IL9 Polarizes Macrophages to M1 and Induces the Infiltration of Antitumor Immune Cells via MIP-1 and CXCR3 Chemokines

Tumor-associated macrophages (TAM) are involved in tumor progression, metastasis, and immunosuppression. Because TAMs are highly plastic and could alter their phenotypes to proinflammatory M1 in response to environmental stimuli, reeducating TAMs has emerged as a promising approach to overcoming the challenges of solid cancer treatment. This study investigated the effect of IL9 on macrophage M1 polarization and verified its antitumor potential to retrain TAMs and promote chemokine secretion. We demonstrated that IL9 stimulated macrophage proliferation and polarized them toward the proinflammatory M1 phenotype in an IFNγ-dependent manner. Tumor-localized IL9 also polarized TAMs toward M1 in vivo and made them release CCL3/4 and CXCL9/10 to recruit antitumor immune cells, including T and natural killer cells, into the tumor microenvironment. Furthermore, peritoneal treatment with recombinant IL9 delayed the growth of macrophage-enriched B16F10 melanoma and 4T1 breast cancer in syngeneic mice, although IL9 treatment did not reduce tumor growth in the absence of macrophage enrichment. These results demonstrate the efficacy of IL9 in macrophage polarization to trigger antitumor immunity. Significance: These findings clarified the effect of IL9 on macrophage M1 polarization and verified its antitumor potential through retraining TAMs and chemokine secretion.


Introduction
Macrophages play critical roles in the innate immune response through the phagocytosis of and cytotoxic effects on foreign substances and serve as a bridge between the innate and acquired immune responses (1). In mice, activated macrophages can be classified into two different polarization states related to different stimuli, classically activated M1 and alternatively activated M2 subtypes (2). M1 macrophages, induced by Th1 cytokines such as IFNγ, IL12, and IL18, or the stimulation of Toll-like receptors, are proinflammatory and play growth of several types of pancreatic, colitis-associated, colon, or breast cancers by negatively regulating the antitumor function of T cells and promoting the immunosuppressive function of regulatory T cells (8). In contrast, IL9 can inhibit tumor progression directly by inhibiting the proliferation and migration of cancer cells or indirectly by activating innate and adaptive immunity to trigger antitumor responses (8,10). High expression of IL9 in tumor tissues was related to prolonged survival of patients with colon carcinoma (11). Furthermore, we found that the number of tumor nodules in lungs intravenously injected with IL9-expressing B16F10 was 5-fold less than that of control groups and the percentages of T cells, natural killer (NK) cells, and M1 macrophages considerably increased in the lungs of the mice injected with IL9-expressing cells (12).
Therefore, we set out to conduct further experiments on the effect of IL9 on macrophage function. We found that IL9R was widely expressed on macrophages, and IL9-IL9R signaling stimulated the plasticity of macrophages toward the M1 phenotype and induced their proliferation. IL9 treatment also reeducated M2 macrophages and TAMs toward the M1 phenotype in vitro and in vivo. In macrophage-enriched cancers, IL9 polarizes TAMs to the M1 phenotype and triggers antitumor immunity by inducing chemotactic recruitment and the in situ accumulation of antitumor immune cells, including dendritic cells (DC), macrophages, T cells, and NK cells in CXCL9-and CCL3dependent manners. Thus, we present for the first time that IL9 cooperated with macrophages to induce antitumor immunity through M1 polarization and chemokine induction.

Recombinant IL9 Expression and Purification
The recombinant mouse IL9 (rIL9) gene with a C-terminal hexahistidine and ALFA tag was cloned between the BamHI and NotI site in the pVL1393 baculovirus transfer vector. The recombinant baculovirus was generated by cotransfection of the BestBac 2.0v linearized baculovirus genome (Expression Systems) into Sf9 insect cells (Thermo Fisher Scientific). The rIL9 protein was produced by infecting High Five insect cells (Thermo Fisher Scientific) with the recombinant baculovirus. The secreted rIL9 protein was purified by Ni-Sepharose chromatography (GE Healthcare). After cleavage by thrombin to remove the hexahistidine and ALFA tag, the mIL9 protein was further purified by SP-Sepharose (GE Healthcare) cation-exchange chromatography and Superdex-200 gel filtration chromatography (GE Healthcare) equilibrated with a buffer containing 20 mmol/L Tris pH 8.0, 200 mmol/L NaCl, and 0.1 mmol/L phenylmethylsulfonyl fluoride. The protein sample buffer was changed into PBS (Gibco) using PD MiniTrap G-25 desalting columns (Cytiva). Endotoxin was removed from rIL9 using Pierce High-Capacity Endotoxin Removal Resin (Thermo Fisher Scientific), and the protein samples were sterilized by 0.2 μmol/L sterile membrane filtration. The protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific).

Tumor Cell Lines and Mice
Female 6-to 8-week-old C57BL/6JBomTac and BALB/C mice were obtained from Daehan Biolink (Eumseong, Korea). All animal procedures were approved and guided by the Institutional Animal Care and Use Committee of Pohang University of Science and Technology (POSTECH-2020-0072). The murine melanoma B16F10 (80008) and RAW264.7 (40071), J774A.1 (40067), and P388D1 (10046) macrophage cell lines were purchased from Korean Cell Line Bank. The Mycoplasma-free cell lines were authenticated by the provider using DNA fingerprinting short tandem repeat profiling. Authenticated mammary carcinoma 4T1 cell lines were kindly provided by Prof. Young Sang Kim (Chungnam National University, Daejeon, Republic of South Korea). All cell lines were cultured in DMEM (Gibco) supplemented with 10% heat-inactivated FBS (Gibco), 2 mmol/L l-glutamine, and 100 U/mL penicillin plus 100 μg/mL streptomycin (1% Pen/Strep, Sigma-Aldrich) in a humidified 5% CO 2 chamber at 37°C. Each cell line was expanded to produce frozen aliquots. After resuscitation, they were used for no more than 15 passages and no longer than 2 months.

