Malato Y, Naqvi S, Schurmann N, Ng R, Wang B, Zape J, et al. Fate tracing of mature hepatocytes in mouse liver homeostasis and regeneration. J Clin Invest 2011;121:4851-4860. (Reprinted with permission.)
Recent evidence has contradicted the prevailing view that homeostasis and regeneration of the adult liver are mediated by self duplication of lineage-restricted hepatocytes and biliary epithelial cells. These new data suggest that liver progenitor cells do not function solely as a backup system in chronic liver injury; rather, they also produce hepatocytes after acute injury and are in fact the main source of new hepatocytes during normal hepatocyte turnover. In addition, other evidence suggests that hepatocytes are capable of lineage conversion, acting as precursors of biliary epithelial cells during biliary injury. To test these concepts, we generated a hepatocyte fate-tracing model based on timed and specific Cre recombinase expression and marker gene activation in all hepatocytes of adult Rosa26 reporter mice with an adenoassociated viral (AAV) vector. We found that newly formed hepatocytes derived from preexisting hepatocytes in the normal liver and that liver progenitor cells contributed minimally to acute hepatocyte regeneration. Further, we found no evidence that biliary injury induced conversion of hepatocytes into biliary epithelial cells. These results therefore restore the previously prevailing paradigms of liver homeostasis and regeneration. In addition, our new vector system will be a valuable tool for timed, efficient, and specific loop out of floxed sequences in hepatocytes.
Few phenomena have attracted the attention of tissue biologists as has the capacity of liver to regenerate. There are several intriguing aspects of this phenomenon, of which perhaps the most important is that the regenerated liver returns to almost exactly 100% of the original liver weight, as though governed by a “hepatostat.”1-3
Tissue damage leading to loss of liver is usually either diffuse (viruses, etc.) or localized to specific areas of the hepatic lobule, most commonly in the centrilobular region (chemicals requiring metabolic activation, such as acetaminophen, etc.). In order to distinguish between phenomena truly related to regeneration and those related to the inflammatory response due to hepatocyte necrosis, liver regeneration after partial hepatectomy has been a very popular model with investigators to study liver regeneration. It is generally accepted that following hepatectomy, hepatocytes, biliary cells, stellate cells, Kupffer cells, and endothelial cells replicate to make more of their own type. It has been argued, however, that liver regeneration after partial hepatectomy may unduly emphasize the capacity of the cells of the liver to take care of their own regeneration, entering into proliferation and replacing the lost cell type with phenotypic fidelity. In the last three decades, however, reproducible experimental models have been developed in which proliferation of hepatocytes during regeneration is suppressed.4, 5 Under those circumstances, a population of cells coming under the names of “oval” or “progenitor” cells emerge in the periportal areas, expand within the lobule, and eventually differentiate to become hepatocytes. Several studies have argued that the progenitor cells arise from a specific, preexisting, cell population distinct from either hepatocytes or biliary epithelial cells.6 The preponderance of recent studies, however, has provided strong evidence that the origin of the progenitor cells is in the biliary compartment, either in the ducts of the portal area or from the canals of Hering.7-9 In rats, when proliferation of hepatocytes is suppressed by acetylaminofluoren (AAF), soon after hepatectomy there is expression of hepatocyte-associated transcription factors in the biliary compartment, immediately prior to the appearance of the progenitor cells.7 Elimination of the biliary compartment by chemical toxins prior to AAF + hepatectomy also results in elimination of the progenitor cells, even when proliferation of hepatocytes is suppressed.10 The above should not be construed as implying that all biliary cells have the capacity to function as progenitor cells. Grompe and coworkers recently demonstrated that there are selective subpopulations of biliary cells having a distinct clonogenic capacity and which are capable of generating hepatocytes and biliary cells in culture.11 Recent work by Reid and coworkers also demonstrated that biliary cells play a role in this process.9, 12
Several recent publications have emerged, however, which suggest that the previous dogma of “phenotypic fidelity” of cellular events related to liver growth biology may need to be reconsidered. Furuyama et al.13 utilized a Sox-9 based lineage tagging approach to label biliary epithelial cells and duct cells in the pancreas. The article demonstrated that under their experimental conditions, hepatocytes and pancreatic acinar cells gradually emerged over the life of the mouse to replace more than half of the parenchymal cells in these organs. The same publication demonstrated that there was a substantial contribution of new hepatocytes generated by the biliary compartment even in the standard liver regeneration after partial hepatectomy. Using a similar but not identical cell lineage tagging for cells expressing Sox9 during embryonic development, Lemaigre and coworkers demonstrated that a small percentage of the periportal hepatocytes derives from remnants of the embryonic ductal plate. They were unable, however, to find any evidence of contribution of the biliary epithelium to production of hepatocytes in postnatal life.14 Taking a different approach, a publication by Iverson et al.15 demonstrated that a finite percentage of hepatocytes on a daily basis derive from cells that have never before expressed albumin. The most recent article addressing these issues was published by Willenbring and coworkers.16 In that publication, AAV vectors expressing CRE recombinase under the control of a (hepatocyte-specific) transthyretin promoter were injected into mice in which yellow fluorescent protein expression was held in check by a “stop” sequence surrounded by LoxP sites. Cre recombinase was activated only in hepatocytes and resulted in tagging hepatocytes red. The authors then performed several thoughtful experiments to critically examine whether biliary cells could contribute to formation of new hepatocytes and whether hepatocytes could contribute to formation of new biliary cells. These studies were conducted following chemical injury with carbon tetrachloride (CCl4), partial hepatectomy, bile duct ligation, and administration of a 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet. The illustrations highlighted that cell renewal in the liver under all these situations occurs predominantly (but not exclusively) with phenotypic fidelity, with only a small percentage of hepatocytes during liver regeneration potentially being contributed by biliary epithelial cells.
