Tissue Kallikrein mK13 Is a Candidate Processing Enzyme for the Precursor of Interleukin-1β in the Submandibular Gland of Mice*

By using Western blot analysis, high levels of 17.5- and 20-kDa interleukin-1β (IL-1β) proteins were detected in the submandibular gland (SMG) of mice. Despite this fact, the amount of pro-IL-1β protein, a precursor of IL-1β, with a molecular size of 35 kDa in this tissue was below the detectable level, although strong expression of pro-IL-1β mRNA was observed. A large amount of 17.5-kDa IL-1β also appeared in the saliva of mice injected with lipopolysaccharide, suggesting that this IL-1β is a secretory form produced by the SMG. The protein for IL-1β-converting enzyme, a processing enzyme for pro-IL-1β, was expressed only at a low level in the SMG as compared with its level in various epithelial tissues or lipopolysaccharide-stimulated macrophages. On the other hand, mK1, mK9, mK13, and mK22, members of the kallikrein family, were detected strongly in the SMG but not in other tissues. By incubation with mK13, but not with mK1, mK9, or mK22, the 35-kDa pro-IL-1β was cleaved into two major products with molecular masses of 17.5 and 22 kDa, and production was inhibited by phenylmethylsulfonyl fluoride, a serine protease inhibitor, but not by IL-1β-converting enzyme inhibitors. A peptide segment corresponding to amino acid residues 107–121 of mouse pro-IL-1β (107WDDDDNLLVCDVPIR) was cleaved by incubation with mK13, generating two peptides, 107WDDDDNL and 114LVCDVPIR. Therefore, kallikrein mK13 would appear to hydrolyze pro-IL-1β between its Leu113and Leu114 residues. The results of immunohistochemistry and an autonomic therapy experiment showed that IL-1β and kallikrein mK13 were co-localized in the secretory granules of granular convoluted tubular cells. Our present results thus suggest kallikrein mK13 is a plausible candidate for the processing enzyme for pro-IL-1β in the SMG of mice.

Members of the interleukin-1 (IL-1) 4 family are thought to be the key mediator when the host responds to microbial invasion, inflammation, tissue injury, and immunological reactions. Both directly and indirectly, IL-1 plays a role in the activation of T cells and hematopoietic stem cells, the induction of fever and acute-phase proteins, the degradation of cartilage, and the healing of wounds. Two biochemically distinct but structurally related IL-1 molecules have been cloned, IL-1␣ (1) and IL-1␤ (2). IL-1␣ and IL-1␤ have similar but not identical multiple biological effects, which have been well characterized by in vivo and in vitro experiments (3,4). Both forms of IL-1 are synthesized by a wide variety of tissues and affect almost all types of cells, which share a common receptor. These cells are keratinocytes, lymphocytes, natural killer cells, macrophages, and others. Each IL-1 is encoded by a separate gene, both of which are located on chromosome 2; and each gene contains seven exons. Their initial protein products are pro-IL-1s of large molecular weight. When mRNAs for these precursor cytokines are translated in vitro, only the IL-1␣ precursor (actually pro-IL-1␣) is active (5) and binds to the IL-1 receptor (6). Removal of the NH 2 -terminal half of the IL-1␤ precursor (pro-IL-1␤) is essential for the biological activity of this form, suggesting the involvement of a processing step. Thus pro-IL-1␤ is a 35-kDa protein synthesized from IL-1␤ mRNA and is an inactive precursor without a conventional signal sequence. It is cleaved to generate the 17.5-kDa bioactive, secreted form of the cytokine (7,8). One of the enzymes responsible for the generation of 17.5-kDa IL-1␤ is IL-1␤converting enzyme (ICE), a cytoplasmic cysteine protease characterized after isolation from cells of monocytic origin and shown to have the ability to process pro-IL-1␤ to generate the active inflammatory cytokine IL-1␤ (9,10).
On the other hand, the kallikreins are a group of serine proteases that are found in diverse tissues and biological fluids, and they are now divided into two major categories, plasma kallikreins and tissue kallikreins. The function of plasma kallikreins includes their participation in the process of blood clotting and fibrinolysis and in the regulation of vascular tone and inflammation reactions through the release of bradykinin (11). The tissue kallikreins were originally characterized by their ability to generate kallidin or bradykinin from kininogens (12). They are expressed in the glandular tissues such as the salivary glands and pancreas as well as in the kidney and other tissues. Tissue kallikreins are also involved in the processing of growth factor precursors (13), activation of other kallikreins in the family, or in the hydrolysis of gel-forming proteins such as semenogelin and fibronectin (14). There are many tissue kallikrein enzymes, and their genes make up a large family, the tissue kallikrein gene family (15,16). All of these genes code for putative serine proteases and share important similarities, including the same chromosome mapping, significant homology at both the nucleotide and protein level, and similar genomic organization. Four members of the tissue kallikrein family, mK1, mK9, mK13, and mK22, are found in the mouse submandibular gland (SMG). We recently found that the 17.5-kDa form of IL-1␤ was present in the SMG (17) and secreted into the saliva. In this study, we demonstrate that one of the kallikreins, mK13, also known as pro-renin-converting enzyme, is also able to process pro-IL-1␤ to form 17.5-kDa IL-1␤.

