Introduction

Transforming growth factor-beta (TGF-β) is a key regulator of liver physiology and pathology, contributing to all stages of disease progression, from initial liver injury through inflammation and fibrosis to cirrhosis and hepatocellular carcinoma (HCC) [1]. The liver is a unique organ, exhibiting high regenerative capacity crucial for homeostasis and tissue repair. Strong evidence points to a crucial role of TGF-β as cytostatic and apoptotic in hepatocytes, which is critical for the control of liver mass [2], so that loss of TGF-β activities results in hyperproliferative disorders and cancer [3]. However, in addition to the well-known TGF-β suppressor effect during the early stages of tumour development, there is strong evidence that TGF-β might later contribute to tumour progression when cells acquire resistance to its suppressive effects, conferring cell survival, inducing cell migration and invasion, mediating immune suppression and microenvironment modification [4, 5]. Targeting the TGF-β signalling pathway has been explored to inhibit liver disease progression [6]. On the basis of recent literature, here we summarize the crucial, and sometimes controversial, roles of TGF-β in liver physiology and pathology.

TGF-β signalling

TGF-β signals its multifunctional effects on liver cells by binding to single transmembrane type I and type II receptors (i.e. TβRI and TβRII) that are endowed with serine threonine kinase activity. Upon ligand-induced heteromeric complex formation, the type II kinase trans-phosphorylates the type I receptor at specific serine and threonine residues in the juxtamembrane glycine–serine domain. This activates the type I kinase, upon which the extracellular signal is transmitted across the membrane, and intracellular signalling is initiated by the phosphorylation of specific proteins, among which the receptor-regulated (R)-SMADs, i.e. SMAD2 and SMAD3, have a prominent role (Fig. 1). Activated SMAD2 and SMAD3, can partner with common mediator SMAD4, and these heteromeric complexes can be translocated to the nucleus and function as transcription factors that regulate specific gene transcriptional responses (TGF-β canonical pathway) [7]. SMAD4 is a critical effector of intracellular signalling. Like TGF-β, SMAD4 has been associated with tumour suppression, but also tumour promotion in HCC [8]. SMADs act together with coactivators and corepressors to regulate gene expression. Recently, orphan receptor NR4A1 was found to act as an endogenous inhibitor of the TGF-β pathway, by recruiting a repressor complex to TGF-β target genes, thereby antagonizing TGF-β-induced liver (and other tissues) fibrosis [9].

New mechanisms by which TGF-β exerts its cellular effects by changing genomic responses keep on being discovered. TGF-β was recently shown to induce genome-wide changes in DNA methylation, thereby enabling stable changes in liver cancer cell subpopulations [10]. Activation of long non-coding RNA–ATB by TGF-β in HCC has a powerful effector role in mediating invasion and metastasis [11]. In addition, repression of miR-122 in hepatic stellate cells (HSC) by TGF-β is important for its profibrotic response on these cells [12].

Of note, to function as transcription factors, SMADs need to interact with other DNA-binding transcription factors. SMAD2 cannot bind directly to DNA, and the affinity of SMAD3 for DNA is weak. The highly contextual actions of TGF-β on different cell types are in part explained by the many SMAD-interacting transcription factors that have been identified, which themselves are often regulated by extracellular cues [7]. This provides a mechanistic basis by which, SMAD activity can be carefully controlled. In addition, SMADs are regulated by various reversible post-translational modifications, including phosphorylation, ubiquitination, SUMOylation and poly(ADP-ribosyl)ation [13]. The picture becomes more complicated with a variety of phospho-SMAD isoforms, which are formed by phosphorylation at terminal carboxyl groups, at the intermediate linker region or at both, depending on the surrounding microenvironment and available growth factors, which then mediate differential roles for TGF-β during acute and chronic liver injury [14, 15]. Similar modes of positive and negative regulation occur for each step in the TGF-β signalling pathway, thereby creating numerous opportunities for cross-talk with other signalling pathways. Recent examples of such interplay are the stimulatory effect of the tyrosine kinase Axl/14-3-3ζ axis on TGF-β/SMAD3-driven invasion and metastasis of HCC [16]. In addition to the SMAD pathway, TGF-β receptors can also initiate so-called non-SMAD signalling responses in the liver, through interaction with other alternative pathways including MAP kinases, phosphatidylinositol-3-kinase (PI3K)/AKT, Ras and Rho-like small GTPases, among others [17, 18] (TGF-β noncanonical pathways) (Fig. 1). These pathways and their cross-talk with the SMAD pathway are also being investigated more and more in liver cells.

