Introduction
Liver fibrosis is caused by various chronic liver diseases, can progress to cirrhosis and hepatocellular carcinoma, and it is associated with significant morbidity and mortality.1,2 However, the molecular mechanisms underlying the condition are complex and not fully understood. Apart from etiological treatments, there are no direct antiherpetic fibrosis therapeutics, thus additional research is required.
Liver fibrosis characterized by the accumulation of proteins in the extracellular matrix (ECM) following the activation and proliferation of hepatic stellate cells (HSCs).3 Changes in collagen are the most important ECM changes associated with fibrogenesis and fibrinolysis.4 Many post-translational modifications occur during procollagen biosynthesis in the rough endoplasmic reticulum (rER), facilitating proper collagen folding, secretion, and biological functions.5 Of these modifications, proline and lysine hydroxylation are well-recognized and studied.6,7 However, collagen glycosylation has not been extensively investigated.
In 1935, Grassmann and Schleich first described collagen glycosylation. Approximately 30 years later, Spiro described the glycan structure: Glc(α1-2) Gal(β1-O) Hyl.8 The corresponding glycosyltransferase enzymes collagen β(1-O) galactosyltransferase 1 (GLT25D1) and collagen β(1-O) galactosyltransferase 2 (GLT25D2) were identified by Schegg et al. in 2009.9 GLT25D1 adds glucosylgalactosyl-hydroxylysine (GG-Hyl) or galactosyl-hydroxylysine (G-Hyl) to specific procollagen hydroxylysine molecules prior to the formation of the triple helix structure in the endoplasmic reticulum (ER).9–11 GLT25D1 regulates cross-linking in bone collagen I.12 Knocking out GLT25D1 in osteosarcoma SaOS-2 cells impaired cell proliferation and viability and induced the upregulation and intracellular accumulation of collagen I.13 In MC3T3-E1 preosteoblast cells, loss of GLT25D1 affects the maturation of collagen cross-linking, fibrillogenesis, and mineralization of collagen fibrils.14 In N-ethyl-N-nitrosourea (ENU)-induced mutant mice carrying a GLT25D1 mutation, collagen IV accumulates inside embryonic fibroblasts and within the ECM.15 GLT25D1 also decreases adiponectin secretion in early obesity and is associated with cerebral small vessel disease through its effect on COL4A1,16,17 indicating that it has a vital role in collagen fibrillogenesis, cross-linking, mineralization, and collagen-cell interaction.12
In light of the findings, we hypothesized that GLT25D1 may play an essential role in activating HSCs, the primary collagen-producing cells associated with liver fibrosis. We used GLT25D1-knockout (KO) mice (C57BL/6J) to investigate this role. However, knocking out GLT25D1 was lethal to the embryos.18 Therefore, heterozygous GLT25D1+/− mice, obtained using Cre-loxP conditional knockout technique, were used. Modeling studies suggest that GLT25D1 may play a role in acute hepatic injury and liver fibrosis.18,19 GLT25D2 deficiency contributes to lipodystrophy and promotes NAFLD.20 We previously showed that GLT25D1 has an essential role in liver disease, although its role and underlying mechanism in liver fibrosis are still poorly understood.
In this study, we demonstrated the role of GLT25D1 in primary mouse HSCs and LX-2 cells for the first time. We found that GLT25D1 was upregulated in activated HSCs and played an essential role in HSC activation. GLT25D1 also potentially affected collagen stability. The results suggest that GLT25D1 is a novel and potential antifibrotic target.
Methods
Human liver samples
Liver samples were collected from eight patients with chronic hepatitis B who underwent hepatic surgery or liver puncture biopsy at Beijing Ditan Hospital between June and November 2020. Three of the patients had mild chronic portal inflammation (S0–S1, control) and five patients had cirrhosis. All biopsy specimens were histologically scored via blinded assessment using the Ishak system.21 Patient baseline information is shown in Supplementary Table 1.
Bile duct ligation (BLD)-induced liver fibrosis
The experimental Animal Welfare Committee of Capital Medical University approved all animal procedures. We used GLT25D1+/− mice, which were generated as previously described.18,19 6–8-week-old male C57BL/6J wild-type (WT) and GLT25D1+/− mice were anesthetized via intraperitoneal injection with 0.8% pentobarbital. After a scalpel midline laparotomy, the liver lobules were turned down and the intestines exteriorized carefully to expose the bile duct. The common bile duct was ligated twice using 6–0 silk sutures and the abdomen was then closed. A sham operation was performed in a similar manner, except that the bile duct was not ligated. The animals were sacrificed on postoperative day 21 and the plasma and liver samples were collected for further experiments.
