2.1 Public database analysis

The Cancer Genome Atlas (TCGA, http://www.cbioportal.org, Liver Hepatocellular Carcinoma of PanCancer Atlas) and Gene Expression Omnibus (GEO;, GSE135251) databases were used to predict the potential roles of LGALS3BP and TGFB1 in patients with liver cancer and metabolic dysfunction-associated steatohepatitis (MASH).

2.2 Clinical samples from Chonnam National University Hwasun Hospital

Tumoral and peri-tumoral normal tissues were collected from 83 patients with HCC (date of visit: from 2017-01-25 to 2021-07-14) at Chonnam National University Hwasun Hospital (CNUHH, Hwasun, Jeollanam-do, South Korea). Clinicopathological characteristics of patients are described in Supplementary Table S1. The present study was approved by the ethics committee of CNUHH (Approval number: IRB CNUHH-2021-187). Written informed consent was obtained from all the patients.

2.3 Cell culture

Mouse primary hepatocytes were isolated via in situ collagenase perfusion and separation using the Percoll (Sigma-Aldrich, Burlington, MA, USA) density-gradient method from the livers of 10-week-old male LGALS3BP-KI mice and their controllittermates or LGALS3BP fl/wt; Alb-Cre mice and their LGALS3BP fl/wt littermates. Isolated mouse primary hepatocytes were cultured in Dulbecco's modified Eagle's medium (Gibco, Burlington, MA, USA) containing 10% fetal bovine serum (FBS) with 5×10⁵ cells per well in six-well plates for 40 h before analysis. Mouse HCC (Hepa-1c1c7) and human HCC cells (HepG2, Hep3B and SNU449) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) or Korean cell line bank (KCLB, Seoul, South Korea). All cells in this study were cultured in the recommended medium as alpha minimum essential medium (Gibco) for Hepa-1c1c7, Dulbecco's Modified Eagle's Medium for Hep3B, Eagle's minimum essential medium (Sigma-Aldrich) with 25 mmol/L sodium bicarbonate for HepG2, or RPMI-1640 medium (Sigma-Aldrich) for SNU449 cells, containing 10% fetal bovine serum (Sigma-Aldrich) and 1% penicillin and streptomycin (Invitrogen). To evaluate the gene regulation effects of recombinant proteins, cells were serum-starved for 16 h followed by the treatment with different concentrations of recombinant IFN-α (rIFN-α), recombinant LGALS3BP (rLGALS3BP) or recombinant TGF-β1 (rTGF-β1) for subsequent whole-cell extract purificaiton or total RNA isolations.

2.4 Animals

All animal experiments were performed in specific pathogen-free facilities at Chonnam National University Medical School, according to the guidelines of the Institutional Animal Care Committee (CNU IACUC-H-2021-7). LGALS3BP-KO mice were generated using the clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 editing system (Macrogen, Seoul, South Korea) with two single-guide RNAs (sgRNAs, 5′-GCCAGGCAATGGCTCTCCTGTGG-3′ and 5′-CTGTGTTCTTGCTGGTTCCAGGG-3′) targeting exon 3 of LGALS3BP as previously described [31]. Briefly, the sgRNAs and Cas9 recombinant proteins were co-injected into the embryos of C57BL/6N mice (Orient Bio, Seongnam, South Korea), which were then implanted into the oviducts of foster mothers to obtain newborn pups. LGALS3BP-KO mice were screened using PCR, the T7E1 assay, and sequencing of tail genomic DNA. The PCR primers used for genotyping are listed in Supplementary Table S2. The T7E1 assay was performed using T7 Endonuclease I (M0302L) according to the manufacturer's instructions (NEB, Ipswich, MA, USA). Heterozyous LGALS3BP-KO mice were further bred on C57BL/6J genetic background mice for at least six generations and intercrossed to obtain homozygous LGALS3BP-KO mice and its control wild type littermates. The hepatocyte-specific LGALS3BP-knockin (KI) mouse was developed using the Cre-/Loxp recombinant enzyme system (Macrogen). Briefly, a silenced C-terminal Myc-tagged LGALS3BP-KI cassette under the control of LoxP sites was introduced into the ubiquitously expressed Rosa26 locus of C57BL/6J mouse embryos. Newborn heterozygous LGALS3BP-KI pups were screened and bred on a C57BL/6J genetic background for at least four generations. Alb-Cre mice were used to obtain mice with hepatocyte-specific LGALS3BP overexpression in LGALS3BP fl/wt; Alb-Cre mice. For the carbon tetrachloride (CCl₄)-induced liver fibrosis model [32, 33], 18-week-old male C57BL/6J mice were injected intraperitoneally (IP) with CCl₄ (0.5µL/g, 1:3, v/v in corn oil; #289116, Sigma) or vehicle (dimethyl sulfoxide in corn oil) twice per week for 4 or 6 weeks. Animals were euthanized by cervical dislocation 72 h after the final CCl₄ injection, and whole livers and sera were collected. For the mouse MASH model, 8-week-old male wild type C57BL/6J mice were purchased (Orient Bio, Seongnam, South Korea) and acclimated to a 12 h light/dark cycle at 22 ± 2°C for 2 weeks. The mice were divided into two groups: the control group was fed a normal chow diet and the MASH group was fed a high-fat methionine-choline-deficient (HFMCD) diet with l-amino acids, 60% calories from fat, 0.1% methionine, and no added choline (A06071302, Research Diets, New Brunswick, NJ, USA) for 5 weeks before being sacrificed. For the diethylnitrosamine (DEN)-high-fat diet (HFD) model of liver carcinogenesis [34], 2-week-old male LGALS3BP-KO mice and their littermates were injected with DEN (25 mg/kg body weights, 442687, Merck-Millipore, Burlington, MA, USA). The mice were fed HFD (D12492, Research Diets, New Brunswick, NJ, USA) for 26 weeks until they were sacrificed. The tumors in each liver lobe were counted and measured. Tumors, non-tumor tissues, and serum samples were collected for further analysis.

