HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Castle, M. Articles by Tempst, P. Search for Related Content PubMed PubMed Citation Articles by Castle, M. Articles by Tempst, P. Social Bookmarking                      What's this?

J Biol Chem, Vol. 274, Issue 46, 32555-32564, November 12, 1999

Lethal Effects of Apidaecin on Escherichia coli Involve Sequential Molecular Interactions with Diverse Targets^*

Madalyn Castle, Arpi Nazarian, San San Yi, and Paul Tempst^§¶

From the  Molecular Biology Program, Memorial Sloan-Kettering Cancer Center and the ^§ Weill Graduate School of Medical Sciences, Cornell University, New York, New York 10021

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Apidaecins, short proline-arginine-rich peptides from insects, are highly bactericidal through a mechanism that includes stereoselective elements but is completely devoid of any pore-forming activity. The spectrum of antibacterial activity, always limited to Gram-negatives, is further dependent on a small number of variable residues and can be manipulated. We show here that mutations in the evolutionary conserved regions result in a more general loss of function, and we have used such analogs to probe molecular interactions in Escherichia coli. First, an assay was developed to measure selectively chiral association with cellular targets. By using this method, we find that apidaecin uptake is energy-driven and irreversible and yet can be partially competed by proline in a stereospecific fashion, results upholding a model of a permease/transporter-mediated mechanism. This putative transporter is not the end point of apidaecin action, for failure of certain peptide analogs to kill cells after entering indicates the existence of another downstream target. Tetracycline-induced loss of bactericidal activity and dose-dependent in vivo inhibition of translation by apidaecin point at components of the protein synthesis machinery as likely candidates. These findings provide new insights into the antibacterial mechanism of a unique group of peptides and perhaps, by extension, for distant mammalian relatives such as PR-39.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Insects have evolved a diverse repertoire of innate, antibacterial substances that in many ways compares well with vertebrate immune defenses (1-13). Merely fascinating at first, insect peptide antibiotics began to attract considerable attention with the realization that functional and structural similarities existed with peptides from mammalian blood cells, the primary effectors for eradicating bacteria (14-21). Related molecules are also present in anatomical locations where direct exposure to microorganisms is likely to occur, such as mouth, airways, gut, skin, wounds, and blisters (22-32). Most ribosomally synthesized peptide antibiotics kill bacteria in basically the same way, by making pores in cell (cytoplasmic) membranes, often indiscriminately on Gram-positives and -negatives (2, 15, 33-39). Pore-forming is not dependent on recognition of chiral molecules, for it is well established that all-D enantiomers are equally active (40-42). Only a few exceptions to this theme are known, most notably a small group of proline-arginine-rich polypeptides found in bovine neutrophils, pig gut, and in the hemolymph of Drosophila and several stinging insects (Hymenoptera) (7, 9, 11, 18, 26). Best characterized to date is apidaecin, a short peptide (18 amino acids) originally isolated from honeybees, that is lethal for many Gram-negative bacteria (7, 11, 43). Killing occurs very rapidly, through a not as yet understood mechanism (7). The most unique feature about its mode-of-action is a total lack of pore-forming activity. In addition, all-D enantiomers are devoid of any measurable function, indicating a significant degree of stereospecificity (44).

To begin to understand the functional role of each amino acid in the apidaecin structure, we have previously isolated 13 isotypes from various hymenopteran insects (11). Sequence alignment delineated conserved "core" regions, most likely essential for activity. Subsequent functional analysis revealed striking differences between the antibacterial spectra of individual isotypes, implying that the structural elements underlying such specificities resided within the short stretches of "variable" sequence (11). In some instances, subtle changes would switch the spectra to mutually exclusive patterns. For example, replacing one Arg and one Gln in isotype Cd1 with lysines (isotype Cd3-) abolished all original activity against Yersinia enterocolitica, yet created a peptide that was lethal for Campylobacter jejuni, whereas the initial "Cd1" was not. Thus, apidaecin-type peptides consist of "constant" regions, conferring general antibacterial capacity (i.e. any modifications are predicted to reduce or eliminate function), and variable regions, determining specificity of the spectrum.

Collectively, the data suggest that apidaecin antibacterial function may be transporter-based. If indeed the case, species-related differences in homologous "receptor/docking" molecules, proteins perhaps, would explain the contrasting antibacterial spectra of different isotypes; and novel mutations could conceivably result in unique bacterial mutants, resistant to one particular isotype but still sensitive to others, as observed (11).

Here, we present strong corroborating evidence for a transporter-mediated mechanism. By using a biological assay to selectively measure chiral interactions with bacterial targets, we show that uptake is energy-driven and virtually irreversible and yet partially competable by proline in a stereospecific fashion. Failure of certain peptide mutants to kill cells after entering indicates the existence of at least one other downstream target. Several lines of evidence, including in vivo inhibition of translation, suggest this could be a component of the protein synthesis machinery.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Synthetic Peptides-- Chemical synthesis of apidaecin (isoform Ho+, GKPRPQQVPPRPPHPRL, and analogs)-type peptides and magainin 2 (GIGKFLHSAKKFGKAFVGEIMNS) was performed with an automated instrument, model 433A (PE Applied Biosystems, Foster City, CA), as described (11). N-(9-Fluoroenyl)methoxycarbonyl (Fmoc)¹-protected L-amino acids were obtained from Anaspec (San Jose, CA); Fmoc-D-amino acids, Fmoc-L-amino acid resins, and Fmoc-D-Leu-Wang and Fmoc-D-Ser(t-butyl)-Wang resins were from Novabiochem (San Diego, CA). Trifluoroacetic acid promoted cleavage from the resin, and extraction and purification by one or more rounds of preparative reversed-phase high performance liquid chromatography (RP-HPLC) were also as documented (11). Stocks were initially quantitated by amino acid composition analysis (Protein/DNA Technology Center, Rockefeller University, New York, NY) and periodically thereafter by RP-HPLC. In case of radiolabeled peptides, the automated synthesis process was stopped with just one cycle to go (before addition of the N-terminal glycine, common to apidaecin and analogs, and magainin 2), after removal of the Fmoc group. Peptide resin was then washed five times with dimethylformamide (DMF), and 910 nmol were taken for manual addition of [¹⁴C]glycine, as follows. Two ml of an Fmoc-[1-¹⁴C]Gly-OH (55 mCi/mmol, 0.1 mCi/ml ethanol; from American Radiolabeled Chemicals, St. Louis, MO) solution were dried down in a SpeedVac and redissolved by adding 1 ml of DMF, 0.25 ml of 1 M 1,3-diisopropylcarbodiimide in DMF, 33.75 mg of 1-hydroxybenzotriazole, and 5 mg of dimethylaminopyridine, before mixing with the resin. Coupling was done for 15 h at room temperature while shaking vigorously, followed by several washes with DMF, dichloromethane/DMF (3:2 by volume), and finally with dichloromethane, before drying. After transfer to a 10-ml PolyPrep disposable chromatography column (Bio-Rad), peptides were cleaved from the resin, and deprotected, for 2 h at room temperature in a solution (2-ml total volume) consisting of 1.67 ml of trifluoroacetic acid, 125 µl of phenol, 42 µl of 1,2-ethanedithiol, 84 µl of thioanisole, and 84 µl of water. Peptide solutions were then vacuum filtered away from the resin, straight into ice-cold tert-methyl butyl ether for precipitation; peptides were collected by centrifugation, followed by four more washes with ether, lyophilized, and redissolved in 10% acetic acid before RP-HPLC purification. Peak fractions (from UV detection) were screened for the correct product by matrix-assisted laser-desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS), using a Voyager RP system (Perspective Biosystems, Framingham, MA), and lyophilized. A 20% solution of piperidine in DMF was then added to cover the dried powder and stirred for 30 min at room temperature to remove the N-terminal Fmoc-protecting group, followed by 20-fold dilution with 0.1% trifluoroacetic acid in water and another round of RP-HPLC. Quality control was done by analytical RP-HPLC and MALDI-TOF MS; stocks were quantitated (see above), and radioactivity was measured by liquid scintillation counting (model LS600IC; Beckman Coulter, Fullerton, CA). Typically, specific activity was on the order of 60,000 cpm/µg peptide.