RNA Extraction and qPCR
RNA was extracted from 1 × 10 6 BMDMs, RAW264.7 cells, and TAMs using the AccuPrep Universal RNA Extraction Kit according to the manufacturer's instructions (K-3140, Bioneer Corporation). The RNA concentration was determined by Ultrospec 8000 (GE Healthcare). Total RNA (1 μg) was reverse transcribed using AccuPower RT premix (Bioneer Corporation). For qPCR, RNA expression was analyzed on the QuantStudio 1 Real-Time PCR System (Thermo Fisher Scientific) using SFCgreen included in the 2X Realtime PCR Master Mix (DQ362-40h, BioFACT) at a concentration of 0.4 μL cDNA/sample. The primers for each of the genes are listed in Supplementary   Table S1. Quantitative gene expression data were normalized to the expression levels of β-actin.

FACS Analysis
The tissues were minced into small pieces and digested in DMEM containing 80 U/mL collagenase IV (17104-019, Gibco) and 80 U/mL DNase I (D5025, Sigma-Aldrich) at 37°C for 30 minutes. The cells were subsequently filtered through a 40-μm cell strainer (93040, SPL). After centrifuging, cells were blocked with 5 μg/mL of purified rat anti-mouse CD16/CD32 (553142, BD Biosciences) for 10 minutes at 4°C. For surface staining, the cells were stained with the appropriate antigen-specific antibodies conjugated to fluorophores or with an isotype control antibody for 1 hour at 4°C in the dark. All antibodies were diluted in staining buffer (1X PBS containing 0.02% sodium azide and 2% FBS). After washing off the unbound antibody with the staining buffer, the cells were analyzed using a flow cytometer (BD FACSCanto, BD Biosciences). To measure intracellular expression, the cells were cultured with a mixture of Brefeldin A (B5936, Merck, 1 μg/μL) and Monensin (M5273, Sigma-Aldrich, 3 μmol/L) for 4 hours. Then, the cells were fixed and permeabilized with BD Cytofix/Cytoperm Fixation and Permeabilization Solution (554722, BD Biosciences) according to the manufacturer's instructions. The permeabilized cells were incubated with an antibody diluted in staining buffer including 0.5% saponin for 1 hour at 4°C in the dark. For carboxyfluorescein succinimidyl ester (CFSE) staining, the cells were incubated with 2.5 μmol/L CFSE (423801, BioLegend) for 20 minutes. To distinguish dead cells, 10 μg/mL of propidium iodide (PI) was added to each sample just before the analysis.
The cells were treated with either clodronate liposomes, control liposomes, or PBS at the indicated concentrations for 24 hours. Then, the cell viability was assessed by the MTT assay.

Multiplex Bead Array Assay
Culture supernatants of IL9-treated BMDMs were used. To prepare cell lysates, the treated cells were incubated with 3 μmol/L monensin for the last 6 hours and harvested. The proinflammatory chemokine expression was determined using the LEGENDplex Mouse Proinflammatory Chemokine Panel (740007, BioLegend), following the manufacturer's instructions.

Chemotaxis Assay
Chemotaxis assay was tested in 24-well Transwell chambers (5.0 μmol/L, polycarbonate membrane, Corning). BMDMs were treated with 100 ng/mL rIL9, rIL9 + 2.5 μg/mL anti-IL9R, or PBS for 48 hours. Then, the culture supernatants from the treated BMDMs were collected and placed in the lower chambers. Splenocytes were isolated from C57BL/6J mice. After red blood cell removal, 3 × 10 6 cells were resuspended in 150 μL of DMEM and plated in the upper charmers. After assembling the chambers, the cells were incubated at 37°C for 2.5 hours. Then, the medium was aspirated from the upper chamber, and the nonmigrating cells in the upper chamber were scraped off with a cotton swab. The chambers were washed with PBS and then incubated with trypsin-ethylenediaminetetraacetic acid (EDTA) (Gibco) for an additional 5 minutes to collect all of the migrated cells. The cells were transferred to 5-mL U-bottom FACS tubes for further staining, and the number of migrated cell subsets was determined by FACS analysis. To analyze the chemotactic role of CCL3 secreted by IL9-treated macrophages, the culture supernatant from IL9treated BMDMs was incubated with either 1.5 μg/mL of neutralizing anti-CCL3 or control IgG antibody for 30 minutes, and then loaded into the lower chamber for the chemotaxis assay. To examine the effect of CXCR3, splenocytes were pretreated with 3.8 μg/mL of a neutralizing anti-CXCR3 antibody or the isotype control for 30 minutes and then loaded into the upper chamber.

Cytotoxicity Assay
Macrophage cytotoxic activity was measured using the MTT assay as described previously (14). To examine the cytotoxicity of macrophages against B16F10 melanoma, 1.5 × 10 6 BMDMs were treated with either 100 ng/mL rIL9 or PBS for 48 hours and the IL9-containing medium was washed off and replaced with fresh culture medium. Then, the treated BMDMs were coincubated with 2 × 10 3 target B16F10 cells at ratios of 2:1, 10:1, and 50:1 for 48 hours at 37°C. Simultaneously, the respective B16F10 target and effector BMDMs were also cultured. The activity was determined by measuring residual target cell viability using the MTT assay, and the percentages were calculated as follows: %Cytotoxicity = 100 − optical density of effectors and targets) − (optical density of effectors optical density of targets × 100

Phagocytosis Assay
The macrophage phagocytosis of apoptotic cells has been described previously (15,16). Apoptotic B16F10 tumor cells were derived by treatment with 10 μg/mL mitomycin C (CAS 50-07-7, Sigma-Aldrich) for 24 hours at 37°C. The efficacy of apoptotic induction was confirmed by annexing V (640905, BioLegend) and PI staining. After incubation, the tumor cells were washed twice with PBS and labeled with CFSE for 15 minutes. The CFSE-labeled cells were resuspended at 4 × 10 6 cells/mL in complete DMEM. Then, the macrophages and apoptotic B16F10 cells were coincubated at a 1:2 ratio at 37°C for 1 to 4 hours. Flow cytometry was performed to obtain the percentage of CFSE + F4/80 + cells.

