Background Chronic liver injury triggers a progenitor cell repair response, and liver fibrosis occurs when repair becomes deregulated. Previously, we reported that reactivation of the hedgehog pathway promotes fibrogenic liver repair. Osteopontin (OPN) is a hedgehog-target, and a cytokine that is highly upregulated in fibrotic tissues, and regulates stem-cell fate. Thus, we hypothesised that OPN may modulate liver progenitor cell response, and thereby, modulate fibrotic outcomes. We further evaluated the impact of OPN-neutralisation on murine liver fibrosis.
Methods Liver progenitors (603B and bipotential mouse oval liver) were treated with OPN-neutralising aptamers in the presence or absence of transforming growth factor (TGF)-β, to determine if (and how) OPN modulates liver progenitor function. Effects of OPN-neutralisation (using OPN-aptamers or OPN-neutralising antibodies) on liver progenitor cell response and fibrogenesis were assessed in three models of liver fibrosis (carbon tetrachloride, methionine-choline deficient diet, 3,5,-diethoxycarbonyl-1,4-dihydrocollidine diet) by quantitative real time (qRT) PCR, Sirius-Red staining, hydroxyproline assay, and semiquantitative double-immunohistochemistry. Finally, OPN expression and liver progenitor response were corroborated in liver tissues obtained from patients with chronic liver disease.
Results OPN is overexpressed by liver progenitors in humans and mice. In cultured progenitors, OPN enhances viability and wound healing by modulating TGF-β signalling. In vivo, OPN-neutralisation attenuates the liver progenitor cell response, reverses epithelial-mesenchymal-transition in Sox9+ cells, and abrogates liver fibrogenesis.
Conclusions OPN upregulation during liver injury is a conserved repair response, and influences liver progenitor cell function. OPN-neutralisation abrogates the liver progenitor cell response and fibrogenesis in mouse models of liver fibrosis.
Statistics from Altmetric.com
Significance of this study
What is already known about the subject?
Progressive liver fibrosis is associated with a liver progenitor cell response (also known as the ductular reaction in human).
Liver progenitor cells participate in the wound healing (fibrogenic) response through direct epithelial-mesenchymal transition changes and/or indirect paracrine crosstalk with hepatic stellate cells.
Osteopontin is a cytokine and matrix molecule that is highly expressed in inflamed and fibrotic tissues throughout the body, and directly activates hepatic stellate cells.
There is currently no licensed antifibrotic therapy for use in patients with advanced liver fibrosis or cirrhosis.
What are the new findings?
Osteopontin is an important regulator of the liver progenitor cell response.
Neutralising osteopontin leads to the reversal of the liver progenitor cell response (ie, reduced progenitor cell numbers and impaired wound healing/transmigration).
Neutralising osteopontin modulates transforming growth factor (TGF)-β signalling by downregulating TGF-β mRNA and protecting levels of Ski and SnoN (two transcriptional co-repressors of TGF-β signalling).
Neutralising osteopontin using two different approaches (aptamer or neutralising antibody) results in a dramatically attenuated liver progenitor cell response and liver fibrosis in three different mouse models of liver fibrosis.
Osteopontin neutralisation is safe in mice with liver fibrosis.
How might it impact on clinical practice in the foreseeable future?
Osteopontin-specific aptamers are available and other neutralising antibodies/small molecule compounds are being developed.
Osteopontin neutralisation may be a useful strategy for the treatment of individuals with advanced liver fibrosis.
The occurrence of fibrosis (scar tissue accumulation) in chronic liver disease (CLD) presents a vast unmet clinical challenge. At present, there is no proven antifibrotic treatment that halts or reverses the progression of liver fibrosis.1 ,2 Individuals with liver fibrosis are therefore, at risk of developing cirrhosis, and complications such as liver cancer and liver failure, for which the only potential treatment is a liver transplant.3 ,4 Furthermore, the prevalence of CLD is predicted to increase in the coming decades due to the global epidemic of major risk factors for non-alcoholic fatty liver disease, including type 2 diabetes mellitus and obesity.5 As such, fibrosis and cirrhosis complicating CLD is a major health and economic burden, and critically requires the identification and development of novel antifibrotic strategies.
Liver fibrosis is an excessive wound healing or repair response to CLD,6 ,7 which includes non-alcoholic fatty liver disease or alcoholic liver disease (ALD), viral hepatitis (B and C), primary biliary cirrhosis (PBC), and primary sclerosing cholangitis. During CLD restoration of liver mass and function in response to hepatocyte-loss involves activation of progenitor cells within the liver (ie, progenitor-associated repair response or ductular reaction),8–10 which proliferate and differentiate into new hepatocytes and cholangiocytes.11 ,12 This pool of progenitors is heterogeneous, compromising the resident liver progenitor cell (LPC) or oval cell residing in the canals of Hering,12 ,13 bone marrow-derived progenitors,14 ,15 as well as hepatic stellate cells (HSCs).16 HSCs are liver fibroblasts which normally transition into collagen-producing myofibroblasts when activated,17 ,18 but have recently been recognised as multipotent progenitors that interact with the liver progenitor pool16; thus, providing an explanation for the fibrogenic outcomes which accompany progenitor cell expansion during persistent liver injury (ie, fibrogenic repair). These cell culture and mouse observations are supported by longitudinal human studies, which show that the ductular reaction is predictive of subsequent fibrosis.9 ,19 Hence, targeting the progenitor-associated repair response may be of value in inhibiting fibrosis in CLD.
