The Interplay Between Regular T Cells and Immunotherapy in Cervical Cancer

Background ： Immune checkpoint blockade inhibitors have aroused great expectation on many types of tumor eradication. However, the therapeutic effect of anti-PD-L1 treatment on cervical cancer is unsatisfactory and the potential antagonist is not very clear. Here, we investigated the therapeutic effect of anti-PD-L1 in cervical tumor mouse model and identified the potential threats for anti-PD-L1 therapeutic efficacy. Results : we found that PD-L1 had a moderate expression in human and mouse cervical tumor cell lines and clinical samples compared to other tumor types and para-tumor tissue. Interestingly, our results showed that the anti-PD-L1 treated mice were dichotomously divided into responsive and unresponsive group even with the same genome background C57BL/6 syngeneic tumor model. The unresponsive tumors showed less immune cell infiltration and higher Tregs population induced immunosuppression activity than the responsive ones. Furthermore, we found that anti-PD-L1 autonomously upregulated Tregs proliferation and frequency in multiple immune organs, and, importantly, Tregs depletion more significantly depressed the tumor growth rate and tumor weight than either anti-PD-L1 or anti-CD25 alone. Finally, we observed that the upregulating effector CD8 + T cell is associated with the better therapeutic effect of anti-PD-L1 therapy post Tregs depletion. Conclusion: In conclusion, anti-PD-L1 therapy upregulates Tregs frequency and proliferation in tumor model, and the depletion of Tregs may be a useful adjuvant strategy for anti-PD-L1 therapy in the immunotherapy of cervical cancer.

106 cells were treated by RAPA buffer on ice for 30 min. Then, the lysis solution was centrifuged for 15 min at 12000g at 4℃. Then, the protein concentration was measured by BSA method. The loading sample was made for 1μg/μl and loaded for 10μl in SDS-PAGE gel. For tumors, the tumor tissue was sufficiently cut into pieces with sterile scissors, and digested with collagenase and hyaluronate for 1h at 37℃. The collected single-cell suspension was lysate and quantified as mentioned for the protein extraction of U14 and Hela cell lines.
In vivo tumor progression and immunotherapy models.
3×106 logarithmic growth phase U14 cells were transplanted subcutaneously into the flanks of 5-to 6-week-old C57BL/6 female mice. 3-4 mice were assigned for every group. PBS, Anti-PD-L1 or anti-CD25 were injected every two days. The tumor size was measured seven days post tumor challenge with a caliper every 2-3 days, and tumor volume was calculated by width 2 × length × 0.5. Mice were sacrificed according to the animal welfare requirement at the endpoint (The maximum tumor less than 20 mm in diameter). All animal protocols were approved by Shanghai JiaoTong University Institutional Animal Care and Use Committee.

RNA extraction and RT-PCR
The cells were harvested and washed for two time with cold PBS. 1ml Trizol reagent was added in 2 x 106 cells and sufficiently suspended. The total RNA was extract according established protocol. In quantitative PCR (q-PCR), the reverse transcription of 1.5μg total RNA were conducted by using SuperScript III First-Strand Synthesis System. The harvested cDNA was diluted for five times by ddH2O. The SYBR Green PCR Master Mix (Applied Biosystems) was used for qPCR, and three repeats were assigned in a Real-Time PCR System (Applied Biosystems). All used primers for qPCR are listed as follow:

Immunohistochemistry
Tumor and spleen tissue samples were carefully extracted and immediately fixed in 4% paraformaldehyde overnight at room temperature. The fixed tissues were embedded in standard paraffin wax to product 5-μm sections for HE and immunohistochemistry assay. In brief, the tissue sections were deparaffinized in xylene for 3 times (10min/time) and rehydrated via an ethanol gradient (100%, 95%, 80%, 75%, 50%). After antigen retrieval with pH 6.0 citrate buffer, sections were incubated in a 0.3% H2O2 solution to remove peroxidase at room temperature for 10min. Then, the sections were washed by PBS for 3times (10min/time) and blocked by normal goat serum or 5% BSA for 1h at 37℃. The sections were then incubated with rat anti-mouse Foxp3 monoclonal antibody (1:200) at 4ºC overnight. On the second day, the tissue sections were treated with instant SABC kit according to provided protocol. Finally, the sections were stained with hematoxylin and sealed for observation under microscope.

Cell and tissue FACS analysis
The peripheral blood, tumor, draining lymph node and spleen were isolated from mice. Then, the single cell suspension for these sample were prepared. The single cell suspension was stained with the following antibodies: anti-mouse CD3, anti-mouse CD4, rabbit anti-mouse CD25at 4 °C for 30minutes in dark. For intracellular cytokine staining (FOXP3 and Ki67), the cells were resuspended in Fixation/ Permeabilization solution (Cytofix/Cytoperm Kit; BD Biosciences) and incubate at 4℃for 40min. The samples were centrifuged for 500g for 5min. Then, anti-mouse FOXP3 and anti-Ki67 were stained for 30min at 4℃ and washed for two times by cold PBS. For the sorting of Tregs, the staining panel is the same as mentioned above. The isolated cells were resuspended with 1ml Trizol regent and the total RNA was extract for RT-PCR.

