Abstract
The plasticity of innate lymphoid cells (ILCs) has been reported in vitro and in the microenvironment of the intestine. However, whether ILC plasticity contributes to regulation of the tumor microenvironment remains unknown. In this study, we explored plasticity of ILCs in human lung cancer.
We analyzed immune subsets and cytokine expression in lung cancers freshly obtained from 80 patients and explored conversion of ILC1 into ILC3 in coculture with lung cancer cells. Prognostic effects of converted ILC3 and related pathway were evaluated by retrospective cohort composed of 875 patients with lung cancer.
Low percentages of ILC1, and high percentages of ILC3 were found in pulmonary squamous cell carcinomas (SqCC) but not adenocarcinomas (ADC). In non–small-cell lung cancers, the percentage of ILC3 was associated with IL23 expression in tumor cells but not immune cells. In cocultures, tumor cells of SqCCs converted ILC1 into ILC3 by producing IL23, thus promoting IL17-mediated tumor cell proliferation. Consistently, among IL17+ immune cells, the percentages of ILCs were higher in SqCCs than ADCs. Furthermore, the numbers of CD3−RORγt+ ILC3, IL17 expression level, and IL23- or IL17RA-expressing tumor cells were associated with short survival of patients with SqCC but not ADC.
Conversion from ILC1 into ILC3 by IL23-producing SqCCs promotes IL17-mediated tumor progression, resulting in a poor prognosis.
Innate lymphoid cells (ILCs) play a critical role in regulating immune responses in various immune-related diseases, including tumor microenvironment (TME). The plasticity between ILC1 and ILC3 has been demonstrated in vitro and in specific microenvironment of intestine. However, the plasticity of ILCs in the TME has not yet been reported. Here, we show an inverse correlation of ILC1 and ILC3 in squamous cell carcinomas (SqCC) and conversion of ILC1 into ILC3 by IL23-producing tumor cells of SqCCs, thus promoting IL17-mediated tumor cell proliferation. Furthermore, the numbers of CD3−RORγt+ ILC3, expression level of IL17 in TME, and IL23- or IL17RA-expressing tumor cells were associated with short survival of patients with SqCC but not adenocarcinoma. Thus, IL23-producing tumor-mediated conversion from ILC1 into ILC3 is a novel strategy that favors tumor growth and expansion in the TME, which suggests that the IL23–ILC3–IL17 axis may be a useful therapeutic target of IL23-producing lung cancers.
Introduction
Non–small-cell lung cancer (NSCLC), which accounts for 85% of all lung cancers, is the leading cause of cancer-related death worldwide (1). Among NSCLCs, adenocarcinoma (ADC) and squamous cell carcinoma (SqCC) are two major histologic subtypes that exhibit distinct epidemiologic, biological, and genetic characteristics (2, 3). Immunotherapy has recently emerged as a strong therapeutic strategy for the treatment of lung cancer (4). However, the responsiveness of immunotherapy differs in two types of NSCLCs (5, 6). Therefore, a comprehensive understanding of the immune regulatory network in the tumor microenvironment (TME) of individual type of NSCLC is critical for achieving efficient and successful treatment outcomes using immune reagents. However, the differential immune regulatory networks in SqCCs and ADCs have not yet been clarified.
IL17 is a proinflammatory cytokine that is generated by various immune cells, including CD4+ Th cells, γδ T cells, natural killer T (NKT) cells, NK cells, and group 3 innate lymphoid cell (ILC) 3 (7), thereby regulating various immune responses in autoimmune diseases, cancer, infection, and transplantation (7). Regarding the functional roles of IL17 in the TME, emerging evidence has demonstrated that IL17 has protumor effects by regulating immune evasion, angiogenesis, and tumor proliferation, although antitumor effects have also been reported (8). However, among IL17-producing immune cells, the contribution of ILC plasticity to IL17-mediated tumor growth and its downstream mechanisms in human lung cancer remain elusive.
