Antifibrotic effects of a tissue inhibitor of metalloproteinase-1 antibody on established liver fibrosis in rats

Authors


Abstract

Liver fibrosis is characterized by increased synthesis, and decreased degradation, of extracellular matrix (ECM) within the injured tissue. Decreased ECM degradation results, in part, from increased expression of tissue inhibitor of metalloproteinase-1 (TIMP-1), which blocks matrix metalloproteinase (MMP) activity. TIMP-1 is also involved in promoting survival of activated hepatic stellate cells (HSCs), a major source of ECM. This study examined the effects of blocking TIMP-1 activity in a clinically relevant model of established liver fibrosis. Rats were treated with carbon tetrachloride (CCl4), or olive oil control, for 6 weeks; 24 days into the treatment, the rats were administered a neutralizing anti–TIMP-1 antibody derived from a fully human combinatorial antibody library (HuCAL), PBS, or an isotype control antibody. Livers from CCl4-treated rats exhibited substantial damage, including bridging fibrosis, inflammation, and extensive expression of smooth muscle α-actin (α-SMA). Compared to controls, rats administered anti–TIMP-1 showed a reduction in collagen accumulation by histological examination and hydroxyproline content. Administration of anti–TIMP-1 resulted in a marked decrease in α-SMA staining. Zymography analysis showed antibody treatment decreased the activity of MMP-2. In conclusion, administration of a TIMP-1 antibody attenuated CCl4-induced liver fibrosis and decreased HSC activation and MMP-2 activity. (HEPATOLOGY 2004.)

Fibrosis is associated with many liver diseases, including hepatitis C virus infection, iron deposition, alcohol consumption, and nonalcoholic fatty liver disease. Hepatic fibrosis results from a net increased synthesis and decreased degradation of extracellular matrix (ECM) proteins. Type I collagen is the most prevalent ECM protein deposited,1 with activated hepatic stellate cells (HSCs) serving as the primary source. Following a fibrogenic stimulus, HSCs activate from their normal quiescent state, whereby they increase synthesis of pro–collagen type I messenger RNA (mRNA) and protein,1, 2 and increase cellular proliferation, migration, and contractility.3, 4 Excess ECM accumulation results in scarring within the tissue. Our understanding of ECM degradation during hepatic fibrosis is still very limited. ECM degradation is mediated by matrix metalloproteinases (MMPs), a family of zinc-dependent enzymes grouped into collagenases, gelatinases, stromelysins, and membrane-type MMPs,5 based upon their substrates. Interstitial collagenases (MMP-1 and MMP-13 in humans, MMP-13 in rats) degrade type I collagen, while gelatinases (MMP-2 and -9) regulate type IV collagen accumulation; MMP-2 also exhibits some interstitial collagenase activity.6

MMP activity is regulated by the tissue inhibitors of metalloproteinases (TIMPs-1-4), which bind in substrate- and tissue-specific manners to MMPs, blocking their proteolytic activity.5, 7, 8 During fibrosis, TIMP mRNA and protein levels dramatically increase; previous studies have shown this in experimentally-induced (carbon tetrachloride (CCl4) administration and bile duct ligation)9 and disease-related hepatic fibrosis.10 Meanwhile MMP levels increase modestly or remain relatively static.9–11 MMP-2 is an exception, in that its levels dramatically increase during fibrogenesis,9–13 and is believed to be involved in degrading normal liver architecture.5 Thus, the net effect of a rapid increase in ECM synthesis, in combination with increased TIMP activity, is formation of the fibrotic scar. Therefore, the interplay between MMPs and TIMPs provides a key point of regulation to target therapies for treating patients with ongoing fibrosis.

Previous studies have shown TIMP-1 may play an important role in the progression of liver fibrosis. Transgenic mice over-expressing human TIMP-1 showed increased fibrosis in response to chronic CCl4 administration.14 The mice also had slower resolution of fibrosis,15 as determined by immunostaining for smooth muscle α-actin (α -SMA), a biomarker of activated HSCs. This correlates with studies showing TIMP-1 expression can inhibit apoptosis of activated HSCs in vitro.15, 16 Furthermore, persistent expression of TIMP-1 in vivo was associated with persistence of activated HSCs, and during resolution of fibrosis, a decrease in TIMP-1 protein levels correlated with decreased numbers of activated HSCs.16 Thus, not only can TIMP-1 act to prevent degradation of the rapidly-accumulating ECM by blocking MMP function, it may also inhibit apoptosis of activated HSCs, the cellular source of type I collagen.

