Abstract
Antibodies targeting “immune checkpoints” have revolutionized cancer therapy by reactivating tumor-resident cytotoxic lymphocytes, primarily CD8+ T cells. Interest in targeting analogous pathways in other cytotoxic lymphocytes is growing. Natural killer (NK) cells are key to cancer immunosurveillance by eradicating metastases and driving solid tumor inflammation. NK-cell antitumor function is dependent on the cytokine IL15. Ablation of the IL15 signaling inhibitor CIS (Cish) enhances NK-cell antitumor immunity by increasing NK-cell metabolism and persistence within the tumor microenvironment (TME). The TME has also been shown to impair NK-cell fitness via the production of immunosuppressive transforming growth factor β (TGFβ), a suppression which occurs even in the presence of high IL15 signaling. Here, we identified an unexpected interaction between CIS and the TGFβ signaling pathway in NK cells. Independently, Cish- and Tgfbr2-deficient NK cells are both hyperresponsive to IL15 and hyporesponsive to TGFβ, with dramatically enhanced antitumor immunity. Remarkably, when both these immunosuppressive genes are simultaneously deleted in NK cells, mice are largely resistant to tumor development, suggesting that combining suppression of these two pathways might represent a novel therapeutic strategy to enhance innate anticancer immunity.
Introduction
Immunotherapies that reengage key immune effectors, such as T cells, to provide them with an enhanced ability to recognize and kill tumor cells are improving outcomes for patients with cancer. Yet, T-cell therapies have a high risk of graft-versus-host disease (GvHD), cytokine release syndrome (CRS), and autoimmune syndromes, are challenging to manufacture as an “off-the-self” product and have a limited ability to infiltrate solid tumors (1). Natural killer (NK) cells offer several advantages over T cells. They are highly adept at killing cancer cells, especially tumor metastases, promote solid tumor inflammation by recruiting cDC1 and CD8+ T cells, and due to their low risk of GvHD/CRS, they can also be manufactured as an “off-the-shelf” allogeneic cell therapy (2). The subsequent integration of novel and clinically proven chimeric antigen receptors (CAR) also allows the generation of allogeneic CAR-NK cells with the ability to target tumor antigens with the added advantages of enhanced recognition of stress ligands and no inhibition via self MHC in the allogeneic setting, thus maximizing the potential of eliminating recurrent and resistant cancers. The intracellular checkpoint CIS has been demonstrated to limit the responses of CAR-NK cells, and its suppression enhances antitumor responses in models of B-cell leukemia (3). In addition, TGFβ signaling was also described by us (as reviewed in ref. 1) and others (4) as a critical inhibitory pathway of NK-cell functions. More recently, the suppression of TGFβ signaling was also shown as an approach to enhance maximal CAR-NK cell antitumor responses in models of glioblastoma (5).
NK cells are emerging as an attractive target cell for cancer immunotherapy due to their ability to directly lyse tumor cells without the need of prior sensitization. Targeting NK cells also has a reduced risk of inducing autoimmune syndromes compared with current T cell–targeting agents. NK cells have been shown to be particularly effective at restraining hematogenous spread of tumor metastases in animal models (1). Clinically, NK-cell activity is inversely correlated with cancer incidence in multiple cancer types [as revisited in The Cancer Genome Atlas cohorts for bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), head and neck squamous cell carcinoma (HNSC), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), melanoma, etc. (2, 6)], and there is now emerging evidence that NK-cell infiltration in human tumors is associated with better prognosis.
Immunosuppressive factors within the tumor microenvironment (TME) can impair the antitumor responses of NK cells (1). We have previously identified TGFβ as an inhibitory checkpoint of IL15-induced activation of mTOR in NK cells, resulting in metabolic blockade, upregulation of tissue residency–like characteristics, and loss of antitumor effector function (7–9). In addition, the SOCS protein CIS (encoded by Cish) was discovered by our group as an inhibitory checkpoint of IL15 signaling in NK cells. Cish-deficient NK cells are hyperresponsive to IL15, with dramatically enhanced survival, proliferation, proinflammatory cytokine production, and cytotoxicity. As a result, Cish-deficient mice are largely resistant to tumor initiation and metastasis in multiple models, and this resistance is dependent on the hyperactivity of their NK cells (10, 11). Other studies have further revealed a metabolic and effector function benefit by deleting Cish in human inducible pluripotent stem cell–derived NK cells or cord blood–derived CD19-CAR NK cells (3, 12). In our current study, we investigate whether simultaneous immune checkpoint suppression of TGFβ and CIS signaling could synergistically increase the magnitude of NK-cell effector functions against a variety of cancer scenarios.
