Abstract
Clinical responses to high-dose IL2 therapy are limited due to selective expansion of CD4+CD25+Foxp3+ T-regulatory cells (Treg), especially ICOS+ Tregs, rather than natural killer (NK) cells and effector T cells. These ICOS+ Tregs are highly suppressive and constitutively express high levels of IL2Rα (CD25) and CD39. Here, we characterized the effect of a mutant form of IL2 (F42K), which preferentially binds to the lower affinity IL2Rβγ with reduced binding to CD25, on Tregs, effector NK cells, and T-cell subsets. Unlike wild-type (WT) IL2, F42K did not efficiently induce the expansion of highly suppressive ICOS+ Tregs in peripheral blood mononuclear cells (PBMC) from healthy controls and melanoma patients. Instead, it promoted the expansion of CD16+CD56+ NK cells and CD56hiCD16− NK cell subsets in both short- and long-term cultures, with enhanced Bcl-2 expression. Stimulation of PBMCs with F42K induced expression of more NK cell activation molecules, such as NKp30, NKp44, DNAM-1, NKG2D, 4-1BB/CD137, and Tim-3, than WT IL2. F42K induced greater upregulation of TRAIL, and NK-mediated cytolytic activity was increased against both autologous and HLA-mismatched melanoma cells compared with WT IL2. Gene expression analysis revealed distinct gene expression profiles stimulated by F42K, WT IL2, and IL15. F42K therapy in vivo also induced a dramatic reduction in the expansion of ICOS+ Tregs, promoted NK cell expansion, and inhibited melanoma tumor growth more efficiently than WT IL2 and more effectively than anti–CTLA-4. Our findings suggest that F42K could be a potential substitute for WT IL2 as a cytokine therapy for cancer. Cancer Immunol Res; 4(11); 983–94. ©2016 AACR.
Introduction
IL2 is one of the most well studied cytokines since its initial discovery as “T cell growth factor” and later for its immunostimulatory effects on natural killer (NK) cells and NK T cells. IL2 belongs to the γc cytokine family and binds to the IL2 receptor, which is composed of the three subunits IL2Rα (CD25), IL2Rβ (CD122), and IL2Rγ (CD132). IL2 binds with different affinities, depending on the subunit composition, with binding to the IL2Rαβγ trimeric complex of greater affinity than binding to either a single IL2 receptor subunit or to the IL2Rβγ heterodimer (1–4). The IL2Rαβγ trimeric complex is expressed constitutively at high level by CD4+Foxp3+ regulatory T cells (Treg), especially highly suppressive ICOS+ Tregs, and some activated CD8+ T and NK cells, whereas the intermediate IL2Rβγ complex is constitutively expressed on effector T cells and CD16+CD56+ NK cells. The discovery that IL2 regulates the survival, proliferation, and differentiation of activated T and NK cells has led to the clinical development of IL2 therapy and was one of the first FDA-approved immunotherapies for metastatic melanoma and renal cell cancer (5–7). It is also the first successful cancer immunotherapy to demonstrate that modulating the self-immune system with cytokine therapy could completely eradicate tumor cells under certain conditions. This success has also led to extensive clinical studies for treatment of metastatic melanoma that combine IL2 with adoptive T-cell transfer (8–10), peptide vaccines (11), and chemotherapeutic agents (12).
Although IL2 therapy can induce long-lasting complete remissions in metastatic melanoma and renal cell carcinoma patients, these effects are observed only in about 5% to 6% of treated patients. For many years, the use of high-dose (HD) IL2 therapy (Proleukin, 720,000 IU/kg) has remained relatively limited due to its severe toxicities (13–16), such as the release of large amounts of proinflammatory cytokines, including IL1β, TNFα, IL6, and IFNγ (15, 16), and the direct binding of IL2 to CD25+ endothelial cells that induce acute vasodilation effects (14). Another key issue with IL2 therapy is the dual functions of IL2 that acts not only as a driver of effector lymphocyte responses, but at the same time it paradoxically drives the expansion and suppressive function of CD4+ Foxp3+ regulatory T cells (Treg). Evidence from many clinical studies suggests that both high- and low-dose IL2 therapy preferentially expand CD4+CD25+Foxp3+ Tregs (17–21), which remain elevated after each cycle of HD IL2 therapy (19). This is a major factor that compromises the therapeutic efficacy of IL2 and thereby limits its clinical application. Our previous data show that HD IL2 treatment induces expansion of a more activated and suppressive subset of Tregs that are CD45RA−/low and express higher levels of the inducible T-cell costimulator (ICOS) molecule (17, 22) and CD39 than ICOS− Treg subset.