2.5 Enzyme-linked immunosorbent assay (ELISA)

The levels of LGALS3BP protein were assessed in mouse serum using a mouse Mac-2 binding protein assay kit (27796, Immuno-Biological Lab, Japan). Conditioned media (CM) derived from Hepa-1c1c7, SNU449, or HepG2 cell cultures under the indicated conditions with rLGALS3BP treatment were collected and stored at −20°C. To evaluate the influence of integrin αV or FAK inhibition on the regulation of TGF-β1 by LGALS3BP, cells were pre-incubated with 0.5 ng/mL for Hepa-1c1c7 or 1 ng/mL for human HCC cell lines of GLPG0187 (HY-100506, MedChemExpress, Monmouth Junction, NJ, USA) or 1 µmol/L Defactinib (HY-12289, MedChemExpress) for 1 hour before being exposed to rLGALS3BP. The concentration of TGF-β1 in the conditioned media was assessed using TGF-β1 Duoset ELISA kits (DY1679 for mouse or DY240 for human TGF-β1, R&D Systems) according to the manufacturer's instructions. Briefly, to measure concentrations of total latent TGF-β1, the CM were activated with acids treatment and stopped for the subsequent quantification. The free active forms of TGF-β1 in the CM were quantified without acid activation.

2.6 Western blotting

Whole cell extracts from mouse liver tissues, cultured mouse primary hepatocytes, or Hepa-1c1c7, HepG2, or SNU449 cell lines were prepared in standard radio-immunoprecipitation assay buffer (Invitrogen) and separated by Sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes (Invitrogen). Membranes were probed with primary antibodies and secondary peroxidase-conjugated antibodies (Jackson Immuno-Research, West Grove, PA, USA) and developed using Western BLoT Ultra-Sensitive Horseradish peroxidase Substrate (Takara, Kusatsu, Japan). Signal intensities were detected using an Image Quant LAS 4000 mini (Minato City, Tokyo, Japan). Primary antibodies against phosphorylated Mothers against decapentaplegic homolog 2 (SMAD) family member 2 (p-Smad2, #3108, 1:3,000, Cell signaling technology [CST], Danvers, MA, USA), Smad2 (#5339, 1:3,000, CST), JunB (#3753, 1:3,000, CST), β-actin (#8457, 1:3,000, CST), Myc tag (M192-3, 1:3,000, Medical & Biological Laboratories, Tokyo, Japan), Cyclophilin C-associated protein (CyCAP)/mouse LGALS3BP (#28128, 1:1,000, Immuno-Biological Lab, Japan, Fujioka, Japan), TGF-β1 (ab215715, 1:3,000, Abcam), Integrin subunit alpha V (ITGaV, #60896, 1:3,000, CST), Phosphorylated focal adhesion kinase (FAK) (#3283, 1:3,000, CST), FAK (#3285, 1:3,000, CST) and vascular cell adhesion molecule 1 (VCAM1, ab134047, 1:3,000, Abcam).

