Identication of GDF15 as A Pro-Fibrotic Factor in Mouse Liver Fibrosis Progression

Aim: To elucidate the inhibitory role of growth differentiation factor 15 (GDF15) in liver brosis and its possible activation mechanism in hepatic stellate cells of mice. Methods: We generated a GDF15-neutralizing antibody that can inhibit TGF-β1-induced activation of the TGF-β/Smad2/3 pathway in LX-2 cells. All the mice in this study were induced by carbon tetrachloride and thioacetamide. In addition, primary hepatic stellate cells from mice were isolated from fresh livers using Nycodenz density gradient separation. The severity and extent of liver brosis in mice were evaluated by Sirius Red and Masson staining. The effect of GDF15 on the activation of the TGF-β pathway was detected using dual-luciferase reporter assays and Western blotting assays. The expression of GDF15 in cirrhotic liver tissue was higher than that in normal liver tissue. Blocking GDF15 with a neutralizing antibody resulted in a delay in primary hepatic stellate cell activation and remission of liver brosis induced by carbon tetrachloride or thioacetamide. Meanwhile, TGF-β pathway activation was partly inhibited by a GDF15-neutralizing antibody in primary hepatic stellate cells. These results indicated that GDF15 plays an important role in regulating HSC activation and liver brosis progression. attenuates chemical-inducible liver brosis and delays hepatic stellate cell activation, to its in TGF-β


Introduction
Liver brosis is a common chronic liver disease that is characterized by the accumulation of collagen and activation of hepatic stellate cells (HSCs) in the liver [1]. Liver brosis is a common precursor of liver cirrhosis and even hepatocellular carcinoma [1]. In addition, liver cirrhosis is irreversible and can only be effectively cured by liver implantation [2]. Therefore, preventing the progression from liver brosis to liver cirrhosis is highly bene cial to patients with chronic liver disease.
TGF-β signalling plays a key role in the progression of liver brosis and activation of HSCs [3], which may theoretically result in the blockage of liver brosis progression [3]. Nevertheless, the direct blockade of TGF-β signalling has been shown to cause many unacceptable adverse effects in animal trials [4]; thus, it is important to nd new targets that can regulate TGF-β signalling as well as liver brosis progression.
Growth differentiation factor 15 (GDF15) is also known as macrophage inhibitory cytokine-1 (MIC-1), which plays an important role in regulating TGF-β signalling [5]. GDF15 is also known as a distant member of the TGF-β superfamily. It has been reported that GDF15 is closely related to lung brosis progression [6]. Moreover, GDF15 can ameliorate renal brosis by inhibiting broblast growth and activation [6]. The results above suggested that GDF15 could potentially regulate liver brosis due to similar mechanisms in the pathogenesis of pulmonary brosis and renal brosis.
We conducted this study and found that blocking GDF15 attenuated liver brosis in mice. In addition, we con rmed the effect of GDF15 on TGF-β signalling in mouse liver tissues and primary HSCs.

Materials And Methods
Clinical samples.
All human normal liver tissues and cirrhotic liver tissues were provided by the Department of Liver Surgery and Liver Transplantation at Shandong Provincial Hospital. The tissues were obtained with informed consent, and the study was approved by the Ethical Committee of Shandong Provincial Hospital.
Cell culture.
LX-2 cells were purchased from Millipore Corporation. These cells were cultured in Dulbecco's-modi ed Eagle medium (DMEM) supplemented with 10% foetal calf serum and 1% antibiotics at 37°C in a Thermo incubator containing 5% carbon dioxide.

