Abstract
In cancers with tumor-infiltrating lymphocytes (TILs), monoclonal antibodies (mAbs) that block immune checkpoints such as CTLA-4 and PD-1/PD-L1 promote antitumor T-cell immunity. Unfortunately, most cancers fail to respond to single-agent immunotherapies. T regulatory cells, myeloid derived suppressor cells (MDSCs), and extensive stromal networks within the tumor microenvironment (TME) dampen antitumor immune responses by preventing T-cell infiltration and/or activation. Few studies have explored combinations of immune-checkpoint antibodies that target multiple suppressive cell populations within the TME, and fewer have studied the combinations of both agonist and antagonist mAbs on changes within the TME. Here, we test the hypothesis that combining a T-cell–inducing vaccine with both a PD-1 antagonist and CD40 agonist mAbs (triple therapy) will induce T-cell priming and TIL activation in mouse models of nonimmunogenic solid malignancies. In an orthotopic breast cancer model and both subcutaneous and metastatic pancreatic cancer mouse models, only triple therapy was able to eradicate most tumors. The survival benefit was accompanied by significant tumor infiltration of IFNγ-, Granzyme B-, and TNFα-secreting effector T cells. Further characterization of immune populations was carried out by high-dimensional flow-cytometric clustering analysis and visualized by t-distributed stochastic neighbor embedding (t-SNE). Triple therapy also resulted in increased infiltration of dendritic cells, maturation of antigen-presenting cells, and a significant decrease in granulocytic MDSCs. These studies reveal that combination CD40 agonist and PD-1 antagonist mAbs reprogram immune resistant tumors in favor of antitumor immunity.
Introduction
Development of T-cell immune-checkpoint inhibitors (ICIs) that block the programmed death 1 (PD-1) pathway has led to substantial responses in several tumor types (1–4). However, these responses are limited to cancers primed for antitumor immunity. Tumors responsive to anti–PD-1 therapy are likely hypermutated, highly infiltrated with effector T cells, and highly express PD-1 ligand (PD-L1) on tumor cells within the tumor microenvironment (TME; ref. 5). However, most tumors develop strategies to evade immune surveillance, leading to immune tolerance, a major obstacle in implementation of successful immunotherapy (6). The majority of immune suppression mechanisms suppress effector T-cell infiltration and function in the TME. These include T regulatory cells (Treg), myeloid derived suppressor cells (MDSCs), tumor-associated macrophages, immature dendritic cells (DCs), and an extensive stromal network of cells and signals that prevent T-cell infiltration (7, 8). Many tumors also fail to elicit antigen-specific T-cell responses due to early tumor immune evasion mechanisms or lack of tumor immunogenicity (9).
Most pancreatic adenocarcinomas (PDACs) and breast cancers have multiple immune-suppressive cell populations in their TME, including monocytic subpopulations and immature DCs (10–12). The majority of patients with these cancers fail to mount an effector T-cell response capable of recognizing cancer cells. Preclinical studies report substantial synergism in combining cancer vaccines with ICIs (13–17). However, few clinical trials have demonstrated benefit (18–21). We previously reported that a human PDAC vaccine alone or in combination with the ICI, ipilimumab, promoted development of lymphoid aggregates and led to transient antigen-specific T-cell responses associated with short-term increases in survival (22, 23). Repeated antigen stimulation in the absence of durable antigen presentation and/or appropriate costimulatory signals may lead to T-cell exhaustion and/or anergy and tolerance. However, if the proper signal is provided to antigen-presenting cells (APCs) for continued T-cell activation, it should lead to more durable responses. A few studies have begun to test agonist immune-oncology (I-O) mAbs that target costimulatory pathways such as anti-CD40, anti–OX-40, and anti-CD137, either alone or in combination with either chemotherapy or ICIs (24–29). CD40, a tumor necrosis factor (TNF) receptor superfamily member, is expressed on many immune cell types, including DCs, macrophages, B cells, and NK cells (30). CD40 signaling induced by CD40L-expressing T helper (Th1) cells on CD40-expressing APCs is critical for APC licensing (31). Administration of CD40 agonist mAb upregulates MHCII, CD80, and CD86 on DCs and macrophages (31), can promote CD8+ T-cell activation in vivo (32), and likely alters the TME myeloid component (25, 33). Therapeutic strategies incorporating CD40 pathway stimulation have been successful in preclinical studies in promoting antigen-specific T-cell expansion (15, 34–36).
Early clinical trials of dacetuzumab, a humanized CD40 mAb, demonstrated responses in hematologic malignancy patients and has entered phase II studies (37). CP870,893, a fully human mAb studied in a number of solid tumors, has shown objective responses in about 20% of melanoma and PDAC patients (27, 31). Preclinical studies demonstrated synergy with anti–PD-1 and CD40 mAb (33, 38, 39) by altering the TME myeloid component, and clinical trials combining CD40 mAb with gemcitabine-based chemotherapy in PDAC are ongoing (25).
We tested the hypothesis that combining a T-cell generating vaccine with a CD40 agonist mAb and anti–PD-1 can induce long-term survival by inducing antitumor CTL trafficking into nonimmunogenic solid malignancies. We show that triple therapy can enhance CTL infiltration in the TME in a tolerance model of breast cancer and a metastatic model of PDAC. We also provide evidence that granulocytic MDSCs (G-MDSC) are reduced and macrophage and dendritic cell populations become mature APCs capable of activating effector CD8+ T cells and Th cells.