ILCs are a distinct cell type in the innate immune system that have a lymphoid morphology but lack the expression of somatically rearranged antigen receptors and cell-surface markers for myeloid or dendritic cells (DC; ref. 9). ILCs are divided into three distinct subclasses (ILC1, ILC2, and ILC3) based on their cytokine profiles and set of expressed transcription factors (9). ILCs play a critical role in regulating the immune responses in various immune-related diseases, including colitis, asthma, and obesity-induced glucose intolerance (10–12). Experimental evidence suggests that ILCs act as a double-edged sword in tumor progression and regression (13). ILC1 inhibits tumor development and growth, whereas ILC2 promotes tumor progression (14, 15). In contrast, ILC3 has both pro- and antitumor effects (16–18). With respect to ILC in human lung cancer, NCR+ILC3 cells have been detected at the edge of the tumor-associated tertiary lymphoid structure in NSCLCs from patients (17). However, the functional roles of ILCs in human lung cancers remain unclear. ILC and Th cells express identical sets of effector cytokines and master transcriptional factors (19), which suggests that these two cell subsets share common regulatory mechanisms for their plasticity. The switches and trans-differentiation of Th cells have been demonstrated both in vitro and in vivo (19). Similar to Th cells, distinct stimulations regulate the plasticity between ILC1 and ILC3 in vitro and in the specific microenvironment of the intestine in patients with Crohn's disease (20). However, the plasticity of ILCs in the TME and its clinical impact have not yet been reported.
To address these issues, we investigated whether the plasticity of ILC affects IL17-mediated regulation of the TME in lung cancer. We found that the IL17–IL17R axis is activated in SqCCs but not ADCs, and IL23-producing SqCCs promotes IL17-mediated tumor growth by converting ILC1 into ILC3 in the TME, thereby shortening patient survival.
Materials and Methods
The cohorts of patients with NSCLC
Fresh tumor and nontumor lung tissues (NTLTs) were obtained from 80 patients who underwent surgery for NSCLCs (40 ADCs and 40 SqCCs) without neoadjuvant chemotherapy or any other lung disease at the Department of Thoracic Surgery of Seoul National University Hospital (SNUH, Seoul, Korea). All patients provided written informed consents for this study. For tissue microarray analyses, tissues were collected from 875 patients who underwent surgery for pulmonary ADC (497 cases) or SqCC (378 cases) and had been followed-up at SNUH from 2001 to 2012. No patients had received chemotherapy before surgery or had distant metastasis at the time of diagnosis. Clinicopathologic data and pathologic tumor node metastasis staging from the 7th American Joint Committee on Cancer were obtained from medical and pathologic records. This study followed the World Medical Association Declaration of Helsinki recommendations and was approved by the Institutional Review Board of SNUH (H-1404-100-572).
Preparation of fresh tissues obtained from patients with NSCLC and tissue microarray
To perform flow cytometric analysis, tumors of at least 1 cm3 in size and matched NTLT were obtained immediately after surgical resection. Tissue was mechanically dissociated using a blade and subsequently digested in RPMI1640 medium supplemented with 80 U/mL DNase I, 300 U/mL collagenase I, and 60 U/mL hyaluronidase at 37°C for 30 minutes. The single-cell suspension was passed through a 70-μm cell strainer. After red blood cell lysis, cells were washed and used for cell sorting, RNA extraction for RT-PCR, and flow cytometry. In addition, a tissue microarray was constructed from 2-mm diameter cores derived from representative tumor areas of formalin-fixed paraffin-embedded (FFPE) tissue blocks.
Antibodies and cell lines
The antibodies used in this study were summarized in Supplementary Materials and Methods. Human lung SqCC cell lines (H1703 and SK-MES1) and ADC cell lines (H522 and A549), and mouse SqCC cell line (KLN-205) cells were purchased from the ATCC, while HCC-15 (human lung SqCC cell line) was purchased from Korean Cell Line Bank. These cell lines were regularly tested by PCR, to ensure they were Mycoplasma negative. The cell lines were authenticated by short-tandem repeat DNA fingerprinting (Korean Cell Line Bank).
Lentiviral transduction and transfection study
The recombinant lentivirus was generated by using the construct of the mouse IL23p19 open reading frame expression clone (EX-Mm12283-Lv105; GeneCopoeia) and the Lenti-Pac HIV expression Packaging Kit (GeneCopoeia) according to the manufacturer's instructions. 293Ta packaging cell line was transfected overnight with the mouse IL23p19 expression clone and packaging plasmids. After 16 hours, fresh medium was added to the cells and incubated overnight. On three consecutive days after transfection, the pseudotyped virus–containing culture medium was harvested, filtered, and supplemented with 8 μg/mL polybrene (Sigma-Aldrich). After KLN-205 cells were infected, stable transfected cells were selected in puromycin. Control cells were transduced with the control lentiviral vector. Fluorescence analysis or RT-PCR was performed to evaluate the IL23 overexpression efficiency.
