Role of B Cell-Activating Factor in Fibrosis Progression in a Murine Model of Non-Alcoholic Steatohepatitis

Non-alcoholic fatty liver disease (NAFLD) is the most prevalent chronic liver disease all over the world. Therapeutic strategies targeting its multidirectional pathways are required. Particularly, fibrosis is closely associated with its prognosis. We previously found that B cell-activating factor (BAFF) is associated with severity of NAFLD. Here, we determined the direct in vivo role of BAFF in the development of liver fibrosis. Histological and biochemical analyses were performed using wild-type and BAFF-deficient mice. We established a murine model of non-alcoholic steatohepatitis (NASH) using carbon tetrachloride injection accompanied by high-fat/high-cholesterol diet feeding. Additionally, in vitro analysis using mouse macrophage-like cell line RAW264.7 and primary hepatic stellate cells was performed. Hepatic steatosis and inflammation, and most importantly, the progression of liver fibrosis, were ameliorated in BAFF-deficient mice compared to those wild-type mice in our model. Additionally, BAFF deficiency reduced the number of CD11c+ M1-type macrophages in the liver. Moreover, BAFF stimulated RAW264.7 cells to secrete nitric oxide and tumor necrosis factor α, which drove the activation of hepatic stellate cells. This indicates that BAFF plays a crucial role in NASH development and may be a promising therapeutic target for NASH.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is an exhibit of systemic disease based on lifestyle-related diseases such as obesity, diabetes, and metabolic syndrome. The number of patients with NAFLD is rapidly increasing with the rising prevalence of obesity, and it has become a global health problem [1,2]. NAFLD is not just a manifestation of a metabolic syndrome but has emerged as a driver of systemic disease [3]. Previous studies have suggested that the fibrosis stage is an independent and significant predictor of overall mortality and liver-related events [4][5][6]. Therefore, it is critical to identify the drivers that mediate the progression of non-alcoholic fatty liver (NAFL) to non-alcoholic steatohepatitis (NASH).
As chronic hepatic inflammation has been shown to contribute to disease progression, one of the potential therapeutic targets of NAFLD is the suppression of inflammation and fibrosis by the regulation of immune abnormalities. We have been conducting research on the immunological aspects of NAFLD [7][8][9] development. B cell-activating factor (BAFF) (CD257) is a factor that promotes the expansion and differentiation of the B cell population [10,11]. BAFF belongs to the tumor necrosis factor (TNF)-ligand family and is secreted from activated T cells, B cells, and myeloid lineage cells such as macrophages and dendritic cells. Previously, we had reported that serum BAFF levels were increased, and that BAFF was preferentially expressed, in the visceral adipose tissue of high-fat diet (HFD)induced obese mice [12][13][14]. In addition, hepatic steatosis was attenuated in BAFF-deficient mice fed an HFD compared to in control mice [15]. Furthermore, serum BAFF levels in patients with NASH were higher than those in patients with NAFL [16]. Collectively, these data indicate that BAFF may be closely related to hepatic steatosis, inflammation, and fibrosis in NASH. However, the direct role of BAFF in hepatic fibrosis has not yet been elucidated.
In this study, we established a murine model of NASH that led to the reproducible progression of steatohepatitis with significant fibrosis using carbon tetrachloride (CCl 4 ) injection accompanied by high-fat/high-cholesterol diet (HFHCD) feeding. To determine the in vivo role played by BAFF in the development of liver fibrosis, we studied the NASH model using BAFF-deficient mice and found that BAFF-deficient mice were protected from developing not only steatosis but also NASH and fibrosis. Our findings indicate that BAFF plays an important role in the development of liver fibrosis and may be a therapeutic target for NASH.

