The Effect of Eicosapentaenoic Acid on Hepatic Metabolism and Intracellular Lipid Droplet Dynamics in An in Vitro Model of Hepatic Steatosis

Katie Morgan The University of Edinburgh Samantha Lee Suppiah The University of Edinburgh Gail Masterton The University of Edinburgh Kay Samuel Scottish National Blood Transfusion Service Shonna Johnston The University of Edinburgh Peter C Hayes The University of Edinburgh Khalida A Lockman The University of Edinburgh Vasileios Koutsos The University of Edinburgh John N Plevris (  J.plevris@ed.ac.uk ) The University of Edinburgh


Introduction
The global epidemic of obesity has triggered an inexorable rise in the prevalence of non-alcoholic fatty liver disease (NAFLD). As a result NAFLD is now the third commonest reason for receiving a liver transplant in the United States [1]. Intrahepatic triglyceride accumulation is the hallmark of NAFLD. Stored within the lipid droplets in the hepatocyte cytoplasm, triglyceride is an e cient fatty acid storage molecule, not only crucial for energy homeostasis but also vital for protecting cells against the deleterious effects of free fatty acids. Lipid droplets are highly dynamic, which in response to the appropriate signals, can be esteri ed, hydrolyzed or reesteri ed to meet cellular demand [2]. Altered liver metabolism with an accumulation of lipid droplets is an early pathophysiological marker of insulin resistance and steatosis leading to cirrhosis and possibly hepatocellular carcinoma [3,4,5]. The size of lipid droplets is maintained by a tightly regulated mechanism and they are known to take part in various biological processes such as neutral lipid storage and the storage and degradation of proteins [4,5].
Despite the abundance of lipid droplets, surprisingly little is known about their role in the pathogenesis of NAFLD.
Omega-3 polyunsaturated fatty acids, of which eicosapentaenoic acid (EPA) is among the most physiologically active and has been suggested as a potential treatment for NAFLD [6,7]. Cell and animal studies have suggested that omega-3 fatty acids may act as regulators of hepatic gene expression [8,9].
The mechanisms whereby omega-3 fatty acids affect gene expression are complex but may involve interplay between the activation of peroxisome proliferator-activated receptor (PPAR) a, and the suppression of sterol regulatory element binding protein (SREBP) [10,11]. Such in uence on lipid homeostasis can potentially alter the characteristics of lipid droplets. For instance, PPAR activation has been shown to modulate lipid droplet proteins-perilipin, adipophilin, and TIP47 (PAT proteins) [12].
However, the effect of omega-3 fatty acids on physical characteristics and biological properties of lipid droplets in NAFLD is yet to be established.
We have previously shown that C3A cells treated with energy substrates; lactate (L), pyruvate (P), octanoate (O) and ammonia (N) in vitro recapitulate the events that have been proposed to occur in human NAFLD [15]. While oleate mimics simple steatosis in vitro, octanoate has been shown to mimic the pathophysiology of NAFLD through increasing β oxidation and upregulating genes causing brosis and synthesis of cholesterol [15,16] Indeed, LPON-induced cellular steatosis manifests many of the key features associated with steatohepatitis such as impaired mitochondrial function, enhanced oxidative stress with production of reactive oxygen species (ROS), increased ketogenesis and altered glucose metabolism [15,16]. However, when cells are pre-treated with oleate, we have shown no mitochondrial dysfunction despite triglyceride accumulation [15].
The main mechanism that underpins the triglyceride accumulation is different between these models.
The medium chain free fatty acid, in LPON, octanoate, cannot be directly esteri ed thus the triglyceride accumulation with LPON relies on de novo lipogenesis [13]. By comparison, triglyceride accumulation with oleate is the result of direct esteri cation in proportion to its concentration [13,14]. Whether such a contrast can in uence the characteristics of lipid droplets remains unknown.
In this study we also explore the effects of both oleate and LPON with and without EPA on human hepatic C3A cells and evaluate several markers on liver function alongside the dynamics of lipid droplets for each model. Liver function was assessed by measuring lactate dehydrogenase (LDH), albumin and aspartate transaminase (AST). Further assessment of the effect of EPA on glucose, ketone bodies and reactive oxygen species donor was performed to assess how EPA may effect basic functionality of the cell and be of importance in the study of type 2 diabetes which is often associated with NAFLD and metabolic disease. Finally, we de ned the dynamics of lipid droplets within our models to see if this could be a useful marker in the context of histopathological characterization of NAFLD examining triglyceride concentrations, lipid droplet spherocity and volume to size ratio which has been implicated in lipid homeostasis and insulin sensitivity [17].
Optimal concentration of EPA in LPON model using LDH measurement C3A cells were grown in 6 well plates and incubated with LPON plus either 10, 50, 100, 250 and 500µM EPA. extracellular LDH concentration was used to assess cell viability. This assay is based on the procedure of Gay et al., [18]Here the LDH concentration in the cell lysates and supernatant were determined by following the rate at which NAD is reduced to NADH measured as an increase in absorbance at 340nm in the presence of lactate using a LDH kit method (Sentinel Diagnostics, Italy) modi ed for use on the Cobas-Fara centrifugal analyser (Roche Diagnostics, Welwyn Garden City, UK). The rate of decrease in absorbance at 340nm, measured at 37C, is directly proportional to LDH activity in the sample. Results were expressed as % LDH released calculated as follows -intracellular LDH/ (extracellular LDH + Intracellular LDH) x 100.The data is presented as % of LDH in the supernatant/ total LDH. The results are also presented in Figure 3.1 as means and standard error of the mean (SEM).

