4-1BB Agonism Averts TIL Exhaustion and Licenses PD-1 Blockade in Glioblastoma and Other Intracranial Cancers

This article is featured in Highlights of This Issue, p. 1201

Immune checkpoint blockade targeting PD-1 ± CTLA-4 is now an FDA-approved strategy for a number of solid tumors. Despite its promise, checkpoint blockade has demonstrated limited success as a monotherapy in glioblastoma (GBM; ref. 1). Efficacy against other intracranially situated tumors, such as melanoma brain metastases, has been modest, but responses remain limited to a minority of patients (2). GBM in particular may hinder checkpoint blockade strategies by proffering a low mutational burden (3); broad tumor heterogeneity (4); restricted central nervous system (CNS) drug/immune access (5); and perhaps most saliently, rampant T-cell dysfunction with little baseline effective T-cell activation to perpetuate with therapy (6–12). Resistance to immune checkpoint blockade at the T-cell level is marked by the upregulation of multiple alternative immune checkpoints (13), a state that frequently signals T-cell exhaustion. Our group recently demonstrated severe T-cell exhaustion among tumor-infiltrating lymphocytes (TIL) isolated from patients and mice with GBM and the corresponding inability to respond to PD-1 blockade (10). Improving and sustaining T-cell activation within the CNS thus remains an important goal for licensing checkpoint blockade strategies against cancers harbored within the intracranial compartment.

Costimulatory receptors of the TNF receptor superfamily (TNFRSF) such as 4-1BB (CD137, TNFRSF9) accumulate on T cells upon activation. Costimulation through 4-1BB powerfully augments T-cell activation via several downstream signaling pathways, including JNK (14), ERK (15), and PI3K and Akt (protein kinase B; ref. 16). Stimulation-induced signaling ultimately converges on the master transcription factor NF-κB (17), profoundly augmenting T-cell proliferation, cytokine production, and cytolytic effector function. In addition, 4-1BB signaling inhibits activation-induced cell death (AICD) in T cells (18) and promotes long-term survival and immunologic memory (19). These features make 4-1BB a potentially attractive target for cancer immunotherapies. Targeting 4-1BB with agonist antibodies in clinical trials, however, has yielded only modest benefit in patients with solid tumors, while also conferring risks for hepatotoxicity (20). Newer therapeutic tactics, such as with 4-1BB–stimulating aptamers, offer the ability to limit off-target toxicity and have renewed interest in this approach (21–23). Furthermore, combining 4-1BB agonism with immune checkpoint blockade has demonstrated substantial synergy, and may diminish the toxicities associated with either treatment alone (24). 4-1BB agonism, however, remains an unexplored approach within the intracranial compartment, where checkpoint blockade failures have been notable to date.

We investigated 4-1BB agonism as both a monotherapeutic and a checkpoint blockade adjuvant in GBM and other intracranial tumors. We characterized 4-1BB expression on TILs isolated from human and murine GBM, finding that 4-1BB is expressed disproportionately and at high levels on nonexhausted, activated GBM TILs. 4-1BB agonism on 4-1BB–expressing GBM TILs improves their function and averts their exhaustion. In murine models of both intracranial and subcutaneous glioma (CT2A), melanoma (B16), breast (E0771), and lung carcinomas (LLC), intracranial CT2A glioma TILs demonstrated the highest levels of 4-1BB expression. Likewise, we found that 4-1BB agonism substantially licensed PD-1 blockade in intracranial CT2A, producing 50% long-term survivors. Success for this combination strategy was dependent on CD8⁺ T cells and directly correlated with TIL 4-1BB expression across tumor histologies and location, with efficacy being most pronounced for intracranial GBM, even compared with its subcutaneous counterpart. Furthermore, in models where TIL 4-1BB levels had previously been low and treatment ineffective, proffering 4-1BB expression to T cells was sufficient to newly license the combination of 4-1BB agonism and PD-1 blockade.

