Hepatocytes take up fatty acids from the circulation both via simple diffusion (non-saturable) and facilitated transport (saturable). Facilitated transport accounts for more than two-thirds of total fatty acid uptake under most circumstances and is mediated by various fatty acid carrier proteins, e.g., fatty acid binding protein (FABP), fatty acid translocase (FAT/CD36), and fatty acid transport polypeptide (FATP) [71, 72]. The total uptake of FFA by hepatocytes directly depends on the concentration of FFA in plasma (within the physiological range, i.e., <1 mmol/l) as well as on the capacity of hepatocytes for FFA uptake which is predominantly determined by the number and/or activity of transporter molecules [71]. During post-absorptive conditions, the major source of hepatic fatty acids is the systemic plasma FFA pool. These are FFA released predominantly from subcutaneous adipose tissue, but also fatty acids generated during the hydrolysis of circulating lipoproteins in the capillary endothelia of peripheral tissues, which escape tissue uptake, spill over into the systemic circulation, and reach the liver via the hepatic artery and the portal vein after passage through splanchnic tissues. Although lipolysis of visceral adipose tissue triglyceride releases additional FFA directly into the portal vein, the relative contribution of visceral adipose tissue is small: only about 5 % and 20 % of portal vein FFA originate from visceral fat in lean and obese subjects, respectively [73].

The basal rate of adipose tissue fatty acid release into the systemic circulation increases directly with increasing fat mass in both men and women [74]. Independent of the degree of obesity (BMI and total body fat), however, NAFLD is associated with 35–45 % greater basal lipolytic rates [52, 75, 76], and also with impaired insulin-mediated suppression of adipose tissue lipolysis (indicative of adipose tissue insulin resistance) [8, 52, 75]. Hence the release of FFA into the circulation proceeds at greater rates in patients with NAFLD than those without NAFLD during most physiological states (i.e., fasting and feeding). Furthermore, the gene expression of hepatic lipase and hepatic lipoprotein lipase are greater in obese subjects with NAFLD than those without NAFLD [77, 78]. These data suggest that NAFLD is associated with a substantially greater delivery of systemic plasma FFA to the liver, derived from lipolysis of subcutaneous adipose tissue triglyceride and lipoprotein-triglyceride. The capacity of the liver in NAFLD to take up fatty acids is also likely augmented, because of increased hepatic gene expression of several proteins involved in lipid uptake and intracellular transport [79]. For instance, hepatic mRNA and protein levels of CD36 are 65–85 % greater (and proportionally more of the CD36 protein is localized in the plasma membrane of hepatocytes) in subjects with NAFLD than in BMI-matched subjects without NAFLD [80]. These findings indicate that alterations in adipose tissue lipolytic activity, regional hepatic lipolysis of circulating lipoprotein-triglyceride, and capacity of the liver for FFA uptake contribute to increased intrahepatic fatty acid availability, and possibly also triglyceride accumulation, and therefore to the pathogenesis of steatosis.

Besides FFA made available to the liver by the hepatic artery and the portal vein, the liver can also synthesize fatty acids intracellularly from acetyl-CoA that serves as the principal building block. In a complex polymerization process taking place in the cytosol (Lynen cycle), acetyl-CoA is first activated to malonyl-CoA by acetyl-CoA carboxylase (ACC), and undergoes several cycles of condensation, decarboxylation, and reduction reactions to form palmitate [81]. The overall synthesis of fatty acids is catalyzed by the fatty acid synthase (FAS) complex, a single polypeptide containing seven distinct enzymatic activities. Regulation of de novo lipogenesis occurs at a variety of steps and is accomplished by the amount and activity of lipogenic enzymes such as FAS, ACC 1 and 2 (and its regulatory enzyme, AMP-activated protein kinase), diacylglycerol acyltransferase (DGAT) 1 and 2, and stearoyl-CoA desaturase 1 (SCD1), by the expression and activation state of nuclear transcription factors such as sterol regulatory element binding proteins (SREBPs), carbohydrate responsive element binding protein (ChREBP), liver X receptor α (LXRα), farnesoid X receptor (FXR), and peroxisome proliferator-activated receptors (PPARs), as well as by the rate of delivery of acetyl-CoA to the cytosol [72, 81]. In normal subjects, de novo lipogenesis accounts for less than 5 % (in the post-absorptive state) or 10 % (in the postprandial state) of fatty acids incorporated in IHTG and VLDL-triglyceride (~1–2 g/day) [82, 83]. However, the contribution of de novo lipogenesis in subjects with NAFLD is much greater and accounts for 15–23 % of the fatty acids in IHTG and VLDL-triglyceride in the fasting state [82, 84]. De novo lipogenesis increases temporally in the postprandial state [85] and results from a study that used sophisticated MRS techniques suggest that an abnormal increase in hepatic de novo lipogenesis following meal ingestion may precede liver fat accumulation and possibly the development of NAFLD [70]. These observations collectively indicate a key role of hepatic de novo fatty acid synthesis in IHTG accumulation.