Materials and Methods
Mice
Male C57BL/6 mice, age 7 to 8 weeks, were obtained from The Jackson Laboratories. Female neu-N mice were bred and maintained at the Animal Core Facility at Johns Hopkins University as described previously (40). T-cell receptor (TCR) transgenic mice were engineered to express a neu-specific TCR and were maintained as described previously (41, 42). All mice were housed in microisolator cages in a pathogen-free animal facility. All animal procedures were conducted in accordance with, and with approval of, the policy of the Johns Hopkins University School of Medicine Institutional Animal Care and Use Committee.
Cell lines
NT2.5 cells are a syngeneic neu-expressing mammary tumor cell line derived from the neu-N mouse model (42–44). 3T3-neuGM (vaccine) cells are 3T3 cells that have been transfected to constitutively express GM-CSF and HER-2/neu (43, 45). Both cell lines were authenticated by flow cytometry to confirm HER-2/neu expression, and the 3T3 cells were confirmed by ELISA to express GM-CSF. T2Dq cells are T2 B lymphoblast/T lymphoblast hybrid human cells deficient in the MHC-I TAP transporter molecule and engineered to express Dq (46). They were authenticated by sequencing and flow cytometry to confirm HLA-Dq expression. PANC02 mouse pancreatic tumor cells were derived and maintained as previously described (47). The cell line was authenticated by whole-exome sequencing for tumor genes and neoantigens. B78H1-GM cells are an MHC class I–negative variant of B16 melanoma cell line engineered to secrete GM-CSF, and authenticated by ELISA (48–50). All cell lines were obtained between 2010 and 2015 and passaged no more than 10 times between thawing and use in experiments. All cell lines were tested for Mycoplasma prior to use in experiments (last tested June 2017). In vitro stimulation was performed using CTL medium, which consisted of RPMI media with 10% FBS, 1% L-glutamine, 0.5% Pen/Strep, and 0.1% 2-mercaptoethanol (Life Technologies).
Reagents and antibodies
Therapeutic monoclonal antibodies (mAb) were obtained from Bio X Cell. Anti–PD-1 (clone RMP1-14), anti-CD40 (clone FGK4.5), and rat IgG isotype (clone 2A3) were administered intraperitoneally (i.p.) at 100 μg in 100 μL phosphate-buffered saline (PBS). Anti-CD8α (clone 2.43), anti-CD4 (clone GK1.5), and isotype (clone LTF-2) were administered i.p. at 200 μg in 100 μL PBS on day −2, day 0, and twice weekly thereafter (51, 52). Cyclophosphamide (Cy) was obtained from Baxter Healthcare Corp. and prepared as a 20 mg/mL stock solution in water. Any unused solution was discarded after 2 weeks. Mice were administered Cy at 100 mg/kg in 500 μL PBS i.p.
A complete list of fluorescent-conjugated antibodies for flow cytometry can be found in Supplementary Table S1.
In vivo tumor models and therapy
For neu-N experiments, 8-week-old female mice were inoculated with NT2.5 mammary tumor cells in the right mammary fat pad. Cells (5 × 104) were administered on day −3 for survival studies, whereas 1 × 106 cells were injected on day −8 for TIL analysis studies to allow for establishment of 5 × 5 mm tumors prior to treatment (42, 53, 54). The 3T3neuGM vaccine was irradiated at 50 Gy and three subcutaneous (s.c.) injections of 106 vaccine cells each were given in 100 μL, in the left upper limb and both lower limbs on day 0 (41–43, 48). Cyclophosphamide (Cy, 100 mg/kg i.p. in 500 μL PBS) was given i.p. on day −1 to all mice receiving vaccine. For survival studies, all mice received 2 × 106 neu-specific CD8+ T cells i.v. on day 1, unless otherwise specified (42, 53, 54). For TIL and TDLN analysis studies, all mice received 6 × 106 neu-specific CD8+ T cells i.v. on day 1. Anti–PD-1 mAb, CD40 mAb, or rat IgG isotype control was given on day 2. For TIL studies, all tissues were harvested at day 6 unless otherwise specified. For survival studies, anti–PD-1 mAb or isotype control mAb was continued twice weekly thereafter. Tumors were measured twice weekly, and tumor volume was calculated as follows: volume = (L × W2)/2. Mice were euthanized if tumors reached >500 mm3.
For the PDAC metastatic model, 8-week-old male C57BL/6 mice underwent general anesthesia with isofluorane then received 2 × 106 PANC02 cells on day −15 via intrasplenic injection followed by hemisplenectomy (47, 50). A 1:1 mixture of PANC02 and B78H1-GM cells (PANC02vac) was irradiated (50 Gy) and given on day 0 and day 14 (50). Cy (100 mg/kg i.p. in 500 μL PBS) was given i.p. on day −1 to all mice receiving PANC02vac. Anti–PD-1, CD40, and rat IgG isotype mAb were administered per the schedule described for neu-N. Livers were harvested for TIL analysis on day 21. Mice were monitored for survival and considered moribund upon signs of lethargy and/or ascites.