Progenitor cell activation is driven by a milieu of growth factors, and cytokines which accompany CLD.20 ,21 Recently, we reported that injury-related reactivation of the hedgehog (Hh) pathway (a morphogen important for embryonic development) induces liver progenitors to proliferate, undergo epithelial-mesenchymal transition (EMT) (ie, upregulating mesenchymal, while repressing epithelial genes), and secrete factors that activate neighbouring progenitors and matrix-producing cells.22–25 Osteopontin (OPN) is one such molecule secreted by liver progenitors.26
OPN is a Hh-target, and a matricellular protein that is highly upregulated in fibrotic skin, lungs, kidneys and joints.27–30 Mice genetically deficient in OPN develop less fibrosis after certain injuries, suggesting that OPN may be a direct effector of the fibrotic process. Not surprisingly, HSCs express high levels of OPN to autoregulate their fibrogenic phenotype.26 However, the highest expression of OPN is seen in cells located in the liver periportal regions, and recent studies show that OPN is a key regulator of bone-marrow progenitor/stem-cell fate.31 These led us to hypothesise that OPN modulates the progenitor-associated fibrogenic repair response during liver injury.
Despite prevailing data giving credence to OPN being an attractive antifibrotic target, and humanised antibodies to OPN being developed for inflammatory-joint diseases,30 no study has yet evaluated the impact of OPN neutralisation in the treatment of fibrosis complicating CLD. Therefore, to evaluate these hypotheses, we studied the direct effects of OPN in cultures of liver progenitors, examined the effects of OPN-neutralisation (by OPN-specific aptamers or OPN-neutralising antibodies) in three murine models of liver fibrosis, and corroborated findings with analysis of liver tissues from patients with CLD.
Materials and methods
Mice: Adult C57BL/6 wild type.
Models of hepatic fibrosis
Carbon tetrachloride (CCL4)
Mice (n=5/group) received twice-weekly intraperitoneal injections of CCl4 (0.5 mg/kg, Sigma-Aldrich) for 6 weeks to induce liver fibrosis,32 or vehicle (mineral oil).
Methionine-choline deficient diet
Mice (n=5/group) were fed the methionine-choline deficient (MCD) diet for 5 weeks to induce non-alcoholic steatohepatitis (NASH) fibrosis, or control chow.24
Model of biliary fibrosis
Mice (n=5/group) were fed the dihydrocollidine (DDC) diet for 3 weeks to induce biliary-type fibrosis.33
Three additional studies were performed (fourth study: CCL4; fifth study: MCD; sixth study: DDC) (n=10/study; 5/group). OPN-specific aptamers (which specifically neutralise circulating-extracellular OPN) or sham-aptamers (negative control with mutated active binding site)34 were administered to mice by tail-vein injections (total of four injections per mouse), during the final week of dietary or chemical challenge. A 200 µg dose of sham or OPN-aptamers (in 100 µL of PBS) was used as this was the dose previously shown to exhibit efficacy in vivo.34 ,35 All mice were sacrificed 24 h after the final dose of aptamers.
MCD diet fed mice (n=5/group) were injected either control (IgG) or anti-OPN (R&D) in the final week, as described above (four injections; 50 µg/injection), using an amount of anti-OPN previously shown to be effective in reducing insulin resistance in obese mice,36 and sacrificed 24 h after the final injection.
Mice were housed in 12-h-light/dark cycle with food and water ad libitum. Liver samples were obtained for RNA analyses and immunohistochemistry. Animal care and procedures were as per the National Institute of Health ‘Guide for the Care and Use of Laboratory Animals’, and approved by relevant institutions: Duke University Institutional Animal Care and Use Committees, Vrije Universiteit Brussel, Belgium (LA 123 02 12), University of Calgary Animal Care Committee, and the UK Home Office approval in accordance with the Animals (Scientific Procedures) Act of 1986 (University of Birmingham, PPL 40/3201).
Formalin-fixed, paraffin-embedded (FFPE) liver sections were from deidentified controls and explanted liver tissues from individuals undergoing liver transplantation for NASH-cirrhosis, ALD-cirrhosis or PBC. Normal tissues were obtained from excess split-liver grafts. Freshly explanted and snap-frozen NASH, ALD and PBC liver tissues (n=5/group) were used for total liver RNA analyses.