2.7 Hematoxylin and eosin (H&E), Sirius Red, Masson's trichrome (MT), and immunohistochemistry staining

Paraffin-embedded mouse liver tissues were sectioned (4-µm thick) and subjected to H&E (BBC Biochemical, Mount Vernon, WA, USA), Sirius Red (ab150681, Abcam), MT (ab150686, Abcam), and immunohistochemical staining with alpha-smooth muscle actin (α-SMA) antibodies (ab5694, 1:1,000, Abcam, Waltham, Boston, USA), TGF-β1 (ab215715, 1:1,000, Abcam), JunB (#3753, 1:3,000, CST), Ki-67 (ab16667, 1:3,000, Abcam) and a secondary antibody (ab6721, 1:1,000, Abcam). To ensure the specificity of staining, negative controls were prepared by incubating the sections with the secondary antibody alone. Additionally, hydrogen peroxidase and BSA were applied to block endogenous peroxidase activity and prevent non-specific binding. Bright-field images of histological sections were captured using a Zeiss Axio Scan Z1 slide scanner (Carl Zeiss, Baden-Württemberg, Germany). The quantification of the intensities of Sirius Red, MT, α-SMA, and TGF-β1 positively stained areas was carried out using Fiji software (version 2.14.0; National Institutes of Health, Bethesda, MD, USA) by setting and adjusting intensity thresholds based on control samples to ensure consistent detection and measurement of the mean pixel intensity. Positive nuclei for JunB and Ki-67 staining were counted and quantified using Fiji software. The intensity threshold was set to differentiate positive nuclei from the background, and the mean intensity of pixels above this threshold was measured.

2.8 Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

Total RNA from human tumoral or peri-tumoral normal tissues of CNUHH patients with HCC, mouse liver tissues, cultured mouse primary hepatocytes, or Hepa-1c1c7, SNU449, Hep3B, and HepG2 cell lines were isolated using a Ribospin II Kit (314-103, GeneAll Biotechnology, Seoul, South Korea) and reverse-transcribed to cDNA using a GoScript reverse transcriptase kit (A5001, Invitrogen, Carlsbad, CA, USA). qRT-PCR was performed using SYBR Green PCR Master Mix (PB20, PCR Biosystems, PA, USA) and the appropriate primers (Supplementary Table S3) with cDNAs using the CFX Connect Real-Time PCR Detection System (Bio-Rad Hercules, CA, USA). The thermocycling conditions were set as follows: 95°C for 3 min, followed by 40 cycles of amplification at 95°C for 10 s, 60°C for 10 s and 72°C for 30 s. The reaction was completed by determining the dissociation curve of all amplicons.

2.9 RNA-sequencing (RNA-seq) analysis

Total RNA was extracted from cultured mouse primary hepatocytes from LGALS3BP-KOand its control mice or Hepa-1c1c7 cells treated with vehicle or 5 µg rLGALS3BP (5608-GA; R&D Systems, Minneapolis, MN, USA) using a Ribospin II Kit (Geneall Biotechnology, Seoul, South Korea). Total RNA (1 µg) was used to prepare an mRNA sequencing library using the MGIEasy RNA Directional Library Prep Kit (MGI Tech, San Jose, CA, USA). The synthesized DNA-nanoball library was quantified using a QuantiFluor DNA System (Promega, Madison, WI, USA). Sequencing was conducted using the MGIseq-2000 system (MGI Tech) with paired-end 100-base pair reads. Adapter sequences and low-quality bases in the raw reads were trimmed using Skewer (Version 0.2.2, https://sourceforge.net/projects/skewer). The remaining high-quality reads were mapped to the reference genome using STAR (2.5). Differentially expressed genes (DEGs) were defined as those showing a >2-fold change between control hepatocytes and LGALS3BP-KO primary hepatocytes and a >1.5-fold change between vehicle-treated and rLGALS3BP-treated Hepa-1c1c7 cells. Volcano plots were constructed using the DEGs with Prizm software, and Reactome and Bioplanet biological pathways and transcription factor, protein-protein interaction (PPI) network were predicted using the Enrichr website [35]. RNA-seq data from Hepa-1c1c7 or primary hepatocytes were deposited in the GEO database under accession codes GSE271352 and GSE271355, including all raw and processed data.

2.10 Short-interfering RNA (siRNA) experiments

siRNA-mediated knockdown experiments were performed with two independent mouse JUNB, mouse LGALS3BP, or human LGALS3BP siRNAs (Bioneer, Daejeon, South Korea; Supplementary Table S4) using Lipofectamine RNAi-MAX (Thermo Fisher Scientific, MA, USA). Following transfection, the cells were incubated for 24 h, serum-starved for 16 h, and treated with 5 µg/mL of rLGALS3BP and/or 0.5 ng/mL of GLPG0187. The total RNA and whole-cell extracts were prepared for qRT-PCR and western blotting, respectively.

