Available online at www.sciencedirect.com

www.EuropeanJournalPain.com

European Journal of Pain 12 (2008) 1059–1068

Role of cytokines in mediating mechanical hypernociceptionin a model of delayed-type hypersensitivity in mice

Thiago M. Cunha a, Waldiceu A. Verri Jr. a, Daniel A. Valerio a, Ana T. Guerrero a,Luciana G. Nogueira a, Silvio M. Vieira a, Danielle G. Souza b, Mauro M. Teixeira b,

Stephen Poole c, Sergio H. Ferreira a, Fernando Q. Cunha a,*

a Department of Pharmacology, Faculty of Medicine of Ribeirao Preto University of Sao Paulo, Av. Bandeirantes, 3900,

14049-900 Ribeirao Preto, SP, Brazilb Department of Biochemistry and Immunology, Institute of Biological Science (ICB), Federal University of Minas Gerais,

Av. Antonio Carlos, 6627, 31270-901 Belo Horizonte, MG, Brazilc Division of Immunology and Endocrinology, National Institute for Biological Standards and Control, Blanche Lane, South Mimms,

Potters Bar, Herts EN6 3 QG, UK

Received 22 August 2007; received in revised form 30 January 2008; accepted 2 February 2008Available online 26 March 2008

Abstract

In the present study, we used the electronic version of the von Frey test to investigate the role of cytokines (TNF-a and IL-1b)and chemokines (KC/CXCL-1) in the genesis of mechanical hypernociception during antigen-induced inflammation in mice. Thenociceptive test consisted of evoking a hindpaw flexion reflex with a hand-held force transducer (electronic anesthesiometer) adaptedwith a 0.5 mm2 polypropylene tip. The intraplantar administration of methylated bovine serum albumin (mBSA) in previouslyimmunized (IM), but not in sham-immunized (SI) mice, induced mechanical hypernociception in a dose-dependent manner. Hypern-ociception induced by antigen was reduced in animals pretreated with IL-1ra and reparixin (a non-competitive allosteric inhibitor ofCXCR2), and in TNF receptor type 1 deficient (TNFR1�/�) mice. Consistently, antigen challenge induced a time-dependentrelease of TNF-a, IL-1b and KC/CXCL-1 in IM, but not in SI, mice. The increase in TNF-a levels preceded the increase in IL-1b and KC/CXCL1. Antigen-induced release of IL-1b and KC/CXCL1 was reduced in TNFR1�/� mice, and TNF-a-inducedhypernociception was inhibited by IL-1ra and reparixin. Hypernociception induced by IL-1b in immunized mice was inhibitedby indomethacin, whereas KC/CXCL1-induced hypernociception was inhibited by indomethacin and guanethidine. Antigen-induced hypernociception was reduced by indomethacin and guanethidine and abolished by the two drugs combined. Together,these results suggest that inflammation associated with an adaptive immune response induces hypernociception that is mediatedby an initial release of TNF-a, which triggers the subsequent release of IL-1b and KC/CXCL1. The latter cytokines in turn stimulatethe release of the direct-acting final mediators, prostanoids and sympathetic amines.� 2008 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All

rights reserved.

Keywords: Cytokines; Hypernociception; Hyperalgesia; Inflammatory pain; Delayed-type hypersensitivity; TNF-a

1090-3801/$34 � 2008 European Federation of Chapters of the International

reserved.

doi:10.1016/j.ejpain.2008.02.003

* Corresponding author. Tel.: +55 16 3602 3204; fax: +55 16 36330021.

E-mail address: fdqcunha@fmrp.usp.br (F.Q. Cunha).

1. Introduction

Inflammatory hyperalgesia results from the sensitiza-tion of primary afferent neurons which is betterdescribed as hypernociception (decrease in nociceptive

Association for the Study of Pain. Published by Elsevier Ltd. All rights

1060 T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068

threshold) in animal models (Cunha et al., 2008). It isinduced by inflammatory mediators, such as prostaglan-dins and sympathetic amines, which directly sensitizeperipheral nociceptive neurons (Ferreira and Nakam-ura, 1979; Khasar et al., 1999; Nakamura and Ferreira,1987).

The release of these direct-acting hypernociceptivemediators is generally preceded by a cascade of cyto-kines (Verri et al., 2006b). In rats, tumor necrosis factor(TNF)-a is the first cytokine released, which stimulatestwo independent hypernociceptive pathways: (a) Inter-leukin (IL)-1b production which induces prostanoidproduction and (b) CXC chemokines which induce therelease of sympathetic amines (Cunha et al., 1992; Ferre-ira et al., 1988; Lorenzetti et al., 2002). As in rats,inflammatory hypernociception in mice is also mediatedby a cascade of cytokines (Cunha et al., 2005). In mice,carrageenin induces a concomitant release of TNF-aand keratinocyte-derived chemokine (KC/CXCL1). Bothmediators stimulate the subsequent release of IL-1bwhich in turn induces prostanoid production. Moreover,KC/CXCL1 also stimulates the sympathetic componentof inflammatory hypernociception.

The majority of human inflammatory diseases inwhich pain is a critical symptom, such as rheumatoidarthritis, are the consequence of an adaptive immuneresponse (Schaible et al., 2002). However, most of theinflammatory models used to investigate the genesis ofhypernociception are based on an innate inflammatoryresponse (Verri et al., 2006a). A major differencebetween adaptive and innate immune responses residesin the cells and cytokines that trigger the process. Whilein adaptive response the lymphocytes and the cytokinesIL-2, IL-12, Interferon-gamma and IL-23 are crucial, ininnate response this process depends on macrophages,mast cells and the cytokines IL-1b and TNF-a. It isnoteworthy that the latter cytokines are also importantin the effector phase of adaptive immune response(Langrish et al., 2004).