TGF-β is part of a larger family of 33 structurally and functionally related proteins, which also includes the activins and bone morphogenetic proteins (BMPs). Activin A and B, which are highly expressed in both acute and chronic inflammation, are emerging as important mediators of liver (and other tissues) fibrosis [19]. BMP9, which has a high and selective liver expression, is actively being investigated. It was recently shown to have pro-oncogenic effects on liver tumour cells; BMP9 stimulated the survival of liver cancer cells via activation of p38MAP kinase [20, 21]. However, their functions, and those of many other related family members, in chronic liver disease are not clear.

The TGF-β as a guardian of liver homeostasis

A very finely tuned TGF-β-driven signalling, in terms of both dosage and spatiotemporal activity, seems to orchestrate hepatic gene expression and be a critical determinant in the zonal specification and differentiation of hepatoblasts to either hepatocytes or cholangiocytes in the developing liver parenchyma (Fig. 2). TGF-β is also key in the control of liver architecture and biliary morphogenesis (reviewed in Ref. [2]). But TGF-β alone does not direct normal liver development. A hepatocyte growth factor (HGF)-mediated SMAD-independent pathway is able to rescue the liver phenotype in SMAD2/3 mutants [22]. This is not very surprising considering the well-known role of the HGF/c-MET pathway in liver development, but the exact interplay between the two pathways might deserve a bit more attention. Several other pathways that regulate biliary fate, namely the Wnt/β-catenin, the Notch and the Hippo-Yap signalling pathways, might interact directly with TGF-β. Hence, TGF-β/BMP signalling has been suggested to activate Notch signalling by regulating Notch signalling intermediates [23]. Functional approaches are missing and hopefully will be taken in the coming years to clarify the functional relationship and/or cross-talk between the TGF-β pathway and these other pathways in this context.

Although there is no doubt about the critical role of TGF-β signalling acting in collaboration with additional signals to regulate liver development, how is this fine-tune signalling determined is still under study. Several microRNAs (miRNAs) have been evidenced to regulate cell fate decisions by modulating TGF-β signalling. First, Rogler et al. [24] showed that SMADs are targets of the miR-23b cluster, whose function seems to be suppressing the expression of the biliary differentiation programme in hepatoblasts. More recently, miR-302b and miR-20a, two miRNAs abundantly expressed in endoderm, have also been shown to suppress TGF-β signalling by targeting Tgfbr2; and more importantly, forced expression of miR-302b during embryonic stem cell differentiation is associated with decreased expression of liver markers [25] (Fig. 2). In addition to miRNAs, other upstream regulators of TGF-β signalling are the SRY-related high mobility group box transcription factors (SOX) 9 and 4. They cooperate to control bile duct morphogenesis and seem to do it by acting as components of a biliary gene network modulating TGF-β, Notch and Hippo-Yap signalling, as evidenced by the fact that SOX4/SOX9 doubly mutant livers display alterations in the expression of the signalling mediators of these pathways, like TβRII, Hes1 and Tead2 [26, 27], although the impact on each of the cascades appears unequal. Strikingly, the TGF-β/SOX9 signalling axis could be bidirectional and operate also in adult liver stem/progenitor cells, because TGF-β-mediated induction of SOX9 in adult liver mesenchymal stem/progenitor cells is associated with decreased hepatocytic differentiation [28].