2.11 Assay for transposase-accessible chromatin with sequencing (ATAC-seq)

ATAC-seq was performed with Hepa-1c1c7 cells treated with vehicle or 5 µg/mL rLGALS3BP using an ATAC-seq Kit (C01080002, Diagenode, NJ, USA). Briefly, the cells were lysed and genomic DNA was transposed simultaneously in a transposition mix. After tagmentation, the transposed DNA fragments were purified and amplified using PCR. The final libraries were purified using AmPure XP Magnetic Beads (Beckman Coulter, Brea, CA, USA), quantified using a Qubit 2.0 Fluorometer (Thermo Fisher Scientific), and sequenced using the NovaSeq platform (Illumina, CA, USA). Potential sequence adapters and low-quality reads were trimmed using the Skewer software. The cleaned quality reads were mapped to the reference genome using bowtie2 (2.3.2). We performed strand cross-correlation analysis to check the suitability of ATAC-seq data for further analysis. MACS2 (2.1) with ‘default’ or ‘broad’ parameters was used to identify the significantly enriched regions. Motif enrichment in peaks was performed using findMotifGenome.pl in Homer (4.9) [36]. ATAC-seq data were deposited in the GEO database under the accession code GSE271354, including all raw and processed data.

2.12 Activator protein-1 (AP-1) reporter assays

Hepa-1c1c7, SNU449, or HepG2 cells were seeded to transiently transfect AP-1 reporter (631906, Clontech, CA, USA) in 96-well plates using Lipofectamine 3000 (Thermo Fisher Scientific). Culture media was changed into serum free medium 20 h following transfection and then treated with increased amounts of rLGALS3BP. Cells were subsequently washed with PBS, lysed for Dual luciferase assay (E1910, Promega, MI, USA), and reporter activities were measured by luminometer (SpectraMax, i3X, Molecular devises, CA, USA).

2.13 Chromatin immunoprecipitation (ChIP) analysis

ChIP assays to validate JunB binding at TGFB1 promoter were performed using Hepa-1c1c7 cell lysates and an EZ-ChIP Kit (17-371, Merck-Millipore, Burlington, MA, USA) with α-JunB (#3753, CST) or normal rabbit IgG (#2729, CST). Briefly, formaldehyde-cross-linked cells were lysed and sonicated to prepare genomic DNA fragments (200-500 bp). For each assay, 4 µg of antibody was used to immunoprecipitate 40 µg of input chromatin. Immune complexes were recovered using 30 µL of resuspended protein A/G agarose. After washing, the DNA was de-crosslinked and eluted, as described in the manufacturer's instructions. qPCR was performed in a 7500 Fast Thermal Cycler (Life Technologies, Carlsbad, CA, USA), using the ChIP primers listed in Supplementary Table S3.

2.14 Liquid chromatography-tandem mass spectrometry (LC-MS/MS)

Whole-cell extracts of primary hepatocytes isolated from LGALS3BP-KI mice were prepared to identify the PPI network using Myc-tagged LGALS3BP. One-milliliter pellets of primary hepatocytes were sonicated in 2.5 mL lysis buffer (25 mmol/L Tris-Cl pH 7.4, 150 mmol/L KCl, 1 mmol/L Ethylenediamine tetraacetic acid, 0.5% NP40, protease inhibitors) and immunoprecipitated. Following incubation with agarose beads or anti-Myc-conjugated agarose beads at 4°C for 2 h, the pellets were washed three times with lysis buffer and twice with phosphate-buffered saline. The precipitated proteins were digested and reconstituted with 0.1% formic acid in LC-MS grade water and analyzed using a Dionex UltiMate 3000 nano LC system connected to an Orbitrap Fusion Lumos MS equipped with an EASY-Spray ion source (Thermo Fisher Scientific).

2.15 Co-immunoprecipitation assay

Hepa-1c1c7 and SNU449 cells were transfected with a vector encoding Myc-tagged LGALS3BP or the parental vector, cultured for 40 h, and incubated in serum-free medium for 16 h. Ten milliliters of conditioned medium from each plate was incubated with anti-Myc-conjugated agarose beads, washed, and then incubated with Hepa-1c1c7 whole-cell extracts prepared with Pierce IP lysis buffer (#87787, Thermo Fisher Scientific). Following incubation at 4°C for 2 h, the beads were washed twice with Pierce IP lysis buffer (25 mmol/L Tris-Cl, pH 7.4, 150 mmol/L KCl, 1 mmol/L Ethylenediaminetetraacetic acid, 0.5% NP40, and protease inhibitors) and once with lysis buffer containing 300 mmol/L NaCl. Bead-bound proteins were resolved by SDS-PAGE followed by western blotting with antibodies against ITGαV (#60896, 1:3,000), ITGα5 (#4705, 1:3,000), and ITGβ1 (#34971, 1:3,000) from CST or an anti-Myc tag antibody (M192-3) from MBL with indicated dilutions.