Novel IL2-like cytokines are needed that do not induce cytokine storms and can circumvent the expansion as well as effect on Tregs while differentially activating NK and T effector cells. Attempts at improving IL2 have been developed by generating IL2 variants with altered IL2Rα binding domains, such as F42K and R38A (23–25), which greatly decreases their affinity for IL2Rα, while having a similar affinity as native IL2 for binding to the IL2Rβγ complex. However, the differential effects of how these IL2 variants on specific Treg subsets, effector T cells, and NK cells remain largely undefined. Based on the findings by Heaton and colleagues (23–25), we sought in this study to compare the effects of F42K and WT IL2 on the expansion of Tregs, NK cells, and T-cell subsets from healthy controls and patients with metastatic cancer. We also present preclinical findings on the therapeutic potential of F42K and on the differential gene expression profiles elicited by F42K, IL2, or IL15 in immune cells.
Materials and Methods
PBMC expansion and immunophenotyping analysis
Freshly isolated peripheral blood mononuclear cells (PBMC) from normal healthy controls (HC) and melanoma patients were seeded at 1 × 106 cell/mL in 24-well plate and were stimulated with WT IL2 (Proleukin; Prometheus Therapeutics and Diagnostics), F42K (provided by Dr. Elizabeth Grimm, PhD, The University of Texas MD Anderson Cancer Center) or IL15 (R&D Systems) at 0.4 nmol/L or otherwise indicated for 6 days. Cells were harvested, counted, and stained for Treg, T, and NK cell markers. The expansion of immune cell subsets was determined by combination of total cell count measurement and flow cytometry staining and was calculated by dividing the cell number obtained on day 6 by the cell number obtained on day 0. Cells were subjected to acquisition on a FACSCanto II flow cytometer (Becton Dickinson) and were analyzed using FlowJo software (Treestar).
Cloning of IL2
The full-length original IL2 gene amplified from human PBMCs was cloned into the commercial vector pcDNA3.1/V5-His-TOPO (Life Technologies) and IL2 mutant with single-site mutation of phenylalanine to lysine at position 42 in the mature pre-protein (F42K) was generated using the QuickChange II Site Directed mutagenesis kit (Agilent Technologies) according to manufacturer's instructions. Endotoxin-free plasmids encoding WT IL2 or F42K and the control empty pcDNA3.1/V5-His/lacZ plasmid were prepared and purified using the EndoFree Plasmid Mega kit (Qiagen).
Blood sampling and PMBC isolation
PBMCs from age-matched healthy controls (HC) and melanoma patients (Pt) were either freshly isolated or cryopreserved as described in a previous study (17). PBMCs from patients were collected 2 to 3 days after the last dose of HD IL2 treatment. All patients enrolled in this study provided informed consent, and the collection and use of patient blood samples for laboratory analysis were approved by the Institutional Review Board at The University of Texas MD Anderson Cancer Center (Protocol# LAB06-0762).
Microarray gene profiling analysis
PBMCs (2 × 106 cell/mL) from HC or melanoma patients were stimulated with WT IL2, F42K, or IL15 at 0.4 nmol/L in 24-well plate for 24 hours. Cells were harvested and total RNA were isolated using AllPrep DNA/RNA/miRNA Universal Kit (Qiagen) according to the manufacturer's instruction. The gene profiling process and analysis were described in Supplementary Information.
Caspase-3 cleavage tumor target killing assay
WT IL2- or F42K-treated PBMCs from melanoma patients for 6 days were assayed for cytotoxicity activity by measuring active caspase-3 in tumor target cells (26). Autologous or HLA-mismatched melanoma cell lines and PBMCs derived from patients were used for the cytotoxicity assay and the details were described in Supplementary Information.
Multiplex cytokine assays
PBMCs (2 × 106 cells) from HC and melanoma patients were plated and stimulated with equal concentration (0.4 nmol/L) of WT IL2 or F42K for 48 hours. Supernatants were collected and cytokines measured using the Luminex multiplex cytokine detection platform (Bio-Rad Laboratories).
Mouse tumor model studies
Eight- to 12-week-old female C57BL/6 mice were purchased from the National Cancer Institute, Bethesda, MD. All mice were maintained in a pathogen-free barrier facility at The University of Texas MD Anderson Cancer Center. All experiments and mice handling were performed according to the protocols approved by the Institutional Animal Care and Use Committee.
Mice were injected subcutaneously (s.c.) with 1 × 105 MCA205 fibrosarcoma cells or intravenously (i.v.) with 3 × 105 B16F10 murine melanoma cells on day 0. Mice were randomized and treated i.v. with 2 mL of WT IL2, F42K, or control vector plasmid in saline on day 7 and 14 using hydrodynamic gene transfer (HGT) technique (27, 28). For the MCA205 tumor model, tumor size was measured 2 to 3 times per week using digital calipers and was calculated by multiplying the length and width of each tumor. For lung metastasis model, mice were treated with plasmid DNA on day 3 with or without second dose on day 5. In some experiments, 100 μg of anti–CTLA-4 (clone 9H10) was injected intraperitoneally alone or with WT IL2, F42K, or control plasmids. Blood was collected and stained for immune cells. Mice were sacrificed, and lungs were removed and fixed in Fekete's solution (70% ethanol, 3.7% paraformaldehyde, 0.75 mol/Lglacial acetic acid) on day 17. Lung tumor nodules were counted under a binocular microscope.
