screening, the condition of the cells was observed, and a second transduction performed with Cas9 lentiviruses. Hygromycin at a minimal toxic dose was added after 3 days. The second screening finished after all negative control cells had died. To minimize off-target errors, monoclonal screening was conducted. Single colonies were spread in a 96-well plate and selected by a colony formation assay. The effect of E6/E7 gene editing on cell proliferation was assessed by Trypan Blue viable cell counting over a 7-day time course. The cell apoptosis induction and proliferation inhibition of Ko E6/E7 SiHa cells were markedly induced in vitro, which were also cultured in RPIM1640 media (Gibco, US) containing 10% FBS in a 5% CO₂ incubator at 37°C. Lentiviral transduction efficiency was proven by employing RT-qPCR and western blotting techniques. Total protein from SiHa cells was extracted using RIPA buffer (Solabio, China) and protein expression was determined by western blotting, probed for E6/E7 (Abcam; ab70, ab30731), GAPDH (Yeason, China) using sensitive ECL detection (Yeason, China). Total RNA extraction and the methods of RT-qPCR are described below and the E6 and E7 primers used are listed in Table S3 .

The diagram illustrating the vector map used to generate transgenic mice was shown in Figure S1A . The gene for HPV16 E6/E7 was placed under the control of the K14 promoter using standard methods. The plasmid DNA employed was purified using the method of plasmid midiprep and then used for microinjection into fertilized eggs of C57BL/6×C57BL/6 hybrid mice. The transgenic (Tg) mice were identified by PCR (Transgene PCR primer F1: TCACTCAGCCAACTGCTCGC; Transgene PCR primer R1: GTCGCAGTAACTGTTGCTTGCAG) ( Figure S1B ). Eight founder lines were established by breeding into C57BL/6×C57BL/6 hybrid mice. The line with the highest transgene expression was selected for further analysis. Most experiments were carried out using heterozygous mice that were housed and fed under specific pathogen free environment. In order to identify the phenotype of the cervix tissues, the tissues from the HPV16 E6/E7 transgenic mice were analyzed histologically. The procedures for care and use of animals were approved by the Ethics Committee of the Shenzhen Hospital of Southern Medical University and all applicable conditions according to strict institutional and governmental regulations concerning the ethical use of animals were followed.

Three copies of SiHa and KoE6/E7 SiHa cell specimens were dissolved in Trizol, and the RNA-seq was determined by the Novo Nordisk Company. The analysis results were evaluated with the aid of the KEGG library. After RNA extraction, the concentration and quality of the total RNA was determined using agarose gel electrophoresis and the RNA stripe was found to be complete, which indicated no degradation. The total RNA purity (OD 260/280) was detected by Nanodrop and the total RNA accurately quantified using Qubit.

Tissue microarrays were obtained from Outdo Biotech Co., Ltd. (Shanghai, People’s Republic of China). IHC studies of PD-L1 were performed on cervical cancer and normal specimens of tissue in a microarray using anti-PD-L1 antibodies (Proteintech). Quantitative analysis of the staining was conducted based on the percentage of positive cells and staining density by three experienced pathologists. For data analysis, staining scores < 2 were defined as low expression and scores > 2 as high expression.

After tissues had been collected and fixed, they were typically dehydrated and embedded in melted paraffin wax, and finally cut into thin slices. The slices were affixed to microscope slides at which point the wax was removed with a solvent and the tissue slices were rehydrated. Arrays were dewaxed and then doused with endogenous peroxidase 3% hydrogen peroxide. Followed by blocking using PBS supplemented with 10% goat serum at 30mins, the specific antibody, E6 (Bioss,bs-1719R) and E7 (Bioss, bs-4623R),were incubated at 4°C overnight, and the secondary antibody was incubated at room temperature for 30 min. Cells were washed with PBS, and then incubated with the appropriate secondary antibody for 30 min at room temperature. Then DAB was added for 10 min incubation and cells were finally stained with hematoxylin. Cells were viewed and photographed using a microscope.

Total RNA was isolated using TRIzol reagent (Invitrogen, US) and cDNA was generated using the Hifair^® II 1st Strand cDNA Synthesis SuperMix (Yeason, China). qPCR was performed using a 7500 system (ABI, USA) with Hieff qPCR SYBR Green Master Mix (Yeason, China). The relative expression levels of the target genes were normalized to the expression level of the internal control GAPDH. The data analyses were performed using the 2^−ΔΔCt method and the primer sequences are listed in Supplementary Table 3 (Table S3) .

