Stiffness-activated hepatic stellate cells boost HCC migration via TGM2/ITGB1-mediated matrix remodeling and mitochondrial transfer

Abstract

Background & Aims

High liver stiffness correlates with poor outcomes in hepatocellular carcinoma (HCC). Prior studies focused on neoplastic cells rather than the tumor microenvironment. This study investigated how the tumor microenvironment, particularly mechanosignaling in hepatic stellate cells (HSCs), drives HCC progression.

Methods

The study examined the roles of transglutaminase 2 (TGM2) and integrin β1 (ITGB1) in HSCs under mechanical stress through proteomics, cell contraction assays, and protein interactions. It also analyzed gene expression data from 178 patients with HCC and cirrhosis to assess the impact of TGM2 and ITGB1 on overall survival (OS). Mitochondrial transfer and cell migration were observed using confocal microscopy, and the effect of TGM2/ITGB1 on extracellular matrix (ECM) remodeling and HCC recurrence was studied in a rat liver cancer model.

Results

We showed that HSC activation under matrix stiffness relied on ITGB1 mechanosignaling, with high cell-surface TGM2 expression required for ITGB1 activation. This process activated downstream CAV1, which in turn stabilized ITGB1 expression. Moreover, high co-expression of TGM2/ITGB1 (R = 0.77, p <2.2 × 10^-16) was negatively correlated with OS. Interestingly, we found massive mitochondrial transfer in hybrid co-cultures between cancer-associated fibroblasts (CAFs) and Huh7 cells by tunneling nanotubes under high stiffness (p = 0.0095), which appeared to be associated with TGM2/ITGB1. Huh7 cells with CAF-derived mitochondria exhibited enhanced migration under increased substrate stiffness (p <0.0001). Accordingly, high liver stiffness activated CAFs, leading to ECM remodeling and postoperative recurrence of HCC. TGM2/ITGB1 was essential for matrix stiffness-driven HCC recurrence following surgery.

Conclusions

This study revealed a novel mechanism by which HSCs facilitate HCC progression under matrix stiffness, which may aid in the design of therapies for the clinical treatment of HCC.

Impact and Implications

Hepatic stellate cells (HSCs) undergo differentiation into cancer-associated fibroblasts (CAFs), which constitute the primary stromal cell population within the liver tumor microenvironment and are associated with poor prognosis in patients with hepatocellular cancer (HCC). The precise mechanisms through which CAFs facilitate HCC progression remain incompletely elucidated. In this study, we emphasize the role of transglutaminase 2-medated integrin β1 in the activation of HSCs induced by increased matrix stiffness. Furthermore, we explore how enhanced matrix stiffness promotes mitochondrial transfer, thereby facilitating the migration of HCC cells. These insights may inform the development of targeted therapeutic strategies for the clinical management of HCC.

Graphical abstract

Highlights

• •High expression of cell-surface TGM2 is essential for the activation of ITGB1 in HSCs.
• •Activated HSCs transfer mitochondria to HCC cells under conditions of high stiffness.
• •Activated HSC-derived mitochondria boost Huh7 cell migration on stiff substrates.
• •TGM2/ITGB1 is essential for matrix stiffness-induced HCC recurrence.

Introduction

The majority of hepatocellular carcinomas (HCCs) arise from advanced liver fibrosis or cirrhosis, with cirrhosis being the primary risk factor.¹ Historically, studies have focused on how matrix stiffness affects hepatoma cells.^2,3 However, the tumor microenvironment, comprising immune cells, fibroblasts, endothelial cells, and extracellular matrix (ECM) components, is crucial to the tumor bulk, and a deeper understanding of its role could lead to better therapeutic outcomes.

Hepatic stellate cells (HSCs) exist in a quiescent state in healthy livers, and transform into myofibroblasts (MFs) in response to stimuli, leading to ECM buildup and liver desmoplasia.⁴ Within the tumor microenvironment, HSCs may differentiate into cancer-associated fibroblasts (CAFs) because of increased ECM stiffness. These CAFs share the contractility and migratory capacity of MFs.⁵ Some clinical research has demonstrated that CAFs correlate to poor outcomes in patients with HCC,[6], [7], [8] but their exact role in HCC progression is not fully understood. CAFs are known to aid less aggressive cancer cell invasion through matrix remodeling and leader-follower dynamics.⁹ Recent studies have demonstrated that CAFs can also promote tumor migration through mitochondrial transfer (MT).¹⁰ Tunneling nanotubes (TNTs) are increasingly recognized as the key transfer route.¹¹ These filamentous actin (F-actin)-rich TNTs form through adhesion- and ECM-driven cytoskeletal changes,^12,13 highlighting the critical role of CAFs in facilitating HCC cell migration and invasion.