Statistical analysis
Two-tailed Mann–Whitney U test, two-tailed Wilcoxon matched paired test, or one-way ANOVA was used to determine statistical significance (95% confidence interval), as indicated. A P value of < 0.05 was considered statistically significant.
Results
F42K is less potent than WT IL2 at inducing Treg expansion, but supported NK cells expansion
F42K has been shown to bind to IL2Rβγ and possessed reduced affinity for IL2Rα chain (23–25). In this study, we further investigate how F42K affects the expansion of Tregs, especially the highly suppressive ICOS+ Treg subset that correlates with clinical outcome during WT IL2 therapy (17). Freshly isolated PBMCs from HC were stimulated either with WT IL2 or F42K at 0.4 nmol/L (based on the initial dose response assays) for 6 days (Supplementary Fig. S1). Because accumulation of suppressive cells (e.g., Tregs) is typically found in patients with metastatic cancers, we performed our analysis also on primary immune cells from patients with advanced cancer to provide insight on how F42K modulates the immune compartment of these patients. We found that WT IL2 facilitated the expansion of highly suppressive CD39+CD127loICOS+ Tregs, but not effector T cells (Fig. 1A–C; Supplementary Figs. S2 and S3), more effectively than F42K in PBMCs from HC and melanoma patients. In contrast, F42K preferentially promoted more expansion of CD16+CD56+ cytolytic NK and CD56hiCD16− cytokine-secreting NK cell subsets than WT IL2 (Fig. 1A–C). F42K stimulation induced expansion of CD8+ T and CD4+Foxp3− cells as efficient as WT IL2 (Fig. 1B and C). By measuring the ratio of different effector cell subsets to ICOS+ Tregs, based on the absolute numbers obtained on day 6, F42K stimulation also skewed the PBMC populations toward a higher T effector:Treg ratio than WT IL2 (Fig. 1D). These biological effects of F42K suggest that it can efficiently drive NK cell expansion while circumventing the expansion of immunosuppressive ICOS+ Tregs better than WT IL2.
F42K is less potent than WT IL2 in supporting the expansion of Tregs but supports the expansion of NK cells. PBMCs (2 × 106) from healthy controls (HC) and patients with stage IV melanoma patients were treated with 0.4 nmol/L of WT IL2 or F42K. Cells were harvested and stained for Treg, T-cell, and NK cell markers on day 0 and 6 after stimulation. Dot and scatter plots depict the gating strategy (A) for the identification of immune cell subsets in response to F42K and WT IL2 from HC (B) and melanoma patients (C). D, graphs depict the ratio of effector to Treg cells upon stimulation. Horizontal bars represent median values, and nine independent experiments were performed. Pt, melanoma patients.
F42K supports expansion and long-term culture of NK cells and induces Bcl-2 expression
Next, we evaluated the potential of F42K to support longer term culture of Tregs and NK cell subsets. After 14 days of culture with F42K, we observed similar differences with a preferential expansion of both NK cell subsets but not ICOS+ and ICOS− Tregs (Fig. 2A and B). The ICOS+ Tregs in long-term culture consistently expressed low levels of CD127 when stimulated with WT IL2 or F42K, suggesting the cells were Tregs rather than T effector cells (Supplementary Fig. S3). The capacity of F42K to sustain long-term expansion of NK cell subsets was associated with increased cell survival and/or proliferation, with a significantly higher levels (P < 0.05) of Bcl-2 (a prosurvival marker) expression and higher percentage of Ki67+CD16+CD56+ and Ki67+CD56hiCD16− NK cells than WT IL2 (Fig. 2C and Supplementary Fig. S4). Both NK cell subsets in bulk PBMCs showed more expansion when treated with F42K than with WT IL2, whereas purified NK cells treated with either cytokine showed similar fold expansions. The fold expansion of ICOS+ Tregs was lower in both bulk PBMCs and sorted CD4+ T cells with F42K stimulation than with WT IL2 (Supplementary Fig. S5). Thus, the biological effect of F42K likely acts indirectly on NK cells by preventing expansion of Tregs.
