Tribbles Pseudokinase 3 Converts Sorafenib Therapy to Neutrophil-Mediated Lung Metastasis in Hepatocellular Carcinoma

Abstract

Rapid development of resistance to sorafenib and subsequent hyperprogression in patients with advanced hepatocellular carcinoma (HCC) pose significant challenges, with the underlying mechanisms still largely unknown. Herein, sorafenib-induced TRIB3 is identified as a liver-specific determinant driving secondary resistance to sorafenib by facilitating the accumulation of protumorigenic neutrophils within tumors. Mechanistically, TRIB3, triggered by the sorafenib-elicited ROS-ER stress axis, operates in an NF-κB-dependent manner to upregulate CXCR1/2 ligands, subsequently promoting neutrophil recruitment into tumors. These enriched neutrophils enhance epithelial-mesenchymal transition processes in malignant cells through the oncostatin M-STAT3 pathway, thereby repurposing the therapeutic efficacy of sorafenib away from anti-angiogenesis and toward lung metastasis. Clinically, elevated TRIB3 expression indicates inferior survival and unfavorable clinical efficacy of sorafenib in HCC patients. Correspondingly, strategies that either inhibiting TRIB3 upregulation or blocking its downstream signaling successfully augment the therapeutic efficacy of sorafenib and prevent sorafenib-induced hyperprogression in vivo. The study thus identifies a pivotal mechanism of sorafenib resistance in HCC, centered on the TRIB3-mediated recruitment of protumorigenic neutrophils and subsequent disease hyperprogression.

Keywords: TRIB3; hepatocellular carcinoma; lung metastasis; neutrophil; sorafenib resistance.

Figure 1

Inducing TRIB3 by sorafenib promotes therapeutic resistance and metastasis. A) Venn diagram showing the overlap between liver‐specific genes (The Human Tissue Specific Proteome section of the Human Protein Atlas), genes associated with an unfavorable prognosis in HCC (The Liver Cancer Proteome section of the Human Protein Atlas), and genes significantly upregulated by sorafenib treatment in HCC (GEO: GSE186280). B,C) Comparison of TRIB3 levels between recurrent tumor tissue and matched primary resected sample from HCC patients who received sorafenib therapy before secondary resection (B, n = 15) or those who did not receive additional treatment before secondary resection (C, n = 15). D,E) Human hepatoma cells were treated with DMSO or sorafenib for 48 h (each n = 5). The expression of TRIB3 was determined by real‐time PCR (D) and immunoblotting (E). F) Huh‐7 cells were treated with DMSO or sorafenib for 48 h. The levels of TRIB3 expression were analyzed by immunofluorescence (n = 3). Scale bar, 50 µm. G) C57BL/6 mice, either bearing orthotopic xenografts of Hepa1‐6 cells or with spontaneous hepatomas, were treated with sorafenib or PBS as described in Figure S1C (Supporting Information) (each n = 8 for orthotopic xenograft; each n = 6 for spontaneous hepatoma). Trib3 expression in hepatoma was determined by real‐time PCR. H–K) Wild‐type or Trib3‐deficient Hepa1‐6 hepatoma‐bearing mice were injected with PBS or sorafenib as described (H) (each n = 5). Tumor growth monitored by bioluminescence imaging (I), tumor volume (J), and lung metastasis (K) of hepatoma were analyzed. Scale bar, 625 µm (K). Data represent mean ± SEM of three independent experiments. ^* p < 0.05, ^*** p < 0.001, Student's t test (B–D, and G), two‐way ANOVA with Tukey's post test (I–K).

