Comprehensive immunophenotyping reveals distinct tumor microenvironment alterations in anti-PD-1 sensitive and resistant syngeneic mouse model

Anti-PD-1 mAb treatment inhibited tumor growth in the MC-38 model but not in the LLC1 model.

To evaluate the therapeutic efficacy of anti-PD-1 mAb treatment, tumor growth was monitored in the MC-38 and LLC1 syngeneic mouse models (Fig. 1A). In the MC-38 model, anti-PD-1 mAb treatment remarkably inhibited tumor growth compared to the isotype control group (Fig. 1B,C)). However, in the LLC1 model, the difference in tumor growth between the anti-PD-1 mAb and isotype control groups (Fig. 1B,C) was minimal. Body weight measurements did not reveal any significant treatment-related toxicity, suggesting no severe side effects in mice (Fig. 1D).

Differential effects of anti-PD-1 mAb treatment on monocytes/granulocytes populations in TdLNs and tumors.

We conducted flow cytometry analysis to characterize distinct immune cell profiles of monocytes/granulocytes among all CD45⁺ leukocytes harvested from TdLNs and tumors in the MC-38 and LLC1 models upon anti-PD-1 mAb treatment. In the TdLNs, both mouse models treated with anti-PD-1 mAb showed a subtle increase in the ratios of dendritic cells (DCs), myeloid-derived suppressor cells (MDSCs) and macrophages (MΦ) compared to the isotype control (Fig. 2A). In the tumor microenvironment, the MC-38 model sensitive to anti-PD-1 mAb exhibited a significant increase in the ratios of DCs and MΦ compared to the isotype control (Fig. 2B). The ratios of DCs and MΦ were 18.8% and 39.6%, respectively, in the isotype control group and increased to 25.7% and 41.0%, respectively, in the anti-PD-1 mAb treated group. In contrast, the ratio of MDSCs decreased from 18.6 to 27.2% upon anti-PD-1 treatment (Fig. 2B). Both rate of increase in DCs and rate of decrease in MDSCs in tumors were greatest in MC38-derived tumors treated with PD-1 (Fig. 2C). Conversely, in the LLC1 model resistant to anti-PD-1 mAb treatment, anti-PD-1 mAb treatment had minimal effect on the ratios of these monocyte/granulocyte populations.

Immunophenotyping analysis of monocytes/granulocytes in TdLNs and tumors from mice intraperitoneally administered anti-PD-1 antibody after inoculation with LLC1 cells or MC-38 cells. Anti-PD-1 mAb or isotype control (each 100 μg/body, n = 5) was administered a total of 3 times to mice every 3 days from 6 days after the inoculation with LLC1 cells and MC-38 cells (each 1 × 10⁶ cells/body). TdLNs and tumors were harvested 7 days after the final dosing, digested into single cell suspensions and analyzed by flow cytometry using fluorescence-conjugated antibodies. The average ratios of several monocytes/granulocytes, including DCs, MΦ and MDSCs, and non-monocytes/granulocytes in CD45⁺ cells from TdLNs (A) and tumors (B) are shown in pie charts with the respective ratios in parentheses. The percent differences between anti-PD-1 mAb treatment and isotype control-treatment for each monocyte/granulocyte in TdLNs and tumors were obtained by subtracting the average ratios of isotype control-treatment from those of anti-PD-1 mAb treatment in panel (C). The average ratios of CD80, CD86, MHC class I (MHC I), MHC class II (MHC II), CD197, and PD-L1 of several monocytes/granulocytes, including DCs, MΦ, and MDSCs, in CD45⁺ cells from TdLNs (D) and tumors (E) are shown in bar graphs. The average ratios of M1 phenotype MΦ (CD11c⁺CD206^–), M2 phenotype MΦ (CD11c^–CD206⁺), and MΦ in CD45⁺ cells from TdLNs (F) and tumors (G) are shown in bar graphs. Bar graphs depict the average values with SD (D–G). Significant differences are denoted with asterisks (*P < 0.05) (E, G).

