A novel form of 4-1BBL prevents cancer development via non-specific activation of CD4⁺ T and Natural Killer cells

Abstract

Costimulation through 4–1BB (CD137) receptor generates robust CD8⁺ T effector and memory responses. The only known ligand, 4–1BBL, is a trimeric transmembrane protein that has no costimulatory activity as a soluble molecule. Thus, agonistic antibodies to the receptor have been used for cancer immunotherapy in preclinical models and are currently being evaluated in the clinic. Here we report that treatment with an oligomeric form of the ligand, SA-4–1BBL, as a single agent is able to protect mice against subsequent tumor challenge irrespective of the tumor type. Protection was long-lasting (> 8 weeks) and a bona fide property of SA-4–1BBL, as treatment with an agonistic antibody to the 4–1BB receptor was ineffective in generating immune protection against tumor challenge. Mechanistically, SA-4–1BBL significantly expanded IFN-γ-expressing, pre-existing memory-like CD44⁺CD4⁺ T cells and NK cells in naïve mice as compared to the agonistic antibody. In vivo blockade of IFN-γ or depletion of CD4⁺ T or NK cells, but not CD8⁺ T or B cells, abrogated the immunopreventive effects of SA-4–1BBL against cancer. SA-4–1BBL as a single agent also exhibited robust efficacy in controlling postsurgical recurrences. This work highlights unexpected features of SA-4–1BBL as a novel immunomodulator with implications for cancer immunoprevention and therapy.

Introduction

4–1BB (CD137; TNFRSF9) is a potent T-cell costimulatory receptor that belongs to the tumor necrosis factor receptor superfamily. 4–1BB is primarily expressed on the surface of activated lymphoid cells, including T cells, B cells, and NK cells (1). Signaling through 4–1BB has been shown to result in the survival, expansion, and differentiation of T cells, particularly CD8⁺ T cells, into effectors and the establishment of long-term memory (2,3). Given the demonstrated role of CD8⁺ T cells as important effectors of cancer immunotherapy, the 4–1BB pathway has been the subject of intense basic and translational research in the immuno-oncology field. Agonistic antibodies (Abs) to 4–1BB alone or in combination with other anti-cancer agents have shown therapeutic efficacy in various preclinical models (2–5), which led to efforts of translating these findings into the clinic. Presently, there are ongoing clinical trials to evaluate the efficacy of agonistic 4–1BB antibodies alone or in combination with other immunotherapy and chemotherapy modalities. However, the use of agonistic antibodies to 4–1BB was reported to cause significant hepatic toxicity and other complications in preclinical (6,7) as well as clinical studies (8). Importantly, treatment with agonistic antibodies was shown to have deleterious effects on various immune cells, including CD4⁺ T cells, humoral immune responses, and NK cells (6,7). It is presently unknown if these adverse effects are bona fide physiological characteristics of 4–1BB receptor signaling or complications associated with the use of agonistic antibodies.

There is only one known natural 4–1BB ligand (4–1BBL) expressed as a type II transmembrane protein primarily on antigen presenting cells, such as dendritic cells (DCs), macrophages, and B cells (9,10). The membranous form of 4–1BBL exists as a trimer, and upon engagement with its receptor on T cells it delivers a robust costimulatory signal (11). In marked contrast, the trimeric soluble form of 4–1BBL lacks costimulatory functions and requires cross-linking to either solid surfaces or by other means to acquire costimulatory function (12). We generated a recombinant chimeric protein, SA-4–1BBL, containing the extracellular domains of murine 4–1BBL fused to a modified form of core streptavidin (7,13). SA-4–1BBL forms tetramers and oligomers with robust T cell costimulatory activity in soluble form. We showed that SA-4–1BBL blocked the conversion of T conventional cells into CD4⁺CD25⁺Foxp3⁺ T regulatory cells (Tregs) that was dictated by the production of IFN-γ in T conventional cells (14). SA-4–1BBL also overcame Treg suppression by stimulating the production of IL-2 in T effector cells (Teffs) (15). Importantly, treatment with SA-4–1BBL did not result in various immune system anomalies in mice, such as systemic cytokine storm, splenomegaly, lymphadenopathy, and hepatitis, otherwise reported for 4–1BB antibodies (7,16).

As an adjuvant component of tumor-associated antigen-based subunit vaccines, SA-4–1BBL generated robust T effector responses with therapeutic efficacy in various preclinical tumor models (17–22). We herein report unexpected findings that treatment with SA-4–1BBL alone protects mice against subsequent tumor challenge. This was a unique feature of the SA-4–1BBL molecule as an agonistic 4–1BB antibody did not protect mice against tumor challenge. This prophylactic effect was long-lasting, tumor type-independent, and required both CD4⁺ T and NK cells as well as IFN-γ. Moreover, treatment with SA-4–1BBL after surgical removal of tumors resulted in control of relapses. To our knowledge, this is the first study to demonstrate that an immune checkpoint stimulator can prime the immune system for cancer prevention. SA-4–1BBL as a novel immune modulator with distinct immune functions from agonistic 4–1BB antibody has implications for cancer prevention and therapy.

Materials and Methods

Mice

C57BL/6 mice were purchased from The Jackson Laboratory. Mice were bred and cared for in a University of Louisville specific pathogen-free animal facility in accordance with NIH guidelines. All animal procedures were conducted under protocols approved by Institutional Animal Care and Use Committee at the University of Louisville.

Antibodies, recombinant proteins, and cell lines

Fluorescent-conjugated antibodies to various cell surface markers were obtained commercially including: α-CD3-V500 (BD Horizon 560771); α-CD4-Alexa700 (BD Pharmingen 557956); α-CD8-APC-Cy7 (BD Pharmingen 557654); α-NK1.1-PE (BD Pharmingen 553165); α-CD19-APC (BD Pharmingen 550992); α-CD19-PE-Cy7 (eBioscience 25-0193-82); α-CD44-APC (eBioscience 17-0441-83); α-CD62L-PerCp-Cy5.5 (eBioscience 45-0621-82); α-CD69-FITC (BD Pharmingen 553236); α−4-1BB-PE (eBioscience 12-1371-83).