Cultured cells were lysed with lysis buffer. Equal amounts of protein were run on 10% SDS-PAGE and electrotransferred to polyvinylidene fluoride membranes (Millipore). After blocking with 5% milk in TBST, membranes were incubated with primary antibodies overnight. The following antibodies were used: anti-HPVE6 (1:1000, Arigo), anti-HPVE7(1:1000, Bioss), (anti-CXCL10 (1:1,000, Abcam), anti-CXCR3 (1:1,000, Boster), anti-PDL1 (1:2,000, proteintech), anti-STAT1 (1:1,000, proteintech), anti-pSTAT1 (1:1,000, Abcam), anti-JAK1 (1:2,000, proteintech), anti-GAPDH (1:6,000, Yeasen) and anti-tublin (1:3,000, Yeasen). Membranes were then incubated with the rabbit peroxidase-conjugated secondary antibody (1:10,000, Abclonal). The blots were detected by sensitive chemiluminescence liquid analysis (Yeasen) and Biorad software was used to capture the images.

The siRNA transfection referred to Ribo Company transfection guidance, and the target sequence of HPVE6 siRNA was GTAGAGAAACCCAGCTGTA. 12 – 16 h prior to transfection, 2 × 10⁵ cells per well were planted on 24-well culture plates. Then a mixture of buffer, siRNA and reagent was prepared and incubated at room temperature for 15 min before being added to cells. After 36 – 48 h post-transfection, the cells were further collected in lysate buffer for western blot analysis.

For transwell co-culture assay, 0.4 μm PC transwell (Corning) was used to do this experiment. The SiHa or Caski cells were planted in upper chamber at a density of 2-3 × 10⁵ cells per well and HFF-1 or Ko E6/E7 SiHa cells were planted in the lower chamber at a density of 2 × 10⁵ cells per well. When all cells were adherent, the cells in the upper chamber and lower chamber were co-cultured together for 48 h though transwell assay. Furthermore, the CXCL10 (AMG487, MCE) (26) and JAK1-STAT1 (Ruxonitilib, MCE) inhibitor (28) were added into the upper chamber at a concentration of 1 μM for 12 h treatment And Recombinant Human CXCL10 (Novoprotein, CX32) were added to the culture media directly at a concentration of 100 ng/ml for 48 h.

Rat IP-10/CXCL10 (Interferon Gamma Induced Protein 10 kDa) ELISA Kit (Elabscience) analysis was performed according to manufacturer’s instructions. For detection of PD-L1 on extracellular vesicles, cell supernatants, 100 µl of extracellular vesicle samples purified from cell culture supernatants, were added to ELISA plates (96-well) (Elabscience) and left to incubate for 1 h and then 100 μL of Biotinylated Detection Ab Diluent against PD-L1 was added and allowed to incubate for 1 h at 37°C. After discarding the liquid in the plate, 100 μL of HRP conjugate was added to each well for a 0.5 h incubation time at 37°C and then 90 μL of substrate was added to each well for 15 min. Plates were stopped with 50 μL of stop solution. The plates were read at 450 nm on a BioTek plate reader and the concentration of ip-10/CXCL10 was in direct proportion to the OD 450 value. The concentration of ip-10/CXCL10 in a sample was calculated against a standard curve.

For CCK8 assay, the cells were plated at a density of 3,000 cells per well in 96-well plates. After adhesion to wall, 10 µl CCK8 was added into the 100 µl cell culture medium and the cells were incubated for 2 h at 37°C. Then cell viability was determined at a 570 nm wavelength using the CCK-8 assay according to the manufacturer’s instructions (Meilunbio, MA0218-3) ( Figure S3A ).

Apoptosis was assessed by Annexin V staining and flow cytometry analysis. Briefly, 5 × 10⁵ SiHa or KoE6/E7 SiHa cells were harvested, washed in PBS, and then analyzed by Annexin V/propidium iodide staining by flow cytometry according to the manufacturer’s protocol (Annexin V-APC PI kit; Biolegend) ( Figure S3B ).