We next investigated the functional status and activation status of monocyte/granulocyte populations. The expression levels of antigen presentation molecules (MHC class I and MHC class II) and immune checkpoint molecules (CD80, CD86, CD197, and PD-L1) on tumor-associated DCs, MΦ, and MDSCs were analyzed. In the MC-38 model, anti-PD-1 mAb treatment significantly upregulated the expression levels (MFI, mean fluorescence intensity) of MHC class I and PD-L1 on DCs and MΦ within the tumor microenvironment but not in TdLNs (Supplementary Fig. 1). However, in the LLC1 model, no significant changes in the expression of these molecules were observed upon anti-PD-1 mAb treatment (Supplementary Fig. 1). Similarly, anti-PD-1 mAb treatment in the MC-38 model significantly increased the ratios of the aforementioned activation-related markers such as CD80, MHC class I, MHC class II, CD197 (CCR7), and CD274 (PD-L1) in DCs among CD45⁺ cells in tumors, whereas anti-PD-1 mAb treatment in the LLC1 model had minimal effect (Fig. 2E). Similarly, increases in these activation markers on monocytes/granulocytes were also observed in TdLNs from the MC-38 model, although they were not significant (Fig. 2D). The anti-PD-1 mAb treatment in the MC-38 model significantly also increased the ratios of MΦ expressing MHC class II and CD197 and those of MDSCs expressing MHC class II in tumors compared to the isotype control (Fig. 2E). For tumor-associated MΦ (TAM) in TdLNs, anti-PD-1 mAb treatment in the MC-38 model increased M1 phenotype macrophages (CD11c⁺CD206⁻), which represent a pro-inflammatory phenotype that plays a crucial role in the host defense against tumor cells, and the same treatment in the LLC1 model showed a significant increase in M2 phenotype macrophages (CD11c⁻CD206⁺), which represented an inhibitory phenotype in anti-tumor immunity (Fig. 2F). Similarly, only anti-PD-1 mAb treatment in the MC-38 model significantly increased the M1 phenotype macrophages, but no change was seen in M2 phenotype macrophages (Fig. 2G).

Differential effects of anti-PD-1 mAb treatment on lymphocyte populations in TdLNs and tumors.

We characterized distinct immune cell profiles of lymphocytes among all CD45⁺ leukocytes harvested from TdLNs and tumors in the MC-38 and LLC1 models upon anti-PD-1 mAb treatment. In the TdLNs, both mouse models treated with anti-PD-1 mAb showed an approximately 3% to 6% decrease in the ratios of CD4⁺T cells and CD8⁺T cells and an approximately 8% increase in the ratio of B cells compared to the isotype control, although there was no statistically significant difference (Fig. 3A,C). In NK cells, there was a slight increase in the ratio upon anti-PD-1 mAb treatment (Fig. 3A,C). Notably, in the tumor microenvironment, the MC-38 model sensitive to anti-PD-1 mAb treatment revealed a statistically significant increase in the ratios of CD4⁺T cells, CD8⁺T cells, Tregs, NK cells, and NKT cells but not B cells compared to the isotype control (Fig. 3B,D), demonstrating a vigorous influx of various immune cell fractions into the tumor bed. Conversely, in the LLC1 model resistant to anti-PD-1 mAb treatment, this intervention had minimal effect on the ratios of these immune cell populations (Fig. 3B,D).

Immunophenotyping analysis of lymphocytes in TdLNs and tumors from mice intraperitoneally administered anti-PD-1 antibodies after inoculation with LLC1 cells or MC-38 cells. Anti-PD-1 mAb or isotype control (100 μg/body, n = 5) was administered a total of 3 times to mice every 3 days from 6 days after inoculation with LLC1 cells or MC-38 cells (1 × 10⁶ cells/body). TdLNs and tumors were harvested 7 days after the final dosing, digested into cell suspensions, and analyzed by flow cytometry using fluorescence-conjugated antibodies. The average ratios of several lymphocytes, including CD4⁺T cells, CD8⁺T cells, regulatory T cells (Tregs), B cells, NK cells, and NKT cells, and non-lymphocytes in CD45⁺ cells from TdLNs (A) and tumors (B) are shown in pie charts with the respective ratios in parentheses. The average ratios of CD3⁺T cells, CD4⁺T cells, CD8⁺T cells, Tregs, B cells, NK cells, and NKT cells in CD45⁺ cells from TdLNs (C) and tumors (D) are shown in bar graphs. Bar graphs depict the average values with SD. Significant differences are denoted with asterisks (*P < 0.05, **P < 0.005).

