Background

Due to the distinctive pulmonary microenvironment, the lungs are frequently identified as a primary site for metastasis in a wide range of malignant tumors,1 encompassing breast cancer, osteosarcoma (OS), prostate cancer, colorectal cancer, and other types. Despite the numerous and increasingly effective therapies for localized tumors, the prognosis for patients with established lung metastases remains highly problematic.2 The treatment of lung metastasis poses significant challenges due to various factors. First, established lung metastases typically indicate an advanced stage of cancer where the primary tumor has already spread to other distant sites. This advanced stage often results in a more aggressive and complex disease presentation. Second, the complexity of lung metastasis management extends to the limited treatment options available. Lung metastasis is known to exhibit resistance to conventional cancer treatments such as chemotherapy and radiation therapy.3 Cancer cells that have metastasized to the lungs can acquire genetic alterations and develop resistance mechanisms, limiting the effectiveness of standard treatment approaches.4 5 Although targeted therapies and immunotherapies have shown promise in certain solid tumors, their applicability depends on the tumor’s specific molecular characteristics and the availability of suitable targeted agents.6–9 Thus, there is still an urgent need to find novel strategies to improve the prognosis of patients with established lung metastases.

T helper 9 (TH9) cells were first identified by Veldhoen and Dardalhon with the priority of secreting interleukin (IL)-9 and IL-1010 11 and related to the development of many autoimmune diseases.12–15 In addition to the proinflammatory ability, TH9 cells exhibit superior efficacy in inhibiting the growth and metastasis of diverse solid tumors compared with TH1 or TH17 cells.16–22 These results suggest the potential of TH9 cells as potent T cells for cancer-adoptive cell therapy. Although studies have evaluated the therapeutic efficacy of TH9 cells against cold tumors such as triple-negative breast cancer (TNBC) and OS, which supports the potential clinical application of TH9-based adoptive cell therapy, the antitumor activity of TH9 cells has been primarily investigated in melanoma.23–25 Additionally, the protumor effect of TH9 cells was also reported in lung cancer.26 27 Therefore, the multifaceted role of TH9 cells remains to be further elucidated.

In the current study, we first tested the antitumor effects of TH9 cells against OS and TNBC in different models. Our results showed that TH9 cells, compared with TH1 and TH17 cells, exhibit superior antitumor efficacy toward the established lung metastases model but not the subcutaneous and primary model. We demonstrated this was associated with the innate lung affinity of TH9 cells driven by the CXCR4-CXCL12 chemoattraction axis. Further experiments proved that tumor necrosis factor (TNF) receptor-associated factor 6 (TRAF6)-mediated activation of NF-κB signaling in TH9 cells led to the phosphorylation of inhibitor of nuclear factor kappa B kinases (IKKs), which inhibited the activity of E3 ubiquitination ligase ITCH, resulting in the attenuated ubiquitination of CXCR4 in TH9 cells. TH9 cells can enhance the proportion of immune cell subsets via IL-9, resulting in a reshaped tumor microenvironment at lung tumor sites. Infusion of TH9 cells increases the number of CD8+T cells at tumor sites, and combining TH9 cells therapy and anti-programmed cell death protein-1 (PD-1) enhances antitumor effects against established lung metastases.

MethodsSex as a biological variable

Our study exclusively examined female mice. It is unknown whether the findings are relevant for male mice.

Human samples

Patients diagnosed with OS lung metastasis between September 2016 and December 2022 at the Musculoskeletal Tumor Center of the Department of Orthopedics at The Second Affiliated Hospital of Zhejiang University School of Medicine were recruited to this study for immunohistochemistry analysis. In total, 22 cases of formalin-fixed, paraffin-embedded tissue blocks of OS lung metastasis were included. Experienced specialists performed all surgeries. All the patients were informed of the usage of their tissue samples.

Mice and cell lines

BALB/C (5–8 week-old) mice were purchased from SLAC (Shanghai, China). The Institutional Animal Care and Use Committee approved all animal studies. DO11.10 mice were donated by Dr Fang Zhang (Medical School of Nanjing University, Nanjing, China). CD45.1 mice were bred by ourselves. Cxcr4flox/flox mice were bought from Cyagen Biosciences (Suzhou, Jiangsu Province, China) and then crossed with Cd4cre mice to produce Cd4cre Cxcr4flox/flox mice. Mice were housed in a specific pathogen-free facility, and experimental protocols were approved by the Animal Care and Use Committee of the School of Medicine, Zhejiang University (Ethical approval number: 26253).

There were three to five animals in each group. Five animals of the same group live in a single animal cage. The sample size was determined based on the number of experimental groups and different time points of analysis. The experiments were conducted in a random manner. In each experiment, mice of the same sex and age were randomly divided into different groups. After allocation, each group comprised an equal number of animals with comparable weights, aiming to minimize experimental error. No criteria were set for including or excluding animals. No data points were excluded from the analysis.

