M2 macrophages drive leukemic transformation by imposing resistance to phagocytosis and improving mitochondrial metabolism

Abstract

It is increasingly becoming clear that cancers are a symbiosis of diverse cell types and tumor clones. Combined single-cell RNA sequencing, flow cytometry, and immunohistochemistry studies of the innate immune compartment in the bone marrow of patients with acute myeloid leukemia (AML) reveal a shift toward a tumor-supportive M2-polarized macrophage landscape with an altered transcriptional program, with enhanced fatty acid oxidation and NAD⁺ generation. Functionally, these AML-associated macrophages display decreased phagocytic activity and intra-bone marrow coinjection of M2 macrophages together with leukemic blasts strongly enhances in vivo transformation potential. A 2-day in vitro exposure to M2 macrophages results in the accumulation of CALR^low leukemic blast cells, which are now protected against phagocytosis. Moreover, M2-exposed "trained" leukemic blasts display increased mitochondrial metabolism, in part mediated via mitochondrial transfer. Our study provides insight into the mechanisms by which the immune landscape contributes to aggressive leukemia development and provides alternatives for targeting strategies aimed at the tumor microenvironment.

Fig. 1.. Heterogeneity within the macrophage landscape in AML.

(A) Heatmap displaying the differentially expressed genes used to perform the SEURAT analysis (scRNA-seq GSE116256). (B) UMAP projection of monocytic/macrophage-like BM cells from nine patients with AML and two HDs, showing the formation of 12 main clusters. (C) Pie charts representing the frequency distribution of cells in each cluster, regarding the sample origin (HD versus AML). (D) Volcano plot demonstrating the differentially expressed genes in monocytes/macrophages from patients with AML compared to healthy BM donors. (E) UMAP projection of differentially expressed genes in AAMs associated with an M2-like phenotype. (F) Representative FACS plots of macrophage marker expression in different AML CD45⁺ subpopulations. (G) The expression of macrophage markers in the AML blasts (SSC^lowCD45^dim) versus the mature myeloid population (SSC^highCD45^high). (H) The amount of CD163⁺CD206⁺ macrophages detected in patients with AML compared to HDs. (I) The amount of CD163⁺ macrophages quantified by immunohistochemistry (IHC) and FACS. Representative IHC pictures for CD163 in AML biopsies. (G and I) Wilcoxon signed rank test (two-sided); n.s., nonsignificant; *P < 0.05 and ****P < 0.0001. (H) Mann-Whitney test for unpaired data (two-sided). *P < 0.05 and ***P < 0.001. (G to I) Each dot represents an individual patient.

Fig. 2.. AAMs display impaired phagocytic activity.

(A) Single-cell scGSEA projection in the macrophage landscape of patients with AML and HD. (B) Density plots displaying the enrichment scores for the scGSEA. (C) Bar plot displaying the fold change of phagocytosed carboxyfluorescein succinimidyl ester (CSFE)–labeled MV4-11 cells by HD (CTRL) and AAM. Oncoprint displaying the baseline mutations of the patients with AML from which macrophages were isolated. (D) Bar plot displaying the fold change compared to control of phagocytosed CSFE-labeled MV4-11 cells by empty vector control and NPM1cyt, MN1-overexpression, BCR-ABL, shDNMT3A, FLT3-ITD (W81) CB-derived transduced macrophages. (E) Cell cycle analyses of macrophages derived under (D). (F) Cumulative cell count of primary AML cells cultured on M0/M2d macrophages (closed circles) or MS5 (control, diamonds) for 14 days. Similar colors indicate similar patient samples. (G) In vivo coinjection experimental setup. (H) Leukocyte count in transplanted mice (n = 4). WBC, white blood cell. (I) Bar graphs of the percentage of human CD45⁺ (left) and human APL blast cells (CD117⁺CD33⁺) (right) detected in the BM of mice without (control) or with coinjected macrophages (n = 4). LL, left leg; RL, right leg, injected with M0 and M2d macrophages, respectively. Representative FACS plot. The material of one mouse was excluded from the BM/spleen chimerism data due to low cell viability. (J) BM representative cytospins of control and coinjected mice. Arrows indicate human APL blast cells. (K) Human CD45⁺ and APL blasts (CD117⁺CD33⁺) cells (%) detected in the spleen (n = 4) and spleen weight. Representative spleen pictures of mice injected without (control) and with macrophages (Mac). (C, D, and E) Kruskal-Wallis test. (I) Independent sample t test. (F and H) Two-way analysis of variance (ANOVA). (K) Mann-Whitney test for unpaired data (two-sided). (D, F, I, and K) Each dot represents an individual patient. Data indicate the SEM. *P < 0.05, **P < 0.01, and ****P < 0.0001.

Fig. 3.. Heterogeneity in intrinsic phagocytosis sensitivity across leukemia samples is associated with CALR expression and maturation stage.

