IL17A Regulates Tumor Latency and Metastasis in Lung Adeno and Squamous SQ.2b and AD.1 Cancer

Lung cancer is responsible for over a quarter of all cancer-related deaths worldwide (1). More than two thirds of lung cancer cases are detected at inoperable stages, and among those who undergo lung resection for treatment of early-stage cancer, approximately 40% develop metastatic recurrence (2). Although individual genetic susceptibility factors affect lung cancer development, little is known about the underlying mechanisms that regulate tumor latency and metastasis. Cigarette smoke has been shown to promote somatic mutations, double-strand DNA breaks, and malignant transformation in airway epithelial cells (3). Despite heightened inflammatory states in the lungs in response to tumor, induction of immune checkpoint molecules (e.g., PD-1 and CTLA4) in T cells inhibits recognition of primary or metastatic cells, rendering local immune responses ineffective (4).

IL17A-expressing CD4⁺ T cells (Th17 cells) are a necessary component of adaptive immune responses in humans and in experimental models of emphysema (5, 6). Although IL17A and Th17 cell have been implicated in human and preclinical models of non–small cell lung cancer (NSCLC; ref. 7), their exact activities remain unclear. For instance, immunohistochemical analysis of primary NSCLC tissue showed increased IL17A expression, and genetic polymorphisms that increase its expression are associated with NSCLC (8, 9). Increased serum concentration of IL17A and Th17 cells in peripheral blood of NCLSC has been reported, albeit inconsistently (10). Furthermore, IL17A can induce angiogenesis and promotes proliferation of lung tumor cell lines and metastasis in immunodeficient mice (11). Similarly, protumor and antitumor roles for Th17 cells have also been proposed in other solid tumors (12–14). Specifically, IL17A has been shown to promote hepatocellular carcinoma, and in a model of breast cancer, IL17A has been shown to induce CXCL12 expression and promote metastasis (13, 15). In contrast, in a model of melanoma, IL17A has been shown to promote antitumor responses mediated by activation of T helper 1 (Th1) cells and induction of cytotoxic T cells (CTL; ref. 12). These divergent (e.g., protumor or antitumor) roles for IL17A might be in part governed by the complex nature of solid organ tumors. However, less is known about the activities of IL17A/Th17 cells in most solid tumors, including NSCLC, in an immune competent host.

NSCLC, the most common histological subtype of lung cancer, is characterized by the presence of an expanding repertoire of oncogenic mutations (e.g., epidermal growth factor receptor) and the loss of tumor suppressors such as KRAS, PTEN, and SMAD4 (16–18). We have previously shown that PTEN and SMAD4 mutations are found in 5.9% and 2.9% of NSCLC, respectively (19). Although deletion of many or most gene copies (approximating near total loss by SNP array) for PTEN and SMAD4 occurred in 5.3% and 2.1% of NSCLC cases, respectively, deletion of a few copies (approximating partial copy loss by SNP array) for PTEN and SMAD4 occurred in 36% and 44% of NSCLC cases, respectively (19). We have previously shown that airway-specific deletion of Pten and Smad4 (Pts4^d/d) using codon-optimized (i)Cre recombinase under the Club Cell Secretory Protein promoter (CCSP^iCre) in Pten and Smad4 floxed (Pts4^f/f) mice results in a single proximal airway tumor (18). At 12 months of age, 100% of Pts4^d/d mice develop lung cancer, which spontaneously metastasizes to distant organs and has histologic features consistent with mixed adeno- and squamous cell carcinoma (18).

NSCLC subtypes have therapeutic implications (19, 20), and little is known about the activities of IL17A in early- or advanced-staged disease. In this study, we observed that tumor-infiltrating lymphocytes (TIL) in early-stage NSCLC (stages I–II) patients harbor increased numbers of Th17 cells when compared with tumor-free lung parenchyma, indicating an increased association of this helper subset with lung cancer. In patients with advanced NSCLC (stages III–IV), Th17 cells present in tumor-positive draining lymph nodes also showed a reverse correlation with PD-1⁺ CD4⁺ T cells. Although loss of Il17a has been shown to protect against several tumors driven by Kras mutations, including NSCLC (14), Kras plays an etiologic role in a subset of all lung tumors. We therefore investigated whether IL17a, which is associated with lung inflammation, could alter tumor latency and metastasis in an autochthonous NSCLC model that is independent of Kras mutation.

Here, we show that the Pts4^d/d-driven mouse model of lung adenosquamous carcinoma displays genetic signatures characteristic of the SQ.2b, and AD.1 subtypes of human NSCLC. However, Il17a deficiency in the Pts4^d/d model resulted in more rapid tumor development and more frequent metastasis. Interrogation of the immune cell profiles within lung tumor, draining lymph nodes, and metastatic organs in Pts4^d/dIl17a^–/– mice showed that the antitumor immune responses were mediated by IL17a. Together, our findings show that Th17 cells function against the tumor in certain subtypes of NSCLC.

Pten^d/dSmad4^d/dmice (Pts4^d/d) have been previously described (18). Il17a^–/– mice (C57BL/6 background) were obtained from Dr. Chen Dong (The University of Texas MD Anderson Cancer Center, Houston, TX). Pts4^d/d mice were crossed three times with Il17a^–/– mice to generate Pts4^d/dIl17a^–/–mice. Both male and female mice were used. All mice were bred in the transgenic animal facility at Baylor College of Medicine, Houston, TX. All experimental protocols used in this study were approved by the Institutional Animal Care and Use Committee of Baylor College of Medicine and followed the National Research Council Guide for the Care and Use of Laboratory Animals.

