Enhancing the Generation of Eomes^hi CD8⁺ T Cells Augments the Efficacy of OX40- and CTLA-4–Targeted Immunotherapy

Generating effective CD8⁺ T-cell responses is critical to support the efficacy of cancer immunotherapies, such as immune-checkpoint blockade (ICB; ref. 1). ICB can generate robust tumor-specific immunity in patients, leading to improved long-term survival (2). However, the therapeutic efficacy of ICB has been limited to a subset of patients, highlighting the need to understand the underlying mechanisms by which they function to inform the design of rational combinations. Blockade of the inhibitory checkpoint receptor cytotoxic T-lymphocyte–associated protein 4 (CTLA-4) effectively releases the brakes on T cells through enhanced priming and inhibition of regulatory FoxP3⁺CD4⁺ T cells (Treg; ref. 3). A CTLA-4-specific monoclonal antibody (mAb), ipilimumab, was the first checkpoint inhibitor to garner FDA approval, and it markedly improves 5-year survival in patients with metastatic melanoma (2, 4, 5). Immune-checkpoint inhibitors targeting the negative regulatory molecule programmed cell death protein 1 (PD1, also known as CD279) and its ligand PD-L1 (also known as B7-H1 and CD274) are approved for the treatment of a variety of cancer types, including melanoma, non–small cell lung cancer, Hodgkin lymphoma, renal cell carcinoma, and others (6, 7). Alternatively, agonist mAbs targeting costimulatory members of the tumor necrosis factor receptor (TNFR) family, including OX40 (also known as CD134), 4-1BB (also known as CD137), and CD27, can boost T-cell responses to augment antitumor immunity (8). Specifically, agonist OX40-specific mAb therapy enhances T-cell proliferation, survival, cytokine production, and the generation of long-lived memory T cells (9, 10).

Combination therapy with anti–CTLA-4 and anti–PD-1 elicited robust tumor regression in preclinical and clinical studies, leading to its FDA approval for treating metastatic melanoma and renal cell carcinoma (11), highlighting the potential of combination immunotherapy. Alternatively, combining ICB with costimulatory receptor stimulation enhances antitumor immunity in numerous preclinical models. For example, we found that anti-OX40/anti–CTLA-4 combination therapy significantly enhances overall treatment efficacy compared with either monotherapy through the generation of effector T cells with increased proliferation and expression of effector cytokines (granzyme B/IFNγ) and the T-box transcription factor Eomesodermin (Eomes; refs. 12, 13). Eomes and T-bet both drive the generation of effector cytokines and cytotoxic molecules, but T-bet preferentially drives the formation of short-lived effector cells and is sufficient for the effector function of T cells, whereas Eomes is required for the generation of longer-lived memory precursor effector cells through increased expression of antiapoptotic molecules and survival (14–16). Increased Eomes expression in CD8⁺ T cells is also associated with better clinical responses to immunotherapy (17, 18), suggesting that inducing Eomes may further improve outcomes.

Eomes expression in CD8⁺ T cells is regulated by multiple signaling pathways triggered by T-cell activation and differentiation, including TCR signaling, NF-κB signaling, and interleukin receptor signaling (19, 20). TCR signaling through Zap70 and Lck phosphorylates interleukin-2–inducible T-cell kinase (ITK) to induce phospho-IRF4 expression, which in turn represses Eomes (21). ITK/IRF4-mediated repression of Eomes is dictated by the strength of TCR signaling because high-affinity TCR binding leads to greater Eomes repression (22). We hypothesized that Eomes^hiCD8⁺ T cells were a critical component of anti-OX40/anti–CTLA-4 therapy and that the efficacy of anti-OX40/anti–CTLA-4 therapy could be enhanced by pharmacologic blockade of ITK, which would enhance CD8⁺ T cell–specific Eomes expression.

In the current study, we demonstrated that combined anti-OX40/anti–CTLA-4 therapy induced a unique population of Eomes^hiCD8⁺ T cells, which resembled central memory cells with high levels of proliferation and effector function, as defined by RNA and protein expression. Furthermore, ITK inhibition with the FDA-approved drug ibrutinib further enhanced Eomes expression in the presence of anti-OX40/anti–CTLA-4 therapy, resulting in significantly improved effector T-cell function, tumor regression, and survival in comparison with anti-OX40/anti–CTLA-4 alone. These data suggest that ITK inhibition enhances the efficacy of combination immunotherapy for the treatment of cancer.

Wild-type (WT) BALB/c, C57BL/6, Eomes^fl/fl, CD8^cre, and Nur77^GFP reporter mice were purchased from The Jackson Laboratory. Eomes-GFP mice (C57BL/6 background) were kindly provided by Dr. John Wherry, University of Pennsylvania, and Dr. Joseph Sun, Memorial Sloan Kettering Cancer Center. Eomes conditional knockout mice were bred in our facility by crossing Eomes^fl/fl mice with CD8^cre mice (all on the C57BL/6 background). All mice were housed under specific pathogen–free conditions in the Providence Portland Medical Center (Portland, OR) vivarium animal facility. All experimental procedures were approved by the Providence Portland Medical Center Institutional Animal Care and Use Committee and were performed under the NIH Guide for the Care and Use of Laboratory Animals.

