Cellular responses in experimental liver injury: Possible cellular origins of regenerative stem-like progenitor cells


  • Potential conflict of interest: Nothing to report.



Mature hepatocytes divide to restore liver mass after injury. However, when hepatocyte division is impaired by retrorsine poisoning, regeneration proceeds from another cell type: the small hepatocyte-like progenitor cells (SHPCs). Our aim was to test whether SHPCs could originate from mature hepatocytes.


Mature hepatocytes were genetically labeled using retroviral vectors harboring the β-galactosidase gene. After labeling, retrorsine was administered to rats followed by partial hepatectomy to trigger regeneration. A liver biopsy was performed one month after surgery and rats were sacrificed one month later.


We observed the proliferation of small hepatocytes arranged in clusters in liver biopsies. These cells expressed Ki67 antigen and displayed high mitotic index. At sacrifice, regeneration was completed and clusters had merged. A significant proportion of clusters also expressed β-galactosidase demonstrating their origin from labeled mature hepatocytes. Finally, the overall proportion of β-galactosidase positive cells was identical at the time of hepatectomy as well as in liver biopsy and at sacrifice.


The constant proportion of β-galactosidase positive cells during the regeneration process demonstrates that mature hepatocytes are randomly recruited to proliferate and compensate parenchyma loss in this model. Furthermore, mature hepatocytes are the source of SHPC after retrorsine injury.

Avril A, Pichard V, Bralet M-P, Ferry N. Mature hepatocytes are the source of small hepatocyte-like progenitor cells in the retrorsine model of liver injury. J Hepatol 2004;41:737-743.(Reprinted with permission from The European Association for the Study of the Liver.)


Liver regeneration typically proceeds through the proliferation of mature hepatocytes and biliary epithelial cells in response to loss of tissue mass related to surgical partial hepatectomy (PH) or necrotic liver injury.1, 2 However, when the capacity of differentiated hepatocytes to proliferate is impaired, undifferentiated progenitor cells are activated to participate in liver regeneration.3, 4 Various types of injury can lead to impaired hepatocyte proliferation, and several experimental models have been extensively studied.3, 4 Each of these models requires inhibition of hepatocyte proliferation and stimulation of liver growth, and each results in the activation of an undifferentiated (or less differentiated) progenitor cell population.4 However, the timing and nature of the cellular responses differ, in some cases substantially.3, 4 In the 2AAF/PH model (where 2-acetylaminofluorene is the mito-inhibitory agent), liver regeneration is accomplished through the activation, expansion, and hepatocytic differentiation of oval cells (Fig. 1).5, 6 In contrast, only modest oval cell proliferation occurs in the retrorsine/PH model (where 12,18-dihydroxysenecionan-11,16-dione; β-longilobine is the mito-inhibitory agent) and liver regeneration is accomplished through the proliferation of small hepatocyte-like progenitor cells (SHPCs) (see Fig. 1).7, 8 It is commonly accepted that oval cells arise from a periductular cell, and perhaps directly from biliary epithelial cells.3 In contrast, the cellular origin of SHPCs is not known, and it is not clear if the SHPC population represents a novel progenitor cell population in the adult rat liver or a transitional cell that is derived from another progenitor cell type, such as oval cells (see Fig. 1).7, 8

Figure 1.

Progenitor cells in liver injury and regeneration. (A-C) Small hepatocyte-like progenitor cells (SHPCs) in the livers of retrorsine-exposed rats (A-B) 5 days and (C) 7 days after PH. Proliferative SHPCs are surrounded by retrorsine-injured hepatocytes (megalocytes). (D-F) Oval cell proliferation in the livers of rats 2AAF-exposed rats (D-E) 7 days and (F) 14 days after PH. At early time points, oval cells proliferate in the periportal regions. By 14 days post-PH, oval cell–derived small basophilic hepatocytes are evident (F). (G) Potential lineage relationships between liver progenitor cell types. Bar = 100 μm.