For the PDAC sc model, which allows for early observation of primary tumor growth changes (55, 56), mice received 106 PANC02 cells s.c. in the left lower limb on day −15. PANC02vac, Cy, and mAb therapy were given as described above. Tumors were measured twice weekly and mice were euthanized if tumors reached >500 mm3.
Tissue processing
Tumors were dissected and chopped into 2 × 2 mm or smaller fragments and transferred to C-tubes containing 2.5 mL tumor dissociation enzyme solution (Mouse Tumor Dissociation Kit, Miltneyi Biotech). Tubes were processed on an Octo-Dissociator (Miltenyi) using the 37C_mTDK_2 protocol designated for hard tumor tissue. Digested tumors were quenched with 20 mL RPMI containing 10% FBS and 1% Pen/Strep (hereafter referred to as complete media) and passed through 70-μm cell strainers (Falcon). Cells were centrifuged at 350 × g for 10 minutes at 4°C, aspirated, and red blood cell lysed with 1 mL Ack lysis solution (Quality Biological) for 2 minutes at room temperature (RT). Cells were quenched with 10 mL complete media, centrifuged, and resuspended in complete media for counting and plating for downstream applications.
Livers were dissected and processed in the same manner as tumors, followed by an additional Percoll isolation step after Ack lysis. Cell suspensions were prepared in 6 mL of 80% Percoll (GE Healthcare) and underlaied with 6 mL 40% Percoll in a 15 mL conical tube. Cells were centrifuged for 25 minutes at 2,400 × g at RT with no brake, and the layer containing mononuclear cells was transferred to a tube containing 30 mL complete media. Cells were centrifuged at 350 × g for 10 minutes, aspirated, and resuspended in complete media.
Lymph nodes (LN) were dissected and firmly pressed through 40-μm cell strainers with a 3 mL syringe plunger in 5 mL complete media. Cell suspensions were centrifuged at 350 × g for 5 minutes at 4°C and resuspended in complete media.
Spleens were processed in the same manner as LN, with an additional Ack lysis step after the first centrifugation. Cells were quenched in complete media, centrifuged and resuspended in complete media. For adoptive T-cell transfer, CD8+ T cells were isolated using the mouse CD8-negative selection kit according to the manufacturer's specifications (Stem Cell Technologies), washed twice with PBS, and resuspended in PBS at 1–2 × 107/mL for intravenous (i.v.) injection.
In vitro cytokine detection
Cell suspensions isolated from tumors or LN of treated neu-N mice were plated in complete media with 50 μmol/L β-mercaptoethanol at 1 × 106 cells/mL media in 24-well tissue culture plates. APCs were prepared by pulsing T2Dq cells for 2 to 3 hours in CTL media containing 20 mg/mL of RNEU420–429 (RNEU, PDSLRDLSVF) or the negative control peptide LCMV NP118–126 (NP, PQASGVYM; ref. 42). Peptides were produced in the Johns Hopkins Biosynthesis and Sequence Facility at a purity of 95%. Peptide-pulsed T2Dq cells were washed and plated at a 1:4 ratio with mouse TIL or LN cells in the presence of protein transport inhibitor (GolgiPlug, BD Biosciences). Cells were incubated at 37°C overnight, followed by intracellular flow cytometry analysis of cytokine expression.
Cell suspensions isolated from livers of treated PANC02 hemispleen mice were plated as described above. Cells were stimulated overnight in the presence of GolgiPlug with CD3/CD28 mouse T-cell activation beads (Gibco) according to the manufacturer's specifications. The following day, cells were resuspended with vigorous pipetting, transferred to 1 mL eppendorf tubes, and separated from activation beads under magnetic isolation. Cells were centrifuged at 350 × g for 5 minutes and plated for intracellular flow cytometry analysis of cytokine expression.
Flow cytometry
Freshly harvested mouse cells were plated in 96-well U-bottom plates (up to 2 × 106/well) in PBS. Cells were stained with Live/Dead near-IR stain (Invitrogen) for 30 minutes, followed by a 30-minute incubation in a cocktail of surface antibodies (target and isotype control cocktails) prepared in FACS buffer (1× HBSS with 2% FCS, 0.1% sodium azide, and 0.1% HEPES). For intracellular staining, cells were permeabilized with the FoxP3 transcription factor staining buffer kit (Invitrogen) according to the manufacturer's specifications. Cells were then incubated in a cocktail of intracellular antibodies (target and isotype cocktails) prepared in 1× perm wash buffer for 30 minutes, and washed twice in perm wash buffer prior to data collection. Fluorescence acquisition was carried out on a Beckman Coulter Gallios flow cytometer and analyzed on FlowJo v10.1 (TreeStar).
Clustering analysis and t-SNE visualization
Clustering analysis was performed in order to analyze and visualize the high-dimensional nature of our flow cytometry data. First, we applied fsc/ssc and live/dead gating, then CD8+ and CD4+ gates for the lymphocyte panel or CD3– CD19– gate for the myeloid panel. We then randomly sampled 3,000 cells from the final gate per animal for our analysis (or all available cells, if less than 3,000) and used a t-distributed stochastic neighbor embedding (t-SNE) Barnes-hut algorithm to reduce the dimensionality of the data set for visualization (57). In order to cluster the data appropriately, a network graph was created using k-nearest neighbors (k = 60), and a clustering solution was solved for by maximizing the modularity of the graph in a method adopted by Levine and colleagues (58). All visualizations of clustering solutions through the use of t-SNE heat maps, conventional heat maps, and high-dimensional flow data were generated with the use of ExCYT, a MATLAB-based graphical user interface for high-dimensional cytometry analysis (59).
