Cisplatin Inhibits Bladder Cancer Proliferation Through cGAS-STING Pathway.

Abstract


Results
We found cisplatin induced immune response in bladder cancer by RNA sequencing, and validated cGAS-STING signal was deeply involved in this response. Cisplatin induced cGAS-STING signal inhibited the proliferation of bladder cancer and increased the in ltration percentages of CD8 + T cells and dendritic cells in transplantation mice tumor model. Accumulation of dsDNA and the release of chromatin bound cGAS are important to activate downstream STING.

Conclusion
Our ndings indicated a cisplatin related immune effects in bladder cancer, cisplatin combined with immunotherapy might have a synergistic effect for bladder cancer therapy.
1. Background 1.1 Bladder cancer is one of the most common tumors in urinary system 1 . Based on pathology, bladder cancer is mainly divided into NMIBC (non-muscle invasive bladder cancer) or MIBC (muscle invasive bladder cancer), and the latter demonstrates a poorer prognosis and are recommended aggressive treatments like cystectomy combined with chemotherapy. More than 80% of bladder cancer chemotherapy is based on platinum drugs, and cisplatin is the most classic drug among them 2 . However, more than 60% of patients receiving chemotherapy strategy like neo-adjuvant chemotherapy are cisplatin resistant 3 . The critical mechanism of cisplatin resistance is still unclear. As a classical DNA cross linker, intracellular cisplatin can inhibit tumor cells by inducing irreversibility structural distortion of DNA, induce DNA damage and result in accumulated productions like dsDNA or micronucleus in tumor cells 4,5 . Over the past decade, immunotherapy (like, PD-L1, PD1 or other) have demonstrated a promising prognostic value for MIBC therapy, and now emergent clinical trials are trying to explore the combination treatment effects of cisplatin with these new therapies. Among these clinical trials, the keynote 189 trials indicated that patients receiving platinum chemotherapy combined with PD1 showed better overall survival rates than patients receiving chemotherapy alone in non-small cell lung cancer (69.2% vs 49.4%) 6 . Cisplatin upregulated cell surface PD-L1 levels on multiple cancer types, and combination of PD-L1, PD-1 targeting therapy with cisplatin gave a synergistic effect in treating several tumor models [7][8][9] . The immunity function of cisplatin provides a scienti c rationale to combine cisplatin with other immunotherapies.
1. 2 Recently it has been reported that cGAS, an intracellular dsDNA receptor, can recognize and catalysis cytosolic dsDNA induced by cisplatin, thus activating the downstream STING signal 4 . The canonical mechanism of cGAS-STING pathway was rstly reported by Chen in 2012 10 . With further research, cGAS-STING pathway has been found to be involved in many disease processes, such as immune defense, tumor progression, autoimmune diseases, neuron degeneration [11][12][13][14] . After recognition of double-stranded DNA (dsDNA), cGAS catalyzes the cyclization of ATP and GTP into the second messenger cyclic GMP-AMP (2, 3 -cGAMP). cGAMP activate STING and result in the translocation of STING from the endoplasmic reticulum to signaling compartments, where STING enables the phosphorylation of kinase TBK1, which mediates the activation of the transcription factor interferon regulatory factor 3 (IRF3) and or nuclear factor κB (NF-κB) 4,15 , and nally leads to the secretion of type interferons, like IFN-β.
1.3 The functional activity of cGAS is the critical step to activate STING pathway. Liu et al. found that the nuclear translocation of cGAS in THP-1 and HeLa cells can affect its functions by suppressing homologous-recombination-mediated repair 16 . Studies have also shown that the binding capacity of cGAS to nucleosomes is much greater than that of cGAS to dsDNA [17][18][19][20][21] . This can explain why the cGAS-STING signal is not signi cantly activated during mitosis when a large amount of dsDNA is exposed to the cytoplasm due to the dissolved nuclear membrane. As cGAS are competitively bound by nucleosomes, it may lose the ability to bind dsDNA and activate the STING downstream 17-21 .
1. 4 We nd that cisplatin treatment induces an cGAS-STING dependent immune responses in bladder cancers. Cisplatin related immune responses promote the secretion of cytokines such as INF-β, IL-6 in vitro and can recruit CD8 + T cells and dendritic cells(DCs) in ltration in mice tumor model. Cisplatin induced DNA damage can activate intracellular cGAS by producing dsDNA fragments and micronuclei. It can also dissociate cGAS from chromatin by distorting the structural of nucleosomes, and restore the ability of cGAS to bind dsDNA and then activate STING signal. Exploration the immune effects of cisplatin in bladder cancer can help us to understand the diverse mechanisms of cisplatin and provide the basis for the combination of cisplatin and immunotherapy.

