Targeted inhibition of ATR kinase promotes antitumor immunity in HNSCC through enhanced STING activation and T cell function Authors: Fabio HB da Costa, Mutsuki Kawabe, Tsung-You Tsai, Kento Okamoto, Mason D. Bartels, Sofia Cortes, Ratna Veeramachaneni, Roberto Rangel, Mitchell J. Frederick, Vlad C. Sandulache, Andrew G. Sikora, Jeffrey N. Myers, Abdullah A. Osman Published: 2026 ABSTRACT Highlights•AZD6738 potentiates anti-PD-1 efficacy by remodeling the tumor immune microenvironment.•Treatment increases cytotoxic T cells, reduces Tregs and suppressive myeloid cells.•ATR inhibition drives M1 polarization via cGAS/STING and decreases STAT3 activation.•Transcriptomic analyses reveal enhanced interferon signaling and suppression of EMT.•Study supports AZD6738 development with anti-PD-1 therapy for HNSCC. Introduction Oral cavity head and neck squamous cell carcinoma (HNSCC) has an incidence of over 40,000 new cases in the US, and over 500,000 worldwide annually, and over half of these patients will ultimately die of their disease [[1], [2], [3]]. The overall 5-yaer survival for patient with oral cavity head and neck squamous cell carcinoma have not improved over the last three decades despite significant progress in treatment modalities consisting of surgery; radiation; chemotherapy; and, more recently, immunotherapy [1,2,4,5]. Although immune checkpoint inhibitor (ICI) immunotherapy prolongs survival for some HNSCC patients, most patients with recurrent/metastatic (R/M) HNSCC showed limited clinical benefit [[6], [7], [8]]. Recently, results from the KEYNOTE-689 clinical trial demonstrated that adding neoadjuvant and adjuvant pembrolizumab to standard care significantly improved event-free survival in participants with locally advanced head and neck squamous cell carcinoma (HNSCC) [9]. Therefore, novel targeted agents that potentiate the response to immune checkpoint inhibitors are urgently needed. Cancer cells are often defective in DNA damage response (DDR) and repair, creating a therapeutic potential for further DNA damage. Growing evidence indicates that alterations in DNA damage response-related genes, including somatic and germline mutations, single nucleotide polymorphisms (SNPs), and promoter methylation are closely associated with the initiation, progression, and therapeutic response of head and neck squamous cell carcinoma (HNSCC) [10]. One potential mechanism for sensitizing HNSCC tumor cells to immunotherapy is targeting the DNA damage response/repair and replication stress (RS) pathways. Several studies have demonstrated that DNA damage can alter systemic immune functions in cancer cells [11,12]. Deficiency in the DDR leads to the accumulation of DNA damage and enhances tumor immunogenicity, suggesting that combining DDR inhibitors with immunotherapy may improve therapeutic outcomes. Head and neck squamous cell carcinomas (HNSCCs) frequently show diminished or lost TP53 function, often resulting from genomic alterations or human papillomavirus (HPV) infection [13]. As a consequence, these tumors rely heavily on the S and G2/M cell cycle checkpoints for proper regulation. Both checkpoints are activated by Ataxia Telangiectasia and Rad3-related protein (ATR), which is typically overexpressed in HNSCC compared to surrounding normal tissue [13]. ATR acts as a serine/threonine protein kinase and it is critical to the cell’s response to replication stress. It is normally activated in response to stalled DNA replication forks to promote G2–M cell-cycle checkpoint transition and fork restart [14]. Pharmacological inhibition of ATR has emerged as a promising strategy for cancer treatment that exploits synthetic lethal interactions with proteins involved in DNA damage repair, overcomes resistance to other therapies and enhances antitumor immunity [15]. AZD6738 (ceralasertib) is a selective, oral inhibitor of the ATR kinase currently in late-stage clinical development [[13], [14], [15], [16]]. It is being evaluated in multiple clinical trials for the treatment of solid tumors [17,18]. However, the effect of AZD6738 on the tumor microenvironment (TME) and immune cells have not been thoroughly investigated in HNSCC. In this study, we evaluated the efficacy of AZD6738 both as a monotherapy and in combination with an anti-PD-1 antibody in the syngeneic mouse oral cancer 1 and 2 (MOC1 and MOC2) cells. Our data demonstrated that AZD6738 significantly enhanced the in vivo response to anti-PD-1 therapy. This enhanced efficacy was attributed to substantial improvements in the tumor microenvironment (TME), including increased numbers and functionality of cytotoxic T cells, along with a reduction in immunosuppressive tumor-infiltrating regulatory T cells (Tregs) and myeloid cells. Furthermore, ATR inhibition with AZD6738 promoted M1 macrophage polarization and modulated a range of cytokines that support T cell activation, both in vivo and ex vivo. Mechanistically, the immune response appeared to be partially driven by the induction of DNA damage response and activation of the cGAS/STING signaling pathway, accompanied by decreased STAT3 phosphorylation. Unbiased RNA sequencing coupled with hallmark gene set enrichment analysis revealed that the combination of AZD6738 and anti-PD-1 antibody treatment significantly enhanced interferon signaling while simultaneously suppressing epithelial-mesenchymal transition (EMT) pathways in the preclinical MOC1 tumor model. Collectively, these findings provide compelling evidence that ATR inhibition represents a bona fide therapeutic vulnerability in HNSCC. Moreover, the concurrent targeting of ATR and immune checkpoints may offer a promising therapeutic strategy for HNSCC. Given that oncogene-induced replication stress is a key driver of genomic instability across multiple cancer types, our findings are likely to have translational relevance beyond HNSCC. Materials and methods Cell lines and cell culture Murine oral cancer 1 and 2 (MOC1 and MOC2) cell lines were obtained from our collaborator, Dr. Ravindra Uppaluri (Dana-Farber Cancer Institute, Harvard University, Boston, MA, USA). The Rangel oral cancer 1 (ROC1) murine cell line was developed in our laboratory. These cell lines were cultured and maintained in mycoplasma-free condition to ensure result reproducibility as previously described [19,20]. The ATR inhibitor, AZD6738 (ceralasertib) used in this study was obtained from AstraZeneca through an existing MTA. The STING antagonist, H-151 was obtained from MedChemExpress. Generation of ATR knockdown stable cell lines The mouse ATR shRNA Lentiviral Particles, cat# sc-29,764-V and Control shRNA Lentiviral Particles, cat# sc-108,080 were obtained from Santa Cruz Biotechnology. The ATR shRNA (m) lentiviral particles is a pool of concentrated, transduction ready viral particles containing 3 target-specific constructs that encode 19–25 nucleotides (plus hairpin) shRNA designed to knock down ATR gene expression. Briefly, the murine MOC1 and MOC2 target cells were infected with the lentiviral ATR shRNA and control non-targeting (NT) particles in the presence of polybrene (polybrene (5 μg/mL) for 48 h. The cells were then grown in fresh media containing puromycin ((1.0 μg/mL) for selection of stably transfected cells. Successful ATR knockdown in the stable cell lines was confirmed by Western blotting. Clonogenic survival assay To determine colony survival in vitro, MOC1, MOC2 and ROC1 cells were seeded in 6-well plates and exposed to various doses of the ATR inhibitor, AZD6738 (0–5.0 μmol/L). To determine whether the growth-suppressive effect of AZD6738 depends on STING signaling, MOC1 cells were pretreated with 0.1 µmol/L of STING antagonist (H-151) followed by treatment with AZD6738 as indicated. Clonogenic survival was performed as previously described by our group [21]. Western blot analysis Minimally toxic drug doses of AZD6738 (0.5–2.0 µmol/L) were identified from clonogenic survival assay IC50 values and used for further in vitro experiments. Briefly, MOC1 and MOC2 cells were treated in 10-cm dishes with AZD6738 (DMSO, 0.5 and 1.0 μmol/L) for 12 h. To assess whether the cytotoxic effect of AZD6738 is mediated through STING signaling, MOC1 cells were pretreated with 0.1 µmol/L of H-151 followed by treatment with AZD6738 as previously indicated. For Whole cell lysates were prepared, and Western blot analyses were performed as described previously [21]. Antibodies used for Western blotting were described in Supplementary Materials and Methods section. Immunofluorescence MOC1 and MOC2 cells were plated in duplicate onto the culture slides and treated with AZD6738 (2.0 µmol/L) for 24 h. Cells were then fixed in 2% paraformaldehyde for 1 hour, washed, permeabilized in 0.2% Triton X-100 in PBS for 20 min, washed, and blocked for 1 hour at room temperature in PBS buffer containing 2% normal goat serum, and 0.3% Triton X-100. The culture slides were incubated with double standard DNA marker antibody (dsDNA Marker from mice (1:100, #sc-58,749, Santa Cruz Biotechnology), Histone H2A.X (Ser139) (20E3, 1:200, cat# 9718, Cell Signaling); overnight at 4 °C followed by Alexa Fluor 594-conjugated secondary antibodies, respectively. Nuclei were counterstained with antifade with DAPI. Fluorescent foci were imaged and acquired on Olympus IX71 microscope. Foci were captured and counted in 4 different fields, and data were presented as percentage of cells and intensity with double standard foci using ImageJ software. For