antifibrotic and fibrolytic properties of celecoxib in liver damage induced by carbon tetrachloride...

BAS IC STUDIES

Anti¢broticand¢brolyticpropertiesof celecoxib in liverdamage inducedby carbon tetrachloride in the ratEnrique Chavez1, Jose Segovia2, Mineko Shibayama3, Victor Tsutsumi3, Paula Vergara2, Luis Castro-Sanchez4,Eduardo Perez Salazar4, Mario G. Moreno1 and Pablo Muriel1

1 Departamento de Farmacologıa, Cinvestav-IPN., Mexico , D.F. Mexico

2 Departamento de Fisiologıa Biofısica y Neurociencias, Cinvestav-IPN., Mexico, D.F. Mexico

3 Departamento de Infectomica y Patogenesis Molecular, Cinvestav-IPN., Mexico, D.F. Mexico

4 Departamento de Biologıa Celular, Cinvestav-IPN., Mexico, D.F. Mexico

Keywords

antifibrogenic – celecoxib – cyclooxygenase –

fibrolytic – fibrosis – metalloproteinase

Correspondence

Pablo Muriel, PhD, Departamento de

Farmacologıa, Cinvestav-I.P.N. Apdo.

Postal 14-740, Mexico 07000, D.F. Mexico

Tel: 15255 5747 3303

Fax: 15255 5747 3394

e-mail: [email protected]

Received 3 December 2009

Accepted 24 March 2010

DOI:10.1111/j.1478-3231.2010.02256.x

AbstractBackground: Transforming growth factor-b (TGF-b) plays a pivotal role inliver fibrosis, because it activates hepatic stellate cells, stimulating extracellularmatrix deposition. Cyclooxygenase-2 (COX-2) has been associated with TGF-b because its inhibition decreases TGF-b expression and collagen productionin some cultured cell types. Aim: The aim of this work was to evaluate theability of celecoxib (a selective COX-2 inhibitor) to prevent and to reverse theliver fibrosis induced by CCl4. Methods: We established experimental groupsof rats including vehicle and drug controls, damage induced by chronicCCl4 administration and CCl4 plus pharmacological treatment in bothprevention and reversion models. We determined: alanine aminotransferase,alkaline phosphatase, g-glutamyl transpeptidase, COX and metalloproteinase-2 and -9 activities, lipid peroxidation, glutathione levels, glycogen and collagencontent and TGF-b expression. Results: Celecoxib prevented and aided tothe recovery of livers with necrotic and cholestatic damage. Celecoxibexhibited anti-oxidant properties by restoring the redox equilibrium (lipidperoxidation and glutathione levels). Glycogen was decreased by CCl4, whilecelecoxib partially prevented and reversed this effect. Celecoxib inhibitedCOX-2 activity, decreased TGF-b expression, induced metalloproteinase-2activity and, consequently, prevented and reversed collagen accumulation.Conclusion: Our findings indicate that celecoxib exerts strong antifibrogenicand fibrolytic effects in the CCl4 model of cirrhosis.

Hepatic chronic diseases are characterized by the gradualdestruction of the parenchyma, accumulation of theextracellular matrix (ECM) (including collagens I, IIIand IV) because of the increased synthesis and theinability to break down these proteins, leading to distor-tion of the hepatic architecture, resulting in liver fibrosisand then in cirrhosis (1). Liver fibrosis is initiated bymechanisms leading to inflammation, which in turnactivates a wound-healing response as a result of theproduction of the fibrogenic cytokine transforminggrowth factor-b (TGF-b). TGF-b appears to be a keycytokine/growth factor mediator in human fibrogenesisbecause it activates hepatic stellate cells (HSC) to increasethe production and accumulation of ECM (2, 3).

Cyclooxygenase (COX) is the enzyme that catalysesthe biosynthesis of prostaglandins (PGs), which areimportant inflammatory mediators. There are two COXisoforms: COX-1 and COX-2 (constitutive or physiological,and induced or pathological respectively). COX-2 is mainlyinduced by stimuli associated with inflammation such as

bacterial lipopolysaccharide (LPS), tumour necrosis factor-a (TNF-a), interleukin 1b and 2 (IL-1b and IL-2) (4).

It has been observed that primary Kupffer cells onlyexpress COX-1 in vitro; however, LPS-treated Kupffercells express both COX-1 and COX-2 (5). Mohammedet al. (6) showed that COX-1 is expressed in both normaland cirrhotic livers. In contrast, COX-2 was not detectedin normal liver, but it was strongly upregulated in post-viral human cirrhosis. Similarly, in rat experimentalmodels of alcohol and CCl4, the same liver COX patternexpression takes place (7, 8). Interestingly, selective COX-2 inhibitors decrease the expression of TGF-b andcollagen I in human peritoneal mesothelial cells, andcollagen III and IV in an experimental model of renalinjury (9, 10). In a choline-deficient L-amino acid-defined diet (CDAA), an experimental model of liverfibrosis, TGF-b is upregulated and this study postulatedthat it induced COX-2 expression; the extent of liverfibrosis was closely associated with the level of COX-2and was prevented by the administration of JTE-522, a

Liver International (2010)c� 2010 John Wiley & Sons A/S 969

Liver International ISSN 1478-3223

mailto:[email protected]

selective COX-2 inhibitor (11). These experimental evi-dences show that the use of pharmacologic inhibitors ofCOX-2 could be useful in the treatment of liver diseases.

Hepatic fibrosis and cirrhosis induced by chronicadministration of CCl4 is the most widely used model tosimulate human cirrhosis in experimental animals be-cause it shares many characteristics of human cirrhosis;therefore, it is useful to study possible antifibrotic drugs.The aim of this work was to evaluate whether celecoxib,a selective COX-2 inhibitor, is able to prevent and toreverse the liver fibrosis induced by chronic administra-tion of CCl4. Furthermore, the antifibrotic effects ofcelecoxib can be explained by its ability to decreaseTGF-b and COX-2 activity, to increase metalloprotei-nase-2 (MMP-2) and its anti-oxidant properties.