Bacterial Strains-- The following bacterial strains (except one) were obtained from the American Type Culture Collection (ATCC, Manassas, VA): Acinetobacter calcoaceticus ATCC 49137, Agrobacterium tumefaciens ATCC 15955, Escherichia coli ATCC 25922 and 11775, Enterobacter cloacae ATCC 529, Klebsiella pneumoniae ATCC 13883, Salmonella typhimurium ATCC 14028, Salmonella typhi ATCC 6539, and Shigella dysenteriae ATCC 13313; E. coli strain D22 is an outer membrane defective mutant, derived from strain K-12 (39, 44). All strains were handled under conditions and in a laboratory environment in compliance with Biosafety Level 2 for Infectious Agents as set forth in Health and Human Services Publication (NIM) 88-8395.

Antibacterial Plate Assays-- For well assays, an aliquot of an overnight culture was inoculated into BHI broth (BBL, Becton-Dickinson, Cockeysville, MD) and grown at 37 °C with shaking at 150 rpm until the cell suspension became moderately turbid. An aliquot of the suspension was then diluted into 10% BHI broth until the turbidity matched the density of a 0.5 McFarland (Remel; Renexa, KS) standard. This procedure produced cells at a density that grow just confluently when plated with a cotton swab on Nutrient Agar (Difco) plates. Subsequent to swabbing the plate with bacterial cells, 10-µl capacity wells (3 mm diameter) were punched into the agar surface. Peptide solutions (10 µl) were then delivered to the wells. In case of filter assays, filters (0.25-inch; BBL) were impregnated with peptide and then placed on the surface of the plate. After 16 h of incubation at 37 °C, all plates were screened for zones of inhibition. Plating was also used to determine number of viable cells. Serial dilutions were made in sterile 0.9% NaCl, and aliquots of the dilution series were removed for triplicate plating on tryptic soy agar (Difco) plates. After spreading, plates were incubated overnight at 37 °C and colonies counted.

Minimal Inhibitory Concentration (MIC) Assays-- An aliquot of an overnight culture was inoculated into BHI broth until the cell suspension became moderately turbid, then diluted into 10% BHI broth to match turbidity of a 0.5 McFarland standard, and diluted 1:1000 in fresh 10% BHI broth. MICs of peptides for cells were determined in round-bottomed, 96-well microtiter plates; total volume of each well was 200 µl. Preliminary assays determined the appropriate peptide concentration range, and then peptides were distributed to wells in successive 2-fold dilutions; control wells did not contain peptide. The final suspension of cells was 1000-5000 colony-forming units (cfu) per ml in 10% BHI. Growth of cells in the microtiter plates was determined by visual inspection after 24 h incubation at 37 °C. MIC is defined as the lowest concentration tested that inhibits bacterial growth.

Uptake Assays-- E. coli ATCC 25922 cells were grown in BHI broth overnight, at 37 °C. An aliquot of this culture was inoculated into fresh BHI and incubated at 37 °C with shaking (150 rpm) until the culture reached 0.5-0.6 A₆₀₀; cells were pelleted and washed with 4 times the original culture volume of 10% BHI. The final pellet was resuspended in 10% BHI at a cell density of 5 × 10⁹ cfu/ml. One-ml aliquots were then incubated with 1-10 µg of radiolabeled (60,000-600,000 cpm) peptide in Uptake Assay Broth (10% BHI) for 30 min at 22 °C, unless otherwise stated. In all cases where higher peptide concentrations were used, unlabeled peptide was added to yield the desired amount. After reaction, cells were pelleted, and the reaction supernatant ("wash 0") was collected into Liquid Scintillation Fluid (Scintiverse, Fisher). The resulting pellet was resuspended in Wash Buffer (fresh 10% BHI, containing 150 mM NaCl) by vortexing, and the procedure was repeated for the remainder of the washes. After removing the last (typically 11th) wash supernatant, the final pellet was resuspended in the original reaction volume of Wash Buffer. If desired, an aliquot of the resuspended pellet is removed for determining viable plate counts. The resuspended pellet and all wash supernatants were inspected for the presence of labeled peptides by liquid scintillation counting. Pellet counts were corrected for removal of the viable plate count aliquot. Experimental variations to the above listed procedure were introduced in the course of this study, as follows. Assay broth was at times supplemented with 0.2 mM MgCl₂ or 25 mM NaCl. L-Pro, D-Pro, or all-L Arg-Pro-Lys-Pro (all from Sigma) were sometimes added to the incubation mixture to give 10 mM final concentrations, just seconds before addition of the labeled peptides. Tetracycline (8 µg/ml; Sigma) or 2,4-dinitrophenol (5 mM final concentration; Sigma) were added to cell suspensions for 15 h, respectively, and 3 h at 30 °C while shaking; cells were then pelleted, washed with 25 ml of 10% BHI, repelleted, and resuspended in 1 ml of assay broth before addition of labeled peptides.

