Dual targeting of hepatic fibrosis and atherogenesis by icosabutate, an engineered eicosapentaenoic acid derivative

Abstract Background & Aims While fibrosis stage predicts liver‐associated mortality, cardiovascular disease (CVD) is still the major overall cause of mortality in patients with NASH. Novel NASH drugs should thus ideally reduce both liver fibrosis and CVD. Icosabutate is a semi‐synthetic, liver‐targeted eicosapentaenoic acid (EPA) derivative in clinical development for NASH. The primary aims of the current studies were to establish both the anti‐fibrotic and anti‐atherogenic efficacy of icosabutate in conjunction with changes in lipotoxic and atherogenic lipids in liver and plasma respectively. Methods The effects of icosabutate on fibrosis progression and lipotoxicity were investigated in amylin liver NASH (AMLN) diet (high fat, cholesterol and fructose) fed ob/ob mice with biopsy‐confirmed steatohepatitis and fibrosis and compared with the activity of obeticholic acid. APOE*3Leiden.CETP mice, a translational model for hyperlipidaemia and atherosclerosis, were used to evaluate the mechanisms underlying the lipid‐lowering effect of icosabutate and its effect on atherosclerosis. Results In AMLN ob/ob mice, icosabutate significantly reduced hepatic fibrosis and myofibroblast content in association with downregulation of the arachidonic acid cascade and a reduction in both hepatic oxidised phospholipids and apoptosis. In APOE*3Leiden.CETP mice, icosabutate reduced plasma cholesterol and TAG levels via increased hepatic uptake, upregulated hepatic lipid metabolism and downregulated inflammation pathways, and effectively decreased atherosclerosis development. Conclusions Icosabutate, a structurally engineered EPA derivative, effectively attenuates both hepatic fibrosis and atherogenesis and offers an attractive therapeutic approach to both liver‐ and CV‐related morbidity and mortality in NASH patients.


| INTRODUC TI ON
Although there is significant hepatic-related morbidity and mortality associated with NASH, including cirrhosis, liver failure and hepatocellular carcinoma, the major overall cause of mortality in patients with NASH is cardiovascular disease (CVD), especially in patients who do not yet have advanced cirrhosis. 1 Novel drugs for the treatment of NASH should thus ideally reduce both liver fibrosis and CVD, or at least avoid negative effects on CV-risk factors. With respect to drugs currently being developed for the treatment of NASH, both beneficial 2,3 and adverse effects on plasma lipids [4][5][6] and glycaemic control 7 are reported.
Icosabutate is a liver-targeted, semi-synthetic, eicosapentaenoic acid (EPA) derivative under clinical development for NASH (NCT04052516). In addition to avoiding esterification via an α-substitution that ensures portal vein uptake from the gut 8 rather than peripheral distribution, an oxygen substitution in the β-position limits its metabolism as a cellular energy source. The goal of the structural changes are to maximise hepatic concentrations of the free-acid form for optimal targeting of both energy metabolism and inflammation via omega-3 fatty acid responsive pathways, for example, nuclear and G-protein coupled receptors 9,10 and the arachidonic acid (AA) cascade. 11 Several lines of evidence, including the recently reported beneficial effect of aspirin in NASH patients, 12 suggest AA metabolism is involved in the progression of liver fibrosis. [13][14][15] The potential of icosabutate to target both energy metabolism and inflammation is suggested by its ability to rapidly reduce both plasma lipids and, of particular relevance to NASH, elevated liver enzymes in hyperlipidaemic subjects. [16][17][18] We recently reported that icosabutate improved early hepatic fibrosis and inflammation in a prevention design NASH rodent model. 8 However, as studies in humans are targeting established NASH, proof of efficacy in a delayed-treatment study design are more relevant. Additionally, the anti-fibrotic effects of icosabutate were compared with rosiglitazone, which has not demonstrated improvements in fibrosis in humans. 19 We have therefore evaluated the dose response effects of delayed treatment with icosabutate in an established biopsy-confirmed AMLN ob/ob mouse model of NASH 20 and compared its activity to a farnesoid X receptor (FXR) agonist, obeticholic acid (OCA), which has demonstrated benefits on liver histology in humans. 21 We have complemented these findings by assessing effects of icosabutate directly on hepatic stellate cells, the key fibrogenic cell in liver. 22 As rodents transport plasma cholesterol primarily in the HDL fraction and are resistant to atherogenesis, transgenic models with human-like lipoprotein metabolism were used. To this end, mice expressing the human ApoE3-Leiden (APOE*3Leiden) isoform and human ApoC1 cross-bred with human Cholesteryl Ester Transfer Protein (CETP) transgenic mice were utilised to study treatment effects on hyperlipidaemia and atherosclerosis. The APOE*3Leiden.
CETP mouse is a well-established model with a human-like response to all lipid-modulating interventions that are being used in the clinic. 23 Inspectorate, Denmark). The animal protocol was designed to minimise pain or discomfort to the animals. Male B6.V-Lepob/JRj (ob/ob) mice, 5-week-old at the arrival, were obtained from Janvier Labs (Le Genest Saint Isle, France) and housed in a controlled environment (12 hour light/dark cycle, light on at 3 am, 21 ± 2°C, humidity 50 ± 10%). Each animal was identified by an implantable Funding information This work was supported in part by the TNO research program 'Preventive Health Technologies'. Laboratory studies using LX-2 cells were supported by a research contract between Scott Friedman's laboratory and Northsea Therapeutics. Hepatic lipidomic studies were supported by a research contract between OWL Metabolomics and Northsea Therapeutics. Ob/ob-NASH mice studies were supported by a research contract between Gubra and Northsea Therapeutics.
Handling Editor: Stefano Romeo via increased hepatic uptake, upregulated hepatic lipid metabolism and downregulated inflammation pathways, and effectively decreased atherosclerosis development.
Conclusions: Icosabutate, a structurally engineered EPA derivative, effectively attenuates both hepatic fibrosis and atherogenesis and offers an attractive therapeutic approach to both liver-and CV-related morbidity and mortality in NASH patients.

