Short-term starvation boosts anti-PD-L1 therapy by reshaping tumor-associated macrophages in hepatocellular carcinoma

Abstract

Background and aims: Immune checkpoint inhibitors have revolutionized systemic HCC treatment. Nevertheless, numerous patients are refractory to immune checkpoint inhibitor therapy. It is currently unknown whether diet therapies such as short-term starvation (STS) combined with immune checkpoint inhibitors can be used to treat HCC. This study aimed to investigate whether STS could sensitize HCC tumors to immunotherapy.

Approach and results: STS was found to attenuate tumor progression by inducing tumor-associated macrophages (TAMs) to switch to an antitumoral phenotype, enhancing phagocytosis of tumor cells, and stimulating subsequent antitumor immunity of CD8 + T cells as demonstrated in 3 HCC mouse models, NCG mice, and Rag2-KO mice. Furthermore, STS combined with anti-programmed cell death 1/ligand 1 (anti-PD-1/L1) suppressed tumor progression, while the efficacy of PD-L1 was improved when combined with STS. Mechanistically, TAM-derived exosomal PD-L1 (exoPD-L1) impairs the efficacy of anti-PD-1/L1. STS attenuates exoPD-L1 secretion from TAM by regulating the fructose diphosphatase 1 (FBP1) /Akt/Rab27a axis. Modulating FBP1/Akt/Rab27a axis potentiates the anti-PD-L1 response using 2 liposomal delivery systems and macrophage adoptive transfer.

Conclusions: This study describes the immunomodulatory effects of STS and provides a rationale for its application as an adjuvant in HCC immunotherapy.

Keywords: diet therapies; exosome; immunotherapy; tumor-associated macrophage.

Graphical abstract

FIGURE 1

STS suppresses the growth of multiple HCC types. (A, B) Schematic diagram of the experimental approach of the STS and SCR regimen (described in Methods section). (C) Tumor size or weight analysis over time (n=6 mice per group). (D) Survival curve of the mice from the ND, SCR, and STS groups (n=10 mice per group). (E) Daily body weight change during the experimental period, expressed as the percentage of body weight relative to pre-experiment levels (n=6 mice per group). (F, G) Schematic diagram of the experimental approach of STS in the orthotopic and hydrodynamic HCC models. (H, I) Gross tumor images and weight statistics in the orthotopic and hydrodynamic HCC models (n=6 mice per group), scale bar=5 mm. (J) Survival curve of the mice from the orthotopic HCC model (n=10 mice per group). (K) Daily body weight of the mice from the orthotopic HCC model (n=6 mice per group). (L) Survival curve of the mice from the hydrodynamic HCC model (n=10 mice per group). (M) Daily body weight of mice from the hydrodynamic HCC model (n=6 mice per group). Error bars represent the mean±SEM. Statistical analyses were performed using Student t test, one-way ANOVA, and two-way ANOVA with Tukey multiple comparisons test, and Log-rank test, *p<0.05, **p<0.01, ***p<0.001. Abbreviations: ND, normal diet; SCR, severe calorie restriction; STS, short-term starvation.