Anti-4–1BB agonistic antibody (clone 3H3) was produced in our laboratory (7). Antibodies used intraperitoneally (i.p.) for the depletion of specific immune cells include CD4 (clone GK1.5, 500 μg/injection), NK1.1 (clone PK136, 500 μg/injection), CD8 (clone 53.6.72, 500 μg/injection), and CD20 (clone 5D2, 200 μg/injection). IFN-γ was blocked in vivo by i.p. injection of 200 μg of an anti-IFN-γ antibody (clone XMG1.2, BioXcell) on days 0, 3, 14, 17, 20 with respect to first SA-41BBL treatment. SA-4–1BBL and streptavidin proteins were produced in our laboratory according to standard protocols as previously reported (13,23). TC-1 and Lewis lung carcinoma (LLC) tumor cell lines were obtained and maintained according to American Type Culture Collection (ATCC). 3LL-huMUC1 cell line was a generous gift from Dr. Jun Yan at University of Louisville, Louisville, KY. All tumor cell lines were passaged at least twice, but not more than six times for injection purposes. Cell lines were authenticated by flow cytometry to check for the expression of mouse MHC class I haplotype (H-2^b) as well as the expression of HPV E7 for TC-1 and human MUC1 for 3LL-huMUC1. Cells were not tested for mycoplasma.

SA-4–1BBL treatment and tumor challenge

Mice were treated s.c. with SA-4–1BBL at the indicated doses once or twice two weeks apart as specified. Mice were challenged s.c. in the left back flank with 1 × 10⁵ live TC-1, LLC, or 3LL-huMUC1 tumor cell lines as indicated. Selected groups were vaccinated 6 days post-tumor challenge with 50 μg of HPV E7 peptide 1 (P1, RAHYNIVTF) serving as the dominant E7 epitope for CD8⁺ T cells adjuvanted with 25 μg SA-4–1BBL protein. Animals were monitored for tumor growth, and tumors were measured twice a week using calipers. Animals were euthanized at a 60-day experimental end-point or when tumors ulcerated or reached a size of ~12 mm in diameter.

To test the therapeutic efficacy of SA-4–1BBL as monotherapy, TC-1 or 3LL-huMUC1 tumors of ~4 mm in diameter were surgically removed under sterile conditions and avertin anesthesia (250 mg/kg). After 48 hours of recovery period, animals were treated s.c. with SA-4–1BBL (25 μg/injection) twice, two weeks apart. Animals without SA-4–1BBL treatment served as controls and were monitored for tumor relapse.

Anti-streptavidin antibody titers

Sera collected at the indicated times from control and treatment groups were assessed for anti-streptavidin antibodies using ELISA. Briefly, 96-well flat-bottom plates were coated with SA-4–1BBL (50 ng/well) or control streptavidin (50 ng/well) proteins in sterile PBS and incubated overnight at 4^oC. Wells were then washed three times with the wash buffer (PBS/Tween-20) then incubated with a nonfat milk blocking buffer for 1 h to block nonspecific binding. After washing the plate three times with the wash buffer, the wells were incubated with serial dilutions of sera at room temperature for 1.5 h. After several washes, the wells were incubated with a secondary antibody conjugated to horseradish peroxidase (HRP) for 1 h. Plates were then incubated for 30 min with TMB substrate (BD Biosciences, Cat#555214) and read on Wallac Victor 1420 Multilabel microplate reader at 450 nm.

Passive serum transfer

Mice were treated s.c. twice with SA-4–1BBL (25 μg/treatment) two weeks apart and serum was collected 27 days after the initial treatment. Serum was assessed for antibody titers against streptavidin and then injected i.v. into C57BL/6 mice (200 μl/animal) 24 hours prior to the TC-1 subcutaneous tumor challenge (1 × 10⁵ cells).

SA-4–1BBL T cell costimulation in vitro assay

C57BL/6 splenocytes (2 × 10⁵ cells/well) were cultured in 96-well U-bottom plates and stimulated with a suboptimal dose of an agonistic antibody to CD3 (0.25 μg/ml). Cultures were then supplemented with various doses of SA-4–1BBL preincubated at room temperature for 1 hr in naïve serum or serum with positive antibody titers against SA. Cultures were then incubated for 48 h and pulsed with [³H]-thymidine for an additional 16 h. Plates were harvested with Tomtec Cell Harvester, and DNA-associated radioactivity was measured using a Beta plate counter and graphed as counts per minute (CPM).

Flow cytometry and phenotyping

Lymphocytes harvested from the spleen and injection site-draining lymph nodes of naïve or various treatment groups were stained with fluorescent-conjugated antibodies to various cell surface and intracellular markers. Cells were analyzed using multiparameter LSRII flow cytometry (BD Biosciences) by gating on live cells. Cell percentages and absolute numbers were calculated and reported.

Statistics

Statistics were performed with GraphPad Prism 6 software (La Jolla, CA). Survival was assessed using Kaplan-Meier method and log-rank test. Student’s t-tests were used to compare differences between two individual groups. Where indicated, one way ANOVA was applied as well. P values of ≤ 0.05 were considered statistically significant.

