Despite advances in the understanding of hepatic fibrogenesis and the central role of activated hepatic stellate cells and other myofibroblasts,1 it remains enigmatic why stellate cell activation and collagen deposition do not necessarily occur at the immediate sites of lobular injury. In diseases such as viral hepatitis and hereditary hemochromatosis, although most of the injury involves the hepatocytes, fibrosis occurs in and around the portal areas. Recent studies have suggested, but not yet proven, that altered hepatic regeneration with activation of hepatic progenitor cells (HPCs) could play a role in this periportal fibrosis.
There are two pathways of hepatic regeneration. In the normal liver, parenchymal loss is repaired by division of normally quiescent, adjacent hepatocytes. However, pre-existing liver diseases such as viral infection, fatty liver disease, excessive iron storage and other conditions associated with oxidative stress significantly impair the ability of hepatocytes to replicate, promoting the activation of a second regenerative pathway.2 This results in the proliferation of a transit-amplifying compartment of cholangiocyte-like HPCs in the periportal canals of Hering. These cells (also called oval cells in rodents) are at least bipotential and capable of differentiation into both hepatocytes and cholangiocytes, and typically form a complex of small ductules and strings of cholangiocytes, termed the oval cell or ductular reaction.3 They have been of increasing interest in recent years because of potential roles not only in regeneration and liver repopulation but also in carcinogenesis and fibrogenesis.4 With regard to the latter, studies in rodent models5–7 and human diseases such as viral hepatitis and steatohepatitis5, 8–10 have noted a close correlation between the degree of hepatocellular injury, HPC expansion, and fibrosis. It has been postulated that the combination of impaired hepatocytic regeneration, causing activation of the HPC compartment, coupled with parenchymal cell loss stimulating proliferation of the HPCs, could in some way drive periportal fibrosis. Secretion of profibrogenic cytokines and growth factors from the ductular reaction similar to biliary fibrosis, or even direct cellular transitioning to myofibroblasts have been suggested as mechanisms.11
The hypothesis is appealing because it can explain the predominance of portal fibrosis, but to date the data have been correlative only. It has not been shown that HPCs directly cause stellate cell activation and fibrosis in vivo, and other interpretations include the opposite, that activated stellate cells promote HPC expansion, or that stimulation of the HPC and stellate cells occurs independently but in tandem, perhaps through similar mediators or stimuli.
In this issue of HEPATOLOGY, van Hul and colleagues examine this.12 Using a choline-deficient ethionine-supplemented (CDE) murine model, the authors performed a time-course study to assess the relationship between HPC expansion, stellate cell activation, and collagen deposition. This model induces a relatively vigorous HPC and fibrogenic response. At the time points studied from day 3 to day 21 there was an early burst of increased expression of extracellular matrix components and stellate cell activation; at the first time point at day 3, there was a 15-fold increase in collagen I expression and a 25-fold increase in alpha-smooth muscle actin messenger RNA, as well as collagen deposition periportally. Conversely, the HPC expansion lagged this, with a statistically significant increase in cytokeratin 19 (CK19) protein staining (a marker for HPCs) and alpha-fetoprotein expression (a marker of hepatoblast transformation) beginning only by day 7. This is interpreted as an indication that stellate cell activation and extracellular matrix deposition precedes the HPC expansion and is deposited ahead of the intralobular migration of the ductular reaction. The authors conclude that the development of a progenitor cell niche, including matrix deposition, is a requisite for HPC activation and differentiation, similar to other stem cell niches (developmental, physiological, and cancer-associated) that are now being characterized.13–15 The study indicates, at least in this model, that fibrosis does not occur simply as a result of the HPC expansion.
However, as designed it still cannot answer whether the HPC and stellate cell activation and proliferation are interdependent, whether they occur in parallel due to some common stimulus (quite possible given the degree of hepatocellular injury in this model) or whether there is true primacy of the myofibroblastic reaction and matrix deposition before HPC expansion can occur. Some of the results do, in fact, suggest that HPC activation may be occurring at day 3, the earliest time point that was studied. Even though the area of CK19-positive HPCs was not significantly increased over control levels by image analysis, quantitative reverse transcription polymerase chain reaction showed a significant increase of CK19 messenger RNA, albeit by only two-fold to three-fold, at day 3 (their figure 5) and it appears that at least two animals had increased CK19 staining in figure 3B. It should also be remembered that this is but a single model, and an aggressive one at that, so that extrapolation to human chronic liver disease should be cautious.
Even to those proponents who believe that activation of the HPC regenerative pathway and the development of a ductular reaction could be an important driver of periportal fibrosis in chronic liver disease, the findings from van Hul and colleagues should not come as any surprise. It is now clear that there is an intimate cross-talk between HPCs and stellate cells.13, 16, 17 Stellate cells are a major source of growth factors for HPCs, expressing transforming growth factor-alpha, hepatocyte growth factor, and acidic fibroblast growth factor. However, other inflammatory cells of the innate immune system such as Kupffer cells are likely to provide additional initiating factors. Reciprocal stimulation was also shown recently.18 Expression of lymphotoxin-beta by HPCs, required for fibrogenesis in the CDE model,6 did not directly activate the stellate cells, but rather caused increased stellate cell expression of intercellular adhesion molecule-1 and the chemokine RANTES. It was postulated that the HPC-stellate cell cross-talk enhanced fibrogenesis indirectly by promoting inflammatory cell recruitment and cellular migration.18
Although the present study still does not answer the ultimate question of whether there is a cause and effect in the relationship between collagen deposition and HPC expansion, it serves to highlight several important points. First, stellate cell activation and periportal fibrosis is not simply a consequence of the ductular reaction, but rather is likely to be an important player in augmentation of the alternate regenerative pathway. Because of this, antifibrotic therapeutic manipulation aimed at stellate cell quiescence could have the undesired consequence of inadequate hepatic regeneration. Second, the fate of expanded HPCs is not invariable. The cells are bipotential, and it is likely that the inflammatory milieu modifies the proportion of cells differentiating along hepatocytic or biliary lineage. It will be necessary to study chronic, clinically relevant models or human liver tissue to determine how to promote hepatocyte replenishment without enhancing those factors that could also induce progressive fibrosis. Finally, although the source of collagen I around the HPC is probably stellate cells or portal myofibroblasts, there is previously unexpected plasticity in many of the parenchymal and stromal cells of the liver. Through epithelial-mesenchymal transition and mesenchymal-epithelial transition, recent studies have suggested that transition of hepatocytes19 and cholangiocytes20 into myofibroblasts, and even stellate cells into HPCs21 may occur, adding complexity to these studies of fibrogenesis. Answers to the age-old question of what comes first will have to wait.