Reversibility of Liver Fibrosis and Inactivation of Fibrogenic Myofibroblasts

Abstract

Many studies have demonstrated that hepatic fibrosis is reversible. Regression of liver fibrosis is associated with resorption of fibrous scar and disappearance of collagen producing myofibroblasts. The fate of these myofibroblasts has been recently revealed: Some myofibroblasts undergo senescence and apoptose during reversal of fibrosis, while other myofibroblasts revert to a quiescent-like phenotype. Inactivation of myofibroblasts is a newly described phenomenon¹ which now requires mechanistic investigation. Understanding of the mechanism of HSC inactivation upon cessation of fibrogenic stimuli may identify new approaches to revert already existing aHSCs/myofibroblasts into a quiescent-like state. This review summarizes the research on the inactivation of hepatic myofibroblasts.

Introduction

Hepatic fibrosis is an outcome of many chronic liver diseases, including hepatitis B virus (HBV), hepatitis C virus (HCV), alcoholic liver disease and non-alcoholic steatohepatitis (NASH) ². Hepatic fibrosis is characterized by extensive deposition of extracellular matrix proteins (ECMs), mostly type I collagen, forming a scar. Chronic liver injury damages hepatocytes. Injured or apoptotic hepatocytes secrete factors that facilitate activation and recruitment of inflammatory cells to the injured liver. Activated macrophages secrete IL-6 and TGF-β1, which in turn, activate hepatic myofibroblasts. Myofibroblasts are not present in the normal liver, but in response to injury they transdifferentiate from hepatic stellate cells (HSCs), upregulate collagen and produce the fibrous scar. Myofibroblasts are characterized by expression of a-smooth muscle actin (α-SMA) and type I collagen, and in all clinical and experimental liver fibrosis serve as a major source of the ECM. Thus, activation and proliferation of hepatic myofibroblasts is a key mechanism in the development of liver cirrhosis.

Myofibroblasts are the primary target of anti-fibrotic therapy

Hepatic myofibroblasts are the major source of collagen Type I in fibrotic liver. Therefore, elimination of myofibroblasts or their inactivation is a goal for therapy. Several sources of myofibroblasts have been identified^3–6. It is believed that hepatic stellate cells (HSCs) are the major source of fibrogenic myofibroblasts, and contribute > 80% of the collagen producing cells ^{2, 1,7}. Under physiological conditions, HSCs reside in the space of Disse and exhibit a quiescent phenotype (qHSCs). They express neural markers, such as GFAP, synemin, synaptophysin, and nerve growth factor receptor p75, and store vitamin A in lipid droplets³. In response to injury, qHSCs decrease vitamin A storage and peroxisome proliferator-activated receptor gamma (PPARγ) expression, and activate into collagen type I and α-SMA expressing myofibroblasts². Although the mechanism of HSC activation has been comprehensively studied, insights into the fate of HSCs during regression of liver fibrosis are new.⁸ In addition to HSCs, portal fibroblasts^9,10 and bone marrow (BM)-derived fibrocytes ¹¹ can also contribute to hepatic myofibroblasts.

Reversal of liver fibrosis

Mechanism of regression of liver fibrosis

Sequential liver biopsies from patients with liver fibrosis have demonstrated that removing the underlying etiological agent may reverse hepatic fibrosis in patients with secondary biliary fibrosis ¹², Hepatitis C ¹³, Hepatitis B ¹⁴, NASH ¹⁵, and autoimmune hepatitis ¹⁶. Withdrawal of the etiological source of the chronic injury (e.g. HBV, HCV) ² results in decrease of pro-inflammatory and fibrogenic (including TGFβ1) cytokines, increased collagenase activity ^2,3, decreased ECM production, and the disappearance of activated myofibroblasts.

Studies of liver fibrosis in rodents have confirmed that established liver fibrosis can reverse upon cessation of etiological agent. Reversal of liver fibrosis has been successfully studied in experimental models CCl₄ ^17,18, alcohol and BDL ¹⁹ induced liver fibrosis ². Several events have been identified to be critical for regression of liver fibrosis. These include disappearance of hepatic myofibroblasts/aHSCs, recruitment of collagenese secreting monocytes/macrophages, and prevalence of matrix degrading metalloproteinases over their natural inhibitors TIMPs.

