Activation of Paneth Cell α-Defensins in Mouse Small Intestine*

Paneth cells in small intestine crypts secrete microbicidal α-defensins, termed cryptdins, as components of enteric innate immunity. The bactericidal activity of cryptdins requires proteolytic activation of precursors by matrix metalloproteinase-7 (MMP-7; matrilysin) (Wilson, C. L., Ouellette, A. J., Satchell, D. P., Ayabe, T., Lopez-Boado, Y. S., Stratman, J. L., Hultgren, S. J., Matrisian, L. M., and Parks, W. C. (1999) Science 286, 113–117). Here, we report on the intracellular processing of cryptdin proforms in mouse Paneth cells. Peptide sequencing of MMP-7 digests of purified natural procryptdins identified conserved cleavage sites in the proregion between Ser43 and Val44 as well as at the cryptdin peptide N terminus between Ser58 and Leu59. Immunostaining co-localized precursor prosegments and mature cryptdin peptides to Paneth cell granules, providing evidence of their secretion. Extensive MMP-7-dependent procryptdin processing occurs in Paneth cells, as shown by Western blot analyses of intestinal crypt proteins and proteins from granule-enriched subcellular fractions. The addition of soluble prosegments to in vitro antimicrobial peptide assays inhibited the bactericidal activities of cryptdin-3 and -4 intrans, suggesting possible cytoprotective effects by prosegments prior to secretion. Levels of activated cryptdins were normal in small bowel of germ-free mice and in sterile implants of fetal mouse small intestine grown subcutaneously. Thus, the initiation of procryptdin processing by MMP-7 does not require direct bacterial exposure, and the basal MMP-7 content of germ-free Paneth cells is sufficient to process and activate α-defensin precursors. MMP-7-dependent procryptdin activation in vivoprovides mouse Paneth cells with functional peptides for apical secretion into the small intestine lumen.

In mouse Paneth cells, matrix metalloproteinase-7 (MMP-7 1 ; matrilysin) mediates the processing and activation of ␣-defensins from 8.4-kDa proforms (21). MMP-7 gene disruption ablates procryptdin processing, resulting in accumulation of cryptdin precursors and the absence of activated mature cryptdin peptides in the small intestine (21). Lacking functional cryptdin peptides, MMP-7-null mice have a defect in clearance of intestinal infections, and they succumb more rapidly and to lower doses of virulent Salmonella typhimurium compared with control mice (21). Thus, the cryptdin deficiency resulting from defective procryptdin activation is associated with a measurable deficit in mucosal immunity and increased risk of systemic disease.
In this study, cryptdin biosynthesis was investigated by characterizing details of intracellular procryptdin processing in mouse Paneth cells. The products of in vitro cleavage of procryptdin-1 and natural procryptdins by MMP-7, the localization of the cryptdin proregion in the exocytotic pathway, and the extent of procryptdin activation in Paneth cells of adult mice have now been characterized. Our results show that extensive intracellular procryptdin activation occurs in mouse Paneth cells and that exposure to bacterial antigens does not induce procryptdin processing in mice.

EXPERIMENTAL PROCEDURES
Animals and Tissue Preparation-All procedures on mice were performed in compliance with the policies of the Institutional Animal Care and Use Committee of the University of California, Irvine (UCI). 45-Day-old male outbred Swiss mice ((Crl:CD-1)(ICR)BR), 6-week-old adult male BALB/cJ and C57/BL6 mice, 6-week-old pregnant female BALB/cJ mice, and 6-week-old adult male germ-free Swiss mice were purchased from Charles River Breeding Laboratories, Inc. (Wilmington, MA). Matrilysin (MMP-7)-null mice were 6 -8-week-old males backcrossed for 10 generations into the C57/BL6 background. Mice were housed under 12-h cycles of light and dark and had free access to standard rat chow and water.
For preparation of fetal small intestine implants, pregnant BALB/cJ mice were killed at 15-17 days of gestation by injection with 600 g of Avertin (500 mg of tribromoethanol and 250 mg of 2-methyl-2-butanol in 39.5 ml of water)/g of body weight. Segments (1-2 cm) of proximal small intestine from each fetus were implanted aseptically under dorsal subcutaneous skin flaps of individual 6-week-old isogenic male BALB/cJ mice (22,23). Approximately 90% of the implants grew and were harvested for isolation of RNA or protein, or they were fixed by immersion in phosphate-buffered Formalin. Fixed tissue was processed into paraffin blocks; sectioned; and stained with hematoxylin/eosin by the Histology Laboratory of the Department of Pathology, University of California Irvine Medical Center.