Mice and tumor inoculation in vivo
BALB/c mice were obtained from Orient Bio Inc. Tumors were established by subcutaneously injecting IL23-overexpressing KLN-205 or control KLN-205 (transduced with empty vector) or wild-type KLN-205 cells (2 × 105 cells/injection) into the flanks of 6- to 8-week-old BALB/c mice in the presence or absence of the indicated dose of control Ig (eBRG1; eBioscience) or anti-IL23p19 antibody (clone 16-7232-81; eBioscience). These antibodies were administered on the first day of tumor injection and repeatedly every 4 days. Tumor growth was estimated by measuring tumor diameters using a digital caliper. On day 21, mice were euthanized and tumors were harvested for further analysis. This study was approved by Institutional Animal Care and Use Committee of Clinical Research Institute, SNUH, and Institutional Biosafety Committee of SNU. This facility was accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International.
Flow cytometric analyses and cell sorting
Single-cell suspensions prepared from fresh tumor and NTLT were preincubated with Fc Receptor Blocking Solution (BioLegend) to reduce nonspecific binding and subsequently stained for various markers. Cells were stained for surface antigens for 30 minutes at 4°C. For intracellular staining, cells were stained with antibodies (Supplementary Materials and Methods) and fixed. An amine reactive dye (Aqua LIVE/DEAD Stain Kit; Life Technologies, Thermo Fisher Scientific) was used to exclude dead cells according to the manufacturer's instructions. To sort total ILCs using flow cytometry according to gating strategy (Fig. 2B), suspended cells from tumor or NTLT were labeled with various antibodies (Supplementary Materials and Methods). After labeling, cells were sorted using a FACSAria (BD Biosciences) to > 95% purity. Tumor and immune cells were isolated from fresh tumor tissues using the magnetic bead sorting MidiMACS System (Miltenyi Biotec) based on the binding of anti-EpCAM-APC (9C4; BioLegend) or anti-CD45-PE (555483; BD Biosciences).
Immunohistochemistry
FFPE tissue blocks were cut into slices of 4 μm thickness and subjected to immunohistochemistry (IHC). All IHC studies were performed using the Ventana Benchmark XT Automated Staining System (Ventana Medical Systems) according to the manufacturer's protocol. To perform double IHC, sliced tissues on slides were washed and incubated with a horseradish peroxidase–linker antibody and an alkaline phosphatase–linker antibody after overnight incubation with primary antibodies. Detection was performed with 3, 3-diaminobenzidine (DAB, brown chromogen) and Vector Blue. Antibodies used for IHC were summarized in Supplementary Materials and Methods. The percentages of stained tumor cells and the pattern of their staining were assessed; 5% of membranous or cytoplasmic staining was considered a cut-off point for positive IL17RA and IL23 expression. ILC3 was defined as cells expressing RORγt in the absence of CD3 and CD20 expression. The numbers of ILC3 were measured in each core of the tissue microarray.
Immunofluorescence
Frozen sections of tumor tissues were attached to coverslips, fixed in acetone solution for 15 minutes, and washed three times with PBS. Sectioned tissues were blocked with 5% BSA in PBS for 30 minutes at room temperature. After washing with PBS, sectioned tissues were incubated at room temperature for 1 hour with antibodies, washed, covered with coverslips, and examined under a fluorescence microscope (Supplementary Materials and Methods).
qRT-PCR
RNA was isolated from sorted tumor cells and immune cells using TRIzol (Life Technologies) and reverse-transcribed into cDNA using RNAM-MLV Reverse Transcriptase (Promega) according to the manufacturer's protocol. Gene-specific PCR products were measured using an Applied Biosystems 7500 Sequence Detection System (PerkinElmer Biosystems). The sequences of primers were summarized in Supplementary Materials and Methods. Relative gene expression was normalized to that of GAPDH (Hs02758991_g1).
Proliferation assay
Lung SqCC cell lines were seeded at a density of 5 × 104 cells/well in 48-well plates. Cells were grown overnight, and the medium was replaced with maintenance medium containing the desired concentrations of cytokines or medium. Cell viability was assessed after 96 hours based on cell number and the Ki-67 index.
In vitro culture of ILCs
Total ILCs or ILC1 were isolated from fresh tissues of ADCs or SqCCs, sorted, plated in a 96-well plate (2 × 104 cells/well), and stimulated with recombinant IL2 (10 U/mL; Novartis) and IL23 (50 ng/mL; R&D Systems) for 5 days.