Hepatic hydroxyproline assay
The total collagen content was assessed by measuring hydroxyproline levels in liver tissue using the kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China).
Insoluble collagen content assay
The concentration of insoluble collagen (cross-linked collagen) in liver tissue was measured using Biocolor Sircol insoluble collagen assay kits (Biocolor Ltd., Carrickfergus, UK).22 Briefly, 20–30 mg (wet weight) of liver samples were crushed in the fragmentation reagent, incubated at 65°C for 2–3 h, and the supernatant was collected and stained for 30 m using Sircol dye reagent. An acid salt wash was used to remove unbound dye and an alkali reagent was then added to release the collagen-bound dye into the solution using a vortex mixer. Absorbance was measured at 550 nm within 2–3 h of the reaction.
Isolation of primary mouse HSCs
Primary mouse hepatic stellate cells (mHSCs) were isolated from 8–10-week-old male C57BL/6J WT and GLT25D1+/− mice by enzymatic digestion and OptiPrep density gradient centrifugation.23 Briefly, each liver was first perfused in situ through the hepatic portal vein using pronase, DNase I, and type IV collagenase. The liver was then excised, homogenized, and digested in external digestive fluid. The mixture was filtered through a 100 µm mesh and centrifuged at 450×g for 7 m. The supernatant was discarded and the cell pellet resuspended in 5 mL of 15% OptiPrep (Sigma-Aldrich, St Louis, MO, USA) and slowly loaded with 5 mL of 11.5% OptiPrep and then 5 mL Gey’s balanced salt solution (GBSS). After centrifuging at 1,400×g and 4 °C for 17 m, the mHSC layer (white flocculus) between the 11.5% OptiPrep and the GBSS was transferred into a new tube, mixed with GBSS, and centrifuged at 450×g for 7 m. The cell pellet was resuspended in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Primary hepatic stellate cells from two mice of the same group were pooled as one sample. The purity of primary HSCs was determined by adding Oil Red O staining to freshly isolated primary cells that had been cultured for 2 days (>97%).24 HSC viability was determined using the Trypan blue exclusion test (>90%). Isolated mHSCs were cultured in DMEM supplemented with 10% FBS at 37°C in a 5% CO2 atmosphere. Activated mHSCs were obtained 7 days after the initial culture.
Immunohistochemistry (IHC) and Immunofluorescence (IF) assays
Human and mouse liver tissues were fixed in 10% neutral buffered formalin and embedded in paraffin and then divided into 4 µm thick sections. Immunohistochemistry assays were conducted using the liver tissue sections. Tissue slides were deparaffinized, hydrated, incubated with 10% H2O2 for 10 m and then steamed in 0.01 mol/L citrate sodium buffer for 30 m. The reaction was blocked by adding 2% bovine serum albumin (BSA) and incubating for 1 h at room temperature. After blocking, a 1:100 dilution of anti-rabbit GLT25D1 antibody was added and the reaction incubated overnight at 4°C. The following day, the slides were washed three times with phosphate buffered saline (PBS) and then incubated with goat anti-rabbit antibody for 1 h at room temperature. Immunoreactive signals were visualized for 5 m using 3,3′-diaminobenzidine.
The cells were grown on 25 mm sterile cell climbing slices for 7 days and then fixed in 4% paraformaldehyde for 15 m at room temperature. After washing three times with PBS, the cells were incubated with 5% BSA for 1 h and then with primary antibody (mouse anti-collagen type I 1:200; Servicebio, Hubei, China) overnight at 4°C. After washing with PBS, the cells were incubated with secondary antibody (CY3; Servicebio, Hubei, China) for 1 h at room temperature. Cell nuclei were stained using diamidino-phenylindole (Biotium, Hayward, CA, USA). Images were acquired using the same microscope parameters (Nikon, Tokyo, Japan) and analyzed using ImageJ software.
Second-harmonic generation/two-photon excited fluorescence (SHG/TPEF) imaging
SHG/TPEF microscopy is a novel optical tissue imaging system that can be used to characterize fibrillar collagen and analyze pathology-relevant collagen architectural features.25,26 Four micron-thick unstained sections of mouse liver tissues were anonymized and imaged at HistoIndex Pte Ltd (Singapore). Four morphological features were investigated: the number of short strings (length ≤20 µm), the number of long strings (length >20 µm), the number of thin strings (axis ratio ≤0.25), and the number of thick strings (axis ratio >0.25).