Results

SA-4–1BBL as a monotherapy protects mice against tumor challenge

We previously demonstrated that a subunit vaccine containing a synthetic peptide (P1) representing the dominant CD8⁺ T cell epitope for human papilloma virus (HPV) E7 protein adjuvanted with SA-4–1BBL had therapeutic efficacy against HPV TC-1 tumor model in C57BL/6 mice (21,22). Streptavidin as a bacterial protein has the potential to generate humoral immune responses that may negatively impact the efficacy of SA-4–1BBL-adjuvanted vaccines. To assess this possibility, mice were pretreated twice with SA-4–1BBL protein alone (25 μg/injection) two weeks apart, followed by TC-1 tumor challenge and vaccination as schematically depicted in Fig. 1A. Treatment with SA-4–1BBL protein generated both humoral (Fig. 1B) and cellular (fig. S1) immune responses against streptavidin. To assess if anti-streptavidin antibodies impact the function of SA-4–1BBL, the protein was preincubated with naïve or immune serum and then tested for T-cell costimulatory activity in vitro. SA-4–1BBL protein pre-incubated with serum containing high titers of anti-streptavidin antibodies had a similar costimulatory activity in driving T cell proliferation as SA-4–1BBL preincubated with naïve serum (Fig. 1C).

Figure 1.

Pretreatment with SA-4–1BBL in the absence of any TAA antigen confers protection against tumor. A, Experimental design and timeline. B, Anti-SA antibody titers. C57BL/6 mice were pretreated twice with 25μg SA-4–1BBL (days 0 and 14), followed by TC-1 tumor challenge (1 × 10⁵) on day 28. A group of mice was also vaccinated s.c. with 50μg E7-P1 peptide + 25μg SA-4–1BBL 8 days post-tumor challenge. Serum was collected on days 21, 34, and 60 post-treatment with the initial dose of SA-4–1BBL and anti-SA antibody titers were assessed with ELISA. C, Anti-SA antibodies do not block the costimulatory function of SA-4–1BBL in vitro. SA-4–1BBL protein was incubated at different doses (μg/ml) with naïve or immune serum with high titers (Log₁₀ = 2.8) of anti-SA antibodies for 1 hour at room temperature. The costimulatory activity of SA-4–1BBL pre-incubated with serum was then assessed using an in vitro CD3-based T cell proliferation assay. SA-4–1BBL and SA-4–1BBL preincubated with naïve serum were used at similar doses as controls. D, Kaplan-Meier survival curves of mice subjected to different treatments. Mice pretreated with SA-4–1BBL alone and those pretreated with SA-4–1BBL followed by immunization with SA-4–1BBL-adjuvanted E7-P1 subunit vaccine according to the experimental scheme in (A) were monitored for tumor growth. Mice were euthanized when average tumor size reached ≥ 12 mm in diameter or on day 60 post-tumor challenge as experimental end-point. Data was analyzed using Kaplan-Meier Survival curve and log-rank statistical method. A P value ≤ 0.05 was considered significant.

Consistent with the in vitro data, pretreatment with SA-4–1BBL protein did not alter the therapeutic efficacy of the subunit vaccine, as all mice in this group remained tumor-free for a 60-day observation period (Fig. 1D). Surprisingly, pretreatment with SA-4–1BBL alone also protected all mice against TC-1 tumor challenge (Fig. 1D). Together, these data demonstrate that the immunogenicity of streptavidin does not negatively impact the therapeutic efficacy of SA-4–1BBL-adjuvanted vaccine and that SA-4–1BBL as a single agent protects mice against tumor challenge, a highly surprising and unexpected finding.

SA-4–1BBL generates a rapid and lengthy window of protection against tumor

We next assessed the kinetics of SA-4–1BBL-conferred protection against tumor. A single treatment with SA-4–1BBL followed by TC-1 tumor challenge one day later did not result in protection, as all mice developed tumor at a rate similar to untreated, control animals (Fig. 2A). Delaying the tumor challenge by one week resulted in significant retardation in tumor growth (p < 0.02), but all mice eventually expired from the tumor burden (Fig. 2A). The SA-4–1BBL mediated protective response was further improved when tumor challenge was delayed by two weeks, resulting in the survival of ~ 40% of mice during a 60-day observation period.

Figure 2.

Time-course dynamics of the SA-4–1BBL generated anti-tumor responses. A, C57BL/6 mice were pretreated once with 25 μg SA-4–1BBL followed by TC-1 challenge (1 × 10⁵ cells) at the indicated time points. B, Pretreatment with SA-4–1BBL twice, two weeks apart followed by TC-1 challenge at the indicated time points. Mice were monitored for tumor growth and euthanized when average tumor size reached ≥ 12 mm in diameter or at the 60-day experimental endpoint. Data was analyzed using Kaplan-Meier Survival curve and log-rank statistical method. A P value ≤ 0.05 was considered significant.

Treatment with SA-4–1BBL twice, two weeks apart improved tumor response kinetics, efficacy, and duration. Forty percent of mice challenged with TC-1 tumor one day after the second SA-4–1BBL treatment survived over an observation period of 60 days (Fig. 2B). Delaying the time of tumor challenge by 1, 2, or 4 weeks resulted in complete protection, as all mice in these three groups survived by the 60-day experimental end-point (Figs. 2B, 1D). Importantly, the tumor-protective effect was long-lasting as 80% of mice challenged with tumor eight weeks post SA-4–1BBL treatment survived by the 60-day experimental end-point. However, the SA-4–1BBL conferred protective effect was lost when tumor challenge was delayed by twelve weeks (Fig. 2B and fig. S2). These data demonstrate that the SA-4–1BBL-induced protective response is rapid, evolves fully within three weeks, and lasts more than eight weeks.

