The Greek myth about Prometheus, who was chained to a rock as punishment for his attempt to steal fire from the gods, has always piqued hepatological interest. Prometheus was tormented by an eagle that returned night after night to peck away at his liver, which during the day restored itself to its previous size. A 1-day recovery time is beyond contemporary measurements of the rate of liver regeneration, but the reality of how the liver responds to removal of functional mass or to acute injury is just as remarkable as this colorful tale. Hepatocytes are normally highly differentiated, metabolically active cells existing in the resting (G0) state. For liver regeneration to occur, they must be first “primed” to acquire proliferative competence (move from G0 to G1) and then, under the influence of growth factors and other mitogens, to progress through the cell cycle and undergo one or more rounds of mitosis (G1 − M) until the mass of the liver is restored.
Liver Regeneration Is Highly Regulated
During the last 3 decades, investigators have revealed many details of that regulation1: The cytokines and transcription factors pivotal to cell cycle induction (priming), the growth factors, signaling pathways and transcription factors that activate and facilitate cell cycle progression,2, 3 and the dozens of genes that are upregulated or downregulated at various stages of this remarkable process of organ recovery.1, 4 Before considering new information that bears on the central issue of how all these changes are coordinated, one could fantasize about why the eagle kept coming back to taste the Promethean liver! For gourmands, this idea is redolent of paté de foie gras (fatted goose liver paté) — fat makes the liver tasty! But is fat good for the liver? Does it help the liver regenerate when injured or removed, as, for example, after surgical resection or transplantation using split liver grafts? To the contrary, it is well known that fatty livers are not ideal for surgical resection or transplantation,5, 6 an issue that hampers application of liver transplantation because up to 40% of donor livers are fatty. While the most likely explanation for reduced viability of steatotic donor livers is their increased susceptibility to ischemia-reperfusion (or preservation) injury,7 concern has been expressed that the fatty liver may respond suboptimally to stimuli that provoke the normal liver to undergo a regenerative response.8, 9
Liver Regeneration in Fatty Liver Disease
Some rodent models of nonalcoholic fatty liver disease (NAFLD) exhibit profound impairment of liver regeneration after partial hepatectomy (PH) or acute toxic liver injury,8, 9 whereas others do not.10–12 Closer scutiny reveals that only models associated with either leptin deficiency (ob/ob mice) or leptin receptor dysfunction (fa/fa rats) exhibit impaired liver regeneration12; nutritional models of NAFLD do not. In some NAFLD models, such as methionine and/or choline deficiency, liver regeneration may actually be enhanced in the presence of steatosis.13 Evidence that leptin itself participates in the control of liver regeneration was provided by experiments in which ob/ob mice were subjected to acute carbon tetrachloride toxicity.14 Hepatocyte proliferation was arrested in these obese mice with fatty livers, and the most “up-stream” site of regulatory impairment was found to be release of tumor necrosis factor-α (TNF). Replacement of physiological leptin levels restored hepatic regenerative capacity to the ob/ob mouse; so too did TNF.14 Thus leptin, a hormone initially characterized as an appetite regulator, is necessary for liver regeneration; this regulator could have to do with effects of leptin on the hepatic inflammatory response to injury.15 However, leptin also has important metabolic effects, including the prevention of lipid accumulation in nonadipose tissues, like the liver.16, 17
Coupling of Regeneration to a Metabolic Response in the Liver
In this issue of HEPATOLOGY, Shteyer, Rudnick, and colleagues provide new insights into metabolic factors and liver regeneration.18 In a key experiment, they administered high dose (pharmacological) leptin to mice and successfully suppressed hepatic fat synthesis mediated by the process termed adipogenesis (because the genes involved are activated in the transition of preadipocytes to mature, fat-storing adipocytes).19, 20 Despite the apparently essential need for some leptin to progress liver regeneration,14 the response here was substantial inhibition of the hepatic regenerative response to PH. In a related experiment, these investigators used a conditional knockout strategy to inhibit glucocorticoid actions in the liver, thereby also reducing hepatic adipogenesis. The purpose of these experiments was to test the hypothesis that fat synthesis may actually be part of an essential metabolic response within the liver that is coupled to the regenerative response. The origins of this hypothesis deserve further consideration. First, the incitement of liver cell proliferation, while apparently initiated by inflammatory cytokines (TNF, interleukin-6 [IL-6]), appears to be in response to loss of functional hepatic mass.1 For example, in experimental PH, removal of one third of the liver is a poor proliferative stimulus (only isolated hepatocytes proliferate) whereas two-thirds removal provokes 