Materials and methods

Chemicals

Commercial capsules of celecoxib (Celebrexs) obtainedfrom the pharmacy were used. Propylene glycol, sodiumthiosulphate, anthrone, thiobarbituric acid, chloramine-T,p-dimethylaminobenzaldehyde, g-glutamyl-p-nitroanilide,L-g–glutamyl-p-nitroaniline, p-nitrophenyl phosphate andbovine serum albumin were purchased from the SigmaChemical Company (St Louis, MO, USA). Carbon tetra-chloride, sodium hydroxide, glacial acetic acid, hydrochloricacid, sulphuric acid, ethanol, methanol, toluene and for-maldehyde were obtained from J. T. Baker (Xalostoc, MexicoCity, Mexico). All the reagents were of analytical quality.

Study design

Wistar male rats initially weighing 90–100 g and fed witha Purina chow rat diet ad libitum were used. Four or fiveanimals were housed per polycarbonate cage undercontrolled conditions (22� 2 1C, 50–60% relative hu-midity and 12 h light–dark cycles).

Cirrhosis was produced by an i.p. administration ofCCl4 (0.4 g/kg of body weight) dissolved in mineral oilthree times per week for 8 weeks. In order to determinethe capacity of celecoxib to prevent liver fibrosis, fourgroups were formed and treated for 8 weeks. Group 1(n = 8) consisted of control animals receiving the vehicleonly (mineral oil, 0.25 ml, i.p.); group 2 (n = 15) wasadministered with CCl4; group 3 (n = 15) was adminis-tered with CCl4 plus celecoxib (6 mg/kg, p.o., daily,dissolved in propylene glycol/saline solution 2:1, v/v);group 4 (n = 8) received celecoxib only. To investigate thefibrolytic effect of celecoxib, we formed groups 5, 6 and 7as follows: groups 5 and 6 (n = 15) were administeredwith CCl4 for 8 weeks, and then CCl4 was discontinuedand celecoxib or its vehicle, respectively, were adminis-tered for 4 weeks; group 7 (n = 8) received mineral oil(0.25 ml, i.p.) for 8 weeks, and then it was discontinuedand the celecoxib vehicle was administered for 4 weeks.All animals were sacrificed under light ether anaesthesia.Animals received humane care according to the institu-

tion’s guidelines and the Mexican Official Norm (NOM-062-ZOO-1999) regarding technical specifications forthe production, care and use of laboratory animals.

Serum enzyme activities

Blood was collected by cardiac puncture and sampleswere kept on ice until analysis. Samples were centrifugedat 1200 g for 15 min to obtain the serum. Then, determi-nation of liver damage by measuring the activities ofalanine aminotransferase (ALT) (12), alkaline phospha-tase (AP) (13) and g-glutamyl transpeptidase (g-GTP)(14) was performed.

Liver determinations

The extent of lipid peroxidation was determined in liverhomogenates by measuring the formation of malondial-dehyde (MDA) using the thiobarbituric acid method(15). Protein was determined according to Bradfordusing bovine serum albumin as a standard (16).

The content of reduced and oxidized glutathione wasdetermined in the liver. Liver samples were homogenizedon ice using a polytron homogenizer. The solution usedfor homogenization consisted of 3.75 ml of the phos-phate-EDTA buffer and 1 ml of 25% H3PO4, which wasused as a protein precipitant. The total homogenate wascentrifuged at 4 1C at 17700 g for 30 min to obtain thesupernatant for the assay of reduced and oxidizedglutathione (GSH and GSSG).

Determination of GSH was performed according toHissin and Hilf (17). To 0.01 ml of the supernatant,4.9 ml of the phosphate-EDTA buffer pH 8.0 was added.The final assay mixture (2.0 ml) contained 100 ml of thediluted tissue supernatant, 1.8 ml of phosphate-EDTAbuffer and 100 ml of the o-phthalaldehyde (1 mg/ml) solu-tion. After thorough mixing and incubation at roomtemperature for 15 min, the solution was transferred to aquartz cuvette. Fluorescence at 420 nm was determinedwith the excitation at 350 nm.

In the GSSG assay (18), 0.5 ml of the original super-natant was incubated at room temperature with 200 ml of0.04 M N-ethylmaleimide for 30 min to interact withGSH present in the tissue. To this mixture, 4.3 ml of0.1 N NaOH was added. A 100 ml portion of this mixturewas taken for measurement of GSSG, using the above-outlined assay for GSH, except that 0.1 N NaOH wasused as a diluent rather than phosphate-EDTA buffer.

Small pieces of liver (0.5 g) were separated for glycogendetermination using the anthrone reagent according toSeifter et al. (19).

In order to quantify the collagen content, fresh liversamples (100 mg) were placed in ampules, 2 ml of 6 NHCl was added and then the samples were sealed andhydrolysed at 100 1C for 48 h. Next, the samples wereevaporated at 50 1C for 24 h and resuspended in 3 ml ofsodium acetate–citric buffer, pH 6.0; 0.5 g of activatedcharcoal was added, the mixture was stirred vigorously

Liver International (2010)970 c� 2010 John Wiley & Sons A/S

Celecoxib prevents and reverses liver fibrosis Chavez et al.

and then it was centrifuged at 1200 g for 15 min. Onemillilitre of chloramine T was added to 1 ml of thesupernatant. The mixture was kept for 20 min at roomtemperature and the reaction was stopped by the addi-tion of 2 M sodium thiosulphate and 1 N sodium hydro-xide. The aqueous layer was transferred into test tubes.The oxidation product from hydroxyproline was con-verted to a pyrrole by boiling the samples. The pyrrole-containing samples were incubated with Ehrlich’s reagentfor 30 min and then, the absorbance was read at 560 nm.Recovery of known amounts of standards was carried outon similar liver samples for quantification (17).

For histological analysis, liver samples were taken fromall the animals and fixed with 10% formaldehyde inphosphate-buffered saline for 24 h. Tissue pieces werewashed with tap water, dehydrated in alcohol andembedded in paraffin. Five micrometres sections weremounted on glass slides previously covered with silane.Masson’s trichromic stain was performed in each slide.

