ARTHRITIS & RHEUMATISMVol. 56, No. 8, August 2007, pp 2585–2594DOI 10.1002/art.22749© 2007, American College of Rheumatology

Pharmacologic Induction of Heme Oxygenase 1 ReducesAcute Inflammatory Arthritis in Mice

Mourad Benallaoua,1 Mathias Francois,1 Frederic Batteux,2 Natacha Thelier,1 John Y.-J. Shyy,3

Catherine Fitting,4 Lydia Tsagris,1 Jorge Boczkowski,5 Jean-Francois Savouret,1

Marie-Therese Corvol,1 Serge Poiraudeau,6 and Francois Rannou6

Objective. To determine the consequences ofpharmacologic up-regulation of heme oxygenase 1 (HO-1), and inhibition of HO-1 by injection of an anti–HO-1small interfering RNA (siRNA), in vivo in the acutephase of a mouse model of nonautoimmune arthritis.

Methods. In the K/BxN mouse serum transfermodel, which mimics human inflammatory arthritiswithout lymphocyte influence, HO-1 was up-regulatedby intraperitoneal injection of cobalt protoporphyrin IX(CoPP), a potent pharmacologic inducer, and was in-hibited using a specific siRNA. The clinical progress ofarthritis was monitored by measurement of paw thick-ness. Interleukin-1� (IL-1�), IL-6, tumor necrosis fac-tor � (TNF�), serum antioxidant, and nitric oxide (NO)levels, prostaglandin E2 (PGE2) production, and matrixmetalloproteinase 9 (MMP-9) activity were measured inserum. At the end of the experiments, joints wereexamined for immunohistopathologic changes.

Results. Intraperitoneal injection of CoPP allevi-ated disease symptoms, such as joint swelling, cartilage

degradation, and proliferation of inflammatory tissue injoints, in the acute phase of inflammatory arthritis. TheCoPP-induced expression of HO-1 in the joints and liverwas associated with marked decreases in IL-1�, IL-6,and TNF� levels, PGE2 secretion, and MMP-9 activityin serum, and with a marked increase in systemicantioxidant activity. In contrast, NO production inserum and inducible NO synthase expression in chon-drocytes were not affected by HO-1 induction. Specificinhibition of HO-1 by in vivo delivery of anti–HO-1siRNA repressed the protective effects.

Conclusion. Our data provide the first evidencethat pharmacologically induced up-regulation of HO-1triggers a robust protective antiinflammatory responsein a model of nonautoimmune arthritis in mice. Thissuggests that exogenously induced HO-1 may havepotential as therapy in the acute phase of inflammatoryarthritis in humans.

The recurrence and the intensity of the acutephase of joint inflammation are important factors in thedestruction of adjacent cartilage and bone in inflamma-tory forms of human arthritis, such as acute gout andpseudogout, infectious arthritis, seronegative spondylar-thritides, rheumatoid arthritis, and juvenile arthritis.Cumulative evidence indicates that systemic proinflam-matory cytokines, such as interleukin-1� (IL-1�),interleukin-6 (IL-6), and tumor necrosis factor �(TNF�), play a central role in the pathogenesis of thisinnate acute phase of joint inflammation (1–3), andsynergistic inhibition of all of these cytokines would bean interesting approach to therapy in this acute inflam-matory phase. Glucocorticoids inhibit the productionand/or effects of all of these cytokines, but their variousadverse effects, mainly in children and in patients withinfectious arthritis, have prompted the search for newantagonistic molecules. Among them, IL-10 has been

Supported in part by Universite Paris Descartes (soutien aequipe en emergence), INSERM, the Fondation de l’Avenir, and theAssociation de Recherche sur la Polyarthrite.

1Mourad Benallaoua, PhD, Mathias Francois, PhD, NatachaThelier, MD, Lydia Tsagris, Jean-Francois Savouret, PhD, Marie-Therese Corvol, MD, PhD: INSERM U 747, and Universite ParisDescartes, Paris, France; 2Frederic Batteux, PhD: Universite ParisDescartes, CNRS (UPRES 1833), and Assistance Publique Hopitauxde Paris, Hôpital Cochin, Paris, France; 3John Y.-J. Shyy, PhD:University of California, Riverside; 4Catherine Fitting, PhD: InstitutPasteur, Paris, France; 5Jorge Boczkowski, MD, PhD: INSERM U 700,and Universite Paris 7 Denis Diderot, Paris, France; 6SergePoiraudeau, PhD, Francois Rannou, MD, PhD: INSERM U 747,Universite Paris Descartes, and Assistance Publique Hopitaux deParis, Hopital Cochin, Paris, France.

Address correspondence and reprint requests to FrancoisRannou, MD, PhD, INSERM UMR-S-747, UFR Biomedicale, Uni-versite Paris Descartes, 45 Rue des Saints Peres, 75006 Paris, France.E-mail: francois.rannou@cch.aphp.fr.

Submitted for publication July 15, 2006; accepted in revisedform April 17, 2007.

2585

shown to inhibit the production of inflammation medi-ators in arthritis (4–7). Heme oxygenase 1 (HO-1),known for its cytoprotective effects against oxidativedamage (ref. 8; for review, see ref. 9), has been sug-gested to exert antiinflammatory actions per se and alsoto mediate the protective effects of IL-10 against inflam-mation (for review, see refs. 10 and 11).

HO-1 catalyzes the rate-limiting step of hemedegradation by oxidative cleavage of the �-meso carbonbridge to produce equimolar quantities of biliverdin IX,free iron, and carbon monoxide. The enzyme is inducedby heme, its synthetic analog cobalt protoporphyrin IX(CoPP), and various cell stress conditions, such ashyperoxia, hypoxia, heat shock, and increased levels ofendotoxin, hydrogen peroxide, oxidized lipoproteins,cytokines, ultraviolet light, and heavy metals. HO-1expression has also been reported to be induced in vitroand in vivo by antiinflammatory molecules, such assimvastatin (12) or resveratrol (13), in human and ratsmooth muscle cells, by IL-10 in lipopolysaccharide(LPS)–activated macrophages (14), by 15-deoxy-�12,14–prostaglandin J2 (15-deoxy-�12,14-PGJ2) in LPS-activated macrophages (15), and by antiarthritic goldcompounds in peritoneal macrophages and U937 andJurkat cell lines (16,17).