Murine K7M2 and 4T1 cells were obtained from the Cell Bank of Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). The K7M2-ovalbumin (OVA) cells, K7M2-Luci-OVA cells, 4T1-OVA cells and 4T1-Luci-OVA cells were transfected with lentiviral transfection in our laboratory. K7M2-OVA cells, K7M2-Luci-OVA cells, 4T1-OVA cells and 4T1-Luci-OVA cells were cultured in Dulbecco’s Modified Eagle’s Medium supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, California, USA), and 1% penicillin/streptomycin. Cells were maintained at 37°C in 5% CO2. All cells were routinely tested for mycoplasma contamination using the Mycoplasma Detection Kit (ab289834, Abcam, Cambridge, Massachusetts, USA) and were found to be negative.

Tumor growth experiments

For K7M2 lung OS and 4T1 lung breast cancer models, mice were injected intravenously with 1×105 4T1-OVA-luci or K7M2-OVA-luci in 100 µl of phosphate-buffered saline (PBS) (P1010, Solarbio, Beijing, China). Mice were transferred with 3×106 OVA-specific TH1, TH9, or TH17 cells resuspended in 100 µl PBS intravenously on day 5, day 12 and day 19. At day 7, day 14 and day 21, mice were anesthetized with 1% pentobarbital sodium and injected intraperitoneally with luciferin (P1043, Promega, Madison, Wisconsin, USA) at 100 μg kg−1 of mice weight. After 10 min of luciferin injection, images were acquired with the IVIS Spectrum In Vivo Imaging System (PerkinElmer, Waltham, Massachusetts, USA). Total photon flux in the lung area was analyzed with Living Image software (PerkinElmer).

For the subcutaneous tumor model, mice were injected with 1×106 4T1 OVA or K7M2-OVA cells in 100 µl of PBS subcutaneously. For the primary tumor model, mice were injected into the mammary fat pad of the mice with 1×106 4T1 OVA or injected into the bone marrow cavity of the mouse tibia with 1×106 K7M2-OVA cells in 20 µl of PBS. Tumor volume was monitored by caliper every other day after 7 days of injection and calculated by the formula V=a × b2/2 (a: the maximal diameter of the tumor, b: the minimal diameter).

In subcutaneous and primary tumor models, mice were transferred with 3×106 OVA-specific TH1, TH9, or TH17 cells resuspended in 100 µl PBS intravenously on day 7 and day 14. For in vivo blockade of IL-9, individual mice were injected with 100 µg InVivoMab anti-mouse IL-9 (BE0181, clone: 9C1, Bio X Cell, West Lebanon, New Hampshire, USA) intraperitoneally every 2 days since the same day of tumor inoculation. To eliminate CD8+ T cells in vivo, individual mice received 40 µg anti-CD8 (BE0061, clone: 2.43, Bio X Cell) via intraperitoneal injection every 2 days since the same day of tumor inoculation. For in vivo blockade of PD-1, individual mice were injected with 100 µg InVivoMab anti-mouse PD-1 (BE0146, clone: RMP1-14, Bio X Cell, West Lebanon, New Hampshire, USA) intraperitoneally every 2 days since the same day of tumor inoculation. According to the criteria of the Animal Care and Use Committee of the School of Medicine, Zhejiang University, when the tumor size was over 2000 mm3, the tumor-bearing mice were euthanized by an intraperitoneal injection of 50 mg kg−1 pentobarbital sodium.

Transfer of CD45.1-TH cells subsets into CD45.2 mice

CD45.1-naïve CD4+ T cells were generated from CD45.1 mice and in vitro polarized to CD45.1-TH1, TH9 and TH17 cells under polarization condition medium for 4 days. In some experiments, 1×107 CD45.1-TH9 cells were intravenously transferred into CD45.2 mice in 100 µl PBS. 48 hours later, CD45.2 mice were sacrificed, and the percentage of CD45.1+ T cells in the lung, liver, spleen, bone marrow and lymph node were analyzed by flow cytometry. In some experiments, 1×106 CD45.1-TH1, TH9 and TH17 cells were intravenously transferred into CD45.2 mice in 100 µl PBS. 48 hours later, CD45.2 mice were sacrificed, and the percentage of CD45.1+ T cells in the lung was analyzed by flow cytometry.

In some experiments, individual mice were intraperitoneally injected with 1 mg kg−1 ML339 (HY-122197), 1 mg kg−1 C-021 (HY-103364), 1 mg kg−1 R243 (HY-122219) or 1 mg kg−1 AMD3100 (HY-10046) in 100 µl PBS 1 day before transfer of TH cell subsets. The inhibitors were all purchased from MCE (New Jersey, USA).

