Introduction

Fibrosis is a progressive disease characterised by the accumulation of scarring extracellular matrix proteins, which disrupt normal tissue architecture. As fibrosis worsens, tissue function is impaired and in vital organs such as the liver—in the absence of transplantation—morbidity and mortality results [1].

Fibrosis is caused by an expansion and an increase in the activity of fibroblast populations within the liver, which—in a chronic damage setting—secrete and promote the accumulation of extracellular matrix proteins [1]. Initial work identified quiescent hepatic stellate cells as a source of cells that cause fibrosis [2], through their trans-differentiation to myofibroblasts (classically identified using α-smooth muscle actin, Fig. 1). However, more recently it has become clear that a distinct population of fibroblasts expand from the portal tract (periductal and portal), particularly when the region of damage is located to this part of the lobule (e.g. in response to cholestasis) [3]. In addition, it has emerged that both these cell populations originate, at least in part during chronic liver injury, from the bone marrow [4, 5] and that fibrogenic cells may even be generated through epithelial-mesenchyme transition [6, 7].

Fig. 1
figure 1

Histology of the cirrhotic human liver. Serial liver sections from a liver biopsy were stained with (a) haematoxylin and eosin. (b) immunochemically with an antibody to α-smooth muscle actin [11], brown staining indicates liver myofibroblasts. (c) Sirius red and haematoxylin [11], red staining indicates extracellular matrix deposition

Full size image

Current and future therapies

At present, there are no therapies indicated for use as anti-fibrogenics in the liver. A major hurdle to the development of drugs for liver fibrosis is the requirement for prolonged treatment times to demonstrate efficacy in clinical trials. Trial length obviously imposes greater cost but an additional disincentive exists that there is more chance for adverse events to show during the trial. Failure in clinical drug trials has major implications for pharmaceutical companies and essentially dictates conservative investment decisions (particularly so in the case of liver disease, a high mortality area and an exclusion criterion for most therapeutic trials). A limited number of clinical trials are currently under way (for review see [1]); in many cases, however, existing drugs licensed for other indications (eg irbesartan and warfarin [1]) have been used. Novel compounds require pre-clinical and clinical safety testing and regulatory authorities require that any new drug must significantly benefit patients. There are fewer drugs emerging to the point of clinical trial as potential anti-fibrogenics. One example, however, is the farnesoid X receptor (FXR) activator 6-ethyl chenodeoxycholic acid (INT-747). This novel compound is known to inhibit both the trans-differentiation of human hepatic stellate cells to myofibroblasts in vitro and the development of fibrosis in animal models [8]. INT-747 is currently undergoing safety testing in persons with type 2 diabetes with presumed fatty liver for safety and tolerability [1].

In vitro models of liver fibrosis (the spontaneous trans-differentiation of hepatic stellate cells to myofibroblasts in culture [1]) and a range of in vivo animal models have aided the identification of potential therapeutic approaches for the treatment of liver fibrosis. Figure 2 illustrates the major parameters affecting myofibroblast activity and Table 1 lists some of the drug targets for therapeutic intervention.

Fig. 2
figure 2

Schematic diagram outlining the major parameters affecting the severity of liver fibrosis and the major points at which potential anti-fibrogenic therapeutics intervene. Numbers refer to numbering for “mode of action” as given in Table 1. Inhibition, T; promoters, ↑

Full size image
Table 1 The major drug targets for potential anti-fibrogenic drugs
Full size table

A striking diversity of signalling pathways is available for pharmacological intervention in myofibroblasts. However, since recovery from fibrosis is accompanied by the apoptosis of liver myofibroblasts [9], this laboratory has concentrated on studying the effects of stimulating the apoptosis of fibrogenic cells in the liver as a potential therapeutic approach for liver fibrosis.

The fungal metabolite gliotoxin had originally been shown to stimulate the apoptosis of leucocytes [10]. Gliotoxin also stimulated the apoptosis of rat and human hepatic stellate cells and liver myofibroblasts in vitro [11, 12]. The effects of gliotoxin in an animal model of liver fibrosis recovery (i.e. gliotoxin was administered after the final liver insult and the effects on the rate of fibrosis reversal were determined) indicated for the first time that a promotion of myofibroblast apoptosis enhanced the resolution of fibrosis [11]. However, directing cell death as a mode of treatment for a disease comes with risk because most therapeutics are unlikely to act specifically on the target cells alone. The risk is therefore the stimulation of cell death in cells other than the target cells, with potential adverse effects.

