Mitochondria‐targeted atovaquone promotes anti‐lung cancer immunity by reshaping tumor microenvironment and enhancing energy metabolism of anti‐tumor immune cells

List of abbreviations

Aco1

aconitase 1

Aco2

aconitase 2

ATO

atovaquone

CD4IL2RAHI

IL2RA‐high CD4⁺ Treg

CD4IL2RALO

IL2RA‐low CD4⁺ Treg

CD4T_Cytotoxic

cytotoxic CD4⁺ T cells

CD4T_Exhausted

exhausted CD4⁺ T cells

CD8T_EffectorMemory

effector memory like CD8⁺ T cells

CD8T_Exhausted

exhausted CD8⁺ T cells

CD8T_MemoryLike

memory like CD8⁺ T cells

CD8T_Naive

naive CD8⁺ T cells

cDC

conventional dendritic cells

FDR

false discovery rate

Glud1

glutamate dehydrogenase 1

G‐MDSC

granulocytic myeloid‐derived suppressor cells

Got1

glutamic‐oxaloacetic transaminase 1

Idh3a

isocitrate dehydrogenase 3 alpha

Idh3b

isocitrate dehydrogenase 3 beta

Idh3g

isocitrate dehydrogenase 3 gamma

Ldha

lactate dehydrogenase A

Mdh1

malate dehydrogenase 1

Mdh2

malate dehydrogenase 2

Mito‐ATO

mitochondria‐targeted atovaquone

Mpc2

mitochondrial pyruvate carrier 2

NES

normalized enrichment score

NK

natural killer cells

OXPHOS

oxidative phosphorylation

pDC

plasmacytoid dendritic cells

Pkm

pyruvate kinase

ROS

reactive oxygen species

TCA

tricarboxylic acid

Tgd

gammadelta T cells

TILPRED

tumor‐infiltrating CD8⁺ lymphocytes states predictor

TPP

triphenylphosphonium

Tregs

regulatory T cells

Dear Editor,

Atovaquone (ATO), a mitochondrial inhibitor, has anti‐cancer effects [1]. Based on ATO, we developed mitochondria‐targeted atovaquone (Mito‐ATO) that had even stronger anti‐tumor efficacy than ATO [2]. We synthesized Mito‐ATO by attaching the bulky triphenylphosphonium (TPP) group to ATO via a ten‐carbon alkyl chain (Supplementary file of methods; Supplementary Figure S1). To assess the effects of Mito‐ATO on tumor microenvironment, we conducted single‐cell RNA‐sequencing (scRNA‐seq) on treated immune cells from mice having lung tumors either treated with or without Mito‐ATO. Seurat was used for clustering and annotation of CD45⁺ immune cells [3]. The detected lymphoid cell populations were CD8⁺ T cells, CD4⁺ T cells, regulatory T cells (Tregs), gamma‐delta T (Tgd) cells, B cells, and natural killer (NK) cells; and the myeloid cells identified were macrophages, neutrophils, plasmacytoid dendritic cells (pDCs), conventional dendritic cells (cDCs) and mast cells (Figure 1A‐C). Clustering of CD4⁺ T cells into seven subpopulations, the separation of neutrophils and granulocytic myeloid‐derived suppressor cells (G‐MDSCs), and the division of macrophages into M1 and M2 subtypes were described in our previous publication [2]. In this study, we further divided CD8⁺ T cells into four subpopulations, i.e., exhausted CD8⁺ T (CD8T_Exhausted) cells, memory like CD8⁺ T (CD8T_MemoryLike) cells, effector memory like CD8⁺ T (CD8T_EffectorMemory) cells and naive CD8⁺ T (CD8T_Naive) cells, using the tumor‐infiltrating CD8⁺ lymphocyte state predictor (TILPRED) method [4] (Figure 1D‐E). Probability scores computed with TILPRED could discriminate CD8T_Exhausted from CD8T_MemoryLike cells despite overlap between the two subsets on UMAP representation (Supplementary Figure S2). Mito‐ATO treatment significantly decreased the proportion of the CD8T_Exhausted cells (7.3% vs. 32.5%, P < 0.001) but increased the proportion of anti‐tumor CD8T_EffectorMemory cells as compared with vehicle treatment (37.3% vs. 11.9%, P < 0.001) (Figure 1F). In comparison, the percentages of CD8⁺ T cells out of total T cells were not different between the two groups (Supplementary Table S1). For validation, we verified that Mito‐ATO treatment induced changes in CD8⁺ T cell repartition by conducting flow cytometry. Mito‐ATO treatment significantly increased the percentage of cytotoxic tumor necrosis factor‐alpha (TNF‐α)⁺CD8⁺ T cells and decreased the percentage of programmed cell death protein‐1 (PD‐1)⁺ T cell immunoglobulin and mucin domain‐containing protein 3 (TIM3)⁺CD8⁺ T cells (Supplementary Figure S3). These matched the scRNA‐seq results. We also observed a slight trend toward the upregulation of genes involved in CD8⁺ T cell recruitment: Ccl25, Ccr7, Cxcl10, Cxcr3, Icam1, and S1pr1 (Supplementary Figure S4).