Subsequently, we assessed the profiles of memory phenotypes of lymphocytes among all CD45⁺ leukocytes harvested from TdLNs and tumors in the MC-38 and LLC1 models upon anti-PD-1 mAb treatment. In TdLNs, there was a robust decrease in the ratio of naïve CD4⁺T cells and CD8⁺T cells but minimal effects on the ratios of CD62L⁺CD44⁺ (central memory: T_CM) and CD62L^–CD44⁺ (effector memory: T_EM) in mice treated with anti-PD-1 mAb compared to the isotype control in both LLC1 and MC-38 models (Fig. 4A). Conversely, in the tumor microenvironment, the MC-38 model sensitive to anti-PD-1 mAb treatment showed statistically significant increases in the ratios of both CD8⁺T_CM cells and CD8⁺T_EM cells but not in the ratios of CD4⁺T_CM cells and CD4⁺T_EM cells compared to the isotype control, whereas minimal changes of these immune cell subsets were observed in the LLC1 model (Fig. 4B). These results illustrated that anti-PD-1 mAb treatment robustly promoted the infiltration of both memory phenotypes of effector CTLs into tumors to exert their cytocidal effects.

Analysis of naïve and memory CD4⁺T cells and CD8⁺T cells in TdLNs and tumors from mice intraperitoneally administered anti-PD-1 antibodies after inoculation with LLC1 cells or MC-38 cells. Anti-PD-1 mAb or isotype control (100 μg/body, n = 5) was administered a total of 3 times to mice every 3 days from 6 days after the inoculation with LLC1 cells or MC-38 cells (1 × 10⁶ cells/body). TdLNs and tumors were harvested 7 days after the final dosing, digested into cell suspensions and analyzed by flow cytometry using fluorescence-conjugated antibodies. The average ratios of CD62L⁺CD44^– (naïve), CD62L⁺CD44⁺ (central memory: T_CM), and CD62L^–CD44⁺ (effector memory: T_EM) in CD4⁺T cells or CD8⁺T cells in CD45⁺ cells from TdLNs (A, left: CD4⁺T cells, right: CD8⁺T cells) and tumors (B, left: CD4⁺T cells, right: CD8⁺T cells) are shown in bar graphs. Bar graphs depict the average values with SD. Significant differences are denoted with asterisks (*P < 0.05).

Anti-PD-1 mAb treatment enhanced T cell, B cell, and NK cell activation in the MC-38 model.