Preparation of Small Intestine Crypts-Crypts were prepared by EDTA treatment of everted small intestine segments as described (15, 24 -26). Briefly, segments of adult mouse small bowel were agitated in buffered 30 mM EDTA (pH 7.4), and eluted crypts were deposited by centrifugation and resuspended in ice-cold Ca 2ϩ /Mg 2ϩ -free buffer. Enteric ␣-defensins derive exclusively from Paneth cells in crypts (15,27,28). Certain experiments were conducted with Protease Inhibitor Mixture Set III from Calbiochem, present in all buffers and solutions to test for the possibility of proteolysis during sample preparation. After protease inhibitors were shown to have no effect on procryptdin recovery or on the state of cryptdin activation (see Fig. 4C), experiments were conducted in the absence of inhibitors.
Extraction of Crypt Proteins-Peptides were prepared by extraction using 30% acetic acid (28,29). For analysis of peptides from cryptenriched fractions, crypts were resuspended in 30% acetic acid, sonicated, and extracted overnight at 4°C. Extracts were centrifuged for 15 min at 10,000 rpm in a Sorvall SA-600 rotor; supernatants were clarified by centrifugation for 2 h at 28,000 rpm in a Beckman SW 28.1 rotor; and high speed supernatants were diluted 10-fold and lyophilized (21).
Preparation of Paneth Cell Secretory Granules-Subcellular fractions enriched in Paneth cell secretory granules were prepared from duodenal and ileal crypts. Crypts deposited by centrifugation at 700 rpm for 5 min in a Beckman GS-6R centrifuge were resuspended in ϳ10 ml of ice-cold Ca 2ϩ /Mg 2ϩ -free phosphate-buffered saline (PBS; Invitrogen) at pH 7.5 and placed under N 2 at 750 p.s.i. for 15 min in a Model 1019HC nitrogen cavitation bomb (Parr Instrument Co., Moline, IL). Cell lysates produced by equilibration to atmospheric pressure were diluted 2-fold with PBS containing 5 mM EDTA and centrifuged at 700 ϫ g for 10 min at 4°C. Low speed supernatants were reserved, and the deposited cell debris was washed by resuspension in ice-cold Hanks' EDTA solution (Invitrogen) and centrifugation at 700 ϫ g for 10 min at 4°C. Granules in the combined supernatants were deposited by centrifugation at 27,000 ϫ g for 40 min at 4°C in the Sorvall SA-600 rotor, and granules in the high speed pellet were washed two to three times by resuspension and centrifugation in Hanks' EDTA solution under the same conditions. Granules were stored frozen or dissolved immediately in 30% acetic acid and extracted as described above.
Acid/Urea-Polyacrylamide Gel Electrophoresis-Lyophilized peptide samples were dissolved in 20 l of 5% acetic acid containing 3.0 M urea and electrophoresed on 12.5% acid/urea-polyacrylamide gels for 6 h at 150 V (29). Resolved proteins were visualized by staining with Coomassie Blue R-250 after fixation in Formalin-containing acetic acid/methanol. ␣-Defensins were identified by their rapid comigration with au-thentic mouse cryptdin peptides on acid/urea-polyacrylamide gels (Ͼ0.6 ϫ R F of methyl green dye) as described (30) and confirmed immunochemically by Western blotting (15,31).
Anti-cryptdin-1 Prosegment Antiserum-The cryptdin-1 prosegment corresponds to residues 19 -58 in preprocryptdin-1 as deduced from cryptdin-1 cDNA (32, 33) (see Fig. 1A). The prosegment (DPIQNTDEET KTEEQPGEDD QAVSVSFGDP EGTSLQEES) was synthesized by Quality Controlled Biochemicals, Inc. (Hopkinton, MA). The composition and concentration of the synthetic prosegment were determined by amino acid analysis on a Waters Model 2690 Alliance Analyzer, and its mass was verified by matrix-assisted laser desorption ionization timeof-flight mass spectrometry (MALDI-TOF-MS) on a Voyager-DE instrument (PE Biosystems, Foster City, CA) in the UCI Biomedical Protein and Mass Spectrometry Resource Facility. Quality Controlled Biochemicals, Inc. produced polyclonal anti-cryptdin-1 prosegment antiserum in a sheep by administering four dorsal subcutaneous injections of prosegment conjugated to bovine serum albumin in complete Freund's adjuvant. Injections were repeated twice, and the antiserum titer was evaluated by enzyme-linked immunosorbent assay by Quality Controlled Biochemicals, Inc. The primary structures of prosegments in all mouse defensin family precursors are highly conserved (see Fig. 1A), and the antibody is likely to cross-react with all mouse defensin family precursors. Rabbit antisera to the cysteine-rich sequence-1c (CRS1C-1) prosegment (34, 35) (see Fig. 1A) react with mouse Paneth cells specifically (36).
