Adoptive immunotherapy utilizing chimeric antigen receptor (CAR) T cells has demonstrated high success rates in hematologic cancers, but results against solid malignancies have been limited to date, due in part to the immunosuppressive tumor microenvironment. Activation of the 4-1BB (CD137) pathway using an agonistic α-4-1BB antibody is known to provide strong costimulatory signals for augmenting and diversifying T-cell responses. We therefore hypothesized that a combination of α-4-1BB and CAR T-cell therapy would result in improved antitumor responses. Using a human-Her2 self-antigen mouse model, we report here that α-4-1BB significantly enhanced CAR T-cell efficacy directed against the Her2 antigen in two different established solid tumor settings. Treatment also increased the expression of IFNγ and the proliferation marker Ki67 in tumor-infiltrating CAR T cells when combined with α-4-1BB. Strikingly, α-4-1BB significantly reduced host immunosuppressive cells at the tumor site, including regulatory T cells and myeloid-derived suppressor cells, correlating with an increased therapeutic response. We conclude that α-4-1BB has a multifunctional role for enhancing CAR T-cell responses and that this combination therapy has high translational potential, given current phase I/II clinical trials with α-4-1BB against various types of cancer. Cancer Res; 77(6); 1296–309. ©2017 AACR.
There is strong clinical evidence that therapy involving an immunologic approach alone can result in the regression of some cancers (1, 2). Recently, it has been reported that adoptive immunotherapy using T cells engineered to express a chimeric antigen receptor (CAR) specific for a tumor-associated antigen can mediate effective responses against some hematologic malignancies. Targeting of the CD19 antigen has been highly successful, with complete responses reported in up to 90% of patients with acute lymphoblastic leukemia (ALL; refs. 3, 4). Despite these results, objective responses reported for solid malignancies have been less frequent (5, 6). This may in part be due to insufficient activation and persistence of CAR T cells and/or the immunosuppressive tumor microenvironment (7). Attempts to optimize this type of therapy for solid malignancies have led to combining CAR T cells with other immunotherapy approaches designed to overcome tumor-induced immunosuppression. For example, we have previously shown that cotreatment with α-PD-1 mAb significantly enhanced CAR T-cell function (8). Other studies have investigated different strategies and demonstrated that genetic modification of signaling and cytokine pathways can increase CAR T-cell efficacy by protecting them from tumor-induced immunosuppression and allowing for activation of innate cells, respectively (9, 10). However, there has been no study reported to date combining CAR T cells with agonist antibodies targeting activation receptors, such as 4-1BB.
Activation of the costimulatory receptor 4-1BB, which is transiently expressed on activated T cells, provides important signals for T-cell proliferation, cytokine secretion, survival, and generation of memory (11, 12). Studies have demonstrated that 4-1BB expression on tumor-infiltrating lymphocytes (TIL) can be used to identify tumor-reactive T cells (13, 14), leading to the use of 4-1BB–selected TILs for adoptive cell therapy in a clinical trial for metastatic melanoma patients (NCT02111863). With regards to activation of the 4-1BB pathway, several preclinical studies have reported increased antitumor responses following treatment with an agonistic α-4-1BB mAb, primarily via CD8+ T-cell activation (15–17). In the setting of CAR T-cell therapy, one approach investigated has been to provide 4-1BB stimulation by incorporating the 4-1BB signaling domain into the CAR. A recent study reported that the CD3ζ-4-1BB CAR ameliorates T-cell exhaustion and improves their persistence in vivo compared with the CD3ζ-CD28 CAR (18). Attempts to further improve CAR function led to the development of third-generation CARs containing CD3ζ and both the CD28 and 4-1BB signaling domains. However, comparisons in preclinical mouse models between second- and third-generation CARs have been largely inconsistent in terms of antitumor effects observed. Although some studies reported increased antitumor responses by third-generation CAR T cells (19, 20), other studies have demonstrated that these CAR T cells may only be as effective or even less effective than CD3ζ-CD28 second-generation CAR T cells (21, 22). The reason for this disparity is presently unclear, although it has been suggested that the incorporation of two costimulatory domains may lead to increased activation-induced cell death (AICD), consequently resulting in a decreased overall antitumor effect (23). In the clinic, treatment with third-generation CD3ζ-4-1BB-CD28 CAR T cells targeting the Her2 antigen led to the death of a patient from a cytokine storm thought to be due to CAR recognition of low levels of Her2 expressed on epithelial cells in the lung (24). Therefore, further preclinical characterization of these third-generation CARs is required to better define their efficacy and safety profile.