(A) Bar plot displaying the percentage of phagocytosed CSFE-labeled primary AML cells by murine (NSG-isolated) macrophages. (B) Bar plot displaying the percentage of phagocytosed CSFE or Incucyte Red–labeled primary AML cells by human PB-derived macrophages. Oncoprint displaying the baseline mutations and ELN2022-risk stratification of the patients with AML. NA, not available. (C) Pearson correlation between the level of phagocytosis measured at diagnosis and the CD47 mean fluorescence intensity (MFI) level, as well as the percentage of CARL, measured on the respective AML blast population (SSC^lowCD45^dimCD34⁺ or CD117⁺ for CD34⁻ AMLs). (D) Phagocytosis of primary AML at diagnosis and after 48 hours of co-culture on M2d macrophages. (E) CALR transcript level of primary AML at diagnosis (D0) and after a 48-hour coculture on M2d macrophages. Data are represented as a fold change to diagnosis (D0) levels. D0, day 0/diagnosis. (F) Heatmap of genes differentially expressed between CALR^low and CALR^high patients analyzed on the transcriptome of CD34⁺-sorted AML cells (GSE30029). (G) GSEA of CALR^low and CALR^high patients analyzed on the transcriptome of CD34⁺-sorted AML cells. NES, normalized enrichment score. (H) FACS measured expression of CALR in CD34⁺ and CD34⁻ leukemic blast cells. (I) M2-AAM score in patients with the highest (10 patients) and lowest (10 patients) expression of CALR in the TCGA set. (J) Colony formation of CALR^neg and CALR^pos cells sorted from primary APL/AML blast cells within the SSC^low CD45^dim fraction. Representative colonies of CALR^neg and CALR^pos cells. CALR^neg, calreticulin negative; CALR^pos, calreticulin positive. (B to E and H to I) Each dot represents an individual patient. (B) Mann-Whitney test for unpaired data (two-sided). (D, E, H, and J) Wilcoxon signed rank test (two-sided). *P < 0.05, **P < 0.01, and ***P < 0.001. Data indicate the SEM.

Fig. 4.. Coculture of non-engrafting APL cells on M2 macrophages induces full-blown leukemia in a xenograft model.

(A) Experimental in vivo setup. (B) OS of mice transplanted with primary APL blast transplanted without pre-culture or after preculture on M0/M2d macrophages for 48 hours. (C to F) Mice transplanted with APL blasts without pre-culture (Ctrl) or after preculture on M0/M2d macrophages were analyzed for WBC counts (C), for human CD45⁺CD117⁺CD33⁺ chimerism (%) measured in the BM (D), for human CD45⁺CD117⁺CD33⁺ chimerism (%) measured in the spleen (E), and spleen weight with representative spleen pictures (F). (G) OS of secondary transplanted mice receiving 1.5 × 10⁶, 5 × 10⁵, or 1 × 10³ sorted human CD33⁺CD117⁺ APL blast cells from the primary transplant described in (B) to (F). (n = 3 to 6 per group). HR, hazard ratio. (H) Human CD45⁺ chimerism levels (%) measured in the BM and spleen of secondary transplanted mice. (I) Representative cytospin of human APL blast cells retrieved from the murine BM. (J) In vivo Long-Term Culture-Initiating Cell (LTC-IC) analyses to determine LSC frequencies. 95% CI, 95% confidence interval. (K) Mice transplanted with AML blasts without pre-culture (Ctrl) or after preculture on M2d macrophages were analyzed for human CD45⁺CD33⁺ chimerism measured in the PB 6, 12, 18, and 24 weeks after transplant. (L) OS of mice transplanted with primary AML blast transplanted without pre-culture (control) or after preculture on M2d macrophages for 48 hours. (M and N) Mice transplanted with AML blasts without pre-culture (Ctrl) or after preculture on M2d macrophages were analyzed for human CD45⁺ chimerism (%) measured in the BM (M), for human CD45⁺ chimerism (%) measured in the spleen and spleen weight (N). (B, G, and L) Each dot represents an individual mouse, OS curves were estimated using the Kaplan-Meier method, and the log-rank test was used for comparison. (C and K) Two-way ANOVA. (D to F and H) Kruskal-Wallis test. (M and N) Wilcoxon signed rank test (two-sided). *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. 5.. M2-derived mitochondrial transfer promotes AML proliferation, which can be targeted.

(A) Liquid cell proliferation of primary AML cells (technical duplicate/AML) and (B) CFU at day 14 of primary AML cells exposed to MS5/M2d macrophages (48 hours). (C) GSEA of the AML proteome with high versus low M2 macrophage (FACS). (D) GSEA/Gene Ontology (GO) analysis of the transcriptome (E) OCR/relative maximal OCR and (F) mitochondrial ATP production rate of primary AML cells exposed to MS5/M2d cells (48 hours). (G) Mitochondrial transfer (MTF) experimental scheme. (H) MTF measured in primary AML cells co-cultured on mitochondrial labeled MS5/M2d cells (48 hours). (I) Experimental scheme of AML cells co-cultured on etomoxir (Eto)–treated MS5/M2d cells. (J) OCR and (K) liquid proliferation of primary AML cells (three biological replicates) exposed to vehicle- or Eto (50 μM)–treated MS5/M2d cells (24 hours). (L) AML cells co-cultured on MS5/M2d cells treated with vehicle or drugs targeting FAO/mitochondrial ETC (24 hours) and then transferred to liquid culture after 48 hours. Heatmap displays AML liquid cell proliferation after exposure to treated MS5/M2d cells. (M) Pearson correlation of M2 macrophage levels (%) and ex vivo sensitivity in primary AML samples (n = 35). Blue and red dots indicate significant negative and positive correlations, respectively. (N) NAMPT expression in AAM. (O) Apoptosis (72 hours) induced in AML blasts and AAM after ex vivo treatment with Ara-C, VEN, KPT-9274 (n = 35), and Daporinad (n = 10). (P) Total NAD⁺ levels in healthy M1/M2d macrophages treated with KPT-9274 or Daporinad. (Q) OCR of M2d macrophages/AAMs (n = 2) exposed to KPT-9274 or Daporinad. (B, E, F, H, J, O, and P) Each dot represents an individual patient. (A, K, L, P, and Q) Two-way ANOVA. (B, E, F, and H) Wilcoxon signed rank test (two-sided). (J) Friedman test. (O) Kruskal-Wallis test. (Q) Represents a technical duplicate. Data indicate the SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and **** P < 0.0001.