Material and Methods
Mice models
NKp46cre/wtTGFβRIIfl/fl mice were used for TGFβRII-deficient NK cells, obtained by crossing, as described previously (7–9). Wild-type (WT) NK cells were isolated from the respective littermate controls (NKp46wt/wtTGFβRIIfl/fl). Cish−/− were generously provided by J. Ihle and E. Parganas (St. Jude Children's Research Hospital) and were maintained on a C57BL/6 background. To obtain a double-deficient mouse model, NKp46cre/wtTGFβRIIfl/fl were back crossed with Cish−/− to obtain a Cish−/− NKp46cre/wtTGFβRIIfl/fl mouse strain. All experiments were performed using cells from age- and sex-matched cohort of mice (age range, 8–12 weeks). Cohort sizes were described in each figure legends to achieve statistical significance. No biological replicate was excluded on the basis of preestablished criteria. Experiments were approved by the Walter and Eliza Hall Institute (WEHI), University of Queensland (Queensland, Australia), and Monash University's Animal Ethics Committees.
NK-cell isolation and culture
NK cells from mouse spleens were isolated by organ maceration, followed by negative selection using a mouse NK-cell isolation kit (Miltenyi Biotec, #130-115-818). NK cells were maintained in RPMI1640 media (Gibco, #11875093) supplemented with 10% FBS (HyClone, #SH30084.03), 1% sodium pyruvate (Gibco, #11360070), 1% Glutamax (Gibco, #35050061), 1% nonessential amino acids (NEAA; Gibco, #11140050), 10 mmol/L HEPES (Gibco, #15630080), 0.1% 2-mercaptoethanol (Gibco, #21985023), 1% penicillin-streptomycin (Gibco, #15240062), and 10 or 50 ng/mL of human recombinant (r)IL15 (PeproTech,#200-15). NK cells were maintained in a humidified incubator at 37°C and 5% CO2.
Preparation for transcriptomics
Single-cell suspensions from indicated organs were prepared for transcriptomics analyses as previously described by our group (9, 11). A total of 106 NK cells per replicate or culture condition were prepared for RNA sequencing. RNA was extracted from sorted ex vivo NK cells using the RNeasy Plus Mini Kit (#74134, QIAGEN), according to the manufacturer's instructions. Purified RNA was measured using an Agilent 2200 TapeStation System (Agilent) with High Sensitivity RNA ScreenTapes (Agilent, #5067-5579,). Next-generation sequencing libraries were generated from 100 ng of high-quality RNA samples with clear 18S and 28S peaks and RNA integrity number values ≥ 9, using the NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs, #E7760L) for Illumina, according to the manufacturer's instructions. Approximately 20 million reads per sample were obtained by pooling RNA libraries and performing single-end 75 bp sequencing. This was performed by the Genomics Laboratory at WEHI on a NextSeq 500 next-generation sequencer (Illumina).
Bioinformatics
Quality of the fastQC files was examined, and trimmomatic was used to trim any bases with sequencing quality less than 20 bp from the end of the reads and to remove Illumina adapters. Only reads larger than 20 bp in length after trimming were kept. Reads were then aligned to mm10 using subread (13). Duplicate reads in the resulting BAM files were marked using picard tools. The aligned reads were sorted and indexed in samtools (14). Gene count quantification was performed using the featureCounts.
Low expressed genes were filtered such that genes with count per million of at least 1 in at least three samples were retained for downstream analyses. The quality of the data was checked by calculating and visualizing the Poisson distance, relative log expression, principal component analysis, and multidimensional scaling. The Poisson distance between samples was calculated using the full matrix of gene counts and the PoissonDistance function from the PoiClaClu R package.
The voom-limma pipeline was used to perform differential expression analysis (15), and voomWithQualityWeightsand duplicateCorrelation functions was used to determine blocking variable of the replicates in each condition (16). Gene ontology and Kyoto Encyclopedia of Genes and Genomes pathway analyses were performed using the goana and kegga functions, and gene set testing were performed using the fry and camera gene set tests from the limma package. The ComplexHeatmap package was used to visualise the results of gene-set test as well as the gene expression values (17). All the analyses were performed in R versions ≥ 3.6.3 and Bioconductor versions ≥ 3.10. General data wrangling and visualizations were performed using base R functions and several core tidyverse packages (mainly dplyr, tidyr, and ggplot2).