We further examined the role of TIMP-1 in hepatic fibrosis through administration of a human anti-rat TIMP-1 antibody in a rat model of CCl4-induced fibrosis. Rats administered antibody showed decreased levels of liver fibrosis compared to rats treated with phosphate-buffered saline (PBS), as determined by histological analysis and hydroxyproline content. In addition, the antibody caused reductions in α-SMA expression and MMP-2 activity. These changes were not due to nonspecific antibody effects, as administration of an isotype control antibody had no effect on fibrosis. Thus, blocking endogenous TIMP-1 activity may provide a therapeutic approach for treating liver fibrosis, in part through decreasing the accumulation of activated HSCs. Our results are encouraging in light of the fact that fibrosis was established prior to antibody introduction, and rats continued to receive profibrogenic stimulus concurrent with therapy, thus better reflecting the clinical approach in treating liver fibrosis.

Abbreviations

ECM, extracellular matrix; HSC, hepatic stellate cell; mRNA, messenger RNA; MMP, matrix metalloproteinase; TIMP-1, tissue inhibitor of metalloproteinase-1; CCl4, carbon tetrachloride; α-SMA, smooth muscle α-actin; HuCAL, human combinatorial antibody library; PBS, phosphate-buffered saline; IgG, immunoglobin G.

Materials and Methods

Antibody Affinity Determination by Surface Plasmon Resonance Measurements.

For affinity determination, monomeric fractions of affinity- and SEC-purified Fab fragments or purified IgG1 molecules were used. Experiments were conducted in HEPES-buffered saline (HBS) buffer at a flow rate of 20 μL/min at 25°C on a BIAcore instrument (BIAcore, Piscataway, NJ). Antigens in 100 mM sodium acetate pH 5.0 were coupled to a CM 5 sensor chip using standard EDC-NHS coupling chemistry. Applying 3-4 μL of 5 μg/mL TIMP-1 typically resulted in 500 resonance units for kinetic measurements. Sensograms were fitted globally using BIA evaluation software. A monovalent fit (Langmuir binding) was applied to monovalent Fab fragments, while a bivalent fit was applied for immunoglobin G (IgG)s.

IC50 Determination in Rat TIMP-1/rat MMP-13 Assay.

To determine the concentration of antibody that resulted in 50% reversal of inhibition (IC50), purified Fab fragments or IgGs were used. Antibodies were diluted in triplicate in assay buffer containing 0.05% bovine serum albumin. Following addition of rat TIMP-1 (final concentration of 1.2 nmol/L), rat MMP-13 (cloned, expressed, produced, and purified by Bayer, Berkeley, CA; final concentration of 1.2 nmol/L), and a fluorogenic peptide for human MMP-1 (Bachem AG, Torrance, CA; final concentration of 50 μmol/L) and incubation for 1-3 hours at 37°C, fluorescence at 320 nm (excited) / 430 nm (emitted) was measured. MMP-13 is the rat homolog of human MMP-1,5 sharing 46.8% identical sequence and an overall 79.2% similarity with MMP-1, as determined by ClustalX multiple sequence analysis software.17

The following controls were included in the assay and used as reference values for IC50 determination:

  • A: MMP + substrate: defined as 100% MMP activity in absence of antibody and TIMP-1.

  • B: MMP + TIMP-1 + substrate: defined as maximum inhibition achieved in the assay and calculated as a % of total MMP activity.

To define the IC50 value, the following procedure was used:

  • The value for 50% reversal of inhibition (expressed as % activity MMP) was calculated as:

  • Y = [(A − B)/2] + B.

  • MMP activity was plotted against concentration of antibody in the assay.

  • The concentration of antibody that resulted in 50% reversal of inhibition (Y) was read on the x-axis and defined as IC50.

  • Error bars in the graphs were derived from triplicate wells in one assay.

  • Standard deviations for IC50 values were calculated from 3 independent assays.

Specificity of the antibody for rat TIMP-1 versus other TIMPs was not tested. However, we found the antibody binds TIMP-1 with subnanomolar affinity (C.Q.P., unpublished data, 2001), and previous studies performed with antibodies generated in a similar manner against human TIMP-1 showed no affinity for human TIMP-2, -3, or -4 (C.Q.P., unpublished data, 2001); thus we expect no cross-reactivity.