Another context in which TGF-β is a critical regulator is liver regeneration. The role of TGF-β as an important inhibitory signal in acute liver regeneration after partial hepatectomy has been explored extensively (reviewed in Ref. [2]) (Fig. 2). A reduced response to the cytostatic and cytotoxic effects of TGF-β in regenerating hepatocytes [29] provided a potential mechanism to promote hepatocyte regeneration. The resistant phenotype is associated with upregulation of antiapoptotic and antioxidant signals, but the underlying mechanism of such changes is still unknown. Interestingly, as in liver development, a perfect spatiotemporal orchestration of TGF-β signalling at different stages of the process is required, in this case, to allow hepatocyte proliferation at the inductive phase and an efficient termination of the regenerative response afterwards. Insights into the mechanism underlying this spatiotemporal control of TGF-β signalling are being provided. Indeed, the same miRNAs shown to regulate TGF-β signalling during liver development might regulate TGF-β1/SMAD3 activation in the termination stage of liver regeneration [30]. Moreover, an intricate cross-talk operating between different hepatic cell types might be essential. Hu et al. [31] have shown that liver sinusoidal endothelial cell-mediated production of angiopoietin-2, which in turn regulates TGF-β production and vascular endothelial growth factor receptor-2 expression, is essential to switch on or off hepatocyte and liver sinusoidal endothelial cell proliferation during the biphasic regenerative response. Together with this, a delicate interplay between TGF-β and tyrosine kinase receptors signalling, like that triggered by HGF or epidermal growth factor receptor (EGFR) ligands, is considered to be fundamental in the control of hepatocyte proliferation and apoptosis during regeneration (Fig. 2). How this cross-talk operates is not clear. On the one hand, reciprocal regulation of expression has been shown between HGF and TGF-β but certainly, cross-talk appears to be more complex and there are a number of candidate mediators. NADPH oxidase (NOX)4 could be one of them. We have demonstrated that it negatively controls hepatocyte proliferation and its expression is downregulated during regeneration after partial hepatectomy in mice [32]. Because NOX4 is known to be an important target of TGF-β that mediates its apoptotic effect and can be repressed by epidermal growth factor (EGF) and HGF, we hypothesize that activation of HGF/Met and/or EGFR signalling in the regenerating liver might contribute to the reduction of NOX4 as a mechanism to promote hepatocyte proliferation and resistance to apoptosis. However, livers lacking EGFR tyrosine kinase activity display delayed proliferation together with an overactivation of the TGF-β/SMAD pathway, but no differences in apoptosis or levels of NOX4 are observed [33]. Another potential mediator is the phosphatase PTP1B that binds Met and EGFR and decreases their activation. PTP1B deficiency results in enhanced EGF- and HGF-mediated signalling in hepatocytes and livers submitted to partial hepatectomy resulting in accelerated regeneration, [34] and also confers resistance on TGF-β-induced suppressor effects in hepatocytes [35]. Whether the effects on tyrosine kinase receptors and TGF-β signalling are mutually dependent has not been clarified. On the other hand, a negative feedback regulation between TGF-β and HGF through the transcription factor nuclear factor I-C has been proposed to have a central regulatory role in the regulation of the phased proliferative response in the liver [36]. Putting all these strands together is not easy, but a proper balance of all these signals might be a good determinant of the efficiency and intensity of the regenerative response.

