Membrane interactions of designed cationic antimicrobial peptides: The two thresholds^†

Materials

Reagents for peptide synthesis, cleavage, and purification included Fmoc-protected amino acids (Novabiochem); dansyl- and dabcyl-chlorides (Molecular probes, Eugene, OR); Fmoc-PAL-PEG-PS-resin, piperidine (Applied Biosystems, Foster City, CA); DMF, methanol, diethyl ether, acetonitrile, (Caledon Laboratories, Georgetown, Ontario, Canada); DIEA, triisopropylsilane (Aldrich); HATU (GL Biochem, Shanghai, China); ultrapure buffer-saturated phenol (Invitrogen). Lipids—POPC, POPE, POPG, and cholesterol (Avanti Polar Lipids, Alabaster, AL). Reagent kits for micro BCA protein assays were from Pierce (Rockford, IL); Mueller Hinton Broth (MHB) and Bacto-Agar were from Difco laboratories. All chemicals were used without further purification. Buffers were prepared in double-distilled water (DDW) with pH adjustment when necessary.

Peptide Synthesis and Characterization

Peptides were synthesized on a PS3 Protein Technologies Inc. peptide synthesizer by standard solid-phase protocols similar to those previously described.17, 18 For cyclization of peptide (K₇-D₁₇)7K-F18-1D₁₇ a low load (>0.15 mmol/g) PAL-PEG-PS resin was used to incorporate an amide function at the peptide C-terminus upon peptide cleavage. DIEA (1.0M in DMF) was used with a four-fold excess of protected amino acids and HATU coupling reagent. N-terminal Fmoc groups were removed at the last step of the synthesis.

The lactam ring cyclization (Scheme 1) was accomplished manually, while N-terminal Fmoc and all side-chain protecting groups were still attached to the peptidyl-resin. Side chain protecting groups—Mtt and -O-2-PhiPr on residues Lys7 and Asp17, respectively—were removed selectively (1% trifluoracetic acid (TFA), 2% triisopropylsilane, 97% dichloromethane) on ice for 30 min.23 The resin was then washed with 5% DIEA (vol/vol) in dichloromethane and the intermolecular amide bond was created through application of five-fold excess of HATU in DMF at room temperature for 24 h.

Following a standard cleavage procedure (89% TFA, 4.5% (vol/vol) phenol, 4.5% DDW, 2% triisopropylsilane), crude deprotected peptides were precipitated in cold ethyl ether, dissolved in water and purified on a reverse-phase C18 high performance liquid chromatography column (Vydac, 250 × 21.5 mm) using a linear gradient of 20–50% B (Buffer A: 0.1% TFA, 95% DDW, 5% acetonitrile; Buffer B: 0.1% TFA, 95% acetonitrile, 5% DDW) for the first 35 min, then 50–100% B for another 10 min at 10 mL/min flow rate. Absorbance was monitored at 215 nm. Major peak fractions containing product were collected, pooled and lyophilized. Molecular masses were confirmed by MALDI mass spectrometry. Peptide concentrations were determined in triplicate by standard micro BCA assay and their purities (above 95%) and retention time (R_t) values were assessed by RP-HPLC at room temperature on a Vydac C₄ column (250 × 4.6 mm) with linear gradients of Buffers A and B at a flow rate of 1 mL/min, starting with 15% B for the first 5 min, then a linear gradient of 15–70% B for another 55 min. Stock peptide solutions were stored at −20°C.

Hydrophobicity

Core segment hydrophobicity (CSH) values were calculated using the Liu-Deber hydrophobicity scale.18

Hemolytic Activity

The propensity of the peptides to cause lysis of human red blood cells (RBCs) was measured similarly as described previously.19 Freshly collected venous human blood with heparin was centrifuged in plastic round-bottom tubes to remove the buffy coat, at 1000g (2500 rpm) for 5 min at 11°C and the erythrocytes obtained were washed three times with PBS: (8.0 g NaCl, 1.44 g Na₂HPO₄ × H₂O, 0.612 g KH₂PO₄ and 0.2 g KCl per litre DDW, pH = 7.36). Lyophilized peptides were diluted in PBS and mixed with 100 μL of (4% (vol/vol) RBCs suspension to a final concentration in 200 μL of 650, 325, 162, and 81 μM in microtiter 96-well polystyrene plates (Costar). The reaction mixtures were incubated for 1 h at 37°C and unlysed RBCs were removed by centrifugation at 1000g on Beckman Model TJ-6 centrifuge for microplates for 5 min. The supernatant aliquots (100 μL) were transferred to new microplates, and the release of hemoglobin was monitored spectrophotometrically by absorbance at 540 nm on a Genesys 5 microplate spectrophotometer (Rochester, NY). PBS and 0.1% Triton X-100 were used as negative and positive lysis controls, respectively. The per cent of hemolysis at each peptide concentration was measured in duplicate according to Eq. (1).24