Animal Use and Care.

Hepatic fibrosis was induced by CCl4 administration, as described.18 Male Wistar rats (Harlan, Indianapolis, IN), 200-225 g, were subjected to a 12:12-hour daylight/darkness environment and allowed unlimited access to chow and water. Two weeks prior to CCl4 treatment, rats were put on water containing 35 mg/dL phenobarbital. The protocol was approved by University of North Carolina's Institutional Animal Care and Use Committee.

At day 0, rats were administered an intragastric dose of CCl4 in olive oil (0.08 mL CCl4/1 mL olive oil; initial dose was 412 mg CCl4/kg body weight) or olive oil alone. Body weights were checked weekly, and dosage of olive oil or CCl4 determined by weight change from the previous week, according to Table 1, to minimize mortality.18 If no dose was given in a particular week, treatment was resumed when body weight again began to increase. At day 24, at which point substantial fibrosis was established (determined by sacrifice of several rats; data not shown), rats were administered a subcutaneous injection of a loading dose of human anti-rat TIMP-1 IgG1, PBS, or a human IgG1 isotype control (conditions and abbreviations described in Table 2). At day 26, rats were administered their first maintenance dose (subcutaneously), and subsequent maintenance doses were administered every 72 hours, with the last dose at day 38. At day 41, rats were sacrificed, body and liver weights assessed, and liver tissue harvested for analysis.

Table 1. CCl4 Dosing Regimen
Weekly Weight ChangeDose Administered
Increase 0–5 gSame as previous dose
Increase 6–10 g125% of previous dose
Increase > 10 g150% of previous dose
Decrease 1–5 g75% of previous dose
Decrease 6–10 g50% of previous dose
Decrease > 10 gNo dose
Table 2. Experimental Groups
Group NameNo. of RatsTreatmentTherapeuticLoading DoseMaintenance Dose
Oil/PBS10Olive OilPBS
Oil/TIMP 1010Olive OilTIMP-1 Mab20 mg/kg10 mg/kg
CCl4/PBS20CCl4PBS
CCl4/TIMP 1020CCl4TIMP-1 Mab20 mg/kg10 mg/kg
CCl4/TIMP 320CCl4TIMP-1 Mab6 mg/kg3 mg/kg
PBS6CCl4PBS
IgG6CCl4IgG Ab3 mg/kg1 mg/kg

Protein Extraction.

Protein was extracted from liver tissue by homogenization in RIPA buffer (50 mM Tris pH 7.2, 150 mM NaCl, 1% deoxycholic acid, 1% Triton X-100, and 0.1% SDS) containing 10 mM 4-nitrophenyl phosphate, 20 mmol/L β-glycerophosphate, 500 μM Pefabloc, 2 μg/mL Aprotinin, 50 μM sodium orthovanadate, and 0.5 μg/mL Leupeptin. Homogenization was followed by a 30-minute tumbling period and a 10-minute centrifugation at 14,000g (both at 4°C). Supernatant was collected and protein concentration determined by Bradford assay (Bio-Rad Laboratories, Hercules, CA).

Western Blotting.

Protein lysate (75 μg) was subjected to SDS-PAGE, and transferred to nitrocellulose membrane (Schleicher & Schuell BioScience, Keene, NH). Equal loading of protein was determined by Ponceau S staining (Sigma-Aldrich, St. Louis, MO) of the transfer blot and by Western blot using a mouse anti-actin antibody (1:1000, ICN, Costa Mesa, CA). Blotting was performed as described,19 using a mouse anti-human α-SMA antibody (1:1000, Dako, Carpinteria, CA), and a goat anti-mouse horseradish peroxidase-conjugated secondary antibody (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA). Protein expression was detected by enhanced chemiluminescence (Amersham, Arlington Heights, IL), and expression levels quantified by densitometry using AlphaEase software (Alpha-Innotech Corporation, San Leandro, CA).

Gelatin Zymography.