Immunolocalization of Cryptdins and Prosegments-Immunoperoxidase staining was performed by the Histology Laboratory of the Department of Pathology, UCI Medical Center. Paraffin sections of Formalin-fixed mouse small bowel were deparaffinized with xylenes, treated for 30 min with 0.3% H 2 O 2 , and washed with water and PBS. Slides were incubated three times for 5 min each in a microwave oven with antigen unmasking solution (Vector Laboratories, Inc., Burlingame, CA) and then cooled in unmasking solution (Vector Laboratories, Inc.) for 30 min at room temperature. After rinsing with PBS, sections were blocked by incubation with normal goat serum for 30 min and with avidin D blocking solution for 15 min, rinsed briefly with PBS, and then incubated with biotin blocking solution (Vector Laboratories, Inc.) for 15 min. Slides were incubated with a 1:100 dilution of sheep anti-cryptdin-1 prosegment immune antiserum or with serum from the sheep prior to immunization. After 30 min, slides were washed three times with PBS, incubated for 30 min with a 1:2000 dilution of biotinylated donkey anti-sheep IgG (28), and washed as described above. After 60 min of incubation with Vectastain ABC peroxidase reagent (Vector Laboratories, Inc.), slides were washed, flooded with diaminobenzidine, washed, and counterstained before mounting.
For immunogold co-localization of the mature peptide and cryptdin-1 prosegment, samples of jejunum were fixed with 2% formaldehyde and 0.1% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) and embedded in Unicryl (Ted Pella, Inc., Redding CA) at Ϫ20°C. Thin sections placed on Formvar and carbon-coated grids were stained with rabbit anticryptdin-1 antibody (28), washed, and reacted with a 1:25 dilution of protein A labeled with 10-nm gold (Ted Pella, Inc.) as described (27). Next, sections were incubated with sheep anti-cryptdin-1 prosegment immune IgG (1:200) overnight at 4°C. After washing, sections were incubated with donkey anti-sheep IgG conjugated to 20-nm gold. Sections were counterstained with uranyl acetate and lead citrate, examined in a Jeol 100 CX electron microscope, and photographed. Equivalent dilutions of preimmune sera provided negative controls in all experiments.
Purification of Mouse Procryptdins-Recombinant procryptdin-1 was prepared as described previously (21). Briefly, mouse procryptdin-1 cDNA cloned in pMalc2 (New England Biolabs Inc. Beverly, MA) was expressed as a maltose-binding protein fusion protein in Escherichia coli BL21(DE3) CodonPlus cells (Stratagene, La Jolla, CA) that was purified by amylose resin affinity chromatography. Procryptdin-1, released from the fusion protein by digestion with 1 g of Factor Xa (New England Biolabs Inc.)/mg of fusion protein at 30°C for 48 h, was separated from maltose-binding protein by C 4 RP-HPLC on a Vydac 214TP1010 column (Vydac, Hesperia, CA) and purified to homogeneity by analytical C 18 RP-HPLC on a Vydac 218TP54 column (21).
For purification of mouse enteric procryptdins, small intestine protein extracts were prepared from MMP-7-null mice by extraction with 30% acetic acid as described above. Protein samples were applied to analytical C 18 RP-HPLC columns (Vydac 218TP54) in aqueous 0.1% trifluoroacetic acid and eluted at ϳ35 min using a 10 -45% acetonitrile gradient developed over 55 min. Protein fractions containing apparent procryptdins were analyzed by acid/urea-polyacrylamide gel electrophoresis (AU-PAGE) as described (21,29). Procryptdins A-C were purified to homogeneity by C 18 RP-HPLC using a 120-min 10 -40% acetonitrile gradient, from which cryptdin precursors eluted between 18 and 30% acetonitrile (data not shown).
