Lewis lung carcinoma (LLC) was obtained from American Type Culture Collection (ATCC) (Sakai et al., 2006). LLC cell was maintained in DMEM+10% FBS+1% P/S, culture in 5% CO₂, 37°C. Six to eight-week-old C57BL/6J mice were purchased from GemPharmatech Co., Ltd. The study protocols were approved by the Institutional Animal Care and Use Committee, Sun Yat-Sen University. Mouse was shaved and subsequently inoculated with 10⁶ LLC subcutaneously. After tumor volume reached to 100 mm³, the treatment was performed. For anti-CSF1R antibody (BP0213, BioXCell, West Lebanon, NH, United States) or anti-IL-6 antibody (BE0046, BioXCell) neutralization, 50 μg/mouse of each antibody was given through i. v. Injection every 2 days. For PD-1 antibody (BP0146, BioXCell), mice were injected intraperitoneally 200 μg in PBS per mouse every 3 days, InVivoPlus rat IgG2a (BP0089, BioXCell) was used as isotype control both for anti-CSF1R antibody group (Isotype: Rat IgG2a, κ) and anti PD1 group (Isotype: Rat IgG2a, κ). For PLX3397 (S7818, Selleck, Shanghai, China), mice were fed with chow containing PLX3397 (50 mg/kg), every 2 days. In the CCL22 compensation experiment, 100 ng/mice of recombinant Mouse CCL22/MDC Protein (439-MD-025, R&D Systems, MN, United States) was given through i. v. Injection at the time of PLX3397 treatment. Mice were sacrificed when the tumor volumes reached approximately 2000 mm³ (tumor volume = 1/2 × a × b2; a = length and b = width), and tumor tissues were collected for subsequent experiments.

For multiple cytokine analysis, Mouse Th1/Th2/Th17 CBA kit (560485, BD Bioscience, CA, United States) was used as instruction described. Mixed capture beads were added into each test tube, and 50 μl diluted serum sample (serum sample was diluted 1:4) was added into each tube. Then, 50 μl mouse Inflammation PE detection reagent was added all tubes, and incubate for 2 h at room temperature in the dark. 1 ml washing buffer was added to each tube, centrifuge at 200 g for 5 min, carefully absorb the supernatant, add 300 μl washing buffer to re-suspend the capture beads. The data collection starts from the low level of standard sample to high, and ensure that the vortex for 3–5 s before sample loading; The BD FACS Canto II flow cytometer was used for data collection.

For mouse chemokine analysis, 50 μl serum was used for ELISA assay after dilute twice using PBS. Mouse MCP-1 ELISA (CCL2) (432704, Biolegend, San Diego, CA, United States), Mouse CCL22/MDC DuoSet ELISA (DY439, R&D Systems, MN, United States), Mouse CXCL9 ELISA Kit (ab203364, Abcom, Atlanta, GA, United States), Mouse IP-10 ELISA Kit (CXCL10) (ab260067, Abcom), were used for chemokine analysis. Optic density was subsequently measured at 450 nm.

For probe for the Ccl22 (NM_009137.2), the probe sequence and tag design are as follows: 5—FITC—GGACATGCATGGGCAGTGAGCAAAGTAGCA—FITC—3. For the detection, the tissue is obtained and washed using PBS, then immediately placed in the fixation solution (prepared with DEPC water) for 2–12 h. After the tissue fixation, it is dehydrated by gradient alcohol, then immersed in wax and embedded, after section was prepared and incubated in 62°C oven for 2 h. Sections was incubated into xylene for 15 min twice, 100% ethanol for 5 min twice, 85% alcohol for 5 min, 75% alcohol for 5 min, then the section was washed with DEPC. The sections maintained in the repair solution in 100°C for 10–15 min and cooled naturally. Protease K (20 μg/mL) was added for digestion at 37°C for 30 min and then washed with PBS for 3 times. Pre-hybridization solution was added and incubated at 37°C for 1 h The pre-hybridization solution containing probe (8 ng/μl) was added and incubated overnight at 37°C in an incubator. After washed with SSC, sections were stained with DAPI dye solution (2 μg/mL), and the sections were sealed for observation, the data was collected using a fluorescent microscope (DMI8, Leica, Wetzlar, GER). ImageJ was used for integrated optical density analysis.

