Adjuvant COX inhibition augments STING signaling and cytolytic T cell infiltration in irradiated 4T1 tumors

Immune therapy is the new frontier of cancer treatment. Therapeutic radiation is a known inducer of immune response and can be limited by immunosuppressive mediators including cyclooxygenase-2 (COX2) that is highly expressed in aggressive triple negative breast cancer (TNBC). A clinical cohort of TNBC tumors revealed poor radiation therapeutic efficacy in tumors expressing high COX2. Herein, we show that radiation combined with adjuvant NSAID (indomethacin) treatment provides a powerful combination to reduce both primary tumor growth and lung metastasis in aggressive 4T1 TNBC tumors, which occurs in part through increased antitumor immune response. Spatial immunological changes including augmented lymphoid infiltration into the tumor epithelium and locally increased cGAS/STING1 and type I IFN gene expression were observed in radiation-indomethacin–treated 4T1 tumors. Thus, radiation and adjuvant NSAID treatment shifts “immune desert phenotypes” toward antitumor M1/TH1 immune mediators in these immunologically challenging tumors. Importantly, radiation-indomethacin combination treatment improved local control of the primary lesion, reduced metastatic burden, and increased median survival when compared with radiation treatment alone. These results show that clinically available NSAIDs can improve radiation therapeutic efficacy through increased antitumor immune response and augmented local generation of cGAS/STING1 and type I IFNs.


Introduction
The immune checkpoints programmed cell death 1/programmed death ligand 1 (PD-1/PD-L1) are key regulators of the immune system that are crucial for self-tolerance and are exploited by tumors for their survival and disease progression.In recent years, checkpoint inhibitors have been employed in cancer therapies to limit immunosuppression and promote antitumor M1/Th1 immune responses; this has improved treatment efficacy and clinical outcomes of some tumors (1).However, there remain a substantial fraction

R E S E A R C H A R T I C L E
JCI Insight 2024;9 (12):e165356 https://doi.org/10.1172/jci.insight.165356 of cancers that do not respond to immunotherapeutic intervention, such as triple-negative breast cancer (TNBC).Recent reports have shown that patients with TNBC present with elevated PD-L1 tumor expression; however, only 8%-20% respond to checkpoint inhibitor therapy (2,3).These observations suggest that other mechanisms prevent maximum tumor immune response.
The efficacies of conventional chemotherapies and focused irradiation can be enhanced by a proactive immune response.For example, augmented radiation-induced tumor growth delay by TGF-β neutralizing antibody is completely abated by CD8 + or CD4 + T cell depletion, implicating the requirement of cytolytic T cells for improved radiation therapeutic efficacy (4).Similarly, inhibition of the immunosuppressive enzyme indoleamine 2,3-dioxygenase (IDO) also improves standard of care therapy (5).This observation is supported by recent studies in melanoma demonstrating that elevated products of tryptophan catabolism by IDO limits responsiveness of immune-based therapies (6).In addition, targeting IL-10 increased the therapeutic efficacies of radiation and CpG treatments (7,8).Importantly, IL-10 blockade increased the survival of tumor-bearing mice by 30% (9).These findings demonstrate roles of alternative immune-suppressive pathways in addition to PD-1/PD-L1 that, when targeted, enhance proinflammatory immune responses and therapeutic efficacies.
In addition to the immunosuppressive pathways described above, the inducible forms of nitric oxide synthase (iNOS or NOS2) and cyclooxygenase-2 (COX2) modulate inflammatory microenvironments (10,11).Interestingly, elevated tumor expression of NOS2 and COX2 predicts poor survival in estrogen receptor-negative (ER -) patients (12)(13)(14).Both NOS2 and COX2 promote drug resistance, metastasis, angiogenesis, and an immunosuppressive tumor microenvironment (TME) (11,(14)(15)(16)(17).Moreover, NOS2 and COX2 fortify their expression in a feed-forward manner as they drive multiple oncogenic pathways, including Erk, Akt, HIF1α, NF-κB, and TGF-β (SMAD), through generation of different cytokines including TNF-α, GM-CSF/G-CSF, IL-6, and IL-8 (11,(17)(18)(19)(20). TCGA pathway analysis of NOS2/COX2 + tumors implicates many immune pathways, including IL-17A, IFN-γ, IL-1β, and TLR signaling in ER -breast cancer (19).These data indicate that NOS2 + and COX2 + cancers induce an active immune response involving Th1-and Th17-related pathways.Given that inflammation is commonly encountered in breast cancers, the poor prognosis predicted by increased tumor expression of NOS2 and/or COX2 may in part be due to an altered tumor immune microenvironment that limits treatment efficacy and clinical outcome.Neoadjuvant therapy is a standard of care with chemo and/or radiation treatments, and this suggests that NOS2/ COX2 inhibition could provide a new opportunity for improved therapeutic efficacies.Toward this end, a recent phase 1/2 clinical trial showed an improved clinical outcome defined by an overall response rate of 45.8% in patients with drug-resistant, locally advanced breast cancer (LABC) and metaplastic TNBC who received the NOS inhibitor L-NMMA and low-dose aspirin combined with taxane (21).Importantly, 27.3% of treated patients with LABC achieved pathological complete response at surgery where remodeling of the tumor immune microenvironment was observed in therapeutic responders (21).
In addition to NOS inhibitors, nonsteroidal antiinflammatory drugs (NSAIDs) have been shown to delay tumor growth, limit chemoresistance, and reduce metastasis (22).When administered in conjunction with irradiated (120 Gy) 4T1 cells, indomethacin (INDO) dramatically induced a potent antitumor immune response (23).Antitumor immunity was long lasting in 4T1 tumor-bearing mice, where 48% of mice were resistant to a second 4T1 challenge.Remarkably, this vaccine response was effective against high but not low COX2-expressing tumor cells.Since COX2 is highly expressed in many tumors, including TNBC, these results suggest that targeting COX2 may be clinically beneficial (23).
Herein, we explored the effects of pan-NOS and -COX inhibitors on radiation therapeutic efficacy in the murine 4T1 TNBC tumor model.When administered after irradiation (IR), the pan-NOS inhibitor L-NAME modestly enhanced radiation-induced tumor growth delay but did not affect lung metastatic burden.In contrast, COX inhibition by INDO reduced primary tumor growth and decreased metastasis as a single agent and in combination with focused tumor irradiation.Using CODEX imaging, elevated densities of infiltrating CD8 + and CD4 + T cells were observed in INDO-and 6 Gy + INDO combination-treated tumors, and these elevated densities were sustained through 23 days after IR.In addition, infiltration patterns differed between treated and control tumors where high leukocyte populations defined fully inflamed tumors in response to treatment.In contrast, untreated control tumors exhibited immune deserts characterized by low levels of leukocyte infiltrates.Examination of leukocyte infiltrates at earlier time points revealed elevated leukocyte infiltration at day 7 after IR.Importantly, these results are supportive of elevated CD8 + T cell density and penetration into tumor parenchyma observed in COX2 lo TNBC tumors.Three

