Fructose-mediated effects on gene expression and epigenetic mechanisms associated with NAFLD pathogenesis

Abstract

Nonalcoholic fatty liver disease (NAFLD) is a chronic, frequently progressive condition that develops in response to excessive hepatocyte fat accumulation (i.e., steatosis) in the absence of significant alcohol consumption. Liver steatosis develops as a result of imbalanced lipid metabolism, driven largely by increased rates of de novo lipogenesis and hepatic fatty acid uptake and reduced fatty acid oxidation and/or disposal to the circulation. Fructose is a naturally occurring simple sugar, which is most commonly consumed in modern diets in the form of sucrose, a disaccharide comprised of one molecule of fructose covalently bonded with one molecule of glucose. A number of observational and experimental studies have demonstrated detrimental effects of dietary fructose consumption not only on diverse metabolic outcomes such as insulin resistance and obesity, but also on hepatic steatosis and NAFLD-related fibrosis. Despite the compelling evidence that excessive fructose consumption is associated with the presence of NAFLD and may even promote the development and progression of NAFLD to more clinically severe phenotypes, the molecular mechanisms by which fructose elicits effects on dysregulated liver metabolism remain unclear. Emerging data suggest that dietary fructose may directly alter the expression of genes involved in lipid metabolism, including those that increase hepatic fat accumulation or reduce hepatic fat removal. The aim of this review is to summarize the current research supporting a role for dietary fructose intake in the modulation of transcriptomic and epigenetic mechanisms underlying the pathogenesis of NAFLD.

Introduction

Excessive hepatic fat accumulation underlies NAFLD pathogenesis. Fat accumulates in the liver as a result of increased lipid accumulation, mostly attributed to enhanced rates of de novo lipogenesis and hepatic fatty acid uptake, or decreased lipid disposal through fatty acid oxidation and removal to the circulation [1]. NAFLD encompasses a histological spectrum from simple steatosis to nonalcoholic steatohepatitis (NASH), with variable degrees of fibrosis [2, 3]. Liver-related morbidity and mortality are increased in NASH patients [4] and rises disproportionately with fibrosis stage [5]. NASH is recognized as the major cause of chronic liver disease and is emerging as the most common indication for liver transplantation [6].

NAFLD patients are at greater risk for developing T2D [7], while individuals with T2D have a higher prevalence of NAFLD and are more likely to experience faster disease progression and higher stage of fibrosis [4, 8, 9]. Further, the greater the number of metabolic syndrome components in NAFLD patients, the higher the risk of death [10]. While the majority of NAFLD patients are overweight or obese [11], NAFLD does develop in normal weight individuals, a clinical subtype referred to as lean NAFLD. Younger age and female sex are independent predictors of lean NAFLD [12]. Obesity-related NAFLD and lean NAFLD have dissimilar clinical profiles, and some studies have shown that lean NAFLD patients present with a severe histological phenotype and may experience higher mortality and morbidity compared to NAFLD patients who are overweight and obese [13–16].

The global prevalence of NAFLD is increasing and is estimated to be ~ 25% in the general population [4]. The prevalence of NAFLD is also rising in children and adolescents, in tandem with rates of obesity [17]. Some studies have reported differential prevalence by sex, although these patterns may be reflective of regional variations in environmental or genetic exposures [10]. The prevalence of NAFLD has been reported to be twice as high in women with prior gestational diabetes (GDM) compared to those with no history of GDM [18]. Offspring of obese women with GDM showed a 68% increase in intrahepatocellular fat at 1–3 weeks of age compared to those born to non-GDM women [19].

NAFLD prevalence varies by ethnicity. In the United States, rates of NAFLD, NASH, and NAFLD-related fibrosis are highest in Latinos, followed by Asians, Non-Latino Whites, and African Americans [20–23]. These interethnic differences in NAFLD prevalence indicate population-specific susceptibility, which could be attributed, in part, to genetic factors. It is well recognized that NAFLD susceptibility is at least partially mediated by genetic factors, based on evidence from familial aggregation [24–26], twin [27–29], and population-based [30–32] studies. Variants in a number of genes, including those that encode patatin-like phospholipase domain-containing protein 3 (PNPLA3) and transmembrane 6 superfamily 2 human gene (TM6SF2) have been associated with NAFLD and NASH across multiple, independent studies, which have been reviewed elsewhere [33, 34].

In addition to genetic factors, dietary components, particularly saturated fat, cholesterol, and sucrose, are associated with the development of NAFLD [35]. The predominant caloric sweeteners in the United States are sucrose, a disaccharide comprised of one molecule of fructose covalently linked with one molecule of glucose, and high fructose corn syrup, a man-made sweetener with a moderately higher fructose: glucose ratio (typically 55%:45%) compared to sucrose. There are distinct differences in metabolism between fructose and glucose in humans [36]. While glucose is metabolized by every cell of the body, the metabolism of fructose is largely restricted to the liver; fructose metabolism is also associated with the depletion of adenosine triphosphate (ATP), which is linked with purine nucleotide turnover and uric acid production. Uric acid may contribute to oxidative stress and mitochondrial dysfunction [36]. Fructose, but not glucose, consumption has also been shown to increase hepatic fractional de novo lipogenesis in humans under conditions of isocaloric feeding [37].