Liver Inflammation Is Attenuated in BAFF −/− Mice in Murine Models of NASH
In addition to steatosis, liver inflammation was attenuated in HFHCD/CCl 4 -treated BAFF −/− mice compared to in HFHCD/CCl 4 -treated WT mice (Figure 1d,e). We per-formed immunohistochemical and flow cytometric analyses of immune cell populations and assessed inflammation-related gene expression in the liver by real-time RT-PCR. In immunohistochemical analysis, the number of macrophages and crown-like structures (CLSs) in the liver was significantly lower in BAFF −/− mice than that in WT mice treated with HFHCD/CCl 4 (Figure 2a). Flow cytometric analysis showed that the proportion of F4/80 + CD11c + M1-like macrophages in the livers of HFHCD/CCl 4 -treated BAFF −/− mice was significantly lower than that in HFHCD/CCl 4 -treated WT mice (Figure 2b). In addition, the expression of TNF-α, interleukin (IL) 6, monocyte chemotactic protein (MCP)-1, and M1-macrophage-related markers, such as CD11c and inducible nitric oxide synthase (iNOS), was significantly lower in the livers of BAFF −/− mice than that in WT mice treated with HFHCD/CCl 4 (Figure 2c). The expression of iNOS in macrophages was confirmed by immunohistochemistry, and iNOS-positive areas in the livers of HFHCD/CCl 4 -treated BAFF −/− mice were significantly lower than in HFHCD/CCl 4 -treated WT mice ( Figure 2d). These data suggest that attenuated liver inflammation in BAFF deficiency is associated with a reduction in proinflammatory cytokines produced by macrophages in our models.
To test whether the role of BAFF is limited to the specific case of HFHCD/CCl 4induced liver injury, we used two alternative model systems: (1) feeding with HFHCD for a longer period ( Figure 4a) and (2) feeding with choline-deficient, L-amino-acid-defined high-fat diet (CDAHFD) (Figure 4b). In both models, collagen deposits in the liver were significantly lower in BAFF −/− mice than in WT mice (Figure 4c,d). Thus, liver fibrosis progression was ameliorated in the absence of BAFF in at least three model systems.

Hepatic Stellate Cell Activation Is Driven by Interaction with Macrophages in Murine Models of NASH
A crucial downstream consequence of hepatic inflammation is the activation of hepatic stellate cells (HSCs), the principal fibrogenic cell type in the liver. To explore the underlying mechanisms of the differences in liver fibrosis in our models, we investigated the direct effects of BAFF on HSCs. However, receptors for BAFF, such as BAFF-receptor (BAFF-R), ransmembrane activator, calcium-modulator, and B-cell maturation antigen, were not expressed in primary cultured mouse HSCs. Furthermore, the addition of BAFF to HSCs in culture did not alter the gene expression of HSCs such as α-SMA. These data indicate that BAFF has little or no direct effect on HSCs.
Thereafter, we focused on macrophages, which are one of the main sources of potent pro-fibrogenic signals [19]. In addition to hepatocytes [13], macrophages express BAFF-R [20].
Hence, we examined whether BAFF-R exerted its signaling function in macrophages

