A STING Agonist Given with OX40 Receptor and PD-L1 Modulators Primes Immunity and Reduces Tumor Growth in Tolerized Mice

The presence of tumor-infiltrating T cells (TIL) is associated with improved clinical outcomes in multiple tumor types (1, 2) and is also necessary for response to immune checkpoint blockade (3, 4). Although T-cell responses to some tumors occur spontaneously, the majority of cancers are not naturally recognized by the immune system. Preclinical and clinical data together have established a central role for type I IFNs in linking innate and adaptive immune responses to mediate tumor rejection (5–7). Non-T cell–inflamed tumors in humans are deficient in type I IFNs. Thus, developing therapeutic strategies for restoring type I IFN signaling is essential for expanding the number of patients who may be effectively treated with immunotherapy.

Innate immune sensing in the tumor microenvironment (TME) is the rate-limiting step in initiating tumor antigen–specific T-cell priming and T-cell migration into the tumor. Cytosolic tumor-derived DNA can trigger IFNβ production by dendritic cells (DC) in vivo by activating the DNA sensor, stimulator of interferon genes (STING; ref. 8). Double-stranded DNA (dsDNA) within the leukocyte cytosol is bound by cyclic GMP-AMP synthase (cGAS), an enzyme that synthesizes cyclic dinucleotides (CDN) with 2′-5′, 3′-5′ mixed linkages (ML) at the internucleotide phosphate bridge (9, 10). This structure confers high-binding affinity for mouse and human STING, triggering a conformational change and downstream signaling cascade that culminates in type I IFN production. In mice, CDN adjuvants augment antigen-specific CD8⁺ T-cell responses in a STING-dependent fashion through type I IFN-mediated innate immune priming (11). Intratumoral injection of the STING ligand R, R dithio ML c-di-AMP (ADU-S100) significantly inhibits the outgrowth of established B16 melanomas, CT26 colon tumors, 4T1 breast tumors, and Panc02 pancreatic tumors in mice (12).

These advances have defined an important role for STING signaling in promoting adaptive tumor immunity. However, data characterizing the immunologic consequences of STING signaling in the setting of tumor antigen-specific immune tolerance are limited. Several mechanisms regulate immune tolerance to the tumor antigen HER-2 in tolerant FVB-Tg (MMTVneu) 202Mul/J (neu/N) mice relative to the nontolerant parental strain FVB/N. The mammary-specific promoter MMTV drives expression of rat HER-2, transforming normal mammary epithelial cells into malignant HER-2 overexpressing breast tumors (13). Neu/N mice have well-established peripheral immune tolerance to HER-2 (14), similar to cancer patients. The immunodominant HER-2 epitope in FVB/N mice is RNEU_420-429, and neu/N mice have a distinct CD8⁺ T-cell repertoire specific for six other HER-2 epitopes (15). Low-dose cyclophosphamide mitigates suppression by regulatory T cells (Treg) in neu/N mice (16), facilitating the priming of high avidity HER-2 specific CD8⁺ T cells. This is further enhanced by modulation of the OX-40 signaling pathway (17, 18).

Here, we evaluate the impact of antigen-specific immune tolerance on the antitumor activity of ADU-S100, using nontolerant FVB/N and tolerant neu/N mice. This tumor model is more stringent than previously reported and more closely recapitulates human cancer than other transplantable tumor models. We demonstrate for the first time that STING signaling effectively activates innate immunity to support T-cell priming in neu/N mice, but that tumor-specific T-cell activation, expansion, and tumor regression do not occur unless secondary signals of T-cell activation are also induced.

FVB/N mice were purchased from The Jackson Laboratory. FVB/N-Tg(MMTVneu)202Mul/J (neu/N mice) were provided by Dr. William Muller (McGill University, Montreal, Canada), and bred to homozygosity as verified by Southern blot. Clone 100 T-cell receptor (TCR) transgenic mice were generated as previously described (19). Experiments were done with 8- to 12-week-old mice using AAALAC-compliant protocols approved by the Animal Care and Use Committee of the Johns Hopkins University School of Medicine.

