Therapeutic Efficacy of 4-1BB Costimulation Is Abrogated by PD-1 Blockade in a Model of Spontaneous B-cell Lymphoma

Immunotherapies that promote anticancer T-cell responses are now at the forefront of clinical oncology research and practice. mAbs that target inhibitory receptors on T cells, such as anti-CTLA-4 and anti-PD-1, have been successfully used in the treatment of lymphomas (1, 2), particularly in advanced Hodgkin lymphoma, for which PD-1 mAbs (nivolumab/Opdivo) gained FDA approval early 2016. In poorly immunogenic cancers, such as some subtypes of non-Hodgkin B lymphomas (B-NHL), vaccine approaches and mAbs targeting costimulatory receptors are additional immunotherapeutic strategies that may promote superior generation and persistence of antitumor T cells. This creates a strong rationale for combining immune stimulatory approaches with inhibitory checkpoint blockade.

TNF receptor superfamily 9 (TNFRSF9), better known as 4-1BB (or CD137), is an inducible, costimulatory molecule expressed on T, B, and natural killer (NK) cells (3). Engagement of 4-1BB on T cells provides costimulation, which promotes activation, differentiation, proliferation, and survival (4–6). Costimulation of T cells with agonistic 4-1BB mAbs has been efficacious against a range of tumors (7), including B-NHL (8, 9). We have previously shown that the efficacy of 4-1BB mAb treatment could be enhanced when combined with an NKT-targeted tumor cell vaccine (10). These results were generated in a transplantable tumor model of B-NHL, where disease initiation is aggressive and progression to terminal disease is rapid.

The major objective of this study was to assess the efficacy of 4-1BB mAb in combination with NKT-targeted tumor cell vaccination or PD-1 mAb, in a model of spontaneous lymphoma initiation and progression. This is an important distinction, because transplanting primary lymphoma cells into immunocompetent recipient mice, although useful for rapid assessment of treatment efficacy and underlying mechanisms, does not accurately represent the evolving interactions between the tumor and immune system that occurs during the ongoing process of immune surveillance. This immune editing process may ultimately affect the ability of the immune cells to respond to immunotherapy and impact therapeutic success.

We found that 4-1BB mAb was capable of generating durable antitumor responses that prevented lymphoma outgrowth long term. Unexpectedly, this protective effect was diminished when 4-1BB mAb was combined with PD-1 blockade. Our data suggest that careful consideration is required when modulating both immune costimulatory and coinhibitory receptors simultaneously with mAb therapies for B-NHL.

Inbred C57BL/6 wild-type mice were obtained from the Animal Resources Centre (Perth, Australia). Eμ-myc transgenic mice on the C57BL/6 background express the c-myc gene under the control of the Ig intron enhancer (Eμ), modeling the chromosomal translocation found in many aggressive NHLs, including Burkitt lymphoma and diffuse large B-cell lymphoma (11). These mice were bred and maintained onsite at the Translational Research Institute Biological Research Facility (TRI-BRF; Brisbane, Australia) and housed under specific pathogen-free conditions. Mice cohorts were aged and sex matched. An equal ratio of male to female mice was used. All experiments were performed in accordance with the animal ethics guidelines provided by the National Health and Medical Research Council of Australia and approved by the University of Queensland Health Sciences Animal Ethics Committee.

α-Galactosylceramide (α-GalCer) was purchased from Avanti Polar Lipids. Fluorochrome-conjugated mAbs to mouse PD-1 (RMP1-30), KLRG1 (2F1), CD127 (A7R34), CD8b (YTS156.7.7), CD62L (MEL-14), CD44 (IM7), TCRβ (H57-597), and CD69 (H1.2F3) were purchased from BioLegend, and mAbs to CD8α (53 6.7) and CD19 (1D3 or 6D5) were purchased from BD Pharmingen and were used to label cells for flow cytometric analysis. Cells were labeled at optimal concentrations of mAb for 25 minutes at 4°C in PBS containing 2% newborn calf serum and 2 mmol/L EDTA. Flow-count fluorospheres (Beckman Coulter) were added to the samples to calculate cell numbers upon acquisition. Labeled cells were acquired on Gallios (Beckman Coulter) or LSR-II (BD Biosciences) flow cytometers and analyzed using the Kaluza version 1.2 (Beckman Coulter) software.

mAb treatment commenced in Eμ-myc transgenic mice at a median age of 7.7 ± 1.2 weeks (range, 4.4–9.3), known to be the approximate age of lymphomagenesis (11). Two cycles of treatments were administered for a total of six mAb injections over a 7-week period, with a 1-week rest period between cycles (Fig. 1A). Mice received either (i) 100 μg 4-1BB mAb (clone 3H3); (ii) 200 μg PD-1 mAb (clone RMP1-14); (iii) 100 μg 4-1BB mAb and 200 μg PD-1 mAb combined in a single injection; or (iv) 100 μg rat IgG2a (clone 2A3), all purchased from Bio X Cell. All antibodies were given via intraperitoneal route.