MMP activity was detected using Novex 10% gelatin gels (Invitrogen, Carlsbad, CA). Briefly, 30 μg of protein lysate were electrophoresed in the gel, which was then incubated for 1 hour at room temperature in 2.5% Triton X-100, followed by a 30-minute incubation at room temperature in developing buffer (0.15 M NaCl, 10 mM CaCl2, 0.02% sodium azide, 50 mM Tris pH 7.5), and a 20-hour incubation at 37°C in the same buffer. Gels were stained for 2 hours in 2% Coomassie Blue, followed by destaining (30% methanol:10% acetic acid). Bands associated with pro-MMP-2 (72 kD) and active MMP-2 (66 kD) activity were quantified as described in Western Blotting. Expression levels were normalized to 10 μg of activated rat HSC lysate, used as a positive control.

Hydroxyproline Content Assay.

A colorimetric assay was performed as described.20 Briefly, lyopholized liver sections (≈ 3 mm3) were hydrolyzed for 20 hours in 6 M HCl at 100°C, redissolved in water, and centrifuged to remove any impurities. Samples were incubated for 10 minutes in 0.05 M chloromine-T (Fisher, Fair Lawn, NJ) at room temperature, followed by 15-minute incubation in Ehrlich's-perchloric acid solution (1 M p-dimethylaminobenzaldehyde in 62% n-propanol and 15.6% perchloric acid) at 65°C. Sample absorbences were assessed at 561 nm and resulting values compared to a hydroxyproline standard curve. Each sample was assayed in duplicate.

Histological Analysis and Immunohistochemistry.

Liver tissues were fixed in formalin at sacrifice, embedded in paraffin, and Masson Trichrome and Sirius Red staining was performed on sections. Additional sections were stained for α-SMA (1:1000) using the EnVision system (Dako) as described,21 or desmin (1:50; Dako) using the VECTASTAIN kit (Vector, Burlingame, CA). For desmin staining, antigen retrieval was performed as described by the manufacturer.

Image Analysis.

Sirius Red, α-SMA, and desmin staining was quantified using MetaMorph v.4.6 software (Universal Imaging Corporation; Downingtown, PA). For Sirius Red global analysis (40× magnification), staining was averaged from four randomly selected fields per slide. For zonal analysis (100× magnification), four randomly selected fields were averaged from each of the following regions: pericentral (including and surrounding the central vein), periportal (including and surrounding the portal triad), and midzonal (between the periportal and pericentral regions). Care was taken to prevent overlap of regions. For immunohistochemistry, staining was averaged from four (α-SMA) or ten (desmin) randomly selected fields per slide (100× magnification).

Statistical Analysis.

The Student's t test was used to determine statistical significance between groups. A P value of less than .05 was considered significant.

Results

Generation of a High-affinity Neutralizing Anti–TIMP-1 Antibody.

HuCAL-Fab1, a fully synthetic human combinatorial antibody library22, 23 was used for in vitro selection of antibodies against rat TIMP-1 (data not shown, B. B., unpublished observations, 2001). Of 36 different rat TIMP-1 specific Fab antibodies identified, 33 were functionally active in a rat TIMP-1/MMP-13 assay. Clone MOR-BAY-17, showing a monovalent KD of 13 nM for rat TIMP-1 based on BIAcore analysis and an IC50 of 11 nM in a rat TIMP-1/MMP-13 assay, was subjected to further affinity optimization by LCDR3 exchange. Affinity-matured clone MOR-BAY-17-1, a monovalent Fab fragment that displays a KD of 0.8 nM for rat TIMP-1 based on BIAcore analysis and an IC50 of 1.3 nM in a rat TIMP-1/MMP-13 assay (Fig. 1), was used in further studies. Upon conversion into IgG1, it has a bivalent KD of 0.04 nM and an IC50 of 0.9 nM.

Figure 1.

Anti-TIMP-1 Fab and IgG1 activities in rat TIMP-1/MMP-13 assay. Fab antibody fragments were diluted in triplicate to the indicated concentrations in assay buffer containing 0.05% bovine serum albumin. After addition of rat TIMP-1 (final concentration of 1.2 nmol/L), rat MMP-13 (final concentration of 1.2 nmol/L) and a fluorogenic substrate for MMP-1 (final concentration of 50 μmol/L), followed by incubation at 37°C, fluorescence at 320 nm (excited) / 430 nm (emitted) was measured. IC50 was calculated as outlined in the Materials and Methods section, using 100% MMP-13 activity (in absence of TIMP-1) and 15% MMP-13 activity (in absence of antibody) as reference values.

CCl4 Adminstration Induced Severe Fibrosis in Rats.