Mechanically, HSC activation occurs through cell-surface receptors-mediated signaling pathway transduction (e.g. integrin cluster into focal adhesions, Rho-kinase activation, and the nuclear translocation of yes-associated protein [YAP] and snail family transcriptional repressor 1).^14,15 Integrin β1 (ITGB1) mediates cancer cell adhesion and cell-ECM interactions, while promoting TNT formation via focal adhesion-dependent actin cytoskeleton regulation.¹⁶ Despite its importance, ITGB1 targeting remains challenging because of its complexity and contradictory characteristics,¹⁷ highlighting the need to explore mechanisms regulating ITGB1. Among transglutaminase (TGM) family members, TGM2 is notably overexpressed in various tumors, promoting metastasis and reducing patient survival.[18], [19], [20] Overexpressed TGM2 interacts with ITGB1, enhancing ITGB1-fibronectin binding and inducing a premalignant phenotype in ovarian cancer stem cells.²¹ Because the TGM2/ITGB1 relationship in HSCs remains unknown, we explored its role in stiffness-activated HSCs. Caveolin-1 (CAV1) regulates ITGB1 endocytosis for fibronectin matrix turnover²² and ECM remodeling, and may cooperate with TGM2 in integrin signaling.²³ We propose that ITGB1-driven HSC activation enhances HCC cell migration via ECM remodeling and MT, with potential crosstalk between TGM2/CAV1 and ITGB1 mechanical signaling in response to microenvironment stiffness.

This study shows that matrix stiffness activates HSC via the TGM2/ITGB1/CAV1 axis, driving cell contraction and ECM remodeling. Activated HSCs transfer mitochondria to Huh7 cells through TNTs, enhancing HCC migration on stiff substrates. In vivo experiments confirmed the role of TGM2/ITGB1 in the postsurgical recurrence of HCC with cirrhosis. Overall, we explored a potential mechanism by which HSCs facilitate HCC progression under matrix stiffness, which may aid in the design of therapies for the clinical treatment of HCC.

Materials and methods

Materials and methods details are provided in the supplementary materials.

Results

Increased matrix stiffness-induced HSC activation and ECM remodeling in an ITGB1-dependent manner

When cultured on cirrhosis-mimicking polyacrylamide hydrogels (20.43 kPa) (Fig. S1A–C), both LX2 and primary rat HSCs exhibited activated morphology (elongated/multiprotrusive) vs. rounded shapes on soft gels simulating normal liver (5.95 kPa) (Fig. 1A). There was a significant increase in the expression of α-SMA, COL1, LOX, and desmin, markers of HSC activation into MFs on hard gel (Fig. 1B and Fig. S1D), whereas quiescent markers (cytoglobin for LX2 and GFAP for rat HSCs) were lower compared with those on soft substrates. Cell proliferation showed minimal differences, with a slight increase in primary rat HSCs on stiffer gels (Fig. S1E). These findings indicate that increased HSC activation markers result from a shift from quiescence to activation rather than changes in proliferation.

Fig. 1.

Increased matrix stiffness activates HSC/MF and remodels the ECM through ITGB1 signaling.

(A) Alteration in appearance and spread of LX2 (left) and primary rat HSCs (right) under different stiffness substrates. Low stiffness, 5.95 kPa; high stiffness, 20.43 kPa. Scale bar, 25 μm. (B) Expression of α-SMA, COL1, LOX, Desmin and cytoglobin or GFAP on different gels. L, low stiffness; H, high stiffness. (C) F-actin polymerization in LX2 (left) and primary rat HSCs (right) via Alexa-488 Phalloidin staining. Scale bar: 20 μm ∗∗∗∗p <0.0001 (Welch’s t test). (D) Gel contraction assay of primary HSCs from rats with different liver stiffness. Normal: HSCs from normal rats. Cirrhosis: HSCs from cirrhotic rats. Down: Quantification of gel contraction (n = 3). ∗∗p = 0.0033 (n = 3, mean ± SD, unpaired t test). (E) Predicted proteins from intersection of ECM-related proteins and differentially expressed proteins in LX2. (F) PPI based on the 63 selected proteins. (G) Expression of ITGB1 in LX2 on low (L) and high (H) stiffness gel. (H) Immunofluorescence of ITGB1 activation in LX2 (scale bar, 20 μm). (I) Expression of active RhoA, p-FAK, FAK, p-PI3K, PI3K, p-AKT and AKT in LX2 on varying substrates stiffness. (J) ITGB1, COL1, LOX, and α-SMA expression in LX2 on stiff substrate post-ITGB1 knockdown. (K) Gel contraction of LX2 from high-stiffness substrate after ITGB1 knockdown. ∗∗p = 0.0059 (n = 3, mean ± SD, unpaired t test). ECM, extracellular matrix; HSC, hepatic stellate cell; ITGB1, integrin β1; MF, myofibroblast; α-SMA: alpha-smooth muscle actin; COL1: collagen type I alpha 1 chain; LOX: lysyl oxidase; GFAP: glial fibrillary acidic protein; PPI, protein-protein interaction; RhoA: ras homolog family member A; FAK: focal adhesion kinase; PI3K: phosphoinositide 3-kinase; AKT: protein kinase B.

Next fibroblast-mediated matrix remodeling was assessed. Both of LX2 cells and rat HSCs exhibited stiffness-dependent cytoskeletal organization – a parallel arrangement of the F-actin network vs. mesh-like networks on soft substrates (Fig. 1C). Furthermore, primary HSCs from cirrhotic rats (activated state) demonstrated stronger gel contraction than those from normal rats (quiescent state) (Fig. 1D).