Murine Tumor Models
For TAM studies, the coinoculation of BMDM and tumor cells has been described previously (17)(18)(19). With this system, the ratio of tumor cells to BMDMs varied from 1:1, 2:1, 4:1, to 5:1. We tested the ratios in the system using B16F10 and 4T1 cells, then we used a 2:1 ratio in the experiments. Briefly, 50 μL of a mixture of 2 × 10 5 syngeneic BMDMs and 4 × 10 5 B16F10 clone cells, either B16F10-IL9 or B16F10-Mock, were mixed with 50 μL of Matrigel (356230, BD Biosciences). Then, 100 μL of the final mixture was subcutaneously implanted into the flanks of C57BL/6J mice. The use of clodronate liposomes for the in vivo depletion of macrophages was described previously (13,20,21). In brief, macrophage depletion was achieved by the intraperitoneal injection of 200 μL of clodronate liposomes (Liposoma) every 3 days. As a control, a group of mice was injected intraperitoneally with control liposomes in PBS (Liposoma). For intraperitoneal treatment with the rIL9 model, BMDMs of syngeneic mice were mixed with B16F10 or 4T1 tumor cells at a 1:2 ratio in Matrigel, and subcutaneously implanted into the flanks of C57BL/6J or BALB/C mice, respectively (n = 3 for the dose-determining study, and n = 7 for the others). Tumor cells only were injected as controls. The tumor-bearing mice were treated intraperitoneally with rIL9 at a dose of 0.5 to 2 μg/mouse or PBS every 3 days. Tumor size was measured with a caliper, and tumor volume was calculated according to the following formula: 0.52 × S 2 × L, where L is the length and S is the width of the tumor. Bodyweight was also monitored daily.

Hematoxylin and Eosin Histologic Staining
Tumor-bearing mice were sampled on day 12 after tumor implantation. The resected tumors were fixed in 4% paraformaldehyde for 11 hours. Tissue samples were mounted in paraffin blocks, and 4-μm sections were prepared. Sections were subjected to hematoxylin and eosin (H&E) staining. The morphologic changes in H&E-stained tissue, as well as tumor-infiltrating immune cells, were analyzed under an uplight microscope (OLYMPUS-BX43, OLYMPUS) at magnifications of 50× and 200×.

Immunoblotting
RAW 264.7 cells were treated with 100 ng/mL rIL9 or PBS. At the indicated timepoints, total protein was extracted using 1X RIPA lysis buffer (20-188, Merck) containing a 1X solution of protease inhibitor cocktail (11836170001, Sigma-Aldrich) for 15 minutes. The protein concentration was determined by the BCA assay (23227, Thermo Fisher Scientific). Samples (20 μg/well) were resolved by 10% SDS-PAGE. After transferring the proteins to polyvinylidene difluoride membranes using an electroblotter (AE-6677 HorizeBLOT, ATTO Corporation), the membranes were blocked with ProNA Phospho-Block Solution (TLP-115.1P, Translab) for 1 hour. Next, the membranes were incubated with primary antibodies overnight at 4°C and then incubated with secondary antibodies for 2 hours at room temperature. The blotted protein bands were detected with Immobilon Western Chemiluminescent HRP Substrate (WBKLS0500, Millipore) in the dark. The images were taken using an Ez Capture II machine (ATTO Corporation) and analyzed using ImageJ software.

Data Availability
The data generated in this study are available upon request from the corresponding authors.
BMDMs were prepared as shown in Supplementary Fig. S1. IL9R was clearly expressed on the cell surface of all of these macrophages (Fig. 1A). Among them, BMDM and J774A.1 cells expressed higher levels of surface IL9R. Because IL9 is known as a growth factor of T cells and hematopoietic progenitor cells, the effect of IL9 on the proliferation of macrophages was tested. As shown in Fig. 1B, in vitro stimulation with rIL9 stimulated the proliferation of J774A.1 cells and BMDMs, and this growth-promoting effect was attenuated by the addition of anti-IL9 antibodies. IL9 mediates the growth-promoting effect in T cells by activating STAT1, STAT3, and STAT5 (22). Hence, we analyzed the phosphorylation of STAT transcription factors to verify the downstream signaling of IL9-IL9R in macrophages. Treatment with rIL9 induced STAT1 and STAT3 phosphorylation within 10 to 30 minutes in BMDMs as observed by FACS analysis of intracellular staining (Fig. 1C). However, no significant increase in p-STAT5

