Intratumoral Activation of 41BB Costimulatory Signals Enhances CD8 T Cell Expansion and Modulates Tumor-Infiltrating Myeloid Cells

This article is featured in Top Reads, p.2555

Activation of 41BB (CD137) costimulatory signals is an effective means to enhance the expansion and function of tumor-infiltrating lymphocytes (TILs) from primary tumor fragments for the purpose of preparing cells for adoptive cell therapy (ACT) (1). Recently, it has been identified that the direct injection of 41BB agonistic Abs into tumors can mount potent immune responses against local and distant untreated tumors (2). Moreover, strategies to engineer therapeutics that selectively activate 41BB within tumors have demonstrated feasibility in mouse models, providing support for advancement of these therapies to clinical trials (3–5). Overall, the targeting of 41BB within tumors can effectively increase T cell proliferation and promote the eradication of tumor cells both in vitro and in vivo.

41BB is widely known as a costimulatory molecule expressed by T cells; however, nearly all subsets of immune cells express 41BB (6). Previous studies identified that 41BB and 41BB ligand (41BBL) play critical yet context-dependent roles in myeloid cell development and function (7, 8). In particular, the knockout of 41BB promotes the accumulation of myeloid cells under steady-state conditions, whereas triggering 41BB activation in dendritic cells (DCs) enhanced their capacity to stimulate T cells in vitro (7, 9). Although accumulating evidence in mice has suggested that both 41BB and 41BBL are critical for directly regulating the function of myeloid cells, little is known about how this receptor–ligand axis potentiates myeloid-mediated antitumor immune responses in humans. Given the importance of the inflammatory context in 41BB–41BBL signaling, a deeper understanding of 41BB–41BBL signaling in human myeloid cells, particularly in the context of tumor-mediated inflammation, is needed (7, 10).

In human biological systems, 41BBL acts as a maturation factor for monocytes, promoting the expression of costimulatory molecules and cytokines, including IL-12, IL-6, IL-8, TNF, and M-CSF (11). The stimulation of 41BBL with 41BB protein induces reverse signaling in monocytes, triggering their maturation into DCs (12). Although 41BB–41BBL bidirectional signaling between T cells and APCs has been shown to promote effector immune responses, it remains unclear how the context of inflammation within human tumors influences this process. At our institution, treatment of melanoma patients using ACT with TIL has resulted in a 38% overall response rate (13, 14). Moreover, 41BB agonists are currently being explored for the ability to enhance TIL expansion for the use ACT (NCT02652455). Hence, the development of therapeutics that exploit immunologic mechanisms to boost ex vivo TIL expansion can greatly benefit from an enhanced understanding of how a supportive immune microenvironment promotes antitumor immune functions. The work outlined in this study highlights the importance of triggering costimulatory signals on T cells and how augmenting the interactions of 41BB–41BBL bidirectional signals provided by APCs ultimately provides support for the improvement of TIL expansion from primary tumor fragments and the promotion of antitumor immune responses in vivo.

Preparation of TIL was performed as previously described (13). Briefly, surgically resected tumors were minced to 1-mm fragments and placed into individual wells of a 24-well plate containing 6000 IU/ml IL-2 (aldesleukin; Prometheus Laboratories). TILs were expanded for up to 5 wk and then tested for IFN-γ production in cocultures with autologous tumor cell lines or cryopreserved tumor digest cell suspensions. IFN-γ⁺ TILs underwent a rapid expansion protocol and were then cryopreserved in 90% human serum with 10% DMSO. A fully human IgG4 monoclonal agonistic anti-41BB Ab (anti-41BB mAb; BMS-663513, urelumab) was a kind gift from Bristol Myers Squibb. The anti-41BB was added with media containing 6000 IU/ml IL-2 to tumor fragments at the initiation of TIL expansion. Thereafter, TILs were fed every 3–4 d with media containing 6000 IU/ml IL-2 only. Cyropreserved TILs were thawed and rested in media containing 3000 IU/ml IL-2 for 3–4 d before being subjected to further stimulation and coculture conditions. For the preparation of tumor digests, the remaining tumor tissue was suspended in digestion media containing collagenase type II and type IV, hyaluronidase, and DNase I (all from Thermo Fisher Scientific) and then subjected to gentleMACS dissociation (Miltenyi Biotec). Tumor digest cell suspensions were incubated at 37°C in a rocking water bath for 1 h and then filtered with 100-μm cell strainers to remove large cellular debris.

PBMCs were obtained from the whole blood of healthy donor volunteers or apheresis product from melanoma patients. For whole blood, PBMCs were prepared by Ficoll-Pacque (GE Healthcare) and then cryopreserved. Thawed PBMCs were labeled with CD11b Microbeads human and mouse (130-49-601; Miltenyi Biotec) for magnetic-activated cell sorting. Purity of CD11b⁺ cells was >90%. For healthy donor myeloid cells, 2.5 × 10⁵ cells were cultured in six-well plates coated with 10 μg/ml anti-41BB or 10 μg/ml 41BB–Fc (R&D Systems) in media alone or media containing 100 ng/ml human GM-CSF (PeproTech) for 3 d. For the generation of APCs, CD11b⁺ cells from the pheresis product of a melanoma patient were cultured with 10 μg/ml immobilized 41BB–Fc for 3 d. Adherent cells were dissociated using ACCUTASE (STEMCELL Technologies) and gentle pipetting. Cell lysate from an autologous melanoma cell line was prepared by suspending cells at 30 × 10⁶/ml in PBS and exposure to repeated and alternating temperatures (solid CO₂ ice and 37°C water). Five cycles of alternating temperature exposure at 5 min intervals were performed. Three cell equivalents of tumor cell lysate were added to 1 × 10⁶ 41BB–Fc–conditioned APCs in media containing 100 ng/ml GM-CSF and incubated overnight. Tumor lysate-pulsed APCs were then dissociated using ACCUTASE and gentle pipetting before coculture with TILs, as described below. Cytokines in supernatants were measured by LEGENDplex HU Essential Immune Response Panel (BioLegend) and acquired via BD FACSCelesta.