As of the most recent estimates in 2007, global average per capita sugar consumption was ~ 65 g/day [38]. Increased sugar consumption has been linked with a range of negative health effects from dental caries to obesity and T2D [39] and organizations including the World Health Organization [40] and the American Heart Association [41] have recommended reducing the intake of free and added sugars. While methods for quantifying fructose consumption in the general population are inaccurate compared to those for sucrose, all estimates indicate that fructose consumption in the United States has increased [38]. Given that neither sucrose, nor fructose, is an essential component of the human diet or necessary for human health, efforts to reduce added sugar consumption in the population in general, and in children specifically, are not unreasonable.

Because many excellent reviews of the epidemiological and observational studies of dietary fructose consumption and NAFLD have already been published [36, 42–48], only a few key studies will be summarized here. One of the first studies to link dietary sugar intake with NAFLD found that self-reported soft drink consumption in NAFLD patients was almost twice that of healthy individuals, and higher soft drink intake (i.e., 31 g/day of sugar intake) was associated with higher risk for NAFLD [odds ratio (OR) 1.45; 95% confidence interval (CI 1.13, 1.85); p = 0.005] [49]. In an unrelated study, approximately 80% of NAFLD patients, in whom classic risk factors such as T2D and overweight were lacking, self-reported excessive soft drink consumption (more than 50 g/day of sugar), while only 20% of healthy individuals reported high consumption [50]. Soft drink consumption was the only independent variable that predicted NAFLD in > 80% of patients with a specificity of 76%, a positive predictive value of 57%, and a negative predictive value of 100%, after controlling for dietary composition and physical activity [50]. Similar findings were reported in NAFLD patients lacking risk factors for metabolic syndrome, in whom self-reported soft drink consumption was three times as prevalent as in controls, and soft drink consumption was significantly associated with fatty liver [OR 2.0 (1.0, 5.0); P = 0.03] [51]. Likewise, self-reported sugar-sweetened beverage consumption was higher in NAFLD patients compared to controls (365 kcal vs 170 kcal; p < 0.05) and hepatic levels of ketohexokinase (KHK) and fatty acid synthase (FASN) in NAFLD patients were significantly higher compared to levels in individuals with normal liver fat [52].

Dietary fructose consumption has also been associated with NAFLD severity. Self-reported sugar-sweetened beverage consumption in 341 individuals from the NASH Clinical Research Network repository was univariately associated with decreased age, male sex, hypertriglyceridemia, low levels of high-density lipoprotein (HDL) cholesterol and serum glucose concentrations, increased daily caloric intake, and hyperuricemia [53]. Daily fructose consumption in these individuals was associated with lower steatosis grade [OR 0.4 (0.2, 0.9); p = 0.02] and higher fibrosis stage [OR 2.6 (1.4, 5.0); p = 0.04]. In 271 obese children and adolescents with NAFLD, fructose consumption was independently associated with the more severe NASH phenotype [OR 1.6 (1.25, 1.86), p = 0.001] [54].

In overweight and obese individuals given glucose- or fructose-sweetened beverages (25% of daily caloric intake) for 10 weeks, hepatic de novo lipogenesis and 23-h postprandial hypertriglyceridemia were increased during fructose consumption [37]. While both glucose- and fructose-consuming groups gained similar amounts of weight, visceral adipose volume was significantly increased only in response to fructose consumption. Plasma lipid and lipoprotein concentrations increased during fructose consumption but remained unaltered during glucose consumption, with the exception of fasting triglyceride concentrations, which significantly increased in response to glucose. Markers of lipid metabolism and lipoprotein remodeling were also elevated during fructose, but not glucose, consumption. In a randomized crossover study comprised of 34 normal weight, healthy men, moderate amounts of fructose supplementation (40 g/day), but not high sucrose (80 g/day) or glucose (80 g/day), increased fatty acid synthesis [55]. A recent randomized controlled trial of adolescent boys with NAFLD reported that 8 weeks of sugar restriction (free sugar intake less than 3% of daily calories) resulted in significant decreases in hepatic fat fraction in intervention youth compared to controls [− 6.23% (95% CI −9.45% to −3.02%; P < 0.001)] and these reductions in liver fat were independent of changes in weight or measures of adiposity [56].

Collectively, these studies provide compelling evidence that excessive fructose consumption is associated with the presence of NAFLD and might promote the development and progression of NAFLD to more clinically severe phenotypes. To date, however, the molecular processes by which fructose elicits effects on liver metabolism have remained only partially understood, with the majority of research efforts focused on animal models. The purpose of this review is to summarize the current research findings supporting a role for dietary fructose intake in the modulation of transcriptomic and epigenetic mechanisms underlying the pathogenesis of NAFLD.

