Potential conflict of interest: Nothing to report.
During liver regeneration after partial hepatectomy, normally quiescent hepatocytes undergo one or two rounds of replication to restore the liver mass by a process of compensatory hyperplasia. A large number of genes are involved in liver regeneration, but the essential circuitry required for the process may be categorized into three networks: cytokine, growth factor and metabolic. There is much redundancy within each network, and intricate interactions exist between them. Thus, loss of function from a single gene rarely leads to complete blockage of liver regeneration. The innate immune system plays an important role in the initiation of liver regeneration after partial hepatectomy, and new cytokines and receptors that participate in initiation mechanisms have been identified. Hepatocytes primed by these agents readily respond to growth factors and enter the cell cycle. Presumably, the increased metabolic demands placed on hepatocytes of the regenerating liver are linked to the machinery needed for hepatocyte replication, and may function as a sensor that calibrates the regenerative response according to body demands. In contrast to the regenerative process after partial hepatectomy, which is driven by the replication of existing hepatocytes, liver repopulation after acute liver failure depends on the differentiation of progenitor cells. Such cells are also present in chronic liver diseases, but their contribution to the production of hepatocytes in those conditions is unknown. Most of the new knowledge about the molecular and cellular mechanisms of liver regeneration is both conceptually important and directly relevant to clinical problems. (Hepatology 2006;43:S45–S53.)
The evolution of ideas pertaining to the mechanisms of liver regeneration may be categorized into three phases: (1) the original view that a single humoral agent could function as a key, capable of unlocking all of the events required for liver regeneration; (2) the idea that the activation of one pathway involving multiple components could be responsible for regeneration; and (3) the more recent idea that the activity of multiple pathways is required for liver regeneration.1–3 This last formulation is still an oversimplification of a more complex reality, as liver regeneration does require the activation of multiple pathways, but these pathways do not act independently of each other. The patterns of interaction between pathways are particularly complex because they may involve simultaneous and/or sequential modes of operation, may occur in different liver cell types, and may be present only at certain stages of liver regeneration.
The recent literature is replete with data showing that the ablation of genes involved in different pathways can inhibit liver regeneration, leading to the notion that this process requires the activation of dozens of different pathways. An alternative view, which we propose, is that the essential circuitry required for liver regeneration is encompassed by three types of pathways: cytokine, growth factor, and metabolic networks that link liver function with cell growth and proliferation (Fig. 1). A characteristic feature of these networks is that redundancy exists among the intracellular components of each network, such that loss of an individual gene rarely leads to complete inhibition of liver regeneration. Instead, a change in the timing of hepatocyte DNA replication or mortality in only a fraction of the animals carrying the defect is typically seen. No single genetically modified mouse model demonstrates 100% mortality and a complete blockage of both DNA replication and cell proliferation after two-thirds partial hepatectomy (PH). Thus, using criteria established by genetic studies in other organisms, no single gene can be considered “essential” for liver regeneration.
Most of the data that we discuss in this review are based on studies of PH in genetically modified mice. It should be noted that the original technique of Higgins and Anderson for PH in rats4 must be modified to be safely and reproducibly performed in mice.5 Ligating both the left and median lobes together (as done in rats) causes necrosis in the remaining right lobe in a mouse, presumably from vascular obstruction. Thus, data reporting the effect of a gene on mouse mortality after PH need to be carefully interpreted.
Before discussing cytokine, growth factor, and metabolic pathways active during liver regeneration, we comment briefly on the definition of the term regeneration as it applies to liver growth processes, about the cell types responsible for this growth, and how synchronous hepatocyte division is coordinated after PH.
