Structure-Activity Determinants in Paneth Cell (cid:1) -Defensins LOSS-OF-FUNCTION IN MOUSE CRYPTDIN-4 BY CHARGE-REVERSAL AT ARGININE RESIDUE POSITIONS*

Paneth cells secrete microbicidal enteric (cid:1) -defensins into the small intestinal lumen, and cryptdin-4 (Crp4) is the most bactericidal of the mouse (cid:1) -defensin peptides in vitro . Here, site-directed Arg to Asp mutations in Crp4 have been shown to attenuate or eliminate microbicidal activity against all of the bacterial species tested regardless of the Arg residue position. R31D/R32D charge-reversal mutagenesis at the C terminus and mutations at R16D/R18D, R16D/R24D, and R18D/R24D in the Crp4 polypeptide chain eliminated in vitro bactericidal activity, blocked peptide-membrane interactions, as well as Crp4-mediated membrane vesicle disruption. Lys for Arg charge-neutral substitutions in (R16K/R18K)-Crp4 did not alter the bactericidal

The ␣-defensins are cationic, amphipathic 3-4-kDa peptides with a defining tridisulfide arrangement and ␤-sheet polypeptide backbone that have a broad spectrum of antimicrobial activities (13)(14)(15). For example, mouse ␣-defensins, termed cryptdins (Crps), 1 account for ϳ70% of the bactericidal peptide activity in secretions elicited from Paneth cells (3) and Crp4 is the most potent of the known mouse ␣-defensin peptides (16,17). Also, Crps became implicated as components of mouse innate enteric immunity in vivo when mice that lacked matrix metalloproteinase-7, the pro-Crp-activating enzyme, were shown to have impaired host defense against oral infections (18,19). Remarkably, mice transgenic for human Paneth cell ␣-defensin HD5 expressed the minigene specifically in Paneth cells and they were immune to oral infection by high doses of Salmonella enterica serovar Typhimurium (S. typhimurium) (20). Accordingly, an understanding of structure-activity relationships in ␣-defensins will improve the understanding of mucosal innate immune mechanisms.
To investigate the role of ␣-defensin primary structure in innate immunity, amino acid substitutions that alter charge were introduced into the Crp4 peptide and tested for effects on microbicidal activity and on Crp4-membrane interactions. Regardless of the site of mutagenesis, charge-reversal substitutions at Arg positions altered Crp4 bactericidal activity profoundly and the loss of activity correlated directly with quantitative effects on peptide binding to phospholipid vesicles and with peptide-induced vesicular permeabilization.
Mutagenesis at Arg Residue Positions-Mutations were introduced into Crp4 molecules by PCR as described previously (21). In the first reaction, a mutant forward primer, e.g. Crp4-R16D-F, containing the mutant codon flanked by three natural codons was paired with reverse primer, [pMALCrp4-R], the normal reverse primer at the 3Ј-end of the desired sequence. In the second reaction, the mutant reverse primer, Crp4-R16D-R, the reverse complement of the mutant forward primer, was paired with the normal forward primer, [ER1-Met-C4-F], at the 5Ј end of Crp4. After amplification at 95°C for 5 min followed by successive cycles at 60°C for 1 min, 72°C for 1 min, and 94°C for 1 min for 40 cycles, samples of purified products from reactions 1 and 2 were combined as templates in PCR in the third reaction using the Crp4 external primers ER1-Met-C4-F and SLpMALCrp4R as amplimers. All of the mutated Crp4 constructs were verified by DNA sequencing, subcloned into pET28a plasmid DNA (Novagen, Inc.), and transformed into E. coli BL21(DE3)-CodonPlus-RIL cells (Stratagene) for recombinant expression. The underlined codons in forward primers denote Met codons introduced upstream of each peptide N terminus to provide a CNBr cleavage site (21,22,24).