Figure 2

TRIB3 is triggered by sorafenib‐elicited ROS‐ER stress axis in HCC. A) Functional annotation of significantly upregulated genes in sorafenib‐treated Hep G2 cells (GSE186280) were analyzed by Metascape. The top 5 enrichment GO terms are listed. B) Gene Set Enrichment Analysis (GSEA) of endoplasmic reticulum unfolded protein response in sorafenib‐treated versus untreated Hep G2 cells (GSE186280). C) Huh‐7 cells were treated with DMSO or sorafenib for indicated time. Activation of indicated pathways was analyzed by immunoblotting (n = 3). D,E) Huh‐7 cells were treated with DMSO or sorafenib in the absence (—) or presence of PERK inhibitor (PERKⁱ) for 48 h (n = 5), the expression of TRIB3 were determined by real‐time PCR (D) and immunoblotting (E), respectively. F) Dynamic assessment of reactive oxygen species (ROS) generation induced by sorafenib. Huh‐7 cells were stained with the fluorescent ROS probe DCFH‐DA, and were subsequently treated with sorafenib (10 µm). Thereafter, the DCF fluorescence intensity in these cells was analyzed by flow cytometry (n = 3). G,H) Huh‐7 cells were treated with DMSO or sorafenib in the absence or presence of the ROS scavenger NAC (n = 3). Thereafter, the DCF fluorescence intensity was analyzed by flow cytometry (48 h, G). PERK activation (1 h) and its downstream effector molecules, TRIB3, eIF2α, ATF4, and CHOP (48 h) were determined by immunoblotting (H). I–L) Hepa1‐6 hepatoma‐bearing mice were left untreated or treated with PERK inhibitor or NAC in the presence of sorafenib for 25 days as described in (I) (each n = 5). Protein levels of TRIB3 and PERK pathway (J), tumor volume (K) and lung metastasis (L) of hepatoma were analyzed. Data represent mean ± SEM of three independent experiments (D, K, and L). ^** p < 0.01, ^*** p < 0.001, one‐way ANOVA with Tukey's post test (D, K, and L).

Figure 3

In vivo milieus dictate TRIB3's protumorigenic properties. A–C) Trib3^WT, Trib3^KO, or Trib3^OE Hepa1‐6 cells were treated with DMSO or sorafenib. Cell proliferation was measured by CCK8 assay (A). Cell apoptosis (48 h) was measured by flow cytometry (B). Migration of the cells (C, 48 h) was determined (n = 5). Scale bar, 200 µm. D) Trib3^WT or Trib3^OE Hepa1‐6 cells were treated with DMSO or sorafenib. Proteins of EMT genes were determined by immunoblotting (n = 3). E–G) Trib3^WT or Trib3^OE Hepa1‐6 hepatoma‐bearing mice were injected with PBS or sorafenib as described in Figure S3B (Supporting Information) (each n = 5). After 25 days of treatment, the proliferation of hepatoma cells (E, scale bar, 50 µm), morphology and area of microvessels in tumor (G, scale bar, 50 µm) were analyzed by immunohistochemistry using αKi67 and αCD31 antibody, respectively. Proteins of EMT genes in mouse tumor tissues were determined by immunoblotting (F). H) Correlation between TRIB3 expression and microvessel area in tumors from 32 HCC patients were analyzed by immunohistochemistry using αTRIB3 and αCD34 antibody. Integrated TRIB3 density and microvessel area was determined by IMMAGE J software. Data represent mean ± SEM of three independent experiments (A–C, E, and G). ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, two‐way ANOVA with Tukey's post test (A, C, E, and G), one‐way ANOVA with Tukey's post test (B). P value and R value were calculated based on the analysis of Pearson's correlation (H).

Figure 4

Neutrophils are essential for TRIB3‐driven pro‐tumorigenesis. A) Effects of Trib3 overexpression on the infiltration of B cell (B220⁺), T helper cell (CD4⁺), cytotoxic T cell (CD8⁺), neutrophil (Gr1⁺), NK cell (NKp46⁺), and macrophage (F4/80⁺) in mouse hepatoma (n = 6, scale bar, 50 µm). B) Association of TRIB3 expression and infiltration of CD15⁺ neutrophils in tumor tissues from HCC patients (n = 32). Scale bar, 50 µm. C–F) Trib3^WT or Trib3^OE Hepa1‐6 hepatoma‐bearing mice were treated with αGr‐1 or isotype antibody as described in Figure S4D (Supporting Information) (each n = 5). Angiogenesis (C), proteins of EMT genes (D), tumor volume (E), and lung metastasis (F) of hepatoma were analyzed. G,H) Hepa1‐6 hepatoma‐bearing mice were injected with sorafenib or PBS in the presence or absence of αGr‐1 antibody as described in Figure S4F (Supporting Information) (each n = 5). Tumor growth monitored by bioluminescence imaging (G) and lung metastasis of hepatoma were analyzed (H). Data represent mean ± SEM of three independent experiments (A–C, and E–H). ^* p < 0.05, ^*** p < 0.001, Student's t test (A,B), two‐way ANOVA with Tukey's post test (C, and E–H).