To address whether mature hepatocytes or some subset of hepatocytes could give rise to regenerative SHPC clusters after retrorsine-mediated liver injury, Ferry and colleagues employed the strategy of genetically labeling hepatocytes in vivo using a retrovirus and then following the fate of marked cells after retrorsine/PH.9 To facilitate infection of hepatocytes with an amphotrophic retroviral vector (encoding the Escherichia coli β-galactosidase gene), rats were treated with cyproterone acetate and triiodotyronine to induce cell division.10 This treatment regimen has been suggested to specifically stimulate hepatocyte division, which should enable specific retroviral marking of hepatocytes.10 In fact, this treatment protocol results in retroviral labeling of approximately 5% of hepatocytes (after two administrations of retrovirus at 2 × 109 infectious units/kg body weight), with a preferential distribution of positive cells in the periportal and midlobular zones.10 Following retroviral infection of rat hepatocytes, rats received two treatments of retrorsine (at 37 mg/kg) 2 weeks apart (at 7 and 9 weeks of age), with PH 4 weeks later.9 This retrorsine dose and administration schedule is similar to previously published procedures, but with notable differences. Almost all published reports that employ the retrorsine/PH model dose rats with 30 mg/kg retrorsine beginning at 6 weeks of age. In this liver injury model, the age of the animals at the time of retrorsine exposure is critical,11 but the timing of PH after retrorsine exposure is more flexible. Ferry and colleagues report that their Sprague-Dawley rats were not as sensitive to retrorsine treatment as Fischer 344 rats, requiring an increase in the dose used. Numerous investigators have successfully employed the conventional retrorsine/PH model (30 mg/kg retrorsine dose) in Fischer 344 rats as well as other rat strains (including Sprague-Dawley, Wistar, and Lewis rats). It is not entirely clear what effect this variation in retrorsine dose might have on our ability to compare the results of Ferry and colleagues to those in the literature. One major difference is the timing of liver regeneration from SHPCs as a function of retrorsine dose. Ferry and colleagues observed SHPC proliferation at early time points after PH, but with clearly isolated cell clusters remaining at 26 days post-PH, and complete regeneration (from the merging of these clusters) requiring nearly 2 months.9 By comparison, in rats treated with 30 mg/kg retrorsine, SHPCs (and their progeny) occupy 50% of the parenchyma by 14 days post-PH, and restoration of normal liver mass is complete by 30 days post-PH.7

Similar to a previous investigation,10 approximately 4% of hepatocytes were found to be β-galactosidase–positive after retroviral infection in retrorsine-exposed rats at the time of PH.9 Subsequently, Avril and colleagues examined the proportion of β-galactosidase–positive SHPCs and derived cell clusters at various times after PH; they found 4.6% of SHPC clusters to be β-galactosidase–positive in liver biopsies collected 26 days after PH.9 Furthermore, approximately 3.5% of all cells were found to be β-galactosidase–positive at this time point, and the occurrence of β-galactosidase–positive megalocytes (resulting from retrorsine-injured hepatocytes) was noted. These results suggest that some of the originally infected cells were retrorsine-resistant (either oval cell progenitors or SHPCs), while others were retrorsine-sensitive (normal) hepatocytes. It is not obvious whether retroviral infection of liver after treatment with cyproterone acetate and triiodotyronine results in β-galactosidase labeling of biliary epithelial cells (or other non-hepatocytic cells). Thus it is difficult to assess the possibility that oval cells give rise to β-galactosidase–positive SHPC clusters. A more comprehensive analysis of the cellular reactions during the first 26 days after PH would help to address some of these important questions. A similar percentage of β-galactosidase–positive hepatocytes was detected in liver tissues collected at the end of the experimental period, 56 days after PH (≈3.8%).9 These β-galactosidase–positive hepatocytes represent the progeny of proliferating SHPCs. In contrast, the majority (or all) of infected normal hepatocytes (retrorsine-injured hepatocytes) should have undergone apoptosis and been cleared from these livers before the completion of regeneration by the SHPCs.12 The finding of a similar percentage of β-galactosidase–positive SHPC clusters at biopsy (26 days post-PH) and β-galactosidase–positive hepatocytes at the completion of regeneration (56 days post-PH) suggests that retrovirally labeled SHPCs proliferated equally with the unlabeled progenitor cells. Given that SHPCs have been shown to completely repopulate the livers of retrorsine-exposed rats,7 this is the expected relationship between labeled progenitor cells and their progeny in the regenerated liver. However, it is not clear how the numbers of β-galactosidase–positive SHPC clusters at biopsy (4.6%) relate to the numbers of β-galactosidase–positive hepatocytes detected in the liver tissues collected at the time of PH (4%). Ferry and colleagues interpreted these results to suggest that random mature hepatocytes are recruited to repopulate the liver after PH in retrorsine-exposed rats.9 This might be an overly simplistic conclusion to draw from these numbers. It is likely that the majority of retrovirally labeled hepatocytes were sensitive to the mito-inhibitory effects of retrorsine and did not proliferate during the initial 26-day period post-PH, but arrested as megalocytes and subsequently underwent apoptosis. Thus, the similarity in percentages of β-galactosidase–positive cells observed between tissues collected at PH and at the various time points after PH might reflect the robust proliferation of a minority of retrovirally labeled cells (SHPCs) coupled with the loss of retrorsine-injured cells (through apoptosis).