The following C-terminal variants of Crp4 were prepared as noted above with the following mutant reverse primers: Similarly, Crp4 variants with C-terminal extensions resembling that of the Crp4(B6b) peptide (21) were prepared using the following primers: Purification of Recombinant Crp4 Proteins-Recombinant proteins were expressed at 37°C in Terrific Broth medium by induction with 0.1 mM isopropyl-␤-D-thiogalactopyranoside for 6 h at 37°C, lysed by sonication in 6 M guanidine-HCl in 100 mM Tris-Cl (pH 8.1), and clarified by centrifugation (21,22,24). His-tagged Crp4 fusion peptides were purified using nickel-nitrilotriacetic acid (Qiagen) resin affinity chromatography from bacterial cells lysed in 6 M guanidine-HCl, 20 mM Tris-HCl (pH 8.1) as described previously (21). After CNBr cleavage, Crp4 peptides were purified by C18 reverse-phase high performance liquid chromatography and quantitated by bicinchoninic acid (Pierce) and molecular masses of purified peptides were determined using matrix-assisted laser desorption ionization mode mass spectrometry (Voyager-DE MALDI-TOF, PE-Biosystems, Foster City, CA) in the Mass Spectroscopy Facility, Department of Chemistry, University of California (Irvine, CA).
Bactericidal Peptide Assays-Recombinant peptides were tested for microbicidal activity against E. coli ML35, S. typhimurium (PhoPϪ), Vibrio cholera, Staphylococcus aureus 710a, and Listeria monocytogenes 104035 (25). Bacteria growing exponentially in trypticase soy broth at 37°C were deposited by centrifugation at 1700 ϫ g for 10 min, washed in 10 mM PIPES (pH 7.4), and resuspended in 10 mM PIPES (pH 7.4) supplemented with 0.01 volume of trypticase soy broth (21,22). Bacteria (ϳ5 ϫ 10 6 CFU/ml) were incubated with test peptides in 50 l for 1 h in a shaking incubator at 37°C, and then 20-l samples of incubation mixtures were diluted 1:100 with 10 mM PIPES (pH 7.4) and 50 l of the diluted samples were plated on trypticase soy agar plates using an Autoplate 4000 (Spiral Biotech Inc., Bethesda, MD). Surviving bacteria were counted as colony forming units per milliliter after incubation at 37°C for 12-18 h.
Peptide Interactions with Phospholipid/Polydiacetylene (PDA) Mixed Vesicles-Crp4 and three mutant peptides with varied Arg charge reversals were investigated for their relative membrane perturbation activities. Colorimetric phospholipid/PDA vesicles were prepared using dimyristoylphosphatidylcholine (Sigma) as described previously (22). Vesicles consisting of two parts phospholipid to three parts PDA were prepared by dissolving the phospholipids and 10,12-tricosadiynoic acid monomer (GFS Chemicals, Powell, OH) together in chloroform/ethanol (1:1), drying in vacuo to constant weight, suspending in H 2 O, probesonicating for 3 min at 70°C, and incubating overnight. PDA was polymerized by irradiation at 254 nm for 10 -20 s, producing suspensions with an intense blue appearance. Peptides (0.2-20 M) were added to 60 l of vesicle solutions (0.5 mM total lipid) in 25 mM Tris-HCl (pH 8) and diluted to 1 ml, and spectra were acquired at 28°C between 400 and 700 nm on a Jasco V-550 spectrophotometer (Jasco Corp., Tokyo, Japan) using a 1-cm optical path cell. Blue-to-red color transitions within the vesicle solutions, defined as the percent colorimetric response (%CR), were calculated as described previously (22,26).

Bactericidal Activities of Recombinant Crp4
Variants-Recombinant peptides (Fig. 1A) were prepared using the pET-28 vector system (21). All of the peptides were purified to homogeneity by reverse-phase high performance liquid chromatography after chemical cleavage with CNBr, migrating as single entities and as expected relative to native Crp4 and pro-Crp4 molecules (21,24). Purity was verified by analytical reversephase high performance liquid chromatography (data not shown) and acid-urea-PAGE analyses (Fig. 1B). The molecular masses of individual recombinant peptides were determined by MALDI-TOF mass spectrometry, and they matched the respective theoretical values. Thus, the purified recombinant peptides were homogeneous (Fig. 1B) and their biochemical features were consistent with the modifications introduced to the natural Crp4 molecule (29).
As a first step toward investigating structure-activity relationships in Crp4, the in vitro microbicidal activities of Crp4, pro-Crp4, and selected N-terminal Crp4 variants were measured against a panel of bacterial test species (data not shown) (24). The overall bactericidal activities of the N-terminal variants differed only slightly over a range of 0.6 -10 g/ml of peptide as observed previously (24). In contrast to the bactericidal activities of these mature Crp4 peptide variants, pro-Crp4 lacks microbicidal activity because matrix metalloproteinase-7-mediated proteolysis is required for Crp4 activation (18,19,21). Although the (G1D)-Crp4 peptide was consistently less active at Յ5 g/ml of peptide (data not shown), modifications at the Crp4 N terminus had very modest effects on bactericidal activity (24).