Targeting myofibroblasts

A number of targeting strategies are being evaluated for their ability to selectively deliver therapeutics to liver myofibroblasts. These include modifying drug-carrying proteins with mannose-6-phosphate for uptake via the insulin-like growth factor II receptor [13], and with peptides that recognise the collagen type VI [14] or platelet-derived growth factor [15] receptors.

This article reviews the generation of a human recombinant single-chain antibody (scAb—termed C1-3 [16]) to an extracellularly exposed region of synaptophysin, a plasma membrane protein expressed in the liver, specifically in human and rodent myofibroblasts [17]. Synaptophysin is primarily associated with neural tissue and is thought to be involved in synaptic vesicle exo-endocytosis. However, synaptophysin knockout mice show no phenotype or detectable effects on synaptic transmission [18]. An advantage of using synaptophysin as a target to deliver therapeutics is that this protein forms part of an endocytosing vesicle, increasing the likelihood for significant scAb and conjugated therapeutic uptake in myofibroblasts.

Phage display

Phage display is a powerful technique for selecting and cloning sequences that encode proteins with a native affinity for a particular ligand [19, 20]. In the case of C1-3, a phage library was employed in which potentially the entire human repertoire of variable heavy and variable light chains (which encode antibody antigen-binding sites) were cloned as single fused domains into the pIII M13 bacteriophage coat protein gene (Fig. 3a). Each phage “displays” a variable domain on its surface, with typically > 107 different clones comprising a phage display library. Variable domains that interact with a particular antigen (for C1-3, the antigen is the peptide 2 sequence—see Fig. 4) can be selected by a process of “panning”, whereby the phage is exposed to an immobilised target sequence. Unbound phage is removed by washing and bound phage is eluted and amplified through re-infection of Escherichia coli (Fig. 3b). The process of panning is repeated several times allowing for the isolation of specific high-affinity antigen-binding phage. Single clones are then isolated and analysed prior to sub-cloning heavy and light chain-encoding regions into an expression vector that generates a single-chain protein (i.e. scAb) fused in frame with other domains that facilitate detection and purification. Using established procedures, such as protein tagging, high levels of pure monoclonal scAb can be generated with relative ease.

Fig. 3
figure 3

(a) Schematic diagram of recombinant M13 bacteriophage incorporating an scAb within its coat protein. (b) Schematic diagram outlining the procedure of phage display

Full size image
Fig. 4
figure 4

Schematic diagram of synaptophysin. Each circle represents an amino acid, with membrane-spanning residues determined using TMpred software [16]. Orange residues correspond to the antigen site for C1-3

Full size image

C1-3 scAb

A range of human recombinant scAbs were generated to synaptophysin by phage display [16]. One of these, C1-3, was selected and its ability to act as a targeting agent for liver myofibroblasts was examined. Fluorescently labelled C1-3 scAb avidly bound to human myofibroblasts (Fig. 5) but not human hepatocytes [16]. The C1-3 was taken up into myofibroblasts through pinocytosis and alone was not toxic to myofibroblasts in vitro, but when conjugated with tributyl tin, the scAb directed the toxin to myofibroblasts (with toxin activity retained) [16].

Fig. 5
figure 5

Human liver myofibroblasts bind FITC-labelled C1-3 scAb (green). Hepatic stellate cells were isolated by pronase and collagenase perfusion from resected human liver and were cultured as outlined [11]. After transdifferentiation and sub-culture, cells were typically more than 95% positive for the classic myofibroblast marker α-smooth muscle actin [1]. The panel shows a typical view of liver myofibroblasts after incubation with FITC-labelled C1-3 (green) in culture as outlined [16], followed by fixation and co-staining for the myofibroblasts marker α-smooth muscle actin (red) and DNA using DAPI (blue)

Full size image

Mice with liver fibrosis (via carbon tetrachloride administration) were injected intraperitoneally with C1-3 scAb to determine its effectiveness as a drug-targeting agent in vivo [21]. Initial studies used fluorescently labelled C1-3 to examine its distribution throughout the body up to 24 h after injection. The results of these studies showed that the C1-3 scAb appeared in the serum within 20 min and was eliminated with an approximate half life of 2 h [21]. The C1-3 scAb was detectable in liver homogenate but undetectable in brain, muscle or spleen [14]. Immunohistochemical analysis indicated that the C1-3 scAb localised to the liver, in regions where scars were present. There was minimal immunohistochemical evidence for the presence of C1-3 scAb in non-fibrotic liver [21]. Co-staining of fibrotic liver sections showed that the C1-3 scAb colocalised with myofibroblast α-smooth muscle actin, but not the monocyte and macrophage marker F4/80 [21]. These data indicate that the C1-3 scAb readily and selectively targeted liver myofibroblasts in an animal model of liver fibrosis.