The NKT-cell–targeting tumor cell vaccine (10, 12) was administered 24 hours prior to first mAb treatment at the beginning of each treatment cycle (Fig. 1A). To prepare the vaccine, tumor cells derived from three independent, de novo tumors developed in Eμ-myc transgenic mice were pooled and loaded with 500 ng/mL of α-GalCer overnight in culture. The α-GalCer–loaded tumor cells were irradiated at 5,000 cGy to arrest proliferation. Mice were vaccinated with 5 × 10⁵ irradiated tumor cells intravenously.

To monitor disseminated tumor burden and peripheral immune parameters, mice were bled every 3 weeks until mice reached disease endpoint or 56 weeks of age, whichever occurred first. Blood was collected into heparinized PBS and lysed with ACK lysis buffer prior to staining for flow cytometry. Mice were sacrificed at disease endpoint, determined by ethics approved disease score sheet.

Tumor-free, transgenic Eμ-myc mice aged between 34 and 56 weeks that had previously received anti-4-1BB, or vaccine plus anti-4-1BB, >30 weeks prior, were rechallenged with 1 × 10⁵ GFP-transduced Eμ-myc 299 tumor cells. To monitor tumor growth and immune parameters, mice were bled via retro-orbital eye bleed 7, 14, and 21 days after tumor inoculation. Specific cell populations were then determined by flow cytometry.

IFNγ in mouse sera was quantified using an ELISA Kit from R&D Systems as per the manufacturer's instructions. Acquisition and analysis were performed on a Multiskan FC Photometer (Thermo Fisher Scientific).

Data are expressed as mean ± SEM. Statistics were generated using GraphPad Prism 6 (GraphPad Prism Software). Significance was accepted at P < 0.05.

Within the TRI-BRF facility approximately 80% of Eμ-myc transgenic mice spontaneously developed B-cell lymphoma–like disease. Disease manifestation was observed from 8-week-old mice with a median survival time for untreated Eμ-myc transgenic mice of 19.3 weeks (Fig. 1B), consistent with previous reports (11, 13). These mice provide an excellent tool for investigating the responsiveness of spontaneous, de novo tumors to immunotherapy. We have previously reported that an irradiated, autologous tumor cell vaccine pulsed with α-GalCer was effective at delaying growth of established transplantable Eμ-myc tumors originally derived from Eμ-myc transgenic mice (12). This could be significantly enhanced by the addition of agonistic 4-1BB mAb, leading to complete remissions (10). We therefore sort to determine the efficacy of an irradiated tumor cell vaccine alone or in combination with 4-1BB mAb therapy against spontaneously arising B-NHL.

Eμ-myc transgenic mice that received an α-GalCer–pulsed tumor cell vaccine between 5.9 and 8.9 weeks of age had no significant survival benefit compared with age-matched, sex-matched untreated Eμ-myc transgenic mice (19.3 vs. 22.2 weeks), with 20% surviving out to 45 weeks (Fig. 1B). The inability to protect against tumor growth was not due to a failure of the vaccine to stimulate innate immunity (14), as determined by an increase in sera IFNγ and upregulation of CD69 on effector cells 24 hours postvaccination (Fig. 1C). When vaccination was combined with 4-1BB mAb, 70% of Eμ-myc transgenic mice were tumor free at 45 weeks (Fig. 1B). A comparable protective benefit was observed in mice that received 4-1BB mAb alone, suggesting 4-1BB costimulation alone was sufficient to control lymphoma outgrowth (Fig. 1B). For 30 weeks posttreatment, mice were bled to monitor immune cell kinetics. The highest numbers of CD8⁺ T cells were observed between 6.7 and 7.0 weeks posttreatment for both vaccine + 4-1BB mAb and 4-1BB mAb only–treated groups. Vaccine + anti-4-1BB drove the activation of significantly more CD8⁺ T cells than did anti-4-1BB alone, as assessed by upregulated CD44⁺ expression (Fig. 1D).