AACRJournals.org
Cancer Res Commun; 3(1) January 2023 (Continued) (IL1β, TNFα, iNOS, and IL12p35) and M2 markers (IL10, VEGF, Arg1, and MRC-1) in BMDMs treated with either rIL9 or rIL9+anti-IL9 for 48 hours. E, FACS analysis of M1 markers (CD80, CD86, TNFα, IFNγ), the activation markers (MHC class I and II) and M2 markers (CD206 and Arg1) on BMDMs treated with either rIL9 or PBS for 48 hours using specific and isotype control antibodies. F, The relative MFI of each marker in E was calculated and the relative fold change compared with PBS was graphed. G, BMDMs was pretreated with anti-IL9R (5 μg/mL) or isotype control antibody for 30 minutes. Then, the pretreated BMDMs were treated with either IL9 or PBS for 48 hours. The expression level of M1 markers (CD80, TNFα, and IFNγ) were analyzed by FACS. H, BMDMs were treated with either rIL9 or rIL9+anti-IL9 for 48 hours. RAW 264.7 or J774A.1 macrophages were treated for 24 hours. Then, qPCR analysis of IFNγ mRNA in the treated macrophages was performed. I, The macrophages were treated with rIL9 or PBS for the indicated days, and IFNγ secretion into the culture medium by the treated macrophages was determined using an IFNγ-specific ELISA. J, qPCR analysis of IL1β and TNFα mRNA in BMDMs and RAW 264.7 macrophages treated with rIL9 or the mixture of rIL9+2.5 μg/mL anti-IFNγ. K, Proliferation assay (MTT assay) of BMDMs and RAW 264.7 or J774A.1 macrophages treated with rIL9, rIL9+anti-IL9, or rIL9+anti-IFNγ was performed 48 hours after treatment. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001. and p-STAT6 was observed in the macrophages. The rIL9 treatment did not significantly change total STAT1 (t-STAT1) and total STAT3 (t-STAT3) compared to PBS treatment. The phosphorylation of STAT1 and STAT3 was further confirmed in IL9-treated RAW264.7 cells using immunoblotting (Supplementary Fig. S2). Taken together, these results showed that IL9 treatment induced the phosphorylation of STAT1 and STAT3 in macrophages.
Macrophages can be activated to M1 or M2 subsets in response to signals in their local environment, and this plasticity is a critical property related to their functions (23). Therefore, we investigated the effect of IL9 stimulation on macrophage polarization. When nonactivated (M0) BMDMs were treated with rIL9, M1 markers, including IL1β, TNFα, Inducible nitric oxide synthase (iNOS), and IL12p35, were significantly induced at the mRNA level, but M2 markers, including IL10, VEGF, Arg1 (arginase 1), and MRC-1 (the CD206coding gene), were not changed (Fig. 1D). The addition of anti-IL9 neutralizing antibodies attenuated the increase of expression of M1 markers but treatment with the isotype control antibody did not ( Fig. 1D; Supplementary Fig. S3A).
The M1 and M2 markers were also examined by FACS analysis. Consistent with the mRNA expression data, the representative cell surface markers of M1 macrophages (CD80 and CD86) were strongly induced by IL9 treatment ( Fig. 1E and F). Other M1 markers (TNFα and IFNγ) and macrophage activation markers (MHC class I and MHC class II) were also induced by rIL9. However, there was no change in the expression of CD206 and Arg1. The M1 polarization effect of IL9 was prevented in BMDMs pretreated with anti-IL9R antibody (Fig. 1G). These results clearly showed that IL9 could induce M1 macrophage polarization. These effects were also found in RAW264.7 cells, showing that IL1β, TNFα, iNOS, and IL12p35 were significantly induced at the mRNA level and that the expression of IL10, VEGF, Arg1, and CCL17, was 2 to 4.5 times lower than in PBS-treated cells ( Supplementary Fig. S3B). Similar to BMDM, the addition of anti-IL9 neutralizing antibody, but not isotype control antibody, abolished the effect ( Supplementary Fig. S3C).
Because IFNγ is responsible for one crucial axis of M1 polarization in both human and murine macrophages (24), we assessed whether IL9-treated macrophages could be a source of IFNγ. Consistent with the FACS analysis results in Fig. 1E, IFNγ mRNA expression was also enhanced 1.5-fold in rIL9-treated macrophages and was attenuated by adding an anti-IL9 antibody (Fig. 1H). To further verify IFNγ secretion from IL9-treated macrophages, the level of IFNγ was analyzed in the culture supernatants from BMDMs, RAW264.7, and J774A.1 cells, and a detectable amount was found upon rIL9 treatment. The amount increased in a time-dependent manner to reach about 10 pg/mL at 72 hours after rIL9 treatment (Fig. 1I). We next analyzed whether IL9-induced M1 polarization was IFNγ dependent. The addition of anti-IFNγ antibody blocked M1 polarization induced by rIL9 treatment as much as the addition of anti-IL9 (Fig. 1J). In contrast, the neutralization of IFNγ did not affect macrophage proliferation stimulated by rIL9 (Fig. 1K). From these results, we conclude that IL9 induced M1 polarization in an IFNγ-dependent manner. However, the efficacy of IL9 on macrophage proliferation appeared to be independent of IFNγ.

IL9 Repolarizes M2-like Macrophages Toward the M1-like Phenotype
To investigate whether IL9 treatment could repolarize macrophages from the M2-to the M1-like phenotype, rIL4-polarized BMDMs were treated with rIL9 as described in Fig. 2A. Treatment of rIL4-pretreated BMDMs with rIL9 induced CD80 and CD86 expression by 1.4-fold and reduced the expression of CD206, an M2 marker, by 1.3-fold compared with rIL4 treatment alone, implying that treatment with IL9 might be efficient in reverting macrophages from the M2 to the M1 phenotype ( Fig 2B). We further analyzed the gene expression of other M1 and M2 markers ( Fig. 2C and D). rIL9 treatment of IL4-pretreated BMDMs significantly provoked the gene expression of M1-associated genes (Fig. 2C) and dramatically lowered M2-related genes (Fig. 2D) compared with IL4 treatment only. To address the role of IL9 signaling on the polarization of TAMs in vitro, we subcutaneously implanted B16F10 melanoma into C57BL/6J mice. Fifteen days after tumor implantation, TAMs from the tumor masses were sorted (Fig. 2E). The high expression of IL9R in sorted TAMs was shown by FACS analysis (Fig. 2F). When TAMs were treated with rIL9, the percentage of CD80 + F4/80 + and CD86 + F4/80 + M1 cells was significantly enhanced, but CD206 + F4/80 + and Arg1 + F4/80 + M2 cells were not changed, indicating that IL9 induced the M1-like polarization of TAMs (Fig. 2G). The mRNA analysis was also used to verify the M1 polarization of TAMs by IL9. IL1β, iNOS, and IL12p35 expression levels were significantly induced, but IL10, Arg1 and VEGF levels were either unchanged or reduced (Fig. 2H). These data collectively indicate that IL9 repolarized IL4-induced M2 macrophages and melanoma-derived TAMs toward the M1-like phenotype.