Total cell number and mean division number determination
Cell numbers were enumerated as published previously (9). For assessment of cellular proliferation by carboxy fluorescein diacetate succinimidyl diester (CFSE; Invitrogen, #C34554), splenic NK cells were labeled with 0.1 μmol/L CFSE and cultured with various doses of hIL15 and 6.25 ng/mL of recombinant (r)TGFβ1 (Peprotech, #100-21) and indicated culture conditions for 5 days, before flow cytometric analysis. Absolute cell counting was performed by adding 123count eBeads (Invitrogen, #01-1234-42) to single-cell suspensions prior to flow cytometry.
Seahorse assays
Freshly isolated NK cells were cultured with 50 ng/mL of rIL15 with or without 6.25 ng/mL of rTGFβ1. After 72 hours, the cells were washed three times with PBS, resuspended in 180 μL of Seahorse XF RPMI Media (unbuffered, glucose free; Agilent, #103576-100), and transferred to 0.5% gelatin (Sigma-Aldrich, #G1890-100G)-coated seahorse plates (Agilent, #103794-100). The cells were then stimulated with 20 μL of the same media with a final concentration of oligomycin (Sigma-Aldrich, #75351) at 1.2 μmol/L, FCCP (Sigma-Aldrich, #C2920) at 1.5 mol/L, and antimycin A (Sigma-Aldrich, #A8675) at 1 μmol/L plus rotenone (Sigma-Aldrich, #R8875) at 1 μmol/L. Oxygen consumption rates (OCR) were measured every 7 minutes using a Seahorse XFe96 analyzer (Agilent).
Tumor models
B16F10 melanoma and MC38 colorectal cancer cells were obtained from ATCC and were maintained in DMEM (Gibco, #11960044) supplemented with 10% FBS, 1% penicillin-streptomycin, 1% Glutamax, 1% NEAA, and 1% sodium pyruvate. SM1WT1-LWT1 melanoma cells (18) were generously provided by M. Smyth (QIMR Berghofer Medical Research Institute) and were cultured in the same condition as described above. Both cell lines were maintained in a humidified incubator at 37°C and 5% CO2 and used with flask confluence not higher than 80%. All cell lines were tested for Mycoplasma contamination by PCR using standard Mycoplasma testing protocol. A total of 2 × 105 B16F10 or 5 × 105 SM1WT1-LWT1 cells were injected into the tail vein of indicated mice strains. SM1WT1-LWT1–bearing mice were treated intraperitoneally either with 500 μg/mice of the neutralizing TGFβ antibody (clone 1D11.16.8, BioXCell) alone or in combination with the specific BRAFV600E inhibitor PLX4720 (ref. 17; Plexxikon Inc). B16F10 or SM1WT1-LWT1–bearing mice were sacrificed, and lungs were harvested on day 14. Lungs from SM1WT1-LWT1–bearing mice were injected with Indian Ink (Speedball, #3338) and washed twice in PBS. Lungs from both models were fixed in Fekete solution (700 mL 100% ethanol, 32 mL 37% formaldehyde solution, 40 mL acetic acid, and 228 mL dH2O) overnight to count metastases as described previously (18). A total of 3 × 105 MC38 cells were injected subcutaneously in the indicated strains of mice, and tumor growth was measured using a digital caliper.