Rats treated with olive oil or CCl4 were weighed weekly for changes in body weight. All rats given CCl4 showed slower weight gain than control rats, with divergence beginning one week after administration of CCl4 (Week 3, Fig. 2A), corresponding with previous findings.24 There was no relationship between amount of CCl4 given and therapeutic antibody administered (data not shown). At sacrifice, the difference between olive oil- and CCl4-treated rats was 70-100 g. CCl4-treated rats showed a 20% increase in liver-to-body weight ratio compared to control rats (Fig. 2B), indicative of the hepatomegaly associated with liver injury.25 Histological examination of CCl4-treated livers showed massive rearrangement of the hepatic architecture, including septal fibrosis, extensive bridging, and inflammation (Fig. 3A, 3B and data not shown). To further verify fibrosis, we performed Western blotting for α-SMA, a biomarker of activated HSCs. All CCl4-treated rats expressed α-SMA, while control rats did not (Fig. 2C).

Figure 2.

Rats administered CCl4 show typical fibrosis-associated characteristics. (A) Average weekly weight chart. Rats treated with olive oil (Oil) or CCl4 were administered PBS (Oil / PBS and CCl4 / PBS) or 3 mg/kg (CCl4 / TIMP 3) or 10 mg/kg anti-TIMP-1 (CCl4 / TIMP 10). Weights were recorded weekly, beginning at administration of phenobarbital to the drinking water (Day 0) and ending the day of sacrifice (Sac). (B) Average liver-to-body weight ratio. Livers were weighed at time of sacrifice and expressed as a percentage of the body weight at the time of sacrifice. For each group in panels A and B, N = at least 8; error bars represent the SEM. *P <.005 vs. olive oil / PBS-treated rats. (C) CCl4-treated, but not control, rats express smooth muscle α-actin. Whole liver extracts were separated by 10% SDS-PAGE, and Western blotting for α-SMA performed. The blot was stripped and reprobed with an actin antibody to ensure equal loading of the gel. N = 5 for each group; dashes indicate individual samples. Protein extract from activated rat hepatic stellate cells (rHSC) used as a positive control.

Figure 3.

Histological analysis of collagen accumulation in livers. Liver sections from control and CCl4-treated rats, administered the indicated therapeutic, were taken at sacrifice and stained for collagen deposition by Masson Trichrome (A) and Sirius Red (B). A representative sample (40× magnification) of each group is depicted. (C) Low-power (40× magnification) analysis of liver sections. *P < .001 vs. olive oil / PBS-treated group; #P < .05 vs. CCl4 / PBS. (D) Zonal analysis of liver sections. Pericentral, periportal, and midzonal regions were analyzed (100× magnification) for collagen deposition. *P < .001 vs. olive oil / PBS-treated rats, $P < .05 vs. olive oil / PBS, #P < .01 vs. CCl4 / PBS, and &P < .05 vs. CCl4 / PBS. Data for both graphs represent the fold-increase in positive staining vs. olive oil / PBS-treated rats. Four randomly-selected fields were chosen per sample (for panels C and D, N = at least 8 per group). Error bars represent the SEM.

TIMP-1 Antibody Treatment Decreased Collagen Deposition.

We next analyzed the effect of anti–TIMP-1 therapy on liver histology in CCl4-induced fibrosis. Masson Trichrome staining showed increased collagen accumulation in CCl4-treated groups compared to control groups (Fig. 3A); accumulation was attenuated in rats treated with TIMP-1 antibody. Sirius Red staining, used to quantify this change, showed similar results to the Trichrome staining. Livers from control rats showed only normal staining around vessels, while massive fibrosis, including bridging, was seen in CCl4-treated livers (Fig. 3B). While the level of fibrosis was not totally eliminated following antibody treatment, the bridging seen in vehicle-treated livers was markedly reduced with the 10 mg/kg dose (compare CCl4 / PBS vs. CCl4/TIMP-10).

Quantification of Sirius Red staining showed a 5-fold increase in collagen accumulation in fibrotic rats compared to control groups (Fig. 3C), while administration of TIMP-1 antibody resulted in a 20% decrease in staining, regardless of the dosage. Zonal analysis showed a high degree of staining in the pericentral region of fibrotic rats (Fig. 3D), consistent with previous CCl4-induced studies of fibrosis.26, 27 Fibrotic rats administered PBS showed 12- and 10-fold increases in staining within the pericentral and midzonal regions, respectively, compared to control rats, while very little staining was found within the periportal region. Accumulation was decreased by approximately 50% in animals treated with TIMP-1 antibody, although no significant difference was detected between the two doses.