F42K support long-term survival and expansion of NK cell subsets. PBMCs (2 × 106) from healthy controls (HC) and melanoma patients (Pt) were treated with 0.4 nmol/L WT IL2 or F42K. Cells were harvested and stained for Treg, T-cell, and NK cell markers on day 0 and 14 after stimulation. The fold expansion of immune cells subsets is indicated for PBMCs from HC (A) and melanoma patients (B). C (left), histograms from one representative melanoma patients and HC and scatter plots (right) show the Bcl-2 expression levels and median fluorescence intensity (MFI) for NK cell subsets after 18 days of WT IL2 or F42K stimulation, and six independent experiments were performed. Pt, melanoma patients.
F42K induced a more activated NK cell phenotype than WT IL2
Next, we evaluated the efficiency of F42K at upregulating NK cell activation markers and costimulatory molecules that are critical in mediating tumor cell recognition and killing by NK cells. After 3 days of stimulation, F42K induced a higher proportion of NK cells that expressed NKp30, NKp44, and CD137 than WT IL2, at different concentrations (Supplementary Fig. S6). Similarly, a greater percentage of CD16+CD56+ and CD56hiCD16− NK cell subsets in PBMC treated with F42K expressed significantly more (P < 0.05) NKp30, NKp44, and DNAM-1 than WT IL2 in HC and melanoma patients after 6 days of culture (Fig. 3A and B and Supplementary Fig. S7). The expression of NK cell–inhibitory KIR molecules CD158a and CD158b on NK cell subsets from HC and melanoma patients was induced to a similar extent (Supplementary Fig. S8). Like T cells, NK cells also express costimulatory molecules such as Tim-3 and 4-1BB (CD137) that regulate NK cell function in response to IL2 and IL15. Both F42K-treated PMBCs from HC and melanoma patients also showed a higher expression and percentage of Tim-3 and 41BB-expressing CD16+CD56+ and CD56hiCD16− NK cells than WT IL2 (Fig. 3C and D and Supplementary Fig. S7). Taken together, F42K induced a more activated and mature phenotype of both NK cell subsets compared with WT IL2.
Phenotype and activation status of NK cells upon F42K treatment. PBMCs (2 × 106) from healthy controls (HC) and melanoma patients (Pt) were treated with 0.4 nmol/L WT IL2 or F42K. Cells were harvested and phenotyped for NK cell-activation markers and costimulatory molecules on day 0 and 6 after stimulation. The MFI of Nkp30, Nkp44, NKG2D, and DNAM-1 of NK cell subsets are depicted in the scatter plots from HC (A) and melanoma patients (B). The MFI of CD137 (C) and TIM-3–expressing NK cell subsets (D) are shown in scatter plots. Horizontal bars represent median values, and 10 independent experiments were performed. Pt, melanoma patients.
Higher TRAIL expression and enhanced NK cell-mediated killing with F42K
The higher expression of NK cell activation molecules prompted us to further ask how F42K modulates the expression of cytolytic molecules and NK cell cytotoxic activity in PBMCs. F42K induced higher percentage of TRAIL-expressing CD16+CD56+ and CD56hiCD16− NK cells and sustained higher proportion of granulysin expression than WT IL2 (Fig. 4A). TRAIL and perforin expression in NK cell subsets was also significantly higher (P < 0.05) upon F42K stimulation than WT IL2 (Fig. 4B and Supplementary Fig. S9). The granzyme B expression level was higher in F42K-stimulated CD16+CD56+ and CD56hiCD16− NK cell subsets from melanoma patients but not from HC (Supplementary Fig. S9).
Functional properties of NK cells upon F42K stimulation. PBMCs (2 × 106) from healthy controls (HC) and melanoma patients (Pt) were treated with 0.4 nmol/L WT IL2 or F42K and were harvested and then stained for cytolytic markers on day 0 and 6 after stimulation. A, the percentage of CD56+CD16+ and CD56hiCD16−NK cells that expressed cytolytic molecules is shown in the scatter plots. B, the MFI of TRAIL by NK cell subsets in response to WT IL2 or F42K is depicted in (left) histograms from one representative patient and (right) in scatter plots. After 6 days of WT IL2 or F42K stimulation, PBMCs from melanoma patients were cocultured with autologous, HLA-matched, or mismatched tumor cell lines in the presence or absence of W6/32 blocking at an effector-to-target ratio of 25:1 for 2 hours. Ten independent experiments were performed. C, the percentages of caspase-3 expressing tumor cells are shown in the graphs. Horizontal bars represent median values, and three independent experiments were performed. Pt, melanoma patients.