Coculture experiments using ILCs
Tumor cell lines (4 × 104 cells/well), or tumor cells isolated from fresh tissues of ADCs or SqCCs (5 × 104 cells/well) were plated in a 48-well plate. After 24 hours, sorted total ILCs or ILC1 (3 × 104 cells/well) from NTLT or NSCLCs were added and cocultured for an additional 5 days in the presence of an anti–IL23-neutralizing (HNU2319; eBioscience), anti–IL1β-neutralizing (B122; eBioscience), anti–IL17-neutralizing (eBio64CAP17; eBioscience), or anti–IL22-neutralizing (22URTI; eBioscience) antibody. During coculture, adherent and nonadherent cells were obtained after stimulation with 10 ng/mL PMA and 0.5 μg/mL ionomycin for 4 hours. Nonadherent cells were stained for 30 minutes using ILC panel antibodies, anti-IL17, or anti-IL22 antibody, while adherent cells were treated with trypsin/EDTA, washed, and stained with anti–Ki-67 antibody or subjected to RNA isolation for RT-PCR. The amounts of IL17 were measured in culture supernatants using ELISA Kit (eBioscience)
RNAseq analyses
CD3+ T cells were obtained from fresh tumor tissues and NTLT of ADCs and SqCCs and sorted using flow cytometry. tRNA was isolated from sorted CD3+ T cells using the RNeasy Mini Kit (Qiagen). RNA quality and quantity were determined on a Nanodrop (Nanodrop Technologies) and the Agilent 2100 Bioanalyzer (Agilent Technologies). RNA-seq analyses were performed at Theragen Bio Institute (Suwon-si, Gyeonggi-do, Korea). Raw read filtering, gene normalization, and differentially expressed gene analyses were performed as described previously (21).
TCGA data analyses
Gene expression data for pulmonary NSCLC were obtained from TCGA, which is available through the cBio Cancer Genomics Portal (http://www.cbioportal.org/public-portal). Clustering-based analyses were performed for cell proliferation, migration, and angiogenesis. Each gene cluster was defined as “angiogenesis; GO: 0001525,” “epithelial cell proliferation; GO: 0050673,” or “positive regulation of epithelial cell migration; GO: 0010634” according to Gene Ontology annotation. Comparisons were made between tumors with and without IL17A expression (mRNA level > 0). Analyses were performed using the R language and core packages (gplots and RColorBrewer).
Statistical analysis
Comparisons between variables were performed using the χ2 test, Student t test, or paired t test. Survival analyses were performed using the Kaplan–Meier method with the log-rank test. Two-sided P values less than 0.05 were considered statistically significant in all analyses. Progression-free survival was measured from the date of surgery to that of recurrent or metastatic disease occurrence. Prism GraphPad software was used for all statistical analyses.
Results
The IL17–IL17 receptor axis is enhanced in pulmonary SqCCs but not ADCs and is associated with short survival of patients with SqCC
To investigate the differential effects of the IL17–IL17R axis on two major types of NSCLC: SqCCs and ADCs, we estimated the expression level of IL17R in tumor cells and the IL17 transcript level in CD45+ immune cells. We isolated CD45+ immune and CD45− EpCAM+ tumor cells from 80 cases of NSCLCs (40 cases of SqCCs and 40 cases of ADCs; Supplementary Table S1) freshly obtained from patients who underwent surgery. The transcriptional levels of IL17RA in tumor cells were higher in SqCCs and the SqCC cell lines (SK-MES, H1703, and HCC15) than in ADCs and the ADC cell lines (A549 and H522), respectively (Fig. 1A and Supplementary Fig. S1A). In tissue array analyses, SqCCs exhibited higher percentages of IL17R-positive tumor cells (378 cases of SqCCs and 497 cases of ADCs, Supplementary Table S2) and higher expression of the IL17 transcript in CD45+ immune cells freshly obtained from tumors than ADCs (Fig. 1B and C). Moreover, the expression of the IL17 transcript in CD45+ immune cells was higher in SqCCs with a high T stage (tumor size > 3 cm) than in those with a low T stage (tumor size < 3 cm), whereas ADCs exhibited similar expression levels of IL17 in the two groups (Fig. 1D). Next, we analyzed The Cancer Genome Atlas (TCGA) data on the functional effects of IL17 on angiogenesis, cell migration, and proliferation of tumor cells in NSCLCs (Fig. 1E and Supplementary Fig. S1B; ref. 22). A heatmap of proliferation-related genes revealed a clear difference in the expression patterns of IL17hi and IL17low NSCLCs (Fig. 1E). The expression levels of proliferation-promoting genes, such as CCNA2, CCNB, Csk2, CDC7, and CDC25C, were significantly higher in IL17hi NSCLCs than IL17low NSCLCs (Fig. 1F). Furthermore, SqCCs expressed higher levels of these proliferation-related markers than ADCs (Fig. 1G), and recombinant IL17 treatment increased the Ki-67 index and the numbers of SK-MES1 cells (Supplementary Fig. S1C and S1D). These findings suggest that IL17-mediated proliferation contributes more to the tumor progression of SqCCs than ADCs. Consistent with this hypothesis, the progression-free survival of patients with IL17R+ or IL17hi SqCC was significantly lower than that of those with IL17R− SqCC or IL17low SqCC, whereas patients with ADC showed similar progression-free survival regardless of IL17R or IL17 expression level in TME (Fig. 1H and I). Together, these findings suggest that the IL17–IL17R axis is involved in tumor progression of pulmonary SqCCs but not ADCs.