SA-4–1BBL-generated protection is tumor-type independent and does not evolve into a long-lasting immune memory

We next investigated if the protective effect of SA-4–1BBL as a single agent against TC-1 extends to other tumor types. C57BL/6 mice were pre-treated with 25 μg SA-4–1BBL twice, 2 weeks apart, followed by a challenge with LLC tumor cells 2 weeks after the second immunization (Fig. 3A). There was significant protection (p < 0.0001) against LLC tumor with 40% of mice surviving over a 60-day observation period (Fig. 3B). The protection against tumor conferred by SA-4–1BBL was dose-dependent. Treatment with 100 μg SA-4–1BBL twice, two weeks apart followed by challenge a week later with 3LL tumor cells expressing the human mucin 1 protein (3LL-huMUC1) resulted in 100% survival over an observation period of 60 days (Fig. 3C). Treatment with smaller 12.5, 25, and 50 μg doses of SA-4–1BBL resulted in lower rates of survival 0%, 30%, and 80%, respectively. Flow cytometric analyses demonstrated the lack of expression of on all the 3 tumor cell lines (fig. S3), ruling out the possibility of SA-4–1BBL directly targeting tumor cells for the observed protective effect.

Figure 3.

Prevention conferred by SA-4–1BBL is dose-dependent and effective against different tumor types. A, Experimental design and timeline. B, Pretreatment with SA-4–1BBL protects mice against different tumor types. C57BL/6 mice were pretreated with 25 μg SA-4–1BBL twice, 2 weeks apart and challenged 2 weeks later with LLC or TC-1 tumor cells. Tumor measurements were recorded twice weekly using calipers and mice were euthanized when tumor size reached ≥ 12 mm in diameter. C, Pretreatment with SA-4–1BBL protected mice against tumor challenge in a dose-dependent manner. C57BL/6 mice were pretreated with the indicated doses of SA-4–1BBL twice, two weeks apart. Mice were challenged with the 3LL-huMUC1 tumor cells one week post-SA-4–1BBL treatment and then monitored for tumor growth. D, The prophylactic efficacy of SA-4–1BBL results in moderate immune memory response against tumor. Mice were pretreated with SA-4–1BBL (25 μg/injection, twice, 2 weeks apart) and two weeks later challenged s.c. with 1×10⁵ TC-1 tumor cells. A second group of animals treated the same were vaccinated s.c. with SA-4–1BBL-adjuvanted E7 P1 peptide (50 μg peptide + 25 μg SA-4–1BBL) 8 days after tumor challenge. Mice in both groups were monitored for 60 days, and those without tumor were re-challenged with a second dose of TC-1 tumor cells to assess immune memory. Animals were monitored for an additional 80 days. Data was analyzed using Kaplan-Meier Survival curve and log-rank statistical method. A P value ≤ 0.05 was considered significant.

We have previously shown that SA-4–1BBL-adjuvanted subunit vaccines generate a long-term immune memory response, primarily driven by CD8⁺ T cells, against tumors (19–22). To test whether the prophylactic efficacy of SA-4–1BBL leads to a long-lasting immune memory, mice free of TC-1 tumor for 60 days were re-challenged with a second dose of tumor cells and monitored for tumor growth. There was a significant (P = 0.0034) delay in tumor growth as compared with the controls, but all mice eventually expired from tumor burden (Fig. 3D). In contrast and consistent with our previously published data (21,22), mice that had eradicated tumors in response to SA-4–1BBL-adjuvanted P1 subunit vaccine showed immune memory with 60% of animals surviving for an observation period of 80 days post second tumor challenge (Fig. 3D). These results show that the preventive immune response generated by SA-4–1BBL as a single agent lacks the characteristics of a classical adaptive immune response against tumors that benefit from a long-lasting memory.

An agonist antibody to 4–1BB does not confer protection against tumor challenge

Agonistic antibodies to 4–1BB have been used for cancer immunotherapy successfully in preclinical models (2,3,5) and are presently being tested in clinical cancer trials (24). Previous studies from our laboratory reported qualitative and quantitative differences between SA-41BBL and an agonistic antibody, 3H3, to 4–1BB receptor (7). We, therefore, assessed if the pretreatment with 3H3 antibody generates tumor preventive immune responses. C57BL/6 mice were treated twice with 3H3 (100 μg/injection, 2 weeks apart) followed by TC-1 tumor challenge two weeks after the second antibody treatment. Surprisingly, pretreatment with the agonistic antibody alone did not protect mice against tumor challenge, as all mice developed tumors in a similar tempo to untreated controls (Fig. 4A). To eliminate the potential contribution of streptavidin as a foreign antigen to the SA-4–1BBL-induced protection, a separate group of mice was pretreated with 3H3 plus an equimolar amount of streptavidin in the treatment with SA-41BBL. Streptavidin alone or in combination with 3H3 antibody did not protect the mice against TC-1 tumor challenge. In marked contrast and consistent with data presented in Fig. 2, pretreatment with SA-4–1BBL (25 μg/injection, 2 weeks apart) resulted in ~ 90% survival for a 60-day observation period (Fig. 4A).

Figure 4.

SA-4–1BBL is a bona fide novel anti-tumor immunomodulator, and humoral immunity is dispensable for its prophylactic effect against tumors. A, Pretreatment with 3H3 agonist 4–1BB antibody does not result in protection against tumor. C57BL/6 mice were pretreated s.c. twice, two weeks apart with streptavidin alone (12.5 μg), 3H3 alone (100 μg), streptavidin + 3H3, or SA-4–1BBL (25μg), followed by TC-1 (1 × 10⁵) challenge s.c. 2 weeks after the last treatment. Animals were monitored for tumor growth and euthanized when tumor size reached ≥ 12 mm or at the 60-day experimental endpoint. B, Anti-streptavidin antibody titers in serum from mice in (A) pretreated with streptavidin alone, or streptavidin + 3H3 antibody collected on day 21 or experimental end-point. C, Passive transfer of immune serum with high titers against streptavidin into naïve mice does not protect against subsequent tumor challenge. Serum collected on day 27 from SA-4–1BBL treated mice (25 μg on days 0 and 14) was injected i.v. into naïve C57BL/6 mice (200 μl/animal) 24 hours prior to TC-1 challenge. D, B-cell depletion does not negate the prophylactic effect of SA-4–1BBL against tumor. C57BL/6 mice were pretreated twice, two weeks apart with SA-4–1BBL (25 μg/injection), followed by TC-1 (1 × 10⁵) challenge s.c. 2 weeks after the last SA-4–1BBL treatment. B cells were depleted using anti-CD20 antibody (clone 5D2; 200 μg/i.p. injection) 1 day before each SA-4–1BBL pretreatment. Data was analyzed using Kaplan-Meier Survival curve and log-rank statistical method. A P value ≤ 0.05 was considered significant.