Regeneration Is Compensatory Growth of the Liver
In biological terms, regeneration means the reconstitution of a structure that has been excised, such as the complete re-growth of the limb of a newt, including skin, muscle, and digits. Regeneration of a lost limb starts with the formation of a blastema at the cut surface, which contains dedifferentiated cells with broad differentiation potential.6 Liver “regeneration” after PH is a very different process, in which the excised parts do not grow back. Rather, the remaining liver expands in mass to compensate for lost tissue. Thus, liver regeneration is technically a process of compensatory growth rather than regeneration. As such, it does not follow the same general steps involved in true regenerative processes, and formation of a blastema containing dedifferentiated cells does not occur.
An important distinction must be made regarding the origin of the cells that replace missing hepatocytes after PH and in the growth processes that follow parenchymal cell necrosis.7, 8 After PH or CCl4-induced injury, liver mass is replenished by replication of existing hepatocytes, without activation of a progenitor cell compartment.9 In the regeneration of the liver that follows loss of parenchymal cells induced by other toxins, such as galactosamine, replication and differentiation of intrahepatic progenitor cells occurs.10–13 The extent to which these cells contribute to regeneration varies according to the nature of the injury, doses of inducing agents, or other experimental variables. It is important to mention that we do not find any compelling evidence that bone marrow cells generate significant numbers of hepatocytes in in vivo physiological or pathological hepatic growth processes.8 Conversely, cells originating in the bone marrow may generate 20% or more of the endothelial cells and other nonparenchymal cells (NPCs) during liver regeneration.14
The extent and timing of liver regeneration are known to vary according to circadian rhythms15; a recent study has identified a mechanism by which these rhythms control hepatocyte proliferation after PH.16, 17 In these experiments, the peak of DNA replication after PH in mice always occurred 36 hours after the operation, regardless of the time of the day at which the procedure was performed. The entry of cells that had replicated their DNA (G2 cells) into mitosis, however, always occurred at the same time of day. This finding suggests that a circadian clock controls the G2/M transition; the authors implicate WEE1 kinase as a candidate circadian regulator of cell division. WEE1 phosphorylates Cdc2 kinase, disrupting the activity of the Cdc2/cyclin B1 complex, which participates in hepatocyte mitosis. Expression of Wee1 follows a circadian cycle, and entry of G2 cells into mitosis varies inversely with levels of WEE1 after PH. This study is consistent with studies in Forkhead boxM 1b (Foxm1b) knockout (KO) mice, which display a deficit in DNA replication after PH. FoxM1b regulates entry into M-phase by coordinating induction of cyclin B1 and activation of cdc25b to dephosphorylate Cdc2.18
A fascinating counterpart to these observations is that the timing of DNA replication, which is not under the control of circadian rhythms, appears to be an intrinsic property of hepatocytes. Rats and mice differ in the timing of DNA replication after PH, which is 12 to 16 hours earlier in rats. Weglarz and Sandgren transplanted rat hepatocytes into the livers of mice after PH and found that rat hepatocytes replicated earlier than mouse hepatocytes in the resultant chimeric liver.19 These results indicate that the timing of hepatocyte DNA replication after PH is an autonomous process, primarily guided by intrinsic signals.