Hepatic Stellate Cell Activation Is Driven by Interaction with Macrophages in Murine Models of NASH
A crucial downstream consequence of hepatic inflammation is the activation of hepatic stellate cells (HSCs), the principal fibrogenic cell type in the liver. To explore the underlying mechanisms of the differences in liver fibrosis in our models, we investigated the direct effects of BAFF on HSCs. However, receptors for BAFF, such as BAFF-receptor (BAFF-R), ransmembrane activator, calcium-modulator, and B-cell maturation antigen, were not expressed in primary cultured mouse HSCs. Furthermore, the addition of BAFF to HSCs in culture did not alter the gene expression of HSCs such as α-SMA. These data indicate that BAFF has little or no direct effect on HSCs.
Thereafter, we focused on macrophages, which are one of the main sources of potent pro-fibrogenic signals [19]. In addition to hepatocytes [13], macrophages express BAFF-R [20].
Hence, we examined whether BAFF-R exerted its signaling function in macrophages and found that nuclear factor (NF)-κB2 and RelB, the non-canonical NF-κB pathway, were induced in the mouse macrophage-like cell line (RAW264.7) treated with BAFF ( Figure 5a). Furthermore, as shown in Figure 5b, BAFF treatment upregulated the expression of genes encoding proteins related to inflammation and tissue repair, including iNOS and IL-10. These changes were inhibited by treatment with the recombinant BAFF-R Fc (Figure 5b). 5a). Furthermore, as shown in Figure 5b, BAFF treatment upregulated the expression of genes encoding proteins related to inflammation and tissue repair, including iNOS and IL-10. These changes were inhibited by treatment with the recombinant BAFF-R Fc ( Figure  5b).
Finally, we investigated whether the effect of BAFF on macrophages influenced HSC activation. Isolated mouse HSCs were cultured with conditioned medium (CM) from BAFF-or phosphate-buffered saline (PBS)-treated-RAW264.7 cells in addition to lipopolysaccharide (LPS). The mRNA levels of TGF-β1 and α-SMA were significantly increased in HSCs following culture with CM from BAFF-treated-RAW264.7 cells (Figure 6a). Moreover, nitric oxide (NO) (LPS 344.7 ± 94.08 μM, LPS+BAFF 423.4 ± 83.28 μM: p < 0.05) and TNF-α levels in CM from BAFF-treated-RAW264.7 cells were significantly higher than those from PBS-treated RAW264.7 cells (Figure 6b). However, the IL-6 levels in CM were not different between the two groups. These results indicate that BAFF-treated macrophages generate inflammatory cytokines such as NO and TNF-α and activate HSCs, leading to the development of liver fibrosis in NASH.  Finally, we investigated whether the effect of BAFF on macrophages influenced HSC activation. Isolated mouse HSCs were cultured with conditioned medium (CM) from BAFFor phosphate-buffered saline (PBS)-treated-RAW264.7 cells in addition to lipopolysaccharide (LPS). The mRNA levels of TGF-β1 and α-SMA were significantly increased in HSCs following culture with CM from BAFF-treated-RAW264.7 cells (Figure 6a). Moreover, nitric oxide (NO) (LPS 344.7 ± 94.08 µM, LPS+BAFF 423.4 ± 83.28 µM: p < 0.05) and TNF-α levels in CM from BAFF-treated-RAW264.7 cells were significantly higher than those from PBS-treated RAW264.7 cells (Figure 6b). However, the IL-6 levels in CM were not different between the two groups. These results indicate that BAFF-treated macrophages generate inflammatory cytokines such as NO and TNF-α and activate HSCs, leading to the development of liver fibrosis in NASH.

Discussion
The prevalence of NAFLD is increasing and liver-related morbidity and mortality will dramatically increase worldwide within the next few decades [2]. At present, lifestyle modification is the mainstay of therapeutic recommendations and no specific pharmacological treatment is available for NAFLD [21]. Drugs targeting metabolism, inflammation, and fibrogenesis are under development; however, several compounds have only improved the investigational endpoints in a subset of patients who are exposed to these drugs, indicating that targeting a single pathway may be insufficient [22]. To overcome these problems, combination therapies and/or therapies targeting multidirectional effects may be promising approaches.
We previously found that BAFF is associated with NAFLD severity in Japanese patients [16]. Furthermore, BAFF deficiency prevents fat accumulation in the liver by suppressing the visceral adipose tissue inflammation and de novo lipogenesis in an HFD-fed mouse model [15]. Although liver fibrosis was not observed in liver specimens from HFDfed mice, the expression of fibrosis-related genes, such as TGF-β1, α-SMA, and Col-1a1, was significantly lower in the livers of BAFF −/− mice than in those of WT mice.
Furthermore, BAFF promotes collagen production by dermal fibroblasts from patients with systemic sclerosis [23], and BAFF inhibition attenuates skin and liver fibrosis in mouse models of scleroderma [24]. These findings suggest that BAFF may play a role not only in metabolism and inflammation but also in fibrosis in the pathogenesis of NASH, which indicates that it is one of the therapeutic targets in NASH.