Treatment with agonistic 4–1BB antibodies was shown to block humoral immune responses by inducing anergy in CD4⁺ T cells (6,25). Consistent with these reports, none of the mice preimmunized with streptavidin and 3H3 had detectable levels of anti-streptavidin antibodies on day 21 post-treatment, and only 1/6 mice scored positive at expiration from tumor burden (Fig. 4B). In marked contrast, mice immunized twice with streptavidin protein alone had high titers of anti-streptavidin antibodies on both day 21 and experimental endpoint. Humoral immunity has been shown to play a role in the efficacy of various cancer immunotherapies (26–28). Antibodies can have a direct effect on the tumor by recognizing and binding to surface antigens, or by helping antigen-presentation and processing by APCs through opsonization, thereby augmenting downstream adaptive immune responses. Given the positive titers of anti-streptavidin antibodies in the pretreatment setting, we asked if such antibodies can contribute to the protective effect observed against tumors. Passive transfer of serum (200 μl/mouse) with high antibody titers against streptavidin into naïve C57BL/6 mice 24 hours prior to TC-1 tumor challenge did not impact tumor growth as compared with controls (Fig. 4C). To further eliminate the role of antibodies in the observed protection against tumor, mice were injected with a depleting antibody against B cells and then treated with 25 μg SA-4–1BBL twice, two weeks apart. The depletion of B cells (fig. S4) did not negate the protective effect of SA-4–1BBL against TC-1 tumor challenge (Fig. 4D). Together, these findings reveal two substantial functional differences between agonistic antibody to 4–1BB and SA-4–1BBL. The antibody blocks humoral responses against streptavidin, and it does not protect mice against tumor challenge, whereas SA-4–1BBL shows opposite effects in both functions.

IFN-γ⁺ producing CD4⁺ T and NK cells as predictors of SA-4–1BBL-mediated immune protection against tumors

To establish immune correlates of protection against tumors and elucidate potential mechanistic differences between SA-4–1BBL and the agonistic 4–1BB antibody, mice were treated twice, two weeks apart with SA-4–1BBL or 3H3 antibody and euthanized 3 days later to collect lymphoid tissues for analyses. Naïve mice had significant percentages of CD4⁺ (~7%) and CD8⁺ (~10%) T cells expressing CD44 molecule as a memory marker (Fig. 5A, B). Treatment with SA-4–1BBL or 3H3 significantly increased the percentage and absolute numbers of both CD4⁺ and CD8⁺ T cells with memory phenotype in draining LNs as compared with naïve mice (Fig. 5B, C). A similar trend, particularly for the 3H3 antibody, was also observed for T cells in the spleen (fig. S5). Considerable percentage of memory-like CD4⁺ (> 18%) and CD8⁺ (> 3%) T cells in naïve mice also expressed 4–1BB on their surface (Fig. 5A, D). Treatment with SA-4–1BBL or 3H3 antibody significantly increased the absolute numbers of lymph node CD4⁺ and CD8⁺ T memory cells expressing 4–1BB on their surface as compared with naïve mice (Fig. 5E). Interestingly, treatment with SA-4–1BBL decreased the percentage of memory CD4⁺ T cells expressing 4–1BB as compared with naïve mice, plausibly due to cell surface modulation of the receptor on actively expanding cells as previously reported (29).

Figure 5.

Pretreatment with SA-4–1BBL increases the frequency of IFN-γ expressing memory-like CD4⁺ T cells and NK cells. C57BL/6 mice were treated s.c. twice, two weeks apart with SA-4–1BBL (n=4; 100 μg/injection) or agonistic 3H3 antibody (n=4). Animals were euthanized 3 days later to harvest injection-site draining lymph nodes for phenotyping using flow cytometry. A, Dot plot showing CD4⁺ and CD8⁺ T cells expressing CD44 and 4–1BB molecules. B and C, The frequency of memory like CD44 expressing CD4⁺ and CD8⁺ T cells. D and E, the frequency of memory like CD4⁺ and CD8⁺ T cells expressing 4–1BB receptor. F, Dot plot showing CD4⁺ and CD8⁺ T cells and NK cells expressing CD44 and IFN-γ. Memory like CD4⁺ and CD8⁺ T cells expressing IL-2 G, and IFN-γ H, and NKT and NK cells expressing IFN-γ I. Each data point is indicative of mean ± SEM, with *P < 0.05, **P < 0.01, ***P < 0.001; One way ANOVA with Bonferroni’s Multiple Comparison Test.