Generation of plasmid constructs and LX-2 single cell-derived clones
Briefly, the lentiviral expression vectors GLT25D1-Plvx-shRNA2-Zsgreen-T2A-puro (Fig. 1E) and GLT25D1-pCDH-CMV-MCS-EF1-GFP-T2A-Puro (Fig. 2A) were successfully constructed at Generay biotech (Beijing, China) and confirmed by sequence analysis. Lentiviruses containing these vectors were prepared and transfected into LX-2 cells using FuGENE 6 transfection reagent (Roche, Basel, Switzerland). Then, single cell-derived clones stably suppressing GLT25D1 (sh-GLT25D1), overexpressing GLT25D1 (OE-GLT25D1), and corresponding negative controls (NC) were selected by subjecting the cells to 2 µg/mL puromycin treatment for one week. Successful section was confirmed using western blot and Quantitative real-time polymerase chain reaction (qPCR).
RNA sequencing and data analysis
Two biological replicates from each mouse group (WT and GLT25D1+/−) and three biological replicates from each clonal group (OE-NC and OE-GLT25D1) were pooled and sequenced on the Illumina NovaSeq 6000. Differential gene expression between the two groups was analyzed using the DESeq2 R package (1.20.0). padj ≤ 0.05 and |log2 (fold change)| ≥1 were set as thresholds for significant differential expression. The ClusterProfiler R package (3.8.1) was used to conduct gene ontology (GO) enrichment analysis of the differentially expressed genes.
Cell culture and treatment
The human hepatic stellate cell line LX-2 (Institute of Infectious Diseases, Beijing Ditan Hospital) was cultured in DMEM (Gibco, Waltham, MA, USA) with 10% FBS and 1% penicillin/streptomycin (Solarbio, Beijing, CN) at 37°C in a humidified atmosphere containing 5% CO2. The LX-2 cells were starved in serum-free DMEM for 12 h prior to commencing the experiments. Transforming growth factor β1 (TGF-β1, 10 ng/mL), a robust fibrosis mediator, was used to stimulate LX-2 cells to simulate liver fibrosis formation. The control group was treated with the same volume of vehicle. Specific signal pathway inhibitors: SIS3 and SD208 (MedChemExpress, Monmouth Junction, NJ, USA), were added 1 h before adding TGF-β1.
Cell proliferation assay
Cell proliferation was measured using the Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) assay following the manufacturer’s protocol. Briefly, LX-2 stable and control cells were seeded in 96-well plates at 2–5×103 cells per well for 12 h and then induced with 10 ng/mL TGF-β1. Thereafter, 10 µL of CCK8 solution was added to each well. Cell growth curves were generated based on optical density absorbance measured at 450 nm using a Varioskan Flash spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).
Wound healing cell migration assay
A wound healing cell migration assay was used to study directional cell migration in vitro.27 Cells were seeded in 12-well culture dishes and cultured for 24 h. The cells were then treated with 10 ng/mL TGF-β1 and scratched with a 10 µL pipette tip. Images were captured under a microscope immediately following wound creation (0 h) and photos were taken every 24 h for 3 days. Images were analyzed in ImageJ and the percentage wound migration area determined.
Cell contraction assay
A standard assay kit (Cell Biolabs, San Diego, CA, USA) was used. LX-2 cells were harvested and resuspended in DMEM. Two parts of the cell suspension was mixed with eight parts of the collagen gel lattice mixture and plated for 1 h at 37°C. After gel polymerization, 1 mL of the medium was added and the mixture incubated for 2 days. Next, 1 mL medium containing TGF-β1 was added and the gels released from the sides of the wells. Images were obtained after 24 h. The changes in collagen gel sizes were analyzed using ImageJ software and normalized to the area of the well.
RNA extraction and quantitative real-time PCR analysis
Total RNA was extracted from cells or tissues using an animal tissue cell total RNA extraction kit (BioSci, Beijing, China), following the manufacturer’s protocol. cDNA was prepared using a PrimeScript RT reagent kit (Takara, Kyoto, Japan), following the manufacturer’s instructions. qPCR was performed using Fast SYBR Green PCR Master Mix (Applied Biosystems, Waltham, MA, USA). Transcript levels were analyzed using the ΔΔCT method and normalized to that of the internal control gene, glyceraldehyde-3-phosphate dehydrogenase. Each reaction was performed in triplicate and each experiment was performed at least three times. The primers used in this study are listed in Supplementary Table 2.