Discussion
The prevalence of NAFLD is increasing and liver-related morbidity and mortality will dramatically increase worldwide within the next few decades [2]. At present, lifestyle modification is the mainstay of therapeutic recommendations and no specific pharmacological treatment is available for NAFLD [21]. Drugs targeting metabolism, inflammation, and fibrogenesis are under development; however, several compounds have only improved the investigational endpoints in a subset of patients who are exposed to these drugs, indicating that targeting a single pathway may be insufficient [22]. To overcome these problems, combination therapies and/or therapies targeting multidirectional effects may be promising approaches.
We previously found that BAFF is associated with NAFLD severity in Japanese patients [16]. Furthermore, BAFF deficiency prevents fat accumulation in the liver by suppressing the visceral adipose tissue inflammation and de novo lipogenesis in an HFD-fed mouse model [15]. Although liver fibrosis was not observed in liver specimens from HFDfed mice, the expression of fibrosis-related genes, such as TGF-β1, α-SMA, and Col-1a1, was significantly lower in the livers of BAFF −/− mice than in those of WT mice.
Furthermore, BAFF promotes collagen production by dermal fibroblasts from patients with systemic sclerosis [23], and BAFF inhibition attenuates skin and liver fibrosis in mouse models of scleroderma [24]. These findings suggest that BAFF may play a role not only in metabolism and inflammation but also in fibrosis in the pathogenesis of NASH, which indicates that it is one of the therapeutic targets in NASH.
In the present study, we report a new murine NASH model using HFHCD and CCl 4 , which develops the histological features of NASH with extensive fibrosis and severe inflammation. HFHCD has been widely used to establish mouse NASH models; however, the major disadvantage of this NASH model is that it does not fully progress to severe steato-hepatitis, even after long-term feeding. CCl 4 has been traditionally used for decades to induce liver injury and fibrosis in rodents [17,18,25]. Although the mechanical stimulation of CCl 4 differs from the natural history of NASH, we used CCl 4 as a fibrosis accelerator in this model. HSCs are a major source of activated myofibroblasts, and their activation is known to drive fibrosis in the liver [26]. HFHCD and CCl 4 treatment induced the proliferation and activation of HSCs, which were responsible for the rapid progression of fibrosis in our NASH model. Consistent with our previous study [15], hepatic steatosis was ameliorated in BAFF −/− mice compared to in WT mice (Figure 1). Moreover, liver inflammation was attenuated in BAFF −/− mice compared to in WT mice (Figure 2). Itoh et al. [27,28] have reported that CD11c + -activated macrophages induce the formation of hepatic CLSs in murine models and patients with NASH, which is consistent with the findings based on our model. BAFF deficiency reduced the number of macrophages, especially CD11c + M1-type macrophages, that produce inflammatory signals such as TNF-α and iNOS, and CLS formation in the livers of HFHCD/CCl 4 -treated mice (Figure 2). It is widely accepted that the M1/M2 balance of macrophages plays an important role in obesity, metabolic syndrome, and fatty liver disease [29]. Macrophages respond to a cocktail of injury signals, including BAFF, and secrete inflammatory substances that may contribute to NASH development and exacerbation.
One of the most relevant findings of our present study is that BAFF-deficient mice are protected from the development of NASH and the progression of fibrosis (Figure 3), which may contribute to the higher liver weight in BAFF −/− mice compared to that in WT mice (Figure 1b). Furthermore, we confirmed the effect of BAFF on liver fibrosis in the other two models (Figure 4). One of the key features of NASH development and fibrosis progression is the activation of HSCs [26,[30][31][32]. In addition to mouse HSC, we observed that the human HSC cell line, LX-2, did not express BAFF-R in our preliminary study. Although the direct effects of BAFF on HSCs in vitro were not observed in our study, BAFF stimulated macrophages to secrete cytokines and other soluble factors that contribute to fibrosis by activating HSC (Figure 6). Similarly, previous reports have demonstrated that macrophages drive HSC activation through the release of soluble factors, such as cytokines, chemokines, and reactive oxygen species, and promote HSC survival [26,31]. Moreover, HSCs and macrophages interact and remodel the extracellular matrix and immune microenvironment among liver cells in the fibrotic liver [32]. Although the influence of other factors has not been disregarded, NO and TNF-α released from macrophages are partly involved in this process. These data indicate that this inflammatory switch is a definitive step in fibrosis progression.
Recent reports have outlined the heterogeneity of hepatic macrophages in NASH [27,28,[33][34][35][36]. Satoh et al. [33] have reported that Ceacam1 + Msr1 + Ly6C − F4/80 − Mac1 + monocytes, which are segregated-nucleus-containing atypical monocytes, are critical for fibrosis. Furthermore, single-cell RNA sequencing has revealed that specific phenotypes of macrophages characterized by high expression of triggering receptor expressed on myeloid cells 2 were expanding in both mouse models and human NASH. These cells, also called NASH-associated macrophages, have been reported to be linked to CLS aggregates and are correlated with disease severity [34]. Initially, we investigated these cells from the liver in our models using flow cytometry; however, the number of these cells did not differ between WT and BAFF −/− mice. Further studies on the functional properties of these cells are necessary in this regard.
The adipose tissue is also a key factor in the pathogenesis of NAFLD. We previously showed that oxidative stress in visceral adipose tissue, which is one of the underlying causes of NAFLD, regulated serum BAFF levels in HFD-fed mice [14]. Although we did not investigate the VAT in this model, similar mechanisms may play a role in regulating the BAFF level.
We primarily focused on the metabolic aspects of BAFF in the liver; however, BAFF further plays an important role in regulating the immune system [37]. Recently, B lymphocytes have been reported as central mediators of the progression of NAFLD and NASH-associated hepatocellular carcinoma [25,[38][39][40][41]. Most of these reports have suggested the pathogenic role of B cells in NAFLD. Interestingly, in mice fed a methionine-choline-deficient diet, B2-cell depletion, including antibody-mediated BAFF neutralization, ameliorated parenchymal damage and lobular inflammation [40]. Furthermore, Novobrantseva et al. [25] have reported that B cells play an important antibody-independent role in tissue repair following liver injury and the development of liver fibrosis. Therefore, it is conceivable that B-celldirected therapies, including B-cell depletion and approaches targeting B-cell inflammatory mediators such as TNF-α, may ameliorate NASH progression [38]. However, Karl et al. [41] have recently reported that B cells have both detrimental and protective effects in a NAFLD mouse model. Therefore, future studies are warranted to investigate the immunological roles of BAFF and B-cell compartments prior to clinical trials.
In summary, we demonstrated that BAFF depletion ameliorates NASH development and fibrosis progression. Although more preclinical evidence is required, our data suggest that targeting BAFF may be beneficial in treating NASH.