Given the current uncertainty of using third-generation CARs, we used an alternative approach of providing 4-1BB costimulation to second-generation CD3ζ-CD28 CAR T cells through the use of an exogenous agonistic α-4-1BB mAb. This strategy has the distinct advantage allowing for dose control avoiding potential AICD and toxicity. Moreover, we hypothesized that the exogenous agonist α-4-1BB mAb may provide the additional benefit of engaging the host immune system unlike utilizing 4-1BB contained within the CAR. Importantly, the addition of α-4-1BB mAb in combination with CAR T-cell therapy has high translational potential given two humanized α-4-1BB mAbs, urelumab (BMS-663513) and PF-05082566, are currently being tested in phase I/II trials against several cancers, including Merkel cell carcinoma and colorectal cancer, with early results indicating that the α-4-1BB mAb is well tolerated and disease stabilization reported in multiple patients (25).
Herein, we examined Her2-specific CAR T-cell and α-4-1BB mAb combination therapy in a self-antigen human-Her2 transgenic mouse model. We demonstrate that the agonist α-4-1BB mAb significantly inhibited the growth of established Her2+ solid tumors when combined with Her2-specific CAR T cells. Moreover, administration of α-4-1BB mAb led to enhanced CAR T-cell function and reduced frequency of host immunosuppressive cells and, importantly, did not lead to on-/off-target toxicity. Our study suggests that the use of agonist antibodies, such as α-4-1BB mAb, is a viable strategy for significantly increasing therapeutic responses by CAR T cells against solid cancers.
Materials and Methods
The C57BL/6 mouse sarcoma cell line 24JK and colon adenocarcinoma MC38 cell line were generously provided by Drs. Patrick Hwu and Jeff Schlom, respectively (NIH, Bethesda, MD). Mouse breast carcinoma cell lines e0771 and AT3 were kindly provided by Professor Robin Anderson (Peter MacCallum Cancer Centre, Victoria, Australia) and Dr. Trina Stewart (Griffith University, Brisbane, Australia), respectively. The parental cell lines 24JK, MC38, e0771, and AT3 were obtained between 2000 and 2008 and were retrovirally transduced to express the human-Her2 antigen under the control of the mouse stem cell virus LTR promoter. These Her2+ cell lines are referred to as 24JK-Her2, MC38-Her2, e0771-Her2, and AT3-Her2, respectively, and were generated from 2008 to 2016. All cell lines were maintained at 37°C 5% CO2 in RPMI media with supplements including 10% heat-inactivated FBS, 1 mmol/L sodium pyruvate, 2 mmol/L glutamine, 0.1 mmol/L nonessential amino acids, 10 mmol/L 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 100 U/mL penicillin, and 100 μg/mL streptomycin. All tumor lines were tested negative for mycoplasma as verified by Victorian Infectious Diseases References, but not tested for genetic authenticity in the past 5 years.
Antibodies and cytokines
The mouse α-4-1BB (3H3 clone, IgG2a, catalog number: BE0239), the isotype control (2A3 clone, IgG2a, catalog number: BE0089), and the α-IFNγ (H22 clone, IgG, catalog number: BE0254) antibodies were purchased from Bio X Cell. IL7 and IL2 cytokines were obtained from Peprotech and the NIH (Bethesda, MD), respectively.
C57BL/6 human-Her2 transgenic mice (6–16 weeks of age; ref. 26) used in this study were bred at the Peter MacCallum Cancer Centre (Victoria, Australia). C57BL/6-Ly5.1 congenic mice and RAG−/−mice (6–10 weeks of age) were purchased from the Walter and Eliza Hall Institute (Victoria, Australia). All animal work in this study was approved in advance by the Peter MacCallum Cancer Centre Animal Experimentation Ethics Committee.
Generation of gene-modified CAR T cells by retroviral transduction and coculture assays
The viral packaging GP+E86 cell line containing empty (control) LXSN or LXSN-anti-Her2 CAR retroviral vector was generated as described previously (27). The CAR construct used in this study was a second-generation CAR comprised of an extracellular single-chain variable fragment (scFv) specific for human-Her2, a CD8 hinge region, transmembrane and intracellular CD28 and CD3ζ domains, and a c-Myc tag domain allowing for CAR T-cell tracking (28). Primary mouse T cells that were genetically modified were collected from splenocytes of C57BL/6 human-Her2 transgenic or Ly5.1 mice. T cells were retrovirally transduced as described previously (29). After transduction, T cells were maintained in supplemented RPMI media with IL2 (100 U/mL) and IL7 (2 ng/mL). For in vitro coculture assays to determine antigen-specific cytokine secretion by the transduced T cells, ELISA was performed to determine the amount of IFNγ secreted following 16-hour coculture of the transduced T cells with tumor cells as described previously (28).
Analysis of immune subsets by flow cytometry
For in vivo immune cell analysis, tumors were harvested and processed postmortem for analysis using flow cytometry as described previously (30). For intracellular staining, cells were fixed and permeabilized using Cytofix/Cytoperm Solution Kit according to the manufacturer's instructions (eBioscience). Analysis was carried out on FACS LSRFortessa (Becton Dickinson) and FCS Express software (De Novo Software).
