Characterization of Luminal Paneth Cell a-Defensins in Mouse Small Intestine ATTENUATED ANTIMICROBIAL ACTIVITIES OF PEPTIDES WITH TRUNCATED AMINO TERMINI*

Paneth cells at the base of small intestinal crypts secrete apical granules that contain antimicrobial peptides including a-defensins, termed cryptdins. Using an antibody specific for mouse cryptdin-1, -2, -3, and -6, immunogold-localization studies demonstrated that cryptdins are constituents of mouse Paneth cell secretory granules. Several cryptdin peptides have been purified from rinses of adult mouse small intestine by gel filtration and reverse-phase high performance liquid chromatography. Their primary structures were determined by peptide sequencing, and their antimicrobial activities were compared with those of the corresponding tissue forms. The isolated luminal cryptdins included peptides identical to the tissue forms of cryptdin-2, -4, and -6 as well as variants of cryptdin-1, -4, and -6 that have N termini truncated by one or two residues. In assays of antimicrobial activity against Staphylococcus aureus, Escherichia coli, and the defensin-sensitive Salmonella typhimurium phoP mutant, full-length cryptdins had the same in vitro antibacterial activities whether isolated from tissue or from the lumen. In contrast, the N-terminal-truncated (des-Leu), (des-LeuArg)-cryptdin-6, and (des-Gly)-cryptdin-4 peptides were markedly less active. The microbicidal activities of recombinant cryptdin-4 and (des-Gly)-cryptdin-4 peptides against E. coli, and S. typhimurium showed that the N-terminal Gly residue or the length of the cryptdin-4 N terminus are determinants of microbicidal activity. Innate immunity in the crypt lumen may be modulated by aminopeptidase modification of a-defensins after peptide secretion.

Studies of mice lacking matrilysin (matrilysin-deficient), MMP-7, 1 a metalloproteinase that activates Paneth cell procryptdins, have shown that the resulting deficiency in activated cryptdins impairs innate mucosal immunity (27). Matrilysindeficient mice lack activated cryptdins, are defective in clearing intestinal infections, and succumb more rapidly and to lower doses of virulent Salmonella typhimurium compared with wildtype mice (27). In addition to their continual release of granules into the lumen, Paneth cells also may be stimulated to degranulate by muscarinic receptor agonists that mobilize intracellular Ca 2ϩ stores (28,29) by general G-protein activators such as NaF/AlCl 3 (30 -32) and by live bacteria and bacterial antigens. 2 The objectives of these studies were to assess the distribution of cryptdins in mouse Paneth cell granules and to investigate the biochemical features of secreted cryptdins by determining the primary structures of cryptdins isolated from the mouse small intestinal lumen and by determining their antimicrobial activities. Relative to the first cysteine residue position in the ␣-defensins, mouse cryptdin N termini are 3-4 amino acids longer than myeloid ␣-defensin orthologs (3,16), and the possibility that the N-terminal extension is an adaptation for extracellular peptide function has been proposed (12,15). The results described here show that cryptdins are present in all mouse Paneth cell secretory granules and that the mouse small intestinal lumen contains full-length as well as N-terminaltruncated cryptdin variants. Cryptdin peptides with shortened N termini have attenuated antimicrobial activities, implicating N-terminal topology as a determinant of the bactericidal activity of these peptides.

EXPERIMENTAL PROCEDURES
Animals-Outbred Swiss mice [(Crl:CD-1)(ICR)BR] were 45-day-old males (30 -35 g) or timed-pregnant dams from Charles River Breeding Laboratories, Inc. (North Wilmington, MA). Mice were housed under 12-h cycles of light and dark and had free access to standard rat chow and water.
Immunogold Localization of Cryptdins-Intestinal tissues were fixed with 2% formaldehyde, 0.1% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4 and embedded in Unicryl at Ϫ20°C (33). Thin sections were placed on formvar and carbon-coated grids and stained with rabbit anti-cryptdin-1 antibody (16), washed, and then reacted with a 1:25 dilution of protein A labeled with 10 nm gold (Ted Pella, Inc., Redding, CA) as described in the legend to Fig. 2. The sections were examined in a JEOL 100 CX electron microscope and photographed.