HO-1 overexpression or induction has been associ-ated with antiinflammatory events. Transgenic mice thatoverexpress HO-1 in the lungs are protected against pul-monary inflammation (18). Intratracheal inoculation ofmice with an adenovirus encoding HO-1 increases the Th2response in macrophages (19). Corneal inflammation isless severe when HO-1 is concomitantly induced (20).IL-10 or 15-deoxy-�12,14-PGJ2, known inducers of HO-1,alleviate adjuvant-induced arthritis (21) and LPS-inducedseptic shock (14) in mice. A human patient who wasdeficient in the HO-1 gene exhibited a severe phenotype ofgeneral inflammation leading to death (22), and a similarobservation was reported in HO-1–null mice (23,24), inwhich inactivation of the HO-1 gene was associated with aTh1-weighted shift in cytokine response.

HO-1 is expressed in synovial tissue and peri-pheral blood monocytes from patients with rheumatoidarthritis (25,26). An in vitro study showed that pharma-cologic up-regulation of HO-1 led to reduced produc-tion of inflammation markers in LPS-treated synovialcell lines (25). In contrast, suppression of endogenousHO-1 by anti–HO-1 small interfering RNA (siRNA)potentiated the proinflammatory effects of cytokines inisolated synovial cells. HO-1 is also expressed in normaland osteoarthritic human chondrocytes (27).

Taken together, the above results support the no-

tion that HO-1 plays a pivotal role in modulating inflam-mation. However, the function of HO-1 is still debated.Devesa et al recently showed, in 2 different models ofchronic autoimmune arthritis, that overexpression of HO-1was ineffective in impeding the progression of the chronicinflammatory disease (28–30). The role of HO-1 in vivo inthe acute phase of forms of arthritis that do not involveimmunologic influence, however, has not been studied indepth. In addition, no previous studies, regardless of theanimal model used, have included specific inhibition ofHO-1 activity. To investigate whether HO-1 might havetherapeutic potential in human inflammatory arthritis, weused a mouse model of nonautoimmune arthritis to deter-mine, in vivo, the consequences of pharmacologic up-regulation of HO-1 and of its inhibition by injection of ananti–HO-1 siRNA.

MATERIALS AND METHODS

Animals. K/BxN mice were generated by crossing KRN-TCR–transgenic mice (B10.BR genetic background) with NODmice (31). Expression of the �V�6-TCR transgene in (KRN �NOD)F1 offspring was identified by flow cytometry; these micewere found to correspond to arthritic K/BxN mice (32). ControlK/BxN mice were arthritis-free F1 offspring that did not expressthe transgene. Two pools of sera were prepared, from arthriticand control K/BxN mice. BALB/c mice were injected intraperi-toneally with arthritic or control K/BxN serum. Mice were keptunder specific pathogen–free conditions in a clean room at theanimal facility of the UFR Biomedicale des Saints Peres (Uni-versite Paris Descartes). Experimental protocols were approvedby the Institut National de la Sante et de la Recherche MedicaleCommittee for Animal Studies.

Induction and evaluation of arthritis. Arthritis wasinduced in BALB/c mice by intraperitoneal injection of 100 �lserum from arthritic K/BxN mice on day 0 and day 1. Controlmice were injected with 100 �l serum from control K/BxNmice. HO-1 was induced by intraperitoneal injection of CoPP(25 mg/kg body weight, in DMSO) in addition to serumtransfer. Vehicle control animals received the same amount ofDMSO alone. Paw thickness was measured daily, on days 0–7,using a dial gauge, and was assessed by investigators who wereblinded with regard to treatment group.

Determination of prostaglandin E2 (PGE2), IL-1�,IL-6, and TNF� concentrations in serum. The PGE2 concen-tration in mouse serum was determined with the use of anenzyme immunoassay kit (EIA Kit-monoclonal; CaymanChemical, Ann Arbor, MI), and IL-1�, IL-6, and TNF�concentrations with the use of enzyme-linked immunosorbentassay (ELISA) kits (Biotrak ELISA System; Amersham Bio-sciences, Orsay, France).

Determination of serum antioxidant status and nitritelevels. The total antioxidant status of serum was determined byevaluating inhibition of the oxidation of ABTS substrate, usinga kit from Calbiochem (Meudon, France). Briefly, in thepresence of strong antioxidant activity in serum, a low level ofoxidated ABTS is detected. Nitrite levels were determined by

2586 BENALLAOUA ET AL

indirect measurement of nitric oxide (NO) production by theGriess reaction, specifically adapted for serum.

Sodium dodecyl sulfate–polyacrylamide gel electrophor-esis (SDS-PAGE) zymography. Mouse serum (2 �l/sample) waselectrophoresed at 4°C in 8% SDS–polyacrylamide gels con-taining 0.1% gelatin, under nonreducing conditions. Proteinsin the gel were denatured by incubation with 2.5% TritonX-100 for 1 hour at 20°C. Gelatinolytic activity was monitoredas previously described (33).

Western blot analysis for HO-1. Mice were killed, andtissue extracts from the liver, spleen, lung, and knee joints wereisolated with the use of a buffer containing 10 mM HEPES (pH7.9), 10 mM KCl, 1.5 mM MgCl2, 0.5% Nonidet P40, 1 �g/mlleupeptin, 10 �g/ml aprotinin, and 1 mM phenylmethylsulfonylfluoride. Samples were centrifuged at 14,000g for 20 minutes at4°C. The supernatants were removed and centrifuged 2 moretimes at 14,000g for 20 minutes at 4°C. Supernatants wereseparated by 10% SDS-PAGE and then subjected to immuno-blotting. Equal transfer of proteins was confirmed by stainingthe nitrocellulose with ponceau red (0.2% [weight/volume]H2O diluted 99:1 in 10% trifluoroacetic acid). After blocking,blots were incubated with rabbit anti–HO-1 antibody (Stress-gen, Victoria, British Columbia, Canada) and then with horse-radish peroxidase (HRP)–conjugated secondary antibodies.The protein bands in the blots were detected using an en-hanced chemiluminescence kit (Pierce, Interchim, France).

