To divide or not to divide: revisiting liver regeneration
© Miyaoka and Miyajima; licensee BioMed Central Ltd. 2013
Received: 6 June 2013
Accepted: 17 June 2013
Published: 20 June 2013
Skip to main content
© Miyaoka and Miyajima; licensee BioMed Central Ltd. 2013
Received: 6 June 2013
Accepted: 17 June 2013
Published: 20 June 2013
The liver has a remarkable capacity to regenerate. Even with surgical removal (partial hepatectomy) of 70% of liver mass, the remnant tissue grows to recover the original mass and functions. Liver regeneration after partial hepatectomy has been studied extensively since the 19th century, establishing the long-standing model that hepatocytes, which account for most of the liver weight, proliferate to recover the original mass of the liver. The basis of this model is the fact that almost all hepatocytes undergo S phase, as shown by the incorporation of radioactive nucleotides during liver regeneration. However, DNA replication does not necessarily indicate the execution of cell division, and a possible change in hepatocyte size is not considered in the model. In addition, as 15–30% of hepatocytes in adult liver are binuclear, the difference in nuclear number may affect the mode of cell division during regeneration. Thus, the traditional model seems to be oversimplified. Recently, we developed new techniques to investigate the process of liver regeneration, and revealed interesting features of hepatocytes. In this review, we first provide a historical overview of how the widely accepted model of liver regeneration was established and then discuss some overlooked observations together with our recent findings. Finally, we describe the revised model and perspectives on liver regeneration research.
The mention of liver regeneration by Prometheus in Greek mythology indicates that ancient people had noticed the regenerative capacity of the liver. Additionally, descriptions of liver regeneration can be traced back to the 19th century when liver mass restoration with spontaneous healing of the scar was recognized after removal of a small portion of the liver . In the early 20th century, it became possible to remove liver lobes by ligating blood vessels to reduce damage to the remnant liver tissue after surgery. In 1931, Higgins and Anderson carefully formulated the currently used procedure for PHx . Notably, they used the term “liver restoration” instead of “liver regeneration” to distinguish clearly between compensatory hyperplasia and true tissue regeneration . Since then, liver regeneration after PHx in rodents has been studied extensively for more than 80 years. Until the 1950s, liver regeneration was analyzed at mainly the tissue or cellular level by microscopic observations [13–15]. In the 1960s, the advent of electron microscopy enabled the analysis of hepatocyte ultrastructure in liver regeneration [16–19]. Almost at the same time, the epoch-making research tool of radioactive isotopes became available for biological studies. This technology was used to show that almost all hepatocytes incorporate radioactive nucleotides during liver regeneration after 70% PHx [20–25]. This landmark observation led to the establishment of the widely accepted concept that all remnant hepatocytes actively divide to recover the original cell number and liver mass (Figure 1C). This long-standing model postulated that all hepatocytes undergo roughly one or two rounds of cell division after 70% PHx [8, 26, 27].
Since the establishment of gene targeting technology in mice in 1989 [28–30], much effort has focused on identifying the genes required for liver regeneration. Many genes have been reported to be involved in liver regeneration after PHx [e.g., β-catenin, methionine adenosyltransferase 1A (MAT1A), oncostatin M (OSM), nuclear factor (erythroid-derived 2)-like 2 (Nrf2) and c-Met] [31–35]. Most of these studies focused on the proliferation or survival of hepatocytes in accordance with the long-standing model. Although an elegant and simple model, accumulating evidence—including our recent findings—suggest that the traditional model of liver regeneration requires revision. We discuss these observations and the proposed revised model in the following sections.
It is well known that hepatocytes accumulate massive amounts of lipids and glycogen immediately after 70% PHx [49–51]. Therefore, the rapid increase in hepatocyte size is at least partly due to lipid and glycogen accumulation. However, the lipid and glycogen amount decreases to normal levels by the completion of liver regeneration, and no obvious change in hepatocyte ultrastructure is observed in regenerated liver after 70% PHx, with the exception of enlarged nuclei . Although several reports have shown that hepatocytes change the size and/or number of various organelles such as mitochondria, lysosomes, endoplasmic reticulum, and ribosomes [16–19, 52], more studies are necessary to reveal the nature of cellular changes in regeneration. Although the detailed mechanism of hypertrophy requires further study, the Akt-mammalian target of rapamycin (mTOR) signaling axis, which regulates the size of various cell types , seems to be an important pathway for hypertrophy in liver regeneration. Akt is a serine-threonine protein kinase with pleiotropic functions such as regulation of cell growth, proliferation, survival, differentiation, and cytoskeletal changes. mTOR is another serine-threonine protein kinase directly phosphorylated by Akt and plays a central role in the functions of Akt. Akt is activated immediately after 30 and 70% PHx, and forced expression of an active form of Akt in hepatocytes increases their size [ and our unpublished data]. Another potential key player is c-Myc, which is a transcription factor involved in cell growth and cell cycle progression. Overexpression or deletion of c-Myc in hepatocytes increases or decreases their size, respectively [54–56]. Interestingly, both Akt-mTOR and c-Myc pathways play critical roles in the enhancement of protein synthesis, indicating that upregulation of gross protein synthesis is one mechanism underlying the hypertrophy of hepatocytes [54, 57]. However, the upstream molecular mechanisms that sense the loss of liver tissue to activate Akt-mTOR and c-Myc pathways remain undefined (see below).
