The IL2 receptor (IL2R) is an attractive cancer immunotherapy target that controls immunosuppressive T regulatory cells (Treg) and antitumor T cells. Here we used IL2Rβ-selective IL2/anti-IL2 complexes (IL2c) to stimulate effector T cells preferentially in the orthotopic mouse ID8agg ovarian cancer model. Despite strong tumor rejection, IL2c unexpectedly lowered the tumor microenvironmental CD8+/Treg ratio. IL2c reduced tumor microenvironmental Treg suppression and induced a fragile Treg phenotype, helping explain improved efficacy despite numerically increased Tregs without affecting Treg in draining lymph nodes. IL2c also reduced Treg-mediated, high-affinity IL2R signaling needed for optimal Treg functions, a likely mechanism for reduced Treg suppression. Effector T-cell IL2R signaling was simultaneously improved, suggesting that IL2c inhibits Treg functions without hindering effector T cells, a limitation of most Treg depletion agents. Anti-PD-L1 antibody did not treat ID8agg, but adding IL2c generated complete tumor regressions and protective immune memory not achieved by either monotherapy. Similar anti-PD-L1 augmentation of IL2c and degradation of Treg functions were seen in subcutaneous B16 melanoma. Thus, IL2c is a multifunctional immunotherapy agent that stimulates immunity, reduces immunosuppression in a site-specific manner, and combines with other immunotherapies to treat distinct tumors in distinct anatomic compartments.
These findings present CD122-targeted IL2 complexes as an advancement in cancer immunotherapy, as they reduce Treg immunosuppression, improve anticancer immunity, and boost PD-L1 immune checkpoint blockade efficacy in distinct tumors and anatomic locations.
Despite recent successes of individual cancer immunotherapy agents, multiple, distinct immune pathways must be corrected for optimal treatment efficacy (1). Immunosuppressive regulatory T cells (Treg) are a fundamental obstacle to effective anticancer immunity in many cancer types (2, 3), including ovarian cancer (4) and melanoma (5). However, the development of drugs that deplete or inhibit Tregs has been difficult, partly due to a scarcity of known targetable, Treg-specific molecules. The IL2 receptor (IL2R) is an attractive immunotherapy target due to its control over Treg development, homeostasis, and functions (6–9), but also antitumor T-cell effector functions (10, 11). An ideal IL2R-targeted agent would reduce Treg suppression, but simultaneously stimulate antitumor T cells and to serve as multifunctional cancer immunotherapy.
The IL2R is comprised of three subunits: high-affinity IL2Rα (CD25), intermediate-affinity IL2Rβ (CD122), and a common gamma chain (CD132, γc). CD25 does not participate in signal transduction but markedly enhances IL2 potency by augmenting IL2 binding to the trimeric αβγ IL2R (11, 12). Mice lacking IL2 (13), CD25 (14), or CD122 (15) die from lethal T cell–driven autoimmunity due to reduced functional Tregs, indicating IL2 signaling dominantly supports Tregs and is dispensable for excessive T-cell activation. Tregs constitutively express high CD25 (16) that is critical for their suppressive function (7) but also enable their preferential depletion by αCD25 antibodies (17). Unlike Tregs, memory-phenotype CD8+CD44+ T cells and natural killer (NK) cells mediating tumor cytotoxicity highly express CD122, not CD25 (11). Selective stimulation of medium-affinity βγ IL2R can be achieved by complexing IL2 with anti-IL2 antibody clones that occlude the CD25-binding epitope of IL2 (CD122-selective IL2 complexes; IL2c; refs. 18–20); and reviewed in ref. 21 to augment CD122+ effector T-cell functions while largely avoiding Treg expansion (18, 22). IL2c induces CD8+ T-cell proliferation associated with tumor rejection in several cancer models (20, 21). However, the effects of CD122-selective IL2c on Treg phenotype and functions are unreported.
Here, we show that IL2c improves antitumor T-cell functions as expected and expand the list of IL2c-responsive tumors, but we unexpectedly found that IL2c also significantly reduce Treg function specifically in the tumor microenvironment and induce a fragile Treg phenotype. We combined αCD25 with reportedly CD25-independent IL2c after hypothesizing IL2c would insulate effector T cells from negative off-target effects of αCD25. However, combination treatment with IL2c plus αCD25 was less efficacious than IL2c monotherapy in the ovarian cancer model, indicating an unappreciated role for CD25 in IL2c actions. Inhibiting programmed cell death-ligand 1 (PD-L1) signals with antagonistic mABs (αPD-L1) has generated high-profile clinical successes in several difficult-to-treat cancers (23–25). Despite these encouraging results, most patients do not benefit from PD-L1 blockade, due in part to the absence of preexisting tumor inflammation (26). Therefore, a strategy to sensitize the large proportion of patients unresponsive to PD-L1 blockade could include pretreatment with an immunotherapy that induces intratumoral T-cell infiltration. Although αPD-L1 had no effect as monotherapy in ID8agg ovarian cancer, combination treatment with αPD-L1 plus IL2c produced complete tumor eradication and elicited durable, protective tumor-specific immunity. IL2c increased intratumoral immune cell infiltration and improved αPD-L1 efficacy in ID8agg ovarian cancer, and similarly improved the αPD-L1 response of subcutaneous B16 melanoma. We conclude that CD122-selective IL2c reduce Treg immunosuppression, promote durable antitumor immunity, and improve αPD-L1 treatment efficacy in distinct cancers and anatomic compartments. Thus, IL2c is a strong candidate for clinical trials in ovarian cancer, melanoma, and other cancers alone and in rational combinations, including with αPD-L1.