Thawed autologous tumor cell suspensions were added to 96-well round-bottom plates at a 1:1 ratio with TILs and cultured for 24 h. Anti-human HLA-A,B,C (W6/32; BioLegend) was added to tumor cells at a concentration of 10 μg/ml and incubated for 1 h at 37°C before adding TILs to the respective wells. Anti-human CD3 (OKT3; Ortho Biotech, Bridgewater, NJ) was immobilized on the bottom of the wells at 5 μg/ml. Supernatants were collected after 24 h of TIL–tumor digest cocultures, and IFN-γ was measured in supernatants via IFN-γ ELISA (BD Biosciences). For TIL cocultures with autologous tumor cell lines or autologous tumor lysate-pulsed APCs, supernatants were collected after 72 h of cultures. Prior to coculture, autologous tumor cell lines were subjected to x-radiation at a dose of 2 × 10⁴ rad. TILs were cocultured at a 10:1 ratio with irradiated tumor cells or tumor lysate-pulsed APCs. Where indicated, soluble 10 μg/ml anti-41BB or anti-41BBL (5F4; BioLegend) was added to TIL cocultures. Cytokines in supernatants were measured by LEGENDplex HU Essential Immune Response Panel (BioLegend) and acquired via BD FACSCelesta. Proliferation was measured by ³H thymidine uptake (1 μCi added per well) during the final 18 h of coculture.

Tumor digest cell suspensions were prepared as described above. One million cells were seeded in 48-well plates containing media with or without 6000 IU/ml IL-2 in combination with soluble anti-41BB (10 μg/ml). After 24 h, cell-free supernatants were collected and stored at −80°C until ready for analysis. Cytokines in supernatants were measured by LEGENDplex HU Essential Immune Response Panel (BioLegend) and acquired via BD FACSCelesta.

Mouse spleens and tumors were harvested under sterile conditions. Spleens were homogenized by applying pressure to tissue on 100-μm cell strainers. Single-cell suspensions were prepared, and RBCs were removed using RBC Lysis Buffer (BioLegend). The resulting suspension was passed through a 70-μm cell strainer and washed once with PBS. Mouse tumor cell suspensions were prepared by enzymatic digestion with media (HBSS; Life Technologies) containing 1 mg/ml collagenase IV, 0.1 mg/ml DNaseI, and 2.5 U/ml hyaluronidase (all from Sigma-Aldrich) and then subjected to GentleMACS dissociation (Miltenyi Biotec). Tumor digest cell suspensions were incubated at 37°C in a rocking water bath for 1 h. RBCs were removed using RBC Lysis Buffer (BioLegend), then cell suspensions were filtered with 100-μm cell strainers to remove large cellular debris. Cells were resuspended into a concentration of 0.5–1 × 10⁶ cells/ml for flow cytometric analysis in FACS buffer containing PBS, 5% FBS, 1 mM EDTA (Sigma-Aldrich), and 0.1% sodium azide (Sigma-Aldrich). Cell viability was measured by staining cell suspensions with Zombie NIR (BioLegend). Prior to surface staining, cells were incubated with Fc Shield (Tonbo Biosciences) for murine specimens and FcR Blocking Reagent (Miltenyi Biotec) for human specimens. For surface staining of murine specimens, cells were stained in FACS buffer with the following Abs: CD3 (145-2C11), CD4 (GK1.5), CD8 (53-6.7), CD11b (M1/70), Ly-6G (1A8), Ly-6C (HK1.4), F4/80 (BM8), CD11c (N418), MHC class II (MHCII) (M5/114.15.2), CD80 (16-10A1), CD86 (GL-1), and PD-1 (29F.1A12) (all from BioLegend); and 41BB (17B5-1H1; Miltenyi Biotec). For intracellular cytokine detection, cells were incubated for 18 h with 1× brefeldin A (BioLegend), stained with cell surface Abs, subjected to fixation and permeabilization via Fixation/Permeabilization Solution Kit (BD Biosciences), then stained with anti-mouse Abs against IL-12p40/p70 (BD Biosciences), IL-6 (MP5-20F3), IL-10 (JES5-16E3), and TNF-α (MP6-XT22) (all from BioLegend). For human specimens, cell surface staining was conducted with the following Abs: CD3 (145-2C11), CD4 (RPA-T4), CD8 CD11c (Bly6), CD14 (MoP9), CD15 (HI98), CD11b (ICRF44), HLA-DR/DP/DQ (Tu39), and CD86 (all from BD Biosciences); PD-L1 (29E-2A3), 41BBL (5F4), 41BB (4B4-1), and CD45 (2D1) (from BioLegend); and CD141 (AD5-14H12) (Miltenyi Biotec). Cells were acquired by FACSCelesta (BD Biosciences), and the data were analyzed with FlowJo (Tree Star).

