The phenotypic fate and functional role for bone marrow-derived stem cells in liver fibrosis

Summary

Liver fibrosis is an outcome of chronic liver injury of any etiology. It is manifested by extensive deposition of extracellular matrix (ECM) proteins that produce a fibrous scar in the injured liver. Bone marrow (BM) cells may play an important role in pathogenesis and resolution of liver fibrosis. BM cells contribute to the inflammatory response by TGF-β1 secretion and activation of liver resident myofibroblasts. Moreover, BM itself can serve as a source of collagen expressing cells, e.g. BM-derived fibrocytes and mesenchymal progenitors, which in turn, have a potential to in situ differentiate into fibrogenic myofibroblasts and facilitate fibrosis. Finally, BM cells play an active part in resolution of liver fibrosis after cessation of fibrogenic stimuli. While natural killer (NK) cells are implicated in apoptosis of activated hepatic stellate cells/myofibroblasts, cells of myelo-monocitic lineage secrete matrix metalloproteinases to actively degrade the fibrous scar. The focus of this review is on the current understanding of the role of different subsets of BM cells in the onset, development and resolution of liver fibrosis.

Introduction

Liver fibrosis is caused by chronic injury which triggers apoptosis of hepatocytes, damage of the endothelial barrier, recruitment of inflammatory cells, increased secretion of TGF-β1, and activation of myofibroblasts responsible for scar formation [10, 50]. However, the contribution of bone marrow (BM) cells to liver fibrosis remains controversial [44, 51]. At the onset of fibrosis, BM cells are recruited to the site of injury to facilitate inflammation. It is believed that monocytes and macrophages are the primary source of TGF-β1, the major fibrogenic cytokine that plays a critical role in activation of fibrogenic myofibroblasts.

Myofibroblasts express type I collagen and other extracellular matrix proteins that constitute the fibrous scar in liver fibrosis. Three sources of myofibroblasts have been identified: hepatic stellate cells (HSCs) in hepatotoxic liver injury [27], portal fibroblasts in cholestatic liver injury [18], and fibrocytes in any inflamed liver (Fig. 1). Most myofibroblasts retain the markers of being originally derived from either fibroblasts (such as Thy1 and elastin), HSCs (such as vitamin A droplets, GFAP, and desmin), or fibrocytes (CD45). Theoretically, myofibroblasts may also be derived directly from a precursor cell, unrelated to stellate cells, fibroblasts, or fibrocytes. Cell fate mapping studies in reporter mice have demonstrated that both hepatic stellate cells and fibroblasts are septum transversum mesenchymal cells that migrate from the mesothelium and submesothelium C [5].

Fig. 1. Possible origins of fibrogenic myofibroblasts.

Hepatic myofibroblasts may originate from liver resident mesenchymal cells. These include hepatic stellate cells, which under physiological conditions reside in the space of Disse in a quiescent state, and in response to injury undergo activation into myofibroblasts. Portal fibroblasts may also be a source of myofibroblasts in the fibrotic liver. In addition, BM-derived hematopoietic and mesenchymal cells may contribute to the myofibroblast population. While the role of mesenchymal stem cells in liver fibrosis is not well characterized due to the lack of specific markers and difficulties with their isolation, hematopoietic stem cells contribute to hepatic fibrocytes in response to liver injury.

Cessation of the injury often causes resolution of liver fibrosis with resorption of the fibrous scar [40, 41]. Under these circumstances, activated myofibroblasts undergo senescence [53, 77, 80], apoptosis and disappear [42, 73]. It has been shown that NK (and NKT) cells facilitate aHSCs apoptosis during regression of fibrosis [72], while newly recruited monocytes actively degrade extracellular matrix proteins (ECM)[20] via upregulation and secretion of matrix metalloproteinases (e.g. MMP13)[95] and collagenases [43].

Stem cell biology has become one of the most intensely studied areas of biomedical research and there is great optimism among scientists and the lay public that stem cells will be used as novel therapies for many incurable chronic diseases. Many institutions, including the State of California, have committed billions of dollars specifically to promote stem cell research with the goal of developing new therapies within a few years. As a result of new insights into stem cells, there is a renewed interest in the role of the bone marrow and its stem cells in liver fibrosis. The information to date is very conflicted, with different studies showing either a contributing effect or a therapeutic effect of bone marrow-derived cells to liver fibrosis.