To carry out Western blot assays, the TriPure reagent(Roche Diagnostics, Indianapolis, IN, USA) was used toisolate total protein from the liver tissue sample. Freshtissue was homogenized in 1 ml of TriPure reagent; then,0.2 ml of chloroform was added to the homogenates andthe lower phase was treated with isopropanol to precipitatethe total protein. Samples were centrifuged at 12 000 r.p.m.for 10 min at 4 1C; then, three washes were performed with0.3 M guanidine hydrochloride in 95% ethanol. A finalwash was performed with 100% ethanol, samples werecentrifuged as described previously and the pellet wasresuspended in 1% sodium dodecyl sulphate (SDS).Volumes equivalent to 50mg of proteins (determined bythe bicinchoninic acid method) were transferred onto a12% polyacrylamide gel; separated proteins were trans-ferred onto an Immuno-BlotTM PVDF membrane (Bio-Rad, Hercules, CA, USA). Next, blots were blocked with5% skim milk and 0.05% Tween-20 for 30 min at roomtemperature and independently incubated overnight at4 1C with a specific antibody against TGF-b (MAB 1032from Chemicon Int. Inc., Pemecula, CA, USA) diluted1:2000. The following day, membranes were washed andthen exposed to a secondary peroxidase-labelled antibody(Zymed, San Francisco, CA, USA) diluted 1:4000 in theblocking solution for 1 h at room temperature. Blots werewashed and protein was developed using the WesternlightningTM Plus-ECL Enhanced Chemiluminescence de-tection system (Perkin Elmer Inc., NEN Life SciencesProducts Elmer Las Inc., Boston, MA, USA). Blots werestripped and incubated with a monoclonal antibodydirected against b-actin (1:500) (20), which was used as acontrol to normalize cytokine protein expression levels.The procedure to strip membranes was as follows: first,blots were washed four times with phosphate-bufferedsaline pH 7.4 (0.015 M, 0.9% NaCl), and then immersedin stripping buffer (2-mercaptoethanol 100 mM, SDS 2%and Tris-HCl 62.5 mM, pH 6.7) for 30 min at 60 1C withgentle shaking; membranes were then washed five timeswith 0.05% Tween-20 in phosphate-buffered saline.

Images were digitalized using the BioDoc-It System(UVP, Upland, CA, USA) and then analysed densitome-trically using the Lab Works 4.0 Image Acquisition andAnalysis software (UVP).

Cyclooxygenase activity was estimated using a com-mercial kit (COX Fluorescent Activity Assay Kit, 700 200Cayman Chemical Company, Mexico City, Mexico) anda fluorescent reader following the instructions providedby the manufacturer.

Proteolytic activity was assayed using gelatin-substrategels as described previously (21). Briefly, the samevolume of non-heated samples was mixed with samplebuffer (2.5% SDS, 1% sucrose and 4mg/ml phenol red),without a reducing agent, and applied to 8% acrylamidegels copolymerized with gelatin at 1 mg/ml. After elec-trophoresis at 72 V for 2 h, the gels were rinsed twice in2.5% Triton X-100 to remove SDS and then incubated in50 mM Tris/HCl at pH 7.4 and 5 mM CaCl2 assay bufferat 37 1C for 48 h. The gels were then fixed and stainedwith 0.25% Coomassie Brilliant Blue G-250 in 10% aceticacid and 30% methanol. Proteolytic activity was detectedas clear bands against the background stain of undigestedsubstrate in the gel. The positive control for MMP-9 andMMP-2 secretion was obtained according to the mod-ified method of Etique et al. (22) using the non-tumori-genic breast epithelial cell line MCF10A. MCF10A cellswere cultured in Dulbecco’s modified Eagle’s medium(DMEM)/F12 (3:1) medium supplemented with 5%FBS, 10 mg/ml insulin, 0.5 mg/ml hydrocortisone, 20 ng/ml recombinant EGF and antibiotics, in a humidifiedatmosphere containing 5% CO2 and 95% air at 37 1C.Briefly, for experimental purposes, confluent cultures ofMCF10A cells were starved in DMEM without FBS,insulin, hydrocortisone and EGF for 12 h before treat-ment and washed twice with DMEM without serum,equilibrated in the same medium at 37 1C for at least30 min and then treated with 400 mg/dl of ethanol for25 h. The supernatant was collected and concentrated.

Statistical analysis

Data are expressed as mean values� SE. Comparisonswere carried out by analysis of variance, followed byTukey’s test, as appropriate, using SIGMA STAT (Systatsoftware Inc., San Jose, CA, USA) for Windows (2.0version). Differences were considered statistically signifi-cant when Po 0.05.

Results

Prevention of liver fibrosis induced by carbontetrachloride administration

The serum enzyme activities of ALT, g-GTP and AP areshown in Table 1. In the rat’s liver injured chronically,ALT activity increased significantly but celecoxib treat-ment prevented this effect partially. Treatment with CCl4for 8 weeks increased the enzyme activities of g-GTP andAP about 5 and 2.6 times respectively. Celecoxib


Chavez et al. Celecoxib prevents and reverses liver fibrosis

administration completely prevented this effect. Celecox-ib by itself did not modify any of these enzyme activities.

Glycogen is a polysaccharide that is the principalstorage form of glucose in animal and human cells;therefore, it is the main source of energy. The content ofhepatic glycogen was depleted because of chronic intox-ication with CCl4 (Table 1), while celecoxib partiallyprevented this effect; however, the statistical analysisindicates that this effect was not significant.

Malondialdehyde is one of the end products of lipidperoxidation and it is known that this process plays apivotal role in CCl4-induced liver injury; thus, thecontent of MDA is useful to assess lipid peroxidation.Treatment with CCl4 led to a significant increase of MDAlevels (Fig. 1a); celecoxib was capable of reducing theincrease of MDA levels.

Glutathione is a molecule that helps to maintain theredox equilibrium in mammals; therefore, measurementof glutathione can be used as an indicator of oxidativestress at the hydrophilic level. The GSH/GSSG ratiodecreased by treatment with CCl4; celecoxib did notprevent this effect (Fig. 1b), but interestingly, the contentof total glutathione increased when the rats were admi-nistered with CCl4 plus celecoxib (Fig. 1c).

