The expression profiles and clinicopathological parameters were obtained from The Cancer Genome Atlas (TCGA), which can be accessed for download through the UCSC website (https://xenabrowser.net/). Additionally, we utilized the Gene Expression Omnibus (GEO) dataset GSE192897 (35). Relevant web addresses for online resources and analysis tools are thoughtfully provided within the context for easy access.

The prognostic analysis of STING downstream genes, which include CCL5, CXCL9, and CXCL10 in cervical cancer, is graciously facilitated by the open-access platform, Kaplan-Meier Plotter (36). Additionally, the expression of these STING downstream genes was stratified based on the clinical stage of cervical cancer patients.

We performed a log₂(x+0.001) transformation of the expression values of STING downstream genes for each sample in the TCGA-CESC database. Additionally, we extracted gene expression profiles for each tumor and mapped them to GeneSymbol. Furthermore, we utilized the R package ESTIMATE and deconvo_CIBERSOR method from the R package IOBR to reevaluate the immune infiltration score and immune cell infiltration in each patient with cervical cancer based on gene expression (37–39). Ultimately, We calculated the Pearson’s correlation coefficient between genes and the aforementioned immune infiltration states in various tumors using the corr.test function from the R software package psych (version 2.1.6). All the data about this section were obtained from TCGA and analyses were processed through the online platform, http://sangerbox.com/home.html (40).

In this study, we analyzed the alterations in the TME after MSA-2 treatment in an immune-excluded tumor model (26). Our analytical journey commenced with the intricate task of annotating cell types using the SingleR package (version 1.8.0) paired with the authoritative ImmGen reference database. To fortify the credibility of our annotations obtained via SingleR, we ventured further by meticulously calculating the expression levels of immune cell-specific markers, drawing from a compendium of previous studies. In our pursuit of accuracy, T cell subclusters were annotated concerning well-established markers, while NK cell subclusters were meticulously defined based on the robust criteria of Itgam and CD27 expression levels. Subsequently, we employed the venerable chi-squared test to scrutinize disparities in cluster proportions among various experimental groups. This statistical examination offered valuable insights into the differences observed. Our visualization and characterization efforts were no less meticulous. We artistically portrayed the distinctive features of T cells, NK cells, and conventional dendritic cells (cDCs) across different experimental groups, invoking the well-respected MSigDB hallmark gene sets (H). The signaling enrichment was assessed through the judicious use of the Gene Set Enrichment Analysis (GSEA), facilitated by the singleseqgset package (version 1.2.9) (26). The evaluation of intercellular communication numbers and intensities utilized the CellChat package (version 1.6.1) with the CellChatDB.mouse database. Visualization of cell communication was accomplished through the netVisual, netVisual_heatmap, the netAnalysis_signalingChanges_scatter, and netVisual_bubble functions.

Murine cervical cancer cell lines U14, TC-1 were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (Gibco) with 10% fetal bovine serum (Gibco) and 1×Penicillin-Streptomycin Solution (Gibco). Murine PD-1 antibody (29F.1A12) was purchased from BioXCell. MSA-2 (HY-136927) was purchased from MedChemExpress (MCE).

We conducted an investigation into the antitumor activity of the combination of MSA-2 and anti-PD-1 in two syngeneic tumor models. U14 and TC-1 are commonly used cervical tumor models to explore immunotherapy efficacy (41, 42). MSA-2 was administered orally at a single dose of 50 mg/kg, while mice received 5 mg/kg of anti-PD-1 treatment on alternate days, for a total of three treatments. Tumor volume was assessed every other day, and mice were euthanized either when their tumor volume exceeded 2000 mm³ or at the termination of the experiment. In the case of subcutaneous tumor models, we initially implanted 5 × 10⁶ U14 or TC-1 cells into the subcutaneous tissue of C57BL/6 mice. All mice carrying tumors were randomly divided into four groups: the control group (CTL), the MSA-2 group, the anti-PD-1 group, and the combination therapy group with MSA-2 and anti-PD-1 (each group consisting of 5 mice). Treatment commenced 10 days after injection into the subcutis or when the tumor volume reached a range of 50 ~ 150 mm³. We repeated the U14 model to investigate the impact of agents on survival, with an initial implantation cell count of 7.5 × 10⁶.