The higher proportion and expression of cytolytic mediators per cell seen in F42K- stimulated NK cells led us to further examine the cytolytic function of F42K-stimulated PBMCs from melanoma patients against autologous or HLA-mismatched melanoma tumor targets. Consistent with the higher TRAIL and perforin expression by CD16+CD56+ and CD56hiCD16− NK cells, F42K-treated PBMCs from melanoma patients had increased cytolytic activity against both autologous and HLA-mismatched melanoma cell lines than WT IL2 (Fig. 4C). By blocking HLA class I molecule, F42K stimulation also induced higher caspase-3 cleavage in the tumor cells than WT IL2 (Fig. 4C). Thus, F42K induced significantly higher (P < 0.05) expression of cytolytic molecules and promoted more NK-mediated cytolytic function, than WT IL2.
Cytokine profiles induced by F42K and WT IL2
One of the major side effects of IL2 therapy is induction of cytokine storms by promoting the secretion of large amounts of proinflammatory cytokines, such as IL6, IL1β, IFNγ, and TNFα (13). F42K stimulation has been shown to induce a lower amount of proinflammatory cytokines (24). We examined the cytokine secretion profile of PBMCs from HC and melanoma patients upon F42K or WT IL2 stimulation using a more extended panel of inflammatory and immunosuppressive cytokines not studied before with IL2 variants. PBMCs from HC and melanoma patients treated with F42K tended to produce less proinflammatory IFNγ and TNFα than WT IL2 (Fig. 5). In addition, the secretion of immunosuppressive cytokines, including IL4, IL5, IL10, and IL13, was significantly lower (P < 0.05) in PBMCs treated with F42K than WT IL2 in melanoma patients samples (Fig. 5). No substantial IL4, IL5, IL10 and IL13 were detected in PBMCs from HC upon F42K or WT IL2 stimulation.
Cytokine production induced by F42K. Cytokine secretion by PBMCs (2 × 106) stimulated with WT IL2 or F42K (0.4 nmol/L) from healthy controls (HC) and melanoma patients was measured using the Luminex multiplex cytokine assay. Cell cultured supernatant was harvested after 48 hours of WT IL2 or F42K stimulation. One independent experiment was performed. Pt, melanoma patients.
Greater in vivo antitumor effects with F42K than WT IL2
To investigate the in vivo effect of F42K versus WT IL2 against tumors, we generated F42K- and WT IL2-encoding plasmids that could be delivered in vivo using an HGT approach (27, 28). Upon plasmid injection, we detected in vivo expression of both soluble WT IL2 and F42K in serum of mice as early as 1 day after treatment (Supplementary Fig. S10). The F42K plasmid (up to 10 μg) treatment was well tolerated and did not induce significant toxicity, no lung edema and other motor function or observable defects in mice (data not shown). Single dose of WT IL2 but not F42K treatment induced a dramatic increase of CD4+CD25+Foxp3+ICOS+ Tregs, whereas there was a trend of higher levels of NK cells in F42K treatment than WT IL2 on days 3, 6, and 9 (Supplementary Fig. S11). F42K was more effective than WT IL2 and control vector in inhibiting MCA205 tumor growth and in controlling B16F10 melanoma lung metastases with significant (P < 0.05) reduction of lung tumor nodules in mice treated with F42K (Fig. 6A and B). Because F42K did not induce Treg expansion (especially highly suppressive ICOS+ Tregs), we also compared its effects against melanoma to anti–CTLA-4, which has been demonstrated to deplete Tregs in melanoma mouse models (29). F42K alone is superior to anti–CTLA-4 monotherapy in inhibiting lung metastases. Furthermore, F42K and anti–CTLA-4 combination did not further enhance the initial superior efficacy of F42K alone over WT IL2 in controlling metastasis. The combination of anti–CTLA-4 with WT IL2 induced moderate increased in controlling metastasis compared with control IgG or anti–CTLA-4 monotherapy (Fig. 6C; ref. 17). The percentage of bulk CD4+CD25+Foxp3+ Tregs, highly activated CD4+CD25+Foxp3+ICOS+ Tregs, and GITR+Lag3+ICOS+OX40+ Treg subsets was markedly lower in blood after two doses of F42K treatment alone or together with anti–CTLA-4, than in mice that received WT IL2 alone or with anti–CTLA-4 (Fig. 6D). F42K also preferentially promoted the expansion of NK cells and CD8+ T cells with a higher frequency of CD3−NK1.1+DX5+ NK cells and CD3+CD8+ T cells found in the blood than WT IL2 or control vector treatment (Fig. 6E). The biological effect of F42K found in vivo is consistent with our in vitro data, suggesting that F42K has the potential to generate a more effective therapeutic antitumor effector response by circumventing the expansion of highly activated and suppressive Treg subsets.
