Human recombinant mature IL-1β was obtained from Endogen. Human recombinant precursor IL-1β (proIL-1β) was purchased from Cistron. Recombinant human PI-8 and PI-9, as well as the respective polyclonal antibodies (raised against the respective full-length protein; affinity-purified), were generated as described previously 41. The anti–IL-1β antibody was obtained from Upstate Biotechnology. The anti-ICE_P20 antibody was provided by Santa Cruz biotechnology. Control mouse and rabbit Ig were obtained from Sigma Chemical Co. Recombinant human caspase-1 was provided by Dr. Nancy Thornberry (Merck Laboratories, Rahway, NJ). Recombinant human caspase-3 was purchased from Transduction Laboratories. Gelsolin was provided by Dr. T. Azuma (Brigham and Women's Hospital). The synthetic tetrapeptide fluorogenic caspase-3 substrate (Ac-DEVD-AMC) and the synthetic caspase inhibitor DEVD-CHO were obtained from PharMingen.

Human vascular SMCs were isolated from human saphenous veins by explant outgrowth 45, and cultured in DMEM (BioWhittaker) supplemented with 1% l-glutamine (BioWhittaker), 1% penicillin/streptomycin (BioWhittaker), and 10% fetal bovine serum (FBS; Atlanta Biologicals). Cells were subcultured after trypsinization (0.5% trypsin [Worthington Biochemicals], 0.2% EDTA [EM Science]), in 75-cm² culture flasks (Becton Dickinson) and used throughout passages two to four. SMCs were cultured 24 h before the experiment in IT (insulin/transferrin) medium lacking FBS, as described previously 46. SMCs were characterized by immunostaining with anti-SMC α-actin antibody (Enzo Diagnostics Inc.).

Mononuclear phagocytes were isolated by density gradient centrifugation 47, using Lymphocyte Separation Medium (Organon-Teknika), and subsequent counterflow elutriation from freshly prepared human PBMCs obtained from leukopacs of healthy donors (provided by Dr. B. Rollins, Dana-Farber Cancer Institute, Boston, MA). The purity of monocytes/macrophages was ≥96%, as determined by FACS^® analysis (anti–human CD68 mAb FITC; PharMingen).

As the source of IL-18, THP-1 cells were obtained from American Type Culture Collection (TIB-202) and were maintained in RPMI 1640 with 2.5 g/liter glucose, 1 mM Hepes, 1 mM sodium pyruvate, 50 μM 2-ME, containing 10% human serum. To obtain lysates, cells were pelleted (500 g, 10 min), resuspended in 10⁶ cells/ml sterile water, and exposed to three freeze–thaw cycles.

Culture media and FBS contained <40 pg endotoxin/ml as determined by chromogenic Limulus amebocyte assay (QLC-1000; BioWhittaker).

For processing, recombinant human proIL-1β (20 nM), proIL-18 (equivalent to 50,000/ml cells), or gelsolin was incubated for 30 min with recombinant human caspase-1 (15 nM), native caspase-1 obtained from lysates of monocyte cultures (obtained by three freeze–thaw cycles), or recombinant human caspase-3 (25 nM) at 37°C in a final volume of 50 μl processing buffer (10 mM Hepes, 2 mM dithiothreitol, 5% glycerol; final concentrations). To analyze inhibition of processing, the indicated concentrations of PI-8 or PI-9 were preincubated with either the enzyme (recombinant or monocyte-derived caspase-1 or caspase-3) or the substrate (proIL-1β, proIL-18, or gelsolin) for the respective duration. For control purposes, a synthetic caspase inhibitor (DEVD-CHO) was used instead of the serpin. Processing was stopped by heating the samples (10 min, 95°C) in 10 μl SDS-PAGE (5×) sample buffer (0.2 M Tris, 5% glycerol, 0.1% SDS, 3% β-ME, 0.1 mg/ml bromophenol blue; final concentrations). Finally, samples were applied to SDS-PAGE and developed by either Coomassie stain (gelsolin) or Western blot analysis using the anti–human IL-1β or IL-18 antibody (1:1,000). To analyze the caspase-1 inhibitory activity in SMCs, culture lysates were harvested in processing buffer. After three freeze–thaw cycles, the lysates were preincubated with the recombinant enzyme (15 nM, 30 min, 37°C) before the IL-1β or IL-18 precursor (20 nM) was added for an additional 30 min.