HO-1 activity assay. HO-1 activity in joint and liverprotein extracts was evaluated spectrophotometrically, using he-min (as substrate) and excess biliverdin reductase from mouseliver cytosol (34). To prepare mouse liver cytosol, samples of liverwere first rinsed in ice-cold 1.15% KCl–20 mM Tris HCl buffer(pH 7.4). The homogenate was then centrifuged at 5,000g for 20minutes at 4°C and then at 105,000g for 1 hour at 4°C. Thesupernatant, which contained microsomal biliverdin reductase,was incubated for 30 minutes in the dark at 37°C, in a reactionmixture containing hemin, NADPH, G6P, G6PDH, and joint orliver protein extracts. Bilirubin production was measured spectro-photometrically and expressed as picomoles of bilirubin permilligram of protein per hour.

Inhibition of HO-1 expression in vivo by siRNA. Smallinterfering RNA specifically targeted against HO-1 and non-specific siRNA were delivered using high pressure (35). Fiftymicrograms of siRNA diluted in a final volume of 1 ml wasdelivered by rapid (10-second) injection into the tail vein onday 0, 1 hour before intraperitoneal injection of CoPP. Anti–HO-1 siRNA nucleotide sequences (36) were as follows: sense5�-AAGCCACACAGCACUAUGUAAdTdT-3�, antisense 5�-UUACAUAGUGCUGUGUGGCUUdTdT-3�. NonsilencingsiRNA was provided by Qiagen (Courtaboeuf, France). Se-quences were as follows: sense 5�-UUCUCCGAACGU-GUCACGUdTdT-3�, and antisense 5�-ACGUGACACGUU-CGGAGAAdTdT-3�.

Figure 1. Induction of heme oxygenase 1 (HO-1) by cobalt protoporphyrin IX (CoPP) and its suppression bysmall interfering RNA (siRNA) in the joints, lung, and liver of BALB/c mice. A, BALB/c mice were injectedintraperitoneally with CoPP (25 mg/kg) or DMSO vehicle. After 24 hours, mice were killed and HO-1 expressionwas assessed by Western blotting. B, HO-1 activity was assayed spectrophotometrically in 3 mice treated asdescribed in A. C, Mice were injected intravenously in the caudal vein with nonspecific (ns) or anti–HO-1 siRNA1 hour before CoPP injection. After 24 hours, mice were killed and HO-1 expression was assessed by Westernblotting. D, HO-1 activity was assayed spectrophotometrically in 3 mice treated as described in C. Equal loadingof protein for Western blotting was confirmed by ponceau red staining. Results shown in A and C are from 1experiment representative of 3 experiments; results shown in B and D are the mean and SEM. �� � P � 0.01.

HO-1 AND INFLAMMATORY ARTHRITIS 2587

Histochemistry and immunohistochemistry. After themice were killed, paws were dissected and fixed in 10%buffered formalin. Fixed tissues were decalcified, dehydrated,and embedded in paraffin. Sagittal sections (5 �m) werestained with hematoxylin and eosin. Some sections were incu-bated with rabbit inducible NO synthase (iNOS) antibodies(1:200 dilution; Transduction Laboratories, Lexington, KY) orKi-67 antibodies (1:1,000 dilution; Novocastra, Newcastle,UK) and then with HRP-conjugated secondary antibodies.Ki-67 immunostaining is positive during the cellular prolifera-tion process. Sections were counterstained with hematoxylin.For immunohistochemical controls, sections were incubatedwith phosphate buffered saline without the primary antibodies.

Determination of cartilage degradation by evaluationof the production of C-propeptide of type II collagen (PIICP).Cartilage degradation in mouse serum was assessed using acommercial ELISA for the PIICP epitope (Ibex, Montreal,Quebec, Canada). This assay measures the neoepitope createdby cleavage of type II collagen following the release of newlysynthesized procollagen into the matrix. Increased PIICPlevels indicate increased cartilage degradation.

Statistical analysis. Each experiment was performedat least 3 times. The results are expressed as the mean � SEM.In experiments comparing results obtained at different timepoints, statistical analysis was performed by analysis of vari-ance and Tukey’s post hoc comparison. In all other experi-ments, the Wilcoxon signed rank test was used. P values lessthan 0.05 were considered significant.

RESULTS

Response of the HO-1 gene to CoPP inductionand siRNA inhibition in BALB/c mice. BALB/c micewere injected intraperitoneally with CoPP or vehicle(DMSO), HO-1 levels in the joints and liver wereanalyzed by Western blotting 24 hours after injection,and the corresponding HO-1 enzymatic activity wasassessed. HO-1 was detected in joint and liver extractsfrom treated mice (Figure 1A). The mean � SEMenzyme activity of HO-1 in joints and liver extracts from3 mice treated with CoPP was 349 � 59 and 354 � 75pmoles/mg/hour, respectively (Figure 1B). Vehicle alonehad no effect on HO-1 expression or activity.

In another set of experiments, anti–HO-1 siRNAor nonspecific siRNA was injected intravenously into thecaudal vein of mice 1 hour before CoPP injection.Compared with nonspecific siRNA, the anti–HO-1siRNA elicited a decrease in HO-1 levels in the joints,lungs, and liver, but not the spleen (Figure 1C). Simi-larly, the anti–HO-1 siRNA elicited a reduction in HO-1activity levels in joint, lung, and liver extracts (102 � 43pmoles/mg/hour, 210 � 60 pmoles/mg/hour, and 166 �48 pmoles/mg/hour, respectively), but not in the spleen(Figure 1D). In further experiments, CoPP and anti–HO-1 siRNA were used to modulate HO-1 expression inmice with K/BxN serum–induced arthritis.

