Abstract
Precursors of α-defensin peptides require activation for bactericidal activity. In mouse small intestine, matrilysin colocalized with α-defensins (cryptdins) in Paneth cell granules, and in vitro it cleaved the pro segment from cryptdin precursors. Matrilysin-deficient (MAT−/−) mice lacked mature cryptdins and accumulated precursor molecules. Intestinal peptide preparations from MAT−/− mice had decreased antimicrobial activity. Orally administered bacteria survived in greater numbers and were more virulent in MAT−/− mice than in MAT+/+ mice. Thus, matrilysin functions in intestinal mucosal defense by regulating the activity of defensins, which may be a common role for this metalloproteinase in its numerous epithelial sites of expression.
One role of the mucosal epithelium is to function as an active barrier against the external environment. Secretion of antibiotic peptides by epithelial cells appears to be an important component of innate immunity (1). The α- and β-defensins comprise a family of cationic peptides that kill bacteria by membrane disruption (2, 3). Granulocytes and several epithelial tissues (4), including the Paneth cells of the small intestine of most mammals (3, 5,6), produce α-defensins as prepropeptides. In mice, Paneth cell α-defensins are termed cryptdins (crypt defensins) (5–7), of which six distinct peptides have been isolated and characterized (6). Cryptdins and other antimicrobial molecules (lysozyme and secretory phospholipase Α2) are released from secretory granules in response to bacteria or cholinergic agents (6,8). The pro segment of cryptdins maintains them in an inactive state (9); the protease that mediates removal of the pro domain, thereby regulating the level of functional peptides, has not been identified.
Previously, we showed that mRNA for matrilysin (MMP-7), a matrix metalloproteinase (MMP), localizes to Paneth cells in mice harboring a conventional microflora (10). Matrilysin expression parallels that of the cryptdins (11) in that it is produced postnatally and is abundant in the unchallenged adult animal (10). Rather than being secreted basally, matrilysin is released at the apical cell surface and is detectable within the crypt lumen (12) (Fig. 1A). No staining for matrilysin was observed in Paneth cells of matrilysin-deficient (MAT−/−) mice (Fig. 1B). MAT+/+ and MAT−/− crypts exhibited identical staining patterns for cryptdin-1, indicating that the absence of matrilysin does not affect cryptdin precursor expression (Fig. 1, C and D). Other MMPs expressed in the mouse small intestine, including collagenase-3 (MMP-13) and stromelysin-2 (MMP-10), were not seen in Paneth cells of MAT+/+ animals (13).
Immunolocalization of matrilysin and cryptdin-1 in MAT+/+ and MAT−/− small intestinal crypts. Frozen sections of small intestine from MAT+/+ (A and C) and MAT−/− (B and D) mice were stained with polyclonal antibodies against matrilysin (20) (A and B) and cryptdin-1 (7) (C and D) as described (12). The blue indicates positive staining; red staining identifies the nuclei. Arrowheads indicate clusters of Paneth cells at the base of adjacent crypts. Photographs were taken with Nomarski optics. Scale bar, 25 μm.
Matrilysin prefers a leucine-arginine dipeptide sequence in the right-hand (P1′ and P2′) position of cleavage sites (14). The NH2-terminal residue of cryptdins 1 to 3, 5, and 6 is a leucine, and, as deduced from the cDNA sequences, this amino acid is predicted to be the NH2-terminal residue in all cryptdin family members (15) (Fig. 2A). Except for cryptdin-5, all cryptdin genes code for an arginine immediately after the NH2-terminal leucine in the mature peptide (15). Finally, matrilysin itself does not have antimicrobial activity (16), supporting a role for it in procryptdin processing.
