Differential hepatotoxicity of dietary and DNL-derived palmitate in the methionine-choline-deficient model of steatohepatitis

Mice were fed custom MCS and MCD diets that differed from commercial MCS and MCD formulas by being nearly completely enriched with a single type of carbohydrate (sucrose or starch) or fat (palmitate or oleate). The custom MCS and MCD mixtures were designed to maximize palmitate accumulation in the liver via DNL (with sucrose) or diet (with palmitate) or both. Starch served as the control to sucrose, whereas oleate served as the control to palmitate. Mice in all 8 dietary groups ate comparable amounts of food during the study period. Animals fed MCS formulas gained weight (14.7 ± 1.3 %), whereas those fed MCD formulas lost weight (28.8 ± 1.0 %), which is characteristic for the dietary model [8]. All MCD-fed mice lost comparable amounts of weight regardless of the macronutrient composition of the diet (Fig. 1b). MCD feeding is unique in that it does not induce insulin resistance or hyperglycemia coincident with steatohepatitis [16]. This pattern did not change with the 4 custom MCD diets; there was no evidence of insulin resistance or hyperglycemia in any MCD group (not shown).

After 3 weeks on the custom diets, MCS-fed mice remained free of histologic hepatic steatosis. By contrast, MCD-fed mice developed markedly different degrees of hepatic steatosis depending on macronutrient composition. This was evident histologically (Fig. 2a) and confirmed by hepatic lipid quantitation [6, 9] (Fig. 2b and c). MCD formulas containing sucrose induced the most pronounced hepatic steatosis regardless of the accompanying type of dietary fat. The worst steatosis occurred in mice fed MCD diets containing both sucrose and palmitate.

Mice fed MCD formulas containing sucrose also exhibited the greatest degrees of liver injury, as shown by TUNEL staining and serum ALT (Fig. 3). Just as it induced the most steatosis, the MCD formula containing both sucrose and palmitate caused the worst hepatotoxicity. Accompanying the liver injury in MCD-fed mice was hepatic activation of Jun-N-terminal kinase (JNK); the greatest degree of JNK activation occurred in the sucrose-palmitate group. In addition to JNK, the necroptosis marker receptor-interacting protein kinase 3 (RIP3) was mildly upregulated in response to MCD feeding. RIP3 was most visible in mice fed sucrose-palmitate. LC3, a marker of autophagosomes, was up-regulated in mice fed MCD sucrose-palmitate, but also in mice fed MCD starch-palmitate. This suggests dietary fat is affecting hepatic autophagy either positively or negatively, but without a firm relationship to liver injury. Overall the data support the notion that dietary sucrose activates cytotoxicity pathways known to be operative in steatohepatitis (JNK, RIP3) [17, 18], and the addition of dietary palmitate accentuates these events.

Hepatocellular injury in MCD-fed mice was accompanied by the induction of pro-inflammatory genes in the liver and the hepatic influx of CD11b-positive leukocytes (Fig. 4). The degree of hepatic inflammation mirrored the degree of hepatocellular injury in all MCD-fed groups. Stellate cell activation, characterized by the induction of type I collagen mRNA in the liver, was also affected by diet; again, MCD sucrose-palmitate provided the greatest stimulus to collagen gene regulation. Despite robust collagen gene induction in the livers of MCD-fed mice, there was no increase in smooth muscle-alpha-actin expression (Fig. 3c). Nor was there any evidence of collagen deposition in the liver by morphometry (<0.5 % Sirius Red-stained area in all groups). This suggests that collagen gene induction reflects acute liver injury rather than fibrosis at the 3-weeks time point, but portends fibrosis over a longer interval.

After characterizing the effects of the 4 custom MCD formulas on liver injury, we explored whether the different outcomes of the 4 diets could be attributed to differences in hepatic palmitate accumulation. Hepatic palmitate levels rose above control values in MCD-fed mice whose diets contained sucrose or palmitate or both (Fig. 5a). Although palmitate levels tended to correlate positively with ALT levels in these MCD groups, the relationship was not strong (Fig. 5b). Indeed, as shown in Fig. 5c, mice fed MCD sucrose-oleate accumulated no more palmitate than those fed MCD starch-palmitate, yet their ALT levels were significantly higher. This suggests that palmitate arising from sucrose in the diet (DNL palmitate) is more hepatotoxic than palmitate coming directly from the diet in the MCD model of liver disease. We assessed lipogenic gene expression in all mice, although previous studies have shown gene expression does not correlate with actual DNL in the MCD model of steatohepatitis [6, 8]. MCD-fed mice displayed marked suppression of mRNA encoding acetyl-CoA carboxylase (ACC) and fatty acid synthase (FAS) compared to MCS controls, as has been reported previously [8] (Fig. 5d). There were no differences in lipogenic gene expression among the 3 custom MCD groups that accumulated hepatic palmitate, but the predictive value of this observation is low.

Noteworthy was that mice fed the MCD formula containing both sucrose and palmitate accumulated more hepatic palmitate and had higher ALT levels than would have been predicted by a mere additive effect of both nutrients (Fig. 5e). Dietary saturated fat is known to stimulate hepatic DNL [19], and thus the excess palmitate in the livers of mice fed MCD sucrose-palmitate likely derives from exaggerated DNL. The extremely high ALT levels in these mice underscores that palmitate arising from DNL is particularly noxious to the liver.

Overall, the current experiments confirm our previous observation that dietary sucrose, through DNL conversion to palmitate in the liver, is an important inducer of liver injury when downstream pathways for fatty acid desaturation and lipid excretion are blocked [6]. More importantly, they extend previous work by demonstrating that dietary palmitate does not induce the same level of hepatotoxicity as DNL palmitate despite accruing to twice the concentration found in control livers (Fig. 5c). This suggests that dietary palmitate is handled differently by the liver than DNL palmitate. Different metabolic fates for DNL vs. exogenous palmitate have been reported in cultured adipocytes and HepG2 cells [20, 21]. The noted differences, however, were in palmitate desaturation and elongation, which in MCD livers would not likely affect lipotoxicity. We searched individual hepatic compartments in MCD-fed mice to determine whether palmitate accumulates preferentially in the more metabolic depots such as free fatty acids or diacylglycerols, but found excess palmitate only in hepatic triglycerides (Fig. 5f). It is possible that liver injury is a function of the lability of the hepatic triglyceride pool in these mice; we could not determine this in the current experiments.

The fact that MCD starch-palmitate mice were relatively free of liver injury, whereas MCD sucrose-palmitate mice had exaggerated liver injury, supports the concept that dietary saturated fat by itself is nearly innocuous to the liver but becomes toxic only in combination with dietary sugar. This is an intriguing theory, but unfortunately, saturated fat consumption is unlikely to be uncoupled from sugar consumption in free-living humans. Our other major finding, that sucrose and palmitate together induce synergistic hepatotoxicity in mice, is more translationally relevant. Indeed, dietary saturated fat has recently been shown to enhance hepatic steatosis, if not steatohepatitis, in humans when added to a mixed-nutrient diet [22]. Given our current experimental results, it will be important to determine whether synergy between sucrose and palmitate is an important inducer of liver injury when taken out of the context of the MCD model. Such experiments are currently underway.