Conditions for reversal of liver fibrosis

BM-derived monocytes and liver resident macrophages play an important role in reversal of liver fibrosis. Selective ablation of CD11b+ cells in mice (CD11b-DTR) during the recovery phase from liver injury significantly attenuates fibrosis resolution, suggesting that macrophages mediate distinct functions at the onset of fibrosis and during recovery. Increased collagenase activity is a primary pathway of fibrosis resolution. At this stage, activated macrophages/Kupffer cells secrete matrix metalloproteinases, e.g. MMP-13 interstitial collagenase, and other enzymes responsible for matrix degradation ^6,20. Moreover, increased activity of collagen degrading enzymes during fibrosis resolution correlates with decreased amount of TIMPs, tissue inhibitors of matrix metalloproteinasesis^21,22. Activated myofibroblasts/HSCs serve as a significant source of TIMPs, and disappearance of myofibroblasts/HSCs during recovery is associated with reduced production of TIMPs.

An established mechanism for the elimination of activated myofibroblasts is due to senescence²³ and apoptosis of activated HSCs ²⁴. Several mechanisms are implicated in the apoptosis of activated HSCs: 1). Activation of death receptor-mediated pathways (Fas or TNFR-1 receptors) and caspases 8 and 3; 2) up-regulation of pro-apoptotic proteins (e.g. p53, Bax, caspase 9); and 3) decrease of pro-survival genes (e.g. Bcl-2) ²². A population of liver associated natural killer (NK) cells and γδT (NKT) cells stimulate apoptosis of activated HSCs. Drugs that induce apoptosis in activated HSCs (glyotoxin, sulfasalazine, IKK inhibitors, and anti-TIMP antibodies) cause liver fibrosis to regress ^2,3.

Conditions for irreversible liver fibrosis

Whether end-stage cirrhosis can reverse to a normal liver architecture remains controversial^25,26. However, significant improvement in hepatic structure and function provide evidence of regression of liver fibrosis^21,27. Perhaps ECM remodeling is limited in cirrhosis by formation of non-reducible cross-linked collagen and an ECM rich with elastin fibers preventing its degradation. This pathophysiological state may lead to a “point of no return” for liver fibrosis^26,27. The characteristics of myofibroblast environment play a critical role in myofibroblast survival. For example, stiffness of extracellular matrix and increased contact between myofibroblasts and collagen scar promote survival of myofibroblasts and facilitate their activation²⁸. In support of this notion, transgenic mice expressing an uncleavable form of Collagen Type I are more susceptible to liver fibrosis, and demonstrate a defect in spontaneous resolution of liver fibrosis upon cessation of liver injury²⁹. Similar to these transgenic mice, prolong duration of fibrogenic liver injury often results in formation of persistent and uncleavable scars caused by irreversible crosslinking of collagen fibers. Furthermore, the presence of elastin fibers distinguishes “biochemically mature scars”, which are more likely to persist in recovering liver¹⁸. In addition, persistent scars are associated with formation of areas of hypocellularity, suggesting that lack of biodegrading macrophages in these areas may contribute to poor desorption of fibrous scars¹⁸.