Immune inflammatory models have been developedby the administration of different protein antigens(methylated bovine serum albumin (mBSA) and ovalbu-min) in previously immunized animals (Brackertz et al.,1977; Canetti et al., 2001). These models are importantfor the study of inflammatory responses in many auto-immune diseases. For instance, mBSA challenge intothe joint of previously immunized mice is a reproducibleexperimental model that has several features of humanarthritis (Brackertz et al., 1977). Models of mBSA-induced Th1-predominant inflammation of immunizedanimals have also been performed in the peritoneum,paw and skin (Cook et al., 2003; Dunn et al., 1989;Teixeira et al., 2001). The latter studies have focusedon mechanisms of leukocyte infiltration and have notinvestigated inflammatory hypernociception. Recently,our group showed that ovalbumin challenge induces

mechanical hypernociception in previously immunizedanimals (Cunha et al., 2003; Verri et al., 2007, 2006b).Hypernociception observed in these models is dependenton TNFa, IL-1b and CXC chemokines in rats anddependent on IL-15 and IL-18 in mice. In the presentstudy, we characterized the cascade of events leadingto mechanical hypernociception in a model of Th1-pre-dominant inflammation induced by challenging immu-nized mice with mBSA.

2. Methods

2.1. Animals

All experiments were performed in 20–30 g C57BL/6(wild type and TNF receptor type 1 deficient, TNFR1�/�) mice housed in the animal care facility of the Schoolof Medicine of Ribeirao Preto University of Sao Paulo.Breeding pairs of mice with targeted disruption ofTNFR1 gene were obtained from Amgen Institute (Tor-onto, Canada; Pfeffer et al., 1993). These mice wereback-crossed with C57BL/6 for 10 generations and thegenotype of TNFR1�/� mice determined by polymer-ase chain reaction of DNA as previously described (Pfef-fer et al., 1993). Mice were taken to the testing room atleast 1 h before experiments and were used only once.Food and water were available ad libitum. Animal careand handling procedures were in accordance with theInternational Association for the Study of Pain (Zim-mermann, 1983) guidelines on the use of animals in painresearch and they were approved by the Committee forEthics in Animal Research of the Faculty of Medicine ofRibeirao Preto-USP.

2.2. Mechanical hypernociceptive test

As mentioned before, the term hypernociceptionrather than hyperalgesia or allodynia is used to definethe decrease of the nociceptive withdrawal threshold(Cunha et al., 2008). Mechanical hypernociception wastested in mice as previously reported (Cunha et al.,2004). In a quiet room, mice were placed in acrylic cages(12 � 10 � 17 cm) with wire grid floors, 15–30 minbefore the start of testing. The test consisted of evokinga hindpaw flexion reflex with a hand-held force trans-ducer (electronic anesthesiometer; IITC Life Science,Woodland Hills, CA) adapted with a 0.5 mm2 polypro-pylene tip. The investigator was trained to apply the tipperpendicular to the central area of the hindpaw with agradual increase in pressure. The end-point was charac-terized by the removal of the paw followed by clearflinching movements. After paw withdrawal, the inten-sity of the pressure was automatically recorded, andthe final value for the response was obtained by averag-ing three measurements. The animals were tested before

T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068 1061

and after treatments. The test is highly objective, since itrecognizes, in a dose-dependent manner, the effect ofclinically used analgesic drugs, such as non-steroidalanti-inflammatory drugs and opioids (Cunha et al.,2004; Verri et al., 2004). It is important to mention thatall experiments were conducted in a double-blind fash-ion in which the person who injected the solutions wasnot same one who made the behavioral assessment.

The following treatments were used based on previ-ous experiments carried out in our laboratory: IL-1ra(an IL-1 receptor antagonist, 2 � 30 mg kg�1 – saline –200 ll - i.p. 30 min before and 3.5 h after stimulus injec-tion) or reparixin (a non-competitive allosteric inhibitorof CXCR2, 30 mg kg�1 saline – 100 ll – i.v. 30 minbefore stimulus injection (Souza et al., 2004)), indometh-acin (a prostaglandin synthesis inhibitor; 5 mg kg�1 –Tris/HCl – 200 ll – i.p. 30 min before stimulus injection;Cunha et al., 2004) and guanethidine (sympathetic neu-ron blocker; 30 mg kg�1 – saline – 100 ll – s.c. 60 minbefore stimulus injection; Cunha et al., 2005). Theresults are expressed by the delta (D) withdrawal thresh-old (in g) calculated by subtracting the mean zero-timemeasurements from the mean time interval measure-ments. The withdrawal threshold was 8.9 ± 0.3 g(mean ± SEM; n = 20) before injection of the hyperno-ciceptive agents. Although the force necessary to inducethe behavioral response in normal paw could promote aslight lift of the hindpaw, it is not observed during theonset of inflammation. For local administration, stimuliwere injectedsubcutaneously into the plantar region ofthe hindpaws. A 26-G hypodermic needle was insertedinto the skin of the second footpad (to avoid backflow)and the tip of the needle was introduced. A volume of25 ll was administered.