Fructose effects on molecular determinants of NAFLD

Fructose-induced changes in hepatic gene expression

Dietary factors have long been recognized to exert both direct and indirect influences on gene expression [57]. A number of studies have sought to better understand the effect of fructose ingestion on the liver by investigating changes in gene expression following dietary exposures. Early reports identified dietary fructose effects on individual regulatory genes involved in glycolysis, lipogenesis, and gluconeogenesis (Fig. 1). Fructose feeding was shown to stimulate hepatic expression of L-type pyruvate kinase (Pklr) in normal [58, 59] and, to a lesser extent, diabetic rats [58]. Lower expression levels in diabetic liver were attributed to insufficiency of insulin, which is required for fructose effects on Pklr [58]. Fructose-feeding in female Sprague–Dawley rats also upregulated hepatic expression of fatty acid synthase (Fasn) and acetyl-coenzyme carboxylase (Acaca), both of which encode important lipogenic enzymes, compared to cornstarch-fed animals [60]. Rats receiving a diet of 67% carbohydrate (comprised of 98% fructose) for 8 weeks also showed higher Fasn and Acaca levels, as well as upregulation of sterol regulatory element-binding transcription factor 1 (Srebp1c, also known as Srebf1) and reduced expression of peroxisome proliferator-activated receptor, alpha (Ppara) [61]. Activation of Ppara by fenofibrate, a PPARA agonist, attenuated insulin resistance, hypertension, hyperlipidemia, and hepatic fat accumulation caused by high fructose feeding. In rats fed a diet containing 68% sucrose, hepatic glucose-6-phosphatase (G6pc) levels increased 1.6-fold compared to animals under food-deprived conditions [62]. In mice treated with a high fructose diet (160 g/kg per day), which resulted in significant liver steatosis, hepatic expression of genes involved in lipolysis [i.e., lipase E, hormone-sensitive type (Lipe), also known as Hsl] and lipid uptake [i.e., low-density lipoprotein receptor (Ldlr), and very low-density lipoprotein receptor (Vldlr)] were reduced, while those involved in lipogenesis [Acaca, Srebp1c, and stearoyl CoA desaturase (Scd)] were increased compared to chow-fed animals [63]. In addition, genes encoding pro-inflammatory proteins, tumor necrosis factor alpha (Tnfa), interleukin 6 (Il6), and C–C motif chemokine ligand 2 (Ccl2) were upregulated ~ 8–25-fold in response to fructose. Fructose feeding also corresponded with reduced expression of sirtuin 1 (Sirt1), Ppara, and fibroblast growth factor 21 (Fgf21). Fisher et al. [64] found increased hepatic expression of carbohydrate-response element-binding protein (Chrebp) and Fgf21, as well as elevated levels of circulating Fgf21 in mice following fructose gavage. Fructose-induced effects on Fgf21 were absent in Chrebp knockout mice, and were accompanied by decreased de novo lipogenesis, VLDL secretion, and hepatic fat compared to fructose-fed control mice. Serum FGF21 levels in humans correlated with rate of de novo lipogenesis, consistent with the animal studies, indicating that a Chrebp-Fgf21 axis may play a role in fructose-induced effects on NAFLD pathogenesis across species.

Fig. 1.

Major metabolic pathways affected by dietary fructose in the liver. Summary of molecular effects of fructose on gene expression results from human, animal models, and in vitro studies

Interestingly, gene expression changes observed in fructose-fed mice were also observed in the human hepatoma HepG2 cell line following treatment with 25 mM fructose [63]. These results are consistent with recent studies showing fructose-mediated alterations in expression of lipogenic genes, as well as mitochondrial respiration [65] and hepatocyte biomechanics [66] in cultured hepatocytes. In male mice fed a high fructose diet (50% kcal from fructose) for 8 weeks, mRNA and protein levels of lipogenic genes, including Srebp1c, Chrebp, Fasn, Cd36, and peroxisome proliferator-activated receptor gamma (Pparg), as well as interleukin 1 beta (Il1) and interleukin 6 (Il6), were significantly elevated relative to animals on a normal chow regime [67]. Levels of beta-oxidation genes, acyl-coenzyme A, oxidase 1, palmitoyl (Acox1), carnitine palmitoyltransferase 1A (Cpt1a), Pparg coactivator 1 alpha (Ppargc1a), and Ppara decreased in response to fructose feeding [67]. In mice fed a diet supplemented with 474.3 g/kg fructose for 10 weeks, hepatic expression of Srebp1c, Chrebp, Ppara, Il1, Il6, tumor necrosis factor alpha (Tnf), and vitamin D receptor (Vdr) increased relative to fructose-free animals [68].

Fructose has been found to induce Srebp1c and other lipogenic genes more strongly than glucose [69]. Srebp1c expression is also regulated by stearoyl-CoA desaturase (Scd), which synthesizes oleate, the fatty acid necessary for triglyceride production. In Scd^−/− mice, fructose induction of Srebp1c, as well as other lipogenic genes was lost and triglyceride levels remained normal [69]. In Srebp1c^−/− mice, fructose-feeding induced the hepatic expression of Scd, which was potentiated by the addition of oleate, suggesting that Scd-derived oleate may be necessary for fructose-mediated induction of lipogenic gene expression and that Srebp1c may mediate this interaction.

Maternal fructose feeding during pregnancy has been demonstrated to affect gene expression in offspring. Clayton et al. [70] fed fructose (20% of the calories) to pregnant rats from day 1 of pregnancy until postnatal day 10 (P10), and then measured hepatic gene expression in mothers, fetuses, and neonates. Prenatal fructose exposure significantly increased expression of the fructose-specific transporter Slc2a5 (solute carrier two member five, also known as Glut5) at embryonic day 21 (E21) and P10, Khk at P10, Srebp1c and Sirt1 levels at E21, Ppara at P10, and markers of endoplasmic reticulum stress, heat shock protein 5 (Hspa5) and X-box binding protein 1 (Xbp1) at E21 and P10, respectively. Levels of phosphoenolpyruvate carboxykinase (Pepck) were decreased at E21 and P10. In offspring, maternal fructose exposure increased hepatic Slc2a5 and reduced Pepck expression as early as E21, and by P10, genes involved with beta-oxidation were downregulated. Maternal exposure to fructose also induced age- and sex-specific changes in expression of pro-inflammatory genes and clock genes [70]. Hepatic gene expression at P10 was differentially regulated by fructose in male and female neonates, with a preponderance of beta-oxidation genes altered in males and genes involved in fatty acid metabolism in females, independent of changes in weight gain, food intake, or serum glucose levels (Table 1). The only differentially expressed gene in common between male and female neonates was lipoprotein lipase (Lpl), which was significantly downregulated in both groups compared to offspring of dams fed a normal chow diet. These results demonstrate that fructose differentially affects hepatic regulation of lipogenesis in males and females, and does so early in life, well before the development of obesity and insulin resistance.