Mechanical dissociation method was used for single cell suspension preparation. Briefly, after tumor dissection, tumor tissues were rinsed in saline to clean off blood and was minced with scissors into 1–2 mm; Then, tissue was put into 70 μm cell strainer (BS-40-XBS, Biosharp, Anhui, China), rubber tip from syringe was used to grinding tissue, single cell suspension was obtained after filtered from the strainer. Single-cell suspensions prepared from tumor or spleen tissues were washed and stained for cell surface phenotyping using the following monoclonal antibodies: CD45-eFluor 450 (48–0451-82, eBioscience), CD11b-APC cy7 (47–0112-82, eBioscience, San Diego, CA, United States), F4/80-PE (12–4801-82, eBioscience), CD115 (CSF1R)-PE-cy7 (25–1152-82 eBioscience), CD8-PE cy7 (25–0081-82, eBioscience), CD4-FITC (11–0042-85, eBioscience), CD25 PerCP-Cyanine5.5 (45–0251-82,eBioscience), FOXP3-APC (17–5773-82, eBioscience), PE anti-mouse CD194 (CCR4) Antibody (131203, Biolegend, San Diego, CA, United States). Single-cell suspensions were stained with antibodies diluted according to the instruction for 30 min, For CD45-eFluor 450 staining, Rat IgG2b kappa Isotype Control (48–4031-82) was used as isotype control, for CD11b-APC cy7, Rat IgG2b kappa Isotype Control (eB149/10H5) (47–4031-82, eBioscience) was used as isotype control, for F4/80-PE and, Rat IgG2a kappa Isotype Control (eBR2a), PE, eBioscience™(12–4321-80, eBioscience) was used as isotype control, for CCR4-PE, PE Armenian Hamster IgG Isotype Ctrl Antibody (400907, Biolegend) was used as Isotype Control; for CD115 (CSF1R)-PE-cy7 and CD8-PE cy7, Rat IgG2a kappa Isotype Control (eBR2a), PE-Cyanine7, eBioscience™(25–4321-82, eBioscience) was used as isotype control, for CD4-FITC, Rat IgG2a kappa Isotype Control (eBR2a)-FITC, eBioscience™(11–4321-80, eBioscience) was used as isotype control, for FOXP3-APC, Rat IgG2a kappa Isotype Control (eBR2a), APC, eBioscience™(17–4321-81, eBioscience) was used as isotype control, for CD25 PerCP-Cyanine5.5, Rat IgG1 kappa Isotype Control (eBRG1), PerCP-Cyanine5.5, eBioscience™(45–4301-80, eBioscience) was used as isotype control. Individual single-color controls were prepared for compensation adjust; samples were washed and resuspended suing PBS. Flow cytometry data was acquisition using CytoFLEX LX ((Beckman Coulter, Brea, CA). Flowjo 10.0 was used for data analysis. For T cell sorting, the CD3⁺CD45⁺ cells were collected using SONY MA900 (Sony, Tokyo, Japan). An example of the gating strategy is given in Supplementary Figure S1.

Anti-F4/80 MicroBeads UltraPure (130–110-443, Miltenyi Biotec, Auburn, CA, United States) was used for tumor-associated macrophage separation according to the instruction. Briefly, the single cell suspension was prepared from tumor tissue using Tumor Dissociation Kit (130–096-730, Miltenyi). 10⁷ total cell was resuspended in 90 μl buffer and 10 μl Anti-F4/80 MicroBeads UltraPure was added, and incubate for 15 min in the dark at 4°C, cells were washed by adding 1 mL washing buffer, and resuspend cells in 500 μl buffer, cells was applied onto the MS column placed in the magnetic field of the MACS separator. MS column was washed with the 3 × 500 μl washing buffer, then the column was removed from the separator and place it on the collection tube. 1 ml of washing buffer was added to the column and immediately flush out the magnetically labeled cells by firmly pushing the plunger into the column.

Data were analyzed by Graphpad 9.0 software (GraphPad, Bethesda, MD, United States). Statistical analysis was performed using unpaired two-tailed Student’s t-test for two group comparation, two-way ANOVA was used for more than two group comparation, two-tailed Mantel-Cox test was used for survival curve analysis, p values < 0.05 were considered statistically significant. ****p < 0.0001, ***p < 0.001; **p < 0.01; *p < 0.05.