FIGURE 2

STS depends on systemic immunity and reconstructs the immune microenvironment. (A) Schematic diagram of the experimental approach. Hepa1-6 cells cultured under high-nutrient conditions (10% FBS and 4.5 g Glu/L) or low-nutrient conditions (1% FBS and 2 g Glu/L) were inoculated subcutaneously in the left flank (immunization side) of immunocompetent mice (C57BL/6J mice) or immunodeficient mice (NCG mice). On the seventh day after the injection, Hepa1-6 cells cultured under high-nutrient conditions (10% FBS and 4.5 g Glu/L) were subcutaneously inoculated in the right flank (naive tumor side). (B) Tumor weight or size analysis of the immunization side of immunocompetent mice (n=6 mice per group). (C) Tumor weight or size analysis of the immunization side of immunodeficient mice (n=3 mice per group). (D) Tumor weight or size analysis of naive tumor side of immunocompetent mice (n=6 mice per group). (E) Tumor weight or size analysis of naive tumor side of immunodeficient mice (n=3 mice per group). (F) Immune infiltration was estimated using xCELL for expression profiles tested from RNA-seq data of the orthotopic HCC model (n=3). (G) tSNE analysis using Flowjo was used to detect differences in the immunophenotypes of infiltrating immune cells detected by flow cytometry in the orthotopic HCC model from both the ND and STS groups. tSNE map derived from flow cytometric analysis indicating the populations of the infiltrating immune cells, including T cells (CD45⁺CD11b⁻CD3⁺), G-MDSCs and M-MDSCs (CD45⁺CD11b^hiGr1^{hi or mid}), NKT cells (CD45⁺CD11b⁻CD3⁺NK1.1⁺), NK cells (CD45⁺CD11b⁻CD3⁻NK1.1⁺), B cells (CD45⁺CD11b⁻CD3⁻CD19⁺), Dendritic cells (CD45⁺CD11b⁺Gr1^loF4/80⁻CD11c⁺IA/IE⁺), Macrophages (CD45⁺CD11b^hiGr1^loF4/80⁺), pTAMs (CD45⁺CD11b^hiGr1^loF4/80⁺CD86^loCD206^hi), aTAMs (CD45⁺CD11b^hiGr1^loF4/80⁺CD86 ^hi CD206^lo). Percentages of immune cell populations within CD45⁺ cells are shown in the right panel (n=6). (H) The functional state of the T cell subpopulation was analyzed by tSNE analysis and flow cytometric quantification; CD4⁺ T cells (CD3⁺CD4⁺CD8⁻), Tregs (CD3⁺CD4⁺CD8⁻ CD25⁺Foxp3⁺), CD8⁺ T cells (CD3⁺CD4⁻CD8⁺), Granzyme B⁺ T cells (CD8⁺ Granzyme B⁺), IFNγ⁺ T cells (CD8⁺ IFNγ⁺), Percentages of immune cell populations within CD45⁺ cells are shown at the right panel (n=6). (I) tSNE analysis by Flowjo for memory or exhaustion T cell subpopulation; T_EM (CD8⁺CD44⁺CD69⁻CD62L⁻), T_CM (CD8⁺CD44⁺CD69⁻CD62L⁺), T_RM (CD8⁺CD44⁺CD69⁺CD62L⁻), Tmem (CD4⁺CD44^hiCD62L^hi), TeffTmem (CD4⁺ CD44^hiCD62L^lo), Tn (CD4⁺ CD44^loCD62L^hi), TIM-3⁺ T cells (CD8⁺TIM-3⁺), LAG-3⁺ T cells (CD8⁺LAG3⁺), PD-1⁺ T cells (CD8⁺PD-1⁺), CTLA-4⁺ T cells (CD8⁺CTLA-4⁺), (J) Percentages of immune cell populations within CD3⁺ cells are shown (n=6). (K) Representative tumor images from the orthotopic HCC model treated using Clophosome or anti-CD8. (L) Tumor weight analysis from the orthotopic HCC model treated using Clophosome or anti-CD8 (n=4 mice per group). (M) Survival curve of the mice from the orthotopic HCC model treated by Clophosome or anti-CD8 (n=10 mice per group). Error bars represent the Mean±SEM. Statistical analyses were performed using Student t test, one-way ANOVA, and two-way ANOVA with Tukey multiple comparisons test, and Log-rank test, *p<0.05, **p<0.01, ***p<0.001. Abbreviations: NCG, NOD/ShiLtJGpt-Prkdcem26Cd52Il2rgem26Cd22/Gpt; ND, normal diet; STS, short-term starvation; T_CM, central memory T cell; T_EM, effector memory T cells; T_RM, tissue-resident memory T cells.