The Combination of IL9 and Macrophages Delays the Growth of B16F10 Melanoma in Mice
To further study the roles of IL9 in the TAM population, we adopted the coinjection system of B16F10 melanoma cells plus macrophages as described in previous studies (17)(18)(19)  not detectable (Fig. 3A). Because the B16F10 melanoma cells did not express IL9R on their surface ( Supplementary Fig. S4), autocrine IL9 expression did not affect the in vitro proliferation of IL9-B16F10 cells (Fig. 3B). The transfected B16F10 cells were implanted subcutaneously into the right flank of C57BL/6J mice (n = 9) along with BMDMs (Fig. 3C). To allow the macrophages to enter the tumor mass, BMDMs and tumor cells were mixed at a 1:2 ratio and then embedded in Matrigel to allow them to solidify immediately upon injection. B16F10-Mock-injected mice showed similar tumor volume growth both in the presence and absence of BMDMs. No statistical difference in the in vivo growth between B16F10-Mock and B16F10-IL9 cells was observed. However, mice implanted with B16F10-IL9+BMDMs showed dramatically reduced tumor growth compared with mice injected with B16F10-Mock+BMDMs or with B16F10-IL9 alone (Fig. 3D). On day 15 after tumor inoculation, the mice were euthanized, and tumor masses were collected (Fig. 3E). The tumor weight in the mice in the B16F10-IL9+BMDM group was 2.8-fold lower than that of the mice in the B16F10-Mock+BMDM group. The survival of the mice in the B16F10-IL9+BMDM group was extended for 10 days compared with that of the B16F10-Mock or B16F10-Mock+BMDM groups (Fig. 3F). No significant change in the bodyweight of the tumor-bearing mice was observed between the groups for 13 days after tumor implantation ( Supplementary Fig. S5). The results suggest that the enrichment of TAMs and IL9 in the TME evoked effective antitumor activity against B16F10 melanoma. To further confirm the TAM contribution, we tried to deplete TAMs using clodronate liposomes (13,20). First, the selective cytotoxicity of clodronate liposomes on BMDMs was tested in vitro ( Supplementary Fig. S6). Treatment with clodronate liposomes selectively killed BMDMs, but not B16F10 cells, in a concentration-dependent manner (IC 50 = 0.415 mg/mL). As illustrated in Fig. 3G, mice (n = 7) were intraperitoneally treated with clodronate liposomes one day before the transfected B16F10 cells were implanted. The treatment was repeated every 3 days until day 12 after tumor implantation. The efficacy of clodronate liposome treatment was validated on day 12 after tumor implantation. Intraperitoneal injection with clodronate liposomes significantly depleted TAMs in vivo (Supplementary Fig. S6B). In B16F10-IL9+BMDM mice treated with clodronate liposomes, tumors grew in 4 of 7 mice and reached a tumor volume similar to that of the B16F10-Mock and B16F10-Mock+BMDM groups, whereas the control liposome treatment failed to prevent the reduction in tumor volume (Fig. 3H). These data imply that macrophage-scavenging diminished the antitumor responses of IL9 in mice injected with B16F10-IL9+BMDMs. Collectively, our data suggest that intratumoral IL9 induced antitumor responses in a macrophage-dependent manner.

The Combination of IL9 and Macrophages Induces the Accumulation of Immune Cells in the Tumor
To investigate the mechanism underlying the antitumor effect of IL9 on the macrophage-enriched tumor model, we collected tumor masses after euthanizing the tumor-bearing mice (n = 4) on day 15 after tumor implantation (Fig. 3C). F4/80 + TAMs from the tumor masses were sorted for further analysis. As shown in Fig. 4A, TAMs from B16F10-IL9+BMDM mice had a higher expression of M1 markers and lower expression of M2 markers, except for MRC-1, compared with mice in the B16F10-Mock+BMDM group. TAMs from growing tumors were also characterized by surface or intracellular staining with anti-CD80, anti-MHC class II, anti-CD206, and anti-Arg1 antibodies (Fig. 4B).
The FACS gating used for this analysis is shown in Supplementary Fig. S7. Consistent with the gene expression data, the surface expression of CD80 and MHC class II on TAMs from the B16F10-IL9+BMDM group was 19-fold and 2.6-fold higher, respectively, than in the B16F10-Mock+BMDM group. No statistical difference in the expression of CD206 and Arg1 was found between these groups. Of note, the absence of CD80 + macrophages in the tumors was related to the immune-suppressive response of TAMs and short overall survival (25), and the repolarization of TAMs to M1 could effectively provoke an antitumor response (2,26). Hence, TAM subsets were further validated by comparing the percentage of TAMs and CD80 + TAMs among the groups (Fig. 4C). The percentage of TAMs was increased 12-fold in the B16F10-IL9+BMDM group. Particularly, the percentage of M1-like TAMs (CD80 + F4/80 + ) was 27-fold higher in the B16F10-IL9+BMDM group. Moreover, the M1/M2 ratio (CD80 + TAM/CD206 + TAM) was 4.4-fold higher in TAMs isolated from the B16F10-IL9+BMDM group. This ratio suggested a tendency toward expressing M1 functional molecules in the TAMs from B16F10-IL9+BMDM-injected mice. Altogether, the data demonstrated that intratumoral supplementation with IL9 polarized TAMs to the M1-like phenotype.
Next, we investigated changes in tumor-infiltrating lymphocyte (TIL) subsets that contributed to antitumor immunity in the same TAM samples   CD3 + CD8 + ), and cytotoxic effector T cells (CD3 + CD8 + granzyme B + ) were stained and analyzed by FACS. E, Mice (n = 4) were challenged as described in Fig. 3. Mice injected with B16F10-IL9+BMDM or B16F10-Mock+BMDM were euthanized on day 11. Tumor mass sections were stained with H&E. The arrows identify immune cells infiltrating the TME. The left and right sections are magnified 50× and 200×, respectively. The number of tumor-infiltrating immune cells per high power field (HPF) is shown in the bar graph. F, Mice (n = 4) were challenged as described in Fig. 3C and euthanized on day 11 after tumor implantation, and the tumor-infiltrating macrophages were isolated using a TAM isolation kit. The chemokine expression in the cell lysate from the TAM fraction was analyzed by multiplex bead array. A heatmap of chemokine profiles (left) and the individual CCL3, CCL4, CXCL9, and CXCL10 concentrations (right) in the cell lysate of TAMs isolated from each group are shown. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