()

The hemolytic activity was determined as minimal hemolytic concentration (MHC)—the maximal peptide concentration that caused no hemolysis of RBCs after 1 h. Since RBCs were in isotonic medium, no detectible release of hemoglobin (<1%) was observed from negative controls.

Antibacterial Activity

The antimicrobial activity of each peptide was tested under aerobic conditions in sterile 96-well microtiter plates (Costar) in a final volume of 100 μL by following standard microtiter dilution protocols25 in MHB. Pseudomonas aeruginosa nonclinical strain wild-type mPAO1 was maintained as a glycerol stock at −80°C. Bacterial cells were grown in MHB at 37°C for overnight and were diluted in the same medium to a final concentration of 2 × 10⁴ to 2 × 10⁵ colony forming units (CFU/mL) as determined by optical density at 600 nm. Ten-microliter aliquots of serial two-fold dilutions in buffer (0.2% BSA, 0.01% acetic acid) of the lyophilyzed peptides were added to microtiter plates followed by 90 μL of bacterial suspension. Peptide antibacterial activity, expressed as the MIC—the lowest peptide concentration that resulted in 100% prevention of microbial growth as evidenced by absence of turbidity after 19 h of incubation at 37°C. Turbidity was measured as OD at 600 nm using a Genesys 5 microplate autoreader Spectrophotometer (Rochester, NY). Positive controls contained no peptide and demonstrated visible turbidity after 19 h of incubation at 37°C. All assays were carried out in triplicate.

Gel Electrophoresis

SDS-PAGE electrophoretic separation was performed at 125 mV with peptide samples dissolved in sample buffer (NuPAGE, Novex, San Diego, CA) to 35 μM, heated at 85°C for 10 min and separated on precast 12% Bis-Tris NuPage gels (1.0 mm × 10 well) in NuPAGE MES SDS buffer. In hetero-oligomerization experiments SPGS-34 was mixed in equal amounts with other peptides before loading to keep the total peptide concentration identical to that used in homo-oligomerization experiments. Coomassie Blue staining was used on all gels to visualize peptides. The MW_exp/MW_theor values were calculated by densitometry using See blue and Mark12 markers and the NIH 1.62 Image Program (software available at http://www-cellbio.med.unc.edu/henson_mrm/pages/NIH.html).26

Circular Dichroism

Circular dichroism (CD) spectra were collected in 1-mm path length quartz cuvettes on a Jasco J-720 spectropolarimeter interfaced with a PC. Spectral scans were performed from 250 to 195 nm, with step resolution of 0.1 nm and bandwidth of 1.0 nm at 50 nm/min at room temperature under a nitrogen atmosphere. Freshly prepared peptide samples (30 and 50 μM) were used in buffer (10 mM Tris-HCl, pH = 7.0, 10 mM NaCl or 1500 mM NaCl) as well as in the presence of 25 mM SDS micelles. The mean residue ellipticity ([θ_n], degree × cm² × dmol⁻¹) values at a given wavelength were calculated with CDpro software as:

()

where [θ_n]_observed is the ellipticity measured, n is number of amino acid residues, l the is optical path length of the cell in cm, and c is peptide molar concentration.

The helical content for each peptide was estimated according to Chen et al.27 as:

()

where n is number of residues. [θ] extrapolated for a 17-residue peptide in a 100% helical conformation by Eq. (3) is −33,529 degrees × cm² × dmol⁻¹. All spectra were corrected by background buffer subtraction. Background samples were prepared using the same protocol except pure water instead of peptide solution was added.