FIGURE 3

STS expedites macrophage polarization to aTAMs and enhances macrophage phagocytosis. (A) A co-culture system of murine macrophages BMDMs (bottom) and Hepa1-6 cells (top). Left: high-nutrient conditions (10% FBS and 4.5 g Glu/L); middle: low-nutrient conditions (1% FBS and 2 g Glu/L); right: culturing BMDMs alone under low-nutrient conditions. (B) Relative expression of the indicated genes using qPCR. (C) The top 6 immuno-related GSEA pathways from the transcriptome of the orthotopic HCC model and the positive regulation of phagocytosis, are presented for both the ND and STS groups. (D) BMDMs were incubated with 1 mg/mL dextran-FITC molecules for 60 minutes at 4°C or at 37°C, and the percentage of FITC⁺ cells and MFI of FITC on F4/80⁺ cells were measured using flow cytometry. Red: macrophage engulfing FITC molecules. Blue: macrophage without FITC molecules. (E) A co-culture system of both CFSE-labeled BMDMs and PKH26-stained Hepa1-6 cells was subjected to starvation for 0, 12, 24, and 48 hours (n=3). Debris and dead cells are washed by PBS. Representative images and the percentage of macrophages engulfing tumor cells in the visual field are shown. Green: BMDMs; Red, Hepa1-6 cells. White asterisks denote the macrophages engulfing tumor cells. Magnification×20, scale bar=100 µm. (F) Statistic analysis and representative histograms showing intratumoral TAMs that phagocytic tumor cells from ND and STS groups. mCherry^hi TAMs were considered to be phagocytosing (n=6). (G) Schematic diagram of the experimental approach and statistical analysis of tumor weight or size. BMDMs or F4/80⁺ macrophages (TAMs) from tumor-bearing mice pretreated by high or low-nutrient conditions were mixed with Hepa 1-6-mCherry-luc⁺ cells and inoculated in the left flank of C57BL/6J mice. (H) Statistical analysis and representative histograms show the phagocytosis of intratumoral macrophages, mCherry^hi TAM was considered to be phagocytosing (n=4). (I) Schematic diagram of the experimental approach. BMDMs from GFP-positive mice pretreated by high or low-nutrient conditions were injected intravenously into C57BL/6J mice with tumors, 3 injections every 6 days. (J) In vivo bioluminescent imaging (BLI) of tumor growth in the orthotopic HCC model at the experimental endpoint. (K) Gross tumor images and weight statistics of mice undergoing macrophage adoptive therapy (n=4 mice per group). (L) Representative immunofluorescence images and analysis of CD86 from the tumor (n=4). (M) Statistical analysis and representative histograms show the phagocytosis of intratumoral macrophages (n=4). Error bars represent the Mean±SEM. Statistical analyses were performed using Student t test, one-way ANOVA, and 2-way ANOVA with Tukey multiple comparisons test. *p<0.05, **p<0.01, ***p<0.001. Abbreviations: aTAMs, antitumoral TAMs; BMDMs, bone marrow-derived macrophages; GSEA, Gene set enrichment analysis; ND, normal diet; pTAMs, protumoral TAMs; STS, short-term starvation; TAMs, tumor-associated macrophages.

FIGURE 4

STS-mediated TAMs improve CD8⁺ T cells-mediated cytotoxic and memory response. (A) A co-culture system of Hepa 1-6 cells and CD8⁺ T cells or BMDMs from OT-1 mice. (B) The functional state of CD8⁺ T cells was analyzed using flow cytometric quantification of GZMB⁺ and IFNγ⁺ cells (n=6). (C) of Hepa1-6 cells caused by CD8⁺ T cells was measured following Annexin V-PI staining (n=6). (D–G) Representative plots and percentages of IFNγ or GZMB -producing OT-1 CD8⁺ T cells co-cultured with the SIINFEKEL peptide or the OVA protein-loaded BMDMs or F4/80⁺ TAMs isolated from tumor-bearing mice (n=3). (H) The functional state of the T-cell subpopulation from tumor-bearing mice was analyzed by flow cytometric quantification. The sample includes tumor tissue from inoculation of BMDMs or F4/80⁺ TAMs pretreated by high or low-nutrient conditions and Hepa 1–6 cells or tumor tissue treated by BMDMs adoptive therapy (n=4). (I) Memory T cell subpopulation was analyzed by flow cytometric quantification (n=4). (J) Schematic diagram of the experimental approach. Rag2^−/− mice bearing Hepa1-6-OVA were injected intravenously with CD8⁺ T cells co-cultured with BMDMs pretreated with high-nutrient or low-nutrient conditions on days 6 and 13. These mice were treated with clodronate liposome or control liposome post-CD8⁺ T cells adoptive therapy. (K) Tumor weight or size statistics of mice Rag2^−/− mice bearing Hepa1-6-OVA treated by CD8⁺ T cells or clodronate liposome (n=3 mice per group). Error bars represent the mean±SEM. Statistical analyses were performed using Student t test, one-way ANOVA with Tukey multiple comparisons test, and two-way ANOVA, *p<0.05, **p<0.01, ***p<0.001. Abbreviations: BMDMs, bone marrow-derived macrophages; ND, normal diet; STS, short-term starvation; TAMs, tumor-associated macrophages.