The HER-2–expressing NT2.5 breast tumor cell line was derived from a spontaneous tumor explanted from a neu/N transgenic mouse (14). NT2.5 and the T2D^q cell lines were grown as previously described (14). NT2.5 cell aliquots were implanted after two passages for each experiment and not maintained in culture for greater than 28 days (10 passages). Routine analysis of HER-2 and MHCI expression was performed monthly and cell lines tested for mycoplasma using MycoProbe Mycoplasma Detection Kit (R&D Systems) yearly. All cell lines used in this study were negative for mycoplasma.

Mice were challenged by subcutaneous injection with NT2.5 tumor cells (5 × 10⁶; FVB/N mice) and NT2.5 tumor cells (5 × 10⁴; neu/N mice) in the right (site of intratumoral injections) or left mammary fat pads, or the subcuticular tissue located on the left rump. NT2.5 cell dosages have been previously established and are based on the immune status of FVB/N and neu/N mice, the capacity to adequately characterize responsiveness to immune therapies, and the establishment of reproducible growth curves. When tumor volumes reached 62.5 mm³ (approximately 5 × 5 mm tumor size), mice were treated with either 3 intratumoral injections with 50 μg of ADU-S100 (Aduro Biotech) or HBSS (Mock) sequenced over 1 week. These tumor volumes were previously established as the lower limit for effective intratumoral injection of 40 μL ADU-S100 (12). Mice were monitored for tumor growth biweekly. Tumor growth was determined by measuring tumor diameter in two perpendicular dimensions with calipers. Tumor volume was calculated by the following formula: tumor volume = ½ (L × W²). Mean tumor size for an experimental group included only those mice with measurable tumors. Mice were euthanized when tumor volumes exceeded 500 mm³ (10 × 10 mm) according to AAALAC-compliant protocols approved by the Animal Care and Use Committee of the Johns Hopkins University School of Medicine.

CD4⁺ and CD8⁺ T cells were continuously depleted using GK1.5 and 2.43 antibodies, respectively, as previously described (20), except that sterile filtered, diluted ascites was used. Natural killer (NK) cells were depleted by twice weekly intraperitoneal injections of anti-NK1.1 (clone PK136, BioXcell). Depletions were initiated 1 week prior to tumor challenge and maintained throughout the duration of the experiment. Anti–PD-L1 (Clone 10F.9G2) and anti-OX40 (clone OXO86) were purchased from BioXcell. The rat IgG2A isotype control antibody was purchased from Jackson Immunoresearch. Antibodies were diluted in 200 μL of PBS and injected intraperitoneally at 0.25 mg per mouse twice weekly beginning at the time of the first intratumoral injection of ADU-S100 or HBSS.

Tumors were weighed and single cell suspensions were generated by passing 2 mm³ pieces of tumor through a nylon cell strainer. Cells were plated on 6-well plates in RPMI-complete medium for 2 hours to allow tumor cells to adhere, and non-adherent cells were collected and filtered several times through glass wool. Cells were counted, analyzed by flow cytometry, and absolute cell numbers normalized to tumor weight.

Anti-CD62L-FITC, anti-Ly6G-FITC, anti-IL2 FITC, anti-MHCII-PE, anti-CD40R-PE, anti-B7.1-PE, anti-B7.2-PE, anti-B7H1 (PD-L1)-PE, anti-B7H4-PE, anti-CD294 (PD-1)-PE, anti-CD152(CTLA4)-PE, anti- CD134(OX40R)-PE, anti-CD223(LAG-3)-PE, anti-Thy1.2 PercPCy5.5, anti-F480 PERCPCy5.5, anti-CD8 PercPCy5.5, anti-CD8-APC, anti-CD11c APC, anti-TIM3-APC and anti-Mac1-Alexa 700, anti-CD4-APCCy7, anti-CD44 Alexa 700, anti-TNFα-Alexa 700, anti-Ly6C–Pacific Blue, anti-Ki67-PECy7, and anti-IFNγ-PECy7 were purchased from BD Pharmingen and eBioscience and used to stain leukocytes. Cells were stained with LIVE/DEAD Aqua (Molecular Probes) to gate out dead cells. Fluorescence-activated cell sorting data were collected using a BD FACSCalibur or LSRII with CellQuest or FACSDIVA software (BD Biosciences) and analyzed with FlowJo software (Tree Star, Inc.).