Quantitative analysis of glucose
Principle: (Hexokinase (HK)) Glucose + ATP ↔ Glucose -6 -phosphate (G6PdH) Glucose-6-phosphate + NAD ↔Gluconolactone 6-phosphate + NADH The assay detects the concentration of NADH produced by the two reactions above. This corresponds to the concentration of glucose in the cuvette. The glucose buffer is prepared by mixing NAD (17mg), ATP (125mL), G6PdH (12.5mL) HK (5mL) and diluting in 50mls of PBS+.200mL of each sample was pipetted into a cuvette and 1ml of glucose buffer added. This was then homogenised by gently inverting the cuvette 3-5 times. A blank well was prepared with 200mL of 95 H2O and 1ml of buffer to act as a control. The samples were then left at room temperature for 60 minutes. Following this absorbance was read at 340nm wavelength for each cuvette and blank on Unicam UV1 spectrophotometer (Unicam Ltd, U.K). Glucose ux was then calculated according to the methods of Bergmeyer [19].

Albumin assay
Albumin working solution was made up from powdered albumin (Albumin blue 580. This was made into solution with isopropanolol (3mg/100mls isopropanolol) to give concentration 30mg/L. The absorbance of the solution was read at 580nm and the solution was diluted such that the OD was 1.00. The solution was diluted with a buffer comprising 0.6g N-morpholino-propanesulfonic acid (Mops free acid), 1.8g Mops sodium salt, 2.4g sodium chloride, 0.2g ethylene-diaminetetraacetic acid and disodium salt (EDTA disodium), 200 mL distilled water, and 20 mL isopropanol. The pH of the resulting solution is 7.4. This solution was then diluted with buffer to create standards with albumin concentration 2.5, 5.0, 10, 20, 40, 50, 75, 100, 150, 200 mg/mL. 80mL of each standard or sample was added in duplicate to wells in a microtitre place. 160mL of dye was added to each well. The plate was then shaken for 30seconds before the uorescence was read (excitation 590nm, emission 645nm) on Cyto uor Series 4000 (PerSeptiveBiosystems). A standard curve was created by inputting the data to Microsoft excel. Test values were then calculated from the standard curve.Concentration of BSA complexed with oleate was subtracted from results.
AST assay AST was determined by a commercial kit (Randox Laboratories, UK) adapted for use on the Cobas-Fara centrifugal analyser (Roche Diagnostic Ltd, Welwyn Garden City, UK). α-oxogluterate reacts with Laspartate in the presence of AST to form Lglutamate plus oxaloacetate. The indicator reaction utilises the oxaloacetate for a kinetic determination of NADH consumption. Within run precision was CV<5%.

Ketone body production
Betahydroxybutyrate + NAD ↔ acetoacetate + NADH The assay detected the concentration of NADH produced by the reaction which corresponds to the concentration of betahydroxybutyrate in the sample. Two readings were taken as the changes in absorbance were small. Method A betahydroxybutyrate buffer was made by mixing NAD (50mg), glycin (3g), hydrate hydrazine (2ml). This was diluted in 100mls sterile water. For the reaction 200mL of sample was added to each cuvette. 200mL of sterile water was added to an additional cuvette to act as control. 1ml of betahydroxybutyrate buffer was then added to each sample and this was homogenised by gentle inversion. A reading was then made at 340nm wavelength on Unicam UV1 spectrophotometer (Unicam Ltd, U.K). A solution of betahydroxybutyrate dehydrogenase was then prepared by diluting 200mL of betahydroxybutyrate dehydrogenase in 1ml of sterile water. 10mL 96 of this solution was added to each cuvette and the sample homogenised. Samples were left for 1 hour at room temperature before a repeat reading was made.