Recombinant murine FLT3L was purchased from eBioscience. Recombinant murine GM-CSF, murine IL4, murine, and human IL17A were purchased from R&D Systems. TLR4 ligand lipopolysaccharide LPS and TLR3 ligand poly (I:C) were purchased from InvivoGen.

Mouse lung, mediastinal lymph node, or spleen single-cell suspensions were prepared by mincing whole organs through a 40-μm cell strainer (BD Falcon) followed by red blood cell (RBC) lysis (ACK lysis buffer, Sigma-Aldrich) for 3 minutes. In some experiments, the lung F4/80⁺ cells were isolated using anti-F4/80 conjugated magnetic beads (Miltenyi Biotec) and were treated with recombinant IL17A as indicated. Alternatively, lungs were fixed with instillation of 4% paraformaldehyde solution via a tracheal cannula at 25-cm H₂O pressure followed by paraffin embedding and were sectioned for histopathologic studies. Hematoxylin and eosin (H&E) staining was performed as described (21). Immunohistochemical staining was performed as previously described (18). TTF-1 (H190) and CK7 (RCK-105) antibodies were purchased from Santa Cruz Biotechnology. SOX2 antibody was purchased from Millipore.

Cells were prepared based on the previously described method for the isolation of small intestinal lamina propria cells (22) with the following modifications. The stomach was cut open into 2 pieces and the contents removed. The tissue was then washed once in a DTT/EDTA solution (1 mmol/L DTT, 30 mmol/L EDTA, and 10 mmol/L HEPES) for 10 minutes in a bacterial shaker at 37 degrees and then followed by an additional EDTA wash (30 mmol/L EDTA and 10 mmol/L HEPES) for 10 minutes. The tissue was washed with complete RPMI and then cut into 1 to 2 cm pieces, digested for 1 hour in a bacterial shaker, in complete RPMI containing 100 U/mL type VIII collagenases (Sigma) and 150 μg/mL DNase I (Sigma). After digestion, cells were eluted from the tissue by shaking. Live cells were then isolated using a Percoll (GE) gradient.

Current or former smokers with early-stage (I–II; n = 17) or advanced stage (II–IV; n = 22) NSCLC serially entered the study; all subjects had a significant (>20 pack-years) history of smoking (Supplementary Table S1). The patients had no history of allergy or asthma, had not received oral or systemic corticosteroids, and were free of acute symptoms suggestive of upper or lower respiratory tract infection during the last 6 weeks.

Studies were approved by the institutional review board (IRB) and were in accordance with U.S. Common Rule, at Baylor College of Medicine. Informed consent was obtained from each patient. Human lung single cell suspensions were prepared as previously described (5). Briefly, fresh lung or tumor tissue were minced into 0.1-cm pieces in Petri dishes and treated with 2 mg/mL of collagenase D (Roche Pharmaceuticals) in HBSS and incubated for 30 to 40 minutes at 37°C. Single cells were collected using a 40-μm cell strainer (BD Falcon) followed by RBC lysis. Locoregional and distal lung lymph node cells were collected from lung cancer patients using endobronchial ultrasound (EBUS)-guided sampling (23), were collected in HBSS, and transported to the lab immediately. Single cells were otherwise harvested using the same protocol as used for collection of cells from lung tissue or tumors (described above).

In total, 1,023 NSCLC cases from TCGA (which represented the cases assayed on at least three different molecular profiling platforms among RNA sequencing, DNA methylation arrays, miRNA sequencing, Affymetrix SNP arrays, whole-exome sequencing, and RPPA arrays) were included in the analysis. Previously, multiplatform analysis of these TCGA data uncovered nine genomic subtypes of NSCLC. RNA-seq data were obtained from the Broad Institute Firehose pipeline (http://gdac.broadinstitute.org/). Copy-number alteration data (originally obtained using Broad Institute's Firehose's “thresholded” calls) and somatic mutation data were obtained (19). Visualization using heat maps was performed using JavaTreeview (version 1.1.6r4; ref. 24). For heat map presentation and for gene signature scoring, gene expression values were log-transformed and then z-normalized to standard deviations from the median across all cancer samples. TCGA RNA-seq profiles were scored based on the manifestation of the Pts4^d/d gene signature, using a “t-score” metric as previously described (25), where the overall t-score is high in a profile where the genes up in the Pts4^d/d expression signature are relatively high, whereas the genes down in the signature are relatively low. The GEO accession number for the microarray data of Pts4^d/d mice in the previous report (18) is GSE57133.