MATERIALS AND METHODS
Reagents-Lipopolysaccharide (LPS) from Escherichia coli (serotype 0111:B4), thioglycollate medium, and TRI Reagent TM were obtained from Sigma. Recombinant mouse IL-1␤ and goat anti-recombinant mouse IL-1␤ (IgG, Western blotting) were products of Peprotech EC (London, UK) and Genzyme-Techne (Cambridge, MA), respectively. RPMI 1640 medium was from Invitrogen. Goat polyclonal antibody raised against a carboxyl-terminal peptide of mouse IL-1␤ (including the blocking peptide; for immunohistochemistry), rabbit anti-ICE polyclonal antibody (specific for the p10 subunit and 45-kDa precursor of caspase-1), and the blocking peptide, donkey anti-goat IgG (H ϩ L) conjugated with fluorescein isothiocyanate (FITC), goat anti-rabbit IgG (H ϩ L) FITC, donkey anti-goat IgG-conjugated with horseradish peroxidase, and donkey anti-rabbit IgG-horseradish peroxidase were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Superscript One-step RT-PCR kit was obtained from Invitrogen, and NuSieve and SeaKem agarose came from Cambrex BioScience (Rockland, ME). ELISA kit was obtained from BIOSOURCE. Caspase-1 inhibitor I (acetyl-Tyr-Val-Ala-Asp-CHO) and caspase-1 inhibitor VIII (Ac-Trp-Glu-His-Asp-CHO) were purchased from Calbiochem. Active recombinant mouse caspase-1 was procured from BioVision (Mountain View, CA). A Historesin Plus kit came from Leica Microsystems (Heidelberg, Germany); and a Bio-Rad protein assay kit was from Bio-Rad. An ECL detection kit was obtained from Amersham Biosciences. A peptide segment of pro-IL-1␤ ( 107 WDDDDNLLVCDVPIR) was custom-synthesized and procured from NeoMPS, Inc. (San Diego).
Preparation of the Four Major Tissue Kallikreins from the SMG of Mice-The tissue kallikreins mK9, mK13, and mK22 were prepared from the SMG of adult male mice, whereas mK1 was obtained from the same gland of female mice (18,19). The purification steps included acid treatment, acetone fractionation, isoelectric focusing, and DEAE-cellulose column chromatography, as described previously (20). Each enzyme was identified by analysis of its complete or partial amino acid sequence. The activity of each purified enzyme for hydrolysis of benzoylarginine ethyl ester at 20°C was 162, 271, 180, and 805 units/mg protein for mK1, mK9, mK13, and mK22, respectively, and these values indicated that the purified enzymes were fully active (18,19). The purity of each enzyme was appreciable, as judged from electrophoresis data and protein sequence analysis.
Preparation of Anti-tissue Kallikrein Antiserum with Restricted Immunoreactivity-Antisera for tissue kallikreins were raised in rabbits by a standard procedure. Because the enzymes of the tissue kallikrein family are highly homologous to each other, these antisera needed to be pre-absorbed with other major SMG kallikreins to eliminate their crossreactivity and enhance their specificity. Following the procedure reported previously (21), a mixture of anti-mK1 antiserum and crossreactive kallikreins (10 l of rabbit polyclonal anti-mK1 antiserum, 10 l each of mK9, mK13, and mK22 (all 0.2 g/l), and 60 l of 25 mM Tris-HCl, pH 7.5) was rocked at 4°C overnight to prepare the specific anti-mK1 antiserum. The antisera against mK9, mK13, and mK22 were similarly treated. All of these antisera are designated as "mK-specific antiserum" in this paper. For a control experiment, each specific antiserum was further absorbed with its antigen, e.g. the mK1-specific antiserum (10 l) was mixed with mK1 (18 g) in 100 l of 25 mM Tris-HCl, pH 7.5, and then rocked at 4°C overnight. The specific antisera against mK9, mK13, and mK22 were also similarly treated. Each of these control antisera is hereafter designated as "pre-absorbed anti-mK antiserum." Preparation of Extract Containing Pro-IL-1␤ from Mouse Peritoneal Macrophages-Mice were killed by cervical dislocation on day 3 after the intraperitoneal injection of 1 ml of sterile thioglycollate medium, and their peritoneal cavity was then lavaged with 8 ml of Hanks' buffer (10 mM HEPES, pH 7.4, containing 137 mM NaCl, 5 mM KCl, 0.3 mM Na 2 HPO 4 , 0.4 mM KH 2 PO 4 , 5.5 mM glucose, 200 units/ml penicillin G, and 200 g/ml streptomycin). The recovered lavage fluid was pooled, and the peritoneal cells in it were collected by centrifugation (200 ϫ g for 10 min at 4°C). The cells were resuspended in RPMI 1640 medium containing 10% fetal bovine serum and plated in 100 mm-dishes (cells from 1 mouse per 1 dish). The macrophages were allowed to adhere for at least 2 h by incubating them at 37°C in an atmosphere of 5% CO 2 , after which the dishes were washed with fresh medium to remove the unattached cells. The cells were then cultured overnight.