According to protocols approved by the Duke Cancer Center Institutional Review Board, tumor samples were obtained from newly diagnosed patients with GBM at the time of resection. All studies were conducted in accordance with recognized ethical guidelines (U.S. Common Rule, 45 CFR 46, 21 CFR 50, 21 CFR 56, 21 CFR 312, 21 CFR 812, and 45 CFR 164.508-514). Written consent was obtained from all subjects when necessary. All blood and tumor specimens were stored at room temperature and processed within 12 hours. Samples were digested and processed as described previously (25). Tumor digests were frozen in 90% FBS/10% DMSO and stored at −80°C. Blood was obtained from healthy individuals or from patients at the time of surgery. Peripheral blood mononuclear cells (PBMC) were isolated, and samples were frozen and stored in the same conditions. All TILs and PBMCs were thawed and analyzed simultaneously.

The Institutional Animal Care and Use Committee approved all experimental procedures. Animal experiments involved the use of female C57BL/6 mice at 6 to 12 weeks of age. C57BL/6 mice were purchased from Charles River Laboratories. Mice were housed at the Duke University Medical Center Cancer Center Isolation Facility under pathogen-free conditions.

Cell lines examined in this study included CT2A malignant glioma, B16F10 melanoma, Lewis lung carcinoma (LLC), and E0771 breast medullary adenocarcinoma. All four cell lines are syngeneic on the C57BL/6 background in which they were studied. In preparation for implantation, CT2A, B16F10, and LLC cells were grown in vitro in DMEM with 2 mmol/L 1-glutamine and 4.5 mg/mL glucose (Gibco) containing 10% FBS. E0771 cells were grown in vitro in RPMI1640 (Gibco) containing 10% FBS plus 1% HEPES (Gibco). Cells were split and harvested in the logarithmic growth phase. For intracranial implantation, tumor cells in PBS were then mixed 1:1 with 3% methylcellulose and loaded into a 250 μL syringe (Hamilton). The needle was positioned 2 mm to the right of the bregma and 4 mm below the surface of the skull at the coronal suture using a stereotactic frame. Note that 1 × 10⁴ SMA-560, CT2A, E0771, and LLC cells or 500 B16F10 cells were delivered in a total volume of 5 μL per mouse. For subcutaneous implantation, 5 × 10⁵ SMA-560, CT2A, E0771, and LLC cells or 2.5 × 10⁵ B16F10 cells were delivered in a total volume of 200 μL per mouse into the subcutaneous tissues of the left flank.

Mouse-specific antibodies were purchased from BD Biosciences, eBioscience, or BioLegend (see Supplementary Table S1). Human-specific antibodies were purchased from BD Biosciences, eBioscience, or BioLegend (see Supplementary Table S2). Tissue processing, T-cell stimulation, flow cytometry, and cytokine detection were performed as described previously (25).

Mouse-specific antibodies were obtained from Bio X Cell including anti-mouse PD-1 (RMP1-14, catalog #BE0146) and anti-mouse 4-1BB (LOB12.3, catalog # BE0169). For PD-1 and 4-1BB antibody treatments, 200 μg of each antibody was diluted with PBS for a total injection volume of 200 μL. Mice received intraperitoneal injections of antibody every 3 days, starting on day 9 after tumor implantation, for a total of 3 to 4 treatments. Control treatments consisted of 200 μL PBS. CD8 and CD4 depletion was performed as described previously (26). In brief, CD8 or CD4 T cells were depleted via intraperitoneal injection of 200 μg anti-CD8 (53-6.7, Bio X Cell catalog #BE0004) or 200 μg anti-CD4 (GK1.5, Bio X Cell catalog #BE0003-1) at days 5, 6, 7, and 14 following tumor implantation for early CD8 depletion or at day 12, 13, and 14 for late-start CD8 depletion.

Stimulation of human TILs was performed as described previously (18). In brief, GBM tumor digests with previously determined 4-1BB expression patterns were thawed, rested overnight in T-cell media (RPMI supplemented with 10% FBS and 1× nonessential amino acids, sodium pyruvate, β-mercaptoethanol, and penicillin/streptomycin), and added into stimulation plates (approximately 200,000 lymphocytes per well in a 96-well plate). Plates were prepared via overnight coating at 4°C with 10 ng/mL OKT3 (Abbott) and 1,000 ng/mL 4-1BB (AF838, R&D Systems) and subsequent blocking with 2% BSA for 30 minutes at room temperature. Cells were cultured for 72 hours at 37°C, then split into fresh uncoated plates and cultured in the presence of IL2 (200 IU/mL) for 48 more hours. Following the 5-day period of stimulation, cells were harvested, stimulated with PMA/ionomycin in the presence of Golgiplug for 6 hours, and intracellular staining was performed via flow cytometry.