Protein extraction and western blot assay
Cells or mouse liver tissues were lysed using RIPA buffer supplemented with protease inhibitor (Roche) and phosphatase inhibitor cocktails 2 and 3 (Sigma). Protein concentrations were measured using a bicinchoninic acid assay kit (Thermo Scientific, Waltham, MA, USA). Proteins were separated on a 10% gel and transferred onto 0.2 µm polyvinylidene difluoride membranes (SDS-PAGE; Millipore, Darmstadt, Germany). After blocking with 5% skimmed milk and washing with TBS-Tween buffer, the membranes were incubated overnight at 4°C with the corresponding antibodies (Supplementary Table 3). The membranes were then incubated with a horseradish peroxidase-conjugated secondary antibody (1:5,000) (ZSGB-bio) for 1 h at room temperature and then with an enhanced chemiluminescence substrate (Millipore, Darmstadt, Germany). Relative protein quantities were determined using ImageJ software. Information on the antibodies is presented in Supplementary Table 3.
Mass spectrometry-based identification of glycosylated Hyl residues
Proteins were extracted from lysates of activated sh-GLT25D1, OE-GLT25D1, and control cells. Type I collagen was separated by SDS-PAGE on 10% Bis-Tris precast gels. After staining with Coomassie SimplyBlue, the target protein band was excised from the gel using a scalpel. The bands were cut and diced into 1 mm3 cubes, digested with trypsin, and subjected to mass spectrometry analysis. HPLC-MS/MS analyses were conducted using a RIGOL L-3000 HPLC system (RIGOL, Beijing, China). The MS scan was obtained over a 500−2,000 m/z range with a 2 Hz frequency in positive ion mode. Peptides were identified by searching the acquired MS/MS spectra against Homo sapiens database using the Proteome Discoverer 2.4 software.
Characterization of collagen fibrils using scanning electron microscopy
Specimens were sampled from the same areas of the liver and fixed using an electron microscope fixation liquid (Solarbio, Beijing, China). The samples were then post-fixed in potassium ferrocyanide-reduced osmium for 1 h at room temperature. Collagen fibrils were observed under a scanning electron microscope. The diameters (µm) of collagen fibers were measured in three different fields of view using Image-Pro Plus software (version 6.0; Media Cybernetics Inc., Rockville, MD, USA).
Statistical analysis
Data were reported as percentages (%) or means±SD of at least three independent experiments. Statistical analyses were performed using Prism 9 software (GraphPad, La Jolla, CA, USA). Pairwise comparisons between groups were performed using unpaired Student’s t-test. The χ2 test was used to analyze pathological characteristics. For p-values, <0.05, <0.01, and <0.001 were considered statistically significant.
Results
GLT25D1 protein is upregulated in the cirrhotic liver and present in nonparenchymal cells
We initially investigated GLT25D1 expression in human liver tissue. α-SMA and GLT25D1 levels were significantly elevated in cirrhotic livers compared with the control livers without fibrosis (Fig. 3A–C). GLT25D1 and α-SMA, a marker of activated HSCs in fibrotic livers, were mainly expressed in nonparenchymal cells (Fig. 3A), suggesting a potential role for GLT25D1 in liver fibrosis and HSCs.
GLT25D1 downregulation attenuates collagen deposition
We used heterozygous GLT25D1+/− and WT mice subjected to BDL-induced liver fibrosis to further investigate the in vivo role of GLT25D1 in liver fibrosis. Sirius Red staining and immunohistochemical analysis of α-SMA showed that GLT25D1+/− mice had reduced collagen deposition compared with WT mice (Fig. 4A), in line with decreased hydroxyproline levels (Fig. 4B). The expression of fibrotic genes, including COL1A1, COL3A1, and ACTA2, was significantly downregulated in GLT25D1+/− mice compared with WT mice (Fig. 4C). The GLT25D1 protein levels in liver was provided in the Supplementary Figure 1. The results demonstrate that GLT25D1 knockdown attenuates BDL-induced liver fibrosis in vivo.