Next the cells in culture were primed with 1 g/ml of LPS for 2 h and then washed with ice-cold phosphate-buffered saline (PBS) to remove any nonadherent cells and the fetal bovine serum. The adherent cells were scraped off the plates and mixed with lysis buffer (1 ml of lysis buffer/dish) consisting of 50 mM Tris-HCl, pH 7.5, 1% Triton X-100, 150 mM NaCl, and 0.1 mM EDTA. The lysates were placed on ice, sonicated, and centrifuged at 105,000 ϫ g for 30 min at 4°C. The post-mitochondrial supernatant fraction was recovered and stored at Ϫ84°C.
Animals, Treatments with Endotoxin, and Autonomic Agents-Male C3H/HeN mice 7-8 weeks old were purchased from Nippon SLC (Shizuoka, Japan) and housed under standard conditions (12-h light/12-h darkness cycle at 22-25°C) with free access to food and water. All animals were sacrificed for experiments at the age of 8 -9 weeks.
For endotoxin experiments, LPS dissolved in saline at 1 mg/ml and stored at Ϫ84°C was used. It was injected intraperitoneally at a dose of 10 g/mouse. For experiments using autonomic agents, 8-week-old male mice were anesthetized with pentobarbital (50 mg/kg, intraperitoneally) and injected intraperitoneally with normal saline or autonomic agents (pilocarpine, isoproterenol, or phenylephrine), each at a dose of 10 mg/kg. The saliva was collected, and the SMG were immediately excised as described below.
Preparation of Tissue Extracts and Collection of Saliva-From normal mice or mice injected with LPS or various autonomic reagents, the saliva was collected; and various tissues were isolated, including SMG, parotid gland (PG), lung, spleen, liver, kidney, and stomach. The saliva from LPS-stimulated mice was collected for 30 min at 6 h after injection of LPS (10 g/mouse, intraperitoneally) by means of the cotton pellet procedure described previously (17). From mice injected with autonomic agents, the saliva was collected for 60 min immediately after injection. The collected saliva was treated with 5% acetic acid. The tissues were homogenized in 9 volumes (w/v) of ice-cold homogenization buffer (50 mM Tris-HCl, pH 7.5, containing 20 mM EDTA, 10 mM PMSF, 2 g/ml aprotinin, and 1 tablet of Complete TM EDTA-free protease inhibitor mixture per 25 ml of the buffer) by using a glass mortar homogenizer fitted with a Teflon pestle. The homogenate was centrifuged at 12,000 ϫ g for 10 min at 4°C to remove the nuclei and cell debris. The supernatant was collected, and the protein concentration was determined by using a Bio-Rad protein assay kit with bovine serum albumin as a standard. The saliva and tissue extracts thus prepared were stored at Ϫ84°C until used in experiments.
Electrophoresis and Immunoblotting-For analysis of kallikreins and ICE, 10-g aliquots of the homogenate from the SMG, PG, lung, spleen, liver, and kidney or 10 l of saliva and macrophage extracts were mixed with an equal volume of 2ϫ SDS sample buffer and denatured by heat-ing at 95°C for 10 min. For IL-1␤ analysis, the same amount of samples mixed with the sampling buffer was incubated at 40°C for 1 h. All samples were resolved by 12% SDS-PAGE. The gel was soaked in transfer buffer for several minutes, and the proteins were transferred onto a nitrocellulose filter by means of a Mini Trans Blot apparatus (Bio-Rad) following the standard procedure. The filters were blocked at room temperature for 2 h with 3% horse serum (for analysis of IL-1␤ and ICE) or with 3% dry milk (for kallikrein analysis) in 0.2% T-TBS (10 mM Tris-HCl, pH 8.0, containing 150 mM NaCl and 0.2% Tween 20). They were next incubated at 4°C overnight with primary antibodies diluted in their respective blocking buffer. The following primary antibodies were used: goat (IgG) anti-recombinant mouse IL-1␤ (0.1 g/ml; Genzyme-Techne); polyclonal rabbit anti-mouse ICE (1/1000); rabbit polyclonal mK1-, mK9-, and mK13-specific antisera (1/25,000); and rabbit polyclonal mK22-specific antiserum (1/10,000). For control reactions, the filter was incubated with the same concentration of antibody/antiserum pre-absorbed with the respective antigen used for immunization. Filters were washed with 0.2% T-TBS, incubated with 3000 times diluted donkey anti-goat IgG-horseradish peroxidase (for IL-1␤ analysis) or with 3000 times diluted donkey anti-rabbit IgG-horseradish peroxidase (for ICE and kallikrein analysis), and washed with 0.2% T-TBS. Finally the filters were reacted with ECL and exposed to an x-ray film for the appropriate time.