The identification of the purified proteins as cryptdin precursors was achieved by N-terminal sequencing and MALDI-TOF-MS. Peptide concentrations were determined using the Bradford assay (Bio-Rad), and the molecular masses of purified putative procryptdins were determined by MALDI-TOF-MS, followed by sequencing in the UCI Biomedical Protein and Mass Spectrometry Resource Facility.
MMP-7 Cleavage of Mouse Procryptdins in Vitro-Recombinant procryptdin-1 and natural procryptdins were digested with MMP-7 and analyzed by AU-PAGE and SDS-PAGE, and mixtures of proteolytic digests from MMP-7 cleavage were analyzed by N-terminal sequencing. Samples (1 g) of recombinant procryptdin-1 (21) and of natural procryptdins A-C purified from MMP-7-null mice were incubated with equimolar quantities of activated recombinant human MMP-7 catalytic domain (Chemicon International, Inc., Temecula, CA) in buffer containing 10 mM HEPES (pH 7.4), 150 mM NaCl, and 5 mM CaCl 2 for 24 h at 37°C. Reactions were analyzed on Tris/Tricine/SDS-15% polyacrylamide gels (Bio-Rad). Curiously, proregions or fragments of proregions were not seen by routine gel staining methods after digestion with MMP-7, even though procryptdins and cryptdin peptides stained well (see Fig. 3A). 2 Samples (ϳ200 ng) of complete digests were subjected to eight cycles of N-terminal peptide sequencing at the UCI Biomedical Protein and Mass Spectrometry Resource Facility.
In small intestine, cryptdin transcripts and peptides previously have been found only in Paneth cells (9, 15, 27, 28, 38 -42); and consistent with those findings, immunoperoxidase detection of the cryptdin-1 prosegment showed that it also is Paneth cell-specific (Fig. 1C). This finding is in agreement with immunolocalization of the related mouse CRS1C prosegment (Fig. 1A) using a rabbit polyclonal antibody to the CRS1C-1 proregion (36). 3 The reactive cryptdin prosegment antigen ap-peared to be associated with secretory granules, prompting immunolocalization studies at the electron microscopic level.
Cryptdin Precursors in Paneth Cell Secretory Granules-␣-Defensin prosegments as well as cryptdin peptides are constituents of mouse Paneth cell secretory granules. The subcellular location of cryptdin prosegments within Paneth cells of mouse mid-small bowel was determined using appropriate gold-conjugated protein A or second antibodies. As shown in Fig. 2 (A-C), both the anti-prosegment and anti-cryptdin peptide antibodies reacted strongly and specifically with Paneth cell granules. Preimmune negative control sera had very low back-2 D. P. Satchell and A. J. Ouellette, unpublished data. 3 A. J. Ouellette, unpublished data.  (Fig. 2C, inset). With both anti-prosegment and anti-cryptdin peptide antisera, the respective antigens first were detected in the trans-Golgi of the Paneth cell exocytotic pathway (Fig. 2B, inset). Cytoplasmic staining was highly specific for the electron-dense region of secretory granules (Fig.  2, B and C). The electron-lucent halos of Paneth cell granules, which contain high levels of O-linked GalNAc glycoconjugates (43), showed very little gold staining (Fig. 2, B and C). All granules were immunoreactive, and staining was uniformly equivalent regardless of subcellular localization, as shown by quantitation of gold particles over apical or supranuclear granules (Fig. 2D). Despite showing that the anti-prosegment antibody reacted with Paneth cell granules, these findings did not distinguish unprocessed procryptdins from soluble proregions generated by MMP-7 proteolysis of cryptdin precursors. To resolve this question, the products of MMP-7 hydrolysis of procryptdins in vitro and the status of procryptdin activation in Paneth cells in vivo were investigated in detail.
Specificity of in Vitro Procryptdin Cleavage by MMP-7-Because Paneth cell ␣-defensin processing intermediates had not been characterized, mouse procryptdins were purified from MMP-7-null mouse small intestine as substrates for analysis of the MMP-7 cleavage products. MMP-7-null mice are an optimal source for cryptdin precursor purification because cryptdin gene expression occurs at wild-type levels, and procryptdins accumulate in MMP-7-deficient Paneth cells (21). Also, studies of natural substrates avoid potential complications of analyzing possibly misfolded recombinant cryptdin precursors.