2.16 Immunofluorescence staining

Hepa-1c1c7, HepG2, and SNU449 cells were subjected to immunofluorescence (IF) staining. Briefly, cells were fixed with 4% paraformaldehyde for 10 min, followed by permeabilization with 0.1% Triton X-100 for 15 min. After rinsing the samples with 1x PBS, a solution of 1% bovine serum albumin and 0.5% Tween-20 was applied to block non-specific antibody-binding sites. Anti-p-FAK (PA5-85602, Invitrogen), Phalloidin-conjugated Alexa Fluor 647 antibody (A22287, Invitrogen), and 4′,6-diamidino-2-phenylindole (DAPI) were used to label the phosphorylated FAK, filamentous actin (F-actin), and nuclei, respectively. To detect the ITGαV accumulation by LGALS3BP overexpression, HepG2 cells were seeded on poly-L-lysine (P4707, Sigma-Aldrich) coated eight well chambers (154534, Thermo Fisher Scientific). After 24 h, when cells achieved 70%–80% confluence, transfections were carried out with Lipofectamine 3000 (Thermo Fisher Scientific) and pcDNA6/myc-LGALS3BP (0.5ug/wells). Cells were fixed with 4% paraformaldehyde, and permeabilized with 0.1% Triton X-100. Myc-tagged LGALS3BP was detected using a mouse Myc-tag antibody (M192-3, MBL) followed by Alexa Fluor 488-conjugated secondary antibody (A32723, Invitrogen). ITGαV (ab179475, Abcam) was detected using rabbit anti-ITGaV antibody followed by Alexa Fluor 647-conjugated secondary antibody (A32733, Invitrogen). For Proximity Ligation Assay (PLA), HepG2 cells were transiently transfected with myc-LGALS3BP expressing vectors on coated eight well chambers (154534, Thermo Fisher Scientific). After 24 h, cells were permeabilized and blocked using a PLA kit (DUO92101, Sigma-Aldrich, St. Louis, MO) according to the manufacturer's instructions. Primary antibodies were incubated with cells for 2 h at 37°C and then washed and incubated with PLA probes. A mounting medium with DAPI (Sigma-Aldrich, St. Louis, MO) was added and cells were visualized under fluorescence with Texas Red filter (excitation 594 nm, emission 624 nm) with a Zeiss LSM 780 laser-scanning confocal microscope. Confocal images were acquired using an Olympus FLUOVIEW FV1000 confocal microscope and quantified using Image J Fiji software (version 2.14.0) and Zen blue 3.1 software (Zeiss, Oberkochen, Germany).

2.17 Triglyceride quantification

Triglyceride levels were measured in 30 mg homogenized liver tissue or mouse plasma samples using a triglyceride assay kit (ab65336, Abcam) according to the manufacturer's instructions.

2.18 Oil Red O staining

For frozen sections, the liver tissues were fixed with 4% paraformaldehyde overnight and then infiltrated with 30% sucrose. The tissues were embedded in OCT compound (Tissue-Tek, Sakura Finetek USA, Torrance, CA) and stored at −80°C until analysis. Frozen tissues were sectioned at 6 µm using a cryostat (CM1860, Leica Biosystems, IL, USA). The sections were then dried, fixed in 10% formalin, and washed with 60% isopropanol. The slides were stained with Oil Red O (O0625, Sigma-Aldrich) working solution for 15 min. The slides were counterstained with hematoxylin and washed with distilled water. Positively stained areas were quantified using Fiji software (version 2.14.0; National Institutes of Health, Bethesda, MD, USA).

2.19 Statistical analysis

All statistical analyses were performed using GraphPad Prism10 (Graph Pad, San Diego, CA, USA) and values are expressed as the mean ± standard deviation from three independent experiments, unless otherwise indicated. Student's t-test was used to determine the statistical significance between two groups, and two-way analysis of variance (ANOVA) for comparing more than two groups. P < 0.05 was considered significant.

3.1 Correlation of TGFB1 expression with LGALS3BP expression in HCC patients

To investigate the role of LGALS3BP in hepatic carcinogenesis, gene sets positively correlated with LGALS3BP expression in TCGA HCC dataset were analyzed using Reactome pathway analysis. As the regulatory role of LGALS3BP in innate immunity and cell-to-cell interactions has been studied, the immune system, signal transduction, and ECM organization genes were found to be significantly co-regulated with LGALS3BP expression (Figure 1A and Supplementary Table S5). As matrix metallopeptidase 2 (MMP2), MMP9, collagen type I alpha 2 (COL1A2), and TGFB1 were common to the pathways (Figure 1B), and MMP and collagen genes are regulated by TGF-β1 signaling [37-39], we hypothesized that LGALS3BP expression was closely associated with TGFB1 expression.

We examined the correlation between LGALS3BP and TGF-β family members TGFB1, TGFB2 and TGFB3 in primary liver cancer tissues from 83 patients who underwent surgery for HCC. As LGALS3BP expression exhibited strong positive correlations with TGFB1 in TCGA HCC data, LGALS3BP expression in HCC tumors at CNUHH also exhibited strong positive correlation with TGFB1. LGALS3BP had relatively weaker positive correlations with TGFB2 and TGFB3 (Figure 1C-E), suggesting its specialized role in TGFB1 regulation.