Female C57BL/6 mice (6–8 wk old) were purchased from Charles River Laboratories (Wilmington, MA). OT-I mice (originally obtained from The Jackson Laboratory) were bred and housed at the Animal Research Facility of the H. Lee Moffitt Cancer Center and Research Institute. Mice were humanely euthanized by CO₂ inhalation and secondary cervical dislocation according to the American Veterinary Medical Association Guidelines. Mice were observed daily and were humanely euthanized if a solitary s.c. tumor exceeded 400 cm² in area or mice showed signs referable to metastatic cancer.

B16 melanoma, Panc02 pancreatic cancer, and MC38 colorectal cancer cell lines (all obtained from American Type Culture Collection), were cultured in complete media: RPMI 1640 media supplemented with 10% heat-inactivated FBS, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 2 mM fresh l-glutamine, 100 mg/ml streptomycin, 100 U/ml penicillin, 50 mg/ml gentamicin, 0.5 mg/ml fungizone (all from Life Technologies, Rockville, MD), and 0.05 mM 2-ME (Sigma-Aldrich, St. Louis, MO). B16 melanoma with pAc-neo-OVA plasmid (B16-OVA) was maintained in media with 0.8 mg/ml G418, as previously described (15). To generate the OVA-expressing fluorescent Panc02 cell line, cells were exposed to supernatants containing a lentiviral vector comprised of a fluorescent ZsGreen (ZsG) protein and OVA. Upon successful transfection, ZsG^hi tumor cells were subjected to FACS using BD FACSAria. OVA–ZsG^hi tumor cells were passaged in vitro four times, whereby OVA expression was validated by staining for H2-K^b bound to SIINFEKL peptide (25-D1.16; BioLegend). The cell lines tested negative for mycoplasma contamination. All cell lines were passaged <10 times after initial revival from frozen stocks. All cell lines were validated in core facilities prior to use. Tumor cells (1 × 10⁵) were implanted s.c. in the flank of mice. When tumors reached ∼25 mm², 75 μg of InVivoPlus anti-mouse 4-1BB (clone LOB12.3) or rat IgG1 isotype control, anti-HRP (both from Bio X Cell), were injected in 50-μl volume intratumorally (i.t.). Injections were repeated twice weekly until experimental end point. In some experiments, anti-mouse 41BB (clone LOB12.3) or rat IgG1 isotype control, anti-HRP were injected with 300 μg of Ab twice weekly until experimental end point. For CD8 T cell depletion, 300 μg of InVivoPlus anti-mouse CD8α (Bio X Cell) were injected i.p. twice weekly for the duration of the experiment. CD8 T cell depletion was initiated prior to treatment with isotype or anti-41BB Abs.

Myeloid cells were isolated from MC38 tumors after treatment with isotype or anti-41BB Abs using an EasySep Mouse CD11b Positive Selection Kit II (STEMCELL Technologies). CD8 T cells were isolated from the spleens of OT-I mice using EasySep Mouse CD8 T cell Isolation Kit (STEMCELL Technologies). OT-I T cells were labeled with CellTrace Violet (Invitrogen) prior to coculture. OT-I T cells were cocultured with myeloid cells in media containing 1 μg/ml OVA_(257–264) peptide with or without neutralizing Abs for IL-10 (JES5-2A5) or IL-12-p75 (R2-9A5) (both from Bio X Cell) at a concentration of 10 μg/ml each. Cells and supernatants were harvested after 72 h incubation. IFN-γ in supernatants was measured with a Quantikine Mouse IFN-gamma ELISA Kit (R&D Systems).

TILs from mice with MC38 tumors were isolated after treatment with isotype or anti-41BB Abs using an EasySep Mouse CD90.2 Positive Selection Kit II (STEMCELL Technologies). TILs were cultured in round-bottom 96-well plates with immobilized anti-CD3 Abs (145-2C11; BD Biosciences) at a concentration of 5 μg/ml or with irradiated tumor cell lines. MC38 or irrelevant target B16 tumor cells were exposed to x-radiation at a dose of 2 × 10⁴ rad and cultured with CD90.2⁺ TILs at a 1:10 (target/TIL) ratio for 48 h. Supernatants were collected and IFN-γ was measured by a Quantikine Mouse IFN-gamma ELISA Kit (R&D Systems).

Graphs were generated using GraphPad Prism software. Graphs represent mean values with SEM. The p values were calculated in each respective figure, in which statistical tests were indicated. For mouse tumor growth studies, tumor growth curves are shown as mean with SEM, and significance was determined by two-way ANOVA and Sidak multiple comparison’s test. Mice were randomized after tumor cell implantation into respective treatment groups. Tumors were measured with vernier calipers. Experimental groups were blinded to the operator throughout the duration of the experiment. For all other experiments, data were compared using either an unpaired two-tailed Student t test corrected for multiple comparisons by a Bonferroni adjustment or Welch correction. The p values are as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Studies were performed under approved Institutional Review Board laboratory protocols at the H. Lee Moffitt Cancer Center (Tampa, FL). TIL, PBMC, and autologous tumors were collected from melanoma patients or PBMC from lung tumor patients as part of TIL ACT clinical trials. All samples were de-identified prior to use in research studies. All patients signed approved consent forms. All animal experiments were approved by the University of South Florida Institutional Animal Care and Use Committee and performed in accordance with the U.S. Public Health Service policy and National Research Council guidelines.