Adoptive transfer experiments
C57BL/6 human-Her2 transgenic mice or RAG−/− mice were injected with e0771-Her2 mammary carcinoma cells (1 × 105 cells/mouse) orthotopically into the fourth mammary fat pad or subcutaneously for colon adenocarcinoma cells MC38-Her2 (5 × 105 cells/mouse). The cells were resuspended in PBS, in a volume of 20 μL (e0771-Her2) or 200 μL (MC38-Her2). Seven days following e0771-Her2 tumor injection, or 10 days following MC38-Her2 tumor injection, mice received total body radiation (5 Gy) prior to receiving treatment. This included adoptive T-cell transfer of anti-Her2 CAR T cells or control (LXSN) T cells injected intravenously (1 × 107 cells/mouse) following total body radiation, a total of five doses of intraperitoneal IL2 injections (50,000 IU/injection) on the same day of T-cell injection and twice daily for the next 2 days, intraperitoneal injections of α-4-1BB (100 μg/injection) or isotype control 2A3 antibody (100 μg/injection) on days 0, 4, and 8 after T-cell injection. Control mice were left untreated. Tumor area was measured by multiplying the maximum perpendicular diameters, and the absence of palpable tumor on day 50 posttreatment was referred to as complete regression. Survival of mice was monitored with the endpoint being when tumor area reached >150 mm2. For FACS experiments, recipient mice received transduced T cells from C57BL/6-Ly5.1 congenic donor mice, allowing tracking of the adoptively transferred T cells in vivo.
Brain and mammary tissues were harvested from untreated or treated human-Her2 transgenic mice receiving anti-Her2 CAR T cells in the presence or absence of α-4-1BB on day 9 posttreatment. Tissues were fixed with 10% neutral-buffered formalin (NBF) prior to being embedded in paraffin and sectioned. Hematoxylin and eosin (H&E) staining was performed on the sections, and evaluation of any potential signs of autoimmunity caused by the therapy was conducted. Images were visualized using Olympus BX61 microscope and obtained using RT SE Diagnostics Instruments SPOT camera jointly with SPOT Advanced Version 4.6.
All statistical analyses in this study were performed using GraphPad Prism (Version 6). Data were analyzed by unpaired Student t test when comparing two sets of data, or one-way/two-way ANOVA for multiple comparisons. Data are presented as mean ± SEM.
High level expression of anti-Her2 CAR on transduced primary mouse T cells
Splenic T cells from human-Her2 transgenic mice were collected and activated prior to being transduced with either the anti-Her2 CAR (Fig. 1A) or control empty LXSN retroviral vectors. We demonstrated that the CAR-transduced but not control T cells expressed the anti-Her2 CAR (Fig. 1A). We observed reproducibly high levels of anti-Her2 CAR on the CAR-transduced T cells (44.1 ± 3.1% SEM, n = 17) compared with control LXSN-transduced T cells (Fig. 1A; 3.8 ± 0.8% SEM, n = 17, P < 0.0001). Similar to our previous work (8), transduced T cells were predominantly CD8+ T cells with only a small proportion of CD4+ T cells (Fig. 1B). Both CD8+ and CD4+ T cells displayed high anti-Her2 CAR expression in the CAR-transduced T-cell population (Fig. 1B; 44.2 ± 3.3% SEM and 33.9 ± 2.9% SEM, respectively, n = 17). Further phenotypic characterization revealed high CD44 T-cell activation marker expression in both transduced T-cell populations (Supplementary Fig. S1A), displaying predominantly effector memory CD44high/CD62Llow and central memory CD44high/CD62Lhigh phenotypes (Supplementary Fig. S1B).
Increased 4-1BB expression on CAR T cells following CAR antigen-specific stimulation
Activation of T cells via their T-cell receptor (TCR) and MHC/peptide complex results in the upregulation of cell surface 4-1BB expression, a costimulatory receptor important in the diversification and magnification of T-cell responses (31). To assess whether T-cell stimulation through the CAR similarly leads to upregulation of 4-1BB expression, anti-Her2 CAR or control (LXSN) T cells were cocultured with Her2− parental or Her2+ tumor cells for 16 hours prior to analysis using flow cytometry. We found an increased proportion of both CD8+ and CD4+ anti-Her2 CAR T cells expressing 4-1BB following stimulation with Her2+ e0771, AT3, MC38, and 24JK tumor cells compared with the respective parental control tumor lines (Fig. 2A). We also observed significantly increased mean fluorescence intensity of 4-1BB surface expression on CAR T cells following Her2-specific stimulation, confirming this observation. We found that the level of 4-1BB expressed on CAR T cells started to decline after this time point, although expression levels still remained significantly higher than that observed on nonstimulated T cells. Upregulation of 4-1BB was greatest on CARhigh cells (gated on c-Myc Taghigh cells) in both CD8+ and CD4+ populations, indicating that 4-1BB expression was likely a direct consequence of CAR activation (Fig. 2B). The upregulation of 4-1BB provided a strong rationale to investigate the effect of α-4-1BB mAb agonist stimulation of CAR T cells in vitro and in vivo.
