Correction of Liver Steatosis by a Hydrophobic Iminosugar Modulating Glycosphingolipids Metabolism

The iminosugar N-(5′-adamantane-1′-yl-methoxy)-pentyl-1-deoxynoijirimycin (AMP-DNM), an inhibitor of glycosphingolipid (GSL) biosynthesis is known to ameliorate diabetes, insulin sensitivity and to prevent liver steatosis in ob/ob mice. Thus far the effect of GSL synthesis inhibition on pre-existing NASH has not yet been assessed. To investigate it, LDLR(−/−) mice were kept on a western-type diet for 12 weeks to induce NASH. Next, the diet was continued for 6 weeks in presence or not of AMP-DNM in the diet. AMP-DNM treated mice showed less liver steatosis, inflammation and fibrosis. Induction of fatty acid beta-oxydation was observed, as well as a reduction of plasma lipids. Our study demonstrates that AMP-DNM treatment is able to significantly correct pre-existing NASH, suggesting that inhibiting GSL synthesis may represent a novel strategy for the treatment of this pathology.


Introduction
The metabolic syndrome represents a combination of health risk factors including abdominal obesity, insulin resistance, dyslipidemia and hypertension. Non Alcoholic Fatty Liver Disease (NAFLD) is the hepatic manifestation of the metabolic syndrome. NALFD includes a large variety of liver derangements ranging from simple fat accumulation in the parenchymal cells (steatosis) to non-alcoholic steatohepatitis (NASH) including inflammation and varying degrees of fibrosis. NAFLD is estimated to affect at least 20% of the general adult population and over 50% of the obese population [1,2]. In about 30% of NAFLD patients, the disease can progress into steatohepatitis and cirrhosis [3]. It is expected that as the prevalence of obesity and metabolic syndrome rises, NAFLD-associated diseases will be an increasing healthcare concern and therapeutic measures are thus needed to address this major health problem [4,5]. Fat accumulation in hepatocytes is the result of an imbalance between triglyceride synthesis and degradation. An increased flux and/or endogenous synthesis of free fatty acids (FFA) may lead to accumulation of triglycerides within the liver when mitochondrial b-oxidation and VLDL production and secretion are not sufficient to handle the FFA load. The molecular mechanisms behind this fat accumulation in hepatocytes leading to NASH still remain unclear. Hepatic inflammatory cell recruitment and inflammatory cytokines appear to play a key role in this process and dietary cholesterol has been proposed to be an important contributor for the development of the pathology in hyperlipidemic mouse models [6,7]. We and others have previously shown that two distinct classes of inhibitors of glucosylceramide (GlcCer) synthase, the rate limiting enzyme involved in glycosphingolipid (GSL) biosynthesis, improved glycemic control, decreased insulin resistance and reduced fatty liver development in animal models of obesity i.e. dietinduced obesity (DIO) mice and ob/ob mice [8][9][10][11][12][13]. A particular role for the ganglioside GM3 in insulin sensitivity has become apparent in recent times. Firstly, Yamashita et al. reported that mice deficient in GM3 synthase, and thus deficient in the ganglioside GM3 and higher gangliosides like GM2-glycol, are protected against insulin resistance and obesity [14]. Inokuchi and co-workers showed that the ganglioside GM3 interacts directly with a lysine residue in the insulin receptor [15]. The role of gangliosides in insulin sensitivity has recently been reviewed [16,17]. The use of the iminosugar N-(59-adamantane-19-ylmethoxy)-pentyl-1-deoxynoijirimycin (AMP-DNM) and ceramide-mimic Genz-123346 [(1R,2R)-nonanoic acid[2-(29,39-dihydro-benzo [1,4] dioxin-69-yl)-2-hydroxy-1-pyrrolidin-1-ylmethylethyl]-amide-L-tartaric acid salt], both inhibitors of GlcCer synthase, clearly improved liver steatosis [13]. However, the ability of AMP-DNM to correct liver steatosis and even NASH when it already has developed, has not yet been investigated.
In the present study, LDLR(2/2) and APOE*3 Leiden mice, two models sensitive to liver steatosis were allowed to develop NASH for 12 weeks on high fat-high cholesterol diet and were subsequently treated with AMP-DNM for 6 weeks. We observed that despite the maintenance of the animals on a high fat-high cholesterol diet, AMP-DNM treatment reduced plasma lipids and that the steatosis, the inflammatory and fibrotic status were profoundly improved.