All studies using material from Duke University Hospital were conducted in accordance with NIH and Institutional guidelines for human subject research. Samples acquired from the Hepatobiliary Unit in Birmingham were studied in accordance with local ethical approval 04/Q2708/41 and REC 2003/242 from the South-Birmingham Research Ethics Committee, UK, and those obtained from University Hospital Essen, Germany under the local ethics commission (09-4252).
Cell culture/treatments and functional studies
See online supplementary Materials and Methods.
Liver immunohistochemistry and molecular techniques
See online supplementary Materials and Methods and tables S1–3.
All data were expressed as mean±SEM. Statistical analysis was performed using Student's t test or one-way ANOVA as indicated. All analysis was conducted using Graph-Pad Prism 4 software (GraphPad Software). Statistically significant differences were considered at p≤0.05
OPN is upregulated in human CLD and is expressed by Sox9+ liver progenitors
Coded liver sections from patients with NASH-cirrhosis (n=5), ALD-cirrhosis (n=5) and PBC (n=5) were stained to demonstrate OPN, Sox9 and K19. Compared with healthy livers, cirrhotic livers exhibit up to 15-fold more OPN protein (figure 1A, see online supplementary figure S1A) and 10-fold more OPN mRNA (see online supplementary figure S1F). The highest level of OPN is expressed by cells located in the periportal regions. As expected, cirrhotic livers express more transforming growth factor (TGF)-β mRNA than healthy livers (see online supplementary figure S1F).
Because Sox9+ cells are bipotent liver progenitors that arise from periportal canals of Hering,37 we examined this LPC (oval cell) marker. As expected, cirrhotic livers contain up to 10 times more Sox9+ cells (figure 1B, see online supplementary figure S1B) and upregulate Sox9 mRNA by up to fivefold (see online supplementary figure S1F). Sox9+ cells also express K19 (another ductular progenitor marker) (see online supplementary figure S1D). Compared with healthy control livers, cirrhotic livers accumulate up to 20-fold more Sox9/K19 double-positive liver progenitors (see online supplementary figure S1D,E). Interestingly, the highest amount of OPN is expressed by liver progenitors (figure 1C–E). Double immunolabelling confirms that OPN+ cells co-localise with Sox9+ cells (figure 1C,D), and cirrhotic livers are enriched with OPN/Sox9 double-positive progenitors by over 15-fold (figure 1E).
OPN regulates viability/proliferation of liver progenitors
As the highest amounts of OPN were expressed by Sox9+ LPC in vivo (figure 1), we evaluated the importance of OPN in LPC viability/proliferation. We used 603B cells (ductular progenitors),24 ,38 which coexpress Sox9, K19 (markers of LPC) and OPN (figure 2A). 603B supernatants contain high levels of OPN (figure 2A). Treating 603B with OPN-aptamers or OPN-neutralising antibodies (R&D) lowered OPN levels in culture supernatants (see online supplementary figure S2A,B). OPN-aptamers reduced cell viability by 24 h, but this was most pronounced at 72 h (up to twofold reduction) when quantified by CCK8 and cell count methods (figure 2B and see online supplementary figure S2C). Because liver fibrosis and cirrhosis are associated with upregulated TGF-β (see online supplementary figure S1F) (and TGF-β modulates progenitor cell differentiation39), we repeated OPN-neutralising experiments under conditions of TGF-β stimulation. OPN-neutralisation under profibrogenic stimulation led to greater suppression of cell proliferation (threefold; figure 2C and see online supplementary figure S2D), and markedly increased 603B apoptosis, as measured by the caspase 3/7 assay (figure 2D, see online supplementary figure S2E); sub-G1 analysis by fluorescence-activated cell sorting / flow cytometry (FACS) (detecting DNA fragmentation) confirmed comparable increases in 603B apoptosis (threefold; figure 2E,F and see online supplementary figure S2F).
OPN enhances progenitor-associated wound healing
OPN effects on LPC
Cumulative data suggest that liver progenitors are capable of reprogramming into myofibroblasts.22 ,24 ,40 Therefore, we evaluated if changes in OPN levels would lead to similar alterations in progenitor phenotype. 603B-LPCs were treated with recombinant OPN or OPN-aptamers. OPN-neutralisation led to a decrease in progenitor-associated fibrogenic genes, Snail and Collagen 1αI (figure 3A), while upregulating epithelial genes, E-cadherin and Id2 (figure 3B). The addition of exogenous OPN, on the other hand, showed minimal effects on pro-EMT genes (data not shown), because 603B already express high levels of OPN (figure 2A) and pro-EMT genes.26 ,41
Increased cell motility/migration is a key phenotypical change that accompanies EMT.22 ,42 Therefore, we compared 603B migration in OPN-neutralising or control conditions. Migration was assessed by semiquantitating the dimensions of a wound dividing the confluent monolayer 12 h after the scratch (figure 3C, see online supplementary figure S3A,B). Treatment with TGF-β led to enhanced wound shrinkage compared with vehicle (data not shown). OPN-neutralisation under basal conditions led to ∼30% less wound healing compared with controls (p=0.057) (see online supplementary figure S3A,B), but this effect was significantly enhanced (>50% less wound healing) after TGF-β (figure 3C). Comparable results were observed in the modified cell-invasion assay (figure 3D and see online supplementary figure S3C,D). OPN-neutralisation under basal conditions led to 20% fewer 603B cells invading across the insert membrane (see online supplementary figure S3C,D), but this was enhanced under TGF-β stimulation, where OPN-neutralisation led to a repression of nearly 70% (figure 3D).