IL9-polarized Macrophages Induce the Expression of Chemokines, Resulting in the Chemotactic Recruitment of Antitumor Immune Cells
To clarify the effect of IL9 treatment on macrophage functions, we conducted macrophage functional assays in vitro. Firstly, we determined whether IL9 treatment increased the phagocytosis of BMDMs. CFSE-labeled apoptotic B16F10 cells were used as target cells for opsonin-independent phagocytosis. As shown in Fig. 5A, the percentage of CFSE + F4/80 + macrophages was similar between rIL9-treated and PBS-treated BMDMs for 1 to 4 hours, so no significant difference in phagocytosis among the groups was observed. Second, to compare the cytotoxic effect of IL9-polarized macrophages, rIL9 and PBS-treated BMDMs were used as effector cells while B16F10 tumor cells were used as the targets. The cytotoxicity of rIL9 or PBS-pretreated BMDMs was measured after 48 hours of incubation with three ratios (2:1, 10:1, and 50:1) of target cells.
As presented in Fig. 5B, there was no significant difference in cytotoxicity against B16F10 cancer cells between IL9-polarized BMDMs and PBS-treated controls.
As shown in Fig. 4, the in vivo data revealed that intratumoral IL9-polarized TAMs could play a central role in recruiting antitumor immune cells and facilitating their infiltration into the tumor. The migration and infiltration of proper immune cells into a solid tumor are crucial for successful antitumor responses (28). Therefore, we sought to identify the critical proinflammatory chemokines originating from IL9-polarized macrophages. BMDMs were treated with PBS, rIL9, or a mixture of rIL9+anti-IL9 antibody. Then, the culture supernatants and monensin-treated cell lysates were tested to evaluate the expression of 13 proinflammatory chemokines using a multiplex bead array. Consistent with the data using TAMs in Fig. 4, the expression of CCL3 and CCL4 was elevated in both culture supernatants and cell lysates of the rIL9-treated group. This expression was abolished by neutralizing IL9 (Fig. 5C and D). The levels of secreted CCL3 and CCL4 chemokines detected in the culture supernatant from rIL9treated BMDMs were about 65 pg/mL. The level of CCL3 and CCL4 in the cell lysates from rIL9-treated BMDMs was 1.6-fold (900 pg/mL) and 1.3-fold (200 pg/mL) higher, respectively, than those from the PBS-treated group. Compared with the chemokine expression profile of TAMs, no expression of CXCL9 and CXCL10 in rIL9-treated BMDMs was detected by the multiplex assay. This finding could be due to their low secretion levels. Therefore, the gene expression levels of the potent chemotactic chemokines were quantitated by qPCR (Fig. 5E). In the BMDMs, the relative expression of CCL3, CCL4, CXCL9, and CXCL10 in the rIL9-treated group was increased 2.0-, 1.5-, 2.6-, and 2.2-fold, respectively, compared with the PBS-treated group. Similar to the BMDMs, the relative expression of CCL3, CCL4, CXCL9, and CXCL10 in rIL9-treated RAW 264.7 cells was increased 2.0-, 1.7-, 2.3-, and 1.6-fold, respectively, compared with the PBS-treated group. The addition of anti-IL9 antibody abolished these stimulation effects of rIL9 on chemokine secretion by macrophages.
To evaluate the chemotaxis of immune cells that occurred in response to chemoattractants derived from IL9-polarized macrophages, we performed the transwell assay to assess the chemotactic effect of culture supernatant from rIL9-treated BMDMs on total splenocytes (Fig. 5F). The migrating cells were collected and labeled for FACS analysis of the immune cell subsets as shown in Supplementary Fig. S8. As shown in Fig. 5G, T cells, NK cells, macrophages, and DCs responded to the culture supernatant from rIL9-treated BMDMs and showed enhanced migration. The increased migration of all of the immune cells was completely inhibited when neutralizing anti-IL9R antibody was added to rIL9-treated BMDMs. CCL3 and CCL4 are the two major forms of macrophage inflammatory protein 1 (MIP-1). Because they have high structural similarity and largely redundant functions, CCL3 was chosen for further analysis. The culture supernatant from rIL9-treated BMDMs was collected and pretreated with neutralizing anti-CCL3 or control IgG antibody before the transwell assay. The inhibition of CCL3 showed a strong inhibitory effect on the migration of macrophages, DCs, and NK cells, but only weakly blocked the chemotaxis of CD4 + or CD8 + T cells (Fig. 5H). To determine whether blocking CXCL9 and CXCL10, ligands of the CXCR3 receptor, influenced the chemotaxis of immune cell subsets, we pretreated splenocytes with either anti-CXCR3 or control antibodies for chemotaxis inhibition assays. As expected, blocking CXCR3 using the anti-CXCR3 antibody significantly inhibited T-cell and NK-cell chemotaxis induced by the culture supernatant of IL9-polarized BMDMs, whereas blocking CXCR3 had a moderate effect on DC migration and almost no effect on the migration of macrophages (Fig. 5H). Notably, the combination of anti-CXCR3 and anti-CCL3 antibodies completely inhibited the chemotaxis of splenic CD4 + /CD8 + T cells, NK cells, macrophages, and DCs induced by IL9 treatment, suggesting that IL9-polarized BMDMs stimulated immune cell chemotaxis in MIP-1 and CXCR3 ligand-mediated manners.
(Continued) anti-IL9R antibody, or PBS for 48 hours, and chemokine profiles were determined by multiplex bead array. The heatmaps and CCL3 and CCL4 concentrations in the culture supernatants (C) and cell lysates obtained from the samples treated with 3 μmol/L momensine for 6 hours (D) are shown. E, qPCR analysis of the mRNA expression of chemokines in BMDMs and RAW264.7 macrophages treated with 100 ng/mL rIL9, rIL9+2.5 μg/mL anti-IL9R, or PBS for 48 and 24 hours, respectively. F-H, Schematic image illustrates the transwell assay evaluating the ability of rIL9-treated BMDMs to induce the chemotactic migration of immune cell subsets (F). BMDMs were treated with rIL9 or the mixture of rIL9+1 μg/mL anti-IL9R for 48 hours. The culture supernatants were collected and placed in the lower transwell chambers. Fresh non-cultured medium containing 100 ng/mL rIL9 was used as the negative control. Splenocytes from C57BL/6J mice were isolated and placed in the upper chambers. After 3 hours of incubation, the migrated cell subsets were harvested, and the cell numbers were determined by FACS analysis (G). BMDMs were treated with 100 ng rIL9 for 48 hours. To determine whether cell migration was affected by blocking CCL3, the culture supernatants were treated with 1.5 μg/mL anti-CCL3 or control IgG antibodies and then loaded into the lower transwell chamber. Splenocytes from C57BL/6J mice were placed in the upper chambers. To examine the role of CXCL9/10 in the migration, their receptor, CXCR3, was blocked. Splenocytes were pretreated with 3.8 μg/mL anti-CXCR3 or isotype control and then loaded into the upper chamber (H). *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.
B16F10+BMDM implantation. Tumor weights on day 15 in the rIL9-treated group implanted with B16F10+BMDMs were lower than those in the controls groups by 1.8-fold at a dosage of 1 μg/mouse (Fig. 6C). No significant difference in body weight of the mice was observed between all treated groups (Supplementary Fig. S9A). In the experiment using more mice at an rIL9 dose of 1 μg/mouse, the growth of tumors in the rIL9-treated B16F10+BMDM group was significantly inhibited (Fig. 6D), and the survival of those mice was also extended by around 5 days compared with the rIL9-treated B16F10 group (Fig. 6E). Because 4T1 breast cancer cells did not express IL9R on their surface ( Supplementary Fig. S4), the therapeutic effect of IL9 therapy on macrophageenriched 4T1 breast cancer was also evaluated (Fig. 6F-H). After subcutaneous implantation with either 4T1 breast cancer cells or a mixture of 4T1+BMDMs, BALB/C mice were treated intraperitoneally with rIL9 or PBS at 2-day intervals ( Fig. 6F). As in the melanoma case, rIL9 treatment delayed tumor growth after implantation in the 4T1+BMDM groups compared with the 4T1-only implantation group (Fig. 6G). Compared with the controls, tumor weights on day 15 in the rIL9-treated 4T1+BMDM group were lower by 2-fold (Fig. 6H). No significant difference in body weights was observed between the experimental groups ( Supplementary Fig. S9B). Collectively, these results demonstrated that recombinant IL9 treatment delayed the growth of cancers and enhanced survival when cancer cells were implanted with macrophages. Therefore, the results indicated that the anticancerous role of macrophages could be evoked when it was signaled and activated by IL9.