The oxidation of fatty acids is a major source of energy for the liver, and occurs primarily in the mitochondria (β-oxidation) and to a much lesser extent in peroxisomes and microsomes. Transport of fatty acids from the cytosol (where they are found following their uptake from plasma or de novo synthesis) to the mitochondrial matrix requires their “activation” by coenzyme A, and is accomplished by a carnitine-dependent enzyme shuttle, sequentially involving carnitine palmitoyl transferase (CPT) 1, carnitine translocase, and CPT2 [86]. Mitochondrial β-oxidation progressively shortens the fatty acyl-CoA chain by two carbon units at each cycle (released as acetyl-CoA) through a series of dehydrogenation, hydration, and cleavage reactions [87]. Several membrane-bound and soluble enzymes are involved in this process, varying in acyl chain length specificity [87]. Acetyl-CoA generated during β-oxidation is disposed either to the tricarboxylic acid cycle (Krebs cycle), where complete oxidation to carbon dioxide generates energy for the liver, or to ketogenesis where acetyl-CoA molecules condense to form ketone bodies (acetoacetate and 3-hydroxybutyrate) [88], which are then released into the bloodstream and provide a source of energy for extrahepatic tissues, including the brain [86]. Control of flux through β-oxidation occurs at many levels; in vivo, it largely depends on the rate of entry of fatty acyl groups into the mitochondria, which is modulated by substrate supply and CPT1 activity [89]. A long-known inhibitor of CPT1 is malonyl-CoA, which is the first metabolic intermediate in the de novo synthesis of fatty acids [90]. The rate of ketogenesis depends on all factors controlling β-oxidation flux, as well as the relative availability of acetyl-CoA to free CoA inside the mitochondrial matrix [88].

Data from studies conducted in rodent models demonstrate that inhibition or activation of intrahepatic fatty acid oxidation by a variety of means can influence IHTG content. Genetic or experimentally induced deficiencies in mitochondrial enzymes involved in the oxidation of fatty acids lead to accumulation of IHTG and hepatic steatosis [91, 92], whereas greater expression or activity of these enzymes reduces IHTG content and ameliorates steatosis [93–96]. Given that hepatic de novo lipogenesis is upregulated in NAFLD [82, 84], it is tempting to speculate that NAFLD is also associated with a coordinate downregulation of hepatic fatty acid oxidation [97] perhaps through the overproduction of malonyl-CoA which would then inhibit CPT1 and the transport of fatty acids into the mitochondria. However, it remains unclear if fatty acid oxidation is reduced in human subjects with NAFLD because there are currently no reliable methods for measuring hepatic fatty acid oxidation in vivo. Indirect measures, such as plasma ketone body concentrations, suggest that hepatic fatty acid oxidation is either normal or, if anything, increased in people with NAFLD [51, 52, 98, 99]. In addition, although CPT1 expression is decreased, gene expression of other hepatic fatty acid oxidative enzymes has generally been found to be greater in subjects with NAFLD than in those with normal IHTG content [79, 100]. In contrast, subjects with NAFLD have evidence of hepatic mitochondrial structural and functional abnormalities, including loss of mitochondrial cristae and paracrystalline inclusions [51, 101], a decrease in mitochondrial respiratory chain activity [102], impaired ability to resynthesize ATP after a fructose challenge [103], and increased hepatic uncoupling protein 2 [100], which could all affect energy production but not fatty acid oxidation. These abnormalities could represent an adaptive uncoupling of fatty acid oxidation and ATP production, which allows the liver to oxidize excessive fatty acid substrates without generating unneeded ATP. At present, therefore, there is little direct evidence to support the notion that decreased hepatic fatty acid oxidation contributes to IHTG accumulation and the pathogenesis of steatosis.