To exclude the possibility that observed responses were 603B-specific, we repeated experiments using the bipotential mouse oval liver (BMOL) line.43 We confirmed that BMOL cells resembled 603B-LPC, expressed OPN, Sox9 and K19 proteins (see online supplementary figure S4A), and proliferated in response to OPN (see online supplementary figure S4B). BMOL exhibited comparable wound-healing responses: OPN-neutralisation inhibited wound closure and transmigration (see online supplementary figure S4C), repressed Collagen 1αI (see online supplementary figure S4D), while inducing E-cadherin and Id2 mRNA (see online supplementary figure S4E,F).
OPN is a complex molecule which consists of intracellular and extracellular/soluble OPN isoforms, which exhibit overlapping, but also different and opposing functions.44 OPN-aptamers neutralise only soluble OPN isoforms. We evaluated if additional loss of intracellular OPN by RNAi might reproduce OPN-aptamer effects on LPC phenotype. Compared with control (603B-shScr) (603B infected with lentiviral particles containing non-targeting scrambled shRNA), OPN knockdown (∼50%) (603B-shOPN) was associated with a twofold to threefold reduction in cell proliferation under basal and TGF-β conditions (see online supplementary figure S5A–D), with minimal changes in LPC apoptosis. shOPN abrogated 603B-LPC transmigration by 50% (see online supplementary figure S5E), but did not affect wound healing response even under TGF-β conditions (data not shown), thus indicating functional differences between extracellular and intracellular OPN.
OPN modulates TGF-β signalling
Because the fibrotic liver is enriched with high levels of TGF-β (see online supplementary figure S1), and the greatest effects of OPN-neutralisation were observed under TGF-β conditions (figures 2 and 3), we evaluated if OPN effects on LPC were mediated by modulating TGF-β signalling. Treating 603-LPC with TGF-β led to the accumulation of phospho-Smad-2/3 proteins (figure 3E, see online supplementary figure S6A), and to the preservation of Ski and SnoN (figure 3E, see online supplementary figure S6B,C), transcriptional co-repressors which inhibit transcriptional activity of TGF-β-dependent Smad-2/3-complexes under basal conditions.45 Smad phosphorylation and Ski and SnoN levels were unaffected by the addition of the sham-aptamers. In contrast, treatment with OPN-specific aptamers led to reduced levels of phospho-Smad-2/3 proteins (figure 3E, see online supplementary figure S6A), while protecting Ski and SnoN from TGF-β-induced degradation (figure 3E, see online supplementary figure S6B,C). No changes were observed with Smad7 protein (figure 3E, see online supplementary figure S6D), another negative feedback mechanism that regulates the TGF-β signal.46
To better define how OPN modulates TGF-β signalling, we targeted putative extracellular OPN-receptors on progenitor cells,47 treating 603B-LPC with CD44-neutralising antibody, and αvβ3 antagonist XJ735. CD44 and αvβ3 blockade resulted in a 30% reduction in phospho-Smad-2/3 expression (see online supplementary figure S6E), resembling the effects of OPN neutralisation with OPN-aptamers. However, general depletion of OPN (intracellular) using shOPN (ie, 603B-shOPN) did not repress phospho-Smad-2/3 expression (data not shown), implying that widespread OPN knockdown could only recapitulate some effects of OPN-neutralisation with OPN-aptamers, and reinforcing the concept that extracellular OPN is different from intracellular OPN.
OPN effects on HSC
Liver progenitors comprise of LPC (oval cell) and HSC, a fibroblast and recently recognised multipotent progenitor.16 When activated, HSCs undergo transition to become myofibroblasts.25 ,40 LPC and HSC are in close proximity, suggesting that both progenitor cell populations are capable of crosstalk. We therefore evaluated whether reduced levels of OPN in the microenvironment alters the ‘LPC- secretome’ that influences HSC phenotype. 603Bs were treated with OPN-aptamers or sham-aptamers, in the presence of TGF-β for 48 h; 603B-conditioned media (CM) were then harvested and added to primary HSC for 24 h. As expected, the addition of 603B-CM to HSC resulted in activation of HSC by upregulating α smooth muscle actin (αSMA) and collagen 1α1 mRNA by twofold (see online supplementary figure S7A,B). HSC activation was enhanced when treated with TGF-β-stimulated 603B-CM. The addition of OPN-aptamers to 603B resulted in a significantly altered secretome: CM from OPN-aptamer-treated 603B induced a significantly attenuated fibrogenic-response in HSC (to almost quiescent) compared with CM obtained from sham-aptamer-treated 603B, with fourfold and fivefold less αSMA and collagen 1α1 mRNA under basal and TGF-β-stimulated conditions, respectively (see online supplementary figure S7A,B). Experiments repeated using the human liver myofibroblast line (Lx2) revealed comparable findings: CM from OPN-aptamer-treated 603B had diminished activating capacity on Lx2 cells (lower αSMA mRNA by 40% and lower collagen 1αI mRNA by 50%) (see online supplementary figure S7C,D).