Figure 2 shows, in the intoxicated group, a 3.5 timesincrease of collagen over the control; importantly, whenthe rats received CCl4 plus celecoxib, there was noincrease in the content of this protein. To corroboratethis result, hepatic fibrosis was also evaluated using ahistological method by visualizing the collagen fibres intrichromic-stained liver sections (Fig. 3). Chronic intox-ication with CCl4 induced a significant deposition ofcollagen around portal spaces, the formation of nodulesof hepatocytes surrounded by collagen bands and loss ofthe normal architecture with extensive necrotic areas(Fig. 3b). Corroborating the biochemical findings, thehistopathological analysis revealed that the simultaneousadministration of CCl4 plus celecoxib prevented theaccumulation of collagen and necrosis; celecoxib partiallypreserved the normal architecture of the parenchyma(Fig. 3c), while hepatic tissue of celecoxib-administeredrats showed a normal appearance (Fig. 3d).

Because celecoxib is a selective inhibitor of COX-2,we evaluated the COX-1 and -2 activities using a com-mercial kit. There was no change in the activity of the

a(a)1.2

0.6

0.8

1

b

(b)10

0

0.2

0.4

nmol

MD

A/m

g pr

otei

n

456789

aa

GS

H/G

SS

G

0123

(c)8 a

3

4

5

6

7a

0

1

2

µmol

glu

tath

ione

/g li

ver

Fig. 1. Oxidative stress in the liver: (a) Lipid peroxidationdetermined as malondialdehyde (MDA) content, (b) reduced/oxidized ratio (GSH/GSSG) and (c) total glutathione (GSH1GSSG)determined in livers from control rats, CCl4-treated rats (CCl4), CCl4plus celecoxib (CCl41celecoxib) and rats administered withcelecoxib alone (celecoxib). Each bar represents the mean value ofexperiments performed in duplicate assays� SE (n = 6). aSignificantlydifferent from the control, Po0.05. bSignificantly different fromCCl4, Po 0.05.

Table 1 Prevention treatment: alanine aminotransferase, g-glutamyl transpeptidase and alkaline phosphatase activities

Parameters (mmol/L min)Groups

Control CCl4 CCl41CLC CLC

ALT 9.87� 2.33 42.84�10.30� 20.46� 2.00w 7.85�0.42g-GTP 17.25� 2.44 60.04�12.50� 15.48� 4.87w 15.17�3.16AP 140.05� 29.93 269.38�25.48� 133.56� 21.19w 92.69�14.39

ALT, g-GTP and AP activities were determined in serum from control rats, CCl4-treated rats (CCl4), CCl4 plus celecoxib-treated rats (CCl41CLC), and rats

administered with celecoxib alone (CLC). Values represent the mean of experiments performed in duplicate assays� SE (n = 6).�Significantly different from control, Po 0.05.

wSignificantly different from CCl4 group, Po 0.05.

ALT, alanine aminotransferase; AP, alkaline phosphatase; CLC, celecoxib; g-GTP, g-glutamyl transpeptidase.



COX-1 isoform in any group, but we found a significantincrease in the COX-2 isoform activity in the ratsreceiving CCl4 with respect to the control group; asexpected, celecoxib prevented the increase of COX-2activity (Fig. 4).

Because TGF-b plays an important role in liver fibro-genesis, we decided to determine whether celecoxib had

any effect on the TGF-b protein expression by Westernblot analysis (Fig. 5). Chronic intoxication with CCl4produced a significant increase in the TGF-b levels butthis effect was totally prevented by simultaneous treat-ment with celecoxib.

Metalloproteinases are enzymes capable of degradingthe components of the ECM. Because MMP-2 and -9 are

5 a

4

3

2

mg

colla

gen

/g li

ver

b

0

1

Fig. 2. Collagen content determined in livers from control rats,CCl4-treated rats (CCl4), CCl4 plus celecoxib (CCl41celecoxib) andrats administered with celecoxib alone (celecoxib). Each barrepresents the mean value of experiments performed in duplicateassays� SE (n = 6). aSignificantly different from control, Po 0.05.bSignificantly different from CCl4, Po 0.05.

(a) (b)

(c) (d)

Fig. 3. Representative Mallory’s trichromic stain of liver sections obtained from rats treated with vehicle (panel A), CCl4 (panel B), CCl4 pluscelecoxib (panel C) or celecoxib alone (panel D). Collagen can be visualized as gray colour. The mean quantitative amount of collagen,determined by measuring liver hydroxyproline, is shown in Fig. 2.

14

10

12

COX-1

a

6

8 COX-2

4

0

2

mM

res

oruf

in/m

in

b

Fig. 4. Cyclooxygenase (COX)-1 and COX-2 activity determined inlivers from control rats, CCl4-treated rats (CCl4), CCl4 plus celecoxib(CCl41celecoxib) and rats administered with celecoxib alone(celecoxib). Each bar represents the mean value of experimentsperformed in duplicate assays� SE (n = 6). aSignificantly differentfrom control, Po 0.05. bSignificantly different from CCl4, Po0.05.



expressed in CCl4-induced liver fibrosis, we evaluatedtheir proteolytic activity using a gelatin zymographyassay. As shown in Figure 6, no group showed anyMMP-9 activity; in contrast, an increase in MMP-2activity was seen in the CCl4 group and it was furtherincreased in the group treated with CCl4 plus celecoxib.

Reversion of liver fibrosis induced by carbontetrachloride administration

The serum enzyme activities of ALT, g-GTP and AP areshown in Table 2. An approximately four-fold increase inALT enzyme activity could be observed after 8 weeks ofchronic CCl4 administration with respect to control rats.When rats received pharmacological treatment withcelecoxib, this increment was restored partially butsignificantly. Celecoxib exerted beneficial effect in rever-sing cholestatic damage, because it restored the g-GTPand AP activities to control levels.

The liver glycogen content decreased nine times byCCl4 administration. When CCl4 intoxication wasstopped, the rats did not restore the glycogen content tocontrol levels after 4 weeks, but those who receivedpharmacological treatment recovered partially but sig-nificantly (Table 2).

As in the prevention model, in the reversion one, asignificant anti-oxidant effect of celecoxib was found;MDA levels increased after CCl4 administration and

celecoxib reversed this effect completely; vehicle admin-istration showed no alteration in this parameter (Fig. 7a).

CCl4 administration for 8 weeks decreased the GSH/GSSG ratio (Fig. 7b); although there was no significanteffect on the content of total glutathione (Fig. 7c),celecoxib completely restored the GSH/GSSG ratio.

