A novel S1PR4 functional antagonist prevents nonalcoholic steatohepatitis by deactivating the NLRP3 inammasome without causing lymphopenia

Sphingosine 1-phosphate (S1P) receptors (S1PRs) are a group of G protein-coupled receptors that confer a broad range of functional effects in chronic inammatory diseases and metabolic diseases. S1PRs may also be involved in the development of non-alcoholic steatohepatitis (NASH), but the specic subtypes involved and the mechanism of action are unclear. Here we show that the livers of various mouse models of NASH as well as Kupffer cells had a particularly strong expression of S1pr4. Accordingly, genetic depletion of S1pr4 protected the mice against hepatic inammation and brosis. Moreover, SLB736, a novel selective functional antagonist of SIPR4 prevented diet-induced NASH in mice without lymphopenia. S1P increased the expression of S1pr4 in Kupffer cells and activated the NLRP3 inammasome through PLC/IP3/IP3R-dependent [Ca++] signaling. SLB736 treatment or S1pr4 depletion in Kupffer cells inhibited LPS-mediated Ca++ release and deactivated the NLRP3 inammasome. S1PR4 antagonism may be a novel therapeutic strategy for NASH.


Introduction
Nonalcoholic fatty liver disease (NAFLD) has become a major health issue worldwide 1 . Approximately 10-20% of patients with NAFLD develop non-alcoholic steatohepatitis (NASH), an advanced stage of NAFLD that may subsequently progress to liver cirrhosis and hepatocellular carcinoma. The mechanism by which simple steatosis progresses to NASH and liver brosis is not completely understood, and an effective treatment for halting the progression of NASH is yet to be discovered 2,3 . Sphingosine 1-phosphate (S1P) is a bioactive sphingolipid that in uences a wide range of important cellular processes by activating ve G protein-coupled receptors (S1PR1-5) 4,5 . Receptor-mediated S1P signaling has become attractive therapeutic targets in several diseases such as chronic in ammatory disease, autoimmunity, cancer, and metabolic disease [6][7][8][9][10] . In the liver, S1PR2 participates in cholestasisinduced liver injury 11,12 and other causes of hepatic brosis 13 . S1PR1 and S1PR3 are involved in hepatic stellate cell motility and activation 14 and play a crucial role in the angiogenic process required for brosis development 15 . Targeting S1PRs was shown to be a promising strategy for treating NASH after the recent preclinical success of FTY720 16 , a drug for multiple sclerosis 17 . FTY720 (Fingolimod; 2-amino-2-[2-(4-noctylphenyl)ethyl]-1, 3-propanediol hydrochloride), a non-selective modulator of S1PRs (S1PR1, 3, 4, and 5), has been shown to prevent the development of alcoholic liver disease 18 , NAFLD 19 and NASH 16 in murine models. However, a widespread use of FTY720 in NASH has been hampered by its lymphopenic effects.
In the present study, we have identi ed S1PR4 as a signi cant novel factor in the pathogenesis of NASH. We found that S1pr4 mRNA expression was signi cantly high in the liver of various diet-induced murine models of NASH. S1pr4 heterozygous knockout (S1pr4 +/-) mice were protected from high-fat, highcholesterol diet (HFHCD)-induced NASH and hepatic brosis by showing minimal NLRP3 in ammasome activation in Kupffer cells. In order to provide further insights into the biological role of S1PR4, we developed and characterized a S1PR4-selective modulator SLB736, which acted as a functional antagonist of S1PR4. SLB736 was effective in preventing the development of NASH and brosis via inhibiting the activation of NLRP3 in ammasome in Kupffer cells. Collectively, our results suggest that S1PR4 is a potential target for the treatment of NASH and hepatic brosis.