Materials and Methods
We performed controlled preclinical experiments to test clinical efficacy and immune effects of selective stimulation of IL2R using IL2c in two tumor models: murine ovarian cancer and melanoma in two anatomic compartments: peritoneum and skin, respectively. We tested αCD25 as an alternative therapy, and the efficacy of combining IL2c with either αCD25 or αPD-L1. Outcomes included antitumor immunity, Treg effects, IL2 signaling, tumor growth control and establishment and durability of effective anti-tumor immunity. All animals were randomized to control/treatment groups and tumor measurements were collected by experimenters blind to treatment conditions.
Proteins and antibodies
Carrier-free recombinant mouse IL2 was purchased from BioLegend and stored at −80°C in aliquots. αIL2 (clone JES6-5H4), αCD25 (clone PC61.5.3), αPD-L1 (10F.9G2) and isotype control antibodies (clone LTF-2 rat IgG2b and clone HRPN rat IgG1) were purchased from Bio X Cell and stored at 4°C.
Mice and cell lines
Wild-type (WT) C57BL/6 mice (all female unless otherwise noted) were purchased from Jackson Labs, bred in our animal facility and used at 7–20 weeks old. Female C57BL/6 FoxP3DTR mice were from Alexander Rudensky (then at University of Washington, Seattle, WA). WT BL6 mice that received diphtheria toxin were used as comparators for FoxP3DTR mice. FoxP3-IRES-RFP (FIR) mice from which viable RFP+ Tregs can be isolated were from Richard Flavell (Yale University, New Haven, CT). Animals were maintained in specific pathogen-free conditions and provided a normal diet and water ad libitum. All tumor cells were cultured in RPMI1640 medium containing 5% FBS, 10 U/mL penicillin, 10 μg/mL streptomycin, 2 mmol/L l-glutamine, and 1 mmol/L HEPES buffer (R5 medium) in a humidified 37°C incubator (5% CO2). The mouse ovarian cancer cell line ID8 was from George Coukos (then at University of Pennsylvania, Philadelphia, PA), from which we developed the highly aggressive ID8agg subline (27). To track ID8agg tumor burden in vivo, we transfected ID8agg cells with the pGL4.51 vector encoding luciferase (Promega) with Attractene transfection reagent (Qiagen) according to manufacturer's instructions. Single cell–derived clones of luciferase-expressing ID8agg (ID8agg-Luc) were isolated by limiting dilution after 1 mg/mL selection with G418 antibiotic. One clone with optimal luciferase activity was cryopreserved and used in all experiments. Upon reculture, ID8agg-Luc was continuously maintained in 0.2 mg/mL G418 to prevent the outgrowth of luciferase-negative clones. B16-F10 was purchased from ATCC. All cell lines were periodically tested and confirmed free of Mycoplasma by Mycoalert PLUS Detection Kit (Lonza Bioscience) before experimental uses but were not genetically authenticated.
In vivo tumor challenge and treatments
Mice were challenged intraperitoneally with 4 × 106 ID8agg or ID8agg-Luc cells, or subcutaneously on one flank with 5 × 105 B16-F10 cells. All cells were in log growth phase at challenge. Upon first tumor measurement and before randomizing into treatment groups, mice were occasionally excluded due to abnormal tumor growth that was verified as an outlier by Grubb test. A total of 200 μg/mouse αCD25 was given intraperitoneally 7 days after ID8agg challenge. A total of 100 μg/mouse αPD-L1 was given intraperitoneally 11, 14, 17, and 20 days after ID8agg and B16 challenge. A total of 1.5 μg/mouse IL2 was complexed with 7.5 μg/mouse αIL2 (clone JES6-5H4) at the optimal 1:2 molar ratio in PBS at 37°C for 15–30 minutes (18) before intraperitoneal administration on days 7, 9, 11, and 13 after ID8agg or B16 challenge. Occasionally, bioluminescent measurements were excluded in the final measurement (week 5) due to obvious interference of ascites with light emission (28). Thus, week 5 bioluminescent mean values for control-treated mice are generally underestimated. For B16-F10 tumors, treatment was initiated when mean tumor volume was approximately 100 mm3 (7 days postchallenge, as indicated) and measured every 2–4 days by calipers in a blinded fashion, with volume calculated as 0.5 × (width × length2). Survival was defined as spontaneous death, moribundity, tumor volume >1,500 mm3 for B16, or weight >130% of baseline (ascites) in ID8agg. In vivo bioluminescence of ID8agg-Luc was assessed with an IVIS Lumina Imaging System (Perkin Elmer) 15 minutes after intraperitoneal injection of 3 mg of PBS-dissolved d-luciferin K+ (Gold Biotechnology) with 30-second exposures, medium binning, and F/stop = 1. Identical regions of interest were drawn over each subject's abdomen and average radiance (photons/second/cm2/sr) was quantified with Living Image software version 4.2. For FoxP3 DTR and WT mice, 5 μg/kg recombinant diphtheria toxin (Santa Cruz Biotechnology) was given intraperitoneally every 3 days, as indicated.
