As mentioned above, insulin-resistant adipose tissue loses its ability to store lipid and releases free fatty acids into the circulation. In an effort to compensate for adipose tissue insulin resistance, the pancreas produces more insulin, which leads to hyperinsulinemia . Thus, in obesity, the liver encounters excesses of both free fatty acids and insulin (Fig. 4.2a). Liver cells take up fatty acids through the transport proteins FATP5 (fatty acid transport protein 5) and CD36 [28, 29]. Both proteins are upregulated in obesity [30, 29], priming the liver to accommodate the enhanced flux from adipose tissue. Once inside the liver, fatty acids are steered toward storage in the form of triglyceride, which provides an immediate and direct route to hepatic steatosis. During the process of triglyceride synthesis, the levels of several intermediates in the triglyceride pathway also rise. Of particular importance are diacylglycerols (DAG), which can activate protein kinase C epsilon (PKCε). PKCε, in turn, can interfere with the normal function of insulin receptors on hepatocytes [31, 32], resulting in a state of hepatic insulin resistance.

In the liver, insulin is an important regulator of two metabolic functions: gluconeogenesis and lipid synthesis. Gluconeogenesis is normally suppressed by insulin, and thus, hepatic insulin resistance is marked by an increase in hepatic glucose production (Fig. 4.2b). Hepatic lipid synthesis is normally enhanced by insulin; thus, hepatic lipid synthesis should be decreased in an insulin-resistant liver. Curiously, this is not the case, and the paradox has been referred to as “selective insulin resistance ” (Fig. 4.2b) [33]. The fact that hepatic lipid synthesis remains sensitive to insulin in a state of hepatic lipid excess whereas gluconeogenesis does not implies that the two metabolic pathways must diverge at some point. Experimental studies indicate that gluconeogenesis is controlled by forkhead box O1 (FOXO1) downstream of the insulin receptor, whereas lipogenesis is controlled by mammalian target of rapamycin (mTOR) [34]. This leaves open the possibility that in a state of hepatic insulin resistance, other factors besides insulin could continue to stimulate hepatic lipogenesis by activating mTOR.

The lipogenic program activated in the liver during obesity involves more than the esterification of adipocyte-derived fatty acids to triglyceride. It also involves the synthesis of new fatty acids from carbohydrate (de novo lipogenesis; DNL) (Fig. 4.3). The enzymes responsible for DNL are regulated in part by insulin, through the action of the transcription factor sterol regulatory element-binding protein-1 (SREBP1) [35]. Importantly, SREBP1 is active in the liver even during times of hepatic insulin resistance (see above, “selective insulin resistance”). DNL enzymes are also regulated by glucose through the action of a second transcription factor, carbohydrate response element-binding protein (ChREBP) . Together, SREBP1 and ChREBP promote hepatic DNL in response to high-carbohydrate feeding, hyperglycemia, and hyperinsulinemia [35]. In the normal liver, only about 4 % of hepatic triglyceride derives from DNL. In obese individuals, this proportion increases as fatty liver progresses, to as much as 25 % [36, 37]. Because the flux of fatty acids to the liver from adipose tissue does not decline as DNL rises [38, 39], adipose tissue-derived fatty acids must over time be diverted to pathways other than triglyceride synthesis. Evidence indicates they are shuttled to oxidation, to provide the energy necessary to drive the processes of gluconeogenesis and DNL [39].

Once lipid accumulation begins in hepatocytes, it can set into motion events that prompt even more steatosis. A key contributor to this process is the endoplasmic reticulum (ER), an organelle that plays a central role in cellular metabolism. The classical function of the ER is to execute posttranslational processing of proteins. The ER can become stressed, either from an increased demand for protein folding or from derangements in the internal environment of the organelle [40]. Stress in the ER prompts adaptive responses involving the activation of one or more ER membrane proteins as shown in Fig. 4.4. Although these responses are designed to maintain ER homeostasis, they may have undesirable side effects. In hepatocytes, for example, exposure to excess fatty acids places stress on the ER by creating demand for the synthesis of apolipoproteins needed for lipid export as VLDL [18]. This activates the RNA-dependent protein kinase-like ER kinase (PERK), which signals downstream events that result in a suppression of cellular protein synthesis. Although well intentioned, this response prevents rather than facilitates fatty acid export; the ensuing outcome is fatty acid retention and steatosis.