Results
Role of S1PR4 in the pathogenesis of NASH. We rst investigated which type of S1PR isoforms is activated in the murine models of NASH. HFHCD-feeding is one of the animal models that closely resemble the clinical characteristics of NASH 1,20 . Despite its recognized effect in body weight 20,21 , the methionine-and choline-de cient diet (MCDD) has long been used as a valuable model of NASH with respect to steatosis, in ammation, and brosis. Among other diet-induced murine models that closely resemble the clinical characteristics of NASH 21,22 , we also used the Western diet (WD) and choline de cient, L-amino acid-de ned, high-fat diet (CDA+HFD) 23 . Interestingly, S1PR4 was only the isoform that consistently showed increased mRNA expression in the livers of mice fed HFHCD, MCDD, WD, or CDA+HFD; the expression of S1pr2 and S1pr1 were only increased in mice fed WD, and that of S1pr3 expression was only increased in mice fed HFHCD or WD (Fig. 1a).
To address which cell types are responsible for this S1pr4 upregulation, we examined the expression levels of S1pr4 in primary hepatocytes, Kupffer cells, and hepatic stellate cells (HSC). S1pr4 expression was rarely detected in hepatocytes and HSC; in contrast, S1pr4 expression was signi cantly higher in Kupffer cells isolated from HFHCD-fed mice than those isolated from control mice (Fig. 1b,   Supplementary Fig. 1). S1PR4 was reported to be speci cally expressed in myeloid cells such as dendritic cells and macrophages 24 , but its role in the pathogenesis of NASH is largely unknown. We thus tested the possible involvement of S1PR4 in the development of NASH by using genetic modulation. As homozygous knockout of S1pr4 resulted in embryonic lethality, we used heterozygous knockout (S1pr4 +/-) mice.
Compared with HFHCD-fed WT mice, HFHCD-fed S1pr4 +/mice showed signi cantly lower degrees of hepatic in ammation and brosis (Fig. 1c-e); however, the degree of hepatic steatosis was similar regardless of the S1pr4 genotype (Fig. 1c,f). S1PR4 is necessary for the activation of NLRP3 in ammasome in Kupffer cells. NLRP3 in ammasome is involved in the pathogenesis of various in ammatory and metabolic diseases including arthritis, diabetes, and atherosclerosis [25][26][27] . Recent evidence also suggested that NLRP3 in ammasome activation in Kupffer cells is an important contributor to NASH and liver brosis 28 , and NLRP3 in ammasome blockade by a small molecule was shown to reduce liver in ammation and brosis in experimental NASH in mice 29 . NLRP3 in ammasome is prominently expressed in Kupffer cells and moderately in HSC, and IL-1β produced by NLRP3 in ammasome promotes the proliferation and transdifferentiation of HSC to induce liver brosis 30 . The rst and the best-studied small molecule acting on S1PRs so far is FTY720, a myriocin-derived sphingolipid-like compound. Previous studies showed that FTY720 prevented the development of NASH 16 . Even though FTY720 affects multiple isoforms of S1PRs, S1PR1 is regarded as the main target of FTY720. Thus, in addition to S1PR4, S1PR1 may be a critical factor in NASH development. However, in our study, shRNA-mediated knockdown of S1pr1 (Fig. 2a) did not signi cantly reduce the HFHCD-induced hepatic in ammation and brosis (Fig. 2b), nor did it signi cantly affect the markers for in ammation (Fig. 2c).
Kupffer cells from S1pr4 +/mice had a signi cantly lower degree of lipopolysaccharide (LPS)-and ATPinduced increases in interleukin-1β (IL-1β) production, whereas mice depleted of S1pr1 by shRNA did not show signi cant differences from WT mice (Fig. 2d, e). These results suggest that S1pr4, but not S1pr1, is necessary for NLRP3 in ammasome activation in Kupffer cells. Collectively, these data indicate that S1PR4 is the critical mediator of the development of NASH.
SLB736 prevents the development of NASH and brosis. Administration of SLB736 to HFHCD-fed mice prevented the development of NASH and hepatic brosis (Fig. 4a, b) without affecting liver triglyceride (TG) levels (Fig. 4c). Interestingly, the administration of SLB736 did not reduce the number of lymphocytes, which is a well-known adverse effect of FTY720 through its effect on S1PR1 39,40 (Fig. 4d).
