Transforming Growth Factor-β and Axl Induce CXCL5 and Neutrophil Recruitment in Hepatocellular Carcinoma

Abstract

Transforming growth factor (TGF)-β suppresses early hepatocellular carcinoma (HCC) development but triggers pro-oncogenic abilities at later stages. Recent data suggest that the receptor tyrosine kinase Axl causes a TGF-β switch toward dedifferentiation and invasion of HCC cells. Here, we analyzed two human cellular HCC models with opposing phenotypes in response to TGF-β. Both HCC models showed reduced proliferation and clonogenic growth behavior following TGF-β stimulation, although they exhibited differences in chemosensitivity and migratory abilities, suggesting that HCC cells evade traits of anti-oncogenic TGF-β. Transcriptome profiling revealed differential regulation of the chemokine CXCL5, which positively correlated with TGF-β expression in HCC patients. The expression and secretion of CXCL5 was dependent on Axl expression, suggesting that CXCL5 is a TGF-β target gene collaborating with Axl signaling. Loss of either TGF-β or Axl signaling abrogated CXCL5-dependent attraction of neutrophils. In mice, tumor formation of transplanted HCC cells relied on CXCL5 expression. In HCC patients, high levels of Axl and CXCL5 correlated with advanced tumor stages, recruitment of neutrophils into HCC tissue, and reduced survival. Conclusion: The synergy of TGF-β and Axl induces CXCL5 secretion, causing the infiltration of neutrophils into HCC tissue. Intervention with TGF-β/Axl/CXCL5 signaling may be an effective therapeutic strategy to combat HCC progression in TGF-β-positive patients.

Figure 1

Role of TGF‐β in mesenchymal‐like HCC cells. (A) Left panel: Confocal immunofluorescence analysis of Smad2/3 in SNU449 cells treated with 2.5 ng/mL TGF‐β1 for 15 minutes. Actin stress fibers are indicated by phalloidin staining (red). Nuclei were counterstained with DAPI (blue). Right panel: Quantification of nuclear Smad2/3 signal intensity using ImageJ. (B) Immunofluorescence analysis of HLF cells as described in (A). (C) Western blot analysis of pSmad2 and total Smad2/3, with and without FCS, and interference with 10 µM Ly for 24 hours. Actin was used as loading control. (D) Migrated areas of HLF and SNU449 cells and those treated with 10 µM Ly in wound healing assays. (E) Levels of Smad4 after treatment with siNT or siSmad4. (F) Migrated areas of HLF and SNU449 cells treated with siNT or siSmad4. Data are expressed as mean ± SD. Error bars depict SD from at least three individual experiments. ***P < 0.001. Abbreviations: c, untreated control; FCS, fetal calf serum.

Figure 2

Duration‐dependent and concentration‐dependent migratory response of mesenchymal‐like HCC cells to TGF‐β treatment. (A) Migrated areas of HLF and SNU449 cells and those treated with 2.5 ng/mL TGF‐β1 for 24 hours in wound healing assays. (B) Migrated areas of SNU449 and HLF cells and those long‐term treated with 1 ng/mL TGF‐β1 (> 10 days, termed SNU449‐T and HLF‐T) in wound healing assays. (C) Western blot analysis of pSmad2 after long‐term treatment (> 10 days) of cells with different concentrations of TGF‐β1 (ng/mL). (D) Migrated areas of SNU449 cells (left panel) and HLF cells (right panel) after long‐term treatment (> 10 days) with 0.125 ng/mL and 1 ng/mL TGF‐β1. Data are expressed as mean ± SD. Error bars depict SD from at least three individual experiments. **P < 0.01, ***P < 0.001. Abbreviations: c, control; n.s., not significant.

Figure 3

Effects of long‐term TGF‐β exposure on mesenchymal‐like HCC cells. (A) Western blot analysis of pSmad2 and Snail after serum starvation and stimulation with 2.5 ng/mL TGF‐β1 (left panel), and after serum starvation and treatment with 10 µM Ly for 24 hours (right panel). Actin was used as loading control. (B) Analysis of TGF‐β mRNA expression by qPCR. (C) Proliferation kinetics of SNU449/SNU449‐T and HLF/HLF‐T cells over 72 hours. (D) Clonogenic survival assay of SNU449/SNU449‐T (upper panel) and HLF/HLF‐T cells (lower panel) after long‐term treatment with TGF‐β1. Representative images are shown. (E) Quantification of clonogenic survival assay shown in (D). (F) IC₅₀ values [µM] of sorafenib (left panel) and doxorubicin (right panel) in SNU449/SNU449‐T and HLF/HLF‐T cells. Data are expressed as mean ± SD. Error bars depict SD from at least three individual experiments. *P < 0.05, **P < 0.01, ***P < 0.001. Abbreviations: n.s., not significant; qPCR, quantitative reverse‐transcriptase polymerase chain reaction.