R E S E A R C H A R T I C L E
JCI Insight 2024;9(12):e165356 https://doi.org/10.1172/jci.insight.165356factors that appear to be important include augmented cGAS/STING and type I IFN, as well as increased tumor infiltration of cytolytic CD8 + T cells in combination-treated tumors.Together, these results suggest that improved antitumor immune response can be achieved by adjuvant COX2 inhibition in conjunction with radiation, which may provide a beneficial therapeutic option for treatment of TNBC and other high COX2-expressing tumors.

Elevated tumor COX2 expression limits radiation therapeutic efficacy and leukocyte infiltration in TNBC tumors.
More than 60% of patients with cancer receive radiation therapy as part of their treatment regimen (24).In TNBC, radiation therapy is considered an option in locally advanced cases (25).Given that elevated NOS2/COX2 tumor expression is a strong predictor of poor survival among ER -patients and patients with TNBC, we postulated that elevated expression of these enzymes might limit clinical outcome in conjunction with radiation treatment.This hypothesis was explored in a cohort of patients with TNBC (n = 147) previously treated with fractionated radiation doses totaling 50 Gy (26).When stratifying for COX2 expression, Kaplan-Meier survival analysis demonstrated that elevated COX2 tumor expression predicted poor survival (HR = 2.09; P = 0.026) (Figure 1).In contrast, elevated tumor NOS2 expression had no predictive value in the same patients (Figure 1).These results suggest that pharmacological COX2 inhibition could improve radiation therapeutic efficacy in patients with TNBC with high COX2-expressing tumors.
Tumor leukocyte infiltration and spatial localization predicts clinical outcomes (2,27).However, elevated tumor COX2 is known to promote an immunosuppressive TME (11,16,28).To further explore the role of COX2 in altered tumor immune responses, we used InSituPlex multiplex imaging to examine the density, infiltration, and localization of CD8 + T cells in COX2 hi versus COX2 lo TNBC tumors (13,14).The expression levels and spatial localization of immune biomarkers including CD3, CD4, CD8, CD68, FOXP3, PD-1, and PD-L1 (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.165356DS1)were examined relative to the tumor marker CK/SOX10 in 16 TNBC COX2 hi versus COX2 lo expressing tumors.Using this approach, Figure 2A shows 3 types of tumor immune microenvironments including (a) increased CD8 + T cell penetration into the tumor core in COX2 lo tumors (Hot-Inflamed), (b) CD8 + T cells that were spatially restricted to tumor stroma in COX2 hi tumors (Cold-Excluded), and (c) the sparse distribution or absence of CD8 + T cells in the tumor epithelium in COX2 hi tumors (Cold-Immune Desert).To further explore correlations between COX2 and CD8 + T cell density, total CD8 + T cells and CD8-to-CK/SOX10 (CK tumor marker) ratios were quantified in COX2 lo (red circles) versus COX2 hi (blue circles) tumors.This quantitative approach revealed approximately 2-to 4-fold increases in total CD8 + T cell number (Figure 2B) and CD8/CK ratio (Figure 2C) respectively, in COX2 lo tumors when compared with COX2 hi tumors.The spatial distribution of CD8 + T cells influences