Whereas treatment with SLB736 did not signi cantly reduce the diet-induced increases in the mRNA level of S1pr4 (Fig. 4e), the protein level of S1PR4 was signi cantly decreased upon treatment with SLB736 ( Fig. 4f), thus signifying that SLB736 carry functional antagonistic roles on S1PR4 in vivo.
We further investigated whether SLB736 shows a similar preventive effect in other diet-induced murine models of NASH. We found that similar to HFHCD-fed mice, mice fed MCDD or CDA+HFD developed NASH along with increases in the expression of in ammation and in ammasome markers, which were effectively nulli ed by the administration of SLB736 (Supplementary Fig. 3a-d). To further demonstrate the therapeutic effect of SLB736, we administered SLB736 to mice fed MCDD for 4 weeks, a time point at which hepatic steatosis was evident ( Supplementary Fig. 3e). Administration of SLB736 for 4 weeks ameliorated NASH and brosis in these mice ( Supplementary Fig. 3f). S1PR4-dependent calcium release from ER plays a pivotal role in the priming of NLRP3 in ammasome in Kupffer cells. The activation of NLRP3 in ammasome is achieved through two sequential steps-signal 1 (priming) and signal 2 (activation) 41 : signal 1 is provided by microbial molecules or endogenous cytokines and leads to the upregulation of NLRP3 and pro-IL-1β through the activation of the transcription factor NF-kB, and signal 2 is triggered by ATP, pore-forming toxins, viral RNA, and particulate matters. Interestingly, we found that treatment with SLB736 signi cantly decreased IL-1β production in Kupffer cells treated with LPS and ATP (Fig. 5a). Interestingly, SLB736 decreased the expression of Nlrp3 and Il-1β (Fig. 5b) and the phosphorylation of NF-kB in LPS-primed primary Kupffer cells (Fig. 5c), suggesting that SLB736 inactivates the NLRP3 in ammasome from signal 1. Similarly, LPS-induced increases in Nlrp3 and IL-1β were signi cantly nulli ed in S1pr4 +/-Kupffer cells (Fig. 5d). Intracellular ions such as K + , Ca ++ , and Clhave signi cant roles in the activation of the NLRP3 in ammasome 42 . Among them, intracellular Ca ++ signaling plays one of the major roles in the activation of NLRP3 in ammasomes 43 . Accordingly, treatment with the [Ca ++ ] chelator BAPTA-AM in Kupffer cells signi cantly decreased the IL-1β production in response to LPS-and ATP-stimulation as well as the LPSinduced increases in the expression of Nlrp3 and IL-1β (Fig. 5e, f).
Phospholipase C (PLC)-dependent changes in [Ca ++ ] are among the downstream signaling of various S1PRs 44 . Activation of PLC triggers the release of inositol trisphosphate (IP 3 ) from phosphatidylinositol 4, 5-bisphosphate (PIP 2 ), and [Ca ++ ] is released to the cytosol when IP 3 interacts with IP 3 receptor (IP 3 R) located at the endoplasmic reticulum membrane 45 . In our experimental setting, treatment with PLC inhibitor U73122 or IP3R inhibitors Xes-c and 2-APB signi cantly decreased the LPS-mediated increases in the expression levels of Nlrp3 and Il-1β (Fig. 5g, h) as well as the production of IL-1β in response to LPS-and ATP-stimulation (Fig. 5i). Taken together, these results indicate that increases in [Ca ++ ] release from the ER through the PLC/IP 3 R axis play an important role in the activation of the NLRP3 in ammasome 43,46 .  