To explore protein expression linked to HSC activation, we performed a proteomic analysis of LX2 cells on hard and soft hydrogels. Gene Ontology (GO) analysis revealed that the differentially expressed proteins were enriched in ECM remodeling (Fig. S1F). Sixty-three proteins that overlapped with differentially expressed proteins in LX2 (Fold change ≥1.5, Table S1) and ECM-related proteins from the database were identified (Fig. 1E). Protein-Protein Interaction (PPI) network analysis of the 63 candidate proteins prioritized ITGB1 because of its central role in adhesion and ECM remodeling (Fig. 1F and Table S2). Western blot and immunofluorescence confirmed elevated ITGB1 expression (Fig. 1G and Fig. S1G) and activation (Fig. 1H) in LX2 cells on stiff vs. soft hydrogels. Key molecules in ITGB1 signaling, such as RhoA, FAK, PI3K, and AKT, were also more activated on hard substrates (Fig. 1I). Furthermore, the expression of COL1, LOX, and α-SMA (Fig. 1J and Fig. S1H), as well as LX2 cell contraction (Fig. 1K), were abolished by ITGB1 knockdown, even on stiff substrates.

High TGM2 expression on the cell surface is required for ITGB1 activation in LX2 cells under high matrix stiffness

TGM2 enhances integrin-fibronectin binding and activates integrin signaling in ovarian cancer.²³ Analysis of 178 patients with liver cancer and cirrhosis from GSE14520, GSE16757, and TCGA-LIHC datasets revealed a strong correlation between TGM2 and ITGB1 expression (R = 0.77, P <2.2 × 10^-16) (Fig. 2A). Patients with high TGM2 and ITGB1 expression had significantly lower overall survival (OS) (Fig. 2B, p = 0.0285). We therefore investigated this interaction in LX2 cells cultured on gels of varying stiffness. Total TGM2 expression (Fig. 2C and Fig. S2A) and surface TGM2 levels detected by flow cytometry (Fig. 2D) were significantly higher in LX2 cells on stiff gels than on soft gels. Confocal microscopy confirmed TGM2 and ITGB1 co-localized on the LX2 cell surface on stiff substrates (Fig. 2E). Knocking down TGM2 on stiff substrates reduced ITGB1 activation and the expression of α-SAM, COL1, and LOX, indicating decreased HSC activation into MFs (Fig. 2F and Fig. S2B). TGM2 knockdown also abolished ECM remodeling-related cell contraction (Fig. 2G). To explore the specific interaction between TGM2 and ITGB1, we analyzed their crystal structures and identified 10 potential docking poses, with one pose being the most likely interaction (Fig. 2H). Immunoprecipitation confirmed TGM2 directly interacted with ITGB1 in LX2 cells (Fig. 2I).

Fig. 2.

High TGM2 expression on the cell surface is required for ITGB1 activation in LX2 cells.

(A) TGM2 and ITGB1 correlation in tumors from 178 patients with liver cancer and cirrhosis. R = 0.77, p <2.2 × 10^-16 (Spearman’s rank correlation coefficient analysis). (B) Overall survival by TGM2/ITGB1 expression (high vs. low, median cut-off; p = 0.0285, Wilcoxon test). (C) TGM2 expression in LX2 cells on different substrates. L, low stiffness; H, high stiffness. (D) Flow cytometry of surface TGM2 in LX2 cells on high-vs. low-stiffness gels. ∗∗p = 0.0014 (Isotype vs. High stiffness); ∗∗p = 0.0049 (Low stiffness vs. High stiffness); ∗∗∗∗p <0.0001; (n = 3, mean ± SD, ANOVA). (E) Immunofluorescence for ITGB1 (green) and TGM2 (red) in LX2 on different stiffness substrates. Protein co-localization is identified by yellow spectra on merged images. ∗∗∗p = 0.0003 (n = 3, mean ± SD, unpaired t test). (F) After TGM2 knockdown, the levels of active-ITGB1, α-SMA, COL1, LOX in LX2 cells on a high-stiffness substrate. (G) Gel contraction of LX2 from high-stiffness substrate after TGM2 knockdown. ∗∗∗p = 0.001 (n = 3, mean ± SD, unpaired t test). (H) Protein–protein interaction poses were identified by analyzing the crystal structure of TGM2 (PDB ID: 2Q3Z) and ITGB1 (PDB ID: 4WK0). (I) Co-IP with anti-TGM2 and anti-ITGB1 antibodies was performed on LX2 cells cultured on different stiffness matrix. L, low stiffness; H, high stiffness. ITGB1, integrin β1; TGM2, transglutaminase 2; α-SMA: alpha-smooth muscle actin; COL1: collagen type I alpha 1 chain; LOX: lysyl oxidase.

CAV1 functioned in TGM2/ITGB1-mediated cell activation and contraction under high-stiffness matrix

Previous studies propose that CAV1 or tetraspanins may enhance integrin signaling through TGM2 interaction.²³ Here we revealed that CAV1 expression was significantly upregulated in LX2 cells cultured on stiff substrates (Fig. 3A and Fig. S2C). TGM2 knockdown abolished this CAV1 induction (Fig. 3B and Fig. S2D), whereas CAV1 silencing did not alter surface TGM2 levels (Fig. 3C) but diminished active ITGB1, total ITGB1, COL1, LOX, α-SMA, RhoA, and p-MLC (Fig. 3D and Fig. S2E). The stiff matrix-induced contraction of LX2 cells was also partially abolished after CAV1 knockdown (Fig. 3E). Additionally, ITGB1 knockdown decreased CAV1 without affecting TGM2 expression under stiffness (Fig. 3F and Fig. S2F). These findings indicate that TGM2 potentiates ITGB1 activation, thereby elevating CAV1 levels, which reciprocally stabilize ITGB1 expression and amplify downstream signaling.

