Caspase 1, a known cysteine protease, is a critical component of the inflammasome. Both caspase 1 and neutrophil serine proteases such as proteinase 3 (PR3) can process pro–interleukin-1β (proIL-1β), a crucial cytokine linked to the pathogenesis of rheumatoid arthritis. This study was undertaken to establish the relative importance of caspase 1 and serine proteases in mouse models of acute and chronic inflammatory arthritis.
Acute and chronic arthritis were induced in caspase 1−/− mice, and the lack of caspase 1 was investigated for its effects on joint swelling, cartilage metabolism, and histopathologic features. In addition, caspase 1 activity was inhibited in mice lacking active cysteine proteases, and the effects of dual blockade of caspase 1 and serine proteases on arthritis severity and histopathologic features were evaluated.
Surprisingly, caspase 1−/− mice, in a model of acute (neutrophil-dominated) arthritis, developed joint swelling to an extent similar to that in wild-type control mice. Joint fluid concentrations of bioactive IL-1β were comparable in caspase 1−/− mice and controls. In contrast, induction of chronic arthritis (characterized by minimal numbers of neutrophils) in caspase 1−/− mice led to reduced joint inflammation and less cartilage damage, implying a caspase 1–dependent role in this process. In mice lacking neutrophil serine PR3, inhibition of caspase 1 activity resulted in decreased bioactive IL-1β concentrations in the synovial tissue and less suppression of chondrocyte anabolic function. In addition, dual blockade of both PR3 and caspase 1 led to protection against cartilage and bone destruction.
Caspase 1 deficiency does not affect neutrophil-dominated joint inflammation, whereas in chronic arthritis, the lack of caspase 1 results in reduced joint inflammation and cartilage destruction. These findings suggest that inhibitors of caspase 1 are not able to interfere with the whole spectrum of IL-1β production, and therefore such inhibitors may be of therapeutic value only in inflammatory conditions in which limited numbers of neutrophils are present.
Cytokines, such as interleukin-1β (IL-1β), that are produced by cells of the innate immune system are induced in response to a variety of pathogen- or damage-associated molecular patterns. Because of its potent inflammatory properties, IL-1β can be deleterious if released in high amounts in various sites of the body (1). Both the production and activity of IL-1β are tightly regulated at several levels, including during transcription and translation (2), conversions of the inactive proIL-1β form into bioactive IL-1β (3), and excretion of IL-1β in microvesicles through K+-dependent mechanisms (4), as well as through interactions with the IL-1β receptors, such as natural antagonists (IL-1 receptor antagonist [IL-1Ra]), decoy receptors (type II IL-1R), and type I IL-1R shedding (5, 6).
Much interest has been generated in recent years with regard to the regulation of IL-1β, especially since the discovery that dysregulated IL-1β production can be linked to many of the manifestations of autoinflammatory disorders, including familial Mediterranean fever, Muckle-Wells syndrome, hyperimmunoglobulinemia D with periodic fever syndrome, familial cold urticaria, juvenile rheumatoid arthritis, adult-onset Still's disease, relapsing polychondritis, and Schnitzler syndrome (7). Consequently, treatments that block the activity of IL-1, with the use of either IL-1Ra or anti–IL-1β antibodies, are highly effective in these disorders (8). Even beta cell destruction in patients with type 2 diabetes mellitus is related to IL-1β–mediated toxicity and can be prevented by IL-1Ra (anakinra) treatment (9). Blockade of IL-1β activity has also been proven to be beneficial in rheumatoid arthritis (10). However, the relatively short half-life of IL-1Ra and the necessity for it to be administered via injection make it a suboptimal drug. Further elucidation of the molecular mechanisms behind production of bioactive IL-1β is needed before a more effective treatment for interfering with IL-1β production and activity can be designed.
