Invasive FoxM1 phosphorylated by PLK1 induces the polarization of tumor-associated macrophages to promote immune escape and metastasis, amplified by IFITM1

Abstract

Background: Understanding the mechanism behind immune cell plasticity in cancer metastasis is crucial for identifying key regulators. Previously we found that mitotic factors regulate epithelial-mesenchymal transition, but how these factors convert to metastatic players in the tumor microenvironment (TME) is not fully understood.

Methods: The clinical importance of mitotic factors was analyzed by heatmap analysis, a KM plot, and immunohistochemistry in lung adenocarcinoma (LUAD) patients. Immunoprecipitation, LC-MS/MS, kinase assay, and site-directed mutagenesis were performed for the interaction and phosphorylation. A tail-vein injection mouse model, Transwell-based 3D culture, microarray analysis, coculture with monocytes, and chromatin immunoprecipitation assays were used to elucidate the function of phosphorylated FoxM1 in metastasis of TME.

Results: The phosphorylated FoxM1 at Ser25 by PLK1 acquires the reprogramming ability to stimulate the invasive traits in cancer and influence immune cell plasticity. This invasive form of p-FoxM1 upregulates the expression of IL1A/1B, VEGFA, and IL6 by direct activation, recruiting monocytes and promoting the polarization of M2d-like tumor-associated macrophages (TAMs). Upregulation of PD-L1 in LUAD having phosphomimetic FoxM1 facilitates immune evasion. In invasive LUAD with phosphomimetic FoxM1, IFITM1 is the most highly expressed through the activation of the STING-TBK1-IRF3 signaling, which enhances FoxM1-mediated signaling. Clinically, higher expression of FOXM1, PLK1, and IFITM1 is inversely correlated with the survival rate of advanced LUAD patients, providing a promising therapeutic strategy for the treatment of LUAD.

Conclusion: FoxM1-based therapy would be a potential therapeutic strategy for LUAD to reduce TAM polarization, immune escape, and metastasis, since FoxM1 functions as a genetic reprogramming factor reinforcing LUAD malignancy in the TME.

Keywords: FoxM1; Invasiveness; PLK1; Phosphorylation; Tumor-associated macrophages.

Fig. 1

Concurrent upregulation of FoxM1 and PLK1 is correlated with poor survival of LUAD patients. a Analysis of Spearman’s coefficient for the correlations between cell cycle-regulatory factors including PLK1, FOXM1, CCNA1, CCNB1, CCND1, CCNE1, CDK1, CDK2, PCNA, and MKI67. b-d The overall survival (OS) of patients with non-small lung cancer (n = 1885) (b), lung adenocarcinoma (LUAD) (n = 703) (c), and stages 3–4 of LUAD (n = 137) (d) were analyzed according to PLK1 and FOXM1 expression levels using KM PLOTTER. High (Hi) and low (Lo) were generated by separating patients according to expression at the median cut-off. e Using human lung tissues from LUAD patients and normal individual, immunohistochemistry analysis was displayed with anti-FoxM1 (Green) and anti-p-Serinie (Red) antibodies. The relative intensity of cells that exhibited positive FoxM1 was analyzed and plotted. WD, well differentiated (grade 1); MD, moderately differentiated (grade 2); PD, poorly differentiated (Grade 3), n > 5000. *p < 0.05; **p < 0.01; ***p < 0.001. Data are presented as mean ± SD. f and g A549, NCI-H358 (H358), and NCI-H460 (H460) NSCLC cells treated with 5 ng/mL of TGF-β for 48 h. f Immunoblotting was performed to measure the expression and phosphorylation of PLK1 using specific antibodies for FoxM1, PLK1, p-PLK1^T210, p-Smad2^S465/S467, Smad2, E-cadherin, N-cadherin, vimentin, SNAI1, SNAI2, and β-actin in A549, NCI-H358 (H358), and NCI-H460 (H460) cells (left panel). The relative band intensities for FoxM1, p-PLK1^T210, PLK1, p-Smad2.^S465/S467, Smad2, E-cadherin, N-cadherin, vimentin, SNAI1, and SNAI2 were quantified using LI-COR Odyssey software (right panel). g qRT-PCR was performed for CDH1, CDH2, FOXM1, PLK1, SNAI1, SNAI2, ZEB1, and TWIST expression in A549 (left panel), NCI-H358 (middle panel), and NCI-H460 (right panel) cells. Data are presented as mean ± SD of at least three independent experiments (significantly different from the experimental control). *p < 0.05; **p < 0.01; ***p < 0.001