The tumor model was established by subcutaneous (s.c.) injection of 5 × 10⁵ SiHa or KoE6/E7 SiHa cells in 120 µl PBS into the flank of nude mice. Tumor volumes were determined from caliper measurements of tumor length (L) and width (W) according to the formula L × W^2/2. Both tumor size and body weight were measured 3 times each week ( Figure S3C ). At the end of the experiment, tumors were removed and fixed in 4% paraformaldehyde, then the structure of the tissue was determined by means of HE staining and immunohistochemistry ( Figures S3D, E ).

For exosome purification from cell culture supernatants, cells were cultured in media supplemented with 10% exosome depleted FBS. Supernatants were collected from 48 – 72 h cell cultures and the extracellular vesicles were purified by a standard differential centrifugation protocol (29). In brief, culture supernatants were centrifuged at 2,000 g for 20 min to remove cell debris and dead cells (Beckman Coulter, Allegra X-14R). Microvesicles were pelleted after centrifugation at 10,000g for 30 min (Beckman Coulter, J2-HS) and resuspended in PBS. Supernatants were then centrifuged at 100,000 g for 2 h at 4°C (Beckman Coulter, Optima XPN-100) to collect exosomes and then 100,000 g centrifugation for 2 h was repeated and exosomes washed. The exosomes were dissolved in PBS and stored at -20°C or -80°C.

In addition, approximately 1 ml blood was collected from the heart of mice in order to obtain exosomes from per mouse, using anticoagulation tubes to prevent blood clotting in 17^th day after subcutaneous (s.c.) injection. Immediately, the blood was centrifuged at 3,000 rpm for 10 min to remove blood cells, then the upper layer of plasma was gently aspirated, and this procedure was repeated twice. The collected plasma was centrifuged at 10,000 g for 30 min to remove extracellular macrovesicles, then the supernatant was filtered through a 0.22 µm filter and finally centrifuged at 120,000 g for 90 min using an ultra-high speed centrifuge to remove the supernatant. The precipitate was then resuspended with PBS and centrifuged at 120,000 g for 90 min and then supernatant was carefully removed. The precipitate was resuspended with PBS which actually were the exosomes and stored at -80°C.

C_T values were exported to an Excel file to create a table of C_T values. This table was then uploaded to the data analysis web portal at http://www.qiagene.com/qiageneglobe. Specimens were assigned to control or test groups. C_T values were normalized based on manual selection of reference genes. Statistical evaluations were also conducted using Graphpad Prism (version 7.0) and IBM SPSS software (version 24.0). The Shapiro-Wilk W test was used to determine whether data were normally distributed continuous variables. Levene’s test was used to justify equality of variances. A t-test was used to analyze mean values for normally distributed continuous variables and a Mann-Whitney U test to compare mean values for non-normally distributed continuous variables. Correlations between relative mRNA expression and quantitative titer for HPV infection were evaluated by Spearman’s correlation. For all statistical tests, a P-value < 0.05 (two-tailed test) was considered to be a statistically significant finding.

For 60 specimens constituting 20 each for the normal cervix (NC, A group), HSIL (B group) and CSCC (C group) groups, HPV positive infections were detected in 2, 17 and 20 cases respectively, with high-risk HPV infections in 0, 14 and 17 cases, and low-risk HPV infections in 2, 3 and 3 cases ( Table 1 ). The distribution of hrHPV infections in different types of cervical lesions in Xinjiang (n%) were: no hrHPV infections found in 20 patients in the NC group; in the HSIL group, hrHPV infection was primarily HPV16 (40 (%), HPV52 (20%), and HPV31 (15%); the infection rate of HPV16 was significantly higher than in the normal control group (P < 0.05) ( Table 2 ).

The results of RT² profiler PCR array confirmed that genes were differentially expressed < 2-fold (downregulation) and > 2-fold (upregulation) in the HSIL (hrHPV+, 14 cases) and CSCC (hrHPV+, 17cases) groups compared to normal control (NC) group, in order to investigate differentially expressed gene of innate and adaptive immune responses related to the microenvironment (receptors, regulatory genes, upstream and downstream effector genes) in HPV+ or HPV- cervical tissues described above ( Table S2A ). After comparison of pairs using the Qiagen tool website, the presented heatmap showed that CXCL10 was significantly overexpressed in CSCC, with a 3.02-fold increase in HSIL vs NC (P < 0.05), 35-fold in CSCC vs NC (P < 0.05) and a 12.21 fold increase in CSCC vs HSIL (P < 0.05) ( Figures 1A, B and Table S2C ). Moreover, upregulation of CXCL10 was confirmed in a second assay measuring gene expression of the TLR signaling pathway, which showed a 4.33-fold increase in HSIL vs NC and 47.42-fold in CSCC vs NC ( Figures 1C, D and Table S2B ). And qRT-PCR result substantiated the high expression levels of CXCL10 in CSCC ( Figure 1E ).