Cell culture
Human muscle invasive bladder cancer cell lines T24, TCCSUP and UMUC-3, mouse derived bladder cell lines MB49, and 293T cell line were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Short tandem repeat DNA pro ling analysis were performed to ensure the stable and reliable of all cell lines during the experiments. T24, TCCSUP, UMUC-3 and MB49 cells were cultured in RPMI-1640 medium (Thermo Fisher Scienti c, Inc.), and 293T cells were cultured in DMEM medium (Thermo Fisher Scienti c, Inc.). Mediums mentioned before were pre-supplemented with 10% fetal bovine serum (Gibco; Thermo Fisher Scienti c, Inc.) without any antibiotics, and all cell lines were incubated under standard conditions (37 °C and 5% CO 2 ).

Cell counting kit-8 (CCK-8) assay and Colony Formation assay
Each groups of T24, TCCSUP and MB49 cell lines were prepared in 96-well plates (1000 cells/well) under standard conditions. Then, premixed medium with a 10% concentration of CCK-8 (Dojindo Molecular Technologies, Inc., Kumamoto Japan) reagent was added into each well and incubated under standard conditions for 1-2 h before measurement at 450 nm. For colony formation assay, different groups of cells were prepared in 6-well plates (1000 cells/well) and cultured for 6-8 days under standard conditions followed by 3 min staining and 6 min xation (Wright -Giemsa Stain Kit, NJJC bioengineering institution, China) at room temperature before colonies comparison.

Plasmids construction and transfection
Sequences of short hairpin RNA (shRNA) against CDS of msSTING were validated from Sigma-Aldrich Online. Scramble sequences were designed using Wizard v3.1 (https://www.invivogen.com/sirnawizard/) to ensure the absent of seed sequence matches. msSTING sequences and scramble sequences were synthesized (Tsingke, Hangzhou, China) and inserted into the plko.1-puro vector. Then the vector was cotransfected with pSPAX2 and pMD2G (with a ratio of 4: 3: 1, respectively) into the 293T cell lines cultured in 100 mm plates (30-60% cell density) using Lipofectamine 3000 (Thermo Fisher Scienti c, Inc.) according to manufacturer's instruction. The supernatant, which contained Lentivirus, was harvested and centrifuged (800 × g, 5 min) 48 or 72 h post-transfection and added into MB49 cells cultured in 6-well plates (30-60% cell density) immediately. Successfully transfected shSTING cell lines were screened with an increasing concentration of puromycin (Sigma-Aldrich; Merck KGaA). The siRNA used here were also transfected using Lipofectamine 3000. Finally, the transfection e ciency was validated by western blotting. The msSTING and siSTING sequences used here were showed in Table S1.