Cleavage of cryptdin precursors by matrilysin. (A) The amino acid sequence alignment of procryptdin-1 (PC-1), procryptdin-15 (PC-15), and the recombinant procryptdin chimera (proCC) is shown. Dots denote residues of identity with PC-1. Residues 40 and 41 are the known or deduced NH2-terminal amino acids of processed cryptdins. (B) One microgram each of PC-1 and proCC (18) was incubated for 24 hours with 1 μg of catalytically active recombinant human matrilysin (+Mat) (Chemicon) in assay buffer [10 mM Hepes (pH 7.4), 0.15 M NaCl, and 5 mM CaCl2] with or without 50 mM EDTA. Reactions were analyzed by tris-tricine SDS-PAGE (15% polyacrylamide). Proteins were visualized by staining the gel with GelCode Blue reagent (Pierce), which does not detect the pro segment. The legend to the right of the gel identifies the bands corresponding to matrilysin, the precursors PC-1 and proCC, and their cleavage products cryptdin-1 and CC, respectively. CC migrates less rapidly than cryptdin-1 because of the additional residues from cryptdin-4 and the 6xHis tag. (C) Matrilysin digestion of proCC was set up as in (B), but the molar ratio of enzyme to substrate was varied as indicated. After tris-tricine SDS-PAGE, proteins were transferred to polyvinylidene difluoride (PVDF) membrane, and reaction products (proCC and CC) were detected with an antibody to His (COOH-terminal) (Invitrogen) at a 1:5000 dilution. (D) A time course of proCC cleavage by matrilysin was analyzed as in (C).
To address this possibility, we tested the ability of activated human matrilysin to cleave the pro segment from recombinant cryptdin precursors. At a 1:2 molar ratio of enzyme to substrate, a ratio typically used in MMP-substrate reactions in vitro (17), purified procryptdin-1 (PC-1) (18) was converted to mature cryptdin-1 (Fig. 2B). This reaction was inhibited by EDTA, a chelator of metal ions required for MMP activity (Fig. 2B). Peptide sequencing confirmed that matrilysin cleaved PC-1 to produce authentic cryptdin-1 with an NH2-terminal sequence of LRDLV (19). To test if matrilysin recognizes heterologous cryptdin precursors, we expressed a procryptdin chimera (proCC) (18) consisting of procryptdin-15 sequence, the last three COOH-terminal residues of cryptdin-4 (PRR), and a 6xHis tag (Fig. 2A). proCC was also efficiently cleaved by matrilysin (Fig. 2B), and NH2-terminal amino acid sequencing verified that catalysis occurred at the same position as in PC-1. Increasing the molar concentration of substrate reduced the amount of proCC converted to CC peptide (Fig. 2C). Although only 50% of the precursor was cleaved by 8 hours (Fig. 2D), the in vitro reaction kinetics may not reflect the in vivo rate.
MAT−/− mice were used to explore the relation between matrilysin expression and processing of cryptdin precursors in vivo. When housed under specific pathogen-free conditions, these mice do not develop spontaneous infections by opportunistic microorganisms. In addition, expression of cryptdin genes is normal in MAT−/− mice (20) (Fig. 1). To assess if procryptdin processing is altered in MAT−/− mice, we analyzed small intestinal extracts by acid-urea polyacrylamide gel electrophoresis (AU-PAGE). In this acidic gel system, cryptdin peptides are the most rapidly migrating of mouse intestinal peptides and separate in a predictable pattern according to their overall positive charge (21, 22). Mature cryptdin peptides were present in MAT+/+ mice (Fig. 3A, bracket); however, there was no evidence of processed cryptdins in intestinal extracts of MAT−/− animals (Fig. 3A) (23). Further analysis of MAT−/− extracts by reversed-phase high-performance liquid chromatography (rpHPLC) verified that peptides in MAT+/+ samples eluting with retention times characteristic of cryptdins were absent from MAT−/−intestinal extracts (13). Finally, although both procryptdins and mature cryptdins were detected in isolated MAT+/+ Paneth cell granules, only precursors were found in MAT−/− granules (24). In MAT+/− small intestine, processed peptide levels were equivalent to those in MAT+/+ animals (13).