2.3. Procedures for active immunization

Mice were immunized (IM) with mBSA. Briefly,mBSA (500 lg/200 ll) emulsified in complete Freund’sadjuvant (CFA; 1 mg/ml of Mycobacterium tuberculo-

sis) plus saline (1:1) was injected subcutaneously (inthe back) in mice on day 0. The mice received a subcu-taneous booster with the same emulsion on day 7.Control mice (sham-immunized, SI) were injected withthis emulsion (200 ll) without mBSA. CFA was usedas adjuvant in this immunization protocol because itinduces mainly a Th1-driven response. After 21 days,the mice were challenged by the intraplantar (i.pl.)administration of mBSA or saline into one of the hind-paws (Brackertz et al., 1977). It is important to men-tion that the immunization process did not alter thenociceptive threshold of the mouse paw compared withnaıve mice at the time of the paw challenge (21 daysafter the first immunization procedure; data notshown). Furthermore, the baselines of the IM and SIgroups did not differ.

2.4. Cytokine measurements

At indicated times after the injection of inflammatorystimulus (mBSA), animals were anesthetized (4%chloral hydrate, i.p.) and decapitated, and the plantartissues (ffi0.4 cm2) were removed from the injected andcontrol paws (saline and naıve). The samples weretriturated and homogenized in 500 ll of the appropriatebuffer (phosphate-buffered saline containing 0.05%Tween 20, 0.1 mM phenylmethylsulphonyl fluoride,0.1 mM benzethonium chloride, 10 mM EDTA and20 kallikrein international units of aprotinin A) fol-lowed by a centrifugation of 10 min/2000g. The super-natants were stored at �70 �C until further analysis.The TNF-a, IL-1b and KC/CXCL1 levels weredetermined as described previously (Cunha et al.,2007, 2005; Safieh-Garabedian et al., 1995) byenzyme-linked immunosorbent assay (ELISA). Theresults are expressed as picograms (pg) of each cytokineper paw. As a control, the concentrations of these cyto-kines were determined in immunized mice injected withsaline as well as in sham-immunized mice injected withmBSA.

2.5. Spleen cell culture and IL-2 measurement

Spleen cells from WT or TNFR1�/� mice were cul-tured at 5 � 105 cells/well in RPMI 1640 (Sigma) supple-mented with 2 mM L-glutamine, 100 IU/ml penicillin,100 lg/ml streptomycin, and 10% fetal bovine serum(all from Sigma) at 37 �C in 5% CO2. Cells were stimu-lated with mBSA (10 lg/ml) for 72 h. Supernatants ofthese cultures were stored at �70 �C until determinationof IL-2 content by ELISA.

2.6. Anti-mBSA ab titer determination

Serum anti-mBSA ab titers in pooled sera from WTor TNFR1�/� mice were measured by ELISA. Briefly,96-well plates were coated with 50 ll of mBSA solution(10 lg/ml, in 0.1 M phosphate buffer) overnight at 4 �C.Thereafter, serial dilutions of sera were added and incu-bated overnight at 4 �C. Bound, total IgG was detectedwith biotin-conjugated anti-mouse IgG (Vector).Finally, 50 ll of avidin-HRP (1:5000 dilution; DAKOA/S, Denmark) were added to each well, and after30 min the plates were washed and the color reagentOPD (200 lg/well; Sigma) was added. After 15 min,the reaction was stopped with 1 M H2SO4 and the opti-cal density (O.D.) read at 490 nm.

2.7. Drugs

The following materials were obtained from thesources indicated: recombinant murine IL-1b, recombi-nant human IL-1ra, recombinant murine TNF-a were

Fig. 1. Dose- and time-response curves of the hypernociceptioninduced by i.pl. injection of mBSA (10, 30, 60, 90 and 120 lg perpaw) in immunized (IM) and sham-immunized (SI) or saline-treated(control) mice. The hypernociceptive effects were determined at 1–144 hafter the stimulus injection. Each time point represents the mean ±SEM (n = 5). *P < 0.05, when compared with SI mice i.pl. injected withmBSA.

1062 T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068

provided by the National Institute for Biological Stan-dards and Control (NIBSC, South Mimms, Hertford-shire, UK). KC/CXCL1 chemokine was purchasedfrom PeproTech (Rocky Hill, NJ). Reparixin was a kindgift from Dr. Riccardo Bertini (Dompe Pharmaceuti-cals, L’Aquila, Italy). Guanethidine, mBSA and CFAwere purchased from Sigma Chemical Co. (St. Louis,MO, USA) and indomethacin from Prodome Quımicae Farmaceutica (Sao Paulo, SP, Brazil).

2.8. Statistical analysis

The results are presented as means ± SEM and arerepresentative of two separate experiments with fiveanimals per group. The data were not combined forstatistical analysis, they were analyzed individuallyand the significances were similar. Two-way analysisof variance (ANOVA) was used to compare the groupsand doses at all times (curves) when the hypernocicep-tive responses were measured at different times afterthe stimulus injection. The factors analyzed were treat-ments, time and time by treatment interaction. Whenthere was a significant time by treatment interaction,one-way ANOVA followed by Bonferroni’s t-test wasperformed for each time. For non-significant time bytreatment interaction curves, the mean of repeatedmeasures at different times was calculated and one-way ANOVA followed by Tukey’s test was used tocompare the groups. Alternatively, when the hyperno-ciceptive responses were measured once after the stim-ulus injection, the differences between responses wereevaluated by one-way ANOVA followed by Bonferron-i’s t-test. Differences were considered to be statisticallysignificant at P < 0.05.

