Numerous experimental and clinical studies have demonstrated a central role of TGF-β, and in particular β1, in liver fibrosis 
. Studies using engineered TGF-β soluble receptors or siRNA interference technique have directly confirmed a causative role of this cytokine in liver fibrosis 
. While the evidence linking TGF-β and fibrogenesis is substantial and raises possibilities for targeting this molecule as an appropriate therapeutic agent, it is necessary to demonstrate whether TGF-β antagonism can effectively treat pre-existing hepatic fibrosis. In the present study, we treated healthy rats with TAA for 8 weeks to establish hepatic fibrosis, which was histologically close to that seen in human liver cirrhosis. With fibrosis established, we examined the effect of administering a TGF-β neutralizing antibody, 1D11, on further fibrosis progression, and also whether there was evidence of regression from pre-existing fibrosis. After TAA withdrawal both control groups showed further deterioration of liver architecture mostly in the form of nodular damage and fibrous foci or septa bridging from portal area to portal area and to central areas. Also noted was substantial evolution of proliferating biliary epithelial neoplasia (CAA) within fibrous foci, consistent with earlier reports of TAA mediated hepatic lesions 
. The degree of TAA-induced lesions correlated with the expression of TGF-β mRNA in the liver. TGF-β1, β2 and β3 were significantly upregulated after 8 weeks of TAA administration (6-fold increase for β1), and remained high in rats that received PBS or 13C4, except for TGF-β3, which declined but was still higher than normal controls upon the cessation of TAA administration.
Treatment with 1D11 reduced transcription of TGF-β1whereas the other two isoforms were unchanged, most likely due to their intracellular location in biliary epithelia. Reduced TGF-β1 expression and apparent interruption of the TGF-β1 autocine regulatory loop likely explain the therapeutic impact on pre-existing hepatic fibrosis. Histologically, a profound improvement in hepatic architecture with reduced number and area of fibrous foci or septa was seen in rats dosed with 1D11 for 8 weeks. Morphometric analysis, along with the measurement of hydroxyproline content, showed a significant decrease in collagen deposition in animals treated with 1D11. Of note, the two control groups showed progressive hepatic injury 4 and 8 weeks following cessation of TAA-induced injury, whereas the 1D11 treated group showed regression of fibrosis, significantly lower than levels recorded just before starting 1D11 treatment. Thus in this model there was no evidence of spontaneous resolution of injury, unlike other models 
and, further, our data argue that antibody-mediated antagonism of TGF-β promotes regression of tissue fibrosis. We conclude that pathological hepatic fibrosis, at least in this preclinical model, is a dynamic situation and is potentially reversible. This may also be true in the human since recent studies examining hepatitis patients with long term antiviral therapy showed reduced liver cirrhosis 
. Our results provided a platform for further mechanistic studies focusing on the reversibility of pathological fibrosis.
Accumulation of ECM proteins is caused not only by increased protein synthesis but also by decreased protein degradation. There is now a significant body of evidence that liver fibrosis is also a consequence of altered ECM degradation, in which PAI-1 may be implicated 
. PAI-1 is the major physiologic inhibitor of tissue-type (t-PA) or urokinase-type plasminogen activator (µ-PA), both of which activate plasminogen to plasmin. Plasmin may also activate unrelated proteases to further promote resolution of fibrosis 
. In the present study, a marked increase of PAI-1 protein occurred in parallel with fibrosis progression. PAI-1 remained elevated in rats treated with PBS or 13C4, but this was normalized in rats dosed with a TGF-β neutralizing antibody. Thus our data suggest that elevated PAI-1 contributed to the development of hepatic fibrosis through a TGF-β dependent manner. These data are consistent with several recent studies showing that; 1) direct manipulation of plasmin/PAI-1 with adenovirus-mediated transfer of siRNA decreased hepatic fibrosis in dimethylnitrosamine-induced and bile duct ligation-induced liver disease models 
; 2) restoration of hepatic plasmin activity by a mutant, noninhibitory PAI-1 was also associated with decreased fibrotic matrix accumulation in this model 
and, 3) higher levels of tPA activation have been associated with fewer hepatic lesions in PAI-1 (−/−) mice 
. Reduced PAI-1 protein, along with attenuated TGF-β1 transcription may indicate an inhibition of TGF-β1/Smad signaling axis by 1D11, presumably with decreased phosphorylation of Smad2/3, a principal effector system of the signaling pathway contributing to fibrosis 
It is also known that TAA is a potent carcinogen and induces CCA in the rat, which recapitulates the histological features and progression of human CCA 
. In the present study, TAA induced hepatic fibrosis, concomitant with development of neoplastic biliary epithelial ductules embedded in fibrotic/stromal tissues. The nature of this type of neoplastic biliary ductules was confirmed morphologically as a typical “intestinal metaplasia”-like appearance, identified by immunostaining of cytokeratins 
, and microscopic diagnosis by a board certified pathologist. The assessment of TAA-induced CCA by immunostaining also provided a way to determine the area of the neoplastic biliary ductules. Neoplastic bile ductules were remarkably fewer in number and in area in livers from rats dosed with 1D11. The diminishment of CCA occurred in parallel with an improved fibrosis index, suggesting a close relationship between these two. Indeed, elaboration of ECM proteins provides both a three-dimensional structure as well as matrix-cellular signals that promote tumorigenesis 
. With specific regard to CCA, Farazi PA et al. 
have shown that increased production of type I and III collagens along with fibroblast recruitment stimulate biliary epithelium hyperplasia and subsequent progression to malignant intrahepatic tumors mice harboring a p53 mutant allele. Thus our findings are consistent with others’ observations suggesting that therapeutic reduction of hepatic fibrosis as a result of TGF-β neutralization represents a potential approach for the treatment of CCA.
To summarize, the present study demonstrates that 1D11, a murine monoclonal TGF-β neutralizing antibody, can reverse pre-existing clinically comparable hepatic fibrosis induced by an extended dosing of TAA. The regression of fibrosis was accompanied with a diminishment of cholangiocarcinoma, presumably linked to a reduction of fibrosis surrounding neoplastic ductules, associated with an inhibition of TGF-β and PAI-1. The data provide the first evidence that reversal of pre-existing hepatic fibrosis can be achieved upon TGF-β neutralization, providing proof-of-concept for considering TGF-β neutralizing antibody in the treatment of liver cirrhosis.