The role of myofibroblasts in reversal of liver fibrosis

Although activated HSCs undergo senescence ²³ and apoptose ¹⁷ during the regression of liver fibrosis, the quantitative contribution aHSC apoptosis to disappearance of activated myofibroblasts remained unknown ²⁷. Meanwhile, the cellular population of HSCs is restored in mice recovering from fibrosis, and the source of these quiescent-like HSCs is unknown. Recent studies have suggested that in addition to apoptosis, activated HSCs/myofibroblasts can be eliminating by undergoing inactivation and reverting to a quiescent-like their phenotype. We¹, and subsequently others³⁰, have used genetic marking to demonstrate an alternative pathway in which myofibroblasts revert to a quiescent-like phenotype in CCl₄-induced liver injury and experimental alcoholic liver disease. These in vivo studies are in concordance with in vitro observations that suggested that cultured HSCs, at least in part, can reverse to a quiescent-like phenotype. The quiescent phenotype of HSCs is associated with expression of lipogenic genes and storage of vitamin A in lipid droplets. Depletion of peroxisome proliferator-activated receptor gamma (PPAR-γ) constitutes a key molecular events for HSC activation, and ectopic expression of this nuclear receptor results in the phenotypic reversal of activated HSC to quiescent cells in culture ³¹. The treatment of activated HSC with an adipocyte differentiation cocktail or over-expression of SREBP-1c results in up-regulation of adipogenic transcription factors and causes morphologic and biochemical reversal of activated HSC to quiescent-like cells ^21,32,33. These in vitro and in vivo studies have provided new insights into the concept of reversibility of liver fibrosis, suggesting that disappearance of activated myofibroblasts is attributed not only to their apoptosis but also to reversal of their phenotype into a quiescent-like.

Methods to study inactivation of HSCs (iHSCs)⁸

The Cre-lox-system³⁴ provides a unique tool to monitor specific cellular populations and their progeny in mice ³⁵. This system is based on the ability of the small 38-kDa bacteriophage protein Cre-recombinase (Cre) to recognize and excise inverted 13 base pair loxP sequences and the DNA it flanks ^34,36. Genetic labeling of a specific cellular population is achieved by crossing mice expressing Cre under control of a cell-specific promoter with reporter Rosa26^f/f-YFP mice, ubiquitously expressing the yfp gene in which transcription is blocked by a floxed Stop cassette ³⁶. Genetic Cre-loxP recombination causes excision of the floxed-Stop-floxed sequence from genomic DNA with activation of YFP transcription in the resulting offspring. The cells of interest now irreversibly express YFP so that their phenotypic changes can be monitored in response to injury or stress ³⁷.

Using Cre-LoxP-based genetic labeling of myofibroblasts, we¹ and the others³⁰ elucidated the fate of activated (a)HSCs/myofibroblasts during recovery from CCl₄-induced liver fibrosis (Figure 1A)⁸. Genetic labeling of aHSC/myofibroblasts resulted from crossing mice expressing Cre under control of the collagen-α1(I) enhancer/promoter (Colα1(I)^Cre mice) with reporter mice (Rosa26^f/f-YFP mice). In the offspring (Colα1(I)^Cre-YFP mice), all collagen Type I-expressing cells expressed YFP. In uninjured mice, collagen Type I is not expressed in the liver, and this correlates with minimal expression of YFP in livers of Colα1(I)^Cre-YFP mice. Following induction of liver injury in these mice, aHSCs and their progeny express Collagen-driven YFP and are permanently labeled by YFP expression¹. Phenotypical changes of aHSCs and the mechanism of their inactivation can now be studied during regression of liver fibrosis (Figure 1B)⁸. Using two models of hepatotoxic-induced liver fibrosis (carbon-tertrachloride (CCl₄) and intragastric alcohol feeding), we demonstrated that half of the myofibroblasts escape apoptosis during regression of liver fibrosis, downregulate fibrogenic genes (Col1a1, Col1a2, SMA, TIMP1, TGFβ-RI) and acquire a phenotype similar to, but distinct from, quiescent HSCs¹. Similar results were obtained using inducible Cre-based systems, in which genetic labeling of aHSCs was achieved in Collagen-α1(I)-ER-Cre and Vimentin-ER-Cre upon tamoxifen administration.

Figure 1.

