Stellate cells: A moving target in hepatic fibrogenesis


  • Scott L. Friedman

    Corresponding author
    1. Division of Liver Disease, Mount Sinai School of Medicine, New York, NY
    • 1123, Mount Sinai School of Medicine, 1425 Madison Ave., Room 1170C, New York, NY 10029
    Search for more papers by this author
    • fax: 212-849-2574

  • See Article on Page 1151

Progress in understanding the cellular and molecular basis of hepatic fibrosis continues to accelerate, leading to an understanding that is increasingly nuanced. A clear example is the paper of Magness and coworkers1 in this issue of HEPATOLOGY, which explores the topic of stellate cell heterogeneity using an elegant genetic approach in mice to unearth subtle differences in cellular phenotype and plasticity.


αSMA, alpha smooth muscle actin; TIMPs, tissue inhibitors of metalloproteinases; GFP, green fluorescent protein; RFP, red fluorescent protein.

First, the “simple” view of stellate cell biology. Interest in these nonparenchymal persinusoidal cells, first described by Kupffer in the 19th century, was resurrected through the morphological studies of Ito, Popper, Wake and others.2 Stellate cells' unique capacity in liver to store vitamin A led to their identification in tissue and their isolation using density gradient methods.3, 4 These advances ushered in an exciting era of fibrosis research in which the phenotype of stellate cells was systematically characterized using culture and in vivo models, establishing them as capable of undergoing “activation” or transdifferentiation into a fibrogenic, proliferative, and contractile cell type that is at the core of the fibrotic response to liver injury.5 Because the initial isolation methods relied on a high content of vitamin A to enable their separation, a relatively homogeneous population of vitamin A–rich cells was isolated that, in retrospect, did not represent the full phenotypic spectrum of the cell type.

Nonetheless, significant advances grew out of these approaches, in particular the characterization of intracellular filaments in stellate cells, cytoskeletal proteins whose composition was thought to represent the tissue of origin. Stellate cells were found to express desmin, a myogenic protein,6 as well as glial fibrillary protein, a marker of neural crest cells.7 Next, the expression of contractile filaments, alpha smooth muscle actin (αSMA), and then myosin were identified in stellate cells that had become activated in culture or in vivo.8, 9 These findings generated a straightforward paradigm in which vitamin A–rich stellate cells undergo activation in response to liver injury, losing their retinoid, and following a one-way path toward a fibrogenic and proliferative phenotype.

The story became considerably more complicated as the heterogeneity of stellate cells and their related mesenchymal cell types became apparent.10 Rather than a single cell type with an identical retinoid and cytoskeletal phenotype, stellate cells appeared to contain variable amounts of vitamin A11 and differing combinations of intracellular filaments, depending on their lobular location, the species studied, and the nature of the liver injury (i.e., biliary vs. parenchymal). For example, an old concept of “portal fibroblasts” reemerged,12, 13 and it also became clear that there is a continuum of changes in gene expression during stellate cell activation, such that the quiescent cell first becomes activated, then continues to evolve into a myofibroblast-like cell.14 Moreover, activated stellate cells are distinct from myofibroblasts in their vitamin content, contractile activity, and relative responsiveness to cytokines, particularly transforming growth factor beta.15 A growing list of neural genes identified in stellate cells16 has broadened the potential combinations of markers that define subpopulations of this cell type. Even when stellate cells reach replicative senescence, the pattern of gene expression continues to evolve, with acquisition of a more inflammatory and less fibrogenic phenotype.17

Stellate cell plasticity was further appreciated with the recognition that their clearance from injured liver could be regulated by apoptosis, a pathway that is proving essential to allow recovery from liver injury and reversal of fibrosis.18 This response is also heterogeneous, with variable sensitivity to apoptosis, reflecting competing activities of apoptotic ligands and metalloproteases that may regulate cell death and their specific inhibitors, especially tissue inhibitors of metalloproteinases (TIMPs).

A related source of conflicting data concerns the embryological origin of stellate cells, a question that has proven even more vexing as the overlapping myogenic, neural, and mesenchymal phenotypes are now fully recognized. The traditional view holds that stellate cells are derived from septum transversum as it invades gut endoderm, giving rise to a common stellate cell/sinusoidal endothelial cell precursor. However, this theory does not explain the extensive expression of neural crest markers, or the capacity for myogenic gene expression, including αSMA, myosin,19 and the transcription factors MyoD20 and MEF2.21

Finally, the anatomic and cellular source of activated stellate cells in liver has provoked controversy and intrigue. Although strong evidence suggests that the bulk of cells arise from resident quiescent stellate cells, more recent data suggest that bone marrow22, 23 and possibly even hepatocytes24 could also give rise to fibrogenic cells in liver.