Table 1.

Sex-dependent effects of prenatal fructose exposure on genes involved in fatty acid metabolism in mouse neonates at postnatal day ten

Symbol	Gene name	M	F	FC*	P value
Acad10	Acyl-CoA dehydrogenase family member 10	x		− 1.69	0.03
Acat1	Acetyl-CoA acetyltransferase 1	x		− 2.42	0.03
Acsl4	Acyl-CoA synthetase long chain family member 4	x		− 1.83	0.006
Crat	Carnitine O-acetyltransferase	x		− 1.93	0.006
Crot	Carnitine O-octanoyltransferase	x		− 1.60	0.017
Cpt1a	Carnitine palmitoyltransferase 1A	x		− 1.38	0.07
Echs1	Enol-CoA hydratase, short chain 1		x	− 1.17	0.02
Eci1	Enoyl-CoA delta isomerase 1	x		− 1.69	0.04
Fabp1	Fatty acid binding protein 1		x	− 1.32	0.003
Fabp2	Fatty acid binding protein 2		x	− 1.50	0.06
Fabp7	Fatty acid binding protein 7	x		− 1.95	0.01
Gpd2	Glycerol-3-phosphate dehydrogenase 2	x		− 1.96	0.03
Hadha	Hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit alpha		x	1.33	0.007
Lpl	Lipoprotein lipase	x	x	− 1.86/− 1.60	0.02/0.02
Prkaa1	Protein kinase AMP-activated catalytic subunit alpha 1		x	1.50	0.003
Slc27a1	Solute carrier family 27 member 1		x	− 1.6	0.07
Slc27a3	Solute carrier family 27 member 3	x		2.80	0.03

*Fold change relative to offspring of chow-fed rats

In mice fed regular chow, high fat (HF 58% kcal fat), or high fat/high carbohydrate (HFHC 58% kcal fat + 55% fructose and 45% sucrose in drinking water) diet, hepatic expression of collagen 1 (Col1A1) mRNA was > 7.0-fold higher and actin alpha 2, smooth muscle (Acta2) was threefold higher in HFHC diet mice compared to HF-fed mice, while Tgfb1 levels were almost fourfold higher in HFHC diet mice compared to chow-fed mice [71]. Hepatic expression of Ccl2 was significantly increased in rats fed a diet comprised of 40% fat, 40% carbohydrate (mostly fructose), and 2% cholesterol (NASH diet), 60% fat, 20% carbohydrate (high fat diet), or 10% fat, 70% carbohydrate (mostly fructose; high fructose diet) compared to chow-fed animals [72]. Interestingly, Ccl2 levels were also elevated in adipose tissue in the fructose-fed rats relative to rats that received either the NASH or high fat diet.

Hepatic stellate cells (HSCs) are the primary storage site for vitamin A under physiologic conditions [73]. In response to hepatic injury, HSCs are transdifferentiated to myofibroblasts, which secrete cytokines and other molecules to attenuate damage to the liver. HSC activation is characterized by loss of intracellular lipid droplets and increased production of ACTA2, COL1A1, and other ECM components. Following resolution of injury, activated HSCs are removed through apoptosis and inactivation [74], but in the presence of chronic hepatic inflammation, these cells continue to produce ECM, which contributes to liver fibrosis. The effects of fructose ingestion on HSC gene expression patterns was investigated in mice consuming a Western diet supplemented with fructose (42 g/L)-infused drinking water (WD). In these animals, hepatic steatosis developed by 6 weeks of feeding and increased with longer treatment duration up to 24 weeks, eventually leading to inflammation and low-grade pericellular fibrosis in the majority of animals [75]. At 12 weeks, Acta2-positive cells began to emerge, indicative of HSC activation. RNA-sequencing of HSCs from WD and chow-fed mice identified differential expression of ~ 5000 genes. K-means and hierarchical clustering demonstrated similar changes in HSC gene expression in WD-fed mice and mice treated with carbon tetrachloride (CCl₄), a common strategy to chemically activate HSCs and induce hepatic fibrosis. Although WD and CCl4 elicit distinct histopathological changes in the liver and contribute to hepatic fibrosis in different ways, the transcriptional programs underlying HSC activation and fibrogenesis were similar between models.