In vitro CD4+ T cell culture

Naive CD4+ CD62L+ cells were isolated using EasySep Mouse CD4+ T Cell Isolation Kit (19852, STEMCELL Technologies, Vancouver, BC, V6A 1B6, Canada) and EasySep Mouse Biotin Positive Selection Kit II (17665, STEMCELL Technologies). Naive CD4+ T cells were seeded into 48-well plates with plate-bound anti-CD3 (BE0001-1, 2 μg mL−1, Bio X Cell) and anti-CD28 (BE0015-5, 2 μg mL−1, Bio X Cell). 1 × 106 cells per well, and polarized into effector CD4+ T lymphocyte subsets for 4 days without cytokines, and with anti-IFN-γ (BE0054, 10 μg mL−1, Bio X cell) and anti-IL-4 (BE0045, 10 μg mL−1, Bio X cell) (TH0 cells); with IL-12 (130-096-707, 20 ng mL−1, Miltenyi Biotec, Bergisch Gladbach, Germany) and anti-IL-4 (10 μg mL−1) for TH1 cells; with TGF-β1 (130-095-067, 2 ng mL−1, Miltenyi Biotec), IL-4 (20 ng mL−1), and anti-IFN-γ (10 μg mL−1) for TH9 cells; with TGF-β1 (2 ng mL−1), IL-6 (130-094-065, 25 ng mL−1, Miltenyi Biotec), anti-IFN-γ (10 μg mL−1), and anti-IL-4 (10 μg mL−1) for TH17 cells.

In vivo fluorescence imaging

For the T cell homing experiment, T cells stained by 1,1-dioctadecyl-3,3,3,3 tetramethylindotricarbocyaine iodide (DiR) (5 µM, Caliper Life Sciences, Boston, USA) for 20 min at 37°C. After being thoroughly washed in PBS three times, 3×106 T cells were delivered intravenously in 100 µl PBS. IVIS Spectrum Animal Imaging System was used to evaluate the homing ability of T cells 48 hours after T cells transfusion.

In vitro tumor-specific cytotoxicity of TH1, TH9 and TH17 cells

OVA-specific naïve CD4+ T cells were generated from OVA-specific DO11.10 mice and in vitro polarized to OVA-specific TH1, TH9 and TH17 cells under polarization condition medium for 4 days. K7M2-OVA or 4T1-OVA and K7M2-wild-type (WT) or 4T1-WT tumor cells were labeled with CFSE (Thermo Fisher Scientific, Waltham, California, USA). K7M2-OVA or 4T1-OVA cells were labeled with 5 µM CFSE (CFSEhigh), while K7M2-WT or 4T1-WT cells were labeled with 0.5 µM CFSE (CFSElow) for 10 min at 37°C. Then the tumor cells were mixed at a 1:1 ratio and seeded into 96-well plates (5×104 cells/well; 2.5×104 K7M2-OVA cells and 2.5×104 K7M2-WT cells; 2.5×104 4T1-OVA cells and 2.5×104 4T1-WT cells). OVA-specific TH1, TH9 and TH17 cells (5×105) were added and incubated for 24 hours. A mixture of tumor cells cultured alone was regarded as the control. 24 hours later, the cells were collected and stained by Fixable Viability Dye (FVD) eFluor™ 780 and CD4 antibodies to dissect the surviving tumor cells (FVD− CD4− CFSE+) and detected by flow cytometry. Ratio = % (CFSEhi) peak/% (CFSElo) peak. Tumor-specific lysis = (1−(Control ratio/Experimental ratio)) × 100.

Flow cytometry

Intranuclear staining was carried out with fixation/permeabilization buffer solution (00-5123-43 and 00-5223-56, eBioscience, San Diego, California, USA). For intracellular staining, cells were stimulated for 4 hours at 37°C in a medium containing PMA (P1585, 50 ng mL−1, Sigma-Aldrich, St Louis, Missouri, USA), ionomycin (I3909, 1 μg mL−1, Sigma-Aldrich), and brefeldin A solution (00-4506-51, eBioscience, California, USA). Then, the cells were subjected to an intracellular staining protocol (00-8222-49, eBioscience) and the stained cells were analyzed using ACEA NovoCyte (Agilent Technologies, California, USA). Data were analyzed using NovoExpress (Agilent Technologies).

Transwell migration assay

Transwell chambers with microporous membranes of 4 µm pore-size (Corning, New York, USA) were used to evaluate the migration ability of TH0, TH1, TH9 and TH17 cells in response to different organ lysates and CXCL12. Briefly, 5×105 differentiated TH cells were seeded onto the upper chamber, and organ lysates or CXCL12 at 100 ng μl−1 were added to the lower chamber. After 4 hours, the cells in the lower chamber were collected and counted by flow cytometry. In some experiments, TH9 cells were pretreated with AMD3100 (5 µg/mL) for 24 hours. Migration index was calculated as the ratio of migrated cells in the presence and absence of organ lysates or CXCL12. Each experiment was done in triplicate.