Discussion
Recently, strategies to reeducate TAMs have attracted increased scientific attention due to the potential benefits of combining them with immune checkpoint inhibitors (CPI; refs. 29,30). In this research, we found that IL9 had a strong potential to reeducate macrophages, including TAMs, to stimulate the antitumor immune response in an IFNγ-dependent manner. IL9-IL9R signaling polarized macrophages to the M1 phenotype and also promoted their proliferation. IL9 retrains M2 macrophages or TAMs to the M1 phenotype. We specifically showed that IL9-polarized TAMs induced antitumor immunity in mice by recruiting antitumor immune cells into the tumor mass via releasing the macrophage-derived chemoattractants, CCL3/4 and CXCL9/10. In addition, when macrophages were enriched, intraperitoneal and intratumoral treatment with rIL9 delayed the growth of B16F10 melanoma and 4T1 breast cancer cells.
Here, we observed that the surface expression of IL9R was common on macrophage cell lines and primary macrophages isolated from mice. In ad-dition, IL9 treatment stimulated proliferation and induced STAT1 and STAT3 phosphorylation in macrophages. Our observations are consistent with previous reports on the growth-promoting effect of IL9 in other hematopoietic cells such as T cells, B cells, and NK cells (7). Regarding the effect of IL9 on macrophages, M1/M2 polarization has not been extensively studied. In B16F10 lung metastasis and allergic lung inflammation models, lung macrophages responded to IL9 stimulation to promote proinflammatory immune responses (12,31). In contrast, in the multiple sclerosis brain system, IL9 acted on macrophages to enhance the production of TGFβ, consequently inducing an anti-inflammatory response (32). In this study, we showed that in vitro stimulation with IL9 polarized macrophages to the proinflammatory M1 phenotype, and IL9 treatment converted M2 plastic BMDMs to the M1 phenotype by dampening M2-associated markers in IL4-treated M2 macrophages. This effect on M1 polarization was found through the autocrine effect of IFNγ secreted from IL9-stimulated macrophages. These IL9 properties, combined with the IL9 activation of macrophage proliferation, are important for the proinflammatory role of TAMs. Because of the reversibility of the macrophage M1/M2 activation state, the newly established microenvironment could influence the TAM reprogramming process. Therefore, retraining M2 macrophages to the M1 phenotype has been proposed as the ideal therapeutic strategy in cancer treatment.
Macrophages represent approximately 5% to 7% of the total lung cells and about 80% of lung immune cells (33). For lung cancer research, although the protumor function of macrophages has been reported, accumulating evidence supports the antitumoral functions of macrophages (34)(35)(36). For example, increases in tumor islet CD68 + macrophage density and the tumor islet and/or stromal macrophage ratio were significant independent predictors of increased survival in patients with surgically resected non-small cell lung cancer (35). Our previous study found that the number of tumor nodules in the lungs of the mice intravenously injected with IL9-expressing B16F10 was significantly less than that of the control groups, and pulmonary M1 macrophages were the population most significantly increased in the lungs of those mice (12). Therefore, we prespecified that intratumoral supplementation with IL9 might alter the TME of TAM-enriched tumors and convert their immunosuppressive phenotype. For this purpose, an IL9-secreting tumor cell line that evenly and continuously distribute IL9 throughout the tumor mass, was used for syngeneic implantation. A coinjection system of BMDMs and tumor cells was applied to mimic TAMs in the TME. The mixed injection of tumor cells and BMDMs was verified in previous studies as a useful tool to investigate the antitumor effect of M1-polarized TAMs (17)(18)(19). In our experiment, we characterized BMDMs and TAMs sorted from B16F10 tumor masses. Both TAMs and BMDMs expressed IL9R on the cell surface. Treatment with IL9 drove both macrophages to the M1 phenotype because IL9 similarly induced the mRNA and protein expression of M1 markers and did not significantly alter M2 markers. Therefore, we used BMDMs as a proper macrophage source to mimic the TAM-enriched microenvironment in an in vivo mice model. Using the coinjection model, we found that the intratumoral combination of IL9 and TAMs delayed the growth of B16F10 melanoma in mice. At first, the presence of IL9 within the tumor area was essential to generating proper antitumor immunity. Even though just a few nanograms of IL9 were secreted by the B16F10-IL9 mice per day, the consistent and even distribution of IL9 in the tumor area resulted in a strong antitumor response.
Notably, without macrophage supplementation, there was no significant difference between the B16F10-Mock and B16F10-IL9 groups, and the depletion of TAMs after tumor cell and BMDM coimplantation by treatment with clodronate liposomes abolished the antitumor responses in B16F10-IL9+BMDMs. Our data presented macrophages as a critical responder to IL9, triggering an antimelanoma immune response.