For immunoprecipitation experiments, lysates of SMC cultures were harvested and incubated with nonimmune rabbit serum (24 h, 4°C; Vector Laboratories) to preclear the samples. The cell extracts were immunoprecipitated with the specific anti–PI-9 antibody (6 h, 4°C) and pelleted by subsequent addition of goat anti–rabbit IgG (18 h, 4°C; Jackson ImmunoResearch Laboratories) as well as protein A–Sepharose beads (2 h, 4°C; Amersham Pharmacia Biotech). Supernatants (corresponding to PI-9–depleted lysates of SMC cultures) and precipitates were either resuspended in SDS-PAGE loading buffer (200 mmol/liter Tris, 5% glycerol, 0.1% SDS, 3% β-ME, 0.1 mg/ml bromophenol blue), separated by SDS-PAGE, and analyzed by Western blotting, or were resuspended in processing buffer and applied to the processing assay.

Antisense PI-9 (5′-GAAAGAGTTTCCATGATGCAG-3′) and control (5′-TTACCGCGCCGTAGACGGGCA-3′) phosphorothioate oligodeoxynucleotides were synthesized and purified via reverse-phase HPLC by Integrated DNA Technologies, Inc. Subconfluent cultures of human vascular SMCs were washed twice with DMEM and subsequently incubated with DMEM containing Lipofectin (1 μg/ml; Life Technologies, Inc.) and the respective oligodeoxynucleotide (5 μM). Cells were maintained in DMEM for 72 h, before the 24-h stimulation (TNF-α, 50 ng/ml). Finally, lysates and supernatants of these SMC cultures were harvested and applied to Western blot analysis for PI-9 or caspase-1, and IL-1β ELISA, respectively.

Cell extracts, equilibrated by total protein (25 μg total protein/lane), culture supernatants, or processing assay preparations were separated by standard SDS-PAGE under reducing conditions and blotted to polyvinylidene difluoride membranes (Bio-Rad) using a semidry blotting apparatus (0.8 mA/cm², 30 min; Bio-Rad). Blots were blocked and first and second mAbs were diluted in 5% defatted dry milk/PBS/0.1% Tween 20. After 1 h of incubation with the respective primary antibody, blots were washed three times (PBS/0.1% Tween) and the secondary, peroxidase-conjugated, goat anti–mouse or goat anti–rabbit antibody (Jackson ImmunoResearch Laboratories) was added for an additional 1 h. Finally, the blots were washed (20 min, PBS/0.1% Tween 20) and immunoreactive proteins were visualized using the Western blot chemiluminescence system (NEN). Densitometric analysis of immunoreactive bands employed Image-Pro^® software (Media Cybernetics) applied to digital images of the respective Western blots.

Release of IL-1β from human vascular SMCs was measured by ELISA, following the recommendations of the manufacturer (Endogen). In brief, SMC supernatants (200 μl) obtained from the antisense experiments were added for 1 h to 96-well modules (Nunc) coated with the capturing antibody (1 μg/ml, 4°C overnight). Subsequently the plates were washed (PBS/0.1% Tween) three times, and the respective biotin-labeled detecting mouse anti–human IL-1β mAb (0.5 μg/ml) was added (1 h). Finally, wells were incubated for 30 min with alkaline phosphatase (Vectastain ABC kit, AK-500; Vector Laboratories) and were washed four times. Antibody binding was detected by adding p-nitrophenyl phosphate (1.39 mg/ml; Sigma Chemical Co.), and absorbance was measured at 405 nm in a Dynatech plate reader. The amount of cytokine detected was calculated from a standard curve prepared from recombinant mature IL-1β. Samples were assayed in triplicate. Error bars represent SD.