Results.
1． PD-L1 shows high expression of PD-L1 in cervical tumor cell lines and tumor tissue. carcinoma.
H. PD-L1 expression in different types of cancer. The original data of this graph comes from the Oncomine database. The expression of PD-1 and PD-L1 are the important predicative biomarker for anti-PD-1/L1 therapy. Therefore, we first investigated the expression of PD-L1 expression in human and mouse cervical cell lines. Our results showed that Hela and U14 cell line has relatively high level of PD-L1 expression compared with other tumor types in mRNA ( Fig. 1A and 1C) and protein level ( Fig.  1B and 1D), indicating the potential therapy of anti-PD-L1 therapy in cervical cancer. To investigate the expression of PD-1 and PD-L1 expression in cervical patient samples, we conduct immunohistochemistry for PD-1/L1 in the tumor tissue and corresponding normal tissue of cervical patients. We found that the expression of PD-1 and PD-L1 significantly increase in tumor tissue compared to the corresponding normal tissue of patients (Fig.1E). In addition, we analyzed the mRNA level of PD-1 and PD-L1 in human cervical squamous cell carcinoma by GEPIA database and Oncomine database. We observed that PD-1 (Fig.1F) and PD-L1 (Fig. 1G) showed higher expression in cancer patients than the health control, and cervical cancer had a relative high PD-L1 expression compared with other common cancer types (Fig. 1H). Taken together, the PD-1/L1signaling pathway may be active in cervical cancer. To investigate the effect of anti-PD-L1 on tumorigenesis of cervical cancer, we constructed the syngeneic tumor model in immune competent C57BL/6 mouse. Anti-PD-L1 or PBS was administrated according the treatment schedule ( Fig. 2A). We found that U14 cell line had 100% tumor formation rate in C57BL/6 mice. Anti-PD-L1 treatment significantly depressed the growth of xenografted tumor in most of mice (Fig.2B). Interestingly, tumors were not response to anti-PD-L1 treatment in about 30% mice. To further investigated the tumor microenvironment situation, we conduct HE dye for the tumors. We found that the tumors responding to anti-PD-L1 treatment showed the higher levels tumor necrosis and immune cell infiltration than the unresponsive ones (Fig.2C), indicating the strong immunosuppressive activity. Therefore, we extract mRNA from responsive and unresponsive tumors to anti-PD-L1 therapy and detected the immunosuppressive activity by several vital molecules, including Foxp3, CD206, Arginase, Ly6c and Ly6G. Our results showed that the unresponsive tumors showed higher immunosuppressive activity than the responsive one (Fig.2D). Collectively, the excessive upregulation of Tregs level after anti-PD-L1 treatment may undermine the therapeutic efficiency.

3． The excessive upregulating Tregs in tumors after anti-PD-L1 treatment is associated with
the compromised therapeutic efficiency.

Figure3. The excessive upregulating Tregs in tumors after anti-PD-L1 treatment is associated with the compromised therapeutic efficiency.
A: The Foxp3 IHC assay for the tumor tissue in (B). The pictures were magnified for 400 times. As shown in the figure 2D, the unresponsive tumor showed very high level of Foxp3 compared to the responsive tumor and other immunosuppressive marker. Therefore, we hypothesized that the upregulated Tregs level may account for the compromised anti-tumor effect in unresponsive ones. The IHC results showed that Tregs had a relatively high level in unresponsive tumors (Fig. 3A) and corresponding spleens (Fig.3B) compared to responsive tumors. To further identified our finding, we conducted the flow cytometry to detect the frequency of Tregs in tumors. Consistently, we indeed observed the highest Tregs frequency in unresponsive tumor (Fig.3C). Of note, we also found that anti-PD-L1 promoted the frequency of Tregs in both responsive and unresponsive tumors at different degree (Fig.3C). Two-tailed unpaired T-test was performed. ns: no significant difference, * p < 0.05, ** p < 0.01compared to the control groups. A P value less than 0.05 was considered to be statistically significant.
Although anti-PD-L1 could effectively depressed the growth rate of tumor in the mouse model, only 20% reduction of tumor weight was achieved. Therefore, we hypothesized that Tregs depletion could enhance anti-PD-L1 efficacy. Therefore, we used PBS, anti-PD-L1, anti-CD25 or anti-PD-L1 plus anti-CD25 to treated cervical tumor mouse model (Fig. S2A). Then, we performed C-flow cytometry of Tregs population in the various immune organs after several five times immunotherapy (Fig. S2B). Our results showed that anti-PD-L1 significantly increased the percentage of Tregs in peripheral blood (Fig. 4A), spleen (Fig. 4B), tumors (Fig. 4C) and lymph node (Fig. 4D). Importantly, anti-PD-L1 plus anti-CD25 treatment significantly inhibited the growth of syngeneic tumor compared to PBS or anti-PD-L1 or anti-CD25 alone (Fig 4E). The tumors were harvest at the endpoint, and the tumor weight in anti-PD-L1 plus anti-CD25 treatment group was significantly smaller than the control group or anti-PD-L1 or anti-CD25 group alone (Fig 4F and 4G). Taken together, Tregs depletion could strengthen the therapeutic effect of anti-PD-L1 treatment by decreasing upregulating immunosuppression after immunotherapy.