Treatment with SA-4–1BBL resulted in a significant increase in the percentage and absolute number of lymph node CD4⁺CD44⁺ T cells expressing IL-2 as compared with naïve mice (Fig. 5F, G and fig. S6). In marked contrast, treatment with 3H3 antibody resulted in an increase of lymph node CD4⁺CD44⁺ T cells expressing IL-2, but it was not significant as compared with naïve mice (Fig. 5G). This trend also applied to CD8⁺CD44⁺ T cells expressing IL-2, except the difference between 3H3 and naïve was significant (Fig. 5G). SA-4–1BBL treatment also resulted in a significant increase in the percentage and absolute number of lymph node CD4⁺CD44⁺ T cells expressing IFN-γ as compared with naive and 3H3 antibody treated mice (Fig. 5H and fig. S6). There was a similar trend for CD8⁺CD44⁺ T cells expressing IFN-γ, but the difference between SA-4–1BBL and 3H3 antibody was not significant (Fig. 5H). Importantly, treatment with SA-4–1BBL resulted in a significant increase in the percentage and absolute numbers of NK1.1⁺CD3⁻ NK cells expressing IFN-γ as compared with naïve and 3H3 antibody treated mice (Fig. 5I). A similar trend was also observed for splenic T and NK cells, but the differences between SA-4–1BBL and 3H3 was not as pronounced as for the LN cells (fig. S7). The observed differences induced by SA-4–1BBL in the number and percentages of CD4⁺CD44⁺ T cells between draining lymph nodes and spleen was not due to the differential expression of 4–1BB on these cells as determined by flow cytometry (fig. S8). Together, these findings demonstrate that the SA-4–1BBL preferentially expands memory like CD4⁺CD44⁺ T cells and CD3⁻ NK cells producing IFN-γ; features that are not shared by the agonist anti-4–1BB antibody.

SA-4–1BBL-mediated protection against tumor requires IFN-γ as a mediator of cross-talk between NK and CD4⁺ T cells

The significant increase in the frequency of IFN-γ production in response to SA-4–1BBL treatment as compared with 3H3 antibody led us to investigate the involvement of this cytokine in the observed tumor prevention. Mice were pretreated s.c. with SA-4–1BBL (100 μg/injection) twice, two weeks apart, followed by TC-1 s.c. challenge (1×10⁵ cells/animal) one week later. Animals were also treated with a blocking antibody to IFN-γ (200 μg/injection) for a total of 5 doses on days 0, 3, 14, 17, 20 in reference to SA-4–1BBL treatment (Fig. 6A). All mice treated with anti-FN-γ antibody developed tumor in a delayed fashion (day 42) as compared with control mice (day 9; Fig. 6B). Importantly, none of the mice pretreated with SA-4–1BBL developed tumor in a 60-day observation period.

Figure 6.

IFN-γ, CD4⁺ T, and NK cells are indispensable for the prophylactic anti-tumor efficacy of SA-4–1BBL. A, Study design. B, Blockade of IFN-γ results in tumor growth. C57BL/6 mice were pretreated s.c. twice with 100 μg SA-4–1BBL 2 weeks apart and challenged with TC-1 tumors 1 week later. One group received saline, while the other was treated with a blocking antibody against IFN-γ on days 0, 3, 14, 17, 20 as shown in (A). Animals were monitored for tumor growth and euthanized when tumor size reached ≥12 mm in diameter or at 60-day experimental end-point. C, Depletion of NK cells results in tumor growth. C57BL/6 mice were pretreated s.c. twice, two weeks apart with SA-4–1BBL (25 μg/injection) followed by TC-1 challenge s.c. (1 × 10⁵) 2 weeks later. NK cells were depleted using an anti-NK1.1 antibody twice, one day before each SA-4–1BBL pretreatment or once one day before TC-1 challenge. D, Depletion of CD4⁺ T, but not CD8⁺ T cells, results in tumor growth. Same experimental design as in (C) except that an antibody to CD4 was used either one day prior to SA-4–1BBL immunizations or one day before TC-1 challenge. CD8⁺ T cells were depleted one day before SA-4–1BBL treatment. Data was analyzed using Kaplan-Meier Survival curve and log-rank statistical method. E and F, Increased frequency of CD4⁺CD44⁺ T cells in lymphoid tissues of tumor-free mice. Mice were pretreated twice, two weeks apart with SA-4–1BBL followed by TC-1 tumor cell challenge 8 weeks later. Mice were then monitored for tumor growth for 60 days. Lymphocytes from spleens and draining lymph nodes of tumor-free mice were phenotyped for CD3⁺CD4⁺CD44⁺ and CD3⁺CD8⁺CD44⁺ cells using flow-cytometry. Each data point is indicative of mean ± SD. P value ≤ 0.05 was considered significant using Student’s t-test.

The significant increase in the absolute number of NK cells and CD4⁺ T cells expressing IFN-γ in SA-4–1BBL treated mice as compared with agonistic 3H3 antibody led us to directly probe the contribution of these cell populations to protection against tumor. Depletion of NK cells (fig. S9) using an antibody to NK1.1 molecule one day before treatment with SA-4–1BBL (priming phase) overcame protection against TC-1 tumor as both SA-4–1BBL-treated and untreated control mice showed a similar tumor growth tempo (Fig. 6C). NK cell depletion one day before tumor challenge (day 27, effector phase) also resulted in complete ablation of the protective effect of SA-4–1BBL, providing direct evidence for the role of NK as effector cells.

Depletion of CD4⁺ T cells (fig. S10A) one day prior to SA-4–1BBL treatment also resulted in the abrogation of anti-tumor protective effect in the TC-1 model (Fig. 6D). Although the depletion of this cell population 1 day before TC-1 challenge resulted in tumor growth in all mice, there was a significant (P = 0.0006) delay in tumor progression as compared with controls. In marked contrast, the depletion of CD8⁺ T cells (fig. S10B) had no impact on the anti-tumor effect of SA-4–1BBL (Fig. 6D). Immune monitoring of T cells in mice that cleared their tumor in the SA-4–1BBL pretreatment group demonstrated significantly increased frequency of memory CD4⁺ T cells in the spleen (Fig. 6E) and in the draining lymph nodes (Fig. 6F) analyzed 60 days post-tumor challenge as compared with naïve mice. The frequency of CD8⁺ T cells remained unchanged in the spleen but was significantly increased in the draining lymph nodes as compared with naive animals (Fig. 6E, F). These data demonstrate the obligatory effect of NK and CD4⁺ T cells as well as IFN-γ to sustain SA-4–1BBL-generated protection against tumors.