TRAMP-C1 prostate adenocarcinoma, MCA-205 fibrosarcoma, and 4T1 mammary carcinoma cell lines were obtained between 2009 and 2011 from Dr. Andrew Weinberg (Earle A. Chiles Research Institute; TRAMP-C1, MCA-205) and Dr. Emmanuel Akporiaye (Earle A. Chiles Research Institute; 4T1). Cell line identity was verified through monthly assessment of morphology and growth kinetics. All cell lines were maintained in 10% complete RPMI and verified Mycoplasma free (MycoAlert Mycoplasma Testing Kit; Lonza) within 6 months of use.

TRAMP-C1 prostate adenocarcinoma cells (1 × 10⁶) were injected subcutaneously into the right flank of naïve male C57BL/6 mice. MCA-205 fibrosarcoma cells (1 × 10⁶) were injected subcutaneously into the right flank of naïve female C57BL/6 mice. 4T1 mammary carcinoma cells (5 × 10⁴) were injected into the mammary fat pad of naïve BALB/c mice. Tumor-bearing mice were measured for tumor growth every 2 to 3 days by tumor area using microcalipers and were sacrificed when tumor area exceeded 175 mm². Tumor-bearing animals were treated with 200 μg rat IgG (Sigma; I4131), and/or 250 μg anti-OX40 (clone OX86; Bio X Cell; BE0031), and/or 200 μg anti–CTLA-4 (clone 9D9; Bio X Cell; BE0164), and/or 200 μg anti–PD-1 (clone RMP1-14, Bio X Cell; BE0146; all i.p.). All mAbs were verified to be endotoxin free. Ibrutinib (Selleck Chemicals; S2680) treatments were administered at a dose of 6 mg/kg (i.p.) as previously described (23). For tumor growth and survival experiments, tumor-bearing mice were treated with anti-OX40 on days 8 and 12; and anti–CTLA-4 was given on days 8, 10, and 12. Control mice were treated with rat IgG on days 8 and 12. Ibrutinib was dosed on days 8, 10, 12, 15, 17, and 19 for tumor growth and survival experiments. For tissue harvest experiments, treatments were initiated when average tumor size exceeded 60 mm², following the same schedule as the survival experiments. FTY720 (Sigma; SML0700) was dosed at 1 mg/kg (i.p.) every 3 days starting 1 week prior to immunotherapy.

Tumors and lymph nodes (inguinal, axillary, and brachial) were harvested from tumor-bearing mice 1 week after the initiation of treatment. Lymph nodes were mechanically fragmented between two frosted slides and filtered through nylon mesh for flow cytometry staining. Tumors were mechanically fragmented into small pieces and digested with 5 mg/mL DNase (Sigma; 4536282001) and 1 mg/mL collagenase (Sigma; C8051) in serum-free RPMI (Lonza; 12-702Q). Digested tumor suspension was filtered through nylon mesh and stained for flow cytometry as described below.

Single-cell suspensions of blood, lymph node, and tumor-infiltrating lymphocytes (TIL) were stained with surface antibodies in flow cytometry wash buffer for 30 minutes at 4°C with surface antibodies: CD45 BV421(BioLegend; Clone 30-F11; 103134), CD4 BV605 (BioLegend; Clone RM4-5; 100548), CD8 BV785 (BioLegend; Clone 53-6.3; 100750), CXCR3 AF488 (BioLegend; Clone CXCR3-173; 126542), PD-1 PE-Cy7 (BioLegend; Clone 29F.1A12; 135216), LAG-3 PerCP-Cy5.5 (BioLegend; Clone C9B7W; 125212), TIM-3 PE (R&D Systems; Clone N/A; FAB1529P), CD11b BV785 (BioLegend; Clone M1/70; 101243), Ly6C PerCP-Cy5.5 (BioLegend; Clone HK1.4; 128012), Ly6G FITC (BD Biosciences; Clone 1A8; 551460), MHC II eF450 (eBioscience; Clone M5/114.15.2; 48-5321-82), PD-L1 BV711 (BioLegend; Clone 10F.9G2; 124319), CD11c APC (eBioscience; Clone N418; 17-0114-82), CD86 PE-Cy7 (BioLegend; Clone PO3; 105116), CD44 APC (eBioscience; Clone IM7; 17-0441-83), and CD62 L AF700 (BioLegend; Clone MEL-14; 104426). Cells were fixed and permeabilized using the Thermo Fisher FoxP3 intracellular staining buffer kit (00-5523-00) for intracellular staining as per the manufacturer's instructions. Cells were stained intracellularly for 30 minutes at 4°C in accordance with the manufacturer's protocol. For intracellular cytokine staining, cells were incubated in 96-well U-bottom plates in 10% complete RPMI and 2.5 μL/mL Golgi Block/Brefeldin A (BD Biosciences; 51-2301KZ). 10% complete RPMI consists of RPMI (Lonza; 12-702Q), 10% fetal bovine serum (Lonza;14-501F), 1% HEPES buffer from 1M stock (Lonza; 17-737E), 1% sodium pyruvate solution from 100 mmol/L stock (Lonza; 13-115E), 1% nonessential amino acids (Lonza; 13-114E), 1% pen/strep/glutamine (Invitrogen; 10378016), 0.005% 2-Mercaptoethanol (Sigma-Aldrich; M3148-100 mL). Cells were stimulated by 2 μg/mL plate-bound anti-CD3 (Bio X Cell; BE0001-1-5MG) and 5 μg/mL anti-CD28 (Bio X Cell; BE000150105MG) in suspension. Following a 4-hour stimulation at 37°C in a 5% CO₂ incubator, cells were stained at 4°C for 30 minutes with: Eomes eF660 (eBioscience; Clone Dan11mag; 50-4875-82), Granzyme A PE-Cy7 (eBioscience; Clone GzA-3G8.5; 25-5831-82), IFNγ PE (eBioscience; Clone XMG1.2; 12-7311-82), TNFα PE-Cy7 (eBioscience; Clone TN3-19.12; 25-7423-82), Ki-67 FITC (BD Biosciences; Clone B56; 556026), T-bet eF660 (eBioscience; Clone eBio4B10; 50-5825-82), pITK PE (BioLegend; Clone A16064A; 646904), IRF4 PE-Cy7 (BioLegend; Clone IRF4.3E4; 646414), CXCL9 AF647 (BioLegend; Clone MIG-2F5.5; 515606), and CCL5 PE (BioLegend; Clone 2E9/CCL5; 149104). Stained cells were analyzed using an LSR II flow cytometer running Diva (BD Biosciences) software and the data were processed using FlowJo 10 software (BD Biosciences). For lymph node and TIL, cells were counted and quantified for viability using the Guava cell counter as per the manufacturer's protocol.

