Matrix Metalloproteinase-7 Activation of Mouse Paneth Cell Pro-α-defensins

Small intestinal Paneth cells secrete α-defensin microbicidal peptides as mediators of innate enteric immunity. In mice, production of mature Paneth cell α-defensins, termed cryptdins (Crps), requires proteolytic activation of inactive precursors (pro-Crps) by the convertase matrix metalloproteinase-7. Proteolysis of mouse (pro-Crp4)20-92 produces the specific cleavage intermediates pro-Crp444-92, pro-Crp454-92, and pro-Crp459-92. To identify which cleavage event enables bactericidal activity, recombinant pro-Crp4-processing intermediates were purified to homogeneity and assayed for bactericidal peptide activity. The in vitro bactericidal activities of pro-Crp4-processing intermediates were very similar to fully processed Crp4, contrasting the lack of bactericidal and membrane-disruptive activity shown by pro-Crp420-92. Thus, cleavage of pro-Crp420-92 at Ser43↓Ile44 is sufficient to activate bactericidal activity, and amino acids in the pro-Crp420-43 of the proregion maintain the precursor in an inactive state. Because cationic Arg residues are determinants of Crp4 bactericidal peptide activity, we hypothesized that Asp and Glu residues in pro-Crp420-43 neutralize Crp4 Arg side chains in pro-Crp420-92. Therefore, a pro-Crp420-92 variant with Gly substitutions at all pro-Crp420-43 Asp and Glu positions ((DE/G)-pro-Crp4) was prepared, and it was bactericidal and lysed phospholipid vesicles under conditions where native pro-Crp420-92 lacks activity. These findings show that MMP-7 proteolysis of pro-Crp420-92 at Ser43↓Ile44 converts inactive precursors to bactericidal forms by removal of covalently associated, inhibitory acidic amino acids from proximity with the Crp4 component of the molecule.

Paneth cell ␣-defensins contribute to innate mucosal immunity in the mammalian small intestine (1,2). For example, ␣-defensins account for ϳ70% of microbicidal peptide activity that is secreted from the base of small bowel crypts by mouse Paneth cells in response to cholinergic stimulation or exposure to bacteria and bacterial antigens (3). Defective dissolution of Paneth cell secretory granules in cystic fibrosis mice is associated with bacterial overgrowth, defective clearance of orally administered bacteria, and marked changes in the resident microflora (4,5). Also, mice lacking matrix metalloproteinase-7 (MMP-7), 2 the mouse Paneth cell ␣-defensin convertase (6,7), have impaired immunity to oral bacterial infections (6). Further evidence implicating ␣-defensins in Paneth cell secretions as mediators of innate enteric immunity derives from studies of mice transgenic for the human Paneth cell ␣-defensin HD-5. The transgenic mice are immune to oral infection by Salmonella enterica serovar Typhimurium (S. Typhimurium), a highly virulent pathogen for wild-type mice (8). These considerations provide rationale for determining the mechanisms regulating the production and functional maturation of these peptide effectors.
The bactericidal activity of mouse Paneth cell ␣-defensins, termed cryptdins (Crps), correlates directly with the ability to interact with and disrupt membranes. Most ␣-defensins kill bacterial cell targets by membrane disruption, although individual peptides differ in their mechanisms of action as shown by studies with model membrane systems in vitro (16,17). Crp4, the most bactericidal of the known mouse ␣-defensins, binds to biomimetic phospholipid membranes (18) and induces graded solute leakage from model membrane vesicles (19). In addition, the bactericidal activities of attenuated Crp4 peptide variants correspond directly with their in vitro membrane-disruptive activities (19,20). Full-length pro-Crp4 20 -92 lacks bactericidal activity in vitro (14), and the molecule binds poorly and induces only modest vesicle lysis (18,20). These findings show that the potent membrane disruptive behavior of mature Crp4 remains inactive in pro-Crp4 20 -92 until MMP-7 proteolysis cleavage enables binding and lysis of target cell membranes (9,14). To identify which cleavage event activates pro-Crp4, recombinant pro-Crp4-processing intermediates were prepared and tested for bactericidal and membrane-disruptive capabilities.