Figure 5

Elevated TRIB3 attracts neutrophils through boosting NF‐κB‐CXCL signaling. A) Expression of CXCR1/2 ligands in wild‐type Huh‐7 cells (TRIB3^WT) or TRIB3‐knockdown Huh‐7 cells (TRIB3^SH) treated with DMSO or sorafenib for 48 h (n = 5). B) Neutrophils were left untreated (Med) or were treated with a conditioned medium (CM‐treated) from sorafenib‐treated wild‐type (TRIB3^WT) or TRIB3‐knockdown (TRIB3^SH) Huh‐7 cells. Migration of neutrophils was determined (n = 4). C) Wild‐type or Trib3‐deficient Hepa1‐6 hepatoma‐bearing mice were injected with PBS or sorafenib as described in (Figure 1H) (each n = 5). Infiltration of Gr1⁺ neutrophils in tumors was analyzed by immunohistochemistry. Scale bar, 50 µm. D) Hepa1‐6 hepatoma‐bearing mice were injected with PBS or sorafenib in the absence or presence of CXCR1/2 inhibitor (Reparixin) as described in Figure S5D (Supporting Information) (each n = 5). Infiltration of Gr1⁺ neutrophils in tumors was analyzed. E) Protein levels of TRIB3 and activation of NF‐κB pathway in sorafenib‐treated Trib3^WT, Trib3^OE, or Trib3^KO Hepa1‐6 cells were determined (n = 3). F–H) Wild‐type (Trib3^WT or Rela^NC), or Trib3‐overexpressing (Trib3^OE), or p65‐knockdown (Rela^SH), or p65‐knockdown plus Trib3‐overexpressing Hepa1‐6 cells were inoculated in liver of C57BL/6 mice as described in Figure S5H (Supporting Information) (each n = 5). Infiltration of Gr1⁺ neutrophils (F), area of microvessels (G), and proteins of EMT genes (H) were determined. Data represent mean ± SEM of three independent experiments (A–D, F, and G). ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, one‐way ANOVA with Tukey's post test (A, and B), two‐way ANOVA with Tukey's post test (C, D, F, and G).

Figure 6

Neutrophils suppress therapeutic efficacy of sorafenib via OSM/STAT3 axis. A–C) Huh‐7 cells were treated with DMSO or sorafenib in the absence or presence of tumor‐associated neutrophils (TAN) in the upper chamber of a transwell plate (n = 5). Cell proliferation (96 h, A), proteins of EMT genes (48 h, B), and expression of angiogenesis‐related genes (4 h, C) were determined. D) Huh‐7 cells were treated with DMSO or sorafenib in the absence or presence of conditioned medium from tumor‐associated neutrophils (N‐CM) for 2 h. Activation of STAT3 pathway was determined by immunoblotting (n = 3). E) Uniform Manifold Approximation and Projection (UMAP) plots showing expression of OSM indifferent cell types in HCC tumors. Each cluster is color‐coded according to cell type. Cluster annotations are indicated in the Figure (http://meta‐cancer.cn:3838/scPLC/). F) OSM concentration in culture supernatant of indicated immune cells isolated from HCC tumors was determined by enzyme‐linked immunosorbent assay (ELISA) (n = 5). G) Huh‐7 cells were left untreated or pre‐treated with isotype or OSM neutralizing antibody (α‐OSM, 3 µg mL⁻¹) in the presence of N‐CM. Activation of STAT3 pathway (2 h) and proteins of EMT genes (48 h) were determined by immunoblotting (n = 3). H–K) Hepa1‐6 hepatoma‐bearing mice were injected with PBS or sorafenib in the absence or presence of OSM neutralizing antibody (α‐OSM) as described in (H) (each n = 5). Tumor volume (I), area of microvessels (J), and lung metastasis (K) of hepatoma were analyzed. Data represent mean ± SEM of three independent experiments (A, F, and I–K). ^* p < 0.05, ^*** p < 0.001, two‐way ANOVA with Tukey's post test (A, and I–K), one‐way ANOVA with Tukey's post test (F).