R E S E A R C H A R T I C L E
JCI Insight 2024;9(12):e165356 https://doi.org/10.1172/jci.insight.165356clinical outcome (2).Next, we examined CD8 + T cell spatial distribution and found elevated CD8 + T cell penetration into COX2 lo tumor epithelium when compared with COX2 hi tumor epithelium (Figure 2D, left).Moreover, CD8 + T cells restricted to COX2 hi tumor stroma (blue squares) were increased when compared with CD8 + T cells that infiltrated COX2 hi tumor epithelium (blue circles) (Figure 2D, middle).However, the number of CD8 + T cells in COX2 lo tumor epithelium (red circles) was not significantly different than those in COX2 lo tumor stroma (red squares) (Figure 2D, right).In addition, CD8 + T cell density in COX2 lo tumor epithelium (850 cells/mm 2 ) was higher than COX2 hi tumor epithelium (189 cells/mm 2 ) (Figure 2E).These results suggest that increased COX2 tumor expression promotes increased areas of immune deserts (100 cells/mm 2 ) as previously described (2).Moreover, density heatmaps show mixed landscapes of stroma and marginally restricted CD8 + T cell aggregates as well as immune desert regions near or below 100 cells/mm 2 in COX2 hi tumors (Figure 2, F and G).In contrast, COX2 lo tumors showed increased CD8 + T cell penetration (Figure 2F) and aggregation approaching 600/mm 2 when quantified at depths of 100 μm beyond the tumor-stroma interface (Figure 2G).Further examination of the COX2/CD8 spatial relationship in COX2 hi tumors demonstrated elevated COX2 expression bordering the tumor margin that appeared to restrict CD8 + T cells penetration into the core (Figure 2H).These observations are consistent with elevated CD8 + T cells restricted to COX2 hi tumor stroma when compared with those that infiltrated into COX2 hi tumor epithelium (Figure 2D).In contrast, dramatically increased CD8 + T cell penetration into the tumor epithelium of COX-2 lo tumors was observed (Figure 2H).Importantly, quantified COX2/CD8 ratios were increased in deceased patients with TNBC when compared with those who survived at 5 years after diagnosis (Figure 2I).In addition to CD8 + T cells, CD4 + T cells were also increased in COX2 lo tumors (Supplemental Figure 2).Together, these results show that elevated COX2 expression correlates with limited CD8 + T cell density and infiltration into the tumor epithelium in TNBC and that this spatial orientation is a limiting factor in clinical outcome.
COX inhibition improves radiation therapeutic efficacy and leukocyte infiltration in 4T1 tumor-bearing mice.Given that elevated COX2 tumor expression limited radiation therapeutic efficacy (Figure 1) and that reduced CD8 + T cell infiltration and increased COX2/CD8 ratios were observed in COX2 hi tumors and deceased patients (Figure 2), COX2 effects on radiation therapeutic efficacy was further examined in 4T1 tumor-bearing mice.The murine 4T1 model was used because the disease progression of 4T1 tumors closely follows TNBC disease progression in humans, as defined by spontaneous metastasis from the primary tumor to lymph nodes, blood, liver, lung, brain, and bone (29).4T1 tumor-bearing mice were treated with 1 dose of 6 Gy x-rays (Supplemental Figure 3), and this effectively induced a tumor growth delay in 4T1 tumor-bearing mice that could be further augmented by combination treatment.A single dose of 6 Gy x-rays gave the same response as 30 Gy total dose administered in 6 Gy dose fractions in 4T1 tumor-bearing mice as reported by Vanpouille-Box et al. (4).Given that the dose enhancement ratio (DER; the slope of tumor growth in treated/control; Supplemental Table 1) in our study was consistent with that of Vanpouille-Box (4), along with limited stress to the mice and the radiation sensitivity of T cells (30), we used the single-dose method.INDO -a potent, clinically available NSAID -was used for combination treatment based upon (a) its accumulation in high COX2-expressing tumors due to its slow rate of release from the COX2 enzyme and (b) its ability to increase expression of the PGE2 consumptive enzyme PGDH (31,32).When compared with control untreated mice, intermediate tumor growth delays were observed in mice treated with 6 Gy or INDO as single agents (Figure 3A).However, when administered in combination, 6 Gy + INDO abated tumor growth through 30 days as shown in Figure 3A.Similar combination effects were observed in EMT-6 (BALB/c) and EO771 (C57BL/6) tumor-bearing mice (Supplemental Figure 4).Our earlier work reported DER in a nonmetastatic squamous cell carcinoma (SCC) murine model that showed improved radiation-induced tumor growth delay by the pan-NOS inhibitor L-NAME (DER 1.8), which involved abated IL-10 expression and increased CD8 + T cell number and activation (7).Herein, the NOS inhibitor L-NAME modestly enhanced the radiation-induced growth delay (DER 1.4) (Figure 3B) with no effect on metastatic burden (data not shown).In contrast, when compared with untreated controls, single-agent INDO and 6 Gy + INDO combination treatment reduced lung metastatic burden (Figure 3C) and improved median survival (Figure 3D) in 4T1 tumor-bearing mice.In addition, RNA-Seq gene expression analysis showed significantly reduced IL-10 gene expression by INDO treatment when compared with control untreated mice (Figure 3E), and this supports earlier studies and further implicates COX2 in the regulation of immune suppression and adaptive immunity (7,9).Importantly, these results support the Kaplan-Meier analysis shown in Figure 1 demonstrating improved radiation therapeutic efficacy in patients with TNBC with low COX2-expressing tumors.JCI Insight 2024;9(12):e165356 https://doi.org/10.1172/jci.insight.165356 Since reduced IL-10 levels improved adaptive immunity (7,9,33), we further explored the effect of COX2 blockade on adaptive immune response in the 4T1 TNBC murine model.The expression, density, and spatial location of lymphocyte biomarkers were evaluated using RNA-Seq, FACS, and Multiplex CODEX imaging in control and 6 Gy ± INDO-treated tumors harvested 7 days after IR.When compared with untreated controls, FACS analysis showed elevated CD8 + CD69 + (active) T cell populations in INDO-treated tumors, while a separate CODEX image analysis showed increased cytolytic/exhausted CD8 + ratios in 6 Gy + INDO treated tumors (Figure 4A).CODEX spatial distribution analysis also revealed increased CD8 + T cell penetration into the tumor core in INDO-treated tumors (Figure 4B).In contrast, control tumors showed sparsely populated CD8 + T cells (Figure 4A) that were largely restricted to the tumor margins (Figure 4B).While a proportional increase in lymphocytes was observed overall, the distribution of CD8 + T cells in the tumor core in response to INDO treatment was reminiscent of fully inflamed TNBC tumors shown in Figure 2A, this fully inflamed phenotype may in part account for the augmented tumor growth delay of the primary lesion (Figure 3A).Next, T cell polarization was examined in control and treated tumors using RNAScope analysis of IFN-γ and Granzyme B (GrnzB) secreted by cytolytic CD8 + T cells, as well as the immunosuppressive biomarker IL-10.When compared with untreated controls, Figure 5 shows increased IFN-γ in all treated tumors.GrnzB was enhanced in 6 Gy-treated tumors, while INDO and 6 Gy + INDO combination treatment trended higher (Figure 5).In contrast, when compared with control, the expression levels of immunosuppressive IL-10 did not change significantly in treated tumors.Interestingly, spatial analyses revealed that CD8 + T cells in treated tumors were surrounded by active CD4 + T cell and CD19 + B cell phenotypes (compare Figure 4B and Supplemental Figure 5).Given that tumor-infiltrating B cells can secrete apoptosis-inducing IgG antibodies as well as