Purification of Cryptdins from Small Intestinal Lumen-Peptides were isolated from the small intestines of adult mice by a modification of a previously described protocol (3,16). Jejunum and ileum from the small intestines of 60 outbred Swiss Webster mice were excised immediately after euthanasia. The lumens were flushed gently with 50 ml ice-cold water, and the rinses were diluted with ice-cold glacial acetic acid to a final concentration of 30% (v/v) acetic acid. Acidified intestinal rinses were clarified by centrifugation at 28,000 rpm for 30 min in the SW 28.1 rotor and lyophilized. Samples dissolved in 30% acetic acid and clarified by filtration through Whatman 541 filter paper were chromatographed on a 10 ϫ 60-cm Bio-Gel P-60 column in 30% acetic acid at a flow rate of 100 ml/h. Fractions containing cryptdins were identified by the presence of rapidly migrating peptides in acid-urea polyacrylamide gel electrophoresis (AU-PAGE), a characteristic of ␣-defensins in this system (34).
Luminal cryptdin peptides were purified to homogeneity by reversephase high performance liquid chromatography (RP-HPLC) (3,16). Pooled fractions containing cryptdins were lyophilized and separated initially on a 1 ϫ 25-cm Vydac C 18 RP-HPLC column using a gradient of water and acetonitrile with 0.13% heptafluorobutyric acid. Solvents were delivered at 3 ml/min generating the following acetonitrile gradient: 0 to 28% (10 min), 28 -34% (20 min), and 34 to 40% (60 min). Peptides that co-eluted under these conditions were lyophilized and resolved by C 18 RP-HPLC using water/acetonitrile with 0.1% trifluoroacetic acid. A 16 -21% acetonitrile gradient was delivered in 35 min at 3 ml/min. All peptides were lyophilized and quantitated by amino acid analysis prior to testing antimicrobial activity.
Peptide Sequencing-Samples (500 pmol) of individual peptides were S-pyridylethylated, desalted by RP-HPLC (16,35), and sequenced on an ABI model 477 system (American Biosystems, Inc., Foster City, CA) configured with on-line phenylthiohydantoin-derivative amino acid analysis in the UCI Biomedical Protein and Mass Spectrometry Resource Facility. Alkylated peptides were sequenced completely through to the C-terminal amino acid as previously reported for full-length cryptdins isolated from mouse small intestinal tissues (3,16) and for full-length synthetic cryptdins (16,36).
Antimicrobial Peptide Assays-In vitro antibacterial activities of cryptdins were performed in buffers containing 10 mM PIPES or 10 mM HEPES as described (3). Samples (5 l) of purified peptides dissolved at 3-300 g/ml in 0.01% acetic acid were pipetted into wells formed in plates of 1% agarose buffered with 10 mM PIPES, pH 7.4, and containing 1 ϫ 10 6 log-phase bacteria grown in trypticase soy broth. After 3 h at 37°C, plates were overlayered with 0.8% agarose containing 2ϫ trypticase soy broth and incubated overnight. Combined bactericidal and bacteriostatic activities were determined as the diameter of growth inhibition around each well as a function of peptide concentration. For measurements of bactericidal activity, exponentially growing bacterial cells were deposited by centrifugation, washed twice with 10 mM PIPES, pH 7.4, and 1 ϫ 10 6 bacteria were exposed to peptide solutions in the PIPES buffer for 1 h at 37°C (3,16). Incubation mixtures were diluted 1:2000 with PIPES buffer and plated using a Spiral Biotech Autoplate 4000 (Spiral Biotech, Inc., Bethesda, MD), and surviving colony forming units (CFU) were quantitated after overnight incubation at 37°C.