To determine the efficacy of 41BB agonists, we validated that systemic treatment via i.p. administration led to tumor growth delay and complete regressions in mice with established MC38 tumors (Fig. 1A, 1C). We observed that i.t. administration with anti-41BB (anti-41BB) Abs exhibited comparable efficacy to the i.p. route of administration, through which the tumor growth kinetics and the frequency of complete regressions were similar (Fig. 1B, 1D). We validated these results in two additional tumor models bearing the model OVA Ag, Panc02–ZsGOVA (Fig. 1E, 1G) and B16-OVA (Fig. 1F, 1H). The i.t. treatment of anti-41BB led to significant growth delay and tumor regression in mice with Panc02–ZsGOVA tumors (Fig. 1E) and B16-OVA tumors (Fig. 1F) compared with control mice that received i.t. isotype Abs. Moreover, the survival of mice treated with i.t. anti-41BB was significantly enhanced in both models (Fig. 1G, 1H). In mice with MC38 tumors, i.t. anti-41BB led to complete regressions in ∼30% of treated mice (Fig. 1D). Likewise, 70% of mice with Panc02–ZsGOVA tumors (Fig. 1G) or B16-OVA tumors (Fig. 1H) had no evidence of tumor growth after tumor inoculation (44 and 70 d, respectively). This suggested that the presence of a highly immunogenic Ag, such as OVA, could potently direct local antitumor immune responses after i.t. treatment with anti-41BB. Together, these data demonstrate that the i.t. treatment with 41BB agonistic Abs is a feasible approach to induce tumor regression in mice.

We determined that within 1 wk after the initial treatment with i.t. anti-41BB, the size of tumors was significantly reduced in mice with MC38 tumors (Fig. 2A) and Panc02–ZsGOVA tumors (Fig. 2B). The reduction in tumor size in response to i.t. anti-41BB treatment was associated with an increase of CD8 T cell infiltration in both tumor models compared with mice treated with isotype Abs (Fig. 2C, 2D). However, the frequency of CD4 TILs was unchanged between mice treated with isotype or anti-41BB (Fig. 2C, 2D). We next determined that the increase of CD8 T cells in MC38 tumors was required for the antitumor efficacy because the depletion of CD8 T cells prior to the start of i.t. anti-41BB treatment abrogated the reduction of tumor growth (Fig. 2E). In contrast, the depletion of CD8 T cells had no effect in mice that received i.t. treatment with isotype Abs, indicating that basal antitumor CD8 T cell responses are ineffective against MC38 tumors (Fig. 2E). Not only was the presence of CD8 T cells necessary for the reduction of tumor growth, we found that TILs isolated from anti-41BB-treated tumors exhibited higher IFN-γ production in response to CD3 stimulation or coculture with irradiated MC38 tumor cells (Fig. 2F). Conversely, TILs from isotype-treated tumors were successfully stimulated with CD3 Abs but failed to produce IFN-γ in cultures with MC38 tumor cells (Fig. 2F). These results demonstrate that i.t. 41BB agonism can rejuvenate CD8 T cell responses, leading to an improvement of antitumor immune responses.

We demonstrated that the triggering of 41BB costimulation via i.t. treatment with anti-41BB converted tumors to a more T cell inflamed environment (Fig. 2). Indeed, the ratio of CD11b⁺ myeloid cells relative to CD8⁺ T cells was significantly reduced in tumors that received anti-41BB treatment compared with mice that received control isotype Abs (Fig. 3A). In contrast to control tumors, in which the majority of CD45⁺ leukocytes were CD11b⁺MHCII⁺F480⁻CD11c⁻ myeloid cells and CD11b⁺F480⁺Ly-6C⁻ tumor-associated macrophages (TAMs), we found that MC38 tumors treated with i.t. anti-41BB had a dramatic reduction in these myeloid cell populations (Fig. 3B). Likewise, anti-41BB treatment decreased the frequency of CD11b⁺MHCII⁺F480⁻CD11c⁺ cells (DCs). In isotype-treated tumors, CD11b⁺F480⁻Ly-6C⁺Ly-6G⁻ monocytes and CD11b⁺F480⁻Ly-6C⁺Ly-6G⁺ polymorphonuclear cells (PMNs), presumably monocytic- and PMN myeloid-derived suppressor cells, comprised <10% of CD45⁺ cells. Conversely, the reduction of TAMs and other myeloid cell populations coincided with significant increases of monocytes and PMNs in tumors that received treatment with anti-41BB (Fig. 3B). Nevertheless, the changes in myeloid cell frequency in response to anti-41BB treatment were associated with an increased abundance of CD80⁺CD86⁺ DCs, TAMs, and monocytes (Fig. 3C). Furthermore, DCs, TAMs, monocytes, and PMNs significantly upregulated CD80 and/or CD86 in tumors treated with anti-41BB compared with isotype controls (Fig. 3D–G). Despite the loss of classical APCs within the tumor microenvironment, the increase in CD80 and CD86 among all tumor-associated myeloid cells may support antitumor T cell responses after i.t. administration of 41BB agonists.