Mice
C57BL/6 mice were purchased from Jackson Labs. 4 week mice were mainly used in this experiment and all mice were maintained in pathogen-free barrier facilities and were approved by Zhejiang University experimental animal welfare ethics review committee. 2 × 10 6 of MB49 cells with shSTING or shScr transfection were subcutaneous injected on the ank of mice. Then cisplatin (3 mg/kg) or PBS were injected subcutaneous injected around the tumors in day 13 and day 16. Tumors size were measured regularly by length (L), width (W) and height (H), respectively, and tumor volume was calculated as 1/2 × L × W × H. Then tumors were harvest in day 23, the weight of tumors were normalized by their corresponding body weight. Fresh tumor issues were then used for ow cytometry analysis.
TRIzol (Thermo Fisher Scienti c, Inc.) was used for cells total RNA extraction according to manufacturer's protocol. Then different groups of RNA were used for RNA-seq or qRT-PCR. T24 cells were treated with cisplatin (2 µg/ml) or PBS for 24 h at standard condition. RNA-sequence was performed by Novogene China. Gene expression levels were quanti ed as Fragments Per Kilobase Million and then Log 2transformed as A value. The signi cant different expression genes were identi ed as absolute fold change of A value > 1.5 combined Padj (adjust P) value < 0.05 between two groups. Then different expression genes were used for plotting heatmap and volcano map by R soltware (Version 4.0.2, https://cran.r-project.org/src/base/R-4/). GO and KEGG analysis were performed on DAVID online (https://david.ncifcrf.gov). Pathways with Padj value < 0.05 were identi ed as signi cant different pathway. As for qRT-PCR, Takara PrimeScript™ RT and SYBR EX Taq™ kits (Takara Bio, Inc., Otsu, Japan) were used for qRT-PCR according to manufacturer's instruction. The standard thermocycling conditions used were as follows: Initiate Step, 95.0 °C: 30 s; cycling Step, 40 cycles of 95 °C: 5 s and 60 °C: 30 s; melt curve analysis Step, 65 °C to 95 °C, increasing in 0.5 °C increments for 5 s. The speci c primers used in this experiment were listed in Table S1. Control groups were conducted to con rm the absence of the agent pollution or primer dimers. Targeting genes were nally normalized to GAPDH expression, and the relative mRNA expressions were calculated using the ΔΔCq method.

Enzyme linked immunosorbent assay (ELISA) and IFN reporter assay
Secreted supernatant IL-6 of T24 and TCCSUP cell lines was measured using the Human IL-6 ELISA kit (DAKEWE, China) according to manufacturer's instructions. IFN-β in cell supernatants with biological activity was compared using a reporter 293T cells stably expressing a pISRE (Genechem Co. Shanghai, China). Reporter cells were co-cultured with supernatant from different groups for 24 h before measured by a uorescent enzyme meter Varioskan™ Flash at 488 nm/520 nm (Thermo Fisher Scienti c, Inc.). Intensities of cellular uorescent were normalized to the total number of T24 and TCCSUP cells.

Flow cytometry analysis
Tumor tissues were digested and incubated with biotinylated anti-CD3, anti-CD8 antibodies, followed by

Immuno uorescence
Different groups of T24, TCCSUP, UMUC-3 and MB49 cells were pre-seeded the day before immuno uorescence assay. 4% paraformaldehyde was used for cell xation and 0.5% Triton-X 100 was used to increase cell membrane permeability for 10-15 min at room temperature. Then 3-5% BSA was used for 30-60 min at room temperature to block antibodies. Cells were further incubated with primary antibodies overnight at 4 °C and subsequently incubated with corresponding uorescence antibodies combined with DAPI staining 30-60 min in darkroom and then wash twice by PBS before observed by NIS A1 laser confocal microscope (Nikon, Japan). The primary antibodies used for immuno uorescence assay were: cGAS (1:200; cat no. A8335; Abclonal), F-actin (1:1, 000; cat no. A12380; Invitrogen™), Rad51(1:300; cat no. ab133534; Abcam), γ-H2A.X (1:300; cat no. ab81299; Abcam). The secondary antibodies used for immuno uorescence assay were: Goat Anti-Rabbit Alexa Fluor 488 (1:200; cat no. ab150077; Abcam) and Goat Anti-Mouse Alexa Fluor 594 (1:200; cat no. ab150116; Abcam). The results were analyzed by NIS-Elements Viewer 4.50 (Nikon, Japan). SPSS 22.0 (IBM Corp. USA) was used to normalize and analyze raw data in this experiment. Data was presented as the mean ± standard deviation. Kaplan-Meier survival method and log-rank test analysis were using to plot (OS) overall survival and (DFS) disease-free survival rates curves. A student's t-test was used to assess the different between two groups. One-way ANOVA test was performed to assess the differences in multiples groups followed by Student-Newman-Keuls post hoc test. IC50 of MB49 cell lines were measured by Probit regression analysis. P < 0.05 was considered to have statistical signi cance.