All peptides were synthesized at >95% purity by the Oncology Peptide Synthesis Facility. CD8⁺ T cells were isolated and analyzed by intracellular cytokine staining (ICS) as previously described (20). The frequency of antigen-specific CD8⁺ T cells is given as the percent of IFNγ⁺ cells in the irrelevant NP_118-126 (RPQASGVYM) sample subtracted from the percent of IFNγ⁺ cells in the neu-specific RNEU_420-429 (PDSLRDLSVF)(p50) sample. The following alternative HER-2 10mer peptides were utilized in this study: (i) 24F (LQLRSLTEIL), (ii) 131E (YVSRLLGICL), (iii) 134C (VTQLMPYGCL), (iv) 158B (RLPQPPICTI), (v) 184E (QSLSPHDLSP), and (vi) 202F (HPSPAFSPAF). These were defined as nondominant epitopes recognized by the neu-specific CD8⁺ T-cell repertoire of vaccinated neu/N mice (15). Total numbers of antigen-specific CD8⁺ T cells were then calculated by multiplying the percentage of peptide-specific, IFNγ-secreting CD8⁺ T cells by the absolute number of CD8⁺ T cells.

To profile the chemokines and cytokines produced by NT2.5 cells in response to stimulation with ADU-S100 in vitro, NT2.5 cells were seeded at 0.25 × 10⁶ cells/mL in a 6-well plate overnight, then treated with either 0.1, 1, 5, or 10 μg/mL of ADU-S100 the next day. Supernatant was harvested at 0, 6, or 24 hours later. To determine the cytokine and chemokine profile in the TME in vivo, NT2.5 tumors from neu/N and FVB/N mice were isolated 24 hours after intratumoral injection with either HBSS or 50 μg of ADU-S100. Tumors were weighed and then processed using a cocktail of MPER lysis buffer (Thermo Fisher Scientific) supplemented with 1X Halt protease and phosphatase inhibitors (Thermo Fisher Scientific). For tumor processing 1 mL of lysis buffer solution was used per 100 mg tumor tissue. Tumors were dissected into 2-mm pieces and passed through a 70 μm Nylon filter (Costar). The aqueous lysate was collected and centrifuged at 20,000 rpm for 15 minutes to remove cellular debris and excess lipid. Total protein was quantitated using BCA protein assay (Thermo Scientific) and lysates were normalized to a protein concentration of 1 mg/mL. IFNβ and IFNα were measured by ELISA (PBL Assay Science); IL1β, IL6, and TNFα were measured by ELISA (R&D Systems). To more comprehensively profile cytokines and chemokines in vivo, the Mouse Th1/Th2/Th17 Cytokines and Mouse Common Chemokine Multi-Analyte ELISArray Kits (Qiagen) were utilized. All ELISAs were all performed according to the manufacturer's protocols.

The statistical significance of differences between treatment groups was determined by a Student's t test (for two groups) and one-way ANOVA (for 3 or more groups). A Bonferroni post test was used to establish statistical significance after confirming significant differences in means using one-way ANOVA. Means with P < 0.05 were considered significant. Analyses were done using GraphPad Prism.

We first examined the efficacy of intratumoral ADU-S100 in nontolerant, tumor-bearing FVB/N mice, where tumors contain significant numbers of intratumoral and peritumoral lymphocytes relative to tolerant neu/N mice (Supplementary Fig. S1). Intratumoral injection of ADU-S100 induced complete regression of injected NT2.5 tumors, protected mice from contralateral tumor challenge (Fig. 1A and B), and induced regression in both uninjected contralateral and distal tumors (Fig. 1C and D) in FVB/N mice. In contrast, intratumoral ADU-S100 only delayed tumor growth in tolerant neu/N mice. Although median survival time was significantly increased [HBSS, MST = 41 days, vs. ADU-S100, MST = 47 days; P = 0.0064 based on log-rank (Mantel–Cox) test], most mice did not clear tumors (Fig. 1E and F). Thus, intratumoral delivery of ADU-S100 induced complete tumor regression in nontolerant FVB/N mice, whereas tumor growth in tolerant, tumor-bearing neu/N mice was merely delayed. Also, in FVB/N mice, intratumoral ADU-S100 both induced regression of a distant tumor mass (the abscopal effect) and established protection from a subsequent tumor challenge, consistent with the induction of tumor-specific systemic immunity.