5．
The increased Tregs proliferation depresses the level of effector CD8 + T cells.

Figure5. The increased Tregs proliferation depresses the level of effector CD8 + T cells.
(A). The change of Ki67 + Tregs in the peripheral blood of mice after anti-PD-L1 treatment. The blood was collected at the endpoint. N=4. Two-tailed unpaired T-test was performed.
(B). The mRNA level of Ki67 in sorted Tregs cells from the spleen of mouse. N=3. Two-tailed unpaired T-test was performed.
(C-E). The change of the frequency of effector CD8 + T cells in tumor (C), the peripheral blood (D) and DLN (E) at the endpoint. N=4. The effector CD8+T cell was defined as CD3 + CD8 + CD62L -CD44 + . Two-tailed unpaired T-test was performed. ns: no significant difference, * p < 0.05, ** p < 0.01compared to the control groups. A P value less than 0.05 was considered to be statistically significant.
To figure out the reason for the increase of Tregs, we analyzed the signature of Tregs after anti-PD-L1 treatment. We found that anti-PD-L1 treatment significantly upregulated the percentage of Ki67 + Tregs, indicating the increasing Tregs proliferation (Fig. 5A). Additionally, we also sorted Tregs for PBS and PD-L1 treated group to detect the mRNA level of Ki67 transcription. Consistently, Ki67 showed the higher mRNA level in anti-PD-L1 group compared with the PBS group (Fig.5B). Therefore, the increased Tregs after anti-PD-L1 therapy may associated with increasing proliferation of Tregs.
Considering of the important role of effector CD8 + T cells (defined by CD3 + CD8 + CD62L -CD44 + ) in anti-tumor response. We respectively analyzed the frequency of effector T cells after PBS, or anti-PD-L1 or anti-CD25 or anti-PD-L1 plus anti-CD25 treatment. Our results that anti-PD-L1 plus anti-CD25 treatment group had a significantly higher level of effector CD8 + T cells than PBS and PD-L1 group (Fig.5C and 5D). In the draining lymph node, we observed the more distinct increase of effector CD8 + T cells in the combination group compared to any of the other three groups (Fig. 5E). In conclusion, the increased effector CD8 + T cells may be associated with the better therapeutic effect after Tregs depletion in cervical tumor model.

Discussion
Currently, immunotherapy had aroused the widely concern of researcher focusing on various tumor types [21]. Although several clinical trials had verified the effect of anti-PD-L1 on advanced tumors, most of patients had great difficulty in maintaining the long-lasting response to immune checkpoint mediated immunotherapy and barely eliminate tumor cells [8,11]. However, the underlying mechanism is not very clear in cervical cancer. Here, we found that anti-PD-L1partially inhibit tumor growth in the syngeneic mouse model, which could be depressed the autonomously increased Tregs proliferation and frequency in multiple immune organs and tumor tissue. We also found that Tregs depletion significantly enhanced the tumor depression effect of anti-PD-L1 treatment in vivo. Therefore, our research provides a novel insight for the limited anti-tumor efficacy in cervical cancer.
The increased Tregs may be one of the important mechanisms of cervical tumors to resist immunotherapy efficacy. Under tumor conditions, various cytokines, such as GM-CSF, IL6, TNFα, and other chemokines [22,23]. Previous studies showed that IL6 and Tumor necrosis factor α (TNFα) could promote Tregs proliferation in tumor sites. The recent study reported that anti-PD-L1 treatment or the Rhein and combination therapy groups upregulated the IL6 level established 4T1 breast cancer xenografts [24]. Consistently, we also observed the slightly increase of IL6 in the tumor tissue after anti-PD-L1 therapy (data not show). TNF is a potent pro-inflammatory cytokine, which played a vital role in the balance of tumor microenvironment. Benoît L Salomon et al reported that TNF is able to increase expansion, stability, and possibly function of Tregs via TNFR2 [23]. In addition, Lack of interleukin-6 in the tumor microenvironment augments type-1 immunity and increases the efficacy of anti-PD-L1 therapy in CT26 cells mouse model [25]. Collectively, we supposed that the increased Tregs in our mouse model probably caused by immunotherapy induced IL6 expression.
Although we observed the enhanced anti-tumor effect after Tregs depletion during anti-PD-L1 treatment in mouse model, we also should be careful for the quickly use of this strategy in cervical cancer patients. A few studies reported that Tregs depleted mice suffered serious autoimmune disease [26]. Furthermore, anti-PD-L1 also may lead to huge immune storm in the host. Therefore, much more attention should by payed on the treatment related adverse event in cervical cancer patients during Tregs depletion combined anti-PD-L1 treatment in cervical cancer patients in the future.
In conclusion, we found that anti-PD-L1 treatment upregulated Tregs levels in cervical cancer mouse model, and Tregs depletion maybe a promising adjuvant treatment of anti-PD-L1therapy for cervical cancer treatment.