The majority of plasma triglyceride in the post-absorptive state is carried in the hydrophobic core of VLDL. Hepatic triglyceride synthesis and secretion provides a means to buffer excess amounts of FFA (which could otherwise be cytotoxic) and a mechanism whereby energy-dense substrates are delivered to peripheral tissues [104]. Hepatic VLDL assembly takes place in two steps, involving (1) the partial lipidation of a newly synthesized apolipoprotein B-100 molecule, and (2) the fusion of this small and dense precursor with a large triglyceride droplet to form mature VLDL, through the action of microsomal triglyceride transfer protein (MTP) [105]. Each VLDL particle contains a single molecule of apolipoprotein B-100 [106], which remains bound to the lipoprotein particle throughout the intravascular remodeling of VLDL, whereas the amount of core triglyceride varies considerably and determines, in part, the metabolic fate of the whole particle [107].

Fatty acids are made available to the liver from the systemic plasma FFA pool and from several non-systemic fatty acid sources, such as hepatic de novo lipogenesis, lipolysis of intrahepatic fat, lipolysis of lipoprotein-triglyceride taken up by the liver, and lipolysis of visceral fat [108]. In healthy post-absorptive subjects, the majority (65–75 %) of fatty acids utilized for VLDL-triglyceride secretion are derived from the systemic circulation, whereas only a small portion (<5 %) originates from hepatic de novo lipogenesis [109, 110]. Intrahepatocellular fatty acids that are not oxidized are esterified to triglyceride, which can then be incorporated into VLDL and secreted into the circulation, or stored within the liver. Therefore, VLDL secretion provides a mechanism for hepatic triglyceride export and thereby also for reducing IHTG content. In fact, results from studies conducted in both human subjects and animal models indicate that an impairment in hepatic VLDL secretion, caused by genetic defects, such as familial hypobetalipoproteinemia [111], pharmacological agents that inhibit MTP [112], or deletion of CD36 [113], is associated with an increase in IHTG content. However, data from most [76, 114] but not all [82] studies in human subjects have found that NAFLD is associated with a marked increase in VLDL-triglyceride secretion rate, independent of BMI and percent body fat. This increase is almost entirely accounted for by a marked increase in the contribution of non-systemic fatty acids (presumably derived from lipolysis of intrahepatic and visceral fat and de novo lipogenesis) to VLDL-triglyceride secretion [76]. The above findings notwithstanding, VLDL-triglyceride secretion rate increases in a linear fashion with increasing IHTG content within the normal range (up to 5–10 % of liver volume), but plateaus thereafter, with IHTG content within the NAFLD range (>10 % of liver volume). This suggests that IHTG content may drive VLDL-triglyceride secretion but eventually VLDL-triglyceride export reaches a biological limit beyond which it cannot adequately compensate for the increase in IHTG, so steatosis cannot be avoided or resolved.

The mechanism responsible for the inadequate increase in hepatic triglyceride export in NAFLD is not known, but might be related to physical limitations in the liver’s ability to secrete VLDL. The secretion rate of VLDL-apolipoprotein B-100 (which is an index of the secretion rate of VLDL particles themselves, and not their core triglyceride) is not different between patients with NAFLD and BMI- and body fat-matched subjects without NAFLD [76], or is only slightly greater in those with high than those with low IHTG content [115]. Therefore, the molar ratio of VLDL-triglyceride to VLDL-apolipoprotein B-100 secretion rates, an index of the average triglyceride content of nascent VLDL particles, and therefore a marker of their size, is substantially greater (e.g., twofold or more) in those with NAFLD [76]. In a study conducted in transgenic mice that overexpress SREBP-1a and develop massive steatosis, it was observed that very large VLDL particles cannot be secreted from the liver because they exceed the diameter of the sinusoidal endothelial pores, resulting in intrahepatocellular accumulation of triglyceride [116]. Therefore, it may be that the failure to upregulate VLDL-apolipoprotein B-100 secretion in obese subjects with NAFLD to an extent adequate to match the surplus of IHTG available for export results in the packaging of many more triglyceride molecules per nascent VLDL particle, and the formation of triglyceride-rich and very large VLDL particles, some of which cannot penetrate sinusoidal endothelial pores for export out of the liver. Consequently, triglyceride gradually accumulates within hepatocytes and this could eventually lead to the development of NAFLD.