We next examined the roles of innate and adaptive immune cell populations in mediating tumor clearance after ADU-S100 treatment in tumor-bearing FVB/N mice, by depleting CD4⁺ and/or CD8⁺ T cells, or NK cells. Depletion of either CD4⁺ or CD8⁺ T cells partially abrogated the ability of intratumoral ADU-S100 to clear tumors in FVB/N mice. Depletion of both CD4⁺ and CD8⁺ T cells led to a complete loss of efficacy, demonstrating the requirement for both T-cell populations for optimal antitumor efficacy (Fig. 2A and B). NK-cell depletion resulted in a modest but significant delay in tumor clearance (Fig. 2C and D). These findings suggest that intratumoral ADU-S100 induces tumor regression in FVB/N mice through priming both innate and adaptive immune cells, and that tumor regression is critically dependent on the activity of both CD4⁺ and CD8⁺ T cells.

In FVB/N mice, the efficacy of intratumoral ADU-S100 was critically dependent upon the presence of functional T cells; therefore, we evaluated the ability of IT ADU-S100 to prime HER-2–specific CD8⁺ T cells in both tumor-bearing nontolerant FVB/N and tolerant neu/N mice. To do this, we quantified the number of IFNγ-producing CD8⁺ T cells responding to in vitro stimulation to T2D^q APCs pulsed with HER-2 peptide epitopes. In nontolerant FVB/N mice, intratumoral ADU-S100 significantly increased the number of IFNγ-producing, CD8⁺ T cells specific for the immunodominant HER-2 epitope RNEU_420-429 beginning at day 7 and persisting over 30 days beyond tumor clearance (Fig. 3A). In contrast, in tolerant neu/N mice intratumoral ADU-S100 induced significantly fewer IFNγ-producing, RNEU_420-429-specific CD8⁺ T cells (Fig. 3B). Intratumoral ADU-S100 treatment did not induce significant numbers of IFNγ-producing CD8⁺ T cells specific for alternative, subdominant epitopes of HER-2 (15) in either FVB/N (Fig. 3C) or neu/N mice (Fig. 3D). Thus, intratumoral ADU S100 induced a durable population of IFNγ-producing, HER-2–specific CD8⁺ T cells in FVB/N, but not neu/N, mice.

Intratumoral ADU-S100 induces production of type I IFN by DCs, resulting in their activation, migration, and presentation of antigen to CD8⁺ T cells. Therefore, we first examined the sensing of ADU-S100 by interrogating type I IFN production within the TME. A day after a single dose of intratumoral ADU-S100, similar concentrations of IFNβ were present within the TME of both FVB/N and neu/N mice (Fig. 4A). However, significantly less IFNα was present within the tumors of neu/N mice relative to FVB/N mice (Fig. 4B). Next, we assessed whether the differences in tumor growth in treated FVB/N and neu/N mice were attributed to differential effects of ADU-S100 on acute tumor cell death. Immunohistochemistry 1 day after intratumoral ADU-S100 injection revealed similar proportions of necrotic and apoptotic (as measured by activated caspase-3 staining) tumor cells in both FVB/N and neu/N mice (Supplementary Fig. S2). Broadening the activity of ADU-S100 in this model, NT2.5 cells produced significant amounts of IFNβ in response to ADU-S100 exposure in vitro, indicating that NT2.5 tumor cells could directly sense ADU-S100 (Fig. 4C).

To further explore the impact of ADU-S100 on innate immunity in neu/N mice, we conducted gene expression analysis of ADU-S100–treated tumors isolated from neu/N mice. We identified 162 genes differentially expressed between untreated tumors and tumors harvested 1 day after intratumoral ADU-S100 (Supplementary Fig. S3A). Analysis of the top 50 genes with >10-fold change revealed significant upregulation of genes related to viral responses (CXCL9, CCL5, OASL2, OAS2, OAS1G, DDX60, DHX58), type I IFN (ISG15, Mx2, IFIT1, IFI44, USP18), and IFN signaling (IFR7, CCL2, CXCL11, IFI204, Slfn4). Ingenuity pathway analysis revealed that the top three canonical pathways associated with ADU-S100 treatment were “Activation of IRF by Cytosolic Pattern Recognition Receptors” (P = 2.5 × 10⁻¹⁴), “Interferon Signaling” (P = 6.9 × 10⁻¹³), and “Role of Pattern Recognition Receptors in Recognition of Bacteria and Viruses” (P = 4.2 × 10⁻⁸; Supplementary Fig. S3B; ref. 3). Overall these findings indicate that intratumoral ADU-S100 induced robust IFNβ production, thus stimulating IFN-responsive pathways.