Preparation of Large Unilamellar Vesicles

LUVs were prepared as described previously.28 The desired mixtures of lipids were dried in glass tubes first under nitrogen and then lyophilized overnight to obtain lipid films. The films were suspended for 1 h on a water-bath at 40–50°C in Tris-HCl buffer, pH = 7.0 (10 mM Tris, 10 mM NaCl), sealed with parafilm under nitrogen, and then freeze-thawed five times under nitrogen to produce large multilamellar vesicles. Each suspension was extruded nine times through polycarbonate membranes with 0.1-mm diameter pores (Nuclepore, Pleasanton, CA) on an Avanti mini-extruder apparatus. Vesicles were used the same day and appropriate aliquots of the desired peptides were added.

Fluorescence Measurements

Fluorescence emission spectra of peptide Trp residues were measured on a Hitachi F-400 Photon Technology International C-60 fluorescence spectrometer at excitation wavelength of 280 nm with a 2nm bandpass, and integrated between 305 and 390 nm with a 6-nm bandpass. Interactions of CAPs with zwitterionic or negatively-charged vesicles were characterized by measuring changes in emission intensity maxima wavelength λ_max (blue shifts) of the peptide intrinsic Trp relative to normal salt aqueous conditions. Samples of peptides (4 μM) in 1 mM of detergent were used in a 0.5 mL semimicro quartz cuvette (10-mm excitation path length and 4-mm emission path length) (Hellma, Concord, ON). All spectra were integrated, corrected for light scattering effects by subtraction of background, and by the correction function of FELIX software provided by the manufacturer.

Peptide Design

To obtain additional details about the mechanism of bioactivity of synthetic CAPs, as well as to improve their antimicrobial activity, while preserving selectivity, we designed a library of 12 peptides based on the 6K-F17 prototype, and comprehensively investigated their antibacterial and cytotoxic activities, secondary structures and folding in various environments, and interactions with model membranes (Table I). The peptide net charge was maintained as +7 in all analogs (including the N-terminal -NH₃⁺ moiety), while minor changes in charge distribution and length were made. In the present study we used 6K-F17 (KKKKKKAAFAAWAAFAA-NH₂) as a framework to alter core hydrophobicity by systematically replacing Ala residues with Leu to increase average core hydrophobicity. The resulting new designs represent mono, double, triple, and quadruple Ala-to-Leu substituted 6K-F17 derivatives (Table I). This designed “Leu-series” of further analogs has two key variables: the number of Leu residues and the relative position(s) of the Leu in hydrophobic core. The full library contains one isomer each with single and quadruple substitutions, five isomers of two, and two isomers of three Leu residues. To distinguish among the isomeric peptides, we specify Leu positions along the sequence by subscript numbers at the end of each peptide designation.

CAPs with inter-molecular disulfide bonds form a major class of innate immune system agents,3, 4 and peptide cyclization is a known tool for metabolic stability improvement through global conformational constraint on otherwise more flexible structure(s).30, 31 This tactic has been applied successfully on magainin II32 and mellitin33 sequences, also demonstrating that linearity of the peptides is not essential for the disruption of the target phospholipid membrane. Thus, we designed two mono Ala-to-Asp substituted peptides—one is an intra-molecular side chain-to-side chain cyclic peptide, termed (K₇-D₁₇)7K-F18-1D₁₇ (see Materials and Methods section and Scheme 1 for details of synthesis), along with its linear analog 7K-F18-1D₁₇ (Table I). The functional groups for lactam ring closure are located 12 residues apart (i, i + 12) from one another to preserve their α-helical secondary structure. To mimic the presumed membrane-bound dimeric oligomerization state of the designed 17-residue CAPs,20 a helix-loop-helix hairpin “covalent anti-parallel dimer” model peptide KKKKAAFAAWAAFAAKSPGSKAAFAAWAAFAAKKKK-NH₂ (termed SPGS-34), was synthesized. It consists of two Lys-tagged nonamphipathic cores of 11 residues each connected by the short hydrophilic β-turn mimic loop of Ser-Pro-Gly-Ser.34

Antimicrobial Activity

P. aeruginosa infections are the major cause of morbidity in adult patients with cystic fibrosis, and also represent a serious problem in patients hospitalized with cancer and burns.35 The MIC values for the present library of peptides against the gram-negative P. aeruginosa mPAO1 strain were measured (Table II). The D-enantiomer 6k-f17 was the most active peptide (MIC = 2 μM) of the series, while the SPGS-34 hairpin was the least active (MIC = 128 μM). Magainin II amide showed intermediate activity (MIC = 16 μM). The remainder of the peptides displayed good-to-moderate antimicrobial effects; their activity in general decreased as increasing Leu residues were incorporated into the sequence. The isomers with the same numbers of Leu residues showed different MICs, as previously noted in other CAPs,36, 37 although the full range was limited (8–64 μM). Cyclic and linear derivatives of 7K-F17-1D₁₇ showed similarly favorable MIC values (MIC = 8 μM), and based on previous results, at least two-fold lower values could be expected for the D-enantiomers of these peptides.