Surgical specimens of human carotid atheroma and aorta were obtained by protocols approved by the Human Investigation Review Committee at the Brigham and Women's Hospital. Nonatherosclerotic tissue was obtained from both carotid arteries and aortae, whereas atherosclerotic tissue employed only carotid specimen. Serial cryostat sections (5 μm) were cut, air dried onto microscope slides (Fisher Scientific), and fixed in acetone at −20°C for 5 min. Sections were preincubated with PBS containing 0.3% hydrogen peroxidase activity. The sections were then incubated (30 min) with primary or control (mouse myeloma protein MOPC-21; Sigma Chemical Co.) antibody, diluted in PBS supplemented with 5% appropriate serum. The subsequent processing was performed according to the manufacturer's recommendations (Universal Dako LSAB kit; Dako). Antibody binding was visualized with 3-amino-9-ethyl carbazole (Vector) according to the recommendations provided by the supplier. For colocalization of caspase-1 with PI-9, or either molecule with the respective cell type, double-immunofluorescence staining was performed. The goat anti–human ICE_P20 antibody (1:100) was applied for 90 min followed by biotinylated anti–mouse secondary antibody for 45 min and Texas red–conjugated streptavidin (Amersham Pharmacia Biotech). Subsequent to application of the avidin/biotin blocking kit (Vector Laboratories), rabbit anti–human PI-9 antibody (1:50), anti–muscle actin mAb for SMCs (Enzo Diagnostics), anti-CD31 mAb for ECs (1:400; Dako), or anti-CD68 mAb for macrophages (Mφ, 1:600; Dako) was added, and sections were incubated overnight at 4°C. Subsequently, the appropriate secondary antibodies were applied for 30 min followed by streptavidin-FITC (Amersham Pharmacia Biotech).

Frozen tissue from nonatherosclerotic and atheromatous specimens, dichotomized by morphological criteria as stable or vulnerable plaques as described previously 48, were homogenized (Ultra-turrax T 25; IKA-Labortechnik) and lysed (0.3 g tissue/ml lysis buffer: 10 mM sodium phosphate, 150 mM sodium chloride, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 0.2% sodium azide). The lysates were clarified (13,000 g, 15 min), and the protein concentration for each tissue extract as well as for the cell culture samples was determined using a bicinchoninic acid (BCA) protein assay according to the instructions of the manufacturer (Pierce Chemical Co.). 50 μg total protein was applied to Western blot analysis.

To determine whether PI-9 indeed inhibits processing of the native substrate by caspase-1, a processing assay was established in which the enzyme cleaves the 33-kD recombinant human IL-1β precursor (proIL-1β) into the expected 1 29 49 predominant 28- and 17-kD (as well as an occasional 25-kD) forms, as detected by Western blot analysis (Fig. 1 A). The inhibitory capacity of PI-9 was explored by addition of the serpin to the assay at various time points, i.e., allowing preincubation with either the enzyme or the substrate. When added after coincubation of proIL-1β with native (monocyte-derived) or recombinant caspase-1, PI-9 did not affect processing (Fig. 1 A). However, simultaneous incubation of PI-9, enzyme, and substrate in this assay yielded partial inhibition of proIL-1β processing, and when preincubated with the enzyme, PI-9 completely inhibited processing of the IL-1β precursor by native or recombinant caspase-1 (Fig. 1A and Fig. B). In addition to the time of preincubation, inhibition of caspase-1 processing activity by PI-9 depended on the temperature during preincubation, requiring preincubation at 37°C (using ≥100 nM PI-9) for maximal inhibition and showing diminished inhibitory activity at room temperature or 4°C. Partial inhibition occurred after times of preincubation as short as 5 min. In contrast to PI-9, a related serpin family member, PI-8 (50; as well as the more distant relative ovalbumin), did not affect caspase-1 activity, even when added at high concentrations (up to 200 μg/ml, ≈5 μM) or preincubated for extended periods of time (up to 24 h). Interestingly, PI-9 differentially inhibited the processing into the two predominant products, the 28- and 17-kD forms (Fig. 2). Complete inhibition of processing into the mature 17-kD form of IL-1β required 4–6 μg/ml (≈100–150 nM) PI-9 (n = 5; with an IC₅₀ ≈ 57 ± 16 nM), whereas inhibition of processing into the 28-kD form required higher concentrations (>20 μg/ml; >500 nM) of PI-9 (n = 5; with an IC₅₀ ≈ 272 ± 69 nM). Partial inhibition occurred when using as little as 0.5–1 nM PI-9. Similar concentration- and time-dependencies of the inhibition of caspase-1 processing activity were obtained when a second native substrate for the enzyme, IL-18, was used (Fig. 3). Interestingly, inhibition of the IL-18 processing activity also required the preincubation of the enzyme rather than the substrate for optimal effectiveness.

We further tested the specificity of the PI-9–mediated inhibition of caspase-1 processing activity by analyzing whether the serpin inhibits other caspases. Neither PI-9 nor PI-8 inhibited caspase-3 activity, using either a native substrate, gelsolin 51, in a processing assay similar to that used for analysis of caspase-1 activity, or a fluorogenic synthetic peptide (data not shown).