The correlations between the relative mRNA expression of differential genes from the PCR array analysis and quantitative titer of HPV infections were evaluated using the Spearman correlation test. Correlation analysis revealed that the viral load of HPV infection only had a statistically significant correlation for the expression of CXCL10 in innate and adaptive immune response signaling pathways (r = 0.662, P < 0.05). There was aggregation of gene expression in the concentration area of different HPV infections, which was just at the beginning of the infection stage (0 – 300), and 800 –1,000, 1,500 and so on ( Figure 1F ).

To investigate further whether CXCL10 in cervical cancer might be associated with epithelial atypical hyperplasia or carcinogenesis caused by HPV infection (29), HPV16 E6/E7 transgenic mice models, 8 founder lines, were established by breeding into C57BL/6×C57BL/6 hybrid mice. The tissues from the HPV16 E6/E7 transgenic mice were analyzed histologically to characterized consequences of E6 and E7 expression in cervix epithelia, which presented mild epidermal hyperplasia through HE staining ( Figure S1C) and Immunohistochemistry ( Figure S1D ). Furthermore, cervix tissues from HPV E6/E7 transgenic mice and control groups were also detected with PCR array for innate and adaptive immune response pathways(QIAGEN, Cat. no. PAMM-052Z). Surprisingly, compared to normal mice, the cervical tissues from HPV E6/E7 transgenic mice expressed a 37.23-fold increase in CXCL10 (P < 0.05) ( Figure 1G ). In sum, our findings demonstrated significantly increased levels of CXCL10 by HPV-transformed cervical epithelium ( Table S2 ). Notably, in the innate immune response pathway, IRF7 and MX1 were greatly increased, which act as IFN regulatory factors and as anti-myxovirus proteins. Others have shown that it was pivotal for the regulation of the IFN responses to evoke antiviral effect in innate immunity (30, 31).

To evaluate whether HPV infection was able to modulate the secretion of CXCL10 in HPV+ cells, KoE6/E7SiHa, a HPV E6/E7 knockout SiHa cell line was established through CRISPR/Cas9 gene editing and the knockout efficiency of the E6 and E7 gene by qPCR was 96.7% and 93.6%, respectively ( Figures 2A, B ). Actually, KoE6/E7 SiHa cells grew at a significantly slower rate than SiHa cells in vitro through CCK-8 assay and Annexin V staining apoptosis assay, suggesting that the cell apoptosis induction and proliferation inhibition were markedly induced by the knockout of E6 and E7 ( Figures S3A, B ). CXCL10 expression was determined by western blotting of HFF-1 (HPV- human epithelial foreskin fibroblasts cells), H8 (HPV+, cervical immortalized cells), SiHa (HPV+, cervical cancer cells), KoE6/E7 SiHa (HPV-E6E7 knockout cells) and the results showed that the expression of CXCL10 was greater in HPV+ compared to HPV-cell lysates ( Figure 2C ). Next, to verify further whether HPV infection was responsible for higher CXCL10 secretion in HPV+ cells, the supernatants of cells including HFF-1, SiHa, Caski as well as KoE6/E7 SiHa cells were collected to detect the expression of CXCL10 using ELISA. The results demonstrated that the secretion of CXCL10 increased in HPV+ cells and decreased in the HPV-group ( Figure 2D ). In contrast, CXCR3, as the receptor of CXCL10, was expressed more abundantly in HPV- cells ( Figure 2C ). Surprisingly, the results indicated that CXCL10 was mainly expressed and secreted by HPV+ cells while CXCR3 was preponderant in HPV- cells ( Figures 2C, D ). Thus, we hypothesized that HPV+ cells might have a remarkable influence on HPV- cells through complex CXCL10-CXCR3 interactions.