Mice were sacrificed by cervical dislocation after induction of deep isoflurane anesthesia. Ascites and peritoneal exudate cells were isolated by recovering 5 mL of intraperitoneally injected ice-cold PBS supplemented with 2% FBS and 1 mmol/L EDTA. For intratumoral analysis, the large omental tumor was dissected, minced with a razor blade, incubated in serum-free RPMI160 with 0.33 mg/mL DNAse I and 3 mg/mL collagenase IV (Worthington Biochemical) for 1 hour, and passed through a 70 μm filter. Ascites and dissected tumor-draining mesenteric lymph node cells were incubated with indicated antibodies and data acquired on a BD LSR II flow cytometer with FACSDiva software version 6.1.2. Dead cells were excluded by staining with Ghost Dye UV 450 (Tonbo). Nonspecific labeling was preblocked by addition of anti-CD16/32 at 1:100 dilution (clone 2.4G2; Tonbo). Cells were stained for surface antigens by incubating at 4°C for 30–45 minutes with αCD3 (clone: 17A2; vendor: Tonbo), αCD4 (RM4-5; Tonbo), αCD8α (5H10; Invitrogen), αCD25 (eBio7D4; eBioscience), αCD107a (1D4B; BioLegend), αCTLA-4 (UC10-4F10-11; Tonbo), αPD-1 (RMP1-30; eBioscience), αCD103 (2E7; BioLegend), at 1:100 dilution. For intracellular staining, cells were fixed and permeablized with a FoxP3/Transcription Factor Buffer Kit (eBioscience) according to the manufacturer's instructions, and incubated at 4°C for 45 minutes with αFoxP3 (FJK16s; eBioscience), αGranzyme B (NGZB; eBioscience), αIFNγ (XMG1.2; Tonbo), TNFα (MP6-XT22; BioLegend), αT-bet (4B10; BioLegend), αRorγt (Q31-378; BD Biosciences), αGata3 (L50-823; BD Biosciences), αIL17 (TC11-18H10.1; BioLegend), αTCF-1 (S33-966; BD Biosciences) at 1:75 dilution in 1× FoxP3 permeabilization buffer. For detection of CD107a, granzyme B, TNFα, and IFNγ, cells were stimulated immediately after isolation with leukocyte activation cocktail + BD GolgiPlug (BD Biosciences) containing phorbol 12-myristate 13-acetate, ionomycin, and brefeldin A at 2 μL cocktail/mL medium CR10 medium (RPMI1640 with 10% FBS, l-glutamine, sodium pyruvate, nonessential amino acids, penicillin/streptomycin, and HEPES buffer) for 5 hours in a 37°C incubator. Absolute numbers of cells were determined by multiplying the ratio of the cell of interest per live, singlet cell in each flow sample by the total number of viable cells from the sample specimen and then, as applicable, normalized to ascites volume (in cases where mice had no measurable ascites, a volume of 50 μL was used for calculations) or tumor weight. Fluorescence minus one (FMO) controls were used to set gates for all antibodies that did not produce clear, bimodal populations. All figures show geometric mean fluorescence intensity (MFI). Because of interexperiment variability, pooled data show normalized MFI (where individual samples were normalized to the control MFI of their respective experiment).
In vitro Treg suppression assay
FIR mice were challenged with ID8agg, treated with IL2c or isotype and sacrificed 5 weeks posttumor challenge. CD45+CD3+CD4+ CD25hiRFP+ Tregs were sorted by flow cytometry (95%–99% purity) from tumor-draining lymph nodes (TDLN) and ascites and incubated with 30,000 carboxyfluorescein diacetate succinimidyl ester (CFSE)-stained, splenic CD45+CD3+CD4+CD25− responder T cells (95%–99% purity) obtained by electronic cell sorting from naïve, age-matched syngeneic females. T cells from naïve WT mice, at graded concentrations plus αCD3/αCD28 Dynabeads (Invitrogen; 1 bead for every 5 effector cells) in a round-bottom 96-well plate for approximately 80 hours. Proliferation was analyzed by flow cytometry based on the dilution of CFSE.
Ex vivo Treg incubation
CD45+CD3+CD4+CD25hiRFP+ Treg from FIR mice spleens were sorted, incubated with mIL2 (20 ng/mL) or IL2c (20 ng/mL mIL2 + 100 ng/mL αIL2) in 100% CR10 medium or 50% CR10+50% ascites collected from ID8agg challenged WT mice in a 48-well plates coated with αCD3 and αCD28 antibodies. Five days later, Tregs from the culture were harvested and subjected to in vitro Treg suppression assays as described above
Cell culture supernatants from Treg suppression assay were appropriately diluted and analyzed with an IL2 ELISA Max Deluxe Kit (BioLegend), according to the manufacturer's instructions.