Analysis of mRNA Expression by RT-PCR-Total RNA was extracted from the SMG and LPS-stimulated macrophages by using Tri-Reagent TM . RT-PCR was performed following the standard protocols described previously (17) by using a Superscript One-step RT-PCR kit. The amplification of mRNAs for IL-1␤ and ␤-actin consisted of RT reaction by incubation at 45°C for 30 min, denaturation at 94°C for 2 min, and PCR amplification. PCR amplifications after the RT reaction were set as follows: denaturation at 94°C for 30 s, annealing at 55°C for 45 s, and extension at 72°C for 1 min for either 35 (IL-1␤) or 22 cycles (␤-actin). At the end of each PCR, an extra extension at 72°C for 5 min was performed. A control PCR without the RT reaction step was run for each pair of primers by using total RNA to ensure no amplification of genomic DNA. RT-PCR without RNA was also conducted as another negative control. These reactions were sequentially conducted in a Takara PCR Thermal Cycler MP, model TP3000 (Takara Biomedicals, Shiga, Japan). The PCR products were separated by electrophoresis in 3% agarose gel (NuSieve, SeaKem ϭ 3/1) following the standard procedure. A mouse pro-IL-1␤ cDNA (nucleotides 71-776 of GenBank TM / NCBI Data Bank accession number M15131) was amplified by RT-PCR from total RNA by using the following primer set, 5Ј-AGCAGCTATG-GCAACTGTTC-3Ј (sense) and 5Ј-CTCTGCAGACTCAAACTC-CAC-3Ј (antisense). Similarly, a mouse ␤-actin cDNA (nucleotides 1-1892 of GenBank TM /NCBI Data Bank accession number NM007393) was amplified by using 5Ј-ACCCACACTGTGC-CCATCTA-3Ј (sense) and 5Ј-CGGAACCGCTCGTTGCC-3Ј (antisense) as the primers.
Restricted Hydrolysis of Pro-IL-1␤ and Its Peptide Segment by Tissue Kallikreins-Ten-microliter extracts from LPS-stimulated macrophages containing 15 g of total protein and pro-IL-1␤ were incubated with various concentrations of tissue kallikrein family enzymes (0.006 -0.6 g) at 37°C for 1 h in 20 l of 25 mM Tris-HCl buffer, pH 7.5. The incubation was terminated by adding 20 l of 2ϫ SDS sampling buffer and was followed by another incubation at 40°C for 1 h. The hydrolytic products were then analyzed by immunoblotting. Similarly, for the time course study, 0.02 g of mK13 was mixed with 5 l of extract of LPSstimulated macrophage, which contained 43.3 fmol of pro-IL-1␤. The reaction mixture (20 l) was incubated at 37°C for 0, 0.5, or 1-3 h.
To examine the effect of inhibitors of ICE and mK13, we mixed the enzyme (1.5 g/ml mK13 or 50 units/ml ICE) with the inhibitors and preincubated the mixture at room temperature for 30 min (final volume, 10 l). Ten microliters of the extract of LPS-stimulated macrophage was then added to the mixture, which was subsequently incubated at 37°C for 3 h. Thereafter, the reaction was terminated by the addition of 20 l of 2ϫ SDS sampling buffer. For determination of the amount of IL-1␤ formed, the band intensity was quantified by using NIH Image software.
To determine the exact position of hydrolysis of pro-IL-1␤ by tissue kallikreins, we mixed 20 g of a peptide segment corresponding to amino acid residues 107-121 of mouse pro-IL-1␤ ( 107 WDDDDNLLVCDVPIR) with 1 g of kallikrein mK13 in a 100-l reaction mixture and incubated it at 37°C for 1 h. The reaction was stopped by adding an equal volume of 1% trifluoroacetic acid, and the precipitated protein was then removed by centrifugation. The supernatant obtained was subjected to HPLC analysis. Namely, the sample was applied to a CrestPak C18S reverse phase-HPLC column (inner diameter 4.6 ϫ 150 mm; Japan Spectroscopic Co., Ltd., Tokyo, Japan), and the column was eluted with a linear gradient of acetonitrile (10 -60%) in 0.1% trifluoroacetic acid at 1 ml/min. The peptide peaks that appeared were collected and dried. Their sequences were analyzed by a protein sequencer, Model PPSQ-10, Shimadzu (Kyoto, Japan).
Detection of IL-1␤ by ELISA-Levels of IL-1␤ in the lysate of LPSstimulated macrophage, saliva, and SMG of normal mice were assessed by ELISA according to the manufacturer's protocol. The minimal detection level was Յ7 pg/ml, and the assay was specific for mouse IL-1␤.
Immunohistochemistry-For IL-1␤ immunostaining, SMGs were fixed and embedded as described (17). For staining of kallikrein mK13, mice were anesthetized, and a fixative (3% paraformaldehyde and 0.1% glutaraldehyde in 50 mM Na-P i buffer, pH 7.4) was circulated throughout the body for 5 min by injection via the left ventricle. The SMGs were dissected and immersed in the same fixative at 4°C for 2 h. The fixed tissue was washed at 4°C overnight in PBS, pH 7.4, containing 6.8% sucrose. These samples were dehydrated in 100% acetone at 4°C for 1 h, and subsequently embedded in Historesin Plus by following the manufacturer's protocol. Sections of 2 m thickness were cut and mounted on 3-aminopropyltriethoxysilanecoated slides. They were dried at 37°C for 2 h and stored at Ϫ20°C until used for immunostaining.