Many studies have raised the issue of whether liver myofibroblasts may be derived from bone marrow stem cells, either hematopoietic or mesenchymal stem cells. Due to their well defined cell lineage markers and methodology for hematopoietic stem cell transfer, the contribution of hematopoietic stem cells to the population of liver myofibroblasts may be readily assessed in experimental murine liver fibrosis.

This review will address three issues: (1) the role of BM-derived macrophage to liver fibrogenesis, (2) the contribution of BM cells to myofibroblasts in the fibrotic liver, and (3) the role of BM stem cells in the resolution of liver fibrosis.

Inflammation

Expression of collagen type I marks fibrogenic/hematopoietic cells

While the fibrogenic properties of fibrocytes will be discussed below, it is important to note that expression of collagen type I by hematopoietic cells has a critical role in the development and maturation of hematopoietic BM cells required to mediate injury-triggered immune responses [22, 78, 89]. First, activated macrophages upregulate collagen type I upon maturation and migration to the site of injury [67]. Surprisingly, the level of collagen expression in activated macrophages is similar to activated cultured fibrocytes or fibrocyte-like cells [67]. Second, upregulation of collagen type I is associated with maturation of hematopoietic stem cells [22, 78, 89]. It is unknown why collagen expression is required for the function of hematopoietic stem cells. A recent study suggested that collagen I regulates the self-renewal of mouse embryonic stem cells through α2β1 integrin- and DDR1-dependent Bmi-1[89]. However, the level of collagen expression in activated macrophages and fibrocytes is relatively low compared to activated myofibroblasts [67, 79], so that these cells are not major sources of collagen, but most likely mediate cellular interaction [78] with extracellular matrix causing cytoskeletal rearrangement [17, 31, 86].

Alternative mechanisms to clear bacterial debris

Increased intestinal permeability has a critical role in the pathogenesis of liver fibrosis [81, 104]. Recent studies have demonstrated that in addition to phagocytosis, neutrophils, macrophages, and fibrocytes may utilize an alternative pathway to combat bacteria, by releasing extracellular DNA-based traps enriched in histones and major antimicrobial enzymes, cathelescidin and myeoloperoxidase [12, 16, 97, 106]. It remains unclear why terminally differentiated cells with phagocytic capacity decide to intake or exterminate bacteria [16]. This mechanism is activated in Vegenar granulomatosus [46] and Lupus nephritis [34]. Although the significance of such phenomenon for liver fibrosis still has to be investigated, fibrocyte-like cells from the spleen (CD45⁺Collagen-α1(I)⁺ BM-derived cells) may form DNA traps following LPS- or CCl₄-induced liver injury [52]. Thus, identification and classification of fibrocytes and fibrocyte-like cells recruited to the injured liver may provide new insights into the pathogenesis of liver fibrosis.

Recruited BM macrophages induce fibrosis

BM macrophages and Kupffer cells (liver resident macrophages) are the major source of TGF-β1 in liver fibrosis [81]. T and B lymphocytes are also recruited to the site of injury and further facilitate secretion of fibrogenic cytokines. Ablation of myolo-monocytic CD11b⁺ cells in mice at the onset of liver fibrosis attenuated activation of fibrogenic myofibroblasts and collagen deposition in liver and kidney fibrosis [20, 21] [45].

Bacterial flora and toll-like receptor (TLRs) signaling are critical in the activation of Kupffer cells/macrophages and TGF-β1 release [81]. For example, TLR4 mutant and knockout mice are resistant to fibrosis of different etiologies [38]. Moreover, genome wide analysis studies have demonstrated that individuals carrying a low efficiency polymorphism in TLR4 gene are less susceptible to HCV-induced liver fibrosis [56]. Toll-like receptors (TLRs) recognize pathogen-associated molecular patterns (PAMP) such as lipopolysaccharide (LPS), bacterial cell wall component, peptideglycan, and bacteria-derived unmethylated CpG-DNA [38]. In addition, endogenous ligands (alarmins, e.g. HMGB-1, hyaluronan) can bind TLR4 in the presence of CD14 and LPS-binding protein [LBP) and transduce similar signals [105]. Upon activation of TLRs, recruited BM cells produce inflammatory cytokines, such as TNF-a, IL-6, IL-1, MCP-1, and RANTES [82]. Moreover, microbial products have a significant impact on fibrogenic progression [104], and LPS synergistically facilitates other fibrogenic factors such as TGFβ-1, oxidative stress, and mechanical injury [2]. TLR4 on BM cells is important in experimental alcoholic liver disease [39], and TL9 on BM cells is important in experimental non-alcoholic steatohepatitis [61].

