The acute myeloid leukemia microenvironment impairs neutrophil maturation and function through NF-κB signaling

Abstract

Acute myeloid leukemia (AML), an aggressive hematological malignancy, is driven by oncogenic mutations in stem and progenitor cells that give rise to AML blasts. Although these mutations are well characterized, their impact on healthy hematopoiesis, those blood cells exposed to AML but not mutated, has not been well characterized. Because the marrow is the major site for granulopoiesis, neutrophils are heavily influenced by AML pathobiology. Indeed, most patients with AML report neutropenia, rendering them susceptible to infections. However, because AML studies use peripheral blood mononuclear cells devoid of neutrophils, the characterization of neutrophil dysfunction remains poorly understood. To investigate AML-exposed neutrophils, a preclinical AML mouse model in which primary leukemic cells were transplanted into nonirradiated neutrophil reporter (Ly6G-tdTomato; Catchup) hosts was used. Neutrophils could not completely mature, suggesting impaired granulopoiesis. Single-cell transcriptomics of AML-exposed neutrophils revealed higher inflammation signatures and expression of CD14, an inflammatory marker. To address the factors contributing to this biology, an ex vivo cytokine screen was performed on marrow neutrophils, and it identified that nuclear factor κB signaling drove CD14 expression. AML-exposed neutrophils displayed widespread chromatin remodeling, and de novo motif discovery predicted increased binding sites for CCAAT enhancer-binding proteins and interferon regulatory factors. Moreover, AML-exposed neutrophils inhibited T-cell proliferation, highlighting their immune-suppressive capability. Finally, a similar biology of immature, inflammatory neutrophils was found in patients with AML, again indicating dysregulated granulopoiesis. Collectively, these data show that AML-associated inflammation alters neutrophil granulopoiesis, impairs neutrophil function, and drives immunosuppression, thereby contributing to patient susceptibility to infection.

Graphical abstract

Figure 1.

AML-exposed neutrophils undergo altered maturation. (A) Experimental design of the Meis1-HoxA9 AML mouse model. (B) Representative confocal imaging of femur cross-sections in control and AML mice displaying laminin and neutrophils (CatchUp). (C-D) Representative gating strategy for C1, C2, and C3 neutrophils in BM (C) and blood (D) in AML. (E) Quantification of neutrophil frequency as a percentage of total CD45 cells in the BM (n > 10). (F) Simple linear regression displaying negative correlation (R² = –0.52; P ≤ .05) between WBC count and frequency of neutrophils as a percentage of total CD45 cells in the blood in our AML mouse model (n > 10). (G) Quantification of C1, C2, and C3 neutrophil subpopulations in healthy and AML mice in the BM and blood (n > 10). Data are shown as mean ± standard deviation. Significance is denoted as ∗∗∗∗P ≤ .0001, calculated using unpaired 2-tailed t test (E), simple linear regression test (F), and 2-way analysis of variance (ANOVA) (G). TX, transplant.

Figure 2.

C1, C2, and C3 neutrophil clusters in AML are more immature than their healthy counterparts and negatively corelate with disease progression. (A) Heat map displaying normalized mean fluorescence intensity (MFI) of various neutrophil markers Ly6G, CD101, CXCR2, CD62L, and SSC in all 3 neutrophil clusters in the BM of AML and healthy mice (n > 9). (B) Ly6G and SSC MFI quantified in BM neutrophils (n > 9). (C) Representative fluorescence-activated cell sorter plot displaying BM neutrophils quantified by SSC, Ly6G, CXCR2, and CD101. (D) Heat map displaying normalized MFI of various neutrophil markers Ly6G, CD101, CXCR2, CD62L, and SSC in neutrophil cluster C3 in the blood of AML and healthy mice (n = 8). (E) CD101 and SSC MFI quantified in blood neutrophils (n = 8). (F) Simple linear regression displaying negative correlation between WBC count and frequency of CD101⁺ neutrophils in the blood in AML (n > 10). (G) Simple linear regression displaying negative correlation between WBC count and frequency of CD62L⁺ neutrophils in the blood in AML (n > 10). Data are shown as mean ± standard deviation. Significance is denoted as ∗∗∗∗P ≤ .0001, not significant (ns) P > .05, calculated using 1-way ANOVA (B) and unpaired 2-tailed t test (E); and P ≤ .05 for simple linear regression test (panels F-G).

Figure 3.

scRNA-seq of AML-exposed neutrophils reveals altered granulopoiesis, decreased maturation score, and increased inflammation scores. (A) Bidimensional UMAP analysis of 4438 BM neutrophils from control mice separated into 3 clusters NEU1, NEU2, and NEU3. (B) Heat map of DEGs from each cluster. (C) Bidimensional UMAP analysis of healthy BM neutrophils (blue) and AML-exposed neutrophils (red). (D) Frequency of each NEU cluster as a percentage of total cells. (E) Maturation score of each NEU cluster. (F) Inflammation score of each NEU cluster. (G) Expression of C/EBPβ transcripts in each NEU cluster. (H) Volcano plot displaying DEG of NEU2 AML-exposed neutrophils (red; 205 genes) vs NEU2 healthy neutrophils (blue; 322 genes). (I) Gene ontology analysis of DEGs in panel H. All DEGs have fold change >1.5; P ≤ .05. Significance is denoted as ∗∗∗∗P ≤ .0001; ∗∗∗P ≤ .001; ∗P ≤ .05; ns P > .05 (using 1-way ANOVA in panels E-G).

