ADAM metallopeptidase with thrombospondin type 1 motif 2 inactivation reduces the extent and stability of carbon tetrachloride–induced hepatic fibrosis in mice


  • Frédéric Kesteloot,

    1. Laboratory of Connective Tissues Biology, Interdisciplinary Cluster for Applied Genoproteomics/Center for Experimental Cancer Research, University of Liège, Liège, Belgium
    Search for more papers by this author
  • Alexis Desmoulière,

    1. Groupe de Recherches pour l'Etude du Foie, Institut National de la Santé et de la Recherche Médicale E0362, Université Victor Segalen Bordeaux 2, Bordeaux, France
    2. Department of Physiology, Faculty of Pharmacy, University of Limoges, Limoges, France
    Search for more papers by this author
  • Isabelle Leclercq,

    1. Gastroenterology Unit, Université Catholique de Louvain, Brussels, Belgium
    Search for more papers by this author
  • Marc Thiry,

    1. Laboratoire de Biologie Cellulaire et Tissulaire, Université de Liège, Liège, Belgium
    Search for more papers by this author
  • Jorge E. Arrese,

    1. Department of Dermatopathology, University Hospital of Liège, Sart Tilman, Belgium
    Search for more papers by this author
  • Darwin J. Prockop,

    1. Center for Gene Therapy, Tulane University Health Science Center, New Orleans, LA
    Search for more papers by this author
  • Charles M. Lapière,

    1. Laboratory of Connective Tissues Biology, Interdisciplinary Cluster for Applied Genoproteomics/Center for Experimental Cancer Research, University of Liège, Liège, Belgium
    Search for more papers by this author
  • Betty V. Nusgens,

    1. Laboratory of Connective Tissues Biology, Interdisciplinary Cluster for Applied Genoproteomics/Center for Experimental Cancer Research, University of Liège, Liège, Belgium
    Search for more papers by this author
  • Alain Colige

    Corresponding author
    1. Laboratory of Connective Tissues Biology, Interdisciplinary Cluster for Applied Genoproteomics/Center for Experimental Cancer Research, University of Liège, Liège, Belgium
    • Laboratory of Connective Tissues Biology, Tour de Pathologie, B23/3, 4000 Sart Tilman, Belgium===

    Search for more papers by this author
    • fax: 32-4-366-2457

  • Potential conflict of interest: Dr. Prockop owns stock in FibroGen.


ADAMTS2 belongs to the “ADAM metallopeptidase with thrombospondin type 1 motif” (ADAMTS) family. Its primary function is to process collagen type I, II, III, and V precursors into mature molecules by excising the aminopropeptide. This process allows the correct assembly of collagen molecules into fibrils and fibers, which confers to connective tissues their architectural structure and mechanical resistance. To evaluate the impact of ADAMTS2 on the pathological accumulation of extracellular matrix proteins, mainly type I and III collagens, we evaluated carbon tetrachloride–induced liver fibrosis in ADAMTS2-deficient (TS2−/−) and wild-type (WT) mice. A single carbon tetrachloride injection caused a similar acute liver injury in deficient and WT mice. A chronic treatment induced collagen deposition in fibrous septa that were made of thinner and irregular fibers in TS2−/− mice. The rate of collagen deposition was slower in TS2−/− mice, and at an equivalent degree of fibrosis, the resorption of fibrous septa was slightly faster. Most of the genes involved in the development and reversion of the fibrosis were similarly regulated in TS2−/− and WT mice. Conclusion: These data indicate that the extent of fibrosis is reduced in TS2−/− mice in comparison with their WT littermates. Inhibiting the maturation of fibrillar collagens may be a beneficial therapeutic approach to interfering with the development of fibrotic lesions. (HEPATOLOGY 2007.)

Chronic damage to the liver in a variety of conditions leads to scar formation resulting from an increased deposition of extracellular matrix (ECM) proteins. The main component of the newly formed ECM is collagen.1 Ultimately, hepatic fibrosis leads to cirrhosis, which is characterized by a pathological accumulation of fibrous tissue impairing hepatic function by the disruption of the organ structure2 and abnormal interactions with resident liver cells. The loss of differentiated functions of hepatocytes and the reduction of sinusoidal cell porosity, leading to the reduced transport of solutes from sinusoidal blood to the subendothelial space, can collectively explain the metabolic dysfunction characteristic of advanced liver disease.3 Currently, no specific therapy for fibrosis is available. Two approaches aimed at the reduction of excessive scar formation are theoretically possible. The first would consist of the stimulation of specific proteolytic pathways in order to accelerate the degradation of the excessive ectopic ECM. The second would be aimed at inhibiting the synthesis of the fibrous extracellular macromolecules and/or altering their structure. This would result in the deposition of defective scar tissue prone to rapid degradation by physiological remodeling processes. Despite the complex regulatory events leading to fibrosis, the biosynthesis of collagen is a well-defined process, and this makes it a suitable target for therapeutic intervention.4 Its turnover in normal adult tissues is quite slow in comparison with the active remodeling occurring in liver fibrosis, which should minimize side effects of the therapy.