R E S E A R C H A R T I C L E
JCI Insight 2024;9(12):e165356 https://doi.org/10.1172/jci.insight.165356function as antigen-presenting cells that prime CD4 + and CD8 + T cells, this spatial lymphoid distribution is consistent with tertiary lymphoid structures, which may suggest a unique orthogonal pattern of immune cell trafficking leading to improved therapeutic efficacy (34,35).
COX inhibition augments cGAS/STING1/type I IFN in irradiated 4T1 tumors.In addition to T cells, other CD45 + immune cells were examined (Supplemental Figure 6), including DC, macrophages, and NK cells that also express CD8.Increased trends in CD11c + CD8 + DCs and CD11c + CD8 + CD169 + cells were observed (Supplemental Figure 7).Also, elevated CD169 + MHCII + , F4/80 + CD169 + , CD11b + CD169 + , and MHCII + C-D8 + CD3 -macrophage populations were observed (Supplemental Figure 7).CD169, a biomarker of STING, is involved in antigen presentation, cross-priming, and expansion of CD8 + T cells (36)(37)(38)(39).This process occurs when CD169 + macrophages bind to sialic acids on CD8α + DCs (36,38).Given that increased CD169 + macrophage populations as well as increased trends in CD11c + CD8 + DC and CD11c + CD8 + CD169 + cells were observed (Supplemental Figure 7), RNA-Seq gene expression was analyzed for a STING signature.Figure 6 and Supplemental Figure 8 show increased expression of MUS81/EME1/PARP biomarkers known to induce cGAS/STING through the cytosolic accumulation of tumor DNA (40,41).In addition, elevated downstream cGAS/STING/type I IFN expression in 6 Gy ± INDO-treated tumors was also observed, and it began as early as day 3 and persisted through day 23 after IR (Figure 6 and Supplemental Figure 8).This cGAS/STING/type I IFN pathway is supported by increased IFN and IFN response gene expression and altered immunosuppressive gene expression shown in Supplemental Figure 9.Because these results strongly implicate a role of cGAS/ STING in the 6 Gy + INDO antitumor immune response, we examined the antitumor effects of STING agonist cGAMP in the presence and absence of INDO in 4T1 tumor-bearing mice.The STING agonist cGAMP was administered 2 times per week for 3 weeks beginning on day 7 as indicated by the arrows in Figure 7A.INDO administration in the drinking water also began on day 7 and was present continuously throughout the experiment.Importantly, the cGAMP + INDO combination treatment completely abated 4T1 tumor growth (Figure 7A) in a remarkably similar fashion to that observed in 6 Gy + INDO-treated mice (Figure 3A).However, tumor growth resumed after cGAMP treatment stopped (Figure 7B), marking abated tumor immune surveillance induced by cGAMP (42).In addition, cGAMP + INDO combination treatment increased median survival when compared with cGAMP alone (Figure 7C).These results were further examined in 4T1 tumorbearing STING1-KO mice.Despite the experiment being done under conditions of systemic STING1 depletion in the STING1-KO mouse, an effective antitumor response was achieved due to the localized response to focused irradiation of the 4T1 tumor that still expressed cGAS/STING1 (Figure 6, Figure 7D, and Supplemental Figure 8).Also, the cGAS/STING1 pathway can promote both pro-and antitumor responses (43), where STING1 signaling has a role in the development of many leukocyte populations, including Treg immunosuppressive phenotypes, which are not induced in the STING1-KO mouse.Moreover, limited tumor growth has been shown in LLC tumor-bearing STING-KO mice when compared with WT controls (44).Together, our results with focused 4T1 tumor irradiation or local administration of cGAMP implicate the importance of the localized induction of STING signaling within the tumor that can be augmented by COX inhibition to enhance antitumor immune response.In addition, 4T1 tumor cells are known to express GM-CSF, which is elevated along with the chemokine CCL2 in 6 Gy + INDO-treated mice (Supplemental Figure 10).This could also contribute to the resumed tumor growth of combination-treated mice, thus providing additional therapeutic targets for improved clinical responses.Importantly, an undesired effect of local radiotherapy involves increased release of GM-CSF/CCL2 and the formation of premetastatic niches by recruitment of M-MDSCs into the lungs of 4T1 tumor-bearing mice (45).Together, this work suggests that the 6 Gy + INDO treatment mediates a temporal progression of CD8 + T cells and antigen presenting cells with proinflammatory and antitumor function that is in part mediated by STING mechanisms, as summarized in Figure 8.This is supported by the GSE37751 database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE264712) probability of survival stratifying for the effects of COX2 in patients with TNBC with high versus low STING1 expression.When compared with high tumor COX2, low tumor COX2 expression increased the probability of survival in these patients (HR = 0.2857, P = 0.0426; Supplemental Figure 11).