We further evaluated changes to i.t. myeloid cells by examining the production of cytokines after i.t. anti-41BB. We harvested tumors from isotype and anti-41BB-treated mice 7 d after treatment initiation and cultured the cells overnight. We found that myeloid cells from tumors produced IL-6 and TNF-α, but no difference was observed between cells from anti-41BB or isotype tumors (Fig. 4A, 4D). In contrast, DCs and TAMs had an elevated expression of IL-10, whereas CD11b⁺Gr-1⁺ cells exhibited a reduced production of IL-10 (Fig. 4B). Similarly, DCs and TAMs exhibited an increased production of IL-12 (Fig. 4C). Although we observed changes in cytokine expression among myeloid cell subsets, the frequency of DCs and TAMs was significantly reduced by anti-41BB treatment (Fig. 3B). Consistent with these data, the number of cytokine-producing DCs and TAMs were largely reduced in cultured tumor cell digests of anti-41BB-treated tumors in comparison with isotype-treated tumors (Fig. 4E–H). Concordantly, the number of cytokine-producing monocytes and PMNs (CD11b⁺Gr-1⁺ cells) was significantly elevated in anti-41BB-treated tumors (Fig. 4E–H). Because IL-10 and IL-12 are key regulators in T cell priming and activation by myeloid cells, we next evaluated the capacity of i.t. myeloid cells to stimulate T cells after treatment. We found that OT-I T cells produced more IFN-γ in cocultures with myeloid cells from anti-41BB-treated tumors, both with and without in vitro IL-10 neutralization (Fig. 4J). Upon examination of the phenotype of OT-I T cells after coculture, we observed that T cells cocultured with myeloid cells from anti-41BB-treated tumors had an elevated expression of cell surface 41BB compared with T cells cultured with myeloid cells from isotype-treated tumors (Fig. 4K–M). Moreover, the in vitro neutralization of IL-10 but not IL-12 enhanced the expression of 41BB in OT-I T cells from cocultures with tumor myeloid cells, suggesting that myeloid cell–derived IL-10 restricted the expression of 41BB (Fig. 4K–M). Similar to the increase of 41BB expression, IL-10 neutralization resulted in an increased expression of PD-1 in cocultures with myeloid cells from isotype tumors (Supplemental Fig. 1). Thus, i.t. anti-41BB treatment enhances the ability of tumor myeloid cells to potentiate T cell responses.

We have previously shown that the addition of 41BB agonistic Abs enhances the expansion and function of melanoma TILs from primary tumor fragments in vitro (16). We obtained two melanoma specimens and attempted to expand TILs from tumor fragments placed in media containing IL-2 or IL-2 in combination with anti-41BB. In patient 1, TILs expanded in cultures with IL-2 plus anti-41BB (5/6 fragments), whereas no expansion occurred in cultures with IL-2 only (0/6 fragments) (Fig. 5A). Similarly, an enhancement of TIL expansion was observed in patient 2 among fragments grown with IL-2 and anti-41BB (23/24 fragments) compared with TILs grown in IL-2 alone (12/24 fragments). Moreover, the number of TIL expanded per fragment was greater in cultures containing IL-2 and anti-41BB compared with IL-2 alone conditions (Fig. 5B). Among fragments grown in IL-2 only, the distribution of CD4⁺ TILs and CD8⁺ TILs was approximately equal. In contrast, the combination of IL-2 plus anti-41BB almost exclusively promoted the expansion of CD8⁺ TILs (Fig. 5C). Because the yield of TILs from IL-2–alone cultures was relatively low, we pooled together TILs from fragments that exhibited the best expansion at the end of the culture. Similarly, we chose TILs expanded from six individual fragments grown in IL-2 plus anti-41BB that yielded the highest number of cells. We then cocultured the selected TILs with autologous tumor digest and determined the magnitude of IFN-γ production. IFN-γ was detected and effectively blocked with Cblocking Abs in TILs grown in IL-2 alone and in three of six selected TILs grown in IL-2 and anti-41BB (Fig. 5D). Notably, the abundance of IFN-γ was higher in cocultures with TILs expanded with IL-2 and anti-41BB compared with TILs grown in IL-2 only (Fig. 5D).

To better understand the contribution of myeloid cells in the process of ex vivo TIL expansion, we evaluated the frequency of leukocyte populations in a fresh tumor sample from melanoma patient 1. We found that 17.1% of all live cells were CD45⁺, of which 73% of CD45⁺ cells consisted of CD11b⁺CD11c⁺CD14⁺HLA-DR⁺ myeloid cells. Approximately 15% of CD45⁺ cells were CD4⁺ and CD8⁺ T cells, and the remainder consisted of a variety of myeloid cell subsets (Fig. 5E). We next evaluated the production of cytokines in fresh tumor digests from patient 1 and patient 2. Fresh tumor cell digests were cultured overnight in IL-2 alone or in combination with anti-41BB. Tumors produced vast amounts of CCL2, IL-6, IL-8, IL-1β, IL-10, and TGF-β. In response to IL-2 and IL-2 in combination with anti-41BB, we detected an increased production of CXCL10 and IFN-γ in comparison with unstimulated tumor digests (Fig. 5F, 5G). Moreover, a trend consistent with an increase of CXCL10 and IFN-γ were observed when tumors were cultured with IL-2 and anti-41BB compared with IL-2 alone (Fig. 5H). Collectively, these data demonstrate that the augmentation of ex vivo TIL expansion via IL-2 and 41BB stimulation is associated with increases in proinflammatory cytokine production.