Experimental animals.
All mice were purchased from Shandong University Laboratory Animal Centre and were housed and fed under speci c pathogen-free (SPF) conditions. Among them, female BALB/c mice were prepared for neutralizing antibodies, and C57BL/6J mice were used for carbon tetrachloride and thioacetamideinduced liver brosis. We ensured that all animals in our study received humane care, and all study protocols complied with the institution's guidelines. The animal experiments were approved by the Ethical Committee of Shandong Provincial Hospital and the approve No. was NSFC:No.2019-033.
Preparation of the carbon tetrachloride-and TAA-induced liver brosis model. Eight C57BL/6J mice were injected with carbon tetrachloride (CCl 4 , 0.5 μL per gram body weight) in 25% olive oil twice per week at equal intervals for eight weeks. Another group of C57BL/6J mice (including eight mice) was injected with thioacetamide (TAA, 0.2 mg/g body weight) in double-distilled water three times per week at equal intervals for eight weeks. All chemical-inducible liver brosis mice were divided into two groups. The GDF15 mAb group mice (including eight mice) were treated with GDF15-neutralizing antibodies at a dose of 20 mg/kg twice a week by abdominal injection, and the control group mice (including eight mice) were injected with homologous IgG for six weeks.
Isolation of primary hepatic stellate cells from mice. C57BL/6J mice were sacri ced using cervical dislocation method, and their livers were removed from the abdominal cavities. The fresh livers were perfused with a digestive solution consisting of 0.1% collagenase, 0.25% pronase E and 0.01% DNase and mechanically dissected. The digested livers were incubated in the same digestive solution at 37°C for approximately 25 to 30 minutes. Next, the suspension was removed and ltered through an iron mesh with 100-μm pores. Then, the ltered suspension was centrifuged through a Nycodenz gradient (Axis-Shield) at 8.2% concentration. The isolated primary HSCs were resuspended in DMEM supplemented with 20% foetal calf serum and 1% antibiotics (penicillin and streptomycin) and cultured at 37°C overnight in an environment containing 5% carbon dioxide. Twenty-four hours later, cell debris and non-adherent cells were removed.
GDF15-neutralizing antibody generation was performed using conventional hybridoma techniques. Histag fusion constructs of GDF15 (which encoded Mouse GDF15 amino acids 10-303) were cloned into PET28 vectors and expressed according to previously established protocols. The puri ed His-tag fusion protein was used as an antigen to immunize female BALB/c mice for the generation of anti-GDF15 monoclonal antibodies. Afterwards, we identi ed one monoclonal antibody that effectively blocked TGF-β1-induced activation of the TGF-β/Smad2/3 signalling pathway and used this mAb as the primary GDF15-neutralizing antibody for all subsequent animal experiments.
Immunohistochemical and Sirius Red staining.
The para n-embedded liver tissues from mice and humans were sliced into 5-μm-thick sections followed by depara nization and stepwise rehydration in preparation for IHC and Sirius Red staining. For Sirius Red staining, the sections were stained with Sirius Red for approximately three minutes and dehydrated stepwise. For immunohistochemical staining, the sections were incubated with 0.3% hydrogen peroxide for thirty minutes and blocked with 10% bovine serum albumin. The sections were then incubated using an antibody targeting GDF15 (1:200, Abcam, ab189358) at 4ºC overnight and labelled with an HRPconjugated secondary mouse antibody (Abcam) at room temperature for approximately one hour. Afterwards, the sections were incubated with a DAB substrate liquid (Thermo Fisher) and stained with haematoxylin for two minutes. All the sealed slides were imaged and recorded on a microscope manufactured by Carl Zeiss.
The slides were depara nized and rehydrated stepwise before Masson staining. The slides were rst stained with Weigert haematoxylin for 5 minutes and washed with water. The slides were then stained with acid ponceau for six minutes followed by quick immersion in 2% glacial acetic acid. Afterwards, the slides were immersed in 1% phosphomolybdic acid for approximately three minutes prior to immediate staining with aniline blue for ve minutes. The slides were subsequently immersed in 0.2% glacial acetic acid brie y and dehydrated stepwise until they were sealed with neutral balsam.
For cell staining, primary mouse HSCs were seeded onto rounded coverslips in 24-well plates and incubated at 37°C for approximately seven days in an environment containing 5%CO 2 . For F-actin staining, cells were incubated with phalloidin-FITC (Sigma) for seventy minutes at room temperature. For α-SMA staining, cells were incubated with α-SMA antibody (Sigma) for seventy-ve minutes at room temperature and subsequently treated with Alexa Fluor 594-conjugated secondary antibody, avoiding light. The nuclei were stained using DAPI (Sigma), and the immuno uorescence images were recorded using a uorescence microscope (Carl Zeiss).

Dual-luciferase reporter assay
We used the pGL4.19-TA-TGFBeta(V2) reporter vector (Promega) as a dual-luciferase reporter vector for TGF-beta signalling. pGL4.19-TA-TGFBeta was transfected into LX-2 cells and treated with GDF15, Ly364947, or TGF-β1. A Thermo Scienti c Microplate Reader was used to detect the absorbance value of each group. Ly364947 is an inhibitor of the TGF-β signalling pathway.
Statistical analysis.
All the data are presented as the means ± standard error of the mean (SEM). Statistical analysis was performed using SPSS 16.0 software. One-way ANOVA method was used for multigroup comparison. The comparisons of absorbance value between different groups were carried out by using ANOVA method. Two-tailed Student's t-test was used for comparisons between two groups. A P value less than 0.05 was considered statistically signi cant.

GDF15 was upregulated in cirrhotic liver tissues and activated HSCs
We rst detected the expression of GDF15 in human liver tissues and found that the expression of GDF15 in cirrhotic liver tissue was signi cantly higher than that in normal liver (Fig 1a and 1b, 1c). GDF15 expression was mainly localized in cirrhotic lesions in liver tissue, which indicated a correlation between GDF15 and liver cirrhosis (Fig 1d). The cell immuno uorescence staining assay showed that the expression of GDF15 in HSCs coincided with α-SMA (Fig 1e). Immuno uorescence staining revealed that the localization of GDF15 expression was nearly coincident with α-SMA, which re ected the activation of HSCs (Fig 1f). Moreover, Western blotting assays showed that GDF15 expression mainly originated from activated HSCs (Fig 1g). IgG-treated mice was signi cantly larger than that in the GDF15 mAb-treated group (Fig 2c,2d).
GDF15 mAb inhibited the activation of mouse HSCs α-SMA expression is closely correlated with HSC activation and is considered an HSC activation marker. We successfully isolated primary HSCs from mouse livers and subjected the cells to aberrant treatment for ve to seven days. Primary HSCs were xed at ve days and stained for immuno uorescence imaging. The results showed that the cell morphology of HSCs treated with GDF15 mAb was different from that of HSCs treated with control IgG. The GDF15 mAb-treated HSCs showed a lower activation state compared with cells in the IgG group (Fig 2e). The α-SMA expression levels in IgG-treated HSCs were signi cantly higher than those in the GDF15 mAb-treated group (Fig 2e).