CCl4 administration induced a four-fold increase inthe liver collagen content and when the aetiological agentwas removed, it showed a partial but significant sponta-neous resolution of liver fibrosis; however, celecoxibreverted the content of collagen to normal levels (Fig. 8).

The MMPs proteolytic activity is shown in Figure 9.Herein, no MMP-9 activity can be seen in any of thegroups, but there is a significant induction of MMP-2activity because of the pharmacological treatment withcelecoxib as compared with the group receiving onlyvehicle.

Discussion

An overexpression of COX-2 has been demonstrated inchronic liver diseases in human and experimental liverfibrosis and cirrhosis (6, 8, 11). However, there arediscrepancies regarding the beneficial effects of drugsthat are capable of inhibiting its activity. Based on this,we evaluated the effects of celecoxib in the hepaticfibrosis induced by the chronic administration of CCl4.This study shows that celecoxib prevents and reversesliver fibrosis produced by 8 weeks of CCl4 intoxication.Furthermore, the antifibrotic effects of celecoxib can beexplained by its ability to decrease TGF-b and COX-2activity, to increase MMP-2 and its anti-oxidant proper-ties. As shown in the ‘Results’, CCl4 chronic administra-tion induces necrosis, cholestatic damage, depletion ofglycogen content and liver fibrosis. Importantly, thesimultaneous treatment with celecoxib prevents all ofthese, indicating several beneficial effects of this com-pound.

Celecoxib is used for its anti-inflammatory properties(13) and little is known about its anti-oxidant effects;

1.8

1.4

1.6 a

0.8

1

1.2

0.2

0.4

0.6

DO

TG

F-β

/DO

β-a

ctin

b0

b

TGF-β

β-actin

ControlCCl4

Celecoxib

+––

–+–

–++

––+

Fig. 5. Celecoxib blockade of TGF-b protein expression in samplesof liver tissue determined by Western blot analysis from control rats,CCl4-treated rats (CCl4), CCl4 plus celecoxib (CCl41celecoxib) andrats administered with celecoxib alone (celecoxib). b-actin was usedas an internal control. Signal intensities were determined bydensitometric analysis of treated blots, and values were calculated asthe ratio of TGF-b/b-actin. Each bar represents the mean value ofexperiments performed in duplicate assays� SE (n = 6). aSignificantlydifferent from control, Po 0.05. bSignificantly different from CCl4,Po 0.05.

MMP-9

MMP-2

Con

trol

CC

l 4

Cel

ecox

ib

CC

l 4 +

cel

ecox

ib

Pos

itive

con

trol

Fig. 6. Metalloproteinase (MMP)-9 and MMP-2 activity wasanalysed by zymography using gelatin-substrate gels. Liver samplesfrom control rats, CCl4-treated rats (CCl4), CCl4 plus celecoxib(CCl41celecoxib) and rats administered with celecoxib alone(celecoxib) were analysed. Positive control was obtained fromconfluent cultures of MCF10A cells treated with 400 mg/dl ofethanol for 25 h.



celecoxib has an inhibitory effect on bladder cancer, inpart because of its anti-oxidant actions (23), by prevent-ing the elevation of superoxide and peroxide productionas well as the reduction in glutathione depletion inoxidatively stressed human lens epithelial cells in culture(HLECs) (24). Herein, celecoxib showed an antioxidanteffect by preventing the elevation of the MDA contentproduced by CCl4 administration. Furthermore, celecox-ib induced de novo glutathione synthesis in the liver.

Liver fibrosis is the result of an imbalance in both thequantity and the composition of ECM produced byHSCs, which undergo activation and transdifferentiationprocesses, acquiring pro-inflammatory and fibrogenicproperties (1). In these progressive processes, TGF-b isan important mediator of ECM deposition. The activeform of TGF-b is a dimer of 25 kDa that, when it binds toits cell-surface receptor, causes phosphorylation andtranslocation into the nucleus of co-Smads, which reg-ulate the expression of target genes like collagen type Iand fibronectin, thus increasing ECM production (3).Our results indicate that celecoxib is able to prevent theexpression of TGF-b. This effect is in agreement withprevious reports using a selective inhibitor of COX-2 thatalso decreases TGF-b and collagen I expression in humanperitoneal mesothelial cells and in a rat model ofunilateral ureteral obstruction (9, 25).

Cyclooxygenase-2 is associated with chronic liver dis-eases like hepatitis C infection (26) and is upregulated innecroinflammatory injury in experimental models in therat (8, 11). These findings suggest that COX-2 can be anovel target in the treatment of liver diseases. In fact, inthis study, we showed that CCl4 selectively induced theenzymatic activity of COX-2. As expected, celecoxib wasable to inhibit COX-2 activity produced by CCl4 admin-istration. It has been demonstrated that PGE2 (a productof COX activity) enhances the level of TGF-b and itsreceptors, and induces a glycogenolytic effect (27, 28).These results can explain the downregulation of TGF-band the partial preservation of glycogen content seen inthis work. Moreover, COX activity generates reactiveoxygen species (8), which are related to oxidative stress;thus, these results help us understand, at least in part, theantioxidant properties of celecoxib reported in this study.

Metalloproteinases are a group of enzymes that help topreserve the homeostasis of ECM, but in the presence of ahepatotoxic agent, HSC is activated, and the tissueinhibitor of metalloproteinase 1 is upregulated, leadingto fibrogenesis (29). MMP-9 (gelatinase B) is producedin CCl4-induced liver injury after a single injection (30)and in chronic hepatitis C infection (31); however, itsactivity was not related to the degree of fibrosis (32). Inthis regard, we did not observe induction of MMP-9activity either by CCl4 administration for 8 weeks or withcelecoxib treatment. This result is in agreement withZhou et al. (33), who did found neither MMP-9 expres-sion nor activity in the same model of chronic liverdisease. This result can be explained by the fact thatMMP-9 seems to have a link with the initial phase ofHSC activation because it has been demonstrated thatMMP-9 is secreted during HSC transdifferentiation andnot by the fully activated cells (34). MMP-2 (gelatinaseA) can cleave collagen type I and IV and is expressed byHSC during their activation following liver injury. Inexperimental liver fibrogenesis induced by CCl4, MMP-2expression is increased and remains elevated during thisprocess (35). We found, by zymography, MMP-2 activityin the group treated with CCl4, and interestingly, it wasmore prominent in the group treated simultaneouslywith celecoxib, thus explaining the inhibition of ECMdeposition.