Next, we dissected how myeloid cells facilitate ex vivo TIL expansion via 41BB stimulation. First, we observed that CD11b⁺ myeloid cells within a fresh melanoma tumor lacked the expression of 41BB (Fig. 6A). Next, we phenotyped peripheral blood myeloid cells and found that CD11b⁺ cells in PBMCs expressed HLA-DR/DP/DQ, CD14, CD86, CD11c, and low levels of PD-L1 and CD141 (Fig. 6B). Moreover, these cells lacked expression of 41BB but did express 41BBL (Fig. 6B, 6C). Because the activation of 41BBL in monocytes is known to promote the maturation to DCs (17, 18) and 41BB expression was poorly expressed by myeloid cells, we examined how the stimulation of 41BBL could differ in activating myeloid cells in comparison with a 41BB agonistic Ab. We stimulated myeloid cells with immobilized anti-41BB to agonize 41BB or immobilized 41BB protein (41BB–Fc) to agonize 41BBL. The viability of donor myeloid cells was greatly reduced in unstimulated cultures or under stimulation with anti-41BB alone. In contrast, stimulation with 41BB–Fc alone maintained cell viability similar to that of cultures containing GM-CSF (positive control) or GM-CSF in combination with anti-41BB or 41BB–Fc (Fig. 6D). This suggested that activation of 41BBL but not 41BB was sufficient to maintain the survival of myeloid cells and that 41BB(L) signals did not augment cell viability in the presence of GM-CSF. Compared with precultured cells, myeloid cells upregulated 41BBL expression but not 41BB expression when incubated with a GM-CSF maturation stimuli (Fig. 6E). In addition to maintaining myeloid cell viability, the stimulation with 41BB–Fc reduced the expression of CD14, whereas it enhanced the expression of PD-L1, CD141, 41BBL, and CD86 compared with preculture myeloid cells (Fig. 6F). Likewise, GM-CSF–stimulated cells exhibited similar phenotypic changes to 41BB–Fc–treated cells; however, GM-CSF failed to upregulate CD86 (Fig. 6G). The increase in CD86 expression was highly consistent between all donors (Fig. 6H). In contrast to 41BB–Fc or GM-CSF stimulation, anti-41BB alone failed to maintain myeloid viability (Fig. 6D), and the phenotype was similar to myeloid cells cultured in media alone (Fig. 6F). Likewise, the addition of anti-41BB or 41BB–Fc failed to augment cell surface marker expression when combined with GM-CSF (Supplemental Fig. 2). Although increases in CD86 and 41BBL expression are indicative of an enhanced costimulatory capacity, we further interrogated the impact of 41BB–41BBL activation in myeloid cells by assessing cytokine production. We found that IL-4 and CCL2 expression was potently induced by GM-CSF and to a lesser extent by anti-41BB and 41BB–Fc (Fig. 6I, 6J). Similarly, IL-1β, IL-6, and IL-8 production were amplified by stimulation with GM-CSF or 41BB–Fc but not anti-41BB (Fig. 6K–M). CXCL10 expression was reduced by 41BB–Fc stimulation but was maintained with GM-CSF or anti-41BB alone (Fig. 6N). In addition, myeloid cells readily produced IL-2 and TGF-β, but the culture conditions maintained or modestly increased the production of these cytokines (Fig. 6O, 6P). In large part, 41BB activation alone via anti-41BB or its addition to GM-CSF stimulus had little effect on cell viability, the expression of cell surface markers, or induction of cytokine expression (Fig. 6D–P, Supplemental Fig. 2). Hence, it is possible that 41BB agonism on myeloid cells provides a weaker stimulus in comparison with reverse signaling through 41BBL.