Anti-4-1BB enhances antigen-specific IFNγ secretion by CAR T cells in vitro
The level of IFNγ secretion by CAR T cells cocultured with parental or Her2+ tumor cells in the presence or absence of α-4-1BB mAb was assessed by ELISA. Following 16-hour stimulation, we observed that activation of 4-1BB by the agonist α-4-1BB mAb significantly enhanced the secretion of IFNγ by CAR T cells following Her2-specific stimulation with 24JK-Her2 and e0771-Her2 target cells (1.1 ± 0.15 μg/mL SEM and 1.7 ± 0.2 μg/mL SEM, respectively) compared with isotype control antibody (0.2 ± 0.07 μg/mL SEM and 0.2 ± 0.09 μg/mL SEM, P < 0.05 and P < 0.05, respectively, representative of five experiments; Fig. 3). We also examined the secretion of other Th1 and Th2 cytokines, including TNFα, IL2, IL10, and IL4, in a cytometric bead array assay, and although Her2-specific stimulation of CAR T cells resulted in the secretion of these cytokines, no significant modulation by α-4-1BB mAb was observed. We next assessed the effects of α-4-1BB mAb on anti-Her2 CAR T-cell cytotoxicity against Her2+ tumor cells in a 51Crrelease assay, and we also tested whether α-4-1BB mAb could modulate CAR T-cell survival in vitro by Annexin V/PI staining as measured by flow cytometry. Interestingly, we found that α-4-1BB mAb did not modulate the cytotoxic response by CAR T cells in both short-term (4 hours; Supplementary Fig. S2A) or long-term (16 hours) assays (Supplementary Fig. S2B) and that CAR T-cell survival was not modulated in the presence of α-4-1BB mAb. These results indicated that stimulation with α-4-1BB mAb increased CAR T-cell function predominantly through enhancing the secretion of the important proinflammatory cytokine IFNγ, but not through direct modulation of T-cell cytotoxicity.
Adoptive transfer of CAR T cells in combination with α-4-1BB mAb enhances growth inhibition of established tumors
Given that α-4-1BB mAb stimulation could enhance IFNγ secretion by CAR T cells (Fig. 3), we next investigated the possibility that α-4-1BB mAb may enhance the antitumor activity of adoptively transferred CAR T cells in vivo. In this experiment, we compared the growth of established subcutaneous MC38-Her2 tumors in human-Her2 transgenic mice following treatment with anti-Her2 CAR T cells or control T cells, with or without α-4-1BB mAb. The treatment group consisting of anti-Her2 CAR T cells and α-4-1BB mAb combined therapy led to significantly enhanced tumor regression (7/9 complete regression) compared with CAR T-cell therapy alone (0/10 complete regression) or α-4-1BB mAb with control T-cell therapy (2/9 complete regression; Fig. 4A), resulting in long-term tumor-free survival of mice (Fig. 4B). Notably, treatment with control T cells and α-4-1BB mAb did not lead to tumor regression to the same extent as CAR T-cell and α-4-1BB mAb combination therapy, highlighting the importance of Her2 antigen recognition by CAR T cells (Fig. 4A). Systemic administration of the α-4-1BB mAb was necessary for the increased antitumor effects by CAR T cells given that incubation of CAR T cells with the α-4-1BB mAb prior to transfer did not increase their therapeutic efficacy. To demonstrate the broad utility of this combination strategy, we examined the antitumor efficacy in human-Her2 transgenic mice bearing established e0771-Her2 tumor in the mammary fat pad. The level of Her2 expressed on this mouse line was physiologically relevant given it was equivalent to the high Her2-expressing human breast cancer cell line, SKBR3 (Supplementary Fig. S3) and was maintained following treatment with CAR T cells and α-4-1BB mAb. Similarly to that observed in the MC38-Her2 model, adoptive transfer of CAR T cells and α-4-1BB mAb significantly inhibited the growth of e0771-Her2 tumors to a greater extent than CAR T-cell therapy alone or α-4-1BB mAb with control T-cell therapy (Fig. 4C). This was consistent with the observation of significantly reduced tumor weight following the CAR T-cell and α-4-1BB mAb combination therapy compared with other treatment groups when tumors were harvested 9 days posttreatment (Supplementary Fig. S4).
To investigate whether the combined effect of α-4-1BB mAb treatment and CAR T-cell therapy was due to direct stimulation of 4-1BB on CAR T cells, we next investigated this combination therapy in RAG−/− mice. As RAG−/− mice lack mature T cells, this model allowed us to exclude the possible effects mediated by α-4-1BB mAb on endogenous T cells. We found that anti-4-1BB enhanced the therapeutic effect of CAR T cells in the tumor-bearing RAG−/− mice as determined by significantly greater inhibition of tumor growth by the combination therapy relative to CAR T-cell treatment alone, suggesting a direct stimulation of 4-1BB on the CAR T cells was important (Fig. 4D). Taken together, our data show for the first time that the combination therapy of CAR T cells and agonist antibody α-4-1BB mAb could significantly enhance antitumor effects against established solid cancers in an immune competent self-antigen setting.
