Ò

PAIN 155 (2014) 1802–1813

www.elsevier.com/locate/pain

Spinal HMGB1 induces TLR4-mediated long-lasting hypersensitivity and glial activation and regulates pain-like behavior in experimental arthritis Nilesh M. Agalave a, Max Larsson a,b,1, Sally Abdelmoaty a, Jie Su a, Azar Baharpoor a, Peter Lundbäck c, Karin Palmblad d, Ulf Andersson d, Helena Harris c, Camilla I. Svensson a,⇑ a

Department of Physiology and Pharmacology, Karolinska Institutet, Stockholm, Sweden Department of Neuroscience, Karolinska Institutet, Stockholm, Sweden Department of Medicine, Karolinska Institutet, Stockholm, Sweden d Women’s and Children’s Health, Karolinska Institutet, Stockholm, Sweden b c

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

a r t i c l e

i n f o

Article history: Received 12 April 2014 Received in revised form 11 June 2014 Accepted 13 June 2014

Keywords: HMGB1 TLR4 Cytokine Rheumatoid arthritis Inflammation Pain Hypersensitivity Spinal cord

a b s t r a c t Extracellular high mobility group box-1 protein (HMGB1) plays important roles in the pathogenesis of nerve injury- and cancer-induced pain. However, the involvement of spinal HMGB1 in arthritis-induced pain has not been examined previously and is the focus of this study. Immunohistochemistry showed that HMGB1 is expressed in neurons and glial cells in the spinal cord. Subsequent to induction of collagen antibody-induced arthritis (CAIA), Hmgb1 mRNA and extranuclear protein levels were significantly increased in the lumbar spinal cord. Intrathecal (i.t.) injection of a neutralizing anti-HMGB1 monoclonal antibody or recombinant HMGB1 box A peptide (Abox), which each prevent extracellular HMGB1 activities, reversed CAIA-induced mechanical hypersensitivity. This occurred during ongoing joint inflammation as well as during the postinflammatory phase, indicating that spinal HMGB1 has an important function in nociception persisting beyond episodes of joint inflammation. Importantly, only HMGB1 in its partially oxidized isoform (disulfide HMGB1), which activates toll-like receptor 4 (TLR4), but not in its fully reduced or fully oxidized isoforms, evoked mechanical hypersensitivity upon i.t. injection. Interestingly, although both male and female mice developed mechanical hypersensitivity in response to i.t. HMGB1, female mice recovered faster. Furthermore, the pro-nociceptive effect of i.t. injection of HMGB1 persisted in Tlr2- and Rage-, but was absent in Tlr4-deficient mice. The same pattern was observed for HMGB1-induced spinal microglia and astrocyte activation and cytokine induction. These results demonstrate that spinal HMGB1 contributes to nociceptive signal transmission via activation of TLR4 and point to disulfide HMGB1 inhibition as a potential therapeutic strategy in treatment of chronic inflammatory pain. Ó 2014 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction Extracellular high mobility group box-1 protein (HMGB1), originally identified as a nuclear chromatin-binding protein, has received attention for its emerging roles in the extracellular environment, acting as a prototypical damage-associated molecular pattern (DAMP) molecule [5]. In response to tissue damage and inflammation, nuclear HMGB1 translocates to the cytoplasm [1]. While necrotic cells release HMGB1 passively [32], immune and ⇑ Corresponding author. Address: Karolinska Institutet, Von Eulers väg 8, 171 77 Stockholm, Sweden. Tel.: +46 8 524 87948; fax: +46 8 310622. E-mail address: [email protected] (C.I. Svensson). 1 Current address: Department of Clinical and Experimental Medicine, Division of Cell Biology, Linköping University, Linköping, Sweden.

other cells, including neurons, undergoing severe stress actively secrete HMGB1 via nonclassical pathways [15,27,49]. Extracellular HMGB1, alone or in complex with other factors, acts as a chemoattractant or proinflammatory mediator, stimulating immune cells to release cytokines such as tumor necrosis factor (TNF), interleukin (IL)-1, and IL-6 [2,18,43]. HMGB1 binds to multiple receptors, including toll-like receptor (TLR)2, TLR4, and receptor for advanced glycation end-products (RAGE) [16]. Recent studies demonstrate that the redox state of HMGB1 is key in determining which receptors HMGB1 activates and thus also the functional consequences of HMGB1 release. When the cysteine in position C106 is in the reduced thiol form and C23 and C45 are engaged in a disulfide bridge (HMGB1C23–C45C106h, disulfide HMGB1), it functions as a cytokine-inducing TLR4 ligand

http://dx.doi.org/10.1016/j.pain.2014.06.007 0304-3959/Ó 2014 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

[45,47]. In contrast, when these 3 cysteines are all reduced (HMGB1C23hC45hC106h, all-thiol HMGB1), HMGB1 potentiates chemotactic activity via CXCR4 by forming a heterocomplex with CXCL12 chemokine [43]. Thus, HMGB1 orchestrates both key events in sterile inflammation, leukocyte recruitment, and their induction to secrete inflammatory cytokines, by adopting mutually exclusive redox states. HMGB1 terminally oxidized to contain sulfonyl groups on all cysteines (HMGB1C23soC45soC106so, sulfonyl HMGB1) has not yet been associated with an in vivo function. Peripheral HMGB1 has been linked to nociception in experimental models of neuropathic [13,34] and low back [29] pain and spinal HMGB1 to hypersensitivity induced by bone cancer [40] and diabetes [31]. However, the role of spinal HMGB1 in arthritis-associated pain has not been examined, nor have studies examined which redox form of HMGB1 drives nociception. Although joint pain is a recognized problem during active rheumatoid arthritis (RA), emerging reports show that a considerable number of patients continue to experience pain even when there is little or no inflammatory disease activity [24,39]. Most studies of RA-induced pain are focused on active inflammatory disease stages, and thus less is known about the mechanisms for residual, postinflammatory pain. We have recently identified 2 animal models of RA, namely, collagen antibody-induced arthritis (CAIA) [4] and K/BxN serumtransferred arthritis [9], in which nociceptive behavior persists for several weeks after the joint inflammation resolves. TLR4 deficiency and i.t. injection of TLR4 antagonists reduced antibody-induced ‘‘post-inflammatory’’ pain-like behavior [11]. As one isoform of HMGB1 serves as an endogenous TLR4 ligand, the aim of the current study was to investigate whether spinal HMGB1 may regulate arthritis-induced pain-like behavior during ongoing joint inflammation and in phases with minor inflammatory activity, and, if so, which redox isoform of HMGB1 is nociceptive and through which receptor HMGB1 exerts its nociceptive action.

1803

highly conserved DNA binding domains of the HMGB1 protein (amino acid 1–89) and has been shown to block extracellular HMGB1 activities, although the exact mechanisms of action are unknown [23,46]. Full-length endotoxin-free recombinant rHMGB1C23–C45C106h (disulfide HMGB1), provided by Dr. K. J. Tracey (Feinstein Institute for Medical Research, Manhasset, NY), was produced in Escherichia coli, purified as previously described [25,42], and injected i.t. (0.1–1 lg). In some experiments, the disulfide bridge between C23 and C45 was readily reduced by exposure to dithiothreitol (DTT, 5 mM) for 1 hour to generate fully reduced HMGB1C23hC45hC106h (all-thiol HMGB1) and exposure to H2O2 (20 mmol/L) for 2 hours to generate fully oxidized HMGB1C23soC45soC106so (sulfonyl HMGB1). Excess DTT and H2O2 were removed by dialysis for 2 hours and overnight, respectively. The i.t. injections were performed as described previously. Briefly, the mice were anesthetized with isoflurane (4%) and placed in the prone position, and the injection needle was inserted between the lumbar vertebrae L5 and L6. The i.t. location of the needle was verified by the tail flick reflex. The control groups were injected with saline solution or an isotype-matched mouse monoclonal antibody against a protein that is not expressed in mice (plum pox virus; ThermoFisher). Saline solution, antibodies, and peptides were injected in a volume of 5 lL i.t. 2.3. Collagen antibody-induced arthritis

2. Methods

Collagen antibody-induced arthritis was established as previously described [4] by intravenous (i.v.) injection of anti-collagen type II (CII) arthritogenic antibody cocktail (1.25 mg per mouse; Chondrex, Redmond, WA) containing 5 monoclonal CII antibodies on day 0. Lipopolysaccharide (LPS; 25 lg per mouse) was injected intraperitoneally (i.p.) on day 5 to synchronize the onset of joint inflammation. Two control groups were included in this study; 1 group received saline solution i.v. on day 0 followed by saline solution i.p. on day 5 (saline control), and the other group received saline solution i.v. on day 0 and LPS i.p. on day 5 (LPS control).

2.1. Animals

2.4. Arthritis score and joint histology

All experiments were carried out in accordance with protocols approved by the local ethics committee for animal experiments in Sweden (Stockholm North Animal Ethics Board). Rage / mice were provided by Bernd Arnold and Angelika Bierhaus (German Cancer Research Center, Heidelberg, Germany) and Tlr2 / and Tlr4 / mice by S. Akira (Osaka University, Osaka, Japan). All knockout mice were on a C57BL/6 background. Wild-type (WT) C57BL/6 male and female mice (10–12 weeks of age, 20–25 g) and BALB/c male and female mice (10–12 weeks of age, 20–25 g) were purchased from Nova SCB, Sweden. Animals were kept and bred at the Karolinska Institutet animal facility in a specific pathogen-free environment with standard temperature and a 12-hour light/dark cycle. Mice were housed in standard cages with 4 to 5 animals per cage, and food and water were provided ad libitum. All behavioral experiments and associated data analysis were carried out during the light period.

The degree of joint inflammation was assessed by visual inspection of the fore and hind paws. The arthritis scores were given based on swelling and redness of toes, knuckles, and ankle, with 1 point given to each inflamed toe or knuckle and 5 points for an inflamed ankle or wrist. Thus the maximum score for each limb is 15 points, generating a total maximum score of 60 points per mouse. Ankle joints were harvested after sacrifice, fixed in 4% paraformaldehyde and decalcified for 3 to 4 weeks in decalcification solution (100 g ethylenediaminetetraacetate (EDTA), 75 g polyvinylpyrolidone, 12.11 g Tris in 1 L of deionized water adjusted to pH 7.0 with Potassium hydroxide (KOH); Sigma-Aldrich, St Louis, MO). The decalcified joints were dehydrated using 70% ethanol and xylene and embedded in paraffin. Joint sections (5-lm) were mounted on glass slides and subjected to hematoxylin and eosin staining.

2.2. Drugs and drug delivery

The mice were habituated to the test environment on 2 occasions before assessment of baseline. After 3 baseline recordings performed on different days, the animals were randomly assigned to the different groups. Mechanical hypersensitivity was determined by assessment of paw withdrawal thresholds in response to application of von Frey optiHair filament (Marstock OptiHair) using the up–down method [7]. A series of filaments with a logarithmically incremental stiffness of 0.5, 1, 2, 4, 8, 16, and 32 mN (converted to 0.051 g, 0.102 g, 0.204 g, 0.408 g, 0.815 g, 1.63 g, and 3.26 g, respectively) were applied to the plantar surface of

To block the action of spinal endogenous HMGB1, the mouse anti-HMGB1 antibody (2G7) (7.25 lg and 15 lg) or the recombinant HMGB1 box A peptide (Abox) (20 lg) was injected i.t. 2G7, an anti-HMGB1 IgG2b noncommercial monoclonal antibody (mAb) [8,14,30,33] (developed at the former Critical Therapeutics, Boston, MA; now Cornerstone Therapeutics, Cary, NC) binds to an epitope within the amino acid region position 53 to 63 of the A box unit. Recombinant A box protein corresponds to 1 of the 2

2.5. Behavioral tests

1804

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

the hind paw and held for 2 to 3 seconds. To avoid tissue damage, a cut-off of 4 g was applied. A brisk withdrawal of the paw was noted as a positive response. The withdrawal thresholds of both paws were measured and averaged. The 50% probability withdrawal threshold (ie, the force of the von Frey filament to which an animal reacts to 50% of the presentations) was calculated and expressed as threshold in grams. In the CAIA study, mechanical hypersensitivity was measured every third day or more frequently during the experiment period. Pharmacological experiments were performed during days 12 to 18 and days 42 to 49. The baselines for pharmacological experiments were assessed the day before injection of drugs. In the i.t. HMGB1 study, mechanical hypersensitivity was assessed 6 hours and day 1, 2, 3, 4, 5, and 7 after i.t. injection. In addition to presenting the results as 50% withdrawal threshold, the data were also presented as a hyperalgesic index, a calculation that defines the magnitude of i.t. HMGB1-induced sensitization. It represents the area (based on withdrawal threshold in grams and time in hours) between the extrapolated baseline and the time– response curve after HMGB1 injection. Increasing values indicate increasing hypersensitivity. The experimenter was blinded to groups and treatments during the behavioral tests and data analysis. 2.6. Western blot Animals were deeply anesthetized, and spinal cords were harvested by hydroextrusion. The tissues were immediately flash-frozen and stored at 70 °C until analysis. Nuclear and cytoplasmic (extranuclear) proteins were extracted from lumbar spinal cords by the NE-PER nuclear and cytoplasmic kit (Thermo Fisher Scientific, Lafayette, CO) according to the manufacturer’s instructions. Proteins were separated by gel electrophoresis (NuPAGE 4–12% Bis Tris and MES running buffer, Invitrogen) and transferred to nitrocellulose membrane (Invitrogen). Nonspecific binding sites were blocked with 5% low-fat milk in Tris-based buffer (50 mmol/L Tris–HCl and 6 mmol/L NaCl with 0.1% Tween 20). Membranes were incubated with primary antibody overnight (48 hours for 2G7), followed by secondary antibody conjugated to horseradish peroxidase (1:7500, Cell Signaling Technology, Danvers, MA). Chemiluminescent reagent (Supersignal, Pierce, Rockford, IL) was used to visualize protein antibody complexes, and signal intensity was measured using Quantity One software (Bio-Rad, Hercules, CA). The antibodies were removed with stripping solution (Millipore, Billerica, MA), and the membranes were reprobed with different antibodies. Positive bands were normalized to their respective GAPDH band. The results were expressed as percentage of control. Primary antibodies used for this part of the study were 2G7 mouse anti-HMGB1 (1:1000; 1.2 lg), mouse anti-TATA binding protein (1:2000, catalog no. 051531, Millipore, Billerica, MA), rabbit anti-histone (1:5000, catalog no. ab1791, Abcam, Cambridge, UK), and mouse anti-GAPDH (1:10,000, catalog no. ab8245, Abcam).

anti-NeuN conjugated to Alexa Fluor 488, 1:100, Millipore, catalog no. MAB377X), astrocytes (mouse anti-Glial fibrillary acidic protein (GFAP), Millipore, MAB360; 1:1000), or microglia (goat anti-Iba-1, 1:2000, Abcam, catalog no. ab107159) overnight. For secondary detection, goat or donkey secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 594 (1:250; Life Technologies, Carlsbad, CA) were used. Immunolabeled sections were coverslipped with Prolong Gold antifade medium with DAPI (Life Technologies) and examined in a Zeiss LSM710 confocal microscope. Micrographs of single optical sections were obtained using 20/0.8 dry and 40/1.20 water immersion objective. 2.8. Quantitative real-time polymerase chain reaction Flash-frozen spinal cords were homogenized by sonication in Trizol (Life Technologies), and mRNA was extracted according to the manufacturer’s protocol. The cDNA obtained by reverse transcription was subjected to real-time quantitative polymerase chain reaction (PCR; StepOne system, Applied Biosystem, Foster City, CA) according to the manufacturer’s protocol, using a hydrolysis probe to measure the relative mRNA levels. Pre-developed specific primer/probe sets for Ccl2 (Mm00441242_m1), Cd11b (Mm00434 455_m1), Gfap (Mm00546086_m1), Hmgb1 (Mm00849805_gH), IL-1b (Mm00434228_m1), Tnf (Mm00443258_m1), Cxcl1 (Mm042 07460_m1), Cxcl2 (Mm00436450_m1), Il6 (Mm00446190_m1), Il33 (Mm00505403_m1), and hypoxanthine guanine phosphoribosyltransferase1 (Hprt1) (Mm01545399_m1) were used for mRNA analysis (all from Applied Biosystems). Threshold cycle values in each sample were used to calculate the number of cell equivalents in the test samples using the standard curve method [6]. The data were normalized to Hprt1 mRNA levels and expressed as relative expression units (REU). 2.9. Statistical analysis Differences in mRNA and protein levels were assessed by 1-way analysis of variance (ANOVA) followed by the Bonferroni post hoc test. Arthritis scores were compared using the Kruskal–Wallis test, followed by Dunn’s multiple comparison post hoc tests. For comparison of CAIA-induced changes in mechanical hypersensitivity and analysis of the effect of drug treatment on mechanical hypersensitivity compared to the CAIA group that received i.t. saline solution, a 2-way repeated-measures ANOVA was used, followed by Bonferroni post hoc test for correction of multiple comparisons. Differences in hyperalgesic index were assessed by 1-way ANOVA, followed by Bonferroni post hoc tests. All data are expressed as mean ± standard error of the mean (SEM), and P values less than .05 were considered statistically significant. Statistical analysis was performed using GraphPad Prism software (GraphPad Prism 6.0, San Diego, CA).

3. Results 2.7. Immunohistochemistry Mice were deeply anesthetized with isoflurane and perfused intracardially with saline solution followed by 4% paraformaldehyde. The lumbar spinal cords (L1–L6) were dissected, postfixed, and cryoprotected in 20% sucrose. Transverse lumbar spinal cord sections (10-lm) were cut on a cryostat and mounted on glass slides. The sections were pre-incubated with 5% normal goat or donkey serum in 0.2% Triton X-100 in phosphate-buffered saline solution to block nonspecific bindings. Subsequently, the sections were incubated with rabbit anti-HMGB1 (1:500, Abcam, catalog no. ab18256), together with markers for neurons (either mouse anti-NeuN, 1:1000, Millipore, catalog no. MAB377 or mouse

3.1. Intravenous injection of collagen type II antibodies induces transient inflammation and persistent mechanical hypersensitivity and glial activation The CAIA mouse model was used to investigate the role of HMGB1 in arthritis-induced pain. Polyarthritis was induced by intravenous injection of a collagen type II antibody cocktail (1.25 mg per mouse), followed by an intraperitoneal LPS injection to boost and synchronize the inflammatory reaction. In agreement with our previous studies, we found that, although the CII antibody-induced arthritis scores were transiently elevated (days 9–24) (n = 6, P < .01) (Fig. 1A), the mechanical hypersensitivity

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

persisted throughout the study (days 6–48) (n = 6, P < .01) (Fig. 1B). Assessment of the pathology in ankle joint sections stained with hematoxylin and eosin showed pronounced cellular infiltrate and bone degradation and mild cartilage serration in the inflammatory phase (day 15) as compared to those in age-matched saline solution- and LPS-injected controls (Fig. 1C). In contrast, in the late phase (day 48), the synovial inflammation had resolved, and only minor residual bone erosion was observed in ankle joints from CAIA mice (Fig. 1C). No signs of joint pathology were detected in ankles from saline solution-injected and LPS-injected control mice (Fig. 1C). Contribution of activated glial cells to arthritis-induced hypersensitivity has been implicated in the CAIA [4] and collageninduced arthritis (CIA) models [20] using QB, CBA, and DBA/1 mice. Glial reactivity in BALB/c mice following CAIA has not been examined previously. Thus mRNA levels for Gfap (astrocytes marker) and Cd11b (microglia marker) were measured in the CAIA and control groups. While there were no differences in Gfap (1.0 ± 0.1 vs 0.4 ± 0.1 REU, n = 5, P = .08) and Cd11b (1.4 ± 0.3 vs 0.7 ± 0.1 REU, n = 5, P = .1) mRNA levels in spinal cords from the CAIA group

1805

compared to controls during the inflammatory phase, Gfap (2.2 ± 0.2 vs 0.9 ± 0.5 REU, n = 4, P = .0004) and Cd11b (2.3 ± 0.4 vs 0.9 ± 0.3 REU, n = 4, P = .007) mRNA were elevated in the CAIA group in that late phase as compared to those in saline solutioninjected controls (Fig. 1D). 3.2. Blocking the action of spinal HMGB1 reverses CAIA-induced mechanical hypersensitivity during the inflammatory and the late phase in male and female mice The role of spinal HMGB1 in established arthritis-induced mechanical hypersensitivity was examined by injecting a neutralizing anti-HMGB1 mAb (2G7) (7.5 and 15 lg) or Abox (20 lg) i.t. on days 12 and 15. The i.t. injection of 2G7 reversed CAIA-induced mechanical hypersensitivity 3 and 6 hours after injection when injected on day 12 in male and female mice, whereas i.t. injection of the vehicle was without effect (Fig. 2A, C). Twenty-four hours after the injection, withdrawal thresholds had returned to preinjection levels (Fig. 2A, C). A second injection on day 15 reversed the mechanical hypersensitivity in a fashion similar to that

Fig. 1. Collagen antibody-induced arthritis (CAIA) induces transient joint inflammation and long-lasting mechanical hypersensitivity and glial activation in the spinal cord. (A) Joint inflammation was assessed by visual scoring (arthritis score 0–60) and (B) mechanical hypersensitivity by von Frey filaments after injection of CII antibody cocktail (1.25 mg) i.v. day 0 and lipopolysaccharide (LPS) (25 lg) i.p. day 5. (C) Ankle joints examined by hematoxylin and eosin staining show infiltration of inflammatory cells (white star), cartilage serration (black arrow), and bone erosion (arrowhead). (D) Gfap and Cd11b mRNA levels in the lumbar spinal cord were measured by quantitative real-time polymerase chain reaction in tissues harvested in the inflammatory phase (day 15) and late phase (day 48). The mRNA data were normalized to Hprt1 mRNA levels and are presented as relative expression units (REU). Data are mean ± standard error of the mean, n = 6 mice per group for arthritis scores and behavioral data and n = 4 or 5 for mRNA data. ⁄⁄P < .01 and ⁄⁄⁄ P < .001 versus the saline solution-injected group.

1806

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

Fig. 2. Intrathecal (i.t.) injection of 2G7 and Abox reverse collagen antibody-induced arthritis (CAIA)-induced mechanical hypersensitivity during the inflammatory and the late phase. Mechanical hypersensitivity was assessed in male (A, B) and female (C, D) mice subjected to CAIA subsequent to i.t. injection of 2G7 (7.5 lg and 15 lg) and Abox (20 lg) days 12 and 15 (A, C) and days 45 and 48 (B, D). Control mice receiving i.p. injection of lipopolysaccharide (LPS) day 5 did not alter mechanical thresholds in male (Fig. 1A, B) or female mice (C, D). Data are presented as mean ± SEM, n = 6 to 8 mice/group; ⁄,°,#P < .05, where ⁄ represents CAIA + 2G7 (7.5 lg), ° CAIA + 2G7 (15 lg), and # CAIA + Abox versus CAIA + saline solution.

observed after the first injection (Fig. 2A, C). The i.t. injection of 2G7 in the late phase, on days 45 and 48, also reversed CAIAinduced mechanical hypersensitivity 3 and 6 hours after injection to male and female mice, compared to the CAIA group that received i.t. vehicle (Fig. 2B, D). The effect of Abox was examined only in male mice and i.t. Abox reversed mechanical hypersensitivity both when injected during the inflammatory phase (on days 12 and 15) and in the late phase (on days 45 and 48) (Fig. 2A, C). A control monoclonal antibody (7.5 lg) was used as a negative control in the male mice in this study and did not reverse CAIA-induced mechanical hypersensitivity in either of the phases (Fig. 2A, B). 3.3. HMGB1 is expressed in neurons, microglia, and astrocytes in the lumbar spinal cord in naive mice Double immunolabeling of HMGB1 and NeuN positive neurons showed varying degrees of HMGB1 immunolabeling in all neurons in the spinal cord (Fig. 3A, B). Notably, not all HMGB1 immunolabeled cells were NeuN positive. Moreover, some DAPI-stained cellular nuclei lacked detectable immunolabeling for HMGB1, and these nuclei were also NeuN immunonegative. Microglial expression of HMGB1 in the dorsal horn was examined using double immunolabeling of HMGB1 and the microglia marker Iba-1 (Fig. 3C, D). Intriguingly, 2 categories of microglia were discerned with respect to HMGB1 and Iba-1 immunolabeling. In some microglia, Iba-1 immunolabeling extended into their nuclei, as judged by DAPI staining. These microglia also showed considerable

nuclear HMGB1 immunolabeling. By contrast, other microglia were essentially devoid of Iba-1 immunolabeling within their nucleus, and these also exhibited undetectable or very weak HMGB1 immunoreactivity. In spinal cord sections co-immunolabeled for HMGB1 and GFAP, we identified astrocytic cell bodies based on their outlining by GFAP immunolabeling. Astrocytes thus identified in the dorsal horn showed varying levels of HMGB1 immunoreactivity within their nuclei (Fig. 3E, F). 3.4. Hmgb1 mRNA levels and extranuclear protein are elevated in the spinal cord after antibody-induced arthritis Spinal cords collected on days 15 and 48 and analysed by qPCR showed that Hmgb1 mRNA levels were elevated both in the inflammatory (1.9 ± 0.3 vs 0.9 ± 0.1 n = 5, P = .05) (Fig. 4A) and late phase (1.7 ± 0.1 vs 0.9 ± 0.1 REU, n = 4, P = .001) (Fig. 4B) in the CAIA group as compared to the saline group. Total HMGB1 protein levels in homogenates of the lumbar spinal cord were not changed in the CAIA compared to the control (saline) group in the inflammatory or late phase (Fig. 4C, D). In contrast to whole homogenate, the extranuclear level of HMGB1 in the lumbar spinal cord was elevated in the CAIA (165 ± 15%, n = 6) compared to the saline (100 ± 13%, n = 6, P = .02) and the LPS (87 ± 13%, n = 7, P = .003) control groups in the inflammatory phase (Fig. 4E). No change in spinal extranuclear HMGB1 protein level was observed in the CAIA group in the late phase (88 ± 14%, n = 8) compared to the saline group (100 ± 14%, n = 8, P = 0.9) (Fig. 4F). The absence of histones in the extranuclear fractionation confirmed that the nuclear compartment

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

1807

Fig. 3. Immunohistochemistry of HMGB1 expression in neurons, microglia, and astrocytes in the lumbar spinal cord dorsal horn. (A, B) Arrowheads indicate examples of DAPI-stained, NeuN-immunopositive neuronal nuclei that also show HMGB1 immunofluorescence. Arrows indicate examples of HMGB1-immunopositive, NeuNimmunonegative nuclei. Dashed arrow points to a rare DAPI-stained cell that lacked immunoreactivity for HMGB1 and NeuN. Dashed frames in the upper panels indicate the position of the field shown in the lower panels. (C, D) Arrow indicates a microglial cell that exhibits substantial nuclear immunostaining of both HMGB1 and Iba-1. Arrowhead indicates a microglia showing weak nuclear Iba-1 immunolabeling and very weak HMGB1 immunofluorescence. Dashed frames in the upper panels indicate the position of the field of view shown in the lower panels. (E, F) HMGB1 immunofluorescence in astrocytes in the lumbar spinal cord dorsal horn. Arrowheads indicate examples of DAPIstained, GFAP-immunopositive neuronal nuclei that also show HMGB1 immunofluorescence. Dashed frames in the upper panels indicate the position of the field shown in the lower panels. Scale bars are 50 lm (upper panels) and 10 lm (lower panels). Micrographs are single optical sections of the medial lumbar dorsal horn obtained with 20 (upper panels) or 40 (lower panels) objectives.

1808

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

Fig. 4. Spinal Hmgb1 mRNA and protein levels are elevated in the spinal cord after induction of collagen antibody-induced arthritis (CAIA). (A) Bar graphs display Hmgb1 mRNA levels in the lumbar spinal cord normalized to Hprt1 and expressed in relative expression units (REU) in the (A) inflammatory and (B) late phase of the CAIA model. (C) Bar graph showing HMGB1 protein levels in whole homogenate from lumbar spinal cord and (D) representative western blot film (S, saline; C, CAIA). (E, F) Bar graphs and representative WB films showing HMGB1 protein levels in the cytoplasmic fraction of the lumbar spinal cord from CAIA, lipopolysaccharide (LPS)-injected control and saline solution-injected control mice in the (E) inflammatory and (F) late phase of the experiment. The HMGB1 signal was normalized against GAPDH and the data expressed as percentage of the saline control group. Data are presented as mean ± standard error of the mean, n = 4 to 8 mice per group. ⁄P < .05 and ⁄⁄P < .01 versus the saline group. Representative western blot images are displayed below the bar graphs. Positive controls for HMGB1 are indicated by + and ++, representing spinal cord homogenate and cell lysate of THP-1 cells stimulated with LPS, respectively.

was removed from the extranuclear fraction (Fig. 4E and F). Two positive controls for HMGB1; spinal cord whole homogenate and cell lysate of THP-1 cells stimulated with 100 ng/mL LPS [48], were included in the western blot analysis and showed a band migrating at 28 kDa (Fig. 4E and F).

3.5. Intrathecal injection of disulfide HMGB1, but not all-thiol HMGB1 and sulfonyl HMGB1, induces mechanical hypersensitivity To further examine whether spinal HMGB1 is coupled to mechanical hypersensitivity, we injected rHMGB1 (0.1–1 lg per

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

mouse) i.t. and assessed tactile thresholds for 7 days. Spinal injection of disulfide HMGB1, but not vehicle (saline solution), evoked hypersensitivity in naive male C57BL/6 (Fig. 5A) and BALB/c (Fig. 5B) mice, which lasted for 4 to 5 days. In contrast, all-thiol HMGB1 (1 lg) and all-oxidized HMGB1 did not evoke mechanical hypersensitivity after i.t. injection (Fig. 5E). The i.t. injection of disulfide HMGB1 to female C57BL/6 (Fig. 5C) and BALB/c (Fig. 5D) also evoked mechanical hypersensitivity. Calculation of the hyperalgesic index over 0 to 3 days showed a significant increase in pain-like behavior in the male and female C57BL/6 and BALB/c mice receiving 1 lg i.t. disulfide HMGB1 compared to their respective controls (Fig. 5F). Comparing the withdrawal thresholds between male and female mice that received i.t. HMGB1 (1 lg) showed that the mechanical hypersensitivity was more pronounced in male C57BL/6 mice at the 4-day time point (male 1.4 ± 0.2 g vs female 2.6 ± 0.3 g, P = .005, n = 6–12) and in male BALB/c mice at the 3 and 4 day time points (day 3: 1.0 ± 0.3 g vs female 2.2 ± 0.2 g, P = .03, P = .001; day 4: male 0.9 ± 0.1 g vs female 2.4 ± 0.2 g, P < .001, n = 6–12). 3.6. Intrathecal injection of disulfide HMGB1 induces mechanical hyper sensitivity in WT, Tlr2 / and Rage / mice, but not in Tlr4 / mice To examine whether TLR2, TLR4, and/or RAGE receptors mediate HMGB1-induced mechanical hypersensitivity, disulfide HMGB1 (1 lg) was injected into male mice deficient in these receptors. Although a robust mechanical hypersensitivity was

1809

observed in Tlr2 / and Rage / mice subsequent to i.t. injection of disulfide HMGB1, pain-like behavior was absent in Tlr4 / mice (Fig. 6C–E). The related hyperalgesic indices were significantly increased in the Tlr2 / (141.4 ± 15.14 vs 18.04. ± 24.53, n = 5, P = .001) and Rage (88.99 ± 18.46 vs 16.68 ± 14.81, n = 9, P = .01) but not Tlr4 / (P > .05) mice receiving 1 ug i.t. disulfide HMGB1 compared to their respective i.t. saline controls. 3.7. Intrathecal injection of dsHMGB1, but not all-thiol HMGB1, induces mRNA of glia-associated factors TLR4 activation is frequently associated with glia activation and synthesis of cytokines [26]. To examine whether HMGB1 activates glia and drives cytokine production in the spinal cord, we measured in male mice Gfap and Cd11b mRNA levels as indicators of astrocyte and microglia activation, respectively, as well as Tnf, Il-1b, Ccl2, Cxcl1 and Cxcl2 mRNA levels, as these cytokines have been indicated to play important roles in pain processing. Six hours after disulfide HMGB1 injection pronounced increases in Gfap (2.0. ± 0.5 vs 1.0. ± 0.1 REU, n = 6, P = .012), Cd11b (1.5 ± 0.3 vs 0.7. ± 0.1 REU, n = 6, P = .003), Tnf (10.2 ± 2.6 vs 2.4. ± 0.3 REU, n = 6, P = .0001), Il-1b (23.1 ± 1.2 vs 3.2 ± 0.7 REU, n = 6, P = .003), Ccl2 (0.25 ± 0.01 vs 0.01 ± 0.01 REU, n = 6, P = .001), Cxcl1 (11.2 ± 1.8 vs 1.6 ± 0.4 REU, n = 6, P = .01), and Cxcl2 (3.5 ± 0.8 vs 0.3 ± 0.1 REU, n = 6, P = .001) mRNA levels were observed in the lumbar spinal cord compared to saline solution-injected mice (n = 12) (Fig. 7). These factors were also significantly elevated

Fig. 5. Single spinal injection of disulfide HMGB1(dsHMGB1) induces mechanical hypersensitivity in C57BL/6 and BALB/c male and female mice. Line graph showing the thresholds with 50% probability of response after i.t. injection of disulfide HMGB1 or saline solution using von Frey filaments in (A) C57BL/6 male, (B) BALB/c male, (C) C57BL/ 6 female, and (D) BALB/c female. (E) Redox forms of HMGB1 (atHMGB1 and oxHMGB1) in male C57BL/6 (F) Hyperalgesic index calculated for 0-3 hours. Data are presented as mean ± standard error of the mean, males: n = 6 mice per group; females n = 12 mice per group. ⁄P < .05, ⁄⁄P < .01, ⁄⁄⁄P < .001 versus saline solution-injected mice.

1810

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

Fig. 6. Single spinal injection of disulfide HMGB1 (dsHMGB1) induces mechanical hypersensitivity in Tlr2 / and Rage / , but not in Tlr4 / , mice. Line graph showing thresholds with 50% probability of response after i.t. injection of disulfide HMGB1 or saline solution using von Frey filaments in (A) Tlr2 / , (B) Tlr4 / , and (C) Rage / mice. (D) Hyperalgesic index calculated for 0-3 hours. Data are presented as mean ± standard error of the mean, n = 6 mice/group, ⁄P < .05, ⁄⁄P < .01, ⁄⁄⁄P < .001 versus saline solution-injected mice.

when compared to mice injected with all-thiol HMGB1 (with exception of Il-1b mRNA) (Fig. 7). All-thiol HMGB1 induced IL-1b (19.9 ± 5.6 vs 3.2 ± 0.7 REU, n = 12, P = .002) but not Tnf (4.2 ± 0.8 REU, n = 12, P > .9), Ccl2 (0.08 ± 0.03 REU, n = 12, P = .09), Cxcl1 (4.3 ± 1.3 REU, n = 12), Cxcl2 (1.2 ± 0.3 REU, n = 11) and Gfap (0.9 ± 0.1 REU, n = 6, P = .9) and Cd11b (0.7 ± 0.1 REU, n = 6, P = .9) gene expression. Disulfide HMGB1 did not increase levels of Tnf (4.5 ± 2.0 vs 2.3 ± 1.1 REU, n = 5, P > .9), IL-1b (6.5 ± 3.8 vs 3.2 ± 1.5 REU, n = 5, P = .9), Ccl2 (0.05 ± 0.03 vs 0.02 ± 0.01 REU, n = 5, P = .9), Cxcl1 (5.6 ± 3.2 vs 4.4 ± 2.5 REU, n = 5), Cxcl2 (0.1 ± 0.09 vs 0.02 ± 0.01 REU, n = 5), Gfap (0.5 ± 0.1 vs 0.6 ± 0.1, REU, n = 5) and Cd11b (0.6 ± 0.1 vs 0.7 ± 0.1, REU, n = 5) mRNA 6 hours after i.t. injection in Tlr4 / mice in comparison to saline solution-injected WT mice (Fig. 7 A–H). 4. Discussion In the present study, we found that spinal levels of HMGB1 mRNA and extranuclear HMGB1 protein levels are increased subsequent to induction of collagen antibody-induced arthritis. Notably, i.t. injection of 2 different HMGB1 antagonists reversed arthritisinduced mechanical hypersensitivity. These beneficial effects occurred not only during ongoing inflammation but also in the late phase, when mechanical hypersensitivity was still pronounced despite joint inflammation resolution, suggesting that spinal HMGB1 contributed to long-lasting pain-like behavior even in the absence of peripheral inflammation. Of importance, our study shows that i.t. injection of disulfide HMGB1, but not all-thiol or all-oxidized HMGB1, induces pain-like behavior in both male and female mice. Thus the redox state of HMGB1 is critical for determining its nociceptive action. We found that the nociceptive activity of disulfide HMGB1 was predominantly mediated via TLR4, and that disulfide HMGB1, but not all-thiol HMGB1, induced glial activation and cytokine gene expression in the spinal cord.

Together these results suggest that disulfide HMGB1 has a significant role in arthritis-induced pain transmission. 4.1. HMGB1 induced nociception is redox state and receptor dependent HMGB1 can activate cells through multiple surface receptors, including TLR2, TLR4 and RAGE [16]. These receptors, in particular TLR2 and TLR4, have been implicated in the regulation of neuropathic and inflammatory pain [11,21,26,38]. Whereas TLR2, TLR4, and RAGE are expressed on inflammatory cells and glia, TLR4 and RAGE are also expressed in sensory neurons [41,44]. Thus, pattern recognition receptors that are crucial for mounting an innate immune reaction are potentially also important in pain processing through their expression on both neurons and glial cells. We here provide evidence that the actions of HMGB1 in spinal pain transmission are TLR4 dependent and do not require TLR2 and RAGE signaling to the same extent, although it should be noted that RAGE-deficient mice recovered from HMGB1-induced hypersensitivity somewhat faster than wild-type mice. The notion of a predominant TLR4 dependence for HMGB1-induced hypersensitivity is further supported by our novel finding that spinal injection of HMGB1, when in the disulfide TLR4-activating form but not the allthiol or sulfonyl non-TLR4 binding forms, induces pronounced and long-lasting mechanical hypersensitivity and glial activation. As HMGB1 exists in its reduced all-thiol state intracellularly [17], our finding implies that to gain nociceptive and glial activating functions, all-thiol HMGB1 has to be converted to disulfide HMGB1. The extracellular milieu is more oxidizing than the nuclear and cytoplasmic environment, allowing formation of the C23 to C45 disulfide bridge in HMGB1 after release. Indeed, it has been shown in models of hepatic inflammation and muscle injury that released HMGB1 may change its redox state from a reduced to first a partially and then a fully oxidized form [43,45]. Thus, it is possible that HMGB1 is released in the spinal cord as a

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

1811

Fig. 7. Intrathecal injection of disulfide HMGB1 (dsHMG), but not all-thiol HMGB1 (atHMG), induces glial activation, cytokine, and chemokine gene expression in the spinal cord. Bar graphs depicting mRNA levels for (A) Tnf, (B) Il1-b, (C) Il6, (D) Ccl2, (E) Cxcl1, (F) Cxcl2, (G) Gfap, and (H) Cd11b. Data are normalized against Hprt1 mRNA levels and presented as mean ± standard error of the mean, n = 6 to 12 mice per group. ⁄P < .05, ⁄⁄P < .01, ⁄⁄⁄P < .001 versus the group receiving i.t. injection of disulfide HMGB1; #P < .05 versus the saline solution-injected WT group.

consequence of increased neuronal activity, and that, when subjected to oxidation, it becomes pro-nociceptive. This hypothesis is in line with recent work in the epilepsy field, showing that pro-excitatory effects of disulfide HMGB1 are coupled to TLR4 but not to RAGE. Via TLR4, disulfide HMGB1 increases phosphorylation of the NR2B subunit of the NMDA receptor, enabling enhanced receptor Ca2+ permeability in hippocampal neurons [3]. Furthermore, another study demonstrated that HMGB1 may elicit intracellular calcium flux and increase the excitability of peripheral DRG neurons [13]. Thus HMGB1 can drive nociceptive signaling directly by activation of neuronally expressed TLR4 and indirectly via activation of TLR4-expressing glial cells. 4.2. HMGB1 is expressed in spinal neurons and glial cells In agreement with previous studies performed in rats [13,28,34], we found ubiquitous HMGB1 expression in neurons, microglia, and astrocytes in the spinal cord of mice. HMGB1 was present at varying degrees in all neuronal nuclei and in most astrocytes and microglia in naive mice. We could not detect increased HMGB1 protein levels in the spinal cord by Western blotting when whole-cell homogenates were analyzed. Previous immunohistochemical studies have demonstrated nuclear HMGB1 translocation to the cytoplasm in spinal neurons after peripheral nerve ligation [28]. We found it technically challenging to measure immunoreactivity only in the cytoplasmic compartment of cells with small soma in the dorsal

horn cells. Instead we assessed HMGB1 protein levels by western blotting after the nuclear compartment of the lumbar spinal cords was removed, which revealed augmented extranuclear HMGB1 protein levels subsequent to CAIA induction. Although we could detect elevated levels of extranuclear HMGB1 protein only during the inflammatory phase, HMGB1 mRNA was elevated during both phases of the CAIA model, indicating that the machinery for de novo production of HMGB1 was engaged in both phases. This differential gene and protein expression in the late phase of the CAIA model is somewhat puzzling, in particular as i.t. injection of HMGB1 neutralizing antibodies, which presumably exert their activity extracellularly, demonstrated antihyperalgesic effect during both phases, suggesting that HMGB1 was present extracellulary during both disease phases. Hence, we conclude that western blotting may not be the optimal method for detection of extranuclear disulfide HMGB1; it is not a very sensitive method, and there are currently no antibodies available that distinguish disulfide from all-thiol and sulfonyl HMGB1. High-resolution mass spectrometry is an alternative approach for quantification of extranuclear levels of HMGB1 in the spinal cord. This technique has been used for measurement of the different HMGB1 redox forms in body fluids of patients [12] and in THP-1 cells [43]. As HMGB1 is expressed in both neurons and glia in the spinal cord, we cannot draw any conclusions about which cell type is responsible for HMGB1 release in our study. However, it has been shown that HMGB1 translocates from the nucleus to the cytoplasm

1812

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813

in DRG and dorsal horn neurons, but not in microglia or astrocytes, subsequent to peripheral nerve ligation [13,28]. HMGB1 is released from immortalized DRG cells in an activity-dependent manner [13], and ethanol exposure triggers dose-dependent HMGB1 release, predominantly from neuronal cells in brain slice cultures [49]. Thus, neurons may indeed be one potential source for extracellular HMGB1 in the transmission and regulation of nociceptive signals.

determining its nociceptive role. Agents interfering with this HMGB1 activity, such as HMBG1 neutralizing antibodies or recombinant HMGB1 box A protein, may therefore be considered in the development of new pain-relieving therapeutics.

4.3. Interference with signaling through the HMGB1-TLR4 axis attenuates arthritis-induced pain-like behavior

Acknowledgements

Previous work has shown that pain-related behavior in different preclinical models of chronic pain, including diabetic, cancer, and neuropathic pain, may be attenuated either by peripheral or i.t. injection of anti-HMGB1 antibodies [28,29,31,34,37,40]. These studies suggest that HMGB1 is involved in the maintenance of chronic hypersensitivity locally at the site of injury and/or at the level of the spinal cord. In the current study, we found that spinal administration of 2 different HMGB1 inhibitors, anti-HMGB1 mAb and Abox, reversed arthritis-induced mechanical hypersensitivity. CAIA-induced mechanical hypersensitivity concurs with the development of joint inflammation, but it is noteworthy that this outlasts the visual and histological signs of joint inflammation. Importantly, the antinociceptive action of i.t. injection of HMGB1 inhibitors was pronounced both during ongoing inflammation and in the late disease phase. Although mechanical hypersensitivity in the postinflammatory phase may be due to residual subclinical inflammation or irreversible articular tissue or nerve damage, our data suggest that it is at least partly also maintained via a facilitated state of spinal nociceptive processing. We previously demonstrated that spinal inhibition of TLR4 attenuated K/BxN transfer arthritis [11]. Furthermore, a role of endogenous TLR4 ligands in pain signaling is supported by work showing that TLR4 deficiency, spinal TLR4 knock-down, and i.t. injection of TLR4 receptor antagonists prevent or reverse pain-like behavior in, for instance, models of nerve injury and antibodyinduced pain [10,19,26,38]. One noteworthy observation in the current study is that i.t. HMGB1 evoked nociceptive behavior in both female and male mice, with nociceptive effect lasting somewhat longer in male mice, and that HMGB1 neutralization attenuated arthritis-induced pain-like behavior in both male and female mice. In contrast to our findings, earlier reports show that i.t. LPS induces mechanical hypersensitivity only in male mice, and TLR4 deficiency protects only male mice from development of mechanical hypersensitivity [35,36]. Although HMGB1, like LPS, interacts directly with the TLR4/myeloid differentiation protein 2 (MD2) complex and requires CD14 for optimal TLR4 activation [22], the exact nature of this interaction remains unclear. Thus it is possible that different TLR4 ligands activate TLR4 differently, and in a sexdependent manner, leading to the varying abilities of, for example, i.t. LPS and HMGB1 to induce hypersensitivity in female mice. In the current work, we did not find a major contribution of TLR2 and RAGE to HMGB1-induced hypersensitivity in male mice; however, it is possible that the nociceptive properties of HMGB1 are mediated to varying degrees via these and other receptors in female mice in other experimental models of pain, leaving female mice less affected by TLR4 deficiency. Further studies exploring the nociceptive mechanisms of TLR4 ligands and coupling to sex are warranted. 4.4. Conclusion In conclusion, our findings provide novel evidence suggesting a TLR4-mediated action of extracellular HMGB1 in arthritis-induced pain, and highlight that the redox state of HMGB1 is critical for

Conflict of interest The authors declare no competing financial interests.

This work was supported by the Swedish Research Council, Swedish Foundation for Strategic Research, Ragnar Söderberg Foundation, Knut and Alice Wallenberg Foundation, the Swedish Rheumatism association and the Karolinska Institutet Foundations. Recombinant HMGB1 was kindly provided by Dr. K J Tracey (Feinstein Institute for Medical Research, Manhasset, NY, USA). References [1] Andersson U, Tracey KJ. HMGB1 is a therapeutic target for sterile inflammation and infection. Annu Rev Immunol 2011;29:139–62. [2] Andersson U, Wang H, Palmblad K, Aveberger AC, Bloom O, Erlandsson-Harris H, Janson A, Kokkola R, Zhang M, Yang H, Tracey KJ. High mobility group 1 protein (HMG-1) stimulates proinflammatory cytokine synthesis in human monocytes. J Exp Med 2000;192:565–70. [3] Balosso S, Liu J, Bianchi ME, Vezzani A. Disulfide-containing HMGB1 promotes NMDA receptor function and excitotoxicity by activating TLR4-dependent signaling in hippocampal neurons. Antioxid Redox Signal 2014 [Epub ahead of print]. [4] Bas DB, Su J, Sandor K, Agalave NM, Pettersson J, Codeluppi S, Baharpoor A, Nandakumar KS, Holmdahl R, Svensson CI. Collagen antibody-induced arthritis evokes persistent pain with spinal glial involvement and transient prostaglandin dependency. Arthritis Rheum 2012;64:3886–96. [5] Bianchi ME, Manfredi AA. High-mobility group box 1 (HMGB1) protein at the crossroads between innate and adaptive immunity. Immunol Rev 2007;220:35–46. [6] Boyle DL, Rosengren S, Bugbee W, Kavanaugh A, Firestein GS. Quantitative biomarker analysis of synovial gene expression by real-time PCR. Arthritis Res Ther 2003;5:R352–60. [7] Chaplan SR, Bach FW, Pogrel JW, Chung JM, Yaksh TL. Quantitative assessment of tactile allodynia in the rat paw. J Neurosci Methods 1994;53:55–63. [8] Chavan SS, Huerta PT, Robbiati S, Valdes-Ferrer SI, Ochani M, Dancho M, Frankfurt M, Volpe BT, Tracey KJ, Diamond B. HMGB1 mediates cognitive impairment in sepsis survivors. Mol Med 2012;18:930–7. [9] Christianson CA, Corr M, Firestein GS, Mobargha A, Yaksh TL, Svensson CI. Characterization of the acute and persistent pain state present in K/BxN serum transfer arthritis. PAINÒ 2010;151:394–403. [10] Christianson CA, Corr M, Yaksh TL, Svensson CI. K/BxN serum transfer arthritis as a model of inflammatory joint pain. Methods Mol Biol 2012;851:249–60. [11] Christianson CA, Dumlao DS, Stokes JA, Dennis EA, Svensson CI, Corr M, Yaksh TL. Spinal TLR4 mediates the transition to a persistent mechanical hypersensitivity after the resolution of inflammation in serum-transferred arthritis. PAINÒ 2011;152:2881–91. [12] Dear JW, Antoine DJ. Stratification of paracetamol overdose patients using new toxicity biomarkers: Current candidates and future challenges. Expert Rev Clin Pharmacol 2014;7:181–9. [13] Feldman P, Due MR, Ripsch MS, Khanna R, White FA. The persistent release of HMGB1 contributes to tactile hyperalgesia in a rodent model of neuropathic pain. J Neuroinflamm 2012;9:180–90. [14] Gao Q, Ma LL, Gao X, Yan W, Williams P, Yin DP. TLR4 mediates early graft failure after intraportal islet transplantation. Am J Transplant 2010;10:1588–96. [15] Gardella S, Andrei C, Ferrera D, Lotti LV, Torrisi MR, Bianchi ME, Rubartelli A. The nuclear protein HMGB1 is secreted by monocytes via a non-classical, vesicle-mediated secretory pathway. EMBO Rep 2002;3:995–1001. [16] Harris HE, Andersson U, Pisetsky DS. HMGB1: A multifunctional alarmin driving autoimmune and inflammatory disease. Nat Rev Rheumatol 2012;8:195–202. [17] Hoppe G, Talcott KE, Bhattacharya SK, Crabb JW, Sears JE. Molecular basis for the redox control of nuclear transport of the structural chromatin protein Hmgb1. Exp Cell Res 2006;312:3526–38. [18] Hreggvidsdottir HS, Lundberg AM, Aveberger AC, Klevenvall L, Andersson U, Harris HE. High mobility group box protein 1 (HMGB1)-partner molecule complexes enhance cytokine production by signaling through the partner molecule receptor. Mol Med 2012;18:224–30. [19] Hutchinson MR, Zhang Y, Brown K, Coats BD, Shridhar M, Sholar PW, Patel SJ, Crysdale NY, Harrison JA, Maier SF, Rice KC, Watkins LR. Non-stereoselective reversal of neuropathic pain by naloxone and naltrexone: involvement of tolllike receptor 4 (TLR4). Eur J Neurosci 2008;28:20–9.

Ò

N.M. Agalave et al. / PAIN 155 (2014) 1802–1813 [20] Inglis JJ, Notley CA, Essex D, Wilson AW, Feldmann M, Anand P, Williams R. Collagen-induced arthritis as a model of hyperalgesia: Functional and cellular analysis of the analgesic actions of tumor necrosis factor blockade. Arthritis Rheum 2007;56:4015–23. [21] Kim D, Kim MA, Cho IH, Kim MS, Lee S, Jo EK, Choi SY, Park K, Kim JS, Akira S, Na HS, Oh SB, Lee SJ. A critical role of toll-like receptor 2 in nerve injuryinduced spinal cord glial cell activation and pain hypersensitivity. J Biol Chem 2007;282:14975–83. [22] Kim S, Kim SY, Pribis JP, Lotze M, Mollen KP, Shapiro R, Loughran P, Scott MJ, Billiar TR. Signaling of high mobility group box 1 (HMGB1) through toll-like receptor 4 in macrophages requires CD14. Mol Med 2013;19:88–98. [23] Kokkola R, Li J, Sundberg E, Aveberger AC, Palmblad K, Yang H, Tracey KJ, Andersson U, Harris HE. Successful treatment of collagen-induced arthritis in mice and rats by targeting extracellular high mobility group box chromosomal protein 1 activity. Arthritis Rheum 2003;48:2052–8. [24] Lee YC, Cui J, Lu B, Frits ML, Iannaccone CK, Shadick NA, Weinblatt ME, Solomon DH. Pain persists in DAS28 rheumatoid arthritis remission but not in ACR/EULAR remission: a longitudinal observational study. Arthritis Res Ther 2011;13:R83–91. [25] Li J, Wang H, Mason JM, Levine J, Yu M, Ulloa L, Czura CJ, Tracey KJ, Yang H. Recombinant HMGB1 with cytokine-stimulating activity. J Immunol Methods 2004;289:211–23. [26] Liu T, Gao YJ, Ji RR. Emerging role of Toll-like receptors in the control of pain and itch. Neurosci Bull 2012;28:131–44. [27] Lu Y, Bhushan S, Tchatalbachev S, Marconi M, Bergmann M, Weidner W, Chakraborty T, Meinhardt A. Necrosis is the dominant cell death pathway in uropathogenic Escherichia coli elicited epididymo-orchitis and is responsible for damage of rat testis. PLoS One 2013;8:e52919. [28] Nakamura Y, Morioka N, Abe H, Zhang FF, Hisaoka-Nakashima K, Liu K, Nishibori M, Nakata Y. Neuropathic pain in rats with a partial sciatic nerve ligation is alleviated by intravenous injection of monoclonal antibody to high mobility group box-1. PLoS One 2013;8:e73640. [29] Otoshi K, Kikuchi S, Kato K, Sekiguchi M, Konno S. Anti-HMGB1 neutralization antibody improves pain-related behavior induced by application of autologous nucleus pulposus onto nerve roots in rats. Spine 2011;36:E692–8. [30] Qin S, Wang H, Yuan R, Li H, Ochani M, Ochani K, Rosas-Ballina M, Czura CJ, Huston JM, Miller E, Lin X, Sherry B, Kumar A, Larosa G, Newman W, Tracey KJ, Yang H. Role of HMGB1 in apoptosis-mediated sepsis lethality. J Exp Med 2006;203:1637–42. [31] Ren PC, Zhang Y, Zhang XD, An LJ, Lv HG, He J, Gao CJ, Sun XD. High-mobility group box 1 contributes to mechanical allodynia and spinal astrocytic activation in a mouse model of type 2 diabetes. Brain Res Bull 2012;88:332–7. [32] Scaffidi P, Misteli T, Bianchi ME. Release of chromatin protein HMGB1 by necrotic cells triggers inflammation. Nature 2002;418:191–5. [33] Schierbeck H, Lundback P, Palmblad K, Klevenvall L, Erlandsson-Harris H, Andersson U, Ottosson L. Monoclonal anti-HMGB1 (high mobility group box chromosomal protein 1) antibody protection in two experimental arthritis models. Mol Med 2011;17:1039–44. [34] Shibasaki M, Sasaki M, Miura M, Mizukoshi K, Ueno H, Hashimoto S, Tanaka Y, Amaya F. Induction of high mobility group box-1 in dorsal root ganglion contributes to pain hypersensitivity after peripheral nerve injury. PAINÒ 2010;149:514–21. [35] Sorge RE, LaCroix-Fralish ML, Tuttle AH, Sotocinal SG, Austin JS, Ritchie J, Chanda ML, Graham AC, Topham L, Beggs S, Salter MW, Mogil JS. Spinal cord

[36]

[37]

[38]

[39]

[40]

[41] [42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

1813

Toll-like receptor 4 mediates inflammatory and neuropathic hypersensitivity in male but not female mice. J Neurosci 2011;31:15450–4. Stokes JA, Cheung J, Eddinger K, Corr M, Yaksh TL. Toll-like receptor signaling adapter proteins govern spread of neuropathic pain and recovery following nerve injury in male mice. J Neuroinflamm 2013;10:1742–2094. Tanaka J, Seki Y, Ishikura H, Tsubota M, Sekiguchi F, Yamaguchi K, Murai A, Umemura T, Kawabata A. Recombinant human soluble thrombomodulin prevents peripheral HMGB1-dependent hyperalgesia in rats. Br J Pharmacol 2013;170:1233–41. Tanga FY, Nutile-McMenemy N, DeLeo JA. The CNS role of Toll-like receptor 4 in innate neuroimmunity and painful neuropathy. Proc Natl Acad Sci U S A 2005;102:5856–61. Taylor P, Manger B, Alvaro-Gracia J, Johnstone R, Gomez-Reino J, Eberhardt E, Wolfe F, Schwartzman S, Furfaro N, Kavanaugh A. Patient perceptions concerning pain management in the treatment of rheumatoid arthritis. J Int Med Res 2010;38:1213–24. Tong W, Wang W, Huang J, Ren N, Wu SX, Li YQ. Spinal high-mobility group box 1 contributes to mechanical allodynia in a rat model of bone cancer pain. Biochem Biophys Res Commun 2010;395:572–6. Wadachi R, Hargreaves KM. Trigeminal nociceptors express TLR-4 and CD14: A mechanism for pain due to infection. J Dent Res 2006;85:49–53. Wang HC, Bloom O, Zhang MH, Vishnubhakat JM, Ombrellino M, Che JT, Frazier A, Yang H, Ivanova S, Borovikova L, Manogue KR, Faist E, Abraham E, Andersson J, Andersson U, Molina PE, Abumrad NN, Sama A, Tracey KJ. HMG-1 as a late mediator of endotoxin lethality in mice. Science 1999;285:248–51. Venereau E, Casalgrandi M, Schiraldi M, Antoine DJ, Cattaneo A, De Marchis F, Liu J, Antonelli A, Preti A, Raeli L, Shams SS, Yang H, Varani L, Andersson U, Tracey KJ, Bachi A, Uguccioni M, Bianchi ME. Mutually exclusive redox forms of HMGB1 promote cell recruitment or proinflammatory cytokine release. J Exp Med 2012;209:1519–28. Vincent AM, Perrone L, Sullivan KA, Backus C, Sastry AM, Lastoskie C, Feldman EL. Receptor for advanced glycation end products activation injures primary sensory neurons via oxidative stress. Endocrinology 2007;148:548–58. Yang H, Lundback P, Ottosson L, Erlandsson-Harris H, Venereau E, Bianchi ME, Al-Abed Y, Andersson U, Tracey KJ, Antoine DJ. Redox modification of cysteine residues regulates the cytokine activity of high mobility group box-1 (HMGB1). Mol Med 2012;18:250–9. Yang H, Ochani M, Li J, Qiang X, Tanovic M, Harris HE, Susarla SM, Ulloa L, Wang H, DiRaimo R, Czura CJ, Roth J, Warren HS, Fink MP, Fenton MJ, Andersson U, Tracey KJ. Reversing established sepsis with antagonists of endogenous high-mobility group box 1. Proc Natl Acad Sci U S A 2004;101: 296–301. Yang HA, Hreggvidsdottir HS, Palmblad K, Wang HC, Ochani M, Li JH, Lu B, Chavan S, Rosas-Ballina M, Al-Abed Y, Akira S, Bierhaus A, Erlandsson-Harris H, Andersson U, Tracey KJ. A critical cysteine is required for HMGB1 binding to Toll-like receptor 4 and activation of macrophage cytokine release. Proc Natl Acad Sci U S A 2010;107:11942–7. Zetterstrom CK, Jiang WW, Wahamaa H, Ostberg T, Aveberger AC, Schierbeck H, Lotze MT, Andersson U, Pisetsky DS, Harris HE. Pivotal advance: Inhibition of HMGB1 nuclear translocation as a mechanism for the anti-rheumatic effects of gold sodium thiomalate. J Leukoc Biol 2008;83:31–8. Zou JY, Crews FT. Release of neuronal HMGB1 by ethanol through decreased HDAC activity activates brain neuroimmune signaling. PLoS One 2014;9: e87915.