Preparation of Recombinant Cryptdins-Recombinant cryptdin-4 and (des-Gly)-cryptdin-4 were isolated after cloning and expression as maltose-binding protein (MBP) fusions using the pMAL bacterial expression system (New England BioLabs, Beverly, MA). A full-length cryptdin-4 cDNA clone was amplified with forward primers containing in the 5Ј-to 3Ј-orientation an EcoRI site, codons for the IEGR Factor Xa (Xa) protease cleavage site, and the 5 N-terminal codons of the mature cryptdin-4 or (des-Gly)-cryptdin-4 peptides. Those primers were paired with a reverse primer complementary to the 5 C-terminal codons of cryptdin-4, including the stop codon, and a 5Ј-flanking SalI site. The sequences were pMalCrp4F (5Ј-GCGCGAATTCATCGAGGGAAGGGG-TTTGTTATGCTATTGT) pMal-desGCrp4F (5Ј-GCGCGAATTCATCGA-GGGAAGGTTGTTATGCTATTGT), and pMalCrp4R (5Ј-ATATATGTC-GACTCAGCGACAGCAGAGCGTGTACAATAAATG). After plasmid constructs were verified by DNA sequence analysis, plasmids were introduced into Escherichia coli BL21(DE3)pLysS cells, and synthesis MBP fusion proteins were affinity purified from cell lysates using amylose resin (New England BioLabs), cleaved with Xa for 96 h at 25°C using 1 g Xa per mg of recombinant MBP fusion. Factor Xa-digested MBP-cryptdin-4 was applied to a C 4 column (Vydac, Hesperia, CA) in an aqueous 0.1% trifluoroacetic acid solution and eluted using a linear 0 -40% acetonitrile gradient for 55 min at a solvent flow rate of 1 ml/min. Under these conditions, recombinant cryptdin-4 eluted with a retention time of 35 min. Recombinant cryptdin-4 was purified to homogeneity from the C-4 fraction containing cryptdin-4 by C 18 RP-HPLC (Vydac). The protein sample was applied to the C 18 -column in aqueous 0.1% trifluoroacetic acid and eluted at a solvent flow rate of 1 ml/min with 0 -10% acetonitrile over 10 min, 10 -45% acetonitrile over 55 min, and 100% over 30 min. Under these conditions, cryptdin-4 had a 51-min retention time. The protein concentration of recombinant cryptdin-4 was determined using Bio-Rad Bradford reagent (Bio-Rad) and by absorbance at 280 nm (⑀ ϭ 3355 M cm Ϫ1 ). The molecular masses of purified recombinant peptides were determined by MALDI-TOF mass spectrometry on a Perkin-Elmer Voyager instrument in the UCI Biomedical Protein and Mass Spectrometry Resource Facility. The homogeneity and appropriate folding of recombinant cryptdin-4 were assessed by comparison with natural cryptdin-4 (16, 36) in 12.5% AU-PAGE gels and found to be identical to the natural peptide.

Cryptdins in Paneth Cell
Granules-Immunogold localization studies demonstrated that cryptdins are constituents of mouse Paneth cell secretory granules. In previous immunohistochemical experiments, cryptdin peptides were found only in Paneth cells (16,36), and anti-cryptdin immunoreactivity was detected in secretory granules and diffusely in the crypt lumen (16,37). Using a rabbit polyclonal anti-cryptdin-1 antibody that reacts with cryptdins 1-3 and -6 but not with cryptdin-4 or -5, 3 the subcellular location of cryptdins in adult mouse Paneth cells was shown to be specific to Paneth cell granules (Fig. 2). The distribution of gold particles was uniform, and all apical granules were labeled to approximately the same extent (Fig.  2). Immunoreactive antigen was also visible in the trans-Golgi (not shown), but there was almost no cytoplasmic staining other than in secretory granules. These results are consistent with the detection of human enteric ␣-defensin HD-5 in human Paneth cells (22) and suggest that cryptdin release depends on Paneth cell degranulation. To test the prediction that functional cryptdin peptides are released into the extracellular space, ␣-defensins were recovered from rinses of adult mouse small intestinal lumen.
Isolation and Characterization of Cryptdins from the Small Intestinal Lumen-The lumen of adult mouse small intestine contains intact cryptdins as well as cryptdin variants with modified N termini. Peptides with AU-PAGE mobilities characteristic of ␣-defensins had been detected previously in saline rinses of adult mouse jejunum and ileum (16). Comparisons of cryptdins purified from intact small bowel with putative cryptdins from luminal washes showed that certain of the luminal peptides did not co-migrate with peptides from tissue, suggesting that cryptdins may undergo modification following secretion (16). Putative luminal cryptdins were purified by combined gel filtration and RP-HPLC (see "Experimental Procedures"). The primary structures of these peptides were determined by Edman sequencing of each cryptdin through to the C terminus in all cases. This experimental approach, reported previously in determining the primary structures of full-length synthetic cryptdins (16,36) as well as cryptdin peptides from mouse small intestinal tissue (3,16), showed that the peptides from the lumen were full-length or N-terminally modified cryptdins.