Discussion
Improved clinical outcomes associated with many cancers including TNBC directly correlate with elevated infiltrating CD8 + T cells, implicating the importance of antitumor immune response for improved treatment efficacy and survival (46).These findings have been extended by a recent study showing that, in addition to increased infiltration and density, the spatial localization of CD8 + T cells was critical for improved survival of patients with TNBC (2).Tumor immune deserts (<100 CD8 + T cells per mm 2 ) exhibited increased expression of immunosuppressive B7-H4 and fibrotic signatures predictive of poor survival (2).In addition, stroma-restricted CD8 + T cells were associated with an immunosuppressed TME through elevated

R E S E A R C H A R T I C L E
JCI Insight 2024;9(12):e165356 https://doi.org/10.1172/jci.insight.165356cholesterol signatures TGF-β, IL17, and TANs, and these elevated signatures predicted poor survival.In contrast, elevated CD8 + T cell penetration into the tumor epithelium was defined as a fully inflamed tumor and correlated with increased GrnzB, type I IFN, IDO, and PD-L1 expression that predicted improved survival.Moreover, type I IFN and cholesterol biosynthesis pathways were shown to negatively regulate one another.Given that type I IFN signatures were associated with CD8 + T cell penetration into the tumor core while cholesterol signatures were associated with the restriction of CD8 + T cells in tumor stroma, these observations implicate a key role of the spatial configuration of CD8 + T cells during polarization of the tumor immune microenvironment that was further validated in a cohort of 579 patients with TNBC (2).These results suggest that CD8 + T cell orientation and increased type I IFN-associated antitumor immunity improves clinical outcomes in TNBC (2).
The inducible isoform COX2 catalyzes the first step in prostanoid synthesis and is expressed at high levels in many tumors, including TNBC (13,16,18,22).While COX2 is known to promote angiogenesis, drug resistance, and metastasis, it also contributes to immune evasion and resistance to cancer immunotherapy (11,28).The COX2/PGE2/EP signaling pathway suppresses DCs and NK cells and inhibits T cell production and responsiveness of IL-2 that limits both activation and expansion of cytolytic T cells and promotes tumor immune evasion, suggesting that COX2 blockade could restore tumor immune surveillance (11,16).Toward this end, NSAIDs have been used in both preventative and adjuvant anticancer applications (47).As adjuvant, the benefits of NSAIDs have been shown when used in conjunction with therapeutic radiation for treatment of prostate and other cancers (22,47).Herein, we extend these observations by showing that 6 Gy radiation combined with the NSAID INDO augmented cGAS/STING1, type I IFNs, and cytolytic CD8 + T cells; this augmentation appeared to restore immune surveillance, limit tumor growth and metastatic burden, and improve survival in the aggressive 4T1 TNBC model.In addition, when combined with INDO, the STING agonist cGAMP abated tumor growth (Figure 7A) in a remarkably similar fashion as the 6 Gy + INDO combination (Figure 3A) and was consistent with augmented antitumor effects of celecoxib/cyclic diadenyl monophosphate (celecoxib/CDA) combination treatment of mice with Lewis lung carcinoma (48).In our work, cGAMP withdrawal was the limiting factor (Figure 7B) as tumor volumes reached the allowable limit after cGAMP administration was stopped and mice had to be euthanized; despite this, the cGAMP/INDO combination treatment improved survival when compared with cGAMP treatment alone (Figure 7, B and C).Despite systemic depletion in STING1-KO mice, COX inhibition by INDO effectively limited tumor growth in irradiated 4T1 tumors, emphasizing the importance of the localized STING response within the tumor (Figure 7D).STING1 signaling also promotes leukocyte populations, including Treg immunosuppressive phenotypes, as well as IL-10, IDO, and COX2 immunosuppressive mediators, which are not induced under conditions of systemic STING depletion in the STING1-KO mouse.Therefore, the results herein suggest the importance of systemic depletion versus localized STING induction for restored antitumor immune surveillance in 4T1 tumor-bearing mice (44).Moreover, our results show synergistic effects of radiation/NSAID combination treatment that augmented the localized tumor induction of STING signaling and type I IFNs when compared with either treatment alone.Together, these results suggest therapeutically beneficial effects of localized STING antitumor response following radiation/NSAID combination treatment and are supported by GSE37751 database probability of survival examining the influence of tumor COX2 on STING1 (Supplemental Figure 11).