Flow cytometry was performed with a BD LSR II (BD Biosciences), and data were analyzed with FlowJo software (TreeStar Inc.). The following antibodies to mouse proteins were purchased from BD Pharmingen and used: Pacific Blue-CD3 (500A2), PE-Cy5-CD4 (RM4-5), FITC-CD4 (RM4-5), APC-Cy7-CD8 (53-6.7), PE-IL17A (TC11-18H10), APC-CD86 (GL-1), PE-CD86 (GL-1), and APC-Cy7-Ly6G/C (RB6-8C5). The following antibodies to mouse proteins were purchased from eBioscience and used: FITC-γδTCR (eBioGL3), eFluro450-B220 (RA3-6B2), PE-CD11b (M1/70), APC-CD11c (N418), APC-CD103 (2E7), PE-Foxp3 (FJK-16s), APC-CD25 (PC61.5), PE-CD217 (IL17 Receptor A; PAJ-17R). The following antibodies to mouse proteins were purchased from BioLegend and used: PerCP/Cy5.5-F4/80 (BM8), PE-CD206 (C068C2), PerCP/Cy5.5-CD152 (CTLA-4; UC10-4B9), PE/Cy7-CD279 (PD-1; 29F.1A12), PE/Cy7-CD11b (M1/70), Pacific Blue-CD45 (30-F11), BV510-CD45 (30-F11), APC-CD45.1 (A20), APC-IFNγ (XMG1.2), and PerCP/Cy5.5-CD11c (N418). The following antibodies to human proteins were purchased from BD Pharmingen and used: PE-IL17A (N49-653), APC-IFNγ (B27), FITC-CD4 (RPA-T4), and APC/Cy7-CD8 (SK1). The following antibodies to human proteins were purchased from BioLegend and used: Pacific Blue-CD45 (HI30), APC/Cy7-CD1c (BDCA-1; L161) and PerCP/Cy5.5-CD279 (PD-1; EH12.2H7). eFluor450-CD3 (OKT3) was purchased from eBioscience. LIVE/DEAD™ Fixable Dead Cell Stain kit (Thermo Fisher) was used to measure the viability of tumor-free adjacent tissue (T-FAT) and tumor infiltrated lymphocytes (TIL) isolated from Stage I-II surgical samples. Anti-mouse CD16/CD32 (eBioscience) or Human TruStain FcX (BioLegend) was added to the cells and were incubated for 5 minutes prior to other antibodies to block Fc receptor. Intracellular cytokine staining was performed by the stimulation of cells with phorbol 12-myristate 13-acetate (PMA, 10 ng/mL; Sigma-Aldrich) and ionomycin (1 μg/mL; Sigma-Aldrich) for overnight supplemented with brefeldin A (10 μg/mL; Sigma-Aldrich) for the last 6 hours. Cells were stained for surface markers and then fixed with FACS lysing solution (BD Biosciences), permeabilized with 0.5% saponin (Sigma-Aldrich), and stained with anti-IFNγ and anti-IL17A.

Mice were anesthetized and scanned using microCT as previously described (26). Briefly, mice were exposed to etomidate (30 mg/kg) and placed in an animal CT scanner (Gamma Medica), and completed images of the chest were obtained and analyzed at the Animal Phenotyping Core in Baylor College of Medicine. Amira 3.1.1 software (FEI) was used to process the images.

Splenic CD4⁺ T cells from Pts4^f/fmice were isolated using anti-CD4 conjugated magnetic beads (Miltenyi Biotec). Pts4^d/dIl17a^–/– mice were intraperitoneally injected with 15 × 10⁶ isolated CD4⁺ T cells at 2.5 and 5 months of age. Mice were then euthanized at 7 or 9 months of age and lung tumor and metastasis were analyzed using lung, spleen, liver, and stomach histology.

Mouse BMDCs were prepared as previously described with some modification (27). Femurs and tibias of 4- to 8-week-old female mice were isolated and freed from the surrounding tissue. Intact bones were kept in 70% ethanol for 3 minutes followed by PBS wash. Both ends of the bones were cut with scissors, and the marrow was flushed out with RPMI-1640 medium through a syringe with 26.5 needle. RBCs were then removed by ACK lysis buffer, and cell debris or tissue clusters were filtered out. Cells from bone marrow were cultured in a 6-well plate with 20 ng/mL recombinant murine GM-CSF and 10 ng/mL recombinant murine IL4 for 5 to 6 days.

Induced CD103⁺ DCs (iCD103 DC) were generated using bone marrow cells as previously described (28). Briefly, 15 × 10⁶ bone marrow cells were cultured in 10 mL RPMI medium with 10% FBS and penicillin/streptomycin supplemented with 200 ng/mL recombinant murine FLT3L and 5 ng/mL recombinant murine GM-CSF (complete medium; CM) for 9 days. New CM was added on day 5 to minimize cell death. Nonadherent cells were harvested on day 9, counted and replated at 3 × 10⁶ cells in 10 mL CM. Differentiated iCD103 DCs were harvested on day 13 or 14.

BMDMs were generated as previously described (29). Bone marrow cells were cultured in DMEM with 20% FBS, and the supernatant was collected from L929 fibroblast cell line culture. After 5 or 6 days of culture, mature macrophages were harvested and stimulated for 24 hours with 10 ng/mL recombinant murine IL4 (M2 macrophage) or with 1 ng/mL LPS (M1 macrophage) for differentiation. Recombinant murine IL17A was added in some groups as indicated.

iCD103 DCs collected on day 13 or 14 of culture were treated with indicated amount of poly (I:C) or IL17A for 24 hours, washed, and placed in coculture assays with splenic CD8⁺ T cells (at 1:10 ratio). CD8⁺ T cells were isolated using anti-CD8 conjugated magnetic beads (Miltenyi Biotec) and were isolated with an auto-MACS cell separator. In some experiments, when OT-I cells were used, iCD103 DCs were pretreated with ovalbumin (10 μg/mL) for 3 hours and washed three times with PBS before coculture with OT-I T cells. ELISA (BD Biosciences) was used for the measurement of concentration of IFNγ in the supernatant collected from cultured cells.