In vitro STAT5 phosphorylation assay
CD3+ T cells were sorted from ascites and TDLN of ID8agg-bearing mice, plated at 100,000 cells/well in CR10 medium in a 96-well plate, and rested for 1 hour. IL2 was pulsed for 30 minutes and cells were immediately fixed and permeablized using BD Phosflow Lyse/Fix and Perm Buffer III, then stained with CD8, CD4, FoxP3, and pSTAT5 [BD Biosciences; clone: 47/Stat5(pY694)]. The pSTAT5 gate was set with reference to a FMO control.
Statistical and data analysis
Data were analyzed and graphed with Graphpad Prism 6.01. All points with error bars represent the mean ± SEM and points without error bars represent individual mice, unless indicated otherwise. For comparison of two means, we used an unpaired Student t test. Three or more means were compared with one-way ANOVA and post hoc Sidak test. Tumor growth curves were compared by two-way ANOVA and first analyzed for an overall effect due to treatment (P value labeled on figure legend) followed by post hoc Sidak test (P value labeled on individual graph points) of discrete time points. Log-rank test was used to compare Kaplan–Maier curves. Occasionally, datasets with suspected outliers were identified by Grubb test (used only once for a given dataset). For all analyses, significance was based on a multiplicity-corrected, two-sided α of 0.05. In each figure, we provide P values to two significant digits and number of independent experiments performed.
All animal work was done under University of Texas Health San Antonio (UTHSA, San Antonio, TX) Institutional Animal Care and Use Committee approved studies in compliance with the Guide for the Care and Use of Laboratory Animal Resources.
IL2c promotes durable antitumor immunity in poorly immunogenic ovarian cancer
We tested a CD122-selective IL2c (18) that induces preferential expansion and activation of effector T cells to assess effects in mouse ID8agg ovarian cancer, a poorly immunogenic tumor refractory to many forms of immunotherapy (29). IL2c extended survival, durably reduced tumor burden, and reduced malignant ascites (Fig. 1A–C). Although IL2c serum half-life is only approximately 24 hours (19), tumor control in mice treated with IL2c alone persisted for several weeks after IL2c administration, suggesting durably augmented antitumor immune function. Accordingly, ascites effector T cells isolated over 3 weeks after the final IL2c dose exhibited increased expression of antitumor cytokines (Fig. 1D and E).
IL2c surprisingly reduces CD8+/Treg ratio in the ascites and affects Treg and effector T-cell numbers distinctly in tumor draining lymph nodes versus ascites
To assess how IL2c altered effector and Treg numbers and thus their relative ratios in various compartments, we sacrificed ID8agg tumor-bearing mice approximately 3 weeks after the final IL2c dose and measured prevalence and numbers of T-cell subsets in ascites and TDLN. IL2c greatly reduced overall ascites cellularity by preventing the formation of malignant ascites, greatly reducing the absolute number of both CD8+ T cells and Tregs in ascites (Fig. 2A, top). However, CD8+ T-cell concentration was unchanged while Treg concentration was elevated (Fig. 2A, bottom). Although an increased CD8+/Treg cell ratio in the tumor microenvironment predicts enhanced survival for patients with ovarian cancer (30, 31), we surprisingly found that IL2c decreased the ascites CD8+/Treg ratio 3 weeks after the final IL2c administration (Fig. 2B). Thus, a surprising IL2c-mediated Treg increase accounted for the reduced CD8+/Treg ratio. IL2c also increased prevalence of nonregulatory CD4+FoxP3− T cells in ascites (Supplementary Fig. S1A). Such changes in T-cell subsets were not observed one week after the final IL2c dose (Supplementary Fig. S1B and S1C). Therefore, we focused further immune studies 3 weeks after final IL2c dose. Strong tumor rejection exhibited by IL2c-treated mice suggested that the low CD8+/Treg ratio could be from improved CD8+ function, reduced Treg function, or both. In TDLN, we saw elevated numbers of CD8+ and nonregulatory CD4+FoxP3− T cells at 3 weeks after the last IL2c dose (Fig. 2C; Supplementary Fig. S1A). However, Treg cell numbers in TDLN were not, or only slightly, elevated (Fig. 2C). Thus, IL2c increased absolute numbers of non-Treg T cells reduced Treg prevalence in TDLN, but not ascites (Fig. 2C and D). We also observed an increase in both CD8+CD44hiCD62L+ central memory and CD8+CD44hiCD62L− effector memory T cell in ascites (Supplementary Fig. S2, top) and an increase in CD8+CD44hiCD62L+ central memory T cells in TDLN (Supplementary Fig. S2, bottom), suggesting IL2c promotes long-term memory.