SqCCs exhibit an inverse correlation between ILC1 and ILC3 compared with ADCs and higher numbers of ILC3 among IL17-producing immune cells
To investigate the mechanism by which SqCCs exhibit IL17-mediated proliferation compared with ADCs, we estimated the percentages of IL17-producing immune cell subsets in SqCCs and ADCs, because various types of immune cells produce IL17 during regulation of the immune response and inflammation (23, 24). Flow cytometric analyses revealed that the percentages of IL17+ ILCs were higher in SqCCs than ADCs, whereas those of IL17-expresssing lineage+, CD4+ T, γδ T, and NKT cells, were similar (Fig. 2A). Moreover, a transcriptome assay revealed that CD3+ T cells from SqCCs and ADCs expressed similar expression levels of various cytokines including IL17 and IL23p19 (Supplementary Fig. S1E). These findings suggest that IL17-producing ILCs regulate the TME of SqCCs but not ADCs. To explore individual ILC subset further, we identified Lin−CD117−ST2−NKp44− ILC1 by excluding NK cells based on CD49a expression, Lin−ST2+ c-kit− ILC2, and Lin−ST2− c-kit+ ILC3 in TME, which expressed their signature transcription factors and cytokines, respectively (Fig. 2B and C; Supplementary Fig. S2A and S2B). Regarding the individual subsets of ILCs, SqCCs exhibited a reduction in the percentage of ILC1 but an increase in both NKp44+ and NKp44− ILC3 compared with NTLT in all cases of prospective cohort, which was also observed in the matched cases of SqCCs and NTLT among all cases (Fig. 2C–F; Supplementary Fig. S2C). In contrast, the percentages of ILC1 and ILC3 were similar in ADCs and NTLT (Fig. 2C and F). Fluorescence microscopy revealed more ILC3 in the intratumor areas of SqCCs than ADCs (Fig. 2D and E). Furthermore, an inverse correlation between ILC1 and ILC3 was detected in SqCCs but not ADCs (Fig. 2G), which correlated with the expression level of the IL17 transcript in CD45+ immune cells (Fig. 2H). However, the percentages of total ILC and ILC2 were similar in the two histologic types of NSCLCs and NTLT (Fig. 2C), and the percentages of ILC1 and ILC3 were not related with those of other immune cell subsets including CD1c+ DCs and CD141+ DCs in NSCLCs (Supplementary Fig. S3A). Collectively, these findings indicate that high numbers of ILC3, but low numbers of ILC1, are located in the TME of SqCCs compared with NTLT, which indicates an inverse correlation between these two subsets in SqCCs but not ADCs.
Conversion from ILC1 into ILC3 is promoted by IL23-producing SqCCs, rather than immune cells, in the TME
IL23 and IL1β induce the conversion of ILC1 into ILC3 (20, 25). Thus, to investigate the mechanism by which SqCCs lead to high numbers of ILC3 but low numbers of ILC1 in the TME, we measured the expression levels of these cytokines in CD45+ immune and CD45− EpCAM+ tumor cells freshly obtained from NSCLCs and analyzed the correlations between these cells and the percentages of ILC1 and ILC3 subsets. In NSCLCs, the expression level of IL23p19 transcript in CD45− EpCAM+ tumor cells, but not CD45+ immune cells, was positively correlated with the percentage of ILC3 but inversely associated with that of ILC1 (Fig. 3A). In contrast, none of the IL1β, IL12, or IL18 transcript levels in tumor and immune cells were significantly associated with ILC subsets in NSCLCs (Fig. 3A; Supplementary Figs. S3B and S4). Moreover, flow cytometric analyses revealed that the percentage of IL23+ tumor cells was positively correlated with that of ILC3 in NSCLCs, whereas that in IL23+ immune cells was not (Fig. 3B and C). These findings suggest that IL23-producing NSCLCs are involved in the plasticity of ILC subsets by converting ILC1 into ILC3 in the TME.