GLT25D1 downregulation inhibits primary HSC activation and collagen expression
IHC results from the analysis of human and mouse liver tissues indicated that GLT25D1 was mainly expressed in nonparenchymal cells. HSCs, which are non- parenchymal cells, are the primary collagen-producing cells and key effectors of liver fibrosis. Therefore, based on the hypothesis that GLT25D1 regulates HSCs in liver fibrosis, we isolated primary HSCs from WT and GLT25D1+/− mouse livers. Quiescent mHSCs showed characteristic lipid droplets (Fig. 5A). mHSCs were activated when the cells were cultured for 9 days, resulting in a spindle-shaped or dendritic morphology (Fig. 5B, left panels). mHSC from WT mice showed more obvious stress fiber-like structures compared with GLT25D1+/− mice after 9 days of culture (Fig. 5B, right panels).
To further explore the effects of GLT25D1 on HSCs, RNA-seq analysis of mHSCs from WT and GLT25D1+/− mice was performed after culturing the cells for 12 days. A total of 360 significantly downregulated genes and 22 significantly upregulated genes were identified (Fig. 5C). The expression of fibrosis-related genes was significantly lower in mHSCs from GLT25D1+/− mice compared with those from WT mice, including transforming growth factor beta (TGFβ1), matrix metallopeptidase 9 (MMP9), and MMP12. GO analysis suggested that the genes that were differentially expressed between GLT25D1+/− and control WT mice were associated with cell activation and corresponding downstream profibrogenic responses, including cell chemotaxis and activation (Fig. 5D).28 Immunofluorescence analysis showed that knocking down GLT25D1 reduced intracellular COL I protein levels (Fig. 5E, F). Western blot and RT-PCR results also confirmed that knocking down GLT25D1 in mHSCs significantly decreased α-SMA and intracellular COL I protein levels (Fig. 5G, H). Taken together, these results show that GLT25D1 may be involved in mHSC activation and ECM production.
GLT25D1 knockdown inhibits the activation of LX-2 cells and their corresponding phenotypes
To further assess the effects of GLT25D1 on HSCs, the human hepatic stellate cell line LX-2, which is commonly used in in vitro hepatic fibrosis studies, was employed. Primary HSCs and LX-2 cells share high gene expression similarities (98.7%).29,30 Western blot analysis revealed that the fibrosis-associated proteins COL I, α-SMA, and GLT25D1 protein were upregulated in activated LX-2 cells (Fig. 1A). qRT-PCR also showed that COL1A1, ACTA2, and GLT25D1 mRNA levels were significantly elevated (Fig. 1B–D). The results demonstrate that GLT25D1 expression was upregulated at both mRNA and protein levels following the TGF-β1-induced activation of LX-2 cells. Thus, GLT25D1 may be involved in LX-2 cell activation and ECM production.
Additionally, we constructed a lentivirus vector expressing shGLT25D1 (Fig. 1E) and generated a stable cell line (sh-GLT25D1, Fig. 1F) that successfully knocked down GLT25D1 gene expression (63.2%) than control (Fig. 1G, H). HSC activation is accompanied by specific phenotypes, including proliferation, contractility, and migration.28 The wound healing assay indicated that knocking down GLT25D1 inhibited LX-2’s migratory capacity (p<0.05, Fig. 1I, J). The CCK-8 assay showed that the viability of sh-GLT25D1 cells decreased significantly compared with that of sh-NC cells (Fig. 1K). Collagen gel contraction was used to determine whether GLT25D1 affects HSC contractility. The results showed that knocking down GLT25D1 repressed TGF-β1-induced LX-2 contraction (p<0.05, Fig. 1L), indicating that suppressing GLT25D1 may affect LX-2 cell activation.
GLT25D1 overexpression promotes the activation of LX-2 cells and their corresponding phenotypes
Based on the results, we postulated that GLT25D1 is involved in LX-2 cell activation. Rescue experiments were performed to confirm these results by testing whether GLT25D1 overexpression has opposite effects in LX-2 cells, including increased cell proliferation, contractility, and migration. We constructed a GLT25D1 overexpressing lentivirus vector (Fig. 2A) and generated cell lines that stably overexpress GLT25D1 (OE-GLT25D1, Fig. 2B), showing high expression of GLT25D1 protein (260%) and mRNA (173%) compared with the OE-NC control (Fig 2C, D). The results showed that GLT25D1 overexpression increased migratory capacity and promoted cell proliferation (Fig. 2E–G) in OE-GLT25D1 compared with the control. GLT25D1 overexpression also enhanced LX-2 cell contraction (Fig. 2H). Taken together, these results confirm that GLT25D1 may mediate LX-2 cell activation.