colocalization analysis of phosphorylated STAT3 and arginase-1 (Arg-1), MOC1 tumors harvested from mice treated with IgG or AZD6738 were stained with an arginase-1 monoclonal antibody (1:200; MA5–31,577, Thermo Fisher Scientific Invitrogen) and a phospho-STAT3 (Tyr705) (D3A7) antibody (1:200; #9145, Cell Signaling Technology). Appropriate secondary antibodies conjugated to Alexa Fluor 594 and Alexa Fluor 488 were used, respectively. Immunofluorescence staining was performed as described above. In vivo syngeneic orthotopic mouse model of oral cancer All animal experimentation was approved by the Institutional Animal Care and use Committee of the University of Texas MD Anderson Cancer Center. To evaluate the combination effect of AZD6738 and anti-PD-1 on tumor growth, 1.5 × 105 cells of MOC1 and MOC2 (7.5 × 104) cells were suspended in 30 μl of medium without supplement and injected into the lateral tongues of 7–8-week-old male C57BL/6 J mice strain because oral cancer is mostly prevalent in male patients. The mice were purchased from the Jackson Laboratory and acclimated for one week prior to experiments. Mice were then randomized into different groups 8 to 10 days after injection. Mice were treated with AZD6738 (50 mg/kg, q. d. 3 days on, 4 days off, given as oral gavage, p.o.) alone, anti-PD-1 antibody, BioXCell # BE0146 (250 μg/body, twice/week, given intraperitoneally, i.p.) alone and in combination with the two drugs after cell inoculation (when tumors were <3 mm3 in size) for 4 weeks. Control mice group received anti-IgG2a antibody, BioXCell #BE0089 (250 μg/mice, twice/week, given intraperitoneally, i.p.). Each treatment group contained 8–10 mice, and 90% of the mice experienced tumor growth under each condition. Mice were monitored twice a week, and their weight and tumor volume were recorded. Tongue tumors were measured in millimeters with digital calipers, and tumor volume was calculated as A × B2 × (π/6), where A is the longest dimension of the tumor and B is the dimension of the tumor perpendicular to A and reported as± standard error mean (±SEM), and the model assumes that the errors are evenly distributed. Tumor measurement continued until end of study or when tumors reached the ethical limit when they lost >20% of their preinjection body weight. Mice were then euthanized and tongues, tumors were harvested and subjected to immunohistochemical and immunoprofiling analyses. For myeloid-derived suppressor cell (MDSC) depletion experiment, a separate group of mice were injected with MOC1 cells and treated with anti-Ly6G/Ly6C (Gr-1) antibody, BioXCell #BE0075 (250 μg/body, twice/week, given intraperitoneally, i.p.) alone, and with the combination of anti-PD-1 and AZD6738 and the experiment was carried out as described above. Immunohistochemical analysis Sections were prepared from formalin-fixed paraffin-embedded (FFPE) specimens of mouse tumor tissues following treatment with the drugs as indicated. The antigen was retrieved by incubation at 95 °C in citric acid antigen repair solution for 30 minutes. Following blocking in methanol and water, sections were incubated in 2.5% normal goat serum made in TBST for 30 min, washed and stained with anti-CD8 (D4W2Z) rabbit mAb (1:200, #98,941, Cell Signaling Technology), anti-FoxP3 (D6O8R) Rabbit mAb (1:200, #9145, Cell Signaling Technology), anti-Phospho-STING (Ser366) polyclonal antibody (1:200, #PA5–105,674, Invitrogen) and anti-Arginase-1 (GT5811) mouse mAb antibody (1:200, #MA5–31,577, Thermo Scientific Invitrogen) at 4 °C overnight. Binding antibodies were detected using conjugated secondary antibodies as appropriate and incubated for 1 h at room temperature. The sections were then developed with DAB solution, and counterstained with hematoxylin. Images were acquired using Leica DMLA microscope (Leica Microsystems). IHC quantification was defined as the density of cells and average optical density per area (n = 3 stained tumors; every antibody staining had 5 field views). Ex vivo differentiation of murine bone marrow (BM)-derived cells and full spectrum flow cytometry For immunoprofiling of tumors, murine tongue tumors (five per group) were excised, minced with scissors, and digested in DMEM containing collagenase type I (1 mg/mL), collagenase type IV (250 U/mL), and DNase I (20 U/mL) (all from EMD Millipore) for 60 min at 37 °C with gentle rocking. Digested tissue was passed through a 70 µm nylon mesh to generate single-cell suspensions. For immunoprofiling of murine bone marrow (BM)-derived cells, cells were isolated from the femurs and tibias of 6–8-week-old C57BL/6 mice. Bones were flushed with sterile PBS, and single-cell suspensions were filtered through a 70 µm strainer, followed by red cell lysis. BM cells were plated in 35 mm dishes, at 1.0 × 105 cells/dish, in a 1:1 mixture of fresh complete DMEM medium (10% FBS, 1% penicillin/streptomycin) and MOC1-derived CM, with or without AZD6738 at 1 µM. Cultures were maintained for 4 days at 37 °C with 5% CO₂, with a medium and drug refresh, on Day 2, equal to as done at plating. See illustration in Fig. 3A, step 2. Single cells from either tumors or bone marrow cultures were prepared and extracellular staining was performed by incubation with anti-mouse CD16/CD32 Fc block (BD Biosciences), for 5 min at RT, to prevent nonspecific binding, followed by staining with a fixable viability dye and fluorophore-conjugated antibodies for 30 min at 4 °C, protected from light, diluted in BD Horizon™ Brilliant Stain Buffer (BD Biosciences). For intracellular staining, cells were fixed and permeabilized using the Intracellular Fixation and Permeabilization Buffer Set (eBioscience), according to the manufacturer’s instructions, and subsequently stained with intracellular antibodies for 30 min at RT. All the antibodies used are listed in the Supplementary Materials and Methods section. Cells were washed two times in permeabilization buffer and a third time in PBS 2% FBS and finally resuspended in PBS 2% FBS for final cell acquisition at the cytometer. Samples were acquired on a 5-laser Cytek Aurora (Cytek Biosciences). Compensation was performed using Cytek’s SpectroFlo software, and data were analyzed using FlowJo v10 (FlowJo LLC). T cell proliferation assay After 4 days of conditioning (Fig. 3A, Step 2), BM cells were harvested, washed, and plated in 96-well U-bottom plates for a T cell suppression assay. Murine splenic T cells were isolated from C57BL/6 mice using negative selection (Pan T Cell Isolation Kit II, mouse, Miltenyi Biotec) and labeled with 5 µM CFSE (carboxyfluorescein succinimidyl ester; Thermo Fisher) for 5 min at RT, in RPMI 10% FBS, according to manufacturer’s recommendations. Labeled T cells were then washed and added to BM cells at a 1:1 ratio (typically 1.0 × 10⁵ cells/well) in the presence of plate-bound anti-CD3 (1 µg/mL) and soluble anti-CD28 (3 µg/mL) to stimulate activation. Cocultures were maintained for 72 h in complete RPMI. At endpoint, cells were stained with anti-CD3 and anti-CD4/CD8 antibodies and analyzed by flow cytometry. T cell proliferation was assessed based on CFSE dilution within the CD3+ T cell population using FlowJo software (BD Biosciences). Proliferation index and division index were calculated using the FlowJo proliferation tool. Cytokine array profiling Conditioned media (CM) was collected from MOC1 cells transduced with either a pLenti-vector control or (m)ATR shRNA and subjected to multiplex analysis of 44 mouse cytokines, chemokines and growth factors using the Luminex™ 200 system (Luminex, Austin, TX, USA) by Eve Technologies crop. (Calgary, Alberta, Canada). The assay was performed on the CM fractions according to the manufacturer’s protocol. The assay sensitivities of these biomarkers range from concentration of 0.3–30.6 pg/mL for the 44-plex. The heat map analysis was generated with Cluster 3.0 and TreeView software (Stanford University). Mouse xenograft bulk RNA sequencing (RNA-seq) MOC1 cells, were injected in oral tongues of male C57BL/6 mice and treated with AZD6738, anti-PD-1 antibody and in combination as described above. Primary tumor tissues from each group were isolated and frozen in liquid nitrogen. Total RNA was extracted from tumor samples using the RNeasy mini kit reagents (QIAGEN) according to the manufacturer's instructions. RNA purity and concentration were determined using a NanoDrop 2000 (Thermo Scientific). RNA-seq was performed by LC Sciences (Houston, TX) with 2 µg of total RNA. Briefly, RNA quality was verified on Bioanalyzer 2100 (Agilent) and RNA of ≥7.07 was used for library preparation. Libraries were prepared using Illumina TruSeq mRNA Prep Kit which involved poly-A enrichment according to manufacturer’s guidelines. The mRNA was fragmented, reverse-transcribed, and strand-specific sequencing libraries were constructed. The resultant library size was about 300±50 bp and sequencing was performed on an Illumina Novaseq™ 6000 sequence platform at 2 × 1 50 pair-end configuration. The normalized expression values of FPKM were used for the analysis. Weakly expressed and noninformative (non-aligned) genes were filtered out before normalization. The sequencing raw reads were trimmed by Cutadapt tool, filtered, and cleaned before being aligned to the mouse reference genome (GRCm39). Alignment was performed by HISAT2 alignment package. The mapped reads of each sample were assembled, and gene expression was quantified by StringTie and ballgown. Gene differential expression analysis