Naïve CD4⁺ T cells were isolated using anti-CD4 conjugated magnetic beads (Miltenyi Biotec) and autoMACS cell separator. Cells were differentiated under Treg polarizing conditions. In brief, 2 to 2.5 × 10⁶/mL cells were activated with 1.5 μg/mL plate-bound anti-CD3 and 1.5 μg/mL soluble anti-CD28 in addition to 10 μg/mL anti-CD4, 10 μg/mL anti-IFNγ, 50 U/mL IL2, and 6 ng/mL TGFβ. Cells were cultured for 3 to 5 days, harvested, and washed for intracellular staining of Foxp3 and the surface staining of CD25 to determine Treg differentiation.

Cell pellets were treated with TRIzol (Life Technologies), and mRNA was extracted as previously described (30). All probes, mouse Arg1 (Mm00475988_m1), mouse Vegf (Mm00437306_m1), and mouse Il10 (Mm01288386_m1) were purchased from Thermo Fisher. All data were normalized to 18S ribosomal RNA (Hs99999901_s1) expression.

In vitro cell migration assay was performed as we have described previously with slight modifications (31). Briefly 24-transwell plates with 5-μm pore-sized membrane inserts (Corning) were used to add 10⁵ iCD103 DCs in 100 μL RPMI medium on the top chambers; lung homogenate collected from Pts4^d/dIl17a^–/– mice or recombinant murine IL17A were added into the bottom chambers. Cells was cultured for indicated times and then the migrated cells in the lower chambers were collected for flow cytometry to detect the cell number of CD103⁺ DCs.

Human broncho-epithelial BEAS-2B cells (ATCCCRL-9609) were purchased from ATCC and cultured according to the ATCC culture condition (BEGM medium (CC-3171, Lonza), growth factors (CC-4175, Lonza), 100 IU/mL of penicillin and streptomycin (100 μg/mL)]. Stable shRNA knockdown of PTEN and SMAD4 in BEAS-2B cells are previously described (18).

Matrigel-coated invasion chambers (24-well; BD Biosciences) were used to measure cell invasion following the manufacturer's suggestions and as previously described (18). Briefly, BEAS-2B cells were trypsinized, resuspended in serum-free BEGM medium, and seeded at 1 × 10⁵ cells per 100 μL per top chamber. Chambers were placed in complete culture medium containing growth factor as chemoattractants. Following 22 hours of incubation, cells in the top chambers were removed; chambers were then fixed with 4% paraformaldehyde solution and stained with crystal violet for cell counting. BEAS-2B cell viability was measured by CellTiter 96 Aqueous Solution Reagent (Promega) according to the manufacturer's instructions. After 1-hour incubation, the cell viability was determined by live/dead dye.

For the comparison of immune cell profiles, gene expression, and cytokine production in mouse experiments or in vitro cell culture, we used the Student t test, the paired Student t test or one-way analysis of variance (ANOVA) test, and Bonferroni multiple comparison test. Linear regression was used to evaluate the correlation between Th17 and PD-1⁺ T cells in lung draining lymph nodes in human subjects. All data shown are the mean ± standard error of the mean (SEM), and all analyses were performed with the Prism software (GraphPad Software).

We examined T helper cell subsets in paired samples using TILs and tumor-free adjacent lung tissue (T-FAT) procured from patients undergoing lung resection for early stages (I–II) of NSCLC. Overall, there were comparable numbers of IFNγ-producing T helper type 1 (Th1) cells and cytotoxic T lymphocytes (CTL) among freshly isolated cells in TILs and T-FAT (Supplementary Fig. S1A and S1B). However, using the same paired samples, we found significantly increased IL17A-producing Th17 cells in early-stage TILs compared with T-FAT (Fig. 1A and B). As expected, we found increased expression of PD-1, an exhaustion marker on CD4⁺ T cells within lung tumors (Fig. 1C). Lung cancer is often diagnosed in late stages (III–IV; ref. 1), when surgical resection is not feasible. Therefore, to gain access to fresh lung tissue with advanced tumor, we examined the association between Th17 cells and late-stage NSCLC using locoregional and distal lymph nodes collected in patients undergoing nodal staging with EBUS-guided sampling (23). Compared with tumor-free lymph nodes (T-FLN), tumor-positive lymph nodes (T-PLN) showed similar abundance of Th1 cells (Supplementary Fig. S1C), whereas Th17 cells were significantly reduced in the latter tissue (Fig. 1D and E). Similarly, in late-stage NSCLC, we found increased expression of PD-1 in CD4⁺ T cells and CTLs in T-PLN when compared with T-FLN (Fig. 1F and G; Supplementary Fig. S1C). Th17 cells and PD-1 expression on CD4⁺ T cells were negatively correlated in T-PLN samples but not in T-FLN (Fig. 1H and I). Together, these findings suggest that Th17 cells may modulate the immune response to NSCLC in humans.