FACS analysis
At the indicated timepoints, primary MC38 tumors were cut into small pieces with a surgical blade and digested with DMEM containing 1 mg/mL of collagenase IV (Worthington Biochem, #LS004186) and 20 μg/mL of DNAse I (Roche, #4716728001) at 37°C for 1 hour, before being passed through a 100-μm cell strainer. Leukocytes were enriched using 37.5% Percoll solution (GE Healthcare, #P1644) and red blood cells lysed with ammonium-chloride-potassium (ACK) lysis buffer (BioLegend, #420302). Dead cells were stained with Fixable Viability Stain 440UV (1:1,000 in PBS; Becton Dickinson, #566332) for 15 minutes at room temperature. Fc receptors were blocked by incubation for 15 minutes in mouse Fc Blocking Reagent (1:100 in FACS buffer; Miltenyi Biotec, #130-092-575). Single-cell suspensions were stained with the indicated fluorescent antibodies on ice for 45 minutes in the presence of 10% Brilliant Stain Buffer Plus (Becton Dickinson, #566385). For intracellular cytokine staining, cells were fixed and permeabilized using the FoxP3/Transcription Factor Staining Buffer Set (eBioscience, # 00-5523-00), then stained for 60 minutes with the indicated fluorescent antibodies. Antibodies targeting NK1.1-BUV395 (PK136), CD45-BUV563 (30-F11), CD4-BUV496 (GK1.5), CD3-BUV737 (145-2C11), CD8a-BUV805 (53-6.7), TIGIT-BV421 (1G9), CD69-BV480 (H1.2F3), CD223/Lag-3-BV750 (C9B7W), CD49b-BV786 (HMa2), CD11b-BUV661 (M1/70), CD44-APC-Cy7 (IM7), PD-1-BB700 (J43), CD49a-PE (HA31/8) were purchased from Becton Dickinson. Antibodies targeting CD226-BV650 (TX42.1), KLRG1-BV605 (2F1/KLRG1), CD19-Biotin (6D5), Ly6G-Biotin (1A8), F4/80-Biotin (BM8), and CD62L-APC-R700 (MEL-14) were purchased from BioLegend. Antibodies targeting Tim-3-FITC (RMT3-23), Eomes-PE-Cy7 (Dan11mag), and FoxP3-PE-Cy5 (FJK-16S) were purchased from eBioscience (San Diego CA). Samples were analyzed on a FACS Symphony A5 flow cytometer (Becton Dickinson). t-distributed stochastic neighbor embedding (tSNE) analysis of data was performed using the Spectre R package (19). Samples were initially prepared in FlowJo, and the populations of interest were exported as batch value CSV files. The dataset was then merged into a single data table, with keywords denoting the sample, group, and other factors added to each row (cell). The FlowSOM algorithm (20) was then run on the merged dataset to cluster the data, where every cell is assigned to a specific cluster and metacluster. Subsequently, the data were downsampled and analyzed by the dimensionality reduction algorithm, tSNE.
Data availability
Raw fastq and counts data generated during this study are available at Gene Expression Omnibus (GSE200134).
Statistical analysis
Statistical analysis was performed using GraphPad Prism Software V9. Statistical tests used were the unpaired t test, multiple unpaired t tests, and two-way ANOVA were performed for experiments according to indication in figure legends and error bars represent SEM as indicated. The precursor cohort method used to determine mean division numbers (MDN) was performed as published previously (21). Plotting the MDN of cell populations against time enables estimates and comparisons of division rates. MDN is the average number of divisions the initial cohort has undergone, where total cohort number describes the number of founding cells initially present within a population. Levels of statistical significance are expressed as P values: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Results
In a lung metastasis model using BRAFV600E LWT1-SM1WT1 murine melanoma cells (22), we observed that Cish−/− mice displayed enhanced anticancer responses by reducing the number of metastatic nodules in the lungs (Fig. 1A and B). Inhibition of TGFβ by the 1D11 (clone) neutralizing antibody, combined with the BRAFV600E inhibitor PLX4720 further enhanced the antitumoral response in both WT and Cish−/− mice. This suggests that deletion of Cish does not alter the sensitivity of NK cells to TGFβ signaling. To study this in depth, we performed transcriptomics (Fig. 2A; Supplementary Fig. S1 and S2) and flow cytometry (Fig. 2B) analysis in WT, Cish−/− and TGFβRII-deficient NK cells either unstimulated or exposed to TGFβ. After 4 hours stimulation by TGFβ, both metabolism-related (e.g., MYC targets and oxidative phosphorylation) and proliferation-related pathways (E2F targets) were downregulated in WT and Cish−/− NK cells, but not in TGFβRII-deficient NK cells. Cish−/− NK cells also displayed a tissue residence–like phenotype in response to TGFβ stimulation by upregulating CD49a, DNAM-1, and TRAIL, similar to the WT NK cells (Fig. 2B). These cells were classified as tissue resident like, as they share markers that are commonly expressed in murine innate lymphoid cells 1 (ILC1; ref. 23), and these cells were originally described as tissue-resident NK cells (24). Previous reports have also shown that murine conventional (cNK) cells exposed to TGFβ1 also displayed ILC1-like markers (e.g., CD49a, DNAM-1, and TRAIL; refs. 8, 9, 25). We previously observed that TGFβ signaling can trigger reduced expression of Eomesodermin (EOMES) in cNK cells (8, 9). In addition, EOMES is a transcriptional factor classically expressed in murine cNK but not in ILC1 cells (26). Here we observed that Cish−/− NK cells also downregulate EOMES upon rTGFβ1 stimulation (Supplementary Fig. S3). In addition, our group observed by computational approaches that exhaustion and tissue residence signatures from transcriptomics datasets are predictive of effector lymphocyte functions in tumors (27). We next took advantage of this bioinformatics approach to measure exhaustion and tissue residency in NK cells upon rTGFβ1 stimulation (Supplementary Fig. S4). However, we did not observe major differences in exhaustion and tissue residency induction by TGFβ signaling in Cish−/− NK cells.