The most abundant fibrillar collagens accumulating during the fibrotic process are type I and type III collagens.5, 6 They are synthesized as precursors (procollagens) formed by a central triple-helix domain extended by propeptides at both extremities. The final step in the posttranslational processing of procollagen molecules is the cleavage of the C-propeptide by bone morphogenetic protein-1 and related mammalian tolloid-like 1 metalloproteinases with an astacin-like protease domain.7, 8 This cleavage is followed by the excision of the aminopropeptide (N-propeptide) by aminoprocollagen (pNcollagen) peptidases9 now identified as ADAMTS2,10 ADAMTS3,11 and ADAMTS14,12 3 closely related metalloproteinases that belong to the “ADAM metallopeptidase with thrombospondin type 1 motif” (ADAMTS) family. Removal of the propeptides is required to generate collagen monomers able to assemble into elongated and cylindrical collagen fibrils to form compact bundles, which confer to the connective tissues mechanical resistance and informational properties. For example, the copolymerization of pNcollagen type I (collagen I that retains its aminoterminal extension) with fully processed type I collagen results in altered fibril organization both in vitro13, 14 and in vivo.15 ADAMTS2 is responsible for most of the pNcollagen type I processing9, 16–18 and is able to process pNcollagen type II in cartilage.11, 19 Recently, it has been shown that ADAMTS2 is also able to process pNcollagen types III20 and V.21 As ADAMTS2 is a key regulator of procollagen maturation and collagen fibril formation, its inhibition should affect collagen deposition and scar tissue stability.

In this study, the potential antifibrotic effect of the inhibition of ADAMTS2 was investigated with an in vivo model of liver fibrosis induced by the administration of carbon tetrachloride (CCl4) in ADAMTS2-deficient (TS2−/−) mice in comparison with wild-type (WT) littermates.


1wr, 1 week of recovery; 2w, 2 weeks; 4w, 4 weeks; α-SMA, alpha-smooth muscle actin; α1(III), α1 chain of type III collagen; ADAMTS, ADAM metallopeptidase with thrombospondin type 1 motif (as defined in the National Center for Biotechnology Information databank for both gene and protein); ALT, alanine aminotransferase; AST, aspartate aminotransferase; BDL, bile duct ligation; C, control; CCl4, carbon tetrachloride; CPS, carbamyl phosphate synthase; CTGF, connective tissue growth factor; CV, centrilobular vein; ECM, extracellular matrix; MMP, matrix metalloproteinase; mRNA, messenger RNA; NI, not injected; N-propeptide, aminopropeptide; OO, olive oil; pNα, α chain containing the aminopropeptide; pNcollagen, aminoprocollagen; PSR, picrosirius red; PT, portal tract; RT-PCR, reverse-transcription polymerase chain reaction; SEM, standard error of the mean; TGF-β1; transforming growth factor-beta 1; TIMP, tissue inhibitor of metalloproteinases; TS2−/−, ADAMTS2-deficient; WT, wild type.

Materials and Methods

Animals and Administration of CCl4.

The generation and characterization of TS2−/− mice of the 129SVJ strain have been described.22 Mice were maintained under standard laboratory conditions, with 12-hour light/12-hour dark cycles and free access to food and water at all stages of the experiments. For acute CCl4-induced liver injury, a single CCl4 dose of 150 μL/kg of body weight (Fluka Chemika, Buchs, Switzerland) as a 3% (vol/vol) solution in olive oil (OO) was administered by intraperitoneal injection, the control animals being not injected (NI). For chronic CCl4-induced liver fibrosis, a CCl4 dose of 150 μL/kg of body weight as a 3% (vol/vol) solution in OO or a CCl4 dose of 300 μL/kg of body weight as a 6% (vol/vol) solution in OO was injected intraperitoneally 3 times a week for 4 weeks. Control animals were injected with OO alone. In some experiments, mice were induced with 300 μL/kg CCl4 for 4 weeks and sacrificed 4 (peak fibrosis), 11 (1 week of recovery), or 18 days (2 weeks of recovery) after the last injection. In other reversal experiments, in order to induce a similar degree of fibrosis in both types of mice, WT mice were injected for 2 weeks and TS2−/− mice were injected for 4 weeks with CCl4 (300 μL/kg) and then sacrificed 4 (peak fibrosis) or 11 days (1 week of recovery) after the last injection. Groups of age-matched and sex-matched WT or TS2−/− mice were sacrificed at the indicated times. After anesthesia and exsanguination, livers were collected and weighed. One lobe was fixed in 4% formalin for histological analyses, and the other was sampled according to a standardized procedure and frozen for biochemical and molecular biological analyses. All procedures were performed in accordance with the institutional, Federation of European Laboratory Animal Science Associations, and American National Institutes of Health guidelines for animal care.

Bile Duct Ligation (BDL) Model.

WT and TS2−/− mice were subjected to BDL as described.23 Briefly, the common bile duct was double-ligated after a midline abdominal incision. The animals were sacrificed 9 days after BDL, and their livers were collected for morphological studies.

Measurement of the Collagen Content.

A frozen piece of the left lobe of the liver was lyophilized, weighed, and hydrolyzed in 6 M hydrochloric acid at 137°C for 3 hours. The collagen content was estimated from hydroxyproline measured with the technique of Bergman and Loxley.24 The hydroxyproline assays performed on 1 lobe were validated by a comparison with measurements performed on a pool of 3 lobes in a series of untreated and CCl4-treated WT and TS2−/− mice (not shown).