To overexpress 4-1BB on T cells, we cloned 4-1BB cDNA (Dharmacon, catalog # MMM1013-202799845) into the MSGV backbone as described previously (27). The following primers were used to amplify the 4-1BB fragment: Forward 5′ TTTTGGATCCCATGGGAAACAACTGTTAC, Reverse 5′ TTTTGCGGCCGCTCACAGCTCATAGCCTCCT. A restriction digest using Nco1 and Not1 (NEB) was performed on both the 4-1BB fragment and the MSGV backbone. Following a gel extraction of both the 4-1BB fragment and the MSGV backbone, the rapid DNA dephosphorylation and ligation kit (Sigma) was used to anneal the fragment into the backbone. The ligation reaction was transformed into One Shot Top10 E. coli (Thermo Fisher Scientific) according to the manufacturer's directions, cells were plated on LB-ampicillin plates, and positive colonies were sent for sequencing to confirm correct insertion. In order to transduce naïve splenocytes with 4-1BB, we transfected 4-1BB into HEK 293 T cells with Lipofectamine Transfection Reagent (Invitrogen). We used 14.1 μg of vector plasmid and 9.9 μg pCL-Eco helper plasmid (Imgenex). On the same day, splenocytes were freshly harvested from OT-1 mice and cultured in T-cell media supplemented with 50 IU/mL IL2 and 2.5 μg/mL Concanavalin A. After 48 hours, splenic T cells were transduced with retroviral supernatant. Transduction was performed on nontissue culture 24-well plates previously coated with 0.5 mL of RetroNectin (Clontech) at a concentration of 25 μg/mL in PBS. Cells were plated at a density of 1 × 10⁶/mL in viral supernatant supplemented with 50 IU/mL IL2. Cells were split every 48 hours for 5 days.

Statistical analysis was conducted in GraphPad Prism version 5.0 (GraphPad Software), primarily using two-tailed, unpaired t tests or one-way ANOVAs to compare means across groups with a designated significance level of 0.05. Analyses were adjusted for multiple comparisons using the Bonferroni adjustment as indicated. Bar graphs and dot plots display the mean ± the SEM. Kaplan–Meier curves were generated for survival analyses and the Gehan–Breslow–Wilcoxon test was used to compare curves. The statistical tests employed for each data presentation are designated in respective figure legends.

We initially examined 4-1BB expression on TILs and PBMCs isolated from patients with GBM. 4-1BB levels were found to be highest amidst GBM CD8⁺ TILs when compared with PBMCs from either patients with GBM or healthy donors (Fig. 1A, representative histogram depicted in Fig. 1B). 4-1BB⁺ TILs were also more likely to exhibit an activated, PD-1⁺ phenotype, rather than to express multiple immune checkpoints (PD-1⁺TIM-3⁺LAG-3⁺; “triple positive”; Fig. 1C). Likewise, 4-1BB surface levels on PD-1⁺ TILs were higher than levels on PD-1⁺TIM-3⁺LAG-3⁺ triple positive TILs, as determined by median fluorescent intensity (Fig. 1D). We have recently shown that the “triple positive” phenotype typifies severe exhaustion among TILs in GBM, while PD-1 “single positive” TILs maintain function (10). Accordingly, among our patient samples here, 4-1BB⁺ TILs expressing PD-1 alone trended toward more IFNγ production than PD-1⁻ TILs and were significantly more likely to produce IFNγ than CD8⁺ T cells coexpressing PD-1, TIM-3, and LAG-3 upon stimulation with PMA and ionomycin (Fig. 1E).

To determine whether 4-1BB expression on CD8⁺ TILs might permit a functional response to 4-1BB agonism, we performed an in vitro stimulation assay with a 4-1BB agonist antibody. To begin, TILs were isolated from patient GBM samples and separated into those expressing high or low levels of 4-1BB (Fig. 1F). We then stimulated these cells in vitro with a 4-1BB agonist antibody (AF838, R&D Systems) as described previously (18), and performed intracellular staining for the production of IFNγ. CD8⁺ TILs with high levels of 4-1BB expression proved uniquely those able to produce IFNγ when stimulated via 4-1BB agonism (Fig. 1G and H) in a manner that appeared to correlate with levels of 4-1BB expression (P < 0.07; Fig. 1I).