Combined CAR T-cell and anti-4-1BB therapy does not cause autoimmunity
We have reported previously that the Her2 transgenic mice used in our experiments express human-Her2 antigen in the mammary tissue and cerebellum of the brain as a self-antigen (29). In our previous study, the level of Her2 expressed in these tissues was found to be considerably reduced compared with Her2+ tumor lines. Therefore, this transgenic model is relevant to the clinical setting given that the target antigen on tumor cells is frequently overexpressed compared with normal tissue. Interestingly, CAR T-cell treatment alone in these mice did not cause autoimmunity to Her2-expressing normal tissue (29). However, it remained possible that the significant increase in antitumor effects following the combined CAR T-cell and α-4-1BB mAb treatment may have induced some pathology to Her2+ normal tissue. In a clinical trial, adoptive transfer of anti-Her2 CD3ζ-4-1BB-CD28 third-generation CAR T cells into a patient with metastatic colon cancer resulted in a fatal pulmonary toxicity thought to be caused by CAR T-cell recognition of Her2 antigen expressed on lung epithelial cells at low levels (24). Given the potential for autoimmune effects against Her2+ normal tissue and to ensure safety of the combination treatment involving 4-1BB stimulation, we examined H&E mammary and brain sections from Her2 transgenic mice following combination treatment. All sections were thoroughly examined by a blinded researcher for any pathology mediated by increased T-cell infiltration. We observed no tissue damage in either mammary or brain tissue from mice treated with CAR T cells with or without α-4-1BB mAb stimulation, as shown by representative sections taken from mice 9 days posttreatment (Fig. 5). The morphology of these sections was comparable with sections taken from untreated mice (Fig. 5). These data importantly demonstrate that the antitumor therapeutic effects of CAR T cells can be significantly enhanced in the absence of pathology against antigen-positive normal tissue.
Anti-4-1BB increases the function of tumor-infiltrating CAR T cells
Having demonstrated that α-4-1BB mAb could enhance the efficacy of CAR T cells, we sought to investigate the mechanism underlying this increased therapeutic effect. To assess whether the enhanced efficacy of the combined therapy correlated with enhanced localization and function of CAR T cells at the tumor site, we performed a similar in vivo experiment as previously described, in which mice bearing e0771-Her2 tumors received anti-Her2 CAR or control T cells derived from donor Ly5.1+ C57BL/6 mice, with administration of either α-4-1BB mAb or isotype control antibody. We then harvested the tumors for FACS analysis 9 days posttreatment using the congenic Ly5.1 marker to specifically gate on the transferred donor T cells. Consistent with our in vitro data, α-4-1BB mAb significantly increased the production of IFNγ by adoptively transferred Ly5.1+ CAR T cells, compared with isotype control antibody (Fig. 6A). This effect was also observed when CAR T cells were transferred into RAG−/− mice, suggesting a direct activation of 4-1BB on CAR T cells (Fig. 6B). Similarly, we observed enhanced expression of the Ki67 proliferation marker by CAR T cells following α-4-1BB mAb therapy compared with treatment with isotype control antibody (Fig. 6C). This increased IFNγ production and Ki67 expression was observed both in CD8+ and CD4+ CAR T-cell populations. However, no difference was observed in CAR T-cell frequency at the tumor site between mice receiving α-4-1BB mAb and those receiving isotype control antibody (Supplementary Fig. S5A). Given the increased IFNγ production observed both in vitro and in vivo following α-4-1BB mAb treatment, we further assessed the importance of IFNγ in the combined CAR T-cell and α-4-1BB mAb therapy by administering α-IFNγ–blocking mAb in subsequent in vivo experiments. We found that blocking IFNγ cytokine abrogated the enhanced therapeutic effects of the combination therapy, confirming the important role of IFNγ in this combination approach (Fig. 6D).