F42K exhibited greater antitumor effect than WT IL2. A, C57BL/6 mice were injected s.c. with 1 × 105 murine MCA205 fibrosarcoma cells on day 0 and were treated with 2 doses of 10 μg WT IL2 or F42K encoding plasmid or control plasmid on days 7 and 14 using HGT. Tumor sizes were shown in the graph. B and C, mice received tail-vein i.v. injection with 3 × 105 B16F10 murine melanoma cells on day 0 and were treated with a single dose of WT IL2, F42K-encoding plasmid, or control plasmid at 10 μg. C, mice also received control IgG, anti–CTLA-4 at 3- to 4-day interval in combination with two doses of WT IL2, F42K, or control plasmid on days 3 and 5 after tumor injection. Mice were sacrificed on day 17, and lung tumor nodules were counted under binocular microscope. Blood was collected 3 or 9 days after first treatment and stained for Treg (D), T-cell (E), and NK cell markers. Horizontal bars represent median values, and three independent experiments were performed.
Gene expression profiling of PBMCs treated with WT IL2, F42K, or IL15
We evaluated how F42K altered the gene expression profile of PBMCs in comparison with WT IL2 to provide more mechanistic insight into these differential effects of F42K. We included IL15 in these experiments, as it also preferentially binds to the IL2Rβγ complex and not IL2Rα/CD25. Principal component analysis (PCA) showed a clear segregation of each treatment group and a gene cluster with F42K treatment was found in between WT IL2 and IL15 gene clusters with close proximity to WT IL2 cluster (Fig. 7A). IL15-treated PBMCs showed a markedly larger number of differentially expressed genes (4,449 genes, P < 0.005) than WT IL2, underlying the distinctness of the two cytokines (Fig. 7B). F42K treatment rendered 374 significant (P < 0.05) differentially expressed genes compared with WT IL2 of which 288 genes overlapped to genes modulated by IL15. A total of 2,206 distinct genes were found upregulated or downregulated when comparing F42K versus IL15 treatment (Fig. 7B). Supervised hierarchical clustering of the 374 differentially regulated genes (P < 0.005) between WT IL2 and F42K treatment revealed distinct upregulated or downregulated gene clusters that were modulated by F42K (Fig. 7C). Cluster visualization showed that the F42K-modulated gene profile had a similar upregulation or downregulation pattern as IL15, with a majority of the genes upregulated rather than downregulated (Fig. 7C). Gene ontology enrichment analysis based on the 374 differentially expressed genes induced by F42K in comparison with WT IL2 revealed a predominant upregulated immune gene signature related to regulation of immune response, immune effector processes, and antigen processing and presentation (Fig. 7D). Among those immune function-related genes, IL2Rα, HMGB1, IL1R1, IL13, TNFRSF21, NRG1, F2RL1, THBS1, MERTK, and CD93 were significantly (P < 0.05) downregulated in F42K-treated PBMCs relative to IL2 treatment (Fig. 7E). In addition, NKp30 (NCR3) and perforin (PRF1) genes were among the immune response genes significantly (P < 0.05) upregulated in F42K-treated PBMCs compared with WT IL2 (Fig. 7E and Supplementary Table S1).
Gene expression profile induced by WT IL2, F42K, or IL15. PBMCs (2 × 106) from healthy controls (HC) were treated with 0.4 nmol/L of WT IL2, F42K, or IL15 for 24 hours. Cells were harvested, and total RNA were isolated for Affymetrix ST 1.0 microarray gene profiling analysis. A, PCA graph of global gene expression data shows that each sample is representative of a single color-coded sphere and each color code corresponds to WT IL2 (green), F42K (red), and IL15 (blue) stimulation. B, Venn diagram shows the number of overlapping and complementary genes that were differentially expressed with P < 0.005 in the comparison between F42K, WT IL2, and IL15 stimulation. C, heat map with hierarchical clustering comparing the log2 fold changes for genes significantly differed in WT IL2-, F42K-, and IL15-treated PBMCs. D, forest plot analysis shows upregulated or downregulated genes induced by F42K in comparison with WT IL2 stimulation based on their gene ontology enrichment. E, heat map shows immune response genes that were differentially regulated in response to F42K as compared with WT IL2 at P < 0.005. One independent experiment was performed.
Discussion
WT IL2 therapy is a relatively potent immunotherapy. However, induction of Treg expansion caused by the preferential binding of IL2 to the high-affinity IL2R, and toxicities induced by high proinflammatory cytokine secretion, have limited the broad application of IL2 in the clinic. In this study, we characterized the effects of F42K, an IL2 mutant that has been shown by others to have reduced affinity to IL2Rα (23, 25), on different subsets of immune cells and assessed its therapeutic potential in generating antitumor responses and controlling tumor growth and metastasis in comparison with WT IL2. Although F42K has been studied previously for its effect on LAK cells (23–25), it is not known how it affects different subpopulations of Tregs, NK, and effector cells and how it regulates antitumor responses. Several IL2 variants have similar reduced binding affinity to CD25 and facilitate antitumor responses in vivo (23–25, 30–32). However, these studies have not addressed how those IL2 mutants affect the balance between specific Treg and NK cell subsets in human PBMCs, especially in patients with metastatic cancer. We characterized the effect of F42K on Treg subsets, including ICOS+ and ICOS− Tregs, NK cell subsets, and T cells, not only from HC but also those from advanced cancer patients. We also compared the effects of IL2 variant to IL15 and WT IL2 on gene expression. As far as we know, no gene expression data has compared the effects of IL2 variants to IL15 and WT IL2 have been presented so far.