GLT25D1 regulates TGF-β1-induced ECM production by LX-2 cells
To elucidate the effects of GLT25D1 overexpression in LX-2 cells, RNA-seq analysis was conducted in OE-GLT25D1 and OE-NC cells 48 h after their stimulation with TGF-β1. The analysis identified 512 upregulated and 167 downregulated genes (Fig. 6A). The expression of fibrosis-related genes, including COL1A2, COL3A1, ACTA2, TIMP3, PDGFRA, COL11A1, CAV1, and IGFBP3 was significantly upregulated in OE-GLT25D1 cells compared with OE-NC cells. GO enrichment analysis was performed to further explore the biological functions of the DEGs. GLT25D1 greatly affected ECM and collagen formation (Fig. 6B). Western blot and RT-PCR verified the accuracy of the transcriptome data. Western blot results showed that the expression of α-SMA (a marker of LX-2 activation), COL I, COL III, and TIMP-1 were suppressed in sh-GLT25D1 (Fig. 6C-G) but activated in OE-GLT25D1 cells (Fig. 6H-L). Expression of the corresponding liver fibrogenic genes: COL1A2, COL3A1, ACTA2, and TIMP-1, showed similar patterns in sh-GLT25D1 (Fig. 6M–P) and OE-GLT25D1 cells (Fig. 6Q-T) 48 h after they were stimulated with TGF-β1. Taken together, these results indicate that GLT25D1 is required, to some extent, for the activation and expression of fibrogenic genes in LX-2 cells.
GLT25D1 drives ECM production through the TGF-β1/SMAD3 signaling pathway
The canonical TGF-β/SMAD pathway and noncanonical pathway (MAPK, PI3K/AKT) were evaluated to study the mechanisms of action of GLT25D1. GLT25D1 knockdown decreased TGFβ1-stimulated SMAD3 phosphorylation (Fig. 7A), whereas GLT25D1 overexpression enhanced TGFβ1-stimulated SMAD3 phosphorylation (Fig. 7B). There was no significant difference in noncanonical pathway. Thus, we hypothesized that GLT25D1 regulated activation and ECM production of LX-2 cells via the TGF-β1/SMAD3 signaling pathway. Smad3 phosphorylation inhibitor (SIS3) and TGF-βRI (ALK5) inhibitor (SD208) were used to confirm whether GLT25D1 regulated LX-2 activation by arresting the TGF-β/SMAD3 signaling pathway.31 Western blot analysis showed that SIS3 and SD208 significantly suppressed GLT25D1-induced collagen synthesis and completely abrogated collagen upregulation (Fig. 7C). The results suggest that GLT25D1 promotes the expression of fibrogenic proteins during liver fibrosis through the TGF-β1/SMAD3 signaling pathway.
GLT25D1 regulates glycosylation of type I collagen and influences collagen properties
GLT25D1 adds the monosaccharide Gal(β1-O) to procollagen’s hydroxylysine residue. We investigated whether GLT25D1 influenced specific collagen post-translational modification of COL I in mHSCs and LX-2 cells. In the α1 chain isolated from WT mouse-derived mHSCs, the glycopeptides α1(501–541) were shown to contain two hydroxylated Lys: Hyl-527 and Hyl-541. The Hyl-527 residue was glycosylated by the G-Hyl glycoforms (Fig. 8A). However, the G-Hyl and GG-Hyl glycoforms were not observed in the α1 chain isolated from GLT25D1+/− mHSCs. The α1 chain in LX-2 cells contained three glycosylation sites: at residues α1-586, α1-594, and α1-862, while the α2 chain contained one glycosylation site at residue α2-657. However, only the α1 chain was glycosylated at site α1-586 in sh-GLT25D1 cells. Three additional glycosylation sites were observed in the OE-GLT25D1 α1 chains: at residues α1-448, α1-781, and α1-934 (Supplemental Table 4). The study results confirmed that the glycosylation of collagen from HSCs was regulated by GLT25D1.