To further understand the broader context of these pathways, we generated a transgenic mouse model with dual suppression of CIS and TGFβRII in NK cells, which did not display major or aberrant phenotypes in steady-state conditions (Supplementary Fig. S5). The enhanced metabolic signatures observed in our computational analysis were then corroborated in WT, Cish−/−, TGFβRII-deficient, and double-deficient NK cells by flow cytometry–based proliferation assays (Fig. 3A). Cish−/− NK cells displayed enhanced proliferative capacity in low concentrations of rIL15 (10 ng/mL; ref. 10); however, here we observed that rTGFβ1 blocked this effect, in both low and high concentrations of rIL15 (Fig. 3A and B), whereas proliferation of TGFβRII-deficient and double-deficient NK cells was not affected by rTGFβ1.
Our group previously reported that rTGFβ1 reduces the metabolism of murine NK cells by suppressing the mTOR metabolic pathway (7). We measured respiratory capacity of our NK cells by Seahorse metabolic flux assay and observed that rTGFβ1 can reduce OCR and the basal and maximal mitochondrial respiration of both WT and CIS-deficient NK cells (Fig. 3C–F). Conversely, deletion of TGFβRII or double deletion of TGFβRII and CIS resulted in NK cells unaffected by rTGFβ1 in this aspect (Fig. 3C–F).
We next assessed the antitumor efficacy of deletion of CIS and/or TGFβRII signaling within NK cells in vivo. We choose two well-defined tumor models to investigate: the uniquely NK cell–dependent rejection of B16F10 metastases (28), and the neoantigen-rich subcutaneous MC38 colorectal cancer model characterized by lymphocytic infiltration, anti–PD-1 responsiveness, and TGFβ−mediated immunosuppression in the TME (29). In both cancer models, the dual deficiency of Cish and Tgfbr2 genes resulted in enhanced antitumor immunity and decreased tumor burden (Fig. 4A and B). T and NK lymphocyte infiltration was detected in all MC38 tumor homogenates at experimental endpoint (Fig. 4C). Dual deficiency of Cish and Tgfbr2 resulted in reduced expression of TIGIT, an important inhibitory checkpoint of NK cells (ref. 30; Fig. 4D). We did not observe differences in other checkpoint inhibitors such as PD-1, TIM-3, and LAG3 or activation and/or tissue residency markers such as CD69, CD226, and CD49a (Supplementary Fig. S6). In addition, the Cish−/−, Tgfbr2-deficient, and dual-deficient NK-cell mouse models all showed an increased infiltration of the terminally mature NK cells M2 (31) in the MC38 tumors, which naturally display an enhanced cytotoxicity capacity (Supplementary Fig. S7). Strikingly, tumor-infiltrating Cish- and Tgfbr2-deficient NK cells expressed greater granzyme B (Fig. 4E) compared with the single-gene mutants and WT NK cells, indicating their activation status and suggesting a greater ability to kill MC38 cells. As MC38 tumor regression is dependent on CD8+ T cells, we also noticed that the lymphocytic infiltration per gram of tumor among NK cell, T CD4+, T CD8+, and regulatory T lymphocytes within the observed MC38 tumors did not display major changes in the double transgenic mice (Supplementary Fig. S8), suggesting that the enhanced immunity to these tumors is likely to be due the enhancement of NK-cell functions.