SA-4–1BBL prevents post-surgical tumor recurrences

Post-surgical tumor recurrence is a significant hurdle in cancer treatment. The observed preventive effect of SA-4–1BBL across various tumor types led us to test the efficacy of this molecule in controlling tumor recurrence in clinically relevant surgical resection models. C57BL/6 mice with established TC-1 tumors (~ 4 mm in diameter) were subjected to surgery to debulk the tumor. Animals were randomly assigned to treatment and control groups. Treatment with SA-4–1BBL (25 μg/injection, 2 weeks apart) resulted in complete blockade of tumor recurrence, while all controls had tumor relapse and expired from tumor burden within 50 days post-tumor resection (Fig. 7A). Importantly, one out of ten mice in the SA-4–1BBL treatment group had relapsed tumor, which was eventually eradicated (Fig. 7A). Treatment with the same SA-4–1BBL regimen also controlled post-surgical recurrences in the 3LL-huMUC1 tumor model (Fig. 7B). Importantly, 80% of mice that cleared primary tumors in response to post-surgical treatment with SA-4–1BBL did not develop tumor when re-challenged with TC-1 cells, demonstrating the establishment of long-term immune memory (Fig. 7C). Collectively, these results demonstrate that SA-4–1BBL as monotherapy is effective in controlling post-surgical tumor recurrences by generating effective and long-lasting adaptive immune responses.

Figure 7.

SA-4–1BBL treatment prevents post-surgical recurrences. A, Experimental design and TC-1 tumor relapse curves. C57BL/6 mice bearing TC-1 tumors (~ 4 mm in diameter) were subjected to surgery to remove the tumor. Two days later, mice were treated s.c. with SA-4–1BBL (25 μg/injection) twice, two weeks apart, and monitored for tumor recurrence. B, Same as in (A) except that 3LL-huMUC1 tumor cells were used. Mice that underwent surgical resection of the tumor without SA-4–1BBL treatment were used as controls. C, SA-4–1BBL therapy is associated with long-term immunological memory. Mice without tumor in (A) were re-challenged s.c. with TC-1 cells (1 × 10⁵) and monitored for an additional 60 days for recurrences. Data was analyzed using Kaplan-Meier Survival curve and log-rank statistical method. A P value ≤ 0.05 was considered significant.

Discussion

We herein report unexpected, novel findings that pretreatment with SA-4–1BBL as a robust agonist of 4–1BB costimulatory pathway protects mice against tumor challenge. Prophylactic efficacy is not restricted to a particular tumor type, is long-lasting, and depends on a cross-communication between CD4⁺CD44⁺ “memory-like” T and NK cells with an absolute requirement for IFN-γ. Importantly, this is a bona fide feature of SA-4–1BBL as an agonistic antibody to the 4–1BB receptor did not alter tumor growth. To the best of our knowledge, this is the first demonstration that an immunomodulator without an antigen primes the immune system for the prevention of tumor growth. As monotherapy SA-4–1BBL also controlled post-surgical tumor recurrences, highlighting its potential for both the prevention as well as treatment of cancer.

Immunotherapy for cancer has gained significant impetus precipitated by the remarkable clinical therapeutic efficacy of immune checkpoint inhibitors (30,31). Costimulatory receptors that initiate and sustain effector and memory immune responses represent the next potential targets for cancer immunotherapy. Among the costimulatory checkpoint receptors, 4–1BB is upregulated on T cells following activation and signaling through this pathway plays a paramount role in promoting T cell survival, expansion, acquisition of effector function, and long-term memory (8,32–35). The importance of this pathway for cancer immunotherapy has already been demonstrated by incorporation of 4–1BB signaling into the CAR T-cell technology (36) and the use of agonistic antibodies to the receptor as mono or combination therapies in preclinical (2,3) as well as clinical settings (24). However, a major drawback for the use of agonistic antibodies has been significant toxicity observed both in the preclinical and clinic settings (6–8,24,37). We, therefore, hypothesized that toxicity might not be an inherent feature of 4–1BB signaling, but rather an adverse effect of agonistic antibodies. Tumor-associated antigen-based subunit vaccines adjuvanted with SA-4–1BBL showed therapeutic efficacy in various preclinical cancer models without toxicity associated with agonistic 4–1BB antibodies (7,17–22,38).

The initial intent of this study was to test if streptavidin portion of SA-4–1BBL as a foreign antigen is immunogenic and assess the impact of anti-streptavidin antibodies on the therapeutic efficacy of SA-4–1BBL-adjuvanted cancer vaccines. Using the well-established TC-1 tumor model (21,22), we demonstrated that pretreatment of naïve mice with SA-4–1BBL as a single agent generated both humoral and cellular immune responses to the streptavidin. However, anti-streptavidin antibodies did not block the costimulatory function of SA-4–1BBL in an in vitro T cell proliferation assay. Vaccination of mice with preexisting high titers of anti-streptavidin antibodies with a subunit vaccine containing a synthetic peptide representing CD8⁺ T cell epitope for E7 (39) adjuvanted with SA-4–1BBL generated a therapeutic response, resulting in eradication of TC-1 tumor in all mice. These in vitro and in vivo findings demonstrating a lack of negative impact of anti-streptavidin antibodies on the costimulatory function of SA-4–1BBL are consistent with the structural design of this molecule. Streptavidin is linked to the N-terminus of the extracellular domain of murine 4–1BBL through a linker that provides flexibility and allows spatial separation of both molecules (13,22). It is, therefore, not surprising that anti-streptavidin antibodies do not interfere with the T cell costimulatory function of SA-4–1BBL.