One of the most important effects of celecoxib was onliver fibrosis. We demonstrated biochemically and histo-logically that celecoxib was capable of completely preser-ving the normal content of collagen. On the one hand,celecoxib is a specific inhibitor of COX-2 and is able toprevent TGF-b expression and accumulation of ECM; onthe other, celecoxib induces MMP-2 activity, and thusECM degradation.

First, we demonstrated that celecoxib possesses anti-fibrotic properties because it is capable of preventing thedeposition of ECM in this experimental model of liverfibrosis. This result prompted us to investigate whethercelecoxib was capable of reversing liver fibrosis inducedby the chronic administration of CCl4.

This work demonstrates the ability of celecoxib toprevent the consequences of necrosis and cholestatic

Table 2 Reversion treatment: alanine aminotransferase, g-glutamyl transpeptidase and alkaline phosphatase

Parameters (mmol/L min)Groups

Control CCl4 CCl41vehicle CCl41CLC

ALT 14.13�0.99 59.25�4.88� 36.38�3.28�,w 26.80�2.89w,z

g-GTP 7.40�0.62 23.12�3.67� 10.41�1.91w 8.65�0.56w

AP 94.69�11.93 403.76�32.47� 242.22�46.62�,w 146.44�16.42w,z

ALT, g-GTP and AP activities were determined in serum from control rats, CCl4-treated rats for 8 weeks (CCl4), CCl4 for 8 weeks, and then vehicle for 4

weeks (CCl41vehicle), and rats administered with CCl4 for 8 weeks then celecoxib for 4 weeks (CCl41CLC). Values represent the mean of experiments

performed in duplicate assays� SE, (n = 6).�Means significantly different from control, Po 0.05.

wMeans significantly different from CCl4 group, Po 0.05.

zMeans significantly different from CCl41vehicle group, Po 0.05.

ALT, alanine aminotransferase; AP, alkaline phosphatase; CLC, celecoxib; g-GTP, g-glutamyl transpeptidase.



damage. With respect to membranal and cytoplasmaticoxidative stress, celecoxib completely restores MDAlevels, reverses the progressive decrement in the GSHconcentration and is able to increase the GSH/GSSGratio values higher than those of the control group.

As occurred in the prevention model, and in contrastwith MMP-2 activity, we did not find MMP-9 activity,probably because of its role just in the initial phase ofHSC activation as we discussed before. Okazaki et al. (36)showed that MMP-2 activity and gene expression areupregulated in the progressive phase of liver fibrosis in

the rat and that it may participate in the breakdown ofperihepatocellular fibrosis, resulting in recovery fromliver fibrosis. MMP-2 activity was found more promi-nently in the group treated with CCl4 plus celecoxib thanin the group treated with CCl4 only and CCl4 plusvehicle. thus demonstrating a significant induction of itsactivity to degrade ECM. In fact, in agreement with thisresult, spontaneous resolution was lower than resolutionbecause of the pharmacological treatment with celecoxib.

Finally, celecoxib also restores partial but significantlyglycogen levels probably because of the antifibrotic effectof the compound because restoration of normal bloodflow allows metabolic exchange of nutritive substances.

This study demonstrates the beneficial effects of thedrug used to prevent and to reverse liver fibrosis inducedwith CCl4 in the rat, although there are studies showing

7

5

6

3

4

1

2

mg

colla

gen

/g li

ver

0

0 2 4 6 8 10 12Weeks

Fig. 8. Collagen content determined in livers from control rats,CCl4-treated rats for 8 weeks (CCl4), CCl4 for 8 weeks and thenvehicle for 4 weeks (Vehicle), and rats administered with CCl4 for 8weeks and then celecoxib for 4 weeks (celecoxib). Each barrepresents the mean value of experiments performed in duplicateassays� SE (n = 6). aSignificantly different from control, Po 0.05.bSignificantly different from the CCl4 group, Po 0.05. cSignificantlydifferent from the CCl41vehicle group, Po 0.05.

MMP-2

MMP-9C

ontr

ol

CC

l 4 (

8 w

eeks

)

CC

l 4 (

8 w

eeks

) pl

usce

leco

xib

(4 w

eeks

)

CC

l 4 (

8 w

eeks

) pl

usve

hicl

e (4

wee

ks)

Pos

itive

con

trol

Fig. 9. Metalloproteinase (MMP)-9 and MMP-2 activity wasanalysed by zymography using gelatin-substrate gels. Liver samplesfrom control rats, CCl4-treated rats for 8 weeks (CCl4), CCl4 for8 weeks and then vehicle for 4 weeks (CCl41vehicle), and ratsadministered with CCl4 for 8 weeks and then celecoxib for 4 weeks(CCl41celecoxib) were analysed. Positive control was obtained fromconfluent cultures of MCF10A cells treated with 400 mg/dl ofethanol for 25 h.

1.2 (a)

0.6

0.8

1.0

(b)12

0 2 4 6 8 10 12Weeks

nmol

MD

A/m

g pr

otei

n

0

0.2

0.4

4

6

8

10

GS

H/G

SS

G

0 2 4 6 8 10 120

2

Weeks

(c)8

4

6

Weeks0 2 4 6 8 10 12

0

2

μmol

glu

tath

ione

/g li

ver

Fig. 7. Oxidative stress in the liver: (a) Lipid peroxidation determinedas malondialdehyde (MDA) content, (b) reduced/oxidized ratio (GSH/GSSG) and (c) total glutathione (GSH1GSSG) determined in liversfrom control rats, CCl4-treated rats for 8 weeks (CCl4), CCl4 for 8weeks and then vehicle for 4 weeks (vehicle), and rats administeredwith CCl4 for 8 weeks and then celecoxib for 4 weeks (celecoxib).Each bar represents the mean value of experiments performed induplicate assays� SE (n = 6). aSignificantly different from control,Po 0.05. bSignificantly different from the CCl4 group, Po 0.05.cSignificantly different from the CCl41vehicle group, Po0.05.



