Neutrophils Recruited by NKX2-1 Suppression via Activation of CXCLs/CXCR2 Axis Promote Lung Adenocarcinoma Progression

Abstract

NK2 Homeobox 1 (NKX2-1) is a well-characterized pathological marker that delineates lung adenocarcinoma (LUAD) progression. The advancement of LUAD is influenced by the immune tumor microenvironment through paracrine signaling. However, the involvement of NKX2-1 in modeling the tumor immune microenvironment is still unclear. Here, the downregulation of NKX2-1 is observed in high-grade LUAD. Meanwhile, single-cell RNA sequencing and Visium in situ capturing profiling revealed the recruitment and infiltration of neutrophils in orthotopic syngeneic tumors exhibiting strong cell-cell communication through the activation of CXCLs/CXCR2 signaling. The depletion of NKX2-1 triggered the expression and secretion of CXCL1, CXCL2, CXCL3, and CXCL5 in LUAD cells. Chemokine secretion is analyzed by chemokine array and validated by qRT-PCR. ATAC-seq revealed the restrictive regulation of NKX2-1 on the promoters of CXCL1, CXCL2, and CXCL5 genes. This phenomenon led to increased tumor growth, and conversely, tumor growth decreased when inhibited by the CXCR2 antagonist SB225002. This study unveils how NKX2-1 modulates the infiltration of tumor-promoting neutrophils by inhibiting CXCLs/CXCR2-dependent mechanisms. Hence, targeting CXCR2 in NKX2-1-low tumors is a potential antitumor therapy that may improve LUAD patient outcomes.

Keywords: cheomkine; lung adenocarccinoma; single cell NGS; tumor microenvironment.

Figure 1

Low expression of NKX2‐1 is associated with aggressive LUAD. A) A representative image of NKX2‐1 IHC staining performed on tumor tissue microarrays of different grades of LUAD. B and C) Histoscore (H‐score) quantification of NKX2‐1 IHC staining performed on a tissue microarray of different LUAD grades (B) and stages (C). The data are presented as means ± SD error bars, N = 3, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns – not significant (Student's t‐test). D and E) Kaplan‐Meier curves showing the overall survival (D) and disease‐free survival of NKX2‐1 LUAD patients from the TCGA dataset (E). p‐values were calculated by the log‐rank test. F) Schematic representation showing the design of experiments performed on both orthotopic and tail vein injected mouse models by using shNKX2‐1/LL2 and shCtrl/LL2 cells. G) Photographs (top panels) and GFP fluorescence images (bottom panels) of tumors derived from orthotopically injected shCtrl/LL2 (left panel) and shNKX2‐1/LL2 cells (middle and right panels). H) Hematoxylin‐eosin staining and IHC staining of NKX2‐1 were performed on the tumors derived from orthotopically injected shCtrl/LL2 and shNKX2‐1/LL2 cells. Left panel: 2197 representative images. Right panel: H‐score quantification. The data are presented as means with SD error bars, N = 3, ****p < 0.0001 (Student's t‐test). I and J) Bioluminescent signal visualization (I) and photographs (J) of tumors derived from shCtrl/LL2 and shNKX2‐1/LL2 cells injected via the tail vein.

Figure 2

NKX2‐1 expression negatively correlates with neutrophil infiltration. A) Scatter plots showing the correlation between NKX2‐1 expression and the indicated immune cell infiltrate levels in LUAD from TIMER dataset analysis. The gene expression levels against tumor purity are displayed first because genes highly expressed in the TME are expected to exhibit negative associations with tumor purity and vice versa. B) H‐score quantification of the expression levels of the selected neutrophil markers (ITGAM, CEACAM8, ELANE, and CXCR2) performed on tumor tissue microarrays of NKX2‐1‐positive and negative LUAD tissues. The data are presented as means with SD error bars, ****p < 0.0001 (Student's t‐test). C) The TCGA dataset analysis shows the correlation between NKX2‐1 and the selected neutrophil markers (ITGAM, CEACAM8, ELANE, and CXCR2) in LUAD. D) Relapse‐free survival curve showing the survival of LUAD patients with high or low neutrophil infiltration from the TCGA dataset analysis. p‐values were calculated by the log‐rank test. E) IHC staining of the indicated neutrophil markers in the cross sections of shCtrl/LL2 and shNKX2‐1/LL2‐derived tumors from mouse orthotopic model. Left panel: representative images. Right panel: H‐score quantification. The data are presented as means with SD error bars, N = 3, ***p < 0.001, ****p < 0.0001 (Student's t‐test). F) A schematic illustration showing that NKX2‐1‐low LUAD tumors are characterized by high infiltration of neutrophils.