Real-time PCR

Total RNA was extracted from cells using RNAiso Plus (9109, Takara Biomedical Technology, Beijing, China) and reverse transcribed into complementary DNA (cDNA) with a HiFiScript cDNA Synthesis Kit (CW2569, Cowin Biotech, Beijing, China) according to the manufacturer’s instructions. RT-PCR was performed by ChamQ Universal SYBR qPCR Master Mix (Q711-02, Vazyme, Nanjing, Jiangsu, China) and specific primers in the applied Bio-Rad real-time PCR system. The following thermal cycling conditions were used for PCR: 1 cycle at 95°C for 30 s, followed by 40 cycles at 95°C for 5 s and 60°C for 34 s. The data were analyzed by the 2−ΔΔCt method.

Immunofluorescent

TH0, TH1, TH9 and TH17 cells were collected and fixed with 4% paraformaldehyde for 30 min and permeabilized with 0.1% Triton X-100 for 10 min. Then, the cells were blocked with 3% bovine serum albumin and 5% goat serum and incubated with anti-CXCR4 (sc-53534, Santa Cruz, Santa Cruz, California, USA) at 4°C overnight. The next day, cells were stained with iFluor 594 Conjugated Goat anti-mouse immunoglobulin G Goat Polyclonal Antibody (HA1126, HUABIO, Hangzhou, Zhejiang, China). Nuclei were stained with 4'6-diamidino-2-phenylindole (DAPI) (H-1900–10, Vectorlabs, San Francisco, USA).

For lung tissue section staining, the lungs were dissected and fixed overnight with 4% paraformaldehyde and 20% sucrose. The lungs were cut into 8-µm-thick sections for immunofluorescence staining and applied to glass slides. After being fixed and stained with DAPI (D9542, Sigma-Aldrich), the sections were washed with PBS and examined with an Olympus IX83-FV3000 confocal microscope (Olympus Corp, Tokyo, Japan).

Western blotting

Cells were collected and washed with 1 mL cold PBS three times. Then, the cells were lysed on ice in Radioimmunoprecipitation assay buffer (RIPA) lysis buffer (P0013B, Beyotime Biotechnology, Shanghai, China) for 30 min. Subsequently, the cell lysates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 10% or 12% gels and were transferred onto polyvinylidene difluoride membranes. The membranes were blocked with 5% bovine serum albumin (BSA) in phosphate buffered solution (PBST) buffer and were incubated with primary antibodies overnight at 4°C. After washing three times, the membranes were incubated with secondary antibodies at room temperature for 1 hour. Each membrane was scanned by a Tanon 4500 imaging system (Shanghai, China).

Immunoprecipitation

TH0, TH1, TH9 and TH17 cells were collected and washed three times with cold PBS. Cell extracts were prepared on ice for 30 min using lysis buffer containing 50 mM Tris (pH 7.4), 150 mM NaCl, 0.5% (vol/vol) Nonidet P-40, and 1 mM EDTA supplemented with protease inhibitor cocktail. Lysates were incubated with indicated antibody-coupled beads at 4°C overnight. Immunoprecipitates were washed three times with lysis buffer and subjected to a western blotting assay.

Mass spectrometry

Each pulldown sample was run just in the separation gel and was cut into approximately 1-mm3 pieces, then subjected to in-gel trypsin digestion and dried. Samples were reconstituted in 5 µl of high-performance liquid chromatography solvent A (2.5% acetonitrile and 0.1% formic acid). A nanoscale reverse-phase high-performance liquid chromatography capillary column was created by packing 5-μm C18 spherical silica beads into a fused silica capillary (100 µm inner diameter × ∼20 cm length) using a flame-drawn tip. After the column was equilibrated, each sample was loaded onto the column using an autosampler. A gradient was formed, and peptides were eluted with increasing concentrations of solvent B (97.5% acetonitrile and 0.1% formic acid). As the peptides eluted, they were subjected to mass spectrometry (MS), as described above. MS analysis of the protein content was performed by using a Q Exactive system (Thermo Fisher, Massachusetts, USA).

Small interference RNA transfection

Naive CD4+ T cells were seeded into 6-well plates at 5×106 per well. Then, the cells were transfected with scramble negative control or targeted small interference RNA using TransIT-TKO Transfection Reagent (Mirus Bio, Madison, Wisconsin, USA) according to the manufacturer’s instructions.

Retroviral infection of CD4+ T cells

Retroviruses were produced by transfecting Plat-E cells with 7.5 μg of pMX-Ires-gfp or pMX-Itch-gfp. The cell culture medium was replaced with fresh medium after 10 hours, and the retrovirus-containing supernatant was collected after an additional 72 hours. Naïve CD4+ T cells were first stimulated with anti-CD3 and anti-CD28 antibodies. At the 24 hours and 36 hours time points, activated T cells were infected with 500 μl of the viral supernatant for 1 hour by centrifugation at 1500×g in the presence of 10 μg ml−1 polybrene and incubated at 37°C for an additional 1 hour before removal from the viral supernatant and resuspension in the corresponding T cell medium for 72 hours.