The failure of the vaccine alone to provide protection against lymphomagenesis could be attributed to the transient nature of innate immune activation and absence of sufficient T-cell activation or persistence. We have previously published that vaccination delayed outgrowth of transplantable Eμ-myc tumors during the initial wave of innate immune activation and IFNγ production (12). Vaccine-induced IFNγ had no direct antitumor effect on Eμ-myc tumors (15), indicating that IFNγ supported initial antilymphoma immune activation. However, vaccine-induced immunity eventually failed in the absence of additional T-cell stimulation. As 4-1BB mAb alone protected against spontaneously arising lymphoma, this indicated that de novo tumors could provide sufficient tumor antigen for presentation to T cells, but the T cells lacked the costimulation required for sufficient effector function. Agonistic 4-1BB mAb provided the necessary costimulation for T-cell differentiation, and the expansion of these cells was boosted by vaccination.

To test the generation of immunologic memory against B-NHL, tumor-free, long-term surviving Eμ-myc transgenic mice from untreated, 4-1BB mAb, or vaccine + anti-4-1BB mAb–treated groups (aged 34–56 weeks) were challenged with transplantable Eμ-myc clone 299 (12). Eμ-myc 299 grew consistently in 12 of 12 untreated Eμ-myc transgenic mice and 7 of 7 mice that had initially responded to 4-1BB mAb. At day 14, only 3 of 7 Eμ-myc transgenic mice previously treated with vaccine + anti-4-1BB had detectable Eμ-myc 299 tumors (Fig. 2A). In addition, when tumors developed in this cohort, the burden was reduced compared with the other treatment groups (Fig. 2B). This protection was remarkable, because the tumor used for the challenge (Eμ-myc 299) was genetically different from both the tumor used to generate the vaccine, and any spontaneous tumors in the Eμ-myc transgenic mouse.

Expansion of KLRG1⁺ activated CD8⁺ T cells was only detectable in the mice that successfully resisted Eμ-myc 299 growth (Fig. 2C). We previously showed that KLRG1⁺ cells are the highest producers of IFNγ (10). Because we see an expansion of KLRG1⁺ cells only in the blood of mice protected from tumor rechallenge, this reaffirms the concept that KLRG1⁺ cells are a protective CD8⁺ T-cell subset in lymphoma induced by immunotherapy. The superior protection observed upon tumor rechallenge in mice that received the vaccine + 4-1BB mAb (Fig. 2A) suggested that enhanced T-cell expansion (Fig. 1D) translated to enhanced quality/persistence of protective CD8⁺ T cells.

As we were not achieving significant expansion/persistence of activated CD8⁺ T cells or durable memory responses with the 4-1BB mAb monotherapy approach, we hypothesized that the T cells were becoming exhausted from repeated costimulation. Chronic upregulation of PD-1 on T cells is a hallmark of dysfunctional/exhausted T cells in the tumor microenvironment (16) and is known to occur after 4-1BB costimulation (17). We have observed upregulation of PD-1 on T cells exposed to transplantable Eμ-myc tumors, which was further increased by administration of 4-1BB mAb (data not shown). Eμ-myc tumors and myeloid cells in the tumor microenvironment express high amounts of PD-L1 (data not shown), engaging a PD-1/PD-L1–inhibitory axis. Impressively, the expression of PD-1 on CD8⁺ T cells was sustained for 20 weeks in Eμ-myc transgenic mice following 4-1BB mAb (Fig. 3A). We hypothesized that the efficacy of 4-1BB mAb treatment could be enhanced by the addition of an antibody to block PD-1 signaling.

Treatment of Eμ-myc transgenic mice with a PD-1–blocking mAb alone did not improve median survival (18.3 weeks) or increase long-term survivors (12.5%) when compared with untreated or isotype Ab–treated mice (Fig. 3B). Unexpectedly, coadministering PD-1 with 4-1BB reduced the median survival and decreased the proportion of long-term survivors to 33%, compared with 65% for those treated with 4-1BB alone (Fig. 3B). The kinetics of effector CD8⁺ T-cell subsets were assessed in the blood. Although generation of both effector memory (CD44⁺CD62L⁻; T_EM) and central memory (CD44⁺CD62L⁺; T_CM) CD8⁺ T cells during treatment was comparable, the persistence of these cells was dramatically decreased in the presence of PD-1 blockade (Fig. 3C).