Another important consequence of ER stress is the transformation of SREBP1 from an inactive to a transcriptionally active form [41]. SREBP1, like the ER stress proteins shown in Fig. 4.4, normally resides in the ER membrane; there it is maintained in an inactive state through interactions with a binding partner named SCAP (SREBP cleavage-activating protein) [42]. Under conditions of ER stress, the SREBP1–SCAP complex moves from the ER to the Golgi apparatus; the complex then dissociates, and SREBP1 is released from the Golgi membrane by proteolytic cleavage, enabling its translocation to the nucleus [43]. A connection between ER stress and SREBP1 activation was first theorized when it was noted that the same steps of Golgi transport and membrane release required to process SREBP1 are also needed to activate the ER stress transducer ATF6 (activating transcription factor 6) [44]. Subsequently, experiments showed that ER stress can activate SREBP1 independently of its classic stimulus, insulin. This was a key finding because it revealed yet another potential mechanism by which SREBP1 and hepatic lipogenesis could persist in fatty livers in the face of hepatic insulin resistance. Importantly, if hepatic steatosis induces ER stress in the liver, and ER stress in turn activates SREBP1, this creates a positive feedback loop for the persistence and progression of fat accumulation in the livers of obese individuals once the process is initiated.

Among the adaptive responses to ER stress is an increase in the synthesis of cellular lipids. Lipid synthesis enables a stressed ER to expand its volume and processing capacity [45]. Stress-related lipid synthesis is mediated primarily through inositol-requiring enzyme 1 (IRE1) and its downstream target X-box protein-1 (XBP1) [46–48]. These two molecules have the potential to influence several aspects of lipid metabolism, including fatty acid and triglyceride synthesis, lipid hydrolysis, and lipid transport [49, 50]. In the liver, the lipid alterations that ensue from activation of IRE1 and XBP1 may not always be adaptive. Instead, they may contribute to hepatic steatosis. This was first discovered when disruption of the pathway by silencing XBP1 in the livers of obese mice improved hepatic steatosis [51]. Recent research into the role of the IRE1–XBP1 axis in hepatic lipid homeostasis has cast some doubt on the ability of IRE1 and XBP1 to stimulate lipogenesis directly. In fact, one group suggests that IRE1–XBP1 signaling is actually necessary for the successful lipidation of VLDL particles [49]. In this case, activation of IRE1–XBP1 would prevent, rather than promote, hepatic steatosis.

Although at present there remains some uncertainty about the importance of ER stress in comparison to other mechanisms of obesity-related hepatic steatosis (Table 4.1), there is ample experimental evidence that ER stress regulates hepatic lipid homeostasis at multiple levels through multiple signal transducers. ER stress signaling can also lead to hepatocyte death in a fatty liver; this will be discussed below.

Autophagy is a process that degrades dysfunctional cellular constituents and maintains cellular energy stores. During autophagy, material is captured in a membrane-bound vesicle called an autophagosome, which then fuses with a lysosome permitting digestion of the contents and releases back into the cytosol. Recently, it was discovered that autophagy plays an important role in lipid homeostasis in adipocytes [52] and hepatocytes [53]. Autophagy is considered particularly important to hepatocytes as a means of accessing the energy stored in cellular lipid droplets, because these cells are not equipped with high concentrations of lipases [54]. Fatty acids mobilized from lipid droplets via autophagy are routed to mitochondria for oxidation. The importance of autophagy to normal hepatic lipid metabolism was documented in experiments demonstrating that inhibition of autophagy leads to spontaneous hepatic steatosis [53]. More importantly, and directly pertinent to fatty liver disease, it was discovered that loading normal liver cells with lipid impairs autophagy [53]. This prompted further investigation to determine whether autophagy is dysregulated in a fatty liver in vivo, and the suspicion was confirmed [53]. At present, it does not appear that abnormal autophagy is a primary cause of hepatic steatosis, but impairments in autophagy that result from steatosis can enhance or perpetuate hepatic lipid accumulation. Dysregulated autophagy can also promote cell death in a fatty liver; this is discussed later (see section “Mechanisms of Cell Death in NASH”).

Bile acids exert important influences on lipid metabolism in the liver [55]. The principal means by which bile acids control lipid homeostasis is through interactions with the farnesoid X receptor (FXR) (Fig. 4.5). Bile acid-induced stimulation of FXR leads to activation of the downstream target, short heterodimer partner (SHP); SHP, in turn, suppresses SREBP1, which results in inhibition of hepatic lipogenesis. FXR also activates PPARα, a master inducer of fatty acid oxidation. This means that bile acid signaling should limit hepatic steatosis by two complementary means. The tonic influence of bile acids on hepatic lipid homeostasis has been demonstrated in mouse studies, which reveal that knockout of FXR leads to hepatic steatosis and steatohepatitis [56]. On the other hand, there is only limited evidence that an alteration in FXR is a root cause of fatty liver disease in humans. One study reports that patients with hepatic steatosis and NASH have low levels of FXR in the liver [57]. Another looked for modulation of hepatic lipid levels as a consequence of a polymorphism in the gene encoding FXR, but found none [58]. Recently, however, it was discovered that high-fat feeding in mice induces a gene named yin yang 1 (YY1) that suppresses FXR [59]. YY1 is upregulated by TNF [60], and thus, the pro-inflammatory state of obesity may be responsible for YY1 induction and FXR suppression.