To determine whether C1-3 could deliver a functional experimental anti-fibrogenic therapeutic agent to myofibroblasts, it was chemically conjugated with gliotoxin. Free gliotoxin has been shown in previous work to stimulate the apoptosis of liver myofibroblasts in vitro and in vivo and enhance recovery from liver fibrosis in vivo [11, 12, 22]. However, gliotoxin also causes—although to a lesser degree—the apoptosis of hepatocytes and stimulates Kupffer cell death [4, 17, 23, 24].

Gliotoxin was chemically conjugated to either a C1-3 scAb or a control CSBD9 scAb (selected for its ability to bind to an irrelevant ligand) using N-[p-maleimidophenyl] isocyanate and S-acetyl thioglycolic acid N-hydroxysuccinimide [21] (for details also see online data: http://dx.doi.org/10.1016/j.jhep.2008.01.032). Conjugation did not significantly alter scAb affinities for their respective antigens (as determined using antigen ELISA and BIAcore [21]) or ablate gliotoxin’s ability to cause liver myofibroblast apoptosis (see Figs. 6 and 7).

Fig. 6
figure 6

Time course for the effects of free gliotoxin or C1-3-gliotoxin (C1-3-GT) on sub-stratum adherence in vitro. Human myofibroblasts (culture-activated hepatic stellate cells [11]) were sub-cultured into 24-well plates in 300 μl of medium and treated with either free gliotoxin added from a 1,000-fold molar concentrated stock in DMSO vehicle (total 450 pmoles/well); 4.5 μg C1-3/well or 4.5 μg C1-3-gliotoxin scAb (to give approximately 450 pmoles gliotoxin/well for conjugated scAb). Data are the mean and standard deviation of 3 separate human cell preparations. Right panels, photomicrographs of typical views of cells at the indicated treatment after 5 h

Full size image
Fig. 7
figure 7

Time course for the effects of C1-3-GT on active caspase 3 levels in human myofibroblasts in vitro. Human myofibroblasts were cultured and treated with C1-3-GT essentially as outlined in Figure 5. At the indicated time points, cells were fixed and stained for α-smooth muscle actin (red), active caspase 3 (green) and DNA using DAPI (blue). The active caspase 3 antibody was purchased from Promega (Southampton, UK). Results are typical of cells isolated from 3 individual patients

Full size image

Mice were administered carbon tetrachloride twice weekly for 8 weeks to generate liver fibrosis. During the last week prior to the final injection of carbon tetrachloride, mice were treated with a single dose of C1-3 scAb or C1-3-gliotoxin scAb. The mice were therefore treated with a potential therapeutic agent and injury was sustained by further treatment with carbon tetrachloride. Fibrotic mice were also treated with an equivalent dose of free gliotoxin or gliotoxin conjugated to CSBD9 (i.e. CSBD9-GT).

The results are summarised in Table 2 and show that free gliotoxin reduced the number of myofibroblasts but did not reduce fibrosis severity in the liver. C1-3-gliotoxin more potently reduced the number of myofibroblasts present than the same dose of free gliotoxin and in addition significantly reduced the severity of fibrosis in the liver. C1-3 scAb, CSBD9 scAb and CSBD9-GT had no effect on the number of liver myofibroblasts or fibrosis severity. Interestingly, free gliotoxin reduced the number of F4/80 positive cells (i.e. monocytes and macrophages/Kupffer cells) and matrix metalloproteinase 13 levels in the liver, whereas C1-3-gliotoxin did not. Macrophage depletion abrogates recovery from liver fibrosis [25]. Retaining macrophages in C1-3-gliotoxin-treated animals may account for the improved anti-fibrogenic effect of gliotoxin when targeted to liver myofibroblasts with C1-3.