These complementary yet sometimes conflicting findings have raised critical questions whose answers could refine our efforts to define novel diagnostics and treatments for liver fibrosis: (1) Are all stellate cell/mesenchymal cell subpopulations equally responsible for extracellular matrix production in liver, and are they all similarly responsive to the signals that regulate hepatic fibrosis progression and regression? (2) Does a given subpopulation have a fixed phenotype, or is it capable of “transdifferentiation,” at least as assayed by intermediate filament expression? (3) Can stellate cells be generated from nonmesenchymal sources, possibly for use in ex vivo gene therapy or even reconstitution of liver function ex vivo in combination with parenchymal cells?

In this context, the study by Magness provides some definitive answers, and it also highlights novel technical approaches that could be exploited in future studies to extend these initial findings. The investigators have created a double transgenic mouse in which a green fluorescent protein (GFP) is driven by the collagen I promoter, while a red fluorescent protein (RFP) is controlled by the αSMA promoter. Expression of GFP and RFP in stellate cells was tracked by flow cytometry and direct visualization following their activation in primary culture and in vivo following bile duct ligation, a standard method for cholestatic fibrosis induction. From a technical perspective alone, the validation of promoters suitable for transgenic expression in activated stellate cells is a recent and important advance that provides new opportunities for the study of this cell type in vivo using not only reporter genes, but also biologically active transgenes. As might be expected, the results indicate that there are mixed populations of activated cells, some of which express GFP or RFP alone, others of which express both. Specific patterns of gene expression were seen in cells expressing αSMA, with higher levels of ICAM-1, MMP-13, reelin, TIMP1, and synaptophysin messenger RNAs (mRNAs). Ideally, these limited mRNA analyses could be complemented by a more comprehensive study using complementary DNA microarray. In vivo, there was some regional localization such that fibrogenic cells expressing only collagen-GFP were peribiliary, whereas those expressing both collagen-GFP and αSMA-RFP were primarily parenchymal. However, in what is among the most important findings of the study, there was significant plasticity of cellular phenotype with regard to the pattern of transgene(s) expression. Specifically, cells initially expressing collagen–GFP reacquired expression of all 3 phenotypes (αSMA-RFP only, collagen-GFP only, or both) following long-term growth in culture, although cells initially expressing αSMA-RFP appeared to be less plastic, and transgene expression in this isolate did not deviate to the same extent as in collagen-GFP cells.

Collectively, these findings reinforce earlier histochemical data highlighting the heterogeneity of stellate cells with respect to classical markers of stellate cell activation. They also underscore their plasticity, although it appears that the capacity to acquire new phenotypic features is not equal among all subpopulations. Whether those with reduced capacity to transdifferentiate are actually terminally differentiated is uncertain. More importantly, it is not known if these differences in collagen-GFP and αSMA-RFP expression reflect important functional differences, possibly rendering some cells less responsive to antifibrotic therapies or more resistant to apoptosis. For example, if a therapy were developed to target a specific cytokine receptor, it would be useful to confirm expression of that receptor by the relevant subpopulation responsible for fibrogenesis in vivo. On the other hand, using the cytokine receptor targeting as an example, rather than regularly performing genetic studies to confirm its upregulation in subpopulations in transgenic animal models, documenting its upregulation in whole liver using either a simple Western blot, immunostaining, or real-time polymerase chain reaction, is likely to sufficiently justify attacking this receptor as a potential therapeutic target.

While important answers have emerged from this informative study, additional questions require further clarification, and follow-up studies are readily apparent from this novel approach. These dual transgenic mice could be studied using other models of liver injury as well as following partial hepatectomy, which would further clarify the impact of different stellate cell subpopulations in liver regeneration and repair. The mice could also be very informative in tracking the fate of the stellate cell subpopulations during resolution of liver injury, possibly identifying specific cells that are relatively resistant to clearance through apoptosis and instead persist in injured liver. Such studies might also indicate, for the first time, the capacity of activated stellate cells to undergo reversion to a quiescent state in vivo rather than undergoing apoptosis during resolution of fibrosis. More broadly, use of transgenic markers of both stellate cells and other liver cell populations (e.g., Kupffer and sinusoidal endothelial cells) could facilitate similar analyses of plasticity and also allow for greater elucidation of cell–cell interactions. Transgenic models of this type in which cells are genetically “marked” might also reveal the full range of source(s) of stellate cells in injured liver, as well as offer new approaches to clarifying their embryological origins.25

In summary, these exciting new data provide direct evidence of functional heterogeneity of stellate cells during activation in vivo and in culture and complement accumulated immunohistochemical findings. The results continue to move us away from the simple paradigm of a single stellate cell type rich in vitamin A undergoing unidirectional activation. However, since significant plasticity between some subpopulations was also documented, developing antifibrotic therapies based on apparent differences among subpopulations of stellate and mesenchymal cells is like trying to hit a moving target. Regardless, since fibrogenic cells in liver express most target molecules currently in the “gun sights” of antifibrotic strategies, the findings are more likely to refine our weaponry rather than disarm our current efforts at antifibrotic therapy.