Figure 4.

ATAC-seq of AML-exposed neutrophils reveals increased DAPs for IRF8 and C/EBPβ affecting neutrophil function. (A) Principal component analysis of ATAC-seq data from sorted healthy (n = 3) and AML-exposed neutrophils (n = 4) for each neutrophil subpopulation. (B) Heat map was generated displaying DAPs in regions closed in AML (decreased) and regions open in AML (increased) for neutrophils from subpopulation NEU2 (Ly6G^HiCD101^–). (C) De novo motif discovery of DAPs open in AML. (D) De novo motif discovery of DAPs closed in AML. (E) Volcano plot displaying genes nearest to DAPs in AML-exposed (red) vs healthy Ly6G^HiCD101 neutrophils (blue). All DAPs have fold change >1.5; P ≤ .05.

Figure 5.

GM-CSF–driven NF-κB signaling drives the formation of AML-exposed neutrophils. (A) Representative histograms of CD14 surface expression in BM neutrophils from healthy and AML mice. (B) Representative histograms of CD14 surface expression in blood neutrophils from healthy and AML mice. (C) Quantification of frequency of CD14⁺ neutrophils from total neutrophils in healthy and AML mice (n > 3). (D) Simple linear regression displaying positive correlation (r = 0.72; P < .05) between WBC count and frequency of CD14⁺ neutrophils as a percentage of total neutrophils in the blood of our AML mouse model (n > 10). (E) Western blot measuring C/EBPβ expression in sorted healthy and AML-exposed neutrophils. (F) Representative histogram of CD14 in our ex vivo cultures. (G) Quantification of CD14 expression in our ex vivo cultures (n > 3). (H) Representative histogram of CD101 in our ex vivo cultures. (I) Quantification of CD101 expression in our ex vivo cultures (n > 3). (J) Representative histogram of SSC in our ex vivo cultures. (K) Quantification of SSC expression in our ex vivo cultures (n > 3). (L) Western blot measuring C/EBPβ expression in ex vivo cultures. (M) Quantification of CD14 expression after 24 hours in various neutrophil and AML blast coculture conditions (n = 3). Significance is denoted as ∗∗∗∗P ≤ .0001 (unpaired 2-tailed t test for panel C); ∗∗∗P ≤ .001 (1-way ANOVA for panels G-K); ∗∗∗P ≤ .001; ∗∗P ≤ .01; ns P > .05 (2-way ANOVA for panel M); P ≤ .05 (simple linear regression test for panel D).

Figure 6.

NF-κB–driven genes are upregulated in AML-exposed neutrophils. (A) Bidimensional t-SNE analysis of 30 215 neutrophils from ex vivo cultures separated into the 3 conditions: G-CSF, GM-CSF, and GMD. (B) Venn diagram representing DEGs in the GMD vs GM-CSF conditions. (C) Volcano plot of DEGs upregulated in GMD conditions (441 genes; blue) or GM-CSF conditions (540 genes; red). (D) Gene ontology analysis of DEGs in panel C. (E) Score of NF-κB–dependent genes in healthy (blue) and AML-exposed neutrophils (red). (F) Inflammation score of healthy neutrophils, AML-exposed neutrophils, and neutrophils from ex vivo cultures. (G) Degranulation score of healthy neutrophils, AML-exposed neutrophils, and neutrophils from ex vivo cultures. (H) Monocyte gene score of healthy neutrophils, AML-exposed neutrophils, and neutrophils from ex vivo cultures. (I) Venn diagram representing DEGs in AML-exposed neutrophils and GM-CSF conditions. (J) Venn diagram representing DEGs in AML-exposed neutrophils and G-CSF conditions. All DEGs have fold change >1.5; P ≤ .05. Significance is denoted as ∗∗∗∗P ≤ .0001 (1-way ANOVA analysis in panels E-H). Cntrl, control.

Figure 7.

Neutrophils from patients with AML display maturation impairment and express CD14. (A) Bidimensional UMAP analysis of 68 214 CD45⁺ cells from healthy and AML-diagnosed fresh patient blood samples (n = 17). (B) Expression of CD15⁺ CD66B⁺ neutrophils (blue, healthy; red, AML). (C) Five hundred neutrophils from each human sample were projected onto a new bidimensional UMAP. (D) Surface expression of CD14 on neutrophils (blue, healthy; red, AML). (E) Surface expression of CD101 on neutrophils (blue, healthy; red, AML). (F) Surface expression of CD10 on neutrophils (blue, healthy; red, AML). (G,J) Representative histogram and quantification of expression of CD14 in healthy and patient neutrophils (n > 10). (H,K) Representative histogram and quantification of expression of CD101 in healthy and patient neutrophils (n > 10). (I,L) Representative histogram and quantification of CD10 expression in healthy and patient neutrophils (n > 10). Significance is denoted as ∗∗∗∗P ≤ .0001; ∗∗P ≤ .01.