Differential effects of fructose and glucose on hepatic gene expression

In addition to the differences in metabolism mentioned earlier, fructose and glucose exert distinct effects on hepatic gene expression and metabolic sequelae. In rats receiving a fructose- or glucose-supplemented diet (63% w/w) for 2 weeks, liver glycogen was higher in fructose-fed animals compared to those on the glucose diet; fructose-fed rats also presented with significant hepatomegaly [76]. Fructose-feeding corresponded with higher plasma levels of triglycerides, VLDL, LDL, and insulin. Hepatic expression of fifteen genes involved with fructose metabolism, glycolysis, gluconeogenesis, and lipogenesis, or encoding transcription factors was compared between fructose- and glucose-fed groups. Compared with the glucose-fed group, the fructose-fed rats showed higher hepatic expression of Khk, aldolase B (Aldob), phosphofructokinase 1 (Pfk1), pyruvate kinase (Pk), fructose-bisphosphatase 2 (Fbp2), glucose-6-phosphatase (G6pc), Fasn, glycerol-3-phosphate acyltransferase (Gpat), microsomal triacylglycerol transfer protein (Mttp), and Chrebp. Comparisons among mice given either chow (21.6% of calories from fat) or high fat diet (HFD: 60% of calories from fat) and ad libitum access to untreated drinking water or water supplemented with either glucose or fructose (30% w/v) revealed no major physiological differences in fructose or glucose supplementation in chow-fed animals, yet HFD diet-fed mice receiving fructose-water gained more weight and developed glucose intolerance and hepatomegaly compared to glucose-supplemented HFD-fed mice, despite similar caloric intake [77]. Fructose supplementation combined with a chow diet induced the expression of ATP citrate lyase (Acly), acetyl CoA carboxylase, alpha (Acaca), Fasn, and Scd by 3- to 12-fold. Glucose supplementation did not increase expression of these genes, with the exception of Scd, which was modestly upregulated compared to fructose-supplemented mice (~ 4-fold vs ~ 12-fold induction compared to chow-fed animals). Interestingly, treatment with HFD did not affect expression of these genes, although it attenuated the effects of fructose on transcript levels. In chow-fed mice, fructose supplementation elicited a ~ 3-fold upregulation of Srebp1c, while glucose appeared to slightly decrease its expression. Similar differential expression patterns were observed in HFD-fed mice, albeit with lower fold-change relative to HFD + normal water-supplemented controls. In chow-fed animals, neither fructose nor glucose significantly altered Chrebp expression, although in the presence of HFD, glucose supplementation increased Chrebp expression > tenfold. However, fructose supplementation did not significantly increase Chrebp levels in HFD-fed mice. Results from this study suggested that glucose supplementation yielded a protective influence on metabolic outcomes, whereas fructose was associated with poor metabolic outcomes, and these effects were mirrored by distinct changes in expression of lipogenic genes. These data are consistent with results from human studies showing distinct metabolic outcomes resulting from short term consumption of fructose or glucose beverages [37].

RNA sequence analysis was performed after 10 weeks of chow, HFD, or HFD supplemented with glucose or fructose to identify diet-induced changes in global hepatic gene expression [77]. Fructose supplementation significantly upregulated expression of genes involved in fatty acid synthesis, including Fasn, Acly, Scd, Acaca, pyruvate dehydrogenase kinase, isoenzyme 3 (Pdk3), elongation of long chain fatty acids member 5 (Elovl5) and member 6 (Elovl6), while glucose supplementation upregulated genes associated with triglyceride synthesis [e.g., glycerol 3-phosphate acyltransferase (Agpt9), glycerol-3-phosphate transporter, member 1 (Slc37a1), and diacyl-glycerol-o-acyltransferase 2 (Dgat2)] and fatty acid oxidation [e.g., carnitine transporter solute carrier family 22, member 5 (Slc22a5) and acyl-CoA thiosterase 1 (Acot1)]. Many of these genes were expressed either in response to supplementation with fructose or glucose, but not both. Fructose or glucose supplementation also exerted differences on hepatic expression of genes encoding proteins involved in insulin signaling. For example, fructose supplementation increased expression of phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha (Pik3ca) and mitogen-activated protein kinase 9 (Mapk9), while glucose supplementation upregulated expression of insulin receptor (Insr) and Akt1. Altered patterns of gene expression were suggested to contribute to the differential effects of fructose and glucose on insulin signaling [77]. Differences in expression between the two sugars were also noted for genes participating in mitochondrial function.

In contrast to results from animal studies showing divergent effects of glucose and fructose supplementation on lipogenic gene expression, these differences were not observed in HepG2 cells [78]. In cells treated with 5.5 mM glucose or 5.5 mM glucose + 5.5 mM fructose for 72 h, levels of Fasn and Acaca were not altered by fructose. This discrepant finding may be at least partially attributed to the use of glucose and fructose together, which may mitigate effects of fructose-only supplementation, the relatively low concentration of fructose used in the cell treatments, and the use of a hepatoma cell line, which may not fully recapitulate the metabolism of human hepatocytes.

Collectively, studies involving fructose feeding in mouse and rat models have identified fructose-mediated effects on genes encoding proteins involved in lipid metabolism (Table 2), as well as glycolysis and inflammation. Upregulation of lipogenic genes, namely Fasn, Srebp1c, and Acaca, have been consistently observed across multiple studies despite the use of different experimental designs, animal models, and treatment protocols, indicating that fructose affects de novo lipogenesis, in part, at a molecular level. Despite these findings, the exact molecular mechanisms by which fructose alters gene expression remains unknown. Fructose regulates the expression of a number of transcription factors, suggesting a mechanism by which transcription of downstream genes might be regulated. In addition, as discussed in the following section, fructose treatment has been found to alter levels of specific microRNAs, which target molecules involved in lipid metabolism and other biological processes related to hepatic function.