was performed between the samples by DESeq2 software and edgeR package. Data were filtered and then transformed to log2 counts per million and principal component analysis was performed in the R-statistical analysis. Differentially expressed genes (DEG) were identified across the samples using the DESeq2 software and edgeR package respectively. Significance of differentially expressed genes was set on the basis of the FDR cutoff below 0.05 and absolute fold change ≥2. The filtering criteria include selection of DEG with an adjusted P value of < 0.05, keeping genes that have at least 50 average reads and 2-fold change in the higher of the comparison among the treatment groups and controls, and exclusion of noncoding RNAs. The top 100 significantly expressed genes (i.e., up- or downregulated among the groups) were considered for further analysis based on the above statistical criteria and depicted in the cluster heatmaps generated by a ggplot2 plotting package. Volcano plots were generated to illustrate significantly upregulated and downregulated genes. Gene set enrichment analysis (GSEA) was conducted using the FGSEA (Fast Gene Set Enrichment Analysis) package and Hallmark gene sets from MSigDB (v7.5.1), with parameters set to minSize = 15, maxSize = 5000, and eps = 0. Enrichment significance was assessed using the normalized enrichment score (NES) and FDR-adjusted q-values. Selected Hallmark pathways were further visualized through enrichment plots, bubble plots, and heatmaps, providing a comparative summary of NES values across different treatment conditions. Statistical analysis GraphPad Prism software (version 8.0) software was used to perform data analysis The Student’s t-test and one-way ANOVA tests were carried out to analyze in vitro data separately. For mouse studies, a 2-way ANOVA test was used to compare tumor volumes between control and treatment groups. For immunohistochemical analyses, a one-way ANOVA and a chi-square test were used to compare immunostaining between control and treatment groups. The Kaplan–Meier method with log-rank test was used to analyze mice survival following drug treatment. Data were expressed as mean ± standard error and P values <0.05 was considered significant. Results Pharmacological inhibition of ATR enhances anti-PD-1 cytotoxicity and leads to tumor regression in vivo in a MOC1 tumor-bearing model To evaluate the effect of ATR inhibitor and in combination with PD-1 blockade, immunocompetent C57BL/6 mice bearing MOC1 and MOC2 tongue xenograft tumors, were treated with AZD6738, anti-PD-1 antibody, and in combination as indicated (Fig. 1A). No single agent activity with the AZD6738 was observed in vivo in mice bearing MOC1 or MOC2 tumors (Fig. 1B, and supplemental figure S1B). Anti-PD-1 antibody alone had modest effect on tumor growth inhibition that did not reach significance compared to IgG treated control in MOC1 (Fig. 1b). Additionally, anti-PD-1 had no effect on tumor growth in MOC2 tumor xenografts (supplemental figure S1B). However, in MOC1 model, the combination of AZD6738 and anti-PD-1 antibody significantly suppressed tumor growth and prolonged survival of mice compared to control and other treatment groups (*P < 0.0001 and *P = 0.001, respectively) (Fig. 1B and C). Reduced tumor growth and increased mice survival were also observed in MOC2 model following treatment with combination compared to other treatment groups (**P = 0.014; supplemental figure S1B and S1C respectively). No weight loss was observed in mice indicating that the combination therapy was well-tolerated (Fig. 1D, and supplemental figure S1D). Additionally, single tumor growth kinetics showed complete tumor response in a large number of mice treated with the combination therapy, compared to the other treatment groups (Fig. 1E–H). Since AZD6738 (ceralasertib) activity has been shown to be dependent on the presence of CD8+T cells [22], we next collected tongue tumors from mice following end of treatment and initially analyzed cytotoxic CD8+T cells and immunosuppressive tumor infiltrating T regulatory cells by immunohistochemistry. The analysis revealed that the CD8+ T lymphocytes were significantly increased in murine tumors treated with combination of the drug compared to other treatment groups (P = 0.0001; Fig. 1I and J). Increased levels of CD8+ T lymphocytes were observed in the MOC2 tumor model (**p < 0.0001, supplemental figure S1E- and S1F). Furthermore, the immunosuppressive FoxP3+ Treg cells were significantly reduced (**P = 0007; Fig. 1K and 1 L) in tumors treated with AZD6738 plus anti-PD-1 antibody compared to other treatment groups, indicating the immunostimulatory effect of the combination therapy. These data suggest that AZD6738 modulates T cell function and effector/suppressor balance, consistent with a need for intact adaptive immunity to benefit from combination therapy.Fig. 1Pharmacological inhibition of ATR enhances anti-PD-1 cytotoxicity and leads to tumor regression in vivo in MOC1 tumor-bearing model. To evaluate the combination effect of AZD6738 and anti-PD-1 on tumor growth, 1.5 × 105 cells of MOC1 cells were inoculated orthotopically in the oral tongue of C57BL/6 mice and treated with drugs as indicated. (A) Diagram showing the treatment schedule. (B) Tumor growth curves calculated after 4 weeks of treatment with the drugs. Statistical analyses were performed by a Two-way ANOVA test and P values < 0.05 were considered significant. (C) Kaplan–Meier survival curve for overall survival of mice injected with MOC1 cells with and without treatment with drugs. The combination therapy significantly reduced tumor growth and prolonged survival in mice injected with MOC1 cells orthotopically in the oral tongue compared to control and other treatment groups (*P < 0.0001 and *P = 0.001, respectively). (D) Percent of relative body weight loss during the treatment with the drugs demonstrating that the combination therapy was well-tolerated. (E-H) Tumor volume growth curves for individual mice in each treatment group. (I) Immunohistochemistry images for CD8+ T cells staining in tissue sections obtained from tongue xenografts bearing tumors with MOC1 cells upon treatment with drugs. (J) Quantification of the CD8+ T cells staining (brown color). (K) Immunohistochemistry images for FoxP3+staining. (L) Quantification of the FoxP3+ staining (brown color). Scale bar = 20 μm for microscopic images. The CD8+ T lymphocytes were elevated in mice tumors treated with combination of the drug compared to other treatment groups (P = 0.0001). Inhibitory FoxP3+ Treg cells were significantly reduced (**P = 0007) in mice tumors treated with the combination of the drugs. Blue color indicates nuclear staining by hematoxylin.Fig 1 dummy alt text Pharmacological inhibition of ATR enhances anti-PD-1 cytotoxicity and leads to tumor regression in vivo in MOC1 tumor-bearing model. To evaluate the combination effect of AZD6738 and anti-PD-1 on tumor growth, 1.5 × 105 cells of MOC1 cells were inoculated orthotopically in the oral tongue of C57BL/6 mice and treated with drugs as indicated. (A) Diagram showing the treatment schedule. (B) Tumor growth curves calculated after 4 weeks of treatment with the drugs. Statistical analyses were performed by a Two-way ANOVA test and P values < 0.05 were considered significant. (C) Kaplan–Meier survival curve for overall survival of mice injected with MOC1 cells with and without treatment with drugs. The combination therapy significantly reduced tumor growth and prolonged survival in mice injected with MOC1 cells orthotopically in the oral tongue compared to control and other treatment groups (*P < 0.0001 and *P = 0.001, respectively). (D) Percent of relative body weight loss during the treatment with the drugs demonstrating that the combination therapy was well-tolerated. (E-H) Tumor volume growth curves for individual mice in each treatment group. (I) Immunohistochemistry images for CD8+ T cells staining in tissue sections obtained from tongue xenografts bearing tumors with MOC1 cells upon treatment with drugs. (J) Quantification of the CD8+ T cells staining (brown color). (K) Immunohistochemistry images for FoxP3+staining. (L) Quantification of the FoxP3+ staining (brown color). Scale bar = 20 μm for microscopic images. The CD8+ T lymphocytes were elevated in mice tumors treated with combination of the drug compared to other treatment groups (P = 0.0001). Inhibitory FoxP3+ Treg cells were significantly reduced (**P = 0007) in mice tumors treated with the combination of the drugs. Blue color indicates nuclear staining by hematoxylin. Combination of AZD6738 and anti-PD-1 modulates the tumor immune microenvironment and enhances the M1-macrophage polarization in vivo in MOC1 tumor-bearing model To comprehensively evaluate the antitumor effects of AZD6738 combined with anti-PD-1 antibody on the tumor microenvironment (TME), we conducted an additional in vivo experiment using the MOC1 mouse model, as previously described. Tongue tumors were collected from five mice per group 18–20 days after treatment, and the composition of immune cell populations was analyzed using full-spectrum flow cytometry. In this experiment, combination treatment resulted in tumor growth reduction and was well tolerated in mice (Fig. 2A and 2 B), consistent with data shown in Fig. 1B and 1 D Among all treatment groups, the combination of AZD6738 and anti-PD-1 antibody significantly increased the proportion of activated CD8⁺ T cells, as