Here, we report that the Ser 43 2Ile 44 cleavage event is sufficient to convert inactive pro-Crp4 to a fully functional state, despite the presence of a 17-amino acid N-terminal extension. Furthermore, 9 acidic amino acid residue positions in pro-Crp4 20 -43 of the proregion are shown to maintain pro-Crp4 in an inactive state prior to MMP-7 proteolysis at Ser 43 2Ile 44 .
Preparation of Recombinant Proteins-Recombinant proteins were expressed as His 6 -tagged peptides and purified to homogeneity. Expression was induced by 0.1 mM isopropyl-␤-D-1-thiogalactopyranoside for 6 h at 37°C in Terrific Broth medium (19). Cells were lysed by sonication in 6 M guanidine-HCl, 100 mM Tris-HCl (pH 8.0) and clarified by centrifugation (14,18). The His 6 -tagged fusion peptides were affinity purified by nickel-nitrilotriacetic acid resin affinity chromatography and eluted with 1 M imidazole, 6 M guanidine-HCl, 100 mM Tris-HCl (pH 5.9) (14). Fusion proteins were reacted with 10 mg/ml CNBr in 80% formic acid for 18 h at 25°C, diluted with water, and lyophilized. Cleavage products were resuspended in 5% acetic acid and were separated by C18 reversed-phase highpressure liquid chromatography with a gradient of acetonitrile with 0.1% trifluoroacetic acid as an ion-pairing agent. Peptide homogeneity was confirmed by analytical AU-PAGE, and the molecular masses were determined by matrix-assisted laser desorption ionization mode mass spectrometry (Voyager-DE MALDI-TOF, PE-Biosystems, Foster City, CA).
Peptide Interactions with Large Unilamellar Vesicles-Recombinant pro-Crp4 and peptide variants were investigated for the ability to induce leakage from large unilamellar phospholipid vesicles (LUV) of defined composition. LUV of palmitoyloleoylphosphatidylglycerol (Avanti Polar Lipids, Birmingham, AL) were loaded with a fluorophore/quencher system (22,23). Aqueous lipid solutions consisting of 17 mM 8-aminonaphthalene-1,3,6-trisulfonic acid (ANTS, Molecular Probes, Eugene, OR), 60.5 mM p-xylenebispyridinium bromide (DPX, Molecular Probes), 10 mM HEPES, 31 mM NaCl, and 19.5 mM NaOH (260 mosM/liter, pH 7.4) were vortexed, frozen, and thawed for 5 cycles and then extruded through polycarbonate filters with 100 nm pores. Vesicles were separated from unencapsulated ANTS/DPX by gel-permeation chromatography with 130 mM NaCl, 10 mM HEPES, and 4.5 mM NaOH (260 mosM/liter, pH 7.4) as column eluant. Vesicular suspensions diluted with eluant buffer to ϳ74 M of total lipid were incubated with peptides at ambient temperature. Time-dependent fluorescence produced by ANTS release was monitored at 520 nm (excitation at 353 nm) as described previously (19,22). The kinetics of vesicular leakage was a function of peptide concentration, and equilibrium was attained Յ4 h. Thus, 4-h values were expressed relative to fluorescence obtained by vesicular solubilization with Triton X-100.
Pro-Crp4 20 -92 , pro-Crp4 44 -92 , and pro-Crp4 54 -92 molecules were shown to have correct C I -C VI , C II -C IV , and C III -C V disulfide connectivities by NMR spectrometry (Fig. 2), ruling out the possibility that peptide bactericidal activities could result from disrupted ␣-defensin cysteine pairings. Sequential assignments were made through each strand of the beta sheet using NOESY and TOCSY CH ␣ -NH connectivities. DQF-COSY and TOCSY data were used to determine the identity of each spin system and confirm backbone assignments. Sections of NOESY spectra that include the cysteine side chain ␣ and ␤ protons for pro-Crp4 20 -92 , pro-Crp4 44 -92 , pro-Crp4 54 -92 , and (DE/G)-pro-Crp4 (Fig. 2) confirmed that the disulfide connectivities correspond to those of native ␣-defensins. In addition, analysis of backbone NOEs demonstrates existence of a ␤-strand conformation with a register identical to that of Crp4. Therefore, all recombinant peptides investigated in these experiments were confirmed by NMR to contain tridisulfide arrays characteristic of ␣-defensins.