Genomic studies have identified specific and overlapping molecular subtypes in NSCLC, indicating that cellular heterogeneity could affect clinical outcome (32). Examination of NSCLC cases in TCGA cohort identified three molecular subtypes associated with squamous-carcinoma (e.g., SQ.1, SQ.2a, and SQ.2b) and six with adenocarcinomas (e.g., AD.1. AD.2, AD.3, AD.4, AD.5a, and AD.5b; ref. 19). As expected, in the TCGA cohort, NSCLC with KRAS mutations segregated with most adenocarcinoma subtypes, whereas PTEN mutations, copy loss, and lower differential expression were associated with SQ subtypes (Fig. 2A). RNAseq profiles in mice with Club cell-specific deletion of Pten and Smad 4 (Pts4^d/d), which spontaneously develop lung cancer beginning at 7 months of age (18), showed gene signature patterns associated with SQ.2b and AD.1 types of human NSCLC, indicating the relevance of this autochthonous model of non-Kras NSCLC to human disease (Fig. 2B and C).

To examine the role of Th17 cells in NSCLC, we immunophenotyped inflammatory cells recruited to early- and late-stage lung cancer using the Pts4^dd/-driven model of lung cancer (18). At 5 months of age, Pts4^d/d mice exhibit lung epithelium hyperplasia without evidence of tumor (18), but showed abnormal expansion of Th17 cells in the lung tissue as compared with wild-type (Pts4^f/f) mice (Fig. 3A and B; Supplementary Fig. S2A). Further, at 11 to 12 months of age, when nearly all mice exhibit lung tumor with distant metastasis (18), we found increased Th17, PD-1⁺ T cells, and CTLs in TILs when compared with T-FAT (Fig. 3C–F; Supplementary Fig. S2B). Consistent with advanced-stage NSCLC in humans, mediastinal lymph nodes also showed decreased Th17 cells and increased PD-1⁺ T cells in Pts4^d/d mice when compared with Pts4^f/f mice (Fig. 3G–J; Supplementary Fig. S2C). In preneoplastic stage, we also detected increased IL6 and IL1β in the lungs of Pts4^d/d mice, indicating that the lung microenvironment is conditioned to promote Th17 cells (Supplementary Fig. S3). These findings further demonstrate an association between advanced NSCLC and the presence of Th17 cells in the lung.

IL17A expression has been implicated in the pathogenesis of several different tumors in humans and in animal models of cancer (12, 14). We next crossed Pts4^d/d with Il17a^–/– mice, to investigate a link between upregulation of the cytokine Il17a and lung tumor invasion capacity. We found that compared with IL17a-sufficient Pts4^d/dmice, global loss of Il17a^–/– (Pts4^d/dIl17a^–/–) resulted in earlier development of tumors (Fig. 4A–C) and more metastasis (Fig. 4D). Although deletion of Il17a resulted in more aggressive tumors, histologic examination of tumors showed similar expressional pattern of tumor markers (Fig. 4E). Further, IL17A failed to increase invasiveness (Supplementary Fig. S4A and S4B) or viability (Supplementary Fig. S4C) of human epithelial cells lacking PTEN and SMAD4, indicating that this cytokine does not directly play a role in transformed epithelial cell function.

Our data thus far indicated that TILs in early-stage NSCLC were enriched in Th17 cells, and Il17a deficiency decreased tumor latency and increased metastasis in the Pts4^d/d model of lung cancer, indicating a potential protective effect of this cytokine in tumor progression. Therefore, we next examined the possible mechanism for IL17A-mediated tumor suppression in the lung tumor microenvironment.

Because we have previously shown that Pts4^d/d mice develop lung cancer at approximately 7 months of age, we comprehensively surveyed lung immune phenotypes during pre- and peri-lung tumor development (e.g., 5- and 7-month-old) in this model of NSCLC. Compared with wild-type (Pts4^f/f) mice, Pts4^d/d mice at 5 months showed more tumor-associated macrophages (TAM), marked by CD11b/CD206 expression, and regulatory T cells (Treg) in the lungs (Supplementary Fig. S5A–S5D). However, examination of single cells isolated from the whole lung in Pts4^d/dIl17a^–/–mice showed more TAMs and Tregs when compared with Pts4^d/d mice of the same age group (Supplementary Fig. S5A–S5D).

At 7 months of age, when approximately 40% of Pts4^d/d and 60% of Pts4^d/dIl17a^–/–mice develop lung tumors, we found increases in PD-1 expression and Treg numbers, and decreases in IFNγ⁺ expression and CD8⁺ cytotoxic T-cell (CTL) numbers in Pts4^d/dIl17a^–/–mice, when compared with Pts4^d/d and Pts4^f/f mice (Fig. 5A and B), indicating an immunosuppressive tumor microenvironment in the absence of IL17A. Immunosuppressive cells did not affect IFNγ expression in CD4⁺ T cells because Th1 responses remained comparable in all three genotypes (Fig. 5A and B). In contrast, CD103⁺CD11b^lowmyeloid DCs, which cross-present antigen and prime CD8⁺ T cells (33, 34), were decreased in Pts4^d/dIl17a^–/–when compared with Pts4^d/d mice (Fig. 5C–E). In addition to the increased abundance of PD-1 expression in T cells, PD-L1–expressing CD11c⁺ CD11b^highmyeloid DCs were more abundant in Pts4^d/dIl17a^–/–mice (Fig. 5C and E). Together, these data suggest IL17A is required for initiation of antitumor responses in this model of NSCLC, possibly through upregulation of immunosuppressive PD-1/PD-L1. Further, consistent with an immunosuppressive microenvironment, we also found significantly increased TAMs (CD11b^highCD206⁺) in Pts4^d/dIl17a^–/–mice (Fig. 5F and G) that expressed significantly more arginase-1 (Arg1), VEGF (Vegf), and IL10 (Il10), factors that are known to promote tumor growth (Fig. 5H; ref. 35). Immunosuppressive myeloid-derived suppressor cells (MDSC) marked by CD11b/Gr-1 were not affected by lack of IL17A in this model (Supplementary Fig. S5E and S5F). These findings suggest that IL17A deficiency promotes TAMs and PD-L1⁺ DCs and limits abundance of CD8-priming CD103⁺ DCs.