Even if fatty liver disease is not attributable to a defect in FXR signaling, FXR agonism improves hepatic steatosis [61–64]. The therapeutic potential of synthetic bile acids for fatty liver disease is discussed in Chap. 9B.

There is a heritable element to NAFLD and NASH, based on evidence that the disease clusters in families, is concordant among monozygotic twins, and exhibits clear ethnic variations in disease risk (reviewed in [65]). The genetics of NAFLD and NASH are discussed in detail in this chapter. Because the current chapter focuses on the pathogenesis of NAFLD and NASH, it is worth highlighting a few examples of NASH-related gene polymorphisms here, focusing on the mechanism by which they might contribute to disease pathogenesis. The most notable genetic alteration that has been linked to NASH is a C → G polymorphism leading to an I148M substitution in the gene PNPLA3 (patatin-like phospholipase domain-containing 3) [66–68]. PNPLA3 encodes adiponutrin, a protein located in the membrane fraction of hepatocytes and adipocytes. Investigators have gone to great lengths to elucidate how mutant adiponutrin promotes steatosis. In one study, mice were engineered to overexpress human I148M adiponutrin. They exhibited enhanced hepatic lipogenesis, which led to the conclusion that adiponutrin functions primarily as lysophosphatidic acid acyltransferase (LPAAT) that generates lysophosphatidic acid in the pathway leading to triacylglycerols [69]. However, in a more recent study, the I148M mutation was “knocked into” the native mouse gene. Using this approach, the mice displayed impaired triglyceride lipolysis, prompting the opposite conclusion that adiponutrin is primarily a triglyceride lipase [70]. Although more research is required to clarify the precise role of mutant adiponutrin in fatty liver disease, some observations are worth noting. First, the I148M knock-in mutation does not cause spontaneous hepatic steatosis in mice, but does enhance steatosis in response to high-carbohydrate feeding [70]. This is consistent with the classification of PNPLA3 as a modifier of NASH in humans [71]. Second, although overexpression or knock-in of mutant adiponutrin induces hepatic steatosis in mice, as yet there is no report of its ability to induce experimental steatohepatitis.

A gene named transmembrane 6 superfamily member 2 (TM6SF2) was recently discovered in an exome-wide association study to be associated with hepatic steatosis [72]. The link between TM6SF2 and NAFLD is believed to supersede that of a nearby gene, neurocan (NCAN), which was uncovered in an earlier GWAS study [67]. Hepatic steatosis is associated with an A → G polymorphism in TM6SF2 leading to a glutamate-to-lysine substitution at amino acid 167. The TM6SF2 polymorphism has already been confirmed by other groups as relevant not only to steatosis [73] but also NASH and cirrhosis [74]. TM6SF2 is a transmembrane protein of unknown function, but in vitro studies indicate the E167K variant has reduced stability. Knockdown of TM6SF2 in liver cell lines and mouse liver in vivo results in steatosis, particularly following a carbohydrate challenge. Importantly, TM6SF2-deficient mice have reduced levels of circulating lipids, which suggests the protein facilitates hepatic VLDL secretion [72]. These examples illustrate how individual gene polymorphisms, each with a strong association to NAFLD but an independent effect on the liver, can impact the NAFLD phenotype. They also underscore how multiple independent modifiers in an individual subject might complicate disease severity and response to treatment.

It is now evident that microbial communities in the intestine are important determinants of metabolism. Pertinent to obesity are observations that diet affects the gut microbiome and the gut microbiome in turn affects nutrient absorption. Specifically, overeating stimulates the intestinal overgrowth of bacteria belonging to the phylum Firmicutes [75, 76], which are efficient at extracting nutrients from the diet [77]. Some of the adverse effects of round-the-clock eating may also stem from alterations in the gut microbiome [78]. Thus, microbial derangements induced by alterations in eating behavior create a situation that enhances the potential for obesity and its complications.