Tissues were fixed in a 4% paraformaldehyde solution overnight. After fixation, the tissues underwent three PBS rinses. Subsequently, they were immersed in ethyl alcohol with an escalating concentration for one hour each. Following this, the tissues underwent dehydration and were then embedded in paraffin wax at 65°C, ultimately facilitating slicing for further analysis. IF staining was conducted using the tyramide signal amplification method. Antibodies targeting PCNA (nucleic staining, BM0104, Boster), anti-CD3 (ab231775, Abcam), and anti-CD8 (ab217344, Abcam) were employed in the assays. The Tunel assay (C1086, Beyotime) was performed following the manufacturer’s recommendations. Captured images were examined using Caseviewer or Hamamatsu Nanozoomer software, and two independent pathologists delineated the regions of interest (ROIs). The integral optical density (IOD) was quantified by Image-Pro Plus 6.0 (Media Cybernetics) used to measure immune cell infiltration status and apoptotic tumor cells.

Statistical analyses were conducted using both GraphPad Prism (version 8.4.2) and R software. Student’s t-test was employed to compare two variables when the data followed a normal distribution. All statistical analyses in this study were two-sided, and statistical significance was defined as p < 0.05.

Previous investigations have identified CCL5, CXCL9, and CXCL10 as downstream targets of STING, which serve as crucial intermediaries connecting DNA sensing with the initiation of the immune response (43, 44). Notably, we uncovered a compelling association between CCL5, CXCL9, and CXCL10 and key prognostic factors, encompassing overall survival and progression-free survival, in cervical cancer, as observed within TCGA dataset. The conspicuous elevation in expression levels of CCL5, CXCL9, and CXCL10 consistently serves as a robust prognostic indicator for prolonged survival and significantly reduced probabilities of recurrence and relapse ( Figures 1A–F ). Moreover, in the context of advanced stages of cervical cancer, there is a noticeable decrease in the expression of these genes ( Figures 1G–I ), and they are closely linked to a conspicuous inverse correlation with metastatic potential ( Figures 1J–L ). Additionally, they present a diminished expression profile in cervical adenocarcinoma compared to squamous cervical carcinoma both in TCGA dataset ( Figures 1M–O ) and the GSE192897 dataset ( Figures 1P–R ). Previous studies demonstrated that cervical adenocarcinoma possessed a heightened malignant propensity, and a greater inclination toward metastatic behavior (45), characterized by the TME with a more robust immune infiltration cells profile (46). These findings underscore the pivotal role of the STING pathway in shaping the prognosis of cervical cancer.

As previously implied, it becomes increasingly evident that various histotypes of cervical cancer exhibit unique and discernible immune infiltration cell profiles. This nuanced variation, in turn, exerts a profound influence on the dynamic landscape of tumor invasiveness and critically impacts the efficacy of drug treatments. With this in mind, our subsequent investigation delves into the intricate correlation between the expression levels of STING downstream genes, CCL5, CXCL9, and CXCL10, and their influence on the overall immune infiltration status within the cervical cancer TME. After analysis, we determine that these genes are all vigorously positively correlated to three dimensions of immune infiltration, namely stromal scores ( Figures 2A–C ), immune scores ( Figures 2D–F ), and estimate scores ( Figures 2G–I ) based on ESTIMATE comprehensive scoring in the context of cervical cancer. Besides, When delving deeper into specific immune cells, we identify consistency of the expression of all these genes and the presence of CD8⁺ T cells ( Figures 2J–L ) and M1-type macrophages ( Figures 2M–O ). Moreover, CCL5 expression, in particular, demonstrates an inverse relationship with M2-type macrophages ( Figure 2P ) and a direct association with NK cells ( Figure 2Q ) and T follicular helper cells ( Figure 2R ). It is firmly established that CD8+ T cells, M1-type macrophages, NK cells, and T follicular helper cells not only serve as instrumental agents in the immune clearance of tumor cells but also serve as indicative barometers of the potential efficacy of ICB therapy (47–52). Conversely, M2-type macrophages exhibit contrasting effects (49).

In our quest for an elaborate comprehension of the nuanced alterations of diverse immune cell subsets in the realm of MSA-2 therapy, we embarked on a secondary analysis within an immune-excluded tumor model (EMT-6), encompassing an array of immune system components ( Figure 3A ). Our initial focus was on the exploration of functional disparities between the MSA-2 treatment group and the CTL group within the broader population of TME cells ( Figure 3B ) (26). Notably, the MSA-2 group exhibited a substantially higher proportion of T cells and NK cells compared to the CTL group, while the proportion of tumor cells exhibited a converse trend ( Figure 3C ). The diverse immune cell populations are distinguished by their specific biomarkers. For example, annotations of Ncr1, Klra7 and Klree1 designate NK cells, and annotations of Cd79a, Cd79b and Cd19 represent B cells ( Figure 3D ). The GSEA analysis is subsequently conducted to explore the inherent discrepancies between the two groups in the context of the immune cell population. After administration of MSA-2, T cells are characterized within the GO annotation by the escalation of the level of Myc target genes, notch signaling associated genes, TNF-α associated genes, E2F target genes, IFN-γ associated genes, and G2/M associated genes ( Figure 3E ). In the scale of KEGG annotation, the levels of immune-activating and response signaling also significantly increase ( Figure 3E ). In the domain of NK cells, as formidable sentinels orchestrating immune surveillance and targeted elimination of tumor cells, the results of GSEA reveal a noteworthy similarity in immune-stimulating signaling between NK cells and T cells after MSA-2 administrated. However, the NK cells signaling profile undergoes the additional influence of IL-6 and IL-2 signals ( Figure 3F ). In the transcriptive profile of cDCs and macrophages, the aforementioned immune-facilitating signal processes are similarly observed, but their specific distinctive activating signaling presents simultaneously ( Figures 3G, H ).