MMP-7 Proteolysis of Pro-Crp4 at Ser 43 2Ile 44 Confers Bactericidal
Activity-The activating cleavage event in pro-Crp4 processing was determined by analysis of recombinant pro-Crp4-processing intermediates. At 3 M final peptide concentration, pro-Crp4 44 -92 and pro-Crp4 54 -92 and Crp4 had equivalent bactericidal activities in in vitro bactericidal assays against E. coli ML35, V. cholerae, S. aureus 502a, L. monocytogenes, and S. Typhimurium ⌬phoP. The biological variability evident in bactericidal assays at low peptide concentrations may be due to steric effects of the N-terminal extensions in the pro-Crp4 44 -92 and pro-Crp4 54 -92 molecules. Characteristically (14), pro-Crp4 20 -92 was inactive or markedly less active than all peptides assayed at concentrations tested. The relative bactericidal activities shown in Fig. 3 against S. aureus 502A, L. monocytogenes, and S. Typhimurium ⌬phoP are representative of results for all bacteria assayed. The fact that the processing intermediates are as active as Crp4 and that native pro-Crp4 20 -92 lacks activity in these assays shows that all MMP-7-mediated cleavage events activate pro-Crp4. More importantly, however, the findings show that proteolysis at Ser 43 2Ile 44 , the cleavage step farthest from the Crp4 N terminus, is sufficient to induce full bactericidal activity. It follows that amino acids located between Asp 20 and Ser 43 must contribute to maintaining pro-Crp4 in an inactive state prior to conversion by MMP-7.
Membrane Binding and Disruption by Pro-Crp4-processing Intermediates-The mechanisms of pro-Crp4 44 -92 and pro-Crp4 54 -92 bactericidal action were investigated by measuring peptide binding to lipid/PDA-mixed vesicles and peptideinduced leakage from LUVs relative to Crp4 and native pro-Crp4 20 -92 . Previous studies have shown that phospholipid/ PDA assemblies undergo rapid blue-red chromatic transformations induced by interactions with membrane-active peptides (18,21,27), and peptides that localize at the lipid/ water interface of the membrane bilayer induce stronger surface perturbations and greater %CR values than peptides that insert into the hydrophobic bilayer core (18,21). Pro-Crp4 44 -92 , pro-Crp4 54 -92 , and mature Crp4 caused similar colorimetric dose-response curves, reaching maximum %CR values of ϳ40% (Fig. 4A). Therefore, these three peptides interact comparably with the phospholipid head-group

Proregion Acidic Amino Acids Inhibit Pro-Crp4 20 -92
region of vesicles, disrupting membranes to similar extents (18). In contrast, native pro-Crp4 20 -92 interacts only weakly with lipid/PDA vesicles as the low colorimetric response indicates (Fig. 4A), again as previously shown (18). Overall, the lipid/PDA binding results are consistent with peptideinduced LUV leakage assays (Fig. 4B) and with relative in vitro bactericidal peptide activities (Fig. 3). Also, these experiments confirm that Crp4 and pro-Crp4 intermediates have similar membrane-binding and disruptive activities in contrast to full-length pro-Crp4 20 -92 , which interacts poorly with lipid vesicles. The bactericidal activities of Crp4 and Crp4 mutants correlate well with the membrane-perturbing activity of the peptides in vesicle leakage assays (19,20).