We next examined whether TIL 4-1BB expression might be recapitulated in murine GBM models, permitting further study in vivo. For these purposes, the CT2A murine glioma model was employed. Mice were implanted intracranially with CT2A glioma cells and blood, tumor, and lymphoid organs subsequently harvested. Levels of 4-1BB were compared on T cells from these tissues in tumor-bearing (TB) and control mice. CD8⁺ TILs isolated from CT2A glioma tumors consistently expressed higher levels of 4-1BB than T cells isolated from blood and lymphoid organs (Fig. 1J). As with patient GBM TILs, 4-1BB-expressing murine CD8⁺ TILs were more likely to express PD-1 alone than to coexpress the immune checkpoints PD-1, TIM-3, and LAG-3 (Fig. 1K). Surface levels of 4-1BB were also lower in TILs coexpressing PD-1, TIM-3, and LAG-3 than TILs expressing PD-1 alone (Fig. 1L). When TIL production was assessed by IFNγ production, PD-1⁺ TILs expressing 4-1BB proved the most functional, while PD-1⁺TIM-3⁺LAG-3⁺ TILs not expressing 4-1BB were least capable of producing IFNγ. The small numbers of PD-1⁺TIM-3⁺LAG-3⁺ TILs that coexpressed 4-1BB demonstrated intermediate functionality (Fig. 1M).

To ascertain whether 4-1BB agonism can restore TIL function in vivo in GBM, we employed an anti–4-1BB agonist antibody in CT2A-bearing mice and assayed TIL surface markers and cytokine producing ability by flow cytometry. Likewise, we assessed the capacity of PD-1 blockade to synergize with or perpetuate the effects of 4-1BB agonism on GBM TIL activation and function. Mice were implanted IC with CT2A and treated with control, PD-1 antibody alone, 4-1BB antibody alone, or PD-1 and 4-1BB antibodies together. Mice were sacrificed at day 17 to 20 following tumor implantation (when control animals were moribund). TILs were isolated, stained for classical and alternative immune checkpoints, and stimulated with PMA and ionomycin to assay for function. Among analyzed CD8⁺ PD-1⁺ TILs, coexpression of the immune checkpoints TIM-3 and LAG-3 decreased dramatically, but exclusively in the group receiving the combination treatment (Fig. 2A). The proportions of severely exhausted CD8⁺ TILs coexpressing PD-1, TIM-3, and LAG-3 were also lower in animals treated with combined 4-1BB agonism and PD-1 blockade compared with untreated animals or to animals treated with either antibody alone (Fig. 2B). Likewise, we observed trends toward increased numbers of CD8⁺ TILs (Fig. 2C) and greater IFNγ production by TILs (Fig. 2D) in animals treated with both anti–4-1BB and anti–PD-1.

Given the observed impact on TIL function, we sought to determine whether 4-1BB agonism might license efficacy for anti–PD-1 in the CT2A model. Much as in human GBM, PD-1 blockade has been ineffective in this model (1). Once again, mice were implanted with CT2A and randomized to the following treatment groups: (i) control (isotype); (ii) PD-1 blockade alone; (iii) 4-1BB agonist alone; or (iv) combination anti–PD-1 and anti–4-1BB. All mice received intraperitoneal injections of the respective treatment beginning at day 9 and every 3 days thereafter until day 18. Consistent with prior experience, mice treated with anti–PD-1 alone did not demonstrate a survival benefit (Fig. 3A). Although 4-1BB agonism did prolong median survival from 28 to 35 days, no long-term survivors were seen (Fig. 3A). In contrast, mice treated with the combination of anti–PD-1 and anti–4-1BB achieved 50% long-term survival, observed even at day 80 after tumor implantation (Fig. 3A).

Given our earlier finding of high 4-1BB expression among CD8⁺ TILs, we investigated whether the observed efficacy for the combination therapy was dependent on CD8⁺ T cells. Mice were implanted with CT2A and randomized to the following treatment groups: (i) control; (ii) CD8⁺ T-cell depletion alone; (iii) combination anti–PD-1 and anti–4-1BB; or (iv) CD8⁺ T-cell depletion in conjunction with combination anti–PD-1 and anti–4-1BB treatment. For groups undergoing CD8⁺ T-cell depletion, respective mice received anti-CD8 antibody at days 5, 6, 7, and 14 following tumor implantation, as described previously (26). CD8⁺ T-cell depletion regimen alone did not impact survival in GBM-bearing mice, but it completely abrogated the long-term survival that was again observed with combination anti–4-1BB and anti–PD-1 treatment (Fig. 3B). When the time course of CD8 depletion was varied, CD8⁺ T cells maintained their importance for efficacy even when depleted later at days 12, 13, and 14 following tumor implantation (Fig. 3C).