Study design to determine the cell fate of aHSCs during regression of CCl₄-induced liver fibrosis. Adapted from Kisseleva and Brenner [8]. A. Cre-loxP based genetic labeling marks the fate of Col-α1(I) expressing aHSCs/ myofibroblasts in Col-α1(I)^Cre-YFP mice generated by crossing Col-α1(I)^Cre and Rosa26^f/f-YFP mice. B. Col-α1(I)^Cre-YFP mice were subjected to CCl₄-induced liver injury (1.5 months), then recuperated upon cessation of injuring agent (for 1 month). Mice were sacrificed and livers were analyzed for the presence of Vitamin A⁺ YFP⁺ and Vitamin A⁺ YFP⁻ HSCs. C. CCl₄ induces qHSC activation into aHSCs/myofibroblasts in Col-α1(I)^Cre-YFP mice. After CCl₄ withdrawal, some aHSCs apoptose while some inactivate (YFP⁺ iHSCs number <100% of aHSCs)⁶.

Generation of these novel transgenic mice provided a unique tool to study the fate of hepatic myofibroblasts in fibrotic liver. In addition, several studies have utilized glial fibrillar acidic protein (GFAP)-Cre mice to successfully target HSCs in the liver. Although these GFAP-Cre mice do not discriminate between quiescent, activated and inactivated HSC phenotypes, they were successfully used to label HSCs for quantification purposes or for HSC specific gene deletion. Since HSCs and astrocytes share expression of several neural markers, including GFAP, deletion of this gene in the brain might affect the phenotype in the liver.

Characterization of novel inactivated HSC phenotype

Inactivated HSCs acquire a novel phenotype which has not been previously described and is now designated iHSCs (Figure 1C)⁸. In particular, iHSCs more rapidly reactivate into myofibroblasts in response to fibrogenic stimuli and more effectively contribute to liver fibrosis. Inactivated HSCs downregulate SMA, Col1a1, Col1a2, TIMP1, TGFβ-RI and obtain features of quiescent-like HSCs due to upregulation of PPAR-γ and Bambi¹. Meanwhile, other quiescent-associated genes such as GFAP, Adipor1, Adpf, and Dbp are not re-expressed in iHSCs, suggesting that despite the similarities between qHSCs and iHSCs and their lack of fibrogenic gene expression, iHSCs possess properties distinct from qHSCs.

Inactivation of HSCs is associated with re-expression of lipogenic genes PPAR-γ, Insig1, and CREBP³¹. Our findings in mice in vivo support previous in vitro studies demonstrating the importance of PPAR-γ for maintaining and re-establishing the quiescent phenotype (qHSCs)^31,33. Based on comparison of the global gene expression in qHSCs, aHSCs and iHSCs, we have identified several genes that are differentially expressed in HSCs depending on their stage of activation, and can be used to distinguish iHSCs from qHSCs and aHSCs¹. This is of a particular importance for identifying inactivation of human HSCs. Our strategy is based on identification of genes similarly and differentially expressed in qHSCs, aHSCs and iHSCs, and that can be easily detected in a pool of HSC by flow cytometry. Differential expression of cell surface antigens by different HSC types is listed below. Since qHSCs and iHSCs share many features, detection of surface markers may be compared in correlation with induction of phenotype specific transcription factors, or other proteins.

Comparative analysis of phenotype-specific HSC signature genes may provide further insight in inactivation of HSCs in vivo (Figure 2). Specifically, we identified 7 qHSC-specific genes (that are expressed only in qHSCs), 12 aHSC-specific genes (expressed only in aHSCs), 12 iHSC-specific genes (upregulated in iHSCs), and 12 genes which are upregulated in both qHSCs and I HSCs (and therefore are associated with non-fibrogenic properties of HSCs). Furthermore, the following genes Nr1d2, Adipor1, IL17ra, Itga5, Egfr, Crlf1, IL1r1, Rnd1, Csf1R, IL7R, Cntfr, Csf2rb, Cx3cr1, IL10RA, Ly86, CD36, Mrc1, Agtr1a, Calcr1, Grap, and Ngfr are cell surface receptors, so that their expression can be detected using flow cytometry (which can simultaneously detect expression of up to 11–12 surface antigens in a single sample) allowing us to confidently identify and isolate live iHSCs. Furthermore, we identified several phenotype-specific transcription factors expressed in HSCs dependent on the stage of activation/inactivation.