Isolation of tumor-infiltrating lymphocytes

Tumors were dissected and subjected to enzymatic digestion with 1 mg mL−1 collagenase IV (CLS-4, Worthington Biochemical Freehold, New Jersey, USA) and 0.5 mg mL−1 DNase I (DN25, Sigma-Aldrich) at 37°C for 90 min followed by Percoll (GE Healthcare, Uppsala, Sweden) gradient purification.

Measurement of cytokine levels

Lung homogenates were assayed by ELISA to measure the levels of CXCL1, CXCL12, CXCL14, CXCL15 and CXCL16 according to the manufacturer’s instructions. ELISA kits for murine CXCL1 (RK0038), CXCL12 (RK00168), CXCL14 (RK02643), CXCL15 (RK07476) and CXCL16 (RK00417) were all purchased from ABclonal (Wuhan, Hubei, China). Absorbance values were measured with a SynergyMx M5 microplate reader (Molecular Devices, Silicon Valley, California, USA).

Statistical analysis

All data are expressed as the mean±SEM. Statistical analyses were performed with GraphPad Prism V.7.0 software. All the data were performed the Shapiro-Wilk test to confirm their distribution. For the normally distributed data, unpaired Student’s t-test analyzed the differences between two groups of data, and differences among three or more groups were analyzed by one-way analysis of variance (ANOVA) or two-way ANOVA followed by a post hoc Tukey test. For the non-normally distributed data, the Mann-Whitney test was used for means comparison between the two groups. The Kruskal-Wallis test was used for the means comparison of multiple individual data sets. Overall survival was assessed by Kaplan-Meier analysis. The Spearman’s rank-order correlation test was used for Pearson’s correlation analysis. A value of p<0.05 was considered statistically significant. The sample size ranges from 3 to 5 according to the effect of a 50% increase and 25% SD compared with the control group. The acceptable error rate is 5%.

ResultsTH9 cells have therapeutic advantages toward established lung metastases

TH9 cells exhibited more substantial antitumor effects than TH1 and TH17 cells in some solid tumors.16 21 22 28 However, the antitumor effects of TH9 cells are primarily evaluated in melanoma and lung adenocarcinoma.29 30 To validate whether TH9 cells are also effective in tumors with poor immune infiltration, such as OS and TNBC, we assessed the antitumor effects of tumor-specific TH1, TH9 and TH17 cells in different mouse models of OS and TNBC. OVA-specific TH1, TH9 and TH17 cells were generated by priming DO11.10 mice derived naïve CD4+ CD62L+ T cells in the corresponding polarized conditions for 4 days (online supplemental figure S1A). Then, we transferred tumor-specific TH1, TH9 or TH17 cells into Balb/c mice bearing K7M2-OVA cells or 4T1-OVA cells, respectively. Although TH1, TH9 and TH17 cell transfusion all resulted in partial tumor inhibition in the subcutaneous models, TH9 cells failed to induce stronger tumor eradication as reported in melanoma16 22 (online supplemental figure S1B,C). TH1, TH9 and TH17 cells consistently extended the survival time of tumor-bearing mice, but TH9 cells showed no priority relative to TH1 and TH17 cells (online supplemental figure S1D). Similarly, we found that the tumor-inhibiting ability of TH9 cells was not better than TH1 and TH17 cells in the primary models of OS and TNBC (online supplemental figure S1E–G). We also tested the control of tumor growth by different TH cell subsets in the setting of established lung metastases (figure 1A). However, we observed that TH9 cells exhibited stronger antitumor effects against pulmonary OS and TNBC than TH1 and TH17 cells. (Figure 1B,C). Besides, TH9 cells further prolonged the survival period of lung tumor-bearing mice (figure 1D). The H&E staining also proved lesser pulmonary tumor burden after TH9 cells treatment as compared with TH1 and TH17 cells (figure 1E,F). Therefore, our results indicate that TH9 cells have a superior therapeutic effect on lung metastatic lesions of OS and TNBC than TH1 and TH17 cells.

Figure 1

TH9 cells exhibited superior antitumor effects in established lung metastasis as compared with TH1 and TH17 cells. (A) Balb/c mice were inoculated with 1×105 4T1-OVA-luci or K7M2-OVA-luci in 100 µl PBS intravenously on day 0. Tumor-bearing mice were transferred with 3×106 OVA-specific TH1, TH9, or TH17 cells in 100 µl PBS intravenously on day 5, day 12 and day 19. (B and C) In vivo bioluminescence images (B) and quantification (C) of the tumor burden in lungs on day 7, day 14 and day 21. (D) Survival of K7M2-OVA or 4T1-OVA tumor-bearing mice with lung metastasis treated with PBS, tumor-specific TH1, TH9, and TH17 cells. (E) H&E staining of lungs bearing with K7M2-OVA or 4T1-OVA tumor on day 14. Scale bar, 2.5 mm. (F) Tumor area and lung weight correspond to (E). Data were analyzed by one-way ANOVA test or unpaired t-test. Representative results from three independent experiments are shown (mean±SEM); n=5 in (B, C and F), n=8 in (D), n=3 in (E and F). ns denote no significant difference,*indicates p<0.05, **indicates p<0.01, ***indicates p<0.001, ****indicates p<0.0001. ANOVA, analysis of variance; OVA, ovalbumin; PBS, phosphate-buffered saline; TH, T helper.