Lack of procryptdin processing in MAT−/− small intestine. (A) Acetic acid extracts of small intestine were prepared and analyzed by AU-PAGE as described (21, 22) without prior knowledge of genotype. Synthetic refolded cryptdin-4 (C-4), the most rapidly migrating of the mouse enteric defensins, was included in the gel as an α-defensin marker. The bracket indicates the position where mature cryptdins migrate. There was complete concordance between genotype and presence (MAT+/+ lanes) or absence (MAT−/−lanes) of detectable peptides. A band comigrating with the cryptdin-4 marker is not readily detected in the extracts, as this peptide is the least abundant of the characterized cryptdins. (B) Lyophilized MAT+/+ and MAT−/− intestinal extracts were dissolved in sample buffer, and equal amounts of protein were analyzed by tris-tricine SDS-PAGE (15% polyacrylamide). Recombinant PC-1 (0.5 μg) was included as a standard. Proteins were transferred to PVDF and immunoblotted with a polyclonal antibody against the pro segment of PC-1 (1:2000 dilution) (25). (C) Extracts from (B) were analyzed by SDS-PAGE (12% polyacrylamide) and immunoblotting with a polyclonal antibody against mouse matrilysin (20). The precursor (Promat) and mature (Mat) forms of matrilysin, at 28 kD and 19 kD, respectively, are indicated. (D) Granules isolated from 3500 crypts (MAT+/+ and MAT−/−) (35) were analyzed as in (C).
The lack of mature cryptdins in MAT−/− mice suggested that procryptdins would accumulate in these animals. Indeed, protein immunoblot analysis of intestinal extracts with an antibody against the cryptdin-1 pro domain (25) showed an increase in cryptdin pro forms in MAT−/− extracts (Fig. 3B). The immunoblot also confirmed that the absence of cryptdin peptides in MAT−/− animals is not due to diminished procryptdin synthesis. Together, the results demonstrate that processing of procryptdins in vivo is dependent on matrilysin and indicate that it occurs intracellularly. Furthermore, the observation that matrilysin cleaves the pro segment from cryptdin precursors in vitro suggests that this metalloenzyme directly activates procryptdins in vivo as well, although potential intermediates in the processing pathway cannot yet be ruled out. The presence of the active form of matrilysin in small intestinal homogenates (26) and acid extracts (Fig. 3C) and within Paneth cell granules (Fig. 3D) supports MMP-mediated intracellular processing.
To assess the biological effect of cryptdin peptide deficiency, we determined the antimicrobial activity of small intestinal extracts in radial diffusion assays that measure combined bactericidal and bacteriostatic activity (22). At protein concentrations at which MAT+/+ extracts showed maximal killing activity against Escherichia coli ML35, MAT−/− extracts were completely inactive (Fig. 4A). Zones of clearance with MAT−/− extracts were detected only with 100 μg of protein, a level 10 times as great as the highest MAT+/+ concentration tested (13). In another approach, intact crypts were isolated from MAT+/+ and MAT−/− small intestines (27), and Paneth cell degranulation was stimulated by the cholinergic agent carbamylcholine (CCh) (28, 29). AU-PAGE confirmed the presence of defensins in supernatants from CCh-treated MAT+/+ crypts; these peptides were absent in untreated supernatants (24). In a liquid-phase assay measuring bactericidal activity, MAT−/− crypt supernatants killed a defensin-sensitive strain of Salmonella typhimurium(30) less efficiently than MAT+/+ supernatants, even at the single-crypt level (Fig. 4B). Although other Paneth cell components released upon degranulation likely contribute to the antibacterial activity of MAT−/− crypt supernatants, the reduced activity of these supernatants is consistent with a deficiency in functional cryptdins.