Cryptdins recovered from the intestinal lumen consisted of both full-length and N-terminal-truncated peptides. Fulllength cryptdin-2, -4, and -6 were present in washings of adult mouse small bowel, and those three peptides were structurally and functionally (Figs. 4 and 5) indistinguishable from the corresponding tissue forms. Variants of cryptdin-1, -4, and -6 with shortened N termini also were recovered from the intes-3 M. E. Selsted, and D. Tran, unpublished observations.

FIG. 3. Primary structures of cryptdins from the mouse small intestinal lumen.
Primary structures determined by peptide sequencing (see "Experimental Procedures") are shown aligned with the primary structure of cryptdin-1 in which the disulfide connectivities are noted. The sequences of full-length luminal cryptdin-2, -4, and -6 are identical to the tissue forms of these peptides. Dashes (-) were introduced in the cryptdin-4 and (des-Gly)-cryptdin-4 sequences as before (3,16) to maintain alignment of cysteine residues. Cryptdin-4 is an unusual ␣-defensin, because the loop formed between the Cys 3 -Cys 5 disulfide bond lacks three amino acids that are found in ␣-defensins as a class. Residues N-terminal of the first Cys residue are boxed. tinal lumen, and they were identified as (des-Leu)-cryptdin-6, (des-Leu-Arg)-cryptdin-6, (des-Leu)-cryptdin-1, and (des-Gly)cryptdin-4 by peptide sequencing (Fig. 3). Thus, following secretion by Paneth cells, cryptdins may be isolated both as intact activated mature peptides and as N-terminal variants that have been altered by apparent aminopeptidase activity that modifies the peptides during or after exocytosis.
N-terminal Truncation Diminishes the Antimicrobial Activities of Luminal Cryptdins-Antimicrobial assays of cryptdin-6, cryptdin-4, and their N-terminally modified congeners showed that removal of the N-terminal residue(s) diminished antibacterial activity. The combined bactericidal and bacteriostatic activities of the peptides were measured in agar diffusion assays (see "Experimental Procedures") and quantitated as zones of clearing (3). Regardless of whether they were cellular or luminal in origin, full-length cryptdin-2 and -6 had equivalent activities against these three species of bacteria. In these assays, cryptdin-2 (not shown) and cryptdin-6 ( Fig. 4) were more active against Staphylococcus aureus than against Gram-negative bacterial species. Loss of Leu and Leu-Arg residues from the cryptdin-6 N terminus caused little effect on activity against S. aureus (Fig. 4A), but the activity of (des-Leu)-cryptdin-6 and (des-Leu-Arg)-cryptdin-6 against E. coli ML35 (not shown) and the attenuated phoP Ϫ mutant of S. typhimurium (Fig. 4B) was greatly reduced. The limited quantity of purified (des-Leu)-cryptdin-1 precluded a comparison with full-length cryptdin-1.
Comparisons of cryptdin-4 and (des-Gly)-cryptdin-4 antimicrobial activities showed further that the N terminus is a determinant of cryptdin activity. Cryptdin-4 is the most potent of the known mouse cryptdins in vitro (3), and it is equally active against Gram-positive and Gram-negative bacteria (Fig.  5). On the other hand, (des-Gly)-cryptdin-4 had only ϳ30% the activity of the full-length peptide against S. aureus (Fig. 5). Furthermore, when the cryptdin-4 congeners were tested against E. coli and S. typhimurium, (des-Gly)-cryptdin-4 completely lacked activity under the assay conditions, even though the peptides differ only by an N-terminal glycine (Fig. 5). Thus, the cryptdin-4 N terminus is a critical determinant of the overall antibacterial activity of this peptide.
Bactericidal Activity of (des-Gly)-Cryptdin-4 Is Ablated-To compare the microbicidal activities of cryptdin-4 and (des-Gly)cryptdin-4, recombinant forms of both peptides were expressed as fusions with MBP in E. coli (see "Experimental Procedures"). After purification to homogeneity by successive C 4 and C 18 RP-HPLC, recombinant cryptdin-4 (rCrp4) and (des-Gly)-cryptdin-4 [r(des-G)-Crp4] peptides had correct molecular masses of 3756 and 3700 atomic mass units, respectively, as determined by MALDI-TOF mass spectrometry. Also, rCrp4 and r(des-G)-Crp4 co-migrated as single bands in AU-PAGE (Fig. 6), a system that resolves defensins effectively (38). The R f values of the two recombinant peptides were indistinguishable from that of natural cryptdin-4, evidence that both peptides were homogeneous and folded correctly (Fig. 6).