Incorporation of [³H]Leu and [³⁵S]Met into Protein and [³H]Uridine into RNA-- E. coli ATCC 25922 cells were grown in Nutrient Broth (Difco) overnight at 37 °C. An aliquot of this culture was inoculated into fresh Nutrient Broth until the cell suspension reached an A₆₆₀ of 0.1. Incorporation reactions were done in round-bottomed 96-well microtiter plates. Total volume in each well was 300 µl, containing 100 µl cells and 200 µl 50% Nutrient Broth with either 2-µCi of L-[4,5-³H]leucine (135 Ci/mmol),30 µCi of L-[³⁵S]methionine (>1000 Ci/mmol), or 2 µCi of [5,6-³H]uridine (39.0 Ci/mmol). Radiolabeled chemicals were purchased from Amersham Pharmacia Biotech. Tetracycline hydrochloride and rifampin were obtained from Sigma. TET stock solutions were prepared at 10 mg/ml in water; RIF stocks were 10 mg/ml in methanol. Reactions that contained these drugs had concentrations of 8.3 µg/ml TET and 83 µg/ml RIF. Reactions that contained apidaecin (LHo+) had concentrations that varied from 1.7 to 17 µg/ml. At various time points, 50-µl aliquots of the reaction mixtures were removed to 200 µl of ice-chilled 20% trichloroacetic acid and allowed to precipitate. After 30 min of precipitation on ice, 200-µl aliquots were delivered into 3 ml of ice-chilled 5% trichloroacetic acid. Glass microfiber filters (24-mm diameter circles; Whatman, Hillsboro, OR) were pre-washed with 1 ml of ice-chilled 5% trichloroacetic acid, and the sample was then filtered under vacuum. The filter was next washed three times with 3 ml of ice-chilled 5% trichloroacetic acid, followed by three 2-ml washes with 100% ethanol. Filters were dried under a heat lamp, immersed in scintillation fluid, and counted in a liquid scintillation counter.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Constant Region Point Mutations Attenuate Apidaecin Activity to Varying Extents-- Taking the model of constant and variable regions as a platform for further study, we carried out an alanine scan of the conserved sequence in apidaecin, isotype Ho+ (sequence from Ref. 11: GKPRPQQVPPRPPHPRL; evolutionary conserved residues in bold). "Ho+" was selected on the basis of its broad and potent activities against clinical isolates from the Enterobacteriaceae family,² including E. coli strain ATCC 25922 commonly used in standardized antimicrobial assays (45). Agar plate diffusion tests were used to assay antibacterial activity of all synthetic analogs. The results of serial dilution experiments using wild type Ho+ (see Table I) initially served to devise a scoring system for loss-of-activity based on shrinking inhibition zones (····· corresponds to the biggest zones; · denotes the smallest visible inhibition); a decrease by one such arbitrary unit corresponds to a roughly 1-2 order of magnitude drop in activity.

View this table: [in this window] [in a new window]   Table I Plate inhibition zone assays of serially diluted apidaecin (Ho+ isotype) Conditions and bacterial strain identification are given under "Experimental Procedures"; all strains were grown on nutrient agar. Peptides were applied in 3-mm wells, 0.05-50 nmol/well. Number of dots express the diameter of the inhibition zones and can be read using the following key: ·, 5-7 mm; ··, 7.5-9.5 mm; ···, 10-12 mm; ····, 12.5-14.5 mm; ·····, 15 or more mm.  denotes that no inhibition zone was observed.

As predicted, and in keeping with earlier observations from a random mutagenesis study (46), replacing conserved Pro or Arg residues (which define the "PR-rich" peptide antibiotics) with Ala reduced apidaecin inhibitory effects on all bacterial strains tested (Table II). Attenuation resulting from each of the four Pro  Ala point mutations varied to some extent, in a species-dependent manner. Analog "P10A" was considerably less potent than "P13A" to inhibit growth of E. coli, Klebsiella, and Shigella strains but, by comparison, worked better against Acinetobacter. Even more pronounced differences were found between the two Arg  Ala analogs. From the inhibition zone data, we estimated that analog "R16A" was at the very least 2 orders of magnitude less active than analog "R11A" and 3 orders less than wild type Ho+. To investigate if the strict Arg conservation was solely based in its ionic nature, a double mutant (R11K/R16K) was synthesized and found to exert greatly reduced activity, suggestive of a role beyond merely contributing to charge state or amphiphilic structure.

View this table: [in this window] [in a new window]   Table II Antibacterial spectra of 15 apidaecin (Ho+ isotype) analogs Listed are the results of agar plate inhibition zone assays. Conditions and bacterial strain identification are given under "Experimental Procedures"; all strains were grown on nutrient agar. Peptide analog sequences are all derived from apidaecin Ho+ (GKPRPQQVPPRPPHPRL; taken from Ref. 11) as follows: P10A, denotes that Pro in position 10 has been replaced with Ala (and likewise for others); R11K/R16K, arginines in positions 11 and 16 replaced by lysines; L17L, Leu in position 17 (C terminus) replaced with D-Leu; Ho+(1-16), C-terminal Leu removed. Peptides were applied in 3-mm wells, 50 nmol/well. Number of dots express the diameter of the inhibition zones and can be read using the following key: ·, 5-7 mm; ··, 7.5-9.5 mm; ···, 10-12 mm; ····, 12.5-14.5 mm; ·····, 15 or more mm.  denotes that no inhibition zone was observed.

His-14 is conserved among fly drosocin and all known apidaecins (9, 11). Since this residue is often located at enzyme active sites and known to function as axial ligand and in charge relay systems (47), we speculated it might play an important role in exercising the broad antibacterial activities. This turned out not to be the case. His  Ala, His  Gln, or His  Arg mutations had only modest effects on inhibiting viabilities of various Enterobacteriaceae (Table II); by contrast, the "H14A" analog, but not "H14R," had lost all lethal activity against Agrobacterium, a plant-associated microorganism and a very susceptible target of apidaecin action (7).

Mutational analysis of the conserved C-terminal Leu yielded valuable insights into apidaecin structure/function as well. The results are shown in Table II and can be summarized as follows. C-terminal truncation by a single amino acid completely abolished all activity against all strains. A Leu  Gly substitution also resulted in a major loss of activity, ranging from an estimated 1 order (against E. coli D22 and S. typhi) to several orders of magnitude (Klebsiella and Agrobacterium). This functional deficit was partially restored in the Leu  Ala analog ("L17A"), indicating that the small CH₃ moiety (Ala versus Gly) has a very enabling effect to enhance activities against Enterobacteriaceae, by an estimated 100-fold in the case of Klebsiella, for example. Addition of a polar functionality, such as an OH group (Ser versus Ala), did not significantly alter those activities, thus dispelling any notions that a hydrophobic tail would be critical. Neither is chirality, in fact. In contrast to the all-D-apidaecin enantiomer, which lacks any measurable activity (44), replacing the C-terminal L-Leu with D-Leu (analog "L17L") did not result in dramatic reduction of activity against most strains (10-fold), except for Agrobacterium.