While tumor killing is ascribed to CD8⁺T cell (CTL) function, pre-clinical and clinical studies have identified intratumoral cytotoxic CD4⁺T cells that can directly kill cancer cells by secreting granzyme B and perforin^11,12,13. To assess the cytotoxic functional status of immune cells, the expression of 3 different cytotoxic markers (CD107a, granzyme B, and perforin) on CD4⁺T cells, CD8⁺T cells, Tregs, B cells, NK cells, and NKT cells in tumors were comparatively analyzed. In cytotoxic CD4⁺T cells, among 3 cytotoxic makers, only a subset of perforin-expressing CD4⁺T cells was significantly higher in the anti-PD-1 mAb treatment compared to the isotype control (Fig. 5A). In CD8⁺CTL, all subpopulations of CD107-, granzyme B-, and perforin-expressing CD8⁺T cells were significantly increased in the MC-38 model, with perforin-positive cells showing the highest absolute number and change of rate (Fig. 5B). Recent studies demonstrated new roles of Tregs and B cells that possess cytotoxic activities^14,15,16,17. Similar to CD8⁺T cells, CD107-, granzyme B-, and perforin-expressing Tregs and B cells were significantly increased in the MC-38 model, with perforin-positive cells showing the highest absolute number and change of rate (Fig. 5C,D). Interestingly, in regard to other innate cytotoxic cells such as NK and NKT cells, only perforin-expressing NK cells were significantly increased in the MC-38 model (Fig. 5E,F). In order to evaluate the most reliable cytolytic activation markers among these 3 molecules, we next performed correlation analyses to investigate the relationships between tumor volumes and average ratios of a range of lymphocytes expressing CD107a, granzyme B, or perforin in tumors treated with anti-PD-1 mAb in both LLC1 cells and MC-38 cells. In the LLC1 model resistant to anti-PD-1 mAb treatment, no data showed a significant negative correlation upon anti-PD-1 mAb treatment (Fig. 6A and Table 1). In contrast, in the MC-38 model sensitive to anti-PD-1 mAb treatment, granzyme B- and perforin-expressing cytotoxic CD4⁺T cells and perforin-expressing NKT cells exhibited a significant negative correlation upon anti-PD-1 mAb treatment (r = − 0.975, − 0.995, − 0.934; P < 0.01, P < 0.01, P < 0.05; respectively, Fig. 6B and Table 1). Regarding CD8⁺CTLs as immunologically representative cytotoxic immune cells, perforin-expressing CD8⁺T cells showed the strongest negative correlation among the 3 markers in the MC-38 model treated with anti-PD-1 mAb, although there was no statistically significant difference (r = − 0.497; Fig. 6B and Table 1). Only perforin-expressing cells showed negative correlation coefficients upon anti-PD-1 mAb treatment in all immune fractions, although there were marked differences in the values (r = − 0.995, − 0.497, − 0.464, − 0.405, − 0.003, − 0.934; Fig. 6B and Table 1). These results collectively implicated that perforin may be the most reliable marker associated with the magnitude of efficacy of anti-PD-1 mAb treatment, and furthermore suggested anti-PD-1 mAb treatment enhances the cytotoxic potential of various lymphocytes, including not only CD8⁺T cells, NK cells and NKT cells, but also CD4⁺T cells, Tregs, and B cells within the tumor microenvironment.

Analysis for average ratios of lymphocytes expressing CD107a, granzyme B or perforin in tumors from mice intraperitoneally administered anti-PD-1 antibodies after inoculation with LLC1 cells or MC-38 cells. Anti-PD-1 mAb or isotype control (100 μg/body, n = 5) was administered a total of 3 times to mice every 3 days from 6 days after the inoculation with LLC1 cells or MC-38 cells (1 × 10⁶ cells/body). Tumors were harvested 7 days after the final dosing, digested into cell suspensions, and analyzed by flow cytometry using fluorescence-conjugated antibodies. The average ratios of CD107a⁺, granzyme B⁺ or perforin⁺ in several lymphocytes, including CD4⁺T cells (A), CD8⁺T cells (B), regulatory T cells (Tregs; C), B cells (D), NK cells (E), and NKT cells (F) in CD45⁺ cells from tumors are shown in bar graphs. Bar graphs depict the average values with SD. Significant differences in bar graphs are denoted with asterisks (*P < 0.05, **P < 0.005).

Correlation analysis between tumor volume and average ratios of lymphocytes expressing CD107a, granzyme B or perforin in tumors from mice intraperitoneally administered anti-PD-1 antibodies after inoculation with LLC1 cells or MC-38 cells. Anti-PD-1 mAb or isotype control (100 μg/body, n = 5) was administered a total of 3 times to mice every 3 days from 6 days after the inoculation with LLC1 cells or MC-38 cells (1 × 10⁶ cells/body). Tumors were harvested 7 days after the final dosing, digested into cell suspensions and analyzed by flow cytometry using fluorescence-conjugated antibodies. The relationships between tumor volumes and the average ratios of CD107a⁺, granzyme B⁺ or perforin⁺ of several lymphocytes including CD4⁺T cells, CD8⁺T cells, regulatory T cells (Tregs), B cells, NK cells, and NKT cells in CD45⁺ cells in tumors from LLC1 tumor-bearing mice (A) and MC-38 tumor-bearing mice (B) are shown in scatter plots with approximation lines.

When looking at TdLNs that foster the antigen-specific effector adaptive immune cells primed by antigen presenting cells, granzyme B- and perforin-expressing CD8⁺T cells were significantly decreased in the MC-38 model treated with anti-PD-1 mAb, with perforin-positive cells showing the highest reduction (P < 0.05) (Supplementary Fig. 2B). Similarly, in Tregs and NKT cells, there were some reductions of granzyme B-, perforin-expressing Tregs, and NKT cells (Supplementary Fig. 2C,F).