The agonistic 41BB Ab urelumab does not compete with the binding of 41BBL to 41BB, thereby preserving native 41BBL-mediated costimulation (19). Consistent with our data in Fig. 6, the maturation of monocytes via 41BBL reverse signaling leads to an increased potential to prime T cells characterized by increased CD86 expression and enhanced production of IL-6 and IL-8 (11). Hence, we hypothesized that the enhancement of ex vivo TIL expansion by 41BB agonists may be aided by additional costimulation mediated by myeloid 41BBL. To determine this, we generated APCs from CD11b⁺ cells isolated from the pheresis product of a melanoma patient. The myeloid cells were incubated with immobilized 41BB–Fc for 3 d, collected, then pulsed with autologous tumor lysate for 24 h in the presence of GM-CSF to generate 41BBL-conditioned APCs (41BBL APCs). First, we examined whether anti-41BB could augment the production of cytokines in TIL cocultures with autologous tumor or pulsed APCs. A variety of cytokines were detected in cultures containing TILs stimulated with anti-CD3 or cocultures of TILs with autologous tumor cells. In particular, the combined stimulation of TILs with anti-CD3 and anti-41BB increased the production CXCL10 and IFN-γ (Fig. 7A). As expected, IFN-γ and TNF-α were induced in TIL cocultures with tumor cells, which were effectively reduced by MHC class I (MHC-I) blockade (Fig. 7B). Next, we determined the cytokine profile in TIL cocultures with autologous APCs. When cultured alone, the 41BBL APCs produced high amounts of CCL2 and IL-8 and modest amounts of IL-2, TGF-β, and IL-1β, which were not impacted by additional stimulation with soluble anti-41BB (Fig. 7C). In comparison with unstimulated TILs and cultures with 41BBL APCs only, CXCL10, IL-2, IFN-γ, IL-1β, TGF-β, and IL-6 were elevated when TILs were cultured with APCs, in which the production of some cytokines was augmented by the addition of agonistic anti-41BB and/or anti-41BBL blocking Abs (anti-41BBL) (Fig. 7D). CXCL10 was only detected in TIL–APC cocultures, which suggested that cell–cell contact facilitated the production of CXCL10. Accordingly, we found that the production of CXCL10 and IL-2 was only augmented by 41BBL blockade, which was enhanced when 41BBL blocking was combined with anti-41BB (Fig. 7E). Similarly, we observed that CXCL10 was elevated in Panc02–ZsGOVA tumors taken from mice treated with i.t. anti-41BB (Supplemental Fig. 3). In human TIL–APC cocultures, the addition of anti-41BB alone, 41BBL blockade alone, and/or the combination increased IFN-γ, IL-1β, and TGF-β. Likewise, IL-6 was absent in all conditions except in TIL–APC cocultures with the addition of anti-41BB in combination with 41BBL blockade (Fig. 7E). Next, we demonstrated that TILs from this patient readily proliferated in cocultures with anti-CD3 or irradiated autologous tumor cells compared with basal proliferation. The combination of anti-41BB with anti-CD3 enhanced TIL proliferation but was reduced when 41BBL Abs were present in culture. However, the addition of anti-41BB and/or anti-41BBL did not alter TIL proliferation in cultures with tumor cells. In parallel, we cocultured TILs with autologous 41BBL APCs pulsed with tumor lysate in combination with soluble anti-41BBL and/or anti-41BB. 41BBL APCs induced the proliferation of TILs, which was negatively impacted by 41BBL blockade alone or in combination with anti-41BB, indicating that the blockade of 41BBL dampened TIL proliferation and that additional costimulation with anti-41BB was not sufficient to reverse this effect. Together, these results demonstrate that myeloid 41BBL can contribute to the effect of 41BB agonists characterized by enhanced TIL proliferation and production of cytokines.

Overall, 41BB agonists are potent immune stimulators, but their translational potential has been heavily restricted by the onset of severe adverse events (20). Significant advancement in the development of tumor-selective 41BB agonists to limit or even eliminate any 41BB-related adverse events has revitalized the therapeutic feasibility of targeting 41BB in humans (4, 5). Although we did not evaluate toxicity in mice that received i.t. injections of anti-41BB, we did not observe any overt toxicity during the administered treatment regimen. Overall, our data support that stimulating 41BB in tumors is a feasible approach to promote antitumor immune responses (Figs. 1, 2). Consistent with other reports, the activity of 41BB agonists in our hands was not independent of changes to the myeloid-tumor immune milieu (21, 22). The antitumor activity of 41BB agonists is greatly reduced in the absence of BATF3-dependent DCs, suggesting that 41BB agonists may act directly on the myeloid compartment to promote the eradication of tumors in mice (21). Indeed, we observed that treatment with 41BB agonists in mice led to increases in monocytes and PMNs coincided with a depletion of macrophages and DCs (Fig. 3A, 3B). Furthermore, myeloid cells from anti-41BB-treated tumors induced the upregulation of 41BB on CD8⁺ T cells and enhanced the production of IFN-γ (Fig. 4J–N). Hence, it is possible that myeloid cells contribute to the efficacy of 41BB activation in vivo by inducing the upregulation of surface 41BB on T cells. We show that IL-10 was essential for restricting both the production of IFN-γ and the expression 41BB on T cells (Fig. 4J, 4L). Although DCs and TAMs increased their production of IL-10 and IL-12 in response to 41BB agonism (Fig. 4B, 4C), the i.t. administration of anti-41BB simultaneously promoted the accumulation of monocytes and PMNs that produced IL-6, IL-12, IL-10, and TNF-α (Fig. 4E–H). Thus, both the proportionality and function of distinct i.t. myeloid cells are likely relevant factors in driving antitumor immune responses elicited by 41BB agonists.