As CD8⁺ T cells do not act in isolation, we additionally assessed the role of CD4⁺ T cell help in mediating anti–4-1BB and anti–PD-1 treatment efficacy. Mice were implanted with CT2A and randomized to the following treatment groups: (i) control; (ii) CD4⁺ T-cell depletion alone; (iii) combination anti–PD-1 and anti–4-1BB; or (iv) CD4⁺ T-cell depletion in conjunction with combination anti–PD-1 and anti–4-1BB treatment. As above, groups undergoing CD4⁺ T-cell depletion received anti-CD4 antibody at days 5, 6, 7, and 14 following tumor implantation. CD4⁺ T-cell depletion did not abrogate the survival benefit obtained with combined antibody treatment, and even trended toward enhancing efficacy (perhaps due to depletion of regulatory T cells; Fig. 3D).

The efficacy observed against GBM led us to question whether the combination strategy of 4-1BB agonism and PD-1 blockade might likewise be effective against other difficult-to-treat brain-situated tumors, particularly those commonly metastatic to the brain. To this end, we employed murine intracranial models of lung cancer (LLC), melanoma (B16), and breast cancer (E0771). The respective mice were treated with either control, anti–PD-1 alone, anti–4-1BB alone, or the combination. Although a mild prolongation of median survival was observed in the intracranial LLC model, and modest long-term survival was observed in the intracranial B16 model, the combination treatment was most effective in intracranial CT2A (Fig. 4A). As the nonglioma models each exhibit shorter median survival than CT2A, even when untreated, the possibility existed that the differences in observed efficacy might be attributable instead to variations in the degree of tumor burden present at treatment outset. To control for this, experiments were repeated with treatment initiated at an earlier time point (day 3 rather than 9) in the intracranial lung, melanoma, and breast cancer models. Despite the earlier treatment, efficacy of the combined antibody treatment remained poor in each model (Fig. 4B).

Given this varying efficacy seen across models, we examined whether levels of 4-1BB expression on CD8⁺ TILs might likewise vary across tumors. We found that CD8⁺ lymphocytes infiltrating CT2A were significantly more likely to express 4-1BB than CD8⁺ lymphocytes infiltrating LLC, B16, or E0771 (Fig. 4C). Furthermore, the percentages of CD8⁺ TILs expressing 4-1BB in a given intracranial tumor appeared to correlate somewhat with the efficacy of combination treatment against that tumor (P < 0.09; Fig. 4D).

Not finding universal efficacy against intracranial tumors, we investigated whether similar variability in both efficacy and TIL 4-1BB expression might be seen in tumors in the periphery. We implanted CT2A, LLC, B16, and E0771 subcutaneously, and treated mice with either control, anti–PD-1 alone, anti–4-1BB alone, or anti–PD-1 and anti–4-1BB. Efficacy was assessed as a function of tumor volume. We found that the combination treatment was no longer more effective than PD-1 treatment alone in either SC CT2A or LLC, but it did confer additive efficacy against SC B16 and E0771 (Fig. 4E). These differences in treatment effect across the intracranial and subcutaneous compartments prompted us to compare TIL 4-1BB expression across locations for a given histology. In so doing, we found that differences in the expression of 4-1BB on CD8⁺ TILs again corresponded to efficacy: for instance, 4-1BB expression was higher in CD8⁺ TILs isolated from intracranial CT2A than in subcutaneous CT2A, but instead was higher in subcutaneous B16 than in intracranial B16 (Fig. 4F), matching the differential efficacy seen for each tumor in each location. For each tumor type then, CD8⁺ TIL 4-1BB expression was the better indicator of treatment efficacy than was tumor location, with intracranial CT2A demonstrating the both the highest TIL 4-1BB levels and the most impressive efficacy across all tumors and locations.