Ultimately, Ferry and colleagues conclude that SHPCs are derived from retrorsine-resistant hepatocytes,9 as previously suggested.7, 8 While it was once thought that the proliferative capacity of hepatocytes was limited, numerous experimental studies have now shown that mature hepatocytes possess have an enormous regenerative potency.3, 4 These studies suggest that the average hepatocyte can generate large numbers of hepatocyte progeny and that subpopulations of hepatocytes could function as progenitor cells under specific pathological conditions. However, it is unlikely that SHPCs are derived from random mature hepatocytes, as suggested by Ferry and colleagues.9 Because the majority of hepatocytes are sensitive to retrorsine, SHPCs may represent a population of retrorsine-resistant hepatocytes lacking expression of specific cytochrome P450 enzymes.7 In fact, SHPCs do not express several cytochrome P450 enzymes, providing a plausible mechanism for retrorsine resistance.8 Moreover, previous studies suggest that SHPCs lack expression of certain genes characteristic of mature hepatocytes (including tyrosine aminotransferase and α1-antitrypsin), but express genes expected for a less mature hepatocytic cell type (including α-fetoprotein).8 These observations leave open the possibility that SHPCs may originate from oval cells or some other undifferentiated progenitor cell type. However, it is intriguing to speculate that small hepatocyte-like progenitor cells might arise from rare α-fetoprotein–positive parenchymal cells13 that morphologically resemble mature hepatocytes but lack the full repertoire of hepatocyte-specific gene expression.8

While it has long been known that the liver is a regenerative organ, it has only recently been recognized that the liver has several options for dealing with injury and functional deficit. Results from several experimental models suggest that there are multiple distinct stem-like compartments in the liver of the adult rodent, including populations of unipotential (mature hepatocytes and biliary epithelial cells) and multipotential (undifferentiated oval cells) stem-like cells, and perhaps other cell types.4 The SHPCs observed in the retrorsine/PH model may represent a novel stem-like cell that may be distinct from the other recognized liver stem cell populations. In addition to several distinct populations of liver stem-like cells, there may be a hierarchical cellular response to liver injury that involves activation of a specific stem-like cell population depending on the nature and extent of liver injury and the capacity of the various stem-like cell compartments to respond. The existence of numerous cellular responses to liver injury is well established.3, 4 The differential cellular responses to liver injury observed in disparate experimental models may reflect complex lineage relationships between undifferentiated stem cells (oval cells) and transitional cell types (SHPCs). The cellular response observed in the 2AAF/PH model varies with the extent of injury related to the dose of 2AAF.6, 14 In fact, at low doses of 2AAF, proliferating oval cells rapidly and synchronously differentiate into small hepatocytes.15 In this low-dose 2AAF model, small hepatocytes represent a dividing transit cell compartment derived from oval cells.15 Thus it is plausible that a similar cellular reaction might take place in the retrorsine/PH model.

The study published by Ferry and colleagues employed an experimental approach involving retroviral labeling of hepatocytes in vivo to address the cells of origin of SHPCs in the retrorsine/PH model of rat liver injury.9 The results of their limited study provide support for the suggestion that SHPCs might derive from rare parenchymal cells that reflect a novel population of retrorsine-resistant hepatocytes.7 However, additional studies using different experimental models of liver injury, in conjunction with genetic marking or cell transplantation approaches, will be required to definitively answer the question of whether SHPCs originate from a subset of parenchymal cells that morphologically resemble hepatocytes.