It has been described that species differences exist between mouse and human myeloid cells in response to 41BBL signaling (23). Although, we did not evaluate the role of 41BBL in murine models, the data we present in human cell culture systems provide relevance for the role of 41BBL in the ability of myeloid cells to potentiate T cell responses. Importantly, the upregulation of costimulatory markers, such as CD80 and/or CD86, in both mouse and human myeloid cells was consistent after exposure to 41BB or 41BBL stimulation. This suggests that myeloid-mediated costimulation could be enhanced in the context of 41BB agonism, leading to the potentiation of TIL expansion and function (Figs. 3C–G, 6G, 6H) (16). In this study, we found that 41BB agonism, contrary to 41BBL stimulation via 41BB–Fc, had little effect in augmenting human myeloid cell phenotypes and cytokine production (Fig. 6). This, perhaps, was not surprising because CD11b⁺ cells expressed little to no 41BB on their cell surface (Fig. 6A–C). Concordantly, the stimulation of myeloid cells with a 41BB agonist alone failed to sustain myeloid cell viability, increase CD86 expression, or provide other maturation stimuli, even when combined with GM-CSF (Fig. 6D–H, Supplemental Fig. 2). Moreover, the addition of anti-41BB to tumor lysate-pulsed APCs did not significantly alter basal cytokine production (Fig. 7C). Hence, at best, the 41BB agonist used in these experiments appears to be a weak stimulator of myeloid cells. In mice, the engagement of 41BBL on myeloid cells by its cognate receptor 41BB restricts the accumulation of IL-12⁺ conventional DCs and TAMs within tumors, leading to a diminished ability to control tumor growth (22). In a contradictory manner, 41BB knockout mice exhibit remarkably similar antitumor immune responses to mice treated with agonistic 41BB Abs, which supports the hypothesis that the lack of interaction between 41BB with 41BBL on myeloid cells promotes antitumor immune responses. However, the evidence we provide in this study demonstrates that reverse 41BBL signaling can promote the immunostimulatory capacity of monocytes and APCs in humans. Consistent with previous reports in human cells (18, 24), we show that the induction of reverse 41BBL signaling in human monocytes via 41BB–Fc promoted the expression of costimulatory markers CD86 and 41BBL while simultaneously increasing the production of IL-8, IL-6, IL-1β, CCL2, and IL-4 (Fig. 6I–P). Thus, differences among species and experimental murine tumor models likely contribute to the contrasting findings in reports investigating the role of the 41BB–41BBL costimulatory axis in antitumor immunity. Together, these results demonstrate that the 41BB agonists may not act directly on human myeloid cells alone to promote the costimulatory capacity of APCs. Rather, the activation of 41BBL and potential bidirectional signaling between myeloid cells and T cells were responsible for providing efficient maturation stimuli to enhance the capacity of APCs to prime T cells.

In contrast to other 41BB agonistic Abs, urelumab facilitates the cross-linking of 41BBL to 41BB, suggesting that bidirectional signaling orchestrated by 41BBL⁺ cells could augment the agonistic activity of anti-41BB (19, 25). We provide evidence in this study that the activation of 41BBL can contribute to the expansion of TILs stimulated with 41BB agonists because the proliferation of TILs cultured with 41BBL APCs was reduced when 41BBL was blocked, even in the presence of urelumab (anti-41BB) (Fig. 7F). Intriguingly, CXCL10 was elevated in TIL–APC cocultures when 41BBL was blocked. Consistent with this data, the production of CXCL10 was reduced in donor myeloid cells conditioned with 41BB–Fc (Fig. 6N), suggesting that the stimulation of 41BBL on myeloid cells represses CXCL10 expression that may have been relieved when blocking Abs were present in TIL–APC cocultures. Moreover, we acknowledge that IFN-γ is a known inducer of CXCL10, a maturation factor for DCs (26), and can enhance the ability of 41BBL–APCs to prime cytotoxic T cell responses (27). Hence, it is possible that the induction of CXCL10 could have been indirectly promoted in the presence of anti-41BB through an increased abundance of IFN-γ. Although the blockade of 41BBL can prevent the induction of T cell proliferation through interaction with its cognate receptor, we cannot rule out the possibility that the production of cytokines, such as IL-6, IL-1β, and TGF-β, by APCs may also have an impact on TIL proliferation (Fig. 7D, 7E). However, the cellular origin and the sp. act. of these cytokines on TIL proliferation and function remain unclear. Hence, future studies need to determine the role of 41BB–41BBL induced cytokines, including CXCL10, IL-6, IL-1β, and TGF-β and their impact on both the expansion of TILs and the activity of 41BB agonists.

In our previous report, the addition of anti-41BB was associated with enhanced TIL expansion and the modulation of tumor-resident DC phenotypes characterized by the upregulation of CD80, CD86, and MHCII (16). We conclusively demonstrated that human myeloid cells upregulated costimulatory markers CD86 and 41BBL and proinflammatory cytokines in response to 41BBL stimulation but not in response to 41BB agonists. Moreover, tumor lysate-pulsed APCs that were matured via reverse 41BBL signaling effectively primed TILs. Together, our findings provide feasibility that 41BB–41BBL bidirectional signaling between immune cells can be exploited to enhance to the expansion and function of TILs.

This work was supported in part by the Flow Cytometry Core Facility, the Cell Therapies Core Facility, and the Tissue Core Facility of the H. Lee Moffitt Cancer Center and Research Institute. We also thank the Comparative Medicine Department of the University of South Florida.

The H. Lee Moffitt Cancer Center and Research Institute has licensed intellectual property related to the proliferation and expansion of tumor infiltrating lymphocytes to Iovance Biotherapeutics. S.P.-T., A.A.S., and M.H. are inventors of such intellectual property. S.P.-T. receives salary support on sponsored research agreements between Moffitt Cancer Center and Iovance Biotherapeutics, Myst Therapeutics, Intellia Therapeutics, and Provectus Biopharmaceuticals. A.A.S. is a paid consultant for Iovance Biotherapeutics and has undertaken sponsored travel. None of these organizations provided funding for this study. The other authors have no financial conflicts of interest.