Suppression of clinical manifestations of jointinflammation in K/BxN serum–induced arthritis byHO-1. BALB/c mice were injected with serum fromtransgenic K/BxN mice exhibiting arthritis symptoms

Figure 2. Inhibitory effect of heme oxygenase 1 (HO-1) on paw thick-ness in BALB/c mice treated with serum from transgenic K/BxN miceexhibiting arthritis symptoms (K/BxN serum). Hind paw thickness wasmeasured at the ankle with a digital gauge, and the mean of both paws wasrecorded. Paw thickness was also assessed visually by an investigator whowas blinded with regard to treatment group. A, Effect of cobalt protopor-phyrin IX (CoPP). Mice (n � 8 per group) were inoculated with K/BxNserum plus vehicle, K/BxN serum plus 25 mg/kg CoPP, or control serumfrom K/BxN mice without symptoms (CTL serum) plus vehicle. Sham-treated mice and mice treated with control serum plus CoPP exhibited nosignificant alteration in paw thickness (data not shown). Values are themean and SEM. � � P � 0.05; �� � P � 0.01, versus mice treated withK/BxN serum plus CoPP. Photographs of a paw from a representativeanimal in each group are also shown. B, Effect of small interfering RNA(siRNA) treatment. Mice (n � 8 per group) were injected intravenouslywith anti–HO-1 siRNA or nonspecific (ns) siRNA 1 hour before injectionof K/BxN serum and CoPP. Values are the mean and SEM. �� � P � 0.01versus mice treated with nonspecific siRNA.

2588 BENALLAOUA ET AL

(K/BxN serum) or with control serum from asymptom-atic K/BxN mice, along with CoPP or the correspondingDMSO vehicle. Local inflammation of the joints wasquantified by measuring paw thickness daily for 7 days(Figure 2A). In mice injected with K/BxN serum plusvehicle, paw thickness was significantly increased on day3 (mean � SEM 4.6 � 0.2 mm, versus 3.9 � 0.1 mm inmice treated with control serum plus vehicle; P � 0.05).Joint inflammation progressed continuously from day 4to day 7 in K/BxN plus vehicle–injected mice (pawthickness 4.7 � 0.2 mm on day 7; P � 0.01 versus micetreated with control serum plus vehicle), but not in micetreated with K/BxN serum plus CoPP (4.1 � 0.1 mm onday 7) or in control mice (3.9 � 0.1 mm on day 7).

The groups administered control serum plus ve-hicle or K/BxN serum plus CoPP exhibited no symptomsthroughout the 7-day period, and no significant differ-ences in paw thickness between these 2 groups wereobserved on day 7. Sham-treated mice and mice treated

with control serum and CoPP showed no significantdifference in paw thickness compared with those treatedwith control serum and vehicle (data not shown).

Since CoPP is not a specific inducer of HO-1, wespecifically inhibited HO-1 in vivo in CoPP- and K/BxN-treated mice by siRNA treatment (Figure 2B). Micewere injected intravenously with an anti–HO-1 siRNAor a nonspecific siRNA sequence as a control, 1 hourbefore injection of K/BxN serum and CoPP. Paw thick-ness was assessed as described above. As shown inFigure 2B, the paws of mice treated with anti–HO-1siRNA were thicker than those of mice receiving non-specific siRNA (P � 0.01). Between day 0 and day 7, pawthickness was not affected in the mice treated with thenonspecific siRNA, which suggests that siRNA treat-ment does not interfere with paw thickness. Theseresults strongly indicate that the protective effect ofCoPP on acute joint inflammation is specifically medi-ated by HO-1.

Figure 3. Inhibitory effect of HO-1 on systemic markers of inflammation in K/BxN serum–treated BALB/c mice,as determined based on the effects observed with CoPP treatment (A and B) and siRNA treatment (C and D).Serum levels of prostaglandin E2 (PGE2) assessed by enzyme immunoassay are shown in A and C. Values are themean and SEM (n � 8 mice per group). � � P � 0.05; �� � P � 0.01. Matrix metalloproteinase (MMP) activitydetected by sodium dodecyl sulfate zymography is shown in B and D. Results shown are from 1 experimentrepresentative of 3 experiments, each performed with 8 mice per group. Sham-treated mice and mice treated withcontrol serum plus CoPP exhibited no significant alteration in serum PGE2 levels or MMP activity (data notshown). See Figure 2 for other definitions.

HO-1 AND INFLAMMATORY ARTHRITIS 2589

HO-1–induced attenuation of the systemic in-flammation elicited by K/BxN serum. In BALB/c miceinjected with K/BxN serum or control serum in conjunc-tion with CoPP or vehicle, serum PGE2 content andgelatinolytic activity were measured as indices of sys-temic inflammation, at the onset of the inflammatoryprocess (days 0–3) and 7 days after arthritis induction.On days 1, 2, and 3 after injection with K/BxN serum, thePGE2 concentration was increased (mean � SEM 111 �10 ng/ml, 80 � 4 ng/ml, and 71 � 6 ng/ml, respectively),as compared with levels in mice injected with controlserum (Figure 3A). Concomitant injection with CoPPsignificantly reduced the level of PGE2 on days 1, 2, and3, to 72 � 5 ng/ml, 60 � 6 ng/ml, and 52 � 3 ng/ml,respectively (P � 0.01 versus levels in mice concomi-tantly injected with vehicle). In parallel, CoPP cotreat-ment reduced 92-kd gelatinolytic activity (matrix metal-loproteinase 9 [MMP-9]), whereas MMP-2 activityremained constant (Figure 3B). The antiinflammatoryeffects of CoPP on serum PGE2 levels and MMP-9activity were also observed on day 7.