For staining of kallikrein mK13, the tissue sections described above were washed in PBS and incubated in 0.05% trypsin in PBS for 10 min to expose the antigen epitope. Next they were blocked by incubation with 5% normal goat serum, washed three times with PBS, for 5 min each time, and then incubated with 25,000 times diluted mK13-specific antiserum at 4°C overnight. After being washed three times with PBS, the sections were reacted with 200 times diluted goat anti-rabbit IgG (H ϩ L) FITC and washed with PBS. For IL-1␤ staining, the sections were blocked with 5% normal donkey serum and incubated with 0.2 g/ml goat anti-IL-1␤ (Santa Cruz Biotechnology) at 4°C overnight. The sections were next reacted with donkey anti-goat IgG (H ϩ L) FITC. For control staining, sections were incubated with pre-absorbed anti-mK13 antiserum. All sections were next incubated for 15 min at room temperature with PBS containing 0.1 g/ml propidium iodide and 20 g/ml RNase A and then washed with PBS to allow the nucleus to become stained. These stained sections were examined under a fluorescence microscope equipped with a DXM 1200 digital camera (Nikon, Tokyo, Japan). Fig. 1 shows the tissue distribution of mouse IL-1␤ as revealed by Western blotting using anti-IL-1␤ antibody (Genzyme-Techne). This antibody strongly detected 17.5-and 20-kDa bands of IL-1␤ in the SMG, which are the same size as active/secretory forms of this cytokine (22,23), thus suggesting that they were produced from pro-IL-1␤ in this tissue. These smaller sizes of the IL-1␤ bands were not detected in tissues other than the SMG. A result similar to that for the SMG was obtained in the experiment in which a saliva sample from LPS-injected mice was assessed (Fig. 1A), again supporting the idea that 17.5-and 20-kDa bands would be secreted from the SMG after having been processed from their precursors.

Expression of 17.5-and 20-kDa IL-1␤ in Mouse Tissue-
On the other hand, a strong 35-kDa band and a less intense 28-kDa band were found in the lane containing the extract from LPS-stimulated macrophages, as reported previously (24). An IL-1␤ immunoreactive band showing a molecular mass of 35 kDa was also seen in the case of lung and spleen, although the level was low; but no such band was detected in the lane containing the SMG extract. All these bands were specific, and they disappeared by pre-absorption of the antibody with the recombinant IL-1␤ (Fig. 1B). These data were confirmed by ELISA, in which the relative level of IL-1␤ in each tissue gave roughly parallel values (Table 1). Furthermore, the pro-IL-1␤ mRNA was detected at high levels not only in the LPS-stimulated macrophages but also in the SMG by RT-PCR analysis (Fig. 2). Because the active form of IL-1␤ is known to be produced from pro-IL-1␤ by protein processing (7,8), it is conceivable that the pro-IL-1␤ protein produced in the SMG was immediately processed to give 17.5-and 20-kDa IL-1␤.
Expression of ICE in Mouse Tissues-ICE, a cytoplasmic cysteine protease that cleaves inactive 35-kDa pro-IL-1␤ to generate the active 17.5-kDa mature IL-1␤ (9, 10), is expressed in many tissues as an inactive pro-enzyme polypeptide consisting of 404 amino acids with a relative molecular mass of 45 kDa (9, 10). To understand the ICE expression in the SMG, we investigated the protein levels of this enzyme by Western blotting and compared the results with those obtained from other tissues (Fig. 3). As shown in Fig. 3A the precursor of ICE, p45, was expressed at a relatively high level in the lung, spleen, liver, kidney, stomach, and in LPS-induced macrophages, but it was expressed at only a minimal level in the SMG. The p10 subunit, one of the functional ICE subunits, was not detected in any tissue but was found in LPS-stimulated macrophages, which sample showed a very faint p10 band. The specificity of these bands was confirmed by pre-absorption of the antibody with the blocking peptide (Fig. 3B). These findings suggest that ICE is probably not the major enzyme for pro-IL-1␤ processing in the SMG.
Expression of Kallikrein Family Enzymes, mK1, mK9, mK13, and mK22, in Mouse Tissues-In the male mouse SMG, many types of serine proteases are expressed; they are ␣and ␥-subunits of nerve growth factor (NGF), epidermal growth factor (EGF)-binding protein, ␥-renin, and ␤-NGF endopeptidase, and others (18), most of which are members of the kallikrein family. By Western blotting, we confirmed that large amounts of kallikreins mK1, mK9, mK13, and mK22 were detectable in the SMG, but we found that these proteases were hardly present in the lung, spleen, liver, kidney, stomach, and nonstimulated or LPS-stimulated macrophages (Fig. 4). All bands that appeared in the case of the SMG extract were specific, because they were not seen when pre-absorbed anti-mK antiserum was used (data not shown).