All-L but Not All-D-Apidaecin Enantiomers Are Stably Retained by E. coli Cells-- Although the broad physiological role of each amino acid in the apidaecin sequence had now been established from the overall effects of various permutations, we couldn't begin to formulate a mechanistic explanation as to why, for instance, mutations of different residues in the core domain resulted in such varied loss of function. To provide a bridge between these initial results and the biochemical methods commonly used to probe binding of peptides to cellular receptors, or transporters in the case of E. coli, we sought to determine conditions under which interactions of wild type apidaecin (isotype Ho+) were primarily specific. Nonspecific binding is largely due to ionic and hydrophobic effects; all-D Ho+ enantiomer is identical to wild type Ho+ in this regard. On the other hand, the all-D enantiomer is also totally inactive (Ref. 44; Table III) and, at least in principle, excluded from chirality-based interactions with cells or any components thereof. To facilitate monitoring, we synthesized ¹⁴C-labeled all-L and all-D peptides (each containing an N-terminal [¹⁴C]Gly) and measured uptake and stable association with E. coli ATCC 25922 cells. Radiolabeled peptides were of the highest purity (by liquid chromatographic and mass spectrometric criteria) and specific activity (measured by scintillation counting) and were structurally and functionally indistinguishable from their unlabeled counterparts (data not shown). We routinely exposed 5 × 10⁹ cells (colony forming units; cfu) to 5-10 µg of labeled peptide for 30 min at 22 °C; cells were then washed free of unbound material as described below. This provided an experimental system with approximately 1 mg of total cellular protein (as assayed) and all-L Ho+ ("LHo+") concentrations that were sufficient to effect a 3-log₁₀ drop in cell viability (to ~5 × 10⁶ cfu) but low enough to allow competition experiments with adequate excess of competitor.

An elaborate washing procedure was developed on the basis of yielding maximal contrast between LHo+ (highest stable association) and DHo+ (minimal association) peptides. Under these conditions, and using 150 mM NaCl in phosphate-free solutions, very few cells (0.1%) were lost during washing and integrity of the outer membrane and its lipopolysaccharide coat remained unaffected (neither stabilization nor disintegration). At the end of the procedure (after 11 consecutive washes; see Fig. 1A), 30.7(±0.6)% of LHo+ and only 4.0(±0.8)% of DHo+ remained associated with the cells, a statistically significant difference (n = 5). Higher salt concentrations (0.5 M) did not yield any better or faster release of peptide (data not shown). Upon additional (5-10 more) washes with standard solution, however, DHo+ associated counts continued to elute in the supernatant, whereas LHo+ retention remained stable. Much of the initial "uptake" of DHo+, then, would appear to represent weak or transient interactions with the cells, perhaps by passively traversing the outer membrane. By contrast, 49.2% of radiolabeled all-D-magainin (D-Mag), a pore-forming peptide (37, 38, 41, 44), was effectively retained by the same number of cells following identical treatment and washing procedures. Fewer (29.5%) all-L-magainin (L-Mag) molecules became stably associated under those conditions, a finding that is consistent with the presence of proteases in bacterial cultures, digesting all-L but not all-D peptide enantiomers. Interestingly, then, the 8-fold higher retention of LHo+ than DHo+ would appear to be occurring despite ongoing proteolysis of the L-form.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Stereoselective apidaecin uptake assay. Elution of 14C-labeled peptides from E. coli ATCC 25922 cells by repeated washing following the initial uptake reaction (A). Cells (5 × 109/ml) were incubated with labeled all-L- or all-D-apidaecin (isotype Ho+) or all-L- or all-D-magainin-2 peptide (10 µg/ml) for 30 min at 22 °C in 1-ml volume reactions. At the end of the incubation, cells were pelleted, washed, and supernatants collected and counted as described under "Experimental Procedures." The percent retention is calculated as shown in Equation 1, (Eq. 1) where T is the total number of 14C-peptide-derived counts (in cpm) added to the cell suspension; Wo is counts recovered in the supernatant following pelleting of the cells after incubation; Wn is counts eluted during the nth (n = 1-11) wash. After the final (i.e. 11th) wash, the cell pellet was resuspended, and the counts associated with the cells were determined directly. The data points shown are averaged from several (n = 5-6) independent experiments. The value of the final percent retention is referred to as the percent uptake in subsequent figures and the text. Elution of labeled peptides from E. coli 25922 wild type and 25922ApR (Ho+)-resistant mutant cells by repeated washing (B). Conditions were as under A, except that 1010 cfu/ml and 5 µg/ml of the peptides were used. wt indicates wild type ATCC25922 cells; mut indicates Ho+-resistant 25922ApR mutant cells.

Post-treatment competition (2 h at 22 °C) with a 1000-fold molar excess of unlabeled LHo+ did not remove any cell-associated peptide (0.2% over the 11th wash), indicating that it either did not reside on the surface or that "off-rates" were very slow. Less than 1% was recovered by cold chase in the presence of Tris·EDTA (50·5 mM), agents known to release lipopolysaccharide from Gram-negative cells and severely permeabilize the outer membrane (48), again suggestive of a near total absence of labeled LHo+ from the periplasmic space or low off-rates to whatever it might be bound. At that point, and under the conditions as described in Fig. 1, we calculated a stable uptake of 187,000 LHo+ molecules per cell. By comparison, about 240,000 D-Mag molecules became permanently associated with each cell within the same period. Yet no detrimental effects on cell viability could be measured, whereas LHo+ caused a 500-fold reduction, establishing a significant difference in specific activity.