We next assessed the phenotype of mediastinal tumor-associated sentinel lymph nodes, in the same group of mice (7-month-old). We found significantly fewer CTLs and Th1 cells in Pts4^d/dIl17a^–/–when compared with IL17A-sufficient Pts4^d/dmice (Supplementary Fig. S6A and S6B). Additionally, CD44, a marker of effector T cells, was upregulated in both CD4⁺ and CD8⁺ T cells in Pts4^d/d when compared with Pts4^d/dIl17a^–/–mice (Supplementary Fig. S6C and S6D), indicating defective T-cell activation in lung tumor sentinel lymph nodes in the absence of IL17A.

We found CTL and Th1 cell responses in metastatic sites (e.g., stomach) in Pts4^d/d mice; similar T cells were fewer in Pts4^d/dIl17a^–/–mice (Supplementary Fig. S6E and S6F). CD103⁺CD11b^low DCs that are capable of efficient antigen cross-presentation and are present in metastatic tissues in Pts4^d/dIl17a^–/–mice were reduced when compared with IL17A-sufficient controls (Supplementary Fig. S6G and S6H). These findings suggest an immunosuppressive microenvironment in the lungs, tumor draining lymph nodes, and distant metastatic sites in Pts4^d/dIl17a^–/–mice.

Our findings thus far pointed to a lack of activated endogenous immune cells in the lungs of Pts4^d/dIl17a^–/–mice that resulted in decreased tumor latency when compared with Il17a-sufficient Pts4^d/d mice. Recombinant (r)IL17A did not directly affect Treg differentiation (Supplementary Fig. S7) or CD8⁺ T-cell proliferation and activation in vitro (36). However, because we had found an increase of CD206⁺ M2 macrophage-like TAM from tumor-bearing Pts4^d/dIl17a^–/–mice, we next examined whether IL17A directly affects differentiation in this cell population. We found that IL4-mediated differentiation of bone marrow-derived macrophages (BMDM) to CD206⁺ M2 cells was inhibited by IL17A in a dose-dependent manner (Supplementary Fig. S8A and S8B). Moreover, lung F4/80⁺ macrophages from Pts4^d/dIl17a^–/– mice exhibited less differentiation of CD206⁺ cells when treated with IL17A in vitro (Supplementary Fig. S8C and S8D). Expression of the M2 macrophage signature genes Arg1 and Vegf were also inhibited by IL17A treatment (Supplementary Fig. S8E). Il17a-deficient bone marrow cells did not display an intrinsic defect in M2 differentiation in vitro (Supplementary Fig. S9). We conclude that IL17A suppressed genes associated with M2-like macrophage differentiation, because its absence resulted in abundant TAM and exaggerated lung tumor progression in Pts4^d/d mice.

Next, we examined whether IL17A could induce differentiation and function of CD103⁺ (iCD103) DCs (Supplementary Fig. S10A). Exogenous addition of rIL17A to bone marrow derived cells failed to promote iCD103 differentiation and/or activation as determined by Clec9a expression, a functional marker for DC-mediated CTL cross-priming (Supplementary Fig. S10B; ref. 37). Because of reduced CD103⁺ DCs in the lungs of Pts4^d/dIl17a^–/–mice, we next examined whether IL17A might act as a chemoattractant and recruit CD103⁺ DCs into the lungs. Using a Transwell assay system, we found that IL17A in a dose-dependent manner enhanced iCD103 DC migration (Fig. 6A). Moreover, iCD103 DCs showed increased migration when Pts4^d/dIl17a^–/–lung homogenates were supplemented with rIL17A (Fig. 6B). These findings support the notion that IL17A may be required for efficient recruitment of CD103⁺ DCs into the lungs of Pts4^d/d mice. In an autocrine/paracrine manner, rIL17A treatment induced expression of IL17A receptor in iCD103 DCs (Fig. 6C and D). Costimulation molecule CD86 was also upregulated in CD103⁺ DCs treated with rIL17A (Fig. 6C and E; Supplementary Fig. S11).

Toll-like receptor (TLR)3 activation is known to promote DC maturation and prime CD8⁺ T cells (38), therefore, we compared IL17A–mediated activation of DCs relative to TLR3 stimulation. Autologous splenic CD8⁺ T cells, cocultured with the TLR3 agonist poly I:C, activated iCD103 DCs and induced CTL responses (Fig. 6F). Treatment with rIL17A in a dose-dependent manner also resulted in iCD103-mediated enhanced CD8⁺ T-cell priming in both a nonspecific (Fig. 6F) and an ovalbumin-specific transgenic T-cell (OT-I) model (Fig. 6G), consistent with IL17A-induced activation of iCD103 DCs. Together, these data suggest that the mechanism for increased IL17A-mediated antitumor responses may be recruitment and activation of CD103⁺ DCs to the primary and metastatic tumors and promotion of CTL development.