ELISA of pro-IL-1␤ and/or IL-1␤ in various mouse tissues
Tissue extracts were subjected to ELISA after 100 -1000 times dilution. The A 450 nm value of a standard curve of the present assay gave 0.336 -1.493 corresponding to 0 -1000 pg/ml recombinant IL-1␤. The A 450 nm value of diluted samples ranged between 0.479 and 0.665, values of which were computed to IL-1␤ concentration and divided by protein concentration. The mean values of the two independent measurements are shown.

Tissue Concentration
Saliva 181 a Submandibular gland a Concentration is given in pg/l. b Concentration is given in pg/g protein. Processing of Pro-IL-1␤ by Kallikreins mK1, mK9, mK13, and mK22-To determine whether members of the kallikrein family could process pro-IL-1␤, we incubated extracts of LPS-stimulated macrophages (containing pro-IL-1␤) with kallikreins mK1, mK9, mK13, and mK22, and we examined the ability of these enzymes to process pro-IL-1␤ by Western blotting using anti-mouse recombinant IL-1␤ antibody (Genzyme-Techne). As shown in Fig. 5, pro-IL-1␤ was degraded by kallikreins mK1, mK13, and mK9, but not by mK22, when LPS-stimulated macrophage extracts were incubated with these enzymes. Furthermore, pro-IL-1␤ (35 kDa) was converted to protein species with molecular masses of 28, 22, and 17.5 kDa by mK13 dose-dependently (Fig. 5C). When pro-IL-1␤ was incubated with mK9, IL-1␤ with molecular masses of 26 and 16 kDa appeared, depending on the amount of mK9 added (Fig. 5B). mK1 degraded pro-IL-1␤ to 25-26 kDa, and a very small amount of 17.5-kDa material was generated at the highest enzyme concentration used (Fig.  5A). We did not see any bands smaller than 28 kDa when pro-IL-1␤ was incubated with mK22 (Fig. 5D). All these bands were specific, because pre-absorption of the antibody with the recombinant IL-1␤ prevented their appearance (data not shown). These data therefore suggest that mK13 was responsible for the production of 17.5-kDa IL-1␤ seen in the SMG and saliva. When pro-IL-1␤ was incubated with 0.3-3 g/ml of mK13, IL-1␤ of intermediate size (22 kDa) appeared. The 22-kDa IL-1␤ protein is therefore supposedly an intermediate product during the generation of 17.5-kDa IL-1␤. On the other hand, no 20-kDa IL-1␤ seen in the SMG and saliva was produced in our in vitro experiments in which the four tissue kallikreins were tested. Therefore, some enzyme(s) other than these kallikreins might have been responsible for the production of the 20-kDa IL-1␤ detected in the SMG.
We next examined the possibility that mK13 may have processed the precursor of ICE present in the macrophage extract, leading to its activation and hydrolysis of pro-IL-1␤. As shown in Fig. 6A, the production of 17.5-kDa IL-1␤ was not inhibited when macrophage extracts were incubated with mK13 in the presence of caspase-1 inhibitor I or caspase-1 inhibitor VIII (5 and 2.5 M, respectively). The inhibitor concentrations used in this experiment were effective to inhibit the ICE activity (Fig. 6B). PMSF at 50 M completely inhibited the production of 17.5-kDa IL-1␤ by mK13, and it did not inhibit the ICE activity (Fig. 6, A  and B). These experiments clearly indicate that mK13 had acted directly on pro-IL-1␤ and processed it to give the 17.5-kDa IL-1␤. ICE activation by mK13 was therefore not likely to have taken place.
The rate at which 17.5 kDa IL-1␤ was produced by mK13 was determined by using a saturating concentration of pro-IL-1␤ (1.5 ng/20 l, being equivalent to 2.2 nM) and 0.02 g of mK13 (Fig. 7). The rate of IL-1␤ formation was constant for 3 h under the present conditions, and the turnover number thus determined was 290 mol/min/mol enzyme (10.8 pmol/min/mg enzyme).

FIGURE 4. Western blot analysis of tissue kallikreins in various mouse tissues.
Extracts of various mouse tissues were analyzed by Western blotting using specific antisera for mK1, mK9, mK13, and mK22, as described in the text. A, mK1-specific antiserum; B, mK9-specific antiserum; C, mK13-specific antiserum; D, mK22-specific antiserum. Lu, lung; Sp, spleen; Li, liver; Ki, kidney; St, stomach; sM, extract of LPS-stimulated macrophages; SG, submandibular gland. Protein amounts applied were 10 g for lung, spleen, liver, kidney, stomach, and stimulated macrophages, and 1 g for submandibular gland. Arrowheads indicate the location of tissue kallikrein bands. To determine the position at which mK13 cleaved pro-IL-1␤ to produce 17.5-kDa IL-1␤, we incubated mK13 and the pro-IL-1␤ peptide segment 107 WDDDDNLLVCDVPIR, a sequence where processing takes place, to generate 17.5-kDa IL-1␤. Kallikrein mK13 hydrolyzed this peptide segment into two peptide fragments, which were separated by reverse phase-HPLC (Fig. 8). As indicated in Fig. 8B, the sequences of these two peptides were determined by using a protein sequencer, and they were shown to be 107 WDDDDNL and 114 LVCDVPIR. These results suggest that mK13 hydrolyzed the bond between 113 Leu-114 Leu of pro-IL-1␤. Human 17.5-kDa IL-1␤ (Val 114 ) generated by cathepsin G, elastase II, chymase, or chymotrypsin (25) was reported to be fully active in the EL4 assay, a standard IL-2 induction bioassay (26). All these data suggest that pro-IL-1␤ synthesized in the SMG was processed to an active form of IL-1␤ in this gland by the action of mK13.