The Cytokine Network and the Initiation of Liver Regeneration
A wide variety of genes are differentially expressed during the first few hours after PH (the “priming phase”); many of these genes are involved in a cytokine network. Evidence for the importance of cytokines during this phase of regeneration includes (1) increases in liver mRNA and serum levels of tumor necrosis factor (TNF) and interleukin-6 (IL-6) after PH20–22; (2) activation of the transcription factors nuclear factor-kappa B (NF-κB) and signal transducer and activator of transcription 3 (STAT3)23, 24; (3) inhibition of DNA replication by anti-TNF antibodies20; (4) blockage of liver regeneration in Il-6 and TNF receptor type I (Tnfr1) KO mice25, 26; and (5) correction of the defect in Tnfr1 KO mice by IL-6 injection.26 The cytokine network is initiated through the binding of TNF to TNFR1, leading to activation of NF-κB in NPCs, production of IL-6, and activation of STAT3 in hepatocytes (Fig. 2). One important STAT3 target gene is Socs3, which acts in a feedback loop to prevent ongoing activation of IL-6 signaling by inhibiting STAT3 phosphorylation.27
Further evidence that cytokines are important for regeneration arises from the fact that certain cytokines have the ability to prime resting hepatocytes for cell division without PH. Hepatocytes in the normal liver are quiescent (G0 phase) and exhibit only a minimal response to potent in vitro mitogens, such as transforming growth factor alpha (TGFα), epidermal growth factor (EGF), and hepatocyte growth factor (HGF). However, growth factor infusion into rats preceded by a single TNF injection induces replication in up to 40% of hepatocytes in the normal liver.28
Although there is no dispute that a cytokine network including the components mentioned above is activated within 30 minutes after PH; there is much debate regarding the precise roles played by individual cytokines. The increase of serum TNF after PH has not been universally observed, although it appears to be higher in rats than in mice, and although TnfrI KO mice have multiple deficits after PH, Tnf KO mice appear to regenerate normally.29, 30 These data suggest that TNF itself may not be required, because other ligands can signal through TNFR1, such as lymphotoxin alpha. Indeed, Knight and Yeoh31 showed that hepatocyte DNA replication after PH is inhibited in Ltα/Tnf double KO mice.
The precise role of IL-6 in liver regeneration has been particularly difficult to define. It has been calculated that almost 40% of the immediate early genes expressed in the regenerating liver32, 33 may be IL-6 dependent,34 suggesting that the role of IL-6 in this process is complex. The primary function of IL-6 in regeneration was originally shown to be proliferative, as Il-6 KO mice had a striking deficit in DNA replication after PH, but subsequent data have suggested that IL-6 also has anti-apoptotic and hepatocyte survival activity.35–39 The proliferative effects of IL-6 have undergone further scrutiny, as some groups have reported that IL-6 KO mice and mice deficient in gp130 (a necessary component of the receptor complex for the IL-6 ligand family) have no defects in DNA replication after PH,39 in direct contrast to the original report.25 These discrepancies may reflect a lack of standardization of experimental conditions, including surgical techniques, anesthetic agents, strain of mice used, or animal maintenance. In addition, it appears that the precise level of IL-6 present after PH may be critical in determining its effects.35, 40
Stem cell factor (SCF) and oncostatin M (OSM) are 2 molecules that may modulate or enhance the effects of IL-6 during liver regeneration. SCF restores DNA replication in Il-6 KO mice after PH,41 and administration of OSM can correct the deficient regeneration seen in Il-6 knockout mice after CCl4-induced injury.42 Conversely, IL-6 cannot restore the defective regeneration after CCl4that is seen in mice deficient for the OSM receptor. The effects of these cytokines are at least in part redundant, as IL-6, SCF, and OSM can all activate STAT3 in hepatocytes, but their intracellular signaling pathways must diverge at some point to explain their apparent differences in biological activity.
Triggering the Cytokine Cascade: Role of Components of the Innate Immune System
Because cytokine activation participates in the initiation of liver regeneration, identifying the mechanisms that trigger the activation of this network is important. A logical candidate for a master upstream molecule is lipopolysaccharide (LPS), which is released from enteric bacteria into the portal circulation.43 Indeed, Cornell et al.44found that rats with restricted production of LPS and mice that are naturally hypo-responsive to LPS (C3H/HeJ mice) have a delay in regeneration after PH. The LPS resistance of C3H/HeJ mice45 was later found to be the consequence of a point mutation in the gene for Toll-like receptor 4 (TLR4), a member of a class of receptors that bind various microbial products. LPS binding to TLR4 activates multiple intracellular signaling pathways, some of which are dependent on myeloid differentiation factor 88 (MyD88), an adapter protein that mediates intracellular signals from several TLRs.46 Contrary to the hypothesis that LPS is the key initiator of regeneration and the findings in C3H/HeJ mice, Tlr4 KO mice showed no abnormalities after PH. Mice deficient for TLR2, TLR9, or CD14 also had normal cytokine activation and regeneration after PH. By contrast, Myd88 KO mice failed to activate TNF and IL-6.47, 48 STAT3 activation and expression of important STAT3 target genes, such as Socs3 and acute phase response genes, were also blocked in Myd88 KO mice after PH. Identifying the ligand and receptor that signal through MyD88 early after PH is an exciting challenge, and perhaps in doing so the mechanisms that initiate the liver regeneration cytokine cascade will be identified.