FIGURE 5

Effect of STS combined with anti-PD-L1 is superior to anti-PD-1. (A) Schematic diagram of the experimental approach. (B) Tumor weight statistics and survival curve of the orthotopic HCC model mice undergoing single regimen or combined therapy (n=8–10 mice per group). (C) Representative plots and percentages of TAMs in the orthotopic HCC model from the ND, STS, anti-PD-1, anti-PD-L1, STS combined with anti-PD-1, and STS combined with anti-PD-L1 groups (n=6). (D) Statistical analysis and representative histograms show the functional state of CD8⁺ T cells in 6 groups (n=6). (E) Statistical analysis of memory CD8⁺ T-cell subpopulations in 6 groups (n=6). (F) Gross tumor images and tumor weight analysis from the orthotopic HCC model from the ND group, STS group, and STS combined with anti-PD-L1 group treated by Clophosome or anti-CD8 (n=5 mice per group). (G) Statistical analysis of CD8⁺ T cell and F4/80⁺ macrophage clearance from mice treated with STS or STS combined with anti-PD-L1. Error bars represent the mean±SEM. Statistical analyses were performed using one-way ANOVA with Tukey multiple comparisons test and Log-rank test, *p<0.05, **p<0.01, ***p<0.001. Abbreviations: aTAMs, antitumoral TAMs; ND, normal diet; PD-L1, programmed cell death ligand 1; pTAMs, protumoral TAMs; STS, short-term starvation; TAMs, tumor-associated macrophages.

FIGURE 6

exoPD-L1 is enriched in TAMs and participates in anti-PD-1/L1 resistance. (A) tSNE map derived from flow cytometric analysis indicating the populations of the infiltrating PD-1⁺ or PD-L1⁺cells in blood, spleen, and tumor tissue. (B) Statistical analyses of PD-1⁺ or PD-L1⁺ myeloid cells (CD45⁺CD11b⁺), lymphocytes (CD45⁺CD11b⁻), and nonimmune cells (CD45⁻) in blood, spleen, and tumor tissue (n=3). (C) tSNE analysis by Flowjo for the expression of PD-1⁺ or PD-L1⁺ in myeloid cells including G-MDSC (CD11b^hiGr1^hiF4/80⁻), TAMs (CD11b^hiGr1^hiF4/80⁺), M-MDSC (CD11b^hiGr1^midF4/80⁻), and DCs (CD11b^hiGr1^loF4/80⁻CD11c⁺IA/IE⁺). The red areas represent TAMs. (D) Statistical analyses and percentage of PD-1⁺ or PD-L1⁺ myeloid cell and percentage of myeloid cell populations within CD11b⁺ cells (n=3). (E) Left: the protein levels of PD-L1, CD63, and TSG101 in whole-cell lysate (Wcl) and exosomes (Exo) from TAMs. Right: the exosome protein levels of PD-L1, CD63, and TSG101 in Hepa1-6 and TAMs. (F) Tumor weight or size statistics from mice inoculated with Hepa1-6 mixed with BMDMs (Rab27a KO or WT) (n=6 mice per group). All mice were treated with anti-PD-L1 or isotype. (G) Tumor weight or size statistics from mice inoculated with Hepa1-6 (Rab27a KO) mixed with BMDMs (Rab27a KO) All mice were treated with anti-PD-L1 or isotype. Some mice were injected intravenously with Hepa1-6 exosome (Hepa1-6 Exo) or TAM exosome (TAM Exo) (n=6 mice per group). (H) Schematic diagram showing the preparation of Rab27a siRNA-loaded liposomes and experimental approach and BLI of tumor growth. (I) Gross tumor images and tumor weight analysis of mice received liposome or anti-PD-L1 (n=6 mice per group). (J) Survival curve of siNC group, siRab27a group, siNC+anti-PD-L1 group, and siRab27a+anti-PD-L1 group (n=6 mice per group). Error bars represent the Mean±SEM. Statistical analyses were performed using one-way ANOVA and two-way ANOVA with Tukey multiple comparisons test, and Log-rank test, *p<0.05, **p<0.01, ***p<0.001. Abbreviations: DC, dendritic cell; DMG-PEG, dimyristoyl glycerol-polyethylene glycol; MDSCs, myeloid-derived suppressor cells; PD-L1, programmed cell death ligand 1; TAMs, tumor-associated macrophages.