Statistical analysis
Data in experiments was repeated at least 3 times.

Cisplatin induces speci c immune effects in bladder cancer cell line.
In order to explore the speci c effect of cisplatin on bladder cancer, a MIBC derived cell line T24 was treated with cisplatin (2 µg/ml, 24 h) followed by RNA SEq. We found a signi cant difference gene expression panels between cisplatin treated groups and control groups (Fig. 1A). The combination of absolute value log 2 (Fold Change) > 1.5 and Padj (adjust P) value < 0.05 was chosen as the criteria for screening signi cant differential genes between two groups, and nally we got 235 up-regulated and 129 down regulated genes (Fig. 1B). After GO and KEGG (https://david.ncifcrf.gov) analysis of these differentially expressed genes, we found that the treatment of cisplatin mainly affect T24 cells in immune response, in ammatory response, cytokine activity, the cellular response to DNA damage stimulus, apoptosis process and some cell growth related pathways (Fig. 1C). Gene set enrichment analysis (GSEA) also con rmed the strong enrichment in immune-and in ammation-related pathways among the top 20 differentially expressed pathways in cisplatin treated group. Interestingly, we noticed type I interferon related pathway and cytosolic DNA sensor pathway were signi cantly up-regulated after cisplatin stimulation (Fig. 1D, Figure S1B, C). The involved cellular response to cytosolic DNA sensor process is closely relevant to cGAS-STING pathway. Recently, it has been reported that cGAS, an intracellular dsDNA receptor, can recognize and catalysis dsDNA induced by cisplatin, thus activating the downstream STING pathway 4 . Then, we explored the activation of cGAS-STING downstream and found that transcription levels of NF-κB and IRF3 related genes were signi cantly up-regulated after cisplatin treatment (Fig. 1E). These results suggest that the immune effects associated with cisplatin are on account of the activation of cGAS-STING pathway 4, 11 . 3.2 The immune effect of cisplatin in bladder cancer is related to the activation of cGAS-STING pathway.
Expression of STING protein is the basic element for functional STING pathway. We rstly explored the basic transcription level of STING in bladder cancer cell lines in CCLE database (https://portals.broadinstitute.org/ccle/about), and found that the transcription level of STING was relative abundant in bladder cancer, including T24 another MIBC cell line TCCSUP cell lines (Fig. 2A). In both ELISA and INF-β reporter gene assay we found the levels of IL-6 and INF-β, the activated downstream of STING pathway, were signi cantly up-regulated with cisplatin treatment in T24 and TCCSUP ( Fig. 2A, B). It was con rmed by qPCR that the transcription levels of cytokines including IL-6, INF-β and TNF-α were signi cantly up-regulated after cisplatin treatment (Fig. 2D). We also observed the translocation of p65 from cytoplasm to nuclear (Fig. 2E). The canonical activation of STING pathway is mainly based on the phosphorylate of STING (Ser366), TBK (Ser173) and IRF3 (Ser396) [22][23][24] . We then performed Western blotting assay and found that the cisplatin treatment signi cantly up-regulated cGAS-STING pathway accompanied by a signi cant increase of DNA damage marker γ-H2A.X in two cell lines, compared with their corresponding control groups (Fig. 2F). The polymer state of STING is the activated result of phosphorylated STING 10,25 , we found the polymer of STING and γ-H2A.X were increased with the time of cisplatin stimulation expanded ( Figure S1A). We also noticed that the cell surface PD-L1 level of two cell lines increased signi cantly with cisplatin inducing (Fig. 2G). In rescue assay, the use of two STING targeting siRNAs (Fig. 3A-C) and STING speci c inhibitor H151 (Fig. 3D, E) signi cantly reversed cisplatin induced IL-6 and INF-β levels compared with their control groups, respectively. These results suggested that cGAS-STING pathway was deeply involved in cisplatin related immune response.