Hemolytic Activity

The toxicity of drug leads to mammalian cells is frequently expressed as hemolytic activity. The hemolytic activities of our synthetic CAPs, as well as of magainin II amide against human RBCs, were determined at four peptide concentrations (650, 352,162, and 81 μM) (Table II), and MHC values were estimated in this range. The total absence of hemolysis at the highest tested concentration (MHC = 650 μM) was observed only for the peptide with the lowest CSH18 of 1.2—7K-F18-1D₁₇ (Table I). Peptides with CSH values between 1.5 through 1.9, as well as cecropin P1, showed slight hemolytic activity at concentrations of 162 μM and above. In contrast, those peptides with CSH of 2.3 and above (with two Leu residues and higher), along with SPGS-34 and magainin II, showed modest or significant hemolytic activities at every concentration tested (MHC values below 81 μM). Overall, the most hydrophobic peptide among the Ala-substituted series of 6K-F17, 6K-F17-4L_8,11,13,16, displayed the highest hemolytic profile (up to 50%), while magainin II amide showed almost complete hemolysis at all concentrations tested. Isomers with the same number of Ala-to-Leu mutations showed similar but not identical activities; for example, 6K-F17-2L_11,13 showed the lowest hemolytic activity among double-substituted isomers.

Therapeutic Index and Selectivity

The therapeutic index is a widely employed parameter to quantify the specificity of antimicrobial reagents. It is expressed as ratio between MHC and MIC values (Table II); thus, larger therapeutic index values indicate greater clinical potential. Calculated therapeutic indexes for each peptide studied in the present work indicate an obvious decrease of selectivity with increasing Ala-to-Leu substitutions (Table II) from 162 times to <1. Among all studied peptides, 6k-f17 and 7K-F18-1D₁₇ showed the largest therapeutic indexes of 162, and >81, respectively, while the remainder of the 6K-F17 derivatives have therapeutic indexes below 10. The naturally occurring magainin II CAP showed a therapeutic index below 5 in our hands, while for cecropin A, indices in the range of 6–17 have been reported.38 Our best designs therefore showed therapeutic indices considerably superior to both of the latter natural CAPs.

Hydrophobicity

RP-HPLC retention behavior is an excellent tool to estimate relative peptide apparent hydrophobicities.39 The peptide retention times (R_t values) are shown in Table III. As expected, the tetra-substituted Ala-to-Leu peptide 6K-F17-4L_8,11,13,16 was retained the longest on the HPLC column among the Ala-substituted series, followed by triple, double, hairpin (SPGS-34), mono-Leu, cyclic, and nonsubstituted peptides, followed finally the least hydrophobic Ala-to-Asp substituted—7K-F18-1D₁₇. Magainin II amide showed the longest R_t value, reading out as the most hydrophobic peptide among those tested.

Importantly, the subset of five isomeric double-Leu peptides displayed a range of R_t values, despite the fact that all five have identical core residue hydrophobicities. For example, among double Ala-to-Leu substituted isomers, the peptide with both Leu residues located at the C-terminal, 6k-f17-2l_16,17, showed the longest R_t (highest apparent hydrophobicity; R_t = 34.2 min; Table III), while 6K-F17-2L_7,8 with both Leu residues at the N-terminal side of the core immediately following the Lys tags had the shortest retention time (31.3 min.). To verify the effect of different interactions of the isomeric peptides with the HPLC column, a mixture of all five double-substituted isomers in equal amounts was injected using the same conditions and all peaks were resolved (data not shown). This result confirms that peptide isomers have different physical properties and potentially different antimicrobial and biophysical properties as well.36

Helicity

TM Finder data analysis29 predicts that nonamphipathic segments of these charge-segregated peptides will insert into cell membranes and adopt an α-helical secondary structure. CD data performed in different media are consistent with these predictions. The studied CAPs are found largely in extended “random coil” conformations in aqueous buffer of low ionic strength (Fig. 1a), except SPGS-34, which is partially helical under these conditions (Fig. 1d). In SDS micelles, we observed typical CAPs transitions to helical-form patterns, indicating membrane insertion of the peptides, including the cyclic derivative (K₆-D₁₇)7K-F18-1D₁₇ (Figs. 1b–1d). Along the Leu-substitution series, the mean residue ellipticity at 222 nm ([θ_n]₂₂₂) generally increases with the number of Leu substitutions and indicates involvement of between 5 and 11 residues in helix formation.