PE-conjugated H-2D^b tetramer to SPAS-1 (STHVNHLHC) peptide and APC-conjugated H-2L^d tetramer to AH1-A5 (SPSYAYHQF) peptide was provided by the NIH Tetramer Core Facility. Cells were surface stained with tetramer at 1 μL per 10⁶ cells in flow cytometry wash buffer for 15 minutes at room temperature prior to surface antibody staining, as described above. Flow cytometry wash buffer consists of 1× PBS, 0.5% FBS (Peak Serum; PS-FB4), 0.4% EDTA (Alfa Aesar; 60-00-4), and 0.05% NaN₃ (Sigma-Aldrich; 26628-22-8).

MCA-205 tumor–bearing mice were treated with anti-OX40/anti–CTLA-4 therapy starting on day 12, and then tumors were harvested 7 days later. Tumors were pooled and processed for surface cell staining and sorting of TIL. Eomes-GFP⁺ and Eomes-GFP⁻ CD8⁺ T cells were sorted from the lymph nodes and cocultured with MCA-205 tumor cells in a 96-well plate at an effector:target ratio of 30:1 along with Caspase-3/7 green dye (Essen Bioscience; 4440) following the manufacturer's protocol and then incubated in an Essen Biosciences IncuCyte imager within a CO₂ incubator. Images were acquired 6 hours later and used to determine the percent cell lysis as a calculation of Cas3/7⁺ MCA-205 cells/total MCA-205 cells (×100).

TRAMP-C1 and 4T1 cells were incubated for 24 hours ± 0.3 μmol/L ibrutinib (Selleck Chemicals; S2680) and stained intracellularly for CXCL9 AF647 (BioLegend; Clone MIG-2F5.5; 515606) and CCL5 PE (BioLegend; Clone 2E9/CCL5; 149104). As a control, WT C57BL/6 splenocytes were incubated for 24 hours ± 5 μg/mL anti-CD3 (Bio X Cell; BE0001-1-5MG) and stained intracellularly for CXCL9 and CCL5.

MCA-205 tumor–bearing mice were treated with anti-OX40/anti–CTLA-4 therapy starting day 12, and tumors were harvested on day 19 post-implant and then pooled and processed for surface cell staining. Eomes-GFP⁺ and Eomes-GFP⁻ CD8⁺ T cells were sorted from the tumor using a FACSAria II flow cytometry cell sorter into RNA-preserving TRIzol buffer (Qiagen). RNA was isolated from the sorted cells using a Qiagen RNeasy mini kit (Qiagen; 74104). Three replicate samples were hybridized to the Affymetrix Mouse 430 2.0 GeneChip microarray. Expression analysis was performed using Affymetrix Transcriptome Analysis Console software (Thermo Fisher). Raw data are available in the Gene Expression Omnibus (GEO) database under accession number GSE166532. To identify cell types with similar transcriptional profiles as Eomes⁺ cells in an unbiased manner, we performed gene set enrichment analysis (GSEA) using GSEA v3.0 (Broad Institute). We compared our gene sets from Eomes⁻ or Eomes⁺ cells against all immunologic signature gene sets from the Molecular Signatures Database, the C7 collection (MSigDB v6.2; refs. 24, 25). We found significant associations with genes down in effector versus memory CD8 T cells (GSE9650_EFFECTOR_VS_MEMORY_CD8_TCELL_DN; ref. 26).

CD8⁺ T cells from the spleens of Nur77-GFP mice were isolated using a CD8 bead isolation kit (Thermo Fisher; 11417D) from a single-cell suspension of splenocytes. Isolated CD8⁺ T cells were stimulated with plate-bound anti-CD3 in 96-well U-bottom plates for 24 hours. 10% complete RPMI media were added containing no drug, ibrutinib (Selleck Chemicals; S2680), acalabrutinib (Selleck Chemicals; S8116), or BMS-509733 (Sigma; 419820) at concentrations ranging from 0.03 to 10 μmol/L. Cells were stained 24 hours post-stimulation for flow cytometry analysis, as described above.