It has been reported that HPV+ tumors tend to have an increased tumor-associated immune cell infiltrate relative to HPV- tumors (32, 33). However, the mechanisms responsible for these differences have not been fully elucidated. In the present study, a total of 25 specimens of cervical squamous cell carcinoma (CSCC) were collected from 14 HPV+ and 11 HPV- patients ( Table S1 ). Immunohistochemistry revealed that in CSCC lesions, the expression of PD-L1 had a strong relationship with HPV infection ( Figure 3A ). In vitro, the expression of PD-L1 showed that upregulated expression of PD-L1 tended to occur mainly in HPV+ cells ( Figure 3B ).

Actually, this study showed a significantly increased production of multiple chemokines including CXCL10 in HPV+ specimens as well as in transgenic mice. It has been reported that CXCL9/10/11 is a regulator of PD-L1 expression in gastric cancer (14, 28). To explore whether the secretion of CXCL10 was responsible for the expression of PD-L1, a transwell co-culture assay was performed for 48 h using HFF-1 and KoE6/E7 SiHa cells. As shown in Figure 3B , after co-culture with HPV+ cells (SiHa, Caski), PD-L1 expression in HFF-1 was upregulated ( Figures 3C, D ). However, when the SiHa or Caski cells cultured in the chamber of a transwell were treated with 1 μM AMG487, which prevents CXCL10-CXCR3 binding, the upregulation of PD-L1 was suppressed ( Figures 3C, D ). To confirm further the specific effect of CXCL10, recombinant human CXCL10 at a concentration of 100 ng/ml was directly added to the culture of HFF-1 and KoE6/E7 SiHa cells for 48 h, respectively and then treatment results after the application of 1 μM AMG487 were subsequently processed. The results showed that PD-L1 was upregulated by CXCL10 and downregulated after AMG487 treatment as expected, suggesting that HPV infection can induce PD-L1 expression by HPV+ cervical cancer through CXCL10-CXCR3 interactions.

According to the map of the KEGG pathway in the DAVID database, the JAK-STAT pathway acts as a downstream pathway for the chemokine-chemokine R downstream pathway (CXCL10-CXCR3) ( Figure 4A ). The difference of pan-phosphorylation among the cell lines including HFF-1, HFF-1 co-cultured with SiHa, KoE6/E7 SiHa, and KoE6E7SiHa co-cultured with SiHa cells revealed that the phosphorylated proteins of the cells co-cultured with SiHa were significantly increased, which indicated that some proteins of the HPV- cells could be phosphorylated and activated by the key protein of HPV+ cells ( Figure 4B ) (34). Interestingly, the PCR array results showed that STAT1 was precisely overexpressed in CSCC vs the NC or HSIL groups (P < 0.05). Therefore, we hypothesized that CXCL10 may regulate PD-L1 expression through the JAK-STAT pathway ( Table S2 ).

In order to verify the JAK-STAT pathway, HFF-1 and KoE6/E7 SiHa cells were treated with recombinant human CXCL10. Notably, targeting JAK proteins or CXCR3 met with success in decreasing the expression of JAK1 and pSTAT1 after treatment with AMG487 ( Figures 4C, D ) and ruxonitilib ( Figures 4E, F ). Therefore, our findings demonstrated a correlation between suppression of STAT1 activation after JAK1 inhibition, evidenced by STAT phosphorylation, and downregulation of PD-L1. Moreover, the effect of the inhibitor on cells was reversed by recombinant human CXCL10, suggesting that CXCL10-CXCR3 interaction play an important role in PD-L1 upregulation by activating the JAK1-STAT1 pathway in HPV- cells.

Interestingly, PD-L1 could not be directly detected in the cell supernatant, mentioned above, by ELISA. Our evidence has shown that PD-L1 existed in cell exosomes ( Figure 5A ) ( Figure S2 ). Next, a tumor model was established by subcutaneous injection of 5 × 10⁵ SiHa or KoE6/E7 SiHa cells in 120 μl of PBS into the flanks of nude mice ( Figures S3C, D ). After collecting the exosomes, the expression level of PD-L1 was higher in the SiHa group, followed by the KoE6/E7SiHa group, a finding which implied that PD-L1 could be packaged into exosomes in CSCC ( Figure 5B ).