Modulation of STING did not affect the proliferation of bladder cancer in vitro.
Since cisplatin activates cGAS-STING signal in bladder cancer cell lines, what role does cGAS-STING plays in bladder cancer progression remains unclear. The CCK8 cell proliferation assay (Fig. 4A) and clonal formation assay indicated the absence of signi cant basic proliferation bias between STING knockdown groups and control groups in two cell lines (Fig. 4D, E). We found that the levels of cGAS-STING pathway related key genes in patients from TCGA bladder cancer cohort were not signi cantly associated with overall survival rate (OS) and disease free survival rate (DFS) (Fig. 4B, C; Figure S1D-I). This imply that the activation states and expression levels of genes in cGAS-STING pathway is not inevitable relevant.

Activation of cGAS-STING suppressed bladder cancer in cisplatin-treated C57 mice.
In order to explore the phenotype in vivo, we constructed a MB49 shSTING cell line and a MB49 scramble control cell line (Mouse derived) by lentivirus transfection (Fig. 4F). To exclude the possible effects caused by different cell proliferation rate and cisplatin sensitivity between two constructed cell lines, we veri ed before that there was no signi cant difference existed in cell proliferation (Fig. 4G, I, J) and basic IC50 of cisplatin treatment in two groups (Fig. 4H). At the same time, we noticed that cisplatin treatment activated cGAS-STING signal in MB49 scramble control group but not MB49 STING knockdown groups (Fig. 4K). We then establish the subcutaneous tumor transplantation model in C57 mice, cisplatin or PBS was injected subcutaneous in day 13 and day 19 (Fig. 5A-left). The tumors were harvest in day 23, and the results showed that the tumor volumes in the cisplatin treated group was signi cantly reduced compared with the corresponding control group (Fig. 5A-right). Compared with STING knockdown group, the tumor volume and relative tumor weight of STING WT group were signi cantly reduced after cisplatin treatment ( Figure S2 A-C).

Activation of cGAS-STING results in the CD8 + T cells and DCs in ltration in cisplatin-treated tumors in xenograft transplantation model.
In order to explore whether the difference in MB49 derived tumor is related to immunity, we analyzed tumor associated in ltrating immune cells by ow cytometry. With cisplatin treatment, we found that compared with STING knockdown group, the percentage of CD8 + CD45+, CD11c + MHCII+ (Dendritic cells, DCs), CD3 + CD45 + was signi cantly higher in STING WT group (Fig. 5B, C, E), but the percentage of F4/80+, CD11b+ (Macrophage) in ltration was not signi cantly different (Fig. 5D). Compared with the control group, the percentage of CD8 + CD45+, CD3 + CD45+, F4/80 + CD11b + in cisplatin treated groups decreased signi cantly, but the percentage of CD11c + MHCII + in ltration cells were signi cantly increased. Studies have reported that CCL20 and CXCL14 are important chemokines for DCs cells. We found that the transcription levels of CCL20 and CXCL14 in T24 and TCCSUP cells were signi cantly increased in cisplatin treatment groups ( Figure S2D, E), correlation analysis showed that CCL20 and CXCL14 were signi cantly correlated with cGAS-STING signal in TCGA bladder cancer cohort ( Figure  S2F). We also performed CIBERSORT (https://cibersort.stanford.edu) to analyzed the percentage of immune cell in ltration in patients with bladder cancer who received cisplatin based chemotherapy in TCGA bladder cancer cohort (https://portal.gdc.cancer.gov/) 26 (Figure S3A-C). Interestingly, we found that the in ltration of DCs activated cells in response group (n = 50, CR or PR) was signi cantly lower than non-response group (n = 30, PD or SD) after cisplatin based chemotherapy, while the in ltration of DCs resting cells in two groups displayed the opposite results (Fig. 5F-I). This nding imply that a unique role that DCs paly in bladder cancer microenvironment after cisplatin treatment 27 . However, the speci c function of cisplatin induced DCs in bladder cancer microenvironment needs to be further explored.
3.6 Cisplatin induced dsDNA may function as an important role in enhancing STING signal.
As an important protein that can recognize dsDNA and catalyze the production of 2, 3 -cGAMPs, expression of cGAS protein is the most important upstream of STING activation. We then explored the basic transcription level of cGAS in bladder cancer cell lines in CCLE database (https://portals.broadinstitute.org/ccle/about), and con rmed the high transcription level of cGAS in bladder cancer cell lines (rank 11th among 40 tumor types), and T24 and TCCSUP also showed relative high transcription levels of cGAS (Fig. 6A, Figure S3D). It was reported that cisplatin induced dsDNA fragments through accumulation of DNA damage and inhibiting DNA homologous complementary repair pathway 16 . We observed the signi cant increase of DNA damage marker γ-H2A.X in T24 and TCCSUP cell lines treated with cisplatin (Fig. 2F). Rad51 is an important protein involved in homologous complementary repair of DNA, the up-regulated and assembled of nuclear Rad51 is also the classical marker of increased DNA damage 28 . We found nuclear Rad51 increased and assembled signi cantly after cisplatin stimulation while the cytosolic Rad51 levels was not changed (Fig. 6C, D). When dsDNA induced by DNA damage leaks into the cytoplasm, histone H3 bound to dsDNA will be brought into the cytoplasm in the same time, resulting in an increase level of cytosolic H3 protein 29 . In consistent with our results, we observed that the level of H3 in the cytoplasm increased signi cantly with the time of cisplatin stimulation expand (Fig. 5A, B), which also suggested that dsDNA might leak from the nucleus continuously after cisplatin treatment. Furthermore, we found that cisplatin also induced micronuclei in T24 and TCCSUP cell lines ( Figure S3E). These are important factors to activate cGAS-STING pathway.