Figure 3

NKX2‐1‐low tumors promote neutrophil infiltration via CXC chemokines. A) Diagram demonstrating the experimental procedure for scRNA‐seq and Visium in situ capturing. B) UMAP analysis from scRNA‐seq showing the distinct cell populations across 2 samples (shCtrl/LL2 and shNKX2‐1/LL2‐derived tumors). C and D) iTALK analysis of scRNA‐seq validating the crosstalk between tumor cells and immune cells (C) and cell‐to‐cell communication (D) between cell types and the involved genes. E) Gene ontology biological process (GO‐BP) enrichment analysis of differentially expressed genes in shNKX2‐1/LL2 tumors compared to shCtrl/LL2 tumors identified by scRNA‐seq. F) UMAP analysis from scRNA‐seq showing myeloid leukocyte migration in shCtrl/LL2 and shNKX2‐1/LL2 tumors. G) Hematoxylin‐eosin staining (H&E) section annotations in shCtrl/LL2 and shNKX2‐1/LL2 tumors (top panel) and representative image from Visium in situ capturing showing the activation of intracellular signal transduction and myeloid leukocyte activation. (H and I) Flow cytometry identification of Ly‐6G⁺CD11b⁺ cells from healthy lungs and metastasized tumor‐bearing lungs (shCtrl/LL2 and shNKX2‐1/LL2) H). The quantification of Ly‐6G⁺CD11b⁺ cells (I) is shown as means with SD error bars, **p < 0.01 (Student's t‐test).

Figure 4

NKX2‐1 downregulation affects the expression of CXC chemokines. A and B) Functional enrichment analysis showing the enrichment of GO biological process (GO‐BP) (A) and molecular function (GO‐MF) (B) terms among the genes positively regulated by NKX2‐1 knockdown. C) Hierarchical clustering heatmap showing the differential gene expression (left panel) and the expression of CXC chemokines (right panel) upon NKX2‐1 knockdown. D and E) qRT‐PCR analysis showing the expression levels of CXC chemokines in HCC827 (D) and H1975 (E) cells subjected to NKX2‐1 knockdown. The mean fold changes (N = 3) relative to scrambled shRNA control (shCtrl) are shown with SD error bars, *p < 0.05, **p < 0.01, ***p < 0.001 (Student's t‐test). F and G) Chemokine array analysis showing the secretion levels of the indicated chemokines following NKX2‐1 knockdown in HCC827 (F) and H1975 cells (G). Left panel: representative dot plot. Right panel: densitometry quantification of dot blots. *p < 0.05, **p < 0.01, ***p < 0.001, ns – not significant (Student's t‐test). H) The TCGA dataset analysis shows the correlation between NKX2‐1 expression and CXCL1, CXCL2, CXCL3, and CXCL5 mRNA expression levels.

Figure 5

NKX2‐1 suppresses the chromatin accessibility at the promoter regions of CXC chemokine genes. A) Heatmap representation of the chromatin accessibility at multiple genes' transcription start sites (TSS) and 10 kb upstream and downstream regions. B) Annotation for ATAC‐seq peak localization showing the distribution of sequenced reads across the indicated functional genomic elements. The peaks located 1 Kb upstream and 100 bp downstream of the TSS were defined as promoter‐TSS. C) The distribution of open chromatin areas across the chromosomes is shown as the ratio of signals from shNKX2‐1/H1975 versus shCtrl/H1975. D) GO‐BP analysis of genes with open (top panel) and closed chromatin structure (bottom panel) at the promoter regions resulting from NKX2‐1 knockdown. E) Integrative Genomics Viewer (IGV) representation of ATAC‐seq signals at the promoter regions of CXC chemokine genes in shCtrl/H1975 and shNKX2‐1/H1975 samples. The red line denotes a given range across the 2 samples. F) ChIP‐qPCR analysis showing the binding of NKX2‐1 at the promoter regions of CXCL1, CXCL2, and CXCL5 genes in shNKX2‐1/H1975 cells and H1975/AZDR overexpressing NKX2‐1. Mean fold changes (N = 3) relative to input are shown with SD error bars *p < 0.05, **p < 0.01, ***p < 0.001 (Student's t‐test). G) The most enriched canonical NKX2‐1 binding motifs across ATAC‐seq reads. H) Schematic illustration showing the potential mechanism of NKX2‐1 regulating the expression of CXC chemokines by altering chromatin structure.