80% to 95% of hepatocytes to undergo coordinated rounds of mitosis. Further, both experimental and clinical studies (the latter after liver transplantation) have convincingly shown that there is an optimal liver to body mass ratio.21 The grafted liver grows or shrinks to meet this apparent metabolic need, but the “sensors” of such ideal “hepatic functional mass”, how these are detected and the response effected remain unclear. Second, altered hepatic gene regulation after PH, both immediate early (1-3 hours), and delayed early (4-12 hours) reveals up- or down-regulation of a host of “metabolic genes”, in addition to the expected expression changes of genes involved with transcriptional and cell cycle control and extracellular matrix formation.2, 4 These metabolic genes include components of the insulin signaling cascade, such as protein kinase (PK)-B/Akt (which is downregulated) and phosphoenolpyruvate carboxykinase (PEPCK) (upregulated many fold), the latter a target of leptin-mediated regulation of gluconeogenesis in the liver.7
One teliological explanation for altered regulation of metabolic pathways during liver regeneration could be that proliferating hepatocytes have a need for energy supply before the normal hepatic microcirculatory relationships are restored during the late phases of liver regeneration,1 as well as providing building blocks for construction of new liver cells. It may also be relevant that acinar zone 1 cells, which appear to be the initial site of proliferating hepatocytes during liver regeneration, prefer fat as their fuel (energy source).22 Further, there appears to be mitochondrial bioenergetic impairment during the prereplicative phase of liver regeneration, a change which has been attributed to Ca2+−induced permeability transition.23 Compensatory pathways for generation of intracellular energy sources may need to be upregulated to account for such changes.
A third reason for considering that the liver's metabolic response may be essential to the regenerative response comes from observations stretching over 65 years. Armed only with structural and ultrastructural techniques and conventional biochemistry, early hepatologs showed that the lean liver actually becomes fatty after PH, and at about the same time as the proliferative response of hepatocytes is in full train (24-72 hours, depending on species and model).24–26 If the role of fat were to inhibit liver regeneration, steatosis might more appropriately be linked to the (poorly explained) termination of liver regeneration after restoration of functional liver mass, rather than seeming to be a component of its initiation and progression.
Shteyer et al.18 have now taken the sharper tools of molecular biology to reprobe the regenerating liver, to time more precisely changes in its hepatic triglyceride content (and steatosis), and to establish the significance of such changes. Using microarray-based gene expression, they first showed that several key genes that are more typically induced during the differentiation of adipocytes (adipogenesis genes) are upregulated in mouse liver after a two-thirds PH prior to peak accumulation of lipid — that is, liver cells become more like fat-storage cells and accumulate triglycerides because of increased hepatic fatty acid synthesis.19, 20 This result contrasts with NAFLD and alcoholic liver disease in which the reasons for steatosis are more complex, involving increased hepatic uptake of fatty acids derived from the periphery, decreased “fat burning” (mitochondrial β-oxidation) and impaired triglyceride processing into very low density lipoproteins for egress from the hepatocyte. Adipogenic genes, such as adipsin, aP2 and FSP-27,19, 20 peaked at 4 to 6 hours post-PH, while hepatocellular fat accumulation was maximal at 12 to 24 hours. Several possible explanations present themselves. Developmentally, hepatocytes and adipocytes are derived from endoderm and mesoderm, respectively, so this phenomenon seems unlikely to be simple reversion to a more “primitive” phenotype of liver cell prior to, or coincident with, proliferative activity. An alternative explanation is that the transcriptional changes essential for cell cycle entry in hepatocytes also commands, as an incidental, “non-essential” metabolic response, upregulation of adipogenic genes. But nature is rarely so careless!
Is Adipogenesis of Hepatocytes Essential for Liver Regeneration?
To examine the central question of whether adipogenesis is a bystander effect or necessary for liver regeneration to occur, Shteyer et al. attempted to block upregulation of adipogenic genes, and examined the effects on the regenerative response. As mentioned earlier, pharmacological administration of leptin or liver-specific deletion of glucocorticocoid receptors both effectively combatted upregulation of adipsin, aP2 and FSP-27.18 This prevented the accumulation of lipid in hepatocytes following PH, as assessed morphologically and by determination of triglyceride. The resultant abrogation of the regenerative response was spectacular. At 48 hours, mitotic counts were reduced by 80% to 90%, and BrdU incorporation into DNA was lowered by 50% to 90%. These results are consistent with the investigators' hypothesis that conferral of adipogenesis is an important part of the proliferative response of hepatocytes during liver regeneration.