Discussion
NK cells are a promising therapeutic target for solid cancers (32). However, the TME (33, 34) and some NK-cell intrinsic inhibitors [e.g., CIS (10)] can impair NK-cell functions. In the current study, we used different mouse models of cancer to show that Cish deletion in NK cells is not enough to overcome TGFβR suppressive signaling. Whereas TGFβRII deficient NK cells showed no response to TGFβ stimulation, Cish−/− NK cells were sensitive to TGFβ in several different priming pathways, including TGFβ signaling-mediated metabolic inhibition (7), and plasticity toward ILC1-like cells (8, 9), phenomena previously described for murine NK cells. In a novel mouse model of dual inhibition of both suppressive pathways in NK cells, we provide a dynamic view of NK-cell responses and reveal enhanced effector functions by simultaneously suppressing CIS and TGFβ signaling.
In mouse models transplanted with lung metastasis–forming B16F10 melanoma cells (28), we could measure a clear synergistic effect by simultaneously targeting both the CIS and TGFβR pathways. Although subcutaneous inoculation of colorectal MC38 cancer cells is a well-known model to measure T-cell responses, NK cells can still recognize and kill these cancer cells under appropriate stimulation and settings (35). Cish is attracting much attention as a target for enhancing endogenous T-cell tumor immunity (36); however, Cish−/− mice failed to significantly control MC38 alone. This suggests that Cish−/− T cells are still suppressed by the TME. Importantly, although immune control of MC38 is usually T cell–mediated, if CIS is removed at the same time of induction of specific loss of TGFβ signaling, NK cells are no longer tightly regulated and can display the ability to make contributions toward killing of the primary tumor. Other studies have shown promise by deleting either CIS (3, 12) or TGFβR (5, 37) in engineered human NK cells, in which single targeting of these pathways was able to augment cellular fitness and anticancer capacity in different settings. Finally, we propose that targeting multiple inhibitory checkpoints of NK cells effector function unveils a potential for more efficient NK cell–based immunotherapy that might have broad therapeutic interventions, particularly in settings of NK-cell adoptive cell therapy.
Authors' Disclosures
P.A. Beavis reports grants from Gilead Sciences, Bristol Myers Squibb, and AstraZeneca outside the submitted work. N.D. Huntington reports personal fees from oNKo-innate Pty Ltd. outside the submitted work. No disclosures were reported by the other authors.
Authors' Contributions
F. Souza-Fonseca Guimaraes: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. G.R. Rossi: Formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration. L.F. Dagley: Conceptualization, resources, data curation, software, formal analysis, funding acquisition, visualization, methodology. M. Foroutan: Data curation, software, formal analysis, visualization, methodology. T.R. McCulloch: Conceptualization, data curation, formal analysis, validation, investigation, methodology. J. Yousef: Software, formal analysis, methodology. H.-Y. Park: Formal analysis, validation, investigation, methodology. J.H. Gunter: Conceptualization, resources, formal analysis, methodology. P.A. Beavis: Formal analysis, methodology. C.-Y. Lin: Investigation, methodology. S. Hediyeh-Zadeh: Resources, data curation, investigation, methodology. T. Camilleri: Investigation, methodology. M.J. Davis: Resources, data curation, investigation. N.D. Huntington: Conceptualization, resources, supervision, investigation, visualization, writing–original draft.
Acknowledgments
This work is supported by project grants from the National Health and Medical Research Council (NHMRC) of Australia (#1140406, to F. Souza-Fonseca-Guimaraes; #115995, to N.D. Huntington). N.D. Huntington is supported by NHMRC Ideas Grant #118461 and Investigator Fellowship #119529. F. Souza-Fonseca-Guimaraes is funded by a UQ Diamantina Institute Laboratory Start-Up Package, a U.S. Department of Defense—Breast Cancer Research Program—Breakthrough Award Level 1 (#BC200025), a grant (1158085) awarded through the Priority-driven Collaborative Cancer Research Scheme and co-funded by Cancer Australia and Cure Cancer, and a ANZSA Sarcoma Research Grant (supported by Kicking Goals for Xav, Stoney's Steps, and Stop Sarcoma). We thank all the members of the Huntington and Guimaraes Laboratories, Dr. S. F. Ngiow, Prof. G. McArthur, and Prof. A. Yoshimura for discussion, comments, and advice on this project; Dr. G. E. Bollag (Plexxikon Inc) for providing PLX4720 for in vivo studies; Prof. S. Karlsson for providing the TgfbRII flox mice; Prof. E. Vivier for providing the NKp46cre mice; and Profs. J. Ihle and E. Parganas for providing the CIS knockout mice.
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).