To extend the above observations, obtained with recombinant material, to extracts of human vascular SMCs, previously found to express a heretofore unidentified caspase-1 inhibitory activity (29; see also Fig. 5 A), we analyzed PI-9 expression in these cells. SMCs indeed expressed immunoreactive PI-9 constitutively as a cell-associated immunoreactive band with an apparent molecular mass of 42 kD (Fig. 4), resembling the biochemical characteristics described for the endogenous caspase-1 inhibitory activity. Stimulation with either IL-1β, TNF-α, or IFN-γ, or combinations thereof, affected neither the cell-associated, constitutive expression of PI-9 nor the release of the serpin. In contrast, constitutive expression of PI-9 in freshly isolated peripheral blood monocytes was diminished after either stimulation with LPS or culture-mediated differentiation (Fig. 4 B).

Immunoprecipitation of SMC lysates with the anti–human PI-9 antibody depleted coordinately the serpin (data not shown) and the caspase-1 inhibitory activity, as demonstrated by the finding that anti–PI-9–treated SMC lysates no longer prevented cleavage of the IL-1β precursor (Fig. 5 A). Interestingly, the anti–PI-9 immunoprecipitate contained both PI-9 (data not shown) and caspase-1 (Fig. 5 B). Control experiments revealed that the anti–PI-9 antibody does not cross-react with recombinant caspase-1. The precipitate did not contain proIL-1β processing activity (data not shown), indicating complete inhibition of endogenously expressed caspase-1 via endogenous PI-9. Neither anti–PI-8 (Fig. 5 B) nor nonimmune rabbit control IgG (data not shown) immunoprecipitated caspase-1.

To explore further the potential importance of PI-9–mediated regulation of caspase-1 processing activity, we treated cultures of unstimulated or TNF-α–stimulated vascular SMCs with PI-9 antisense oligonucleotides. This treatment coordinately reduced PI-9 expression (Fig. 6, top) and augmented the release of active IL-1β from TNF-α–stimulated cultures (Fig. 6, middle). As SMCs retain IL-1β intracellularly under usual conditions 52 53, this finding demonstrates that diminished expression of endogenous PI-9 indeed allows maturation of the IL-1β precursor in intact cells. PI-9 antisense treatment did not affect endogenous caspase-1 expression (Fig. 6, bottom), demonstrating specificity of the observed alterations in IL-1β release.

Since SMCs comprise the most abundant cell type in arteries and participate in vascular diseases, e.g., via expression of proinflammatory cytokines such as IL-1, we examined the expression of PI-9 in normal (n = 5) tissue and colocalized its expression with that from caspase-1 and IL-1β within atherosclerotic (n = 5) arteries. SMCs in normal arteries had low, but detectable, levels of caspase-1 (Fig. 7 A), which colocalized with abundant and homogeneously distributed PI-9 (Fig. 7 B). Normal vessels did not stain for IL-1β (Fig. 7 C). Expression of PI-9 was less intense and homogenous within atherosclerotic lesions compared with unaffected arterial tissue. Furthermore, PI-9 and caspase-1 localize inversely in atheroma. Caspase-1 (Fig. 7 D) as well as IL-1β (Fig. 7 F) localized most prominently in the macrophage-enriched shoulder region, whereas PI-9 localized most prominently in the SMC-enriched fibrous cap of the atherosclerotic plaque (Fig. 7 E), further demonstrated by immunofluorescence double labeling for PI-9 (green) and caspase-1 (red) (Fig. 7 G). The caspase-1 signal predominated over PI-9 in the more macrophage-rich shoulder area, whereas in the SMC-enriched fibrous cap PI-9 staining exceeded immunodetectable caspase-1. These findings corroborate with the reduced constitutive or absent expression of PI-9 in activated and/or differentiated mononuclear phagocytes (Fig. 4 H). In accordance with the immunohistochemical findings, Western blot analysis further demonstrated diminished total PI-9 concentrations in protein extracts of atherosclerotic lesions compared with unaffected tissue (Fig. 7 H). Moreover, Western blot analysis for PI-9 correlated inversely with enhanced immunoreactive IL-1β levels (Fig. 7 H). Notably, mature IL-1β was found primarily in extracts of atherosclerotic lesions, with morphologic characteristics of those prone to rupture and thus likely to cause acute clinical complications (defined by a large lipid core and thin fibrous cap, as described previously [48, 54]).