Intrahepatic lipids accumulate when the rate of hepatic lipid input exceeds the rate of hepatic lipid output (Figure 2
). On the input side, lipids may be synthesized (de novo) from carbohydrates or other precursors, imported as lipolysis-derived FA, or imported as lipoprotein-derived triglyceride. Tracer experiments have demonstrated that all three input sources contribute significantly to intrahepatic lipid stores: ~59% of intrahepatic triglycerides in NAFLD patients derives from circulating FA, whereas ~26% derives from DNL and ~15% from dietary fat [18
]. However, when NAFLD patients were compared to matched controls without steatosis, there was no meaningful difference in the FA flux from adipose tissue to liver [19
]. In contrast, DNL was 3-fold greater in NAFLD compared to the control subjects [19
], suggesting that increased DNL is a distinct pathophysiological feature of human NAFLD.
On the output side, intrahepatic lipids may be utilized in β-oxidation, leading to the production of carbon dioxide or ketone bodies, or secreted as triglyceride in very low-density lipoprotein (VLDL) particles. Defects in these pathways have also been demonstrated, although interpretation of these data seems more complicated. Under fasting conditions, hepatic lipid oxidation rates are unchanged in NAFLD patients [48
]; however, high fructose consumption may promote a shift towards carbohydrate oxidation [39
]. In addition, NAFLD was associated with increased hepatic VLDL secretion in some [49
], but not in all trials [19
], whereas NASH was associated with decreased VLDL production [52
]. In the following sections, we will discuss if and how fructose consumption influences these mechanisms.
3.1. Fructose and Lipid Synthesis
As discussed above, acute hepatic fructose metabolism rapidly results in the availability of DNL substrate, and some authors have suggested that this may largely drive the increase in plasma triglyceride levels after acute fructose ingestion [29
]. This, however, has yet to be confirmed in human metabolic tracer studies [39
]. The available evidence suggests that only a small percentage of fructose-derived carbon is directly converted into lipid within 4–6 h [45
]. However, labeled acetate infusion studies do support DNL-promoting effects of fructose in the subacute-to-long term [46
], suggesting that chronic fructose exposure induces prolipogenic mechanisms (Figure 3
), in addition to increasing the availability of DNL substrate.
Carbohydrate response element-binding protein (ChREBP), also known as Mlx interacting protein-like (MLXIPL), is a key transcription factor for enzymes in the fructolysis, glycolysis, gluconeogenesis, and DNL pathways [56
]. The least potent isoform, ChREBPα, is inhibited under conditions of low glucose [59
]; however, upon activation by intracellular carbohydrate metabolites, ChREBPα induces the transcription of a more potent isoform, ChREBPβ, from an alternative promotor [60
]. Thus, the ChREBPα/β mechanism senses intracellular carbohydrate signals and regulates the expression of metabolic gene programs in response to increased carbohydrate availability [61
As fructose is primarily metabolized by the liver [30
], fructose, more than glucose, may give rise to the intrahepatic carbohydrate metabolites that activate hepatic ChREBP independently of hepatic insulin signaling [29
]. Observations in multiple species now support an important role for hepatic ChREBP in fructose metabolism and metabolic disease. In rats, high-fructose feeding, as compared to glucose, was associated with increased ChREBP activity and expression of its target genes [65
]. In accordance, high-fructose feeding to mice increased hepatocellular carbohydrate metabolites, expression of ChREBP target genes, and hepatic steatosis, and these adverse metabolic effects of the high-fructose diet were fully dependent on hepatic ChREBP activation [58
]. In another recent rodent study, fructose-activated ChREBP induced the hepatic expression and secretion of fibroblast growth factor 21 (FGF21), a response that was essential for adaptive fructose metabolism [66
]. Fructose ingestion also acutely raised circulating FGF21 levels in humans [43
], warranting further investigation into the ChREBP-FGF21 signaling axis in fructose metabolism.
Notably, in the absence of ChREBP, high-fructose diets do not cause hepatic lipid accumulation, but inflammation and early signs of fibrosis instead [68
]. It thus seems that ChREBP-mediated lipogenesis is required to prevent even more adverse hepatotoxicity upon fructose consumption, but that this adaptive mechanism may cause hepatic steatosis when too much fructose is consumed [69
]. Increased hepatic ChREBP expression is also associated with NAFLD and insulin resistance in obese humans [70
], but a direct effect of ingested fructose on ChREBP in human liver has not yet been demonstrated.