Antifibrotic effects of macrophages

Original experiments by Duffield et al. [20] and subsequent studies [60, 102] have demonstrated that, over a period of time, two functionally distinct types of macrophages are recruited to the injured liver. During the injury phase, pro-fibrogenic macrophages mediate recruitment of injury-associated macrophages that promote myofibroblast proliferation and apoptosis [32]. In contrast, during recovery from injury, a population of macrophages predominates that resembles classical macrophages and does not promote HSC survival but mediates matrix degradation. This macrophage population is present during resolution of injury and at a time when pro-fibrogenic and inflammatory mediator levels are decreasing [20]. These two functional phenotypes are separated chronologically in experimental liver fibrosis by several days, suggesting that they may represent different populations.

How do monocytes/macrophages promote matrix degradation? First of all, during the resolution phase, myelo-monocytic cells serve as a source of collagen degrading enzymes, such as collagenase and other matrix metalloproteinases (MMPs)[24, 36, 99]. Thus, increased secretion of MMP13 by hepatic macrophages is critical for dissolution of the fibrous scar in the recovering liver [24]. In addition, macrophages are responsible for clearance of apoptotic cells [10, 58, 69].

Fibrogenic myofibroblasts

Definition of myofibroblasts

Myofibroblasts are characterized phenotypically by a stellate shape and expression of stress fibers, abundant pericellular matrix and fibrotic proteins (α-smooth muscle actin (α-SMA), non-muscle myosin, fibronectin, vimentin, and collagen type I) [23]. Ultrastructurally, myofibroblasts are defined by prominent rough endoplasmic reticulum (rER), a Golgi apparatus producing collagen, peripheral myofilaments, fibronexus (no lamina) and gap junctions [23]. Myofibroblasts are implicated in wound healing and fibroproliferative disorders [28, 57]. Myofibroblasts are produced in response to fibrogenic stimuli, such as TGF-β1 [65]. Classic myofibroblasts differentiate from a mesenchymal lineage and, therefore, lack expression of lymphoid markers such as CD45 or CD34. However, subsets of myofibroblasts can express Thy1.1 (CD90) or CD34. It remains unclear if expression of these genes links (myo)fibroblasts to hematopoietic stem cells, or these antigens have a broader distribution than previously appreciated. Sustained injury may trigger (trans)differentiation of myofibroblasts from other cellular sources, including HSCs [10].

The question remains whether BM-derived cells are capable of giving rise to the functional myofibroblasts in liver fibrosis. Several BM cells have been implicated in fibrogenesis, such as fibrocytes and circulating mesenchymal cells, which could contribute to liver fibrosis.

The origin of fibrogenic myofibroblasts

Although initial reports have suggested that BM may be a source of fibrogenic myofibroblasts [26, 75], most recent studies have reported that the majority of myofibroblasts activated in response to injury are from liver resident cells [37, 48, 49, 51, 81]. These findings are based on BM transplantation techniques in mice, in which the collagen-α1(I) or collagen-α2(I) promoters drive expression of the GFP reporter only in BM cells [37, 51]. Since collagen-α1(I) or collagen-α2(I) fibers are expressed in the same cells to form a triple helix [85], these reporter genes are expected to exhibit identical localization. Indeed, similar results were obtained in both mice in response to two models of liver fibrosis [37, 51], bile duct ligation and toxic liver injury induced by CCl₄, demonstrating that activated myofibroblasts do not originate in the BM but emerge from the liver resident cells, e.g. HSCs and portal fibroblasts. Meanwhile, a small population of collagen type I expressing BM-derived cells, scattered in the liver and spleen of these mice, is composed of fibrocytes [37, 51]. Despite differences in experimental approaches and duration of injury, there was no evidence that BM contributes to replenishment of HSCs and portal fibroblasts or liver stem cells.