FIGURE 1.

Single‐cell RNAseq analysis of immune cells from mice lung tumors to identify mechanisms of Mito‐ATO's anti‐tumor function. (A) The expression of the marker genes for the T and B lymphocytes; (B) The expression of the marker genes for the NK cells, macrophages, neutrophils, pDC, cDC and mast cells; (C) The annotation of all the sequenced immune cells to the above lymphocytes and myeloid cell populations according to marker gene expression analysis; (D) Heatmap of the expression of the markers for each CD8⁺ T cell subset. (E) The clustering of the CD8⁺ T cell subsets, i.e., the exhausted CD8⁺ T cells, memory like CD8⁺ T cells, effector memory like CD8⁺ T cells, naive CD8⁺ T cells subpopulations; (F) Percent changes of the CD8⁺ T cell subsets across the control and the Mito‐ATO treatment groups; (G) The activity of the KEGG metabolism pathways were compared between Mito‐ATO and control group across the anti‐tumor and pro‐tumor CD45⁺ immune cell subpopulations from the mice lung tumors. The red stars mark the ten metabolism pathways whose activity changes were similar to OXPHOS activity changes by Mito‐ATO treatment in these immune cell populations. The blue stars mark the four cell death‐related pathways that were negatively correlated with OXPHOS activity changes by Mito‐ATO treatment; (H) Under Mito‐ATO treatment, the key metabolic reactions of the TCA cycle and glutamine metabolism were up‐regulated across the anti‐tumor immune cell populations of effector memory like CD8⁺ T cells, memory like CD8⁺ T cells, CD4T_Cytotoxic, and M1 macrophage cells; (I) The key metabolic reactions of the TCA cycle and glutamine metabolism were down‐regulated across the immunosuppressive immune cell subpopulations of CD4IL2RALO, G‐MDSC, mast cells, CD4IL2RAHI and CD4T_Exhausted cells by Mito‐ATO treatment; (J) The metabolic reaction steps in TCA cycle and glutamine pathways that exhibited the ‘plasticity’ in reaction activity under Mito‐ATO treatment between anti‐tumor and pro‐tumor immune cells, which involved the key enzymes of Aco1/Aco2, Mdh1/Mdh2, Idh3a/Idh3b/Idh3g in the TCA cycle and the Glud1 enzyme in the glutamine metabolism; (K) The pyruvate metabolism pathway was significantly up‐regulated by Mito‐ATO in the Mito‐ATO‐stimulated OXPHOS‐high cells; (L) The pyruvate metabolism pathway was significantly down‐regulated by Mito‐ATO in the Mito‐ATO‐suppressed OXPHOS‐low cells; (M) The glutamine metabolism pathway was not significantly changed by Mito‐ATO in the Mito‐ATO‐stimulated OXPHOS‐high cells; (N) The glutamine metabolism pathway was significantly down‐regulated by Mito‐ATO in the Mito‐ATO‐suppressed OXPHOS‐low cells; (O) The up‐regulation of the pyruvate metabolism genes by Mito‐ATO in the Mito‐ATO‐stimulated OXPHOS‐high cells and the down‐regulation of these genes in the Mito‐ATO‐suppressed OXPHOS‐low cells; (P) The up‐regulation of the glutamine metabolism genes by Mito‐ATO in the Mito‐ATO‐stimulated OXPHOS‐high cells and the down‐regulation of these genes in the Mito‐ATO‐suppressed OXPHOS‐low cells; (Q) OCR analyses of vehicle‐ versus mito‐ATO treated activated CD8⁺ T cells in vitro; (R) ECAR analyses of vehicle‐ versus mito‐ATO treated activated CD8⁺ T cells in vitro; (S) OCR analyses of vehicle‐ versus mito‐ATO treated G‐MDSC cells in vitro; (T) ECAR analyses of vehicle‐ versus mito‐ATO treated G‐MDSC cells in vitro. Data are shown as the mean ± SE; n = 5‐16 per group. *** P < 0.001; **** P < 0.0001. Abbreviations: Tregs, regulatory T cells; Tgd, gammadelta T cells; NK cells, natural killer cells; pDC, plasmacytoid dendritic cells; cDC, conventional dendritic cells; G‐MDSC, granulocytic myeloid derived suppressor cells; CD8T_Exhausted, exhausted CD8⁺ T cells; CD8T_MemoryLike, memory like CD8⁺ T cells; CD8T_EffectorMemory, effector memory like CD8⁺ T cells; CD8T_Naive, naive CD8⁺ T cells; KEGG, Kyoto Encyclopedia of Genes and Genomes; OXPHOS, oxidative phosphorylation; Mito‐ATO, mitochondria‐targeted atovaquone; CD4T_Cytotoxic, cytotoxic CD4⁺ T cells; CD4IL2RALO, IL2RA‐low CD4⁺ Treg; CD4IL2RAHI, IL2RA‐high CD4⁺ Treg; CD4T_Exhausted, exhausted CD4⁺ T cells; TCA, tricarboxylic acid; ROS, reactive oxygen species; Aco1, aconitase 1; Aco2, aconitase 2; Mdh1, malate dehydrogenase 1; Mdh2, malate dehydrogenase 2; Idh3a, isocitrate dehydrogenase 3 alpha; Idh3b, isocitrate dehydrogenase 3 beta; Idh3g, isocitrate dehydrogenase 3 gamma; Glud1, glutamate dehydrogenase 1; FDR, false discovery rate; NES, normalized enrichment score; Ldha, lactate dehydrogenase A; Mpc2, mitochondrial pyruvate carrier 2.