Gene expression analysis revealed upregulation of several chemokines (CXCL9, CXCL11, and CCL5) essential for T-cell trafficking into the TME (Supplementary Fig. S3C). Therefore, we performed multianalyte ELISA analysis to assess changes in chemokine expression within the TME of untreated, HBSS, or ADU S100-treated FVB/N and neu/N mice. Relative to the untreated tumors of FVB/N mice (T-cell inflamed), untreated tumors from neu/N mice (T-cell deficient) were deficient in CXCL9, CCL3, CCL4, CCL5, but expressed more CCL2 (Supplementary Fig. S4). To assess whether intratumoral ADU-S100 increased the T-cell recruiting chemokines CXCL9 and CCL5 in neu/N mice, we analyzed chemokine expression 1 day after ADU-S100 treatment in FVB/N and neu/N mice. Intratumoral ADU-S100 induced comparable increases in both the chemokines MCP-1 (CCL2) and RANTES (CCL5), and in the cytokines IL6 and TNFα in both FVB/N and neu/N mice (Fig. 4D–F; neu/N) and Supplementary Fig. S5A and B (FVB/N). However, increases in CXCL9 and CXCL10 were only seen in neu/N mice; the high baseline expression of both CXCL9 and CXCL10 in FVB/N mice was unchanged after intratumoral ADU-S100 (Supplementary Fig. S5C). Taken together, these findings indicate that intratumoral ADU-S100 induced increased expression of T cell-recruiting chemokines in the TME of tolerant neu/N mice.

Finally, we examined whether ADU-S100 could modulate expression of the costimulatory molecules necessary for DC-mediated T-cell priming. Expression of CD80, CD86, and CD40 on DCs within the tumor-draining lymph nodes of both FVB/N and neu/N mice were evaluated 1 day after intratumoral ADU-S100 (Fig. 5A and B). DCs isolated from both FVB/N and neu/N mice displayed a similar increase in cell surface expression of CD80, CD86, and CD40 in response to intratumoral ADU-S100 (Fig. 5C). These findings suggest that the STING-signaling pathway was active in CD11c⁺ DCs present in the tumor-draining lymph nodes of both tumor-bearing FVB/N and neu/N mice.

The ability of ADU-S100 to activate innate immunity in neu/N mice was equivalent to FVB/N mice, suggesting that differences in response to ADU-S100 could indicate defective priming of HER-2–specific CD8⁺ T cells. Therefore, we examined whether intratumoral ADU-S100 could prime adoptively transferred naïve Thy 1.2⁺ RNEU_420-429-specific CD8⁺ T cells in tumor-bearing FVB/N and neu/N mice. We evaluated T-cell activation 1 day after intratumoral ADU-S100, and found that Thy1.2⁺ HER-2–specific CD8⁺ T cells isolated from FVB/N but not neu/N recipients displayed significant increases in the activation markers CD25 and CD44 (Fig. 6A). We evaluated T-cell proliferation 1 week after intratumoral ADU-S100, and found that Thy1.2⁺ HER-2 specific CD8⁺ T cells isolated from FVB/N but not neu/N recipients displayed a significant increase in the proliferation marker Ki67 (Fig. 6B). Evaluation of Thy1.2⁺ HER-2–specific CD8⁺ T-cell numbers in both the spleen and tumor-draining lymph nodes revealed more T cells in FVB/N than neu/N mice (Fig. 6C and D). This pattern was also noted in the TME, where tumors explanted from FVB/N mice contained significantly more Thy1.2⁺ HER-2–specific CD8⁺ T cells/mg of tumor tissue both before and after intratumoral ADU-S100 than tumors explanted from neu/N mice (Fig. 6E). Finally, Thy1.2⁺ HER-2–specific CD8⁺ T cells isolated from tumor-draining lymph nodes of FVB/N produced significantly more IFNγ than those isolated from neu/N mice (Fig. 6F). Together, these findings suggest that naïve RNEU_420-429-specific CD8⁺ T cells were unable to undergo sufficient activation, proliferation, and differentiation into effector CD8⁺ T cells in response to intratumoral ADU-S100 therapy in tolerant neu/N mice.