IHC
Tissues were dissected and fixed in 10% formalin solution prior to paraffin embedding, sectioning, and preparing slides. H&E staining was performed on each specimen for tumor pathologic analysis. IHC was performed on Plus Slides for mouse CD8α and FoxP3 expression by the Oncology Tissue Services Core at Johns Hopkins University. Images were acquired on a Nikon Eclipse E600 and imaged on a Nikon DS-Fi2 camera at 10 × magnification.
Statistical analysis
Survival data were plotted using Kaplan–Meier curves and statistical significance was determined using log-rank test. For tumor growth data, statistical significance was determined using a two-way ANOVA with Tukey multiple comparisons test. Bar graphs represent mean ± SD (n = 3–5 mice per group). Dot plots show one mouse per dot, and each bar represents mean ± SEM. Significance was determined by a one-way ANOVA with Tukey multiple comparisons test. Statistically significant P values are abbreviated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. All statistical analyses were performed using GraphPad Prism v7.00.
Results
A CD40 agonist improves antitumor efficacy of anti–PD-1 therapy
In the neu-N model of breast cancer, mice constitutively express rat neu and are tolerized against the neu antigen (44). We previously reported that a vaccine consisting of 3T3 cells expressing HER-2/neu and GM-CSF (3T3neuGM) induced CD8+ neu-specific T-cell responses against the immunodominant epitope RNEU420–429 and resulted in eradication of neu-expressing tumors (NT2.5 cells) in about 20% of HER-2/neu transgenic (neu-N) mice (41, 43). Tumors develop with low baseline immune infiltration with a predominance of macrophages and G-MDSCs, some Tregs, and few effector T cells (Supplementary Fig. S1). We utilized this model to assess the effects of drug combinations on antitumor immune responses. Tumor-bearing mice were treated with either 3T3neuGM + PD-1 mAb alone or in combination with CD40 mAb, or isotype control mAbs, and monitored for tumor clearance and survival (Fig. 1A). Mice treated with 3T3neuGM + PD-1 mAb + CD40 mAb (triple therapy) had delayed tumor progression and increased median survival relative to isotype controls (P < 0.0001), CD40 mAb (P < 0.0001), or 3T3neuGM + PD-1 mAb treatment alone (P < 0.05; Fig. 1B and C).
Although delayed tumor growth was promising, the majority of mice in all groups eventually succumbed to tumors, leaving room for improvement. We previously reported development of CD8+ TCR transgenic mice that recognize the immunodominant epitope RNEU420–429 (41). We showed that the 3T3neuGM vaccine given with adoptive transfer of 6 × 106 TCR transgenic T cells into tumor-bearing neu-N mice leads to tumor clearance in the majority of animals, whereas the 3T3neuGM vaccine given with 2 × 106 T cells eradicates tumors in only 20% of mice (42). We tested whether triple therapy improved the effect of 3T3neuGM + PD-1 if a suboptimal number of TCR transgenic T cells (2 × 106) were introduced. Only mice receiving the TCR transgenic T cells in combination with 3T3neuGM + PD1 + CD40 resulted in 100% tumor clearance. In contrast, only 30% of mice treated with 3T3neuGM + PD1 were able to clear tumors, and none of the mice receiving isotype or CD40 mAbs alone had long-term survival (Fig. 1D and E). The 3T3neuGM vaccine was necessary for the combination antibody treatment to be effective, as treatment with PD1 + CD40 mAb alone failed to clear tumor in this model (Supplementary Fig. S2A and S2B).
Triple therapy was next evaluated in a metastatic PDAC model in which the mouse tumor PANC02 cells were introduced into wild-type recipients via intrasplenic injection followed by hemisplenectomy to allow liver metastases to develop (Fig. 1F; refs. 47, 60). Mice treated with isotype control mAbs succumbed to metastatic disease at 37 to 62 days (median survival 60 days). A PANC02 vaccine (PANC02vac, irradiated PANC02 cells mixed with a GM-CSF secreting cell line) + anti–PD-1 treatment showed some efficacy, although 50% of mice developed fatal liver metastases (median survival 107.5 days). CD40 agonist mAb was active against PANC02 liver metastases, with 70% long-term survival following a single dose (survival >120 days), and mice treated with PANC02vac + PD1 + CD40 had 100% survival at 120 days. To evaluate primary tumor progression in addition to metastatic survival in PDAC, we utilized a third model in which PANC02 cells were introduced subcutaneously. Similarly, only mice receiving triple therapy remained 100% tumor-free over the course of the study. CD40 mAb and PANC02vac + PD1 alone were only able to clear tumor in 30% of mice (Fig. 1G).