Mito‐ATO treatment significantly up‐regulated oxidative phosphorylation (OXPHOS) activity in four anti‐tumor immune cell populations, i.e., CD8T_EffectorMemory cells, CD8T_MemoryLike cells, cytotoxic CD4⁺ T cells (CD4T_Cytotoxic), and M1 macrophage cells (Figure 1G). Particularly, the upregulated genes after Mito‐ATO treatment were significantly enriched for T cell differentiation (Supplementary Figure S5), suggesting that Mito‐ATO treatment may induce CD8⁺ T cell differentiation. In contrast, Mito‐ATO treatment significantly down‐regulated OXPHOS activity in five pro‐tumor immune cell populations, i.e., interleukin‐2 receptor subunit alpha (IL2RA)‐low CD4⁺ (CD4IL2RALO) Tregs, G‐MDSCs, mast cells, IL2RA‐high CD4⁺ (CD4IL2RAHI) Tregs, and exhausted CD4⁺ T cells (CD4T_Exhausted) (Figure 1G). The two types of Treg cells were named following the previous practice [2, 5]. Ten metabolic pathways activity changes were similar to OXPHOS activity changes by Mito‐ATO treatment in the above immune cell populations. These pathways were glycolysis, the tricarboxylic acid (TCA) cycle, pyruvate metabolism, glutamine metabolism, Complex I, Complex III, Complex V, DNA repair, purine metabolism and pyrimidine metabolism (Figure 1G). The changes of four cell death‐related pathways, i.e., DNA damage, apoptosis, cell death, and reactive oxygen species (ROS) pathways, were negatively correlated with OXPHOS activity changes by Mito‐ATO treatment (Figure 1G). These suggested that Mito‐ATO enhances energy metabolism and suppresses cell death in anti‐tumor immune cells while inhibiting energy metabolism and promoting cell death in pro‐tumor immune cells.

Compass metabolism analysis [6] showed similar results (Supplementary Figures S6‐S14) . The key metabolic reactions of the TCA cycle and glutamine metabolism were up‐regulated in the anti‐tumor immune cell populations but down‐regulated in the pro‐tumor immune cell populations (Figure 1H‐I). We identified key metabolic reactions differentially regulated by Mito‐ATO across anti‐tumor and pro‐tumor immune cells, including aconitase 1 (Aco1)/Aco2, malate dehydrogenase 1 (Mdh1)/Mdh2, isocitrate dehydrogenase 3 alpha (Idh3a)/Idh3b/Idh3g in TCA cycle and the glutamate dehydrogenase 1 (Glud1) enzyme in glutamine metabolism. These metabolic reactions were consistently up‐regulated by Mito‐ATO in anti‐tumor immune cells but down‐regulated in pro‐tumor immune cells (Figure 1J).