We examined total liver hydroxyproline levels, a biochemical marker of hepatic collagen content. Control rats expressed low levels of hydroxyproline (Fig. 4). Administration of TIMP-1 antibody to these rats had no effect. Fibrotic rats showed a 5-fold increase in hepatic hydroxyproline levels. Administration of the 3 mg/kg dose of the TIMP-1 antibody showed a modest, but not significant, decrease in hydroxyproline, while the 10 mg/kg dose resulted in a 40% decrease.

Figure 4.

Hydroxyproline content of livers. Liver segments from control or CCl4-treated rats administered the indicated treatments were analyzed for hydroxyproline. *P <.001 vs. olive oil / PBS-treated rats, #P <.001 vs. CCl4 / PBS, and &P <.005 vs. CCl4 / TIMP 3. Each sample was analyzed in duplicate (N = at least 8 per group). Error bars represent the SEM.

Anti–TIMP-1 Therapy Resulted in Decreased α-SMA Expression.

Western blotting for α-SMA (Fig. 2C) revealed expression in all CCl4-treated rats. However, we observed anti–TIMP-1 treatment decreased α-SMA expression. To further address this finding, immunostaining of tissue sections for α-SMA expression revealed intense staining patterns in CCl4-treated rats given PBS, similar to that seen by Sirius Red (compare Figs. 3B and 5A). Administration of TIMP-1 antibody resulted in a dose-dependent reduction in positive staining (Fig. 5A). In addition, fibrotic rats administered the 10 mg/kg dose of antibody showed a 50% decrease in α-SMA protein levels as compared to rats administered PBS (Fig. 5B). Immunostaining for desmin, which detects quiescent and activated HSCs, also showed a dose-dependent decrease in total HSCs (Fig 5A).

Figure 5.

Administration of the TIMP-1 antibody decreases α-SMA expression. (A) The TIMP-1 antibody decreases α-SMA staining in a dose-dependent manner. Liver sections were stained with an α-SMA or desmin antibody and quantitated for positive staining. A representative picture from each group (at 100× magnification) is shown. Data are presented as fold-increase in positive staining vs. olive oil / PBS-treated rats. For both graphs, *P < .005 vs. olive oil / PBS-treated rats, #P < .005 vs. CCl4 / PBS, and &P < .01 vs. CCl4 / TIMP 3. Four (α-SMA) or ten (desmin) randomly-selected fields were chosen per sample (for each group, N = at least 8). (B) TIMP-1 antibody decreases α-SMA protein levels. Whole liver lysates from fibrotic rats administered either PBS (CCl4 / PBS) or 10 mg/kg anti-TIMP-1 (CCl4 / TIMP 10) were separated by 10% SDS-PAGE, and analyzed by Western blotting for α-SMA. Blots were stripped and re-probed with an actin antibody to ensure equal loading of the gel. N = 10 for both groups; dashes represent individual samples. Bands were quantitated by densitometry and normalized to actin levels; *P < .05. The graph represents the average of three individual experiments. For all graphs, error bars represent the SEM.

Anti–TIMP-1 Therapy Resulted in Decreased MMP-2 Activity.

We next assessed the effect of anti-TIMP-1 on MMP-2 activity, using gelatin zymography (Fig. 6A). CCl4 treatment resulted in a 4.5-fold increase in pro-MMP-2 activity compared to controls (Fig. 6B), while administration of TIMP-1 antibody showed a dose-dependent reduction in this activity. Furthermore, we observed a dose-dependent decrease in activity of the active form of MMP-2 (Fig. 6C). Control rats expressed little active MMP-2 (Fig. 6A).

Figure 6.

Administration of TIMP-1 antibody decreases MMP-2 activity. Whole liver lysates were subjected to gelatin zymography (A), and pro- (B) and active (C) MMP-2 activities were measured by densitometry. A representative gel is shown in panel A. N = 5 for each group; dashes indicate individual samples; protein extract from activated rat hepatic stellate cells (rHSC) used as a positive control. Relative MMP-2 expression levels are expressed as a fold-change from control (Oil / PBS; panel B) or fibrotic (CCl4 / PBS; panel C) rats administered PBS. For panel B, *P < .01 vs. olive oil / PBS rats, #P < .05 vs. CCl4 / PBS rats, and &P < .001 vs. CCl4 / TIMP 3 rats. For panel C, *P < .001 vs. CCl4 / PBS rats, and #P < .05 vs. CCl4 / TIMP 3 rats. Both graphs represent the average of 3 individual experiments (N = 5 samples per group); error bars represent the SEM.