Fig. 3.

CAV1 takes part in TGM2/ITGB1-mediated cell contraction under high-stiffness matrix.

(A) CAV1 expression in LX2 on different stiffness gels. L, low stiffness; H, high stiffness. (B) CAV1 expression after TGM2 knockdown on high-stiffness substrate. (C) Flow cytometry of cell-surface TGM2 in LX2 (CAV1 siRNA vs. Control siRNA). ∗∗∗p = 0.0003 (Isotype vs. Control siRNA); ∗∗∗p = 0.0002 (Isotype vs. CAV1 siRNA); ns = 0.5953; (n = 3, mean ± SD, ANOVA). (D) Levels of active ITGB1, ITGB1, COL1, LOX, α-SMA, RhoA, and p-MLC, and (E) Gel contraction of LX2 from high-stiffness substrate after CAV1 knockdown. ∗p = 0.0126 (n = 3, mean ± SD, unpaired t test). (F) TGM2 and CAV1 expression in LX2 cells on high-stiffness gel psot-ITGB1 knockdown. (G) Fluorescence assay for intracellular Ca²⁺ in LX2 cells under different stiffness (cell number: n = 30/3 independent experiments). ∗∗∗∗p <0.0001 (n = 3, mean ± SD, Welch’s t test). (H) Transmission electron microscopy images show plasma membrane cross-sections with caveolae (red arrows). Quantification of caveolae number and size per μm of membrane under different stiffness substrate [caveolae number: n (low stiffness) = 11/3 experiments, n (high stiffness) = 16/3; caveolae size number: n (low stiffness) = 16/3 experiments, n (high stiffness) = 65/3]. ∗p = 0.0128 (Mann–Whitney test); ∗∗∗p = 0.0008 (Welch’s t test). (I) Post-CAV1 knockdown, intracellular Ca²⁺ fluorescence in LX2 cells on high-stiffness substrate (cell number: n = 30/3 experiments). ∗∗∗∗p <0.0001 (Mann–Whitney test). (J) Expression of CAV1, COL1, LOX, α-SMA, active RhoA, and p-MLC in LX2 cells on high-stiffness substrate after calcium ion inhibitors treatment. CAV1, caveolin-1; ITGB1, integrin β1; TGM2, transglutaminase 2; siRNA: small interfering RNA; α-SMA: alpha-smooth muscle actin; COL1: collagen type I alpha 1 chain; LOX: lysyl oxidase; RhoA: ras homolog family member A; p-MLC: phosphorylated myosin light chain.

CAV1, the main coat protein of caveolae, influences cell mechanics and Ca²⁺ influx signaling. To investigate the contribution of TGM2/ITGB1 to matrix stiffness-induced cell contraction, we assessed Ca²⁺ influx in LX2 cells on substrates of varying stiffness. Ca²⁺ fluorescence intensity increased significantly with higher matrix stiffness (Fig. 3G). Transmission electron microscopy (TEM) revealed a more than twofold increase (0.95 per micron) in caveolae invaginations in HSCs on stiff gels compared with soft gels (0.33 per micron), with a slight increase in diameter (Fig. 3H). CAV1 knockdown via siRNA significantly reduced Ca²⁺ fluorescence intensity (Fig. 3I). Furthermore, Ca²⁺ inhibitors reversed stiffness-induced upregulation of COL1, LOX, α-SMA, active RhoA, and p-MLC in LX2 cells on stiff substrates but did not affect CAV1 expression (Fig. 3J and Fig. S2G).

Activated HSC-derived mitochondria potentiated the migration of HCC cells under high-stiffness matrix

Historically, increased matrix stiffness has been linked to HCC cell migration by inducing epithelial–mesenchymal transition (EMT).²⁴ Consistent with this, our study demonstrated that HCC cells cultured on stiffer substrates adopted a fibroblast-like morphology (Fig. 4A), accompanied by decreased E-cadherin and increased N-cadherin, vimentin, Snail, and MMP9 expression (Fig. S2H). TGM2 and ITGB1 levels also rose with stiffness (Fig. 4B). To better simulate the tumor microenvironment, we established monocultured (Huh7 alone) and co-cultured (Huh7+LX2) spheroids on substrates of varying stiffness. Confocal live-cell imaging revealed that Huh7 cells exhibited enhanced migration on stiff matrices, with co-culture systems further amplifying this effect compared with monocultures (Fig. 4C and D). ATP levels, which support HCC cell migration, were higher in co-cultured Huh7 cells on stiff substrates (Fig. 4E). As mitochondria generate most cellular ATP,^25,26 we next asked whether stiffness altered mitochondrial content in co-cultured Huh7 cells. Flow cytometry revealed that Huh7 cells from co-cultured spheroids on stiff substrates contained more mitochondria than those on soft substrates (Fig. 4F). Because protein levels of PGC1A (a key regulator of mitochondrial biosynthesis) and TFAM (a mitochondrial nucleoid marker) remained unchanged (Fig. 4G and Fig. S2I), these data suggested that the increase in mitochondrial content resulted from MT from LX2 cells rather than de novo synthesis.