3. Results

3.1. Antigen challenge induces mechanical

hypernociception in previously immunized mice

In attempt to characterize the nociception induced byan inflammatory response secondary to a Th1-predomi-nant delayed-type hypersensitivity, antigen (mBSA) wasinjected locally in the hindpaw of immunized mice. Itwas observed that intraplantar injection of mBSA pro-duced a dose-dependent (10–120 lg paw�1 – saline�25 ll) mechanical hypernociception in previouslyimmunized but not in sham-immunized mice (Fig. 1).The Two-way analysis of variance indicated a time(F7,222 = 110; P < 0.0001), treatment (F6,222 = 282;P < 0.0001) and time by treatment interaction(F42,222 = 6.9; P < 0.0001). The baseline nociceptivethreshold of immunized mice is similar to that ofsham-immunized mice. BSA-induced hypernociceptionwas already significant 1 h after injection, reached a

maximal response between the 3rd and 5th hour anddecreased only after 72 h post mBSA challenge. ThemBSA dose of 90 lg paw�1 was selected for subsequentexperiments.

3.2. TNF-a, IL-1b and CXCR2 chemokine receptor

ligands mediate hypernociception during antigen-induced

inflammation

Next, the participation of IL-1b, TNF-a andCXCR2-acting chemokines in mBSA-induced hyperno-ciception was investigated by using IL-1ra, TNFR1�/�mice and reparixin. The two-way analysis showed thatthere is no time by treatment interaction (F12,84 = 0.35).However, it showed a treatment interaction (F6,84 = 79).One-way analysis (F6,28 = 23) indicated that pretreat-ment of mice with IL-1ra (P < 0.001) or reparixin(P < 0.001) significantly reduced the hypernocicep-tive effect of mBSA (90 lg paw�1, 25 ll; Fig. 2A).mBSA-induced hypernociception was also reduced inTNFR1�/� mice (P < 0.001; Fig. 2A). Treatment ofmice with the combination of IL-1ra and reparixininhibited mBSA-induced hypernociception to the samelevel observed in mBSA-injected TNFR1�/� mice(P < 0.001; Fig. 2A). The hypernociceptive index is arepresentation of the area under the curve (Fig. 2B).

3.3. Antigen induces cytokine production in the paw ofimmunized mice

To confirm our behavioral experiments, the levels ofthe above-mentioned cytokines were determined in thepaw skin of immunized mice challenged with mBSA.Injection of mBSA in the paw of previously immunized,but not sham-immunized, mice induced TNF-a(F9,40 = 45; P < 0.0001) IL-1b (F9,40 = 58; P < 0.0001)

Fig. 2. (A) Mechanical hypernociception induced by i.pl. injection ofmBSA in WT and TNF-R1�/�immunized mice: effects of IL-1ra,reparixin, and combination of IL-1ra and reparixin. Nociceptiveresponse induced by mBSA (90 lg per paw) in WT mice pretreatedwith vehicle or TNF-R1�/� mice and pretreatment of WT mice with30 mg/Kg of reparixin (RTX) delivered i.v., 30 mg/kg (2�) of IL-1radelivered i.p., or reparixin plus IL-1ra. The hypernociceptive responseswere determined 3, 5 and 7 h after mBSA injection. (B) Hypernoci-ceptive index represents the area under the curve for differenttreatments on mBSA-induced hypernociception over the differenttimes. Each time point and bar represents the mean ± SEM (n = 5).*P < 0.05 when compared with vehicle treatments.

T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068 1063

and KC/CXCL1 (F9,40 = 31; P < 0.0001) chemokineproduction (Fig. 3A–C). The increase in TNF-a levelspreceded the increase in levels of IL-1b and KC/CXCL1

(Fig. 3). In attempt to test whether TNF-a is responsiblefor the production of IL-1b and KC/CXCL1, asobserved during innate inflammatory response in rats(Cunha et al., 1992), the production of IL-1b and KC/CXCL1 stimulated by mBSA was determined inTNFR1�/� mice. The local production of both cyto-kines (IL-1b, F2,12 = 57; P < 0.001; KC/CXCL1,F2,12 = 13; P < 0.01) in response to antigen challengewas inhibited in TNFR1�/� mice (Fig. 4A, B).

In order to confirm that TNF-a could indeed mediatehypernociception via the release of IL-1b and CXCL1,we tested the effect of IL-1ra and reparixin on TNF-a-induced mechanical hypernociception in previouslyimmunized mice. One-way analyses (F4,20 = 61.2) indi-cated that the pretreatment of mice with reparixin(P < 0.001) and IL-1ra (P < 0.001) reduced mechanicalhypernociception induced by TNF-a (100 pg paw�1, sal-ine �25 ll). Moreover, the combination of these twodrugs abolished the hypernociceptive effect of TNF-a(P < 0.001; Fig. 4C).

To investigate whether the reduced response ofTNFR1�/� mice was due to a reduced immunizationstate we determined the levels of antibody againstmBSA in the serum of immunized mice, the IL-2 levelsin the supernatant of spleen-derived cells stimulatedwith mBSA (indirect measure of lymphocyte prolifera-tion) as well as the levels of TNF-a in the mouse pawchallenged with mBSA. It was observed that IL-2(F2,9 = 10; P > 0.05) production in lymph node culturesof previously immunized mice stimulated with mBSAand the serum levels of antibody against mBSA(F2,6 = 9; P > 0.05) were similar in WT and TNFR1�/�mice (Fig. 5A, B). In further support of these findings,intraplantar administration of mBSA increased the levelsof TNF-a in immunized TNFR1�/� as in WT mice(F2,12 = 17; P > 0.05; Fig. 5C). These experiments sug-gested that TNFR1�/� mice were immunized to thesame extent as WT mice.

3.4. Cytokines trigger release of prostanoids and

sympathetic amines which mediate mBSA-induced

inflammatory hypernociception

The hypernociceptive effect of cytokines is indirectand depends on the release of direct-acting hypernoci-ceptive mediators (prostanoids and sympatheticamines), which bind to their respective receptors presenton nociceptors and induce mechanical hypernociception(Verri et al., 2006a). One-way analyses (F3,16 = 40)showed that pretreatment of mice with indomethacin(P < 0.001), but not with guanethidine (P > 0.05)inhibited IL-1b (1 ng paw�1, saline �25 ll)-inducedhypernociception in immunized mice, whereas the hyper-nociceptive effect of KC (10 ng paw�1; saline �25 ll)was reduced by both drugs (F3,16 = 23; P < 0.001,Fig. 6).

Fig. 3. I.pl. injection of mBSA-stimulated TNF-a, KC/CXCL1 and IL-1b production in paw skin of immunized (IM) mice but not in sham-immunized (SI) mice. Concentrations of TNF-a (A), KC/CXCL1 (B), and IL-1b (C) in mouse paws injected with 90 lg of mBSA or saline (Sal) or innaive paws of IM and SI mice. At 0.5, 1, 3, 5, 7 and 24 h after injection, the mice were killed and paw skin samples were extracted for cytokines, whichwere measured by ELISA. The results are expressed as the means ± SEM of four animals per group. *, Statistically significant differences whencompared with the control group (P < 0.05).

1064 T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068

Next, it was evaluated the involvement of prostanoidsand sympathetic amines in mechanical hypernociceptioninduced by mBSA challenge in immunized mice. Thetwo-way analysis showed that there is no time by treat-ment interaction (F8,54 = 1.4). However, it showed atreatment interaction (F4,54 = 122). Consistently, one-way analysis (F4,20 = 52) indicated that mechanicalhypernociception induced by antigen challenge of previ-ously immunized mice was reduced by indomethacin(P < 0.05) and guanethidine (P < 0.05), and almostabolished by the combination of the two drugs(P < 0.001; Fig. 7A). The hypernociceptive index is arepresentation of the area under the curve for all timesevaluated (Fig. 7B).

4. Discussion

As the majority of inflammatory diseases that areaccompanied by pain show components of adaptiveimmunity, it is important that animal models that eval-uate hypernociception try to mimic adaptive immuneresponses. In the present study, we characterized a novelmodel of hypernociception in mice. Mechanical hypern-ociception was induced by antigen challenge of miceimmunized with mBSA in Freund’s complete adjuvant(Brackertz et al., 1977). This model of delayed-typehypersensitivity is characterized by marked infiltrationof leukocytes and is dependent on the activity ofCD4+ lymphocytes (Cher and Mosmann, 1987). The

Fig. 4. (A) Production of cytokines in TNF-R1�/� immunized mice. Concentration of IL-1b (A) and KC (B) in WT and TNF-R1�/� mouse pawsinjected with 90 lg of mBSA or saline (Sal). At 5 h after injection, the mice were killed, and paw skin samples were extracted for cytokines, whichwere measured by ELISA. (C) Nociceptive response induced by TNF-a (100 pg per paw) in WT mice pretreated with vehicle or with 30 mg/Kg ofreparixin (RTX) delivered i.v., 30 mg/kg of IL-1ra delivered i.p., and reparixin plus IL-1ra. The hypernociceptive responses were determined 3 h afterTNF-a injection. The results are expressed as the means ± SEM of four animals per group. *, Statistically significant differences compared with thesaline group (P < 0.05); # statistically significant differences when compared with the WT mouse group and vehicle-treated groups (P < 0.05).

Fig. 5. (A) Production of IL-2 in the supernatant of spleen-derived cells from WT and TNFR1�/� immunized (IM) and sham-immunized (SI) mice.Spleen-derived cells (5 � 105/well) were stimulated with mBSA (10 lg/ml), and after 72 h the levels of IL-2 were determined by ELISA. (B) Levels ofantibody against mBSA in the serum of WT and TNFR1�/� immunized (IM) and sham-immunized (SI) mice. (C) Production of TNF-a in paws ofWT and TNFR1�/� immunized mice 5 h after the i.pl. challenge with mBSA (90 lg/paw). The results are expressed as the means ± SEM of fouranimals per group. *, Statistically significant differences when compared with the SI mice and cell groups (P < 0.05).

T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068 1065

adaptive nature of the present model could be confirmedby the fact that mBSA-induced mechanical hypernoci-ception only in previously immunized mice but not innon-immunized (sham-immunized) mice. It wasobserved that mechanical hypernociception induced byantigen challenge of immunized mice depended on therelease of a cascade of cytokines. TNF-a was the firstcytokine released and, by acting on TNFR1, it triggeredtwo distinct pathways. One pathway was dependent onIL-1b production which in turn activated a prosta-noid-dependent hypernociceptive response. Secondly,TNF-a/TNFR1 also stimulated the release of KC/CXCL1 which activated CXCR2 and induced hyperno-

ciceptive pathways dependent on both prostanoids andsympathetic amines.

The cascade of events involved in the genesis ofinflammatory hypernociception has been extensivelyinvestigated in animal models of innate inflammation.For instance, LPS- or carrageenin-induced mechanicalhypernociception in the rat and mouse paw dependson a sequential release of cytokines and chemokines(Verri et al., 2006a). In mice, carrageenin-inducedhypernociception depends on the concomitant releaseand action of TNF-a and KC/CXCL1, and the twomediators cooperate to induce the subsequent releaseof IL-1b/prostanoids (Cunha et al., 2005). KC/CXCL1

Fig. 6. Effect of indomethacin and guanethidine on IL-1b– andCXCL1/KC-induced mechanical hypernociception. Hypernociceptiveresponse induced by IL-1b (1 ng per paw) or CXCL1/KC wasevaluated in mice pretreated with vehicle (—), indomethacin (Indo)5 mg/kg delivered i.p., or guanethidine (Gua) 30 mg/kg delivered s.c.The hypernociceptive responses were determined 3 h after IL-1b orCXCL1/KC injection. Each bar represents the mean ± SEM (n = 5).* indicates statistically significant differences when compared withvehicle-treated group (P < 0.05).

Fig. 7. (A) Effect of indomethacin and guanethidine on mBSA-induced mechanical hypernociception. Hypernociceptive responseinduced by mBSA (90 lg per paw) was evaluated in WT immunizedmice pretreated with vehicle, indomethacin (Indo) 5 mg/kg deliveredi.p., guanethidine (Gua) delivered s.c., or indomethacin plus guaneth-idine. The hypernociceptive responses were determined 3, 5 and 7 hafter mBSA injection. (B) The hypernociceptive index represents thearea under the curve for different treatments in mBSA-inducedhypernociception over the different times. Each time point and barrepresents the mean ± SEM (n = 5). * indicates statistically significantdifferences when compared with vehicle-treated group (P < 0.05).

1066 T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068

is also responsible for triggering the sympathetic compo-nent of inflammatory hypernociception (Cunha et al.,2005). In the present study, we also showed that a sim-ilar set of mediators, i.e., TNF-a, IL-1b and CXC che-mokines, was involved in hypernociception duringmBSA-induced inflammation. However, the sequenceof events differed following inflammation observed in adelayed-type hypersensitivity reaction.

Initial experiments showed that TNF-a was releasedearly after injection of antigen in immunized animals,and its release preceded that of other cytokines (IL-1band KC/CXCL1), suggesting a sequential release ofthese cytokines, i.e., TNF-a production would mediatethe release of IL-1b and KC/CXCL1. The role ofTNF-a was investigated by using animals that lackTNFR1 (Cunha et al., 2005; Kondo and Sauder,1997). Indeed, this receptor has been implicated in thehypernociceptive effects of TNF-a (Cunha et al., 2005).In these animals, hypernociception induced by antigenwas reduced. There was also marked inhibition of therelease of both IL-1b and CXCL1, suggesting thatTNF-a release and activation of TNFR1 were necessaryfor further cytokine release and consequent inflamma-tory hypernociception. This significant role of TNF-ain mediating further cytokine release, tissue inflam-mation and hypernociception is consistent with thetherapeutic effect of anti-TNF-a antibodies or solubleTNFR1 in the setting of chronic inflammatory diseases,including rheumatoid arthritis (Rankin et al., 1995;Taylor et al., 2001; Williams et al., 1992). The fact thathypernociception induced by mBSA was not abolishedin TNFR1�/� mice suggests that other mediators could

also be involved in adaptive immune response inducedhypernociception. For instance, endothelins (Verriet al., 2006a,b), IL-18 (Verri et al., 2006a), IL-15 (Verriet al., 2006a) and leukotrines (Cunha et al., 2003) medi-ate hypernociception during immune inflammatoryresponse.

Although there is evidence that cytokines may actdirectly on nociceptive neurons (Verri et al., 2006a),the majority of the studies have demonstrated thatcytokines induce hypernociception by an indirectmechanism, stimulating the release of direct-actinghypernociceptive mediators, such as prostanoids andsympathetic amines (Cunha et al., 1992, 2005). In thepresent study, during the delayed-type hypersensitivityreaction after the release of IL-1b and CXC chemokinesby TNF-a, their hypernociceptive actions seem to bedependent on the production of prostaglandins and

T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068 1067

sympathetic amines, respectively. On the other hand, asmentioned, during innate immune response in mice,TNF-a and CXC chemokines are produced concomi-tantly, and they stimulate the IL-1b/prostaglandinhypernociceptive pathway. Thus, although a similar setof mediators appears to be important for innate andimmune-associated hypernociception, the sequence ofevents are different (Cunha et al., 2005).

The reasons for the differences in the mediatorinvolved in the induction of hypernociception duringinnate and adaptive inflammation are not immediatelyapparent but could be due to differential role of residentcell types that trigger the process. In this respect, T cellsare needed for mBSA-induced inflammation and, inaddition to resident macrophages and mast cells, theycould play a role in mediating inflammatory hypernoci-ception observed in the delayed-type hypersensitivityreaction (Canetti et al., 2001; Cher and Mosmann,1987; Teixeira et al., 2001). For instance, whereas mac-rophages are the main source of TNF-a in the innateimmune response, during the adaptive immuneresponse, T cells, especially CD4+ cells, seem to playan important role in the release of this inflammatorycytokine (Canetti et al., 2001). Recently, another typeof T cell has been identified, Th17 cells (Yu and Gaffen,2008). These T cells are characterized by their ability toproduce IL-17 when activated (Yu and Gaffen, 2008).They are involved in the pathogenesis of many inflamma-tory diseases, including rheumatoid arthritis (Chabaudet al., 1999). For instance, the production of IL-17 atthe site of inflammation is responsible for triggeringthe release of different inflammatory mediators, suchas TNF-a and chemokines, culminating in leukocyteinfiltration and tissue lesion (Koenders et al., 2006).Although we did not investigate the role of IL-17 inthe onset of hypernociception during delayed-typehypersensitivity reaction, it is reasonable to suggest thatthe events initiated by TNF-a, observed in the presentstudy, could be stimulated by the IL-17 produced byTh17 cells. Therefore, it could also explain the differ-ences in the events that occur during innate and adaptiveimmune responses.

Besides the role of cellular components of theimmune system in the adaptive immune response, wecannot discard the possible participation of humoralfactors. In this context, antibodies against mBSA aredetected in the serum of immunized mice. They couldalso be involved in the initiation of the inflammatoryresponse, for instance by activation of the classical com-plement cascade (Woodruff et al., 2002). In fact, werecently demonstrated that during adaptive immuneresponse, the activation of the complement system gen-erates C5a, which plays a critical role in the genesis ofhypernociception (Ting et al., 2008).

In conclusion, we extend the concept that cytokinesconstitute a link between cellular injury or immunolog-

ical recognition and the development of inflammatorypain. Although there is a difference in the sequence ofrelease of cytokines following innate and adaptiveinflammation, a common set of cytokines mediateshypernociception via the release of prostanoids andsympathomimetic amines, which are direct-actinghypernociceptive mediators. These findings are consis-tent with the concept that inhibition of cytokine produc-tion or action is an important target for new therapeuticapproaches to control pain that follows immune inflam-matory diseases.

Acknowledgements

We thank Ieda Regina dos Santos Schivo, Sergio Ro-berto Rosa, and Giuliana Bertozi Francisco for excellenttechnical support. This work was supported by grantsfrom Fundac�ao de Amparo a Pesquisa do Estado deSao Paulo, Fundac�ao de Amparo a Pesquisa do Estadode Minas Gerais and Conselho Nacional de Desenvolvi-mento Cientıfico e Technologico. T.M.C. is a recipientof a Ph.D. Scholarship from Fundacao de Amparo aPesquisa do Estado de Sao Paulo. We are also gratefulto Dr.A. Leyva for English editing of the manuscript.

References

Brackertz D, Mitchell GF, Mackay IR. Antigen-induced arthritis inmice. I. Induction of arthritis in various strains of mice. ArthritisRheum 1977;20:841–50.

Canetti C, Silva JS, Ferreira SH, Cunha FQ. Tumour necrosis factor-alpha and leukotriene B(4) mediate the neutrophil migration inimmune inflammation. Brit J Pharmacol 2001;134:1619–28.

Chabaud M, Durand JM, Buchs N, Fossiez F, Page G, Frappart L,et al. Human interleukin-17: a T cell-derived proinflammatorycytokine produced by the rheumatoid synovium. Arthritis Rheum1999;42:963–70.

Cher DJ, Mosmann TR. Two types of murine helper T cell clone. II.Delayed-type hypersensitivity is mediated by TH1 clones. JImmunol 1987;138:3688–94.

Cook AD, Braine EL, Hamilton JA. The phenotype of inflammatorymacrophages is stimulus dependent: implications for the nature ofthe inflammatory response. J Immunol 2003;171:4816–23.

Cunha FQ, Poole S, Lorenzetti BB, Ferreira SH. The pivotal role oftumour necrosis factor alpha in the development of inflammatoryhyperalgesia. Brit J Pharmacol 1992;107:660–4.

Cunha JM, Sachs D, Canetti CA, Poole S, Ferreira SH, Cunha FQ. Thecritical role of leukotriene B4 in antigen-induced mechanicalhyperalgesia in immunised rats. Brit J Pharmacol 2003;139:1135–45.

Cunha TM, Verri Jr WA, Vivancos GG, Moreira IF, Reis S, ParadaCA, et al. An electronic pressure-meter nociception paw test formice. Braz J Med Biol Res 2004;37:401–7.

Cunha TM, Verri Jr WA, Silva JS, Poole S, Cunha FQ, Ferreira SH. Acascade of cytokines mediates mechanical inflammatory hyperno-ciception in mice. Proc Natl Acad Sci USA 2005;102:1755–60.

Cunha TM, Verri Jr WA, Fukada SY, Guerrero AT, Santodomingo-Garzon T, Poole S, et al. TNF-alpha and IL-1beta mediateinflammatory hypernociception in mice triggered by B(1) but notB(2) kinin receptor. Eur J Pharmacol 2007;573:221–9.

1068 T.M. Cunha et al. / European Journal of Pain 12 (2008) 1059–1068

Cunha TM, Verri Jr WA, Poole S, Parada CA, Cunha FQ, FerreiraSH. Pain facilitation by proinflammatory cytokine actions atperipheral nerve terminals. In: DeLeo J, Sorkin L, Watkins L,editors. Immune and glial regulation of pain, vol. 1. Seattle: IASPPress; 2008; p. 67–83.

Dunn CJ, Gibbons AJ, Miller SK. Development of a delayed-typehypersensitivity granuloma model in the mouse for the study ofchronic immune-mediated inflammatory disease. Agents Actions1989;27:365–8.

Ferreira SH, Nakamura M. I – Prostaglandin hyperalgesia, a cAMP/Ca2+ dependent process. Prostaglandins 1979;18:179–90.

Ferreira SH, Lorenzetti BB, Bristow AF, Poole S. Interleukin-1 beta asa potent hyperalgesic agent antagonized by a tripeptide analogue.Nature 1988;334:698–700.

Khasar SG, McCarter G, Levine JD. Epinephrine produces a beta-adrenergic receptor-mediated mechanical hyperalgesia and in vitrosensitization of rat nociceptors. J Neurophysiol 1999;81:1104–12.

Koenders MI, Joosten LA, van den Berg WB. Potential new targets inarthritis therapy: interleukin (IL)-17 and its relation to tumournecrosis factor and IL-1 in experimental arthritis. Ann Rheum Dis.2006;65(Suppl. 3):iii29–=0?>iii33, November.

Kondo S, Sauder DN. Tumor necrosis factor (TNF) receptor type 1(p55) is a main mediator for TNF-alpha-induced skin inflamma-tion. Eur J Immunol 1997;27:1713–8.

Langrish CL, McKenzie BS, Wilson NJ, de Waal Malefyt R, KasteleinRA, Cua DJ. IL-12 and IL-23: master regulators of innate andadaptive immunity. Immunol Rev 2004;202:96–105.

Lorenzetti BB, Veiga FH, Canetti CA, Poole S, Cunha FQ, FerreiraSH. Cytokine-induced neutrophil chemoattractant 1 (CINC-1)mediates the sympathetic component of inflammatory mechanicalhypersensitivity in rats. Eur Cytokine Network 2002;13:456–61.

Nakamura M, Ferreira SH. A peripheral sympathetic component ininflammatory hyperalgesia. Eur J Pharmacol 1987;135:145–53.

Pfeffer K, Matsuyama T, Kundig TM, Wakeham A, Kishihara K,Shahinian A, et al. Mice deficient for the 55 kd tumor necrosisfactor receptor are resistant to endotoxic shock, yet succumb to L.

monocytogenes infection. Cell 1993;73:457–67.Rankin EC, Choy EH, Kassimos D, Kingsley GH, Sopwith AM,

Isenberg DA, et al. The therapeutic effects of an engineered humananti-tumour necrosis factor alpha antibody (CDP571) in rheuma-toid arthritis. Brit J Rheumatol 1995;34:334–42.

Safieh-Garabedian B, Poole S, Allchorne A, Winter J, Woolf CJ.Contribution of interleukin-1 beta to the inflammation-induced

increase in nerve growth factor levels and inflammatory hyperal-gesia. Brit J Pharmacol 1995;115:1265–75.

Schaible HG, Ebersberger A, Von Banchet GS. Mechanisms of pain inarthritis. Ann NY Acad Sci 2002;966:343–54.

Souza DG, Bertini R, Vieira AT, Cunha FQ, Poole S, Allegretti M,et al. Repertaxin, a novel inhibitor of rat CXCR2 function, inhibitsinflammatory responses that follow intestinal ischaemia andreperfusion injury. Brit J Pharmacol 2004;143:132–42.

Taylor PC, Williams RO, Maini RN. Immunotherapy for rheumatoidarthritis. Curr Opin Immunol 2001;13:611–6.

Teixeira MM, Talvani A, Tafuri WL, Lukacs NW, Hellewell PG.Eosinophil recruitment into sites of delayed-type hypersensitivityreactions in mice. J Leukoc Biol 2001;69:353–60.

Ting E, Guerrero AT, Cunha TM, Verri Jr WA, Taylor SM, WoodruffTM, et al. Role of complement C5a in mechanical inflammatoryhypernociception: potential use of C5a receptor antagonists tocontrol inflammatory pain. Brit J Pharmacol 2008;153:1043–53.

Verri Jr WA, Schivo IR, Cunba TM, Liew FY, Ferreira SH, CunhaFQ. Interleukin-18 induces mechanical hypernociception in rats viaendothelin acting on ETB receptors in a morphine-sensitivemanner. J Pharmacol Exp Ther 2004;310:710–7.

Verri Jr WA, Cunha TM, Parada CA, Poole S, Cunha FQ, FerreiraSH. Hypernociceptive role of cytokines and chemokines: targets foranalgesic drug development? Pharmacol Ther 2006a;112:116–38.

Verri Jr WA, Cunha TM, Parada CA, Wei XQ, Ferreira SH, Liew FY,et al. IL-15 mediates immune inflammatory hypernociception bytriggering a sequential release of IFN-gamma, endothelin, andprostaglandin. Proc Natl Acad Sci USA 2006b;103:9721–5.

Verri Jr WA, Cunha TM, Parada CA, Poole S, Liew FY, Ferreira SH,et al. Antigen-induced inflammatory mechanical hypernociceptionin mice is mediated by IL-18. Brain Behav Immun 2007;21:535–43.

Williams RO, Feldmann M, Maini RN. Anti-tumor necrosis factorameliorates joint disease in murine collagen-induced arthritis. ProcNatl Acad Sci USA 1992;89:9784–8.

Woodruff TM, Strachan AJ, Dryburgh N, Shiels IA, Reid RC, FairlieDP, et al. Antiarthritic activity of an orally active C5a receptorantagonist against antigen-induced monarticular arthritis in therat. Arthritis Rheum 2002;46:2476–85.

Yu JJ, Gaffen SL. Interleukin-17: a novel inflammatory cytokine thatbridges innate and adaptive immunity. Front Biosci. 2008;13:170–7.

Zimmermann M. Ethical guidelines for investigations of experimentalpain in conscious animals. Pain 1983;16:109–10.