Sections (5 μm) were stained with hematoxylin/eosin for a general histology and necrosis assessment and immunostained with a mouse monoclonal antibody against alpha-smooth muscle actin (α-SMA; 1/100; Dako, Glostrup, Denmark) biotinylated with an animal research kit (Dako). The quantification of α-SMA–positive cells was performed in 10 fields per liver, focused on the border of a centrilobular vein, in CCl4-treated WT (n = 4) and TS2−/− mice (n = 5).

Quantitative Image Analysis of Picrosirius Red (PSR) Staining.

Fibrosis was evaluated in WT and TS2−/− mice by the quantitative image analysis of sections stained with PSR to visualize collagen fibers.25 Images were acquired with a Zeiss Axiovert 25 microscope (Carl Zeiss Microscopy, Jena, Germany) equipped with an AxioCam camera, and quantitative data were obtained with the KS 400 imaging system. The analysis was performed on 6 fields per section with a 10× objective. Stained vessel walls were systematically excluded to quantify only the fibrotic scar septa [overlaid by green pixels; see Fig. 4A, panels e′ and f′ (shown later)]. Fibrosis was expressed as the mean percentage of the PSR-stained areas of the total field tissue area.

Western Blotting Analysis.

Liver samples were homogenized on ice in an aqueous solution containing a cocktail of protease inhibitors [3 tablets (Complete Mini, Roche) in 20 mL of 5 mM ethylenediamine tetraacetic acid, 2.5 mM N-ethylmaleimide, and 0.5 mM phenylmethylsulfonyl fluoride]. An aliquot of the total homogenate was mixed with the same volume of a 2× concentrated Laemmli buffer and heated at 65°C for 15 minutes. An amount of lysate equivalent to 0.5 (for collagen I) or 2.25 mg of liver (for collagen III) was electrophoresed under nonreducing (for collagen I) or reducing conditions (50 mM dithiothreitol for collagen III) on a 7.5% sodium dodecyl sulfate–polyacrylamide gel according to the technique of Laemmli.26 Western blotting was performed as described.27 Some gels were stained with Coomassie blue to visualize carbamyl phosphate synthase (CPS), which was used as a standard to monitor the protein loading. The polyvinylidene difluoride membranes were probed with a rabbit antiserum directed against type I collagen (1/3000) or with a guinea pig antiserum directed against type III collagen (1/4000).28 The secondary peroxidase-conjugated antibodies were swine anti-rabbit immunoglobulin (1/2000; P0217, Dako) and rabbit anti–guinea pig immunoglobulin (1/2000; P0141, Dako). Peroxidase was revealed with an enhanced chemiluminescence assay (Amersham Biosciences, Ltd., England) and X-ray film exposure.

RNA Purification and Reverse-Transcription Polymerase Chain Reaction (RT-PCR) Analysis.

The total RNA was extracted from the whole liver with a NucleoSpin RNA II kit (Macherey-Nagel, Düren, Germany), according to the manufacturer's recommendations. The RT-PCR amplifications were performed in an automated thermocycler (GeneAmp PCR System 9700, Applied Biosystems, Foster City, CA) with a GeneAmp Thermostable rTth Reverse Transcriptase RNA PCR Kit (Applied Biosystems, Foster City, CA). Primer sequences of genes used for the quantification of messenger RNA (mRNA) levels by RT-PCR are listed in Supplementary Table 1. For the α1 chain of type III collagen [α1(III)], matrix metalloproteinase-2 (MMP-2), MMP-9, MMP-13, MMP-14, tissue inhibitor of metalloproteinases-3 (TIMP-3), and 28S ribosomal RNA, the efficiency of RT-PCR was controlled by a synthetic RNA cotranscribed and coamplified with the same primers as the endogenous RNA to yield an amplification product of a larger size.29, 30 The RT-PCR products were quantified after electrophoresis on a 10% polyacrylamide gel and staining (Gelstar, FMC BioProducts) with a Fluor-S MultiImager (Life Science, Bio-Rad Laboratories) and normalized as a ratio to 28S ribosomal RNA. For all investigated genes, the mRNA levels were similar in control groups subjected to acute (NI mice) and chronic (4 weeks of OO) treatments, and this allowed us to pool the NI and OO groups for statistical tests.

Blood Enzyme Analyses.

As a quantitative measure of liver injury by CCl4, the activity of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) in the serum was determined with commercially available kits (ALT [ALAT/GPT] for ALT and AST (ASAT/GOT) for AST, Roche, Mannheim, Germany) according to the manufacturer's instructions.

Electron Microscopy.

Fragments of livers from WT and TS2−/− mice were fixed for 60 minutes at room temperature in 2.5% glutaraldehyde in Sörensen's buffer (0.1 M, pH 7.4), postfixed for 30 minutes in 0.1% osmium tetroxide in the same buffer, dehydrated in a series of ethanol concentrations, and embedded in an epoxy resin (Epon 812, Fluka). Ultrathin sections were stained with uranyl acetate and lead citrate before being examined with a JEOL CX100 II electron microscope at 60 kV. The measurement of the diameter of at least 10 fibers in a transverse section per photograph was performed for at least 4 micrographs per group of mice (n = 2).