To test the hypothesis that bactericidal pro-Crp4-processing intermediates act by membrane-perturbing mechanisms, peptide-induced vesicle leakage was measured and compared with the activities of pro-Crp4 20 -92 and Crp4 (19,20). Mature Crp4 induces extensive leakage from ANTS/ DPX-loaded anionic LUV (23), but native pro-Crp4 20 -92 has only slight membrane-disruptive capability (Fig. 4B). Consistent with their bactericidal activities, pro-Crp4-processing intermediates perturbed LUV membranes more than pro-Crp4 20 -92 (Fig. 4B), although the extent of pro-Crp4 44 -92 and pro-Crp4 54 -92 induced leakage was slightly less than that of mature Crp4. These results support the conclusion that Ser 43 2Ile 44 proteolysis enables or facilitates increased interactions between these peptides and phospholipid vesicles. Because the bactericidal and membrane-disruptive activities of Crp4 depend, in part, on cationic Arg residues (20), we considered that the Ser 43 2Ile 44 cleavage event may activate pro-Crp4 20 -92 by eliminating steric hindrance or charge neutralization imposed by the electronegative pro-Crp4 20 -43 region.
Acidic Amino Acids in Pro-Crp4 20 -43 Block Pro-Crp4 Bactericidal Activity-To investigate mechanisms that maintain pro-Crp4 in an inactive state, we hypothesized that dissociated carboxyl groups of Asp and Glu side chains in pro-Crp4 20 -43 interact with Arg guanidinium side chains when the pro-Crp4 20 -43 region is bound covalently to the precursor. If so, the electropositive Crp4 peptide surface that interacts with phospholipid head groups in the bacterial cell envelope would be neutralized, preventing membrane binding. To test whether the anionic Asp and Glu residues in pro-Crp4 20 -43 do inhibit pro-Crp4 20 -92 antibacterial action, a recombinant pro-Crp4 20 -92 variant with Gly substitutions at all Asp and Glu residues in pro-Crp4 20 -43 ((DE/ G)-pro-Crp4) was prepared (Fig. 5A). (DE/G)-pro-Crp4 peptide homogeneity was assessed by reversed-phase highpressure liquid chromatography and by AU-PAGE (Fig. 5B), and its mass was confirmed by MALDI-TOF MS (not shown).
Because bactericidal activity could potentially result from artifacts of peptide misfolding, NMR experiments were per-formed on (DE/G)-pro-Crp4 to determine the disulfide connectivities. Sections of NOESY spectra that include the cysteine side-chain ␣ and ␤ protons (Fig. 2) confirmed that the tridisulfide arrays of pro-Crp4 20 -92 , pro-Crp4 44 -92 , pro-Crp4 54 -92 , and (DE/G)-pro-Crp4 are the same as that of native Crp4. A comparison of chemical shifts of backbone NH groups shows that there are no differences Ͼ0.05 ppm between (DE/G)-pro-Crp4 and Crp4 resonances. Thus, the Crp4 domain in (DE/G)pro-Crp4 is correctly folded.
The effects of Asp and Glu residue positions on the inhibition of pro-Crp4 20 -92 bactericidal activity were inferred by determining the in vitro bactericidal peptide activity of (DE/G)-pro-Crp4. Assays performed against Gram-positive and Gram-negative bacterial species showed that deleting acidic side chains in pro-Crp4 20 -43 produced a (DE/G)-pro-Crp4 molecule with the same antibacterial activity as Crp4 (Fig. 6), particularly at high peptide concentrations. Assay variability increased at low peptide levels, perhaps due to steric effects of the 39-amino acid (DE/G)-pro-Crp4 N-terminal extension or some yet undefined structural factor. These findings do not rule out possible inhibitory contributions of additional prosegment amino acid residues that remained unchanged in (DE/G)-pro-Crp4. Despite the presence of a full-length, though charge-modified, prosegment, (DE/G)-pro-Crp4 was lethal to bacteria (Fig. 6). Native pro-Crp4 20 -92 was predictably inactive at almost all concentrations tested, except for low pro-Crp4 activity observed at 15 mM of peptide.
These experiments also demonstrate that variant Crp4 molecules, even molecules with as many as 41 amino acids extending beyond the natural ␣-defensin N terminus, retain bactericidal activity when the defensin fold is correct and peptide charge is not neutralized by residues in the proregion. Furthermore, this biochemical evidence supports the conclusion that the Asp and Glu residues in pro-Crp4 20 -43 maintain native pro-Crp4 20 -92 in an inactive state until MMP-7 catalyzes the Ser 43 2Ile 44 cleavage.