Given the correlation observed between TIL 4-1BB expression and treatment efficacy, we sought to determine whether “providing” 4-1BB to T cells would be sufficient to proffer efficacy in an intracranial tumor model where none had been observed before. We employed the IC B16 melanoma model, as these tumors had poorly responded to anti–PD-1 and anti–4-1BB treatment in prior experiments (Fig. 4A) and had shown correspondingly low levels of 4-1BB on CD8⁺ TILs (Fig. 4B). To provide a model antigen for targeting and for determining treatment impact on antigen-specific T-cell responses, B16 cells were transfected with chicken ovalbumin (OVA). Accordingly, antigen-specific responses were enhanced in relevant host mice by adoptively transferring in OT-1 (OVA-specific CD8) T cells, which could also be engineered to overexpress 4-1BB.

4-1BB was overexpressed on OT-1 T cells by retrovirally transducing OT-1 naïve splenocytes with a 4-1BB construct cloned into the MSGV backbone. 46.4% of OVA-tetramer⁺ CD8⁺ T cells expressed 4-1BB after transduction and 5-day expansion (Fig. 5A). We then implanted B16-OVA intracranial and randomized mice into the following groups (n = 8/group): (i) control (untreated); (ii) anti–PD-1 and anti–4-1BB combination treatment; (iii) adoptive transfer of OT-1 cells only; (iv) adoptive transfer of OT-1 cells overexpressing 4-1BB; (v) adoptive transfer of OT-1 cells followed by anti–PD-1 and anti–4-1BB combination treatment; or (vi) adoptive transfer of OT-1 cells overexpressing 4-1BB followed by combination PD-1/4-1BB antibody treatment. Mice receiving adoptive transfers (ALT) received 10⁷ either OT-1 T cells or OT-1 T cells overexpressing 4-1BB on day 7 following tumor implantation. Relevant antibody treatments were administered at days 9, 12, and 15. In the B16-OVA model, PD-1 and 4-1BB antibody treatment prolonged median survival slightly from 18 to 21 days. Neither the OT-1 ALT nor the OT-1/4-1BB ALT alone prolonged median survival. However, the OT-1 ALT followed by combinatorial anti–PD-1 and anti–4-1BB prolonged median survival from 18 to 30 days, while the OT-1/4-1BB ALT followed by combinatorial anti–PD-1 and anti–4-1BB prolonged median survival from 18 to 38 days and resulted in 40% long-term survival (Fig. 5B). We further assessed the impact of the anti–PD-1 and anti–4-1BB antibodies individually when given with the OT-1/4-1BB ALT. Neither anti–PD-1 nor anti–4-1BB individually enhanced survival when combined with OT-1/4-1BB ALT, where the combination of anti–PD-1 and anti–4-1BB with OT-1/4-1BB ALT significantly prolonged median survival and resulted in long-term survivors (Fig. 5C).

Immune checkpoint blockade has rapidly emerged as a promising therapeutic strategy against a variety of solid tumors, yet has shown limited success in GBM and the intracranial compartment (1, 2, 28). Although checkpoint blockade may be limited in the intracranial setting by several factors (3–5), severe T-cell exhaustion is a prominent saboteur amidst GBM (10, 11). Checkpoint blockade is dependent upon the availability of functional, activated T cells for its effectiveness, an amenity not afforded amongst dysfunctional, exhausted GBM TILs with poor tumor infiltration. We hypothesized that a strategy that could improve TIL numbers and activation while averting T-cell exhaustion might proffer newly licensed efficacy to checkpoint blockade in GBM.

As 4-1BB signaling in T cells can improve T-cell effector functions (16), inhibit AICD (18), and promote long-term T-cell survival and immunologic memory (19), we set out to investigate the effects of 4-1BB agonism on GBM TILs. We examined TIL 4-1BB levels, ultimately finding it present at particularly high levels on activated, nonexhausted TILs within GBM. Although the higher levels of 4-1BB observed on GBM patient TILs did not quite reach significance, certain patients clearly demonstrated high levels of 4-1BB on their tumor-infiltrating T cells. Likewise, we found that 4-1BB agonism is indeed a viable strategy for improving TIL function in both human and murine GBM, when 4-1BB is present. Among the patient GBM TILs we examined, 57% expressed high levels of 4-1BB, suggesting that 4-1BB agonism may be a viable therapeutic outlet in a significant proportion of patients with GBM.