AMP-DNM treatment ameliorates hyperlipidemia and reverses hepatic steatosis in LDLR(2/2) mice
In the present study, 40 LDLR(2/2) mice were fed a westerntype diet to induce NASH and were subsequently treated with two different doses of AMP-DNM to achieve the dosing level of 50 and 100 mg AMP-DNM. kg bw 21 .day 21 . AMP-DNM supplementation did not affect the behaviour of the animals. At the dose of 100 mg AMP-DNM, the bodyweight of the animals was reduced compared to the control group and food consumption slightly decreased after the switch of diet (table S1). This was not observed in the group treated with 50 mg AMP-DNM. AMP-DNM treatment induced a dose-dependent decrease of plasma GlcCer and ceramide (table 1). Whereas the amount of GlcCer was dosedependently decreased in the livers of treated animals, ceramide amounts were not changed (table 1). AMP-DNM also dosedependently decreased plasma triglycerides, FFA, and cholesterol ( fig. 1A-C). We also determined the effect of AMP-DNM treatment on hepatic concentration of higher glycosphingolipids: lactosylceramide (Laccer); globotriaosylceramide (Gb3) and gangliosides (table 2). The neutral glycosphingolipids LacCer and Gb3 were reduced by AMP-DNM treatment. The most abundant ganglioside was GM2-glycol. It was significantly reduced by drug treatment. This was not seen for the less abundant GM3.
We then determined the effect of AMP-DNM on liver steatosis. First liver weight was significantly reduced in a dose-dependent fashion (25% reduction with 50 mg AMP-DNM and 32% reduction with 100 mg AMP-DNM, p,0.01). Analyses of morphology and fat content on liver sections stained with H&E ( fig. 2A) and Oil-red-O ( fig. 2B) showed that after 12 weeks on western-type diet feeding, livers of LDLR(2/2) contained micro and macro-lipid droplets, which were slightly increased in number and in size after 18 weeks of diet. By contrast, the animals treated with 50 mg AMP-DNM showed a dramatic reduction in the amount of hepatic lipid droplets, whereas at the dose of 100 mg AMP-DNM, virtually no lipid droplets were present. Biochemical analysis of lipids confirmed a dose-dependent reduction of hepatic triglycerides ( fig. 2C). Free cholesterol content was significantly reduced at the dose of 100 mg AMP-DNM ( fig. 2D).
AMP-DNM treatment modulates hepatic lipogenesis and glucose production in LDLR(2/2) mice AMP-DNM ameliorates glucose homeostasis and insulin sensitivity in obese rodents [9,11,12,18]. In LDLR(2/2) animals, we observed that blood glucose levels were significantly lower at the end of the treatment compared to control animals ( fig. 3A). Insulin levels also tend to decrease with the treatment (fig. 3B). The homeostasis model assessment (HOMA) index was significantly reduced in treated animals ( fig. 3C). As a consequence of the improved glucose homeostasis, the percentage of glycated haemoglobin was notably reduced as well ( fig. 3D).
Sterol regulatory element binding protein 1c (SREBP1c) is a transcription factor activated by insulin and responsible for the transcription of numerous genes involved in fatty acid synthesis. Despite insulin resistance for the glucose homeostasis pathway, it has been reported that hyperinsulinemia triggers the activation of SREBP-1c and consequently fatty acid synthesis [19]. As anticipated, we observed that the western-type diet induced an up-regulation of SREBP1 ( fig. 4) and some of its target genes, i.e. fatty acid synthase (FAS) and stearoyl-CoA desaturase (SCD1). AMP-DNM treatment normalized expression of these genes. Moreover, we observed that the western-type diet induced expression of PDK4 and glucose-6-phosphatase. Animals treated with AMP-DNM showed corrected expression of both genes. Of interest, CPT1a and PCG1a were increased by AMP-DNM treatment and levels of LCA mRNA tended to be also higher with drug treatment at the highest dose ( fig. 4). These findings suggest that fatty acid beta-oxidation is increased by AMP-DNM treatment. Taken together, our data strongly suggest that the treatment with AMP-DNM increases insulin sensitivity in the livers of treated animals, reduces lipogenesis and promotes fatty acid beta-oxidation. Gene expression profiles of inflammatory markers were also monitored ( fig. 5B). After 18 weeks of western-type diet the expression of the macrophage marker F4/80 was increased 4-fold, the chemokine (C-C motif) ligand 2 (CCL2)/monocyte chemoattractant protein 1 (MCP-1) 15-fold, and CD11c 3-fold.