Reactive oxygen species donor
Cells were grown in 6 well plates in standard MEME until con uent. Media was then replaced with either standard MEME or MEME + oleate, with or without reactive oxygen species donor (100µM tBOOH) and incubated for 72 hours at 37C. Supernatant and cells were harvested and triglyceride levels were determined as per previous methods.

BODIPY 493/503 staining
Cells were treated on chamber slides for 72 hours and xed with 4% paraformaldehyde for 30 minutes before staining with 200ml of BODIPY 493/503 (Invitrogen, NY, USA) for 10 minutes. Cells were mounted in ProLong gold antifade reagent and were left for 24 hours at 4ºC before imaging with a confocal laser microscope (Leica SP5, Mannheim, Germany). Fluorescent intensity in at least 50 cells per image was analysed using ImageJ software (National Institutes of Health, Bethesda, MD). To determine the characteristics of lipid droplets, the 3D confocal image stacks were analysed using Volocity 3D image analysis software (Perkin Elmer, Waltham, MA). Using the z stacks of BODIPY stained lipid droplets, objects smaller than 3μm3 were identi ed and excluded and to determine individual nuclei within the DAPI image stacks the limit was set at 50μm3.

Triglyceride concentration
To determine the triglyceride concentrations, cells were treated with the speci ed combination of energy substrates in 6-well plates (in triplicates) for 3 and 7 days. Intracellular triglyceride concentrations were measured using a commercial kit adapted for use on the Cobas-Fara centrifugal analyzer (Roche Diagnostics Ltd., Welwyn Garden City, UK). Triglyceride concentration was normalized to the total protein and expressed as mg per gram of total protein.

Visual Basic
Visual Basic for Applications was used in MS Excel to analyse lipid droplets from Bodipy imaging. Data was taken from Volocity imaging software and processed uniformly throughout. The dataset was ltered and sorted by volume in cubic microns. Larger objects identi ed as multiple droplets or nuclei were excluded. For images containing lipid droplets, objects with measured volumes of 400μm3 and higher were removed. Total, average, minimum and maximum object volume (μm3) and surface area (μm2) was assessed and average surface area to volume ratio was calculated for the data set. Objects were measured and grouped according to their volumetric size.

Statistical Analysis
All data were analysed on GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA)using one way ANOVA followed by Tukey post-test. P<0.05 was deemed statistically signi cant.

Dose dependent response of glucose and ketone bodies to EPA in LPON model
Endogenous glucose production was selected as the primary outcome to assess e cacy as this can be rapidly assessed by spectrophotometry. We measured glucosein micro mole of glucose per hour per gram of total protein (µmol.h-1.gTP-1). The endogenous glucose production of LPON control is signi cantly higher than untreated cells (p=<0.0001).Likewise, we see a concentration dependent trend showing a decrease in ketone bodies with increasing EPA.

Liver function tests with addition of EPA
There was no signi cant change in AST following incubation with EPA under any condition though the MEME control showed a trend between decreasing AST and increase of EPA (p=0.06) which just missed signi cance.Incubation with EPA signi cantly increased albumin in MEME control vs MEME + 250µM EPA (p=<0.05). Some slight signi cance was seen in oleate, but likely do to outlier seen in oleate control.
No change was seen between other samples.At 500µM EPA had signi cantly elevated LDH leakage when compared with LPON control (p-<0.001) and was also elevated compared to other concentrations of EPA.