TH9 cells possess unique lung tissue affinity

We initially hypothesized that this difference could be attributed to the distinct direct cytotoxic abilities of different TH cell types against tumor cells, as existing literature has reported on the direct killing capability of TH9 cells.21 22 Thus, we first test the tumor-specific killing ability of TH1, TH9 and TH17 cells toward K7M2 and 4T1 cells in vitro. According to our results, TH1, TH9 and TH17 cells seemed to promote rather than inhibit the growth of 4T1 cells. Besides, TH1 and TH9 cells showed slight tumor-specific killing ability toward K7M2 cells with no significant difference, and TH17 cells still promoted the growth of K7M2 cells (online supplemental figure S2A,B). These results indicated that TH9 cells do not possess superior direct cytotoxicity toward tumor cells compared with TH1 and TH17 cells. Next, we speculated more TH9 cells were accumulated in lung tissue than TH1 and TH17 cells since the distribution of T cells contributes to the antitumor effects. To prove our speculation, we labeled exogenous TH cell subsets with DIR to trace the accumulation of TH cells in lungs after adoptively transfusing them into mice with established lung metastases (figure 2A). As expected, higher amounts of TH9 cells were accumulated in the lungs than TH1 and TH17 cells (figure 2B,C). To evaluate whether tumors affect the lung tropism of TH9 cells, we explored the distribution of transferred TH9 cells in multiple organs in tumor-free mice, including the lungs, liver, spleen, lymph nodes and bone marrow. We generated CD45.1-TH9 cells in vitro and then intravenously transfused them into CD45.2 mice (figure 2D). We detected the distribution of CD45.1+ cells in different organs through flow cytometry after 48 hours and found that most CD45.1-TH9 cells were aggregated in lung tissues (figure 2E,F). Next, we compared the accumulation of TH1, TH9, and TH17 cells in the lungs of tumor-free mice and found that TH9 cells presented much higher lung tissue tropism than TH1 and TH17 cells (figure 2G,H). Consistent with these results, lung tissue lysates had a more vital ability to chemoattract TH9 cells in vitro than liver and spleen lysates (figure 2I). Furthermore, lung tissue lysates specifically promoted the chemoattraction of TH9 cells but not TH1 and TH17 cells (figure 2J). Altogether, these results suggested that TH9 cells possess natural lung tropism.

Figure 2

TH9 cells show higher lung affinity as compared with TH1 and TH17 cells. (A) Balb/c mice were inoculated with 1×105 4T1-OVA-luci or K7M2-OVA-luci in 100 µl PBS intravenously on day 0. Tumor-bearing mice were transferred with 3×106 DIR labeled OVA-specific TH1, TH9, or TH17 cells in 100 µl PBS intravenously on day 7. 2 days later, in vivo bioluminescence spectrum was used to detect the accumulation of transferred TH cells in lungs. (B and C) In vivo bioluminescence images (B) and quantification (C) of TH cell subsets accumulated in lungs 48 hours after transfusion of DIR labeled tumor-specific TH1, TH9, and TH17 cells. (D) CD45.1+ naïve CD4+ T cells were derived from CD45.1+ mice and differentiated into CD45.1+-TH cells in vitro for 4 days. CD45.1+-TH (1×107) were transferred intravenously into CD45.2+ C57 mice. 48 hours later, the mice were sacrificed and the CD45.1+-TH cells in different organs were detected by flow cytometry. (E and F) The representative density plot (E) and quantification (F) of CD45.1-TH9 cells in lung, liver, spleen, lymphoma, and bone marrow 48 hours after intravenous injection of CD45.1 derived TH9 cells into healthy CD45.2 mice. (G and H) The representative density plot (G) quantification (H) of CD45.1-TH1, TH9 and TH17 cells in the lung 48 hours after intravenous injection of CD45.1 derived-TH cell subsets into healthy CD45.2 mice. (I) Migration index of TH9 cells toward liver, lung, and spleen lysate (100 µg/mL) detected by transwell. (J) Migration index of TH1, TH9, and TH17 cells toward lung lysate (100 µg/mL) detected by transwell. Data were analyzed by one-way ANOVA test or unpaired t-test. Representative results from three independent experiments are shown (mean±SEM); n=3 in all groups. ns denote no significant difference,*indicates p<0.05, **indicates p<0.01, ***indicates p<0.001, ****indicates p<0.0001. ANOVA, analysis of variance; DiR, 1,1-dioctadecyl-3,3,3,3 tetramethylindotricarbocyaine iodide; OVA, ovalbumin; PBS, phosphate-buffered saline; TH, T helper.