To address concerns that effects on HSC could be related to residual OPN-aptamers in the 603B-CM, we used CM obtained from 603B-shOPN (OPN knockdown) and 603B-shScr (control) cells, treated with or without TGF-β (see online supplementary figure S7A,B). Under basal conditions, CM derived from 603B-shOPN activated primary HSC (αSMA and collagen 1α1 mRNA) approximately twofold less than CM-603B-shScr (see online supplementary figure S7A,B). These differences in CM-effects were reduced under TGF-β conditions (∼1.5-fold). In a further experiment, 603B-shOPN and 603B-shScr were treated with either sham-aptamers or OPN-aptamers for 48 h, and respective CMs collected to treat HSC. The addition of OPN-aptamers virtually abolished HSC activation (αSMA and collagen 1α1 mRNA levels were lower than untreated HSC). In aggregate, these data implicate the importance of OPN in modulating the LPC response.
Finally, to confirm that OPN neutralisation has a direct impact on HSC, we treated primary HSC directly with sham-aptamers or OPN-aptamers. OPN neutralisation with OPN-aptamers led to 50% decrease in αSMA and collagen 1α1 mRNA under basal conditions (see online supplementary figure S7E,F), but up to fourfold downregulation under TGF-β stimulation.
In summary, OPN is a critical factor that modulates the LPC niche, by modulating progenitor cell and HSC responses, via direct and indirect mechanisms.
OPN neutralisation ameliorates the liver progenitor cell response and fibrogenesis in mice
Accumulation of OPN+ (Sox9+) LPC in liver fibrosis (validation)
CCl4 and MCD models
Fibrosis: CCl4 and MCD-treated mice developed significant liver fibrosis (figure 4, see online supplementary figures S8 and S9). This was demonstrated by increased Sirius red (SR) staining (eightfold) (figure 4A,B) and hepatic hydroxyproline quantification. Collagen deposition was accompanied by the accumulation of αSMA+ cells (10-fold) (figure 4C,D), and induction of key fibrogenic genes, αSMA, Collagen 1αI and TGF-β1 (see online supplementary figure S9).
Liver progenitors: Pertinently, liver fibrosis was associated with a 6–10-fold increase in OPN+ cells (figures 5A and 6A, see online supplementary figure S8C), and an exuberant progenitor response: fivefold more Sox9+ LPC (figures 5B and 6B, see online supplementary figure S8D), fourfold more Sox9 mRNA (see online supplementary figure S10A,C), and ∼twofold more K19 mRNA (see online supplementary figure S10B,D). Specifically, there was a 5–10-fold enrichment of OPN+ cells which coexpressed Sox9, the LPC marker (figures 5C and 6C).
Liver progenitor reprogramming: Liver fibrosis was accompanied by the upregulation of mesenchymal markers, OPN and Snail (figure 7A,B, see online supplementary figure S11A,B), and a downregulation of epithelial markers, E-cadherin and Id2 (figure 7C,D, see online supplementary figure S11C,D). Double immunolabelling identified Sox9+ LPC which coexpressed E-cadherin (figure 7E,F, see online supplementary figure S11E,F) under basal conditions; expression of E-cadherin was significantly repressed during fibrogenesis.
Biliary-type fibrosis was induced by the DDC-supplemented diet.33 Increased SR staining and liver hydroxyproline content (see online supplementary figure S12A) was similarly associated with a greater than 10-fold enrichment in the number of αSMA+ cells (see online supplementary figure S12B) and an upregulation in αSMA, Collagen 1αI and TGF-β1 mRNA ( see online supplementary figure S12C–E). There was an accumulation of OPN+ and Sox9+ cells (over 10-fold) (see online supplementary figure S13A,B), and OPN/Sox9 double-positive cells (up to sixfold) (see online supplementary figure S13C). These changes were accompanied by upregulation in OPN (∼14-fold) (see online supplementary figure S13D), and Sox9 mRNA (∼40% increase) (see online supplementary figure S13E).