Animals
All study protocols complied with the guidelines of Ehime University (Ehime, Japan), and the protocol was approved by Ehime University Animal Research (No. 05TI70-16).
Male C57BL/6J WT mice and BAFF −/− mice were purchased from CLEA Japan (Tokyo, Japan) and the Jackson Laboratory (Bar Harbor, ME, USA), respectively. They were maintained at the Department of Biological Resources, Integrated Center for Science, Ehime University, under controlled temperature, humidity, and light (12-h light/dark cycles).
Six-week-old mice, WT mice, and BAFF −/− mice were fed HFHCD (20 g% fat, 18 g% cholesterol, 22 g% protein, and 45 g% carbohydrates; 1883.7 kJ [450 kcal]/100 g; D16010101; Research Diets, New Brunswick, NJ, USA). They received intraperitoneal injections of CCl 4 (Wako, Osaka, Japan) at a dose of 0.4 mL/kg diluted 1:25 in olive oil twice a week for 4 weeks from 9 weeks of age. Mice were maintained on each diet and sacrificed at 13 weeks of age.
Serum and liver samples were stored at −80 • C until use. The liver was submerged in RNA later (Life Technologies, Carlsbad, CA, USA) overnight and stored at −20 • C. Serum AST and ALT levels were measured using a Hitachi 7180 autoanalyzer (Hitachi, Ltd., Tokyo, Japan).

Histological and Morphometric Analysis
Liver tissues were fixed in neutral-buffered formalin and embedded in paraffin. Threemicrometer-thick sections were stained with hematoxylin and eosin, SR, and α-SMA (Thermo Fisher Scientific, Waltham, MA, USA). Histological examination was performed in a blinded manner by two experienced liver pathologists with a histological scoring system for NAFLD [42]. The positive areas of each stain(s) were measured digitally using the ImageJ software (National Institutes of Health, Bethesda, MD, USA). To evaluate the degree of fat accumulation, the livers were fixed with osmium tetroxide (OsO 4 ), as described previously [15]. SR-positive areas or α-SMA-positive areas were measured using histological light-microscopy images (10×; 7 sections per animal, n = 8 animals/group).