K E Y W O R D S
apoptosis, arachidonic acid, atherosclerosis, lipotoxicity, NASH, oxidised phospholipids

Lay summary
Liver scarring associated with obesity and diabetes rarely exists in isolation, and is typically part of a spectrum of disorders, including heart disease. Icosabutate is a novel treatment that, in mouse models, reduces both scarring of the liver and clogging of arteries. It is thus a promising therapy for subjects with both liver and heart disease.

| APOE*3Leiden.CETP transgenic mouse model
All APOE*3Leiden.CETP mice were housed and bred at the animal fa-

| Both icosabutate and OCA reduce hepatic steatosis, but only icosabutate reduces plasma ALT and hepatic macrophage numbers in AMLN ob/ob mice
In the AMLN ob/ob model neither icosabutate nor OCA affected bodyweight ( Figure 1A). OCA significantly reduced liver weight by 19% (P < .05) at 8 weeks whereas icosabutate had no significant effect ( Figure 1B). A dose-dependent decrease in steatosis ( Figure 1C) in response to icosabutate treatment was seen, with the 135 mpk dose achieving a 47% reduction while OCA reduced liver fat by 38% (both P < .001). Icosabutate also elicited a dosedependent effect on hepatic TAG lowering ( Figure 1D), with reductions of 17, 35 and 40% (all P < .001), while OCA achieved a reduction of 16% (P < .05).
With respect to liver injury and inflammation, only icosabutate (90 and 135 mpk) reduced plasma ALT (−32 and −44% respectively, P < .001) ( Figure 1E). To further assess the effects of either treatment on hepatic inflammatory macrophage infiltration, quantitative immunohistochemistry was performed to assess hepatic galectin-3 (Gal-3) content. All doses of icosabutate achieved a significant reduction in hepatic Gal-3 ( Figure 1F). Representative histological photomicrographs of liver cross sections stained with H&E and galectin-3 are shown in Figure 1G

| Icosabutate prevents progression of hepatic fibrosis in AMLN ob/ob mice
To assess the effects of treatment with either icosabutate or OCA on fibrosis, hepatic concentrations of hydroxyproline (HYP) were measured via a biochemical assay in addition to type 1 collagen α1 (col1A1) protein content measured via quantitative immunohistochemistry. Icosabutate reduced hepatic col1A1 content expressed as % area at the 90 mpk dose by 27% (P < .01, Figure 2A) and as total col1A1 at both the 90 and 135 mpk doses (−32%, P < .01 and F I G U R E 1 Both icosabutate and OCA reduce liver fat, but only icosabutate reduces liver enzymes and liver inflammation in AMLN ob/ob mice. Effects of treatment on terminal bodyweight (A), liver weight (B) and liver lipids as measured by % steatosis (C) or hepatic triacylglycerol content (D), plasma ALT (E) and hepatic galectin-3 (F). Values represent mean ± SEM for 12 mice per group. (G) Representative histological photomicrographs of liver cross sections stained with H&E or anti-Galectin 3, magnification 20x. *P < .05, **P < .01, ***P < .001 vs vehicle col1A1 and HYP data demonstrate that icosabutate effectively inhibits the progression of fibrosis.
The decreases in hepatic fibrosis in response to icosabutate treatment occurred in conjunction with significant decreases in mRNA transcripts for multiple genes regulating stellate cell activation, fibrogenesis and fibrolysis in AMLN ob/ob mice after 4 weeks treatment with icosabutate ( Figure S1, section A).