48 , was signi cantly decreased in S1pr4 +/cells or cells treated with SLB736 (Fig. 6e, f). However, the ATP-induced increase in IP-one was not decreased in LPS-primed S1pr4 +/-Kupffer cells or cells treated with SLB736 (Fig. 6g, h). These results collectively indicate that S1PR4 is required for the calcium signaling associated with signal 1 but not signal 2 of the NLRP3 in ammasome activation. S1P activates the priming of NLRP3 in ammasome by the S1PR4/PLC/IP 3 axis. We examined the possible role of the S1P/S1PR4 axis in the priming of the NLRP3 in ammasome. Sphingosine kinase (SK) catalyzes the formation of S1P from the precursor sphingosine 4 . Interestingly, expression of Sk1 was profoundly increased in the liver and in Kupffer cells but not in hepatocytes of HFHCD-fed mice (Fig.  7a-c). S1P signi cantly increased the expression level of S1pr4 in Kupffer cells. S1P also signi cantly increased the expression levels of Nlrp3 and Il-1β, an effect that was dampened by pretreatment with SLB736 (Fig. 7d) and in S1pr4 +/-Kupffer cells (Fig. 7e). S1P also stimulated the phosphorylation of NF-kB in Kupffer cells, and this was reduced by treatment with SLB736 (Fig. 7f) and in S1pr4 +/-Kupffer cells (Fig. 7g). Pretreatment with BAPTA-AM, U73122, XesC, or 2-APB signi cantly reduced the S1P-mediated induction of Nlrp3 and Il-1β expression (Fig. 7h). These results suggest that extracellular S1P may act as a paracrine modulator of the priming of the NLRP3 in ammasome in Kupffer cells through the PLC/IP 3 /IP 3 R signaling axis. Discussion S1PR4 is speci cally expressed in myeloid cells such as dendritic cells and macrophages and regulates antigen-presenting cells to shape the T cell effector functions 49 . S1PR4 is also required for the differentiation of plasmacytoid dendritic cells 24 and regulates the production of interferon-α thereof 50 . However, compared with other S1PRs, our knowledge of the physiological relevance of S1PR4 has been modest 37 . In the present study, we found for the rst time that S1PR4 in Kupffer cells plays an important role in the pathogenesis of NASH by activating the NLRP3 in ammasome. This is in line with a previous study that reported the upregulation of S1PR4 in human samples of liver cirrhosis 51 .
As a major reservoir of intracellular [Ca ++ ], the ER plays a critical role in the regulation of intracellular [Ca ++ ] regulation 45 . Activation of the IP 3 R, a Ca ++ -release channel on the ER surface, is triggered by IP 3 , a product of PLC-mediated PIP 2 cleavage. We found that LPS sequentially activated PLC and IP 3 R in Kupffer cells to increase [Ca ++ ] and to activate the NLRP3 in ammasome, and that this reaction was abrogated by genetic depletion of S1pr4 or treatment with SLB736. In addition to LPS, we found that S1P can also prime the NLRP3 in ammasome to activate it. Interestingly, expression of Sk1 was induced in Kupffer cells by HFHCD feeding, and S1P increased the expression of S1pr4 in Kupffer cells. Accordingly, a previous study showed that the overloading of saturated fatty acids induces Sk1 in hepatocytes to initiate proin ammatory signaling 52 . On the other hand, in HFHCD-fed mice, Sk1 expression was induced in Kupffer cells but not in hepatocytes. We thus suggest that S1P produced by SK1 from Kupffer cells induces S1PR4 in a paracrine manner to activate the NLRP3 in ammasome (Fig. 7i).
NAFLD occurs mostly in obese individuals, and insulin resistance and deregulation of the lipid metabolism increase the risk of NAFLD and NASH 1 . Although lifestyle modi cation is the rst-line treatment for patients with NASH, it is usually unsuccessful. Therefore, many agents for the treatment of NASH by targeting different pathways are under development 2 , and several compounds have shown promising histologic results in phase IIa studies 3,53 . However, it was pointed out that histologic NASH is not an independent predictor of long-term mortality and that the stage of brosis is the only robust and independent predictor of liver-related mortality 3,54 . In this regard, targeting the NLRP3 in ammasome activation, which plays a central role in hepatic in ammation and brosis 30 , is increasingly recognized as a promising strategy for developing an e cient therapy against NASH 29,55,56 . In accordance, our study showed that SLB736 was effective in preventing the development of NASH and brosis by deactivating the NLRP3 in ammasome.