Anti-4-1BB modulates host immune cell populations
As α-4-1BB mAb can also stimulate other immune cell populations that express 4-1BB, we next investigated whether α-4-1BB mAb modulated the frequency of tumor-infiltrating host endogenous immune cells. Strikingly, we observed a significant decrease in the frequency of host (CD45.2+) CD4+ Foxp3+ regulatory T cells (Treg) infiltrating the tumor on day 9 posttreatment in the group of mice that received the combined CAR T-cell and α-4-1BB mAb therapy, compared with the cohort treated with CAR T cells and isotype control antibody (Fig. 7A). We also investigated the effect of combination treatment on the presence of another immunosuppressive cell population, myeloid derived suppressor cells (MDSC). MDSCs were defined as CD11b+ Ly6Chigh and were also shown to be MHCIIlow CD43+, a phenotype previously reported to represent MDSCs. Similar to our previous observation with reduced Treg cell frequency, our data revealed significantly reduced frequency of MDSCs at the tumor site following combined CAR T-cell and α-4-1BB mAb treatment compared with mice that received CAR T cells and isotype control antibody (Fig. 7B). The decrease in the frequency of host Tregs and MDSCs consequently led to an increased ratio of CD8+ transferred CAR T cells to Tregs and MDSCs, respectively, correlating with enhanced therapeutic responses (Fig. 7C and D). Interestingly, we observed no apparent effect of α-4-1BB mAb on the frequency of host CD8+ T cells (Supplementary Fig. S5B) or total number of CD45.2+ cells (Supplementary Fig. S5C). Unlike the effects we observed for adoptively transferred CAR T cells, we found no difference in the cytokine-producing function of host endogenous T cells following α-4-1BB mAb treatment.
We next assessed the effects of α-4-1BB mAb treatment on APCs, including dendritic cells (DC) and macrophages. Although we did not observe changes in CD11b+ F4/80+ macrophage frequency following α-4-1BB administration at the tumor site, interestingly, we found a significant reduction in frequency of MHCII+ CD11b+ CD11c+ mature DCs at the tumor site on day 9 posttreatment following α-4-1BB mAb administration (Fig. 7E). Reduction in DC frequency at the tumor site was observed only after day 6 (Supplementary Fig. S6A), which may reflect engagement of the host immune response and migration of DCs to secondary lymphoid organs. Indeed, analysis of tumor-draining lymph nodes showed increased frequency of the MHCII+ CD11b+ CD11c+ DCs in the α-4-1BB–treated group on day 9 posttreatment (Supplementary Fig. S6B). Interestingly, these changes in myeloid compartment were not observed in RAG−/− mice treated with the combination therapy (Supplementary Fig. S7A and S7B).
Like T cells, natural killer (NK) cells have also been reported to upregulate the expression of 4-1BB upon activation, and stimulation of these cells with α-4-1BB mAb has been shown to promote their function and antibody-dependent cellular cytotoxicity (32, 33). To investigate the potential involvement of NK cells in our combination therapy, we examined the frequency of tumor-infiltrating NK cells 9 days posttreatment. Surprisingly, NK-cell frequency was diminished in the mice receiving treatments containing α-4-1BB mAb (Fig. 7F). To further investigate the possible mechanism leading to this decline in NK-cell number, we performed FACS analysis to examine the 4-stage NK-cell maturation program based on surface expression of CD11b and CD27. The phenotypic progression from CD11blow CD27low → CD11blow CD27high → CD11bhigh CD27high → CD11bhigh CD27low is known to be associated with progressive gain-of-effector function (34). Interestingly, on day 6 posttreatment, when we were able to analyze NK-cell phenotype despite the decrease in frequency following α-4-1BB mAb treatment, we found that the majority of NK cells in the combined CAR T-cell and α-4-1BB treatment group were at the most mature effector stage, as defined by CD11bhigh CD27low expression (Supplementary Fig. S8). These data indicate that α-4-1BB mAb may stimulate the maturation of NK cells into a terminally differentiated state, potentially resulting in an overall reduction in frequency. Taken together, our results strongly suggest that the enhanced antitumor efficacy following combination therapy against Her2+ tumors is primarily due to increased IFNγ and Ki67 expression by CAR T cells and is commensurate with decreased frequency of host immunosuppressive cells infiltrating the tumor, including Tregs and MDSCs.
Adoptive immunotherapy utilizing CAR T cells has emerged as a highly effective and specific strategy for treating hematologic malignancies, such as ALL (4). However, similar CAR T cell trials against established solid malignancies have reported only moderate responses, in part due to the hostile immunosuppressive tumor microenvironment that can significantly dampen the functional activity of tumor-infiltrating T cells (7). To overcome this problem, we explored the potential of using an exogenous agonistic antibody to enhance CAR T-cell therapeutic efficacy, an approach that has never been investigated in the context of CAR T-cell therapy. The α-4-1BB agonistic antibody is a promising and highly translational candidate in cancer therapy, given its potent ability to induce T-cell–mediated antitumor responses in vivo (16, 35, 36), facilitate stable memory generation (37), and its safe use in recent phase I/II clinical trials against various types of cancer (25). Ligation of 4-1BB is known to trigger recruitment of TRAF1 and TRAF2 adapter proteins, resulting in the activation of NF-κB and MAPK signaling pathways leading to T-cell activation (38). The capacity of 4-1BB costimulation to potently augment T-cell activation therefore makes it a promising target for enhancing T-cell antitumor responses.
One approach to providing 4-1BB costimulation to CAR T cells has been via direct incorporation of the 4-1BB costimulatory domain in addition to the existing CD28 costimulatory and CD3ζ signaling domains, referred to as third-generation CAR (19). This strategy has been investigated in several preclinical mouse tumor models with contrasting results (19–22), and therefore, it remains unclear whether the third-generation CAR is superior to the more widely used CD3ζ-CD28 second-generation CAR with regard to antitumor efficacy in vivo. Further preclinical testing targeting a broad range of tumor antigens is required to determine the true potential of third-generation CAR.