Triple therapy enhances T-cell infiltration and function in murine solid tumors
A hallmark of most nonimmunogenic cancers is lack of TIL and/or failure to mount a robust antitumor T-cell response. To determine whether triple therapy induces effective T-cell responses in mammary tumors, we treated tumor-bearing neu-N mice with established (>5 × 5 mm) NT2.5 tumors with either CD40 mAb alone, 3T3neuGM + PD1, 3T3neuGM + PD1 + CD40, or isotype mAb, with 6 × 106 TCR transgenic T cells, and harvested tumors and TDLN at day 6 after vaccine for analysis of T-cell proliferation, infiltration, and function. Flow cytometry analysis of the TDLN revealed that adoptively transferred neu-specific T cells (Thy1.2+) were significantly increased with triple therapy compared with all other groups, with a 25-fold expansion compared with isotype mAb, a 2.4-fold expansion compared with CD40 mAb treatment, and a nearly 100-fold expansion compared with 3T3neuGM + PD1 treatment at day 6 after vaccine (Supplementary Fig. S2C; P < 0.0001, P < 0.01, and P < 0.0001, respectively). Increased absolute T-cell number was associated with increased Ki67 expression (Supplementary Fig. S2D; P < 0.001) and CFSE dilution (Supplementary Fig. S2E; P < 0.0001). CD40 antibody administration significantly affected the capacity of 3T3neuGM + PD1 therapy to expand total T-cell numbers and could partially account for increased tumor clearance and survival observed. Flow cytometry analysis of tumors revealed that neu-specific T cells were significantly increased in 3T3neuGM + PD1 + CD40-treated mice compared with all other groups (P < 0.01), with no change in endogenous CD8+ T cells (Thy1.2−), CD4+ Th cells, or Tregs, relative to isotype mAb–treated mice (Fig. 2A; Supplementary Fig. S2F). Significantly increased tumor CD8+ T-cell/Treg ratio, a measure of immune robustness, was noted with triple therapy compared with all other groups (Fig. 2B; P < 0.001). In contrast, there were twice as many Tregs in the tumors of 3T3neuGM + PD1-treated mice compared with mice receiving triple therapy (Supplementary Fig. S2F; P < 0.05), and overall the CD8+ T-cell/Treg ratio was similar to isotype mAb–treated mice (Fig. 2B). This pattern of T-cell infiltration was observed on IHC analysis (Supplementary Fig. S3). T-cell activation was measured by the number of IFNγ-, Granzyme B-, and TNFα-secreting CD8+ TIL per mg tumor following 24 hours of in vitro stimulation with neu-peptide–pulsed T2Dq APCs. Granzyme B- and IFNγ-expressing cells were increased in TIL following both 3T3neuGM + PD1 and triple therapy, with triple therapy resulting in significantly more cells compared with CD40 mAb alone (Fig. 2C). Only mice receiving triple therapy had increased TNFα+ T cells, as well as dual and triple cytokine–secreting polyfunctional T cells, compared with all other treatment groups (Fig. 2C). In the TDLN, CD40 and 3T3neuGM + PD1 treatment alone led to increased cytokine-expressing CD8+ T cells, with a further increase observed with triple therapy, including Granzyme B+, TNFα+, and polyfunctional T effector cells (Fig. 2D; P < 0.05).
In the PANC02 hemispleen model, we measured T-cell infiltration into liver metastases after 3 weeks of treatment (Fig. 2E). Similarly, endogenous CD8+ T cells were increased in mice receiving triple therapy, along with PANC02vac + PD1-treated mice. CD4+ Th and Tregs were increased with PANC02vac + PD1 and PANC02vac + PD1 + CD40 treatment compared with isotype and CD40 mAb treatment (P < 0.001 and P < 0.01 for Th and Tregs, respectively). Overall, only CD40 mAb treatment led to improvement in CD8+ T-cell/Treg ratio (CD8+ T-cell/Treg ratio of 10.8 vs. 21.6 for isotype mAb and CD40 mAb, respectively, P < 0.01, from data shown in Fig. 2E), but liver TIL had increased cytokine-secreting CD8+ T cells following stimulation with CD3/CD28-activating beads in all treatment groups (Fig. 2F).
Given the efficacy of triple therapy readily observed in the s.c. PDAC model (Fig. 1G) with endogenous T cells alone, we depleted CD4+ or CD8+ T cells in PANC02 sc tumor-bearing mice during treatment, then monitored tumor progression and survival. Depletion of either CD4+ or CD8+ T cells reduced tumor clearance rate of triple therapy to 20% (Fig. 2G; Supplementary Fig. S4A). CD8+ T-cell depletion completely abolished efficacy of CD40 mAb and PANC02vac + PD1 treatment, with 0% tumor clearance compared with 30% in undepleted mice (Supplementary Fig. S4B and S4C). CD4+ T-cell depletion did not affect tumor clearance in the CD40 mAb–alone group, with 50% of mice remaining tumor-free, whereas tumor clearance rate was reduced from 30% to 10% in the PANC02vac + PD1 treatment group with CD4+ T-cell depletion (Supplementary Fig. S4B and C). Isotype mAb–treated mice all developed tumors, regardless of T-cell depletion status (Supplementary Fig. S4D). In the neu-N model, we can conclude that CD8+ T cells are required for tumor clearance, as long-term survival could only be achieved with triple therapy when neu-specific CD8+ T cells were introduced. However, efficacy of triple therapy in the neu-N model is completely independent of CD4+ T cells, with 100% tumor clearance even in their absence (Supplementary Fig. S4E).