Figure 2.

Differential expression of cell surface antigens by different HSC types. Based on the whole genome microarray, we have identified mRNAs that are specifically upregulated in qHSCs: Nr1d2; Adipor1 - Adiponectin receptor 1, Adfp - Adipose differentiation-related protein, GFAP - Glial fibrillary acidic protein, Dbp - D site of albumin promoter (albumin D-box) binding protein, Prei4; Foxj1 - Forkhead box protein J1. The mRNAs were specifically upregulated in aHSCs: αSMA - Alpha-actin-2 (also known as actin, aortic smooth muscle or alpha smooth muscle actin); Col1 1; TIMP-1 - tissue inhibitor of metalloproteinase 1, Crlf1 – cytokine receptor-like factor 1; IL1r1 - Interleukin 1 receptor, type I (IL1R1, CD121a), Itga5 - Integrin alpha-5; Spp1 - secreted phosphoprotein 1 (Osteopontin (OPN), also known as bone sialoprotein I (BSP-1 or BNSP), early T-lymphocyte activation, ETA-1); Lox - lysyl oxidase; LoxL2 - lysyl oxidase-like 1; IL-17ra - Interleukin 17 receptor A; Fosl1 - fos-like antigen 1; Folr1 - folate receptor 1. The following mRNA s were upregulated in iHSCs: Rnd1 - Rho family GTPase 1; Csf1R - colony stimulating factor 1 receptor; IL7R - interleukin-7 receptor; Cntfr - ciliary neurotrophic factor receptor; Csf2rb - colony stimulating factor 2 receptor, beta, low-affinity (granulocyte-macrophage); Cx3cr1 - CX3C chemokine receptor 1 (fractalkine receptor or G-protein coupled receptor 13, GPR13); Ly86 - lymphocyte antigen 86 (CD180/MD-1), Cxcl2 - chemokine (C-X-C motif) ligand 2 (Cxcl2); Fos - FBJ osteosarcoma oncogene; Egr1 - epidermal growth factor receptor (EGFR; ErbB-1; HER1 in humans); Spic - Spi-C transcription factor (Spi-1/PU.1 related); Atf3 - activating transcription factor 3. Several mRNAs were upregulated both in qHSCs and iHSCs: Mrc1 - mannose receptor 1, Agtr1a - Angiotensin II receptor, type 1 (or AT₁ receptor); Calcrl - calcitonin receptor-like; Grap - GRB2-related adapter protein, Ngfr - The Low-Affinity Nerve Growth Factor Receptor (also called the LNGFR or p75 neurotrophin receptor); Insig1 - insulin induced gene 1; PPARγ-peroxisome proliferator activated receptor gamma; CD36; IL-10RA - interleukin 10 receptor, alpha; Foxo1 - forkhead box O1; Tal1 - helix-loop-helix protein; Bambi - BMP and activin membrane-bound inhibitor, homolog (Xenopus laevis).

Conclusions

Inactivation of myofibroblasts during reversal of fibrosis opens new prospects for therapy. Hepatic fibrosis is reversible in patients and in experimental models with decreased fibrous scar and disappearance of the myofibroblast population. However, the fate of the myofibroblasts in patients with liver fibrosis is unknown. Although some myofibroblasts undergo cell death¹⁷, an alternative untested hypothesis is that the myofibroblasts revert to their original quiescent phenotype or obtain a new phenotype. Understanding of the origin and biology of fibrogenic myofibroblasts will provide a new target for anti-fibrotic therapy.²¹

Abbreviations

HSCs

hepatic stellate cells

qHSCs

quiescent HSCs

aHSCs

activated HSCs

iHSCs

inactivated HSCs

CCl₄

carbon tetrachloride

α-SMA

α-smooth muscle actin

Col-α2(I)

Collagen-α2(I)

Col-α1(I)

Collagen-α1(I)

Col-α1(1)^Cre-YFP mice

Col-α1(I)^Cre mice x Rosa26^{flox-Stop-flox-YFP} mice