5 5 6 | E I J K E L K A M P E T A L . | M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2

INTRODUCTIONProper functioning of the nociceptive

system is essential to protect the bodyfrom tissue damage. Inflammation sensi-tizes the nociceptive system, leading to alower threshold to painful stimuli (hy-peralgesia) (1). This process is thought to

serve adaptive purposes, but becomesmaladaptive when hyperalgesia persistsafter resolution of inflammation. Theproinflammatory cytokine interleukin(IL)-1β directly sensitizes nociceptors,leading to transient hyperalgesia (2,3).Peripheral injection of other inflamma-

tory mediators such as prostaglandin E2(PGE2) also increases the sensitivity ofnociceptors and the response to painfulstimuli (4). Moreover, there is evidencethat proinflammatory cytokines such asIL-1β and tumor necrosis factor (TNF)-αcontribute to the genesis of neuropathicpain (5,6).

Many of the signals involved in in-flammatory hyperalgesia are generatedvia activation of G protein–coupled re-ceptors (GPCRs) expressed in sensoryneurons. The activity of GPCRs is regu-lated by the family of GPCR kinases(G protein–coupled receptor kinase[GRK] 1–7). Agonist-activated GPCRs arephosphorylated by GRKs, inducing rapiduncoupling from the G protein, a process

G Protein–Coupled Receptor Kinase 6 Acts as a CriticalRegulator of Cytokine-Induced Hyperalgesia by PromotingPhosphatidylinositol 3-Kinase and Inhibiting p38 Signaling

Niels Eijkelkamp,1,2 Cobi J Heijnen,1 Anibal Garza Carbajal,1 Hanneke L D M Willemen,1 Huijing Wang,1Michael S Minett,2 John N Wood,2 Manfred Schedlowski,3 Robert Dantzer,4 Keith W Kelley,4 andAnnemieke Kavelaars1,4

1Laboratory of Neuroimmunology and Developmental Origins of Disease (NIDOD), University Medical Center Utrecht, Utrecht,the Netherlands; 2Molecular Nociception Group, University College London, London, U.K.; 3Institute of Medical Psychology andBehavioral Immunobiology, University Hospital Essen, University of Duisburg-Essen, Essen, Germany; and 4Integrative Immunologyand Behavior Program, College of Agricultural Consumer, and Environmental Sciences (ACES) and College of Medicine, Universityof Illinois at Urbana Champaign, Urbana, Illinois, United States of America

The molecular mechanisms determining magnitude and duration of inflammatory pain are still unclear. We assessed the con-tribution of G protein–coupled receptor kinase (GRK)-6 to inflammatory hyperalgesia in mice. We showed that GRK6 is a criticalregulator of severity and duration of cytokine-induced hyperalgesia. In GRK6–/– mice, a significantly lower dose (100 times lower)of intraplantar interleukin (IL)-1β was sufficient to induce hyperalgesia compared with wild-type (WT) mice. In addition, IL-1β hy-peralgesia lasted much longer in GRK6–/– mice than in WT mice (8 d in GRK6–/– versus 6 h in WT mice). Tumor necrosis factor (TNF)-α–induced hyperalgesia was also enhanced and prolonged in GRK6–/– mice. In vitro, IL-1β–induced p38 phosphorylation inGRK6–/– dorsal root ganglion (DRG) neurons was increased compared with WT neurons. In contrast, IL-1β only induced activationof the phosphatidylinositol (PI) 3-kinase/Akt pathway in WT neurons, but not in GRK6–/– neurons. In vivo, p38 inhibition attenuatedIL-1β– and TNF-α–induced hyperalgesia in both genotypes. Notably, however, whereas PI 3-kinase inhibition enhanced and pro-longed hyperalgesia in WT mice, it did not have any effect in GRK6-deficient mice. The capacity of GRK6 to regulate pain re-sponses was also apparent in carrageenan-induced hyperalgesia, since thermal and mechanical hypersensitivity was signifi-cantly prolonged in GRK6–/– mice. Finally, GRK6 expression was reduced in DRGs of mice with chronic neuropathic or inflammatorypain. Collectively, these findings underline the potential role of GRK6 in pathological pain. We propose the novel concept thatGRK6 acts as a kinase that constrains neuronal responsiveness to IL-1β and TNF-α and cytokine-induced hyperalgesia via biasedcytokine-induced p38 and PI 3-kinase/Akt activation.Online address: http://www.molmed.orgdoi: 10.2119/molmed.2011.00398

Address correspondence to Annemieke Kavelaars, Laboratory of Neuroimmunology andDevelopmental Origins of Disease (NIDOD), University Medical Center Utrecht, Lundlaan 6,Office KC 03.068.0, 3584 EA Utrecht, Netherlands. Phone: +31 88 755 4360; Fax: +31 88 7555311; E-mail: a.kavelaars@umcutrecht.nl.Submitted October 18, 2011; Accepted for publication February 3, 2012; Epub(www.molmed.org) ahead of print February 6, 2012.

R E S E A R C H A R T I C L E

M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2 | E I J K E L K A M P E T A L . | 5 5 7

called homologous receptor desensitiza-tion. GRK-mediated GPCR phosphoryla-tion facilitates binding of arrestin pro-teins, promoting GPCR internalization(7,8). GRKs are also capable of interact-ing with a variety of downstream signal-ing molecules, thereby regulating cellularsignaling independently of GPCRs (8,9).

GRK6 plays a crucial role in inflamma-tory pathologies. For example, GRK6 de-ficiency increases acute inflammatoryarthritis as well as colitis in male mice(10,11). Recently, we showed that post- inflammatory visceral hyperalgesia is enhanced in female GRK6–/– mice with-out affecting inflammation (12). This previous study indicated that GRK6plays a role in regulating visceral post-inflammatory pain, but did not give in-sight into the mechanisms involved.

Although most of the substrates ofGRK6 are probably still unknown, it hasbeen shown that GRK6 regulates desen-sitization of the chemokine receptorCXCR4, the BLT1 receptor for theleukotriene B4 (LTB4) and the calcitoningene-related peptide (CGRP) receptor(13–16). Furthermore, GRK6 binds andphosphorylates PDZ domains in Na+/H+

exchanger regulatory factor (NHERF)and binds to downstream regulatory ele-ment antagonistic modulator (DREAM),both regulators of ion channels, indicat-ing that GRK6 can also regulate cellularsignaling via mechanisms independentof GPCR desensitization (17,18).

We aimed to determine the contribu-tion of GRK6 to somatic inflammatoryhyperalgesia. As a model, we inducedlocal hyperalgesia by a single injection ofcarrageenan or the proinflammatory cy-tokines IL-1β or TNF-α into the paw. Insearch of the mechanism via which GRK6regulates hyperalgesia, we analyzed theconsequences of GRK6 deficiency for cy-tokine signaling to p38 and PI 3-kinase.

MATERIAL AND METHODS

AnimalsFemale GRK6-deficient C57BL/6 and

wild-type (WT) control littermates werebred in the Utrecht University Central

Animal Facility (12) and genotyped bypolymerase chain reaction (PCR) analysison genomic DNA. All experiments wereperformed in accordance with interna-tional guidelines and approved by theUniversity Medical Center Utrecht exper-imental animal committee or were ap-proved under the United KingdomHome Office Animals (Scientific Proce-dures) Act 1986.

Induction of Cytokine-InducedThermal Hyperalgesia

Mice received an intraplantar injectionof 5 µL 1% λ-carrageenan (Sigma-Aldrich, St. Louis, MO, USA), 5 µL re-

combinant mouse IL-1β (0.2–200 ng/mL;Preprotech, Rocky Hill, NC, USA) orTNF-α (20 ng/mL; Preprotech) in salineor 5 µL saline as a control (19).

Heat withdrawal latency times weredetermined using the Hargreaves test(IITC Life Science, Woodland Hills, CA,USA) as described (20). Mechanical allo-dynia was measured using von Freyhairs (Stoelting, Wood Dale, IL, USA),and the 50% paw withdrawal thresholdwas calculated using the up-and-downmethod (21). The observers were blindedto genotype.

The p38 inhibitor SB239063 (5 µg/paw;Sigma-Aldrich) or the PI 3-kinase in-

Figure 1. Increased and prolonged IL-1β–induced thermal hyperalgesia and mechanicalallodynia in GRK6–/– mice. (A) Heat withdrawal latencies were determined using the Harg-reaves test at three different intensities (n = 8). (B) Thresholds to mechanical stimulationwere determined using von Frey hairs (n = 14). (C) WT and GRK6–/– mice (n = 8–12) re-ceived an intraplantar injection of IL-1β, and the decrease in heat withdrawal latencywas determined 2 h after injection. Two-way ANOVA: genotype: P < 0.001; dose: P < 0.001;interaction: P < 0.001. (D, E) Time course of IL-1β–induced thermal hyperalgesia in WT andGRK6–/– mice (D: 10 pg IL-1β; E: 1,000 pg IL-1β; n = 8–14). (F) Time course of IL-1β–inducedmechanical allodynia in WT and GRK6–/– mice (n = 8). Data are expressed as mean ±SEM. *P < 0.05; **P < 0.01, ***P < 0.001.

5 5 8 | E I J K E L K A M P E T A L . | M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2

G R K 6 R E G U L A T E S C Y T O K I N E - I N D U C E D H Y P E R A L G E S I A

hibitor LY249002 (10 µg/paw; Sigma-Aldrich) was injected intraplantarly,20 min before cytokine injection (22).

Chronic Pain ModelsL5 transection (neuropathic). In iso -

flurane anesthetized mice, the transverseprocesses at the L4-S1 levels was ac-cessed, the L5 transverse process was re-moved and the left L5 spinal nerve wascut. Dorsal root ganglions (DRGs) wereisolated 4 wks later.

Carrageenan (inflammatory). Car-rageenan (2%, 20 µL; Sigma-Aldrich) wasinjected in both hindpaws. DRGs wereisolated 6 d later.

Dorsal Root Ganglia Cell CultureDRGs were digested in collagenase

(Type XI, 0.6 mg/mL; Sigma-Aldrich),protease (Streptomycis Griseus, 0.4 mg/ mL;Sigma-Aldrich) and glucose (1.8 mg/mL;Sigma-Aldrich) in Ca2+ and Mg2+-free

phosphate-buffered saline. Cells werecultured in Dulbecco’s modified Eagle’smedium (Gibco) containing 10% fetalbovine serum (Gibco), 2 mmol/L gluta-mine (Gibco), 10,000 IU/ mL penicillin-streptomycin (Gibco) and 100 ng/mLnerve growth factor (NGF) (Sigma-Aldrich) and poly-L-lysine– and laminin-coated wells. Cells were stimulated withIL-1β for 5 min 15–25 h after plating.

Western Blot AnalysisCells were homogenized in lysis buffer

(200 mmol/L NaCl, 50 mmol/L Tris-HCl,pH 7.5, 10% glycerol, 1% NP-40,2 mmol/L sodium orthovanadate,2 mmol/L phenylmethylsulfonyl fluoride[PMSF], 2 µmol/L leupeptin, protease in-hibitor mix [p3840, 1:200; Sigma-Aldrich]). Proteins were separated bysodium dodecyl sulfate–polyacrylamidegel electrophoresis (SDS-PAGE) andtransferred to polyvinylidene fluoride(PVDF) membranes (Millipore, Bedford,MA, USA). Blots were stained with rabbit-anti-p-p38, rabbit-anti-p38, rabbit-anti-p-Akt and rabbit-anti-Akt (Cell Sig-naling Technology Inc., Danvers, MA,USA) followed by goat anti-mouse- peroxidase (Jackson Laboratories) or don-key anti-rabbit-peroxidase (AmershamInternational) and developed by en-hanced chemiluminescence plus (Amer-sham International). Band density wasquantified using a GS-700 Imaging Den-sitometer (Bio-Rad, Hercules, CA, USA).

mRNA Isolation and Real-Time PCRLumbar DRGs were homogenized in

Trizol (Invitrogen, Paisley, UK). TotalRNA was isolated with RNeasy Mini Kit(Qiagen) and reverse-transcribed usingan iScript™ Select cDNA Synthesis Kit(Invitrogen). Real-time quantitative PCRwas performed with an iQ™ SYBR®

Green Supermix (Invitrogen). Primerpairs used were as follows:

GRK6: CTTGG TCTCA TAGGCGTAGG (forward), GCGGA TAAAGAAGCG AAAGG (reverse); GRK2:CGGGACTTC TGCCT GAACC ATCTG(forward), CTCGG CTGCG GACCACACG (reverse); β-arrestin1: AAGGG

ACACG AGTGT TCAAG A (forward),CCCGC TTTCC CAGGT AGAC (re-verse); β-arrestin2: GGCAA GCGCGACTTT GTAG (forward), GTGAGGGTCA CGAAC ACTTT C (reverse);β-actin: TTCTT TGCAG CTCCT TCGTT(forward), ATGGA GGGGA ATACAGCCC-3′ (reverse); and glyceraldehyde-3-phosphate dehydrogenase (GAPDH):TGCGA CTTCA ACAGC AACTC (for-ward), CTTGC TCAGT GTCCT TGCTG(reverse).

Data AnalysisData are expressed as mean ± standard

error of the mean (SEM) and analyzedusing the Student t test, one-way analy-sis of variance (ANOVA) or two-wayANOVA followed by Bonferroni analysis.A P value <0.05 was considered statisti-cally significant.

All supplementary materials are availableonline at www.molmed.org.

RESULTS

Increased IL-1β–Induced Thermal andMechanical Hyperalgesia in GRK6–/–

MiceHeat-withdrawal latencies of WT and

GRK6–/– mice at baseline were comparedusing the Hargreaves test at increasingintensities. Under baseline conditions,there was no genotype-dependent differ-ence in heat sensitivity (Figure 1A). Simi-larly, sensitivity to mechanical stimuli, asmeasured with the Von Frey test, was notdifferent between genotypes (Figure 1B).

Intraplantar injection of IL-1β dose- dependently increased thermal hyperal-gesia in WT mice, as determined at 2 hafter injection (Figure 1C). In GRK6–/–

mice, the dose-response curve wassharply shifted to the left (see Figure 1C).Notably, the lowest dose of IL-1β (100 pg)that induced detectable thermal hyperal-gesia in WT mice was 100-fold higherthan the lowest dose that induced ther-mal hyperalgesia in GRK6–/– mice (1 pg).

To determine whether GRK6 also regu-lates duration of IL-1β–induced thermalhyperalgesia, we followed intraplantar

Figure 2. Prolonged carrageenan-inducedthermal hyperalgesia and mechanical al-lodynia in GRK6–/– mice. Time course ofcarrageenan-induced thermal hyperalge-sia (A) and mechanical allodynia (B) in WTand GRK6–/– mice. n = 6 per group. Dataare expressed as mean ± SEM. *P < 0.05;**P < 0.01, ***P < 0.001.

R E S E A R C H A R T I C L E

M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2 | E I J K E L K A M P E T A L . | 5 5 9

IL-1β–induced thermal hyperalgesia overtime. Hyperalgesia induced by intraplan-tar injection of either a low (10 pg/paw)or high (1,000 pg/paw) dose of IL-1βwas markedly prolonged in GRK6- deficient mice (3 or 8 d after injection ofa low or high dose of IL-1β in GRK6–/–

mice versus 6 h or 1 d in WT mice; Fig-ures 1D, E). Mechanical allodynia in-duced by IL-1β (1,000 pg/paw) was alsomarkedly prolonged in GRK6–/– micecompared with WT mice (4–6 d after in-jection in GRK6–/– mice versus <1 d inWT mice; Figure 1F). At 24 h after intra-plantar IL-1β (1,000 pg/paw), there wereno differences in expression of COX2,IL-6 and TNF-α mRNA, indicating thatthe prolongation of hyperalgesia was in-dependent of inflammatory activity(Supplementary Figure 1).

Carrageenan-induced thermal hyperal-gesia was also significantly prolonged inGRK6–/– mice in comparison to WT mice(Figure 2A, recovery within 6–8 d inGRK6–/– mice versus 2–3 d in WT mice).

Additionally, carrageenan-induced me-chanical allodynia in GRK6–/– mice lastedthree times longer than in WT mice (Fig-ure 2B, recovery within 8–10 d inGRK6–/– versus 2–3 d in WT mice).

In Vitro Response of GRK6–/– DRGNeurons to IL-1β

Intraplantar IL-1β–induced thermal hy-persensitivity is mediated via activationof p38 in primary sensory neurons (2). Todetermine whether GRK6 deficiency fa-cilitated IL-1β signaling to p38 in primarysensory neurons, DRG neurons werestimulated in vitro for 5 min with increas-ing doses of IL-1β, and the level of p-p38was determined as a measure of p38 acti-vation. IL-1β–induced activation of p38was significantly higher in GRK6–/– DRGneurons than in WT neurons (Figures 3A,B). The dose-response curve for IL-1β–in-duced p-p38 in vitro was shifted to theleft in GRK6–/– DRG cultures comparedwith WT DRG neurons (~40-fold moresensitive). Baseline p-p38 did not differsignificantly between WT and GRK6–/–

mice (1 ± 0.16 versus 0.95 ± 0.14, n = 7).Next, we determined whether this shift inthe dose- response curve of GRK6–/– DRGneurons to IL-1β was also observed at thelevel of activation of the PI 3-kinase/Aktpathway, another important signalingcascade that mediates IL-1β responses.WT DRG neurons responded to IL-1βwith a clear increase in p-Akt (Figures3A, C). Interestingly, IL-1β did not inducea significant increase in p-Akt in GRK6–/–

DRG neurons. Baseline p-Akt did not dif-fer significantly between Wt and GRK6–/–

mice (1 ± 0.18 versus 0.85 ± 0.08, n = 8).These in vitro data indicate that GRK6

deficiency prevents activation of PI 3- kinase/Akt while facilitating p38 activa-tion, thereby inducing a switch in IL-1βsignaling from activation of both the p38and PI 3-kinase/Akt pathways towardactivation of p38 only.

In Vivo Role of p38 and PI 3-Kinase inthe IL-1β–Induced Hyperalgesia

To determine the in vivo relevance ofthe shift in IL-1β–induced activation ofthe p38 and PI 3-kinase/Akt pathway in

DRG neurons of GRK6–/– mice, we com-pared the effect of intraplantar adminis-tration of the p38 inhibitor SB239063 andthe PI 3-kinase inhibitor LY249002 onIL-1β–induced hyperalgesia in WT andGRK6–/– mice.

In line with previous reports (2), intra-plantar administration of SB239063 sig-nificantly attenuated IL-1β–induced hy-peralgesia in WT mice. The p38 inhibitorSB239063 (5 µg/paw) also inhibited themagnitude of acute IL-1β–induced hy-

Figure 3. In vitro IL-1β–induced phosphory-lation of p38 and Akt in DRG neurons ofWT and GRK6–/– mice. (A) Examplar dose-response curves for IL-1β–induced p-p38and p-Akt. Quantification of p-p38 levels(B) and p-Akt levels (C) in DRG neuronsafter stimulation with increasing concen-trations of IL-1β for 5 min (n = 3–7). All dataare expressed as mean ± SEM. *P < 0.05;**P < 0.01.

Figure 4. Role of p38 and PI 3-kinase inIL-1β–induced hyperalgesia. Mice were in-jected intraplantarly with the p38 inhibitorSB239063 (A) or the PI 3-kinase inhibitorLY249002 (B) 20 min before intraplantar in-jection of IL-1β (100 pg/paw; n = 4) or be-fore saline (C). Heat withdrawal latencieswere determined over time. All data areexpressed as mean ± SEM. GRK6–/–, IL-1β +vehicle, versus GRK6–/–, IL-1β + inhibitor:*P < 0.05, **P < 0.01, ***P < 0.001. WT, vehi-cle, versus WT, treatment: #P < 0.05, ##P <0.01, ###P < 0.001.

5 6 0 | E I J K E L K A M P E T A L . | M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2

G R K 6 R E G U L A T E S C Y T O K I N E - I N D U C E D H Y P E R A L G E S I A

peralgesia in GRK6–/– mice and partiallyattenuated the duration of hyperalgesiathat develops in these mice (Figure 4A).

In contrast, intraplantar administrationof the PI 3-kinase inhibitor LY249002(10 µg/paw) significantly enhanced themagnitude of acute (0.5–6 h) IL-1β– induced hyperalgesia in WT mice but didnot have any effect on acute IL-1β hyper-algesia in GRK6–/– mice (Figure 4B). Ad-ditionally, IL-1β–induced hyperalgesia inWT mice was significantly prolongedafter inhibition of PI 3-kinase. LY249002did not have any effect on the duration ofIL-1β–induced hyperalgesia in GRK6–/–

mice. Injection of SB239063 or LY249002alone did not have any effect on heatwithdrawal latencies in WT or GRK6–/–

mice (Figure 4C). These findings indicatethat GRK6–/– mice lack an inhibiting sig-nal that is provided in WT mice by acti-vation of the PI 3-kinase pathway.

PGE2- and TNF-α–InducedHyperalgesia in WT and GRK6–/– Mice

We also determined the effect of GRK6deletion on hyperalgesia induced byTNF-α, another proinflammatory cy-tokine that also signals to p38 and PI3-kinase/Akt (23). Acute (<1 d) TNF-α–induced hyperalgesia was enhanced inGRK6–/– mice when compared with WTmice (Figure 5A). In addition, GRK6–/–

mice also developed long-lasting hyper-algesia after a single injection of TNF-α,whereas WT mice recovered within 1 d.Inhibition of p38 with SB239063 signifi-cantly attenuated acute TNF-α–inducedhyperalgesia in both WT and GRK6–/–

mice (Figure 5B). Moreover, inhibition ofPI 3-kinase by intraplantar administrationof LY249002 increased and prolongedTNF-α–induced hyperalgesia in WT micebut did not affect TNF-α hyperalgesia inGRK6–/– mice (Figure 5C). These data in-dicate that a similar mechanism, a switchfrom activation of both p38 and PI 3- kinase to p38 only, is operative inGRK6–/– mice leading to the prolongationof TNF-α–induced hyperalgesia.

To determine whether the effect ofGRK6 deficiency on severity and dura-tion of hyperalgesia was limited to medi-

ators signaling via p38, we also analyzedPGE2-induced hyperalgesia that is knownto be cAMP-dependent protein kinase A(PKA) dependent (24,25). The data in Fig-ure 6C show that PGE2- induced thermalhyperalgesia was similar in WT andGRK6–/– mice (Figure 5D), indicating thatGRK6 deficiency does not affect hyperal-gesia induced by an inflammatory media-tor that signals via the cAMP/PKA path-way and independently of p38.

GRK6 mRNA Expression Levels inDRG of Mice with Neuropathic orInflammatory Pain

To investigate whether changes inGRK6 do occur in conditions of chronicpain, we investigated GRK6 mRNA ex-pression levels in DRGs of mice withchronic neuropathic or inflammatorypain. Four weeks after unilateral L5 nervetransection (L5 SNT), mice were more

sensitive to mechanical stimulation of theipsilateral paw (Figure 6A). Importantly,at this same time point, GRK6 mRNAlevels were significantly reduced in ipsi-lateral DRGs compared with contralateralDRGs from sham-operated mice (Figure6B). L5 SNT did not induce changes inmRNA levels for GRK2, β-arrestin1 or β-arrestin2 (Supplementary Figures 2A–C).Chronic inflammatory pain was inducedby intraplantar injection of carrageenan.Six days after carrageenan injection, heat-withdrawal latencies were reduced (Fig-ure 6C). At this time, GRK6 mRNA levelswere significantly decreased in the DRGsof carrageenan-treated mice comparedwith vehicle-treated mice (Figure 6D).mRNA levels for GRK2 and β-arrestin2did not differ between carrageenan- andvehicle-treated mice, whereas β-arrestin1mRNA levels were slightly reduced (Sup-plementary Figures 2D–F).

Figure 5. Role of p38 and PI 3-kinase in TNF-α–induced thermal hyperalgesia in GRK6–/–

mice. (A) Mice received an intraplantar injection of TNF-α (100 pg/paw; n = 8), and thechange in heat withdrawal latency was determined over time. Mice were injected intra-plantarly with the p38 inhibitor SB239063 (B) or the PI 3-kinase inhibitor LY249002 (C) 20 minbefore intraplantar injection of TNF-α (100 pg/paw; n = 4). Heat withdrawal latencies weredetermined over time. (D) WT and GRK6–/– mice (n = 8) received an intraplantar injectionof PGE2 (100 ng/2.5 µL), and the decrease in heat withdrawal latency was determinedover time. All data are expressed as mean ± SEM. GRK6–/–, vehicle, versus GRK6–/–, treat-ment: *P < 0.05, **P < 0.01, ***P < 0.001. WT, vehicle, versus WT, treatment: #P < 0.05, ##P < 0.01, ###P < 0.001.

R E S E A R C H A R T I C L E

M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2 | E I J K E L K A M P E T A L . | 5 6 1

Unfortunately, we were unable to testwhether the decrease in GRK6 mRNAwas associated with a reduction in GRK6protein, since no reliable GRK6 antibod-ies were available.

DISCUSSIONIn this study, we present the novel

concept that the kinase GRK6 plays apivotal role in regulating the durationand intensity of inflammatory hyperalge-

sia. GRK6 deficiency strongly enhancedand prolonged thermal hyperalgesia andmechanical allodynia induced by intra-plantar injection of either IL-1β orTNF-α. Similarly, hyperalgesia inducedby intraplantar injection of carrageenanwas markedly prolonged in GRK6–/–

mice. We also show that GRK6 deficiencypromotes activation of p38 while the acti-vation of PI 3-kinase/Akt is dampened.Thus, the novelty and significance ofthese results is that GRK6 emerges hereas a kinase that constrains neuronal re-sponsiveness to IL-1β and TNF-α and ul-timately cytokine-induced hyperalgesiavia biased cytokine-induced p38 and PI3-kinase/Akt activation (Figure 6E). Thepotential pathophysiological significanceof these findings is substantiated by ourfinding that DRG GRK6 expression lev-els are reduced in a model of chronicneuropathic pain and inflammatory pain.

Recent evidence indicates that hyperal-gesia induced by intraplantar adminis-tration of IL-1β is mediated via activationof p38 and subsequent modulation of theactivity of tetrodoxin-resistant sodiumchannels (2). TNF-α is also known to sig-nal to p38 and is capable of inducing hy-peralgesia via mechanisms that involveincreased transient receptor potentialcation channel subfamily V member 1(TRPV1) expression and modulation oftetrodoxin-resistant sodium channels(26,27). We show here for the first timethat GRK6 constrains IL-1β–induced p38activation in sensory neurons. This con-clusion is on the basis of our in vitro find-ing that IL-1β–induced p38 phosphoryla-tion was enhanced in GRK6–/– DRGcultures. Additionally, our in vivo studiesshowed that p38 activation is requiredfor IL-1β and TNF-α hyperalgesia both inWT and GRK6–/– mice. Finally, we showthat in vivo IL-1β– and TNF-α–inducedhyperalgesia is increased in GRK6- deficient mice. Collectively, these find-ings support a central role of p38 in cytokine-induced hyperalgesia and posi-tion GRK6 as a pivotal regulator of cytokine-induced hyperalgesia.

Our findings indicate that GRK6 defi-ciency enhances IL-1β signaling to p38 in

Figure 6. GRK6 mRNA levels in DRGs during chronic neuropathic and inflammatory pain.(A) The sensitivity to mechanical stimulation was determined in sham-operated mice (n = 8)or mice subjected to unilateral L5 SNT 4 wks after surgery (n = 10). (B) GRK6 mRNA levels inDRGs innervating the contralateral (-C) or ipsilateral (-I) side of sham-operated (n = 4)and L5 SNL (n = 5) mice 4 wks after surgery. The sensitivity to thermal stimulation was de-termined after intraplantar injection of vehicle (n = 5) or carrageenan (n = 5) (C), and 6 dlater, DRG GRK6 mRNA expression levels were determined (n = 5) (D). GRK6 mRNA expres-sion levels were corrected for GAPDH and β-actin mRNA expression levels. All data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. (E) In WT mice, IL-1β inducesactivation of the p38 and PI 3-kinase/Akt signaling cascade leading to transient hyperal-gesia. The activation of p38 promotes hyperalgesia, whereas activation of the PI 3-kinase/Akt signaling cascade constrains hyperalgesia. Loss of GRK6 enhances p38 activity in sen-sory neurons, whereas activation of the PI 3-kinase/Akt pathway is attenuated, ultimatelyleading to enhanced and prolonged cytokine-induced hyperalgesia.

5 6 2 | E I J K E L K A M P E T A L . | M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2

G R K 6 R E G U L A T E S C Y T O K I N E - I N D U C E D H Y P E R A L G E S I A

nociceptors. The IL-1 receptor is ex-pressed in virtually all sensory neurons(28). Moreover, IL-1β injection into therat paw enhances p-p38 in peripherin-positive sensory nerves (2). Finally, inmice in which Nav1.8-positive nocicep-tors are deleted, peripherin-positive neu-rons are reduced >85% and inflamma-tory pain does not develop (29).Therefore, we propose that GRK6 defi-ciency enhances IL-1β–induced hyperal-gesia via promoting p38 signaling in pe-ripherin-positive nociceptors.

Our experiments identified GRK6 as aprerequisite for IL-1β–induced activationof the PI 3-kinase/Akt pathway in pri-mary sensory neurons. This conclusion ison the basis of our finding that IL-1β in-duced a significant increase in Akt phos-phorylation in WT mice, but was unableto do so in GRK6–/– mice. Second, weshow here that IL-1β– and TNF-α– inducedPI 3-kinase activity is required to con-strain cytokine-induced hyperalgesia inmagnitude as well as duration. Inhibitionof PI 3-kinase enhanced and prolongedIL-1β and TNF-α hyperalgesia in WTmice, whereas inhibition of PI 3- kinasedid not affect hyperalgesia in GRK6–/–

mice. Thus, our studies in GRK6-defi-cient mice reveal a completely novel roleof the PI 3-kinase/Akt signaling cascadein attenuating the severity and durationof cytokine-induced hyperalgesia. Inter-estingly, recent evidence shows that PI 3-kinase also mediates a negativefeedback loop in preventing neuronalhyperexcitability in the Drosophila neuro-muscular junction (30). In mice, it wasshown that PI 3-kinase inhibition blockscapsaicin- and NGF- induced increases inpain sensitivity (hyperalgesia) (31),whereas the morphine-induced reduc-tion in pain sensitivity (analgesia) is me-diated via PI 3-kinase (32). Our presentdata demonstrate that PI 3-kinase inhibi-tion increases and prolongs cytokine-in-duced hyperalgesia in WT mice. The dif-ferential effects of PI 3-kinase activity indetermining pain sensitivity may dependon the fact that the isoform of PI 3-kinasethat is activated in response to NGF is ofa different subtype than the PI 3-kinase

activated by cytokines or morphine,PI3Kγ (PI 3- kinase γ) (32). Thus, if GRK6is only required for activation of thePI3Kγ isoform of PI 3-kinase activated byTNF-α and IL-1β, one would expect thatNGF-induced hyperalgesia would not beaffected by GRK6. Indeed, we show thatNGF-induced hyperalgesia was similarin GRK6–/– and WT mice (SupplementaryFigures 3A, B).

Opposing effects of the mitogen- activated protein kinase p38 and PI 3- kinases on cellular functioning have beenshown previously. Lipopolysaccharide(LPS)-induced IL-6 release is inhibitedwhen p38 is blocked, whereas inhibitionof the PI 3-kinase/ Akt pathway en-hances IL-6 production (33). Addition-ally, TNF-α and IL-1β can both induce aPI 3-kinase/Akt- dependent decrease inpotassium currents (34), whereas p38 ac-tivation enhances potassium currents(35). The exact molecular mechanismsthat explain how p38 and PI 3-kinase define the balance in the effects on cytokine-induced hyperalgesia remain tobe elucidated (Figure 6E).

The question arises through whichmechanisms GRK6 is required for the ac-tivation of PI 3-kinase and how GRK6 in-hibits p38 activation. IL-1β–induced acti-vation of PI 3-kinase requires recruitmentof PI 3-kinase to the interleukin-1 recep-tor (IL-1R) (36). Another member of theGRK family, GRK2, was implicated in fa-cilitating agonist-induced PI3Kγ recruit-ment to the β2- adrenergic receptor (37).Similarly, it is possible that GRK6 regu-lates recruitment of PI3Kγ to the IL-1β re-ceptor, enabling IL-1β–induced activationof PI 3-kinase. In addition, GRK2 inhibitsp38 activation by phosphorylation of p38at Thr-123, a residue located at its dock-ing groove (38). Importantly, the centralserine/ threonine kinase catalytic domainas well as the N-terminal regulator of Gprotein signaling (RGS)-like domain arehighly conserved in all GRKs (39). Thus,it is possible that GRK6 interferes withp38 activation via binding and phospho-rylation of p38 similar to what has beendescribed for GRK2. Future studies willhave to unravel the precise mechanisms.

We recently described that a partial re-duction in GRK2 increases hyperalgesiainduced by carrageenan and the GPCRligands CCL3 and PGE2 (24,40). How-ever, reduced GRK2 did not affect themagnitude of IL-1β–induced acute hy-peralgesia (40,41). Vice versa, we showhere that GRK6 deficiency does not affectPGE2-induced hyperalgesia. These find-ings indicate that GRK6 and GRK2 regu-late hyperalgesia induced by inflamma-tory mediators via separate mechanisms.These distinct effects of GRK6 and GRK2on inflammatory hyperalgesia conformto the notion that the different GRKshave specific and often contrasting ef-fects on signaling pathways (42).

GRK6 and other members of the GRKfamily have been originally identifiedbecause of their capacity to phosphory-late agonist-occupied GPCRs (43). How-ever, to the best of our knowledge, thereis no evidence that IL-1β and TNF-α sig-nal directly via a GPCR. One possibleexplanation for our findings could bethat IL-1β induces the local productionof GPCR ligands (for example, PGE2)and that GRK6 regulates signaling viathis receptor with consequences for hy-peralgesia. However, inhibition of COX2did not affect IL-1β–induced hyperalge-sia in WT or GRK6–/– mice (Supplemen-tary Fig ure 4). Additionally, we showedthat the magnitude and duration ofPGE2 hyperalgesia was similar in WTand GRK6–/– mice. We cannot excludethat other GPCR ligands are produced atthe site of injection of the cytokines.However, this explanation seems un-likely in view of our finding that thechange in phosphorylation of p38 andAkt in sensory neurons was already ob-served 5 min after stimulation with IL-1β in vitro. Overall, these data point to arole of GRK6 in regulating cytokine-in-duced events that are independent ofGPCR signaling.

The data presented herein indicatethat GRK6 deficiency enhances the sen-sitivity to IL-1β–induced hyperalgesia.It may therefore be that GRK6 defi-ciency also facilitates other inflamma-tory processes involving IL-1β. Indeed,

R E S E A R C H A R T I C L E

M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2 | E I J K E L K A M P E T A L . | 5 6 3

this hypothesis is supported by our ear-lier studies. For example, we haveshown that increased inflammatory vis-ceral hyperalgesia in GRK6–/– mice ischaracterized by increased neuronal andbehavioral responses to noxious colonicstimulation with capsaicin. We alsoshowed that overexpression of GRK6 at-tenuated IL-1β– induced TRPV1 sensiti-zation (12). Interestingly, blockingTRPV1 with capsazepine partially inhib-ited the ongoing IL-1β hyperalgesia 24 hafter IL-1β administration (Supplemen-tary Figure 5), suggesting enhancedTRPV1 sensitization may underlie theenhanced intraplantar IL-1β–inducedheat hyperalgesia as well. Finally, wedemonstrated that in vivo, GRK6 defi-ciency leads to increased severity aswell as duration of colitis (11). In thesemodels of visceral hyperalgesia andcolon inflammation, IL-1β is producedin the colon, and thus increased sensi-tivity to IL-1β–induced events may con-tribute to the increased and prolongedpain and colitis. In addition, Tarrant etal. (10) showed that GRK6–/– mice de-velop more severe joint inflammation inthe K/BxN model of acute inflamma-tory arthritis (10). These authorsshowed that serum IL-1 levels are nor-mal, but IL-6 levels are elevated inGRK6–/– mice in this model of inflam-matory arthritis. Because IL-1 is a potentIL-6 inducer, it may be that increasedIL-1 signaling also underlies the higherlevels of IL-6 and increased joint inflam-mation in this model.

The pathophysiological relevance ofour finding that GRK6 deficiency increases severity and duration of cy-tokine-induced hyperalgesia is under-lined by our findings on GRK6 expres-sion in DRGs in two different chronicpain models. We show that sensory neu-ron GRK6 expression is reduced in DRGsinnervating the sensitized paw of micewith chronic neuropathic or inflamma-tory pain. Moreover, chronic inflamma-tory diseases that are accompanied bypain, such as rheumatoid arthritis, are as-sociated with a decrease in intracellularGRK6 levels (44).

CONCLUSIONIn summary, here we have identified

for the first time GRK6 as a crucial ki-nase that is required to constrain cy-tokine signaling and cytokine-inducedhyperalgesia. This contribution of GRK6to hyperalgesia is likely to be mediatedby regulating the balance of cytokine- induced p38 and PI 3-kinase/Akt activa-tion in normal mice. This important rele-vant function of GRK6 in modulatingcytokine-induced signaling events mayalso play a role in chronic inflammatoryconditions.

ACKNOWLEDGMENTSWe thank Richard T Premont and Rob-

ert J Lefkowitz, Duke University, Dur-ham, North Carolina, for providing theGRK6–/– mice. This study was supportedin part by National Institutes of Healthgrants RO1-NS073939 to R Dantzer, KW Kelley and A Kavelaars and R01-NS074999 to A Kavelaars. Part of thework of N Eijkelkamp was supported bya Rubicon fellowship of the NetherlandsOrganization for Scientific Research.

DISCLOSUREThe authors declare that they have no

competing interests as defined by Molec-ular Medicine, or other interests thatmight be perceived to influence the re-sults and discussion reported in thispaper.

REFERENCES1. Marchand F, Perretti M, McMahon SB. (2005)

Role of the immune system in chronic pain. Nat.Rev. Neurosci. 6:521–32.

2. Binshtok AM, et al. (2008) Nociceptors are inter-leukin-1beta sensors. J. Neurosci. 28:14062–73.

3. Schafers M , Sorkin L. (2008) Effect of cytokineson neuronal excitability. Neurosci.Lett. 437:188–93.

4. Hucho T, Levine JD. (2007) Signaling pathwaysin sensitization: toward a nociceptor cell biology.Neuron. 55:365–76.

5. Myers RR, Campana WM, Shubayev VI. (2006)The role of neuroinflammation in neuropathicpain: mechanisms and therapeutic targets. DrugDiscov. Today 11:8–20.

6. Wolf G, Gabay E, Tal M, Yirmiya R, Shavit Y.(2006) Genetic impairment of interleukin-1 sig-naling attenuates neuropathic pain, autotomy,and spontaneous ectopic neuronal activity, fol-lowing nerve injury in mice. Pain. 120:315–24.

7. Zhang J, et al. (1997) Molecular mechanisms of Gprotein-coupled receptor signaling: role of G pro-tein-coupled receptor kinases and arrestins in re-ceptor desensitization and resensitization. Recep-tors Channels. 5:193–9.

8. Ribas C, et al. (2007) The G protein-coupled re-ceptor kinase (GRK) interactome: role of GRKs inGPCR regulation and signaling. Biochim. Biophys.Acta. 1768:913–22.

9. Reiter E, Lefkowitz RJ. (2006) GRKs and beta- arrestins: roles in receptor silencing, traffickingand signaling. Trends Endocrinol. Metab. 17:159–65.

10. Tarrant TK, et al. (2008) Granulocyte chemotaxisand disease expression are differentially regulatedby GRK subtype in an acute inflammatory arthri-tis model (K/BxN). Clin. Immunol. 129:115–22.

11. Eijkelkamp N, et al. (2007) G protein-coupled re-ceptor kinase 6 controls chronicity and severityof dextran sodium sulphate-induced colitis inmice. Gut. 56:847–54.

12. Eijkelkamp N, et al. (2009) G protein-coupled re-ceptor kinase 6 controls post-inflammatory vis-ceral hyperalgesia. Brain Behav. Immun. 23:18–26.

13. Vroon A, et al. (2004) GRK6 deficiency is associ-ated with enhanced CXCR4-mediated neutrophilchemotaxis in vitro and impaired responsivenessto G-CSF in vivo. J. Leukoc. Biol. 75:698–704.

14. Kavelaars A, et al. (2003) Increased acute inflam-mation, leukotriene B4-induced chemotaxis, andsignaling in mice deficient for G protein-coupledreceptor kinase 6. J. Immunol. 171:6128–34.

15. Fong AM, et al. (2002) Defective lymphocytechemotaxis in beta-arrestin2- and GRK6-deficientmice. Proc. Natl. Acad. Sci. U. S. A. 99:7478–83.

16. Aiyar N, et al. (2000) Involvement of G protein-coupled receptor kinase-6 in desensitization ofCGRP receptors. Eur. J. Pharmacol. 403:1–7.

17. Ruiz-Gomez A, et al. (2007) G protein-coupled re-ceptor kinase 2-mediated phosphorylation ofdownstream regulatory element antagonist mod-ulator regulates membrane trafficking of Kv4.2potassium channel. J. Biol. Chem. 282:1205–15.

18. Hall RA, et al. (1999) G protein-coupled receptorkinase 6A phosphorylates the Na(+)/H(+) ex-changer regulatory factor via a PDZ domain- mediated interaction. J. Biol. Chem. 274:24328–34.

19. Fukuoka H, Kawatani M, Hisamitsu T, TakeshigeC. (1994) Cutaneous hyperalgesia induced by pe-ripheral injection of interleukin-1 beta in the rat.Brain Res. 657:133–40.

20. Hargreaves K, Dubner R, Brown F, Flores C, JorisJ. (1988) A new and sensitive method for measur-ing thermal nociception in cutaneous hyperalge-sia. Pain. 32:77–88.

21. Chaplan SR, Bach FW, Pogrel JW, Chung JM,Yaksh TL. (1994) Quantitative assessment of tac-tile allodynia in the rat paw. J. Neurosci. Methods.53:55–63.

22. Woolf CJ, Allchorne A, Safieh-Garabedian B,Poole S. (1997) Cytokines, nerve growth factorand inflammatory hyperalgesia: the contributionof tumour necrosis factor alpha. Br. J. Pharmacol.121:417–24.

23. Wajant H, Pfizenmaier K, Scheurich P. (2003)Tumor necrosis factor signaling. Cell Death Differ.10:45–65.

24. Eijkelkamp N, et al. (2010) Low nociceptor GRK2prolongs prostaglandin E2 hyperalgesia via bi-ased cAMP signaling to Epac/Rap1, protein ki-nase Cepsilon, and MEK/ERK. J. Neurosci.30:12806–15.

25. Aley KO, Levine JD. (1999) Role of protein kinaseA in the maintenance of inflammatory pain.J. Neurosci. 19:2181–6.

26. Constantin CE, et al. (2008) Endogenous tumornecrosis factor alpha (TNFalpha) requires TNFreceptor type 2 to generate heat hyperalgesia in amouse cancer model. J. Neurosci. 28:5072–81.

27. Jin X, Gereau RW. (2006) Acute p38-mediatedmodulation of tetrodotoxin-resistant sodiumchannels in mouse sensory neurons by tumornecrosis factor-alpha. J. Neurosci. 26:246–55.

28. Obreja O, Rathee PK, Lips KS, Distler C, KressM. (2002) IL-1 beta potentiates heat-activatedcurrents in rat sensory neurons: involvement ofIL-1RI, tyrosine kinase, and protein kinase C.FASEB J. 16:1497–503.

29. Abrahamsen B, et al. (2008) The cell and molecu-lar basis of mechanical, cold, and inflammatorypain. Science. 321:702–5.

30. Howlett E, Lin CC, Lavery W, Stern M. (2008) API3-kinase-mediated negative feedback regulatesneuronal excitability. PLoS. Genet. 4:e1000277.

31. Zhuang ZY, Xu H, Clapham DE, Ji RR. (2004)Phosphatidylinositol 3-kinase activates ERK inprimary sensory neurons and mediates inflam-matory heat hyperalgesia through TRPV1 sensiti-zation. J. Neurosci. 24:8300–9.

32. Cunha TM, et al. (2010) Morphine peripheralanalgesia depends on activation of thePI3Kgamma/AKT/nNOS/NO/KATP signalingpathway. Proc. Natl. Acad. Sci. U. S. A. 107:4442–7.

33. Qian Y, et al. (2009) Regulation of TLR4-inducedIL-6 response in bladder cancer cells by opposingactions of MAPK and PI3K signaling. J. CancerRes. Clin. Oncol. 135:379–86.

34. Diem R, Meyer R, Weishaupt JH, Bahr M. (2001)Reduction of potassium currents and phos-phatidylinositol 3-kinase-dependent AKT phos-phorylation by tumor necrosis factor-(alpha) res-cues axotomized retinal ganglion cells fromretrograde cell death in vivo. J. Neurosci.21:2058–66.

35. Redman PT, et al. (2007) Apoptotic surge ofpotassium currents is mediated by p38 phospho-rylation of Kv2.1. Proc. Natl. Acad. Sci. U. S. A.104:3568–73.

36. Marmiroli S, et al. (1998) Phosphatidylinositol3-kinase is recruited to a specific site in the acti-vated IL-1 receptor I. FEBS Lett. 438:49–54.

37. Naga Prasad SV, et al. (2002) Phosphoinositide3-kinase regulates beta2-adrenergic receptor endo-cytosis by AP-2 recruitment to the receptor/beta-arrestin complex. J. Cell Biol. 158:563–75.

38. Peregrin S, et al. (2006) Phosphorylation of p38by GRK2 at the docking groove unveils a novel

mechanism for inactivating p38MAPK. Curr. Biol.16:2042–7.

39. Willets JM, Challiss RA, Nahorski SR. (2003)Non-visual GRKs: are we seeing the whole pic-ture? Trends Pharmacol. Sci. 24:626–33.

40. Eijkelkamp N, et al. (2010) GRK2: a novel cell-specific regulator of severity and duration of in-flammatory pain. J. Neurosci. 30:2138–49.

41. Willemen HL, et al. (2010) Microglial/ macro -phage GRK2 determines duration of peripheralIL-1beta-induced hyperalgesia: contribution ofspinal cord CX3CR1, p38 and IL-1 signaling.Pain. 150:550–60.

42. Gurevich EV, Tesmer JJ, Mushegian A, GurevichVV. (2012) G protein-coupled receptor kinases:more than just kinases and not only for GPCRs.Pharmacol. Ther. 133:40–69.

43. Lefkowitz RJ. (1998) G protein-coupled receptors.III. New roles for receptor kinases and beta- arrestins in receptor signaling and desensitiza-tion. J. Biol. Chem. 273:18677–80.

44. Lombardi MS, et al. (1999) Decreased expressionand activity of G-protein-coupled receptor ki-nases in peripheral blood mononuclear cells ofpatients with rheumatoid arthritis. FASEB J.13:715–25.

G R K 6 R E G U L A T E S C Y T O K I N E - I N D U C E D H Y P E R A L G E S I A

5 6 4 | E I J K E L K A M P E T A L . | M O L M E D 1 8 : 5 5 6 - 5 6 4 , 2 0 1 2