Post-translational modifications often affect either the molecular weight or the charge of the protein. In this study, the glycosylation changes did not result in an apparent change in the electrophoretic bands (Fig. 8B). To explore whether the glycosylation affected the properties of the fibrotic ECM, collagen fiber diameters and insoluble collagen (cross-linked collagen) were investigated in BDL-induced liver fibrosis in WT and GLT25D1+/− mice. Collagen fibrils in the portal area viewed by SEM (Fig. 8C). After analyzed by Image-Pro Plus software, we found that heterozygous mice had fibers with larger diameters compared with WT mice (Fig. 8D). Insoluble collagen did not differ significantly between heterozygous knockdown and WT mice (Fig. 8E). SHG/TPEF imaging can characterize the architectural features of fibrosis at the individual collagen fiber level.25 We used SHG/TPEF microscopy to capture morphological differences in liver fibrosis between WT and GLT25D1+/− mice. GLT25D1+/− mice had more long strings than WT mice. The results confirm that GLT25D1 regulated glycosylation in type I collagen and influence collagen properties.
Discussion
We previously showed that GLT25D1 was associated with liver injury, autoimmune liver disease, and nonalcoholic steatohepatitis (NASH).19,20 However, little is known about the role of GLT25D1 in liver fibrosis and HSCs, which are mainly myofibroblast progenitor cells, and the central collagen-producing cells and key effectors in liver fibrosis.26,31 To our knowledge, this is the first study to systematically identify the specific biological functions of GLT25D1 in liver HSCs. We found that GLT25D1 was upregulated in activated HSCs and played an essential role in HSC activation, ECM production, and collagen stabilization.
HSCs reside in the space of Disse, between parenchymal cells and endothelial cells (EC) of the hepatic lobule.32 In this study, GLT25D1 and α-SMA (a marker of activated HSCs) were expressed mainly in nonparenchymal liver cells in fibrotic livers. Several studies have reported that GLT25D1 is involved in collagen formation in MC3T3-E1, MEFs, SaOS-2 and collagen-producing cells.13–15 Therefore, we concluded that GLT25D1 expression was also present in activated HSCs. We previously demonstrated that the concentration of GLT25D1 in serum from patients with cirrhosis was significantly higher than in serum from patients without cirrhosis.33 The study results showed that GLT25D1 expression was upregulated in IHCs the cirrhotic liver. Importantly, partial deletion of GLT25D1 significantly reduced BDL-induced collagen deposition and α-SMA expression, indicating that GLT25D1 has an important role in liver fibrosis and HSC activation.
HSC activation leads it to transdifferentiate into MFBs, characterized by expression changes and specific phenotypes, including proliferation, contractility, migration, fibrogenesis, matrix degradation, chemotaxis, and inflammatory signaling.34–36 In this study, GLT25D1 deficiency suppressed mRNA and protein expression of specific phenotype- and fibrosis-related genes in both LX-2 and mHSCs. Interestingly, GLT25D1 overexpression in LX-2 cells enhanced gene expression and the phenotypes, illustrating that GLT25D1 is required to activate HSCs.
Activated HSCs within the space of Disse migrate toward sites of inflammation, and deposit excessive ECM during the progression of liver fibrosis.37,38 Therefore, enhanced migration is an important aspect of GLT25D1 that contributes to liver fibrosis. Meanwhile, GLT25D1 upregulation affects cell proliferation and contractility. The high number of activated HSCs and contractibility of myofibroblasts in the fibrotic liver promote the constriction of hepatic sinusoids, resulting in blood flow and nutrient exchange.39 Clinically, HSC contraction causes and aggravates portal hypertension that predominantly determines liver stiffness.40,41 In this regard, GLT25D1 may be directly involved in portal hypertension formation, hence indirectly influencing liver stiffness, which correlates with liver fibrosis severity. Therefore, under conditions of chronic inflammation, GLT25D1 upregulation accelerates the progression of liver fibrosis. Baumann et al. reported that the loss of GLT25D1 in SaOS-2 cells induced high expression and accumulation of collagen I in the ER, although this did not result in ER stress.13 Geister et al. found that, in MEFs the loss of GLT25D1 induced the accumulation of type I collagen in the cell but led to its decrease in the culture medium.15 We obtained different results in this study. For both sh-GLT25D1 and mHSCs from GLT25D1+/− mice, COL I levels were significantly suppressed in the cytoplasm. This discrepancy may be due to differences in cell types, the clinical stage of the disease, the levels of GLT25D1 protein, and the knockdown and overexpression methods used. Nevertheless, it is imperative to emphasize that, in the current study, similar results were obtained using mHSC and LX-2. GLT25D1 overexpression induces the production of large quantities of collagen that need proper folding prior to secretion into the extracellular space, challenging the cell’s ER folding capacity. However, it is not clear whether COL I was correctly folded and secreted.