Ketone body production across all conditions
No signi cant difference in production of ketone bodies could be seen in control (MEME media only) + EPA or oleate + EPA, however, there was signi cant reduction of ketone bodies between LPON control and LPON + EPA 50µM (p<0.05) and LPON + EPA 250 µM (p<0.01)

Reactive oxygen species donor
Our LPON model is associated with increased ROS while control and oleate models are not. This experiment was designed to measure total triglyceride concentration in control and oleate models in the presence of a ROS donor to rule out ROS in the LPON model preventing the lowering of triglycerides.The presence of a reactive oxygen species donor did not prevent the reduction in hepatocyte triglycerides in the Control (MEME media only) model: both with and without a ROS donor there was a signi cant reduction in hepatocyte triglyceride content (p<0.0001)

Triglyceride concentrations
We have previously shown that treatment with either oleate or LPON for 24 and 72 hours resulted in signi cant triglyceride accumulation which resulted in two models of NAFLD [15]. Oleate shows a model whereby lipid accumulation is the result of direct esteri cation [15]. Whereas, LPON shows a model characterized by de novo lipogenesis [15]. Here, we examined the impact of EPA on triglyceride concentration in these models of cellular steatosis. Cells were treated with the speci ed combinations of energy substrates, with or without EPA for 72 hours. Triglyceride concentrations were measured at the end of the treatment period. As shown in Figure 5, there was a signi cant reduction in triglyceride content when untreated cells were incubated with 250μM EPA at both 3 (p<0.05) and 7 days (p<0.05). These equate to 21.9% (95%CI 9-35%) and 23.1% (95%CI 5-41%) reduction in triglycerides respectively. Incubation with 50μM had no effect on hepatocyte triglyceride content. A linear trend between increasing EPA concentration and reduced hepatocyte triglyceride content was con rmed on post test analysis in the standard media model for both day 3 (p=0.005) and day 7 (p=0.006). In contrast, EPA did not alter triglyceride concentrations in either oleate or LPON treated cells.

Lipid droplets number per nucleus
Next, we examined the effect of EPA on the quantity of lipid droplets per cell (quanti ed as ratio of lipid droplets to each nucleus) induced by the speci ed treatment. Cells were treated with LPON or oleate in chamber slides for 24 and 72-hours. The effect of adding EPA was assessed: here cells were treated for 72 hours. Using the Volocity 3D image analysis software, the number and characteristics of the lipid droplets captured by confocal microscopy were analyzed. There was no difference in the number of lipid droplets per cell in the varying treatment groups at either time point (Figure 6). The addition of EPA resulted in an increase in the number of lipid droplets per nucleus in the LPON group [p<0.05].

Surface Area to Volume Ratio and spherocity of lipid droplets
We then sought to determine the surface area to volume ratio (surface area (mm2)/volume (mm3)). The surface area to volume ratio of lipid droplets has been shown to in uence crucial processes involved in lipid homeostasis including triglyceride hydrolysis and insulin sensitivity. Table1 shows the effect of each treatment on the surface area to volume ratio. Oleate and LPON-treated cells had lower surface area to volume ratio when compared to the untreated cells after 24 hours suggesting that the initial response to lipid loading in these cells was predominantly volume expansion. After 72 hours, surface area to volume ratio of droplets induced by oleate increased. This is consistent with higher reduction in volume rather than surface area in these cells. Although a similar pattern was also seen with LPON, the ratio remained lower than that seen in oleate-treated cells. There was no difference in surface area to volume ratio on EPA treated cells.
The surface area to volume ratio in uences the sphericity of lipid droplets. Here, we calculated the sphericity of the lipid droplets as: p1/3(6(volume))2/3/ surface area. The sphericity of lipid droplets induced after a 24-hour treatment with oleate was lower than that seen in LPON-treated cells though it did not vary much from untreated (Table 1). After 72 hours, the rise in surface area to volume with oleate was paralleled by increased sphericity. Similarly, the reduction in the surface area to volume in oleate + EPAtreated cells was associated with diminished sphericity. In contrast, the impact of LPON treatment on sphericity was diametrical to its effect on surface area to volume ratio; the modest rise in surface area to volume ratio was mirrored by a reduction in sphericity. The sphericity of lipid droplets resulted from LPON treatment was lower than that seen in oleate and signi cantly lower than untreated cells with EPA(p<0.0001). Though the addition of EPA did not alter the sphericity of LPON treated cells overall.

Overall comparison of characteristics
Tables 1 and 2 are a summation of characteristincs of hepatic lipid droplets developed under different conditions. Relative lipid volume per cell in oleate decreased by day 3 but increased with addition of EPA. Within the LPON model, relative lipid volume per cell did not change with addition of EPA. This demonstrates that EPA has no signi cant effect within our in vitro model of steatosis with ROS (LPON).