indicated by the ratio of CD44⁺PD-1⁻ to CD44⁺PD-1⁺ cells (Fig. 2C). Compared to untreated controls, all treatments reduced the frequency of PD-1⁺ CD8⁺ T cells, with the most significant decrease observed in the combination treatment group (Fig. 2D, and supplemental figure S2A and S2B). Anti-PD-1 antibody monotherapy led to a modest reduction in PD-1 expression on CD4⁺ T cells, whereas the combination treatment resulted in a more pronounced decrease in PD-1⁺ CD4⁺ cells (Fig. 2E, and supplemental figure S2A and S2B). Macrophage polarization shifted toward an anti-tumor state, as evidenced by an increased M1-like/M2-like ratio, with the most pronounced effect observed in the combination treatment group (Fig. 2F, and supplemental figure S2A and S2B). Functionally, treatment led to an increase in Granzyme B⁺ CD8⁺ T cells, with the highest levels observed in the combination group (Fig. 2G, and supplemental figure S2A and S2B), indicating enhanced cytotoxic activity. The CD8⁺/CD4⁺ T cell ratio increased across all treatment groups, with the highest levels observed in the anti-PD-1 antibody and combination treatments (Fig. 2H, and supplemental figure S2A and S2B). Collectively, these findings indicate that combined ATR inhibition and PD-1 blockade reduces T cell exhaustion, enhances the population of cytotoxic CD8⁺ effector T cells, and promotes macrophage polarization toward an M1-like phenotype, features consistent with a more immunostimulatory tumor microenvironment than that induced by either agent alone. This enhanced immune activation is associated with significantly reduced tumor progression and improved survival. Myeloid-derived suppressor cells (MDSCs) are known to suppress tumor-reactive T cell responses and limit the efficacy of immunotherapies. To investigate whether efficacy of AZD6738 + anti-PD-1 combination therapy is limited by MDSCs, mice bearing MOC1 orthotopic tongue tumors were treated with an anti-Gr-1 antibody to selectively deplete granulocytic MDSCs (G-MDSCs). Gr-1⁺ cell depletion significantly enhanced the therapeutic effect of AZD6738 combined with anti-PD-1. Compared to the AZD6738 plus anti-PD-1 group, the addition of anti-Gr-1 antibody further suppressed tumor growth (P < 0.0011, supplemental figure S3A). This triple combination also significantly improved survival relative to monotherapy and dual treatment groups (log-rank P = 0.031 supplemental figure S3B). Flow cytometry confirmed effective target engagement: CD11b⁺Ly6G⁺ G-MDSCs within the CD45⁺ tumor-infiltrating leukocyte population were markedly reduced with anti–Gr-1 treatment but remained elevated in the AZD6738 plus anti–PD-1 group (Supplemental figure S2B and S3C). Together, these results indicate that Gr-1⁺ MDSCs limit the efficacy of ATR inhibition combined with PD-1 blockade, and that their depletion further enhances anti-tumor responses in vivo.Fig. 2Combination of AZD6738 and anti-PD-1 modulates the tumor immune microenvironment and enhances the M1-macrophage polarization in vivo in MOC1 tumor-bearing model. MOC1 cells were inoculated in the tongues of C57BL6 mice as indicated and treatments started nine days post-inoculation and followed by full-spectrum flow cytometry to assess the composition of immune cell populations in the mice tumors. (A) Tumor growth curves calculated after treatment with drugs. (B) Percent of relative body weight loss during the treatment with the drugs demonstrating that the combination therapy was well-tolerated. (C) The proportion of activated CD8⁺ T cells in mice tumors, as indicated by the ratio of CD44⁺PD-1⁻ to CD44⁺PD-1⁺ cells following treatment with the combination (combo) of AZD6738 and anti-PD-1 antibody. (D-E) Reduced frequency of PD-1⁺ CD8⁺ T and PD-1⁺ CD4⁺ T cells respectively in mice tumors after treatment with the combination. (F) M1-like to M2-like macrophage ratio demonstrating increased M1- macrophage polarization in mice tumors upon treatment with the combination. (G) Increased population of Granzyme B+ CD8+ T cells indicates activation of these cytotoxic T cells in mice tumors following treatment with AZD6738 and anti-PD-1 antibody. H, Combination treatment with AZD6738 and anti-PD-1 enhanced the ratio of CD8+to CD4+ T cells in mice tumors.Fig 2 dummy alt text Combination of AZD6738 and anti-PD-1 modulates the tumor immune microenvironment and enhances the M1-macrophage polarization in vivo in MOC1 tumor-bearing model. MOC1 cells were inoculated in the tongues of C57BL6 mice as indicated and treatments started nine days post-inoculation and followed by full-spectrum flow cytometry to assess the composition of immune cell populations in the mice tumors. (A) Tumor growth curves calculated after treatment with drugs. (B) Percent of relative body weight loss during the treatment with the drugs demonstrating that the combination therapy was well-tolerated. (C) The proportion of activated CD8⁺ T cells in mice tumors, as indicated by the ratio of CD44⁺PD-1⁻ to CD44⁺PD-1⁺ cells following treatment with the combination (combo) of AZD6738 and anti-PD-1 antibody. (D-E) Reduced frequency of PD-1⁺ CD8⁺ T and PD-1⁺ CD4⁺ T cells respectively in mice tumors after treatment with the combination. (F) M1-like to M2-like macrophage ratio demonstrating increased M1- macrophage polarization in mice tumors upon treatment with the combination. (G) Increased population of Granzyme B+ CD8+ T cells indicates activation of these cytotoxic T cells in mice tumors following treatment with AZD6738 and anti-PD-1 antibody. H, Combination treatment with AZD6738 and anti-PD-1 enhanced the ratio of CD8+to CD4+ T cells in mice tumors. AZD6738 increases T cell function ex vivo and negatively modulates proliferation of M2 macrophages and MDSC cells in mouse oral tumor cells To investigate the direct effects of ATR inhibition on myeloid cell differentiation and immunosuppressive phenotypes, murine bone marrow (BM)-derived cells were treated ex vivo with conditioned media (CM) from MOC1 or MOC2 tumor cells, in the presence or absence of the ATR inhibitor AZD6738. CM was harvested after 48 h of tumor cell culture in serum-free medium and mixed 1:1 with fresh complete medium to stimulate BM cells for four days, with media and drug refreshed on day two (Fig. 3A). Flow cytometric analysis of the resulting BM-derived populations revealed that AZD6738 treatment significantly reduced the overall proportion of macrophages, particularly in cultures stimulated with MOC2 CM (Fig. 3B). Within the macrophage compartment, the frequency of M2-like macrophages induced by MOC 1 conditioned medium was significantly decreased following ATR inhibition (Fig. 3C). In addition, expression of immunoregulatory markers such as IDO in both macrophages and granulocytic myeloid-derived suppressor cells (G-MDSCs) was significantly downregulated in the presence of AZD6738 (Fig. 3D and 3 E, respectively). To assess the effect of ATR inhibition during myeloid differentiation on myeloid cell T cell suppressive function, BM-derived cells were co-cultured with CFSE-labeled murine splenic T cells for three days. BM cells conditioned with either MOC1 or MOC2 CM in the presence of AZD6738 exhibited significantly reduced suppressive activity, resulting in enhanced T cell proliferation compared to vehicle-treated controls (Fig. 3F). Collectively, these results suggest that ATR inhibition directly modulates myeloid cell differentiation and impairs the acquisition of immunosuppressive phenotypes that restrict T cell responses.Fig. 3AZD6738 increases T cell function Ex vivo and negatively modulates proliferation of M2 macrophages and MDSC cells in mouse oral tumor cells. (A) Graphic methods summary for collection of conditioned medium (CM) from MOC1 and MOC2 cancer cells (Step 1), and for the culture conditions of murine BM cells stimulated with CM and AZD6738 (Step 2). BM cells were cultured for four days in a 1:1 mixture of fresh complete DMEM and CM, with and without AZD6738 at 1 µM. On day 4, BM cells are harvested for flow cytometry and T cell suppression assays. (B-E) Flow cytometry analysis of BM cells treated with CM and AZD6738 showing populations of macrophages (F4/80+), M2 macrophages (CD206+, F4/80+), Indoleamine 2,3-dioxygenase (IDO)-expressing macrophages and granulocytic MDSCs (G-MDSC; Ly6G+, CD11b+). (F) Harvested BM cells were cocultured with splenic T cells (1:1 ratio) to assess suppression of T cell activation and proliferation. Student t-tests were performed for statistical analysis, where *p < 0.05, **p < 0.01, and ***p < 0.001 considered significant.Fig 3 dummy alt text AZD6738 increases T cell function Ex vivo and negatively modulates proliferation of M2 macrophages and MDSC cells in mouse oral tumor cells. (A) Graphic methods summary for collection of conditioned medium (CM) from MOC1 and MOC2 cancer cells (Step 1), and for the culture conditions of murine BM cells stimulated with CM and AZD6738 (Step 2). BM cells were cultured for four days in a 1:1 mixture of fresh complete DMEM and CM, with and without AZD6738 at 1 µM. On day 4, BM cells are harvested for flow cytometry and T cell suppression assays. (B-E) Flow cytometry analysis of BM cells treated with CM and AZD6738 showing populations of macrophages (F4/80+), M2 macrophages (CD206+, F4/80+), Indoleamine 2,3-dioxygenase (IDO)-expressing macrophages and granulocytic MDSCs (G-MDSC; Ly6G+, CD11b+). (F) Harvested BM cells were cocultured with splenic T cells (1:1 ratio) to assess suppression of T cell activation