Figure 4

Expression profiling of genes involved in tumor‐promoting mechanisms of TGF‐β. (A) Heat map of selected target genes. (B) qPCR validation of CXCL5 expression in SNU449/SNU449‐T and HLF/HLF‐T cells. (C) Kaplan‐Meier survival curves showing higher (red) or lower levels (blue) of CXCL5 expression and corresponding overall survival in 360 HCC patients from TCGA RNAseqV2. (D) RPKM values of the selected target genes in high TGF‐β (RPKM > 20, left panel) and low TGF‐β‐expressing samples (RPKM < 5, right panel). Data are expressed as mean ± SD. Error bars depict SD from at least three individual experiments. ***P < 0.001. Abbreviations: RPKM, mean reads per kilobase per million mapped reads; qPCR, quantitative reverse‐transcriptase polymerase chain reaction; TCGA, the Cancer Genome Atlas.

Figure 5

Regulation of CXCL5 and its role in cell invasion and tumor formation. (A) CXCL5 secretion of cells was assessed by ELISA. (B) CXCL5 secretion in HLF/HLF‐T cells and those treated with 10 µM Ly for 72 hours, as well as in HLF‐Axl‐KO1 and HLF‐Axl‐KO1‐T cells. (C) Representative images of hepatospheres consisting of HLF/HLF‐T/HLF‐CXCL5 cells. (D) Quantitative analyses of respective hepatosphere invasion into collagen gels. (E) Volumes of HLF/HLF‐T/HLF‐CXCL5‐derived tumors. (F) Immunohistochemical analysis showing consecutive tumor sections of HLF‐T‐derived and HLF‐CXCL5‐derived tumors stained with anti‐CXCL5 or anti‐Flag antibody. The secondary antibody was used only as control. Error bars depict SD from three individual experiments carried out in triplicates. *P < 0.05, **P < 0.01, ***P < 0.001. Abbreviations: ELISA, enzyme‐linked immunosorbent assay; n.s., not significant.

Figure 6

Long‐term TGF‐β treatment causes neutrophil migration. (A) Quantification of neutrophil migration as assessed by under‐agarose assay. Cell Tracker green‐labeled neutrophils were exposed to supernatants of HLF, HLF‐T, and HLF‐CXCL5 cells, those treated with 10 µM Ly for 24 hours, and long‐term TGF‐β‐treated (>10 days) HLF‐Axl‐KO1‐T and HLF‐Axl‐KO2‐T cells. (B) Representative immunofluorescence images of the under‐agarose assay shown in (A). (C) Quantification of neutrophil migration after exposure to supernatants from HLF (left panel), HLF‐T (middle panel), and HLF‐CXCL5 cells (right panel) treated with siNT or siSmad4. (D) Quantification of neutrophil migration after exposure to supernatants of HLF, HLF‐T, and HLF‐CXCL5 cells and those treated with 1 µM Axl inhibitor TP0903 for 48 hours. (E) Quantification of neutrophil migration after exposure to supernatants of SNU449, SNU449‐T, and SNU449‐CXCL5. Data are expressed as mean ± SD. ***P < 0.001.

Figure 7

Correlation of CXCL5 with tumor staging and neutrophil attraction as well as TGF‐β and Axl expression in HCC patient samples. Immunohistochemical staining intensities of CXCL5 and elastase were scored with low, medium, and high protein levels, whereas TGF‐β, Axl, and Smad3L were scored with no, low, medium, and high. (A) Correlation of CXCL5 expression with tumor stages. (B) Correlation of elastase with tumor stages. (C‐F) Correlation of CXCL5 expression with elastase (C), TGF‐β1 (D), Smad3L (E), and Axl expression (F). Data are expressed as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001. Abbreviation: n.s., not significant.