3.2 IFN-α induced LGALS3BP for TGFB1 expression

Treatment with IFN-α, a critical regulator of LGALS3BP expression, increased TGFB1 mRNA levels in rat preneoplastic liver cells, and IFN-α and TGF-β1 cooperated to induce liver fibrosis [40, 41]. To determine whether LGALS3BP regulates TGFB1 in this context, the effects of IFN-α on murine hepatic carcinoma Hepa-1c1c7 cells were tested. The rIFN-α treatment induced the expression of LGALS3BP and interferon regulatory factor 7 (IRF7) significantly, along with that of TGFB1 (Supplementary Figure S1A), and the knockdown of LGALS3BP using two separate siRNA constructs blocked IFN-α-dependent TGFB1 expression significantly (Supplementary Figure S1B). Notably, the mRNA expression of IFN-α-induced genes (IRF7 and MX2) and TGF-β1 target genes (VCAM1 and COL1A1) were positively associated with LGALS3BP expression in the tumoral and peri-tumoral normal tissues of HCC patients (Supplementary Figure S2).

3.3 Effect of hepatic LGALS3BP overexpression on TGFB1 upregulation and exacerbation of CCl₄-induced hepatic fibrosis

To investigate the role of LGALS3BP in TGFB1 regulation in vivo, we generated a conditional knockin (KI) mouse model wherein hepatocytes specifically overexpressed LGALS3BP (Figure 2A). LGAL3BP was exclusively expressed in the liver tissues of transgenic mice when mated with Alb-Cre mice (Figure 2B). Primary hepatocytes isolated from LGALS3BP-KI mice exhibited significantly elevated TGFB1 levels along with LGALS3BP overexpression (Supplementary Figure S3A). Known TGF-β1-target genes, such as VCAM1, LIF, PDGFB, and COL1A1, were induced in LGALS3BP-KI primary hepatocytes (Supplementary Figure S3A). Culture medium and whole-cell extracts from primary LGALS3BP-KI hepatocytes showed the induced LGALS3BP protein secretion and elevated pro-TGF-β1 levels (Supplementary Figure S3B).

We used a CCl₄-induced model of hepatic fibrosis to demonstrate the role of LGALS3BP in hepatic fibrosis in vivo (Figure 2C) as dysregulated TGF-β1 signaling regulates the progression of the disease [26]. Hepatocyte-specific overexpression in LGALS3BP-KI mice resulted in higher circulating serum LGALS3BP levels than in their control littermates (Figure 2D). The liver tissues of CCl₄-treated mice exhibited a distorted structure, with micronodules noted throughout the liver parenchyma along with fibrosis symptoms, as determined by measuring the accumulation of collagen fibers and α-SMA. The LGALS3BP-KI mice displayed a more pronounced fibrosis profile in their livers than in those of WT mice (Figure 2E). We found that the expression of TGF-β1 was significantly higher in the livers of LGALS3BP-KI mice than in those of control mice following intraperitoneal CCl₄ injection (Figure 2F) along with the expression of TGF-β1 target genes (Figure 2G).

3.4 Induction of TGFB1 and its target gene expression by rLGALS3BP treatment

The fibrosis-promoting role of LGALS3BP prompted us to investigate whether secreted LGALS3BP directly regulates TGFB1 expression. Surprisingly, treatment with rLGALS3BP alone was sufficient to stimulate TGFB1 expression and its canonical downstream signaling mediator, Smad2 phosphorylation, in Hepa-1c1c7 cells (Figure 3A-B). Compared with 5 µg/mL, 20 µg/mL of rLGALS3BP did not enhance the TGFB1 induction probably due to the induced SMAD6, a negative regulator for SMAD signaling at this condition (Supplementary Figure S4).

We examined the impact of gene expression in Hepa-1c1c7 cells using RNA-seq following exposure to rLGALS3BP (Figure 3C). Pathway analysis of the upregulated gene sets revealed the activation of multiple signaling pathways by inflammatory cytokines, including TNF-α, interleukins, and TGF-β1 (Figure 3D and Supplementary Table S6). Genes downregulated by rLGALS3BP treatment indicated that the metabolism of free fatty acids, such as the regulation of the pyruvate dehydrogenase complex and import of palmitoyl-coenzyme A into the mitochondrial matrix, was decreased (Figure 3E and Supplementary Table S7). The genes associated with the regulation of TGF-β on ECM was increased significantly, including known targets of TGF-β1, such as SERPINE1, VCAM1, and IL6 (Figure 3F), along with the upregulation of another TGF-β1 target gene, PDGFB (Figure 3G).