Table 2 The effects of gliotoxin and C1-3-gliotoxin on parameters of liver fibrosis in a sustained carbon tetrachloride model of liver fibrosis [21]
Full size table

Diagnostic and therapeutic perspectives

Two major hurdles must be overcome to advance successful treatments for fibrosis. The first is the effective assessment of fibrosis in both experimental and clinical studies. It remains problematic to accurately screen for fibrosis using a battery of serum markers, particularly for those at early and intermediate stages of disease [26]. Biopsy remains the “gold standard” diagnostic tool for liver disease [27] but has significant drawbacks (the procedure is risky; repeated biopsy is inadvisable; and it does not give a globally accurate picture of organ fibrosis). The second hurdle is specificity and efficacy of potential treatments (in the liver, the myofibroblasts lie adjacent to the major drug metabolising cell of the body, the hepatocyte).

This laboratory has examined the potential of an antibody-based approach to tackle both imaging and therapeutic delivery in fibrosis. Although the C1-3 scAb does not directly bind to fibrosis proteins (it binds to the major fibrogenic cell), an image-visible scAb derivative may be effective in providing an indication of fibrosis severity. If successful as an imaging agent, experimental and clinical imaging studies should aid in its development as a targeting agent for anti-fibrogenic therapeutics.

The utility or therapeutic efficacy of whole antibody, or related antibody structures, in vivo is influenced by a number of competing factors. These include specificity, affinity (including avidity) and serum half life. The C1-3 has already been expressed successfully as a monomeric scAb (37 kDa). In addition, it could also be used in a slightly smaller but related scFv formulation (25 kDa) or expressed as a dimeric fragment (60 kDa, which will typically show one order of magnitude improvement in affinity for antigen through increased avidity). Finally, the C1-3 could be expressed as a full (150 kDa) mAb incorporating either a mouse or a human Fc region. This construct will have a similar affinity to the C1-3 dimer but an increased serum half life and the capacity to initiate cell killing upon binding in vivo.

Typically, for imaging agents, a shorter serum half life is favoured, to maximise signal-to-noise ratios in the diseased tissue. For therapeutic applications, a whole antibody is often preferred where antibody binding and presentation of the Fc region can recruit components of the mammalian immune system and induce cell killing. However, for certain applications, a smaller antibody fragment conjugated to a cell-killing agent may be a more potent therapeutic formulation. This type of agent can provide a powerful alternative to Fc-induced cell death, offering greater tissue penetration and rapid clearance of toxic drugs from non-target tissues.

The administration of a recombinant antibody specific for the surface of liver myofibroblasts could be used to develop a safer, less invasive and more effective diagnostic tool for the assessment of fibrosis throughout the entire organ. In the short term, this could lead to more refined experimental animal studies; provide a more effective assessment of potential anti-fibrogenics; and reduce the number of animals required in pre-clinical studies. Experimental animal studies will also contribute to the translation of the diagnostic to the clinical reality in the long term, as part of in vivo imaging (diagnostic and diseases management). In addition, the C1-3 antibody could be used to direct anti-fibrogenics to the myofibroblasts (or analogous pro-fibrogenic cells in other tissues). Directing therapeutics specifically to the target myofibroblast—and away from other cells—will enhance the efficacy of an anti-fibrogenic since it will reduce the amount of therapeutic required and the likelihood of adverse effects. However, a drawback of using C1-3 is the potential effect on any cells that also express synaptophysin, particularly when the conjugated therapeutic mediates cell death. The major concern would therefore be neural and neuroendocrine cells. Although there is no evidence that the C1-3 scAb crossed the blood-brain barrier in animal studies [21], imaging studies should confirm whether this is the case. However, it should be borne in mind that there is currently no treatments available for use as anti-fibrotic in patients [28]. The major causes of liver fibrosis in the developed world are non-alcoholic steatohepatitis (NASH), hepatitis C infection and alcoholism. NASH has an incidence of at least 2% in the European Union and United States and the incidence of Hepatitis C infection is between 1% and 2% [29, 30]. The incidence of both of these diseases is predicted to rise in the future and further reinforces the urgent need to diagnose, manage and treat this “silent killer”. It may be the case with some patients with cirrhosis—in whom there are no therapeutic options available—that an experimental and more risky approach may be needed in order to find effective treatments.