| Icosabutate reduces hepatic myofibroblast content in AMLN ob/ob mice in vivo and proliferative responses of human stellate (LX-2) cells in vitro
To gain further insight into underlying drivers of the decrease in fibrosis with icosabutate treatment, α-SMA content was measured as a marker of myofibroblast content. Icosabutate 90 and 135 mpk reduced α-SMA content expressed both as % area and total (all P < .01, Figure 3A and B respectively) whereas OCA and icosabutate 45 mpk had no significant effect. Icosabutate 135 mpk also led to a significant reduction in post-vs prebiopsy α-SMA content ( Figure 3C).
Representative histological photomicrographs of liver cross sections stained with α-SMA pre-or post-treatment are shown in Figure 3D.
Overall, α-SMA data demonstrate that icosabutate prevents the development of fibrosis in association with a decline in myofibroblast numbers.
In association with the finding of decreased myofibroblast content in response to treatment with icosabutate, we performed addi-
Expression of genes regulating hepatic HETE formation demonstrated a significant decrease in ALOX5AP (regulating 5-HETE and LTB4 generation) mRNA transcripts after treatment with icosabutate.
Transcripts for the lipoxygenase genes, ALOX5 and ALOX15, were reduced but constitutive expression levels were low ( Figure S1,

| Icosabutate reduces hepatic oxidative stress and oxidised phospholipids in AMLN ob/ ob mice
As highly unsaturated fatty acids are susceptible to peroxidation and can increase hepatic oxidative stress, 30 we also measured hepatic oxidised (GSSG) glutathione and the GSSG/reduced glutathione (GSH) ratio as a marker of oxidative stress and cellular redox status. 31 Icosabutate (90 and 135 mpk) significantly decreased hepatic GSSG concentrations ( Figure 4G), which in turn was largely responsible for the markedly increased GSH/GSSG ratio (84% increase, P < .01) seen with the highest icosabutate dose ( Figure 4H). In conjunction with the increased GSH/GSSH ratio, significant decreases in the hepatic expression of genes regulating enzymatic antioxidants catalase (CAT), and superoxide dismutase 1 and 2 were also observed ( Figure   S1, section C).
With respect to the functional consequence of increased hepatic oxidative stress, it has recently been demonstrated that hepatic oxidised phospholipids (oxPL) are a causal driver of NASH. 32 We therefore also measured a mixture of the dominant oxPLs (PC 16:0/AA-OH and LPC 16:0/LA-OH) and found that icosabutate significantly reduced their concentration by 31% (P < .05) and 46% (P < .001) for the 90 and 135 mpk doses respectively, whereas no effects were seen in response to OCA treatment ( Figure 4I).

| Icosabutate primarily targets highly unsaturated DAG and TAG species in AMLN ob/ ob mice
As specific DAG and ceramide species are associated with steatohepatitis 30 and insulin resistance 31 in addition to serving as the with a deleterious effect on glycemic control. 31 Overall the data suggest major differences between icosabutate and OCA with respect to both quantitative and qualitative changes in hepatic lipotoxic lipid species.  Figure 6B. Overall the data demonstrate that icosabutate effectively decreases apoptosis in AMLN ob/ob mice.

| Icosabutate decreases plasma TAG and total cholesterol but does not affect HDL cholesterol in APOE*3Leiden.CETP mice
The effects of 4 weeks of treatment with either icosabutate or fenofibrate on the main circulating plasma lipids were assessed in APOE*3Leiden.CETP mice. Both icosabutate and fenofibrate significantly decreased plasma TAG ( Figure 7A) to a similar extent (−70% for both compounds, P < .001), whereas icosabutate had a more pronounced effect on total cholesterol (−68% and −47% respectively, both P < .001, Figure 7B). Fenofibrate increased HDL cholesterol by 62% (P < .05) whereas icosabutate was without effect ( Figure 7C).