Fibrocytes are implicated in fibrogenesis of parenchymal organs

Fibrocytes are defined as spindle shaped “CD45 and collagen type I (Col⁺) expressing leukocytes that mediate tissue repair and are capable of antigen presentation to naive T cells” [13]. Due to their ability to differentiate into myofibroblasts, fibrocytes are implicated in the fibrogenesis of skin, lungs, kidneys, and the liver [1, 48, 87]. In addition to collagen type I, fibronectin and vimentin, fibrocytes express CD45, CD34, MHCII, MHCI, CD11b, Gr-1, and secrete growth factors (TGF-β1, MCP-1) that promote deposition of ECM [11, 70]. Upon injury or stress, fibrocytes proliferate in the BM and migrate to the injured organ [13, 70]. The reported number of recruited fibrocytes varies from 25% (lung fibrosis) [48, 88] to ≈3–5% (liver fibrosis, e.g. BDL and CCl₄) [49] of the collagen expressing cells, suggesting that the magnitude of fibrocyte differentiation into myofibroblasts depends on the organ and the type of injury. Fibrocytes originate from hematopoietic cells and differentiate in the liver into typical myofibroblasts [79]. Mice treated with human serum amyloid protein (hSAP)[66], a natural inhibitor of fibrocyte differentiation and maturation, develop less fibrosis in response to injury. Our data and studies in other parenchymal organs [14, 63, 68] clearly demonstrate that fibrocytes play an important role in pathogenesis of many fibrogenic disorders, including lungs. Elevated levels of circulating fibrocytes in peripheral blood in patients with lung fibrosis have a poor prognostic value [62]. Moreover, hSAP has been successfully tested in limited clinical trials in patients with skin, kidney and lung fibrosis [14, 59, 63, 68].

BM mesenchymal stem cells (MSCs)

MSCs are defined as self-renewable, multipotent progenitor cells with the capacity to differentiate into lineage specific cells that form bone, cartilage, fat, tendon and muscle tissue [44, 84]. Unlike hematopoietic stem cells, MSCs are radio- and chemoresistant [9] and do not express progenitor markers (CD45, CD34, 133 [44]) or myelo-monocytic markers (CD11b, MHCII, and F4/80). Hepatic myofibroblasts may arise from BM-derived mesenchymal progenitors [26, 75]. BM-derived mesenchymal progenitors can give rise to myofibroblasts in the injured liver [6, 19, 55]. BM-derived cells may populate fibrotic lungs [35] and the liver [75] and contribute to fibrosis by differentiating into tissue myofibroblasts [40, 48, 75]. By subfractionating the BM stem cell compartment, the hepatic BM-derived myofibroblast-like cells were reported to be of mesenchymal stem cell origin [44, 75]. Cultured mesenchymal stem cells have the potential to become myofibroblast-like cells when transplanted into mouse livers [6, 19].

Whether circulating mesenchymal progenitors significantly contribute to ECM deposition in the course of experimental liver fibrosis remains to be determined, but they most likely represent a population, distinct from hematopoietic-derived fibrocytes [51]. Unlike hematopoietic stem cells, the definitive markers for mesenchymal stem cells have not been identified, and ablative radiation protocols to establish donor cell transfer have not been standardized. Therefore, a definitive murine liver fibrosis experiment with documented transfer of all bone marrow constituents expressing a myofibroblast specific marker has not been reported. Although initial enthusiasm about the BM origin of myofibroblasts declined in the recent years, further studies are required to re-evaluate this phenomenon.

Liver fibrosis precedes development of hepatocellular carcinoma [10, 38]. Recruitment of BM-derived fibroblasts (including fibrocytes) has been implicated in the pathogenesis of liver cancer, and cancer of other organs [8]. Thus, using collagen-α1(I)-GFP and α-smooth muscle actin (SMA)-RFP mice, BM-derived myofibroblasts were shown to contribute to neoplasia in gut and intestine [71]. Since most liver injury models in mice develop within a short period of time, it is possible that these experimental conditions are too short for the recruitment of BM myofibroblasts, as seen in cancer during 8–9 months of development.

Contribution of BM cells in genetically altered mice

BM cells may have different roles in different mouse models of genetically-induced liver injury. Since these phenomena are usually not observed in the wild type mice, contribution of BM-derived cells to hepatic cells is discussed separately.