Lung tropism of TH9 cells depends on the CXCR4-CXCL12 chemokine axis

Since TH9 cells were significantly attracted by lung lysates (figure 2I,J), we supposed there are probably high levels of chemokine(s) responsible for TH9 cell chemotaxis in the lungs. We analyzed transcriptomic data from normal mouse lung tissues sourced from the GSE179554 data set to evaluate the basal expression levels of common chemokines. Our analysis revealed that Cxcl15, Ccl6, Cxcl12, Cxcl14 and Cxcl16 exhibited the highest basal expression (figure 3A). High levels of CXCL15, CXCL12, CXCL14 and CXCL16 proteins in the lungs were further confirmed (online supplemental figure S3A). Next, the global transcriptional profile of TH0 and TH9 cells were analyzed in triplicate to determine the upregulated chemokine receptors in TH9 cells and revealed that the genes encoding chemokine receptors, including CC motif chemokine receptor 4 (Ccr4), Ccr6, Ccr7, Ccr8, Cxcr4, Cxcr5 and Cxcr6, were upregulated in TH9 cells (figure 3B). Then, we analyzed the messenger RNA (mRNA) levels of chemokine receptors in TH9 cells as compared with TH0 cells and revealed significant upregulation of Ccr8 and Ccr6, followed by Cxcr4, Cxcr2 and Cxcr1. Whereas the expression levels of Cxcr5 and Cxcr6 in TH9 cells did not exhibit significant upregulation compared with those in TH0 cells (figure 3C). To compare the potential of different chemokine axes in the lung tropism of TH9 cells more intuitively, we standardized the expression chemokines in lung tissues and the chemokine receptors on TH9 cells and multiplied chemokines with paired chemokine receptors. Among all these paired chemokine-chemokine receptor axes, CXCL15 was ruled out since the corresponding receptor of CXCL15 was still elusive. In this way, we noticed that the CXCL12-CXCR4 axis ranked first, suggesting the potential to recruit TH9 cells into the lungs (figure 3D). In addition, we confirmed the upregulated CXCL12 in lung tumor-bearing mice relative to tumor-free mice (online supplemental figure S3B).

Figure 3

The CXCR4-CXCL12 axis mediates the lung tropism of TH9 cells. (A) The expression of chemokines in normal lung tissue was analyzed from the GSE179554 data set. (B) Heat map of chemokine receptor genes in TH0 and TH9 cells. (C) Real-time PCR analysis of the expression of the chemokine receptors in TH0 and TH9 cells. (D) Bubble plot depicting ranked chemokine–chemokine receptors by the production of chemokines in lungs and chemokine receptors in TH9 cells. (E and F) The representative density plot (E) and quantification (F) of CD45.1-TH9 cells in the lungs 48 hours after intravenous injection of CD45.1-TH9 cells into healthy CD45.2 mice. Healthy CD45.2 mice were received intraperitoneal (i.p.) injection of 1 mg kg−1ML339, C-021, R243 and AMD3100, 24 hours before T cell transfer. (G) Migration index of TH1, TH9 and TH17 cells with the presence or absence of lung lysate (100 µg/mL). The TH cell subsets were pretreated with AMD3100 (5 µg/mL) or not. (H) Migration index of TH1, TH9 and TH17 cells toward recombinant CXCL12 (10 ng/mL). (I) Immunofluorescence staining of CXCR4 in TH1, TH9 and TH17 cells. Scale bar, 15 µm. (J) Statistical analysis of mean fluorescence intensity of CXCR4 in TH1, TH9 and TH17 cells. (K) The intensity of CXCR4 expression detected in TH0, TH1, TH9 and TH17 cells by flow cytometry. (L) Western blotting analysis of CXCR4 protein levels in TH0, TH1, TH9 and TH17 cells. (M and N) The representative density plot (M) and statistical analysis (N) of CD45.2-TH9 cells in the lung 48 hours after intravenous injection of CD45.2 WT derived TH9 cells and CD45.2-Cd4cre Cxcr4flox/flox derived TH9 cells into healthy CD45.1 mice. (O) H&E staining of lungs bearing with K7M2-OVA tumor on day 14. The mice were intravenously injected with TH9 cells or AMD3100 (5 µg/mL) pretreated TH9 cells on day 5 and day 12. Scale bar, 2.5 mm. (P) Statistical analysis of tumor weight and tumor area corresponds to (O). Data were analyzed by one-way ANOVA test or unpaired t-test. Representative results from three independent experiments are shown (mean±SEM); n=3 in (C, E–H and M–P), n=4 or 6 in (I and J). ns denote no significant difference,*indicates p<0.05, **indicates p<0.01, ***indicates p<0.001, ****indicates p<0.0001. ANOVA, analysis of variance; DAPI, 4'6-diamidino-2-phenylindole; OVA, ovalbumin; PBS, phosphate-buffered saline; TH, T helper; WT, wild-type.