Liver samples were ground into liquid nitrogen, homogenized in an extraction buffer [50 mM trishydroxymethylaminomethane, 0.1% Brij, 2 M urea, 1 M NaCl (pH 7.6), and 0.1 mM phenylmethylsulfonyl fluoride] and centrifuged. The supernatant was dialyzed 3 times against 50 mM trishydroxymethylaminomethane, 0.1% Brij, 10 mM CaCl2 (pH 7.6), and 0.1 mM phenylmethylsulfonyl fluoride, and the total protein content was assessed with a bicinchoninic acid assay (Micro BCA Protein Assay Kit, Pierce, Rockford, IL). Equal amounts of proteins were loaded onto gelatin sodium dodecyl sulfate–polyacrylamide gels, and zymography was performed as described.30

Statistical Analysis.

Animals were randomly assigned to control and treatment groups. The results are presented as the means ± the standard error of the mean (SEM). The significance of the differences between means was assessed with a 2-tailed Student t test or a Mann-Whitney test, when appropriate. P < 0.05 was considered significant.


ADAMTS2 Deficiency Does Not Alter the Acute Reactivity to Liver Injury Induced by a Single CCl4 Injection.

In order to determine whether the absence of ADAMTS2 had any influence on the acute liver response to CCl4 exposure, the extent of the initial necrotic reaction and its resolution were analyzed after a single dose of CCl4 (Fig. 1). In both WT and TS2−/− mice, centrilobular necrotic lesions were already present on day 1, peaked on day 2, and had almost completely recovered on day 4 after the CCl4 treatment. After 7 days, inflammatory cells were almost absent. At all investigated times, no histological difference could be found between WT and TS2−/− mice. Consistent with these results, both the ALT and AST activities, hallmarks of liver injury,31 measured in the serum (Fig. 2) showed similar peaks in WT and TS2−/− mice 1 day after CCl4 administration that rapidly and similarly vanished.

Figure 1.

The extent of CCl4-induced hepatic necrosis and its resolution are similar in WT and TS2−/− mice. (A-J) Representative photographs of hematoxylin/eosin staining of liver tissue sections are shown (scale bars: 100 μm). (A,C,E,G,I) WT and (B,D,F,H,J) TS2−/− mice were (A,B) not injected or injected once with 150 μL/kg CCl4 and sacrificed after (C,D) 1 day, (E,F) 2 days, (G,H) 4 days, or (I,J) 7 days.

Figure 2.

The lack of ADAMTS2 activity does not modulate the serum ALT and AST levels following the injection of CCl4. (A) ALT and (B) AST activity in serum collected from (○) WT and (●) TS2−/− mice at the indicated times after a single injection of CCl4 (150 μL/kg). The ALT and AST levels are shown as international units per liter ± the SEM. Each time represents the mean value measured in a minimum of 3 animals, except at days 4 and 7 (n = 2). No difference was observed between WT and TS2−/− mice at any time.

The expression of several genes related to connective tissue synthesis, maturation, and remodeling was assessed with RT-PCR. The level of mRNA coding for transforming growth factor-beta 1 (TGF-β1), connective tissue growth factor (CTGF), and α-SMA (Fig. 3A) was already up-regulated on day 1 to similar extents in WT and TS2−/− mice and remained elevated on day 4. The mRNA of fibrillar collagen types I, III, and V (Fig. 3B) started to increase on day 1, the maximal increase being observed after 4 days. Enzymes involved in the processing and maturation of procollagen (ADAMTS2 in WT mice and ADAMTS14 in WT and TS2−/− mice), the proteolytic enzymes (MMP-2, MMP-9, MMP-13, and MMP-14; Fig. 3C) and some of their inhibitors (TIMP-1, TIMP-2, and TIMP-3; Fig. 3D) were also up-regulated on days 1 and 4 following liver injury. ADAMTS3 expression was not detected in the liver. The strongest induction for both genotypes was observed for MMP-9 and TIMP-1 (see the scale). Except for ADAMTS14 and TIMP-3 after 1 day, no significant difference was observed between WT and TS2−/− mice at any time and for any of the investigated genes, and this means that an ADAMTS2 deficiency does not modify the acute reactivity to liver injury.

Figure 3.

The genes related to acute toxicity liver injury are similarly modulated in WT and TS2−/− livers. The mRNA expression of several genes related to fibrogenesis was measured in WT (empty bars) and TS2−/− mice (filled bars) 1 and 4 days after a single injection of CCl4 (150 μL/kg) and in control (C) animals not receiving CCl4. The investigated genes were related to (A) fibrosis initiation (TGF-β1, CTGF, and α-SMA), (B) fibrillar collagens (types I, III, and V), (C) metalloproteinases (MMP-2, MMP-9, MMP-13, MMP-14, ADAMTS2, and ADAMTS14), and (D) TIMPs (TIMP-1, TIMP-2, and TIMP-3). The results are expressed as the means (± the SEM) of groups from 4-11 mice, except at day 4 (n = 2), in a percentage of the control WT mice. No significant difference was observed between the WT and TS2−/− mice for any gene at any time, except for ADAMTS14 and TIMP-3 at day 1.