and proliferation. Student t-tests were performed for statistical analysis, where *p < 0.05, **p < 0.01, and ***p < 0.001 considered significant. ATR inhibition in oral cancer cells promotes a secretome that supports activation of the immune microenvironment To investigate how ATR depletion in MOC1 tumor cells alters their secretome and modulates the immune microenvironment, we performed a cytokine array using conditioned media (CM) collected from MOC1 cells transduced with either a pLenti-vector control or mouse ATR (mATR) shRNA, as described in the Methods section. Cytokine profiling revealed distinct changes following ATR knockdown, with several pro-inflammatory cytokines—including IFN-γ, MIP-3α (CCL20), and IL-9—markedly elevated in CM from ATR knockdown (KD) cells compared to controls (Fig. 4A–D). Conversely, inflammatory cytokines such as G-CSF, IL-11, and IL-16 were significantly reduced following ATR depletion (Fig. 4A, and E–G). To assess the functional impact of these secretome changes, murine bone marrow (BM)-derived cells were preconditioned with CM from control or ATR KD MOC1 and MOC2 cells, followed by co-culture with murine splenic T cells to evaluate CD4⁺ and CD8⁺ T cell activation and proliferation. Notably, BM cells preconditioned with CM from ATR KD cells induced robust T cell proliferation and activation, whereas CM from control cells suppressed T cell responses (Fig. 4H). Successful ATR knockdown in mouse tumor cells was confirmed by Western blotting (Fig. 4I). Collectively, these findings indicate that ATR depletion in oral tumor cells enhances a lymphoid-driven immune response while suppressing a myeloid-mediated inflammatory response, thereby promoting immune activation within the tumor microenvironment.Fig. 4ATR Inhibition in oral cancer cells promotes a secretome that supports activation of the immune microenvironment. (A) Heatmap showing relative expression levels of cytokines detected in the conditioned media (CM) from MOC1 cells expressing either a pLenti-vector control or mouse ATR (mATR) shRNA knockdown (KD) as determined by the cytokine array. (B-D) Bar graphs showing significant and selectively differentially expressed cytokines upregulated in ATR KD conditioned media. (E-G) Bar graphs showing significant and selectively differentially expressed cytokines downregulated in ATR KD conditioned media. Data are shown as mean ± SD from two independent biological samples. (H) Bar graph showing proliferation of CFSE-labeled murine splenic T cells after 3-day coculture with bone marrow-derived cells pre-treated with CM from pLenti-vector control or ATR KD MOC1 cells. I, Western-blot analysis confirming successful knockdown of ATR in MOC1 and MOC2 cell lines. Data are representative of at least two independent experiments. Student t-tests were performed for statistical analysis, where *p < 0.05, **p < 0.01, and ***p < 0.001 considered significant.Fig 4 dummy alt text ATR Inhibition in oral cancer cells promotes a secretome that supports activation of the immune microenvironment. (A) Heatmap showing relative expression levels of cytokines detected in the conditioned media (CM) from MOC1 cells expressing either a pLenti-vector control or mouse ATR (mATR) shRNA knockdown (KD) as determined by the cytokine array. (B-D) Bar graphs showing significant and selectively differentially expressed cytokines upregulated in ATR KD conditioned media. (E-G) Bar graphs showing significant and selectively differentially expressed cytokines downregulated in ATR KD conditioned media. Data are shown as mean ± SD from two independent biological samples. (H) Bar graph showing proliferation of CFSE-labeled murine splenic T cells after 3-day coculture with bone marrow-derived cells pre-treated with CM from pLenti-vector control or ATR KD MOC1 cells. I, Western-blot analysis confirming successful knockdown of ATR in MOC1 and MOC2 cell lines. Data are representative of at least two independent experiments. Student t-tests were performed for statistical analysis, where *p < 0.05, **p < 0.01, and ***p < 0.001 considered significant. ATR inhibition induces DNA damage response and cGAS-STING activation associated with decreased STAT3 phosphorylation in syngeneic oral cancer cells. We evaluated the effect of AZD6738 monotherapy on three representative syngeneic murine oral cancer cell lines using a clonogenic survival assay, as previously described. All cell lines exhibited sensitivity to AZD6738, with half-maximal inhibitory concentrations (IC₅₀) ranging from 0.18 to 0.74 µmol/L. Among them, MOC1 cells were the most sensitive to treatment (Fig. 5A and B). The accumulation of cytosolic double-stranded DNA (dsDNA) resulting from unrepaired DNA damage can activate the cyclic GMP-AMP synthase (cGAS)–stimulator of interferon genes (STING) pathway [[23], [24], [25]]. This activation triggers the production of type I interferons and other pro-inflammatory cytokines, contributing to an immunostimulatory tumor microenvironment [[23], [24], [25]]. As shown in Fig. 4, AZD6738 treatment elevated levels of pro-inflammatory cytokines. Given its ability to exacerbate DNA damage and promote cytosolic DNA accumulation [[23], [24], [25]], we hypothesize that ATR inhibition may enhance cGAS/STING pathway activation, thereby amplifying anti-tumor immune responses in addition to its direct cytotoxic effects. Treatment of MOC1 and MOC2 cells with the ATR inhibitor, AZD6738 in vitro led to a significant induction of DNA damage and activation of STING, as indicated by a time and dose-dependent increase in phosphorylation of γH2AX and STING, respectively (Fig. 5C, and supplemental figure S1A). Phosphorylated H2AX (known as γ-H2AX) is a well-established and highly sensitive cellular marker for DNA double-strand breaks (DSBs). Therefore, rapid phosphorylation of H2AX on serine 139, forming visible nuclear foci at the sites of damaged chromatin was observed, confirming induction of DNA damage response following treatment with AZD6738 in vitro in MOC1 cells (supplemental figure S4A and S4B). Additionally, AZD6738 treatment inhibited CHK1 activity, suggesting the induction of replication stress (Fig. 5C, and supplemental figure S1A). Notably, the drug also suppressed STAT3 activation in MOC1 and MOC2 cells, indicating that ATR inhibition may interfere with immunosuppressive signaling and potentially shift the tumor microenvironment toward a more pro-inflammatory, immune-permissive state. Evidence of cytosolic double-stranded DNA (dsDNA), detected through immunostaining using a dsDNA-specific antibody, further supports activation of innate immune signaling following AZD6738 treatment (Fig. 5D–F). To determine whether the tumor cell-killing effects of AZD6738 are mediated through the STING-TBK1 signaling axis, we utilized H-151, a potent and selective STING antagonist. H-151 inhibits STING signaling primarily by suppressing STING palmitoylation and preventing TBK1 phosphorylation. MOC1 cells were pretreated with H-151 for 8 h prior to AZD6738 exposure (supplemental figure S5A). Pharmacological inhibition of STING partially rescued colony survival following AZD6738 treatment, as demonstrated by colony formation assays (Supplemental figure S5B and S5C). In addition, western blot analysis revealed that H-151 pretreatment attenuated AZD6738-induced TBK1 phosphorylation, a key downstream mediator of STING signaling, in MOC1 cells in vitro (Supplementary Fig. S5D). Taken together, these results indicate that ATR inhibition induces MOC1 tumor cell death, at least partially through a STING-dependent mechanism, further confirming the importance of STING signaling in regulating antitumor immune responses. In vivo, immunohistochemical analysis of MOC1 tumors from mice treated with AZD6738, anti-PD-1 antibody, or the combination revealed increased phosphorylation of STING, with the combination therapy showing the most pronounced effect compared to IgG-treated controls (Fig. 5G and H). This in vivo increase in phospho-STING levels corroborates the enhanced STING phosphorylation observed in vitro by western blotting. Collectively, these findings demonstrate that ATR inhibition either alone or in combination with PD-1 blockade, activates the cGAS/STING innate immune pathway, contributing to antitumor responses in immunocompetent mouse model of oral cancerFig. 