Table 2.

Fructose-mediated effects on hepatic expression of genes encoding proteins involved in lipid metabolism

Symbol	Gene name	Pathway	Direction	Ref
Acaca	Acetyl-CoA carboxylase alpha	Lipogenesis	Up	[60, 61, 63, 77]
Acly	ATP citrate lyase	Lipogenesis	Up	[77]
Acox1	Acyl-CoA oxidase 1	Fatty acid oxidation	Down	[67]
Chrebp	Carbohydrate-response element-binding protein	Triglyceride synthesis	Up	[64, 67, 68, 76]
Cpt1a	Carnitine palmitoyltransferase 1A	Fatty acid oxidation	Down	[67]
Elovl5/6	EVOVL fatty acid elongase 5/6	Fatty acid elongation	Up	[77]
Fasn	Fatty acid synthase	Lipogenesis	Up	[60, 61, 67, 76, 77]
Fgf21	Fibroblast growth factor 21	Fatty acid oxidation	Down	[63]
			Up	[64]
Gpat	Glycerol-3-phosphate acyltransferase, mitochondrial	Lipogenesis	Up	[76]
Lipe/Lpl	Lipase E, hormone sensitive type/lipoprotein lipase	Triglyceride hydrolysis	Down	[63, 70]
Ldlr	LDL receptor	Lipid uptake	Down	[63]
Mttp	Microsomal triglyceride transfer protein	Lipoprotein assembly	Up	[76]
Ppara	Ppar alpha	Fatty acid catabolism	Down	[61, 63, 67]
			Up	[68, 70]
Pparg	Ppar gamma	Lipogenesis	Up	[67]
Ppargc1a	Pparg coactivator 1 alpha	Fatty acid oxidation	Down	[67]
Scd	Stearoyl-CoA desaturase	Lipogenesis	Up	[63, 69, 77]
Sirt1	Sirtuin 1	Promotes lipolysis	Down	[63]
			Up	[70]
Srebp1c	Sterol regulatory element-binding transcription factor 1	Lipogenesis	Up	[61, 63, 67–70, 77]
Vdr	Vitamin D receptor	Regulates genes involved in lipid metabolism	Up	[68]
Vldlr	VLDL receptor	Lipid uptake	Down	[63]

Aberrant miRNA expression in fructose-enriched diets

MicroRNAs (miRNAs) are small noncoding RNA molecules that modulate gene expression through binding with complementary sequences on target mRNA transcripts [79, 80]. The resulting miRNA–mRNA interactions result in gene silencing by promoting mRNA destabilization and degradation or interfering with mRNA translation into protein [80, 81]. Recent analyses estimate that there are approximately 2300 true human mature miRNAs [82], with more than 45 000 miRNA target sites conserved within the 3′ untranslated regions of human genes [83]. Over 60% of human protein-coding genes appear to be under selective pressure to maintain pairing with miRNAs [83, 84], indicating that miRNA-mediated silencing represents an important mechanism for the regulation of gene expression.

MiRNAs are involved in a variety of disease processes. In the liver, miRNAs are abundantly expressed, where they regulate a diverse array of functions [85, 86] and play important roles in the pathogenesis of many liver diseases including chronic viral hepatitis, NAFLD, drug- or alcohol-induced liver toxicity, autoimmune hepatitis, and hepatocellular carcinoma [86, 87]. We [88] and others [89–91] have reported altered hepatic miRNA levels in patients with NAFLD and NASH, suggesting that these molecules may translate important information to liver cells in response to chronic liver injury.

A number of studies have investigated the relationship between fructose and miRNA expression in the pathogenesis of NAFLD. An early study by Alisi et al. [92] profiled expression patterns of 350 miRNAs in liver tissue from rats fed a standard diet (SD), SD with fructose-supplemented drinking water (30% w/v; SD-HF), a high fat, low carbohydrate diet (HFD) and HFD with fructose-supplemented water (HFD-HF). While all diets resulted in higher insulin levels compared to SD, SD-HF and HFD-HF showed the highest levels, indicating a greater impact of fructose on hyperinsulinemia compared to HFD alone. Histologically, livers from the HFD group showed mild steatosis, while those from the SD-HF group presented with hepatocyte ballooning, signifying liver parenchymal cell death. Liver fibrosis was only observed in animals fed the HFD-HF regime. Fourteen miRNAs showed at least a 1.4-fold change in expression (P < 0.05) in the three diets compared to SD; of these, miR-140 and miR-93 showed expression changes in the SD-HF group opposite to those reported in HFD and HFD-HF, suggesting either a modulatory role of fat on fructose effects or a spurious finding. In addition, miR-200a showed the highest upregulation in the SD-HF group (> 3.5-fold) compared to the HFD and HFD-HF groups (~ 2.5-fold), consistent with findings of miR-200a upregulation in fructose-treated HepG2 cells compared to untreated cells [63]. Interestingly, overexpression or knockdown of miR-200a in HepG2 cells corresponded with reduced or elevated levels of SIRT1 and PPARA, respectively [63].