Last but not least, TGF-β also regulates phenotypic conversions, in particular, it drives epithelial-to-mesenchymal transition (EMT). Although there is now no question that TGF-β is able to induce EMT in hepatocytes, fetal, neonatal and adult (at least in vitro and under certain conditions), and that this process confers resistance to the TGF-β-mediated suppressor effects [37], whether TGF-β actually drives this conversion in vivo remains controversial and awaits further investigation. Furthermore, data lead us to conclude that this process gains relevance in the context of liver disease, as discussed later. Nevertheless, some progress is being made in relation to how EMT is induced. Collaboration between SMAD and non-SMAD signalling is known, but new signalling interplays and regulatory molecules are being identified. Again, miRNAs seem to have a role. Specifically, miR-101 has been shown to inhibit TGF-β-induced EMT in hepatocytes [38]. As to the cross-talk between TGF-β and tyrosine kinase receptor signalling during EMT induction, the picture is still quite obscure. Although TGF-β-mediated EGFR activation appears to be dispensable for the EMT process in primary fetal hepatocytes [39], it has been proposed as a requirement in immortalized adult hepatocytes cell lines [40]. In hepatic progenitors, the EGFR pathway could inhibit, or even reverse, the effects of TGF-β on EMT [41], suggesting a context- and/or cell-type-specific response.

TGF-β in liver fibrosis: from simple to complex

TGF-β was recognized as a profibrogenic cytokine due to its role in HSC activation and extracellular matrix production [42]. The key fibrogenic effects of TGF-β are mainly mediated through activation of HSC, the principal producer of extracellular matrix during liver fibrosis (Fig. 3). Interestingly, different functions have been attributed to SMAD2 and SMAD3 in HSC; SMAD2 was found to antagonize TGF-β/SMAD3-induced collagen expression [12]. Moreover, there is a cohort of fibrotic mediators downstream of TGF-β, most importantly NOX4 [43] and connective tissue growth factor [44, 45]. Migration and proliferation are essential features of activated HSC, and are at least partially induced by TGF-β through various mechanisms. For example, TGF-β-induced HSC migration is mediated by guanosine triphosphate (Rho GTPase) signalling [46]. Activation of platelet-derived growth factor receptor-β expression and oxidative stress induction by TGF-β mediates the proliferative actions of TGF-β on HSC [47]. miRNAs have also been observed in TGF-β-mediated regulation of HSCs. For example, miR-29 mediates the regulation of liver fibrosis and forms part of a signalling nexus involving TGF-β- and NF-κB-dependent downregulation of miR-29 family members in HSC, with the subsequent upregulation of extracellular matrix genes [48]. TGF-β also downregulates miR-30c and miR-193 in HSCs, hence modifying its effects on extracellular matrix remodelling, a critical process during liver fibrosis [49]. Importantly, it became evident that TGF-β also participates in the development of fibrosis through signalling in other liver cell types, such as hepatocytes and hepatic progenitor cells. For example, we have utilized cell-type-specific overexpression of SMAD7 (a known negative regulator of TGF-β signalling by interfering with SMAD2/SMAD3 phosphorylation [7]) in transgenic mice to demonstrate that blunting TGF-β signalling in hepatocytes interferes with liver damage and fibrogenesis [50]. Interestingly, TGF-β signalling in hepatocytes under metabolic stress mediates hepatocyte death and lipid accumulation [43, 51], processes that require SMAD signalling and reactive oxygen species production, leading to the development of nonalcoholic steatohepatitis [51].

Different regulators of this important pathway were recently identified, for example, vasodilator-stimulated phosphoprotein [52], platelet-derived growth factor receptor-α [53] and budding uninhibited by benzimidazoles-1 kinase [54], which can modulate TGF-β signalling directly through forming complexes with TGF-β receptors, or lipopolysaccharide, which modulates TGF-β signalling indirectly by sensitizing HSC to TGF-β through downregulation of BAMBI, a transmembrane TGF-β decoy receptor [55]. Moreover, such regulators might be further regulated through different mechanisms. For instance, SMAD7 can be regulated through epigenetic- [56] or miRNA-based mechanisms [57, 58], resulting in a steady increase in the complexity of this pathway.

Role of TGF-β on liver carcinogenesis: angel or evil?