Fig. 2

TGF-β-treated EMT results in phosphorylation of FoxM1 by PLK1 by direct interaction. a, b A549 (a) and NCI-H460 (b) were treated with TGF-β (5 ng/mL) for 48 h. Immunoprecipitation of cell lysates was performed with normal IgG or anti-PLK1 antibody and then immunoblotting was performed with anti-FoxM1 antibody. c An in vitro kinase assay was performed with an active version of PLK1 with T210D (PLK1-TD), radioactive ATP, and purified GST-FoxM1. GST-tagged TCTP was used as the positive control. d In LC–MS/MS analysis, possible phosphorylation residues of FoxM1 by PLK1 were newly detected at the S25, S360, S361, and S393 sites. e, Purified GST-tagged wild-type, S25A, S361A, S715A, and S25/S361/S715A (AAA) FoxM1 mutants were used for a PLK1 kinase assay with radioactive ATP. f, g Phosphorylation of FoxM1 in A549 (f) and NCI-H358 (g) cells treated with TGF-β for 48 h. Treatment with calf intestinal alkaline phosphatase (CIP) reduced the phosphorylation of FoxM1 and PLK1 in TGF-β-induced EMT. Immunoprecipitation was performed with anti-normal IgG (Fig. S4d-e) or anti-FoxM1 antibody, and then immunoblotting was conducted with anti-p-Serine antibodies. Immunoblotting was performed for FoxM1, PLK1, p-PLK1^T210, TCTP, and p-TCTP^S46 using specific antibodies. TCTP was used as a positive control of the PLK1 substrate. Data are presented as mean ± SD of at least three independent experiments (significantly different from the experimental control). *p < 0.05; **p < 0.01; ***p < 0.001

Fig. 3

Phosphorylation of FoxM1 at Ser25 facilitates cancer cell migration and invasiveness but not cell proliferation. RFP-tagged wild-type (WT) FoxM1 and S25A, S25E, S361A, S361E, S715A, and S715E mutants were expressed in A549 cells. A549 cells were treated with doxycycline to express RFP-tagged FoxM1. a Immunoblotting was performed using specific antibodies for RFP, N-cadherin (N-Cad), E-cadherin (E-Cad), vimentin, and β-actin (left panel). The band intensity values were quantified using LI-COR Odyssey software, normalized, and plotted (right panel). b qRT-PCR was performed for FOXM1, CDH1, CDH2, and VIM in A549 cells expressing wild-type or mutants FoxM1. *p < 0.05; **p < 0.01; ***p < 0.001; (n = 3). Data are presented as mean ± SD. c Cell proliferation assay was performed (n = 3). Data are presented as mean ± SD of at least three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001 compared with experimental control; ^#p < 0.05 compared with indicated groups of cells. d Cells expressing wild-type or mutants of FoxM1 were subjected to a Transwell migration assay. As a positive control for migration, cells were treated with TGF-β. Three days after seeding, the cells on the bottom surface were stained with 0.05% crystal violet dye. Images of the Transwell cell migration assay were collected and analyzed with an Odyssey infrared imaging system and plotted. e An invasion assay was performed using A549 cells expressing wild-type or mutants of FoxM1. Seven days after seeding, the cells that invaded the bottom surface were stained with 0.05% crystal violet dye, and the relative absorbance was plotted. Data are presented as mean ± SD of at least three independent experiments (significantly different from the experimental control). *p < 0.05; **p < 0.01; ***p < 0.001 compared with experimental control. ^#p < 0.05; ^##p < 0.01; ^###p < 0.001 compared with S25A FoxM1