Regression of fibrosis was not due to a nonspecific antibody effect. To assess whether the decrease in fibrosis resulted from nonspecific antibody effects, we performed additional experiments on fibrotic rats treated with PBS or an anti-human IgG1 antibody. As seen in Fig. 7, there were no significant differences in weekly average body weight (A), liver-to-body weight ratio (B), α-SMA expression (C), or hydroxyproline content (D) between PBS- and IgG1-treated rats. In addition, no statistically-significant differences in collagen accumulation between the two groups were found by visual (E) and computer-assisted (F, G) examination of the liver histology by Sirius Red staining. Furthermore, administration of isotype control resulted in nonsignificant elevation of pro- and active MMP-2 activities compared to PBS (H).

Figure 7.

A control IgG1 antibody has no effect on fibrosis compared to vehicle treatment. (A) Average weekly body weight chart. PBS, CCl4-treated rats administered PBS; IgG, CCl4-treated rats administered an isotype control antibody as a therapeutic. (B) Average liver-to-body weight ratio taken at time of sacrifice. (C) Western blot for α-SMA expression. Experimental conditions are as described in Fig. 2; dashes indicate individual samples. (D) Hydroxyproline content of whole liver. Conditions are as described in Fig. 4. (E) Histological analysis of liver sections by Masson Trichrome and Sirius Red staining. Sirius Red staining was quantitated at 40× (F) and zonal (100×; G) magnification as described in Fig. 3. *P < .05 vs. olive oil / PBS-treated rats. (H) Quantification of zymography. Experimental conditions are described in Materials and Methods, except that pro- and active MMP-2 activities were compared directly between PBS- and IgG1-treated samples. For all figures, error bars represent the SEM, and NS indicates no statistically-significant difference between groups; for all experiments, N = 4.

Discussion

Hepatic fibrosis is the final common pathway for many chronic liver diseases, with hepatitis C virus infection and alcohol the two leading causes.28 The only effective approach to treating advanced liver fibrosis is transplantation.29 However, several potential anti-fibrotic therapies have been investigated, including interferon α and γ,30, 31 hepatocyte growth factor gene therapy,32 and inhibitors of transforming growth factor-β-1,33 Rho GTPase-associated kinase,34 and the renin-angiotensin system.35 Two approaches to removing the fibrotic scar involve increasing the levels of ECM-degrading MMPs or decreasing the levels of their natural inhibitors, the TIMPs. Recent studies using adenoviral introduction of human MMP-1 in a model of rat liver fibrosis36 transfection of antisense TIMP-1 plasmids into rat HSCs in vitro37 have shown up to a 66% reduction in fibrotic levels. “Knocking out” TIMP-1 function by genetic manipulation is possible in mouse models38, 39 but clearly not in humans. However, directly inhibiting TIMP-1 activity through antibody therapy may prove highly effective in treating human liver fibrosis.

We have shown that CCl4-induced hepatic fibrosis is markedly attenuated by administration of an antibody against TIMP-1, a regulator of MMP activity. Fibrotic rats administered the antibody showed decreased hydroxyproline content and Sirius Red staining, two quantitative measures of collagen deposition, compared to control rats. The decrease in hydroxyproline is less impressive than the change in Sirius Red staining. However the hydroxyproline assay detects total collagen protein in the liver (including degraded protein), while Sirius Red detects only intact collagen accumulated within the tissue.36 In addition, rats administered 10 mg/kg of the antibody showed an approximate 50% decrease in expression of α-SMA a well-established biomarker of activated HSCs. Expression of desmin showed a similar response, indicating the decrease in α-SMA was due to a net loss of activated HSCs, and not reversion from the activated to the quiescent phenotype. The decrease in desmin does not completely mirror that observed in α-SMA expression, since quiescent HSCs, which may be unaffected by the presence of the antibody, also express desmin. Furthermore, these effects are not due to a nonspecific immunoglobulin response, as a control IgG1 antibody had no beneficial effects.