(DE/G)-pro-Crp4 Binds to Membrane Vesicles and Disrupts Model Membranes-Studies of many Crp4 peptide variants have shown a direct correspondence between relative peptide bactericidal activity and an ability to bind to and induce leakage from model membrane vesicles. To test whether (DE/G)-pro-Crp4 bactericidal action correlates similarly with membranedisruptive behavior, peptide binding to phospholipid/PDA vesicles and peptide-induced LUV leakage were assayed and compared with native Crp4 and pro-Crp4 20 -92 (Fig. 7). Replacement of pro-Crp4 20 -43 Asp and Glu residues with Gly markedly improved the association of pro-Crp4 20 -92 with lipid/PDA vesicles, as evident from the 30% difference in %CR plateaus of native pro-Crp4 20 -92 and (DE/G)-pro-Crp4 (  onic membrane vesicles was similar to that of Crp4, greatly exceeding leakage induced by native pro-Crp4 20 -92 exposure (Fig. 7B). Collectively, these results support the conclusion that acidic amino acid side chains in the N-terminal region of pro-Crp4 20 -92 prevent the precursor's Crp4 moiety from interact-ing with anionic membranes, thereby blocking bactericidal activity by inhibiting interactions with the bacterial cell envelope.
Recombinant Pro-Crp4 44 -92 , Pro-Crp4 54 -92 , and (DE/G)pro-Crp4 Permeabilize Live E. coli ML35 Cells-The relative membrane disruptive activities of pro-Crp4 44 -92 , pro-Crp4 54 -92 , and (DE/G)-pro-Crp4 also were compared with Crp4 and pro-Crp4 20 -92 in E. coli cell permeabilization assays performed using 6, 3, and 1.5 M peptide concentrations. At all peptide concentrations, permeabilization induced by pro-Crp4 44 -92 and pro-Crp4 54 -92 were comparable to Crp4 controls (Fig. 8). (DE/G)-pro-Crp4 induced 86% the extent of Crp4 permeabilization activity at 60 min of peptide exposure but was the same as Crp4 by 120 min. The relative permeabilization data shown in Fig. 8 for 6 M peptide levels are representative of all peptide concentrations and are consistent with results of bactericidal peptide assays and phospholipid vesicle perturbation studies. Evidence of a low level of pro-Crp4 20 -92 -induced E. coli permeabilization is apparent after 1 h (Fig. 8). Because low bactericidal activity can be detected at 15 M pro-  L. monocytogenes (B), and S. Typhimurium ⌬phoP (C ) were exposed to the peptide concentrations shown at 37°C in 50 l of PIPES-TSB buffer for 1 h (see "Experimental Procedures"). Following peptide exposure, the bacteria were plated on TSB-agar and incubated overnight at 37°C. Surviving bacteria were counted as CFU/ml at each peptide concentration. Plate counts below 1 ϫ 10 3 CFU/ml signify that no colonies were counted. Symbols: pro-Crp4 20 -92 (E), Crp4 (F), pro-Crp4 44 -92 (), and pro-Crp4 54 -92 (ƒ). The activities of MMP-7-processing intermediates were consistent against all bacteria tested. mole ratio) were exposed to the peptides (see "Experimental Procedures") to test peptide-membrane interactions. The relative increase in %CR is proportional to the extent of bilayer disruption. B, increasing peptide concentrations were exposed to ANTS/DPX-loaded anionic vesicles and leakage of ANTS fluorophore was measured at 520 nm from quenched conditions. Replicate assays of membrane activity of each peptide correlated with bactericidal activity. Symbols: pro-Crp4 20 -92 (E), Crp4 (F), pro-Crp4 44 -92 (OE), and pro-Crp4 54 -92 (‚).
Crp4 20 -92 concentrations (Fig. 6), it may be that the native pro-Crp4 20 -92 population may exist in equilibrium between partially membrane disruptive and inhibited forms.