5ATR inhibition induces DNA damage response and cGAS-STING activation associated with decreased STAT3 phosphorylation in syngeneic oral cancer cells. To explore the degree of sensitivity of the syngeneic mouse model to ATR inhibition, MOC1, MOC2 and ROC1 murine cell lines were treated in vitro with various concentrations of single agent AZD6738 as indicated. (A) Representative images of clonogenic survival assays. (B) Survival curves demonstrating calculated IC50 values ranging from 0.18 to 0.74 µmol/L across all tested cell lines, with MOC1 cells exhibiting the highest sensitivity to treatment. (C) Western blot analysis showing induction of DNA damage markers and increased phosphorylation of the cGAS/STING pathway and decreased STAT3 activity in MOC1 cells treated with various doses of AZD6738 in vitro at the time indicated. (D-F) Enhanced levels of cytosolic double-stranded DNA (dsDNA), detected through immunostaining using a dsDNA-specific antibody following AZD6738 treatment presented as percentage of cells with ds-DNA per nucleus or fluorescence intensity divided by nuclear area respectively. (G) Immunostaining demonstrating increased STING phosphorylation in vivo in tumors bearing MOC1 cells. (H) Quantification of the STING phosphorylation in mice tumors presented as percent of immunostaining intensity. Scale bar for all microscopic images is 100 µm.Fig 5 dummy alt text ATR inhibition induces DNA damage response and cGAS-STING activation associated with decreased STAT3 phosphorylation in syngeneic oral cancer cells. To explore the degree of sensitivity of the syngeneic mouse model to ATR inhibition, MOC1, MOC2 and ROC1 murine cell lines were treated in vitro with various concentrations of single agent AZD6738 as indicated. (A) Representative images of clonogenic survival assays. (B) Survival curves demonstrating calculated IC50 values ranging from 0.18 to 0.74 µmol/L across all tested cell lines, with MOC1 cells exhibiting the highest sensitivity to treatment. (C) Western blot analysis showing induction of DNA damage markers and increased phosphorylation of the cGAS/STING pathway and decreased STAT3 activity in MOC1 cells treated with various doses of AZD6738 in vitro at the time indicated. (D-F) Enhanced levels of cytosolic double-stranded DNA (dsDNA), detected through immunostaining using a dsDNA-specific antibody following AZD6738 treatment presented as percentage of cells with ds-DNA per nucleus or fluorescence intensity divided by nuclear area respectively. (G) Immunostaining demonstrating increased STING phosphorylation in vivo in tumors bearing MOC1 cells. (H) Quantification of the STING phosphorylation in mice tumors presented as percent of immunostaining intensity. Scale bar for all microscopic images is 100 µm. Transcriptomic profiling identifies interferon signaling, cell cycle regulation, and EMT suppression as major features of the response to ATR inhibition in vivo in MOC1 tumors To characterize the transcriptional effects of AZD6738-based treatment regimens in vivo, bulk RNA sequencing was performed on MOC1 orthotopic tongue tumors harvested from C57BL/6 mice following treatment with IgG control, AZD6738, anti-PD-1, or the combination of AZD6738 and anti-PD-1. Unsupervised hierarchical clustering of the top 100 differentially expressed genes demonstrated distinct transcriptomic profiles across treatment groups, indicating substantial treatment-induced transcriptional remodeling (Fig. 6A; and supplemental Table S1). Notably, tumors treated with AZD6738 alone or in combination with anti-PD-1 segregated from control tumors, suggesting that ATR inhibition profoundly alters the tumor transcriptional landscape. Differential gene expression analysis further revealed extensive transcriptional changes in AZD6738-treated tumors relative to IgG controls (Fig. 6B; and supplemental Table S2). Among the most significantly altered genes were those associated with DNA damage response (DDR), immune signaling, and cell cycle regulation, consistent with the established role of ATR inhibition in modulating replication stress responses and downstream cellular programs. To define the biological pathways affected by treatment, we performed gene set enrichment analysis (GSEA) using Hallmark gene sets. Both AZD6738 monotherapy and combination therapy were associated with significant enrichment of interferon-γ and interferon-α response pathways compared with IgG-treated tumors (Fig. 6C and D; and supplemental Table S2), indicating robust activation of innate and adaptive immune signaling networks. In contrast, epithelial-mesenchymal transition (EMT) signatures were significantly suppressed in tumors receiving combination therapy, suggesting that ATR inhibition may also modulate tumor cell plasticity and phenotypes associated with invasion and immune evasion. Given the substantial overlap in transcriptional programs induced by AZD6738-containing regimens, we next compared the combination treatment directly with anti-PD-1 monotherapy and IgG controls to identify molecular changes attributable to ATR inhibition. These analyses revealed distinct pathway-level alterations associated with the addition of AZD6738 (Fig. 6E; and supplemental figure S6). Specifically, Hallmark pathway analysis demonstrated enhanced enrichment of immune activation and apoptosis-related programs in the combination group, whereas pathways governing cell cycle progression remained among the most prominently regulated transcriptional networks (Fig. 6E; and supplemental figure S6; and supplemental Table S2). To facilitate a systematic comparison across treatment groups, normalized enrichment scores (NES) for 15 selected Hallmark pathways were summarized in a comparative heatmap (Fig. 6F). Interferon-α and interferon-γ response signatures emerged as the most consistently and strongly enriched pathways following AZD6738 treatment, with the greatest enrichment observed in the combination arm. These findings are consistent with previous reports linking ATR inhibition-induced replication stress to activation of inflammatory and immune signaling pathways [14,22,23]. In parallel, cell cycle-associated programs, including E2F targets and the G2/M checkpoint, were positively enriched in AZD6738-treated tumors, reflecting activation of checkpoint and DDR-related processes. Metabolic pathways such as glycolysis exhibited context-dependent regulation, showing differential enrichment across treatment comparisons. Collectively, these transcriptomic analyses identify interferon signaling, cell cycle regulation, and suppression of EMT-associated programs as major features of the transcriptional response to AZD6738 in MOC1 tumors. These findings provide mechanistic insight into the antitumor activity of ATR inhibition and support its ability to remodel both tumor-intrinsic and immune-related signaling pathways, particularly when combined with PD-1 blockade.Fig. 6Transcriptomic profiling identifies interferon signaling, cell cycle regulation, and EMT suppression as major features of the response to ATR inhibition in vivo in MOC1 tumors. Tumor samples were collected from black mouse xenografts following treatment with drugs as indicated and subjected to RNA sequencing followed by differential expression and pathway enrichment analyses. (A) Heatmap showing the top 100 differentially expressed genes across the four treatment groups. Gene expression values are Z-score normalized across samples, with clustering highlighting treatment-specific transcriptional patterns. (B) Volcano plots of differentially expressed genes expression comparing control IgG vs AZD6738 (left) and control IgG vs combination (combo) (right). Significantly upregulated and downregulated genes are highlighted (red and blue, respectively), with selected genes annotated. (C-D) Representative GSEA enrichment plots showing hallmark pathways significantly enriched in AZD6738 vs control IgG (top) and Combo vs control IgG (bottom). Key enriched pathways include interferon gamma response, interferon alpha response, and epithelial–mesenchymal transition. (E) Bubble plot summarizing hallmark pathway enrichment for control IgG vs Combo. The normalized enrichment score (NES) is shown on the x-axis, bubble size indicates the fraction of core genes contributing to enrichment, and bubble color reflects the NES direction and magnitude. Dashed lines indicate FDR thresholds of 0.25, 0.05, and 0.01 respectively. (F) Heatmap of hallmark pathway enrichment scores (NES) across comparisons. Pathways related to immune signaling (interferon alpha/gamma response), cell cycle regulation (E2F targets, G2M checkpoint), and stress responses (unfolded protein response, apoptosis) show differential activation among treatment groups.Fig 6 dummy alt text Transcriptomic profiling identifies interferon signaling, cell cycle regulation, and EMT suppression as major features of the response to ATR inhibition in vivo in MOC1 tumors. Tumor samples were collected from black mouse xenografts following treatment with drugs as indicated and subjected to RNA sequencing followed by differential expression and pathway enrichment analyses. (A) Heatmap showing the top 100 differentially expressed genes across the four treatment groups. Gene expression values are Z-score normalized across samples, with clustering highlighting treatment-specific transcriptional patterns. (B) Volcano plots of differentially expressed genes expression comparing control IgG vs AZD6738 (left) and control IgG vs combination (combo) (right). Significantly upregulated and downregulated genes are highlighted (red and blue, respectively), with selected genes annotated. (C-D) Representative GSEA enrichment plots showing hallmark pathways significantly enriched in AZD6738 vs control IgG (top) and Combo vs control IgG (bottom). Key enriched pathways include interferon gamma response, interferon alpha response, and epithelial–mesenchymal transition. (E) Bubble plot summarizing hallmark pathway enrichment for control IgG vs Combo. The normalized enrichment score (NES) is shown on the x-axis, bubble size indicates the fraction of core genes contributing to enrichment, and bubble color reflects the NES direction and magnitude. Dashed lines indicate FDR thresholds of 0.25, 0.05, and 0.01 respectively. (F) Heatmap of hallmark pathway enrichment scores (NES) across comparisons. Pathways related to immune signaling (interferon alpha/gamma response), cell cycle regulation (E2F targets, G2M checkpoint), and stress responses (unfolded protein response, apoptosis) show differential activation among treatment groups. ATR inhibition