Fig. 4.

Activated HSCs-derived mitochondria potentiates HCC cell migration.

(A) Alexa-488 Phalloidin staining shows that high matrix stiffness promotes F-actin polymerization in Huh7 and 97H cells (scale bar, 30 μm). (B) Expressions of TGM2, ITGB1, p-FAK, FAK, p-PI3K, PI3K, p-AKT, and AKT in Huh7 cells on different stiffness substrates. L, low stiffness; H, high stiffness. (C) Live cell images of monoculture Huh7 (green) spheroid and co-culture spheroid with Huh7 (green) and LX2 (red) at 2:1 ratio, at 4 h, 24 h, and 48 h on different stiffness substrates. (D) Cell movement analysis under different stiffness using a real-time cell monitoring system, ∗p = 0.0463 (High-Huh7 vs. High-Huh7 + LX2, at 12 h); ∗p = 0.0136 (High-Huh7 vs. High-Huh7 + LX2, at 18 h); ^#p = 0.0144 (High-Huh7 vs. Low-Huh7); ∗∗∗p = 0.0006; ^##p = 0.0033; ^###p = 0.0007; ∗∗∗∗p <0.0001; ^####p <0.0001; by ANOVA. (E) ATP levels in Huh 7 cells from the co-culture system on different stiffness substrates (n = 3, ∗∗p = 0.0043, unpaired t test). (F) MitoTracker uptake in Huh7 cells is measured by flow cytometry, and MFI is quantified and graphed for each group (co-culture spheroid on different stiffness substrate). n = 3, ∗∗∗∗p < 0.0001, unpaired t test. (G) Expression of PGC1A and TFAM in Huh 7 cells from co-culture on different stiffness substrates. F-actin, filamentous actin; HCC, hepatocellular carcinoma; HSCs, hepatic stellate cells; ITGB1, integrin β1; TGM2, transglutaminase 2; FAK: focal adhesion kinase; PI3K: phosphoinositide 3-kinase; AKT: protein kinase B; MFI: mean fluorescence intensity; ATP: adenosine triphosphate; PGC1A: peroxisome proliferator-activated receptor gamma coactivator 1-alpha; TFAM: mitochondrial transcription factor A.

Intercellular MT occurs mainly through TNTs, microvesicles (MVs), and gap junctions.²⁷ Using inhibitors targeting TNTs (Cyt-B), gap junctions (18-α-GA), and MVs (dynasore), we found Cyt-B significantly blocked MT (Fig. 5A), confirming TNTs as the primary route for mechanical stress-induced transfer. Using confocal microscopy, we observed that MT from activated LX2 to Huh7 cells via TNTs on stiffness, confirmed by 3D Z-stack imaging (Fig. 5B, white arrows). Cyt-B inhibited stiffness-induced TNT formation, which are structurally composed of F-actin and β-tubulin, reducing MT (Fig. 5C) and Huh7 migration from co-cultured spheroids (Fig. 5D and E). Because of the role of ITGB1 in regulating the dynamics of the actin cytoskeleton, we knocked down TGM2 or ITGB1 and observed a reduction in TNT formation (Fig. 5F), as well as decreased mitochondrial acquisition by Huh7 cells on stiff substrates (Fig. 5G).

Fig. 5.

Activated HSCs transfer mitochondria to HCC cells via TNTs under high-stiffness substrate.

(A) Flow cytometry measures MitoTracker (red) uptake in Huh 7 cells from co-culture spheroids on high-stiffness gels, treated with 18α-glycyrrhetinic acid (50 μM), dynasore (50 μM), or cytochalasin B (2.5 μM) (n = 3). Fluorescence microscopy visualized Huh7 (Cell-Tracker – blue) and LX2 (MitoTracker – red), with F-actin (Fluor^TM488 Phalloidin). ∗∗p = 0.0095; ns, p = 0.5552 (Control vs. 18α-Gly); ns, p = 0.9702 (Control vs. Dyna); mean ± SD, by ANOVA. (B) Cells on different stiffness gels and (C) cells on high-stiffness substrate with/without Cyt-B treatment. The white arrows indicate LX2-derived mitochondria moving into Huh7 cells via TNTs. Scale bar, 50 μm. (D) Live imaging of co-culture spheroids (Huh7 in Green; LX2 in red) at 4 h, 24 h, and 48 h on stiff substrates with or without Cyt-B treatment. (E) Analysis of cell movement under different stiffness by the real-time cell monitoring system. ∗∗p = 0.0043; ∗∗∗∗p <0.0001 (mean ± SD, ANOVA). (F) Fluorescence microscopy of Huh7 (blue) and LX2 (red) with F-actin (Phalloidin-488) on stiffness gels following ITGB1 or TGM2 depletion. (G) Flow cytometry analysis of mitochondrial uptake (MitoTracker – red) in Huh7 cells co-cultured with LX2 on stiff gels after ITGB1 or TGM2 knockdown. ∗∗∗∗p <0.0001 (mean ± SD, ANOVA). F-actin, filamentous actin; HCC, hepatocellular carcinoma; HSCs, hepatic stellate cells; ITGB1, integrin β1; TGM2, transglutaminase 2; TNTs, tunneling nanotubes; Cyt-B: cytochalasin B.