Other components of the innate immune system appear to be critical for normal regeneration as well; mice deficient in the C3 and C5 components of complement display significant deficits after PH.49 In these animals, diminished activation of the cytokine pathway is manifested by lack of increases in TNF and IL-6 levels, and in impaired NF-κB and STAT3 activity. Whether and how these two aspects of innate immunity, TLR-MyD88 signaling and the complement cascade, converge to initiate cytokine signaling in liver regeneration is not clear.
Growth Factors and Cell Cycle Progression
The cytokine network acts at the priming phase of liver regeneration, which corresponds to the passage of quiescent hepatocytes into the cell cycle (G0to G1). Cell cycle progression is then driven by growth factors, which override a restriction point in late G1. Passage from G1to S phase is associated with Rb phosphorylation, increased expression of the Rb family member p107 and of cyclins D, E, and A, and formation of cdk4/cyclin D and cdk2/cyclin E complexes.50–52
HGF and the EGF receptor (EGFR) ligand family are important growth factors that drive cell cycle progression during liver regeneration.53, 54 HGF is produced by mesenchymal cells and acts on hepatocytes in a paracrine or endocrine fashion. Its effects are multiple and have been grouped into morphogenic, motogenic, and mitogenic categories. Studies of liver regeneration in mice with hepatocyte-specific deletion of c-met, the gene for the HGF receptor, were recently conducted.55, 56 Borowiak et al.56 demonstrated that HGF/c-met signaling is essential for cell cycle entry after PH, and that it is responsible for the activation of extracellular signal–regulated kinase 1/2 (ERK1/2). In contrast, Huh et al.55 reported that hepatocyte c-met–deficient mice had massive mortality after PH, and thus examined the role of this pathway in other liver injury models. They conclude that HGF/c-met signaling is important in hepatoprotection from apoptosis, and in facilitating healing after CCl4 administration. The discrepancy in post-operative survival between the two reports is most likely related to the different surgical techniques used by the two groups, as noted by Borowiak et al.56 Until additional data become available, deciding whether HGF/c-met signaling functions primarily in mitogenesis, or whether it maintains hepatocyte homeostasis and thus facilitates cell replication, is not possible.
The family of ligands that bind the EGFR, in addition to EGF, includes TGFα, heparin-binding EGF-like growth factor (HB-EGF), and amphiregulin (AR). TGFα is an autocrine growth factor, both produced by and active on hepatocytes.57 Although TGFα has effects on cell motility and vascularization, its main effect is the stimulation of cell proliferation. Transgenic mice that overexpress TGFα display constitutive hepatocyte proliferation and eventually develop cancer.58Tgfα expression increases after PH in wild-type mice, but Tgfα KO mice have no defects in liver regeneration.59 The normal regeneration seen in these animals is likely a consequence of compensation by other EGFR ligands, although the roles of these growth factors after PH are not entirely redundant, as discussed below.
HB-EGF is expressed earlier than HGF and TGFα after PH and appears to have a unique role in liver regeneration.60, 61 A 30% PH does not result in coordinated DNA replication, despite activation of the cytokine cascade.62, 63 A single injection of HB-EGF 24 hours after 30% PH can override this blockage between priming and cell cycle progression, eliciting a wave of DNA replication. Interestingly, this effect cannot be accomplished by similarly injecting HGF or TGFα.63 In addition, Hb-egf KO mice have a delay in DNA replication after 70% PH, although this deficiency is partially compensated by an earlier increase in Tgfα expression in these animals.