AMP-DNM treatment reduces hepatic inflammation and fibrosis
We also analysed expression levels of Glycoprotein Nonmetastatic Melanoma Protein B (GPNMB) also called osteoactivin, a new marker of monocyte to macrophage differentiation [20]. On normal chow diet, GPNMB was poorly expressed but feeding with the western-type diet resulted in a strong up-regulation of its expression (i.e. 300-fold increase after 18 weeks of western-type diet). AMP-DNM treatment resulted in a dose-dependent reduction of expression of each of these markers. At the dose of 100 mg AMP-DNM, the reduction was even larger. It suggests that the highest dose of AMP-DNM completely corrected the effect of the western-type diet on liver inflammation.
An important feature in NASH is the progression towards liver fibrosis. To monitor this aspect, we stained sections with Siriusred. In all groups of animals, positive collagen staining was observed and was localized near vessels of periportal and centrolobular regions. After 12 and 18 weeks of western-type diet, collagen staining in these two areas was higher than in animals on chow diet. Collagen content was slightly reduced as a consequence of AMP-DNM treatment ( fig. 6A). Immunohistochemistry staining of activated stellate cells with anti-SMA antibody and gene expression analysis of SMA confirmed this finding ( fig. 6B, C). Gene expression profiles of the collagen 1 alpha (col 1a) and transforming growth factor beta (TGFb) were also monitored ( fig. 6C). Both these markers were considerably increased with the western-type diet, whereas the AMP-DNM treatment normalized their expressions levels.

AMP-DNM also corrects hepatic steatosis in APOE*3 Leiden
We examined also the impact of AMP-DNM treatment on fatty liver of APOE*3 Leiden mice that were exposed for 12 weeks to a 1% cholesterol, 15% fat diet. The switch of diet induced a slight reduction of the animal bodyweights treated with either 50 or 100 mg AMP-DNM. No major changes were observed concerning the food consumption (table S2). The liver of these animals after 12 weeks of diet showed steatosis, but no clear fibrosis as compared to the LDLR(2/2) mice ( fig. 7A). Animals were subsequently switched for 6 weeks to 0.25% cholesterol, 15% fat diet with or without AMP-DNM. The drug treatment lowered as expected the levels of glucosylceramide in plasma and liver of treated animals (tables S3 and S4) and it dramatically corrected liver steatosis that was still prominent in the control animals ( fig. 7A). AMP-DNM also resulted in corrections in liver triglyceride and cholesterol content ( fig. 7B).