FIGURE 7

STS inhibits Rab27a and exoPD-L1 of TAMs through activating FBP1 (A) TAMs were isolated from mice subjected to ND or STS diet. The protein levels show Rab27a in Wcl and PD-L1, CD63, and TSG101 in exosomes (Exo). (n=3 mice per group) (B) Representative immunofluorescent staining for F4/80⁺ or PD-L1⁺ cells from mice undergoing ND or STS regimen. F4/80 (green), PD-L1 (red), and DAPI (blue), Magnification ×80, scale bar=25 µm. (C) KEGG pathway analysis and term candidate gene number from the orthotopic HCC model with ND or STS assessed by RNA-seq (n=3). The differences in FPKM values in carbon metabolism genes are shown in the right panel. (D) ScRNA-seq data (GSE149614) were analyzed. UMAP plots showing multiple cell clusters including antitumoral TAM, protumoral TAM, MDSC, CD4 T cell, CD8 T cytotoxic cell, CD8 T exhausted cell, CD8 T Naive cell, NK cell, B cell, Endothelial, Fibroblast, Hepatocarcinoma cell, MAST cell (left upper), the expression of FBP1 (right upper), FBP1 high and low origins (left lower) and cell proportion statistics (Right lower panel). (E) Protein levels of the key enzymes in glycolysis (HK2, PFKFB3, PKM2, LDHA) and the key enzymes in gluconeogenesis (G6PC3, FBP1, PCK1, PCK2) were assessed by western blot in TAMs collected by the Hepa1-6 cells and BMDMs co-culture system (left). Protein levels of the indicated enzymes or exosomes following silencing HK2 and LDHA or overexpressing G6PC3 and FBP1 in TAMs (right). (F) The Rab27a and exoPD-L1 protein levels after silencing or overexpressing FBP1 in TAMs subjected to ND or STS-like conditions. (G) Protein levels of the Akt/mTOR pathway in TAMs subjected to ND or STS-like conditions. (H) Co-immunoprecipitation of FBP1 and Akt in TAMs or TAMs FBP1-KO isolated from mice subjected to ND or STS diet (n=3). (I) Co-immunoprecipitation of FBP1 and Akt from BMDMs co-cultured with Hepa 1-6 cells (upper). Immunofluorescence assays for FBP1 (green) with Akt (red) in TAMs (below). Magnification ×80, scale bar=25 µm. (J) Protein levels of the Akt/mTOR/Rab27a axis following silencing or overexpressing of FBP1 in TAMs added with MK2206, Rapamycin, and rIGF-1. (K) The mRNA expression of Rab27a and indicated protein levels in TAMs treated by MK2206, Rapamycin, and rIGF-1. (L) Tumor weight analysis and survival curve of four group mice (n=6 mice per group). (M) Schematic diagram showing STS reshapes the TME by triggering a series of antitumor immune responses. Error bars represent the mean±SEM. Statistical analyses were performed using Student t test, one-way ANOVA with Tukey multiple comparisons test, and Log-rank test, *p<0.05, **p<0.01, ***p<0.001. Abbreviations: BMDMs, bone marrow-derived macrophages; ND, normal diet; STS, short-term starvation; TAMs, tumor-associated macrophages; Wcl, whole-cell lysate.