In addition to the clear evidence for TGF-β signalling as a tumour suppressor, many studies have reported the overexpression of TGF-β1 in various types of human cancer, which correlates with a poor prognostic outcome [6]. Although TGF-β suppresses the early stages of tumour development, it later contributes to tumour progression when cells become resistant to its suppressive effects, conferring resistance to cell death, inducing cell migration and invasion, mediating immune suppression and microenvironment modification [4, 5] (Fig. 1).

To make the situation even more complex, TGF-β plays a dual role in the control of growth and death in immature/proliferating hepatocytes and in liver tumour cells. As a tumour suppressor, it early induces cell-cycle arrest and apoptosis. However, it later activates proliferative and antiapoptotic signals through the transactivation of other promitogenic pathways, such as PDGF or EGF [39, 59-61] (Fig. 1). PDGF links TGF-β signalling to nuclear β-catenin accumulation in liver tumour cells [62] and EGF is an important survival signal in HCC, counteracting TGF-β-induced cell death [61]. Downstream signals, such as the PI3K/AKT axis, counteract TGF-β-induced apoptosis, through impairing up-regulation of the NADPH-oxidase NOX4, which is required for reactive oxygen species production and TGF-β-induced mitochondrial-dependent apoptosis [63, 64]. The transactivation of the EGFR pathway by TGF-β in liver cells requires the activity of the metalloprotease tumour necrosis factor-α-converting enzyme (TACE)/ADAM17, which is responsible for shedding and activation of the EGF receptor ligands [39, 59]. Recent evidence suggests that caveolin-1 and lipid raft domains in the cell membrane are required for this process [65]. Interestingly, caveolin-1 overexpression is associated with HCC tumorigenesis and metastasis [66].

Liver tumour cells that overcome TGF-β-induced apoptosis respond to this factor by undergoing EMT, concurring with downregulation of E-cadherin and other epithelial-related genes, upregulation of vimentin and other mesenchymal-related genes, as well as cytoskeleton reorganization, altogether contributing to increased cell migration and invasion [67]. Interestingly, cross-talk between EMT and resistance to TGF-β-mediated apoptosis has been described. Upregulation of SNAI-1, the gene that codifies for Snail (a repressor of E-cadherin expression, upregulated by TGF-β) mediates antiapoptotic signals counteracting TGF-β-induced cell death [68] (Fig. 1). Concomitantly, once EMT has been undergone, cells express high levels of TGF-β, which acts as an autocrine loop, transactivating the EGFR pathway, which in turn mediates cell survival signals [69]. It has been suggested that SMAD-dependent regulation of EMT might also be under the control of miRNAs [70]. The miR-200 family is an important regulator of EMT and recent evidence suggests that it plays relevant roles in suppressing metastasis in human HCC [11, 71]. Furthermore, miR-125b has been shown to suppress EMT and EMT-associated properties by targeting SMAD2 and SMAD4 [72]. This dual, and in some aspects controversial, response of liver tumour cells to TGF-β explains the complexity when studying the role of this pathway in liver tumour progression. A classification established according to the TGF-β-gene signature in HCCs showed that a ‘late’ TGF-β signature (coincident with the expression of EMT-related, antiapoptotic and invasion-related genes) correlates with increased tumour recurrence, as well as a higher invasive phenotype, when compared with tumours that show an early TGF-β signature (suppressor genes) [73].