Sterol regulatory element-binding protein 1c (SREPB1c) is another key transcription factor for genes in the DNL pathway and implicated in the development of NAFLD [71
]. The expression and post-translational activation of SREBP1c are strongly stimulated by insulin signaling; this mechanism ensures a coordinated DNL response under conditions of high glucose availability [57
]. High-fructose diets commonly induce systemic insulin resistance and fasting hyperinsulinemia [58
], thereby promoting insulin-mediated SREBP1c activation and hepatic lipid synthesis. However, the observations that insulin-depleted [75
] and liver-specific insulin receptor knockout [76
] mice also display an induction of SREBP1c upon acute sugar administration or on a short-term high-fructose diet indicate that this transcription factor is also regulated by nutrient signals, independent of insulin signaling. Finally, monosaccharides activate other transcriptional coactivators, including liver X receptor (LXR) [77
] and peroxisome proliferator-activated receptor γ coactivator 1β (PPARGC1B) [78
], that may amplify the ChREBP and SREBP1c-mediated lipogenesis response.
In summary, ingested fructose, unlike glucose, is primarily handled by hepatocytes, increasing the availability of intrahepatic carbohydrate metabolites. These provide hepatocytes with both DNL substrate and regulatory signals, mediated via several key lipogenesis transcription factors, for effective hepatic lipid synthesis.
3.2. Fructose and Lipolysis
In healthy human subjects, acute fructose ingestion is associated with a decrease in circulating FA levels [79
], suggesting inhibition of adipose tissue lipolysis or increased FA clearance, but the signaling mechanism is currently unknown. An antilipolytic effect of fructose was also demonstrated in isolated rat adipocytes [80
] and in healthy subjects after a 7-day high-fructose diet [81
]. Adipose tissue lipolysis is primarily suppressed by insulin [82
], but fructose ingestion does not stimulate a strong insulin excursion [43
]. Another as-yet-unknown signal likely mediates this physiological effect of fructose ingestion.
Obesity, inflammation, and other mechanisms promote resistance to insulin’s antilipolytic effect [83
]. This phenomenon is commonly referred to as adipose tissue insulin resistance and contributes to the increased release of FA from fat depots [84
], which may contribute to intrahepatic lipid accumulation and the pathogenesis of NAFLD [85
]. In obese insulin-resistant subjects, a relatively small increase in insulin following a fructose-rich meal or drink may not be sufficient to suppress adipose tissue lipolysis [86
]. This would result in the continuous mobilization of endogenous energy stores under conditions of sufficient nutrient (ingested fructose) availability. These calories may then be diverted to the liver.
Increased visceral adiposity is strongly associated with adipocyte insulin resistance [87
]. Since visceral adipose tissue releases FA into the portal circulation, which directly drains to the liver, this adipose compartment is particularly important in the context of NAFLD [26
]. In one 10-week diet-intervention trial, the consumption of fructose, but not glucose, promoted the accumulation of visceral adipose tissue [55
]. At the same time, 24-h average plasma FA levels were unchanged after the fructose diet, and the authors conclude that the harmful effects of fructose are not likely due to an effect on FA mobilization. However, since visceral lipolysis and/or portal FA levels were not quantified, these data do not exclude the possibility that chronic high fructose consumption contributes to an increased adipose tissue-liver FA flux.
There is some indirect evidence in support of this possibility. Insulin signaling in visceral white adipose tissue of rats is attenuated after 2 months of fructose, but not glucose, supplementation [88
]. In another rat study, the administration of a high-fructose diet increased FA release from isolated adipocytes, as compared to glucose [89
]. Human observational studies link the increased consumption of sugar-sweetened (fructose-containing) beverages to the accumulation of visceral adipose tissue [90
] as well as intrahepatic lipids [91
]. In addition, the consumption of excess calories from fructose contributes to weight gain [92
], which promotes visceral adiposity, inflammation, and insulin resistance [83
]. Thus, it remains debated whether fructose consumption directly influences adipocyte lipolysis and/or visceral adiposity, but it may contribute to an increase in adipose-to-liver FA trafficking through several indirect mechanisms.