Among NSCLCs, SqCCs express higher transcript levels of IL23 in CD45− EpCAM+ tumor cells than ADCs, whereas CD45+ immune cells similarly express these cytokines in the two histologic types of NSCLCs (Fig. 3D), consistent with data from TCGA (Supplementary Fig. S3C). Flow cytometric analyses demonstrated higher percentages of IL23-expressing CD45−EpCAM+ tumor cells and their cytosolic expression of IL23 in SqCCs than ADCs, whereas the percentages of IL23-positive CD45+ immune cells were similar in the two types of NSCLCs and were not associated with the percentage of ILC3 (Fig. 3E; Supplementary Fig. S3D). These findings suggest that IL23-producing SqCCs, rather than immune cells, promote the conversion of ILC1 into ILC3 in the TME. To address this, we cultured total ILCs or ILC1 sorted from NTLT in the presence of IL2 and IL23. Recombinant IL2 + IL23 or IL23 increased the percentage of ILC3 after coculture with total ILCs or ILC1 for 5 days, which was decreased by treatment with an anti–IL23-neutralizing mAb, which indicates that IL23 promotes the conversion from ILC1 into ILC3 in lung tissue (Supplementary Fig. S5A). The percentage of ILC3 was higher in coculture of total ILCs from ADCs and CD45−EpCAM+ tumor cells of SqCC, but not those from ADCs (Fig. 4A and B; Supplementary Fig. S5B and S5C). ILC subsets were defined according to expression pattern of transcription factors (RORγt and GATA3) as well as cell-surface markers (c-kit and IL7Rα; Fig. 4A–D; Supplementary Fig. S5A–S5F). Coculture of ILCs from SqCC and CD45−EpCAM+ tumor cells from SqCC or ADC did not alter the ILC3 population (Fig. 4C and D; Supplementary S5D and S5E). Furthermore, coculture with sorted ILC1 from NTLT or ADC tumor cells, and CD45−EpCAM+ tumor cells of SqCC increased the percentage of ILC3 but reduced that of ILC1, which was inhibited by adding the neutralizing IL23 mAb (Fig. 4E and F; Supplementary Fig. S5F and S5G). In contrast, CD45−EpCAM+ tumor cells of ADCs did not alter the percentages of ILC1 and ILC3 during coculture with sorted ILC1. These findings indicate that SqCCs rather than ADCs drive the conversion from ILC1 into ILC3 subsets by producing IL23 in NTLT and TME.
ILC3 promotes IL17-mediated tumor proliferation and contributes to the poor prognosis of patients with pulmonary SqCC
To explore IL23–ILC3–IL17 axis in tumor proliferation in vitro, we cultured ILC1 from the IL23low ADC + IL23high SqCC cell line (SK-MES1, H1703, or HCC15) or tumor cells freshly isolated from SqCCs in the presence of the anti-IL23, anti-IL17, or anti-IL22 mAb (Supplementary Fig. S6). The percentages of ILC3 among total ILCs and amounts of IL17 in culture supernatants were increased after coculture of ILC1 and the SqCC cell line, whereas those of ILC1 were decreased. These alterations were inhibited by adding the anti-IL23 antibody, but not by the anti-IL22, anti-IL17, or anti–IL1β-neutralizing antibody (Supplementary Fig. S6A–S6C). The expression levels of proliferation markers in tumor cells were also increased during coculture of ILC1 and the SqCC cell line, which was inhibited by adding the anti-IL23 or anti-IL17 antibody (Fig. 5A and B; Supplementary Fig. S6D). Moreover, IL17 production was detected in ILC3, but not ILC2 (Supplementary Fig. S6E). The percentage of IL22+ ILCs was similar in SqCCs and ADCs, and the expression level of the IL22 transcript was not significantly associated with percentage of ILC3 in NSCLCs (Supplementary Fig. S6F and S6G). These findings indicate that IL23-producing SqCCs promote tumor growth by inducing ILC3-mediated IL17 production, although they increase the production of IL17 and IL22. To confirm this in vivo, we injected wild-type (WT) mice subcutaneously with IL23-overexpressing mouse SqCC cell line (KLN-205 transduced by IL23-overexpressing lentivirus) or control cells (KLN 205 with empty vector). IL23-overexpressing KLN-205 cells increased tumor sizes more than control and WT KLN-205 cells, which was abolished by injecting with anti-IL23 blocking antibody (Fig. 5C and D). Moreover, the percentages of ILC3 and IL17-expressing ILCs were higher in tumors from mice injected IL23-expressing KLN-205 cells than control or WT KLN-205 cells, whereas those of ILC1 were lower. This inverse correlation between ILC3 and ILC1 and enhancement of IL17-expressing ILCs in tumors were also inhibited by anti-IL23 antibody injection (Fig. 5E and H). However, the percentages of ILC2 were similar in the tumors of all groups. These findings indicate that IL23 produced by SqCC cells promotes tumor growth by increasing IL17-producing ILC3 in TME. To address this in patients with SqCC, we analyzed the survival of patients with SqCC or ADC by estimating the numbers of CD3−CD20−RORγt+ ILC3 and the expression of IL23 in tumor cells using IHC (Fig. 6A–E). SqCCs exhibited higher numbers of IL23-expessing tumor cells and CD3−CD20−RORγt+ILC3 in the TME than ADCs (Fig. 6C and D). The numbers of CD3−CD20−RORγt+ILC3 were higher in IL23+ tumor cells than IL23− cells in NSCLCs (Fig. 6E). Moreover, the numbers of CD3−CD20−RORγt+ILC3 and the expression of IL23 in tumor cells were significantly associated with short progression-free survival of patients with SqCC, but not ADC (Fig. 6F). These findings indicate that the IL23–ILC3–IL17 axis contributes to a poor prognosis of patients with SqCCs among NSCLCs.