Our results have revealed the biological effects of F42K, which supports short- and long-term expansion and survival of both human CD16+CD56+ and CD56hiCD16− NK cell subsets, while circumventing immunosuppression. The balance in the immune compartment shifts toward high effector to Treg ratios in both HC and patients. The biological effects of F42K on Tregs imply that it could also circumvent the activation of IL2Rα-expressing endothelial cells that mediate WT IL2 toxicity (14). Additionally, F42K induced significantly less (P < 0.05) proinflammatory and immunosuppressive cytokines than WT IL2 in vitro and was nontoxic at the doses tested in vivo using HGT. Our findings add to our knowledge on the effects of F42K, in comparison with WT IL2 on Tregs and NK cell subsets from both HC and melanoma patients. Although HD IL2 therapy activates and expands effector T and NK cells, IL2 increases the numbers of circulating Tregs even more, which potently suppress activation and proliferation of effector CD4+, CD8+ T, and NK cells, rather than inducing antitumor responses (17, 33, 34). Indeed, failure of HD IL2 therapy in large numbers of patients is attributed to expansion of CD4+CD25+Foxp3+ Tregs, particularly the highly suppressive ICOS+ Treg subset that we have shown to correlate with IL2 therapy outcome (17, 18). F42K, in contrast, limited the expansion of ICOS+ Treg subset, the major Treg subset in the tumor microenvironment of melanoma (35, 36). ICOS+ Tregs are activated and more suppressive than ICOS− Tregs, with memory phenotype and coexpressing low level of CD127 but with high levels of CD25, Foxp3, PD-1, CD39, TGF-β, and IL10 compared with the bulk Treg population. ICOS+ Tregs may use a wider variety of suppressive mechanisms, including ATP depletion and adenosine generation through the high expression of the CD39/CD73 axis (37). Thus, the ability of F42K to circumvent the expansion of these highly suppressive ICOS+ Tregs is a previously unknown feature of this IL2 variant that can have important impact for improving IL2 therapy.
Increasing evidence demonstrates a significant role of NK cells in tumor immunosurveillance and in preventing metastases in melanoma and other cancer types (38, 39). NKp30 and NKp44 have critical roles in facilitating melanoma recognition by NK cells, indicating that a higher expression of these molecules induced by F42K is advantageous (39). A higher gene and protein expression of NKp30 induced by F42K may further enhance NK–dendritic cell (DC) interaction and DC maturation via the NKp30/NKp30L axis (40, 41). It could be advantageous that F42K maintained a larger proportion of both NK cell subsets that expressed higher TIM-3 and 4-1BB expression than IL2. The expression of 4-1BB by NK cells is key in enhancing antibody-dependent cytolytic function of NK cells in the presence of agonistic 41BB antibody (42–44). TIM-3 expression is known to be an “exhaustion” marker on T cells, but it is associated with a more mature NK phenotype and may induce activation rather than an inhibitory function in NK cells.
Although F42K binds similarly to IL15 to the IL2Rβγ heterodimer complex and shares some unique gene expression patterns, F42K and IL15 induce a distinct overall gene expression profile in PBMCs of HC and patients. These differences may be attributed due to IL15 being trans-presented by IL15Rα expressed on DC and monocytes/macrophages, whereas F42K is not trans-presented to the βγ receptor complex. The trans-presentation of IL15 to IL15Rα induces different dynamics of receptor occupancy and activation leading to stronger and longer-lasting downstream signaling events compared with soluble cytokine binding (46). Although it might be advantageous that IL15 can be trans-presented in the in vivo setting, one potential caveat is the availability of DC and macrophages for IL15 trans-presentation as cancer patients frequently have reduced DC numbers and dysregulated DC and macrophage functions. Also, both IL15Rα and IL2Rα are shed via proteolytic cleavage by matrix metalloproteinases and can act as endogenous antagonists inhibiting IL15 and IL2 activity, respectively (47, 48). Therefore, F42K may have some distinct advantages in cancer immunotherapy through its interaction with IL2Rβγ in a non-complexed fashion without the need of trans-presentation and the requirements of accessory DC, while also bypassing the potential adverse effects posed by soluble IL15Rα and IL2Rα mentioned above.