Putative mouse procryptdins A-C were purified to homogeneity by combined C 4 and C 18 RP-HPLC (see "Experimental epithelial cells (C). Paneth cells contain granules (G), an extensive system of rough endoplasmic reticulum (rER), and prominent Golgi (Gp). L, lumen of the crypt. Original magnification ϫ5375; bar ϭ 5 m. B, the boxed supranuclear region in A is shown at higher magnification to visualize gold-labeled structures after staining with rabbit anti-cryptdin antibody and gold (10 nm)-labeled protein A and sheep anti-prosegment antibody and gold (20 nm)-conjugated anti-sheep IgG. Note that the Golgi and granules in the Paneth cell were labeled with both the 10-nm (short arrows) and 20-nm (long arrows) gold probes, demonstrating that both peptides are present (inset). In contrast, the rough endoplasmic reticulum was not labeled. Gold labeling of the Paneth cell Golgi was much more extensive than that of the Golgi from adjacent undifferentiated crypt epithelial cells. Original magnification ϫ22,306; bar ϭ 1 m. C, a high magnification electron micrograph shows co-localization of the prosegment and cryptdin peptides in Paneth cell granules. This section was labeled with rabbit anti-cryptdin-1 antibody/gold (10 nm)-labeled protein A and sheep anti-prosegment antibody/gold (20 nm)-labeled anti-sheep IgG, and both 10-nm (short arrows) and 20-nm (long arrows) gold particles were present in the electron dense (Gd), but not electron lucent (Gl), zones of Paneth cell granules. The inset shows the relative lack of background staining in Paneth cell granules that were incubated with preimmune sera prior to incubation with 10-and 20-nm gold conjugates. Original magnification ϫ53,750; bar ϭ 0.5 m. D, sections labeled with rabbit anti-cryptdin-1 antibody/gold (10 nm)-labeled protein A were evaluated by counting gold particles in 38 apical and 40 supranuclear granules. The labeling density of apical and supranuclear granules was identical. Data are presented as means Ϯ S.E.
Procedures") (Fig. 3A). Candidate molecules were deduced to be cryptdin precursors based on elution times from C 18 RP-HPLC columns and comigration with recombinant procryptdin-1 on SDS-polyacrylamide gels (Figs. 1B and 3A) and acid/urea-polyacrylamide gels (data not shown). MALDI-TOF-MS of putative procryptdins A-C provided atomic masses of 8543, 8478, and 8277, respectively, values that did not correspond to previously deduced procryptdin sequences (32,33,38). Despite this apparent discrepancy, procryptdins A-C were shown to be ␣-defensin precursors by N-terminal sequencing, because they had N termini identical to that of procryptdin-1, DPIQNTD (Table I), the consensus N terminus of all known mouse procryptdins (32, 33, 38) (Fig. 1A). Analysis of procryptdins A-C by SDS-PAGE following cleavage with MMP-7 in vitro produced only one evident primary cleavage product of appropriate mobility for mature ␣-defensin peptides (Fig. 3A).
The peptide bonds cleaved in procryptdins A-C by MMP-7 were determined by direct N-terminal sequencing of MMP-7 digests of the precursors (Fig. 3B). For each putative procryptdin, only three N termini were detected besides that of the activated MMP-7 enzyme ( Fig. 3B and Table I). The first Nterminal sequence was DPIQNTD . . . , the consensus procryptdin N terminus (Table I). The second sequence was VSFGD-PEG . . . , an internal cleavage site between Ser 43 and Val 44 in the prosegment (Fig. 3B and Table I). The VS2VSFG sequence flanking the cleavage site (Fig. 3B, asterisk) within the prosegment is conserved in all mouse defensin family precursors (32,33,37) (Fig. 1A). The third sequence was LRDLV_Y_ . . . , where the underscore characters represent deduced cysteines, and that N-terminal sequence results from proteolysis between Ser 58 and Leu 59 in all related procryptdins (21) (Fig. 3B). LRDLV is the consensus N terminus for all cryptdin peptides, except cryptdin-4 and -5 (33). The masses determined for procryptdins A-C were not in concordance with known mouse procryptdins, perhaps because the MMP-7 knockout is in the C57/BL6 genetic background. Previous clones of procryptdin cDNAs and genes were from inbred C3H/HeJ and 129/SvJ mouse strains or from outbred Swiss mice, and these strains may have unreported proregion or cryptdin peptide amino acid substitutions that differ from those in C57/BL6 mice. Collectively, these results both confirmed procryptdins A-C as cryptdin precursors and defined an MMP-7-catalyzed processing site within the proregions of these precursors.