Electrophoresis

SDS-PAGE electrophoresis has proven to be an efficient qualitative method for determining association states of many TM proteins, and indeed, many small membrane proteins maintain their oligomeric structure in the presence of SDS.31, 34 Interestingly, all 6K-F17 derivatives including cyclic (K₇-D₁₇)7K-F18-1D₁₇ appeared as discrete SDS-resistant dimers on SDS-PAGE based on observed MW_exp/MW_theor values of 2.1–2.5 (Table III). The monomeric behavior of hairpin SPGS-34 was the lone exception and likely represents formation of a tightly packed hairpin. Hetero-oligomerization was not observed when mixtures of SPGS-34 with either 6K-F17-2L_11,13 or 6K-F17-2L_7,8 were analyzed by SDS-PAGE (not shown), indicating only a single potential interface for peptide-peptide interactions along the hydrophobic core. In contrast, magainin II amide and cecropin P1 ran as monomers on SDS-PAGE20 (Table III).

Fluorescence

Measurement of fluorescence emission shifts in the λ_max of the single Trp residue present in the nonamphipathic cores of all Leu-substituted analogues of 6K-F17 is a particularly useful tool for analysis of their membrane insertion/conformation in various media. As well, peptide Trp fluorescence emission positions upon interaction with model cell membranes has been observed to be closely correlated with their hemolytic interactions.40 In the present work, additional evidence of CAP selectivity for bacterial vs. mammalian cell membranes was obtained by exposure of the Trp-containing 6K-F17 derivatives to freshly prepared anionic phospholipid vesicles, corresponding to a typical bacterial membrane (LUV-POPC:POPG = 75:25) vs. zwitterionic vesicles corresponding to a typical mammalian membrane (LUV-POPC, and LUV-POPE, in the presence or absence of cholesterol) (see Fig. 2). When introduced to anionic membrane models, all peptides displayed significant blue shifts in Trp λ_max values, ranging from 14 to 25 nm. These results provide evidence of peptide insertion into the bilayer41 and effective removal of the peptides from the bulk aqueous environment even at concentrations below their MIC values. In contrast, the same series of peptides displays variable behavior in the presence of zwitterionic vesicles, which generally correlates with their R_t values. Thus, minor blue shifts (<5 nm) were seen for the less hydrophobic peptides (6k-f17, 6K-F17-1L₁₁ and most of the Ala-to-Leu double-substituted isomers) indicating a minimal level of attachment of the peptides to the bilayer surface. Under the same conditions, 6K-F17-2L_8,16 and 6k-f17-2l_16,17, along with both triple-substituted isomers and 6K-F17-4L_8,11,13,16, showed an indication of partial/shallow Trp insertion into the zwitterionic bilayer moiety with larger blue shifts of 5–15 nm. In correlation with R_t values, double-substituted peptides with Leu residues close to the C-terminus showed higher insertion potential relative to their other isomers. Though some minor differences in fluorescence spectra were noted for PE and PC phospholipids (the former caused smaller blue shift values), the presence of cholesterol in the bilayer core did cause a discernible decrease in blue shifts in λ_max values, by a few nm in general (see Fig. 3). Even though the total lipid compositions of LUVs employed in fluorescence studies do not completely resemble those of human RBCs or bacteria, nor maintain the innate asymmetry of outer versus inner RBC bilayer leaflets, our fluorescence-based LUV insertion data were nevertheless consistent with the corresponding observation of hemolytic effects as Leu-content increased, legitimizing comparison of the biophysical and biological approaches.