We also demonstrated the mechanisms by which the intratumoral combination of IL9 and macrophages triggered antitumor immune responses. First, intratumoral IL9 shifted the resident macrophages at the newly developing tumor site to the M1 phenotype, thus, minimizing the protumor effects of general TAMs. Second, the intratumoral combination of TAMs and IL9 significantly increased the total number of tumor-infiltrated immune cells, which included DCs, CD103 + DCs, macrophages, NKs, T cells, and especially cytotoxic effector T cells. The enhancement of T-and NK-cell infiltration into solid tumors, as well as the maintenance of an adequate TME for infiltrated immune cells, are critical points for efficacious cancer immunotherapy (37,38). Recently, intratumoral CD103 + DCs have been robustly reported for their critical roles in trafficking tumor antigen to lymph nodes for both direct CD8 + T-cell stimulation and antigen transfer to resident myeloid cells (39,40). CD103 + DC deficiency resulted in the depletion of tumor-infiltrated CD8 + Tcell populations in melanoma (41). Collectively, our results demonstrated that IL9-polarized macrophages produced a proper TME to recruit and facilitate antitumor immune cell infiltration into the tumor mass, as well as provide an antigen-presenting cell-enriched microenvironment for the cross-presentation of tumor antigens to CD8 + T cells.
Next, we further elucidated how immune cells could be recruited by colocalizing IL9 and macrophages in the tumor mass. We hypothesized that de novo infiltrated immune cells were recruited into the TME by signals from IL9-polarized TAMs. TAMs isolated from mice bearing B16F10-IL9+BMDMs had significantly higher expression levels of chemotactic chemokines, including CCL3, CCL4, CXCL9, and CXCL10. Our observations are consistent with previous observations highlighting the role of CCL3 and CCL4 chemokine systems in recruiting antitumor CD103 + DCs to the tumor, especially in the tumor models that respond poorly to CPIs (27,(41)(42)(43). Similarly, the central role of CXCR3 chemokines in the tumor-targeting migration of antitumor effector T cells was also reported in adoptive T-cell transfer models (44)(45)(46). In patients with melanoma or colorectal carcinoma, the recruitment of immune cells through CXCR3/CCR5 ligands is critical for the immune-mediated rejection of tumors (47). Furthermore, tumor-targeting approaches to recruit T cells, NK cells, and CD103 + DCs via tumor-targeted chemokine delivery significantly improved the antitumor efficacy of CPI immunotherapy (27,37,38). Collectively, our results proved that the presence of TAMs at the early timepoint of tumor progression is important for the antitumor effect of IL9. The absence of macrophages at the initial time, as in the case of the B16F10-IL9-only group, or the partial depletion of TAMs during tumor progression, as in the case of treatment with clodronate, abolished the antitumor effect of IL9. The resident macrophages at the newly developing tumor site were polarized toward the M1-like phenotype by IL9. In turn, the IL9-polarized macrophages secreted chemokines, including CCL3, CCL4, CXCL9, and CXCL10, to recruit antitumor T cells, NKs, DCs, and macrophages to infiltrate the tumor. On the basis of our data, the antitumor effect of IL9 mainly depended on the surrounding immune environment, and the presence of TAMs may be essential. Our findings support the conclusion in previous studies (17)(18)(19) macrophage niches could have a decisive role in the TME, and TAM reprogramming is a promising strategy for cancer therapy.
Finally, we extended our observations using rIL9 as an antitumor therapy for macrophage-enriched tumors. Intraperitoneal treatment with rIL9 delayed tumor growth and extended the overall survival of mice bearing macrophageenriched B16F10 melanoma and 4T1 breast cancer in a dose-dependent manner. However, the survival extension observed by intraperitoneal treatment with rIL9 was shorter than that in the intratumoral IL9 secretion model. We can infer that the amount of IL9 reaching the tumor region was small in the systemic circulation model and not evenly distributed within the tumor. Altogether, our data suggest that intratumoral supplementation with IL9 might be a viable therapeutic strategy for macrophage-enriched tumors. A very recent study on B16 melanoma metastasis showed that IL9 from CD4 + T cells exerted protumor functions by inducing the expression of Arg1, an M2 marker, in pulmonary interstitial macrophages (48). However, in the experimental conditions in our current study, IL9 treatment did not alter or reduced Arg1 expression at either the mRNA or protein level in subcutaneous B16F10 TAMs, BMDMs, or the RAW264.7 cell line. Furthermore, no enhancement of Arg1 expression was found in F4/80 + TAMs from B16F10-IL9+BMDM-coinjected mice compared with the B16F10-Mock+BMDM group. The plausible conflicting results may be attributed to the lineage diversification of macrophages, the intensity and duration of IL9 signals, and different immune cell pools of IL9 responders in the TME.
It is essential to note that until now, except for IL2 and IFNγ, the clinical use of cytokines as monotherapy has shown modest antitumor efficacy. IL15 superagonists are promising candidates in preclinical and clinical development for anticancer therapy (49). Although the remarkable antitumor efficacy of IL12 has yet to be replicated in preclinical human models, preclinical studies on the combination of IL12, IL15, and IL18 might represent ideal approaches to generating NK and DCs for cancer therapy (50,51). Above all, combining immunostimulatory cytokines with other immunotherapeutic strategies like adoptive cell transfer or CPI has shown promising clinical data. Because we focused on clarifying the mechanism underlying the antitumor effect of IL9 in macrophage-enriched tumor models in this study, further studies are needed to develop new strategies for generating locally targeted IL9 and define the therapeutic effects of targeting IL9 in TAM-enriched cancer.