Activated ␣-Defensins in Mouse Paneth Cell Secretory Granules-The co-localization of prosegments and cryptdins in secretory granules (Figs. 1C and 2) prompted an evaluation of the processing status of cryptdin precursors in Paneth cell granules. The distribution of cryptdins and procryptdins in Paneth cell secretory granules was determined by AU-PAGE and Western blotting (15,31). On acid/urea-polyacrylamide gels, activated ␣-defensins were the most rapidly migrating intestinal peptides (see "Experimental Procedures"), and they are lacking in MMP-7-null mice (21) (Fig. 4A). Previously, only low levels of procryptdins were detected in secretions elicited from Paneth cells by carbamylcholine exposure (15). Partially purified Paneth cell secretory granules contained abundant activated cryptdins at levels equivalent to those in intact crypts (Fig. 4B, lane  2). Inclusion of a complex of potent proteinase inhibitors in all solutions and buffers during crypt isolation, granule sedimentation, and protein extraction (see "Experimental Procedures") had no effect on the apparent levels of activated cryptdins (Fig.  4C), a fact taken as evidence that procryptdin processing was not caused by experimental manipulation.
The relative distribution of cryptdins to procryptdins was evaluated by Western blot analysis of Paneth cell granule proteins from wild-type and MMP-7-null mice using an anti-cryptdin-1 peptide antibody (15,28). As predicted from previous analyses of whole mouse small bowel proteins (21), Paneth cell granules from MMP-7-null mice lacked rapidly migrating activated cryptdins, but contained high levels of procryptdins (Fig.  4D, lane 2). In contrast, granule proteins from wild-type C57/ BL6 mice gave strong immunoreactivity at the position of cryptdin mobility, where the signal strength was approximately twice that of the procryptdin region (Fig. 4D). From these considerations and because cryptdins gave weaker immunostaining upon Western blotting compared with equimolar quantities of procryptdins, 4 we estimate that 60 -70% of the procryptdins in Paneth cells are processed to functional peptides before secretion (see "Discussion"). Because extensive procryptdin processing is intracellular (Fig. 4), prosegments in granules (Fig. 2) are subject to secretion, suggesting that proregions might inhibit the bactericidal activities of activated cryptdin peptides.
Soluble Prosegment Neutralizes Cryptdin Bactericidal Activity in Vitro-The cryptdin-1 prosegment lacks antimicrobial activity, but it inhibited the bactericidal activity of mature cryptdin peptides in trans. The ability of the soluble cryptdin-1 propeptide, corresponding to residues 19 -58 in the cryptdin-1 precursor (Fig. 1A), to inhibit the activity of cryptdin-3 and -4 was tested in bactericidal assays against E. coli ML35 cells. In agreement with the inhibition of myeloid ␣-defensins by human neutrophil proregions (20), prosegment/cryptdin molar ratios of 0.5-1 or greater inhibited both peptides to approxi-4 M. E. Selsted, personal communication.

TABLE I MMP-7 cleavage sites in mouse cryptdin precursors
The molecular masses of putative natural cryptdin precursors were determined by MALDI-TOF-MS, and N-terminal sequences were determined by Edman degradation of samples before and after digestion with MMP-7 (see "Experimental Procedures"). N-terminal residue assignments are based on the primary structure of preprocryptdin-1 (Fig. 4B), with residue 1 at the initiating Met residue. ND, not determined. mately the same extent (Fig. 5). Perhaps as other authors have suggested (17), the acidic proregions (pI ϳ3.4) may neutralize the activity of the cationic defensins by charge neutralization. These in vitro inhibitory activities of secreted prosegments in trans are consistent with a possible cytoprotective role. Procryptdin Activation in Germ-free Mice-Because MMP-7 is required for procryptdin activation (21) (Fig. 4A), we evaluated the extent of cryptdin activation in Paneth cells of germfree mice to test whether their basal MMP-7 levels are adequate for cryptdin processing. Germ-free mice contain less Paneth cell MMP-7 than conventional mice, and monocolonization of germ-free mice with Bacteroides thetaiotaomicron induces expression of Paneth cell MMP-7 to levels found in mice harboring conventional microflora (36). Whether raised conventionally or germ-free, mouse intestinal extracts contained comparable levels of activated cryptdins (Fig. 6); and thus, sufficient MMP-7 exists under germ-free conditions to activate cryptdins normally. Although the mice were germ-free and consumed sterile chow, exposure to dietary bacterial antigens may have been responsible for inducing elevated MMP-7 levels, even in the germ-free state. For that reason, the level of cryptdin activation was examined in sterile implants of fetal mouse small intestine grown subcutaneously (22,23).