Is it possible to reconcile the different conclusions arising from these models of careful cell lineage tagging? Are all the assumptions made for each model fully validated? Is it possible that the limitations of wildtype animal manipulations, decried for many years as subject to multiple interpretations, have been replaced by more elegant methodologies with genetically modified mice, which nonetheless have limitations of their own that are more difficult to expose? If we examine the studies of the last two decades, and employing only wildtype nongenetically modified rats and mice, transdifferentiation of cells from the biliary compartment to form progenitor cells that eventually also transdifferentiate to hepatocytes occurs only when hepatocyte proliferation is suppressed or when hepatocyte death is so overwhelming that there no residual hepatocytes sufficient to provide restoration of the lost liver tissue. The publication by Furuyama et al. reaches different conclusions from the articles by Lemaigre and colleagues and by Willenbring and colleagues, who argue that in the absence of the above limits to hepatocyte proliferation, contribution of biliary cells to formation of new hepatocytes is either absent or miniscule. Currently, there is no “clean” model to suppress hepatocyte proliferation after partial hepatectomy in the mouse as it exists for the rat (i.e., AAF plus partial hepatectomy) and the rat model cannot be evaluated by lineage tagging. Despite the apparently contradictory studies with genetic mouse models, the majority of workers in liver growth biology seem to agree that the biliary compartment (portal ductules, canals of Hering, glands around gallbladder) is the source of progenitor cells and the formation of hepatocytes from biliary-derived progenitor cells under extreme conditions mentioned above is also generally accepted. The demonstration of expression of HNF4α and HEPPAR in proliferating biliary cells in fulminant hepatic failure in humans also strongly argues that this pathway is a clinically important SOS mechanism to salvage the liver from total collapse under extreme circumstances.17
The transdifferentiation in the opposite direction, i.e., hepatocytes giving rise to biliary epithelial cells, is much debated. The article by Willenbring and coworkers, using simple bile duct ligation, did not observe evidence for formation of biliary epithelial cells from hepatocytes. Previous studies using chimeric rat livers also demonstrated that under simple bile duct ligation there is only a small number of immediately periportal of hepatocytes transdifferentiating to biliary cells. However, when bile duct ligation is combined with exposure to the biliary toxin DAPM, thus causing loss of most of the biliary epithelium, more than 50% of the biliary ductules apparently derived from hepatocytes18 and that the receptors EGFR and MET play a unique role.19 The failure to observe this phenomenon by Willenbring and colleagues, also commented on by the authors, probably reflects the fact that in their study the biliary cell capacity to proliferate is not compromised. Of interest, in chronic biliary disease in humans caused by a variety of conditions, biliary-associated transcription factors appear in hepatocytes, suggesting that pathways of transdifferentiation of hepatocytes to biliary cells may also occur in humans under mechanisms operating in situations of compromised biliary cell proliferation during liver disease (e.g., primary biliary cirrhosis).17 The different scenarios for activation of proliferative compartments within liver are shown in Fig. 1.
Although the complete suppression of proliferation of hepatocytes and massive hepatocyte necrosis are extreme conditions that are easily detected, it is also conceivable that some of the discrepancies in results between the different genetic lineage tagging mouse models may be explained by some interference with the capacity of hepatocytes to proliferate. Such interference may not be an “all or none situation” but a more subtle restricting effect. Under such circumstances it would not be unreasonable to expect that progenitor cells may slowly and gradually come to the rescue. It would be wrong to conclude from such studies, however, that similar phenomena are necessarily occurring under normal circumstances in wildtype mice with no genetic manipulation, when clear and simple evidence obtained from straightforward regenerative models using accepted cell proliferation markers suggests that phenotypic fidelity of cell proliferation is the overwhelming norm. Nonetheless, it is not possible to completely exclude some degree of phenotypic promiscuity in small numbers, and critically examined lineage tagging experiments will continue to be helpful to resolve such issues.