Consistent with the diminished activity of natural (des-Gly)cryptdin-4 isolated from the small intestinal lumen, r(des-G)-Crp4 had no in vitro bactericidal activity, whereas rCrp4 was potently microbicidal. Against E. coli ML35 and S. typhimurium (39,40), r(des-G)-Crp4 had no measurable microbicidal activity under conditions (5 g/ml or greater) in which rCrp4 sterilized the culture of 1 ϫ 10 6 bacteria/ml (Fig. 7). These results confirm that the recombinant molecules are equivalent to the natural forms and provide an opportunity for further investigations of mechanisms by which N-terminal residues modulate cryptdin-4 microbicidal activity. DISCUSSION These experiments show that full-length cryptdins are released into the intestinal lumen and that functional peptides can be isolated from the lumen of the small intestine. Immunogold-detection experiments localized mouse cryptdins to Paneth cell secretory granules (Figs. 1 and 2), and, therefore, luminal cryptdins result from the secretory responses of Paneth cells and not those of other enteric epithelial lineages. Also, immunohistochemical experiments provided no evidence that cryptdins were associated with apical microvilli or the brush border of the mucosal epithelium (data not shown). These findings suggest that the peptides remain soluble following secre- E. coli ML35 (circles) or S. typhimurium phoP Ϫ (triangles) cells were exposed to peptide concentrations as shown for 1 h at 37°C, Reactions were plated using a Spiral Biotech Autoplate 4000 and surviving CFU were quantitated after overnight incubation at 37°C. Filled symbols, bacterial cells exposed to rCrp4; open symbols, bacterial cells exposed to r(des-G)-Crp4. tion or that they diffuse rapidly out of the crypt. Although approximately equal quantities of full-length cryptdins and cryptdins with N-terminal truncations of 1 or 2 amino acids were isolated, truncated variants have not been detected among cryptdins purified from mouse intestinal tissue. Whether the truncations occur in the crypt lumen or after diffusion above the crypt-villus junction is unknown. Because no Paneth cell secretory stimuli (30 -32) were administered to the mice in these studies, the luminal peptides that were isolated represent molecules released by Paneth cells under steady-state conditions.
In the human neutrophil ␣-defensins, the N terminus is known to modulate certain activities. For example, the respective N termini of human neutrophil ␣-defensins HNP-1-3 are ACYC, CYC, and DCYC, but otherwise the primary structures are identical (41). Despite these similarities, HNP-1 and 2 are chemotactic for monocytes and T lymphocytes, but HNP-3 lacks both activities (42,43). Similarly, HNP-3 is the only peptide of the three that does not kill Candida albicans (44). Interestingly, those comparisons showed that the charge at the N terminus affected activity but that removal of the N-terminal Ala had little effect. Perhaps these differences reflect the differential activities of peptides that function primarily in neutrophil phagolysosomes as compared with the external environment of the intestinal lumen.
One or two residue truncations of cryptdin N termini attenuate or eliminate peptide antimicrobial activity against Grampositive and Gram-negative bacteria (Figs. 4, 5, 7). For both cryptdin-6 and cryptdin-4, the effect of truncation was greater against Gram-negative bacteria than against S. aureus, a Gram-positive bacterium. Because the molecular details of cryptdin bactericidal activity are not known, it is not immediately apparent how the deletion of a single glycine residue from the cryptdin-4 N terminus could eliminate bactericidal activity so thoroughly. ␣-Defensin tertiary structures show that the N and C termini are juxtaposed (45)(46)(47)(48), and we speculate that N-terminal truncation may disrupt hypothetical associations with the C terminus, leading to impaired peptide interactions with target cell membranes.
The variability of N-terminal length among certain luminal cryptdins does not appear to result from inaccurate precursor processing and activation by MMP-7, the procryptdin activating enzyme in mouse Paneth cells. In vitro, MMP-7 cleaves recombinant procryptdin substrates with high fidelity (27). Also, extensive processing of cryptdin precursors occurs intracellularly, because mouse Paneth cell granules contain high levels of fully activated cryptdins. 4 Furthermore, N-terminal cryptdin variants have not been detected in mouse intestinal tissue (3,16). For these reasons, it is unlikely that an errorprone MMP-7 introduces variation at cryptdin N termini during post-translational activation. Current evidence supports the view that N-terminal truncation of cryptdin peptides in the lumen is caused by the modification of activated peptides following secretion.