Treatment with SA-4–1BBL as a single agent conferred protection against subsequent tumor challenge in mice. This highly novel and unexpected prophylactic efficacy was dose-dependent and effective against 3 different tumor types in the present study. The tumor preventive effect required 3 weeks to fully evolve following first SA-4–1BBL treatment and lasted for more than 8 weeks. In marked contrast, pretreatment with an agonistic 4–1BB antibody did not impact tumor growth. The immunogenicity of streptavidin in SA-4–1BBL was not responsible for the observed protective effect against tumors, as treatment with the agonistic antibody combined with streptavidin also failed to generate a tumor-preventive response. Importantly, SA-4–1BBL and the agonistic 3H3 antibody had opposite effects on the generation of streptavidin antibodies; SA-4–1BBL generating high titers, while the agonistic 3H3 antibody blocking such a response. This observation is consistent with a previous study reporting 4–1BB agonistic antibodies inducing anergy in CD4⁺ T cells that resulted in the blockade of humoral responses (25). However, lack of a humoral response to streptavidin was not responsible for the inability of agonistic antibody in protecting mice against tumor challenge. Passive transfer of serum with high titers of streptavidin antibodies did not prevent naïve mice against tumor challenge. Consistent with the lack of a humoral response, B cell depletion did not negate the prophylactic efficacy of SA-4–1BBL against tumor challenge.

We demonstrated that a subpopulation of CD4⁺ T cells in naïve mice express both CD44 memory marker and 4–1BB receptor on their surface and respond to SA-4–1BBL treatment by significant expansion. The presence of “memory-like” CD4⁺ T cells have previously been reported by others (40,41). Although, the exact nature of these cells remains to be fully elucidated, they may represent cells that had responded to pathogenic/environmental antigens or activated due to physiological homeostatic proliferation. Consistent with a previously published study (42) demonstrating that agonistic antibodies to 4–1BB deliver antigen-independent growth signal in T cells having memory-like phenotype in naïve mice, the agonistic 3H3 antibody used in our study also expanded memory-like T cells. However, SA-4–1BBL differed from the agonistic antibody by significantly increasing the number of CD4⁺ memory T cells and NK cells producing IFN-γ. Treatment with SA-4–1BBL also increased the frequency of CD4⁺CD44⁺ T cells expressing IL-2 as compared with naïve, significant, and 3H3 treated mice, trending towards significance.

Cross-communication between CD4⁺ T cells and NK cells has been reported in various infection and tumor models (43–45) and orchestrated by IL-2 and IFN-γ. IL-2 produced by antigen-activated CD4⁺ T cells plays an important role in the activation, expansion, and production of cytokines, particularly IFN-γ by NK cells (46). Once IFN-γ is produced, it drives T cell responses towards a Th1 response (47), which is critical for tumor eradication (17). Consistent with these previously published studies, both CD4⁺ T and NK cells as well as IFN-γ were required for SA-4–1BBL-generated cancer preventive effect. Depletion of either cell population at the priming or effector phases or blockade of IFN-γ in vivo resulted in the abrogation of SA-4–1BBL conferred prophylactic effect against tumor. The importance of CD4⁺ T and NK cell interplay in immune responses against infections and cancer has previously been reported. Antigen-primed T cells were shown to play a requisite role for the activation of NK cells and production of IFN-γ in a Leishmania major infection preclinical model, primarily through the secretion of IL-2 (45). Collaboration between CD4⁺ T cells and NK cells was also shown in a B16 melanoma preclinical model lacking CD8⁺ T cells (48) and in patients with HIV-1 viral infection that compromises CD4⁺ T cell number and IL-2 production, resulting in NK cell anergy and irresponsiveness to infection (49). Immunization with an HIV-1 subunit vaccine resulted in increased IL-2 production by antigen-specific CD4⁺ T cells and IFN-γ by NK cells. Depletion of CD8⁺ T cells, which we have shown to play a critical role in SA-4–1BBL-adjuvated subunit vaccines (19,21,22), had no impact on the preventive effect of SA-4–1BBL, which is consistent with the lack of sustained immune memory in this model.

In addition to its unexpected preventive effect, monotherapy with SA-4–1BBL was effective in preventing post-surgical tumor recurrences. It has been shown that although spontaneous T cell responses are inherently generated with tumor growth, these cells are not truly functional and effective (50). SA-4–1BBL treatment post-surgical treatment could be targeting these T cells that are activated against tumor antigens and express high levels of the 4–1BB receptor. Indeed, 4–1BB is used as a bona fide marker to sort tumor-specific T cells for ex vivo expansion and adoptive cell therapy (51,52). Thus, the engagement of SA-4–1BBL as monotherapy with its receptor 4–1BB serves as a convenient and effective way of expanding T cells that are primed by tumor neoantigens, leading to the acquisition of effector functions and controlling recurrences. This notion is consistent with our observations that SA-4–1BBL-mediated control of tumor recurrences was associated with long-term memory. Our findings are also consistent with a previously published report demonstrating that agonistic 4–1BB antibodies protect mice against post-surgical tumor challenge by expanding tumor-primed CD8⁺ T cells (53).

In conclusion, our data demonstrate unique and unexpected immunomodulatory features of SA-4–1BBL that bridge innate and adaptive immune responses with both preventive as well as therapeutic efficacy against cancer. The tumor type-independent protection conferred by SA-4–1BBL is significant with important clinical implications for primary and secondary cancer prevention modalities. This agent may have utility for the treatment of individuals at high risk for tumorigeneses, such as patients with chronic liver diseases, individuals with hereditary mutations in p53 or breast cancer (BRCA1 and BRCA2) genes, and those affected by specific cancer risks, such as preneoplastic/early neoplastic lesions. The safety profile of SA-4–1BBL and its cancer immunoprevention attributes, both qualities not shared by agonistic antibodies to the 4–1BB receptor, highlight its potential for cancer immunoprevention and therapy.

Significance Statement.

This study demonstrates the unique and unexpected immunomodulatory features of SA-4–1BBL that bridge innate and adaptive immune responses with both preventive and therapeutic efficacy against cancer.