Because hepatocytes increase their size by 1.5-fold and then proliferate after 70% PHx, the 1.5-fold increase in cell size seems to be the threshold for hepatocytes to switch their response from hypertrophy to proliferation. As discussed above, this 1.5-fold size increase is sufficient to restore a 30% loss in liver mass, and hepatocytes do not proliferate after 30% PHx. Therefore, it would be interesting to determine precisely how much liver mass must be removed to induce proliferation. The molecular trigger for hepatocyte proliferation in liver regeneration is unknown. One possible explanation is that the size of the hepatocyte itself is the sensor to drive its cell division cycle, which is considered a general mechanism for activating cell division . Further studies are required to address this question.
Polyploidy is a characteristic feature in mammalian hepatocytes, and about 70% of adult hepatocytes in rodents are tetraploid . In general, polyploid cells can arise from failed cytokinesis, mitotic slippage, cell fusion or endoreplication. Polyploid hepatocytes can be either mononuclear or binuclear. Polyploidization of hepatocytes is initiated in postnatal liver growth by incomplete cytokinesis, that produces binuclear polyploid hepatocytes, endoreplication that produces mononuclear polyploid hepatocytes, or both that produce binuclear polyploid hepatocytes [37, 59, 60]. Insulin signaling has been implicated in the polyploidization and binucleation at the weaning stage, as discussed below [37, 60, 61].
While it has long been known that ploidy of hepatocytes increases after PHx [14, 36, 37, 62], its mechanism remains unknown. Although a majority of hepatocytes undergo S phase in regenerating liver after 70% PHx, not all hepatocytes undergo cell division, resulting in an increase in ploidy. We noticed that the ratio of hepatocytes that were positive for phosphorylated histone H3 (an M phase marker) to those that were positive for Ki67 (a G1 to M phase marker) in liver regeneration is much lower than that in postnatal liver, where hepatocytes actively undergo cell division . These results suggest that M phase progression is compromised in liver regeneration.
M phase promoting factor (MPF), composed of cyclin B and Cdk1, regulates entry into M phase [63, 64], and MPF must be activated for cell transition from G2 to M phase. Cdk1 is phosphorylated at three amino acid residues (Thr-14, Tyr-15, and Thr-161) in the inactive form of MPF, whereas Cdk1 dephosphorylation at Thr-14 and Tyr-15 by cell division cycle 25 (Cdc25) activates MPF . We found that the phosphorylation level of Cdk1 at Tyr-15 was much higher in regenerating liver compared to postnatal liver. Therefore, the lower activity of MPF in regenerating liver could be a cause of the infrequent entry into M phase. In fact, MPF activity is dispensable for liver regeneration; hepatocytes increase their size to regenerate liver after 70% PHx without cell division, even in the absence of Cdk1 .
In contrast to the G2 to M phase transition, the G1 to S phase transition is driven mainly by cyclin D/A2 and Cdk2 in normal cell division and endoreplication . Therefore, it is intriguing to compare the activity of Cdk2 in liver regeneration with that in liver development. As hepatocytes increase their ploidy in both postnatal liver development and regeneration, the cell cycle regulators driving the G1 to S phase transition seem to dominate those driving the G2 to M phase transition in mature hepatocytes. Although the exact molecular mechanism that blocks the entry of hepatocytes into M phase in regenerating liver remains elusive, a crucial role of the E2F family transcription factors has been reported recently [67, 68]. The E2F family consists of E2F transcription activators and transcription repressors and regulates cell cycle progression. Using mouse genetic models, these studies clearly showed that atypical E2F repressors E2F7 and E2F8 inhibit the completion of cell division to enhance polyploidy and binucleation in hepatocytes both in liver development and regeneration, whereas the canonical activator E2F1 counteracts their activities. These E2Fs differentially control the transcription of cell cycle regulators to either enhance or inhibit the G2 to M phase transition.