Co-localization of IL-1␤ and Kallikrein mK13 in the SMG-Because mature IL-1␤ and mK13 proteins were detected in the SMG and pro-IL-1␤ was suggested to be converted to the active form by mK13, we investigated the cellular localization of these proteins by immunohistochemical staining (Fig. 9). As shown in Fig. 9, A and D, the granular convoluted tubular (GCT) cells of the SMG showed a positive immunoreaction for both IL-1␤ and mK13. High power magnification of a part of the same section showed that both IL-1␤ and mK13 were colocalized in the secretory granules of these cells (Fig. 9, B and E). Such immunoreactions were not seen when the section was incubated with pre-absorbed anti-mK13 antiserum (Fig. 9C) or with pre-absorbed antirecombinant IL-1␤ antibody (Fig. 9F), indicating the specificity of the present staining. Other cells, such as acinar cells, did not show a positive reaction. These results suggest that IL-1␤ and kallikrein mK13 are exclusively co-localized in the secretory granules of the GCT cells.
The granular components of the secretory granules in the GCT cells are known to be reduced by injection of ␣-adrenergic agents (27). To confirm the above data showing that mK13 and IL-1␤ were co-localized in the granules in the GCT cells, we examined the effects of various autonomic agents on the secretion of mK13 and IL-1␤. By Western blot analysis, both mK13 and IL-1␤ were shown to be increased in the saliva and decreased in the SMG after an injection of phenylephrine, an ␣-adrenergic agent (Fig. 10, A and B). The mice injected with saline, pilocarpine (cholinergic agent), or isoproterenol (␤-adrenergic agent) did The pro-IL-1␤ peptide segment 107 WDDDDNLLVCDVPIR was incubated with kallikrein mK13 at 37°C for 1 h, and the reaction was stopped by adding an equal volume of 1% trifluoroacetic acid. The sample was then treated as described under "Materials and Methods" and subjected to HPLC as described there. The sequences of the two resulting peptide fragments were determined. A, 4 g of original peptide; B, 4 g of original peptide incubated with 0.2 g of kallikrein mK13; C, 0.2 g of kallikrein mK13.  not show such changes. These data support the finding that both mK13 and IL-1␤ were co-localized in the secretory granules of GCT cells and the probability that kallikrein mK13 processed pro-IL-1␤ in the SMG to produce the active/secretory form of IL-1␤.

DISCUSSION
Several studies have demonstrated that kallikrein mK13 is a serine protease that cleaves pro-renin, resulting in active renin; therefore it is designated as pro-renin-converting enzyme (28,29). In this study, we found that kallikrein mK13 was able to process pro-IL-1␤ to its active form, 17.5-kDa IL-1␤. IL-1␣ and IL-1␤ are macrophage-derived cytokines involved in the regulation of the immune system. Although mature IL-1␣ and the precursor of IL-1␣, pro-IL-1␣ (5,6), are both biologically active, pro-IL-1␤ is inactive and requires processing for its biological activity (6,23,26). Sufficient production of active IL-1␤ appears to require the following two separate steps: 1) a priming stimulus, e.g. LPS, to promote synthesis of the pro-cytokine, and 2) a secretion stimulus to initiate post-translational processing and/or release (30). In this study we obtained the first data showing the presence of mature IL-1␤ (17.5 and 20 kDa) in the SMG and saliva of LPS-injected mice.
In our experiments, 17.5-and 20-kDa bands for the IL-1␤ protein were detected only in the case of the SMG of normal mice, whereas the 35-kDa band obtained for other tissues, such as the lung and spleen as well as for LPS-stimulated macrophages, was not detected in the lane containing the SMG extract, which agrees with the results of others (24,31). The same 17.5-and 20-kDa bands for IL-1␤ were also detected when the saliva collected from LPS-injected mice was examined. Neither pro-IL-1␤ protein nor mature IL-1␤ protein was detected in the PG of normal mice, suggesting that the majority of 17.5-and 20-kDa IL-1␤ in the saliva originated from the SMG and not from the PG, although the PG is also known to produce IL-1␤ (32, 33).
The mouse SMG contains a number of bioactive substances such as NGF, EGF, and renin, as well as the proteases that are involved in the processing/formation of these factors. Some of these proteases have been identified as members of the kallikrein gene family and are abundant in the glands of male or female mice (19,20). In the present study, we confirmed that these tissue kallikreins were strongly expressed in the SMG of C3H/HeN mice as well. Western blot analysis of the digestion products of extracts of LPS-stimulated macrophage (which contained pro-IL-1␤) formed by each of the four SMG kallikreins suggested that only kallikrein mK13, the processing enzyme of pro-renin (34), generated the 17.5-kDa IL-1␤ following processing of pro-IL-1␤.