The transient delay in tumor growth observed after intratumoral ADU-S100 in neu/N mice suggested that small numbers of functional tumor antigen-specific CD8⁺ T cells were induced in response to intratumoral ADU-S100. To further elucidate potential barriers to the activation of HER-2–specific CD8⁺ T cells in neu/N mice, we examined the expression of immune checkpoint molecules on T cells from untreated, tumor-bearing neu/N mice. Tumor-infiltrating CD4⁺ and CD8⁺ T cells both highly expressed PD-1 in the TME, though CD8⁺ T cells expressed more. Expression of other immune checkpoint receptors was low on these CD8⁺ T cells except for the costimulatory receptor OX40 (OX40R), which was expressed on approximately 30% of tumor-infiltrating CD8⁺ T cells (Fig. 7A; Supplementary Fig. S6A). Evaluation of their complementary ligands revealed increased expression of PD-L1 within the TME. (Supplementary Fig. S6B). Furthermore, intratumoral ADU-S100 increased the expression of PD-1 on T cells and PD-L1 on myeloid cells (Supplementary Fig. S6C and D). These data, together with our prior studies of OX40 modulation in neu/N mice (18, 19), suggested that the PD-1 and OX40 pathways could play pivotal roles in regulating tolerance to HER-2.

We examined whether PD-L1 blockade and/or OX40R activation could enhance the ability of intratumoral ADU-S100 to prime HER-2–specific CD8⁺ T cells in tolerant neu/N mice. Combining OX40R ligation with PD-L1 blockade in the absence of ADU-S100 did not delay tumor growth or prolong median survival (Fig. 7B and C). However, adding intratumoral ADU-S100 to concurrent OX40R stimulation and PD-L1 blockade (triple therapy) significantly delayed tumor growth and rendered 40% of neu/N mice tumor free at 60 days (Fig. 7D and E). Evaluation of the TME revealed a significant increase in CD8⁺ T cells (Fig. 7F), and increased IFNγ production (Fig. 7G). Evaluation of purified CD8⁺ T cells from the spleens of treated neu/N mice revealed a significant increase in the numbers of IFNγ-secreting RNEU_420-429-specific CD8⁺ T cells (Fig. 7H). Triple therapy also induced systemic immune responses in neu/N mice, resulting in control of both injected tumors within the right cranial mammary fat pad and distal tumors located in the contralateral rump (Supplementary Fig. S7). To evaluate whether control of tumor outgrowth was solely attributed to the generation of functional tumor antigen-specific CD8⁺ T cells, we performed T-cell depletions in neu/N mice receiving triple therapy. Depletion of either CD4⁺ or CD8⁺ T cells significantly decreased the efficacy of triple therapy. Depletion of both CD4⁺ and CD8⁺ T-cell subsets rendered neu/N mice completely unresponsive to triple therapy (Supplementary Fig. S8A), demonstrating a critical dependence on T cells for therapeutic efficacy.

Lastly, we examined the induction of HER-2–specific IgG in neu/N mice treated with either ADU-S100 alone or with triple therapy. Although ADU-S100 monotherapy induced a modest, but significant, increase in HER-2–specific IgG, triple therapy induced the highest concentration of HER-2–specific IgG (Supplementary Fig. S8B). Depletion of CD4⁺ T cells abrogated the HER-2–specific antibody response, indicating a CD4⁺ T-cell dependence (Supplementary Fig. S8C). Together, these findings indicate that modulation of PD-1 and OX40 pathways can enhance HER-2–specific CD8⁺ T-cell activation, HER-2–specific antibody production, and survival in tolerant neu/N mice.

This study systematically examined pathways of STING-mediated innate immune sensing and T-cell priming in the toleragenic TME. Our findings elucidated therapeutically relevant mechanisms of innate immune sensing and antigen-specific CD8⁺ T-cell priming in tumor-bearing hosts. First, STING signaling in the TME induced T-cell–mediated immunity that cleared both injected and distant tumors in immune competent mice, but only delayed the growth of injected tumors in mice with antigen-specific immune tolerance. Second, proximal STING-signaling events (IFNβ production and DC priming) in the TME and T-cell recruitment to the tumor were intact in both nontolerant and tolerant mice treated with a STING agonist. Third, whereas STING signaling in the TME induced robust CD8⁺ T-cell priming, activation, and expansion in immune competent mice, it could not sufficiently engage tumor antigen-specific CD8⁺ T cells when immune tolerance was present. Fourth, restoration of secondary signals of T-cell activation in the setting of STING-mediated immune priming could overcome immune tolerance, effectively activating tumor-specific CD8⁺ T cells and generating HER-2–specific IgG that could mediate tumor regression and cure 40% of tumor bearing neu/N mice.