Overall, HCC cells showed increased migration on stiff substrates after acquiring mitochondria from activated HSCs via TNTs. TGM2/ITGB1 likely promotes TNT formation by regulating the actin cytoskeleton through mechanotransduction.

Stiffness-activated TGM2/ITGB1 promoted HSC activation into CAFs and the postoperative recurrence of HCC with cirrhosis in vivo

The migratory and invasive capacity of HCC cells underlies postoperative recurrence. To assess the role of mechanical force-mediated TGM2/ITGB1 in tumor recurrence, we used a rat HCC model with tumors implanted in livers of different stiffness. We created rat models with varying liver stiffness (Fig. S3) and implanted subcutaneous tumors into the liver subcapsular region. After tumor formation, partial liver resection was performed (Fig. 6A). Within 28 days after surgical resection, recurrence rates reached 100% in cirrhotic livers vs. 60% in normal liver groups (Fig. 6B and Table S3). Atomic force microscopy confirmed elevated tissue stiffness in tumors from cirrhotic livers (Fig. 6C), and Masson staining further demonstrated enhanced collagen deposition in these tumors (Fig. S4A). Consistently, high-stiffness tissue presented significantly greater expression of TGM2, ITGB1, CAV1, α-SMA, COL1, and LOX than low-stiffness tumor tissue (Fig. 6D and Fig. S4B). Notably, TGM2/ITGB1 co-expression was significantly elevated in high-vs. low-stiffness tumor tissue (Fig. 6E), consistent with our in vitro findings. This pattern extended to adjacent non-tumor tissues (Fig. S4C and Fig. S4D). Western blot confirmed upregulation of TGM2, active ITGB1, ITGB1, CAV1, α-SMA, COL1, LOX, and desmin, with concomitant E-cadherin reduction in high-stiffness tissue (Fig. 6F and Fig. S4E). The above data suggest a potential link between TGM2/ITGB1-mediated CAFs activation, ECM remodeling, and postoperative HCC recurrence under high substrate stiffness conditions.

Fig. 6.

Stiffness-activated TGM2/ITGB1 promotes ECM remodeling effects of CAFs and tumor recurrence post-surgery in vivo.

(A) A flowsheet details the creation of orthotopic HCC models with varying liver stiffness, followed by surgical resection and recurrence monitoring in rats. (B) Positron emission tomography-computed tomography scans for HCC recurrence in rats with varying liver stiffness post-liver cancer resection (n = 5). (C) Tumor stiffness from rats with different liver hardness. ∗∗∗∗p <0.0001 (mean ± SD, Welch’s t test). (D) Immunohistochemistry images showing TGM2, ITGB1, CAV1, α-SMA, COL1, and LOX in tumors from rats with/without cirrhosis. (E) Co-expression of TGM2 and ITGB1 in tumors from rats without/with cirrhosis. (F) Expression of TGM2, active-ITGB1, ITGB1, CAV1, α-SMA, COL1, LOX, Desmin, and E-cadherin in tumors from rats with/without cirrhosis by Western blot. CAFs, cancer-associated fibroblasts; CAV1, caveolin-1; ECM, extracellular matrix; HCC, hepatocellular carcinoma; ITGB1, integrin β1; TGM2, transglutaminase 2; α-SMA: alpha-smooth muscle actin; COL1: collagen type I alpha 1 chain; LOX: lysyl oxidase.

TGM2 and ITGB1 repression partly abolished the postsurgical recurrence of HCC in rats with high liver stiffness

To investigate the function of TGM2/ITGB1 in stiffness-driven HCC recurrence, we treated cirrhotic rats bearing orthotopically implanted tumors with luciferase-labeled adeno-associated virus 1 (AAV1)-TGM2 RNAi via tail vein injection and/or the ITGB1 inhibitor ATN-161 orally (Fig. S5A and Fig. S5B). Twenty-eight days post-tumor resection, all rats with liver-implanted tumors in control group relapsed. Recurrence occurred in 80% of rats treated with either ATN-161 or AAV1-TGM2 RNAi alone, but only 60% relapsed when given both treatments (Fig. 7A and Table S4). Masson staining showed reduced collagen deposition in tumors from the combination therapy group compared with other groups (Fig. S5C). Western blot and immunohistochemistry showed that the AAV1-TGM2 RNAi and ATN-161 therapy group had reduced levels of TGM2, ITGB1, CAV1, α-SMA, desmin, COL1, LOX, and N-cadherin, and increased E-cadherin compared with other groups (Fig. 7B and C and Fig. S5D). Immunohistochemistry also detected COL1, LOX, and α-SMA in adjacent non-tumor tissues (Fig. S5E). Collectively, these results confirm the role of TGM2 and ITGB1 in matrix stiffness-promoted postsurgical recurrence of HCC in an orthotopic liver tumor rat model.