Then, the CellChat package was utilized to infer biologically significant cell communication events ( Supplementary Files: Supplementary Figures S1 , S2 ). The MSA-2 group, intriguingly, demonstrated a comprehensive enhancement in cellular communications, encompassing both the number and strength of interactions ( Figures 4A–C ). Significantly, MSA-2 augmented intercellular communication within immune response pathways, including C-C motif chemokine ligand (CCL), Secreted Phosphoprotein 1 (SPP1), C-X-C motif chemokine ligand (CXCL), TNF, Growth Differentiation Factor (GDF), and IFN-II ( Figures 4D, E ). Notably, CCL, SPP1, and CXCL were specific to MSA-2 ( Figure 4F ). Subsequently, we conducted a comprehensive examination of cell signaling pathways unique to the MSA-2 and CTL groups across different cellular subtypes ( Figures 4G–I ). Of all the signaling patterns, CXCL exhibits modest cellular interaction but presented a relatively higher correlation with prognosis in our previous analyses. More interaction analysis on CXCL signaling patterns was conducted, and it revealed that there was higher communication strength between antigen presentation cells and other components with the MSA treatment through autocrine and paracrine effects under the CXCL signaling pattern ( Figure 5A ). In T cells, besides biologically significant cell communications were observed involving CCL, SPP1, and CXCL pathways, exclusive to the MSA-2 group. Recognizing the crucial role of T cells in the cancer-immunity circle, intercellular interactions centered around T cells were highlighted within the CCL, SPP1, and CXCL pathways. Within CCL signaling, MSA-2 significantly strengthened communications between T cells and macrophages/monocytes/neutrophils, and improved the communications between T cells and NK cells/cDC/pDC/T cells compared to CTL ( Figures 5C, D, I, J ). In the realm of SPP1 signaling, MSA-2 strikingly consolidated communications between T cells and macrophages/monocytes/tumor cells, particularly tumor cells. Additionally, it amplified intercellular interactions of T cells with NK cells/cDC/pDC/B cells/mast cells/T cells in comparison with CTL ( Figures 5E, F, K, L ). Analogously, CXCL signaling moderately fortified the interactions between T cells and monocytes/cDC/tumor cells/macrophages ( Figures 5G, H, M, N ).

To assess the feasibility of using MSA-2 as a monotherapy or as a synergistic agent with ICB in cervical cancer treatment, we then conducted an in vivo murine assay ( Figure 6A ). After the administration of MSA-2, the tumor volume was significantly reduced in mice, whether they were treated with anti-PD-1 or not, as compared to the CTL group ( Figures 6B, D ). Furthermore, the combination treatment of MSA-2 and anti-PD-1 exhibited a remarkably suppressive effect on tumor growth when compared to either MSA-2 administration or anti-PD-1 monotherapy ( Figures 6B, D ). Upon termination of the experiment, following euthanasia, the weight of each tumor was assessed, confirming the previous observation that MSA-2 administration effectively suppressed tumor growth and synergistically enhanced the therapeutic efficacy of anti-PD-1 treatment ( Figures 6C, E ). The overall survival of C57BL/6 mice with subcutaneous U14 cervical cancer cells was significantly prolonged by both MSA-2 and anti-PD-1 monotherapy. Moreover, the combination strategy of MSA-2 and anti-PD-1 demonstrated a remarkable extension in overall survival of U14 C57BL/6 models, surpassing the efficacy observed with anti-PD-1 monotherapy ( Figure 6F ). The IF analysis of murine U14 tumor tissue was subsequently conducted, revealing a more extensive distribution of CD3+ and CD8+ cells in the combination groups compared to the anti-PD-1, MSA-2, or CTL group ( Figures 6G–I ). Meanwhile, the TUNEL assay was performed to assess the apoptotic status of tumor tissue, illuminating a higher abundance of apoptotic tumor cells in the synergistic therapeutic approach involving MSA-2 and anti-PD-1 ( Figures 6G, J ).