Alterations at TGF-β receptors or SMAD2/3/4 levels are rare in HCC and expression of TGF-β is upregulated in a large percentage of these patients [4, 5]. Nodal, a TGF-β-related morphogen, has been also shown to be upregulated in HCC, and is associated with aggressive characteristics [74]. The TGF-β-mediated SMAD-dependent cytostasis may be altered in HCC, as demonstrated in different cell lines that express high amounts of SMAD7 and show reduced SMAD3 signalling [75]. Loss of adaptor protein embryonic liver foldrin, an important SMAD3/4 adaptor has been also related to loss of TGF-β tumour suppressor actions in HCC [76]. Furthermore, HCC cells overexpress a specific set of miRNAs that would allow escape from TGF-β-induced apoptosis [77-79]. Most importantly, overactivation of growth factor-mediated survival signals, such as MAPK/ERKs, PI3K/AKT or NF-κB, in HCC cells would impair the TGF-β suppressor arm observed in liver tumour cells. In particular, during hepatocarcinogenesis, an increase in the activity of the metalloprotease TACE/ADAM17 may promote the antiapoptotic arm of the TGF-β signalling [59] (Fig. 1).

Different conditions in the tumour microenvironment may also provide the signals for the switch in the role of TGF-β in cancer progression [5, 80]. Direct cross-talk between TGF-β and chemokine signalling has recently been proposed. TGF-β induces the expression and polarization of the chemokine receptor CXCR4 in HCC cells, which facilitates the response to its ligand (CXCL12/SDF-1, produced by the stroma cells), mediating HCC cell migration and survival [81]. Through regulation of connective tissue growth factor expression in HCC invasive cells, TGF-β also mediates tumour/stroma cross-talk [82]. Furthermore, current evidence suggests that TGF-β is a critical mediator of the cancer-associated fibroblasts phenotype, which plays a relevant role in facilitating the production of growth factors and cytokines that contribute to cell proliferation, invasion and neoangiogenesis [83]. Elevated TGF-β activity also creates an immune-subversive microenvironment by which immune privilege develops [84] and that may also favour colonization of disseminated HCC cells in the portal venous system [85]. Indeed, in the late stages, TGF-β contributes to liver tumour progression (Fig. 3).

Targeting TGF-β in liver diseases: a new challenge for the future

Although many successful approaches used to inhibit TGF-β signalling in animal models have given very promising results, the major problem facing translation into the clinic is the complex role of TGF-β in the liver, not only in liver fibrogenesis, but also in EMT, cell proliferation, carcinogenesis and immune modulation [6, 86]. Careful understanding and analysis of TGF-β signalling in the liver, its various regulators along with surrounding microenvironment and the time point of disease progression will help in developing better personalized therapy and successful treatment for fibrotic patients in the future.

Drug-based therapies for the treatment of HCC are currently restricted to patients with an advanced stage of disease. However, this approach has a strong impact on clinical practice both because, at the time of diagnosis, the majority of patients are already in too advanced a disease stage to receive so-called ‘curative therapies’ (liver transplantation, surgery, radiofrequency ablation), and because ~ 50% of patients who do undergo these therapies suffer disease recurrence within 3 years, limiting further invasive treatment. At present, Sorafenib is still the only treatment approved for therapy in patients with advanced stage HCC, but new therapies are an urgent medical necessity [87, 88], owing to the high number of drop-outs due to side effects, or to drug suspension because of lack of effectiveness. Recently, a large number of phase III clinical trials have failed to achieve the primary end points, despite the encouraging results obtained in the phase II clinical trials and the promising scientific rationale for inhibiting the relative molecular target. The main reason for these frustrating results is that heterogeneity is a hallmark of HCC, so the absence of proper patient stratification to select those more likely to benefit from that specific therapy heavily affects the clinical outcome [89]. For instance, Tivantinib, a small molecule inhibitor of c-Met, has been tested in a phase II clinical trial in patients with advanced HCC. The overall survival of patients who received therapy was similar to that of the patients who received placebo, but when stratifying patients according to high/low levels of c-Met a strong difference was observed between the two groups, the treatment being more successful in patients with higher expression levels of c-Met [90]. This is a convincing example of the great importance of personalized therapy based on the molecular characteristics.