The release of chromatin bound cGAS is important to activate downstream STING.
We tried to explore the role of cGAS in cisplatin related STING activation and found that the expression level of total cGAS did not change signi cantly after cisplatin stimulation in two cell lines (Fig. 2E). The subcellular distribution of cGAS is closely related to its function [17][18][19][20][21] . At rst we con rmed that cGAS protein of T24 and TCCSUP cell lines was mainly located in nucleus (Fig. 6C, E). And same results were also showed in MB49 and UMUC3 cell lines (Fig. 6E). Then we tried to explore whether the subcellular distribution of cGAS was changed after the cisplatin treatment. Though the morphology of T24 and TCCSUP cells became enlarged after cisplatin stimulation, the distribution of cGAS did not change signi cantly (Fig. 6F). Western blotting results of nuclear-cytoplasmic separation also showed that cGAS distribution in cytoplasm and nucleus did not change signi cantly in two cell lines (Fig. 6C). Recently, many scholars have found that cGAS can bind with histones H2 and H3 on nucleosome to form stable structure and prevent cGAS from binding with dsDNA [17][18][19][20][21] . Based on this ndings, we performed subcellular component western blotting assay and found that most of cGAS were bound to the nucleosome in the nucleus. After cisplatin stimulation, especially at 24 h, the binding of cGAS in nucleosome in two cell lines was signi cantly reduced accompanied by a signi cant increase of γ-H2A.X, compared with their corresponding control groups (Fig. 6G). Thus the free cGAS might reacquire the ability to bind to dsDNA and then activate downstream STING dependent immune signal.
4. Discussion 4.1 In this work, we reported that cisplatin treatment of bladder cancer cells did produce speci c immune effects, mainly including the secretion of important cytokines (such as interferon type I cytokines, INF-β, IL6) and the increase of PD-L1 level on the cell surface. Recently, the remarkable keynote 189 trials showed that platinum based chemotherapy combined with PD1 showed better prognosis than chemotherapy alone in non-small cell lung cancer. We then con rmed that cGAS-STING signal were deeply involved in these immune effects. Cisplatin relative cGAS-STING signals have also been reported in ovarian cancer, NSCLC and epithelial cell carcinoma [30][31] . These results provided the potential therapeutic choice of combined cisplatin and immune therapy in bladder cancer therapy.