contrasting results: Yu et al. (37) and Hui et al. (38)found that celecoxib did not prevent liver fibrosis in bileduct-ligated (BDL) rats or in rats treated with CCl4. Bothstudies administered 15 mg/kg of celecoxib during 1 and2 weeks of BDL or CCl4 administration for 5 weeks. It iswell-documented that BDL causes liver fibrosis 4 weeksafter surgery (39) and that CCl4 treatment for at least 8weeks (40). Therefore, these differences may be explainedby the different exposition time to the aetiological agentand the drug doses used. In fact, Neiderberger et al. (41)showed that celecoxib has different effects at low andhigh doses. These results can be because of the mechan-isms of action showed for this drug at the dose used asthey found in their study. Interestingly, clinical studiesrevealed that an increase of the dosage did not improveits beneficial effects; the patients responded better tolower doses (42). In agreement with our results, someinvestigations demonstrated that the administration ofCOX inhibitors prevents liver damage induced by CCl4,thioacetamide (TAA) and CDAA in rats (11, 43, 44).Moreover, Paik et al. (45) showed that celecoxib inducesHSC apoptosis, inhibits type I collagen in HSC and hasantifibrogenic effects in BDL rats and TAA experimentalmodels.

In summary, the present study shows that chronicadministration of CCl4 does not stimulate COX-1 activ-ity but COX-2. Celecoxib is a drug that is able to preventand reverse liver damage induced by chronic administra-tion of CCl4 that produces a well-established hepaticfibrosis. The antifibrotic effect of celecoxib can be ex-plained by its anti-inflammatory and anti-oxidant prop-erties. Once the toxic agent is discontinued, celecoxibshowed a fibrolytic effect, stimulating MMP-2 activityand restoring the redox equilibrium.

Acknowledgements

The authors thank Mr. Benjamın Salinas Hernandez, Mr.Ramon Hernandez, Q. F. B. Silvia Galindo and M. V. Z.Rafael Leyva for their excellent technical assistance.Enrique Chavez was a fellow of Conacyt. This work wassupported in part by Conacyt grant 54756 (J. S.).

References

1. Svegliati-Baroni G, De Minicis S, Marzioni M. Hepaticfibrogenesis in response to chronic liver injury: novelinsights on the role of cell-to-cell interaction and transi-tion. Liver Int 2008; 28: 1052–64.

2. Muriel P. Cytokines in liver diseases. In: Sahu S, ed.Hepatotoxicity: from Genomics to in vitro and in vivoModels. West Sussex, UK: John Wiley & Sons Ltd, 2008;371–89.

3. Gressner AM, Weiskirchen R. Modern pathogenic conceptsof liver fibrosis suggest stellate cells and TGF b as majorplayers and therapeutics targets. J Cell Mol Med 2006; 10:76–99.

4. Vane JR, Bakhle YS, Botting RM. Cyclooxygenases 1 and 2.Annu Rev Pharmacol Toxicol 1998; 38: 97–120.

5. Dieter P, Scheibe R, Jakobsson P, et al. Functional couplingof cyclooxygenase 1 and 2 to discrete prostanoid synthasesin liver macrophages. Biochem Biophys Res Commun 2000;276: 488–92.

6. Mohammed NA, Abd El-Aleem SA, El-Hafiz HA, McMa-hon RF. Distribution of constitutive (COX-1) and induci-ble (COX-2) cyclooxygenase in postviral human livercirrhosis: a possible role for COX-2 in the pathogenesis ofliver cirrhosis. J Clin Pathol 2004; 57: 350–4.

7. Nanji AA, Miao L, Thomas P, et al. Enhanced cyclooxygen-ase-2 gene expression in alcoholic liver disease in the rat.Gastroenterology 1997; 112: 943–51.

8. Kim SH, Chu HJ, Kang DH, et al. NF-kappa B bindingactivity and cyclooxygenase-2 expression in persistentCCl(4)-treated rat liver injury. J Korean Med Sci 2002; 17:193–200.

9. Liu H, Peng Y, Liu F, et al. A selective cyclooxygenase-2inhibitor decreases transforming growth factor-b1 synth-esis and matrix production in human peritoneal mesothe-lial cells. Cell Biol Int 2007; 31: 508–15.

10. Wang JL, Cheng HF, Shappell S, Harris RC. A selectivecyclooxygenase-2 inhibitor decreases proteinuria and re-tards progressive renal injury in rats. Kidney Int 2000; 57:2334–42.

11. Yamamoto H, Kondo M, Nakamori S, et al. JTE-522, acyclooxygenase-2 inhibitor, is an effective chemopreventiveagent against rat experimental liver fibrosis. Gastroenterol-ogy 2003; 1252: 556–71.

12. Reitman S, Frankel S. A colorimetric method for determi-nation of serum oxaloacetic and glutamic pyruvic transa-minases. Am J Clin Pathol 1957; 28: 56–63.

13. Bergmeyer HU, Grabl M, Walter HE. Enzymes. In: Berg-meyer J, Grabl M, eds. Methods of Enzymatic Analysis.Weinheim: Verlag-Chemie, 1983; 269–70.

14. Glossman M, Neville DM. Glutamyl transferase in kidneybrush border membranes. FEBS Lett 1972; 19: 340–4.

15. Okawa H, Ohmishi N, Yagi K. Assay for lipid peroxides inanimal tissues by the thiobarbituric acid reaction. AnalBiochem 1979; 95: 351–8.

16. Bradford MM. A rapid and sensitive method for thequantitation of microgram quantities of protein utilizingthe principle of protein dye binding. Anal Biochem 1976;72: 248–54.

17. Prockop DJ, Underfriend S. A specific method for theanalysis of hydroxyproline in tissues and urine. AnalBiochem 1960; 1: 228–39.

18. Hissin PJ, Hilf R. A fluorometric method for determinationof oxidized and reduced glutathione in tissues. AnalBiochem 1976; 74: 214–26.

19. Seifter S, Dayton S, Novic B, Muntwyler E. The estimationof glycogen with the anthrone reagent. Arch Biochem 1950;25: 191–200.

20. Garcia-Tovar CG, Perez A, Luna J, et al. Biochemicaland histochemical analysis of 71 kDa dystrophin iso-form (Dp71f) in rat brain. Acta Histochem 2001; 103:209–24.



21. Heussen C, Dowdle EB. Electrophoretic analysis of plasmi-nogen activators in polyacrylamide gels containing sodiumdodecyl sulfate and copolymerized substrates. Anal Bio-chem 1980; 102: 196–202.