Our demonstration that anti–TIMP-1 decreases the amount of activated HSCs is consistent with a recent study that reported persistent expression of TIMP-1 mRNA resulted in the continued presence of activated HSCs.16 During resolution of fibrosis, decreased TIMP-1 expression was associated with reduced numbers of α-SMA–positive HSCs, presumably through increased apoptosis. Both rat and human HSCs are resistant to various promoters of apoptosis when incubated with TIMP-1 in vitro.15, 16 TIMP-1 inhibits apoptosis in other cell types, such as mesangial cells,40 Burkitt's lymphoma cell lines, and human breast epithelial cells,7, 8 In our study, anti–TIMP-1 decreased the number of activated HSCs, perhaps by inducing their apoptosis.

TIMPs are associated with a variety of biological processes, including uterine reorganization during pregnancy,41 cell cycle–dependent association with the nucleus as a potential mediator of cell growth,42 erythroid cell potentiation,7 and anticancer activities, such as inhibiting tumor angiogenesis and metastasis.7, 8 In liver fibrosis, expression of TIMPs-1 and -2 increases9–11; activated HSCs are the source of this increase.43, 44 Activated HSCs are also a source of MMP-2, whose expression and activity are greatly increased during fibrogenesis and remain elevated during fibrotic progression,9–12, 43, 44 unlike expression of other MMPs, which increase immediately after insult, but diminish over time.11 Increased MMP-2 activity is believed to be associated with an increase in destruction of the normal liver architecture, allowing for increased HSC activation.5

Both MMP-2 mRNA steady-state levels and enzymatic activity dramatically increase during fibrogenesis and remain elevated during fibrotic progression. However, there have been conflicting reports of MMP-2 expression during the resolution of fibrosis. While some studies show that treating HSCs with potential antifibrotic agents results in increased MMP-2 mRNA and protein activity,45 others have observed gradual decreases in these same markers during recovery.12, 46 Still other studies show both increases and decreases, depending on the time-frame of the study.47 We show that administration of TIMP-1 antibody reduces, dose dependently, the activity of both the pro- and active forms of MMP-2. As TIMP-1 expression seems an important event in the survival of activated HSCs (perhaps through an autocrine signaling pathway), our findings indicate that neutralization of TIMP-1 may stimulate apoptosis of the HSCs in vivo, resulting in decreased α-SMA expression and MMP-2 activity. Our study was not designed to examine early changes in the mRNA levels or enzymatic activity of MMP-2 or other MMPs, such as MMP-13, the major collagenase in rats.5 Such an examination may help to better understand the dynamic changes of MMP activity during fibrotic progression.

We have assessed a new therapy to treat established liver fibrosis, using a stringent in vivo model of fibrosis. Many studies in experimentally induced fibrosis administer therapies prior to fibrotic induction,18, 48 while others remove the causative agent before treatment.36, 49 Our approach was to establish fibrosis prior to treatment, and to continue to challenge the liver with CCl4 during anti–TIMP-1 therapy, a model that better reflects the clinical situation where a patient would be specifically treated for ongoing hepatic fibrosis. Studies have shown that resolution and recovery from CCl4-induced experimental fibrosis occurs within 28 days of withdrawal from the agent.50 We saw a marked reduction in fibrotic indices (decreased collagen accumulation by histology and hydroxyproline content) just 6 days after the final CCl4 treatment and 3 days after the final antibody injection. In summary, treatment with TIMP-1 antibody was able to attenuate fibrosis, clearly demonstrating the potential of such an anti-fibrotic therapeutic approach.

Acknowledgements

We thank Drs. Mitsuo Yamauchi, William Reed, and Robert Schoonhoven, Gloria Chandler, Jennifer Dulyx, Melanie Choidas, Meike Eisenmann, Yvonne Stark, Sieglinde Müller, Christine Rottenberger, Mel Wong, John Murphy, Liming Yang, DaDong Chen, Tom Thompson, Helena Yee, Colleen Brown, Carol Mirenda, Sarah Patrick, Vivian Lee, Stephanie Yung, Sarah Hamren, Jianmin Chen, Lauren Thorner, Chun-Mei Ji, Toshi Takeuchi, and Jeffrey Greve for technical assistance, and Drs. Ramón Bataller and Richard Rippe for manuscript assistance.

Ancillary