Apidaecin Antibacterial Effect Is Proportional to the Number of Molecules Retained per Cell-- Uptake rates as shown in Fig. 1A were very consistent. In fact, the data in Fig. 2 indicated that, within certain limits, experimental variables such as incubation time (5-180 min), temperature (0-37 °C), and growth phase of the cells (0.2-1.0 A₆₀₀) did not influence the outcome of the uptake assays by much. The time course experiments deserve some comment. LHo+ caused major lethality within 5 min of exposure (2.5-log₁₀ decline in cell viability); the kill rose to 2.7-log₁₀ at 30 min, but no further increase was noted after that. However, kill could be drastically reduced by washing the cells with saline solution 1-2 min after addition of the peptide (data not shown). Apparently, during this short exposure, LHo+ had not yet entered all cells, and it washed out after disrupting ionic interaction with the negatively charged outer envelope. This could no longer be done after 10 min, indicating that full lethal activity had been exerted.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Effects of incubation time, temperature, and cell growth phase on stereoselective (L-form) apidaecin uptake by E. coli. Conditions were as described under Fig. 1, except for the experimental modifications listed hereinafter. A, incubation time. 1010 cfu/ml cells were incubated with 5 µg of LHo+ for the times indicated. A 30-min incubation time was arbitrarily selected as the standard (100%), and other time points are expressed relative to this value. Results are averaged from two independent experiments. Viability plating was done in triplicate for all experiments (data not shown). B, temperature. 3 × 109 cfu/ml cells were incubated with 5 µg of LHo+ at the temperatures indicated. The 22 °C temperature was arbitrarily selected as the standard (100% value). C, cell growth phase. Cells were grown in 100% BHI to an A660 as indicated and then prepared for the uptake assay as described under "Experimental Procedures." Cells from each culture were resuspended at a density of 5 × 109 cfu/ml and incubated with 9 µg of LHo+. The A660 = 0.6 was arbitrarily selected as the standard (100% value).

Amount of peptide and number of cells, on the other hand, were the critical variables in our uptake experiments, as shown in Fig. 3. In case of uniform cell density (7 × 10⁹ cfu/ml; Fig. 3B), we noticed a logarithmic relationship between the amount of peptide that was added and either uptake (molecules per cell) or kill (fold reduction of cfu). For a fixed amount of peptide (10 µg of LHo+; Fig. 3A), an inverse logarithmic relationship existed between cell number (cfu) and uptake or kill. Taking all these data points into consideration, a near-perfect logarithmic relationship (y = 0.0003 x^1.1788) was found to exist between the number of LHo+ molecules taken up per cell (under standard assay conditions) and the kill (Fig. 3C). This equation, in turn, also allowed us to screen particular conditions, peptide analogs, or mutant cells for an uptake-versus-kill relationship that might deviate from the norm and therefore could provide useful tools to elucidate mechanistic features of LHo+ activity.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Apidaecin-induced kill is proportional to the number of stereoselectively (L-form) retained molecules per E. coli cell. Conditions were as described under Fig. 1, except for the changes in cell density and peptide amount listed hereinafter. A, uptake and bactericidal activity of 10 µg/ml LHo+ as a function of cell density. Data points, plotted as molecules LHo+ incorporated (on a per cell basis), or kill, against density (cfu/ml; 1-ml reaction volumes), are corrected for removal of viable plate aliquot. Cell kill by LHo+ is obtained by subtracting viable cell numbers (cfu) at the end of the incubation from those at t = 0. Molecules incorporated per cell are calculated as follows: LHo+ input (in moles) × % uptake (relative to input) × Avogadro's number × (t = 0) cfu1. Data points are averaged from two or three independent experiments. B, uptake and bactericidal activity of LHo+ as a function of peptide concentration at a uniform cell density of 7 × 109 cfu/ml. Data points (corrected as in A) are plotted as molecules LHo+ incorporated (on a per cell basis), or kill, against peptide concentration (µg/ml; 1-ml reaction volumes). C, relationship between cell kill and uptake of LHo+ molecules per cell. The best fit curve for the data points (black circles) in A and B is shown. For illustration, the data for Ho+-resistant mutant described in Fig. 1B is plotted on the curve (open square).

A first such case was presented by the isolation of a spontaneously emerging E. coli ATCC25922 mutant (25922 ApR) that was partially resistant to LHo+ activity. Apidaecin MIC for 25922ApR was 100 times higher as compared with the parent strain; by contrast, D-Mag MICs were similar for both parent and mutant cells (data not shown). As shown in Fig. 1B, LHo+ was inadequately retained by the ApR mutant ("mut") in our uptake assay, 24.6% of 5 µg of peptide (37,600 molecules/cell) versus 47.9% (73,300 molecules/cell) stably retained by the parent. From the slopes of the wash-out curves in Fig. 1B, it could be extrapolated that the difference in LHo+ "elution" from mutant and wild type cells would continue to grow with additional washes. Stable uptake of D-Mag was comparable for both cell types (>60%), suggesting that reduced LHo+ uptake by 25922ApR may not have been the result of a generic barrier mutation alone (49-51) but rather of an altered "receptor." Importantly, the kill (as measured in the uptake assay) was also reduced by more than 2-fold (from 2.25 log₁₀ to 1.9 log₁₀), in accordance with the previously established molecules/cell-kill relationship (Fig. 3C); the data point obtained for the mutant strain fits the curve when superimposed on the graph in Fig. 3C. It would appear therefore that, after impeded entry into the cells, LHo+ peptides inhibited viability with typical efficacies.

Reduced uptake (40% of the control) is also the sole reason for diminished LHo+ effects on wild type ATCC 25922 cells in the presence of either Mg²⁺ ions (0.5 mM) or moderate concentrations (25 mM) of salt, conditions that apparently did not impede D-Mag activity by much (Ref. 41; Fig. 4A). Again, the molecules/cell-kill data points fit the standard curve quite well (Fig. 4B), indicating undiminished activity of those molecules that had crossed the cell membrane.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Mg2+ ions and salt inhibit stereoselective (L-form) apidaecin uptake by E. coli. Conditions were as described under Fig. 1, except that the uptake assay medium was supplemented with 0.5 mM MgCl2 or 25 mM NaCl, in separate experiments. Uptake of 9 µg/ml LHo+ and 2.5 µg/ml D-magainin by 5 × 109 cfu/ml cells is expressed relative to the corresponding control experiments (no Mg2+ ions; no NaCl) (A). Data shown are averaged from duplicate experiments. Viability plating was done in triplicate for all three experiments where LHo+ was added. Cell kill versus uptake of LHo+ molecules per cell is plotted (open symbols) in B, superimposed on the best fit curve from Fig. 3C (black circles).