22. Etique N, Grillier-Vuissoz I, Flament S. Ethanol stimulatesthe secretion of matrix metalloproteinases 2 and 9 in MCF-7 human breast cancer cells. Oncol Rep 2006; 15: 603–8.

23. Parada B, Sereno J, Reis F, et al. Anti-inflammatory, anti-proliferative and antioxidant profiles of selective cycloox-ygenase-2 inhibition as chemoprevention for rat bladdercarcinogenesis. Cancer Biol Ther 2009; 8: 7–14.

24. Petersen A, Carlsson T, Karlsson JO, Zetterberg M. Intra-cellular effects of NSAIDs/ASA in oxidatively stressed hu-man lens epithelial cells in culture. Ophthalmic Res 2008;40: 77–85.

25. Miyajima A, Ito K, Asino T, et al. Does cyclooxygenase-2inhibitor prevent renal tissue damage in unilateral obstruc-tion? J Urol 2001; 166: 1124–9.

26. Nunez O, Fernandez-Martınez A, Majano PL, et al. In-creased intrahepatic cyclooxygenase 2, matrix metallopro-teinase 2, and matrix metalloproteinase 9 expression isassociated with progressive liver disease in chronic hepatitisC virus infection: role of viral core and NS5A proteins. Gut2004; 53: 1665–72.

27. Ramirez-Yanez GO, Hamlet S, Jonarta A, Seymour GJ,Symons AL. Prostaglandin E2 enhances transforminggrowth factor-beta 1 and TGF-beta receptors synthesis: anin vivo and in vitro study. Prostaglandins Leukot EssentFatty Acids 2006; 74: 183–92.

28. Puschel GP, Kirchner C, Schroder A, Jungermann K.Glycogenolytic and antiglycogenolytic prostaglandin E2actions in rat hepatocytes are mediated via different signal-ling pathways. Eur J Biochem 1993; 218: 1083–9.

29. Schuppan D, Ruehl M, Somasundaram R, Hahn EG.Matrix as a modulator of hepatic fibrogenesis. Semin LiverDis 2001; 21: 351–72.

30. Roderfeld M, Geier A, Dietrich CG, et al. Cytokine block-ade inhibits hepatic tissue inhibitor of metalloproteinase-1expression and up-regulates matrix metalloproteinase-9 intoxic liver injury. Liver Int 2006; 26: 579–86.

31. Lichtinghagen R, Bahr MJ, Wehmeier M, et al. Expressionand coordinated regulation of matrix metalloproteinases inchronic hepatitis C and hepatitis C virus-induced livercirrhosis. Clin Sci (Lond) 2003; 105: 373–82.

32. Reif S, Somech R, Brazovski E, et al. Matrix metalloprotei-nases 2 and 9 are markers of inflammation but not of thedegree of fibrosis in chronic hepatitis C. Digestion 2005; 71:124–30.

33. Zhou X, Hovell CJ, Pawley S, et al. Expression of matrixmetalloproteinase-2 and -14 persists during early resolu-

tion of experimental liver fibrosis and might contribute tofibrolysis. Liver Int 2006; 26: 380–1.

34. Han YP, Zhou L, Wang J, et al. Essential role of matrixmetalloproteinases in interleukin-1-induced myofibroblas-tic activation of hepatic stellate cell in collagen. J Biol Chem2004; 279: 4820–8.

35. Hemmann S, Graf J, Roderfeld M, Roeb E. Expression ofMMPs and TIMPs in liver fibrosis – a systematic reviewwith special emphasis on anti-fibrotic strategies. J Hepatol2007; 46: 955–75.

36. Okazaki I, Watanabe T, Niioka M, Sugioka Y, Inagaki Y.Reversibility of liver fibrosis: role of matrix metalloprotei-nases. In: Razzaque MS., ed. Fibrogenesis: Cellular andMolecular Basis. Kluwer Academic Plenum Publishers,2005; 143–59.

37. Yu J, Hui AY, Chu ES, et al. The anti-inflammatory effect ofcelecoxib does not prevent liver fibrosis in bile duct-ligatedrats. Liver Int 2009; 29: 25–36.

38. Hui AY, Leung WK, Chan HL, et al. Effect of celecoxib onexperimental liver fibrosis in rat. Liver Int 2006; 26: 125–36.

39. Kountouras J, Billing BH, Scheuer P. Prolonged bile ductobstruction: a new experimental model for cirrhosis in therat. Br J Exp Pathol 1984; 65: 305–11.

40. Muriel P, Moreno MG, Hernandez M del C, Chavez E,Alcantar LK. Resolution of liver fibrosis in chronic CCl4administration in the rat after discontinuation of treat-ment: effect of silymarin, silibinin, colchicine and tri-methylcolchicinic acid. Basic Clin Pharmacol Toxicol 2005;96: 375–80.

41. Niederberger E, Tegeder I, Vetter G, et al. Celecoxib loses itsanti-inflammatory efficacy at high doses through activationof NF-kB. FASEB J 2001; 15: 1622–4.

42. Simon LS, Weaver AL, Graham DY, et al. Anti-inflamma-tory and upper gastrointestinal effects of celecoxib inrheumatoid arthritis: a randomized controlled trial. JAMA1999; 282: 1921–8.

43. Planaguma A, Claria J, Miquel R, et al. The selectivecyclooxygenase-2 inhibitor SC-236 reduces liver fibrosis bymechanisms involving non-parenchymal cell apoptosis andPPARgamma activation. FASEB J 2005; 19: 1120–2.

44. Denda A, Tang Q, Endoh T, et al. Prevention by acetylsa-licylic acid of liver cirrhosis and carcinogenesis as well asgenerations of 8-hydroxydeoxyguanosine and thiobarbitu-ric acid-reactive substances caused by a choline-deficient, L-amino acid-defined diet in rats. Carcinogenesis 1994; 15:1279–83.

45. Paik YH, Kim JK, Lee JI, et al. Celecoxib induces hepaticstellate cell apoptosis through inhibition of Akt activationand suppresses hepatic fibrosis in rats. Gut 2009; 58:1517–27.



antifibrotic and fibrolytic properties of celecoxib in liver damage induced by carbon tetrachloride...

Documents