Association of matrilysin deficiency with a decrease in microbicidal activity. (A) Total small intestinal protein extracts from MAT+/+ and MAT−/− mice were dissolved at concentrations ranging from 0.3 to 8.3 μg/μl in sterile 0.01% acetic acid for performance of agar diffusion antibacterial assays (22). Three microliters of each solution was applied to wells in solidified agarose containingE. coli ML35. Antibacterial activity (zone of inhibition) was determined by subtracting the radius of the well from the total radius of the zone of clearance. The results shown here are representative of several independent experiments. (B) One thousand logarithmic-phase S. typhimurium phoP bacteria (30) (estimated by optical density at 620 nm) in isotonic PIPES buffer were combined with 5 μl of CCh-stimulated secretion supernatants from the indicated number of crypts (29). After incubation for 1 hour at 37°C with shaking, CFUs were determined by plating in triplicate on nutrient agar. The mean number of bacteria killed is expressed as a percentage (±SD) of controls from which crypts were omitted. P ≤ 0.01 by Student's t test. (C) MAT+/+ and MAT−/− mice were inoculated intragastrically with 1 to 2 × 1010 CFUs of E. coli KBC-236 and killed 2 hours later for small intestine extraction and quantification of surviving bacteria in the small intestine (32). The data were obtained from two separate experiments, with a total of seven to eight mice of each genotype. We calculated the median, or 50th, percentile (denoted by bisected triangles) and analyzed the statistical significance of the data using the nonparametric Wilcoxon/Mann-Whitney Rank Sum Test (one-tailed). (D) To assess mouse survival after oral challenge with a virulent strain of S. typhimurium (ATCC 14028s), we inoculated MAT+/+ and MAT−/− females (C57BL/6 background) at about 6 weeks of age with 2.89 × 105 CFUs as described (32). LD50 values were determined with the moving average interpolation method (33).
To ascertain the effects of cryptdin deficiency in vivo, we assessed the fate of exogenous bacteria in MAT+/+ and MAT−/− small intestine. The indicator strain of bacteria, KBC-236, is a kanamycin-resistant E. coli derivative expressing type-1 adherent pili that mediate binding to mucosal epithelium (31). MAT+/+ and MAT−/−mice were orally infected with KBC-236, and the number of viable microorganisms remaining in the small intestine was determined by growth on selective media (32). Because these bacteria are noninvasive in the intestine (16), in contrast to enteric pathogens such as Salmonella, they are particularly appropriate for quantification of bacterial survival. Infected mice were killed 2 hours after inoculation to minimize changes in bacterial numbers by cell division. The median number of viable KBC-236 recovered from the proximal small intestine, where Paneth cell density is relatively low, was about the same for both MAT+/+ and MAT−/− mice (Fig. 4C) and represented only a small percentage of the input bacteria. In contrast, the median number of bacteria recovered from MAT−/− mid and distal segments was higher than that from MAT+/+ mice (Fig. 4C). In both MAT+/+ and MAT−/− mice, bacterial recovery increased from segment to segment along the proximal-distal axis, excluding the possibility that bacterial transit is altered in MAT−/− mice. In addition, little difference in bacterial numbers was noted at 30 min after inoculation, when the majority of bacteria were still in the proximal intestine (13). Thus, despite the presence of other antimicrobial molecules, survival of exogenous bacteria in the small intestine is enhanced in the absence of matrilysin. In addition, the oral 50% lethal dose (LD50) (33) of a virulent, invasive strain of S. typhimurium for MAT−/− mice was one-tenth that for MAT+/+ mice (Fig. 4D). MAT−/− mice succumbed more rapidly to Salmonella infection than did MAT+/+ animals at a bacterial dose slightly above the LD50 for MAT+/+ mice (Fig. 4D). Although we cannot exclude a potential matrilysin-dependent systemic mechanism, these results support our conclusion that epithelial expression of matrilysin is important in restricting bacterial colonization and access to the intestinal mucosa.
These findings show involvement of an MMP in host defense against bacteria and extend the activity of matrilysin beyond that of connective tissue remodeling. The absence of any defect in matrix turnover in MAT−/− mice and the ability of matrilysin to cleave nonmatrix substrates (34) support a role for this metalloenzyme in protein processing. The wide spectrum of constitutive matrilysin expression in different organs suggests that the enzyme has a common function among mucosal tissues. The role we have described for matrilysin in the small intestine, specifically regulation of defensin activation, may be the shared function of this metalloproteinase in other epithelia.
↵* To whom correspondence should be addressed. E-mail: wilson_c{at}kids.wustl.edu