To determine whether mutagenesis of pro-Crp4 20 -43 had effects on activation of bactericidal activity, Crp4, pro-Crp4, and (DE/G)-pro-Crp4 were exposed in vitro to MMP-7 and assayed for bactericidal activity against E. coli ML35. Mock digests lacking the convertase provided appropriate controls (Fig. 9B) (30,31). As anticipated, the bactericidal activity of Crp4 digests was unaffected by incubation with MMP-7, and MMP-7 digestion of native pro-Crp4 20 -92 activated bactericidal activity to a level equivalent to native Crp4 (14,18). Also, MMP-7 digests of (DE/G)-pro-Crp4 were as active as mock digests of the peptide, consistent with the activity of both forms being attributable to the Crp4 component of the precursor. Previous studies had established that the pro-Crp prosegment lacks antimicrobial activity in our in vitro assays (7), and MMP-7 itself is not bactericidal (6,14). Because MMP-7 digests of pro-Crp4 20 -92 still contain the pro-Crp4 20 -43 , pro-Crp4 44 -53 , and pro-Crp4 54 -58 cleavage fragments, effective inhibition of pro-Crp4 bactericidal activity by the prosegment requires covalent association of pro-Crp4 20 -43 with the Crp4 molecule.

DISCUSSION
The pro-Crp4 proregion maintains the precursor molecule in an inactive state until MMP-7-mediated cleavage occurs at one of three sites. Recombinant peptides corresponding to MMP-7-processing intermediates have in vitro bactericidal activities that are equivalent to that of native Crp4, identifying MMP-7 proteolysis at Ser 43 2Ile 44 in pro-Crp4 20 -92 as sufficient for activation. (DE/ G)-pro-Crp4, prepared by substituting acidic amino acids N-terminal of Ser 43 with Gly, has bactericidal activity equivalent to that of Crp4, and (DE/G)-pro-Crp4 is processed normally by MMP-7 in vitro. In membrane binding and vesicle leakage assays, the native pro-Crp4-processing intermediates and (DE/G)-pro-Crp4 have membrane-disruptive activities similar to Crp4, sharply contrasting that of inactive pro-Crp4 20 -92 . The active Crp4 intermediates appear to function by inducing graded leakage similar to Crp4 and not by alternative mechanisms potentially resulting from mutagenesis. Preliminary bactericidal peptide assays of (DE/NQ)-pro-Crp4 20 -92 , a pro-Crp4 variant in which pro-Crp4 20 -43 Asp and Glu residues were respectively converted to uncharged Asn and Gln, have shown that (DE/NQ)-pro-Crp4 and (DE/G)-pro-Crp4 are similar in activity. Thus, potential effects peculiar to Gly substitutions are excluded. 3 Collectively, these experiments strongly support the conclusion that acidic amino acids in pro-Crp4 20 -43 block the bactericidal action of pro-Crp4 by inhibiting membrane-disruptive behavior. It is the presence of acidic amino acids in the prosegment rather than the absolute length of the peptide extension bound to the Crp4 N terminus that inhibits pro-Crp4 membrane-disruptive behavior and microbicidal action.
NMR results confirm that the defensin domain folds correctly in the absence of acidic residues within the proregion. In fact, comparison of NMR chemical shifts shows essentially no change between defensin peak positions of pro-Crp4 44 -92 , pro-Crp4 54 -92 , (DE/G)-pro-Crp4, and Crp4. In contrast, the defensin domain of native pro-Crp4 20 -92 does show backbone chemical shift changes that are consistent with Crp4 interactions of the prosegment. 4 Thus the pro-region of the DE/G variant no longer masks the intact Crp4 ␣-defensin sufficiently to inactivate the molecule. We speculate that direct interactions between Arg residues, critical for Crp4 bactericidal activity (20), and acidic residues in the prosegment may be evident in the pro-Crp4 solution structure. Also, the amino acid composition of prosegment residue positions 40 -59 may contribute appropriate spacing and side-chain composition to provide optimal interactions between the anionic pro-Crp4 20 -43 region and the Crp4 peptide.