One of the interesting ndings is that treatment with SLB736 signi cantly mitigated hepatic in ammation or brosis while not signi cantly affecting the degree of hepatic steatosis. Similarly, S1pr4 +/mice did not show improvement in hepatic steatosis, despite the notable bene cial effects on in ammation and brosis. This suggests that S1PR4 speci cally targets Kupffer cells and not hepatocytes. Lipotoxic hepatocyte injury may be the primary lesion that triggers the activation of NLRP3 in ammasome in Kupffer cells 57,58 . Thus, even though SLB736 shows promising effects on preventing hepatic in ammation and brosis, combination with other drugs that can reduce steatosis or lipotoxic injury of hepatocytes 2 may be more e cacious and may be used as an ideal therapy for the treatment of NASH.  Fig. 2).
Mice and diet. Mice were housed in ambient temperature (22 ± 1°C) with a 12:12 h light-dark cycle and free access to water and food. After the indicated time of diet feeding, the mice were fasted for overnight before they were euthanized. All animal use and experiment protocols were approved by the Institutional Animal Care and Use Committee of Asan Institute for Life Sciences, Seoul, Korea.
Kupffer cells isolation and identi cation. Kupffer cells were isolated from mice by collagenase digestion, gradient centrifugation, and selective adherence, 59  were removed and placed in 60-mm petri dishes. The livers were frittered with forceps in RPMI1640 (LM 011-01, Welgene) supplemented with 10% (vol/vol) FBS (16000-044, Gibco). The cell suspensions were ltered through a sterile 100-μm nylon cell strainer (352360, Falcon) to remove undigested tissues and connective tissues. The cells were centrifuged for 5 min at 50 × g at room temperature to remove hepatocytes. The supernatants were transferred to clean 50 ml tubes. The supernatants were centrifuged at 1600 rpm (4℃) for 10 min, and the cell pellets were re-suspended in 20% OptiPrep and gently layered on OptiPrep gradient (20,11.5% and HBSS) and centrifuged at 3000 rpm at 4℃ for 17 min with the brake option off. Subsequently, the upper layers were removed and the cell fraction between 20% OptiPrep and 11.5% OptiPrep gradient were collected without contamination from the pellets. The collected layers were washed twice with RPMI1640 supplemented with 10% (vol/vol) FBS, and plated into 12-well or 24-well tissue culture plates. At 10 min after seeding, non-adherent cells (cell debris or blood cells) were removed by aspiration and fresh media were added. The next day, the cells were washed twice with 1 × PBS, and the attached Kupffer cells were cultured for another 48 h, at which point they were ready for experimental use.
Primary HSC isolation. Selective macrophage depletion was achieved with a single intraperotoneal injection of clodronate (20 mg/ ml) according to manufacturer's instructions (FormuMax, Sunnyvale, CA, USA). After 24 hrs, primary HSCs were isolated using the same protocol used in the isolation of Kupffer cells, and the cell fraction between the upper layer and the 20% OptiPrep gradient were collected without contamination from the pellets. After centrifugation, the cells were seeded into culture plates in Dulbecco's modi ed Eagle's medium containing fetal serum at 37°C. The culture medium was changed and the RNA was isolated from primary HSCs 60 .
Treatment with SLB736 or FTY720 in vivo. Mice were administrated SLB736, FTY720 (1 mg/kg body weight each) or vehicle (0.9% NaCl) via oral gavage every day for 5 days/week for the indicated periods. After the indicated period of treatment, the mice were fasted for overnight and sacri ced. The liver tissues were quickly removed and kept frozen at -70°C for subsequent analysis.
Liver TG contents. TG content in the liver was determined in duplicate using the Sigma Triglyceride (GPO-Trinder) kit.
SLB736 treatment in vitro. Cells were treated with chemicals at the indicated doses or sterile water (control) for 2 h. After washing twice with PBS, the cells were stimulated with LPS (L2880, Sigma-Aldrich) at a concentration of 100 ng/ml for 3 h, and then 1 mM ATP (A6419, Sigma-Aldrich) was added for 30 min.