Type I IFNs mediate diverse antitumor effects, including immune surveillance of precancerous lesions, inhibitory effects against the disease progression of established tumors by augmented cell cycle arrest and apoptosis, and abated invasion and metastasis of established tumors (49).During immune surveillance, type I IFN generated by immunogenic precancerous cells activates immune cells, including DC, macrophages, and cytolytic T cells that induce IFN-γ for elimination of the neoplastic lesion (49).A plethora of studies have shown that conventional therapies, including chemo and therapeutic radiation, promote cytosolic tumor DNA accumulation and cGAS/STING-mediated type I IFN antitumor immune responses (49)(50)(51).Moreover, chimeric antigen receptor (CAR) T cell-based therapies have demonstrated the requirement of IFNAR1/2 receptor signaling for both survival and cytolytic activity of CAR T cells (52).While these and other studies have demonstrated the importance of type I IFN for anticancer immunity and improved patient survival, most tumors exhibit dysregulated type I IFN signaling, leading to protumor resistance mechanisms (53).Interestingly, low, basal levels of type I IFN expressed in cancer cells have accounted for enhanced immunity in response to immunotherapies that is driven by DNA leakage and cGAS/STING (54)(55)(56).Indeed, most tumor-relevant type I IFNs are induced by cytosolic or endolysosomal sensing of nucleic acids, where DNA that has leaked from the nucleus or  50).In addition to the direct generation of DNA fragments from double stranded (ds) DNA breaks caused by radiation, leaked DNA can arise from mutations in oncogenes and tumor suppressor genes causing dysregulated cancer cell replication and stalling of replication forks.A replication stress response is then triggered, which restarts the stalled replication forks (40).During this process, the DNA endonuclease MUS81/EME1 complex and PARP-dependent DNA repair pathways mediate shedding and cytosolic accumulation of genomic tumor DNA (40,57).The accumulation of cytosolic DNA induced STING signaling, type I and II IFN, cytolytic T cell immune responses, and macrophage-dependent tumor cell rejection (40).Moreover, CD169-expressing macrophages promote DC antigen presentation during CD8 + T cell cross-priming and activation, and DC are thought to be critical for type I IFN antitumor responses (36)(37)(38)(39).Herein, gene expression analysis showed the induction of MUS81/EME1/PARP, CD169, cGAS, STING1, type I and II IFN, and type I IFN receptors IFNAR1/2 in INDO and 6 Gy + INDO-treated tumors on days 15 and 23 after IR (Figure 6 and Supplemental Figure 8).In addition, cGAMP + INDO combination abated tumor growth in a remarkably similar fashion as 6 Gy + INDO treatment (compare Figure 3A and Figure 7).The flattened tumor growth associated with cGAMP + INDO treatment demonstrated in Figure 7A required cGAMP as tumor growth resumed upon withdrawal of the STING agonist (Figure 7B).Together, these results show that the 6 Gy + INDO combination treatment restored tumor immune surveillance against aggressive 4T1 tumors, which involved at least in part enhanced antitumor innate and adaptive immune responses through cGAS/STING/type I IFN signaling.
Herein, we provide evidence demonstrating that INDO-mediated COX2 inhibition augments MUS81-mediated cytoplasmic tumor DNA accumulation and cGAS/STING/type I IFN, which restored tumor immune surveillance via potent antitumor innate and adaptive immune responses in irradiated 4T1 TNBC tumors.Together, these results demonstrate immune changes that occur through abated COX2 signaling, which promotes altered spatial organization of M1/Th1 immune phenotypes and augmented antitumor response.Whole tumor RNA-Seq revealed important changes beginning as early as day 3 after IR that persisted throughout the experiments, showing that 6 Gy + INDO treatment supports increased cGAS/ STING signaling and augmented type I IFN expression including IFN-αβ, IFN-β1, IFN-α4, and IFN-α13 as well as increased IRF3 and type I IFN receptors IFNAR1/2 (Figure 6 and Supplemental Figure 8).In addition, increased IRF7 (Figure 6 and Supplemental Figure 8), which is a master regulator in the induction of type I IFN-β such as IFN-αβ (58), was also observed.These results highlight major changes in the spatial immune landscape of 6 Gy + INDO-treated tumors that augments antitumor immunity and reverses tumor immune suppression.Altered spatial immune landscapes were confirmed in TNBC where high CD8 + T cell penetration was observed in COX2 lo fully inflamed tumor epithelium.In contrast, stroma restricted CD8 + T cells and immune deserts were found in COX2 hi TNBC tumors (Figure 2).Importantly, COX2/CD8 ratios were elevated in deceased patients with TNBC (Figure 2H) and elevated tumor COX2 expression reduced radiation therapeutic efficacy (Figure 1).These results strongly implicate COX2 as a key mediator of immunosuppression, limited adaptive immunity, and poor clinical outcomes.Early studies show that type I IFN can provide an important antitumor tool that improves efficacies of standard-of-care therapies.However, systemic administration of these IFNs and other cytokines has severe side effects that limit their clinical use (49).Here, we show that NSAIDs and focused tumor irradiation can provide a localized tumor-specific cGAS/STING-mediated generation of type I IFN, as summarized in Figure 8, which had profound effects on the spatial localization of CD8 + T cells, favoring antitumor immune response in aggressive 4T1 murine tumors.Targeting COX2 in combination with radiation enhances local control of the primary lesion by augmented M1/Th1 immune mediators, restored immune surveillance, and reduced metastatic burden.Given the challenges associated with the clinical administration of STING agonists in cancer and immune therapy and the clinical availability of COX inhibitors, these results suggest that therapeutic radiation and COX inhibition may provide a readily available approach for the localized induction of STING antitumor response that could improve clinical outcomes in patients with aggressive TNBC tumors.