Discussion
We have previously demonstrated in leptin-deficient ob/ob mice that AMP-DNM treatment reduced liver steatosis associated with a restoration of liver insulin signaling [9]. Also hepatic fatty acid synthesis and glycogen storage were normalized. Two other inhibitors of GlcCer synthase, Genz-123346 and Genz-112638, also corrected liver steatosis in diet-induced obese (DIO) mice [10,13].
A limitation of the leptin-deficient ob/ob and DIO models is that the mice do not develop steatohepatitis or liver fibrosis as is observed in humans [21]. LDLR(2/2) mice fed a western-type diet consisting of a moderate amount of fat (15%) and cholesterol (0.25%) offer a better model to investigate NASH. The animals,    showing a human ''like'' lipoprotein profile, rapidly develop hepatic steatosis and inflammation [22]. We have previously observed in this model that treatment with AMP-DNM from the start of the diet reduces liver triglyceride content [23]. To study the ability of AMP-DNM to also correct hepatic manifestations of the metabolic syndrome, LDLR(2/2) mice were kept on a western-type diet for 12 weeks to induce NASH. Next, the diet was continued for 6 weeks in the presence or absence of AMP-DNM in the diet. Our study demonstrates that AMP-DNM treatment is not only able to prevent liver steatosis, but is also able to significantly correct pre-existing NASH. AMP-DNM treated mice showed less liver steatosis and inflammation with a liver phenotype closely resembling that of matched LDLR(2/2) mice fed a standard chow diet. In APOE*3 Leiden mice, another hyperlipidemic model sensitive to hepatic steatosis, AMP-DNM treatment corrected liver accumulation of fat and reduced inflammation in a similar fashion to what was observed with LDLR(2/2) mice. In this experiment set-up, animals were first fed with a high-fat high cholesterol diet (1% cholesterol, 15% fat) for 12 weeks and fed for 6 following weeks a western-type diet (0.25% cholesterol, 15% fat) supplemented with either 0, 50 or 100 mg AMP-DNM. It has to be noted that the decrease in cholesterol in the diet tends by itself to decrease liver lipids in the control animals (table S3) whereas we observed a poor effect of the reduction of lipids in the diet on plasma cholesterol and triglycerides. The AMP-DNM treatment induced a significant and prominent decrease of these parameters in both compartment. The underlying mechanism by which AMP-DNM treatment is able to correct NASH manifestation is intriguing. In the LDLR(2/2) mice the most prominent ganglioside in liver is GM2-glycol, rather than GM3. AMP-DNM treatment led to a major reduction in GM2-glycol, accompany the correction of liver steatosis. It may be speculated that GM2-glycol, rather than GM3, causes insulin resistance in the LDLR(2/2) mice liver. Hyperinsulinemia has been proposed to promote liver steatosis [19] and hyperinsulinemic patients with NASH treated with insulinsensitizing agents show improvements of liver steatosis [24]. It is well known that inhibitors of GlcCer synthase like AMP-DNM and Genz-123346 are also able to improve insulin sensitivity in various rodent models by virtue of lowering glycosphingolipids [16,17]. LDLR(2/2) mice treated with AMP-DNM in this study showed an improvement of the HOMA index and a concomitant decrease of hepatic lipogenesis and increased beta-oxidation. Recent experiments with ob/ob mice also revealed that AMP-DNM causes major changes in hepatic lipid metabolism. The rates of fatty acid beta-oxidation and glucose oxidation were determined using metabolic cages, revealing that shortly after exposure to AMP-DNM, fat oxidation was nearly doubled (0.07 to 0.16 kcal.h 21 in the nocturnal period and 0.10 to 0.19 kcal.h 21 in the diurnal period). On the opposite, carbohydrate oxidation was markedly reduced by AMP-DNM treatment (0.43 to 0.28 kcal.h 21 in the nocturnal period and 0.34 to 0.20 kcal.h 21 in the diurnal period [25]. Other researchers have also reported beneficial effect of iminosugars on hepatic fatty acid oxidation. Similar to our findings with AMP-DNM, Tsuduki et al reported that intake of 1-deoxynojirimycin promotes hepatic beta-oxidation and suppresses lipid accumulation in rat liver [26]. Kobayashi and co-worker reported similar effects induced by mulberry extracts known to be rich in simple iminosugars [27]. It is likely that these metabolic changes, following improved insulin sensitivity, contrib- ute to the lowering of triglycerides in hepatocytes and the correction of steatosis observed in LDLR(2/2) and APOE*3 Leiden mice (figure S1 and S2).
Of note, lowering of glycosphingolipids by specific genetic knock-down of GlcCer synthase in hepatocytes does not reduce liver triglycerides in mice exposed to a high fat diet [28]. It seems unlikely that the insulin-sensitizing effect of AMP-DNM alone explains the observed effects on hepatic inflammation and fibrosis. Presumably the earlier described anti-inflammatory actions of AMP-DNM play a key role as well [12,29].
In the past years, several studies showed that ER stress response plays an important role in lipid metabolism and is linked to fatty liver disease [30][31][32]. The role of glycosphingolipids in ER stress response has not been extensively studied yet. A recent study on pancreatic MIN6 b-cell has revealed that ER stress is associated with increased GlcCer and that it is ameliorated by transfection of cells with glucosylceramide synthase [33]. Ceramide concentration is not changed. In our previous study on AMP-DNM treatment of ob/ob mice, microarray data for liver did not reveal any significant changes in expression of proteins involved in pathways associated with ER stress [9]. It nevertheless remains of interest to study in more detail the possible role of (glyco)sphingolipids in ER stress responses and related effects on lipid homeostasis.
Liver natural killer T (NKT) cells are thought to be able to ameliorate steatosis and NKT cells numbers are reduced in steatotic livers of human subjects and ob/ob mice [34,35]. We anticipated that AMP-DNM treatment might modulate the hepatic NKT cell population. Non-parenchymal cells were therefore isolated from livers of treated animals and screened for NKT markers by flow cytometry (fig. S3A). The percentage of hepatic NKT cells upon AMP-DNM treatment did not change, suggesting that this cell population was not causing the observed correction of NASH. CD4+ -and CD8+ T-cell populations were also unaffected by AMP-DNM ( fig. S3B).
AMP-DNM exerts a positive effect on cholesterol homeostasis by promoting biliary sterol secretion and fecal excretion [23]. Increased hepatic free cholesterol in obese, diabetic mice is a known critical factor in the development of nonalcoholic  [36,37]. Of interest, AMP-DNM treatment at the highest dose of 100 mg. kg bw 21 . day 21 led to a significant decrease in free cholesterol content of liver of the LDLR(2/2) mice. We also observed a strong dose-dependent reduction of plasma cholesterol in LDLR(2/2) animals treated with AMP-DNM. All this was accompanied by reduction of F4/80 positive cells in the liver ( fig. S3C). AMP-DNM treatment clearly decreased (macrophage) inflammation in the liver. The positive effects on cholesterol homeostasis exerted by AMP-DNM may contribute to the reversal in hepatosteatosis. High cholesterol in the western-type diet is detrimental since it drives inflammation due to uptake of modified cholesterol-rich lipoproteins by Kuppfer cells [38]. We therefore would like to propose that AMP-DNM, by improving cholesterol homeostasis, causes a beneficial reduction of Kuppfer cell activation.
In conclusion, our study shows clearly that drug treatment is sufficient to swiftly correct the liver manifestations of NASH in two different animal models with improved insulin sensitivity, promotion of fatty acid beta-oxidation and reduction of inflammation.