3.5 Regulation of TGFB1 expression by JunB with rLGALS3BP treatment

To define the underlying mechanisms via which TGFB1 expression is controlled directly, the alternation of transcription factor binding upon rLGALS3BP treatment was analyzed using ATAC-seq. Differentially accessible regions were curated for motif analysis, and one distinct transcription factor that exhibited changes in accessibility was JunB (Figure 4A), which has been studied for its role in regulating TGFB1 expression [42-45]. Transactivation of AP-1 luciferase reporter activated by transcription factors including JunB-binding sites was induced by rLGALS3BP treatment in various HCC cell lines (Figure 4B). Notably, the chromatin-accessible region around the TGFB1 locus contained three JunB-response sequences (Figure 4C). The rLGALS3BP-inducible JunB binding to ATAC peak regions was validated independently by JunB ChIP-qPCR (Figure 4D). JUNB knockdown using siRNA abolished the rLGALS3BP-promoted TGF-β1 expression completely (Figure 4E). Primary hepatocytes isolated from LGALS3BP-KI cells showed higher JUNB expression than those from the controls (Figure 4F). Using TCGA HCC data and CNUHH HCC patient tissues, a positive correlation between JUNB and LGALS3BP expression was identified (Figure 4G-I).

3.6 Interaction of LGALS3BP with ITGαV and promotion of ITGαV accumulation on the cell membrane to release TGF-β1

We observed an unexpected trend in TGFB1 mRNA expression and the amount of secreted total TGF-β1 in the culture medium upon rLGALS3BP treatment. The released active form of TGF-β1 peaked at 3 h, and even secreted total TGF-β1 levels were not altered, whereas TGFB1 mRNA expression increased gradually over time (Figure 5A, Supplementary Figure S4).

LC-MS/MS proteomics was conducted to gain insight into the mechanism of TGF-β1 release (Figure 5B). Interestingly, one network (cluster 4) interacting with LGALS3BP contained ITGαV among the 30 LGALS3BP-interacting protein clusters, and the enhanced network revealed the physical interactions between ITGαV and TGF-β1 (Figure 5C and Supplementary Figure S5A-B). The physical interaction was validated by an immunoprecipitation assay using the isolated LGALS3BP from the culture medium and ITGαV in SNU449 and Hepa-1c1c7 cell extracts (Figure 5D and Supplementary Figure S5C).

To determine the role of ITGαV in LGALS3BP-mediated TGF-β1 activation, we utilized GLPG0187, an ITGαV inhibitor that has been studied as a drug that suppresses the growth of various metastatic tumors [46] and more recently SARS-CoV-2 delta infections [46, 47]. GLPG0187 effectively inhibited the release of active TGF-β1 triggered by LGALS3BP in HCC cells, (Figure 5E). Physical interaction between transiently transfected LGALS3BP and ITGαV was observed (Figure 5F), and ectopic expression of LGALS3BP led to the ITGαV accumulation (Figure 5G). Upon rLGALS3BP treatment, ITGαV on the cell membrane was accumulated, and the number of filamentous actins (F-actin) in the cytosolic region was increased significantly (Figure 5H-I). The inhibition of ITGαV caused the reduction of the increased JunB binding to the promoter region of TGFB1 by rLGALS3BP treatment (Figure 5J). FAK is a primary intracellular downstream mediator that delivers signals from ITGαV [48, 49]. Notably, when Hepa-1c1c7 cells were treated with Defactinib, a chemical inhibitor of FAK, free TGF-β1 levels was significantly suppressed (Figure 5K)

3.7 JunB-TGF-β1 feedback loop initiation by LGALS3BP via the activation of ITGαV-mediated rearrangement of F-actin cytoskeleton

The release of active TGF-β1 from its latent complex requires the application of tensile force, which is generated by the formation of a stiff cytoskeleton modulated by ITGαV-FAK signaling [28-30, 50]. Interestingly, rLGALS3BP treatment led to the increased FAK phosphorylation and the subsequent rearrangement of F-actin cytoskeleton in Hepa-1c1c7 cells (Figure 6A). GLPG0187 was found to efficiently inhibit the activation of p-FAK and F-actin stretching(Figure 6A). Inhibition of ITGαV resulted in decreased TGFB1 and JUNB mRNA expression (Figure 6B). When the cells were treated with rLGALS3BP, expression levels of ITGαV, JunB, and cytosolic pro-TGF-β1 were significantly induced; the elevated protein levels along with increased phosphorylated FAK were blocked by ITGαV inhibition (Figure 6C). The rearrangement of the cytoskeleton and subsequent gene regulation dependent on ITGαV were also observed in human HCC SNU449 cells upon treatment with rLGALS3BP (Figure 6D-F). Collectively, LGALS3BP triggered active TGF-β1 releasement from its latent complex through ITGαV-FAK-mediated stiff F-actin cytoskeleton which is critical for the tensile force generation. rTGF-β1 treatment promoted the expression of JUNB and TGFB1 (Supplementary Figure S6A), suggesting the positive feedback loop by released TGF-β1. Since TGF-β1 transcriptionally promoted LGALS3BP expression in rat thyroid FRTL-5 cells [51], we also tested whether rTGF-β1 treatment increased LGALS3BP expression. Although rLGALS3BP or rTGF-β1 treatment promoted TGFB1 expression in Hepa-1c1c7 cells, LGALS3BP levels were not changed at these conditions (Supplementary Figure S6B).