The classical example is FAH^−/− mice, in which a mutation in fumaryl-aceto-acetate hydrolase (FAH) gene causes a metabolic disorder equivalent to hereditary tyrosinemia type 1. Withdrawal of the protective drug 2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione (NTBC) from drinking water causes extensive apoptosis of FAH^−/− hepatocytes in these mice. Transplantation of wild type BM into these mice results in rescue from fatal liver failure by FAH-deficient hepatocytes. Wild type BM myelo-monocytes cells fuse with damaged hepatocytes to give rise to colonies of functional hepatocytes [54, 98]. Moreover, infusion of myeloid cells alone is sufficient to give rise to functional hepatocytes [3, 92, 101]. However, fusion of hepatocytes with macrophages was only rarely observed in wild type mice in response to other types of liver injury (CCl₄, BDL), suggesting that hematopoietic cells have a limited contribution to hepatocyte population under physiological conditions or in response to injury [3, 92].

Recruitment of fibrocytes into the injured liver representing a high percentage of myofibroblasts has been observed in Abcb4-deficient mice (Abcb4^−/− mice), and has been shown to substantially ameliorate development of liver fibrosis [74]. However, only a modest contribution of fibrocytes to liver fibrosis (3–5% of fibrogenic myofibroblasts) has been observed in wild type mice in response to CCl₄ and BDL [51, 79]. Although Abcb4-deficient mice provide a unique opportunity to study recruitment of fibrocytes in great detail, they do not reflect the fibrocyte contribution to the population of myofibroblasts in hepatotoxic or cholestatic injury.

Resolution of liver fibrosis

Dissapearance of myofibroblasts

Reversal of fibrosis is associated with increased collagenase activity, activation of macrophages/Kupffer cells that secrete matrix metalloproteinases, e.g. MMP-13, and matrix degradation [24, 95]. Senescence and apoptosis of activated HSCs play a significant role in resolution of liver fibrosis by eliminating the cell type responsible for producing the fibrotic scar [41, 53]. Several mechanisms are implicated in the apoptosis of activated HSC: (1) activation of death receptor-mediated pathways (Fas or TNFR-1 receptors) and caspases 8 and 3; (2) upregulation of pro-apoptotic proteins (e.g., p53, Bax, caspase 9); and (3) decrease of pro-survival genes (e.g., Bcl-2) [50]. A population of liver-associated natural killer (NK) cells and NKT cells mediate apoptosis of activated HSCs [72]. Kupffer cells and BM macrophages actively participate in clearance of apoptotic cells and degradation of extracellular matrix proteins.

Studies in culture suggest that aHSCs, at least in part, can revert to a quiescent phenotype. Therefore, the disappearance of activated α-SMA⁺ Col⁺ HSCs in the course of fibrosis reversal may indicate that activated HSCs return to their quiescent state, which is associated with expression of lipogenic genes (Adfp, Adipor1, CREBP, PPAR-γ) [83] and storage of vitamin A in lipid droplets. Depletion of peroxisome proliferator-activated receptor gamma (PPAR-γ) constitutes a key molecular event for HSC activation, and ectopic over-expression of this nuclear receptor results in the phenotypic reversal of activated HSC to quiescent cells in culture [83]. The treatment of activated HSCs with an adipocyte differentiation cocktail, over-expression of SREBP-1c, or culturing on basement membrane-like ECM [29, 100] result in up-regulation of adipogenic transcription factors and cause morphologic and biochemical reversal of activated HSCs to quiescent cells [93, 94]. Although these results suggest that activated HSCs can revert to a quiescent state, these findings have only been documented in cultured cells.

Therapy

Many studies have demonstrated that transplantation of bone marrow cells reduces experimental liver fibrosis ([76, 90] and others). The mechanism is not trans-differentiation of bone marrow cells into hepatocytes. More likely, hematological stem cells may contribute to the reversal of liver fibrosis via macrophages that produce collagenases [91] and phagocytose dead parenchymal cells [69]. More unexpectedly, mesenchymal stem cells, even though they have the potential to become myofibroblasts, also have functions that may contribute to the reversal of fibrosis. Cultured mesenchymal stem cells secrete agonists that inhibit hepatocyte apoptosis, induce hepatocyte proliferation, and increase hepatocyte specific gene expression [96]. Also, mesenchymal stem cells may be induced in culture to become endothelial progenitor cells (EPCs). Transplantation of EPCs reverses hepatic fibrosis and improves survival in CCl₄-induced cirrhosis in rats [64].