Modifications at the Crp4 C terminus-Because the Cys-1 to Cys-6-disulfide bond common to ␣-defensins places the Crp4 N and C termini in proximity (13), we prepared a series of Crp4 peptide variants with charge-modified C termini including (R31D/R32D)-Crp4, (R31G/R32G)-Crp4, (R31V/R32V)-Crp4, and (⌬R31/R32)-Crp4 (Fig. 1). Bactericidal activity assays of these variants showed that deleting the two C-terminal Arg residues or converting them to Gly or Val altered peptide activity variably depending on the bacterial target species (Fig. 2, compare B with D). In contrast, however, the R31D/R32D charge-reversal mutation eliminated bactericidal activity, even against the S. typhimurium PhoPϪ strain (data not shown), which is very sensitive to membrane-active cationic peptides (Fig. 2) (30, 31). As expected, wild-type S. typhimurium and the PhoP-constitutive (PhoP c ) CS022 strain were less sensitive to all of the peptides (data not shown).
To investigate the effect of added C-terminal Arg residues on Crp4 bactericidal activity, additional modifications were introduced at the Crp4 C terminus. C57/BL6 mice express a Crp4 variant (Crp4(B6b)) that has four Arg residues at its C terminus (21), but because Crp4 and Crp4(B6b) differ at several other positions, the specific role of C-terminal charge in these peptides cannot be compared directly. Accordingly, the natural Crp4 C terminus was extended by two Arg residues to place the Crp4-(B6b) C terminus in the context of the Crp4 primary structure. The peptide, (Arg-33/Arg-34)-Crp4, had greater in vitro bactericidal peptide activity than Crp4 against E. coli and wild-type and PhoP c S. typhimurium strains (Fig. 4), suggesting that C-terminal charge could be a determinant of bactericidal activity. However, the Crp4(B6b) molecule lacks the Pro-30 residue present in Crp4 between Cys-29 and Arg-31. Therefore, the effect of the Pro-30 residue was tested by assaying the bactericidal activity of (⌬Pro-30/Arg-33/Arg-34)-Crp4, a variant of (Arg-33/Arg-34)-Crp4 from which Pro-30 was deleted (Fig. 1). (⌬Pro-30/Arg-33/Arg-34)-Crp4 was less active than Crp4 and (Arg-33/Arg-34)-Crp4 against E. coli and S. typhimurium PhoP c (Fig. 4), and the results in Fig. 4 also are representative of the relative bactericidal activities of these peptides against V. cholera, S. aureus, and L. monocytogenes (data not shown). Thus, addition of Arg residues at the C terminus per se is not sufficient to improve peptide activity in the absence of proline at residue position 30. Because the R31D/R32D C-terminal charge reversal eliminated Crp4 bactericidal activity, the effects of charge alterations at other residue positions in the Crp4 polypeptide chain were studied to test whether the effects were specific to C-terminal modification.

Charge Reversal at Arg Positions Result in Loss-of-function-
Crp4 mutants with double Asp for Arg substitutions were introduced at three additional Arg pairs, purified, and assayed for bactericidal peptide activity (see "Experimental Procedures"). As was observed for (R31D/R32D)-Crp4, the R16D/ R18D, R16D/R24D, and R18D/R24D variants of Crp4 all lacked bactericidal activity under the conditions of these in vitro assays (Fig. 5). Thus, the loss of activity by charge reversal is not a specific effect at the C terminus, because all of the peptides with double charge-reversal mutations were inactive. Perhaps, Asp for Arg substitutions eliminate bactericidal activity by reducing the overall electropositivity of the Crp4 peptide. Nevertheless, it is unlikely that Arg residues at specific positions are required for Crp4 bactericidal activity because all of the mutations introduced resulted in equivalent loss-of-function independent of the position of the replacements.