Hepatic Fibrosis Is Reduced in TS2−/− Mice.

After chronic injections of CCl4 for 4 weeks, the ALT and AST levels were slightly but similarly up-regulated in WT and TS2−/− mice (data not shown). A histological evaluation of hepatic tissues after PSR staining indicated that WT and TS2−/− mice that were NI (not shown) or were injected with OO alone (Fig. 4A, panels a and b) displayed similar basal amounts of fibrous collagen staining restricted to the vessel walls. After chronic injections with CCl4 (Fig. 4A, panels c and e), WT developed fibrous septa around the centrilobular veins that were already quite visible at the lowest concentration of CCl4 (150 μL/kg) and accentuated at the highest dosage (300 μL/kg). In contrast, the collagen deposition in TS2−/− mice injected with the same amount of CCl4 (Fig. 4A, panels d and f) was less dense. Digital image analysis (Fig. 4A, panels e′ and f′) was used to quantify the stained hepatic collagen accumulating specifically in newly formed scar septa after the systematic exclusion of the stained vessel walls. This quantification (Fig. 4B) confirmed the visual observation that a lower collagen accumulation occurred in TS2−/− animals versus WT animals. The extent of collagen deposition was also compared between males and females during the development of the fibrosis. No significant difference was observed at any investigated time for either WT or TS2−/− mice (data not shown); this allowed us to consider mixed males and females as a group and to compare larger groups of mice from a lower number of litters, improving the statistical strength of the data. The weight of the liver similarly increased by 30% in TS2−/− and WT mice upon a chronic fibrogenic treatment. Despite the elevated background due to the presence of type IV collagen and fibrillar collagens in the vessel walls, hydroxyproline measurements (Fig. 4C) confirmed these results, with significantly reduced levels in TS2−/− animals in comparison with WT animals after chronic injections of CCl4 for 4 weeks.

Figure 4.

Reduced fibrosis is observed in TS2−/− mice after chronic CCl4 administration. (A) Representative PSR staining of liver sections (scale bars: 250 μm). (a) OO-treated WT mouse. (b) OO-treated TS2−/− mouse. (c,e) WT or (d,f) TS2−/− mice were subjected to a chronic CCl4 treatment at a dose of (c,d) 150 or (e,f) 300 μL/kg in OO as described in the Materials and Methods section. (c-f) Livers were collected after 4 weeks of the chronic treatment. (e′,f′) Digital image analyses of PSR staining: fibrotic zones (in green, excluding vessel walls) have been normalized to the field tissue surface (in blue, excluding nontissue area). (B) Digital image analysis quantification of PSR staining. The fibrosis areas were measured as described in the Materials and Methods section in WT (empty bars) and TS2−/− mice (filled bars) after 4 weeks of the OO vehicle only or 4 weeks of CCl4 injected at a dose of 150 or 300 μL/kg, as indicated. The degree of fibrosis was expressed as a mean percentage (± the SEM) of the field tissue area. (C) The hydroxyproline content in the livers of WT (empty bars) and TS2−/− mice (filled bars) was assessed in the same samples on a microgram/milligram dry weight basis after 150 or 300 μL/kg CCl4 exposure as described in the Materials and Methods section. The bars show the means ± the SEM of groups from 6-36 mice. **P < 0.01 and ***P < 0.001 versus the WT group. #P < 0.05, ##P < 0.01, and ###P < 0.001 versus the corresponding control group treated with OO.

These results were confirmed with the BDL model. The PSR staining of liver sections from WT and TS2−/− mice 9 days after BDL displayed extensive peribiliary and interstitial collagen staining around portal tracts that was much less intense in TS2−/− mice (Fig. 5).

Figure 5.

Portal fibrosis induced by BDL is reduced in TS2−/− mice. PSR staining of liver sections from WT (left side) and TS2−/− mice (right side) 9 days after BDL is shown (scale bars: 100 μm). CV indicates centrilobular vein; and PT, portal tract.

Fibrosis Reversal Is More Efficient in TS2−/− Mice.

Further investigations examined the effects of an ADAMTS2 deficiency on the stability of the fibrosis (Fig. 6) by analyzing livers from cohorts of mice of each genotype after 1 week of recovery following the end of CCl4 administration. To test more precisely the rate of regression of fibrosis, 2 groups of WT and TS2−/− mice were treated with CCl4 for 2 and 4 weeks, respectively, to obtain levels of fibrosis as similar as possible. The degree of fibrosis in the TS2−/− mice after 4 weeks of CCl4 was 80% of that of the WT mice treated for only 2 weeks, supporting a lower rate of collagen deposition in the deficient mice. The regression in the TS2−/− mice was somewhat faster (−68%) than that in the WT mice (−54%).

Figure 6.

Reversal of fibrosis in ADAMTS2−/−. (A) Representative PSR staining of liver sections is shown (scale bars: 150 μm). A chronic CCl4 treatment at a dose of 300 μL/kg in OO, as described in the Materials and Methods section, was performed (a) for 2w in WT or (b) for 4w in TS2−/− mice to attain a similar degree of fibrosis. (c,d) The treatment was then discontinued, and the mice were sacrificed after 1wr. (B) PSR staining was quantified by digital image analysis. The degree of fibrosis was expressed as the mean percentage (± the SEM) of the field tissue area (n = 5-7 mice in each group). *P < 0.05 versus the WT mice. 1wr indicates 1 week of recovery; 2w, 2 weeks; and 4w, 4 weeks.