OPN neutralisation attenuates the liver progenitor response and ameliorates fibrogenesis (interventional)
CCl4 and MCD models
Fibrosis: Aptamers were administered during the final week of liver injury. No mice died. OPN-neutralisation ameliorated liver fibrosis in CCl4-treated and MCD-fed mice. OPN-aptamers significantly repressed hepatic hydroxyproline content and the amount of SR-stained fibrils (figure 4A,B) by fourfold to fivefold, reduced αSMA+ cells by 50–80% (figure 4C,D), and downregulated αSMA (80–90%), Collagen 1αI (50–100%) and TGF-β1 (50–90%) mRNA (see online supplementary figure S9).
Liver progenitors: Inhibited fibrogenesis was accompanied by fewer liver progenitors (figures 5 and 6, see online supplementary figure S10). OPN+ cells and Sox9+ cells were ∼fourfold and ∼twofold to threefold fewer, respectively, after OPN-neutralisation (figures 5A,B and 6A,B). There was parallel repression of Sox9 and K19 mRNA levels (see online supplementary figure S10).
Liver progenitor reprogramming: OPN-neutralisation also downregulated mesenchymal markers, OPN and Snail (figure 7A,B, see online supplementary figure S11A,B) while upregulating epithelial markers, E-cadherin and Id2 (figure 7C,D, see online supplementary figure S11C,D). Immunostaining further revealed the restored-expression of membranous-E-cadherin, and greater number of Sox9/E-cadherin double-positive cells in OPN-aptamer-treated mice, to near normal levels (figure 7E,F, see online supplementary figure S11E,F). This was mirrored by a reduction in Sox9/OPN double-positive cells to basal levels (figures 5C and 6C), thus confirming reversal of the fibrogenic phenotype.
Comparable outcomes were noted in DDC-fed mice: OPN-aptamer treatment led to ∼fourfold less hepatic hydroxyproline, ∼twofold less SR-stained fibrils (see online supplementary figure S12A), ∼threefold fewer αSMA+ cells (see online supplementary figure S12B), and repression of αSMA (threefold), Collagen 1αI (∼25%), and TGF-β (∼30%) mRNA (see online supplementary figure S12C–E). This was accompanied by an attenuated LPC response: fewer OPN+ (∼fourfold) (see online supplementary figure S13A), Sox9+ (∼threefold) (see online supplementary figure S13B), and Sox9/OPN double-positive cells (∼twofold) (see online supplementary figure S13C). There was a comparable downregulation in OPN (threefold), and Sox9 (2.5-fold) mRNA (see online supplementary figure S13D,E). By contrast, Sox9/E-cadherin double-positive cells increased nearly twofold (see online supplementary figure S14A,B) during OPN-neutralisation.
Because OPN can potentially bind to matrix metalloproteinases (MMPs),48 we evaluated expression of extracellular matrix regulation genes. OPN neutralisation led to the downregulation of tissue inhibitor of metalloprotease-1 (TIMP1) and lysyl oxidase (mediates collagen cross-linking) in all three models (by ∼threefold; p<0.05), MMP2 in CCL4 (by ∼twofold; p<0.05), MMP9 in all three models (by ∼threefold to fivefold; p<0.05), and MMP13 in the MCD-treated mice (by ∼twofold; p<0.05). Furthermore, OPN-neutralisation increased MMP2:TIMP1 ratio in the MCD and DDC models (by ∼twofold), and increased MMP13:TIMP1 ratio in the CCL4 and DDC models (by ∼twofold).
OPN is an immune-cell chemoattractant,44 and could modulate the degree of hepatic injury. Thus, we evaluated if OPN-neutralisation could lead to differences in serum alanine aminotransferase. Treatment with OPN-aptamers led to greater reductions in alanine aminotransferase levels in MCD (385±42 IU/L to 102±22 IU/L; p<0.05) )-fed mice and DDC (846±75 IU/L to 133±15 IU/L; p<0.05)-fed mice, compared with CCL4-treated mice (438±44 IU/L to 321±71 IU/L). These aggregate data demonstrate that OPN-aptamers ameliorate liver fibrogenesis via multiple pathways: (1) directly—by modulating the LPC and HSC responses, (2) indirectly—by reducing hepatic injury and/or regulating matrix degradation.
The outcomes of OPN-aptamer-studies were verified using OPN-neutralising antibodies. MCD-fed mice that received OPN-antibodies accumulated fourfold to fivefold less SR-stained fibrils (see online supplementary figure S15A), 60% fewer αSMA+ cells (see online supplementary figure S15B), and downregulated αSMA, Collagen 1αI, TGF-β1 mRNA by up to 60% (see online supplementary figure S15C–E). This was associated with fewer OPN+ and Sox9+ cells (∼80%) (see online supplementary figure S16A,B), and reduced OPN and Sox9 mRNA (∼50%) (see online supplementary figure S16C,D).
In summary, targeting OPN using two neutralising approaches, and in three liver disease models, led to an attenuated liver progenitor cell response, and an amelioration of murine liver fibrosis.