To determine whether a specific requirement exists for the guanidinium group of the Arg side chain in Crp4 or whether charge-neutral substitutions at Arg positions would result in peptides with equivalent bactericidal activity, we prepared (R16K/R18K)-Crp4 (Fig. 1A) and evaluated its antimicrobial activity. Because all of the Crp4 charge-reversal mutants were attenuated similarly including (R16D/R18D)-Crp4, the (R16K/ R18K)-Crp4 internal charge-neutral mutant was taken to be representative of such peptide alterations. As stated earlier, the Lys for Arg substituted peptide was purified to homogeneity as in Fig. 1B (data not shown) prior to comparing its bactericidal peptide activity relative to Crp4. The bactericidal peptide activities of Crp4 and (R16K/R18K)-Crp4 were not distinguishable against E. coli or L. monocytogenes (Fig. 6) or against other bacterial species (data not shown). Thus, with regard to bactericidal peptide activity, Lys could replace Arg in Crp4 without effect.

Charge-reversal Mutations Block Crp4-Membrane
Interactions-To investigate the mechanisms by which charge-reversal mutagenesis of Crp4 eliminates peptide activity, we compared interactions of Crp4 and charge-reversal Crp4 mutants
Crp4 loss-of-function by Arg to Asp mutagenesis correlated with the inability of mutant peptides to interact with membranes at the vesicular lipid/water interface and to perturb the hydrophobic membrane core. In the dimyristoylphosphatidylcholine/PDA-mixed vesicle system, higher %CR values corresponded to interfacial lipid binding because peptides that localized at the lipid bilayer surface induced greater perturbation in the head-group region of the lipid/polymer assembly than peptides that penetrate deeper into the hydrophobic core (32,33). Crp4 induced blue-red chromatic transitions, reaching a plateau of ϳ35%CR upon interaction with the phospholipid/ PDA vesicles. The %CR value was dependent on peptide concentration (Fig. 7), and previous work has determined that the colorimetric response indicates an interface localization of Crp4 (22,32). In contrast and without exception, only base-line %CR values were induced by inactive Crp4 mutants with Asp for Arg substitutions (Fig. 7). Consistent with its intermediate bactericidal activity (Fig. 4), (G1R/R31D/R32D)-Crp4 induced somewhat higher intermediate %CR values between those of Crp4 and the Crp4 variants (Fig. 7). Thus, the loss of bactericidal activity by Arg charge reversals was consistent with defective Crp4 interaction with the vesicular lipid/water interface and an inability to perturb the membrane core of the leaflet.
The membrane-disruptive activities of Crp4 and inactive Crp4 variants were analyzed by comparing induced leakage of the low molecular weight fluorophore, ANTS, from LUV (see "Experimental Procedures"). Crp4 induces leakage from LUV via a "graded leakage" mechanism (24, 28), and pro-Crp4, con-sistent with its lack of bactericidal activity (21,22), induces markedly less leakage than Crp4. For both peptides, leakage is dependent on LUV phospholipid composition (Fig. 8) (28). When exposed to (R31D/R32D)-Crp4, (G1D/R31D/R32D)-Crp4, and (R16D/R24D)-Crp4, ANTS-loaded LUV prepared with the anionic lipid palmitoyl-oleoyl-phosphatidylglycerol exhibited leakage that was Յ5% of that induced by Crp4, much less than leakage induced by pro-Crp4 (Fig. 8). Thus, regardless of the position of Arg charge reversals in the polypeptide chain, the loss-of-function was the same and corresponded directly to a loss of peptide binding to membranes and an inability to disrupt phospholipid vesicles in vitro.

DISCUSSION
In this report, site-directed Arg to Asp mutations in Crp4, a mouse Paneth cell ␣-defensin, were shown to attenuate or eliminate microbicidal activity against all of the bacterial target cells examined. Whether charge reversals occurred at the C terminus, at R16D/R18D, R16D/R24D, or at R18D/R24D, mutagenesis eliminated Crp4 in vitro bactericidal activity, blocked peptide-membrane interactions, and eliminated Crp4-mediated membrane vesicle disruption. Because Lys for Arg chargeneutral substitutions in the (R16K/R18K)-Crp4 peptide did not modify Crp4 bactericidal peptide activity, it appears that it is the charge rather than a specific requirement for Arg per se that determines bactericidal activity in Crp4. Fluorophore leakage from anionic phospholipid vesicles induced by the charge-reversal variants was negligible relative to Crp4 and lower than that induced by pro-Crp4, the inactive Crp4 precursor. Thus, Arg residues function as determinants of bactericidal activity in native Crp4 by facilitating or enabling target cell membrane disruption but the role of the Arg residues appears to be independent of their position in the polypeptide chain.