Given its capacity to improve TIL function, we investigated 4-1BB agonism as both a monotherapeutic and as an adjuvant to checkpoint blockade in GBM and other intracranial tumors. Our study is the first to demonstrate that 4-1BB agonism can license PD-1 blockade in glioma. Previous preclinical studies show that PD-1 blockade can be efficacious in murine glioma (29, 30), yet these studies have not translated to clinical efficacy. We utilized an immunologically stringent model of murine glioma that is resistant to PD-1 blockade to better mimic human GBM, and showed that 4-1BB agonism licensed responses to PD-1 blockade in this model. Traditionally, PD-1 blockade is thought to rescue the function of exhausted T cells that have sustained a progressive, hierarchical, yet sometimes reversible loss of function. A pair of recent studies, however, has demonstrated that PD-1 blockade may instead act on the activation and effector phases of T-cell responses (31, 32). The implication is that PD-1 blockade may require T cells that are not yet exhausted (33). In light of these findings, the failure of PD-1 blockade in GBM is altogether unsurprising, and may reflect the severe T-cell exhaustion we have previously shown (10, 11).

We demonstrate here that 4-1BB is expressed on activated, nonexhausted TILs, and that agonizing the 4-1BB on these TILs appears to improve their function while simultaneously avoiding their exhaustion. Although further studies are needed to provide definitive evidence that 4-1BB agonism when combined with PD-1 blockade is capable of reversing T-cell exhaustion, we do see some evidence that “triple positive” T cells may retain some functional activity (IFNγ production) when 4-1BB is present. These findings suggest that 4-1BB agonism may play a tailored, complementary role improving baseline T-cell activation and averting exhaustion, thereby overcoming resistance to immune checkpoint blockade.

Surprisingly, the efficacy of combined 4-1BB agonism and checkpoint blockade was independent of (and unfazed by) tumor type or location (brain vs. flank), but instead dependent simply on levels of TIL 4-1BB expression. This finding has multiple implications. First, this strategy is not seemingly subject to the typical limitations conferred by intracranial tumor locale. An exceedingly unusual finding, for instance, was the greater efficacy we observed with the combination treatment when used against intracranial gliomas compared with the subcutaneous gliomas implanted within the flank. This may make strategies combining 4-1BB agonism and checkpoint blockade especially well suited to the intracranial compartment (and perhaps GBM), where few systemically delivered treatment options are effective.

A second implication of therapeutic dependence on TIL 4-1BB levels (rather than tumor type or location) is that if TILs can be either provided with or coaxed into 4-1BB expression, then 4-1BB agonism may be utilized somewhat broadly in tumors, even in those whose TILs otherwise express low levels of 4-1BB. This was indeed our finding, as providing 4-1BB–expressing T cells to mice with intracranial melanoma proffered new efficacy to the combination of 4-1BB agonism and PD-1 blockade. Although adoptive transfer of T cells overexpressing 4-1BB may improve responses, such strategies are expensive and subject to significant clinical variability. Further studies then to assess mechanistic determinants of TIL 4-1BB expression are needed. Such investigations may yield needed approaches to increase TIL 4-1BB expression and advance them as viable strategies for reducing resistance to PD-1 blockade in a variety of tumors.

No potential conflicts of interest were disclosed.

Conception and design: K.I. Woroniecka, L. Sanchez-Perez, P.E. Fecci

Development of methodology: K.I. Woroniecka, P.E. Fecci

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.I. Woroniecka, K.E. Rhodin, C. Dechant, X. Cui, P. Chongsathidkiet, J. Waibl-Polania

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.I. Woroniecka, K.E. Rhodin, C. Dechant, D. Wilkinson, L. Sanchez-Perez, P.E. Fecci

Writing, review, and/or revision of the manuscript: K.I. Woroniecka, P. Chongsathidkiet, D. Wilkinson, J. Waibl-Polania, P.E. Fecci

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.I. Woroniecka, P. Chongsathidkiet, D. Wilkinson, J. Waibl-Polania, P.E. Fecci

Study supervision: P.E. Fecci

The work was supported in part by the National Institutes of Health Duke Brain SPORE Developmental Research Program (P.E. Fecci) and the Medical Scientist Training Program at Duke University School of Medicine (K. Woroniecka).

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.