AR also appears to contribute to regeneration, as mice deficient for this growth factor have a significant deficit in DNA replication after PH.54, 64 Direct comparisons between AR and other EGFR ligands have not been made, but it is likely that the different growth factors have independent but partially overlapping functions in liver regeneration. The complexity of EGFR signaling after PH is most likely attributable to the number of EGF ligands involved, their specificity for different receptor heterodimers, and the nuances of subsequent activation of intracellular signaling pathways.54, 65
Both c-met and the EGFR are receptor tyrosine kinases, which recruit enzymes and scaffolding proteins to phosphorylated intracellular domains of each receptor. Multiple intracellular signaling pathways are thus activated, which regulate a multitude of transcription factors, initiate translation, and regulate metabolic pathways (Fig. 3). One mitogenic signal transduction pathway that is of particular interest, because it may integrate cytokine signals as well as growth factor signals, is the Ras-Raf-MEK cascade, which results in the activation of ERK1/2. ERK1/2 activation is correlated with hepatocyte DNA replication in vivo and hepatocyte proliferation in vitro.66–69 Moreover, growth factors such as HGF and TGFα and cytokines such as TNF and IL-6 stimulate ERK1/2 activity in primary hepatocytes and hepatocyte cell lines.70–72
A family of nuclear receptor ligands, including thyroxine, 1,2-bis [2-(3.5-dichloropyridyloxy)] benzene (TCPOBOP), and retinoic acid, appears to be primary hepatocyte mitogens in vivo, capable of inducing replication without tissue loss. TCPOBOP, which binds the constitutive androstane receptor (CAR), in particular has been shown to stimulate hepatocyte DNA synthesis independent of many molecules that are critical to regeneration after PH, including TNF, IL-6, NF-κB, STAT3, and cyclin D1.73, 74 However, the pathways through which CAR activation directs hepatocyte replication are yet undefined.
Metabolic Pathways and Liver Regeneration
Liver regeneration after PH is a perfectly calibrated response whose apparent sensor is the body's requirement for liver function. Identifying molecular mechanisms that account for the capacity of the liver to modulate its growth in accordance to the needs of the whole organism has been difficult. Nevertheless, the increased metabolic demands imposed on the liver remnant after PH are likely connected with activation of the machinery directly involved in DNA replication.
The administration of an amino acid mixture to intact rats induces a wave of hepatocyte replication, and protein deprivation blocks liver regeneration after PH.75, 76 Recent studies have shown that amino acids regulate hepatocyte proliferation through cyclin D1 expression.77 The initiation of protein translation is a critical control point that may integrate nutrient and energy levels with mitogenic signals.78 After PH, the activity of p70 S6 kinase increases, and the activation 4E-BP1 (a translational repressor) decreases, leading to an increase in translation. Both of these proteins are thought to be downstream effectors of mTOR (mammalian target of rapamycin), which is part of a complex that senses nutrient or energy status, and also integrates growth factor signals, resulting in the regulation of protein translation and cell growth.79, 80 The importance of translation in liver regeneration has been illustrated by a study of PH in S6 KO mice, in which a near complete loss of hepatocyte DNA replication was demonstrated.81 The mTOR complex may regulate liver regeneration by modulating cell growth and proliferation in response to the energy demands of the remaining liver, given that rapamycin, an inhibitor of mTOR, inhibits DNA replication after PH.81–83 If so, liver regeneration poses a challenge for this complex, because nutrient-sensing mechanisms in mammalian cells appear to modulate cell growth, depending on the availability of nutrients. After PH, the liver needs to regulate systemic nutrient homeostasis while its own cells are undergoing cell growth and proliferation.