Another interesting approach to provide 4-1BB costimulation has involved the overexpression of the 4-1BB ligand (4-1BBL) on the surface of CD3ζ-CD28 CAR T cells targeting the CD19 antigen (39). Data from this study indicated that CD28 contained within the CAR construct was a strong driver of the antitumor response. In contrast, 4-1BB signaling was more effectively delivered following coexpression of 4-1BBL and the CD3ζ-CD28 CAR compared with 4-1BB signaling provided through a third-generation CD3ζ-4-1BB-CD28 CAR, owing to its ability to provide trans-costimulation to other T cells and potentially recruit host immune cells (39). This supports our hypothesis that unlike having the 4-1BB domain incorporated into the CAR, exogenous α-4-1BB mAb may additionally provide stimulation to not only the host endogenous T cells, but also other 4-1BB–expressing cells, such as NK cells, DCs, and neutrophils (12), potentially resulting in a multipronged immune attack against the tumor. Indeed, our studies reveal that the provision of α-4-1BB mAb modulates several host immune cell subsets, including Tregs, DCs, and MDSCs. Furthermore, it has been reported that endothelial cells express 4-1BB, and stimulation with α-4-1BB mAb can improve recruitment of activated T cells to the tumor site (40). Although we did not observe increased CAR T-cell frequency at the tumor site following α-4-1BB mAb treatment, we cannot exclude the possibility that α-4-1BB mAb may help with the initial trafficking of CAR T cells to the tumor site. The current technology to efficiently express both CAR and 4-1BBL transgenes is challenging and raises safety concerns; therefore, the use of exogenous α-4-1BB mAb is another potential strategy to enhance both the recruitment and activation of T cells, two important determinants for efficient tumor elimination.
To test our hypothesis that the use of α-4-1BB mAb could enhance CAR T-cell antitumor responses, we first showed that 4-1BB is upregulated following Her2-specific stimulation through the CAR, similar to as observed upon TCR activation. Indeed, coculture assays revealed a marked increase in the production of IFNγ secreted by CAR T cells following α-4-1BB mAb stimulation, although interestingly, no modulation of other cytokines or CAR T-cell cytotoxic capacity was observed in these in vitro assays. In adoptive transfer experiments in mice bearing established Her2+ tumors, we demonstrated striking tumor growth inhibition in two different tumor models following combined anti-Her2 CAR T-cell and α-4-1BB mAb therapy, compared with CAR T cells alone or control T cells with α-4-1BB mAb. Overall, this is the first study to demonstrate that activating a major costimulatory pathway using an exogenous immune agonist can potently increase antitumor efficacy by adoptively transferred CAR T cells.
Investigation into the mechanism underlying the increased therapeutic effects revealed that α-4-1BB mAb enhanced the production of IFNγ by both CD8+ and CD4+ CAR T cells at the tumor site. In addition, we also observed increased expression of the proliferation marker Ki67 in both CD8+ and CD4+ CAR T cells following α-4-1BB mAb stimulation, indicating that α-4-1BB could help maintain the proliferation of both CD8+ and CD4+ CAR T cells. This is an interesting finding given that signaling through 4-1BB is reportedly biased toward CD8+ T cells (12), and a previous study reported increased IFNγ secretion and proliferation of TCR-activated CD8+ T cells to a much greater extent than CD4+ T cells upon stimulation with α-4-1BB (41). Given the emerging data indicating that CD4+ CAR T cells play an important role in antitumor responses (42) and that T cell expansion and persistence is a major requirement for effective adoptive immunotherapy (43), this effect of α-4-1BB mAb on these parameters is potentially important for augmenting CAR T-cell responses, particularly for solid cancers where their activity has been limited to date. Similar to another study that used MHC-I–restricted OT-I T cells (36), we found no significant increase in the frequency of adoptively transferred Ly5.1+ CAR T cells at the tumor site following α-4-1BB mAb administration at early or later time points. Collectively, our data suggest that α-4-1BB mAb plays a significant role in enhancing the function of CAR T cells at the tumor site following antigen engagement.