Discussion
Effects of EPA on ketone bodies, glucose and liver function Our rst experiments were designed to establish the optimal concentration of eicosapentaenoic acid (EPA) for use in this model. Two markers were evaluated: LDH leakage was used to assess toxicity and membrane leakageleading to cell death, and glucose production to re ect e cacy. The results showed that EPA was effective at altering endogenous glucose production in the LPON model but that large, i.e. 500μM, doses of EPA resulted in a signi cant increase in cell deathand/or membrane leakage shown by high levels of LDH ( Figure 2). With these results 50μM and 250μM doses were chosen as they represented a balance of desirable and undesirable effects as well as allowing a dose-response relationship to be demonstrated.
Increase in albumin synthesis could be seen in control model supporting the suggestion that EPA has bene cial effects on the cell beyond a sole reduction of intracellular triglyceride content. There was a reduction of albumin in the oleate model compared to control, but albumin production remains higher than media only control and LPON control suggesting that overall albumin is higher in oleate loaded cells than control or LPON under all conditions. (Figure 2 Ketone bodies are a byproduct of the breakdown of fatty acids. They have been shown to lower ROS [20], stimulate insulin release and cause lipid peroxidation which may play a role in vascular disease in diabetes [21]. However, high levels of ketone bodies can lead to life threatening ketoacidosis. As such ketone bodies were measured in our model to determine if fat loading increased production of ketones. Within controls ketone bodies were measured at 3.8 umol.hr MEME only control, 6 umol.hr oleate and 21umol.hr LPON (Figure 3). This shows the LPON model to have a signi cantly higher level of ketone bodies at baseline than oleate or control. Addition of EPA did not change levels of ketone bodies in the standard model and oleate but signi cantly reduced ketones in LPON model in a dose dependent manner ( Figure 1). This alongside the reduction of glucose, also in a dose dependent manner, suggest EPA may have a qualitative bene t beyond mere reduction of triglycerides ( Figure 1) We have previously shown that by adding lactate, pyruvate and insulin to C3A cells that you get a dose dependent drop of insulin(Anns thesis). This is likely due to the increase of lactate and pyruvate fueling the TCA cycle and diverting pyruvate towards the gluconeogenicpathway.The presence of oxidative stress in this model augments this theory as mitochondrial dysfunction in the presence of oxidative stress can accelerate acetyl-co-A production. (Anns thesis)While more investigation is needed into the mechanisms behind this decrease of glucose, other studies have proven the effects of EPA on insulin resistance and glycemic control in in vivo models. [22,23,24] It was considered whether the lack of triglyceride reduction in the LPON group could be a result of the increased reactive oxygen species seen in this model. Cells in the control and oleate groups were therefore co-incubated with a non-lethal dose of a reactive oxygen species donor (tBOOH). EPA continued to have an effect in cells in the standard media containing ROS donor, but there was still no signi cant change in the oleate group when incubated with EPA and tBOOH. Co-incubation with a reactive oxygen species (ROS) donor did not mitigate the triglyceride lowering effect seen in untreated hepatocytes when incubated with EPA. The bene cial effect on the control in terms of triglyceride reduction with ROS suggests that the ineffectiveness of EPA in the LPON model to lower triglyceride content is not purely as a result of the increased ROS in this model. Since ROS does not affect the concentration of triglycerides in fat loaded models, the reduction seen in control must be through a different mechanism.
Lipid droplet characteristics Characterisation of lipid droplet formation, stability and breakdown is important in understanding how fragmentation of droplets contribute to free fatty acids and ROS released within the cell. Previous work highlighted regulation of proteins involved in lipid droplet stability within the LPON and oleate models [25]Notably, we show an upregulation of perilipin-2 (PLIN2), a gene related to stability of lipid droplets, is upregulated in our LPON model. PLIN2 is known to be involved reduction of lipolysis and decrease triglyceride turnover, which we see in Figure 5. PLIN2 is also involved in the regulation of lipid droplet stability [25]. This suggests a protective mechanism whereby stability of lipid droplet is prioritized to prevent the release of FFA and ROS which cause cytotoxicity [25].This led us to the current study which highlights two important points.