attenuates arginase-1 expression and its co-localization with phosphorylated STAT3 in vivo in the MOC1 tumor model The ATR inhibitor AZD6738 has been shown to inhibit CHK1, leading to reduced phosphorylation of STAT3, particularly at Tyr705 site, and consequent attenuation of STAT signaling, which impacts response to therapy and cancer cell survival [27,28]. Myeloid-derived suppressor cells (MDSCs) are key mediators of immunosuppression in multiple cancers, including head and neck squamous cell carcinoma (HNSCC) [27,28]. STAT3 activation is central to MDSC expansion and immunosuppressive function within the tumor microenvironment, promoting MDSC differentiation through upregulation of arginase-1 (Arg1), depletion of l-arginine, increased oxidative stress, and modulation of iNOS, ultimately suppressing T-cell responses [29]. Based on these findings, we investigated whether the reduced MDSC populations observed in the MOC1 tumor model are mechanistically driven by AZD6738-mediated suppression of the STAT3–arginase-1 signaling axis. To evaluate the impact of ATR inhibition on myeloid differentiation, arginase-1 (Arg1) and iNOS expression were assessed in murine bone marrow (BM)–derived cells treated ex vivo with conditioned media (CM) from MOC1 cells or MOC1 cells transduced with either pLenti control vector or mouse ATR (mATR) shRNA, in the presence or absence of AZD6738, as described in the Methods. Flow cytometric analysis revealed that both AZD6738 treatment and ATR knockdown significantly reduced arginase-1 mean fluorescence intensity (MFI) in monocytic (M-MDSC) and granulocytic (G-MDSC) populations, as well as in macrophages (Fig. 7A). In addition, CM from ATR shRNA–transduced MOC1 cells induced a significant increase in iNOS-positive G-MDSCs (Fig. 7A, left panels). To determine whether ATR inhibition directly modulates myeloid differentiation via suppression of the STAT3–arginase signaling axis, immunofluorescence analysis was performed to assess STAT3 and Arg1 co-localization in MOC1 tumors. AZD6738-treated tumors exhibited minimal co-localization of phosphorylated (active) STAT3 with Arg1 compared with IgG-treated controls, which showed higher levels of co-localization (Fig. 7B). Consistently, immunohistochemical analysis demonstrated significantly reduced Arg1 expression in AZD6738-treated tumors relative to IgG-treated controls (Fig. 7C and D). Collectively, our data provide evidence that ATR inhibition, in addition to activating the cGAS–STING pathway, also suppresses the STAT3-Arg1 axis in oral tumors, thereby reducing the immunosuppressive activity of MDSCs, key drivers of resistance to immunotherapy (Fig. 7E).Fig. 7ATR inhibition attenuates arginase-1 expression and its co-localization with phosphorylated STAT3 in vivo in the MOC1 tumor model. (A) Flow cytometric analysis of bone marrow (BM) cells cultured with conditioned media (CM) derived from parental MOC1 cells or MOC1 cells transduced with either a pLenti control vector or mouse ATR (mATR) shRNA, in the presence or absence of AZD6738. Mean fluorescence intensity (MFI) was quantified for Arg1⁺ monocytic MDSCs (M-MDSCs), Arg1⁺ granulocytic MDSCs (G-MDSCs), Arg1⁺ macrophages, and iNOS⁺ G-MDSCs. MFI represents the average fluorescence intensity of the stained cell population and reflects per-cell expression levels of Arg1 or iNOS. Increased Arg1 MFI concomitant with reduced iNOS MFI is indicative of enhanced immunosuppressive capacity. (B) Immunofluorescence staining demonstrating co-localization of phosphorylated STAT3 and Arg1 in mouse tumors treated in vivo with IgG control vehicle or AZD6738. (C) Immunohistochemical detection of Arg1 expression in mouse MOC1 tumors. Hematoxylin and eosin (H&E) staining of corresponding tumor sections is shown. (D) Quantification of Arg1 expression shown in panel (D), presented as the percentage of immunostaining intensity. Student t-tests were performed for statistical analysis, and data were expressed as mean ± standard error where p < 0.05, considered significant. (E) A proposed mechanistic model illustrating a potential synergistic immune effect of combined ATR and PD-1 inhibition, driving diverse alterations in the tumor microenvironment in the MOC1 tumor model. ATR is activated in response to DNA damage, replication stress, accumulation of single-stranded DNA, and increased stalled replication forks, initiating a kinase signaling cascade involving CHK1. This cascade promotes activation of signal transducer and activator of transcription 3 (STAT3) and arginase-1 in both tumor cells and myeloid-derived suppressor cells (MDSCs). Consequently, ATR signaling enhances the expansion and differentiation of MDSCs, leading to suppression of T-cell function and impairment of innate immune responses. AZD6738 inhibits ATR signaling and disrupts these immunosuppressive mechanisms. In addition, ATR inhibition results in the accumulation of cytosolic DNA, which may activate the cGAS–STING pathway and induce a type I interferon response, thereby enhancing innate immunity. Combination treatment with AZD6738 and immune checkpoint blockade (anti-PD-1) acts synergistically to enhance tumor cell killing.Fig 7 dummy alt text ATR inhibition attenuates arginase-1 expression and its co-localization with phosphorylated STAT3 in vivo in the MOC1 tumor model. (A) Flow cytometric analysis of bone marrow (BM) cells cultured with conditioned media (CM) derived from parental MOC1 cells or MOC1 cells transduced with either a pLenti control vector or mouse ATR (mATR) shRNA, in the presence or absence of AZD6738. Mean fluorescence intensity (MFI) was quantified for Arg1⁺ monocytic MDSCs (M-MDSCs), Arg1⁺ granulocytic MDSCs (G-MDSCs), Arg1⁺ macrophages, and iNOS⁺ G-MDSCs. MFI represents the average fluorescence intensity of the stained cell population and reflects per-cell expression levels of Arg1 or iNOS. Increased Arg1 MFI concomitant with reduced iNOS MFI is indicative of enhanced immunosuppressive capacity. (B) Immunofluorescence staining demonstrating co-localization of phosphorylated STAT3 and Arg1 in mouse tumors treated in vivo with IgG control vehicle or AZD6738. (C) Immunohistochemical detection of Arg1 expression in mouse MOC1 tumors. Hematoxylin and eosin (H&E) staining of corresponding tumor sections is shown. (D) Quantification of Arg1 expression shown in panel (D), presented as the percentage of immunostaining intensity. Student t-tests were performed for statistical analysis, and data were expressed as mean ± standard error where p < 0.05, considered significant. (E) A proposed mechanistic model illustrating a potential synergistic immune effect of combined ATR and PD-1 inhibition, driving diverse alterations in the tumor microenvironment in the MOC1 tumor model. ATR is activated in response to DNA damage, replication stress, accumulation of single-stranded DNA, and increased stalled replication forks, initiating a kinase signaling cascade involving CHK1. This cascade promotes activation of signal transducer and activator of transcription 3 (STAT3) and arginase-1 in both tumor cells and myeloid-derived suppressor cells (MDSCs). Consequently, ATR signaling enhances the expansion and differentiation of MDSCs, leading to suppression of T-cell function and impairment of innate immune responses. AZD6738 inhibits ATR signaling and disrupts these immunosuppressive mechanisms. In addition, ATR inhibition results in the accumulation of cytosolic DNA, which may activate the cGAS–STING pathway and induce a type I interferon response, thereby enhancing innate immunity. Combination treatment with AZD6738 and immune checkpoint blockade (anti-PD-1) acts synergistically to enhance tumor cell killing. Discussion Although PD-1-directed immunotherapy benefits some patients with head and neck squamous cell carcinoma (HNSCC), most patients do not meaningfully respond, and numerous clinical trials have failed to identify effective targeted therapies. Recent findings from the KEYNOTE-689 clinical trial revealed that incorporating neoadjuvant and adjuvant pembrolizumab into standard therapy significantly enhanced event-free survival in patients with locally advanced head and neck squamous cell carcinoma (HNSCC) [9]. Ataxia telangiectasia and Rad3-related protein (ATR) is a key component of the cellular response to DNA damage and replication stress. Inhibition of ATR using AZD6738 (ceralasertib) impairs DNA repair in cancer cells, thereby increasing genomic instability and promoting apoptosis [22,23]. Beyond its direct anti-cancer effects, AZD6738 has also been shown to modulate the tumor microenvironment, rendering it more immunogenic. Specifically, it enhances the infiltration and activation of cytotoxic CD8⁺ T cells, which play a crucial role in tumor cell elimination. In this study, we demonstrate that AZD6738 exerts broad immunomodulatory effects within the tumor immune microenvironment (TIME) of a preclinical head and neck cancer model. The treatment not only enhances T-cell activation but also alters the number and function of myeloid cells. Recent preclinical studies have shown that the pleiotropic effects of AZD6738 are highly dependent on dose and treatment schedule, with intermittent dosing yielding maximal efficacy through CD8⁺ T-cell and type I interferon (IFN)-dependent mechanisms [22]. In our model, co-administration of AZD6738 with anti–PD-1 antibodies at comparable dosing levels led to optimal anti-tumor activity, characterized by increased T-cell proliferation