Previous studies on GLT25D1 have focused on the structure and secretion of collagen. However, the regulation of signal pathways by GLT25D1 has not been explored. An important finding in this study is that GLT25D1 mediates the activation of LX-2 cells by enhancing the activation of the canonical TGF-β1/SMAD3 pathway. This suggests that GLT25D1, either directly or indirectly, enhances the phosphorylation of Smad3. TGF-β1-induced HSC activation is involved in several signaling pathways, including the canonical TGF-β1/SMAD and non-SMAD-dependent TGF-β signaling. Non-SMAD-dependent TGF-β signaling includes the MAPK, mTOR, PI3K/AKT, and Rho/TPase pathways.42 The canonical TGF-β1/SMAD pathway, where SMAD3 is phosphorylated at the C-terminus, is the main fibrogenic pathway.
GLT25D1 is localized in the ER43 while the TGF-β1/SMAD signaling pathway operates in the membrane, cytoplasm and nucleus.42 thus, theoretically, there is no a direct interaction between GLT25D1 and proteins in the TGF-β1/SMAD pathway (TGFβRI, TGFβRII, Smad2/3/4). It is worth noting that collagen, which is regulated by GLT25D1, is also an important molecule mediating HSC activation. Col I binds to the integrin α11β1 to affect cell function.44 However, whether the process involves the TGF-β1/SMAD signaling pathway remains undetermined. Knocking down Gal-3(galectin-3) was correlated with the downregulation of GLT25D1 in CMT cells.45 Gal-3(galectin-3) regulates proliferation and migration of human pulmonary artery smooth muscle cells via the TGF-β1/Smad2/3 signaling pathway.46 Therefore, Gal-3 may participate in the pathway regulated by GLT25D1. Cellular communication network factor 1 (CCN1), which is secreted by fibroblasts, is glycosylated by GLT25D1,47 thus its function is affected by GLT25D1. CCN1 regulates the expression of genes associated with fibrosis in lung fibroblasts and is dependent on the TGF-β1/Smad3 signaling pathway.48 This suggests that CCN1 also participates in the pathway regulated by GLT25D1 in LX-2 cells. The possible mechanism diagram was shown in Figure 9.
We confirmed that GLT25D1 regulates glycosylation (GG-Hyl, G-Hyl) in COL I from mHSCs and LX-2 cells by mass spectrometry (Supplementary Table 4). Terajima et al reported that this glycosylation might regulate cross-link maturation and the growth of collagen fibrils in bone.14 Baumann and Hennet also reported that it affects the kinetics of triple helix formation.13 However, the content of insoluble collagen (cross-linked collagen) did not differ significantly between GLT25D1+/−and WT mice in this study. That may be due partly to the short disease course in our fibrosis model. GLT25D1+/− mice had larger fibrils compared with WT mice. This is consistent with the results of previous studies showing that loss of this glycosylation induces larger mean fibril diameters compared with controls.14,49,50 In general, collagen molecules with a higher degree of glycosylation form smaller fibrils that are involved in collagen stability.50,51 This finding may reflect that the alteration in collagen morphology was, to some extent, regulated by GLT25D1, which may be involved in collagen degradation and the reversal of fibrosis. Thus, GLT25D1 affects liver fibrosis by altering the collagen structure.
The study has some limitations. GLT25D1 knockdown in GLT25D1+/− mice is not restricted to HSCs, where the biological function of GLT25D1 in hepatocytes during liver fibrosis should not be ignored18,19 and may be accompanied by a compensatory effect of GLT25D2, another galactosyltransferase enzyme isoform.13 Additionally, the expression of GLT25D1 in other hepatic cells remains unclear, including Kupffer cells, macrophages, cholangiocytes, and liver sinusoidal EC. Data on the role of GLT25D1 in the signaling pathway are fairly limited. When investigating collagen properties, a high purity extract of collagen was required, which need to further improve in the further study. Therefore the role of GLT25D1 in hepatic fibrosis requires further investigation.
In conclusion, our data showed that GLT25D1 regulated HSC activation and collagen stability. Additionally, GLT25D1 downregulation may alleviate liver fibrosis. Targeting GLT25D1 may be more feasible and safer than directly targeting TGF-β in liver fibrosis and offers a promising therapeutic target for treating liver fibrosis.