Real-time PCR analysis. Total RNA isolated from each sample was reverse-transcribed and the target cDNA levels were quanti ed by real-time PCR analysis using gene-speci c primers (Supplementary Table  1). Total RNA was isolated using TRIzol (Invitrogen), and 1 μg of each sample was reverse-transcribed with random primers using the Reverse Aid M-MuLV Reverse Transcription Kit (Fermentas, Amherst, NY, USA). The relative expression levels of each gene were normalized to that of 18S rRNA or Tbp.
Western blot analysis. Cell and liver samples were homogenized in lysis buffer ( (40-50 μg) was subjected to immunoblotting with primary antibodies: antibodies against phosphorylated NF-kB (#3033) and NF-kB (#6956) were purchased from Cell Signaling (Danvers, MA, USA) and the antibody against S1PR4 was purchased from NOVUS (NBP2-24500; Centennial, CO, USA). β-actin (A5441, Sigma-Aldrich) was used as housekeeping control (Supplementary Table 2). The signal intensities of protein bands were quanti ed with the ImageJ software (NIH, Bethesda, MD, USA) and normalized using the intensity of the loading control.
IL-1β measurement. Mouse IL-1β in cell culture supernatants were measured using the mouse IL-1β /IL-1F2 Quantikine ELISA kit (DY401, R&D Systems, Minneapolis, MN, USA). Intracellular IP-one measurement. PLC activity was tested with the IP-one ELISA (72IP1PEA; Cisbio, Bedford, MA, USA), in which Kupffer cells were stimulated with LPS or S1P and then the cell culture medium was replaced with fresh medium. Intracellular IP-one, a surrogate measure for the level of inositol triphosphate, was measured after treatment with LiCl (50 mM) to prevent the degradation of IPone into myo-inositol. The level of inositol triphosphate in cell lysates was measured using ELISA.
Determination of S1PR4 localization in C6 glioma cell line. Stable C6 glioma cells expressing EGFPconjugated S1PR4 were prepared by infection with retrovirus bearing S1PR4-EGFP fusion construct (kindly provided by Dr. Jerold Chun at The Sanford Burnham Prebys). S1PR4 internalization and recycling were assessed as previously described 36 . In brief, cells were plated on poly-L -lysine (100 mg/mL)-coated coverslips, cultivated, serum-deprived, and then used for experiments. The cells were treated with vehicle (0.1% fatty acid-free BSA), S1P, FTY720-P, or SLB736 for 0.5 h; in some cases, the cells were washed and further incubated in the presence of cycloheximide (5 mg/mL) for 2 h or 4 h. At the end of each experiment, the cells were xed in 4% paraformaldehyde and mounted with Vectashield. S1PR4 localization in cells was assessed by detecting the EGFP signal using a laser scanning confocal microscopy (Eclipse A1+, Nikon, Tokyo, Japan). S1PR β-Arrestin Assay. β-Arrestin recruitment assays for S1PR activity were performed by DiscoveRx (Fremont, CA, USA).
Calcium analysis by confocal microscopy. Kupffer cells were plated on 35 mm imaging dish (81156, Ibidi, Gräfel ng, Germany) at a density of 0.1 × 10 6 cells and incubated with Fluo-4/AM. Images of untreated cells were acquired at t = 0, and the cells were treated with 1 mg/ml LPS or 1 mM ATP in RPMI 1640. The cells were imaged for 5 min at 5 s intervals on a Zeiss LSM780 Confocal Imaging System using the 488 nm laser and emission in the range of 500-600 nm. The images were analyzed using Zen 2012 SP5 software by creating surfaces to encompass the volume of each cell. The absolute intensity for all cells in a eld at different time points was obtained, and normalized to t=0 to calculate the fold increases in intensity. Data are displayed as the relative intensity of cells in a eld.
Statistical analysis. Data are expressed as mean ± standard error of the mean (SEM). Unpaired two-tailed Student's t-tests were used to compare variables between groups, and one-way ANOVA was used to compare variables among multiple groups. For the comparison of multiple measurements made at different time points, one-way repeated-measures ANOVA was used. Bonferroni correction was applied for post hoc analysis of the multiple comparisons. All statistical tests were conducted according to two-sided sample sizes and were determined on the basis of previous experiments that used similar methodologies.