and activation. Consistent with prior reports [22,23], we found that the combination of ATR inhibition and PD-1 blockade reduces T-cell exhaustion, expands the population of activated cytotoxic CD8⁺ effector T cells, and promotes macrophage polarization toward an M1-like phenotype. These features reflect a shift toward a more immunostimulatory tumor microenvironment compared to monotherapy. This enhanced immune modulation correlated with significantly reduced tumor progression and improved survival. The M1- macrophages, known for their pro-inflammatory and anti-tumor functions, are typically induced by cytokines such as tumor necrosis factor-α (TNF-α) and interferon-γ (IFN-γ) [26,30]. It is plausible that the observed increase in M1 macrophages in our model results from elevated IFN-γ levels following combination treatment. Myeloid-derived suppressor cells (MDSCs), which suppress tumor-reactive T-cell responses, are major obstacles to effective immunotherapy. Our in vivo and co-culture experiments provide strong evidence that ATR inhibition impairs myeloid cell differentiation and reduces the induction of immunosuppressive phenotypes that hinder T-cell activity. Furthermore, depletion of Gr-1⁺ cells enhanced the therapeutic efficacy of the AZD6738 and anti-PD-1 combination, underscoring the central role of immunosuppressive myeloid cells in limiting treatment response. Although the mechanism by which ATR inhibition diminishes immunosuppressive MDSC populations in the MOC1 model is not fully elucidated, our data suggest involvement of a CHK1-STAT3-dependent inhibitory signaling axis. Inhibition of this pathway is associated with reduced arginase-1 activity and a reciprocal increase in inducible nitric oxide synthase (iNOS) expression, consistent with previously reported findings [29]. Several studies have reported that ATR inhibition leads to the accumulation of cytosolic DNA due to unresolved DNA damage following S-phase replication stress [23,31,32]. This cytosolic DNA can trigger the cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) pathway, initiating innate immune responses. In line with these findings, we observed increased cytosolic DNA, increased DNA damage induction following AZD6738 treatment in vitro and activation of the cGAS-STING-IFN-cytokine signaling axis in vivo. This activation likely contributes to the enhanced therapeutic response observed with ATR inhibition and PD-1 blockade in our HNSCC model. In our study, we also detected an increase in MAC CD45 expression, which further supports the idea that AZD6738 treatment facilitates the maturation of NK cells. These findings suggest that the anti-tumor activity of the combination therapy may, in part, be mediated through the activation of both innate and adaptive immune pathways. Collectively, our results, along with other recent studies, highlight the cGAS-STING pathway as a critical downstream mediator for maximizing the immunotherapeutic potential of ATR-targeted therapies. Furthermore, in our study, the partial restoration of colony-forming capacity following STING blockade, together with the reduction in TBK1 phosphorylation, provides functional evidence supporting a role for STING signaling in AZD6738-induced tumor cell death. These results further strengthen our conclusion that activation of the cGAS-STING axis contributes to the immunomodulatory and antitumor effects of ATR inhibition. Bulk RNA-sequencing analysis of in vivo tumors further illuminated the molecular effects of ATR inhibition combined with PD-1 blockade in the MOC1 model. Gene set enrichment analysis (GSEA) and hallmark pathway analysis revealed robust induction of IFN-α and IFN-γ signaling, apoptotic pathways, E2F targets, and G2-M checkpoint activation, consistent with ATR inhibition–induced replication stress and enhanced immune responses. These interferon responses may be essential for effective T cell-mediated anti-tumor activity in this context. Interestingly, we also observed a significant downregulation of epithelial-to-mesenchymal transition (EMT) signatures in tumors treated with the combination therapy. This suggests that the dual treatment not only stimulates immune activation but also suppresses transcriptional programs associated with tumor invasiveness and metastatic potential. In contrast, metabolic pathways such as glycolysis displayed variable enrichment patterns, potentially reflecting differences in immune cell infiltration and stromal composition. Together, these findings indicate that ATR inhibition, particularly in combination with PD-1 blockade, induces profound transcriptional remodeling in tumors. Key pathways impacted include interferon signaling and cell cycle regulation, alongside context-dependent changes in metabolism. These transcriptomic alterations offer important mechanistic insights into the therapeutic efficacy of AZD6738 in HNSCC. However, single-cell RNA sequencing and spatial transcriptomics will be essential to identify the specific cell populations driving the observed transcriptional changes in tumors following treatment. Other ATR inhibitors, including M6620 (berzosertib), BAY 1895,344 (elimusertib), M4344 (gartisertib), and RP-3500 (camonsertib), are available. We selected AZD6738 because of its well-characterized pharmacological profile, extensive preclinical and translational validation, and established use in combination therapy studies, making it an appropriate tool compound to investigate ATR inhibition in our model. Further studies are needed to dissect the relative contributions of neoantigen load and immunologic memory, such as through tumor rechallenge experiments in mice treated with ATR inhibitor-mediated T-cell activation. In our study, STING activation by AZD6738 is unlikely to result from off-target effects and is more likely a direct consequence of CHK1 inhibition in tumors. Supporting this, Tang and colleagues have recently demonstrated that ATR inhibition enhances STING phosphorylation by disrupting ATR-CHK1 signaling, which normally suppresses the CDK1-SPOP axis and enforces cell cycle arrest [33]. In the absence of ATR activity, CDK1-SPOP signaling is activated, leading to PD-L1 degradation and accumulation of cytoplasmic DNA fragments. These fragments trigger the cGAS-STING pathway, resulting in STING phosphorylation and downstream immune activation [33]. IFN-γ is well known to activate tumor-intrinsic STING signaling by inducing DNA damage and promoting the accumulation of cytosolic double-stranded DNA (dsDNA) [34]. Therefore, future investigations will examine whether targeted inhibition of IFN-γ and/or cGAS-STING pathways can abrogate the anti-tumor immune responses elicited by AZD6738. A limitation of this study is that the initial in vivo experiments were conducted using only a single murine syngeneic oral cancer model (MOC1). To enhance the robustness and generalizability of our findings, we subsequently evaluated the therapeutic efficacy of AZD6738 in combination with anti-PD-1 therapy in the immunotherapy resistant MOC2 model. In contrast to the MOC1 model, the combination treatment elicited only modest effects on tumor growth inhibition and immune activation in MOC2 tumors (supplemental figure S1). This attenuated response may reflect the intrinsically immunosuppressive or “cold” tumor microenvironment characteristic of MOC2 tumors, which are known to exhibit relative resistance to immunotherapeutic interventions [35].In conclusion, our findings provide a strong molecular and therapeutic rationale for combining ATR inhibition with immune checkpoint blockade in the treatment of previously untreated patients with advanced head and neck cancer. Declarations Ethics approval and consent to participate All animal studies described herein were performed following approval by, and in accordance with the rules of the University of Texas MD Anderson Cancer Center Institutional Animal Care and Use Committee (Study protocol No 00,001,499-RN03). Consent for publication Not applicable. Availability of data and materials The data generated in this study are available within the article and its supplementary files. Financial support This work was supported by the National Institute of Dental and Craniofacial Research R01DE030104 (A.A. Osman and M.J. Frederick), and generous philanthropic gifts from Melissa & Albert Grobmyer, IV and Carol & Steve Mattingly (awarded to A.A. Osman and J. N. Myers). CRediT authorship contribution statement Fabio HB da Costa: Writing – review & editing, Writing – original draft, Validation, Software, Methodology, Investigation, Formal analysis, Conceptualization. Mutsuki Kawabe: Writing – review & editing, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Conceptualization. Tsung-You Tsai: Methodology, Investigation, Formal analysis. Kento Okamoto: Methodology, Investigation, Funding acquisition. Mason D. Bartels: Methodology, Investigation. Sofia Cortes: Methodology. Ratna Veeramachaneni: Methodology. Roberto Rangel: Resources, Methodology. Mitchell J. Frederick: Writing – review & editing, Visualization. Vlad C. Sandulache: Writing – review & editing. Andrew G. Sikora: Supervision, Conceptualization. Jeffrey N. Myers: Writing – review & editing, Writing – original draft, Supervision, Resources, Investigation, Funding acquisition, Conceptualization. Abdullah A. Osman: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization. Declaration of competing interest The authors declare no conflict of interest.