Dysregulation of hepatic expression of 45 miRNAs was identified by RNA-sequencing analysis in mice fed either a high fructose (60% fructose) or chow diet [93]. Ten of these miRNAs, miR-223-3p, miR-33-5p, miR-145a-3p, miR-125b-5p, miR-128-3p, miR-19b-3p, miR-101a-3p, miR-30a-5p, miR-378a-3p, and miR-582-3p were found to form a regulatory network targeting molecules involved in insulin signaling [i.e., insulin receptor substrate 1 (Irs1) and forkhead box O1 (Foxo1)], hepatic de novo lipogenesis [Srebp1c, Srebp2, Chrebp, insulin induced gene 1 (Insig1), and insulin induced gene 2 (Insig2)], and VLDL assembly and secretion [i.e., microsomal triglyceride transfer protein (Mttp) and apolipoprotein B (apoB)]. In fructose-fed mice, hepatic levels of Irs1 and Insig2 were decreased, while those of Foxo1, Srebp1c, Srebp2, Chrebp, Mttp, and apoB were increased compared to chow-fed animals; expression of these genes was also inversely correlated with levels of predicted targeting miRNAs. Consistent with the findings reported by Alisi et al. [92], hepatic expression patterns of many fructose-induced miRNAs in this study were different from those regulated by HFD, providing further evidence that fructose and fat exert effects on liver metabolism through distinct mechanisms.

In an analysis of 19 miRNAs selected on the basis of prior association with obesity, lipid metabolism, and NAFLD, fructose feeding (23 g fructose/L of water) in lean mice corresponded with hepatic upregulation of miR-33a, miR-221, and miR-27a compared to lean mice supplemented with fructose-free drinking water [94]. In obese mice, fructose-induced expression of miR-34a, miR-21, and miR-27a and downregulated levels of miR-1 and miR-224 compared to fructose-free feeding. Of the dysregulated miRNAs, miR-27a was the only miRNA shared between lean and obese mice treated with fructose, implying that metabolic status has distinct effects in translating fructose-induced changes in miRNA expression. In plasma, fructose-feeding increased levels of miR-27a, miR-335, and miR-19b in lean mice, while miR-21 was the only plasma-derived miRNA upregulated by fructose in obese mice. To date, only these four studies have examined the effect of fructose on hepatic miRNA expression. While a few miRNAs were observed to be dysregulated in at least two studies (i.e., miR-200a and miR-27), for the most part, there was little replication among the published studies. The lack of validation may be reflective of differences in fructose concentrations used, duration of treatment period, and animal model. More studies utilizing physiologically relevant fructose treatments, comparable treatment periods, and similar experimental designs are necessary as a first step toward demonstrating biological relevance of fructose-regulated miRNAs in the pathogenesis of NAFLD.

Fructose effects on DNA methylation

DNA methylation of CpG dinucleotides is a well-characterized epigenetic modification that occurs mostly within CpG islands (CGI) in promoter regions and functions to regulate gene expression. In general, high levels of methylation (i.e., hypermethylation) are associated with repressed gene expression, while low levels (i.e., hypomethylation) are associated with transcriptional upregulation. Environmental exposures, including dietary factors, are known to exert effects on DNA methylation [95, 96] and several groups have examined the specific relationship between fructose consumption and altered patterns of DNA methylation. An early study evaluated the DNA methylation status of the promoters of Ppara and Cpt1a in male rats given 20% fructose solution [97]. After 14 weeks of fructose-feeding, serum levels of triglycerides and total cholesterol were elevated compared to control rats given chow, and total hepatic DNA methylation levels were significantly increased, while those of the control animals were not altered. Because of the relationship between fructose and lipogenesis, the authors selected four genes (Ppara, Cpt1a, Ldlr, and Mttp) for further study. Levels of Ppara and Cpt1a were downregulated in fructose-fed rats, while levels of Ldlr and Mttp were not significantly different from those in control animals. Correspondingly, DNA methylation in the upstream sequences of the two dysregulated genes was significantly increased by fructose-feeding compared to control conditions. Fructose-induced hypermethylation of Ppara and Cpt1a and reduced hepatic expression of these genes provide evidence that pathways involved in lipid metabolism are regulated in part through epigenetic mechanisms. In contrast to these results, global hypomethylation of hepatic mitochondrial DNA, corresponding to upregulation of mitochondrial genes and increased relative mitochondrial DNA content were observed in fructose-fed rats compared to those receiving a control diet [98], consistent with earlier reports showing that excessive fructose exposure affects the liver mitochondrial compartment [61, 99–101]. In the liver, mitochondrial efficiency may increase to produce more ATP for de novo lipogenesis and gluconeogenesis.