Firstly, we evaluated the inhibition of TAM using different inhibitors. In addition to PLX3397 (19), a small-molecule inhibitor of CSF1R, CSF1R antibodies have also been validated in a variety of tumor models to inhibit tumor growth by inhibiting the proliferation of TAM(15). We used IL-6 antibody as another candidate for IL-6 also plays an important role in tumorigenesis and progression. It has been reported that macrophage colony-stimulating factors (M-CSF, CSF-1) can significantly promote tumor tissue growth in tumor models constructed from non-small cell lung cancer cell line (Okazaki et al., 2005). We examined the CSF1R phenotype of macrophages in the subcutaneous LLC tumor model. The results showed that TAM had a significantly positive CSF1R phenotype, while macrophages in the spleen of tumor-bearing mice had a negative CSF1R phenotype (Figure 1A). We use αCSF1R (anti-CSF1R antibody), PLX3397, and αIL-6 (anti-IL-6 antibody) to treat subcutaneous lung cancer as method described. The results showed that the αCSF1R neutralization group and PLX3397 treatment group inhibited tumor growth to a certain extent, while there was no significant difference in tumor growth rate between the αIL-6 group and the PBS group (Figure 1B). There was no anti-tumor effect in the αCSF1R isotype control group (Figure 1B). The IL-6 antibody could well neutralize the IL-6 level in the blood (Supplementary Figure S2A), but only played a slight anti-tumor effect, which was consistent with previous reports (Song et al., 2014). In the survival analysis, the survival rate of the α-IL-6 group was similar to that of the PBS group, while the PLX3397 treatment group and the αCSF1R neutralization group reached the humane end point on 27 days post treatment (DPT) began and 25 DPT, respectively (Figure 1C). We analyzed the tumor macrophages in mice on day 7 after treatment began. As expected, the percentage of macrophages was significantly lower in the group receiving CSF1R antibody or PLX3397 (Figures 1D,E). We separated the TAM and analyzed the phenotype changes. As the phenotype of macrophages in tumors tends to be M2c subtype, the main markers include Tgb1, Il10 (Roszer, 2015). The results showed that the levels of Arg1 and Tgfb1 decreased significantly after αCSF1R and PLX3397 treatment (Figures 1F,G), but the level of Il10 did not decrease significantly (Figure 1H), which indicated that the phenotype of TAM changed from M2-like to M1-like during the treatment.

Due to the important crosstalk between TAM and T cells in tumor immune environment (Bremnes et al., 2016; Cannarile et al., 2017), and in some studies, the use of PLX3397 in nude mice can achieve a better therapeutic effect (Liu et al., 2018; Wu et al., 2020). We examined the changes in the proportion of T cells during PLX3397 and αCSF1R treatment strategy (Figure 2A). The results showed that the ratio of T cells infiltration decreased to a certain extent in the treatment group, and the decrease level in the PLX3397 group was significantly on 15 DPT (Figure 2B). Analysis of CD8⁺ T cells and CD4⁺ cells in tumor after treatment showed that the proportion of CD8⁺ T cells decreased both in 9 and 15 DPT (Figures 2C,D). At the same time, the proportion of Treg cells decreased significantly in both treatment groups on 15 DPT (Figures 2E,F). In the change of immune cells ratio, Treg cells in the αCSF1R group decreased by 72.2% compared with the PBS group (0.18% CD25⁺FOXP3⁺CD4⁺CD45⁺ cells of total CD45⁺ cells in PBS group vs. 0.05% in αCSF1R group) and decreased by 77.8% in the PLX3397 group (0.18% in PBS group vs. 0.04% in PLX3397 group). The ratio of CD8⁺ T cells versus Treg cells, a key prognostic factor for cancer, increased in the treated tumors, with a significant increase in levels on both 9 and 15 DPT, with 1.16 times increase in the αCSF1R group compared to the control group (Figure 2G). Next, we want to explore the reasons for the decrease in the proportion of T cells, CD8⁺ T cells and Treg cells during CSF1R targeted therapy. One possibility is that CSF1R targeted therapy acts directly on T cells. The level of CSF1R on the surface of T cells in the tumor was analyzed and shown to be low, so the possibility of CSF1R acting directly on T cells was less likely (Supplementary Figure S2B). The information above indicated that during the inhibition of TAM, the proportion of CD8⁺ T cells and Treg cells has decreased, while the ratio of CD8⁺ T cells versus Treg cells was significantly increased. A previous article about CSF1R⁺ macrophages and Foxp3⁺ Treg cells, showed that a slight increase in both CD8⁺ T cells and Treg cells in tumors after PLX3397 treatment (Gyori et al., 2018). One possible reason is that the tumor model and the time point is vary between two research.