Apidaecin Uptake Is Dependent on Cell Energetic State and Can Be Blocked by L-Pro but Not by D-Pro-- Contrary to stationary insertion in lipid bilayers, active peptide transport through the periplasm and permease/transporter-mediated uptake by Gram-negative bacteria requires an electrical potential (i.e. an energized cytoplasmic membrane) or some source of expendable cellular energy, usually ATP (52-56). Thus, E. coli cells were starved for 48 h by vigorous aeration (shaking at 150 rpm) in minimal medium and then collected and resuspended for the standard uptake assay. Alternatively, cells were pretreated in assay medium containing 5 mM 2,4-dinitrophenol, a known uncoupler of oxidative phosphorylation (55, 56). Pretreatments caused no measurable loss of viability (data not shown). In each case, however, LHo+ uptake was reduced to less than half of the control experiment (no starvation/uncoupling), with corresponding loss in lethal activity (Fig. 5A), i.e. lower uptake fully explained partial loss of function (Fig. 5C). Neither starvation nor uncoupling had any apparent effect on the uptake of D-Mag, a natural pore-forming peptide.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Cell energetic state and molar excess L-proline affect stereoselective (L-form) apidaecin uptake by E. coli. Conditions were as described under Fig. 1, except for the experimental changes listed hereinafter. A, effects of starvation and uncoupling of oxidative phosphorylation. Cells were grown as described under "Experimental Procedures," harvested, and resuspended at a density of 2 × 1010 cfu/ml. One 5-ml aliquot of the cells was immediately incubated with LHo+ or D-magainin (2.5 µg/ml) to determine control uptake levels. A second aliquot was diluted to pre-harvest density in fresh 10% BHI and incubated for 2 days at 30 °C, with shaking (150 rpm); starved cells were then washed, resuspended, and assayed for uptake as for control cells. Uptake is expressed relative to the corresponding controls. Numerous replicate viability plate counts indicated that over the 2-day starvation period, cells either did not grow or underwent at most one cell doubling. Viability plating was also done in triplicate for all experiments where LHo+ had been added (control and starved cells). In separate experiments, 2,4-dinitrophenol (5 mM final concentration) was added to the cell suspension for 3 h at 30 °C, with shaking; cells were then pelleted, washed with 10% BHI, and resuspended (1010 cfu/ml) in assay broth, before addition of the peptides. Uptake and viability assays were as above. B, competition experiments. In separate experiments, the uptake assay medium was supplemented with 10 mM (final concentration) all-L Arg-Pro-Lys-Pro, or L-proline, or D-proline. Uptake of 2 µg/ml LHo+ or 2.5 µg/ml D-magainin (~1 µM concentrations) by 2 × 1010 cfu/ml cells is expressed relative to the corresponding control experiments (no additives). Data are averaged from duplicate experiments. Cell kill versus uptake of LHo+ molecules per cell, resulting from the experiments shown in (A and B), is plotted (open symbols) in C, superimposed on the best fit curve from Fig. 3C (black circles).

To substantiate further the transporter-mediated model, we carried out LHo+ uptake experiments in the presence of a 10,000-fold molar excess of a commercially available tetrapeptide, RPKP, chosen for its fitting resemblance of the apidaecin sequence. A 10-fold reduction in uptake was noted, whereas, importantly, the uptake of D-Mag was very little affected (Fig. 5B). A 2-fold reduction was also achieved by the addition of L-Pro (10,000-fold molar excess), an effect that could not be reproduced by the addition of D-Pro and that did not affect D-Mag uptake either (Fig. 5B). The stereospecific element of proline interference with LHo+ uptake by E. coli cells was also manifest in collateral loss of antibacterial activities; L-Pro was antagonistic, whereas D-Pro was not (Fig. 5C).

Mutations in the Apidaecin Structure Can Suppress Function at Two Stages of Interaction with E. coli Cells-- The uptake assay had now been optimized to the point where the issue of analyzing the functional role of individual amino acids in the apidaecin structure could be revisited. To limit laborious chemical synthesis of radiolabeled peptides, MIC values of 18 different LHo+ analogs/truncations were first determined for E. coli 25922 (10³ cells/ml). The data in Table III were in good agreement with the results of the earlier disc diffusion tests (Table II) and provided a more accurate and quantitative estimate of reduction in activity, which varied over at least 4 orders of magnitude. Five peptides were then selected for further study as follows: one Pro mutant, one Arg mutant, two C-terminal Leu mutants, and one C-terminal truncation. LHo+ and DHo+ served as positive and negative controls, respectively. The results of the uptake assays (shown relative to LHo+ uptake) are compared, side-by-side, to the reduction in killing activity (expressed as x-fold increase over the MIC of LHo+) in Fig. 6.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Effects of amino acid substitutions in apidaecin (L-form) on stereoselective uptake and antibacterial activity. Upper panel shows the percent uptake of selected LHo+ analogs by E. coli ATCC 25922 cells, relative to that of wild type peptide (100% value); DHo+ was included as a negative control. Uptake assays were performed as described under "Experimental Procedures" and Fig. 1; data points are the average of two independent experiments, each carried out in triplicate, for each peptide. Cell density was 5 × 109 cfu/ml; and peptide concentration was 2.5 µg/ml. Corresponding results of MIC assays, for each peptide, are taken from Table III and shown in the bottom panel. MIC assays were as described under "Experimental Procedures," using a cell density of 1,000 cfu/ml. Data are presented as x-fold increase of MIC values (i.e. reduction of antibacterial activity) as compared with the wild type peptide (MIC = 0.01 µg/ml; from Table III) which was assigned an arbitrary value of 1. Key to the analog labels is to be found in Tables II and III; D-17 corresponds to L17D-L in Table III (C-terminal Leu replaced by D-leucine).

Evidently, proper peptide uptake is necessary but not sufficient for full antibacterial activity. Indeed, uptake of analogs L17L (Leu-17 replaced with D-Leu) and L17G was unabated (122 and 92% of the control, respectively), yet their activities were 10- and 80-fold reduced, respectively; and a C-terminally truncated peptide (missing only Leu-17) that had no measurable activity at all (at least 4-log₁₀ less than the control) was also quite well retained by the cells (47% of control). Analog P10A, on the other hand, was very poorly retained (about the same as DHo+) but still 100 times more active than the all-D enantiomer, at least, indicating that those P10A molecules that made it inside the cells exerted activity, albeit less than usual. By contrast, analog R16A was not only inadequately retained (22% of control) but also appeared deficient in subsequent lethal action, resulting in a 2,500-fold overall reduction of activity. Collectively, the data established apidaecin lethal function on E. coli cells as a sequential process of at least two steps, transporter-mediated uptake and subsequent interference with a vital cellular process, and hence is likely to involve diverse molecular targets as well.