IL2c reduces ascites Treg suppressive function and induces a fragile Treg phenotype
Because IL2c reduced both tumor burden and ascites CD8+/Treg ratio (Figs. 1A and 2B) due to increased Treg concentration in ascites (Fig. 2A), we hypothesized that IL2c reduced Treg function in the tumor microenvironment. Thus, we examined IL2c effects on expression of proteins that mediate Treg function. IL2c increased FoxP3 MFI (per cell expression) among ascites, but not TDLN Tregs (Fig. 3A). However, despite increased FoxP3 MFI, IL2c greatly reduced ascites Treg CD25 MFI, and slightly reduced CD25+ ascites Treg prevalence (Fig. 3B; Supplementary Fig. S3A), suggesting reduced Treg function (7). Despite reducing Treg CD25, IL2c increased CD25 expression in CD8+ and CD4+FoxP3− T cells (Fig. 3B), indicating differential effects on effector versus Tregs, and suggesting favored IL2 availability to antitumor T cells over Tregs as expected. IL2c elicited a similar, opposing expression pattern for the cytotoxic effector protein granzyme B, important for Treg suppressive function (32, 33) and antitumor T-cell cytotoxicity, in effector versus Tregs from ascites (Fig. 3C). IL2c also reduced Treg CTLA-4 expression (Fig. 3D, top), which Tregs use to prevent effector T-cell costimulation (34), reduced CD103 (Fig. 3D, bottom), a marker of highly suppressive Tregs (35) and mildly reduced CD39 (Supplementary Fig. S3B), another marker of highly suppressive Tregs (36). We did not see a significant difference in expression of TIGIT, LAG3, the transcription factor HELIOS or production of IL10 (Supplementary Fig. S3C) in IL2c-treated ascites Tregs. These data indicate that IL2c therapy durably reduces expression of proteins that mediate Treg function, while increasing some of those same proteins on effector T cells. Thus, CD122-selective IL2c likely degrades Treg functions, but improves effector T-cell functions.
Recent reports have described a fragile Treg phenotype in the tumor microenvironment, which is defined as Tregs maintaining FoxP3 expression but with loss of suppressive function. These fragile Tregs produce IFNγ, upregulate the transcription factor Tbet, can have elevated PD-1 expression and are associated with responsiveness to αPD-1 cancer immunotherapy (37). We observed that IL2c increased PD-1 (Fig. 3E), T-bet (Fig. 3F), and IFNγ production (Fig. 3G) in Tregs in ascites, but not TDLN (Supplementary Fig. S3D), demonstrating that IL2c induces a fragile Treg phenotype in ascites.
To assess Treg functions directly, we challenged FIR mice (38) with ID8agg, treated with IL2c as before, recovered CD3+CD4+CD25hi RFP+ Tregs by electronic cell sorting and tested their ability to suppress cellular proliferation and cytokine secretion. Although IL2c did not change TDLN Treg functions, IL2c compromised ascites Treg suppression of naïve T-cell proliferation and IL2 secretion (Fig. 4A–C). Thus, IL2c is a novel Treg inhibitor that does not compromise effector T-cell functions, and specifically reduces tumor microenvironmental Treg functions in the ID8agg ovarian cancer model. To determine whether IL2c directly affects Treg suppressive function, we sorted live Tregs from naïve FIR mouse spleens and incubated them with IL2 or IL2c ex vivo. There was no significant difference in Treg numbers or suppressive function after ex vivo IL2c incubation in complete culture medium (Supplementary Fig. S4A), but we observed a minor reduction of Treg suppression after culturing in medium containing 50% ascites collected from ID8agg challenged mice treated with IL2c (Supplementary Fig. S4B). These data suggest an indirect mechanism for IL2c-mediated reduction of Treg suppression, although direct mechanisms are not excluded.
IL2c differentially regulates Treg and effector T-cell IL2-dependent STAT5 phosphorylation
Because both CD25 expression and Treg suppression are primarily STAT5 dependent (7), we assessed intracellular IL2 signal transduction, as differential CD25 expression (Fig. 3B) suggested potentiation of signaling cascades in effector T cells but reduced IL2 signaling in Tregs. We challenged mice with ID8agg and treated with IL2c as in Fig. 1, sorted total CD3+ T cells from ascites and TDLN, and pulsed with IL2 ex vivo. As expected, ascites-derived CD8+ and non-Treg CD4+FoxP3− T cells from IL2c-treated mice versus control-treated mice exhibited increased STAT5 phosphorylation (Fig. 4D), indicating that IL2c enhanced sensitivity to IL2-dependent intracellular signaling. In contrast, tumor microenvironmental Tregs showed no change in pSTAT5 following incubation with up to 100 ng/mL IL2 (Fig. 4D). However, Tregs exhibited downregulation of CD25 that mediates high-affinity IL2 signaling, possibly only compromising their response to lower IL2 concentrations. In support, after pulsing with low IL2 ex vivo, we observed reduced pSTAT5 in Tregs in contrast to potentiated responses among CD8+ and CD4+FoxP3− effector T cells (Fig. 4D) while no difference was observed in pSTAT5 of Tregs from TDLN (Supplementary Fig. S5). These data indicate that CD122-selective IL2c compromises high-affinity IL2 signal transduction in Tregs, which helps explain their reduced function, and supports a role for improved IL2 signaling as a basis for IL2c-mediated improved effector T-cell functions as suggested (18, 20–22, 39), but not previously demonstrated.