High fructose (60%) or high fat (10%) administration for 9 weeks caused global hepatic methylation changes in mice; methylation levels in the dietary groups were reduced around the transcriptional start site in all samples compared to normal chow-fed animals [102]. Interestingly, after reverting animals back to a normal chow diet, DNA methylation changes persisted in a subset of genes. An analysis of the most common differentially hypomethylated promoters in this set of genes revealed 52 CpG loci in the high fat-fed group and 99 CpG loci in the high-fructose-fed group, with only two loci shared between the two data sets. In the hypermethylated promoters, there were 93 and 119 CpG loci in the high fat and high fructose-fed groups, respectively, five of which were overlapping. In the high fructose-fed group, the top persistently hypomethylated genes were apolipoprotein A4 (Apoa4), ATPase Na +/K + transporting subunit alpha 1 (Atp1a1), and potassium inwardly rectifying channel subfamily J member 16 (Kcnj16), while the top hypermethylated genes were fibroblast growth factor 1 (Fgfr1), protein tyrosine phosphatase non-receptor type 11 (Ptpn11), and collagen type IV alpha 2 chain (Col4a2). Apoa4 showed the highest fold change in both the high fat and the high fructose groups compared to normal chow, and these high Apoa4 levels persisted following diet reversion. One of the key findings of this study is that changes in DNA methylation levels resulting from high fat or high fructose diets persisted after reverting back to a normal chow diet. These results have implications for long term consumption of a Western-style diet in methylation-dependent mechanisms underlying NAFLD pathogenesis. While this study compared high fat and high fructose diets independently, a recent study examined effects of a perinatal obesity-inducing diet (FFC), comprised of 12% saturated fat, 0.2% cholesterol, and 23.1 g/L of fructose solution, on NASH fibrosis and DNA methylation changes in offspring [103]. Offspring of FFC diet-fed mice who were weaned to an FFC diet (pFFC–FFC mice) presented with higher body weight, higher fat mass, and higher levels of visceral fat compared to the other perinatal diet groups, despite similar daily caloric intakes among groups. Histologically, this group also presented with a significant increase in hepatic steatosis, Mallory-Denk bodies (a component of ballooned hepatocytes), and early onset fibrosis, while biochemical assessment showed higher levels of macrophage-associated inflammation, and serum ALT levels. Whole liver transcriptomic and methylomic profiling identified 189 genes that were differentially expressed and methylated in pFFC–FFC mice compared to mice from chow-fed parents and weaned to an FFC diet (pChow–FFC). The top canonical pathway identified by gene set enrichment analysis was hepatic fibrosis/hepatic stellate cell activation, indicating that alterations in DNA methylation of mice exposed to the FFC diet perinatally corresponded with a profibrogenic transcriptional program. These findings suggest that perinatal exposures to high fat/high fructose feeding facilitates a fibrogenic environment for the developing liver, possibly through methylomic programming.

Intestinal absorption of fructose

Isotope tracing studies in mice have recently demonstrated that the small intestine metabolizes a significant proportion of dietary fructose, when present in low amounts, but when the dose of fructose increases, the absorption capacity of the intestine is exceeded, leading to spillover to the liver and colonic microbiota [104]. The small intestine converts approximately 40% of fructose into glucose, 20% and 10% are converted to lactate and alanine, respectively, and < 15% leaves the small intestine as fructose [104]. Fructose was also found to increase expression of Glut5, G6pc, triokinase (Tkfc), Aldob, and Fbp1; most of these genes also show to fructose-mediated changes in the liver, as discussed above. What these findings mean with respect to fructose and pathogenic underpinnings of NAFLD is unclear, although the generation of a hepatotoxic metabolite is one possibility put forth by which the small intestine may contribute to fatty liver [104]. It is also possible that excess fructose itself exerts direct effects on the liver, by exceeding the absorption capacity of the small intestine, an explanation that is consistent with the wealth of data supporting effects of fructose on specific hepatic gene expression and epigenetic patterns, as well as development of hepatic steatosis and inflammation. More studies are warranted to explore potential fructose-induced crosstalk between the small intestine and liver with respect to NAFLD pathoetiology.

Conclusions

The prevalence of NAFLD in adults is increasing worldwide [4] and is escalating in the pediatric population in parallel with rising rates of obesity [17]. The most likely culprits for these relatively sudden changes in the prevalence of NAFLD and related metabolic disorders like obesity, insulin resistance, and T2D are higher levels of overall daily energy intake and specific dietary composition. With respect to these changes, consumption of sugar-sweetened beverages in the United States has increased over the last several decades [105–108]. While some have argued against any detrimental effects of high dietary fructose consumption on human health [38, 109, 110], the studies summarized in the present manuscript provide compelling evidence that fructose consumption can indeed lead to disruptions in metabolic pathways that can eventually lead to the development of excessive fat accumulation in the liver.

Apart from the studies by Stanhope et al. [37] and Hochuli et al. [55], few studies have examined the effects of fructose on human metabolism under semi-controlled conditions and instead have relied on cross-sectional observation or animal experiments. However, to truly decipher the metabolic consequences of prolonged consumption of high amounts of fructose and tease out the specific effects on liver metabolism, well-controlled, prospective studies that include unbiased techniques to accurately measure fructose intake and integrate molecular profiling are needed.

Based on the experimental studies conducted to date, fructose regulates a number of genes involved in de novo lipogenesis and fatty acid oxidation through transcriptomic and epigenetic mechanisms. Some evidence suggests that these effects of fructose may depend on metabolic status, and sex, persist despite removal of fructose from the diet, are transmitted from fructose-fed mothers to offspring, and can potentially impact hepatic regulation of lipogenesis well before the development of obesity and insulin resistance.

While there remains much to be learned about the effects of dietary fructose on molecular mechanisms contributing to NAFLD pathogenesis in humans, efforts to reduce global consumption of added sugars in the diet can be expected to yield a beneficial impact on overall health in humans. Reductions are particularly important for school-aged children and adolescents, in whom daily intakes of added sugar are highest [111]. Policy strategies to limit consumption of sugar-sweetened beverages (e.g., soft drinks, sports drinks, energy drinks, and fruit juices) may include public awareness campaigns, warning labels, and limiting accessibility.