Materials
AMP-DNM was synthesized as previously described [39,40]. All solvents and reagents used were of analytical grade.

Mice and diets
Experiments were performed with the approval of the local Ethical Committee for Animal Experiments. 40 female LDLR(2/2) mice (8-12 weeks-old) were fed a western-type diet (0.25% cholesterol, 15% fatty acids; Arie Blok, Woerden, the Netherlands) for 12 weeks to induce NASH and 10 mice were sacrificed. The 30 remaining mice were subdivided into 3 groups of 10 animals, one group continued the same diet for another 6 weeks, the second group received the same diet supplemented with 0.3 g AMP-DNM per kg diet to obtain the calculated dose of 50 mg AMP-DNM. kg bw 21 . day 21 and the third group received 0.6 g AMP-DNM per kg diet to obtain 100 mg AMP-DNM. kg bw 21 . day 21 . In parallel, one group of 5 LDLR(2/2) mice was kept on chow diet. In a second experiment, 40 female APOE*3 Leiden (8-12 weeks-old) were fed a high fathigh cholesterol diet (1% cholesterol, 15% fatty acids) for 12 weeks and 10 mice were sacrificed. The 30 remaining mice continued with a western-type diet (0.25% cholesterol, 15% fatty acids) supplemented or not with two different doses of AMP-DNM to obtain the calculated dose of 50 mg and 100 mg AMP-DNM. kg bw 21 . day 21 .  puncture and plasma samples were stored at 220uC. Livers were quickly excised and weighed, and parts were snap-frozen in liquid nitrogen and stored at 280uC, or fixed in 10% buffered formalin and embedded in paraffin.

Analytical procedures
Blood glucose levels were determined in plasma of fasted animals using a hand-held Glucometer (Ascensia Elite; Bayer A.G., Leverkusen, Germany) one week before the sacrifice of the animals. Hemoglobin A1c levels were measured using a single measurement A1C now device (Metrika, Sunnyvale, CA). Insulin levels were determined by enzyme-linked immunosorbent assay (Crystal Chem Inc., Downers Grove, IL). Free cholesterol, total cholesterol, and triglycerides in liver samples were determined after lipid extraction according to Folch [41] using a colorimetric enzymatic kit (Biolabo, Maizy, France). Cholesterol and triglyceride levels in plasma were determined using the same assay. FFA level in plasma were determined using a colorimetric enzymatic kit (Wako Chemicals Gmbh, Neuss, Germany). Ceramide and GlcCer were determined after Folch extraction as previously describe [11]. Plasma cholesterol concentrations in the main lipoprotein classes were determined in pooled plasma samples of each group (5 mice) separated by high performance gel filtration chromatography [42].