Statistical significance was determined using an unpaired t test (for comparisons between two groups, two-tailed), one-way ANOVA (for comparisons of three or more groups), and Kaplan–Meier survival where appropriate using GraphPad Prism software (GraphPad). A P value less than 0.05 was considered significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

MCA-205 (sarcoma) or TRAMP-C1 (prostate adenocarcinoma) cells were implanted into cohorts of WT C57BL/6 mice. Eight days following tumor implant, mice were treated with rat IgG (control), agonist anti-OX40, CTLA-4 blockade, or combined anti-OX40/anti–CTLA-4 therapy (Supplementary Fig. S1A). Monotherapy did not significantly reduce tumor growth or increase survival. However, combined anti-OX40/anti–CTLA-4 therapy led to complete tumor regression in 63% of MCA-205 tumor–bearing mice (Fig. 1A and B; Supplementary S1B), confirming our previous finding (12). Combined anti-OX40/anti–CTLA-4 therapy significantly increased CD8⁺ T cell–specific Eomes expression in the lymph nodes in MCA-205 (Fig. 1C–E) and TRAMP-C1 (Fig. 1F) tumor–bearing mice as compared with monotherapy-treated controls. Eomes expression was not induced in the CD4⁺ T-cell compartment (Fig. 1D) and no changes were observed in natural killer (NK) cells (Supplementary Fig. S1C). The strong synergistic induction of Eomes^hiCD8⁺ T cells that we observed following anti-OX40/anti–CTLA-4 therapy occurred primarily in the lymph nodes; the same effect was not observed in the tumor (Supplementary Fig. S1D and S1E). Although OX40-mediated therapy has demonstrated therapeutic potential in combination with anti–PD-1 (27, 28), combined anti–PD-1/anti-OX40 therapy did not induce the same robust Eomes expression as anti-OX40/anti–CTLA-4 (Fig. 1G).

Because the levels of Eomes^hiCD8⁺ T cells correlated with tumor regression following anti-OX40/anti–CTLA-4 and have been associated with improved clinical responses in patients (17, 18), we investigated whether Eomes^hiCD8⁺ T cells were necessary for the efficacy of anti-OX40/anti–CTLA-4 therapy. We crossed CD8^cre mice with Eomes^fl/fl transgenic mice to generate CD8⁺ T cell–specific Eomes-deficient conditional knockout mice (Eomes^CKO; refs. 29, 30). Eomes expression was ablated in the CD8⁺ T-cell compartment without changes to the frequency of CD8⁺ T cells in the Eomes^CKO mice (Fig. 2A). There was no impairment of Eomes expression in NK cells in Eomes^CKO mice (Fig. 2B and C). Next, we implanted MCA-205 tumors into Eomes^CKO and WT mice and assessed tumor growth and response to immunotherapy. There was no difference in tumor growth between Eomes^CKO and WT mice treated with IgG (Fig. 2D). However, the efficacy of anti-OX40/anti–CTLA-4 therapy was significantly abrogated in Eomes^CKO mice compared with WT mice (Fig. 2E and F). Similar results were obtained in the TRAMP-C1 model (Supplementary Fig. S2A and S2B). Anti-OX40/anti–CTLA-4–treated Eomes^CKO mice were found to exhibit reduced granzyme A expression compared with WT mice (Fig. 2G). Together, these data indicate that Eomes^hi CD8⁺ T cells are required for the full therapeutic efficacy of anti-OX40/anti–CTLA-4 therapy.

The expansion of Eomes^hiCD8⁺ T cells by anti-OX40/anti–CTLA-4 was primarily observed in the lymph nodes, suggesting an important role for T-cell priming in this process. TRAMP-C1 tumor–bearing mice were treated starting at day 3 with FTY720, a small-molecule agonist of sphingosine-1-phosphate receptor (S1PR) that blocks lymph node egress, anti-OX40/anti–CTLA-4 (starting at day 12), or the combination of FTY720 and anti-OX40/anti–CTLA-4. Tumors and lymph nodes were harvested on day 19 (Fig. 3A). As expected, circulating numbers of CD8⁺ T cells were significantly reduced following FTY720 treatment (Fig. 3B). The frequencies of Eomes^hiCD8⁺ T cells were increased in the lymph nodes but reduced in the tumors after FTY720 and anti-OX40/anti–CTLA-4 treatment (Fig. 3C). Similar changes were observed in GzmA⁺CD8⁺ T cells (Fig. 3D). Together, these data suggest that anti-OX40/anti–CTLA-4 therapy drives the priming of effector Eomes^hiCD8⁺ T cells within the lymph nodes of tumor-bearing mice.