The identification of promising target pathways involved in HCC tumour progression is fundamental to design new therapies. In this scenario, as previously discussed, TGF-β seems to be a promising therapeutic target, bearing in mind that increased TGF-β levels have been reported in patients with HCC, and are considered a hallmark of this disease [91, 92]. Using in vitro assays, it was first proved that inhibitors of the TGF-β receptor pathway, such as Galunisertib (LY2157299), do not induce apoptosis in HCC cells (at the intended pharmacological levels for patients), but rather block their invasion and migration, which correlates with an increase in E-cadherin expression [93]. Galunisertib also inhibits the stromal reaction and neoangiogenesis, reducing tumour growth in experimental models [82, 94]. Moreover, Galunisertib blocks β1-integrin activation in cancer cells and consequently impairs intravasation of HCC cells into the blood vessels [95]. All these previous results offered the scientific rationale for testing the drug in clinical trials [96]. Robust experimental preclinical data suggest that the compound acts in HCC mainly through inhibiting the canonical TGF-β pathway, although other noncanonical TGF-β signalling pathways are also involved in HCC tumour progression [97]. Several different strategies have been proposed for inhibiting the TGF-β pathway in human cancer, including the use of chimeric proteins, monoclonal antibodies, small molecules (such as Galunisertib) and antisense oligonucleotides (reviewed in Refs [6, 98]). In the case of monoclonal antibodies such as D10, directed against the type II receptor of TGF-β they lacked effectiveness in HCC experimental models compared with Galunisertib [99]. Biomarkers would be extremely useful in the proper selection of those patients who might benefit from receiving Galunisertib in clinical trials, but so far no biomarkers have been validated for this purpose. However, it has been reported that in patients with HCC, higher circulating levels of TGF-β are correlated with lower levels of E-cadherin, regardless of the Barcelona Clinic Liver Cancer classification [100]. Taken together, these data are also important in view of the preliminary results of a phase II clinical trial the results of which will be released in the coming year, showing that patients with higher levels of circulating TGF-β are more likely to respond to therapy with Galunisertib, a small molecule inhibitor of the TβRI kinase. The preliminary results of this trial, NCT01246986 (http://clinicaltrials.gov) are very encouraging in terms of both the clinical outcome and the safety of the treatment. Development of this therapy as a second line treatment would fulfil an important need in medical practice. The final data on this trial will be available next year and should be an important step toward better framing the role of TGF-β in HCC, and will likely orient future research in the field. Targeting the TGF-β pathway may be relevant not only in HCC, but also in other types of liver tumours, such as cholangiocarcinomas or combined hepatocellular–cholangiocarcinomas where the TGF-β pathway has been shown to be deregulated as well [101-103].

Concluding remarks

Given the plethora of functions attributed to TGF-β in regulating liver pathological processes (Fig. 3), it is considered as a targetable molecule in liver disease. However, as mentioned in this review, TGF-β also has numerous and important roles in liver physiology (Fig. 2), thus transforming the scenario into a less bright panorama. The mechanism of action of any therapeutic approach should consider impairment of the profibrotic and tumorigenic properties of this cytokine, while amplifying its proapoptotic and tumour suppressor effects. Encouraging results in clinical trials using TGF-β receptor inhibitors prove that research is moving forwards despite the complexity of the situation.

Year after year, new regulators of TGF-β functions are discovered. This allows the possibility of targeting secondary regulators in order to alter the TGF-β response in the liver. Among these regulators, miRNAs have a great potential. Last, but not least, the multiple cells responsive to TGF-β acting in liver pathology, and the different cell-type-dependent functions it exerts should also be taken into consideration. During the last decade, the importance of cross-talk between hepatocytes/liver cancer cells and the stroma has been evidenced, meaning that more specific cell-targeted therapies are required in the future.

Authors contributions

All the authors participated in the elaboration of the text. Figures were designed by IF, JM-C and AS. Final edition of the first manuscript and preparation of the revised version were performed by IF and JM-C.

Podcast

This article is accompanied by a podcast, listen now. Or listen in iTunes.