Figure 6

Neutrophils attracted by NKX2‐1‐low tumors exhibit cancer‐promoting properties. A) The schematic illustration demonstrating the experimental procedure for chemotaxis assay using HL‐60 and conditioned medium from LUAD cell lines. B) Chemotaxis assay showing the migratory capacity of HL‐60 cells in response to the conditioned media (CM) derived from H1975 (left panel) and HCC827 (right panel) subjected to NKX2‐1 knockdown. Mean numbers of migrated cells (N = 3) are shown with SD error bars, *p < 0.05, **p < 0.01, ***p < 0.001 (Student's t‐test). C) Revigo scatter plot showing enriched GO‐BP terms among the genes upregulated in HL‐60 cells cultured with the conditioned media derived from shNKX2‐1/H1975 as compared to shCtrl/H1975 as shown by RNA‐seq analysis. D) Hierarchical clustering heatmap from RNA‐seq showing the differential gene expression (left panel) and neutrophil‐related genes (right panel) in HL‐60 cultured with the conditioned media derived from shCtrl/H1975 and shNKX2‐1/H1975 cells. E) qRT‐PCR analysis showing the expression levels of pro‐inflammatory genes and anti‐inflammatory genes in HL‐60 cultured with the conditioned media derived from shCtrl/H1975 and shNKX2‐1/H1975 cells. The mean fold changes (N = 3) relative to shCtrl are shown with SD error bars, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns – not significant (Student's t‐test). F) GO‐BP enrichment analysis of the genes upregulated in the tumor regions derived from shNKX2‐1/LL2 cells compared to shCtrl/LL2 cells from Visium in situ capturing profiling. G) GO‐BP enrichment analysis of the upregulated genes in tumors derived from shNKX2‐1/LL2 cells compared to shCtrl/LL2 from Visium in situ capturing profiling. The genes upregulated with ‐log2 (fold‐change) of more than 1 were defined as infiltration‐positive, while the genes with below‐threshold expression were defined as infiltration‐negative. H) Violin plots showing the expression levels of the most upregulated genes annotated by GO‐BP terms “regulation of cell communication” and “regulation of cell population proliferation” in shNKX2‐1 and shCtrl infiltrated neutrophils. I) Visium in situ capturing representative images showing the neutrophil‐positive regions and overlapping regions of the expression of genes annotated by “regulation of cell communication” and “regulation of cell population proliferation” GO‐BP terms.

Figure 7

In vivo targeting of CXCLs/CXCR2 signaling with SB225002 reduces tumor growth and neutrophil infiltration. A) Subcutaneous mouse model. Top panel – schematic representation of the experimental design, bottom panel – tumor growth curve for 18 days following subcutaneous injection. Administration of SB225002 and vehicle control in the shNKX2‐1/LL2 tumor. Data are shown as means with SD error bars, N = 4, ***p < 0.001 (Student's t‐test). B) Orthotopic mouse model. Top panel – schematic representation of the experimental design, bottom panel – bioluminescent signal visualization (top), photographs and GFP fluorescence images of tumors (bottom) derived from shCtrl/LL2, shNKX2‐1/LL2, and shNKX2‐1/LL2 injected with or without (vehicle) SB225002 administration. C and D) Unsupervised clustering of scRNA‐seq data showing subdivision of infiltrated neutrophils into six distinct subpopulations (C) and in separate samples of shCtrl+vehicle, shNKX2‐1+vehicle, and shNKX2‐1+SB225002 cells (D). E) Pseudo‐time visualization showing the single‐cell lineage order based on the gene expression profile from scRNA‐seq analysis. F) Graphical representation of cell percentages according to neutrophil clusters with more than one read for the corresponding gene. G) Violin plot showing the gene expression score of upregulated cancer‐promoting genes, expressed in shCtrl+vehicle, shNKX2‐1+vehicle, and shNKX2‐1+SB225002 infiltrated neutrophils. H) Ridgeline plot visualization of the expression distributions of the indicated differentially expressed cancer‐promoting genes from cluster 3 across 3 samples (shCtrl+vehicle, shNKX2‐1+vehicle, and shNKX2‐1+SB225002 cells). I) The Venn diagram showing the numbers of NKX2‐1‐dependent genes involved in CXCR2‐mediated signaling. The left circle denotes the upregulated genes in shNKX2‐1+vehicle versus shCtrl+vehicle, and the right circle represents the downregulated genes in shNKX2‐1+SB225002 versus shNKX2‐1+vehicle samples. The overlapping genes are NKX2‐1‐regulated genes in the context of CXCR2‐mediated signaling. J) GO‐BP enrichment analysis of the indicated groups of 101, 14, and 263 genes shown in (I).

Figure 8

Networks engaged by NKX2‐1‐low tumor‐activated neutrophils in vitro and in vivo models. A) Pie chart representation showing the enriched GO‐BP terms with the upregulated gene percentage obtained from scRNA‐seq analysis of NKX2‐1‐low sample compared to the control group and RNA‐seq from HL‐60 cultured with the conditioned medium from shNKX2‐1/H1975 cells compared with shCtrl. B) Graphical representation of the enriched GO‐BP analysis from scRNA‐seq analysis of NKX2‐1‐low tumors compared to the control group and RNA‐seq from HL‐60 cultured with the conditioned medium from shNKX2‐1/H1975 cells compared with shCtrl. D and C) Network analysis showing different genes involved in the enriched GO‐BP analysis from scRNA‐seq analysis of NKX2‐1‐low sample compared to the control group (C) and RNA‐seq from HL‐60 cultured with conditioned medium from shNKX2‐1/H1975 cells compared with shCtrl (D).