Interactions Between Cytokines and Growth Factors During Liver Regeneration
We have thus far discussed separately the cytokine, growth factor, and metabolic pathways active after PH, but these pathways interact during different phases of liver regeneration. One important linkage between cytokines and growth factors may be the activation of matrix metalloproteinases (MMPs) by cytokines, such as TNF. The activity of several MMPs increases after PH,84, 85 and a recent study in tissue inhibitor of metalloproteinases 3 (Timp3) KO mice implicates an MMP called TGFα converting enzyme (TACE, also known as ADAM17).86 TIMP3 is thought to be a specific inhibitor of TACE, and Timp3 KO mice displayed elevated hepatic TNF protein levels and an earlier entry into S phase after PH than wild-type mice. However, 144 hours after PH, a proportion of these animals die, possibly because of loss of regulation of TNF signaling pathways.
We recently demonstrated that TNF itself can activate TACE, resulting in release of TGFα, activation of EGFR, and cell proliferation in cultured hepatocytes.70 The sequential activation of cytokine and growth factor receptors may stimulate numerous intracellular signaling pathways needed for cell survival and proliferation (Fig. 4). Because TACE has been shown to cleave the precursor forms of cytokines and many EGFR ligands,87 including HB-EGF, AR, and TGFα, further studies are needed to elucidate its role(s) in liver regeneration.
In elegant studies of the proliferation of hepatocytes co-cultured with liver epithelial cells, Serandour et al.87 showed that EGF alone does not initiate hepatocyte replication, but exposure of the system to EGF and TNF induces replication in up to 30% of the hepatocytes. The authors demonstrate that the primary effect of TNF is to activate MMPs, which subsequently degrade components of the extracellular matrix, allowing hepatocyte proliferation. This regimen of EGF and TNF was capable of inducing repeated waves of hepatocyte replication if given over multiple 10-day cycles. These remarkable results are an important extension of the above-mentioned data on the interactions between TNF and EGFR signal transduction, and provide an interesting in vitro system for studying liver regeneration.
During the last decade, a surge of interest in the mechanisms of liver regeneration has been seen, generated from both a biological science and a clinical perspective. Laboratory scientists, at long last, realized that liver regeneration constitutes a unique model to study signal transduction and cell cycle events in a synchronized manner in vivo. From a clinical perspective, understanding the mechanisms of liver regeneration is crucial for the appropriate management of and the development of new therapies for a number of important conditions, such as acute liver failure and cirrhosis. The interaction between basic biological knowledge and clinical issues is particularly close in studies dealing with liver progenitor cells and liver repopulation during liver development and hepatic disease.
The regulation of growth of liver transplants, and particularly that of the donor liver in living donor transplantation, appears to follow the same principles as those that regulate liver regeneration after PH in laboratory animals.89, 90 What is needed, however, is a more vigorous effort to apply the knowledge gained in experimental work to solve clinical problems, such as the failure of small-for-size transplants, and to obtain more rapid liver growth after transplantation. Another important issue is the replicative activity of hepatocytes during the progression of cirrhosis.91 Do hepatocytes at the late stages of cirrhosis truly exhaust their replicative capacity? If this is the case, does the “replicative senescence” of hepatocytes contradict the experimental data obtained by serial cell transplantation in mice, which demonstrated that hepatocytes are capable of at least 80 doublings92? And finally, would the restoration of proliferative activity in these hepatocytes correct functional defects, or might it lead to enhanced tumorigenesis?
An important gap in our knowledge is the lack of understanding of the factors that determine whether hepatocyte production during diverse regenerative processes originates from the replication of mature hepatocytes, as in liver regeneration after PH, or the differentiation of liver progenitor cells, as occurs after massive liver necrosis. Moreover, more knowledge needs to be gained on the role of the innate immune system and lymphoid cells in the initiation of these processes. Major advances in the genetic manipulation of mice now make possible the construction of appropriate animal models to study unresolved issues of major biological and clinical importance involving the molecular and cellular aspects of liver regeneration.