Triple therapy promotes the development of functional T-cell memory
We evaluated the contribution of memory T cells to long-term survival with triple therapy. TIL and TDLN isolated from neu-N mice at day 15 after treatment were analyzed by flow cytometry for abundance of naïve, central memory (TCM), and effector memory (TEM) CD4+ and CD8+ T cells (Fig. 3A–D). As anticipated, few naïve and TCM cells were among the TIL in all cohorts (Fig. 3A and B). However, CD4+ TEM cells were increased with triple therapy compared with isotype and CD40 mAb therapy (Fig. 3A, P < 0.05), whereas CD8+ TEM cells were increased in tumors of mice treated with CD40 mAb alone (Fig. 3B, P < 0.01). To test durability of T-cell responses, tumor-free mice were rechallenged with NT2.5 tumors in the contralateral mammary fat pad. Mice that initially received triple therapy showed some resistance to tumor rechallenge compared with previously untreated (naïve) tumor-challenged neu-N mice, with a median survival of 48 days after rechallenge vs. 32 days for naïve mice (Fig. 3E and F). Similar results were observed in the PANC02 subcutaneous model, but all mice that received prior triple therapy remained tumor-free following rechallenge with PANC02 cells on the contralateral lower limb (Fig. 3G and H). This indicates development of a persistent functional memory T-cell population that prevents formation of new tumors.
Triple therapy converts TME T cells from naïve T cells and Tregs to activated CTLs
To conduct a more detailed and unbiased analysis of immune populations in the tumors of treated mice, we applied high-dimensional clustering and t-SNE visualization to multicolor flow cytometry data. We collected TIL from neu-N mice treated with 3T3neuGM + PD1 with and without CD40 mAb and stained them with a panel of T-cell marker mAbs. Live CD4+ and CD8+ T cells were identified, and a fixed number of T cells from each tumor were clustered based on the mean fluorescence intensity (MFI) of each parameter, including FoxP3, PD-1, CTLA4, and Tim3 (Fig. 4A; Supplementary Fig. S5A–S5C). Fourteen unique clusters were identified, with 6 key clusters labeled on the t-SNE plots and phenotypically characterized on high-dimensional flow plots (Fig. 4A and B). We observed a significant shift in the lymphoid composition of the tumors in each treatment group (Fig. 4A and B). Isotype mAb–treated mice had predominantly T cells that cluster into a CD4+ Th and Treg phenotype (clusters 10 and 6, respectively), as well as naïve CD8+ endogenous T cells (cluster 1). In contrast, CD40 mAb–treated mice had relatively decreased Th and Tregs and a larger proportion of Thy1.2+ CTLs with high CTLA4 expression (cluster 4). 3T3neuGM + PD1–treated tumors demonstrated a significant frequency of Thy1.2+ CTLs (clusters 4, 5, and 14). Cluster 5, which had the highest PD-1 and Tim3 coexpression, was the most highly represented in this treatment group. Triple therapy–treated mice had the greatest frequency of activated CD8+ CTLs overall (clusters 4 and 14), which moderately expressed PD-1 and CD40L, but were largely Tim3 and Gal3-negative. These populations correlated with response to treatment and likely indicate T-cell activation and not exhaustion.
To confirm these CD8+ CTLs represented activation rather than exhaustion, we explored dynamics of the cell populations identified by clusters 4, 5, and 14. Specifically, expression of PD-1 and Tim3 in CD8+ TIL was assessed at days 4, 6, and 8 after treatment (Fig. 5A). We expected Tim3 expression to increase over time if cells were becoming exhausted following gain of PD-1 expression. In all treatment groups of mice, Thy1.2+ TIL were mostly PD-1+ with minimal expression of Tim3 (i.e., clusters 4 and 14). There was a transient increase in Tim3 expression in a small subset of PD-1+ Thy1.2+ T cells at day 6, analogous to the PD-1+ Tim3+ population in cluster 5. This population is somewhat diminished by day 8 and is replaced by PD-1+ Tim3– CTLs. Thy1.2– CD8+ T cells had increased PD-1 expression at day 6 in mice treated with triple therapy (Fig. 5B, P < 0.01), and overall the proportion of Thy1.2– CD8+ T cells lacking PD-1 and Tim3 expression was reduced in mice treated with triple therapy relative to isotype or CD40 mAb therapy (P < 0.01 and P < 0.05, respectively). By day 8, Thy1.2– T cells continued to gain PD-1 expression, a marker of activated CTLs, while losing Tim3 expression. This PD-1+ population was highest in mice receiving triple therapy, indicating that combination treatment improves activation of endogenous CD8+ T cells as well as TCR transgenic T cells.
Triple therapy alters the myeloid component of the TME
We tested whether increased T-cell activation and infiltration observed in mice receiving triple therapy was associated with changes in myeloid cell phenotype and/or function. We performed a similar t-SNE clustering analysis on the myeloid population (CD3– CD19–) in tumors from treated neu-N mice (Fig. 6A; Supplementary Fig. S6A–S6C). The most notable difference observed was a significant increase in CD8+ CD11c+ DCs with triple therapy (clusters 4 and 7; Fig. 6A and B), which highly expressed MHCII and CD86 and represent licensed APCs. Mice treated with 3T3neuGM + PD1 also had an increase in macrophages (cells expressing CD11b and F4/80; clusters 8 and 10) as well as a DC population, but it was a CD8– CD11c+ phenotype, with low MHCII and CD86 expression (cluster 5). G-MDSCs were increased with CD40 mAb treatment compared with isotype mAb and triple therapy (cluster 12). A similar pattern was observed in cluster 2, which contained a population of monocytic cells with dim Ly6C expression. Isotype mAb–treated tumors contained predominantly nonmyeloid cells (i.e., negative expression of all myeloid markers in the panel), and most likely represent contaminating tumor cells (cluster 3).