| Icosabutate increases VLDL plasma clearance and hepatic uptake of cholesterol and TAG in APOE*3Leiden.CETP mice
To investigate the mechanism/s by which icosabutate lowers plasma lipids, we evaluated very low-density lipoprotein (VLDL) production and clearance as determinants of plasma VLDL-TAG levels as compared to fenofibrate. 28 In contrast to icosabutate, fenofibrate increased VLDL-TAG production compared to controls (+25%) ( Figure 7D). Since VLDL-associated ApoB production and VLDL composition were similar in both groups (data not shown) these data indicate production of larger TAG-rich VLDL by fenofibrate as reported previously. 28 Compared to controls, the half-life of glycerol tri cholesteryl oleate by the liver was increased by both treatments compared to controls ( Figure S2B). Post-heparin activity of hepatic lipase ( Figure 7G) and lipoprotein lipase ( Figure 7H) did not significantly differ between control-and icosabutate-treated animals, whereas fenofibrate increased lipoprotein lipase activity as reported previously. 28 Combined, these data show that the lipid-lowering effect of icosabutate is mediated through increased hepatic VLDL remnant clearance and may be independent from lipase-mediated hydrolysis of VLDL-TAG.
LDL-R-mediated hepatic uptake of ApoB-containing lipoproteins is the most important pathway for removal of these atherogenic particles from the circulation. Both icosabutate and fenofibrate increased hepatic LDL-R protein expression compared to controls ( Figure 7I). These findings indicate that icosabutate increases LDL-R-mediated cholesterol uptake in the liver.

| Effect of icosabutate on liver lipid content and faecal bile acid excretion in APOE*3Leiden.CETP mice
To assess if reductions in plasma lipids were associated with hepatic cholesterol/TAG accumulation, decreased cholesterol absorption or increased excretion, hepatic lipid concentrations of cholesterol and F I G U R E 5 Icosabutate preferentially targets highly unsaturated hepatic TAG, DAG and reduces ceramides in AMLN ob/ob mice. Influence of the number of carbons and double bond content in the decrement of TAG (A). Change in specific DAG (B) and ceramide (C) species. Colour code represents the transformed ratio between means of the groups: green sections denote metabolites that were reduced (negative log 2 fold-changes) and red sections denote increased metabolites (positive log 2 fold-changes). For TAG the y axis denotes the number of carbons and the x axis the number of double bonds. Data are presented as mean ± SEM, *P < .05, **P < .01, ***P < .001 vs vehicle TAG along with faecal excretion of bile/neutral sterols were measured. Both treatments decreased hepatic cholesteryl ester content ( Figure 7J). Compared to controls, faecal excretion of bile acids was reduced by icosabutate and fenofibrate, whereas neutral sterol contents were unaffected ( Figure 7K). These findings negate the possibility that liver accumulation accounts for the decrease in plasma lipids.