To understand the functional role of Eomes^hiCD8⁺ T cells in regulating anti-OX40/anti–CTLA-4 responses, we compared the transcriptional profile of Eomes^hi versus Eomes^lo CD8⁺ T cells following anti-OX40/anti–CTLA-4 therapy. Tumor-infiltrating Eomes-GFP⁺ and GFP⁻ CD8⁺ T cells were sorted for mRNA analysis following anti-OX40/anti–CTLA-4 therapy of Eomes-GFP mice (Fig. 4A). Hierarchical clustering revealed distinct phenotypic populations with 378 (P < 0.05) differentially expressed transcripts (Fig. 4B). We also used GSEA to compare Eomes^hiCD8⁺ T cells to known cell types. The Eomes^hiCD8⁺ T-cell profile correlated most closely with that of memory T cells, suggesting that Eomes^hiCD8⁺ T cells have a more memory-like than effector-like phenotype (Fig. 4C).

We next compared the relative expression of Eomes^hiCD8⁺ T cell–associated genes to naïve, effector, and effector memory CD8⁺ T-cell populations responding to acute lymphocytic choriomeningitis virus (LCMV) infection available in the GEO database (Fig. 4D–G; ref. 31). We found that, with the exception of PD1 expression, our Eomes^hiCD8⁺ T-cell phenotype most closely resembled effector memory cells (Fig. 4D). For example, TCF1 was increased in Eomes^hiCD8⁺ T cells to similar levels as observed in effector memory cells, suggesting self-renewal memory formation (ref. 32; Fig. 4E). We also observed increased CD62L and granzyme A expression, which was similar to the levels seen in the effector memory population (Fig. 4F and G). Analysis of CD62L and CD44 showed that Eomes was most highly expressed in the central memory (CD44^hi, CD62L^hi) and effector/effector memory (CD44^hi; CD62L^lo) CD8⁺ T-cell population in the lymph nodes and tumors following anti-OX40/anti–CTLA-4 therapy (Supplementary Fig. S3A and S3B). Thus, Eomes^hiCD8⁺ T cells induced by anti-OX40/anti–CTLA-4 therapy have a phenotypic profile similar to central memory CD8⁺ T cells with strong effector potential. In addition, Eomes^hiCD8⁺ T cells expressed significantly lower levels of critical inhibitory and exhaustion markers including PD1, LAG-3, and TIM-3 compared with Eomes^loCD8⁺ T cells (Fig. 4H). We confirmed these markers by protein expression, including verification of checkpoint inhibitors (PD1, TIM-3, LAG-3), proliferation (Ki-67), and effector cytokine production (granzyme A, polyfunctional IFNγ/TNFα population; Fig. 4I).

High granzyme A expression among Eomes^hi compared with Eomes^lo CD8⁺ T cells following anti-OX40/anti–CTLA-4 therapy raised concern over whether they might be functionally exhausted, as increased Eomes expression has been associated with the development of CD8⁺ T-cell exhaustion (33, 34). Therefore, the cytolytic activity of Eomes^hi and Eomes^lo CD8⁺ T cells isolated from anti-OX40/anti–CTLA-4–treated tumor-bearing mice was determined. Eomes^hiCD8⁺ T cells trended toward lysing more target cells (tumor) compared with Eomes^lo cells (Supplementary Fig. S4), demonstrating that Eomes^hi CD8⁺ T cells retain comparable or slightly increased cytotoxic function along with producing higher levels of granzyme A.

Considering that CD8⁺ T cell–specific deletion of Eomes abrogated the efficacy of anti-OX40/anti–CTLA-4 therapy, we next asked whether enhancing CD8⁺ T-cell Eomes expression would augment treatment efficacy. Because CD8⁺ T cells deficient in ITK or IRF4 express high levels of Eomes, even in the absence of stimulation (22), we hypothesized that pharmacologic inhibition of ITK may further increase Eomes expression following anti-OX40/anti–CTLA-4 therapy. Ibrutinib is the only FDA-approved drug with activity against ITK, as well as Bruton's tyrosine kinase (BTK; ref. 35). Because ITK plays an important role in TCR signaling and the downstream regulation of Eomes, we investigated whether ITK inhibition could impair TCR signaling in CD8⁺ T cells in vitro. CD8⁺ T cells are not completely dependent on ITK because they express a redundant kinase (RLK) that activates NFAT, AP-1, and NF-κB in the absence of ITK (22, 36). To investigate the impact of ibrutinib on TCR signaling, we utilized the Nur77-GFP mouse model, which expresses GFP in proportion to the extent of TCR stimulation, independent of inflammatory stimuli (37). Naïve Nur77-GFP CD8⁺ T cells were stimulated in the presence of ibrutinib (ITK/BTK specific; ITK/BTKi), BMS-509733 (ITK specific; ITKi), or acalabrutinib (BTK specific; BTKi). Ibrutinib did not inhibit Nur77-GFP expression at or below clinical concentrations (0.3 μmol/L; Fig. 5A and B). However, we did observe a reduction in Nur77-GFP expression at ibrutinib concentrations higher than 0.3 μmol/L (Fig. 5C). BTK inhibition facilitated by acalabrutinib had no effect on Nur77 expression; however, the ITK-specific inhibitor (BMS-509744) inhibited Nur77 expression above 0.3 μmol/L (Fig. 5D). Neither BTKi nor ITKi affected cell viability following 24 hours of coculture (0.3 μmol/L; Fig. 5E). For these reasons, we sought to investigate whether ITK inhibition by ibrutinib in combination with anti-OX40/anti–CTLA-4 therapy would further enhance Eomes expression and therapeutic efficacy in vivo.