3.3. Fructose and Hepatic Lipoprotein-Triglyceride Uptake
Approximately 15% of intrahepatic lipids are derived from dietary fat [18
], and specific types of dietary fat, such as saturated FA [93
], may increase intrahepatic lipid accumulation. Triglycerides from chylomicrons are hydrolyzed by lipoprotein lipase (LPL) for storage or oxidation in peripheral tissues. Spillover of FA from lipolysis of chylomicron-triglycerides is one route by which dietary lipids may end up in the liver [94
]. Another is the clearance of chylomicron remnants, which carry triglycerides in addition to cholesteryl esters, after binding to the low-density lipoprotein (LDL) receptor on hepatocytes [95
Hepatic lipoprotein metabolism also involves the secretion of VLDL particles (discussed below), remodeling of VLDL remnants, and clearance of cholesterol-rich LDL particles through LDL receptor-mediated endocytosis (for reviews, see [94
]). This last route may also contribute, albeit limitedly, to hepatic triglyceride import. However, the hypertriglyceridemia associated with visceral adiposity is primarily due to impaired clearance of VLDL-triglycerides and lipoprotein remnants [94
], suggesting that lipoprotein-triglyceride uptake is not necessarily increased in the context of the metabolic syndrome. In fact, tracer studies of triglyceride kinetics in obese humans suggest that liver fat is a determinant of lipoprotein synthesis, but not of lipoprotein clearance [50
Short-to-medium-term high-fructose diets (conditionally) raise VLDL-associated plasma triglycerides in humans [55
]. It is therefore likely that such diets also increase VLDL remnant concentrations and, by extension, hepatic remnant uptake [100
]. We are, however, not aware of studies that have directly investigated the effect of an acute fructose load or chronic fructose consumption on hepatic lipoprotein (remnant) uptake in humans. In hamsters, high-fructose feeding was recently associated with the development of dyslipidemia and increased PCSK9-mediated hepatic LDL receptor degradation [101
]. This suggests that decreased hepatic lipoprotein-triglyceride uptake may be one mechanism by which fructose contributes to dyslipidemia. However, the effect of fructose on hepatic lipoprotein handling in the context of NAFLD needs further exploration.
3.4. Fructose and β-Oxidation
Fructose is a high-energy nutrient. Its triose phosphate metabolites can enter the citric acid cycle for intrahepatic oxidation or the synthesis of glucose, lactate, and/or lipids (Figure 1
). Most human cells lack the fructolysis machinery, but will readily use the newly synthesized glucose and lactate as energy source. Thus, via this two-step mechanism [102
], fructose produces an energy source shift from lipid to carbohydrate oxidation [39
], and this may influence the accumulation of intrahepatic lipids by decreasing hepatic lipid oxidation and output (Figure 2
). This was demonstrated in rat livers, where perfusion with fructose-spiked blood inhibited hepatic β-oxidation of FA by substrate competition [103
], as well as in healthy subjects, where the ingestion of fructose acutely increased carbohydrate oxidation and suppressed fat oxidation rates [79
]. It is noteworthy that the inhibitory effect of fructose on FA oxidation in rat livers was amplified by co-administration of insulin [103
], suggesting that it may be altered in the context of insulin resistance. Nevertheless, whether impaired β-oxidation is quantitatively relevant to clinical NAFLD remains inconclusive [26
3.5. Fructose and Hepatic VLDL Secretion
Excess triglycerides are secreted from the liver in triglyceride-rich VLDL, which functions as a transporter for endogenous lipids in otherwise hydrophobic plasma. Most [49
], but not all [19
] of the available evidence indicates that the increased availability of liver fat in the context of NAFLD drives up hepatic VLDL secretion. Thus, in order for hepatic steatosis to progress, the rate of triglyceride synthesis (from DNL or FA re-esterification) must overcome the increased rate of VLDL secretion [105
]. To our knowledge, it is unknown if fructose metabolites directly influence hepatic VLDL synthesis or secretion pathways. It is, however, generally thought that chronic fructose consumption and the increase in lipid synthesis will provide the liver with excess triglycerides, allowing for increased VLDL secretion [106
Under postprandial conditions, insulin enhances LPL activity and triglyceride uptake in adipose tissue [106
]. Fructose ingestion, however, does not induce high insulin levels and is, therefore, associated with relatively lower postprandial (insulin-stimulated) triglyceride clearance. This leads to large chylomicron and VLDL particles in circulation, hypertriglyceridemia, and, possibly, more remnants to circle back to the liver.
Finally, long-term exposure to intrahepatic lipid species results in endoplasmic reticulum (ER) and oxidative stress, which increases the degradation of apolipoprotein B100 and reduces VLDL secretion [107
], suggesting that impaired hepatic triglyceride export may worsen hepatic steatosis in the long term. Accordingly, patients with NASH are characterized by distinctly impaired apolipoprotein B100, but not global protein, synthesis [52