Histology
Paraffin embedded liver sections (7 mM) were de-waxed and stained with hematoxylin-eosin for general histology, or with 0.2% picro-sirius red to detect fibrillar collagen deposits. To detect neutral lipids, cryostat sections of 7 mM were stained with 0.3% Oil Red O. For detection of macrophages/monocytes, a rat polyclonal anti-CD68 was used (Serotec, Oxford UK) and an antialpha smooth muscle actin (SMA) antibody to visualize activated stellate cells (1A4, Abcam, Cambridge, UK). Frozen sections of livers fixed in cold acetone were incubated one hour at room temperature with CD68 (1/100 dilution) or SMA antibody (1/100 dilution) followed by incubation for 30 min at room temperature with the corresponding secondary antibodies. Visualization of the complex was done with 3,39-diaminobenzidine tetrahydrochloride (DAB, Immunologic, Duiven, The Netherlands) for 5 min. For all stainings, haematoxylin (Sigma-Aldrich, Zwijndrecht, the Netherlands) was used to counterstain. Primary antibodies were omitted in negative control samples. Images were captured with a Leica DFC 420 video camera.

Determination of mRNA levels
Total RNA was isolated from 50-100 mg of liver tissue using Trizol reagent (Invitrogen, Breda, The Netherlands) and reverse transcribed with SuperScript II Reverse Transcriptase and random hexamers (Invitrogen, Breda, The Netherlands) after treatment with RQ1 Rnase-free Dnase (1 units/2 mg of total RNA, Promega, Leiden, The Netherlands). Gene expression analysis was performed on a Bio-Rad MyIQ Single-color Real-Time PCR Detection System by using the Bio-Rad IQ SYBR Green Supermix (Bio-Rad Laboratories Inc., Hercules, CA). Expression levels were normalized to Acidic ribosomal phosphoprotein (36B4).

Statistical analysis
Values presented in figures concerning the LDLR(2/2) mice represent mean 6 SEM. Dunnett's comparison test was performed between 18 weeks controls and each other group using GraphPad Prism software. Values presented in figures concerning the APOE*3 Leiden mice represent mean 6 SEM. Dunnett's comparison test was performed between baseline at 12 weeks and each other group using GraphPad Prism software *, p,0.05; **, p,0.01; ***, P,0.001. Figure S1 Genes expression in livers of APOE*3 Leiden mice fed a high cholesterol-high fat diet (1% cholesterol, 15% fat) for 12 weeks and fed for 6 more weeks a western-type diet (0.25% cholesterol, 15% fat) supplemented with either 0, 50 or 100 mg AMP-DNM. Expression levels normalized to Acidic ribosomal phosphoprotein (36B4). Data are expressed as mean 6 SEM, n = 5. Statistical significance determined between baseline 12w and other groups with Dunnett's comparison test. *p,0.05, **p,0.01, p,0.001. Table S1 Effect of AMP-DNM treatment on bodyweight and food intake in LDLR(2/2) mice fed a western-type diet for 18 weeks, receiving in the last 6 weeks either 0, 50 or 100 mg AMP-DNM. Data are expressed as mean 6 SEM, n = 10 for bodyweight. Food intake based on the amount of food left in each cage of treatment (2 to 3 cages per treatment) at the end of each dosing week. (DOC) Table S2 Effect of AMP-DNM treatment on bodyweight and food intake in APOE*3 Leiden mice fed a high cholesterol-high fat diet (1% cholesterol, 15% fat) for 12 weeks and fed for 6 following weeks a western-type diet (0.25% cholesterol, 15% fat) supplemented with either 0, 50 or 100 mg AMP-DNM. Data are expressed as mean 6 SEM, n = 5 for bodyweight. Food intake based on the amount of food left in each cage of treatment (2 to 3 cages per treatment) at the end of each dosing week. (DOC)