Hydrophobicity Thresholds for Membrane Insertion of CAPs

The gradual increase in CSH values along the series of nonamphipathic CAPs from mono- to tetra-substituted Ala-to-Leu peptides causes a gradual increase of their RP-HPLC retention times and roughly speaking, in their MIC values (i.e., a decrease in antimicrobial activity). However, the onset of hemolytic effects on RBCs, paralleled by peptide insertion into mammalian model membranes, appears only once CSH reaches prescribed levels. In previous studies from our laboratory, a “hydrophobicity threshold” (termed here the “first threshold”) was defined as the minimal average segmental hydrophobicity required for peptide insertion from water into micelles or anionic bilayer membranes42, and therefore for antimicrobial activity.16 This threshold was shown to correspond approximately to just above that of a poly-Ala segment. A similar pattern was noted earlier for Lys-tagged 25-residue peptides with 19-residue core hydrophobic segments.17 However, when this paradigm was applied to zwitterionic mammalian-like membranes, we found for similar peptides containing highly charged Lys tags attached to a core hydrophobic sequence,20 that peptides ultimately cannot leave the bulk water for attachment/insertion into erythrocyte-like bilayers until their CSH begins to approach sufficiently high levels, as in the case where the sequence contains at least two Ala-to-Leu substitutions (see Fig. 2). One can thus define a “second hydrophobicity threshold” for peptide insertion into zwitterionic (mammalian) membranes—and as observed in the overall studies reported here—generally correspondingly for hemolytic activity. In the present work, we are able to localize this “second threshold” to the gap between CSH values of 1.9–2.3 among the sets of 6K-F17 analogues (Table I).

The highest therapeutic index (MHC/MIC = 162) of peptide 6K-F17, can also be explained by this “two threshold” paradigm. Since hemolysis of mammalian cells results in principle from insertion of the peptide into the hydrophobic core of the erythrocyte membrane, it seems reasonable to assume that relatively low core hydrophobicity would prohibit direct peptide attachment to the membrane via hydrophobic interactions. On the other hand, because the mechanism of action against bacterial cells does allow the attachment of multiple Lys side chains to the anionic phospholipid head groups in the interface region, insertion into the hydrophobic core of the membrane by peptides with average CSH above the “first threshold” will be facilitated by the initial incorporation of the peptide into the membrane surface region.

Role of Positional Hydrophobicity

The set of peptides with systematic variation of Leu residue location in the sequence allows us to assess a possible key role of positional hydrophobicity on CAP bioactivity and hemolytic strength, in a series where peptide composition, charge, and length are otherwise identical. In our experiments, lipid:peptide (L:P) ratios were maintained high enough (250:1) at low peptide concentration (4 μM, well below the majority of MIC values and below the hemolytic threshold for the conditions tested) to mimic the initial steps of peptide-mediated membrane disruption. Among five isomers of double-Ala-to-Leu substituted peptides (6k-f17-2l_16,17 is studied as the D-enantiomer), CSH values are correspondingly equal in each case, but R_t values range from 31.3 to 34.2 min, and hemolytic activity ranges from 2.2 to 11.4% at 81 mM and 2.4 to 23.7% at 650 mM peptide. Yet the antimicrobial activities of these peptides are broadly similar: allowing for a doubling of the MIC for the L-version of the Leu_16,17 isomer, four of the five isomers have MICs = 8– 16 μM, and only the Leu_8,16 isomer is higher at MIC = 64 μM. The 3-Leu and 4-Leu isomers are discernibly higher with MICs of 32–64 μM. The trend toward decreasing antimicrobial activity with increasing Leu content may reflect the circumstances that (i) the “active” interface needed for insertion becomes involved instead in peptide-peptide interactions, such that at least a portion of the peptide population becomes unable to penetrate the microbial cell membrane despite electrostatic attachment; and/or (ii) the increased segmental hydrophobicity imparts an increased potential to peptide self-association/oligomerization in the aqueous phase, thus limiting the concentration of peptide actually impacting on the bacterial cell surface. This latter situation would have little effect on the access of the peptides to zwitterionic mammalian membranes, thus uncoupling, at least in part, any direct correlation between hemolysis levels and antimicrobial activity.