In contrast to many other proinflammatory cytokines, IL-1β lacks a signal peptide, and its processing and secretion depend on cleavage by proteolytic enzymes such as caspase 1. Activation of caspase 1 has been proposed to be mediated, in turn, by protein platforms called the inflammasomes (11). Several of such inflammasomes are capable of activating caspase 1, and all of them include members of the NOD-like receptor (NLR) family of proteins, such as NALP3, NALP1, and IPAF (12). Mutations in NALP3 (cryopyrin) are a known cause of Muckle-Wells syndrome, neonatal-onset multisystem inflammatory disease, and chronic infantile neurologic, cutaneous, articular syndrome (13, 14), whereas NALP1 polymorphisms are associated with vitiligo and autoimmune diseases (15). Not only caspase 1 but also serine proteases, such as proteinase 3 (PR3) (16, 17), elastase, or cathepsin G, can process proIL-1β, but the role of these proteins in joint inflammation and cartilage destruction is limited (18). In addition, the mast cell proteases granzyme A and chymase have also been implicated in the activation of proIL-1β (19, 20).
Although the role of the NLR inflammasomes in the activation of IL-1β is supported by observations from in vitro studies and clinical data obtained from patients with autoinflammatory disorders, it is unclear whether activation of the inflammasome is also involved in other inflammatory disorders, such as arthritis. Moreover, the relative role of caspase 1 and serine proteases in the activation of proIL-1β in arthritis is not known. Since both caspase 1 and PR3 are considered to be potential targets in inflammation, discerning their roles in arthritis is important for the design of novel anti–IL-1β therapies. In the present study, we investigated the specific contribution of caspase 1 and serine proteases to both acute inflammation and chronic inflammation in experimental models of arthritis. To this end, we studied mice deficient in the caspase 1 gene (caspase 1−/− mice) or the dipeptidyl peptidase I (DPPI) gene (DPPI−/− mice, which lack neutrophil proteases), and performed blockade experiments with a potent caspase 1 inhibitor.
MATERIALS AND METHODS
Male C57BL/6 and BALB/c mice were obtained from Charles River Wiga (Sulzfeld, Germany). Mice deficient in the IL-1β gene (IL-1β−/− mice) were kindly provided by J. Mudgett (Merck, Rahway, NJ). Caspase 1−/− mice were originally obtained from R. A. Flavell (New Haven, CT) (21). DPPI−/− mice (BALB/c background) were obtained from Christine T. Pham. Homozygous matrix metalloproteinase 9 (MMP-9)–deficient mice were obtained from Robert Thompson (Washington University, St. Louis, MO). C57BL/6-Beige/Beige mice (lacking neutrophil elastase and cathepsin G) were obtained from Harlan/Olac (Blackthorn, UK). All mice were bred at the Central Animal Laboratory of Radboud University Nijmegen Medical Centre (Nijmegen, The Netherlands). All mice were housed in filter-top cages, and water and food were supplied ad libitum. The mice were used at the ages of 10–14 weeks. All animal experiments were approved by the Radboud University Animal Ethics Committee.
All polymerase chain reaction (PCR) primers were purchased from Biolegio (Malden, The Netherlands). Bovine serum albumin was purchased from Sigma-Aldrich (St. Louis, MO). TRIzol reagent, Taq DNA polymerase, and RPMI 1640 medium were obtained from Life Technologies (Breda, The Netherlands). SYBR Green was purchased from Applied Biosystems (Foster City, CA). Cytokine kits for the Luminex x-MAP technique were purchased from Bio-Rad (Hercules, CA). The tumor necrosis factor α (TNFα) bead kit was obtained from R&D Systems (Abingdon, UK). The caspase 1 inhibitor was provided by Novartis (Basel, Switzerland).
K/BxN serum transfer arthritis model.
The experimental model of K/BxN serum transfer arthritis is based on antibodies that recognize glucose-6-phosphate isomerase (22). Transfer of serum or purified IgG from K/BxN mice leads to induction of a robust and reproducible acute disease in several mouse strains, including C57BL/6 mice. Caspase 1−/− mice and wild-type control mice were injected with 200 μl of arthritic serum, and the expression of arthritis was scored macroscopically on a scale ranging from 0 to 2 per hind paw.
Streptococcal cell wall (SCW) preparation for the induction of acute or chronic arthritis.
Streptococcus pyogenes T12 organisms were cultured overnight in Todd-Hewitt broth. Cell walls were prepared as described previously (23). The resulting supernatant, centrifuged at 10,000g, was used throughout the experiments. These preparations contained 11% muramic acid. Unilateral arthritis was induced by intraarticular injection of 25 μg SCW fragments (rhamnose content) in 6 μl of phosphate buffered saline into the right knee joint of naive mice. To create SCW-induced chronic arthritis, multiple intraarticular injections of SCW into the right knee joint were administered on days 0, 7, 14, and 21. These repeated injections resulted in the development of chronic inflammation by day 28.
Treatment of SCW-induced arthritis with caspase 1 inhibitor.
Mice were treated with a potent caspase 1 inhibitor, pralnacasan (24), in 0.5% tylose, which was administered daily at a dose of 100 mg/kg orally. In the SCW-induced acute arthritis model, mice were treated from day 0 until day 2 with vehicle or the caspase 1 inhibitor. To block the activity of caspase 1 during SCW-induced chronic arthritis, mice were treated for 14 days with the caspase 1 inhibitor, starting on day 14 and continuing up to day 28 of arthritis. As controls, mice were treated with vehicle for 14 days.
Measurement of joint swelling.
Joint inflammation was measured by the 99mTc-uptake method, as described previously (25, 26). Briefly, 20 μCi of the radioisotope 99mTc in 200 μl saline was injected subcutaneously, which resulted in dispersal of the 99mTc over the whole body of each mouse in a few minutes. Due to increased blood flow and edema formation, accumulation of 99mTc in the right (inflamed) joint could be measured with external gamma counting. Values for joint swelling were expressed as the ratio of 99mTc uptake in the inflamed joint to that in the control joint (the left knee). All ratios exceeding 1.10 indicated the presence of inflammation.
Determination of chondrocyte proteoglycan (PG) synthesis.
Patellae were isolated with minimal surrounding synovial tissue, and chondrocyte PG synthesis in the patellae was determined by ex vivo 35S-sulfate incorporation. Briefly, the patellae were incubated in RPMI 1640 medium complemented with 740 μBq/ml 35S-sulfate for 2 hours at 37°C. Thereafter, the patellae were washed 3 times in saline and fixed in 4% buffered formaldehyde. After decalcification in 5% formic acid, the patellae were separated from the surrounding tissue and dissolved in Lumasolve (25). The radioactivity was determined by liquid scintillation counting.
RNA isolation and PCR amplification.
Immediately after cervical dislocation, synovial tissue was isolated from the inflamed knee joints. The samples of synovium were immediately stored in N2 until total RNA isolation. Using the Magnalyzer system (Roche, Basel, Switzerland), the synovium was grinded and total RNA was extracted in 1 ml TRIzol reagent (Life Technologies). Subsequently, 200 μl chloroform and 500 μl 2-propanol (Merck, Darmstadt, Germany) were used to separate the RNA from the DNA and proteins. Finally, after a wash step with 75% ethanol (Merck), the dry RNA was dissolved in 30 μl of water. To obtain double-stranded complementary DNA, standard reverse transcription–PCR was performed using oligo(dT) primers. Subsequently, quantitative PCR was performed using an ABI Prism 7000 Sequence Detection System (Applied Biosystems). PCR analyses of GAPDH, F4/80, and myeloperoxidase (MPO) were performed with SYBR Green PCR Master Mix (Applied Biosystems). Quantification of the PCR signals of each sample was performed by comparing the threshold cycle (Ct) values of the gene of interest, in duplicate, with the Ct values of the GAPDH housekeeping gene. We validated all primers according to the protocol, and the standard curves were all within the tolerable range.
Determination of cytokine levels.
Protein levels of IL-1β or TNFα were measured in washouts of patellae. At several time points after the induction of acute or chronic arthritis by SCW injection, the patellae were isolated from the inflamed knee joints and cultured for 1 hour at room temperature in RPMI 1640 medium containing 0.1% bovine serum albumin (200 μl/patella). Thereafter, the supernatants were harvested and centrifuged for 5 minutes at 1,000g. IL-1β and TNFα were measured in 50 μl of the patellar washouts, using Luminex xMAP technology. Bioactive IL-1β was measured using a previously described NOB-1 bioassay (27) with a modification of the IL-2 measurement, with results assessed utilizing the Luminex xMAP technique with the Bio-Plex IL-2 kit (Bio-Rad, Hercules, CA).
Processing of proIL-1β by PR3 or caspase 1.
Human recombinant proIL-1β (R&D Systems) was incubated for 120 minutes with 10 μg/ml purified human PR3 or caspase 1 (Athens Research & Technology, Athens, GA). Thereafter, cleaved IL-1β was tested for bioactivity by assessing the IL-1β–induced production of IL-8 in A549 cells. Cells were incubated for 24 hours with proIL-1β, cleaved product, or mature IL-1β. As a comparison, proIL-1β was processed in cell cultures with caspase 1. Specific enzyme-linked immunosorbent assays (ELISAs) were used to detect proIL-1β or total IL-1β (DLBP00; R&D Systems).
Mice were killed by cervical dislocation, and the whole knee joints were removed and fixed in 4% formaldehyde for 7 days before decalcification in 5% formic acid and processing for paraffin embedding. Tissue sections (7 μm) were stained with hematoxylin and eosin or Safranin O–fast green. Histopathologic changes in the knee joints were scored in the patella/femur region on 5 semiserial sections, spaced 140 μm apart. Scoring was performed on decoded slides by 2 separate observers (LABJ and MIK). The amount of cells infiltrating the synovial lining and the amount of cells infiltrating the joint cavity were each scored on hematoxylin and eosin–stained slides on a scale of 0–3. PG depletion was scored on Safranin O–stained slides on a scale of 0–3, ranging from stained cartilage to fully destained cartilage. In addition, cartilage damage was scored on a scale of 0–3, ranging from unaffected cartilage to the maximum effects of chondrocyte death and cartilage destruction (28).
Differences between experimental groups were tested using the Mann-Whitney U test. Data are expressed as the mean ± SEM, unless stated otherwise. P values less than 0.05 in comparison with wild-type controls were considered significant.
Lack of effect of caspase 1 deficiency on joint inflammation or cartilage damage during acute arthritis.
Although previous studies have analyzed the effects of caspase 1 inhibitors in models of osteoarthritis, the precise role of caspase 1 in the production of active IL-1β in inflammatory arthritis is not known. To investigate the function of caspase 1 in an experimental model of acute arthritis, we examined the passive K/BxN serum transfer model, an IL-1–dependent model of acute arthritis, in caspase 1−/− mice (22). As shown in Figure 1A, caspase 1 activity was not needed for arthritis to be induced by injection of arthritogenic K/BxN serum.
To corroborate these findings, we explored another model of arthritis in which IL-1β is produced and IL-1–mediated cartilage damage can be analyzed. As reported previously, intraarticular injection of cell wall fragments of Streptococcus pyogenes (SCW) results in joint inflammation and concomitant cartilage damage, the latter being highly dependent on the presence of IL-1 (25, 26, 29). When we analyzed the development of SCW-induced acute arthritis in caspase 1−/− mice, we found no evidence of significantly reduced joint swelling as compared with that in wild-type mice (Figure 1B).
Since active caspase 1 converts proIL-1β to active IL-1β, we compared the findings in caspase 1−/− mice with those in IL-1β−/− mice (Figures 1B–D). As has been shown previously, the response in terms of SCW induction of acute arthritis in IL-1β−/− mice is similar to that in wild-type mice (26). Interestingly, caspase 1−/− mice produced more local IL-1β (pro- and mature IL-1β) than did wild-type mice (Figure 1C). This might be explained by the excessive storage of proIL-1β attributable to the lack of caspase 1.
Inhibition of chondrocyte PG synthesis is a hallmark of joint inflammation and is a process that is almost exclusively dependent on IL-1 (25, 29). As illustrated in Figure 1D, chondrocyte PG synthesis was strongly inhibited in cartilage from wild-type mice on days 1, 2, and 4 after the initiation of SCW-induced acute arthritis, whereas chondrocyte PG synthesis in cartilage from IL-1β−/− mice was fully protected. Lack of caspase 1, in contrast, did not result in protection against the inhibition of chondrocyte PG synthesis, which is consistent with the increased local concentration of IL-1β in these mice (Figures 1C and D). However, on day 4 of SCW-induced acute arthritis, a partial recovery of chondrocyte PG synthesis was noted in the caspase 1−/− mice (Figure 1D).
In addition, histologic analysis of the cartilage on day 7 of SCW-induced arthritis revealed that the loss of PG from the cartilage matrix, attributable to the arthritic process, was ameliorated in caspase 1−/− mice compared with wild-type mice (Table 1 and Figures 1E and F). As has been reported previously (23), IL-1β−/− mice were completely protected against inflammation-induced cartilage PG loss (Table 1).
Table 1. Histopathologic features of the joints of mice deficient in caspase 1, IL-1β, DPPI, or MMP-9, compared with C57BL/6 or BALB/c wild-type and Beige/Beige controls, with or without caspase 1 inhibition*
Histopathologic changes were examined on day 2 or 7 of streptococcal cell wall (SCW)–induced acute arthritis and day 28 of SCW-induced chronic arthritis. For acute arthritis, mice were intraarticularly injected only on day 0. For chronic arthritis, SCW fragments were intraarticularly injected on days 0, 7, 14, and 21. Values are the mean ± SD score in at least 6 mice per group for the number of inflammatory cells in synovial tissue (scale 0–3), loss of Safranin O staining of cartilage (reflecting the extent of proteoglycan depletion) (scale 0–3), erosion of the cartilage surface and chondrocyte death (scale 0–3), and erosion of the patellar, tibial, and femoral bone (scale 0–5, ranging from no damage to complete loss of the bone). IL-1β = interleukin-1β; DPPI = dipeptidyl peptidase I; MMP-9 = matrix metalloproteinase 9; Beige/Beige = mice lacking neutrophil elastase and cathepsin G.
P < 0.01 versus wild-type (C57BL/6 or BALB/c), by Mann-Whitney U test.
Association of caspase 1 deficiency with partial protection against severe cartilage damage during chronic arthritis.
To investigate the role of caspase 1 in chronic destructive arthritis, we utilized a model of SCW-induced chronic arthritis in caspase 1−/− mice. It was recently demonstrated that this model is highly dependent on IL-1β (26), and we compared the role of caspase 1 with that of IL-1β. Figure 2A displays the protocol for the induction of chronic arthritis by multiple SCW fragment injections. After the fourth local injection of bacterial fragments (SCW), chronic inflammation, with concomitant joint destruction, developed on day 28 (Table 1 and Figure 2A). Microscopic analysis of the cellular infiltrate in both the joint cavity and the synovial lining showed that the percentage of polymorphonuclear neutrophils (PMNs) was notably different during the acute phase of SCW-induced arthritis compared with the chronic phase (Figure 2B). The neutrophil content in the joint cavity was >90% on day 1 after the intraarticular injection of SCW fragments, whereas less than 20% of the cells in the joint cavity were PMNs within 24 hours after the last reactivation on day 23 of SCW-induced arthritis (Figure 2B). Figures 2C and D show the PMN content of the synovial cavity of arthritic wild-type mice on day 1 as compared with day 23.
To demonstrate the distinct phases in this particular arthritis model, we investigated the cellular content of the inflamed synovium by using cell markers. Both microarray and real-time PCR technology were used. As shown in Figure 2E, PMN influx (determined as MPO gene expression) was clearly higher in the acute phase than in the late (chronic) phase. In contrast to MPO gene expression, we found that F4/80 (a macrophage marker) was strongly up-regulated in the chronic stage.
Mice deficient in caspase 1 tended to have less joint swelling, although the severity of the inflammation was not significantly different from that in wild-type mice (Figure 3A). In contrast, IL-1β−/− mice, as has been previously described (26), did not develop the chronic stage of arthritis in this model. Histologic analysis on day 28 demonstrated that caspase 1−/− mice displayed significantly less synovial inflammation, cartilage damage, and bone erosion than was observed in the wild-type controls (Table 1). Of interest, enhanced cartilage matrix PG production was noted in caspase 1−/− mice (compare Figures 3E and F), indicating that there was less IL-1β activity, since IL-1β−/− mice were completely protected against cartilage PG loss (25). Moreover, synovial messenger RNA expression of IL-1β, inducible nitric oxide synthase, and cyclooxygenase 2 was strongly reduced on day 28 (results not shown); all 3 of these mediators are known to be involved in suppression of chondrocyte PG synthesis.
Enhanced synovial concentrations of IL-1β in caspase 1–deficient mice in both the acute and chronic stages of SCW-induced arthritis.
Several studies have indicated that IL-1β is the pivotal cytokine with respect to cartilage catabolism during experimental arthritis (26, 29, 30). Since we observed striking differences in the way caspase 1−/− mice responded during acute and chronic experimental arthritis, in terms of chondrocyte metabolism and cartilage damage, we measured the levels of immune-reactive and bioactive IL-1β in synovial tissue washouts. As shown in Figure 3B, shortly after the injection of SCW fragments, IL-1β protein concentrations were 2–3-fold higher, both in acute arthritis and in chronic arthritis, in caspase 1−/− mice than in wild-type mice (P < 0.01). Of interest, IL-1β protein concentrations declined rapidly in caspase 1−/− mice with SCW-induced chronic arthritis but not in those with SCW-induced acute arthritis (Figure 3B).
Analysis of bioactive IL-1β concentrations using a cell-based assay showed results similar to the total protein measurement of IL-1β, although the synovial concentrations of bioactive IL-1β were 2-fold lower than the total IL-1β concentrations (Figure 3C). The metabolic function of chondrocytes was rapidly restored in mice deficient in caspase 1 in the chronic phase of SCW-induced arthritis (Figure 3D). In contrast, during SCW-induced acute arthritis, chondrocyte anabolic function was strongly affected in both caspase 1−/− and wild-type mice (Figure 3D).
Processing of precursor IL-1β to active IL-1β by PR3.
Enzymes other than caspase 1 can process IL- 1β in vitro, and these enzymes are predominantly neutrophil-derived serine proteases (16, 31). Since we noted a strong influx of neutrophils into the joint cavity after SCW induction of arthritis, we investigated whether one of these serine proteases, namely PR3, can process proIL-1β into bioactive IL-1β. As shown in Figure 4A, in experiments using A549 cells to assess IL-1β–induced IL-8 production, PR3 cleaved inactive proIL-1β into bioactive IL-1β. Processing of proIL-1β by PR3 was further confirmed by the use of a specific ELISA. After incubation with PR3, the concentration of proIL-1β clearly decreased, whereas concentrations of mature IL-1β were increased (Figures 4B and C).
Reduction of bioactive IL-1β by blockade of both caspase 1 and PR3.
To investigate whether blockade of PR3 and caspase 1 leads to less active IL-1β in SCW-induced acute arthritis as well as SCW-induced chronic arthritis, we performed studies in mice lacking neutrophil proteases (32), including those deficient in active PR3 (DPPI−/− mice). The mice were treated with a potent caspase 1 inhibitor (24). Deficiency of PR3 in combination with inhibition of caspase 1 did not reduce the joint swelling in SCW-induced acute arthritis. However, a moderate reduction of joint swelling was evident in caspase 1 inhibitor–treated DPPI−/− mice with SCW-induced chronic arthritis (results not shown). However, chondrocyte PG synthesis was completely protected in DPPI−/− mice exposed to the caspase 1 inhibitor, in contrast to that in the wild-type control mice (Figure 5A). These results are comparable with those found in IL-1β−/− mice (Figure 1B).
Analysis of total IL-1β protein concentrations revealed that in wild-type and DPPI−/− mice, the total IL-1β protein content was roughly similar between mice treated with the caspase 1 inhibitor and vehicle- treated control mice (results not shown). As indicated by analysis of the chondrocyte metabolic function, the amount of bioactive IL-1β was strongly reduced in the DPPI−/− mice treated with the caspase 1 inhibitor (Figure 5B).
Moreover, histologic analysis of the articular cartilage on day 2 after arthritis induction showed that the PG content of the articular cartilage displayed intense loss of matrix PG in wild-type mice, caspase 1 inhibitor–treated wild-type mice, and DPPI−/− vehicle-treated control mice (Figures 5C, D, and E). Of great interest, DPPI−/− mice injected with the caspase 1 inhibitor showed no loss of cartilage PG (Table 1 and Figure 5F). Furthermore, there was protection against severe cartilage damage on day 28 in DPPI−/− mice treated with the caspase 1 inhibitor, as compared with DPPI−/− vehicle-treated control mice and caspase 1 inhibitor–treated or vehicle-treated control wild-type mice (Table 1). In addition, the bone erosion noted in SCW-induced chronic arthritis on day 28 was reduced when caspase 1 was inhibited in DPPI−/− mice, as compared with that in DPPI−/− vehicle-treated control mice (Table 1). This latter effect was highly dependent on IL-1β, since IL-1β−/− mice were fully protected against bone erosion (Table 1). The SCW induction of acute or chronic arthritis in mice deficient in other classes of neutrophil serine proteases revealed that MMP-9 (in experiments using MMP-9−/− mice) as well as cathepsin G and neutrophil elastase (in experiments using Beige/Beige mice, which lack neutrophil elastase and cathepsin G) were not involved in joint inflammation or cartilage damage (Table 1).
In the present study, we dissected the mechanisms of IL-1β activation in an experimental model of arthritis and described the differential role played in this process by caspase 1 and the serine proteases, especially PR3. Consistent with the results from previous studies, we demonstrated that although joint swelling is mainly mediated by TNFα, production of IL-1β is crucial for PG loss. Caspase 1 and serine proteases such as PR3 are redundant in terms of their contribution to the release of bioactive IL-1β during joint inflammation, but blocking both processing systems results in an almost complete inhibition of the activation of IL-1β and protection against severe articular cartilage damage.
IL-1β is a proinflammatory cytokine that lacks a signal peptide and needs cleavage in order to be activated and released (33). The cysteine protease caspase 1 as well as the neutrophil serine proteases cathepsin G, elastase, and, in particular, PR3 are known to cleave proIL-1β (16, 34, 35). We demonstrated that cleavage of proIL-1β by caspase 1 and by PR3 results in the release of bioactive IL-1β. Although IL-1β may not be required for the induction of edema and swelling during the acute phase of the inflammatory reaction, as was shown in IL-1β−/− mice, its role is crucial for chondrocyte PG synthesis and tissue destruction (36). In contrast to its role in the acute phase, IL-1β strongly contributes to joint swelling and inflammation during the chronic phase of arthritis.
During inflammation, IL-1β can be released by a variety of leukocytes, such as monocytes, macrophages, and neutrophils (2, 8). The acute inflammation of arthritis is characterized by a rich infiltrate consisting of both neutrophils and monocyte/macrophages. When caspase 1 is absent in knockout mice, little effect on IL-1β production and PG loss during acute arthritis is observed, demonstrating that caspase 1 activation of IL-1β is not necessary in that stage. Most probably, the abundance of neutrophils is responsible for the release of serine proteases during acute inflammation. These enzymes are produced as zymogens that require enzymatic processing to become active.
The signal peptide of the serine proteases is cleaved by a signal peptidase called DPPI. This activation occurs just before the serine proteases are stored in the granules of neutrophils (32). It has been shown that DPPI−/− mice do not develop anticollagen antibody– and type II collagen–induced arthritis, being 2 experimental models in which initiation is based on immune complex formation in the joint (37, 38). One explanation for the absence of immune complex–induced arthritis in DPPI−/− mice may be the fact that mast cells of DPPI−/− mice lack active tryptase. Tryptase needs to be activated by DPPI for the production of chemokines, such as monocyte chemotactic protein 1 and IL-8, by mast cells. It is known that mast cells are crucial for the development of immune complex–induced arthritis in animal models (39, 40). However, we demonstrated herein that DPPI−/− mice develop aggravated joint inflammation when arthritis is induced by intraarticular injection of arthritis stimuli, such as SCW fragments or zymosan (Table 1 and Pham CT: unpublished observations). Because active serine protease is absent in DPPI−/− mice, the clearance of either SCW fragments or zymosan particles from the joint may be delayed.
The mere absence of PR3 alone in the DPPI−/− mice did not lead to decreased IL-1β production, likely due to the presence of caspase 1. However, when the activity of both enzymes was inhibited by administration of a potent caspase 1 inhibitor in DPPI−/− mice, IL-1β bioactivity was completely blocked and PG loss was reversed. During the chronic phase of arthritis, the relative importance of caspase 1 increases, as was suggested by the partial protection observed in caspase 1−/− mice, although a clear role of PR3 is still present, as was shown in the DPPI−/− mice.
The importance of caspase 1 in chronic inflammation is most likely a consequence of the predominance of macrophages and the presence of only a few neutrophils in the infiltrate in this phase of arthritis. The dichotomy between the role of caspase 1 in acute versus chronic inflammation is supported by experimental colitis models in which disruption of the caspase 1 gene leads to protection in chronic disease (41). These colitis models are also characterized by a minor role for neutrophils and a crucial role for monocytes and T cells (42). However, in models of lipopolysaccharide (LPS)–induced endotoxic shock, it was shown that caspase 1 was essential. LPS shock models are models of hyperacute inflammation in which IL-1β production peaks at 90 minutes. Monocytes are crucial for the LPS-induced IL-1β production, due to the fact that monocytes express already-active caspase 1 and can rapidly produce mature IL-1β (43). In most arthritis models, the local production and activation of IL-1β are dependent on macrophages or granulocytes (PR3).
Results from our present study and prior studies (16, 35) point to PR3 as the pivotal serine protease responsible for the effects observed in the DPPI−/− mice, since the 2 other major neutrophil serine proteases, neutrophil elastase and cathepsin G, were not able to process proIL-1β in vitro. In addition, in another model of experimental arthritis, it was shown that mice lacking elastase and cathepsin G were not protected against severe cartilage destruction (18). The present study showed that lack of MMP-9, neutrophil elastase (in Beige/Beige mice), and cathepsin G (in Beige/Beige mice) does not affect the development of SCW-induced arthritis (Table 1), and this suggests that PR3 plays the most important role. It has been demonstrated that PR3 is the most important serine protease released by neutrophils that process IL-1β (44) as well as IL-32 (45). From this perspective, PR3 is important in inflammation. The final proof for the role of PR3 in arthritis should come from studies of PR3-deficient mice, which are not available at this moment. However, we cannot exclude the possibility that DPPI has a PR3-unrelated effect on IL-1β release.
The findings of this study have clear clinical relevance. They provide an explanation for the clinical effects of anti–IL-1–based therapy, such as anakinra (recombinant IL-1Ra) and the fully human anti–IL-1β antibody ACZ885 (46, 47), and could explain the failure of caspase 1 inhibitors such as VX-765 in early clinical trials. In the present study, it was shown that under inflammatory conditions, the activities of caspase 1 and neutrophil serine proteases can process proIL-1β to mature IL-1β. Moreover, our results clearly demonstrate that processing of proIL-1β by serine proteases can compensate for the inhibition of caspase 1, especially in inflammatory foci in which neutrophils are abundant. From this perspective, one should probably apply therapies directed to caspase 1 only in inflammatory conditions in which neutrophils do not play a major role. A therapy that might be favored is likely to be one in which a combination of caspase 1 and serine protease inhibitors is used.
We thus have shown that caspase 1–independent processing of IL-1β occurs in arthritis by serine proteases such as PR3. Activation of IL-1β in a manner independent of caspase 1 is especially apparent in the acute phase of inflammation, characterized by a predominantly neutrophilic infiltrate that serves as a source for PR3. It is therefore likely that only therapies based on the dual inhibition of caspase 1 and serine proteases will be successful in the management of complex inflammatory diseases in humans.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Joosten had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Joosten, Netea, Van der Meer, Van den Berg.
Acquisition of data. Joosten, Fantuzzi, Koenders, Helsen, Sparrer, Pham.
Analysis and interpretation of data. Joosten, Netea, Fantuzzi, Koenders, Sparrer, Pham, Van der Meer, Dinarello, Van den Berg.
We thank Birgitte Walgreen and Liduine van der Bersselaar for providing support in the in vivo studies and histologic analyses.