TRAMP-C1 and 4T1 tumor–bearing mice were treated with rat IgG (control), ibrutinib, anti-OX40/anti–CTLA-4, or the combination of ibrutinib and anti-OX40/anti–CTLA-4 (referred to as triple therapy). Eomes expression was unaffected by ibrutinib monotherapy, but it was enhanced by anti-OX40/anti–CTLA-4 therapy. Notably, triple therapy significantly increased Eomes expression (percentage and mean fluorescence intensity) compared with anti-OX40/anti–CTLA-4, suggesting that there is synergy between ibrutinib and anti-OX40/anti–CTLA-4 (Fig. 6A and B). Similar CD8⁺ T-cell Eomes expression was also observed in the 4T1 model (Supplementary Fig. S5A and S5B), suggesting that this is a general phenomenon of triple therapy in different models of cancer.

The Eomes^hi population induced by triple therapy exhibited significantly increased proliferation (Ki-67⁺), reduced inhibitory receptor expression (PD1), and more potent effector function (granzyme A and IFNγ) within the lymph nodes (Fig. 6C; Supplementary Fig. S5C) and tumors (Fig. 6D; Supplementary Fig. S5D) as compared with Eomes^loCD8⁺ T cells, similar to the phenotype we observed following combination therapy (Fig. 4). Eomes^hiCD8⁺ T cells generated by triple therapy also expressed higher levels of CXCR3 in the lymph nodes compared with Eomes^loCD8⁺ T cells, supporting increased trafficking to inflamed tissues (Fig. 6C). Triple therapy did not significantly affect CD8⁺ T-cell frequency or T-bet expression in lymph nodes or TIL compared with anti-OX40/anti–CTLA-4 therapy (Supplementary Fig. S5E and S5F). Ibrutinib alone also had no effect on CXCL9 or CCL5 expression in TRAMP-C1 or 4T1 tumor cells (Supplementary Fig. S6A and S6B).

Phospho-ITK and IRF4 suppress Eomes expression downstream of TCR stimulation in CD8⁺ T cells. Because ibrutinib acts as an ITK inhibitor, we measured pITK and IRF4 expression in the tumor following anti-OX40/anti–CTLA-4 and triple therapy. Triple therapy significantly reduced expression of both pITK and IRF4 (Supplementary Fig. S7A and S7B). The reduction in Eomes repressors likely led to the observed increase in Eomes with triple therapy. Interestingly, triple therapy did not significantly affect myeloid cell frequency in the tumors (Supplementary Fig. S8).

We also investigated the impact of triple therapy on T-cell differentiation. TRAMP-C1 or 4T1 tumor–bearing mice were treated with rat IgG (control), ibrutinib, anti-OX40/anti–CTLA-4, or triple therapy and then 1 week later, lymph nodes and tumors were harvested for flow cytometry analysis. Proliferation (Ki-67) of effector (FoxP3⁻) CD4⁺ T cells and CD8⁺ T cells in the lymph nodes was enhanced by triple therapy compared with either anti-OX40/anti–CTLA-4 therapy or ibrutinib alone (Supplementary Fig. S9A). However, Treg frequencies and proliferation in the lymph nodes and TIL were unchanged (Supplementary Fig. S9A and S9B). In addition, ibrutinib monotherapy had minimal impact on CXCR3 expression, whereas triple therapy significantly enhanced the generation of CXCR3⁺ cells within the CD4⁺ and CD8⁺ T-cell compartment in the lymph nodes and the tumors (Fig. 7A and B; Supplementary Fig. S9A and S9B). Triple therapy also resulted in increased granzyme A⁺ and IFNγ⁺ CD8⁺ T cells and IFNγ⁺CD4⁺ T cells (Fig. 7A and B; Supplementary Fig. S9A and S9B), further indicating the generation of potent effector-cell responses. TRAMP-C1 tumors endogenously express the MHC class I–restricted antigen SPAS-1, which is an ortholog to the human prostate cancer antigen SH3GLB2 (38). We utilized SPAS-1–specific MHC I tetramers to determine the impact of triple therapy on these tumor antigen–specific CD8⁺ T cells. Neither ibrutinib nor anti-OX40/anti–CTLA-4 therapy increased the frequency of SPAS-1⁺CD8⁺ T cells. However, triple therapy significantly increased the frequency of these cells in the lymph nodes and tumors (Fig. 7A and B).

Given that triple therapy induced the expansion of granzyme A⁺PD-1^loCXCR3⁺Eomes^hiCD8⁺ T cells, we asked whether it also improved tumor control and survival over anti-OX40/anti–CTLA-4 combination therapy. Indeed, triple therapy led to a significant reduction in tumor growth and increased survival in the TRAMP-C1 model, highlighting the therapeutic efficacy of this approach (Fig. 7C; Supplementary Fig. S10A). We also evaluated triple therapy in the 4T1 model, which revealed an initial strong antitumor effect within the first 2 weeks after treatment, but minimal improvement in long-term survival (Supplementary Fig. S10B and S10C). Furthermore, all tumor-free mice following triple combination therapy rejected autologous tumor rechallenge, demonstrating the formation of a durable memory response (Supplementary Fig. S10D). The inability to maintain complete tumor clearance in 4T1 is likely the result of the myeloid cell infiltration into the tumor that is associated with the 4T1 model (39). These results indicate that the addition of ibrutinib to anti-OX40/anti–CTLA-4 therapy not only expands the critical Eomes^hiCD8⁺ T-cell population, but also leads to decreased tumor growth and improved survival.