Neu/N transgenic mice overexpress the rat HER-2 protein, which promotes spontaneous breast tumor development and progression. We previously observed profound immune tolerance to HER-2 in neu/N mice (14), and have used the neu/N model to dissect mechanisms of immune regulation by which tumor antigen-specific tolerance is maintained (15–20). The priming of HER-2–specific T cells is defective in neu/N mice, as we found that spontaneous mammary tumors explanted from neu/N mice contain a paucity of TIL. At least four possibilities could explain the lack of detectable TIL in neu/N mice. One explanation is that innate immune sensing does not occur in tolerant neu/N mice, and tumor-specific T cells are not primed. Another possibility is that the migration of effector T cells into the tumor could be defective due to insufficient chemokine gradients. A third explanation is that the vast majority of tumor antigens are perceived as self by the immune system, and T cells specific for them have been deleted in the thymus. Finally, tumor-specific T cells may not be fully activated and migrate to the TME due to peripheral tolerance mechanisms.

STING signaling is critical for innate immune sensing, leading to type I IFN production in the TME and ultimately the priming of tumor antigen-specific CD8⁺ T cells by DC (6–8). Intratumoral injection of the synthetic STING agonist ADU-S100 could delay tumor growth but could not sufficiently prime CD8⁺ T cells specific for the HER-2 self-antigen in the majority of tolerant neu/N mice. Despite its low antitumor efficacy in neu/N mice, the proximal STING signaling response to ADU-S100 was essentially intact. Although concentrations of IFNα were lower in neu/N mice relative to FVB/N mice, we detected similar amounts of IFNβ within the TME of FVB/N and neu/N mice. Also, intratumoral ADU-S100 induced a significant amount of acute tumor cell death in both FVB/N and neu/N mice. These results were consistent with previous studies indicating that acute growth inhibition of tumors injected with STING agonists is largely mediated by cytokines and not T cells (21, 22). Stimulation of NT2.5 cells with ADU-S100 in vitro resulted in significant IFNβ production, indicating that NT2.5 tumor cells themselves likely contribute to IFNβ production in the TME in vivo. We also found that intratumoral ADU-S100 induced comparable expression of costimulatory molecules on CD11c⁺ DCs in the TME of FVB/N and neu/N mice, effectively activating DCs to initiate T-cell priming in each environment.

The lack of T cells in the untreated neu/N TME could be due to lack of T-cell migration into the tumor, so we profiled T-cell recruiting cytokines in the TME of immune competent FVB/N and tolerant neu/N mice. Although neu/N mice were deficient in the T-cell–recruiting chemokines CCL5, CXCL9, and CXCL10, relative to FVB/N mice, production of CCL5, CXCL9, and CXCL10 was restored with intratumoral ADU-S100 treatment in neu/N mice. The ability of ADU-S100 to induce production of CXCL9 and CXCL10 in neu/N mice was consistent with earlier reports demonstrating a role for type I IFNs in promoting production of chemokines necessary for T-cell recruitment (8, 23). Despite the ability of intratumoral ADU-S100 to establish necessary chemokine gradients, the priming HER-2–specific CD8⁺ T cells remained largely ineffective.

The rat HER-2 protein is a self-antigen in neu/N mice, and is likely subjected to both central and peripheral tolerance mechanisms similar to other models of tumor antigen enforced immune tolerance. Rat HER-2 is transiently expressed within the thymus in neonatal and lactating neu/N mice (24), and displays 97% amino acid identity to murine HER-2. Therefore, it may be that negative selection within the thymus of neu/N mice influences HER-2–specific CD8⁺ T-cell responses. However, we previously reported that low dose cyclophosphamide (CY) could mitigate the suppressive influence of regulatory T cells in neu/N mice, allowing the recruitment of RNEU_420-429-specific CD8⁺ T cells to the vaccine-induced immune response and clearing tumors in some neu/N mice (16). Evaluation of these responding HER-2–specific CD8⁺ T cells revealed co-expression of an additional TCRα chain indicating that RNEU_420-429-specific CD8⁺ T cells underwent receptor editing prior to their egress into the periphery (Todd Armstrong and Elizabeth Jaffee, personal communication); also evaluation of peripheral HER-2–specific T cells revealed that these T cells are under constant negative selection leading to their eventual removal (18). Overall, these findings suggest that peripheral tolerance plays an active and constant role in shaping the activity of the HER-2–specific CD8⁺ T cells in neu/N mice, consistent with similar findings in other genetically engineered mouse models of cancer where antigen-specific tolerance is present (25, 26).