IL2c increases exhaustion receptor expression on antitumor T cells
IL2c treatment promoted tumor rejection that greatly extended survival, but was not curative. As multiple immune pathways must be corrected for optimal anticancer efficacy, we hypothesized that IL2c promoted chronic antigenic stimulation critical for the expression of exhaustion receptors that could impede cure. PD-1 expression was strongly upregulated on both CD8+ and CD4+FoxP3− effector T cells after IL2c treatment (Fig. 5A and B). Coexpression of multiple inhibitory receptors could indicate greater T-cell activation state or exhaustion (40). We observed increased prevalence of both double-positive lymphocyte activation gene (LAG)-3+PD-1+ CD8+ and CD4+FoxP3− T cells and triple-positive T-cell immunoreceptor with Ig and ITIM domains (TIGIT)+LAG-3+PD-1+ CD8+ T cells after IL2c treatment of ID8agg ascites (Fig. 5C and D). Beneficial IFNγ expression was enriched among PD-1+ T cells in IL2c-treated animals (Fig. 5E), consistent with enhanced effector function, proliferation and expression of costimulatory proteins on PD-1+ CD8+ T cells in the peripheral blood of patients with lung cancer following αPD-1 treatment (41) and similar to enhanced T-cell functions after treating tumor-bearing mice with a human IL2c construct, although PD-1+ T cells had poor IFNγ expression in that study (21). Thus, distinct IL2c constructs appear to mediate distinct immune effects in different tumors.
IL2c plus αPD-L1 promotes complete ovarian cancer tumor response with durable, protective immune memory
We found that mice experienced complete tumor eradication after ID8agg challenge using FoxP3DTR mice for specific and near total Treg depletion (Supplementary Fig. S6A–S6C), demonstrating the importance of Tregs to immunopathology in this model, as we showed in human ovarian cancer (4). As IL2c alone was effective but not curative against ID8agg, we considered adding αCD25 to improve Treg depletion by further preventing IL2 access to CD25. However, αCD25 unexpectedly reduced IL2c efficacy on ID8agg based on tumor growth, ascites accumulation and antitumor T-cell functions (Supplementary Fig. S7A–S7E) and thus was not considered further.
αPD-L1 immunotherapy is thought to be less effective in patients without preexisting T-cell infiltration (26) and generally more effective in PD-L1+ tumors (27). We showed that both B16 mouse melanoma and ID8agg express PD-L1 (27). In ID8agg, while control mice had essentially no tumor-infiltrating lymphocytes (TIL), IL2c greatly increased intratumoral infiltration of effector T cells as reported in renal cell carcinoma (42) but also increased intratumoral Tregs, although the former to a greater degree (Supplementary Fig. S8A). Because IL2c promoted TILs and the expression of exhaustion receptors, particularly PD-1, on antitumor T cells we hypothesized that IL2c-treated mice would derive additional benefit from PD-L1 blockade. Thus, we treated ID8agg-bearing mice with IL2c as before plus αPD-L1 starting with the third IL2c dose, to allow time for TIL to accumulate. While αPD-L1 had no effect on ID8agg tumor growth alone, we observed complete reduction of tumor bioluminescence with αPD-L1 addition in nearly all subjects and overall survival significantly exceeding that for IL2c alone (Fig. 6A–C). For those IL2c + αPD-L1 treated ID8agg-bearing mice with complete responses >60 days past the original tumor challenge, we rechallenged with ID8agg at the initial challenge inoculum. Tumors grew progressively in untreated, naïve WT mice as expected. In contrast, tumors were fully rejected by complete responder mice in the absence of additional treatment (Fig. 6D), consistent with IL2c-mediate induction of memory T cells (Supplementary Fig. S2). Similarly, αPD-L1 improved IL2c treatment response and survival in B16 melanoma (Fig. 6E–G). Similar to ID8agg, IL2c increased B16-infiltrating Treg prevalence and numbers (Supplementary Fig. S8B), but increased CD8+ T cells to a greater extent, thereby increasing the CD8/Treg ratio (Supplementary Fig. S8C), contrasting with the reduced CD8/Treg ratio after IL2c in ID8agg (Fig. 2A), but consistent with reported effects of a human IL2c construct in B16 melanoma (21). IL2c also reduced Treg functional markers (Supplementary Fig. S9A and S9B) without affecting their FoxP3 expression (Supplementary Fig. S9C) in B16 tumors, similar to effects in ID8agg tumors (Fig. 3B–E). We also observed increased TIGIT+LAG-3+PD-1+ CD8+ T cells in B16 tumors 1 day after IL2c treatment (Supplementary Fig. S10), differing from lower T-cell exhaustion markers after treating B16 tumor-bearing mice with a human IL2c construct (21).Comparable times after IL2c treatment could not be assessed between B16 and ID8agg tumors owing to different kinetics of tumor progression. In both ID8agg and B16, αPD-L1 had little effect on tumor growth when used alone, suggesting pretreatment with IL2c altered the immune milieu, including promoting TIL as shown, to promote αPD-L1 response. Thus, IL2c sensitizes tumors to αPD-L1 in two distinct tumors in distinct anatomic compartments and is curative even in poorly immunogenic tumors with accompanying durable, protective immune memory.