3.8 Downregulation of TGF-β1 signaling and CCl₄-induced fibrosis by LGALS3BP depletion

The observation that the JunB-TGF-β1 axis was directly regulated by secreted LGALS3BP prompted us to examine the effect of LGALS3BP depletion on the regulation of TGF-β1 signaling in the livers using LGALS3BP KO mice (Supplementary Figure S7). As an initial step, we analyzed the expression of altered genes in the isolated primary hepatocytes. DEG analyses between control and LGALS3BP-KO primary hepatocytes from RNA-seq analysis revealed significant downregulation of TGF-β1 and its target genes (Figure 7A). Upregulated genes were associated with multiple metabolic processes, including phenylalanine and tyrosine amino acid metabolism (Figure 7B and Supplementary Table S8). Notably, Reactome gene sets with downregulated genes showed decreased non-ITG membrane and ECM interactions and organization genes (Figure 7C and Supplementary Table S9). Moreover, the upregulated gene sets of Bioplanet pathways in rLGALS3BP-treated cells (Figure 3D) was reversed by the depletion of LGALS3BP (Figure 7C). TGF-β signaling mediators, Smad2/3, and its active transcriptional partner Smad4 were predicted to be factors associated with decreased gene sets (Supplementary Figure S8A), and TGF-β1 target genes, VCAM1, PDGFB, LIF, and COL1A1, were downregulated along with TGFB1 (Supplementary Figure S8B).

We further examined a CCl₄-induced fibrosis mouse model to demonstrate the role of LGALS3BP depletion in vivo (Figure 7D). As expected, induced expression levels of TGFB1 by hepatic fibrosis were reduced by depleting LGALS3BP along with JUNB and COL1A1 significantly (Figure 7E). The livers of LGALS3BP-KO mice demonstrated reduced fibrosis compared to those of WT mice, as evidenced by decreased collagen fiber deposition and α-SMA expression (Figure 7F). TGF-β1, ITGαV, p-FAK, and JunB levels were significantly reduced in LGALS3BP-KO mouse livers, (Figure 7G and supplementary Figure S9) as opposed to the stimulation seen in LGALS3BP-KI mouse livers (Supplementary Figure S3C).

3.9 Reduction of the steatohepatitis-induced hepatocarcinogenesis by LGALS3BP depletion

We found that MASH severity in human patients was associated with increased LGALS3BP along with TGFB1 and COL1A1 expression, and LGALS3BP and TGFB1 levels were positively correlated in the patients [52] (GSE135251, Supplementary Figure S10A-C). In fact, LGALS3BP expression was elevated significantly along with TGFB1 in the livers of mice fed a HFMCD diet (Supplementary Figure S10D-E), which is a mouse model for studying MASH [53, 54].

As uncontrolled chronic inflammation-induced hepatic fibrosis is closely associated with the development of hepatic carcinogenesis [32], we investigated whether LGAL3BP depletion ameliorates DEN- and HFD-induced hepatic carcinogenesis via regulating TGF-β1 signaling (Figure 8A). The livers of LGALS3BP-KO mice showed a noticeable decrease in tumor score, despite having body weight and lipid contents similar to WT mice (Figure 8B-C and Supplementary Figure S11). We found that active TGF-β1 was not detectable in the plasma probably due to its short half-life of 2-3 min [55]. However, circulating total TGF-β1 levels were significantly less in the DEN+HFD KO mice (Figure 8D). Serum levels of LGALS3BP were positively correlated with tumor scores (Figure 8E), and TGFB1 and LGALS3BP expression was correlated positively in the liver tumors of WT mice (Figure 8F). Furthermore, the accumulation of collagen and α-SMA around the tumor nodules decreased in the KO mouse livers (Figure 8G). Levels of TGFB1 and its associated genes were decreased in LGALS3BP-KO mouse liver tumors (Figure 8H-I). TGF-β1-stained area near the tumor sites, JunB-positive nuclei, and the proliferation of tumor cells were all decreased (Figure 8J). LGALS3BP was required for the expression of TGFB1 and its target genes, including JUNB, PDGFB, VCAM1, SERPINE1, and COL1A1 in various HCC cell lines (Supplementary Figure S12).