Fig. 7.

TGM2 and ITGB1 suppression abolished ECM remodeling and tumor recurrence in rats with high liver stiffness.

(A) PET-CT for HCC recurrence in cirrhosis rats treated with TGM2-RNAi and/or ATN161 treatment (n = 5). (B) Expression of TGM2, ITGB1, CAV1, α-SMA, Desmin, COL1, and E-cadherin in tumors from control, ATN161, TGM2-RNAi and combined treatment groups with high liver stiffness. (C) Representative immunohistochemistry images of COL1, LOX, α-SMA, E-cadherin and N-cadherin in tumors from control, ATN161, TGM2-RNAi, and combined treatment groups with high liver stiffness. CAV1, caveolin-1; ECM, extracellular matrix; HCC, hepatocellular carcinoma; ITGB1, integrin β1; TGM2, transglutaminase 2; α-SMA: alpha-smooth muscle actin; COL1: collagen type I alpha 1 chain; LOX: lysyl oxidase.

Discussion

The tumor microenvironment, especially the arrangement and rigidity of the ECM, plays a pivotal role in cancer progression and has attracted substantial research interest. As the dominant stromal cells in the HCC microenvironment, CAFs exert profound effects on hepatoma cell behavior via both biological and mechanical signaling. Nevertheless, the precise mechanisms by which CAFs contribute to HCC progression remain incompletely characterized, owing to limited investigation in this area.

The first discovery in this study was that microenvironmental stiffness triggers HSC activation into MFs, evidenced by morphological alterations and changes in characteristic markers. F-actin reorganization and enhanced gel contraction confirmed the ECM remodeling capacity of HSCs on stiff substrates, consistent with prior reports.[28], [29], [30] Proteomic profiling in LX2 cells identified stiffness-dependent ECM remodeling proteins, with ITGB1 selected for its established roles in adhesion, migration, and mechanotransduction.³¹ ITGB1 emerged as a mechanoresponsive regulator of HSC activation, demonstrating both upregulated expression and enhanced activation under high-stiffness conditions, thereby amplifying RhoA/FAK-mediated contractile signaling – a central driver of pathological ECM remodeling (Fig. 1). Unfortunately, targeting ITGB1 directly is somewhat challenging because of its complexity and contradictory characteristics.

The co-elevation of TGM2 and ITGB1 in patients with HCC and cirrhosis correlated with poor OS, suggesting a synergistic pro-tumorigenic role. We demonstrated that substrate stiffening upregulated TGM2 in LX2 cells, which subsequently potentiated ITGB1 activation via direct physical interaction (validated by co-IP and molecular docking) and spatial co-localization at the cell-matrix interface (Fig. 2). This finding aligns with prior evidence that membrane-localized TGM2 facilitates integrin engagement in fibrotic microenvironments,³² as demonstrated by our observation of stiffness-dependent enrichment of TGM2-ITGB1 complexes on the cell surface. Notably, the TGM2-ITGB1 interaction in HCC had not been previously reported. Although TGM2 deficiency was shown to minimally affect fibrotic ECM stability in vivo,³³ our data propose that the crosstalk between TGM2-ITGB1 – rather than individual roles – drives pathological ECM remodeling. This hypothesis was reinforced by clinical and preclinical evidence: both TGM2 and ITGB1 were overexpressed in patients with HCC and cirrhosis and orthotopic rat models with high liver stiffness (Fig. 2, Fig. 6, Fig. 7), correlating with adverse outcomes. Collectively, our study identifies matrix stiffness as a critical inducer of TGM2-ITGB1 synergy in HCC progression. This mechanism may explain the limited efficacy of therapies targeting TGM2 alone in fibrosis, emphasizing the need for combinatorial strategies to disrupt their interaction.

Previous studies suggest that TGM2 oligomerizes with integrin-binding proteins (e.g. CAV1, tetraspanins) to modulate integrin activity,³⁴ potentially through their co-localization in lipid rafts.³⁵ Our data reveal that CAV1, acting downstream of TGM2/ITGB1, is essential for stiffness-induced ITGB1 signaling and LX2 cell contraction (Fig. 3). Mechanistically, TGM2 may promote ITGB1 activating, thereby enhancing CAV1 expression to stabilize integrin levels and activate downstream pathways (e.g. RhoA/MLC). This reciprocal regulation aligns with reports that ITGB1 signaling upregulates CAV1, while CAV1 controls ITGB1 membrane recycling and protein turnover.^36,37 Consistent with elevated CAV1 expression, LX2 cells on stiff substrates exhibited increased caveolae formation and Ca²⁺ influx. As CAV1 is a structural component of caveolae³⁷ – key membrane domains integrating mechanical and biochemical signals – this observation suggests that CAV1-enriched caveolae amplify mechanoadaptation. Notably, Ca²⁺ influx, which is stiffness-dependent and oncogenic,^38,39 activates RhoA/MLC to drive ECM remodeling.^40,41 Thus, CAV1 may serve as a mechanotransducer linking matrix stiffness to Ca²⁺-RhoA signaling in HCC progression. Indeed, CAV1 knockdown or calcium deletion abolished LX2 contraction even under high stiffness, underscoring the central role of CAV1 in TGM2/ITGB1-mediated ECM remodeling.