Designed CAPs Form Dimers in Hydrophobic Environments

Essentially all Lys-tagged 11-residue core peptides we have designed have a strong tendency toward dimerization under membrane-mimetic conditions, as confirmed by SDS-PAGE gels (Table I) and FRET spectroscopy.20 In the present work, all 2-, 3-, and Leu isomeric peptides migrate on SDS-PAGE as dimers. Such dimerization must relate to the presence of highly potent helix-helix interaction motifs in their sequence. Dimerization in this peptide series may arise largely from the fact that their nonamphipathic cores contain at least one AxxxA in their sequences. Whether or not other oligomerization promoting motifs may be present, two “small” residues separated by three residues, such as AxxxA43-45 represent likely candidates for driving dimerization. In the context of antimicrobial and hemolytic activities, such strong dimer-forming ability in hydrophobic environments might serve as a key determinant, i.e., the promotion of even higher oligomerization states within the membrane surface region, which in turn may be expected to create substantially larger disturbances in the bilayer once insertion occurs. It may be further noted that in CAPs that act through barrel-stave and toroidal pore mechanisms, peptides are believed to oligomerize before or during bilayer insertion,10, 46 so that their ability to oligomerize enhances the effectiveness of membrane disruption. Although we did not undertake to measure the affinity constants of this series of dimers, which may span a considerable range given the variations in core hydrophobicity, the SDS-resistant behavior of these dimers suggests high affinity for all of them. It is therefore unlikely that this affinity correlates with antimicrobial activity, although the more hydrophobic peptides may associate to higher oligomers as a function of Leu residue placement, particularly in aqueous media.

Novel Conformationally Constrained CAPs Derived from 6K-F17

Neither of our attempts to significantly improve the antimicrobial activity by either intermolecular cyclization or hairpin dimerization succeeded. Cyclic (K₇-D₁₇)7K-F18-1D₁₇ showed values equal to those of its linear precursor 7K-F18-1D₁₇, and MIC values slightly higher than 6k-f17 with P. aeruginosa.

Although induced dimers (through formation of intermolecular disulfide bridges) of many antimicrobial peptides have shown increased activity,47 the SPGS-34 hairpin—that resembles a parallel dimer of 6K-F17 cross-linked via the C-terminals with a short loop—has the least antimicrobial activity of any “monomeric” peptide in the present study. We suspect that the monomeric state of the helix-loop-helix SPGS-34 hairpin itself, and the lack of its hetero-oligomerization in mixtures with other single “TM” CAPs on SDS-PAGE (not shown) is explicable by the presence of just a single potential SDS-resistant oligomerization face upon the nonamphipathic core helix.

Insertion of CAPs into Anionic and Zwitterionic Bilayer Membranes

As proposed earlier in a “grip and dip” model for antimicrobial action,16 the Lys residues anchor the peptides into the anionic bacterial membrane surface, and the uninterrupted core of hydrophobic residues (with amidated C-terminus) can then substantially penetrate the membrane. Given the strong dimerization properties of this category of CAPs, one can envisage an antiparallel dimer (with Lys residues maximally separated) as a powerful perturbant, disrupting the packing of lipid chains as it diffuses laterally—largely parallel to the bilayer surface—through the bacterial membrane in “submarine” fashion.

In contrast, zwitterionic membranes offer no direct route to penetration by these CAPs until CSH exceeds the “second threshold”; this is manifested by the generally increasing blue shifts of Trp fluorescence spectra as a function of increasing Leu content from one to four residues (see Fig. 2), which in turn are also broadly correlated with HPLC retention times and hemolytic activity (Fig. 3a). The data acquired from the double-Leu-substituted subset of our peptide library provide some further mechanistic insights into this process. In the latter series, evidence was also obtained of Trp insertion into the hydrophobic phase of zwitterionic lipid bilayers, but principally when one or both Leu residues are located close to the uncharged C-terminus of the sequence (Table I, Fig. 3b); hemolytic activity displayed no correlation with this process. These latter results suggest initially that in the absence of electrostatic forces (i.e., the membranes are zwitterionic, the C-terminal positions are most remote from Lys sites), reorientation of the peptide perpendicularly to the bilayer surface, with accompanying “corkscrew-like” insertion, becomes possible, while the Lys-tag of six consecutive positively charged residues remains above the surface in the aqueous phase. Insertion of peptides in this manner would likely orient them similarly to the transmembrane segments of native membrane proteins, and as such would be expected to cause less disruption to the bilayers than the parallel insertion experienced by the bacterial membrane. This latter situation probably allows many hydrophobic isomers of the present CAP series to induce minimal hemolysis even at relatively high concentrations. Peptides that have amphipathic sequences, including most natural CAPs like magainin, may bind zwitterionic membrane phosphate groups weakly through distributed cationic residues, producing a situation geometrically more similar to their parallel orientation in bacterial membranes, and hence increasing their hemolytic propensity. Interestingly, no specific correlations were observed in the present work between Leu-residue positions and hemolytic activity.