BM cells for anti-fibrotic therapy

The improvement of liver function following transplantation of hematopoietic progenitors in mice and rats with injured livers provided the basis for clinical trials [25]. Clinical studies with adoptive transfer of autologous CD133⁺ BM cells in patients have been reported to stimulate liver regeneration [4]. Similar to that, autologous infusion of CD34⁺ blood cells, or even monocytes, improved biochemical parameters and stimulated liver regeneration [33]. Within the limits of these small, uncontrolled clinical trials, evidence is starting to accumulate that transplantation of hematopoietic progenitors may be beneficial in patients. However, the mechanism of their action remains to be defined. Such improvement may result from release of cytokines and growth factors by transplanted hematopoietic cells, or occur due to infusion of scar-resorbing monocytes. In concordance with these observations, treatment with granulocyte-colony stimulating factor (G-CSF) was used to mobilize the BM cells and demonstrated a positive histological effect in patients with alcoholic steatohepatitis [30].

Mesenchymal stem cells serve as another potential target for the liver stem cell therapy. In addition, mesenchymal cells are readily available (for example, from fat tissue) and relatively easy to expand in vitro. A recent study investigated the ability of purified hematopoietic stem cells (HSCs), mesenchymal stem cells (MSCs), and mononuclear cells to engraft and contribute to liver regeneration in response to injury in mice [15]. However, only a low level of engraftment with the MSCs and reconstitution of the liver mass has been reported [7].

In concordance with this notion, injection of MSC-derived conditioning media into a liver-assist device decreased hepatocyte apoptosis and increased their proliferation [96, 103]. However, recent studies have raised a safety question on MSCs transplantation, demonstrating that MSCs can give rise to myofibroblasts in mice in response to liver injury. For example, BM-derived MSCs contributed to the development of liver fibrosis in chimeric mice that received bone marrow transplantation with an enriched BM mesenchymal fraction, and subjected to the CCl₄-liver injury [75]. Taken together, both hematopoietic and mesenchymal stem cells demonstrate a limited, if any, contribution to hepatocyte replenishment, but may stimulate liver function by providing soluble growth factors or cytokines [3, 49, 92].

A few clinical trials have been performed in patients with end-stage liver disease caused by hepatitis B, hepatitis C, alcoholic liver disease, and cryptogenic fibrosis. These patients were transplanted with autologous MSCs harvested from the iliac crest. The tested parameters (albumin, creatinine) demonstrated a modest but significant improvement without severe adverse effects, suggesting that MSCs might be useful for the treatment of end-stage liver disease with satisfactory tolerability [47].

Conclusions

The literature provides evidence that bone marrow cells might contribute to increase or to inhibit experimental liver fibrosis (Fig. 2). Although there is clearly a need for additional, better defined studies, some conclusions can be made from our current information. Hematological stem cells are the source of monocytes, Kupffer cells and recruited macrophage. Overall, these cells contribute to the initial inflammation in the injured liver that progresses to liver fibrosis. However, recruited macrophages may also secrete agonists such as IL-10 that inhibit stellate cell activation as well as collagenases that cause regression of fibrosis. Hematological stem cells are also the source of fibrocytes, which are recruited to the injured liver and function in the innate immune response as well as differentiate into myofibroblasts. Mesenchymal stem cells have the capacity to become myofibroblasts, but studies to follow their cell fate in vivo are limited by the lack of specific markers.

Fig. 2. Potential roles of BM-derived progenitor populations in liver injury.

BM is the source of hematopoietic and mesenchymal stem cells, which may participate in the response to liver injury.

Most, but not all, studies using BM transplantation have demonstrated a beneficial effect on experimental liver fibrosis. The mechanism for this benefit is unclear, and in particular BM-derived cells do not constitute a significant source of hepatocytes in the injured liver. However, both mesenchymal stem cells and hematopoietic stem cells are reported to contribute to the regression of liver fibrosis. On the basis of these studies, small, mostly uncontrolled clinical studies have treated cirrhotic patients with autologous transplantation of BM derived cells. Although these studies have established the feasibility of this approach, the mechanism and long term benefit of transplantation of BM-derived cells in cirrhosis is unknown.