Regulation of the Gene Expression in WT and TS2−/− Mice During Fibrosis and Its Reversal.

After a long-term treatment (4 weeks), most of the investigated genes were strongly up-regulated, with the exception of TGF-β1, CTGF, and α-SMA (Table 1, left) α-SMA was also assessed by immunohistology in control and fibrotic WT and TS2−/− livers (Fig. 7A). Besides an intense staining of the vessel walls of the portal tracts, few (but similar numbers of) α-SMA–positive cells were seen around the centrilobular veins (Fig. 7B). The mRNA levels of fibrillar collagens I, III, and V were significantly increased and similar to those recorded 4 days after a single CCl4 injection in both genotypes. ADAMTS2 and, to a lesser extent, ADAMTS14 were increased in fibrotic WT livers. An increased expression of the investigated MMPs was noted, which was quite extensive for MMP-2, and confirmed at the protein level by zymography analyses (Supplementary Fig. 1). The up-regulation of MMP-9 was not as strong in the chronic condition as in the acute reaction. No significant difference was noted between the WT and TS2−/− mice. The inhibitors of MMPs, TIMP-1, TIMP-2, and TIMP-3 were also overexpressed in the fibrotic livers, but again, no significant difference was observed between WT and TS2−/− mice. Upon the cessation of the treatment (Table 1, right), a progressive return to a normal level of expression was observed for the fibrillar collagens and their processing and degrading enzymes. Altogether, these results show that selected gene expression in reaction to CCl4-induced hepatic injury is similar in WT and TS2−/− mice.

Table 1. Steady-State Levels of mRNAs After CCl4 Administration in WT and TS2−/− Livers and During the Reversal of Fibrosis
  Control (OO + NI)4 Weeks of CCl4 (300 μL/kg)+1 Week of Recovery+2 Weeks of Recovery
  1. The results (means ± SEM) are expressed as arbitrary units per unit of 28S ribosomal RNA in a percentage of the control WT mice. n was 3-17 in each group, except at 4 weeks + 1 week of recovery, for which n was 2-4. No statistical difference between WT and TS2−/− mice was observed. Obvious differences during the evolution of the process have not been labeled to prevent overloading of the table.

Fibrosis initiationTGF-β1100 ± 13117 ± 1371 ± 998 ± 17114 ± 1382 ± 17125 ± 982 ± 22
CTGF100 ± 24100 ± 20173 ± 20178 ± 16230 ± 24149 ± 44133 ± 5299 ± 20
 α-SMA100 ± 14121 ± 1483 ± 21131 ± 29173 ± 93212 ± 121110 ± 21187 ± 64
Structural proteinsColα1(I)100 ± 1294 ± 241328 ± 203940 ± 129605 ± 88542 ± 12367 ± 106486 ± 188
Colα1(III)100 ± 1387 ± 13471 ± 48369 ± 19306 ± 39267 ± 6236 ± 39271 ± 52
 Colα1(V)100 ± 1568 ± 25824 ± 304374 ± 152315 ± 127269 ± 25162 ± 25151 ± 15
ProteinasesMMP-2100 ± 2792 ± 232919 ± 2492366 ± 226622 ± 122477 ± 9231 ± 50417 ± 100
 MMP-9100 ± 1263 ± 31248 ± 120115 ± 12267 ± 23202 ± 113146 ± 18225 ± 116
 MMP-13100 ± 5380 ± 451727 ± 1021736 ± 212213 ± 4554 ± 23201 ± 45176 ± 68
 MMP-14100 ± 3296 ± 36232 ± 71144 ± 43122 ± 28132 ± 36118 ± 32143 ± 71
 ADAMTS2100 ± 9444 ± 28298 ± 79151 ± 7
 ADAMTS14100 ± 22109 ± 22238 ± 44159 ± 30126 ± 7192 ± 30124 ± 30101 ± 44
InhibitorsTIMP-1100 ± 10080 ± 32200 ± 32416 ± 16016 ± 16217 ± 73112 ± 5771 ± 46
 TIMP-2100 ± 1797 ± 20278 ± 48237 ± 54178 ± 64201 ± 15127 ± 20123 ± 32
 TIMP-3100 ± 19127 ± 38152 ± 5131 ± 23129 ± 24127 ± 2074 ± 695 ± 20
Figure 7.

Immunohistochemical staining of α-SMA. (A) Representative α-SMA immunostaining of liver sections is shown (scale bars: 50 μm). (a) NI WT mouse. (b) NI TS2−/− mouse. (c) WT or (d) TS2−/− mice subjected to a chronic 150 μL/kg CCl4 treatment and sacrificed after 4 weeks. Besides the intense staining of the vessel walls (black arrows), α-SMA staining can be observed in elongated cells after the chronic CCl4 treatment (white arrows). (B) The number of α-SMA-positive cells (± SEM) was quantified in 10 fields per mouse (n = 4-5 mice in each group).

Structural Architecture and Characterization of Newly Deposited Collagen in Fibrotic Livers.