Sex as a biological variable
Pearson correlation coefficients for the incidence rates of female versus male breast cancer have been reported (59).Male breast cancer rates were generally less than 1 per 100,000 man years, in contrast to the much higher rates of female breast cancer of 122.The differences in both incidence rates and time trends between males and females may reflect sex differences in underlying risk factors, including differences in ducts and lobules and the absence of p53 mutation (60).While most males are ER + and ductal carcinoma in situ (DCIS) represents 10% of male breast cancers, TNBC is less frequent with poorer prognosis due to higher histopathological grade (61).Given these low occurrences in males, female patients with TNBC were examined for the effects of tumor NOS2 and COX2 expression on radiation therapeutic efficacy.

RNA-Seq of bulk tumor
In brief, fresh frozen tissue samples were homogenized in the presence of TRIzol (Thermo Fisher Scientific) and further purified with affinity column (RNeasy Mini Kit, Qiagen) following the manufacturer's recommendation.The RNA quality following extraction was checked in a bioanalyzer, and only samples with a RNA integrity number (RIN) larger than 6 were used to make the RNA-Seq library prep.Sample libraries were prepped with the Illumina Stranded Total RNA Prep, and paired-end sequencing was performed according to the manufacturer protocol and sequenced in a NovaSeq 600 sequencing system.Reads of the samples were trimmed for adapters and low-quality bases using Cutadapt.Sequencing data were exported and then uploaded to the Partek Flow server for subsequent sample normalization and QC steps using the built-in RNA-Seq Data Analysis workflow.Differentially expressed gene lists were generated with the Partek GSA algorithm, which applies multiple statistical models to each individual gene in order to account for each gene's varying responses to different experimental factors and to different data distributions.Prior to heatmap development in Microsoft 64-bit Excel 365, transcript counts were normalized using the default "counts per million + 0.0001" method in the Partek Flow software (Build 10.0.22.0428).All data were normalized to internal housekeeping genes.A 2-fold cutoff and P value < 0.05 filter was applied to finalize the gene lists.

Figure 1 .
Figure 1.Association between tumor COX2 expression and breast cancer survival.Kaplan-Meier and log rank test were used to determine cumulative disease-free survival curve of patients with TNBC (n = 147) by COX2 status; when compared with low COX2 tumor expression (n = 96), high COX2 (n = 51) predicted poor survival among patients who had received fractionated radiation doses totaling 50 Gy.P = 0.026.Elevated NOS2 tumor expression had no predictive value in the same patients.