Fig. 4

Phosphorylation of FoxM1 at Ser25 facilitates metastasis of NSCLC in a tail-vein injection model. A549 cells expressing RFP-tagged WT, S25A, and S25E of FoxM1 were injected intravenously into the tail-veins of four-week-old BALB/c nude mice, and the tumorigenic and metastatic properties were evaluated after 15 weeks. a Representative lung tumor from the mouse model (left panel). The number of metastatic lung tumors was counted and plotted. (n = 4 or 5) (right panel). Data are presented as mean ± SD. b, c Representative H&E staining (b, left panel) and Ki-67 staining (c, left panel) were performed using lung tissue from the mice.The relative density of H&E staining (b, right panel) and Ki-67-positive cells (c, right panel) was analyzed and plotted. *p < 0.05; **p < 0.01; ***p < 0.001. Data are presented as mean ± SD. d Immunoblotting was performed using lung tissue lysates from each mouse model. FoxM1, RFP, E-cadherin, N-cadherin, vimentin, SNAI1, PD-L1, and β-actin were detected using specific antibodies (left panel). The band intensity values were quantified using LI-COR Odyssey software, normalized, and plotted (right panel). e qRT-PCR was performed for CDH1, CDH2, SNAI1, VIM, and CD274 using lung tissue lysates from each mouse model. The relative mRNA expression was plotted. *p < 0.05; **p < 0.01; ***p < 0.001 compared with experimental control. ^#p < 0.05; ^###p < 0.001 compared with S25A FoxM1

Fig. 5

Interferon signaling is mainly activated in invasive cells having phosphorylated FoxM1 at Ser25. a, b Using non-invasive and invasive A549 cells expressing S25E of FoxM1, transcriptome profiles were analyzed by microarray. The transcriptome data were clustered by gene probes with fold change > 1.5 in cells expressing S25E FoxM1. Gene Ontology (GO) analysis of transcriptome profiles was performed as biological processes (a) and KEGG pathways (b). KEGG pathways were analyzed, and the signaling pathways had higher gene numbers in cells expressing invasive S25E than in those expressing mock. c Transcriptome comparison between gene profiles of invasive and non-invasive A549 cells expressing WT, S25A, and S25E. The MORPHEUS program was used to visualize the expression levels of genes related to the EMT, metastasis, interferon-stimulated genes, cytokines and chemokines, JAK-STAT, MAPK, NF-kB, and TLR signaling in non-invasive and invasive A549 cells expressing WT, S25A, and S25E of FoxM1. d Relative gene expression profile of the top 12 genes in invasive and non-invasive A549 cells expressing WT, S25E, and S25A FoxM1. e, f qRT-PCR was performed for the top five genes IFITM1, XAF1, IF44L, MX1, and IL1A in invasive A549 cells expressing FoxM1 (e) and total A549 cells expressing FoxM1 (f). *p < 0.05; **p < 0.01; ***p < 0.001; (n = 3). Data are presented as mean ± SD