The structural architecture of newly deposited collagen in the septa was visualized with electron microscopy (Fig. 8A). In comparison with the fibrotic livers in WT mice, in which regular and cylindrical collagen fibers were parallel and were densely packed into bundles, the fibrillar collagen network in TS2−/− livers was less abundant and less dense. The fibers presented slightly irregular contours and a significantly smaller cross-section diameter (Fig. 8B).

Figure 8.

Electron microscopy images of livers from WT (n = 2) and TS2−/− mice (n = 2) treated with CCl4 at 300 μL/kg for 4 weeks. (A) The photographs are representative of several sections per liver in each group (scale bars: 200 nm). (B) The diameter of the fibers, measured in transverse sections in 4 fields per group, is significantly reduced in the fibrous septa of TS2−/− mice. ***P < 0.001 versus the WT mice.

The extent of processing of hepatic fibrillar pNcollagen deposited during the fibrotic process was investigated with western blotting. Figure 9A illustrates the patterns of type I collagen polypeptides extracted from the skin (lane 1) and liver (lane 3) of TS2−/− mice and from the same amounts of skin and liver extracts mixed together (lane 2). It first demonstrates that the anti-collagen I antibody reacted more efficiently with α2 and α2 containing the N-propeptide (pNα2) than with α1 and pNα1. Furthermore, although pNα1 was quite visible in the skin extract, it became undetectable in the mixture with the liver extract, which contained, as revealed by Coomassie blue staining (not shown), a significant amount of a liver protein displaying the same electrophoretic migration as pNα1. This protein was not affected by bacterial collagenase digestion that removed all the collagen bands nor was it affected by reduction. It was potentially identified as CPS. Its large abundance in the liver extract probably prevented the efficient transfer of pNα1 during blotting, as suggested by Fig. 9A. This protein stained by Coomassie blue was used to monitor the amount of protein loading in Fig. 9B (top panel). In nontreated animals, the α2 chain was clearly visible in WT mice and was accompanied by very little staining of α1. These protein bands were barely apparent in the TS2−/− mice. In CCl4 fibrotic animals, both α1 and α2 chains were increased in the WT mice, whereas in the TS2−/− mice, type I collagen was lower, and only pNα2 and α2 were clearly identified. α1(III) and pNα1(III) in mouse livers were identified by a comparison with bovine purified dermatosparactic collagen III (positive control; Fig. 9C). Almost only α1(III) was observed in nontreated WT or TS2−/− animals. After the CCl4 treatment, type III collagen significantly accumulated as fully processed α1(III) and little pNα1(III) in WT mice, whereas in the TS2−/− mice, pNα1(III) and α1(III) similarly increased.

Figure 9.

Western blot analysis of type I and type III collagens. (A) Patterns of type I collagen polypeptides extracted from TS2−/− skin (lane 1) and liver (lane 3) with an antiserum against type I collagen are shown. In lane 2, skin and liver extracts were mixed. (B) Liver extracts of 3 WT and TS2−/− mice harvested after 4 weeks of an OO treatment or 300 μL/kg CCl4 injections were submitted to western blotting with anti–type I collagen antiserum. Collagen purified from TS2−/− mouse skin was used as a positive control. The abundant hepatic protein (CPS) was used to monitor the amount of loaded proteins. (C) Patterns of type III collagen polypeptides in the same livers of WT and TS2−/− mice treated with OO or CCl4 are shown. Purified type III collagen from dermatosparactic bovine skin containing both pNα1(III) and α1(III) chains was used as a positive control.


The excessive deposition of fibrous collagen within hepatic tissue is a frequent result of chronic injury to the liver by a number of causative agents, leading to cirrhosis as an endpoint. Effective treatments for liver fibrosis are still lacking, except for the suppression of the initial fibrogenic stimulus. The biosynthesis and polymerization of collagen, the main component of the fibrous scar, require various successive steps, thus providing a large panel of possibilities for therapeutic intervention.

In contrast to the complex regulatory events leading to fibrosis, the biosynthesis of collagen is a well-described multistep process that is quite similar in all tissues. In this study, we focused our attention on ADAMTS2, the main enzyme catalyzing the N-propeptide excision of procollagens I, III, and V,21, 32 a crucial step that allows the polymerization of the triple-helix collagen monomers to assemble and form mechanically resistant fibers and bundles. The lack of activity of ADAMTS2 is responsible for a heritable disease in humans (Ehlers-Danlos syndrome type VIIC) and animals (dermatosparaxis) characterized by extreme fragility of the skin16, 17 and other organs. Mice in which the ADAMTS2 gene has been invalidated (TS2−/−) display the same defect in skin and some other connective tissues.22, 33 These mice were used as an in vivo model to evaluate a potential therapeutic benefit of inhibiting ADAMTS2 function during the development of liver fibrosis.