| Icosabutate modulates hepatic expression of genes regulating lipid metabolism and inflammation in APOE*3Leiden.CETP mice
A partial overlap between icosabutate and fenofibrate was observed for differentially expressed genes (DEG) (Figure S3A), which could be expected given that fatty acids and their metabolites act as endogenous ligands for PPAR-α. 9 In line with the physiological data, upregulated DEG in the icosabutate group were enriched for pathways involved in lipid metabolism, including fatty acid metabolism and β-oxidation, and (chole)sterol biosynthetic processes ( Figure S3B). Downregulated pathways include complement activation, the AA cascade, innate immune response and acute phase response-related pathways ( Figure S3C). Using computational pathway analysis, Z factor scores were calculated to identify the most dominant transcription factors involved in transcriptional control following icosabutate treatment. Transcription factors that are predicted to be involved in icosabutate-mediated pathway activation are PPAR-α and -γ, and sterol regulatory elementbinding transcription factor (SREBF)-1 and −2 and PPAR gamma coactivator (PGC)-1α ( Figure S3D). tive responses in response to incubation with PPAR-α agonists, the mechanism is likely PPAR-α independent. 34 Decreases in specific hepatic ceramides also provide a mechanistic link not only to NASH but also to atherogenesis and insulin resistance. 35 In particular, the 30% decrease in Cer as DAG is potentially a more important source of AA than membrane PC for eicosanoid synthesis in the liver. 32 The reason for the decrease in hepatic AA stores in response to icosabutate are beyond the scope of the current work but inhibition of AA synthesis in vitro by both EPA and α-substituted EPA has previously been reported. 41 With respect to the improvements in hepatic oxidative stress, incorporation of highly unsaturated omega-3 fatty acids into hepatocyte membranes could potentially increase susceptibility of hepatocytes to lipid peroxidation. 42 However, icosabutate was able to markedly decrease both hepatic GSSG and hepatic oxPLs in AMLN ob/ob mice, an effect that likely benefits from the fact that α-substituted EPA avoids incorporation into phospholipids. 41 The decrease in PC-AA in response to icosabutate may also contribute to the Interestingly, in AMLN diet-induced NASH in Ldlr −/− mice, it has recently been shown that neutralisation of oxPLs substantially reduces hepatic apoptosis, 43 a crucial driver of liver injury in NASH. 44 This corresponds well with our findings in AMLN ob/ob mice, with a decrease in oxPLs occurring in conjunction with a significant reduction in apoptosis in response to icosabutate treatment. The overall changes induced by icosabutate also correspond with the substrate-overload liver injury model of NASH pathogenesis, where lipotoxic lipid species sequentially trigger oxidative stress, apoptosis and stellate cell activation. 1 Furthermore, although OCA was effective in reducing steatosis, there was no effect on oxidative stress, oxPLs or apoptosis, suggesting that a quantitative reduction in liver fat alone is insufficient for reducing fibrosis in the AMLN ob/ob mouse model.

F I G U R E 6 Icosabutate reduces hepatic apoptosis in AMLN
In relation to the effects upon atherogenic plasma lipids and hepatic lipid metabolism, the comprehensive data obtained in the APOE*3Leiden.CETP mice studies clearly demonstrate that icosabutate improves lipid and lipoprotein kinetics in conjunction with transcriptional activation of hepatic genes involved in fatty acid metabolism and handling. When combined with the biochemical measurement of plasma and liver TAG contents, the data demonstrate that TAGs are taken up by the liver and metabolically processed, resulting in a continuous removal of plasma TAGs in icosabutate-treated animals. We also observed increased transcription of genes involved in cholesterol synthesis, likely regulated by increased SREBF2-dependent pathway activation 45 in response to a decrease in hepatic cholesterol in icosabutate-treated animals. Using pathway analysis, a role for SREBF2 activation was predicted in icosabutate-treated animals, but not by treatment with fenofibrate. This suggests that SREBF2-mediated effects on LDL-R expression and cholesterol synthesis are among the PPARα-independent effects induced by icosabutate. Overall, the data suggest that the decreases in atherogenic lipids and lipoproteins previously described in the clinic are secondary to increased hepatic clearance with minimal contributions from a decrease in hepatic output and/or increased peripheral TAG lipolysis. 16 Importantly, the effects observed in the APOE*3Leiden.CETP mice confirm that icosabutate significantly inhibits the development of atherosclerosis in association with reductions in atherogenic plasma lipids.
A limitation of the current work in relation to clinical translatability is that OCA displayed minimal effects on fibrosis in the current AMLN ob/ob mouse model of NASH, yet has demonstrated beneficial efficacy in humans. 21 The reason for the lack of effect seen with OCA treatment in the current model is uncertain but as improvements in fibrosis have been observed after 16 weeks treatment with OCA in the same model, 46 the 8 week treatment period employed in the current study may have been too short.
Another potential limitation is that although the data demonstrate reductions in lipid species that have established negative effects on NASH and CVD, potential positive contributions of icosabutate metabolites are not accounted for. Specific omega-3 metabolites, referred to as specialised pro-resolving mediators (SPMs), have anti-fibrotic effects and have even been shown to reverse established fibrosis (see review by Musso et al 47 ). Although the main oxygenated metabolites of icosabutate have been identified in vivo, their comparative function vs oxygenated EPA metabolites are still being established.
In summary, the data demonstrate that icosabutate, a liver-targeted, semi-synthetic EPA derivative, effectively targets both atherogenesis and hepatic fibrosis in association with increased clearance of atherogenic plasma lipids and decreased hepatic stellate cell activation respectively. Icosabutate may thus provide a promising and novel therapeutic approach to the dual treatment of liver-and CV-related morbidity and mortality in NASH patients.

ACK N OWLED G EM ENTS
We thank Simone van der Drift-Droog, Anita van Nieuwkoop-van