This is the first study to show that OPN-neutralisation is effective in treating murine liver fibrosis. Treatment with OPN-aptamers or OPN-neutralising antibodies attenuated liver progenitor cell response, and repressed fibrogenesis to levels comparable with control-fed mice, suggesting that 1 week of ‘anti-OPN’ treatment is safe, and effective in reversing fibrogenic outcomes. Mechanistically, we show that OPN is an important viability factor for liver progenitors, and directly regulates progenitor cell phenotype by upregulating mesenchymal genes while repressing epithelial ones. Importantly, OPN-neutralisation significantly inhibited progenitor cell migration in wound healing and transmigration assays, key features of EMT.42
EMT describes the process by which epithelial-progenitors acquire a more mesenchymal phenotype that facilitates their migration into the stroma.49 This process is critical for development and is characteristic of invasive cancers. Fate-mapping studies in three distinct mouse strains and in two models of chronic liver injury provide compelling evidence that EMT occurs during liver regeneration and repair.16 ,40 ,50 Comparable features of EMT are detected in human diseased livers.22 ,24 In this study, OPN-neutralisation led to fewer Sox9+ LPC, and significantly less hepatic K19 and Sox9 mRNA (ie, attenuated liver progenitor response) than sham-treated mice. Importantly, Sox9+ LPC lost E-cadherin expression with fibrosis progression, but regained epithelial-type expression during OPN-neutralisation. Sox9+ LPC also lost expression of the mesenchymal marker, OPN, during fibrosis regression. The collective in vitro and in vivo data support the hypothesis that OPN regulates progenitor cell-wound healing responses.
The liver progenitor niche comprises the LPC (oval cell), bone marrow-derived progenitors and liver fibroblasts (HSC).51 Previously, we showed that cocultures of LPC with HSC led to enhanced progenitor cell proliferation and EMT,22 while LPC-derived Hh ligands and OPN activate HSC into myofibroblasts.26 Here, we showed that OPN-neutralisation resulted in a ‘less-fibrogenic’ LPC-secretome in vitro and a ‘less-fibrogenic’ progenitor cell microenvironment in vivo, highlighting the importance of OPN within the progenitor niche. This finding is particularly relevant given our recent study identifying HSC as a resident multipotent progenitor,16 ,40 and provides an explanation for their shared phenotype (ie, LPC and HSC express similar progenitor markers, and undergo EMT).24 ,25 Thus, LPC and HSC interact to replenish the progenitor pool,16 and identify OPN as a key modulator of the liver progenitor-associated response during injury.
OPN binds to multiple receptors.44 As such, it is likely that OPN-neutralisation has other effects, apart from modulating LPC and stellate cell responses. Preliminary studies show that OPN can promote immune cell entry into the liver, and perpetuate hepatic injury. Others have reported that OPN can directly regulate immune-cell functions.52 Our findings that OPN-neutralisation resulted in the downregulation of key extracellular matrix regulators, TIMP1, lysyl oxidase and MMPs, while increasing the ratios of MMP:TIMP1 is unsurprising, as OPN and its family members are known to bind to MMPs.48 Nevertheless, these observations explain in part, how OPN-neutralisation could lead to such impressive reversibility in fibrosis. Further studies however, will be needed to evaluate how OPN modulates MMP activities and whether OPN regulates specific immune-subsets recruitment and function (fibrosis-promoting vs fibrosis-regressing) in vivo.53
TGF-β is a profibrogenic cytokine, and promotes progenitor cell EMT.39 Given that OPN also regulates the liver progenitor response and EMT, it is not surprising that TGF-β and OPN may interact. Previously, we showed that TGF-β mRNA expression is Hh-regulated.24 In this study, we confirmed that OPN-neutralisation in mice led to reduced TGF-β1 mRNA. Intriguingly, OPN-neutralisation also decreased levels of phospho-Smad-2/3-complexes in LPC. This was mirrored by increased Ski and SnoN, two potent transcriptional co-repressors of the TGF-β pathway.45 Under basal conditions, Ski and SnoN inhibit gene transcription; TGF-β stimulation leads to Ski and SnoN degradation, thereby, allowing phospho-Smad-2/3 to bind to target genes. The results imply that OPN may be a novel regulator of SnoN and Ski levels during fibrogenic liver repair, and OPN-neutralisation decreases levels of phospho-Smad-2/3 which, in turn, triggers proteasomal degradation of Ski and SnoN.54 This is consistent with observations that Ski and SnoN overexpression is associated with amelioration of renal fibrosis, and resistance to renal-tubular EMT.55 Our preliminary studies further suggest that OPN-mediated effects may occur via OPN-CD44 and OPN-αvβ3 interactions (putative OPN-receptors), and/or via MZF1 regulation of TGF-β mRNA (a zinc finger transcription factor) (Mi Z, personal communications). The aggregate findings show that OPN effects are mediated in part, by modulating TGF-β signalling, complementing and extending earlier evidence which positioned OPN and TGF-β downstream of Hh during liver fibrogenesis.24 ,26