Measurement of Hepatic Triglyceride and Cholesterol
Hepatic triglyceride and cholesterol levels were measured at Skylight Biotech (Akita, Japan) using the Folch technique with Cholestest TG and Cholestest CHO kits (Sekisui Medical, Tokyo, Japan), respectively.

Collagen Content in Hepatic Tissue
The total collagen content in the hepatic tissue was measured using a commercially available kit (QuickZyme Total Collagen Assay; QuickZyme Biosciences, Leiden, Netherlands).

Quantitative Real-Time RT-PCR
RNA was extracted using a RNeasy Plus Mini Kit (Qiagen, Hilden, Germany). Reverse transcription was performed using a High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA), and real-time RT-PCR analysis was performed using SYBR Green I (Roche Diagnostics, Basel, Switzerland) on a LightCycler ® 96 (Roche Diagnostics). Primer sequences and annealing temperatures are provided in Table 1. Gene expression data were normalized to the housekeeping gene encoding hypoxanthine phosphoribosyltransferase (HPRT) 1 and expressed as a ratio of the values obtained for WT mice and control RAW264.7, respectively.
The CellAmpTM Direct TB Green RT-qPCR Kit (TAKARA Bio, Shiga, Japan) was used for primary cultured stellate cells according to the manufacturer's protocol. Gene expression data were normalized to hypoxanthine phosphoribosyltransferase 1 and expressed as a ratio of the values obtained for the CM from PBS-treated-RAW264.7 cells in addition to LPS (Sigma-Aldrich, St. Louis, MO, USA).

Isolation of Primary HSCs
Primary HSCs were isolated from male C57BL/6 mice using the pronase-collagenase digestion method [43] and were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Merck, Darmstadt, Germany). One day after culturing, the cells were treated with 100 ng/mL murine recombinant BAFF (R&D Systems, Minneapolis, MN, USA) for 24 h.

Cell Lines
Mouse macrophage-like cell line RAW264.7 was purchased from DS Pharma Biomedical Japan (Osaka, Japan). RAW264.7 cells were cultured in DMEM with high-glucose, L-glutamine, and sodium pyruvate (Thermo Fisher Scientific), supplemented with 10% FBS and 1% penicillin and streptomycin. The cells were treated with 2 µg/mL murine BAFF-R Fc (Enzo Life Sciences, Plymouth Meeting, PA, USA) or PBS. After 2 h, RAW264.7 cells were treated with 100 ng/mL murine recombinant BAFF or PBS for 24 h. To assess the soluble factors released from macrophages, RAW264.7 cells were treated with 100 ng/mL murine recombinant BAFF and 100 ng/mL LPS for 20 h, and their supernatants were added to primary cultured HSCs on the day of collection.

NF-κB Activity Assay
Nuclear protein extracts were prepared from RAW264.7 cells using the Nuclear Extract kit (Active Motif, Carlsbad, CA, USA) according to the manufacturer's protocol. As previously described, NF-κB activation was analyzed with the TransAM NF-κB Family kit (Active Motif) [13,14].

Nitrate and TNF-α Determination
The NO concentration in the culture supernatants was determined using the Quan-tiChrom Nitric Oxide Assay Kit (BioAssay Systems, Hayward, CA, USA) according to the manufacturer's protocol. TNF-α concentration in the culture supernatants was measured using the enzyme-linked immunosorbent assay (R&D Systems).

Statistical Analysis
Data were analyzed using the JMP version 11.2.0 software (SAS Institute, Cary, NC, USA). Values are shown as the mean ± standard error of the mean or standard deviation. Normally distributed, skewed data and categorical data were analyzed using unpaired t-tests, Mann-Whitney U tests, and χ 2 tests, respectively. Differences were considered statistically significant at p < 0.05.