Firstly, LPON induced steatosis, as previously shown, is associated with oxidative stress which appears to have an effect on the characteristics of lipid droplets [15]. (Figure 8) This is evidenced by the differences in lipid droplet accumulation, surface area and volume. In the oleate model there was immediate lipid volume expansion while the number of lipid droplets remained the same. By day 3 lipid volume decreased with a slight increase in the number of lipid droplets. In contrast, addition of EPA to oleate treated cells resulted in a rise of total lipid volume. This suggests that addition of EPA increases lipid accumulation to a level higher than control (Table 1). This suggests that cells pre-treated with oleate have an initial increase in fat which is metabolized by day 3 resulting in lower total volume, though addition of EPA increases total lipid volume to a level higher than control. While in contrast, LPON treatment was associated with a steady rise in number of lipid droplets and lipid accumulation but had no change in volume at day 3, which was unaffected by EPA treatment. We postulate the difference in triglyceride acquisition between LPON and oleate may be responsible for the above. Octanoate (the chosen FFA in LPON) cannot be directly esteri ed thus the triglyceride accumulation is largely mediated by increased de novo lipogenesis. This may explain the time dependent increase in volume and quantity of lipid droplets with LPON.
As fatty acids can become incorporated in the cell and lipid droplet membrane, another potential mechanism to explain our results is that the different fatty acids may have resulted in different phospholipids and membrane proteins. Diminished total and lipid droplet volume with oleate after 72 hours may have been attributable to enhanced PAT protein expression (an essential surfactant positioned on the surface of lipids] which promotes the packaging of lipids in smaller units thus increasing the surface area to volume ratio [26]. Such an increase in the surface area to volume ratio allows access to lipases for triglyceride hydrolysis and in uences insulin sensitivity. Conversely, current hypothesis stipulates that a reduction in the surface area to volume ratio can lead to an incomplete triglyceride hydrolysis leading to DAG (diacylglycerol) formation [27]. DAG has been shown to impair insulin signaling via the PKC isoform ε which phosphorylates the insulin receptor effecting all downstream effects of insulin signaling, upregulation of de novo lipogenic genes and glycogen synthesis [27,28,29]. Oleate induced cellular steatosis has been shown previously to have preserved insulin sensitivity whereas LPON (with a higher surface area to volume ratio than oleate) was assciated with increased gluconeogenesis [15,30].
The second observation is that the effects of EPA on lipid droplets differ between oleate and LPON. In untreated cells, EPA signi cantly reduced triglyceride concentration and the volume of lipid droplets resulting in a rise in the surface area to volume ratio. This nding of reduced hepatocyte triglyceride with EPA is consistent with several previous studies [31,32,33]. However, similar effects have not been observed in either oleate or LPON groups. Contrary to the ndings in the untreated cells, EPA increased the volume of lipid droplets hence reducing the surface area to volume ratio in LPON and oleate treated cells. How might these results be accounted for? It is likely that the differences lie in the mechanisms of EPA in modulating fatty acid metabolism. EPA is thought to promote free fatty acid oxidation by activating the PPAR-a oxidative pathways [11,34,35]. The bene cial effect of EPA in adult NAFLD remains unproven. The effect of EPA on PPAR oxidative pathways encompasses the activation of both w-oxidation and peroxisomalb-oxidation. Unlike mitochondrial b-oxidation, these pathways can generate a signi cant amount of ROS [39,40]. In support of this, a metabolomic analysis has shown that an omega-3 fatty acid, DHA, increases lipid peroxidation resulting in increased isoprostanes formation mirrored by a decline in hepatic α-tocopherol and ascorbate [41]. Nevertheless, the by-product of n-3 lipid peroxidation such as 4-hydroxyhexenal from DHA but not EPA is thought to confer cardioprotective effect by enhancing antioxidative pathways mediated by NRF-2 [42]. Furthermore, DHA has been shown to suppress hepatic markers of in ammation without a reduction in hepatic steatosis [43]. It is therefore possible that any bene cial effects of omega-3 fatty acids are independent of their abilities to reduce hepatic steatosis.
The question remains whether EPA would confer bene t in vivo in the presence of mitochondrial dysfunction with high ROS burden. Mitochondrial beta-oxidation requires intact mitochondrial respiration.      This image is not available with this version.

Figure 8
Graphical representation of changes in EPA model Media only control shows lower triglycerides with no change to AST or ketone body production and an increase of albumin in the presence of EPA. The oleate model, representing simple steatosis, shows an increase in size of lipid droplets which decrease over time leading to higher lipid volume, but no increase in ROS. This model shows no change in the level of triglycerides even with an addition of a ROS donor. There was also no change to AST, albumin or levels of ketone bodies. The LPON, NAFLD, model shows a higher number of lipid droplets with increased spherocity and area compared to control. There was no change to LDH, AST or albumin, though a decrease in glucose and ketone bodies was seen in the presence of EPA.