Fig. 6

p-FoxM1.^S25 functions in recruitment of macrophages and triggers polarization of M2-like TAM. a, b Monocyte THP-1 cells were co-cultured with A549 cells expressing mock, WT, S25A, and S25E FoxM1 for 48 h. In THP-1 cells, qRT-PCR was performed for markers of M1 (INOS, IL12B), M2 (IL10, CD163, CD206), and TAM (TGFB1, VEGFA) (a). In A549 cells, qRT-PCR was performed for IL4, IL6, IL10, VEGFA, and CD274 (b). *p < 0.05; **p < 0.01; ***p < 0.001; (n = 3). Data are presented as mean ± SD. c, THP-1 cells were co-cultured with A549 cells expressing mock, WT, S25A, and S25E FoxM1. The secreted levels of TGF-β1 and VEGFA from THP-1 cells co-cultured with A549 cells were detected using ELISA. d Monocyte THP-1 cells were co-cultured with A549 cells expressing mock or S25E depleted FoxM1 using shRNA for 48 h. Using THP-1 cells, qRT-PCR was performed for CD163, CD206, and VEGFA. *p < 0.05; **p < 0.01; ***p < 0.001; (n = 3). e, f Representative CD68 (pan-macrophage marker) staining (e, upper panel) and CD163 (TAM marker) staining (e, lower panel) were performed using lung tissue from mice. The relative density of CD68 staining (f, left panel) and CD163 staining (f, right panel) was analyzed and plotted. *p < 0.05; **p < 0.01; ***p < 0.001. Data are presented as mean ± SD. g Immunoblotting was performed using lung tissue lysates from each mouse model. FoxM1, CD68, CD163 and β-actin were detected using specific antibodies (upper panel). The relative protein intensities were analyzed and plotted (lower panel). h The viability of A549 cells expressing mock, WT, S25A, and S25E FoxM1 was measured when the cells were co-cultured with monocyte THP-1 cells. The ratio between A549 cells and THP-1 cells was 1:0, 1:2, 1:4, and 1:6, as indicated. i Monocyte THP-1 cells were co-cultured with A549 cells expressing mock, WT, S25A, and S25E FoxM1. qRT-PCR was performed for CD279 mRNA level in THP-1 cells. Data are presented as mean ± SD of at least three independent experiments (significantly different from the experimental control). **p < 0.01; ***p < 0.001; (n = 3)

Fig. 7

p-FoxM1^S25 translocates to the nucleus and activates genes for monocyte recruitment, TAM polarization, immune escape, and angiogenesis by direct activation. a, b A549^S25E were treated with 1 μM trametinib, an inhibitor of MEK, for 48 h. a qRT-PCR was performed for interferon-stimulated genes (IFITM1, IF44L, and MX1), IL1A, IL1B, IL6, VEGFA, CXCL1, and FOXM1 in A549^S25E cells. b Immunoblot analyses were performed using specific antibodies for FoxM1, p-Erk1/2, Erk1/2, c-Fos, c-Jun, IFITM1, IL1A, IL6, and β-actin. c, d A549^S25E cells were treated with 15 μM ruxolitinib, a JAK inhibitor, for 48 h. c qRT-PCR was performed for IFITM1, IF44L, MX1, IL1A, IL1B, IL6, VEGFA, CXCL1, and FOXM1 in A549^S25E cells. d Immunoblot analyses were performed using anti-FoxM1, anti-STAT1, anti-p-STAT1, anti-IFITM1, anti-IL1A, anti-IL6, and anti-β-actin. e Immunofluorescence was performed with A549 cells expressing WT, S25A, or S25E mutant of FoxM1. FoxM1 (green), RFP (red), and DNA (DAPI, blue) was displayed. Scale bar, 5 μm. f The quantification of the population of cells in the cytoplasm, nucleus, and both is presented on the left. The percentage of cells that exhibited positive RFP (red) staining was assessed with the following categories (N > C, RFP staining predominantly in the nucleus; N = C, similar RFF levels in both the nucleus and cytoplasm; N < C, RFP staining mainly in the cytoplasm). The population of RFP-positive cells specifically in the nucleus was plotted (right). n > 800. g-i ChIP assays for FoxM1 binding to the promoters of FOS (g, left), STAT1 (g, right), IL6 (h, left), VEGFA (h, right), and IFITM1 (i). Assays were performed on chromatin fragments using antibody to FoxM1 and normalized to pre-immune normal IgG. Immunoprecipitated fractions were assayed by qRT-PCR for binding the promoters of FOS, STAT1, IL6, VEGFA, and IFITM1. The qRT-PCR products were visualized in agarose-gel. j qRT-PCR was performed for STING, TBK1, and IRF3 in A549^S25E cells. k Immunoblot analyses were performed using specific antibodies for STING, p-TBK1, TBK1, IRF3 and anti-β-actin. l ChIP assays were performed for FoxM1 binding to the promoters of STING in A549^S25E cells. m ChIP assays for IRF3 binding to the promoters of IFITM1 in A549.^S25E cells. Assays were performed on chromatin fragments using antibody to IRF3 and normalized to pre-immune normal IgG. Immunoprecipitated fractions were assayed by qRT-PCR for binding the promoters of IFITM1. Data are presented as mean ± SD of three independent experiments (significantly different from the experimental control). *p < 0.05; **p < 0.01; ***p < 0.001; (n = 3)