Apidaecin Blocks Protein Synthesis in Vivo-- To begin identifying processes and molecules that might be targeted by apidaecin after entering E. coli cells, we studied its effects on biopolymer synthesis, a logical choice because many non--lactam antibiotics function this way (48). In a first experiment, subinhibitory concentrations (1/2 MIC) of antibiotics known to block replication (nalidixic acid; inhibitor of DNA gyrase), RNA synthesis (rifampin; inhibitor of RNA polymerase), and protein synthesis (tetracycline; blocks aminoacyl-tRNA binding to the ribosome) were evaluated for possible synergistic, or antagonistic, activities on the ability of apidaecin to inhibit cell growth in culture. Under those conditions, only tetracycline was found to lower the apidaecin MIC by at least 1 order of magnitude (Table IV); no such cooperative effect was observed on the MIC of all-D-magainin.

In a subsequent experiment, 8 × 10⁹ cells were pretreated with 8 µg of tetracycline for 15 h before addition of radiolabeled LHo+, or D-Mag, in otherwise unmodified uptake assays. As shown in Fig. 7A, uptake was reduced by 70 and 60%, respectively, as compared with the controls (no pretreatment). Lower apidaecin uptake could readily be explained, although no proof for this exists, by assuming gradual turnover of a transporter-type protein, without replenishment because of blocked synthesis by tetracyline. The underlying reason for a tetracycline-dependent decrease of D-Mag uptake was unclear and was not further investigated. A rather unexpected observation, however, was made in the course of these experiments, namely a 50-fold drop in apidaecin-induced kill (from 2.6 log₁₀ to less than 1 log₁₀) of tetracycline-pretreated E. coli cells. Remarkably, this is an order of magnitude less than what might have been expected from the uptake result (in molecules/cell), according to the earlier established relationship; the data point maps way below the curve (Fig. 7B). Why is it that wild type Ho+ peptides are seemingly not working under those conditions? Viability plating, as used here, determines bactericidal activity only; hence, tetracycline bacteristatic effects, although manifest in MIC assays, will essentially go unnoticed (8 µg caused no reduction in cfu; data not shown). The fact remains, however, that the drug will bind to the ribosome (30 S subunit), which could possibly result in competition with any bactericidal substance targeted at the same or overlapping site, ultimately causing a net loss of lethal activity (48, 57). If this would indeed be the case for apidaecin, one might also speculate that it could have adverse effects on protein synthesis of its own, a theory entirely consistent with the earlier established apidaecin/tetracycline cooperative inhibition of growth.

View larger version (16K): [in this window] [in a new window]   Fig. 7.   Stereoselectively (L-form) retained apidaecin by tetracycline-pretreated E. coli cells is poorly bactericidal. Conditions were as described under Fig. 1, except that 8.3 µg/ml tetracycline was added to the cell suspension for 15 h at 30 °C, with shaking; cells were then pelleted, washed with 10% BHI, and resuspended in assay broth before addition of the peptides. Uptake of 9 µg/ml LHo+ and 2.5 µg/ml D-magainin by 8 × 109 cfu/ml cells is expressed relative to the corresponding control experiments (15 h at 30 °C; no tetracycline added) (A). Data shown are averaged from two independent experiments. Viability plating was done in triplicate for both experiments where LHo+ was added. Cell kill versus uptake of LHo+ molecules per cell is plotted (open symbols) in B, superimposed on the best fit curve from Fig. 3C (black circles).

To verify this model, in vivo effects of apidaecin (LHo+; 1.7 µg/ml) on protein synthesis were studied directly by monitoring incorporation of radiolabeled amino acids, [³H]leucine and [³⁵S]methionine; tetracycline (8.3 µg/ml) was used as a positive control for inhibition. To exclude the possibility that any observed block of protein synthesis could have been the consequence of a block in RNA synthesis or the abrupt result of any otherwise induced arrest of cell functions, incorporation of [³H]uridine was also monitored in separate experiments. Rifampin (83 µg/ml) served as control for RNA synthesis inhibition. The results presented in Fig. 8 clearly show that, unlike rifampin that blocked both RNA and protein synthesis, LHo+ and tetracycline selectively arrested protein synthesis. Whereas, compared with the control (no drugs added), leucine incorporation was about five times lower after a 60-min exposure to LHo+ (Fig. 8B), uridine incorporation was reduced by only 10% (Fig. 3A). To rule out that apidaecin somehow might have prevented leucine from entering the cells (by interference with permease activity, for example), the experiments were repeated with methionine and yielded the exact same results; LHo+ blocked protein synthesis very rapidly, in a dose-dependent manner (Fig. 8C).

View larger version (27K): [in this window] [in a new window]   Fig. 8.   Apidaecin blocks protein synthesis in E. coli cells. Incorporations of [3H]uridine into RNA (A), [3H]leucine (B), and [35S]methionine into protein (C) were monitored as described under "Experimental Procedures." LHo+ peptide concentrations were 1.7 (1×) or 17 µg/ml (10×); 8.3 µg/ml tetracycline (Tet) and 83 µg/ml rifampin (Rif) were used; controls didn't contain peptide or drug. Aliquots were withdrawn at the indicated time points, and trichloroacetic acid-precipitable counts were measured (listed as cpm).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Apidaecins are the largest group of proline-arginine-rich antimicrobials known to date. Small, gene-encoded, unmodified peptides, they work rapidly with essentially irreversible effects (7, 11, 43). Like PR-39, a mammalian relative, apidaecin is non-lytic for bacterial cells (44, 58). Earlier studies suggested critical dependence on a conserved core sequence and on chiral interactions with undefined targets, receptors, in E. coli cells (11, 44). It was hard to conceive, however, that all the peptide needed to do was just bind to a surface protein for the cells to get killed. Unlike mammalian cells, Gram-negative bacteria have a cell envelope that consists of an outer (OM) and inner (IM) membrane, separated by the periplasmic space, thus making uptake of peptides, be it a natural or lethal process, more complicated. Whereas the IM resembles a conventional phospholipid bilayer, the OM has pores and a substantial lipopolysaccharide component that endows the bacterial surface with strong hydrophilicity and negative charge (59, 60). Cationic polypeptides are attracted and, in some cases, can disrupt the OM and enter the periplasm, as shown for polymyxin, defensin, hymenoptaecin, and others (10, 34, 35, 51, 60). It was possible that apidaecin might do the same and, therefore, would have both chiral and non-chiral elements to its actions. To separate specific from such nonspecific interactions, an uptake assay was developed that allowed us to discriminate between wild type apidaecin and its inactive all-D enantiomer. Because of the very nature of the L- and D-forms, all observed differences had to be stereospecific and not due to ionic or hydrophobic binding.