Developing rational combinatorial immunotherapy treatments with synergistic effects on antitumor immunity is critical to provide clinical benefit for a greater percentage of patients. In the current study, we demonstrated that anti-OX40/anti–CTLA-4 therapy synergized to enhance the survival of tumor-bearing mice across multiple tumor models. This increased efficacy corresponded with the generation of Eomes^hiCD8⁺ T cells and was abrogated in the absence of CD8⁺ T cell–specific Eomes expression, which correlated with significantly decreased effector function (granzyme A) in Eomes^CKO as compared with WT mice, highlighting the critical role of these cells in mediating antitumor responses. Several questions were raised by these findings such as what is the expression profile and function of Eomes^hi CD8⁺ T cells and does a further increase in Eomes^hiCD8⁺ T cells improve therapeutic efficacy? The increased expression of Eomes in CD8⁺ T cell may be a unique property of the anti-OX40/anti–CTLA-4 combination, as anti-OX40/anti–PD-1 did not elicit the same response, indicating that anti–CTLA-4 has unique properties when used concurrently with anti-OX40. This may reflect PD1 blockade functioning through the rescue of exhausted T cells, whereas anti-OX40 and anti–CTLA-4–mediated therapies promote effector T-cell priming and, in some cases, Treg depletion (3, 40, 41). Thus, it seems likely that T-cell priming and TNFR stimulation may be critical for the generation of Eomes^hiCD8⁺ T cells. Future studies will investigate whether ligation of other TNFR family members in the presence of ICB similarly induces Eomes^hiCD8⁺ T cells. Our findings suggest that the benefit of anti–CTLA-4 therapy may be enhanced through anti-OX40 stimulation and the generation of Eomes^hiCD8⁺ T cells, which may be an effective alternative approach for treating PD1 refractory patients.

The observed synergy between these two therapies may be a unique property of how they signal and influence costimulatory receptor expression. CTLA-4 blockade aids in T-cell stimulation through enhanced CD28 costimulatory receptor signaling by blocking CTLA-4 from binding to the ligands of CD28 (B7-1, B7-2). The CD28–B7 ligand interaction directly regulates OX40 receptor expression through enhanced IL2 production, while also promoting OX40 receptor expression (12, 42). OX40 signaling, in turn, sensitizes tumor-reactive CD8⁺ T cells to the direct effects of CTLA-4 blockade by increasing CTLA-4 expression on tumor-reactive CD8⁺ T cells (41, 43). Our data demonstrated that when T-cell lymph node egress was inhibited by FTY720 treatment, Eomes^hiCD8⁺ T cells accumulated in the lymph node and were diminished in number in the tumor, further suggesting that Eomes⁺CD8⁺ T cells were driven by T-cell activation and priming in the lymph nodes prior to traffic to the tumor site (Fig. 3C). This finding may have implications for lymphadenectomy, common in surgery for many types of cancers, which may be removing critical sites of Eomes^hiCD8⁺ T-cell generation.

Compared with their Eomes^lo counterparts, Eomes^hiCD8⁺ T cells had lower RNA and protein expression of several key inhibitory receptors including PD1, LAG-3, and TIM-3, as well as higher expression of several key effector markers such as Ki-67, granzyme A, TNFα, and IFNγ, indicating a unique effector memory phenotype. IRF4, a key negative regulator of Eomes expression in the TCR signaling pathway (22), was also reduced in Eomes^hiCD8⁺ T cells, suggesting that Eomes expression may be regulated by TCR signaling as a result of anti-OX40/anti–CTLA-4 therapy. Additionally, we observed increased expression of TCF1 in Eomes^hiCD8⁺ T cells. TCF1 is a transcription factor that identifies CD8⁺ T cells that are stimulated by antigen. TCF1 maintains lymphoid recirculation and self-renewal potential of T cells and positively regulates the transcription of Eomes (44). TCF1 has also been linked to responsiveness to PD1 blockade immunotherapy through the development of memory T-cell responses (45). This was supported by our finding of high Eomes expression in central memory CD8⁺ T cells and the ability of these cells to kill tumor cells in vitro, which indicates that this subset is not exhausted.

TCR ligation signals downstream through ITK and inhibits Eomes through IRF4. Importantly, ITK is redundant to RLK signaling in CD8⁺ T cells. Ibrutinib covalently binds to BTK and ITK, but does not bind to RLK in CD8⁺ T cells (46). Ibrutinib's effects on treating B-cell malignancies through BTK inhibition are well documented; however, its effects mediated by ITK blockade, including enhancing T-cell function, have only recently been appreciated (47, 48). For example, ITK inhibition was shown to improve T-cell function, number, and TCR diversity in chronic lymphocytic leukemia (CLL) patients receiving long-term ibrutinib therapy (49, 50). This effect was the result of the ITK-specific effects of ibrutinib (but not the BTK-specific inhibitor acalabrutinib) to increase T-cell numbers in CLL patients, suggesting a specific role for ITK inhibition in enhancing T-cell function in cancer therapy (23).