In addition, F42K has therapeutic potential for cytokine therapy for cancer with an increased capacity of F42K over WT IL2 to prevent in vivo melanoma lung metastasis development, a key role of NK cells in tumor immunosurveillance. Similar F42K-mediated effects in the human system were recapitulated in our in vivo mouse tumor models where F42K circumvented the expansion of activated Treg subsets, particularly ICOS+ Tregs, without depleting the endogenous Treg pool, and was associated with activation and expansion of NK and T cells. HGT is an efficient way to induce production of cytokines by hepatocytes and to allow assessment of in vivo activity of cytokines. One caveat with the HGT approach is its limitation on frequent, repeated, multiple tail-vein injections of plasmid to induce higher systemic cytokine levels to assess toxicity that is achievable using recombinant protein injection. However, HGT allowed us to investigate efficiently the biological effect of F42K in vivo when a large numbers of purified recombinant cytokines were not available due to the expense and difficulty in manufacturing. Although F42K concentrations in the serum were slightly higher than in WT IL2, the difference was not statistically significant (P < 0.05), and the percentage of ICOS+ Tregs induced by F42K remained at least 3-fold lower than WT IL2. The slightly lower concentrations of WT IL2 in the serum could be due to several factors, including longer half-life of F42K than WT IL2 and/or cytokine sink effect due to WT IL2 consumptions by CD25-expressing Tregs. Although our in vivo studies revealed no synergistic effect of the combination of F42K with anti–CTLA-4, our findings indicated that F42K alone was more efficient than WT IL2 and was more efficient than anti–CTLA-4 in controlling lung metastasis. This is a critical point considering the toxicity of anti–CTLA-4 therapy and the demonstrated mechanism of action of anti–CTLA-4 in depleting Treg in patients and mouse models (29). F42K has a similar mechanism of action by leaving Tregs “un-touched” while expanding effector NK cells and would significantly increase (P < 0.05) the NK cell-to-Treg ratio. Moreover, unlike anti–CTLA-4, F42K would not cause an indiscriminant depletion of natural (unactivated) Tregs, which are also needed in preventing autoimmune reactions in patients.
Taken together, the data presented here suggest that F42K can circumvent the expansion and negative immunoregulatory effects of highly suppressive ICOS+ Tregs, while promoting NK cell expansion and function. F42K also induces a unique gene expression profile and does not activate many IL2-induced genes, although it has the capacity to activate NK cells and NK cell–associated activation genes, costimulatory molecules, and NK-mediated cytolytic function. As such, it may serve as a way of activating and maintaining NK cell function in cancer patients to prevent metastasis and as immunosurveillance enhancer in a nontoxic way, avoiding Treg expansion during long-term therapy. F42K may serve as a unique niche to overcome defects in NK cell function in advanced cancer patients through its selectivity for NK cell activation.
Disclosure of Potential Conflicts of Interest
W.W. Overwijk is a consultant at GLG Consulting; reports receiving a commercial research grant from Nektar, Inc., 7 Hills Pharmaceuticals, and Immatics, Inc.; and is a consultant/advisory board member for Immatics, Inc. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: G.C. Sim, P. Hwu, E. Grimm, L. Radvanyi
Development of methodology: G.C. Sim, C. Liu, E. Grimm, L. Radvanyi
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G.C. Sim, C. Liu, E. Wang, C. Creasy, Z. Dai, W.W. Overwijk, E. Grimm
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G.C. Sim, E. Wang, Z. Dai, J. Roszik, F. Marincola, L. Radvanyi
Writing, review, and/or revision of the manuscript: G.C. Sim, E. Wang, W.W. Overwijk, P. Hwu, L. Radvanyi
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G.C. Sim, C. Liu, H. Liu, P. Hwu
Study supervision: E. Grimm, L. Radvanyi
Grant Support
This work was supported by The University of Texas MD Anderson Cancer Center SPORE in MelanomaP50 CA093459funded from the NCI (to E. Grimm and G.C. Sim), the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation (to L. Radvanyi), and The University of Texas MD Anderson Cancer CenterP30 CA016627CCSG grant (to UTMDACC Flow Cytometry Core Facility and Immunomonitoring Core Facility).
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.
Acknowledgments
We thank Dr. Karen Dwyer and her team from the Flow Cytometry Core Facility and Immunomonitoring Core Lab of The University of Texas MD Anderson Cancer Center for their support. We are also grateful for the hard work in acquiring patient samples by the research nurses at the Melanoma Medical Oncology Department, in particular Edwina Washington and Amber Richardson. We also thank the support from Dr. Nallaparaju Kalyan of the Lion Biotechnologies and Dr. Richard Wu Cheng-Han of The University of Texas MD Anderson Cancer Center.
Footnotes
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).
- Received August 9, 2015.
- Revision received August 22, 2016.
- Accepted August 23, 2016.
- ©2016 American Association for Cancer Research.