Binucleation is another interesting feature in adult hepatocytes that begins from the neonatal liver . It has long been known that the number of binuclear hepatocytes decreases during liver regeneration after 70% PHx, as assessed by microscopic observations and manual counting [13–15, 69–71]. Weaning increases the amount of circulating insulin to activate Akt signaling, which induces incomplete cytokinesis to generate binuclear hepatocytes during liver maturation [37, 60, 61]. In contrast, even though Akt is activated by PHx, the number of binuclear hepatocytes decreases in regenerating liver, suggesting that Akt has different functions in liver maturation and regeneration. Indeed, Akt signaling induces hypertrophy of hepatocytes in liver regeneration (our unpublished data). The different responses to Akt may be due to the molecular targets of the Akt signaling pathway differing according to the cellular context. mTOR is a major downstream molecule of Akt that functions in induction of hypertrophy. Because the E2F family transcription factors regulate the progression of M phase [67, 68], it is tempting to speculate a link between Akt signaling and E2Fs in binucleation. Indeed, it is already known that this link exits in other cell types [72, 73].
Binucleation is generally considered a sign of terminal differentiation in both hepatocytes and cardiomyocytes [77, 78]; however, in contrast to binuclear hepatocytes, binuclear cardiomyocytes do not divide. Mononuclear and binuclear cardiomyocytes seem to have distinct functions, and it has been proposed that only mononuclear cardiomyocytes maintain their proliferative potential to serve as stem or progenitor cells in heart muscle [79, 80]. It is unclear whether hepatocytes with different numbers of nuclei have different functions. This issue is discussed further below.
Cellular properties of hepatocytes in regenerated liver after 30 and 70% PHx, compared to normal liver
Adult hepatocytes can be binuclear, polyploid and even aneuploid under normal conditions [76, 81]. Furthermore, cell number, cell size, nuclear number and ploidy of hepatocytes are significantly different in normal liver and regenerated liver after 30% or 70% PHx (Table 1). Despite these differences, liver seems to function almost equally in different conditions, which raises an intriguing question whether such differences in the cellular properties affect hepatocytes. One study using transcriptomic analysis showed that hepatocytes with different ploidy were basically indistinguishable ; however, another study indicated that polyploid cells were more resistant to stressful conditions . Hepatocytes with different ploidy were shown to be equally susceptible to interferon-γ (IFN-γ)-induced apoptosis . Proliferation of polyploid hepatocytes was compromised and they exhibited more characteristics of senescence . No consensus has been reached on the functional differences in hepatocytes of different ploidy or number of nuclei. The volume of hepatocytes is basically proportional to their ploidy, which is often the case with other cell types [42, 84, 86, 87]. However, we noticed that hepatocytes increase their size without increasing their DNA content after 30% PHx , suggesting that hypertrophy without increased ploidy allows hepatocytes to function properly. Naturally occurring aneuploidy is another feature of hepatocytes, which seems to arise from inaccurate chromosome segregation [76, 81]. Aneuploidy is often associated with genetic disorders and is observed in various cancers [88, 89]. However, aneuploidy does not seem to be tumorigenic in hepatocytes, and it may even provide genetic diversity in hepatocytes to perform different functions .
In addition to these characteristic features exhibited by wild-type mice, genetically modified mice show rather extreme phenotypes of hepatocytes. Mutant mice with impaired cell cycle progression showed a fully functional liver with extraordinarily enlarged hepatocytes after 70% PHx [44–47]. A loss of E2F7 and E2F8 reduced ploidy or nuclear number but did not affect hepatocyte function and regeneration after several liver injuries including PHx [67, 68]. Furthermore, hepatocytes were resistant to DNA damage caused by a lack of telomeric repeat binding factor 2 (TRF2). In the absence of TRF2, liver regenerated by increasing the size and ploidy of hepatocytes and was fully functional after 70% PHx . These observations indicate collectively that hepatocytes have a “cellular robustness” which allows them to perform their functions in a variety of settings; these differ in terms of cell size, ploidy or nuclear number. The observed extreme plasticity in ploidy of hepatocytes supports their robustness . It is tempting to speculate that the cellular robustness of hepatocytes is one reason why of mammalian organs, only the liver has such a marked regenerative capacity.