The proportion of T cells in tumor immune cells is also influenced by several other factors. Cytokines and chemokines also have significant influence on T cell infiltration, including CXCL9, CXCL10, CCL20, CCL22, CCL2 and CCL5, etc. (Nagarsheth et al., 2017). Treg cells can be induced by a variety of chemokines, such as CCR4‐CCL17/22, CCR8‐CCL1, CCR10‐CCL28 and CXCR3‐CCL9/10/11. Cytokines/chemokines involved in T cells infiltration were detected in the serum on 15 DPT (Figures 3A,B), and the results showed that changes in IL-4, IL-17A and IFNγ were not significant. IL-2 and IL-6 increased after αCSF1R treatment, IL-2 and TNF increased after PLX3397 treatment, and serum IL-10 decreased from 40 pg/mL to 20 pg/mL in both treatment group (Figure 3A). The analysis of chemokines related to T cell chemotaxis showed that CCL2 has no significant change after treatment, CXCL10 and CXCL9 were a little upregulated after PLX3397 treatment (Figure 3B). The level of CCL22 with the highest relative level in PBS group, was significantly decreased in the treatment group (1114.7 pg/mL vs. 493.54 pg/mL vs. 517.16 pg/mL) (Figure 3B). Previous reports have shown that CCL22 is a cytokine from macrophages that induces chemotactic activation of T cells, and its receptor is CCR4, which is mainly expressed on activated CD4⁺ T and Treg cells (Iellem et al., 2001; Sather et al., 2007). It has been reported that tumor-derived CCL22 can recruit Treg cells and promote tumor development. We further identified the time course of serum CCL22 change from the beginning of tumor inoculation. CCL22 has a relatively high level in the early stage of tumor development. After treatment began (7 days post inoculation), the level of serum CCL22 decreased significantly, and the inhibitory effect of PLX3397 on CCL22 was more acute (Figure 3C).

The sources of cytokines and chemokines may be tumor cells, stromal cells and immune cells (Curiel et al., 2004; Liu et al., 2021). We further sorted potential cell subpopulations to identify the source of CCL22. The results showed that the Ccl22 transcription level of CD45⁻ cells and macrophages increased gradually with the time of tumor progression, while the Ccl22 mRNA of T cells remained at a low level (Figure 3D). Previous reports have shown that Ccl22 is a typical marker of the M2-type phenotype of macrophages (Martinez and Gordon, 2014). It has also been reported that macrophage-derived CCL22 can induce infiltration of Treg cells (Gordon and Martinez, 2010; Wang et al., 2019). In order to make it clear that CCL22 was involved in Treg infiltration during treatment, we analysis the CCR4 level in tumor infiltrated Treg, a relatively high CCR4 level was observed in Treg (Figure 3E), we conducted a compensation experiment. The results showed that the CD8T/Treg ratio in the PLX3397 treatment group with CCL22 was significantly inhibited on 13 DPT (Figure 3F). Meanwhile, the antitumor effect of PLX3397 was partially antagonized by the use of CCL22 protein (Figure 3G). The above information indicates that PLX3397 can improve the CD8T/Treg ratio by inhibiting the TAM-derived CCL22 during tumor therapy.

Base on the PLX3397 effect on Treg cells, we treated the tumor model with PD-1 antibody in combination with PLX3397 in the subcutaneous LLC tumor model. It showed a similar tumor growth inhibition between the PD-1 monotherapy group and combination groups within 15 DPT (Figure 4A). After 40 days observation, there was a significant improvement in the survival rate (Figure 4B), and a significant reduction in tumor weight (Figures 4C,D). There was no anti-tumor effect in the αPD-1 isotype control group (Figure 4A). After combination therapy, the ratio of CD8⁺ T cells versus Treg cells was significantly increased (Figure 4E). Another independent treatment experiments consisting five mice per group showed a similar result (Figures 4F–H). The Ccl22 transcription level and expression level of in the tumor were analyzed on the 40 DPT, the transcription level (Figures 4I–K) of Ccl22 were significantly decreased. It has been reported that PD-1 antibody also has a certain effect on TAM. Our results showed that in the combination treatment group, the polarization level of macrophages was further transformed to M1, and the levels of Arg1 and Tgf1b were decreased significantly on the 15 DPT (Figure 4L).