In addition to having an altered colonic microbiome, overweight individuals with fatty livers often display small intestinal bacterial overgrowth (SIBO) [79, 80]. SIBO is thought to arise from impaired intestinal motility in obesity, which can result from a reduction in the hormone ghrelin [81]. Toxins or inflammatory mediators elaborated by these abundant organisms can compromise intestinal barrier function by degrading tight junction proteins [80, 82]. This allows potentially harmful compounds from the intestine to invade the portal circulation and reach the liver, where they can stimulate NAFLD and NASH.

Two intestinal compounds that have been implicated in the development of NAFLD and NASH are bacterial endotoxin and ethanol. Endotoxin , which can enter the portal circulation through a compromised intestinal barrier, is readily detectable in the blood of patients with hepatic steatosis and NASH [80, 83]. Importantly, endotoxin has even been identified in the blood of healthy persons fed a high-energy diet for a brief period [82, 84]. This suggests that alterations in the intestine can occur very early in the evolution of obesity and NAFLD. Ethanol, a by-product of bacterial fermentation in the gut, can be absorbed through an intact intestinal epithelium. Like endotoxin, ethanol is detectable in the circulation of patients with NAFLD [85, 86] and has the potential to promote hepatic steatosis or steatohepatitis (see Chap. 9A on Alcoholic Liver Disease).

Intestinal bacteria can also perform other functions that lead to undesirable effects on hepatic lipid homeostasis. One is the ability to convert choline into methylamines. This property, which is characteristic of microbes that overgrow in obesity [87], can lead to choline depletion in the intestine and choline deficiency in the host. Choline deficiency is well known to promote hepatic steatosis [88]; it does so by limiting the synthesis of phosphatidylcholine, critical to the assembly of VLDL particles [89]. Intestinal bacteria also suppress the synthesis of fasting-induced adipocyte factor (FIAF), an inhibitor of lipoprotein lipase [90, 91]. When FIAF is reduced, adipose tissue lipolysis is increased, which results in an increase in fatty acid delivery to the liver.

Lastly, gut microbes can influence metabolism through effects on bile acids. Gut bacteria can modify bile acid structure and activity; conversely, bile acids can influence microbial homeostasis in the gut. The interplay between microbes and bile acids, as well as other topics relevant to the gut microbiome in fatty liver disease, is reviewed in [92–94].

The term NASH, in contrast to NAFLD, connotes the presence of hepatic steatosis with accompanying liver cell injury and hepatic inflammation. Depending on the severity of disease, NASH can also feature hepatic fibrosis and even liver cancer. Hepatic steatosis is at the heart of both NAFLD and NASH; consequently, there is a tendency to assume these two entities form a continuum in which disease progresses from steatosis to steatohepatitis. Indeed, this is the basis for the “two-hit” hypothesis, in which the first hit to the liver is fat accumulation and a second hit, which can take many forms, triggers liver damage [122]. Although the two-hit concept has validity, there is also reason to consider that NASH represents the consequence of multiple parallel hits to the liver [123]. In this paradigm, NAFLD does not inevitably become NASH, and NASH may not be preceded by the more benign NAFLD. The discussion that follows does not specifically enlighten this debate but does address the cellular mechanisms of liver injury in NASH. The focus is on how liver cells are damaged by lipid accumulation and how resident liver cells and recruited inflammatory cells react to this damage to stimulate inflammation, fibrosis, and cancer.

As knowledge about cell biology advances, so does our concept of the pathogenesis of fatty liver disease. This is immediately evident in the fact that we now include ER stress, autophagy, and inflammasome activation among the pathways contributing to hepatic steatosis and NASH. These new paradigms are not replacing old theories; rather, they are putting them into new perspective and in many cases permitting consolidation of seemingly unrelated events into single pathogenic schemes (e.g., consider fatty acids, dead cells, and endotoxin all as inducers of the inflammasome in a fatty liver). Although for simplicity, specific disease mechanisms were often addressed individually in this chapter, they frequently intersect. For example, there is crosstalk between ER stress and autophagy [259], and JNK activation can result from numerous insults including ER stress, death receptor signaling, and ROS induction. This underscores the true complexity of disease pathogenesis in a fatty liver.

With regard to the pathophysiology of NASH and its complications, one theme that emerges from the above discussion is the central role of hepatocyte death in the process. Indeed, hepatocyte death triggers not only inflammation but also fibrosis and even cancer. With this in mind, it remains a conundrum why some individuals at risk for fatty liver disease develop only hepatic steatosis, whereas others develop NASH. Further exploration of genetic and behavioral modifiers is likely to shed light on this issue.