Liver regeneration serves as an excellent model for regulation of organ size. Generally, differences in organ size among animals reflect differences in cell number rather than cell size [91, 92]. Limb regeneration in amphibians also depends on an increase in cell number to fully recover the original tissues. However, the size and number of hepatocytes in liver regeneration after PHx contribute differentially to the recovery of liver mass. It has been suggested that removal of one kidney induces the enlargement of the other by increasing the size of kidney cells . Moreover, physiological and pathological cardiac hyperplasia is induced by hypertrophy of cardiomyocytes [94, 95]. Therefore, cellular hypertrophy could be a general mechanism for increasing organ size. Hippo/mammalian Ste-20 like kinase (Mst)1/2-Yorkie/Yes-associated protein (YAP) signaling plays an indispensable role in the regulation of organ size [96–99]. In normal adult hepatocytes, Mst1/2 kinase (the mammalian homologue of Drosophila Hippo) phosphorylates and inactivates YAP (the mammalian homologue of Drosophila Yorkie), which is a transcription activator that induces cell proliferation and suppresses apoptosis. Transgenic expression of human YAP in mouse hepatocytes drastically increased the liver size . This regulation of organ size by Hippo/Mst1/2-Yorkie/YAP signaling might be due mainly to the control of proliferation and apoptosis in cells . However, Hippo/Mst1/2-Yorkie/YAP signaling affects cell size by tuning Akt-mTOR signaling via miRNA , demonstrating that cell size plays a critical role in organ size regulation by Hippo/Mst1/2-Yorkie/YAP signaling as well.
What senses and regulates liver size is a fundamental question. The decreased number of hepatocytes and increased size of hepatocytes in regenerated liver suggests that liver size is determined by the total mass of hepatocytes. PHx drastically changes the blood flow into the liver. This increased blood flow generates shear stress that induces nitric oxide production, triggering regeneration [102–106]. In addition, the amount of bile acid in the blood might serve as a mechanism of monitoring the size of liver because it reflects the total mass of the hepatocytes [107, 108]. Additionally, because the liver serves as a major reservoir of glycogen, the blood glucose level reflects the liver mass and so might also be a sensor. Consistent with this hypothesis, it has long been known that rodents become hypoglycemic after PHx, and supplementation of glucose inhibits liver regeneration [109–111]. This inhibitory effect of glucose is suggested to be mediated by p21 . Other factors in the regulation of liver size might be cytokines and serum proteins secreted from hepatocytes. A key contributing feature of these factors is that they must reflect the total mass of hepatocytes, but not the number or size of individual hepatocytes. Although these factors may sense the liver size, the mechanism of initiating and promoting regenerative responses remains unknown. Furthermore, liver regeneration must terminate when the liver recovers its original mass. Several molecules have been suggested to be involved in the termination of liver regeneration including transforming growth factor-β (TGF-β), a mitoinhibitory cytokine for hepatocytes ; extracellular matrix, which might inhibit proliferation of hepatocytes via integrin-linked kinase (ILK) and glypican 3 [114–116]; and peroxisome proliferator-activated receptor-γ (PPAR-γ) a mitoinhibitory transcription factor for hepatocytes . However, the termination of liver regeneration has been inadequately studied compared to the initiation process. Recent transcriptome analyses of termination may shed light on its underlying molecular mechanisms [118, 119]. A future challenge is to elucidate the molecular links between the sensors of liver size, the factors that regulate the hypertrophic and proliferative responses of hepatocytes, and the termination process of liver regeneration that acts to maintain the appropriate liver size.
Although liver regeneration has been studied extensively, many important fundamental mechanisms remain undefined such as the mechanisms of cellular hypertrophy, cell division, nuclear division, ploidy changes and organ size control. Liver regeneration after PHx provides an excellent experimental system to tackle such basic biological questions. Understanding the mechanisms underlying liver regeneration is clinically important because hepatectomy is a practical treatment for liver tumors, and liver transplantation is an important therapeutic option in patients with severe liver diseases. Understanding the mechanism of liver regeneration will lead to the development of promising therapeutic strategies.
We thank Dr. T. Itoh for his critical reading of this manuscript. YM is a recipient of a JSPS Postdoctoral Fellowship for Research Abroad and an Uehara Memorial Foundation Research Fellowship. This work was supported in part by research grants to AM from the Ministry of Education, Culture, Sports, Science and Technology of Japan, Ministry of Health, Labour and Welfare of Japan, and the CREST program from Japan Science and Technology Agency.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.