Furthermore, we performed gene set enrichment analysis (GSEA) [7]. We combined datasets of anti‐tumor immune cells, which were termed Mito‐ATO‐stimulated OXPHOS‐high cells since their OXPHOS was enhanced by Mito‐ATO treatment. Analogously, we combined the pro‐tumor immune cells and named them Mito‐ATO‐suppressed OXPHOS‐low cells. The pyruvate pathway was up‐regulated by Mito‐ATO in Mito‐ATO‐stimulated OXPHOS‐high cells (false discovery rate [FDR] = 0.18, normalized enrichment score [NES] = 1.23, Figure 1K) while down‐regulated in Mito‐ATO‐suppressed OXPHOS‐low cells (FDR = 0.13, NES = ‐1.22) (Figure 1L). The glutamine pathway was not significantly changed by Mito‐ATO in Mito‐ATO‐stimulated OXPHOS‐high cells (FDR = 0.65, NES = ‐0.86, Figure 1M) but down‐regulated in Mito‐ATO‐suppressed OXPHOS‐low cells (FDR = 0.03, NES = ‐1.40, Figure 1N). For pyruvate metabolism, Mito‐ATO treatment up‐regulated lactate dehydrogenase A (Ldha), mitochondrial pyruvate carrier 2 (Mpc2), pyruvate kinase (Pkm) expression in Mito‐ATO‐stimulated OXPHOS‐high cells while down‐regulated their expression in Mito‐ATO‐suppressed OXPHOS‐low cells (Figure 1O). For the glutamine metabolism pathway, Mito‐ATO treatment up‐regulated Glud1, glutamic‐oxaloacetic transaminase 1 (Got1) and Mdh1 expression in Mito‐ATO‐stimulated OXPHOS‐high cells while down‐regulated their expression in Mito‐ATO‐suppressed OXPHOS‐low cells (Figure 1P). Interestingly, Mdh1 and Glud1 were also identified by Compass analysis to be significant TCA cycle and glutamine metabolic reaction regulators (Figure 1J). Therefore, the anti‐cancer efficacy of Mito‐ATO treatment may be realized through differential regulation of TCA and glutamine metabolism between anti‐tumor and pro‐tumor immune cells, in which the expression changes of Mdh1 and Glud1 could be critical. As orthogonal validation, the Seahorse metabolic flux assay showed that Mito‐ATO treatment significantly increased OXPHOS activity and aerobic glycolysis in activated CD8⁺ T (Figure 1Q‐R). In contrast, Mito‐ATO significantly suppressed OXPHOS and glycolysis in G‐MDSCs (Figure 1S‐T). These supported the predicted higher OXPHOS and glycolysis in effector memory CD8⁺ T cells and lower OXPHOS and glycolysis in G‐MDSCs upon Mito‐ATO treatment.

The metabolic plasticity of different types of cells in the tumor microenvironment (TME) in response to a glutaminase inhibitor Ethyl 2‐(2‐amino‐4‐methylpentanamido)‐DON (JHU083) has been reported [8]. In the present study, we found that Mito‐ATO may differentially regulate pyruvate metabolism, glutamine metabolism and TCA cycle across immune cells with distinct roles in the TME. The metabolic plasticity effects caused by Mito‐ATO treatment may contribute to the overall efficacy of this drug on lung tumors. Mito‐ATO's parental compound – ATO has begun to be applied to anti‐cancer clinical trials [9]. Given that Mito‐ATO is much more potent against human cancer cell lines compared to ATO [10], it is reasonable to predict that Mito‐ATO has great potential in clinics.

DECLARATIONS AUTHOR CONTRIBUTIONS

DHX did the data analysis of this project and drafted the manuscript. ZY helped with Compass analysis. MFH did the mice experiment and scRNA‐seq experiment. MH synthesized the Mito‐ATO compound used in this study. YW, BK, and STW helped in preparing experimental samples and worked with MY to revise the manuscript. MY also designed this study.

CONFLICT OF INTEREST

The authors declare no conflict of Interest.

FUNDING INFORMATION

This research was supported by National Institutes of Health (NIH): R01CA223804, R01CA232433, R01CA205633, and R01CA280746.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

The animal study was reviewed, and all procedures were approved by the Medical College of Wisconsin (MCW) Institutional Animal Care and Use Committee (Ethics approval number of AUA00001807).

CONSENT FOR PUBLICATION

Not applicable.

Supporting information

Supporting Information

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.