Ibrutinib has primarily been utilized in trials focused on BTK-expressing cancers such as CLL, diffuse large B-cell lymphoma, follicular lymphoma, and mantle cell lymphoma. However, the immune-modulating activity of ibrutinib provides a strong rationale for its combination with immunotherapies. In preclinical murine models, ibrutinib plus anti–PD-L1 leads to a greater frequency of antigen-experienced CD8⁺ T cells (23). Our studies revealed that ibrutinib plus anti-OX40/anti–CTLA-4 therapy led to a synergistic increase in Eomes expression, reduced pITK and IRF4 expression, enhanced therapeutic efficacy, and markedly enhanced tumor regression.

Triple therapy also enhanced several aspects of T-cell effector function including proliferation of CD8⁺ T cells and effector (FoxP3⁻) CD4⁺ T cells, which was associated with a striking increase in CXCR3 expression. The induction of CXCR3 is typically associated with type 1 helper T cell (Th1)–polarized immune responses, which are critical to an effective anticancer response through enhanced tumor trafficking via the CXCR3 ligands CXCL9 and CXCL10 (51). Ibrutinib alone had a small, but significant, impact on CXCR3 expression in FoxP3⁻ CD4⁺ T cells and CD8⁺ T cells within the tumor, which likely stems from ITK-mediated regulation of CXCR3 expression, as Itk^−/− mice are characterized by high expression of CXCR3 and Eomes (22). The generation of a Th1-polarized population by triple therapy was further supported by increased expression of IFNγ, TNFα, and granzyme A by effector CD4⁺ T cells and CD8⁺ T cells. ITK inhibition likely plays a role in the generation of Th1 immunity beyond its role in regulating Eomes. For Th2-polarized CD4⁺ T cells, ITK is a required TEC kinase as Th2 cells do not express the redundant RLK found in Th1 CD4⁺ and CD8⁺ T cells. As a result, inhibition of ITK by ibrutinib may inhibit the formation of Th2 CD4⁺ T cells while leaving Th1-polarized CD4⁺ and CD8⁺ T-cell populations intact.

ITK signaling acts as a rheostat for T-cell function and differentiation. The quantity of phosphorylated ITK (pITK) is determined by LCK and Zap70-mediated phosphorylation through stable TCR complexes based on antigen binding affinity. The downstream effects of pITK can be divided into digital and graded responses (52). The activation of the NFAT and NF-κB pathways is a digital response, meaning that any activity of pITK triggers these pathways. AP-1 activation, however, is a graded response, meaning the amount of upstream pITK determines the extent to which AP-1 is activated. By inhibiting ITK with ibrutinib, we are likely shifting CD8⁺ T cells receiving strong TCR stimulation to expresses higher levels of Eomes. This functional shift would occur without inhibiting NF-κB and NFAT signaling. Recently, it has been shown that submaximal TCR affinity stimulation leads to a survival benefit through Eomes expression (53). This in turn allows for Eomes^hiCD8⁺ T cells to compete for a memory niche that would otherwise be occupied by high-affinity cells in the absence of Eomes. We are currently investigating the role of TCR stimulation in the Eomes response to understand how the strength of TCR signal determines and influences CD8⁺ T-cell function and response to immunotherapy.

As new cancer immunotherapies continue to emerge, opportunities for novel combinations of therapies have expanded. Although a greater percentage of patients may benefit from combination immunotherapy, particular attention needs to be given to the development of adverse events that may result from new combinations. Adverse effects may be ameliorated, in part, by finding combinations that have synergistic, rather than additive or antagonist effects. Our studies revealed unique synergy between ibrutinib and anti-OX40/anti–CTLA-4 immunotherapy, through a CD8⁺ T-cell Eomes-specific mechanism. How ibrutinib may combine and possibly synergize with the plethora of other immunotherapies currently being evaluated clinically (e.g., vaccines, oncolytic viruses, checkpoint inhibitors, T-cell agonists, cytokines, and myeloid-targeted therapies) remains unclear. However, our data demonstrate that ibrutinib/anti-OX40/anti–CTLA-4 therapy warrants further investigation and evaluation in the clinic.

D.A. Emerson reports grants from Providence Portland Medical Foundation and Susan G. Komen during the conduct of the study. W.L. Redmond reports grants from Susan G. Komen and Providence Portland Medical Foundation during the conduct of the study, as well as grants from Galectin Therapeutics, Nektar Therapeutics, Veana Therapeutics, Aeglea Biotherapeutics, Merck, Bristol-Myers Squibb, GlaxoSmithKline, Inhibrx, MiNA Therapeutics, Calibr, Shimadzu, and AstraZeneca and other from Vesselon outside the submitted work. No disclosures were reported by the other author.

D.A. Emerson: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. A.S. Rolig: Formal analysis, visualization, writing–review and editing. W.L. Redmond: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing.

The authors thank members of the Earle A. Chiles Research Institute (EACRI) Cancer Research Animal Division and Miranda Gilchrist (EACRI Flow Cytometry Core) for excellent technical assistance. This work was supported by the Providence Portland Medical Foundation and Susan G. Komen Career Catalyst Grant CCR15329664 (W.L. Redmond).

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.