To date, studies have used OPN-knockout (genetically deficient) mice.26 ,33 ,56 Despite their usefulness in providing proof of concept, translating findings from these animals to humans are limited, as genetic modifications may result in contradictory outcomes when subjected to chronic injury. These disparities could be explained by the presence of distinct OPN isoforms (intracellular and extracellular/soluble) which exhibit overlapping, but also differing functions.44 Soluble OPN behaves as a cytokine while intracellular OPN is an important viability and cytoskeletal protein that regulates intracellular protein functions. In support, our cell culture data show that the reduction of intracellular OPN expression (in 603B-shOPN) leads to similar but non-identical outcomes. In vivo, OPN-knockout mice developed more fibrosis after chronic CCL4,56 but less fibrosis in dietary-induced NASH.26 Similar divergent outcomes have been reported in OPN-knockout mice with lung or rheumatological disease.57 ,58 OPN-aptamers or OPN-neutralising antibodies used in this study are potentially safer because they negate the excess circulating OPN without directly abolishing the expression of intracellular OPN necessary for key cellular function.47 Furthermore, in clinical practice, individuals are more likely to need treatment for advanced-fibrosis, as opposed to prophylactic antifibrotics; therefore, the administration of OPN-neutralising therapies once fibrosis has developed is more likely to be clinically relevant.1
Although no mice died in this study, future studies need to specifically evaluate the potential risks of OPN-neutralisation. OPN overexpression, however, occurs during tissue inflammation and fibrosis (skin, lung, kidneys, bone marrow), and in individuals with metabolic risk factors such as obesity, endothelial dysfunction and diabetes.44 ,59 Thus, neutralising and lowering excess extracellular OPN under such circumstances could be beneficial. This concept and therapeutic safety is supported by a recent phase 1/2 study of OPN-neutralisation in patients with advanced rheumatoid arthritis.30
In summary, our analyses in cell culture, mice and humans show that OPN upregulation during liver injury is a conserved repair response, and influences liver progenitor cell function by modulating TGF-β signalling. OPN neutralisation using two neutralising modalities, and in three different models of mice abrogated the liver progenitor cell response and liver fibrosis. Future studies will be necessary to evaluate the importance of alternative mechanisms by which OPN modulates fibrogenesis, and to evaluate if extended periods of treatment could lead to even better antifibrotic outcomes. As humanised antibodies to OPN and OPN-specific aptamers are currently being developed, future studies will also be needed in humans to evaluate the safety and efficacy of anti-OPN treatment.
Dr G J Gores (Mayo Clinic, Rochester, Minnesota, USA) and Yoshiyuki Ueno (Tohoku University, Sendai, Japan) for providing the murine immature ductular cell line (603B), Dr Scott L Friedman (Mount Sinai School of Medicine, New York, USA) for providing the human HSC line, LX-2, and Dr George Yeoh (University of Western Australia) for providing the BMOL line. The authors are grateful to Dr Mari Shinohara (Duke University Medical Center, North Carolina, USA) and Dr T Uede (Hokkaido University, Japan) for helpful discussions.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
DR, BE and LC are shared authorship.
Contributors JC, MS-S, LD, DR, BE, LC, MAB-O, SS, YHO, AR, SC, SSC performed experiments and contributed intellectually to the study; SP, ZM, PCK, RW, AC, DHA, LAvG, AMD, SSC provided samples, and contributed to the design and funding of the studies. JC, MAB-O, RW, and AMD cowrote the manuscript. WKS is the lead investigator, who designed and supervised the overall project and substudies, performed experiments, wrote the manuscript, and is the senior author and guarantor of the manuscript.
Competing interests None.
Funding This study was funded predominantly by CORE-UK (WKS), BRET (WKS), EASL (WKS), The Foundation for Liver Research London (WKS), and the University of Birmingham (WKS). Additional funding was provided by the National Institute of Health 5K08DK080980 (SSC), R01 DK077794 (AMD), Belgian Federal Science Policy Office (Interuniversity Attraction Poles programme—P6/20 and P7/83-HEPRO) (LD, LAvG), the Brussels Capital Region (INNOVIRIS Impulse programme-Life Sciences 2007 and 2011; BruStem) (LD, LAvG), the Institute for the Promotion of Innovation through Science and Technology in Flanders (SBO-IWT-090066 HEPSTEM) (LD, LAvG), the Natural Sciences and Engineering Research Council of Canada Postgraduate Doctoral Scholarship (DR, BE), the Alberta Innovates Technology Futures Graduate Scholarship (DR, BE), DFG (German Research Association) CA267/8-1 (AC) and the Wilhelm Lapitz Foundation (AC).
Ethics approval University of Birmingham Local ethical approval 04/Q2708/41 and REC 2003/242 from the South-Birmingham Research Ethics Committee, UK; and University Hospital Essen, Germany under the local ethics commission (09-4252).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.