With respect to the impact of mechanical force on the behavior of HCC cells, a 3D environment more accurately simulates in vivo conditions. Under this condition, Huh7 cells exhibited enhanced migration on stiff substrates, particularly when co-cultured with LX2 cells (Fig. 4), suggesting that HSCs may enhance HCC cell migration through intercellular communication, in addition to ECM remodeling. Notably, co-cultured Huh7 cells showed elevated ATP levels, produced mainly through mitochondrial oxidative phosphorylation.⁴² However, mitochondrial abnormalities are common in cancer, and replenishing damaged mitochondria is crucial for tumor cells energy needs.²⁵ Strikingly, under high substrate stiffness, Huh7 cells acquired functional mitochondria predominantly through transfer from LX2 cells rather than de novo synthesis. Mechanistically, we identified TNTs as the primary route for MT from LX2 to Huh7 cells under high matrix stiffness. TNT formation is driven by cytoskeletal reorganization, influenced by cellular adhesion and ECM remodeling. Indeed, disrupting actin polymerization with Cyt-B under high-stiffness conditions reduced Huh7 cells migration and MT from LX2 to Huh7 cells, similar to MT from adipose stem cells to breast cancer cells.⁴³ This suggests high matrix stiffness aids TNT formation via cytoskeletal changes, facilitating MT. ITGB1 is crucial in TNT formation for MT by regulating actin dynamics.¹⁶ Knocking down TGM2 or ITGB1 reduced TNT formation and MT from LX2 cells to Huh7 cells, even on stiffer substrates, likely because of impaired mechanical signal transduction and cytoskeletal reorganization. These findings highlight the role of ITGB1 in high matrix stiffness-mediated HSC activation and HCC cell migration.

Migrating cancer cells may serve as the ‘seeds’ that drive tumor recurrence and metastasis.⁴⁴ To further investigate the role of TGM2/ITGB1 in matrix stiffness-induced postoperative recurrence, we developed HCC models with varying liver stiffness backgrounds and validated their function in vivo. In rats with high liver stiffness, the recurrence rate reached 100%, whereas those with normal stiffness exhibited only 60% recurrence within 28 days post-surgery. Reduced E-cadherin in high-stiffness tumors indicated enhanced migratory potential. Atomic force microscopy confirmed elevated tissue stiffness in tumors from cirrhotic rats. Consistently, tumors with elevated stiffness exhibited more collagen deposition, higher co-expression of ITGB1/TGM2, and increased markers of CAF activation and ECM remodeling, such as α-SMA, LOX, and COL1. Notably, dual knockdown of TGM2 and ITGB1 reduced post-resection recurrence to 60% in high-stiffness rat models, compared with 80% recurrence with single knockdown, suggesting compensatory pathways in HCC progression. This was supported by E-cadherin upregulation and N-cadherin downregulation in dual knockdown group. These results correspond to the data from patients with HCC and cirrhosis, showing that the OS of patients with high TGM2 and ITGB1 expression was significantly lower than that of patients with low TGM2 and ITGB1 expression. Furthermore, TGM2/ITGB1 repressing attenuated collagen deposition, and reduced CAF activation markers (α-SMA, COL1, and LOX) in tumors from cirrhotic livers. These results establish that elevated liver stiffness drives CAF-mediated ECM remodeling and HCC recurrence, with TGM2/ITGB1 serving as critical mechanotransductive mediators.

Nonetheless, it remains possible that other mechanisms are involved in this pathological process. As research into the tumor mechanical microenvironment advances and more ideal animal models are developed, other molecular mechanisms underlying stiffness-driven HCC progression will be better elucidated.

Conclusions

In summary, activated HSCs facilitate HCC cell migration by ECM remodeling and MT under high ECM stiffness. This is the first attempt to elucidate how increased matrix stiffness drives MT, thereby propelling the migration of HCC cells, and highlights the role of TGM2-mediated ITGB1 activation in high matrix stiffness-induced HSC activation. This study may explore potential therapies for anticancer therapy targeting CAFs.

Abbreviations

AAV1, adeno-associated virus 1; CAFs, cancer-associated fibroblasts; CAV1, caveolin-1; ECM, extracellular matrix; EMT, epithelial–mesenchymal transition; F-actin, filamentous actin; HCCs, hepatocellular carcinomas; HSCs, hepatic stellate cells; ITGB1, integrin β1; MFs, myofibroblasts; MT, mitochondrial transfer; MVs, microvesicles; OS, overall survival; TGM, transglutaminase; TNTs, tunneling nanotubes; YAP, yes-associated protein.

Financial support

This work was supported by the National Natural Science Foundation of China [grant numbers No. 82074138 and 82374174].

Authors' contributions

Contributed to the study conception and design: all authors. Proposed conceptualizations and designed the study: JD. Developed the experiment methods and completed all the experiments: MW, YX, WX. Helped complete the animal experiments: JC. Provided guidance and suggestions for the experiments: YM. Wrote the original draft and revised it: JD, MW. Read and approved the final manuscript: all authors.

Conflicts of interest

The authors declare that they have no competing interests.

Please refer to the accompanying ICMJE disclosure forms for further details.

Supplementary data

The following are the supplementary data to this article: