Secretory Leukocyte Protease Inhibitor Suppresses the Inflammation and Joint Damage of Bacterial Cell Wall–Induced Arthritis

RNA extracted from peritoneal macrophages (PMs) was reverse-transcribed to cDNA using an oligo-dT primer (Promega). Two pairs of PCR primers were synthesized 22 to generate the complete open reading frame. The first pair: upstream primer 5′-GGAGGCAAAAATGATGCTATC-3′, downstream primer 5′-CCGAGCACGAGTCCAGAGCCG-3′; and the second pair: upstream primer 5′-CACCATGAAGTCCTGCGGCCT-3′, downstream primer 5′-GGCGCCAATGTCAGGGATCAG-3′. cDNA amplifications were performed using a Perkin-Elmer PCR kit. 10 μl of the PCR product using the first primer pair was reamplified with the second primer pair. PCR products were resolved on an agarose gel, transferred to a nitrocellulose membrane, and probed with murine SLPI (mSLPI) cDNA. The cDNA fragment that hybridized to mSLPI was subcloned and used to probe a rat macrophage cDNA library in Lambda ZAP II (Stratagene). Inserts of the positive clones were subcloned and sequenced with T3 and T7 primers at the National Institute of Dental and Craniofacial Research DNA core facility.

The EMBL, Swissprot, and GenBank molecular biology databases were searched using the network service (National Center for Biotechnology Information, NLM, Bethesda, MD) and the FASTA program from the Genetics Computer Group (GCG) Wisconsin Sequence Analysis Software Package (University of Wisconsin). Multiple sequence alignments were performed using ClustalW alignment in MacVector software (Oxford Molecular Group, Oxford, UK).

PBMCs and polymorphonuclear neutrophils (PMNs) were isolated from female Lewis rats as previously described 25. Resident PMs were collected by PBS lavage of the peritoneal cavity. Total RNA was isolated by the RNeasy protocol (Qiagen). 8 μg of total RNA was then subjected to Northern blot analysis using [³²P]dCTP-labeled rSLPI and rat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA as the probes.

The rSLPI cDNA encoding the mature secreted protein was generated by PCR with an upstream primer 5′-GGCAAAAATGATGCTATC-3′ and a downstream primer 5′-TTACACTGGGGGAAGGCAGA-3′ with BamHI and SalI adaptors, respectively. The BamHI–SalI cDNA fragment was subcloned into a prokaryotic expression vector pQE30 (Qiagen Inc.) with an in-frame ATG and the hexahistidine tag to the NH₂-terminus. Bidirectional sequencing verified the correct sequence and reading frame. Escherichia coli strain M15 was transformed with the rSLPI expression construct and grown in a 10-liter fermentation system at 37°C with ampicillin (100 μg/ml) and kanamycin (25 μg/ml) until an OD of 10 was reached. rSLPI expression was induced by addition of 1 mM IPTG for 5 h at 37°C. The cells were then harvested and the pellet was resuspended in 50 mM Na-phosphate buffer (pH 8.0) containing 300 mM NaCl, and in 10 mM imidazole. After sonication and RNase A and DNase I treatment, sequential 15,000 and 50,000 g centrifugations were performed. The supernate was mixed with pre-equilibrated Ni-NTA resin and incubated at 4°C for 1 h before being packed into a chromatographic column. Step-elution with 50, 75, 100, 125, and 150 mM imidazole was then carried out. The eluate from 75 mM imidazole, which showed anti-elastase, anti-cathepsin G, and anti-chymotrypsin activities, was dialyzed against sodium phosphate buffer (PBS, 10 mM sodium phosphate, 0.9% saline, pH 7.4), concentrated, filter-sterilized, and stored in aliquots at −80°C. 25 mg of purified rrSLPI protein was isolated from 50 g of wet weight of E. coli, which constituted 0.6% of the total soluble protein. rrSLPI was sequenced by Edman degradation (CBER, FDA, Bethesda, MD) to confirm identity. The molecular weights of the purified proteins were determined by mass spectrometry (Dr. Lewis Pannel, NIDDK, NIH, Bethesda, MD). An 8.4-kD truncated HIS-rSLPI recombinant protein (the COOH-terminus ends at Arg⁶⁴ of the mature rSLPI protein) was obtained using the same purification scheme, which has no protease reactive site and lacked anti-elastase, anti-cathepsin G, and anti-chymotrypsin activities. Endotoxin levels of both the full-length and truncated rrSLPI protein preparations were found to be ≤25 pg/ml (detection limit).

Rabbit polyclonal antibodies generated against the rSLPI peptide EGGKNDAIKIGAC (Quality Controlled Biochemicals, Inc.) were affinity purified and used for immunoblotting, ELISA, and immunohistochemical studies.

Elastase activity was measured as the amidolytic effect of human neutrophil elastase (15 nM; Calbiochem) on pyroGlu-Pro-Val-pNA (0.5 mM; Chromogenix) after 10 min at 37°C 26. Trypsin activity was determined by measuring the amidolytic effect of l-1-Tosylamide-2-phenylethyl chloromethyl ketone (TPCK)-treated bovine pancreatic trypsin (Sigma Chemical Co.) on the chromogenic substrate N-α-benzoyl-l-Arg-pNA (Boehringer Mannheim, Inc.) 27. Chymotrypsin (12.5 nM) and cathepsin G (32 nM) (Calbiochem) activities were measured using the chromogenic substrate N-suc-Ala-Ala-Pro-Phe-pNA (Sigma Chemical Co.) at 0.1 and 0.4 mM concentrations for 15 and 20 min, respectively 28,29. In all assays, the respective enzymes were incubated with or without different concentrations of purified rrSLPI at 37°C for 20 min. Then the substrate for each enzyme was added and the residual activity was measured as the change in absorbence at 405 nm after an appropriate incubation time at 37°C.

Arthritis was initiated in female Lewis rats (Charles River Breeding Labs., Inc.) by an intraperitoneal injection of group A SCW (Lee Labs.) 25. The severity of arthritis (AI) was determined by blind scoring of each ankle and wrist joint, based on the degree of swelling, erythema, and disfigurement on a scale of 0–4 and adding the scores for all four limbs during the course of the study (26–28 d). Rats were randomly assigned to five groups for two separate experiments (I and II) as follows: control animals, which received PBS on day 0 (n = 6 for both experiments); SCW-injected animals (n = 6 for both experiments); SCW-injected animals treated with rrSLPI on days 1 and 9 (n = 6 for each dosage group in experiment I, and n = 12 for experiment II); SCW-injected animals treated with rrSLPI on day 13 (n = 3 for each dosage group for experiment I, and n = 12 for experiment II); and SCW-injected animals treated with truncated rrSLPI on days 1 and 9 (n = 3 for each dosage group for experiment I, and n = 9 for experiment II). In experiment I, 100 μg or 1 or 5 mg of purified full-length or truncated rrSLPI was injected intraperitoneally at the times specified for each group of animals; in experiment II, 100 μg of full-length or truncated rrSLPI was used. Statistical significance was ascertained using the analysis of variance followed by Scheffe's post-hoc test. Radiographs were taken as previously described 25. Excised ankle joints were processed for histopathology 25, and sections were stained with anti-rSLPI antibody (5 μg/ml) using Vectastain-ABC kit with DAB substrate (Vector Labs.) and counterstained with methyl green 25.

Rat plasma samples (1:2) were diluted in the appropriate dilution reagent and the levels of TNF-α were measured by ELISA (Biosource International). The plasma levels of rrSLPI were measured by our ELISA assay. In brief, Immulon 4 HBX microtiter plates (DYNEX Technologies, Inc.) were coated with 200 μl of anti-tetra-HIS antibody (3 μg/ml; Qiagen), washed four times in PBS, and incubated in blocking buffer (2% sucrose, 0.1% BSA, and 0.9% sodium chloride) for 2 h. Standard ELISA procedure was then performed using rabbit anti-rSLPI (5 μg/ml) as the primary antibody, alkaline phosphatase–conjugated anti–rabbit antibody as secondary antibody (1:4,000; Boehringer Mannheim Biochemicals), and p-nitrophenyl phosphate as substrate (Sigma Chemical Co.).

Joint samples were obtained at the indicated times and processed in a freezer/mill (SPEX CertiPrep) in the presence of liquid nitrogen 25. Nuclear proteins were isolated from the powdered joint tissue by homogenization in lysis buffer (20 mM Tris, pH 7.6, 120 mM NaCl, 1% NP-40, 10% glycerol, 10 mM NaPPi, 100 mM NaF, 2 mM Na-orthovanadate, 1 mM AEBSF, and 5 μg/ml leupeptin). Electromobility shift assay (EMSA) for nuclear factor (NF)-κB was performed using a radiolabeled NF-κB consensus oligonucleotide probe (Promega) as previously described 30.

Rat plasma samples were collected by tail bleeding or intracardiac puncture. Plasma levels of type II collagen cleavage product Col2-3/4C long neoepitope were monitored by a solution phase ELISA assay based on a published method 31. The Col2-3/4C long neoepitope resides at the COOH terminus of the TC^A piece produced by cleavage of the α1(II) chain by collagenase. The mouse mAb used in the assay is distinct from the rabbit antibody described previously 31 in that it recognizes only the cleaved α1(II) chain and thus is absolutely specific for type II collagen (Billinghurst, R.C., M. Ionescu, M.-A. Fitzcharles, E. Keystone, and A.R. Poole, manuscript in preparation).

rSLPI mRNA was identified by reverse transcriptase PCR using nested primers designed against the mSLPI open reading frame 22. A 400-bp cDNA fragment was isolated and used to probe a cDNA library from rat PMs. A 490-bp cDNA fragment was isolated, sequenced, and found to contain an open reading frame (bp 9–402, Fig. 1 A) encoding a predicted translation product of 131 amino acids with an estimated molecular mass of 14 kD. The putative rSLPI coding sequence shares 88% homology with murine SLPI (mSLPI) and 75% with hSLPI. At the amino acid level, rSLPI exhibits 83 and 58% identity to mSLPI and hSLPI, respectively (Fig. 1 B). 16 cysteine residues are conserved in the mature SLPI (residues 26–131, Fig. 1 B) and the serine protease binding site in hSLPI (Leu⁷² of the mature protein) is identical in rSLPI (Fig. 1 B). However, by Northern blot analysis, distinct tissue distribution patterns compared with humans were observed (Song, X.y., L. Zeng, W. Jin, and S.M. Wahl, manuscript in preparation). Moreover, although no detectable hSLPI mRNA has yet been observed in human mononuclear cells and macrophages, rSLPI was constitutively present at very high levels in naive PMs, less abundant in PBMCs, and essentially absent in resting PMNs (Fig. 2 A).

When rSLPI mRNA was monitored during the course of synovial inflammation and compared with the low constitutive SLPI mRNA levels in control joints, a striking biphasic expression pattern was observed during the course of the disease. The increase in rSLPI mRNA coincided with the peak acute (day 4 after SCW injection) and chronic (day 30) inflammatory responses (Fig. 2 B), but decreased significantly at the onset of the remission phase (day 7), and was negligible at the onset of chronic disease (day 14). These findings suggest that failure to produce sufficient SLPI or to maintain the initial elevated levels of the protease inhibitor preceding chronic disease may facilitate joint destruction.

Assessment of the origins of SLPI in the inflamed synovium revealed that rSLPI protein was found in infiltrating inflammatory cells consistent with SCW-induced mRNA expression (Fig. 2 A), and was also associated with some chondrocytes and cells in the bone marrow (Fig. 2D and Fig. E).

Because SLPI levels were minimal during the interval before joint destruction becomes most pronounced, we attempted to restore levels of the protease inhibitor by exogenous delivery of active rrSLPI. For this purpose, a mature rSLPI–HIS fusion protein was expressed, purified to homogeneity (Fig. 3A–C), and found to contain a polypeptide recognized by both anti-rSLPI (Fig. 3 C, lane 4) and anti-HIS tag antibodies (data not shown). This full-length fusion protein (molecular mass ∼12 kD; Fig. 3 C, lane 4) exhibited antiprotease activity toward human neutrophil elastase (Fig. 3 D), cathepsin G, and bovine chymotrypsin, but not trypsin (data not shown), and was used in the subsequent in vivo study. A COOH-terminally truncated form of the rrSLPI protein without anti-elastase, anti-cathepsin G, or anti-chymotrypsin activities was used as the control agent in the following animal experiments.

Based on the SLPI expression profile (Fig. 2B–F), full-length or truncated rrSLPI (0.1, 1, and 5 mg) was administered intraperitoneally before the clinically evident acute response to SCW or preceding the chronic inflammatory response, when endogenous SLPI was at a low ebb. As early as 2 d after the first full-length SLPI-injection (0.1 mg), a reduction in AI was evident that was more significant by day 4 of the acute response (AI = 3.17 ± 0.86, n = 12; vs. 5.88 ± 0.98 in SCW animals, n = 12, P = 0.035, Fig. 4 A). The effects of 0.1, 1, and 5 mg of rrSLPI were similar based on AI (data not shown). However, the impact of a single injection of only 0.1 mg of full-length rrSLPI was transient, as the clinical and histopathological symptoms reappeared during the late chronic disease stage, albeit at reduced levels (data not shown). Importantly, joint swelling, erythema, and disfigurement, the hallmarks of chronic arthritis, a response typically considered T cell– and macrophage- rather than neutrophil-dependent, were drastically curtailed after a second rrSLPI injection 9 d after SCW (AI = 2.83 ± 1.17 vs. 9.58 ± 1.19, n = 12, day 26, P = 0.0065, Fig. 4 A) and remained suppressed for the duration of the experiment. Radiographic analysis revealed that compared with SCW-injected animals, which exhibit overt erosion of cartilage and bone (Fig. 4 B), full-length rrSLPI-treated animals had minimal pathological changes with only mild soft tissue swelling (Fig. 4 C). In contrast, treatment of SCW-injected animals with the antiproteolytically inactive truncated rrSLPI caused no significant changes in disease severity (Fig. 4 D). Moreover, a single injection of full-length rrSLPI at the beginning of chronic arthritis (day 13) appeared to reduce the disease severity, achieving statistical significance within 10 d (Fig. 4 A). When plasma levels of rrSLPI were examined in rrSLPI-treated (1 mg) and untreated SCW animals, 2,652 pg/ml of rrSLPI was detected in rrSLPI-treated animals 2 h after injection, which declined over the next 16 h. SLPI levels were not detected in control or untreated arthritic animals. Finally, no apparent adverse effects were observed in rrSLPI-treated (1 mg) animals, which maintained their body weight at a level similar to the nonarthritic control animals (body wt = 175.7 ± 1.8 g in rrSLPI-treated arthritic animals vs. 158.6 ± 1.76 g in untreated SCW arthritic animals and 184.5 ± 2.4 in nonarthritic control animals, day 26).

Circulating type II collagen collagenase–generated cleavage products reflect the increase in cleavage in degenerate articular cartilage in human osteoarthritis 31, but have not been analyzed in experimental animal models. A new assay to detect a hidden type II collagen epitope in the COOH-terminal three-quarter–length fragment (Col2-3/4C long neoepitope), which is exposed only when native triple helical type II collagen is cleaved by collagenase, enabled us to quantify cartilage degradation in SCW arthritic rats. We demonstrate that Col2-3/4C long neoepitope increased in plasma with disease progression (Fig. 4 E), paralleling the evolution of cartilage destruction observed in these animals (Fig. 4 B) 25. Interestingly, during the remission phase, when inflammation and swelling typically decline, collagen cleavage appears to persist with the spill-over of cleavage products into the circulation. After a single full-length rrSLPI treatment on day 1 after an arthritogenic injection of SCW, a significant reduction (two- to threefold) in Col2-3/4C long neoepitope plasma levels was detected even during acute arthritis (Fig. 4 E). The lack of significant decrease in cleavage products in rrSLPI-treated arthritic animals during remission (day 10) is consistent with the clinical and pathological data in which a single treatment with full-length rrSLPI does not fully sustain suppression of the arthritis through the chronic phase. Although a second rrSLPI injection was administered on day 9, this probably represents an inadequate interval of time to effect a further reduction in enzymatic profile on day 10. However, the injection of SLPI on day 9 did have a profound impact on the subsequent evolution of cartilage and matrix destruction associated with chronicity. In fact, full-length rrSLPI appeared to nearly eliminate the joint destruction as quantified by multiple parameters, including AI, radiologic assessment of joint destruction, and the release of type II collagen cleavage products (Fig. 4).

In addition to proteolytic blockade of matrix degradation, histopathological analysis revealed reduced evidence of synovial inflammation in rrSLPI-treated arthritic animals. It has been shown that TNF-α can induce SLPI production in epithelial cells 32, and SLPI can in turn inhibit TNF-α production in macrophages 21,22. To determine whether exogenously administered active rrSLPI could reinforce the endogenously produced SLPI to antagonize SCW-induced TNF-α 25, we compared plasma levels of TNF-α in full-length rrSLPI-treated and untreated arthritic animals over the course of the disease. Plasma levels of TNF-α in arthritic animals were elevated during the acute phase, decreased in remission, and markedly increased in the chronic phases of the disease (reference 33; Fig. 5 A). However, after administration of full-length rrSLPI on days 1 and 9, there was a rapid and sustained decrease in detectable levels of circulating TNF-α (Fig. 5 A). Since SCW 30 and TNF-α both activate NF-κB, which is responsible for the transcription of multiple inflammatory mediators including TNF-α 34, the consequences of rrSLPI injection on NF-κB activity in active rrSLPI-treated and untreated animals during both acute (Fig. 5 B) and chronic (data not shown) arthritis were examined. By EMSA, NF-κB was strongly activated in SCW-induced arthritic joint tissue as compared with PBS-injected control animals, and this activation was suppressed by rrSLPI (Fig. 5 B). These data implicate a potential role for rrSLPI in breaking the positive feedback loop between TNF-α and NF-κB activation, thereby dampening SCW-induced joint inflammation.

SLPI, a serine protease inhibitor with potent neutralizing activity for leukocyte enzymes such as neutrophil elastase, is upregulated in response to bacteria and bacterial products 21,22,24. Here we demonstrate that, in contrast to hSLPI, rSLPI, like its murine counterpart 24, is expressed in monocytes and PMs and, apparently, other cells exposed to bacterial cell wall components. Our identification of SLPI in articular chondrocytes is consistent with recent reports, although it is still controversial whether SLPI is produced by these cells in situ or is sequestered into the cartilage matrix 35,36,37. Moreover, we have expressed biologically active recombinant rSLPI, and show that when delivered to animals in which bacterial products induce arthritic lesions, SLPI can profoundly inhibit proteolytic tissue destruction and joint inflammation. This effect of rrSLPI on SCW-induced arthritis and the parameters monitored implies that SLPI inhibits multiple pathways either directly as a consequence of its antiprotease activity or via yet to be determined regulatory pathways.

SCW-induced arthritis presents a biphasic disease process with a neutrophil-dependent acute phase and a prolonged T cell–dependent, macrophage-mediated chronic phase that manifests ∼2 wk after SCW injection 25. Although the acute phase is considered reversible, the chronic destructive disease presents irreversible changes with progressive pannus invasion of the cartilage and bone, and eventual loss of normal joint structure and function. The increasing differences between nonarthritic and SCW arthritic animals in type II collagen cleavage as disease progresses is, in part, due to a gradual decrease in normal type II collagen cleavage in young control animals as they age, associated with the closure of the growth plate 38. Parallel to the biphasic disease manifestation, the expression of endogenous SLPI is substantially upregulated at both mRNA and protein levels. The unexpected increases in endogenous SLPI observed during both acute and chronic arthritis most likely represent an attempted defense against the elevated serine proteases triggered by SCW activation. However, expression of SLPI nearly vanishes in the interval preceding chronic destructive disease, suggesting that this loss of SLPI activity may contribute to the joint destruction typical of chronic arthritis. In this regard, the therapeutic effect of exogenously delivered rrSLPI might result from direct inhibition of the serine protease cascade, initiated not only by neutrophil elastase, but also by that derived from both lymphocytes and macrophages 39,40,41. Consistent with this hypothesis, denaturation of type II collagen in articular cartilage as reflected by quantifiable circulating levels of a neoepitope resulting from cleavage of collagen by collagenase, the Col2-3/4C long neoepitope, was significantly suppressed by exogenously delivered rrSLPI during both the acute and chronic phases of the disease. The persistent collagen cleavage during remission, when inflammatory activity is low, indicates that the devastating cartilage destruction associated with chronicity is the consequence of smoldering proteolysis. Consequently, the therapeutic effect of exogenous SLPI on chronic arthritis may also involve suppression of protease activities existing at the transition between acute and chronic arthritis.

By providing exogenous active rrSLPI, the severity of disease at the clinical, histological, and molecular levels can be reduced. By neutralizing elastase, which cleaves not only elastin, but also collagen, fibronectin, and proteoglycans leading to cartilage and matrix degradation, this protease inhibitor may have multiple beneficial actions 8,9. Moreover, because elastase 3,5 and the plasminogen–plasmin system 3 can activate latent matrix metalloproteinases, SLPI may also indirectly block matrix metalloproteinase–mediated matrix cleavage. Recent evidence indicates that SLPI inhibits synthesis of macrophage collagenase and impairs PGE synthesis 42, implicating SLPI in the interruption of a proteolytic cascade of inflammatory events that evolve to tissue destruction 43,44.

In addition to inhibiting proteolytic tissue destruction, SLPI appears to have multiple antiinflammatory actions. SLPI inhibits TNF-α production, probably through its ability to block NF-κB activation 22,45. Our demonstration of SLPI's effect on TNF-α production and NF-κB activity in vivo complements the in vitro findings that SLPI is upregulated by proinflammatory stimuli including LPS, TNF-α, IL-6, and IL-1β, and subsequently inhibits TNF-α and nitric oxide production. These pathways may underlie SLPI's antiinflammatory functions in vivo 22,24,32.

Based on the similarities between human rheumatoid arthritis and SCW-induced experimental arthritis, and the shared protease reactive site in hSLPI and rSLPI, our study provides initial preclinical evidence in support of the use of this serine protease inhibitor in alleviating arthritic pathology. SLPI may represent a prototype therapeutic approach for multiple inflammatory destructive diseases.

We thank Dr. Y. Shiloach for his assistance with the fermentation system, and Drs. L. Wahl and L. Fisher for their critical reading of the manuscript.