In a typical experiment, all-D-apidaecin would become rapidly associated with the cells but could then be almost entirely recovered by exhaustive washing, whereas the all-L forms and both all-L- and all-D-magainins could not. The need for repeated (10-15 times) washings, dialysis style, to remove all-D-apidaecin implied a slow outward diffusion through the OM, in turn offering some rational ideas about the initial phase of apidaecin action. The peptide likely permeates and traverses the OM in a nonspecific fashion, an apparently irreversible process due to sequential stable binding to a periplasmic or IM component, or through further translocation. This secondary interaction has to be stereospecific for it does not occur with the D-form; it also doesn't apply to magainins since both enantiomers insert equally well in the IM (37, 41), preventing washout. It was this chiral aspect of apidaecin uptake that was specifically measured in all further experiments, leading to the observation that a direct logarithmic relationship existed between the number of peptides per cell thus retained and the overall induced kill. This allowed us to evaluate a series of mutants and conditions known to attenuate apidaecin bactericidal activity. The results unfailingly upheld the model of permease/transporter-mediated uptake. First, a stereospecific reduction in uptake resulted from induced depolarization of the IM and from general energy depletion of the cells. Second, specific binding could be competed in part with an excess L-Pro but not D-Pro. Pretreatments or competition had no effect on all-L- and all-D-magainins. Furthermore, an E. coli mutant exhibiting increased resistance to apidaecin showed also reduced specific uptake of the peptide, whereas magainin uptake was unchanged, thus excluding a simple OM barrier mutation.

Convincing evidence was also obtained that specific engagement of a bacterial binding protein, or transporter, is not the end point of apidaecin action. Uptake experiments using apidaecin point mutants, previously determined to be highly inactive, revealed that some analogs entered the cells poorly, thus explaining lack of function. Others, however, entered just fine, and yet caused much lower mortality than expected. Mutation or truncation of the C-terminal leucine, especially, had moderate to no effect on specific uptake but resulted in 1 to several orders of magnitude reduced kill. We postulated that these particular analogs had to be defective in disrupting some vital cellular process. An indication to what this putative downstream target might be came from the observation that a similar block in lethal activity of internalized, wild type peptides could be achieved by pretreatment of E. coli cells with tetracycline, a bacteristatic agent known to bind the 30 S ribosome. This antagonistic effect led us to test, and demonstrate, the ability of apidaecin to inhibit specifically protein synthesis in vivo in a dose-dependent manner, an observation that was not the result of a block in either amino acid uptake or RNA synthesis.

All of the findings summarized argue for a multistep pathway of apidaecin entry into the cells. The proposed mechanism involves an initial, nonspecific encounter of peptide with an OM component, followed by invasion of the periplasmic space, and by a specific and essentially irreversible engagement with a receptor/docking molecule that may be IM-bound or otherwise associated, most likely a component of a permease-type transporter system. In the final step, the peptide is translocated into the interior of the cell where its ultimate target, components of the protein synthesis machinery, most likely the ribosome, are adversely encountered.

Because of location (IM), function (transport of amino acids, small peptides, and many other small solutes), and ATP-dependent activity, members of the larger family of traffic "ABC transporters" (61) are all potential candidates to mediate apidaecin uptake. Many of these systems also involve periplasmic binding-proteins, permeases, which serve as the specific receptor for the molecule to be transported (62, 63). We speculate that one of these could be utilized by apidaecin to secure untimely entry into the cells. Two types of amino acid permeases exist in E. coli, shock-resistant (membrane-associated) and shock-sensitive (free in periplasm). Previous studies have indicated that, in general, shock-resistant systems are coupled directly to energized membrane state, whereas shock-sensitive ones require synthesis of ATP (56). So, our data do not rule out either one or the other. It is interesting, though, that Pro permeases (shock-resistant) have been found to be inhibited significantly by Arg-Pro dipeptides (RP; the signature sequence of both apidaecin and PR-39), whereas other Pro-containing dipeptides had no such effect (64). It is therefore possible that a proline porter (65) could be involved, but it remains speculative at this time. Another candidate may be the product of the oppA gene, known to bind a wide range of peptides (two to five amino acids in length, without regard to sequence) and to serve as an initial receptor for peptide transport across the cell membrane in Gram-negative bacteria (66-69). However, it has been shown that the OppA protein (i) completely engulfs the peptides (70), yet (ii) fails to bind single amino acids (69), which is difficult to reconcile with binding of a 17-residue long apidaecin and with our finding that L-proline effectively competes for uptake in a stereospecific manner. A final consideration that apidaecin might enter E. coli cells by utilizing host OM proteins for translocation in the same fashion as colicins and phages do (59, 71) can essentially be eliminated because all-D enantiomers seem to cross the OM just as well, indicative of self-promoted entry.

Several peptide antibiotics have already been shown to inhibit bacterial protein synthesis. However, it almost entirely concerns molecules of bacterial origin that are rather complex biosynthetic products, some cyclic and containing very unusual and D-amino acids (72, 73). Microcin C7 (7 amino acids), even though synthesized on the ribosome, is also uniquely modified afterward, containing a phosphodiester of 5'-adenylic acid and n-amino propyl alcohol (74, 75). PR-39 comes closest to apidaecin in structure (Pro-Arg-rich), source (multicellular organism), and ability to stop protein synthesis (26, 39, 58). The peptide is much longer, however, and little is known about how it enters the cells. Thus, our studies have provided important new insights into the antibacterial mechanism of a distinctive group of peptides, distributed among insects and vertebrates. We introduce a general model for mode-of-action to be used as a platform for future biochemical and genetic studies.

	ACKNOWLEDGEMENTS

We thank Scott Geromanos and Brian Marley for expert advice and help with peptide synthesis and quantitations and Gordon Freckleton for assistance with computer graphics. The Sloan-Kettering Core Facilities are supported by NCI Core Grant P30 CA08748 from the National Institutes of Health.

	FOOTNOTES

^* This work was supported in part by the Hirschl Trust (to P. T.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ To whom correspondence should be addressed: Memorial Sloan-Kettering Cancer Center, 1275 York Ave., New York, NY 10021. Tel.: 212-639-8923; E-mail: p-tempst@mskcc.org.

² M. Castle and P. Tempst, in preparation.

	ABBREVIATIONS

The abbreviations used are: Fmoc, N-(9-fluoroenyl)methoxycarbonyl; cfu, colony-forming units; RP-HPLC, reversed phase-high performance liquid chromatography; MALDI-TOF MS, matrix-assisted laser-desorption/ionization time-of-flight mass spectrometry; DMF, dimethylformamide; BHI, brain heart infusion; MIC, minimal inhibitory concentration; D-Mag, all-D-magainin; L-Mag, all-L-magainin; OM, outer membrane; IM, inner membrane.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES