Introduction

Gout is characterized by recurrent episodes of acute arthritis with redness, swelling, and intense pain of the joint, in which shedding of monosodium urate (MSU) crystals from preexisting articular deposits into joint cavities triggers an acute inflammation (Alexander, 1986). Studies have shown that an initial interaction between MSU crystals and cells in the joints is essential for the initiation and development of acute gouty arthritis. MSU crystals activate synovial cells, which release inflammatory mediators, leading to the accumulation of cells, neutrophils in particular, from the circulation. Once activated, the cells further release various inflammatory mediators that intensify the inflammatory responses (Gordon et al, 1988; McCarty, 1994). A wide variety of inflammatory mediators can be produced in the joints, including histamine, leukotriene B4, prostaglandins, platelet activating factor, proinflammatory cytokines, and chemokines (Terkeltaub, 1993; Weinberger, 1995).

Among chemokines, evidence suggests that the CXC chemokines play critical roles in the accumulation of neutrophils (Baggiolini et al, 1997; Rollins, 1997). The CXC chemokines are a structurally related and functionally redundant family of proteins with specific leukocyte chemotactic activity that can be divided into two subsets based on the presence or absence of specific amino acid residues Glu-Leu-Arg (ELR) (Baggiolini et al, 1997; Zlotnik and Yoshie, 2000). IL-8 (CXCL8) and growth-related oncogene-α (GROα: CXCL1) are members of ELR CXC chemokines that attract neutrophils. These chemokines have been detected in a variety of clinical diseases characterized by a massive neutrophil infiltration that include acute respiratory distress syndrome, bacterial inflammation, and rheumatoid arthritis (Kunkel et al, 1995; Luster, 1998). We have shown that these chemokines are essential in the recruitment of neutrophils in animal models of inflammation. Equivalent levels of IL-8 and GROα were detected at the foci of LPS-inflammation, and blockade of each chemokine resulted in a decrease in the number of infiltrating neutrophils. The inhibition was enhanced when these chemokines were together neutralized (Fukumoto et al, 1998; Matsukawa et al, 1997, 1998; Mo et al, 1999, 2000).

A large amount of IL-8 was detected in the synovial fluids of patients with gout (Terkeltaub et al, 1991). We and others have shown that IL-8 is involved in the neutrophil infiltration in a rabbit model of gouty arthritis (Matsukawa et al, 1998; Nishimura et al, 1997), suggesting an important role of IL-8 in gouty arthritis. However, little is known about the role of GROα in gouty arthritis. The aim of the present study was to elucidate the role of GROα and its possible cooperation with IL-8 in gouty arthritis. To address this, rabbits received intra-articular injection of MSU crystals and the subsequent inflammatory responses were investigated. We also analyzed the interaction of GROα with TNFα, IL-1β, and IL-8 in this model of gouty arthritis.

Results

Detection of GROα in MSU Crystal-Induced Arthritis

Intra-articular injection of MSU crystals resulted in an increase in the synovial level of GROα. GROα was first detectable at 1 hour and peaked at 2 hours after the injection. Once decreased at 4 hours, the level increased again and reached a second peak at 9 hours, after which the level decreased gradually and returned to the basal level at 36 hours (Fig. 1). We previously demonstrated the production of IL-8 in this model (Matsukawa et al, 1998). To compare the kinetics between these chemokines, the IL-8 level was measured in these newly harvested samples. As shown in Figure 1, the kinetics of IL-8 was similar to that of GROα, peaking at 2 and 12 hours after the injection. Interestingly, GROα level was higher than that of IL-8, demonstrating a two-fold increase at 2 hours (Fig. 1).

Figure 1
figure 1

Generation kinetics of GROα and IL-8 in MSU-induced arthritis. At indicated times after the intra-articular injection of 10 mg of monosodium urate (MSU) crystals, rabbits were killed. The joints were washed with saline and cleared synovial fluids were harvested. The levels of GROα (closed circles) and IL-8 (open circles) in the samples were then measured. Each point represents the mean ± se of 8 to 30 estimations from separate joints. * indicates below detection level.

Full size image

The cell-associated GROα level was also measured quantitatively. The data showed that infiltrating leukocytes contained abundant GROα (Table 1). The peak level of GROα in the infiltrating leukocytes was observed at 9 hours after MSU crystal injection, at a time when the number of infiltrating leukocytes (neutrophils > 95%) reached the maximal level (Matsukawa et al, 1998). The GROα level in the infiltrating leukocytes was notably lower than that in the synovial fluids at 2 hours, but significantly higher after 4 hours (Fig. 1; Table 1). In neutrophil-depleted rabbits, the synovial level of GROα at 2 hours was comparable to that in control rabbits, whereas the GROα level at 9 hours was significantly lower than control (Fig. 2), suggesting that synovial resident cells were the main source of GROα in an early phase, whereas synovial resident cells and infiltrating leukocytes were responsible for GROα production in a late phase. Immunohistochemistry demonstrated that synovial lining cells were positively stained for GROα at 2 hours (Fig. 3). These data clearly demonstrated that GROα was produced in the joints after MSU crystal injection.

Table 1 Production of GROα in Synovial Exudate Cells of MSU Crystal-Induced Arthritis
Full size table
Figure 2
figure 2

GROα level in normal and neutrophil-depleted rabbits. Neutrophil-depleted rabbits were prepared by intravenous injection of vinblastine (0.75 mg/kg). At 2 hours (A) and 9 hours (B) after the intra-articular injection of 10 mg of MSU crystals, the joints were washed with saline, cleared synovial fluids were harvested, and the levels of GROα in samples were measured. The data represent the mean ± se of 8 to 20 estimations from separate joints. # p < 0.05 when compared with normal rabbits.

Full size image
Figure 3
figure 3

Immunohistochemistry for GROα. Synovial tissues were obtained 2 hours after intra-articular injection of 10 mg of MSU crystals. The sections were stained with anti-GROα IgG, as described in the text. A, Synovial lining cells were positively stained for GROα. B, No staining was found in control (preimmune IgG). Original magnification, × 1000.

Full size image

Involvement of GROα and IL-8 in the Recruitment of Neutrophils

To investigate the involvement of GROα and its cooperation with IL-8 in the recruitment of neutrophils in this model, anti-GROα mAb, anti–IL-8 IgG, or their combination was administered together with MSU crystals, and the subsequent neutrophil infiltration was investigated. Administration of anti-GROα mAb resulted in a decrease in the number of infiltrating neutrophils after 6 hours. A maximal level of neutrophil infiltration observed at 9 hours after the injection was inhibited by 16% (Fig. 4). Consistent with our previous observations (Matsukawa et al, 1998), administration of anti–IL-8 IgG attenuated the neutrophil infiltration, which was a 21% inhibition at 9 hours. When these antibodies were together injected with MSU crystals, the neutrophil infiltration was further inhibited, resulting in a 33% reduction at 9 hours (Fig. 4). Thus, the contribution of GROα to the recruitment of neutrophils was comparable to IL-8, and GROα attracted neutrophils in collaboration with IL-8.

Figure 4
figure 4

Involvement of GROα and IL-8 in neutrophil influx. Anti-GROα mAb, anti–IL-8 IgG (10 μg each), or their combination was injected simultaneously with 10 mg of MSU crystals into rabbit knee joints. At indicated time intervals, the infiltrating leukocytes in the joints were counted. The data represent the mean ± se of 8 to 20 estimations from separate joints. # p < 0.05, § p < 0.01, when compared with MSU crystals alone. * p < 0.05, when compared with MSU crystals + anti-GROα mAb. † p < 0.05, when compared with MSU crystals + anti–IL-8 IgG.

Full size image

Regulation of GROα by TNFα, IL-1, and IL-8

We previously demonstrated the production of TNFα, IL-1β, and IL-8 in this model, all of which peaked at 2 hours after MSU crystal injection (Matsukawa et al, 1998). To investigate whether these cytokines/chemokines could induce GROα in this model, anti-TNFα mAb, IL-1 receptor antagonist (IL-1Ra), or anti–IL-8 IgG was injected with MSU crystals. As shown in Figure 5, the GROα level at 2 hours was not altered by anti-TNFα mAb, whereas the level was significantly inhibited (by 71%) by IL-1Ra. Administration of anti–IL-8 IgG also reduced the level, which was a 40% inhibition at 2 hours. On the other hand, either inhibitor decreased the second GROα peak seen at 9 hours after the injection (36% by anti-TNFα mAb, 32% by IL-1Ra, and 43% by anti–IL-8 IgG) (Fig. 5). These data demonstrated the differential regulation of GROα during inflammation.

Figure 5
figure 5

Regulation of GROα by TNFα, IL-1, and IL-8. Anti-TNFα mAb, IL-1 receptor antagonist (IL-1Ra), or anti–IL-8 IgG (10 μg each) was injected with 10 mg of MSU crystals into rabbit knee joints. At 2 hours and 9 hours after the injection, the synovial fluids were harvested and levels of GROα were measured. The data represent the mean ± se of 8 to 20 estimations from separate joints. # p < 0.05, §p < 0.01, when compared with MSU crystals alone.

Full size image

Regulation of TNFα, IL-1β, and IL-8 by GROα

Experiments were next conducted to examine the role of GROα in the production of TNFα, IL-1β, and IL-8. To do this, anti-GROα mAb was injected with MSU crystals, and levels of TNFα, IL-1β, and IL-8 were then measured at their peak (2 hours). Levels of IL-1β and IL-8 were also measured at their second peak (9 and 12 hours, respectively). The data demonstrated that administration of anti-GROα mAb failed to reduce the level of TNFα, IL-1β, and IL-8 at 2 hours after MSU crystal-injection (Fig. 6A). In contrast, the treatment significantly decreased the IL-1β level by 27% at 9 hours and the IL-8 level by 23% at 12 hours (Fig. 6, B and C). Thus, production of TNFα, IL-1β, and IL-8 in an early phase was independent of GROα, whereas GROα seemed to have a role in regulating the expression of both IL-1β and IL-8 in a late phase.

Figure 6
figure 6

Regulation of TNFα, IL-1β, and IL-8 by GROα. Anti-GROα mAb (10 μg) was injected with 10 mg of MSU crystals into rabbit knee joints. At 2 hours (A), 9 hours (B), and 12 hours (C) after the injection, the synovial fluids were harvested and levels of TNFα, IL-1β, and IL-8 were measured. The data represent the mean ± se of 8 to 20 estimations from separate joints. # p < 0.05, § p < 0.01, when compared with MSU crystals alone.

Full size image

Discussion

In addition to IL-8, MSU crystals are capable of inducing GROα in vitro (Terkeltaub et al, 1991, 1998). Mice lacking CXC chemokine receptor 2 (CXCR2) had impaired neutrophil infiltration in the air pouch model of MSU crystal-induced inflammation (Terkeltaub et al, 1998). These results suggest that GROα, a ligand for CXCR2, may be involved in the recruitment of neutrophils in gouty arthritis. In the present study, we demonstrated that GROα was detected in large concentrations in a rabbit model of gouty arthritis. Neutralization of GROα resulted in a decrease in the number of infiltrating neutrophils. The data clearly demonstrated the involvement of GROα in this model.

The participation of GROα in the recruitment of neutrophils was equivalent to IL-8, because the inhibition induced by anti-GROα mAb was comparable to that induced by anti–IL-8 IgG. Thus, GROα seems to be involved in the recruitment of neutrophils at the same level as IL-8. However, the GROα level was two-fold higher than that of IL-8, suggesting that IL-8 might be more effective than GROα in attracting neutrophils. We recently demonstrated that the number of infiltrating neutrophils induced by IL-8 was more than that induced by GROα, despite the same molar concentrations of chemokines injected into the joints (Fujiwara et al, 2002). This could be explained by the fact that IL-8 binds CXCR1 and CXCR2, whereas GROα binds only CXCR2 (Hall et al, 1999).

When these antibodies were together injected with MSU crystals, the neutrophil infiltration was further inhibited, indicating that GROα and IL-8 acted in concert to mediate neutrophil infiltration. We had initially assumed that these two chemokines would be the main factors for neutrophil infiltration in this model. However, in contrast to our assumption, the participation of these chemokines in the recruitment of neutrophils in this model was relatively small; the maximal inhibition induced by these antibodies was only 33% at the peak. Likewise, neutrophil infiltration in an early phase was not altered with these antibodies. These results suggest that chemoattractants other than GROα and IL-8 may play important role(s) in the accumulation of neutrophils. Webster et al (1972) demonstrated evidence for the combined actions of kinins, histamine, and complement in MSU crystal-induced inflammation model in the rat paw. Prostaglandins and leukotriene B4 were reported to play important roles in MSU crystal-induced arthritis (Rae et al, 1982; Rosenthale et al, 1972). A PAF antagonist exerted an anti-inflammatory effect in a rabbit model of gouty arthritis (Miguelez et al, 1996). Other mediators would be expected to include chemokines other than GROα and IL-8 (Zlotnik and Yoshie, 2000). These mediators, probably acting in cooperation, are likely to be responsible for the other part of the neutrophil infiltration. Some of these seemed to be regulated by IL-8 and IL-1, because administration of anti–IL-8 IgG together with IL-1Ra resulted in a decrease in the number of neutrophils of up to 62% (Matsukawa et al, 1998). Further study is necessary to address these points.

We have analyzed the cytokine network among TNFα, IL-1β, and IL-8 in LPS and MSU crystal-induced arthritis (Matsukawa et al, 1997, 1998). A characterization of GROα in the network was also investigated in LPS-induced arthritis and uveitis (Matsukawa et al, 1999; Mo et al, 2000). In MSU crystal-induced arthritis, production of GROα peaked at 2 hours, at the same time that TNFα, IL-1β, and IL-8 peaked (Matsukawa et al, 1998), suggesting a possible interaction of GROα with TNFα, IL-1β, and IL-8. Our present data suggest that TNFα does not initiate the production of GROα despite the production by synovial cells in vitro (Golds et al, 1989). Either IL-1 or IL-8 blockade decreased the level of GROα at 2 hours. The effect of IL-8 seemed to be independent of IL-1, because anti–IL-8 IgG failed to reduce the level of IL-1β at 2 hours (Matsukawa et al, 1998). Thus, GROα seems to be regulated in part by IL-1 and IL-8, but not TNFα, in an initial phase. In a late phase, GROα level was reduced with anti-TNFα mAb, IL-1Ra, or anti–IL-8 IgG by 30% to 40%. We demonstrated in our previous study that anti-TNFα mAb, IL-1Ra, or anti–IL-8 IgG decreased the infiltrating neutrophils (Matsukawa et al, 1998) at the same level as seen in GROα inhibition, and infiltrating neutrophils were another source of GROα in this phase. Thus, TNFα, IL-1, and IL-8 seem to induce GROα in a late phase, indirectly via the accumulation of neutrophils. On the other hand, levels of TNFα, IL-1β, and IL-8 at 2 hours were not altered with anti-GROα mAb, indicating that these cytokines/chemokine were independent of GROα in an initial phase. In contrast, anti-GROα mAb decreased IL-1β and IL-8 level in a late phase. As shown in this study, administration of anti-GROα mAb attenuated the neutrophil infiltration. Thus, the reduced IL-1β and IL-8 levels in a late phase by anti-GROα mAb were correlated with the decreased number of neutrophils at these time points by the treatment. We showed that infiltrating neutrophils were responsible for the production of IL-1β and IL-8 in a late phase (Matsukawa et al, 1998). Altogether, it is conceivable that GROα induces IL-1β and IL-8 production in a late phase indirectly via the accumulation of neutrophils. However, the degree of cytokine inhibition by anti-GROα mAb (IL-1β, 27%; IL-8, 30%) was greater than that of neutrophil infiltration by the treatment (16% at 9 hours and 15% at 12 hours). The discrepancy might be explained by the activation of neutrophils by GROα; that is, GROα may activate neutrophils, resulting in the stimulation of IL-1β and IL-8 release from the cells.

Studies have built the concept that chemokines are crucial in the pathogenesis of inflammatory responses. In LPS-induced arthritis, IL-8 as well as GROα blockade resulted in a considerable decrease in the number of infiltrating neutrophils, which was a 65% and 70% inhibition at 2 and 9 hours, respectively (Matsukawa et al, 1997, 1999);. In contrast, the participation of these chemokines to the neutrophil infiltration in MSU crystal-induced arthritis was limited, up to 33% at 9 hours. In addition, the neutrophil infiltration at 2 hours was unchanged by IL-8 and GROα blockade. Our series of experiments revealed the distinct contribution of these chemokines to the neutrophil infiltration in different models of inflammation.

Materials and Methods

Reagents

Anti-rabbit IL-8 and anti-rabbit GROα IgG were raised in a goat, and purified with affinity column coupled with each chemokine (Matsukawa et al, 1997, 1999). Anti-rabbit GROα mAb was prepared as described (Matsukawa et al, 1999). Anti-rabbit TNFα mAb was kindly provided by Dr. H. Nariuchi (University of Tokyo, Japan) (Haranaka et al, 1985). Monoclonal and polyclonal antibodies used in this study had specific neutralizing activity and did not cross-react with other rabbit cytokines/chemokines available (data not shown). Recombinant rabbit IL-1Ra was purified, as described previously (Matsukawa et al, 1992). Endotoxin contaminations in reagents were removed with Kurimover II (Kurita Kogyo Company, Tokyo, Japan), according to the manufacturer's instruction. Endotoxin contents in the resultant samples were less than 0.02 ng/mg protein (QCL-1000; Daiichi Pure Chemicals, Tokyo, Japan), respectively.

Induction of Arthritis

MSU crystals were prepared according to the methods reported by McCarty and Faires (1963). All animal experiments were performed under the guidelines and permission of the Animal Care Committee of Kumamoto University School of Medicine. Female New Zealand white rabbits (weight, 2.2 to 2.5 kg) were anesthetized by pentobarbital sodium (30 mg/kg) given intravenously, and MSU crystals (10 mg/500 μl saline) were injected in the presence of 10 U of polymixin B (PB; Pfizer Pharmaceutical, Tokyo, Japan) into the knee joints through the suprapatellar ligament. At various intervals after the injection, rabbit were anesthetized, bled, and killed. The knee joints were washed with 1 ml saline and the joint fluids were collected and centrifuged at 6,000 × g for 1 minute at 4° C. The cell-free synovial fluids were harvested and stored at −80° C until subsequent analyses. Cell pellets were resuspended in saline and the numbers were counted. Differential cell analysis was made after Wright-Giemsa staining.

Blocking the Activities of Cytokines

To neutralize TNFα, IL-1β, IL-8, and GROα activity in knee joints, 10 μg of anti-TNFα mAb, IL-1Ra, anti–IL-8 IgG, anti-GROα mAb were injected simultaneously with MSU crystals in a total volume of 500 μl through the suprapatellar ligament into knee joints. Ten micrograms of each reagent were regarded as sufficient, because higher doses (30 to 100 μg) did not further reduce leukocyte infiltration (not shown). These agents were randomly chosen and injected into bilateral knee joints. Injection of inhibitors into one side of knee joints did not affect the leukocyte infiltration induced in the contralateral knee joints. Ten micrograms of mouse IgG and preimmune goat IgG were used as controls. The control IgG did not induce leukocyte infiltration and cytokine production. Leukocyte influx and cytokine/chemokine level between MSU crystals alone and MSU crystals plus control IgG were very similar (not shown). Therefore, injection of MSU crystals alone was used as control.

Preparation of Neutrophil-Depleted Rabbits

Neutrophil-depleted rabbits were prepared by intravenous injection of 0.75 mg/kg vinblastine (Sigma Chemical, St. Louis, Missouri) 3 days before the experiments (Rosenshein et al, 1979). The numbers of peripheral leukocytes before and after vinblastine treatment were 12.7 ± 0.7 × 106/ml and 3.4 ± 0.8 × 106/ml, respectively. The majority of leukocytes in neutrophil-depleted rabbits were mononuclear cells, whereas 30% to 50% of the leukocytes in normal rabbits were neutrophils.

Immunohistochemistry

Freshly isolated synovial tissues were fixed in 4% paraformaldehyde, dehydrated, embedded in paraffin, and 4-μm–thick sections were made. After blocking endogenous peroxidase with 0.3% H2O2 in methanol, the tissue sections were treated with 5% normal rabbit serum and incubated overnight at 4° C with goat polyclonal anti-rabbit GROα IgG. Preimmune goat IgG was used as a control. After washing, the sections were then incubated for 30 minutes with 5 μg/ml biotinylated rabbit anti-goat IgG (Vector Laboratories, Burlingame, California), rinsed, and incubated for 30 minutes with avidin-biotin-peroxidase complex (Vector). As a chromogen, diaminobenzidine (DAB; Wako Chemical Industries, Osaka, Japan) was used. Counterstaining was performed with hematoxylin.

Measurement of Cytokines

The IL-1β and IL-8 levels were measured quantitatively by ELISA, as described (Matsukawa et al, 1997). The detection limits were 10 pg/ml and 30 pg/ml, respectively. GROα was measured quantitatively by time-resolved fluoroimmunoassay, as described previously (Matsukawa et al, 1999). The detection limit was 3 pg/ml. TNFα activity was determined by L929 cell cytotoxic assay, as described (Flick and Gifford, 1984). This cytotoxic assay was specific to TNFα, as preincubation of the samples with neutralizing anti-TNFα mAb completely abolished the cytotoxicity for L929 cells. The detection limit was 30 pg/ml. Synovial fluids were digested with 10 U/ml hyaluronidase (Sigma) for 1 hour at 37° C before assays.

Statistics

Statistical significance was evaluated by two-tailed unpaired Student's t test; p < 0.05 was regarded as statistically significant. All data were expressed as mean ± se.