We then specifically inhibited HO-1 in vivo inCoPP and K/BxN serum–treated mice, by siRNA treat-ment as described above. PGE2 production (Figure 3C)and MMP-9 activity (Figure 3D) were significantlyhigher in the presence of the anti–HO-1 siRNA ascompared with the nonspecific siRNA. Again, these dataprovide strong evidence that the protective effect ofCoPP on systemic inflammation is essentially mediatedby HO-1.

HO-1 attenuates systemic cytokine productionand increases systemic antioxidant status but has noeffect on systemic NO production elicited by K/BxNserum. A similar protocol involving CoPP and siRNAtreatment of K/BxN serum–treated mice was carried outto investigate serum levels of the major inflammatorycytokines involved in the acute phase of nonautoimmunearthritis. As shown in Figure 4A, K/BxN serum treat-ment induced production of IL-1�, TNF�, and IL-6(mean � SEM 45 � 2 pg/ml, 23 � 3 pg/ml, and 1,767 �300 pg/ml, respectively), which was reduced to basallevels by CoPP cotreatment. The antiinflammatory ef-fects of CoPP were suppressed by treatment with anti–HO-1 siRNA injection (Figure 4B). The nonspecificsiRNA had no effect. A similar phenomenon was ob-served with regard to the antioxidant status of serum. Asshown in Figure 4A, cotreatment with CoPP and K/BxNincreased the antioxidant activity by a factor of 2.7 ascompared with treatment with K/BxN alone. This sys-temic effect of CoPP on antioxidant status was sup-

pressed by injection of anti–HO-1 siRNA (Figure 4B).No similar effect on systemic NO production was ob-served (Figures 4A and B).

These findings demonstrate that the protectiveeffect of HO-1 is mediated mainly by significant reduc-tion of proinflammatory cytokine production and up-regulation of systemic antioxidant activity, but not byregulation of systemic NO production. In addition, thesedata, along with results presented above, indicate thatthe protective effects of CoPP occur in the early phase of

Figure 4. Effects of HO-1 on systemic cytokine production, serumantioxidant status, and systemic nitric oxide (NO) production in K/BxNserum–treated BALB/c mice. Serum levels of interleukin-1� (IL-1�),IL-6, and tumor necrosis factor � (TNF�) were measured by enzyme-linked immunosorbent assay. The total antioxidant status of serum wasdetermined by evaluating inhibition of the oxidation of ABTS. Serumlevels of NO were measured by the Griess reaction. A, Effect of CoPP.Sham-treated mice and mice treated with control serum plus CoPPexhibited no significant alteration in serum levels of IL-1�, IL-6,TNF�, NO, or antioxidant contents (data not shown). B, Effect ofsiRNA. Values are the mean and SEM (n � 8 mice per group). �� �P � 0.01. ND � not determined (see Figure 2 for other definitions).

2590 BENALLAOUA ET AL

induced arthritis and are sustained for at least 7 days andmediated mainly by HO-1.

HO-1–induced reduction of local tissue inflam-mation and cellular recruitment elicited by K/BxNserum. Examination of histologic specimens fromBALB/c mice injected with K/BxN serum or controlserum in conjunction with CoPP or vehicle revealed thepresence of inflamed joint tissue in animals treated withK/BxN serum, but a total absence of inflammation in thejoints of K/BxN serum–treated animals that receivedconcomitant CoPP treatment or control mice (Figure 5).In parallel, we observed iNOS staining only in inflamedjoint tissue from mice treated with K/BxN serum alone(Figure 5). No staining for Ki-67, a usual marker of cellproliferation, was observed under any of the experimen-tal conditions (results not shown), indicating that therewas no modulation of cell proliferation and that theeffect of HO-1 more likely involves cell recruitment.These findings suggest that HO-1 could protect againstjoint tissue inflammation at the local level by down-regulating iNOS and reducing cell recruitment.

HO-1 attenuates cartilage degradation but doesnot inhibit chondrocyte iNOS expression elicited byK/BxN serum. Cartilage degradation was assessed bymeasuring serum levels of PIICP on day 7. Circulatinglevels of the PIICP epitope were reduced in CoPP-cotreated mice (mean � SEM 46 � 6 ng/ml) as com-pared with mice treated with K/BxN serum alone (68 �8 ng/ml) (Figure 6A), showing that CoPP attenuatescartilage degradation induced by K/BxN serum. Finally,positive immunostaining of articular chondrocytes foriNOS was observed in mice treated with K/BxN serumalone and also, surprisingly, in mice treated with K/BxNserum plus CoPP (Figure 6B). These results and thosedescribed above indicate that, although it is locallyprotective against joint tissue inflammation and cartilagedegradation, HO-1 is still unable to inhibit iNOS expres-sion in articular chondrocytes.

DISCUSSION

The purpose of the present work was to analyzewhether in vivo induction of HO-1 during the acute

Figure 5. Inhibitory effect of HO-1 on local joint inflammation in K/BxN serum–treated BALB/c mice. Upperpanels, Histologic findings in the joints on day 7. Mouse paws were dissected, fixed, decalcified, dehydrated, andembedded in paraffin. Sagittal sections (5 �m) were stained with hematoxylin and eosin. Arrows showinflammatory infiltration in the joint space. TB � inferior extremity of the tibia; TL � talus (originalmagnification � 100). Lower panels, Immunostaining for inducible nitric oxide synthase (iNOS) in the joints onday 7. Sagittal sections (5 �m) were immunostained with iNOS antibody and stained with hematoxylin. Arrowsshow positive immunostaining for iNOS in cells from the inflammatory infiltrate. Sections depicted correspondto the boxed areas in the upper panels (original magnification � 400). Results shown are representative offindings in studies of 8 mice. See Figure 2 for other definitions.

HO-1 AND INFLAMMATORY ARTHRITIS 2591

phase of inflammatory arthritis in mice leads to protec-tive antiinflammatory effects, including attenuation ofsystemic proinflammatory cytokine induction and jointdestruction. We used the K/BxN mouse model of in-duced arthritis, which mimics human inflammatory ar-thritis without any lymphocyte influence. HO-1 induc-tion by CoPP suppressed the development of arthritisinduced by K/BxN serum in BALB/c mice. CoPP alsoattenuated joint destruction, PGE2 production, andMMP-9 activity, as well as serum IL-1�, IL-6, and TNF�production. These effects were largely abolished in vivoin the presence of an anti–HO-1 siRNA.

The original K/BxN mouse model was a model ofspontaneous arthritis, described in 1996 by Kouskoff etal (37). In 1999, an alternative model based on serumtransfer from arthritic K/BxN mice into healthy animalswas introduced by the same group (31). As pointed outin a recent report by Ohmura et al (38), this latter modelprovides a powerful experimental tool for analyzing theacute effector-phase pathways in arthritis, without the

confounding influence of the immunologic phase. Thisserum transfer model is simple, robust, produces 100%arthritis incidence in BALB/c mice, and provides a rapid(24-hour) response (31). We used the K/BxN mouseserum transfer model because the primary intent of thepresent study was to investigate in a broad sense theinnate acute inflammatory arthritic process observed inhuman inflammatory arthritis, without lymphocyte influ-ence.

The poor specificity of HO-1 modulators (CoPP,hemin, zinc protoporphyrin IX, and tin protoporphyrinIX) used either in vivo or in vitro, as well as the lack ofefficient breeding of HO-1–null mice, are sources ofexperimental discrepancies. Metalloporphyrin inhibitorsof HO-1 robustly affect enzymes such as NOS andguanylate cyclase, in addition to their various othernonspecific effects (39,40). Thus, to evaluate the exactrole of HO-1 in the effects of CoPP observed in theK/BxN mouse model of arthritis, we developed ananti–HO-1 siRNA and a method for its systemic deliv-

Figure 6. Effect of HO-1 on articular cartilage in K/BxN serum–treated BALB/c mice. A, Effect of CoPP oncartilage degradation, as evaluated spectrophotometrically by detection of the C-propeptide of type II collagen(CPII) epitope in the serum on day 7. Sham-treated mice and mice treated with control serum plus CoPPexhibited no significant alteration in serum levels of the CPII epitope (data not shown). Values are the mean andSEM (n � 8 mice per group). �� � P � 0.01. B, Immunostaining for inducible nitric oxide synthase (iNOS) inarticular chondrocytes on day 7. Sagittal sections (5 �m) were immunostained with iNOS antibody and stainedwith hematoxylin. Positive immunostaining for iNOS in articular chondrocytes was observed in mice treated withK/BxN serum plus vehicle and mice treated with K/BxN serum plus CoPP, but not in control mice (originalmagnification � 400). Results shown are representative of findings in studies of 8 mice. See Figure 2 for otherdefinitions.

2592 BENALLAOUA ET AL

ery. The efficient suppression of the beneficial effects ofCoPP by coinjection of the anti–HO-1 siRNA stronglysupports the notion that the antiinflammatory effects ofCoPP are mediated mainly by HO-1.

Our data are partially at variance with the find-ings of in vivo studies by Devesa et al (28–30). Thoseauthors showed that pharmacologic up-regulation ofHO-1 by CoPP was not sufficient to control the progres-sion of chronic inflammation in rat adjuvant-inducedarthritis and mouse collagen-induced arthritis (28,29),whereas inhibition of the endogenous form of HO-1significantly reduced the severity of arthritis and theproduction of inflammation mediators in rat adjuvant-induced arthritis (30). These discrepancies appear to usto be essentially related to the different models ofarthritis. Devesa and colleagues used classic modelsspecifically designed to study the late immunologicphase of arthritis while we chose the K/BxN serumtransfer model, which mimics the acute, nonauto-immune phase of inflammatory arthritis.

Besides the lymphocyte influence, another expla-nation for the discrepancies observed with the auto-immune and nonautoimmune arthritis models may re-late to qualitative and quantitative differences in theHO-1 inducers used. Devesa et al used CoPP at 2.5mg/kg body weight in the collagen-induced arthritismodel and 5 mg/kg in the rat adjuvant-induced arthritismodel (28,29), whereas we used 25 mg/kg in our model.Using hemin, another HO-1 inducer, Devesa et alobserved no response in rat adjuvant-induced arthritis(28), a result which differed from findings reported byKobayashi et al (25). Conversely, our results on HO-1–induced protective effects and their suppression by aspecific siRNA parallel the in vitro findings of Koba-yashi and colleagues in studies performed on an LPS-treated fibroblast-like synoviocyte cell line in culture(25). Overall, our findings, in accordance with theexisting literature, suggest that strategies aimed at HO-1up-regulation may not be beneficial in chronic auto-immune arthritis but may have real therapeutic potentialduring the acute phase of arthritis, in which lymphocyteinfluence is not implicated, e.g., in acute gout, pseudo-gout, and infectious arthritis.

Acute joint inflammation mainly involves 3 tissueprocesses: proliferation of inflammatory tissue in thejoints, degradation of cartilage, and destruction of sub-chondral bone tissue. In the present study, HO-1 up-regulation reduced inflammatory joint tissue prolifera-tion and cartilage degradation (Figures 5 and 6A) butthe effect on cartilage was only partial, as evidenced bythe finding that iNOS was still expressed in chondrocytes

in the presence of CoPP (Figure 6B). Zwerina et aldemonstrated inhibition of osteoclastogenesis and bonedestruction in mice with TNF-mediated acute inflamma-tion in the presence of hemin, known as an HO-1inducer (41). These results and ours suggest that HO-1potentially has a pivotal role in protection against jointtissue proliferation and subchondral bone tissue destruc-tion. However, its effect on cartilage is less clear, and thefindings suggest differential regulation of NO produc-tion in cartilage and serum by HO-1.

In conclusion, this study is the first to demon-strate that the acute joint inflammatory response in-duced by K/BxN serum in BALB/c mice can be counter-acted by pharmacologic up-regulation of HO-1 by CoPP.This chemical inducer elicits a robust decrease in jointdestruction. This decrease is abolished in the presence ofanti–HO-1 siRNA, which demonstrates the essentialinvolvement of HO-1 expression in the process. Ourresults strongly support the notion that HO-1 has apotential role in treatment during the acute phase ofhuman inflammatory types of arthritis, such as acutegout and pseudogout, infectious arthritis, seronegativespondylarthritides, rheumatoid arthritis, and juvenilearthritis.

ACKNOWLEDGMENTSThe authors thank Carole Nocco, Irene Balguy, Marie-

Anne Gougerot-Pocidalo, Abdelhamid Alomolki, and Tzong-Shyuan Lee for technical assistance.

AUTHOR CONTRIBUTIONS

Dr. Rannou had full access to all of the data in the study andtakes responsibility for the integrity of the data and the accuracy of thedata analysis.Study design. Francois, Batteux, Shyy, Fitting, Boczkowski, Savouret,Corvol, Poiraudeau, Rannou.Acquisition of data. Benallaoua, Francois, Batteux, Thelier, Fitting,Tsagris, Rannou.Analysis and interpretation of data. Benallaoua, Francois, Thelier,Shyy, Tsagris, Boczkowski, Savouret, Corvol, Poiraudeau, Rannou.Manuscript preparation. Benallaoua, Francois, Batteux, Shyy, Bocz-kowski, Savouret, Corvol, Poiraudeau, Rannou.Statistical analysis. Benallaoua, Rannou.

REFERENCES1. Choy EH, Panayi GS. Cytokine pathways and joint inflammation

in rheumatoid arthritis. N Engl J Med 2001;344:907–16.2. Liu-Bryan R, Pritzker K, Firestein GS, Terkeltaub R. TLR2

signaling in chondrocytes drives calcium pyrophosphate dihydrateand monosodium urate crystal-induced nitric oxide generation.J Immunol 2005;174:5016–23.

3. Kyburz D, Rethage J, Seibl R, Lauener R, Gay RE, Carson DA, etal. Bacterial peptidoglycans but not CpG oligodeoxynucleotidesactivate synovial fibroblasts by Toll-like receptor signaling. Arthri-tis Rheum 2003;48:642–50.

4. Asadullah K, Sterry W, Volk HD. Interleukin-10 therapy—reviewof a new approach. Pharmacol Rev 2003;55:241–69.

HO-1 AND INFLAMMATORY ARTHRITIS 2593

5. Wamsley M, Katsikis PD, Abney E, Parry S, Williams RO, MainiRN, et al. Interleukin-10 inhibition of the progression of estab-lished collagen-induced arthritis. Arthritis Rheum 1996;39:495–503.

6. Murakami Y, Akahoshi T, Kawai S, Inoue M, Kitasato H. Anti-inflammatory effect of retrovirally transfected interleukin-10 onmonosodium urate monohydrate crystal–induced acute inflamma-tion in murine air pouches. Arthritis Rheum 2002;46:2504–13.

7. Gjertsson I, Hultgren OH, Tarkowski A. Interleukin-10 amelio-rates the outcome of Staphylococcus aureus arthritis by promotingbacterial clearance. Clin Exp Immunol 2002;130:409–14.

8. Otterbein LE, Otterbein SL, Ifedigbo E, Liu F, Morse DE, FearnsC, et al. MKK3 mitogen-activated protein kinase pathway medi-ates carbon monoxide-induced protection against oxidant-inducedlung injury. Am J Pathol 2003;163:2555–63.

9. Abraham NG, Kappas A. Heme oxygenase and the cardiovascular-renal system. Free Radic Biol Med 2005;39:1–25.

10. Otterbein LE, Soares MP, Yamashita K, Bach FH. Heme oxygen-ase-1: unleashing the protective properties of heme. Trends Im-munol 2003;24:449–55.

11. Chen S, Kapturczak MH, Wasserfall C, Glushakova OY, Camp-bell-Thompson M, Deshane JS, et al. Interleukin 10 attenuatesneointimal proliferation and inflammation in aortic allografts by aheme oxygenase-dependent pathway. Proc Natl Acad Sci U S A2005;102:7251–6.

12. Lee TS, Chang CC, Zhu Y, Shyy JY. Simvastatin induces hemeoxygenase-1: a novel mechanism of vessel protection. Circulation2004;110:1296–302.

13. Juan SH, Cheng TH, Lin HC, Chu YL, Lee WS. Mechanism ofconcentration-dependent induction of heme oxygenase-1 by res-veratrol in human aortic smooth muscle cells. Biochem Pharmacol2005;69:41–8.

14. Lee TS, Chau LY. Heme oxygenase-1 mediates the anti-inflam-matory effect of interleukin-10 in mice. Nat Med 2002;8:240–6.

15. Lee TS, Tsai HL, Chau LY. Induction of heme oxygenase-1expression in murine macrophages is essential for the anti-inflammatory effect of low dose 15-deoxy-�12,14-prostaglandin J2.J Biol Chem 2003;278:19325–30.

16. Kataoka K, Handa H, Nishizawa M. Induction of cellular antioxi-dative stress genes through heterodimeric transcription factorNrf2/small Maf by antirheumatic gold(I) compounds. J Biol Chem2001;276:34074–81.

17. Sato H, Yamaguchi M, Shibasaki T, Ishii T, Bannai S. Induction ofstress proteins in mouse peritoneal macrophages by the anti-rheumatic agents gold sodium thiomalate and auranofin. BiochemPharmacol 1995;49:1453–7.

18. Minamino T, Christou H, Hsieh CM, Liu Y, Dhawan V, AbrahamNG, et al. Targeted expression of heme oxygenase-1 prevents thepulmonary inflammatory and vascular responses to hypoxia. ProcNatl Acad Sci U S A 2001;98:8798–803.

19. Inoue S, Suzuki M, Nagashima Y, Suzuki S, Hashiba T, TsuburaiT, et al. Transfer of heme oxygenase 1 cDNA by a replication-deficient adenovirus enhances interleukin 10 production fromalveolar macrophages that attenuates lipopolysaccharide-inducedacute lung injury in mice. Hum Gene Ther 2001;12:967–79.

20. Laniado-Schwartzman M, Abraham NG, Conners M, Dunn MW,Levere RD, Kappas A. Heme oxygenase induction with attenua-tion of experimentally induced corneal inflammation. BiochemPharmacol 1997;53:1069–75.

21. Kawahito Y, Kondo M, Tsubouchi Y, Hashiramoto A, Bishop-Bailey D, Inoue K, et al. 15-deoxy-�12,14-PGJ2 induces synoviocyteapoptosis and suppresses adjuvant-induced arthritis in rats. J ClinInvest 2000;106:189–97.

22. Yachie A, Niida Y, Wada T, Igarashi N, Kaneda H. Oxidativestress causes enhanced endothelial cell injury in human hemeoxygenase-1 deficiency. J Clin Invest 1999;103:129–35.

23. Poss KD, Tonegawa S. Reduced stress defense in heme oxygenasedeficient cells. Proc Natl Acad Sci U S A 1997;94:10925–30.

24. Kapturczak MH, Wasserfall C, Brusko T, Campbell-Thompson M,Ellis TM, Atkinson MA, et al. Heme oxygenase-1 modulates earlyinflammatory responses: evidence from the heme oxygenase-1-deficient mouse. Am J Pathol 2004;165:1045–53.

25. Kobayashi H, Takeno M, Saito T, Takeda Y, Kirino Y, Noyori K,et al. Regulatory role of heme oxygenase 1 in inflammation ofrheumatoid arthritis. Arthritis Rheum 2006;54:1132–42.

26. Kirino Y, Takeno M, Murakami S, Kobayashi M, Kobayashi H,Miura K, et al. Tumor necrosis factor � acceleration of inflamma-tory responses by down-regulating heme oxygenase 1 in humanperipheral monocytes. Arthritis Rheum 2007;56:464–75.

27. Fernandez P, Guillen MI, Gomar F, Alcaraz MJ. Expression ofheme oxygenase-1 and regulation by cytokines in human osteoar-thritic chondrocytes. Biochem Pharmacol 2003;66:2049–52.

28. Devesa I, Ferrandiz ML, Busserolles J, Alcaraz MJ. Effects ofheme oxygenase-1 inducers on established rat adjuvant arthritis.Cell Mol Biol (Noisy-le-grand) 2005;51:479–85.

29. Devesa I, Ferrandiz ML, Terencio MC, Joosten LA, van den BergWB, Alcaraz MJ. Influence of heme oxygenase 1 modulation onthe progression of murine collagen-induced arthritis. ArthritisRheum 2005;52:3230–8.

30. Devesa I, Ferrandiz ML, Guillen I, Cerda JM, Alcaraz MJ.Potential role of heme oxygenase-1 in the progression of ratadjuvant arthritis. Lab Invest 2005;85:34–44.

31. Korganow A, Ji H, Mangialaio S, Duchatelle V, Pelanda R, MartinT, et al. From systemic T cell self-reactivity to organ-specific auto-immune disease via immunoglobulins. Immunity 1999;10:451–61.

32. De Bandt M, Ben Mahdi MH, Ollivier V, Grossin M, Dupuis M,Gaudry M, et al. Blockade of vascular endothelial growth factorreceptor I (VEGF-RI), but not VEGF-RII, suppresses jointdestruction in the K/BxN model of rheumatoid arthritis. J Immu-nol 2003;171:4853–9.

33. Francois M, Richette P, Tsagris L, Raymondjean M, Fulchignoni-Lataud MC, Forest C, et al. Peroxisome proliferator-activatedreceptor-� down-regulates chondrocyte matrix metalloprotein-ase-1 via a novel composite element. J Biol Chem 2004;279:28411–8.

34. Taille C, Foresti R, Lanone S, Zedda C, Green C, Aubier M, et al.Protective role of heme oxygenases against endotoxin-induceddiaphragmatic dysfunction in rats. Am J Respir Crit Care Med2001;163:753–61.

35. Lewis DL, Hagstrom JE, Loomis AG, Wolff JA, Herweijer H.Efficient delivery of siRNA for inhibition of gene expression inpostnatal mice. Nat Genet 2002;32:107–8.

36. Zhang X, Shan P, Jiang D, Noble PW, Abraham NG, Kappas A,et al. Small interfering RNA targeting heme oxygenase-1 enhancesischemia-reperfusion-induced lung apoptosis. J Biol Chem 2004;279:10677–84.

37. Kouskoff V, Korganow AS, Duchatelle V, Degott C, Benoist C,Mathis D. Organ-specific disease provoked by systemic autoimmu-nity. Cell 1996;87:811–22.

38. Ohmura K, Johnsen A, Ortiz-Lopez A, Desany P, Roy M, BesseW, et al. Variation in IL-1� gene expression is a major determi-nant of genetic differences in arthritis aggressivity in mice. ProcNatl Acad Sci U S A 2005;102:12489–94.

39. Meffert MK, Haley JE, Schuman EM, Schulman H, Madison DV.Inhibition of hippocampal heme oxygenase, nitric oxide synthase,and long-term potentiation by metalloporphyrins. Neuron 1994;13:1225–33.

40. Grundemar L, Ny L. Pitfalls using metalloporphyrins in carbonmonoxide research. Trends Pharmacol Sci 1997;18:193–5.

41. Zwerina J, Tzima S, Hayer S, Redlich K, Hoffmann O, Hanslik-Schnabel B, et al. Heme oxygenase 1 (HO-1) regulates osteoclas-togenesis and bone resorption. FASEB J 2005;19:2011–3.

2594 BENALLAOUA ET AL