Discussion
In our study, an inverse correlation between ILC1 and ILC3 was found in pulmonary SqCCs and the conversion from ILC1 into ILC3 was promoted by IL23-producing tumor cells. These findings indicate that a specific TME regulates the functional plasticity of ILCs, in particular, the conversion from ILC1 into ILC3. Regarding the mechanism of the conversion of ILCs, Bernink and colleagues reported that the combination treatment of IL2, IL23, and IL1β promoted the conversion of CD127+ ILC1 into IL22-producing ILC3 in vitro, which was reversible and dependent on RORγt (20). They also demonstrated that CD14+ DCs promoted polarization from ILC3 to ILC1, whereas CD14− DCs promoted the conversion from ILC1 into ILC3 (20). Moreover, the numbers of CD127+ ILC1 increased at the expense of ILC3 in inflamed intestinal tissues from patients with Crohn's disease, whereas CD127+ ILC1 converted into ILC3 in humanized mice in the absence of inflammation (20, 25). CX3CR1+ DCs contribute to the homeostasis of CXCR6+ NKp46− ILC3 by producing IL23 and CXCL16 in the intestine (26). These findings suggest that the conversion between ILC1 and ILC3 and homeostasis of ILCs in the intestine may be regulated by hematopoietic cells rather than nonhematopoietic cells, depending on the inflammatory status of the microenvironment. In contrast to the intestine, the percentages of CD141+ DCs were not correlated with ILC subsets in NSCLCs. Furthermore, tumor cells from SqCCs, rather than ADCs, produced high levels of IL23 and promoted the conversion of ILC1 into ILC3 in an IL23-dependent manner. Thus, tumor cells, rather than hematopoietic cells, in the TME of SqCCs directly regulate the conversion from ILC1 into ILC3 by producing IL23. However, SqCC tumor cells or recombinant IL23 minimally affect conversion of SqCC-derived ILCs. Moreover, tumor cells of ADC failed to maintain high levels of ILC3 in SqCC-derived ILCs in coculture, which was restored by adding recombinant IL23. To explain these findings, it is hypothesized that conversion of ILC1 into ILC3 has been maximally achieved and maintained as plateau status in TME of SqCC. Combined, these findings suggest that IL23 might exert effect on maintenance of ILC3 in addition to conversion of ILC1 into ILC3 in TME.
In TME, tumor cells provide tumor cell proliferation signals and a suitable environment for tumor progression by modulating immune cells (27, 28). To this end, tumor cells employ diverse strategies, including decreasing the expression of antigen-presenting proteins, increasing the expression of inhibitory proteins in immune cells, and regulating the plasticity of Th cells and M1/M2 macrophages (29–32). In addition to these strategies, we suggest that the IL23-producing tumor-mediated conversion from ILC1 into ILC3 is a novel strategy of tumors that favors tumor progression and expansion in the TME. Similar to lung cancer, SqCCs in the esophagus also express IL23, contributing to the epithelial–mesenchymal transition via the Wnt/β-catenin pathway (33). Thus, these findings suggest that IL23 produced by SqCCs has diverse effects on tumorigenesis and progression by regulating intrinsic tumor biology and cross-talk with extrinsic immune cells, which suggests that the IL23–ILC3 axis may be a therapeutic target for IL23-producing SqCCs. Biological reagents that inhibit the IL23 pathway have demonstrated efficacy in clinical trials for the treatment of inflammatory diseases including Crohn's disease and psoriasis (34, 35), which suggests that IL23 blockade may be a useful therapeutic strategy for IL23-mediated human diseases, including IL23-producing lung cancers. However, no clinical trials have addressed human cancers using an IL23 inhibitor or blockade. Moreover, the effects of IL23 blockade on tumor progression may be diverse, because the biological effects of IL23 on both tumor and immune cells are more complex, depending on the activation status of immune and tumor cells and the patient's clinical stages. Nevertheless, IL23 blockade or an IL23 inhibitor may efficiently suppress the progression of IL23-producing lung cancers due to dual effects on the plasticity of both ILC3 and Th17 cells in the TME, although further study is needed to validate this strategy.