Results from over three decades of ovarian cancer immunotherapy clinical trials have largely been disappointing (43). Ovarian cancer has a relatively lower mutation load compared with other carcinomas that have exhibited immunotherapy trial successes (e.g., melanoma, lung, renal cell, and bladder cancer; ref. 44), possibly making ovarian cancer intrinsically more difficult to treat with immunotherapy. Nonetheless, intratumoral T-cell infiltration predicts improved ovarian cancer patient survival (45) that is reduced by intratumoral Tregs (4), providing a strong rationale for pursuit of effective anti-ovarian cancer immunotherapy.
Despite its substantial side effects, treatment of cancer patients with IL2 is among the first immunotherapies to generate cures in patients with treatment-resistant cancers (46). Nonetheless, aside from limited efficacy, IL2 also suffers from significant toxicities and its ability to promote Tregs (46, 47). The IL2R is an excellent immunotherapy target for further studies of agents mediating multiple, beneficial immune outcomes, owing to its control of both antitumor and Tregs. In preclinical studies, genetic ablation of the IL2/IL2R/STAT5 axis leads to life-threatening T cell–driven autoimmunity due to lack of functional Tregs (8, 13–15), indicating dominant IL2 control over Treg function and IL2 dispensability to control excessive T-cell activation. Thus, targeting the IL2R for Treg ablation has a clear rationale to boost antitumor immunity, although an effective strategy in that regard remains to be defined.
Here we tested the effect of IL2 complex comprised of IL2 and αIL2 (clone JES6-5H4), a S4B6-like clone, which selectively activated CD122 to improve antitumor effector functions while avoiding stimulating Tregs (20). IL2c are known to promote antitumor immunity by direct stimulation of effector T cells (18, 20–22, 39), but we provide much insight into additional mechanisms that can improve clinical translation and optimal use of IL2c and related agents. In contrast to studies in B16 melanoma, where tumor growth accelerates quickly after withdrawal of IL2c treatment (22) (Fig. 6), we show that ID8agg tumor burden remains stable for weeks following IL2c clearance, suggesting durably augmented antitumor immunity particularly for ovarian cancer, the peritoneal microenvironment or both, or could reflect insufficient time for effective memory development in rapidly growing B16 tumors. Such effects could be directly on antitumor T cells, or through reducing microenvironmental immunosuppression (such as reducing Treg suppression) and could involve special features of the microenvironment of the cells circulating there, an area requiring additional study. Our data suggest that both mechanisms are operative, and that IL2c promote long-lasting immunity to ovarian cancer, a tumor notoriously refractory to immunotherapy (43). IL2c reduces ovarian cancer microenvironmental Treg CD25 expression and suppressive functions, and simultaneously increases CD25 expression and IL2 responsiveness of effector T cells. Thus, IL2c could promote antitumor immunity by decreasing IL2 sequestration from antitumor T cells by Treg CD25 (7), inhibiting IL2R-mediated Treg functions (7), promoting IL2-driven antitumor effector cell functions (10) or some combination of these effects.
Multiple reports confirm that CD122-selective stimulation of the IL2R promotes expansion of CD8+ T cells and NK cells much greater than proliferation of Tregs (18, 22). However, none examined Treg functions in detail, despite the potential for IL2c effects on all IL2-dependent cells. We showed that IL2c reduced Treg effector molecules for a prolonged period after the final dose in the peritoneum. CD122-selective αIL2 antibodies block the CD25-binding epitope on IL2 (39), which appears to deprive Tregs of IL2 critical to differentiation and effector functions (6, 7). We further found that IL2c induced Treg fragility, which is associated with improved immune checkpoint blockade efficacy (37). As Treg fragility is induced by local IFNγ (37), IL2c-generated IFNγ in local T cells or NK cells could contribute to this effect. Alternatively, improved effector T-cell functions could promote sufficient inflammation to destabilize Treg differentiation secondarily in conjunction with IL2c and ablate their suppressive functions as seen in toxoplasmosis (48). Finally, we cannot exclude the possibility that IL2c selectively reduced a population of suppressive Tregs aside from inducing fragility. These issues and the basis for the site-specific effect of IL2c on Tregs merit additional investigations.
IL2c are thought to activate antitumor immunity in a CD25-independent manner, suggesting combination treatment with αCD25 could improve clinical responses. Unexpectedly, we found that αCD25 reduced IL2c efficacy. Thus, either high-affinity IL2 signaling via CD25 or the actions of CD25+ cells (which are depleted by αCD25) are crucial for optimal antitumor activities of IL2c. Recent work has shown that CD25 is responsible for intracellular recycling of IL2 and prolonging STAT5 activation when antitumor T cells leave IL2-rich TDLN and enter relatively IL2-poor peripheral tissues such as the tumor microenvironment (10). αCD25 depletion of newly formed CD25+ effector cells induced by IL2c treatment likely helps explain our finding of αCD25-mediated reduction in IL2c efficacy.