Fig. 8

IFITM1 functions as a regulator of metastasis and polarization of M2d-TAM in p-FoxM1-induced metastasis. a, b IFITM1 was depleted in A549^S25E cells using human IFITM1 shRNA for 48 h. a qRT-PCR was performed for IFITM1, FOXM1, CDH2, PLK1, TGFB1, IL6, and VEGFA. b Immunoblot analyses were performed using anti-IFITM1, anti-RFP, anti-E-cadherin, anti-p-Smad2, anti-Smad2/3, anti-SNAI1, anti-SNAI2, and anti-GAPDH antibodies. c, d 5 ng/mL TGF-β was applied to A549 cells depleting IFITM1 with shRNA. c Immunoblotting was performed using anti-IFITM1, anti-FoxM1, anti-E-cadherin, anti-N-cadherin, anti-p-Smad2, anti-Smad2/3, anti-SNAI2, anti-vimentin, and anti-GAPDH antibodies. d qRT-PCR was performed for IFITM1, FOXM1, CDH2, PLK1, TGFB1, IL6, and VEGFA. e IFITM1 was depleted in A549^S25E cells using human IFITM1 shRNA for 48 h. Cells were subjected to a Transwell migration assay. Three days after seeding, the cells on the bottom surface were stained with 0.05% crystal violet dye. Images of the Transwell cell migration assay were collected and analyzed with an Odyssey infrared imaging system and plotted. f An invasion assay was performed using A549 cells expressing wild-type or mutants of FoxM1. Seven days after seeding, the cells that invaded the bottom surface were stained with 0.05% crystal violet dye, and the relative absorbance was plotted (n = 3). g THP-1 cells were co-cultured with A549^S25E cells depleting IFITM1 for 48 h. qRT-PCR was performed for CD163, CD206, and VEGFA. h The overall survival (OS) of all LUAD patients (n = 719) (h, upper) and stage 3 LUAD patients (n = 24) (h, lower) was analyzed according to IFITM1 expression level using KM PLOTTER. High (Hi) and low (Lo) were generated based on the expression at the median cut-off. i The OS of all LUAD patients (n = 703) (i, left) and stage 3–4 LUAD patients (n = 137) (i, right) was analyzed according to IFITM1, FOXM1, and PLK1 expression levels using KM PLOTTER. High (Hi) and low (Lo) were generated based on the expression at the median cut-off. Data are presented as mean ± SD of three independent experiments (significantly different from the experimental control). *p < 0.05; ***p < 0.001; (n = 3)

Fig. 9

A plausible model of p-FoxM1^S25-based monocyte recruitment, TAM polarization, angiogenesis, immune escape, and metastasis. In the cytoplasm, PLK1 phosphorylates FoxM1 at S25 in TGF-β-induced EMT, which triggers the nuclear translocation of p-FoxM1. p-FoxM1^S25 directly activates STING, FOS, STAT1, IL1A, IL1B, IL6, VEGFA, CD274, and SNAI1 by direct binding to their promoters in the nucleus. Additionally, activated STAT1 and AP-1 (A complex of c-Fos/c-Jun) signaling facilitates the expression of IFITM1, CXCL1, IL1A, IL1B, IL6, VEGFA, CD274, and SNAI1. Upregulated IL1A, IL1B, CXCL1, and VEGFA trigger the recruitment of monocytes. IL6 induces TAM polarization. IFITM1 amplifies signaling through the upregulation of FoxM1. TGF-β and VEGFA secreted by TAM strengthen TGF-β-induced EMT of LUAD and angiogenesis in the TME, respectively. Expressed PD-L1 in LUAD escapes the immune checkpoint by binding with PD1 of TAM. Upregulated SNAI1 regulates the EMT and metastasis in LUAD