4.2
The expression levels of STING and other components involved in STING pathway were not associated with OS and DFS in bladder cancer patients. But in mice tumor model, we con rmed that knockdown of STING in bladder cell had no signi cant effect on tumor volume and weight, but the tumor volume and weight in group of STING-knockdown combined cisplatin treatment was signi cantly larger than in group of STING WT combined cisplatin treatment. These results suggest that cisplatin related STING dependent signal strong inhibits bladder cancer progression, which is consistent with many previous studies 7-9, 30, 31 . These difference might be explained by higher in ltration levels of CD8 + T cells and DCs cells in STING-WT combined with cisplatin group, compared with that in STING-knockdown combined with cisplatin group. The in ltration levels of CD8 + T cells in tumor tissues is recently proved to be consistent with functional cGAS-STING pathway 32 . The in ltration and activation of DCs cells in tumor tissues is closely related to speci c chemokines in tumor microenvironment. Chemokines like CCL20 and CXCL14 can recruit DCs to tumor tissue to inhibit tumor proliferation and metastasis 33,34 . In our experiment, we also observed the increased transcription levels of CCL20 and CXCL14 in T24 and TCCSUP cell lines after cisplatin treatment. It is also reported that intrinsic cGAMPs in tumor cells can be transferred to DCs through gap junction and activate the immune response of DCs cells 35,36 . By CIBERSORT analysis, we predicted that the in ltration percentage of DCs activated cells in response group (n = 50, CR or PR) was signi cantly lower than unresponse group (n = 30, PD or SD) in patient receiving platinum based chemotherapy in TCGA bladder cancer cohort. Thus, DCs may play an important role in cisplatin related cGAS-STING immune effects in bladder cancer, but its underlying mechanism and speci c functions towards tumor microenvironment remain to be clari ed.

Urinary tract tumors including bladder cancer has relative high expression of cGAS. Researches have
shown that tumor cells with high expression of cGAS protein have relatively active cGAS-STING pathway activity 35 . This is because that tumor cells often have higher levels of intracellular dsDNA due to more frequent DNA damage and repair cycles, which can activate cGAS and generate plenty of cGAMPs and nally activate STING downstream 35,36 . As a classical DNA crosslinking agent, cisplatin can strongly induce DNA damage in multiple tumor types. DNA damage lead to dsDNA leakage from nucleus to cytoplasm 4,5,29 . We observed that with the time of cisplatin inducing expand, the level of histone H3 in cytoplasm increased gradually, which indicated that dsDNA was constantly leaking from the nucleus 29 .
At the same time, we observed the production of micronuclei in cisplatin treated groups. Micronuclei was reported not only as a marker of DNA damage, but also as a way to activate STING signal 5 .

4.4
The functional state of cGAS is closely related to its subcellular localization. For instance, translocation of cGAS from cytoplasm to cell membrane helps THP-1 and HeLa cells to recognize the exogenous virus DNA 37 . And nuclear translocation of cGAS can affect its functions by suppressing homologous recombination mediated repair 16 . But we found that cGAS was mainly located in nucleus in T24 and TCCSUP cell lines, and this subcellular distribution was not changed after cisplatin treatment. Studies have shown that the binding capacity of cGAS to nucleosomes is much greater than that of cGAS to dsDNA [17][18][19][20][21]38 . Based on this important nding, many research groups further discovery that cGAS can bind with histones H2 and H3 on nucleosome to form stable structure and prevent cGAS from binding with dsDNA [17][18][19][20][21] . So, we then explored the state of chromatin binding cGAS after cisplatin treatment, and found that with the prolongation of cisplatin treatment, the DNA damage marker γ-H2A.X increased signi cantly, while the chromatin bound cGAS decreased signi cantly. In multiple tumor types, the activation of cGAS is thought to have the function of inhibiting tumor proliferation by activating STING dependent immune response, but this function will be weakened due to the binding of nucleosome [7][8][9] . These results indicate that cisplatin may hamper the stability of nucleosome by distorting the natural structural of DNA 39,40 . The free cGAS may reacquire the ability to bind to dsDNA and then activate downstream STING dependent immune signal. However, whether the dissociated cGAS is transferred to the cytoplasm and further activate STING still need to be proved.

Conclusion
In summary, our ndings indicated a cisplatin dependent cGAS-STING signal in bladder cancer. This Availability of data and material: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Competing interests: The authors declare that they have no competing interests.