Although Th17 cells are considered a potent source of IL17A, other cells (e.g., γδ T cells, monocytes, and neutrophils) also express IL17A under homeostatic or activated conditions. To determine whether Th17 cells specifically contribute to the antitumor immune response in the Pts4^–/– model of NSCLC, CD4⁺ T cells isolated from splenocytes of WT (Pts4^f/f) mice were adoptively transferred to Pts4^d/dIl17a^–/–mice at 2.5 and 5 months of age. IL17A-sufficient T cells persisted over 5 months and significantly reduced lung tumor incidence and stomach metastasis in Pts4^d/dIl17a^–/–mice at 7 to 8 months of age (Fig. 7A–C). Furthermore, CTL responses and CD103⁺ DCs were upregulated in Pts4^d/dIl17a^–/–mice after adoptive transfer of IL17A-sufficient CD4⁺ T cells as compared with control (Fig. 7D and E). Consistently, CTL responses were stronger in lung mediastinal nodes in response to CD4⁺ T-cell transfer, whereas CTLA-4 expression, a marker of T-cell exhaustion, was reduced (Fig. 7F and G). Together, these findings support a homeostatic effect of IL17A-expressing T cells that is required to activate CTLs and suppress lung tumor progression and metastasis in Pts4^d/dIl17a^–/–mice.

Th1 and Th17 cells are associated with emphysema development (5), which is linked to increased risk of lung cancer (39). Carcinogens in cigarette smoke promote large numbers of somatic mutations that result in production of altered proteins in premalignant and transformed airway epithelial cells (40). Studies of tumor survivors and animal models (41) show that CTLs that target mutated tumor antigens are more abundant and are sustained longer than CTLs that target nonmutated tumor antigens (42). However, how accumulation of altered proteins shapes early protective CTL tumor responses remains unclear. CD4⁺ T cells within TILs are thought to play a role in development of effective CTLs (43). To assess their role, we examined the relative abundance of different subset of CD4⁺ and CD8⁺ T cells within and outside the lung tumor microenvironment in early and late stages in NSCLC. We found increased relative abundance of Th17 cells in early-stage NSCLC and an inverse relation between the T-cell exhaustion marker (PD-1) and Th17 cells in late-stage tumor, indicating a role for Th17 cells in human lung cancer. This premise was further supported in our preclinical model of adenosquamous lung cancer that shares features with human NSCLC (18).

We have previously shown that loss of Pten in mice with a codon-optimized Cre recombinase results in hyperplasic airway epithelia, whereas concurrent loss of Smad4 in these mice (Pts4^d/d) results in proximal airway adenosquamous lung tumors (18) and promotes recruitment of acquired immune cells. Lung tumors in Pts4^d/d mice share characteristics with the human SQ2b and AD.1 subtypes of NSCLC as determined by RNAseq signature. Using the TCGA database, we have previously shown that the fraction of estimated mutations varies for mutational signatures and by smoking status in NSCLC molecular subtypes (19). Compared with other subtypes, SQ.2b and AD.1 show the highest proportion of patients who were active- or short-term former-smokers and have the highest manifestation of the SI4 transversion signature, a smoking-related signature of cytosine to adenine (C > A) transversions (19). Our work here underlines the importance of antitumor immunity with respect to tumor molecular diversity (19). Prior work has shown that global loss of Il17a can prevent tumor formation in the Kras-driven model of NSCLC (14). Here, we show an opposite response, whereby Pts4^d/d/Il17a^–/– mice display reduced tumor latency and increased metastasis. We found that IL17A expression is upregulated early in response to tumor in the lungs and constitutes an immune response to hyperplastic and premalignant airway epithelial cells in NSCLC distinct molecular subtype. These observations are supported by our findings that global loss of Il17a results in earlier malignant transformation and increased distant organ metastasis when compared with Il17a-sufficient Pts4^d/d mice. Adoptive transfer of Il17a-sufficient CD4⁺ T cells to Pts4^d/d/Il17a^–/– mice reversed early tumor development and metastasis, indicating the antitumor function of Th17 cells.

Several studies have identified protumor roles for Th17 cells and IL17A in preclinical models and in human of NSCLC (14, 44), whereas some have implicated antimalignant function (10). Although preclinical studies have examined the activity of IL17A and Th17 cells in the context of mice with immune deficiency (45), human NSCLC develops under immune-competent conditions. Studies in mouse models and human breast (46), colon (47), and pancreatic (48) cancers have suggested that IL17A activates several transcriptional factors and can promote tumor growth and metastasis. In our study, however, we found no evidence for IL17A to directly induce cell proliferation or an invasion phenotype of human airway epithelial cells that lack PTEN and SMAD4. However, consistent with the proinflammatory nature of pten/smad4-driven tumors, we have previously found suppressed TGFβ1 signaling (18) and increased recruitment of Th17 cells prior to tumor development the lungs of Pts4^d/d mice, suggesting that the endogenous immune system suppresses lung tumors.

We and others have previously shown that cigarette smoke induces Th17 cell–mediated lung inflammation, and IL17A and its receptor are critical for emphysema development in mice (5, 26, 49). Although induction of smoke-induced Th17 cells seemly counters the antitumor function of IL17A as shown here, several carcinogens in cigarette smoke activate proinflammatory cytokines, in particular IL6, which induces tumor growth (50, 51). Thus, we propose that induction of Th17 cells may represent an immune response to tumor subtypes and to the protumorigenic effect of inflammatory cytokines. Using a non-Kras–driven preclinical model of lung cancer, here, we have provided insight into the role of Th17 cell–bias in a subtype of NSCLC. Our findings show that Th17 cells increase early in human NSCLC, and that Th17 cells correlate negatively with an immune suppressive microenvironment in tumors. Adoptive transfer of T cells isolated from TILs to patients with metastatic disease has shown promising results, although such immune-targeted therapy is limited to solid tumors (7, 52). Our findings here suggest that an induced bias of TILs toward Th17 cells in non-Kras–driven tumors may boost adoptive T-cell therapeutic functions in NSCLC.