Activated Cryptdins in Implants of Fetal Small Intestine-To test whether procryptdin activation requires exposure to bacterial antigens, the state of cryptdin processing was investigated during Paneth cell ontogeny in BALB/cJ isogenic implants. In mice, the ontogeny of the small intestine epithelium occurs during the first 3 weeks postpartum (45,46). Interest- ingly, subcutaneous growth of fetal intestinal implants provides conditions that favor epithelial cell differentiation in structures that develop to resemble the morphology of normal adult small intestine (22,23,47). In our experiments, ϳ90% of the implants grew, and Paneth cells were evident at the base of crypts by ϳ12 days post-transplantation (PT12) as judged by hematoxylin/eosin staining. Paneth cell granules increased in number and size between PT12 and PT19 (Fig.  7A). Reverse transcriptase PCR amplification assays for Paneth cell-specific mRNAs in PT7 to PT28 implant RNAs detected lysozyme; MMP-7; and cryptdin-1, -4, and -5 mRNAs at all time points (Fig. 7B). AU-PAGE analysis of proteins extracted from implants removed on PT7 to PT19 showed that activated cryptdins were evident from PT12 onward, resembling adult levels by PT19. Protein extracts from PT7 implants and MMP-7-null mice lacked activated cryptdins (Fig. 7). Because Paneth cells in implants PT12 or older contain processed cryptdins, luminal exposure to bacterial antigens cannot be required to initiate procryptdin processing. Furthermore, Paneth cells that are naive to luminal bacterial antigen exposure contain MMP-7 in adequate quantities to provide functional cryptdins for secretion.

DISCUSSION
In mouse small intestine, a substantial fraction of procryptdin activation occurs in Paneth cells and prior to secretion. This conclusion is supported by evidence from electrophoretic and Western blot analyses of proteins extracted from subcellular fractions enriched in secretory granules, where 60 -70% of procryptdins exist already activated by MMP-7-dependent proteolytic cleavage (Fig. 3). This value for the fraction of processed precursors represents an overall average for all cryptdin precursors, with the exception of cryptdin-4 and -5, which do not react with the anti-cryptdin-1 antibody. Interpretation of these data is complicated, though, by the dynamics of crypt cell biology and by the ongoing processes of Paneth cell differentiation and granule biogenesis in the regulated secretory pathway. For example, our findings do not distinguish between the procryptdin activation state in mature granules poised for vesicular fusion at the Paneth cell apical membrane from that in nascent granules that are forming in the trans-Golgi. In addition, we cannot discount the possibility of post-secretory activation of the procryptdin molecules that are secreted. Also, Paneth cells differentiate in crypts over ϳ8 days as they emerge from the stem cell zone and descend to the base of the crypt (25). The extent of procryptdin processing in granules of maturing Paneth cells may differ relative to that in fully differentiated cells at the crypt base.