The findings reported here support the conclusion that Arg residues in the Crp4 polypeptide chain facilitate peptide-membrane binding and interactions that lead to or are coincident with disruption of phospholipid bilayers (Figs. 7 and 8). not be folded properly or form a correct disulfide array is a potential concern. However, preliminary one-and two-dimensional TOCSY NMR determinations indicate that the mutant peptides have proper disulfide connectivities. 2 Also, when natural ␣-defensins have been available as standards, recombinant mouse ␣-defensins and their precursors that we have prepared co-elute from reverse-phase high performance liquid chromatography. The recombinant peptides also co-migrate in acid-urea gels with the native molecules, a sensitive system for resolving misfolded defensins. It is generally accepted that peptide amphipathicity confers an ability on antimicrobial peptides to interact with and disrupt microbial cell membranes (37)(38)(39)(40)(41)(42)(43), leading to dissipation of cellular electrochemical gradients and microbial cell death (44). Because early events in bacterial cell killing by human neutrophil ␣-defensins depend on electrostatic interactions, our results are consistent with the Arg residues in Crp4 facilitating such interactions. Also, implicating the particular Arg residues noted does not exclude the possibility that additional polar or hydrophobic amino acids may be equally important for Crp4 bactericidal activity. The evidence shows that Crp4 bactericidal activity does not require that arginines occupy specific positions in the peptide primary structure, because all of the charge reversals caused equivalent loss-of-function independent of individual replacements.
The bactericidal activity of mouse Paneth cell Crps is dependent on the activation of pro-Crps from inactive 8.4-kDa proforms by matrix metalloproteinase-7-mediated proteolysis (18,19,21). For example, full-length pro-Crp4 (pro-Crp4 20 -92 ) lacks bactericidal activity and that lack of activity corresponds with a diminished ability to bind to model membranes (22) and to permeabilize LUV (Fig. 7) (28). Interestingly, the two pro-Crp4 processing-intermediates, pro-Crp4 44 -92 and pro-Crp4 54 -92 have the same bactericidal activity as the mature Crp4 peptide. 3 That finding implicates amino acids in the N-terminal 24 residues of the pro-Crp4 20 -92 molecule as inhibitory to Crp4 peptide activity in the unprocessed pro-Crp4 molecule. Because nine of the residues in the N-terminal moiety of the pro-region are Asp or Glu, it is reasonable to hypothesize that those acidic side chains in the pro-region interact with the Arg residues blocking their interactions with membranes. As has been proposed for human neutrophil ␣-defensin precursors (45,46), we speculate that the formation of favorable yet hypothetical Asp-Arg or Glu-Arg salt bridges (47) could disable the ability of the Crp4 moiety of the pro-Crp4 20 -92 molecule to bind and disrupt target cell membranes by neutralizing their cationicity at the peptide surface.
The ␣-defensins from human and rabbit neutrophils achieve bacterial cell killing by distinctive membrane disruptive mechanisms. Neutrophil ␣-defensins permeabilize the outer and inner membranes of E. coli sequentially, inducing the formation of ion channels in lipid bilayers (48), and both of these peptide-elicited effects are influenced by membrane energetics (49). The crystal structure of human neutrophil peptide 2 is a noncovalent dimer (14), and human neutrophil peptide 2 dimers form stable 20-Å multimeric pores in LUV after insertion into model membranes (40). In contrast, rabbit neutrophil peptide 1 is a monomer in solution (35) and neutrophil peptide 1 permeabilizes the membrane by creating large short-lived defects in model phospholipid bilayers (50). Although the structure of Crp4 is not yet reported, the peptide, which has the greatest in vitro bactericidal activity among the mouse ␣-defensins, exhibits strong interfacial binding to model membranes (22) and induces graded fluorophore leakage from LUV (24,28). Thus, Crp4 resembles the rabbit neutrophil peptide 1-type ␣-defensin mechanism of bactericidal action and sitedirected Arg to Asp mutagenesis targets Crp4 residues that facilitate or enable that membrane-disruptive mechanism.