GDF15 modulated HSC activation and liver brosis by regulating TGF-beta signalling
Compared with IgG-treated primary mouse hepatic stellate cells, the hepatic stellate cells treated with GDF15 mAb showed lower phospho-Smad2 and phospho-Smad3 levels, which re ected the activation of TGF-β signalling (Fig 3a). In addition, chemical-inducible mouse liver tissues from the GDF15 mAb group showed lower phospho-Smad2 and phospho-Smad3 levels than those from the IgG group (Fig 3b).
Moreover, a luciferase reporter assay showed that TGF-β signalling in LX-2 cells can be activated by the GDF15 protein, and this activation effect can be blocked by the TGF-β pathway inhibitor Ly364947 (Fig  4).

Discussion
Liver brosis occurs during the early stages of liver cirrhosis and is mainly characterized by intrahepatic structural rearrangement, hepatic cell regeneration and excess accumulation of collagen [7][8][9][10]. Liver brosis is a reversible process until liver cirrhosis develops. Progressive liver brosis causes increased morbidity and mortality in patients with chronic liver disease [11,12]. The therapeutic methods for liver brosis are limited, with minimal effectiveness to date [13].
GDF15 was rst isolated from macrophages after cytokine stimulation [43]. It has been reported that GDF15 expression can be regulated by several signalling pathways, including the p53/DEC1, GSK-3β, and EGR-1 pathways [44]. Moreover, it was reported that GDF15 plays important roles in regulating renal brosis and lung brosis [6]. However, the role of GDF15 in liver brosis progression remains unclear. Therefore, elucidation of the biological functions of increased GDF15 in liver disease pathogenesis may promote the potential application of GDF15 in diagnosis and targeted therapy. Similar to other TGF-β superfamily cytokines, GDF15 is synthesized as a precursor protein. N-terminal propeptides originating from mature GDF15 through proteolysis can be secreted as disulphide-linked homodimers [45]. The membrane receptors of GDF15 have not been identi ed. Moreover, it has been reported that GDF15 activates smad2 and smad3 through TGF-β I type receptors and promotes colorectal cancer metastasis [46].
In our study, we discovered evidence that illustrates the pro-brotic effect of GDF15 in liver brogenesis. The expression levels of GDF15 in cirrhotic liver tissue and activated HSCs were higher than those in normal liver tissue and quiescent HSCs. The differential expression of GDF15 in cirrhotic liver tissue and normal liver tissue suggested its potential role in liver brosis regulation.
Blocking GDF15 with a neutralizing antibody markedly delayed HSC activation and attenuated liver brosis in the chemical-inducible liver brosis mouse model. These results revealed the important role of GDF15 in promoting liver brosis progression. The activation of primary mouse HSCs treated with the GDF15-neutralizing antibody was slower than that of the IgG-treated control group. Therefore, GDF15 mAb may inhibit liver brosis progression by regulating HSC activation. The activation of TGF-beta signalling in primary HSCs treated with GDF15-neutralizing antibody was inhibited compared with the control group. These results suggested that GDF15 may play an important role in regulating TGF-beta signalling. In addition, we also found that GDF15 regulated liver brosis in vivo via TGF-beta signalling.
Activation of TGF-beta signalling has already been shown to play a central role in regulating the progression of tissue brosis. According to previous reports and our study results, the positive regulation of TGF-beta signalling by GDF15 can be recognized as the main mechanism explaining the progressive effect of GDF15 in liver brosis.
In summary, our research supports the role of GDF15 as a therapeutic target for attenuating liver brosis. In our study, no severe adverse effects caused by GDF15 inhibition were observed in either liver brosis mouse model. We con rmed the positive effects of GDF15 on HSC activation and liver brosis. According to our study, GDF15 can be further explored as an anti-brosis agent.
Declarations Figure 1 (a), qPCR assay showed that the expression of GDF15 in human brotic liver tissues was higher than that in normal liver. (b) Grey value analysis of Western blot assay showed the expression of GDF15 in human brotic liver tissues was higher than that in normal liver. (c) Western blot assay showed the GDF15 protein expression in brotic liver tissues (Fib) was higher than that in normal liver (Nor). (d) IHC images of GDF15 expression in human brotic liver tissues and normal liver tissues. (e) Cell immuno uorescence staining showed that GDF15 was expressed at high levels in activated HSCs. (f) GDF15 expression was mainly localized to cirrhotic lesions of liver tissues. (g) Western blot assay showed that GDF15 protein expression mainly originated from activated HSCs compared with HEPs (hepatocytes), PFs (portal broblasts), KCs (Kupffer cells), and SECs (sinusoidal endothelial cells).