Questions remain regarding the biology of cryptdin prosegments. For example, the inhibition of cryptdin bactericidal activity by addition of the complete prosegment in trans (Fig. 5) is consistent with comparable dose-dependent inhibition of HNP-1 activity by the recombinant HNP-1 prosegment (pro-HNP-1-(20 -64)) (20), but paradoxical in view of the bactericidal activity of Paneth cell secretions (15). Perhaps, as suggested for myeloid ␣-defensin prosegments (20), cryptdin propeptides may interact with additional chaperones to neu- FIG. 7. Activated cryptdins in fetal mouse intestinal implants grown subcutaneously. A, implanted tissue removed 5-28 days after implantation (PT5 to PT28) was fixed in buffered Formalin, processed, and stained with hematoxylin and eosin (see "Experimental Procedures"). Arrows indicate granule-containing Paneth cells in crypts of developed implants. B, RNAs from PT7 to PT28 implants were amplified by reverse transcriptase PCR using primers specific for lysozyme; MMP-7; and cryptdin-1, -4, and -5 as reported previously (48,54). As in neonatal small bowel (48), the Paneth cell marker mRNAs were present in the implanted tissues prior to the appearance of recognizable Paneth cells. The Ad lanes contained products amplified from total RNA from adult mouse small bowel, and the W lanes contained equivalent samples of amplification reactions in which water was substituted for template RNA. C, samples (700 g) of implant protein extracts were analyzed by AU-PAGE as described in the legend to Fig. 6. Lanes contained proteins from implants taken 7 to 19 days after implantation (PT7 to PT19), intestinal protein extracts from MMP-7-null (؊/؊) and control wildtype (؉/؉) mice, and 1 g of cryptdin-3 (C1) and cryptdin-1 (C3) as noted. The boxed region of the gel shows the position of activated cryptdin peptides. tralize the potential membrane-disruptive activities of mouse ␣-defensins as they traverse the Paneth cell Golgi stack during granulogenesis. Because secretory granules containing activated cryptdins also react with anti-prosegment antibodies (36) (Figs. 1 and 2), the processed proregions or proregion fragments (Fig. 3B) are likely to be released along with activated cryptdins as Paneth cells degranulate. The high bactericidal peptide activity in Paneth cell secretions (15) suggests, however, that proregion inhibitory activity may be neutralized before or during secretion. Possibly, MMP-7-catalyzed proteolysis of proregions between Ser 43 and Val 44 , Ser 53 and Leu 54 (44), and Ser 58 and Leu 59 during precursor activation may eliminate the inhibitory capabilities of the complete 39-amino acid proregion tested in our studies. Also, MMP-7 cleaved procryptdins A-C reproducibly, but additional cleavage steps may exist, as suggested by the isolation of apparent procryptdin processing intermediates with LQEESLRDLV N termini from mouse small intestine (44). Those intermediates may be MMP-7 cleavage products that our sequencing experiments did not detect, or they may be procryptdin cleavage products of an MMP-7-dependent enteric protease(s) capable of cleaving the precursors in vivo. Interestingly, preliminary studies of MMP-7 digests of recombinant procryptdin-4 have not detected the proregion cleavage site between Ser 42 and Ile 43 . Instead, an abundant LHEKS N-terminal sequence was found, showing that MMP-7 cleaves procryptdin-4 between Ala 52 and Leu 53 , 5 a site that corresponds to the intermediates purified by Putsep et al. (44). Thus, in vitro, MMP-7 appears to be capable of generating all known cryptdin processing intermediates.
Paneth cells in germ-free mice have almost undetectable MMP-7 levels, as previously determined immunohistochemically (36). Nevertheless, the base-line level of MMP-7 suffices to ensure normal cryptdin activation (Figs. 6 and 7). Similarly, procryptdin processing in sterile intestinal implants shows that enough MMP-7 exists (Fig. 7B) to activate the pool of cryptdin precursors without microbial stimuli in the lumen (Fig. 7C). Although reverse transcriptase PCR amplification of implant RNAs detected lysozyme, MMP-7, and cryptdin mRNAs in all implants (Fig. 7B), cryptdin RNAs were not detected by Northern blot hybridization before PT12 (data not shown). Similar findings have been obtained in fetal and newborn mouse intestine, which also lacks Paneth cells prior to crypt ontogeny and where cryptdins accumulate in apparent secretory cells of the maturing epithelial monolayer (48).
Paneth cell differentiation inherent to small bowel development includes programmed mechanisms for procryptdin activation and the secretion of functional ␣-defensins without environmental cues from the lumen. The evidence in support of this conclusion does not exclude responses of Paneth cells or their progenitors to pro-inflammatory mediators, including tumor necrosis factor-␣ or interferon-␥, that may be released by neighboring epithelial cells or by stromal cells. In fact, Trichinella spiralis infection of mouse small intestine stimulates an increase in Paneth cell numbers as well as recruitment of intermediate cells to accumulate cryptdins in dense granules, and both outcomes are mediated by T lymphocytes (49,50). Similarly, Paneth cells increase rapidly in number when T cells are activated by CD3 ligation, and those events are partly dependent on tumor necrosis factor-␣ (51). Perhaps pro-inflammatory cytokines influence the inherent plasticity of the gastrointestinal epithelium by redirecting lineage determination programs in the short term and modulating Paneth cell numbers during inflammatory episodes. The responsiveness of MMP-7 biosynthesis and activation to pro-inflammatory cyto-kines (52,53) is consistent with this possibility. Regardless of the mechanisms regulating MMP-7 expression in Paneth cells, MMP-7-dependent procryptdin processing ensures the secretion of active ␣-defensins to facilitate innate mucosal immunity.