Changes in the myeloid lineage shown by clustering analysis led us to further investigate infiltration and function of these cell populations. We performed additional characterization of TIL and TDLN cells by flow cytometry to look at expression of myeloid maturation markers. Here we observed a significant decrease in G-MDSCs per mg of tumor in treated neu-N mice, whereas monocytic MDSC (M-MDSC) abundance was relatively unchanged (Fig. 7A). As with clusters 8 and 10, CD11b+ F4/80+ macrophages were increased with treatment, particularly CD40 mAb and triple therapy (Fig. 7B). CD8+ CD11c+ MHCII+ DCs (analogous to clusters 4 and 7) were significantly increased with triple therapy relative to isotype and CD40 mAb alone, and increased with 3T3neuGM + PD1 only relative to isotype mAb (Fig. 7C). However, immature M-MDSCs and mature macrophages in tumors of mice treated with CD40 mAb and triple therapy had significantly decreased expression of arginase 1 (Arg1), an enzyme that contributes to immunosuppressive function (Fig. 7D). In addition, DCs assessed in the TDLN of CD40 mAb and triple therapy–treated mice had significantly higher expression of the costimulatory molecules CD80 and CD86 (Fig. 7E), and a greater abundance of CD103+ DCs (Fig. 7F). CD103+ DCs are Batf3-dependent cross-presenting DCs thought to play a critical role in tumor antigen presentation (61, 62). There was evidence of M1 macrophage polarization occurring in the TDLN with CD40 mAb and triple therapy, demonstrated by increased CD80 and CD86 expression (Fig. 7G; ref. 63). Frequency of CD80+ macrophages was further increased with triple therapy compared with CD40 mAb alone. Finally, CD40 mAb and triple therapy led to a greater abundance of CD169+ (Siglec1) macrophages in the TDLN, which are thought to serve as LN-resident APCs that activate tumor antigen–specific CTLs (Fig. 7H; ref. 63).
Based on these data, triple therapy enhances maturation of multiple myeloid populations in the LN and TME. Distinct patterns were noted in expression of PD-L1 and MHCII in myeloid cell types. In the neu-N model, treatment with 3T3neuGM + PD1 + CD40 led to increased MHCII expression in conventional/classic nonplasmacytoid DCs and M-MDSCs (Supplementary Fig. S7A and S7B). M-MDSCs and G-MDSCs also gained expression of PD-L1, a ligand of PD-1 upregulated in response to IFNγ signaling (Supplementary Fig. S7B and S7C). There was a significant increase in coexpression of MHCII and PD-L1 in CD11b+ F4/80+ macrophages (Supplementary Fig. S7D). Similar patterns in myeloid cells were observed infiltrating PANC02 liver metastases (Supplementary Fig. S7E–H), although coexpression of MHCII and PD-L1 was more predominant in cDCs and CD11b+ F4/80+ macrophages in mice receiving triple therapy.
Discussion
Development of I-O mAbs has revolutionized cancer therapy and reinvigorated the cancer immunology and immunotherapy field. However, most cancer types fail to respond to current single-agent and combination I-O mAbs because of diversity among and within different tumors. New strategies that account for biological differences are needed to convert immune resistant tumors into immunologically primed tumors that can respond to ICIs. Our data show that naturally nonimmunogenic tumors can be reprogrammed by combining a T-cell–inducing vaccine with a costimulatory molecule agonist and an ICI to elicit antitumor immunity.
We found that optimal T-cell activation required triple therapy with a T-cell–inducing vaccine in combination with the DC maturing CD40 agonist and anti–PD-1. Coinciding with increased costimulatory molecule expression and MHCII expression on LN DCs, triple therapy promoted trafficking and/or DC maturation in mammary tumor TME, with greater than a 30-fold increase relative to isotype mAb–treated mice. The critical role of DCs in CD40 agonist combination therapy has been reported (32, 64). These findings, together with observations that triple therapy is dependent on larger numbers of infiltrating CD8+ T cells and less on CD4+ T cells, supported by T-cell depletion studies, suggest that mature DCs may directly prime CTLs in the TME. This possibility is supported by our prior observation of infiltrating lymphoid aggregates containing mature DCs in PDAC patients treated with a similar vaccine (65).
In addition to the role of maturing DCs for enhanced T-cell activation in the TDLN and the tumor, CD40 agonist activity alters infiltration and function of immunosuppressive cells in the TME. Triple therapy led to decreased tumor-infiltrating G-MDSCs and Tregs to further promote an active CTL response. Although there was expansion of MHCIIhi CD86hi CD11b+ F4/80+ macrophages in the TME, expression of arginase 1 in macrophages and immature M-MDSCs was decreased, supporting our findings that CD40 agonist disarms functionality of multiple suppressor cell populations. Studies have reported macrophage-dependent tumor regression (66), particularly in PDAC, in which CD40-activated macrophages deplete the stromal cells within the TME (25). We showed that CD40 agonist activity matures multiple monocyte and DC populations to attract activated CD8+ T cells. We observed increased PD-L1 expression on G-MDSCs, M-MDSCs, and macrophages, likely another indication of myeloid maturation occurring in these previously immunosuppressive cells. With a PD-1 antagonist as part of this treatment regimen, it is unlikely that these cells are able to exert a suppressive function through PD-L1.