Profiling the expression of immune checkpoint molecules on intratumoral T cells in untreated tumor-bearing neu/N mice revealed increased expression of the OX40 and PD-1 receptors on CD8⁺ T cells. These findings were consistent with our previous studies in neu/N mice, which demonstrate an essential role for OX40R in both promoting the survival and function of HER-2–specific CD8⁺ T cells (17, 18) in tolerant neu/N mice, and with an ovalbumin-based model of hepatocellular carcinoma (26). In our current study, we observed increased PD-1 expression on CD8⁺ T cells and increased PD-L1 expression on both myeloid cells and tumor cells; PD-L1 expression was further upregulated by intratumoral ADU-S100. These findings were consistent with previous studies demonstrating that type I IFNs upregulate the PD-1 pathway (27, 28). Consistent with this immune profile, we found that adding both an OX40 agonist and a PD-L1 antagonist to intratumoral ADU-S100 more effectively activated HER-2–specific T cells in the setting of intratumoral ADU-S100 than either immune checkpoint modulator alone. These data argue that high-quality CD8⁺ T cells are present in neu/N mice, and can be effectively recruited to lyse tumors if appropriate priming and costimulatory signals that can overcome peripheral tolerance mechanisms are delivered.

The use of STING agonists as immune adjuvants to promote the priming of tumor antigen-specific CD8⁺ T cells represents a potential cancer therapy. ADU-S100 has been evaluated in preclinical tumor transplantation models as an intratumoral injection (STINGIT), an adjuvant with GM-CSF–secreting vaccines (STINGVAX), and in combination with radiation therapy (12, 22, 29, 30). However, these previous studies are limited by the absence of tumor-specific immune tolerance in the models utilized. A unique strength of our model is the ability to evaluate the impact of immune tolerance on cancer immunotherapy strategies, and develop strategies for abrogating pathways of immune tolerance and suppression. Our findings have implications for future clinical studies of ADU-S100–based therapies in cancer patients, where established immune tolerance to tumor antigens presents a major challenge to the efficacy of immunotherapies. A clinical trial testing single agent intratumoral ADU-S100 in advanced cancer patients with solid tumors and lymphomas is already underway (NCT02675439). Our data support combining intratumoral ADU S100 with relevant immune checkpoint modulation to boost its clinical efficacy in future clinical studies.

E.M. Jaffee reports receiving commercial research grant from Aduro Biotech, Roche, and Bristol-Myers Squibb and is a consultant/advisory board member for BMS, Adaptive Biotech, MedImmune, and Incyte. T.W. Dubensky Jr is the chief scientific officer at Aduro Biotech. L.A. Emens is SGE at FDA, reports receiving commercial research grant from Genentech, Roche, AstraZeneca, Aduro Biotech, EMD Serono, Corvus, and Merck, is a consultant/advisory board member for Celgene, Vaccinex, Amgen, AstraZeneca, Syndax, Peregrine, Bayer, MolecuVax, and eTheRNA. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.B. Foote, T.W. Dubensky Jr, L.A. Emens

Development of methodology: J.B. Foote, L.A. Emens

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.B. Foote, M. Kok, J.M. Leatherman, T.D. Armstrong, B.C. Marcinkowski, E.M. Jaffee, L.A. Emens

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.B. Foote, M. Kok, B.C. Marcinkowski, L.A. Emens

Writing, review, and/or revision of the manuscript: J.B. Foote, M. Kok, T.D. Armstrong, B.C. Marcinkowski, E.M. Jaffee, T.W. Dubensky Jr, L.A. Emens

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.M. Leatherman, L.A. Emens

Study supervision: E.M. Jaffee

This work is in part supported by Breast Cancer Research Foundation Grant # 116656 (L.A. Emens and E.M. Jaffee), NIH T32 RR07002-37 (J.B. Foote), and NCI P30 CA006973.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.