This work is elegant, with a rigorous approach to hypothesis testing by using two different types of intervention to block hepatic adipogenesis. Does a little fat indeed “fuel the fire” and provide energy for (Promethean) liver recovery, at least after surgical removal? It still remains possible that the nexus between adipogenesis and hepatocyte proliferation during liver regeneration is circumstantial; high dose leptin and glucocorticoid receptor dysruption could both block adipogenesis and cell cycle control by pathways that are “structurally” connected (via the same transcription factors) but with consequences that are not “functionally” essential. For example, high dose leptin upregulates PEPCK, thereby stimulating gluconeogenesis as well as blocking fatty acid synthesis by downregulating steroid regulatory element binding protein (SREBP)-1.17 At the same time, leptin signals through a gp130 receptor analogous to IL-6. IL-6 primes hepatocytes to enter the cell cycle,1, 27 an effect mediated by the STAT3 (signal transducer and activator of transcription), but sustained activation of STAT3 (as would result from pharmacological leptin infusion) paradoxically inhibits hepatocyte proliferation.14, 28 Likewise, glucocortcoid receptor signaling upregulates expression of adipogenic genes,18, 19 but is also pertinent to C/EBPβ accumulation,29 another transcription factor that is pivotal in the control of liver cell proliferation.3 The link between adipogenic genes and the control of liver regeneration could still be circumstantial, rather than lipid being a key precursor for hepatocyte proliferation.
On the other hand, there are at known ways in which the right fatty acids at the right time in the liver could enhance hepatocellular replicative efficiency. For example, Rudnick et al. had earlier shown that prostaglandins are required for CREBβ activation and cell proliferation during liver regeneration.30 Another connecting thread between hepatic regulation of lipid synthesis and liver cell proliferation could be the peroxisome proliferator-activated receptors (PPAR) and their heterodimeric partner in transcriptional regulation, retinoid X receptor (RXR). PPARα is an important regulator of hepatic lipid turnover in the rodent liver (less so in humans), governing genes involved in fatty acid uptake, synthesis, mitochondrial and peroxisomal β-oxidation, and ω and ω-1 oxidation of longchain fatty acids. The latter reactions are catalyzed by cytochrome P4502E1 (CYP2E1) and members of the CYP4A family; the latter are under PPARα regulatory control. Both CYPs 2E1 and 4A can generate reactive oxygen species (ROS) that spark lipid peroxidation and activate nuclear factor-kappaB (NF-κB), a transcription factor that appears to be the target of TNF and leads to IL-6 synthesis during the priming of resting hepatocytes for proliferative competence.1, 2 At 24 hours after PH, Shteyer et al. found Cyp4a14 was increased more than 8-fold,18 a time when many other P450s are suppressed.31 This could reflect activation of PPARα, although it is surprising that another PPARα-regulated gene, acyl-CoA oxidase was not up-regulated at this time.18 The delay of liver regeneration in PPARα nullizygous mice (which also exhibit steatosis) is consistent with a role for PPARα (rather than fat) in hepatic cell cycle control,32, 33 as is the well-known mitogenic effects in rodent liver of peroxisomal proliferators. It seems possible that PPARα can upregulate cyclin D1,34 thereby potentially progressing hepatocytes through G1 and into the cell cycle without the need for a TNF/NF-κB-mediated priming phase.1 However, it is also known that PP-mediated activation of PPARs is also associated with ROS-mediated NF-κB activation, at least in hepatoma cell lines.34 Other potential connections between hepatic lipid accumulation and mitogenesis of heptocytes have come from work with transgenic mice that express the hepatitis C virus core gene35–37; activation of PPARα RXR has been implicated to explain this association.
Potential Clinical Relevance
Further molecular studies using approaches to restore expression of adipogenesis genes in the models studied by Shteyer et al. could clarify whether hepatocytes really need to synthesize fat in order to proliferate. Coupling transcriptional regulation of hepatic physiology (energy and lipid metabolism) to proliferative drive makes sense, and could eventually have clinical applications, such as when it is desirable to overcome what appear to be blocks in the liver's regenerative capacity (Prometheus stoned?). Examples include massive (>80%) loss of hepatic functional mass, cirrhosis, and alcohol toxicity.1 Earlier work showing that infusion of lipid and carnitine (which may stimulate mitochondrial uptake of long chain fatty acids via carnitine palmitoyl transferase-1) following PH could enhance liver regeneration in rats gives support to this possibility.38 While too much fat (especially at the wrong time, for the wrong reasons, for too long and too rancid [oxidized]) is bad for the liver, a little bit (at the right time, and just ripe [possibly a little oxidized]) may be crucial fuel to fire hepatic recovery.
The author is grateful to Narci Teoh, M.D., Ph.D., Penny and Michael Farrell for stimulating discussion and helpful suggestions during the conceptualisation of this article. Work in the Storr Liver Unit is supported by the Storr Bequest of the Medical Foundation, The University of Sydney.