Subsequently, we employed inhibitors ML339, C-021, R243, and AMD3100 to obstruct the chemokine–chemokine receptor axis in the lungs, targeting the CXCL16-CXCR6, CCL17-CCR4, CCL1-CCR8, and CXCL12-CXCR4 axes, respectively. We found that CXCR4 antagonist AMD3100 specifically inhibited TH9 cell accumulation in the lungs, while other inhibitors had no obvious influence (figure 3E,F). Besides, AMD3100 did not impede the accumulation of TH1 and TH17 cells in the lungs (online supplemental figure S3C). Consistently, CXCR4 antagonist AMD3100 abolished the increased TH9 rather than TH1 and TH17 cells chemoattraction induced by lung tissue lysates in vitro (figure 3G). To further confirm the role of the CXCL12-CXCR4 axis in TH9 cells chemotaxis, recombinant CXCL12 protein was used to test the chemoattraction ability of TH1, TH9 and TH17 cells. TH9 cells were more susceptible to recombinant CXC12-induced chemoattraction than TH1 and TH17 cells (figure 3H). Altogether, these results suggested that the CXCL12-CXCR4 axis is central to TH9 cell recruitment.

CXCL12 is a critical factor in both physiological and pathological processes, including embryogenesis, hematopoiesis, angiogenesis, and inflammation, as it activates and induces the migration of hematopoietic progenitor and stem cells, endothelial cells, and a majority of leukocytes. CXCL12 is predominantly expressed in the lungs, liver, and bone marrow.31 32 In healthy tissues, CXCL12 is secreted by stromal cells, including fibroblasts, macrophages, and endothelial cells.33 34 Considering the responsive difference between TH cell subsets toward CXCL12, we wondered whether TH9 cells express increased CXCR4 compared with TH1 and TH17 cells. Our results indicated that the total and surface CXCR4 levels of TH9 cells were higher compared to other TH cell subsets (figure 3I–L and online supplemental figure S3D). Furthermore, we constructed Cd4CreCxcr4flox/flox C57BL/6 mice and obtained Cd4CreCxcr4flox/flox -TH9 cells (online supplemental figure S3E) . Compared with TH9 cells differentiated from WT naïve CD4+ T cells (Cxcr4flox/flox-TH9 cells), the accumulation of Cd4CreCxcr4flox/flox -TH9 cells in lung tissues was significantly inhibited (figure 3M,N). Besides, the chemoattraction of lung lysate to TH9 cells was also blunted after Cxcr4 knockout (online supplemental figure S3F). To further confirm the role of the CXCR4-CXCL12 axis in the inhibitory effect of lung metastatic OS, tumor-bearing mice received a transfer of TH9 cells pretreated by AMD3100. We found that AMD3100 greatly blunted the tumor inhibitory effect of TH9 cells (figure 3O,P). However, AMD3100 pretreatment did not affect the antitumor effect of TH1 and TH17 cells (online supplemental figure S3G,H). Thus, these results indicate that the CXCR4-CXCL12 axis attracts TH9 cells to the lungs.

ITCH regulates CXCR4 levels in TH9 cells by ubiquitinating CXCR4

Then, we investigated how CXCR4 is upregulated in TH9 cells. First, we compared the mRNA levels of Cxcr4 among TH0, TH1, TH9 and TH17 cells. We found that Cxcr4 expression was significantly upregulated in TH9 cells compared with TH0 cells, consistent with the bulk RNA sequencing of TH0 and TH9 cells mentioned in figure 3B. However, Cxcr4 mRNA levels a TH1, TH9 and TH17 cells showed no difference (figure 4A). In addition, we found that lysosome inhibitor chloroquine treatment led to comparable CXCR4 protein levels among TH1, TH9 and TH17 cells (figure 4B). However, treatment with the proteasome inhibitor MG132 (Z-Leu-Leu-Leu-al) did not eliminate the differences in CXCR4 expression among the three subsets, as TH9 cells continued to exhibit the highest CXCR4 levels (figure 4C). These results indicate that different CXCR4 protein levels among these TH cell subsets were due to the post-translational modification. Consistently, previous studies reported CXCR4 protein levels were regulated by the process of ubiquitination and lysosome trafficking.35 36 Thus, we detected the levels of ubiquitinated CXCR4 in TH0, TH1, TH9 and TH17 cells. Interestingly, TH9 cells contained the lowest levels of ubiquitinated CXCR4 proteins (figure 4D). To find out the E3 ligase responsible for CXCR4 ubiquitination degradation, we overexpressed CXCR4 in 3T3 cells (figure 4E) and performed mass spectrum analysis to identify the E3 ligases pulled down by CXCR4. We identified 10 E3 ligases in 3T3 cells with empty vector or Cxcr4 overexpression, among which ITCH was upregulated in CXCR4-overexpressing cells (figure 4F). ITCH was reported to degrade CXCR4 and prevent the formation of immune synapse required for T cell activation.35 37 Consistently, ITCH was verified to be immunoprecipitated by CXCR4 (figure 4G). To confirm the function of ITCH on CXCR4 regulation in TH9 cells, we silenced Itch and f