We also found other differences here versus prior studies of IL2c. For example, we found that IL2c promoted T-cell exhaustion markers, whereas others report little increase in these with a human IL2c construct (22), which could be from differences in details of the constructs, including actions of human versus mouse IL2 on mouse immune cells, the fact that we studied tumors later in their evolution and different timing of treatment initiation and interval until immune studies were performed. Nonetheless, our studies here accord with prior work showing improved CD8+ T-cell functions using similar IL2c constructs.
The use of combinations of multiple chemotherapy agents, radiation, surgery, and/or other specific targeted therapies is used to achieve optimal responses in patients with cancer. αPD-L1 (and other) immunotherapy works best in inflamed tumors with T cell–enriched TIL (49). We identified improving TIL numbers as another novel IL2c mechanism and thus tested the ability of IL2c to improve αPD-L1. We showed that in two distinct tumors (ID8agg ovarian cancer and B16 melanoma) in two distinct anatomic compartments (peritoneum and skin, respectively), αPD-L1, which alone at these doses and schedule had little effect on tumor growth, effectively improved tumor responses in each model. Furthermore, αPD-L1 was curative in the aggressive, poorly antigenic ID8agg model for ovarian cancer when combined with IL2c. Increased PD-1 expression on antitumor T cells could improve response to αPD-L1 (50). IL2c also augmented local T-cell expression of exhaustion markers including PD-1, although we and others (49, 51) have shown that some PD-1+ cells have significant effector functions after immunotherapy for ovarian cancer, including data presented here. IL2c differentially affected the intratumoral CD8/Treg ratio in the ovarian cancer versus melanoma models studied here, suggesting tumor or compartment effects of IL2c that merit additional investigations. IL2c induction of fragile Tregs, in which fragile Tregs promotes αPD-1 efficacy in other tumor models (37, 52) and could contribute to IL2c-mediated potentiation of αPD-L1 efficacy in these studies.
Our study has some limitations. We only tested IL2c effects in vivo in mice. However, there is significant homology between rodent and human IL2 and IL2R subunits confirmed in vivo by effectiveness of human IL2 in mouse models of melanoma (21), consistent with relevance of these data to humans. A potential obstacle in the translation of IL2c could be discordance in the distribution of high- or medium-affinity IL2R among effector and Tregs between patients with cancer and rodents (e.g., if human Tregs have a greater ability to bind CD122-selective IL2c) that could undermine the selectivity of IL2c to promote the strongest stimulation in effector T cells and thus mediate antitumor immunity. Targeting IL2 to medium-affinity IL2R with human IL2c was recently reported that replicated major mouse IL2c effects in murine cancer models (21). Nonetheless, no Treg effects were reported. There is additional preclinical evidence for the use of CD122-selective IL2 and related strategies to reduce IL2/CD25 interactions in clinical trials, including pegylated IL2, IL2 with the CD25-binding site genetically deleted and IL2 bound to CD25 (21, 53–55). Pegylated IL2 (NKTR214, bempegaldesleukin) shows promise in metastatic melanoma when combined with nivolumab (αPD-1; ref. 56), and just received FDA Breakthrough designation on that basis. On the basis of our data, these specific constructs could have differential effects on Tregs or other immune cells, a subject meriting additional studies. Our data support a trial of IL2c or similar agent in the treatment of ovarian cancer, especially in combination with αPD-L1.
Disclosure of Potential Conflicts of Interest
J.M. Drerup reports grants from NIH during the conduct of the study. J.R. Conejo-Garcia reports grants from NCI during the conduct of the study, grants, personal fees, and nonfinancial support from Compass Therapeutics, Anixa Bioscience, and personal fees from Leidos outside the submitted work; in addition, J.R. Conejo-Garcia has a patent for Anti-CD277 antibodies pending to Compass Therapeutics. No potential conflicts of interest were disclosed by the other authors.
J.M. Drerup: Conceptualization, data curation, formal analysis, validation, investigation, writing-original draft. Y. Deng: Formal analysis, validation, investigation, writing-review and editing. S.L. Pandeswara: Investigation. Á.S. Padrón: Investigation. R.M. Reyes: Investigation. X. Zhang: Investigation. J. Mendez: Investigation. A. Liu: Investigation. C.A. Clark: Data curation, investigation. W. Chen: Investigation. J.R. Conejo-Garcia: Conceptualization, validation. V. Hurez: Data curation, formal analysis, supervision, investigation. H. Gupta: Formal analysis, supervision, writing-review and editing. T.J. Curiel: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, methodology, writing-original draft, project administration, writing-review and editing.
This study was supported by grants from NCI (CA164122, CA054174, CA205965), the Ovarian Cancer Research Fund Alliance (290498), the Skinner endowment, The Holly Beach Public Library, The Owens Foundation, and The Barker Foundation to T.J. Curiel; the NCATS (TL1 TR001119) to J.M. Drerup, the NCI (CA205568) to C.A. Clark, and CPRIT (RP170345) and Ovarian Cancer Research Alliance to Y. Deng, J.R. Conejo-Garcia, and T.J. Curiel. We thank the UTHSA MD/PhD program for financial and administrative support to J.M. Drerup and C.A. Clark. We thank Shunhua Lao and June Deng for animal colony management.
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.