To determine whether mK13 gives the most rapid hydrolysis rate and therefore is likely to be the most physiologically significant tissue kallikrein among SMG kallikreins, we separated these enzymes in the SMG extract by isoelectric focusing and detected the tissue kallikrein peaks by their ability to hydrolyze a synthetic substrate (N ␣ -benzoyl-L-arginine ethyl ester). The activity to hydrolyze pro-IL-1␤ to form 17.5-kDa IL-1␤ and the immunoreactivity toward the mK13-specific antiserum or other antisera were measured. Only kallikrein corresponding to mK13 gave 17.5-kDa IL-1␤. When the activity to form 17.5-kDa IL-1␤ was determined by conducting a time course study, we found that such an activity in the SMG extract was 15.3 fmol/ min/mg extract protein.
The role of ICE in the release of mature IL-1␤ has been extensively studied (7,8). Cells that do not express ICE are able to release mature IL-1␤ upon transfection with ICE cDNA (23,35,36). In the present study, the protein level of ICE in the SMG was minimal compared with that in LPS-stimulated macrophages and other epithelial tissues. This finding suggests that ICE may not be the major enzyme responsible for pro-IL-1␤ processing in the SMG. Rather, our data indicate that kallikrein mK13 is responsible in the processing of pro-IL-1␤ in the SMG.   of mK13 in B). Ten micrograms of SMG extract and 1/150 volume of saliva collected over a 1-h period after injection were applied for IL-1␤ analysis, and the respective amounts were 0.2 g and 1/3000 volume for mK13 analysis.
Actually this is not unusual, as some proteases other than ICE are reported to process pro-IL-1␤ (25,37).
It was possible that mK13 cleaved pro-IL-1␤ at the 113 Leu-114 Leu bond in vitro, because the pro-IL-1␤ peptide segment 107 WDDDD-NLLVCDVPIR was hydrolyzed by mK13 at this position. It was reported previously that kallikrein mK13 specifically cleaves the peptide bond on the carboxyl side of the Arg at the Lys-Arg or Arg-Arg pair of mouse Ren2 pro-renin to yield mature renin (34,38). As a serine proteinase, the substrate specificity shown here appears to be unusual and has not been widely reported. However, there is a report describing a similar enzyme (39), suggesting that catalysis by certain serine proteinases actually involves hydrolysis of a Leu-Leu bond. In the SMG, the production of a large amount of pro-IL-1␤ protein, the precursor of IL-1␤, was strongly suggested because its mRNA was present at a high level in this tissue. Thus, we speculate that pro-IL-1␤ protein synthesized in this tissue would be immediately processed to active IL-1␤ and secreted into the saliva if the SMG were stimulated with LPS (17).
In the present study, we found by Western blotting that a 20-kDa form of IL-1␤ was present in the SMG and saliva besides the 17.5-kDa IL-1␤. None of the four tissue kallikreins tested produced this size of IL-1␤ from macrophage pro-IL-1␤, indicating that these kallikreins were not involved in production of this 20-kDa form. The enzyme involved in the formation of this molecular species should be present in the SMG, and so we are searching for it by a biochemical procedure.
By both immunohistochemistry and pharmacological/biochemical experiments, we demonstrated the co-localization of IL-1␤ and tissue kallikrein mK13 in the secretory granules of GCT cells. These results raise the following questions. 1) How does pro-IL-1␤, which lacks a conventional secretory sequence, enter the secretory granules? 2) How is pro-IL-1␤ processed specifically to the 17.5-kDa form in secretory granules in which many other kallikreins are also co-localized? Related to the first query, the mechanism as to how IL-1␤ is secreted is still unclear. Three possible mechanisms have been proposed for the secretory mechanism of IL-1␤ in human monocytes. The first mechanism is simple cell lysis leading to the release of cytoplasmic IL-1␤ (40). The second mechanism involves bleb formation and subsequent shedding of membrane vesicles (41). In the third mechanism, fusion with endolysosome-related vesicles is proposed (42). In this third mechanism, the existence of IL-1␤ transporters on the vesicle membrane has been suggested. Although no evidence is yet available for it at present, we speculate that a pathway like the last one may be involved in the packaging of IL-1␤ in the secretory granules of GCT cells in the SMG. We do so because IL-1␤ was obviously localized in the secretory granules in this tissue and a similar localization has been suggested for the parotid gland (32), which is also an exocrine gland.
Regarding the second query, there is a possibility that the mature IL-1␤ that appeared after cleavage may have been bound to its processing enzyme and then stored in the secretory granules. Such speculation would seem reasonable, because a similar mechanism has been proposed and is very much likely for the processing and storage of EGF and NGF. These molecules are also localized in the same secretory granules as their processing kallikreins and are known to be formed from their precursors by these enzymes (43,44). The mature growth factors bind with their processing enzymes (EGF binds mK9, whereas NGF binds mK3 and mK4) and are stored within the secretory granules (43,44). The possibility that a similar mechanism exists for mK13/IL-1␤ system needs to be determined in the future.