These studies suggest that triple therapy efficacy may occur via different mechanisms in PDAC and breast tumor models. This may be due to the tumor type itself or the site of tumor development or metastasis. Understanding the role of monocyte and T-cell populations between tumor types and sites may inform the optimal combination immunotherapies for patients. Both models showed increased Th1 cytokine production in the TME that corresponded with improved survival. Yet the CD8+ T-cell/Treg ratio, significantly increased with triple therapy in neu-N mice, was not elevated in the PDAC model. The increased rate of tumor clearance between triple therapy and PANC02vac + PD1 in the PDAC model (100% vs. 30%–50%) could not be accounted for by differences in T-cell infiltration alone. In vivo T-cell depletion is useful to delineate the contributions of CD4+ and CD8+ T-cell subsets in treatment efficacy. Both CD4+ and CD8+ T cells were required for tumor clearance with triple therapy in the subcutaneous PDAC model. CD40 agonists may act via a CD4+ T-cell–independent mechanism, and as anticipated, depletion of CD4+ T cells did not affect survival of mice treated with CD40 mAb alone. There was a trend toward increased survival in the absence of CD4+ T cells, with 50% of mice achieving long-term tumor-free survival, compared with 30% of undepleted mice, similar to other reports involving therapeutic CD40 pathway stimulation (15, 67). This could be due to the fact that Th cells and Tregs are being depleted simultaneously with anti-CD4, the latter of which may account for the trend toward improved survival. In the neu-N model, CD8+ T cells were required for tumor clearance (triple therapy only results in long-term survival with adoptive T-cell therapy of neu-specific CD8+ T cells). Triple therapy efficacy was completely independent of CD4+ T cells in this model, with 100% tumor clearance even in their absence. Thus, triple therapy efficacy relies on CD8+ T cells in both models, but the role of CD4+ T-cell help might be more important in the PDAC model, in which endogenous antigen-specific T cells need to be induced and therefore require APC and Th cell cross-talk.
In summary, T-cell–inducing vaccines in combination with an agonist monocyte/DC maturing antibody and PD-1 antagonist mAb synergize to induce durable antitumor immunity. CD40 mAb improves T-cell activation in the TDLN but is not sufficient to promote tumor infiltration on its own. CD40 mAb and the combination of CD40 + PD-1 mAbs fail to substantially increase CD8+ TIL. Therefore, vaccine administration is required to promote T-cell infiltration into tumor tissue. Triple therapy with vaccine and dual agonist–antagonist mAbs potentially synergizes by means of the vaccine providing abundant tumor antigen and early APC maturation in the lymph node. This is followed by CD40-induced upregulation of costimulatory molecules and enhanced APC maturation in the lymph node, which promotes CD8+ T-cell activation, proliferation, and trafficking into the TME. These results provide a model for converting nonimmunogenic tumors into immune responsive cancers.
Disclosure of Potential Conflicts of Interest
E.M. Jaffee reports receiving commercial research funding from Aduro Biotech, Bristol-Myers Squibb, Amgen, Hertix, and Corvus; has ownership interest in Aduro Biotech; and is a consultant/advisory board member for Genocea, DragonFly, CSTONE, and Adaptive Biotech. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: H.S. Ma, E. Roussos Torres, E.M. Jaffee
Development of methodology: H.S. Ma, E. Roussos Torres, B. Scott, T. Armstrong, E.M. Jaffee
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H.S. Ma, B. Poudel, E. Roussos Torres, T.M. Robinson, B. Christmas, B. Scott, K. Cruz, S. Woolman, T. Armstrong
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H.S. Ma, B. Poudel, J.-W. Sidhom, V.Z. Wall, E.M. Jaffee
Writing, review, and/or revision of the manuscript: H.S. Ma, B. Poudel, E. Roussos Torres, T.M. Robinson, B. Christmas, V.Z. Wall, E.M. Jaffee
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Armstrong
Study supervision: E. Roussos Torres, T. Armstrong, E.M. Jaffee
Acknowledgments
This study was supported by NIH R01CA184926, NIH P30CA006973 (Sidney Kimmel Comprehensive Cancer Center grant), NIH P50CA062924 (Gastrointestinal cancers Spore grant), and by a Lustergarten Foundation Pancreatic Cancer Convergence Dream Team Translational Research Grant (Grant Number: SU2C-AACR-DT14-14). Stand Up To Cancer is a program of the Entertainment Industry Foundation administered by the American Association for Cancer Research. E.M. Jaffee also acknowledges funding from the Broccoli Foundation, The Bloomberg–Kimmel Institute for Cancer Immunotherapy, and The Skip Viragh Center for Pancreas Cancer Clinical Research and Patient Care. E. Roussos Torres is funded through the MacMillan Pathway to Independence Fellowship as well as a training grant (5T32 CA009071-34). T.M. Robinson has salary support through the institutional hematology training grant (T32 HL 7525-33). B. Scott is supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-1232825. Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.
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.