Figure 2 .
Figure 2. CD8 + T cell spatial distribution in COX2 hi and COX2 lo expressing tumors.(A) CD8 + T cells (red stain), CK tumor marker (blue stain), and DAPI (white stain) showing COX2 lo "Hot-Inflamed" tumor with high CD8 + T cell penetration into tumor epithelium, COX2 hi "Cold-Excluded" tumor showing CD8 + T cells restricted to stroma, and COX2 hi "Cold-Immune Desert" showing few or absence of CD8 + T cells in the tumor epithelium.Scale bars: 200 μm.(B) CD8 + T cell quantification showing increased total CD8 + T cells in COX2 lo (red circles n = 10) versus COX2 hi (blue circles n = 6) tumors.(C) Increased CD8 + T cell/tumor CKSOX10 ratio in COX2 lo (red) versus COX2 hi (blue) tumors.(D) The left graph shows significantly elevated CD8 + T cell infiltration in COX2 lo versus COX2 hi tumors.The middle graph shows significantly reduced CD8 + T cell infiltration in COX2 hi annotated tumor regions where CD8 + T cells are highly stroma restricted.The right graph shows no significant difference between CD8 + T cells in tumor versus stroma regions in COX2 lo tumors.(E) CD8 + T cell density per mm 2 localized in tumor-(left) or stroma-annotated (right) regions in COX2 lo (red) versus COX2 hi (blue) tumors.(F) Density heat map showing elevated CD8 + T cell aggregation in COX2 lo versus COX2 hi tumors.Scale bar: 1 mm.(G) Increased number of CD8 + T cells infiltrating from tumor-stroma interface into tumor epithelium in COX2 lo (red bar) versus COX2 hi (blue bar) tumors.(H) COX2 lo expressing tumors (left panel) exhibit dramatically increased number and penetration of CD8 + T cells into tumor epithelium (white arrows).In contrast, CD8 + T cell (white arrowhead) in COX2hi tumors (right panel) are stroma restricted.DAPI (white), COX2 (green), CD8 + T cell (red), and CKSOX10 tumor marker (blue) are shown.(I) Increased COX2/CD8 + T cell ratios in patients with TNBC who succumbed to disease versus those who survived (Deceased versus Alive) at 5 years after diagnosis.*P ≤ 0.05, **P ≤ 0.0075 using Mann-Whitney U or Welch's test.

Figure 3 .
Figure 3. Antitumor effect of 6 Gy radiation ± INDO in 4T1 tumor-bearing mice.(A) A single dose of 6 Gy irradiation as well as daily INDO treatments were given on day 7 following tumor injection.After tumor irradiation, the mice were returned to their cage and given INDO (30 mg/L) in the drinking water, which continued for the duration of the experiment.The tumor growth curve shows intermediate growth delays associated with single agent 6 Gy and INDO treatments while 6 Gy + INDO combination treatment abated tumor growth.Two-way ANOVA with Tukey's multiple-comparison test was used to determine significant changes in tumor growth.(B) Modest enhancement of 6 Gy-induced growth delay by the pan-NOS inhibitor L-NAME.(C) INDO alone and 6 Gy + INDO treatments reduce lung metastatic burden when compared with control untreated mice.One-way ANOVA with Dunnett's post hoc test was used was used.(D) Improved median survival associated with 6 Gy + INDO combination treatment.P = 0.0271 log rank test for trend.(E) RNA-Seq gene expression showing reduced IL-10 gene expression in INDO and 6 Gy + INDO-treated tumors.One-way ANOVA with Dunnett's post hoc test were used *P < 0.05.

Figure 4 .
Figure 4. FACS and CODEX analyses show increased tumor infiltrating CD8 + T cells on day 7 after IR. (A) FACS and CODEX analysis show increased CD8 + CD69 + (active) T cells and increased cytolytic/exhausted CD8 + T cell ratios, respectively.(B) CD8 + T cell spatial distribution, where red dots represent the detection of > 1 CD8 + cell marker in a 25 μm diameter circle and the spatial location of the migrating cells.*P < 0.05 using Kruskal-Wallis with Dunn's post hoc test.

Figure 6 .
Figure 6.Increased cGAS/STING1 in 6 Gy + INDO-treated tumors.Heatmaps analysis of individual genes related to the cGAS/STING pathway leading to augmented type I IFN in 6 Gy-, INDO-, and 6 Gy+INDO-treated samples.The green-tored (low-to-high) color scale indicates the number of transcript counts.White boxes indicate no transcripts were found.Prior to heatmap development in Microsoft 64-bit Excel 365, transcript counts were normalized using the default "counts per million + 0.0001" method in the Partek Flow software (Build 10.0.22.0428).

Figure 7 .
Figure 7. STING agonist cGAMP + INDO abates 4T1 tumor growth.(A) The STING agonist cGAMP was administered intratumorally (5 μg/50 μL in endotoxin-free H 2 0) 2 times per week for 3 weeks beginning on day 7 as indicated by the arrows.INDO administration in the drinking water also began on day 7 and was present continuously throughout the experiment.The cGAMP + INDO combination treatment completely abated tumor growth.Two-way ANOVA with Tukey's multiple-comparison test was used.(B) Tumor growth in cGAMP + INDO resumed after cGAMP treatment stopped on day 29.(C) cGAMP + INDO combination treatment improved median survival when compared with cGAMP treatment alone.Statistical Log-rank (Mantel-Cox) test was used.(D) INDO treatment promoted antitumor effects independent of radiation in STING-KO mice indicating the importance of local tumor STING response versus systemic STING depletion in mice.

Figure 8 .
Figure 8. Indomethacin augments cytoplasmic tumor DNA accumulation and downstream induction of cGAS/ STING1, and type I IFN in treated tumors.Radiation + indomethacin for COX2 inhibition increases cytoplasmic tumor DNA accumulation through MUS81/EME1/PARP.Accumulated cytoplasmic tumor DNA then induces cGAS/STING, type I IFNs, IFNAR, and increases cytolytic T cells.