One possible role of the prosegment is to minimize potential cytotoxicity or membrane-disruptive behavior of ␣-defensin peptides in the biosynthetic pathway. For example, certain tumor cells and also nonmalignant cells undergo cytolysis when exposed to ␣-defensins from human and rabbit granulocytes (32,33). HNPs from polymorphonuclear leukocytes are cytotoxic to human lung fetal fibroblast MRC-5 cells, A549 lung adenocarcinoma cells, and human umbilical vein endothelial cell primary cultures (34). By inhibiting ␣-defensin-induced membrane disruption, the prosegment may prevent deleterious effects to the host cell during pro-␣-defensin biosynthesis, folding, and packaging into granules. Mouse Crp4 has markedly greater affinity for anionic phospholipid bilayers than for neu-   ratio) were exposed to the peptides (see "Experimental Procedures") to test peptide-membrane interactions. The relative increase in %CR is proportional to the extent of bilayer disruption. B, increasing peptide concentrations were exposed to ANTS/DPX-loaded anionic vesicles, and leakage of ANTS fluorophore was measured at 520 nm from quenched conditions. Membrane activity of each peptide correlated with bactericidal activity. Symbols: Crp4 (F), pro-Crp4 20 -92 (E), and (DE/G)-pro-Crp4 (f).
tral model membranes that more closely resemble mammalian plasma membranes. Nevertheless, the prosegment may inhibit ␣-defensin precursors from inserting into the cis-and trans-Golgi prior to being sequestered into secretory granules (7). Human Paneth cells store ␣-defensins as proforms, which are cleaved by trypsin after secretion into the gut lumen, a further suggestion that inhibition of defensin-mediated intracellular membrane disruption may be under positive selection. Although ␣-defensins in the gut lumen may potentially be cytotoxic to epithelial cells after secretion, preliminary data show that pro-Crp4 20 -92 and Crp4 are not cytotoxic to human fibroblast cell line HS68 at 25 M. 5 Although mouse Crps are released in millimolar concentrations at the point of secretion (3), peptide concentrations may dissipate rapidly as granules dissolve and peptides diffuse from the crypt compartment to the markedly greater volume of the small intestinal lumen above the crypt-villus boundary.
The pro-Crp4 prosegment efficiently inhibits the bactericidal activity of the C-terminal Crp4 ␣-defensin moiety only when present in cis and covalently bound in the precursor molecule. When pro-Crp4 20 -92 is cleaved by MMP-7 and tested for in vitro bactericidal activity, the combined products of the cleavage reaction are equivalent in activity to Crp4 peptide controls, even though the cleaved prosegment products remain in solution at equimolar concentrations. HNP-1 has been shown to interact with its soluble propeptide in trans as evidenced by decreased peptide fluorescence and coincident with a blue shift (35). At proregion concentrations of maximal prosegment inhibition of HNP-1, significant HNP-1 bacteriostatic activity was retained in comparison with full-length pro-HNP-1 (36). Crp4 also is inhibited by its prosegment when incubated in trans (7), but the proregion inhibition of Crp4 bactericidal activity is only 2-fold rather than the 1000-fold difference between Crp4 and native, full-length pro-Crp4 20 -92 (14). Although HNP-1 bound to soluble propeptide in trans and partially inhibited bactericidal activity, we speculate that components of the bacterial cell envelope may have greater affinity for the HNP-1 than for binding to prosegment in solution. Activated Crps and prosegments co-localize in Paneth cell granules, and the majority of pro-Crps are activated before or during granulogenesis (7), resulting in the secretion of both Crps and prosegment proteolytic fragments into the small intestinal lumen. Because MMP-7-generated prosegment fragments lose inhibitory activity once cleaved from the ␣-defensin peptide component of the proform, we infer that secreted Crps will be active following secretion despite the presence of potentially inhibitory proregion fragments. However, if post-translational processing of pro-␣defensins were defective, the covalently linked proregion would 5 C. S. Weeks, D. Tran, and A. J. Ouellette, unpublished observation. remain in proximity and inhibit Crp interactions with target cells. The phenotype of MMP-7-null mice is characterized by a lack of processed Crps and increased susceptibility to oral bacterial infection (6). We speculate that human populations with corresponding defects in pro-␣-defensin activation also may have impaired enteric immunity and be predisposed to mucosal infections and inflammation.