A striking aspect of the current study was that the strong antitumor responses observed following CAR T-cell and α-4-1BB combined treatment correlated with a marked decrease in the frequency of host immunosuppressive CD4+ Foxp3+ Tregs and CD11b+ Ly6Chigh MDSCs at the tumor site. This consequently led to an increased ratio of transferred CD8+ CAR T cells to Tregs and MDSCs, respectively. This is a highly significant observation given the clinical evidence that elevated levels of Tregs and MDSCs have a negative impact on patient survival (44, 45), and a high CD8+:Treg ratio is associated with a favorable prognosis in cancer patients (46). The mechanisms underlying this decrease in Treg frequency following α-4-1BB mAb stimulation are not fully understood and may be either direct and/or indirect (25). It has been reported that Tregs express 4-1BB (47); hence, it is possible that α-4-1BB mAb could have a direct effect on these cells. However, other studies have also demonstrated that various cytokines in the tumor microenvironment, such as IL2, IL6, TGFβ, and others, can influence Treg number and function (48), and therefore, modulation of Treg frequency may be indirect as a consequence of enhanced proinflammatory cytokine production from CAR T cells. The decrease in Treg frequency observed in our study could be an important factor contributing to the enhanced therapeutic efficacy following combined CAR T-cell and α-4-1BB mAb treatment. In contrast to Tregs, MDSCs are not reported to express 4-1BB (49), and therefore, we hypothesize that the effects of α-4-1BB mAb on the MDSC population in our tumor models are indirect and most likely cytokine driven. Modulation of MDSCs by α-4-1BB mAb was not observed when CAR T cells were transferred to RAG−/− mice. This could indicate a role for host T cells in this effect. However, we observed that the number of CAR T cells trafficking to the tumor site was reduced in RAG−/− mice compared with hHer2 transgenic mice, potentially impacting on their ability to modulate the cytokine microenvironment and therefore MDSC maturation and/or trafficking.
Although α-4-1BB mAb has been reported to increase NK-cell proliferation and function (32), we surprisingly found in our study that NK cells were almost completely diminished at the tumor site following α-4-1BB mAb administration. The remaining NK cells were predominantly of the mature phenotype CD11bhigh CD27low following α-4-1BB mAb treatment, suggesting that α-4-1BB could potentially induce these NK cells to be more cytotoxic. It remains unclear whether these NK cells play a role in our combined CAR T-cell and α-4-1BB mAb therapy. However, Weigelin and colleagues demonstrated in an OT-I adoptive transfer model that NK cells were not necessary for the therapeutic efficacy of α-4-1BB mAb as shown by similar efficacy in an NK-cell–depleted setting (36). Another interesting observation was the decrease in mature DCs at the tumor site that was concomitant with an increase in this subset in the draining lymph node. Although we found no evidence of increased function of endogenous T cells following α-4-1BB mAb therapy, we cannot exclude the possibility that these DCs are presenting tumor-derived antigen to both host and adoptively transferred T cells that may have contributed to the overall response.
In summary, this study has shown for the first time that the use of an immune agonist α-4-1BB mAb has a multifunctional role for significantly enhancing CAR T-cell adoptive therapy by augmenting CAR T-cell function as well as alleviating the host immunosuppressive tumor microenvironment. This combination is likely to have a significant impact, as it highlights the potential for improving adoptive therapy for various solid cancers that are refractory to first-line treatment. Importantly, this combination has high translational potential given that two fully humanized α-4-1BB mAbs have been developed for clinical use, and the safe clinical use of Her2-specific second-generation CD3ζ-CD28 CAR T cells has been previously demonstrated in cancer patients (50).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: S. Mardiana, C.Y. Slaney, J.A. Trapani, P.J. Neeson, P.A. Beavis, P.K. Darcy
Development of methodology: S. Mardiana, C.Y. Slaney, P.A. Beavis, P.K. Darcy
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Mardiana, L.B. John, M.A. Henderson, C.Y. Slaney, B. von Scheidt, L. Giuffrida, P.A. Beavis, P.K. Darcy
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Mardiana, L.B. John, C.Y. Slaney, S. Loi, M.H. Kershaw, P.A. Beavis, P.K. Darcy
Writing, review, and/or revision of the manuscript: S. Mardiana, L.B. John, C.Y. Slaney, A.J. Davenport, J.A. Trapani, P.J. Neeson, S. Loi, N.M. Haynes, M.H. Kershaw, P.A. Beavis, P.K. Darcy
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Mardiana, M.A. Henderson, B. von Scheidt, L. Giuffrida, P.K. Darcy
Study supervision: L.B. John, N.M. Haynes, M.H. Kershaw, P.A. Beavis, P.K. Darcy
This work was funded by a project and program grant from the National Health and Medical Research Council (NHMRC; grant number 1062580) and a Cancer Council Victoria project grant (APP1084420). P.A. Beavis and C.Y. Slaney were supported by National Breast Cancer Foundation fellowships (PF-14-008 and ECF-16-005). P.K. Darcy and M.H. Kershaw were supported by NHMRC senior research fellowships (APP1041828 and APP1058388).
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.
The authors thank animal facility technicians at the Peter MacCallum Cancer Centre for the help with animal care and the histology department for processing and staining the H&E sections. The authors also acknowledge CRC Cancer Therapeutics (CTx) for the top-up PhD scholarship supporting Sherly Mardiana.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received July 4, 2016.
- Revision received November 23, 2016.
- Accepted December 12, 2016.
- ©2017 American Association for Cancer Research.