Others have shown that 4-1BB and PD-1 together provide a synergistic therapeutic effect in transplantable solid tumor models, via enhanced IFNγ production (17, 18). We have also observed therapeutic benefits from combining these antibodies in the transplantable Eμ-myc setting (data not shown). Therefore, our findings that PD-1 blockade abrogated the therapeutic effect of 4-1BB mAb treatment in Eμ-myc transgenic mice was surprising. However, it highlights the potential differences in immunologic responses in short-term transplantable tumor models versus spontaneously arising cancers in transgenic mice, and the importance of testing immunotherapeutics in multiple settings. In an attempt to address the impact of PD-1 mAb treatment on CD8⁺ T cells in Eμ-myc transgenic mice, we further characterized the CD8⁺ T-cell population. We have previously shown that CD44⁺ CD8⁺ T cells with the phenotype KLRG1⁺ CD127⁻ [short-lived effector cells (SLEC)] or KLRG1⁺ CD127⁺ [double-positive effector cells (DPEC)] are increased in response to vaccine + 4-1BB mAb treatment in mice challenged with transplantable Eμ-myc tumors (10) and are the putative protective subsets based on enhanced functionality. In Eμ-myc transgenic mice, we observed that 4-1BB mAb treatment transiently increased both SLEC and DPEC subsets, with the addition of vaccine skewing the population toward DPEC (Fig. 4B). There was no evidence that PD-1 mAb treatment alone induced effector T-cell populations. The addition of PD-1 blockade to 4-1BB mAb treatment led to reduced generation of both SLEC and DPEC populations (Fig. 4B and C) and severe abrogation of total CD8⁺ T-cell expansion (Fig. 4A). Vezys and colleagues investigated how 4-1BB mAb could enhance PDL1 mAb–mediated reinvigoration of exhausted CD8⁺ T cells in a model of chronic LCMV infection (19). Repeated dosing with 4-1BB mAb in combination with aPDL1 caused more CD8⁺ T cells to undergo apoptosis compared with a single dose of 4-1BB mAb. We did not observe any adverse effect on CD8⁺ T-cell numbers with repeated doses of 4-1BB mAb. Instead, elevated CD8⁺ T cells numbers were sustained for up to 12 weeks posttreatment (Fig. 3C). This suggests that overstimulation of CD8⁺ T cells is a concern when reinvigorating exhausted T cells; however, when 4-1BB mAb is being used to support priming/activation of CD8⁺ T cells, multiple doses are not deleterious.

Altogether, these data suggest that 4-1BB–mediated costimulation was an effective means of generating antilymphoma T-cell immunity during B-NHL development/outgrowth. The addition of NKT-cell–targeting vaccination enhances the expansion of effector CD8⁺ T-cell subsets and promotes persistence of these cells and/or better generation of memory cells, which can protect against tumor rechallenge. Generation of memory may be important in clinical settings of tumor relapse that may occur at various periods following treatment. These findings also show that PD-1 blockade is deleterious to the therapeutic efficacy of 4-1BB mAb when administered simultaneously, which is associated with a rapid and dramatic decrease in peripheral effector CD8⁺ T cells. Given that CD8⁺ T cells are critical for therapeutic responses to these treatments in lymphoma (10, 12), a better understanding of the mechanism that underlies this reduction in T cells is required. We speculate that CD8⁺ T cells require initial signals through PD-1 upon activation to become KLRG1⁺ effector cells. In this scenario, sequential administration of 4-1BB mAb (±vaccine), followed by PD-1 mAb would be predicted to provide superior protection. We also cannot rule out the possibility that the addition of PD-1 mAb is driving programmed cell death within the CD8⁺ T effector cells. Future studies addressing the outcomes of PD-1 signaling in CD8⁺ T cells at different stages of antitumor immunity in B-NHL are warranted. Until we have an improved understanding of the tightly regulated effects of costimulatory and coinhibitory receptor signaling in T cells, caution should be applied when combining antibody-based immunotherapeutic approaches targeting these receptors in different cancers.

No potential conflicts of interest were disclosed.

Conception and design: S.R. Mattarollo

Development of methodology: S.R. Mattarollo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.L. Doff, M.S.F. Soon

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.J. McKee, B.L. Doff, M.S.F. Soon, S.R. Mattarollo

Writing, review, and/or revision of the manuscript: S.J. McKee, M.S.F. Soon, S.R. Mattarollo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.S.F. Soon

Study supervision: S.R. Mattarollo

The authors thank Rebecca West, Ben Harvie, Kendall Hepple, and Kamil Sokolowski for mouse husbandry and technical assistance and Prof. Ricky Johnstone (Peter MacCallum Cancer Centre) for originally providing the Eμ-myc transgenic mice.

This work was supported by Project Grant (APP1044355) from the National Health and Medical Research Council (NHMRC) of Australia and a Priority-driven Young Investigator Project Grant (APP1097691) coawarded by Cancer Australia/Cure Cancer Australia. S.R. Mattarollo was supported by an NHMRC Career Development Fellowship (APP1061429).