Current adjuvant cancer therapies that target immune checkpoint molecules (e.g., PD-1/PD-L1, CTLA-4) can elicit durable, albeit low (20%), clinical responses in NSCLC (53). Although expression of these markers in the tumor microenvironment does not correlate with clinical responses to immune checkpoint inhibition in humans (54), increased somatic mutation, which can promote expression of mutated proteins and their presentation by activating CD103⁺ DCs in the tumor landscape, has been a reliable predictor of favorable response to immune adjuvant therapy (55). High frequency of failed clinical response to immune checkpoint inhibitors is thought to be due to the presence of TAMs, MDSCs, and cancer-associated fibroblasts that mediate T-cell exclusion in CD4⁺ T cells and CTLs in solid tumors (4). We showed here that IL17A can inhibit TAM differentiation and promote activation of CD103 DCs, which are capable of antigen cross-presentation, alter the immunosuppressive tumor microenvironment, and promote a more effective antitumor immune response. We showed that IL17A, in a dose-dependent manner, inhibited Arg-1 and Vegf expression in TAMs, and Th17 cells restored tumor latency in Pts4^d/dIl17a^–/– mice comparable with Pts4^d/dmice. Moreover, we showed that Th17 cells recruited CTL-priming CD103⁺ DCs to the lungs and activated CD8⁺ T cells, necessary for antitumor responses in vivo.

In summary, we show that the immune phenotype of CD4⁺ T cells in early- and advanced-stage human NSCLC varies, and that PD-1 expression in CD4⁺ T cells and Th17 cells shows an inverse correlation. Using Pts4^d/dmice, a model of NSCLC that shares with human NSCLC a propensity for spontaneous distant metastasis, we show that IL17A functions in early endogenous antitumor responses. Global deficiency in Il17a in Pts4^d/dmice resulted in early and more invasive lung tumors, indicating a protective role for this cytokine in NSCLC. The mechanisms for Th17 cell–mediated antitumor activities are in part through inhibition of TAMs, whereby CD103⁺ DCs can be activated and recruited in the lungs. The immunosuppressive role of TAM and CTL-priming role of CD103 DCs have been reported in other cancers (56) and may also extend to the Pts4^d/d model and/or human NSCLC. Adoptive transfer of IL17A-sufficient CD4⁺ T cells reduced lung cancer and metastasis in IL17-deficient Pts4^d/d mice. Although IL17A has been shown to promote tumors associated with Kras mutations, our studies here support a role for Th17 cells in antitumor immune responses in tumors independent of the Kras mutation. Thus, molecular phenotyping of lung cancer should be considered prior to immune therapy. We propose that Th17-based immune therapy could be considered in some cases of NSCLC. Adoptive transfer of TILs differentiated to express IL17A in NSCLC patients with metastatic disease could provide a more effective result in immune-targeted cancer therapy.

R.F. Casal reports receiving commercial research grants from Olympus, Siemens, and Concordia and is a consultant/advisory board member for Olympus. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R. You, R.F Casal, W. Lu, H.-Y. Tung, S. Paust, F. Kheradmand

Development of methodology: R. You, F.J. DeMayo, W. Lu, A. Hill-McALester, F. Kheradmand

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. You, J. Liu, B. Burt, R.F. Casal, D.R. Lazarus, W. Lu, X. Yuan, A. Hill-McALester, M. Kim, S. Perusich, L. Cornwell, S. Paust

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. You, C.J. Creighton, W. Lu, A. Hill-McALester, D. Corry, F. Kheradmand

Writing, review, and/or revision of the manuscript: R. You, F.J. DeMayo, J. Liu, B. Burt, R.F Casal, D.R. Lazarus, W. Lu, A. Hill-McALester, L. Cornwell, D. Rosen, S. Paust, G. Diehl, D. Corry, F. Kheradmand

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.-N. Cho

Study supervision: L.-z. Song, G. Diehl

This work was supported in part by the R01HL117181-01 (to F. Kheradmand and D. Corry), CX000104 (to F. Kheradmand), and BX002221 (to D. Corry) VA Merit Awards. This project was supported by the Cytometry and Cell Sorting Core at Baylor College of Medicine with funding from the NIH (AI036211, CA125123, and RR024574), Mouse Phenotyping Core at Baylor College of Medicine with funding from the NIH (UM1HG006348), and the expert assistance of Joel M. Sederstrom and Dr. Corey Reynolds. This project was done in collaboration with the Adrienne Helis Malvin Medical Research Foundation through its direct engagement in the continuous active conduct of medical research in conjunction with Baylor College of Medicine and the Immune Therapy in a Novel Humanized Model of Lung Cancer Program (to F.J. DeMayo and F. Kheradmand) and the Natural Killer Cell Immunotherapy to Cure Lung Cancer (to S. Perusich and F. Kheradmand). C.J. Creighton was supported in part by NIH grant CA125123 and Cancer Prevention and Research Institute of Texas grant RP120713.

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.