The acute responses of WT and TS2−/− mice to a single injection of CCl4 were similar, as shown by the histology of the necrotic lesions, the serum level of hepatic enzymes, and the mRNA levels of fibrogenic cytokines, collagen, MMPs, and TIMPs. In agreement with previous reports, we observed a clear peak of overexpression of genes, such as MMP-13, MMP-9, and TIMP-1, within the first day of acute toxic liver injury in both WT and TS2−/− mice, which was probably related to the release of inflammatory cytokines.34 A similar peak of TIMP-3 was observed 1 day after injury in the WT mice. It could be regarded as an acute phase protein together with TIMP-1.35 Besides its activity as an inhibitor of MMPs, TIMP-3 controls the activity of ADAM metallopeptidase domain 17, a pro–tumor necrosis factor α processing enzyme. Its lack of activity in TIMP-3 null mice leads to chronic hepatic inflammation and failure of liver regeneration.36 Its higher induction in WT may be related to ADAMTS2 activity on novel substrates that are under investigation.

PSR staining evaluated by automated digital image analysis, hydroxyproline content quantification, and a transcriptomic analysis of a large series of genes potentially participating in the remodeling process were used to assess the development of fibrosis upon chronic exposure to CCl4 and its resolution. All investigated parameters of the liver response to chronic injury were also similarly modulated in WT and TS2−/− animals. It was clearly demonstrated that TS2−/− mice developed less extensive hepatic fibrosis than their WT littermates, and this was related to a slower rate of deposition of poorly structured fibrils of partly processed collagen precursors.

The edification of a fibrous collagen matrix is a complex process of remodeling regulated by a positive balance between the production of new structures and their degradation. This balance is clearly illustrated by the steady-state level of the mRNA; both the WT and the TS2−/− animals display a large and similar increase in collagen mRNAs. The degrading enzymes are also quite similarly increased. The removal of such structures during, for example, fibrosis resolution operates by similar mechanisms in which degradation is greater than deposition. The increased expression of the genes for producing matrix collagens I and III and that of the genes involved in their degradation are similarly modulated in WT and TS2−/− mice. Collectively, these results indicate that the reduced fibrosis in TS2−/− mice is not due to reduced collagen synthesis or up-regulation of proteolytic enzymes but seems directly related to the absence of ADAMTS2 activity responsible for the persistence of the N-propeptide in a significant proportion of type I and III collagen molecules. The steric hindrance introduced by the N-propeptide leads to delayed and/or altered assembly of collagen molecules into fibrils and the formation of thinner and loosely packed fibers. The structural alterations are, however, less marked than those observed in the skin of human beings suffering from Ehlers-Danlos syndrome type VIIC, dermatosparactic animals, and TS2−/− mice.16, 22 These poorly organized polymers are more susceptible to degradation by collagenase (F.K., unpublished data, 2007). This hypothesis is largely supported by the observation that transgenic mice expressing a mutated collagen resistant to collagenase degradation display a higher level of hepatic fibrosis.37

An alternative model for inducing a fibrotic reaction in the liver by ligation of the bile duct23, 38 was also investigated. The extreme fragility of the connective tissues in TS2−/− mice caused early tears in the distended bile duct walls and a high mortality rate before the long-term development of fibrosis. However, a few animals could be sacrificed 9 days after BDL. In WT mice, extensive bile duct proliferation around portal tracts was accompanied by fibrous PSR-stained extensions within the hepatic parenchyma. In TS2−/− mice, however, the PSR staining around the bile ducts in the portal tract was lighter than that in the WT mice, and no fibrotic extension was observed toward the hepatic tissue. This observation is consistent with the reduced fibrosis observed in TS2−/− mice in the CCl4 model.

ADAMTS2, ADAMTS3, and ADAMTS14 are the 3 proteases responsible for the pNcollagen peptidase activity in the organism.12 ADAMTS3 expression was not detected in the livers of our mice, whereas ADAMTS14, expressed and stimulated by CCl4, may compensate for the absence of ADAMTS2, explaining partial processing of the pNcollagen in the TS2−/− livers. The simultaneous inhibition of the 2 pNcollagen peptidases by the use of small interfering RNA or blocking antibodies directed against their catalytic site, for example, would probably prevent processing of the N-propeptide of all fibrillar collagens, resulting in an even higher instability of the collagen fibers. Targeted activation of TIMP-3, recently identified as an inhibitor of ADAMTS2 in vitro,39 may represent an alternative approach.

The inhibition of fibrosis by targeting the far end of its development through the reduction of the stability of fibrous collagen, as described in this article, seems to be a safe strategy. Indeed, the physiological turnover of collagen in adults is a very slow process in comparison with collagen remodeling during fibrogenesis. The side effects resulting from the inhibition of ADAMTS2 activity should therefore be limited and probably fully reversible at the end of the treatment. Reducing collagen stability has already been investigated with inhibitors of prolyl-4-hydroxylase, the enzyme catalyzing the formation of hydroxyprolyl residues responsible for the collagen triple-helix stabilization. The tested inhibitors of prolyl-4-hydroxylase effectively inhibited experimental fibrosis development,40, 41 and this supported the concept of posttranslational modifications of procollagen as suitable targets for therapeutic intervention.

In conclusion, the suppression of ADAMTS2, a critical metalloproteinase for collagen maturation and stabilization, reduces hepatic fibrosis in vivo. Strategies to knock down its expression or activity in the liver may be considered alternative, effective, and basically only slightly aggressive antifibrotic therapies.


We gratefully thank J. Delwaide for his interest and advice and M.-J. Nix and T. Heyeres for their expert technical assistance.