Emerging studies have demonstrated that IL17 promotes the progression and metastasis of lung cancer by increasing tumor angiogenesis, and cell proliferation but inhibiting apoptosis (36). However, whether IL17 regulates tumor growth in specific histologic types of NSCLCs, such as SqCCs and ADCs remains unknown. Our experiments demonstrated that IL17–IL17R axis was enhanced in SqCCs relative to ADCs and contributed to a short survival of patients with SqCC, which suggests that IL17–IL17R axis contributes to tumor growth in SqCCs but not ADCs. Among the IL17-expresssing immune cells, IL17+ ILCs were higher in number in SqCCs than ADCs, whereas the expression level of IL17 in CD3+ T cells from the two histologic types was similar in transcriptome analyses. Moreover, coculture experiments revealed that converted ILC3 promoted tumor growth in SqCCs by producing IL17. These findings suggest that the tumor growth of SqCCs rather than ADCs may be attributable to higher numbers of IL17-producing ILC3, although IL17 produced by other immune cells similarly affects tumor growth of the two histologic types of NSCLCs. Thus, it is feasible that increased numbers of ILC3 in SqCCs contribute more to tumor growth than ADCs. In contrast, it was suggested that NKp46-expressing ILC3 may inhibit tumor growth via the IL12-dependent enhancement of adhesion molecules in blood vessels in melanoma (18). Moreover, Carrega and colleagues hypothesized that NCR+ ILC3 has an antitumor effect on human lung cancer, because NCR+ ILC3 upregulated the expression of adhesion molecules in endothelial and mesenchymal stem cells in vitro and they were found in tertiary lymphoid structures (17). These contradictory outcomes between that study and our study may be due to the different cohorts of patients with different numbers of cases (50 cases in the study by Carrega and colleagues versus 80 cases in this study) and/or histologic types of NSCLC. However, it is unclear how many cases of SqCCs and ADCs were involved in previous study, although the authors reported no difference in the ILC population between the two histologic types. Furthermore, they did not identify an inverse correlation between ILC1 and ILC3 in NSCLCs and did not directly address the functional role of ILC3 in tumor progression and the survival of patients with NSCLC. In contrast, our experiments clearly demonstrated that the numbers of CD3−RORγt+ ILC3 in the TME, and IL17RA-expressing tumor cells were associated with a poor prognosis of patients with SqCCs but not ADCs. These findings suggest that the IL23–ILC3–IL17 axis is a critical pathway that promotes tumor growth of SqCCs, thereby shortening progression-free survival.
In conclusion, our results demonstrate that IL23-producing SqCCs promote IL17-mediated tumor growth by converting ILC1 into ILC3 in the TME, thereby shortening patient survival. Furthermore, we believe that lung cancers may utilize the IL23–ILC3–IL17 axis to favor tumor growth and expansion in the TME, which suggests that this axis may be a useful therapeutic target of IL23-producing lung cancers.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: H.Y. Kim, D.H. Chung
Development of methodology: J. Koh, W.-W Lee
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Koh, I.K. Park, C.H. Kang, Y.T. Kim, Y.K. Jeon
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Koh, Y. Lee, M. Choi, W.-W Lee, Y.K. Jeon, D.H. Chung
Writing, review, and/or revision of the manuscript: J. Koh, I.K. Park, M. Choi, Y.K. Jeon, D.H. Chung
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.Y. Kim, J.-E. Kim
Study supervision: D.H. Chung
Acknowledgments
We thank all members of the Hye Hwa Forum, in particular, Prof. Eui-Cheol Shin for helpful comments on the manuscript. We also thank the NIAID Tetramer Facility at the NIH, USA, for providing PBS-57–loaded CD1d tetramers. This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI14C1277).
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.