Kinetics And Mechanism Of Membrane Interactions With Antimicrobial .

Table 1: One-Letter Amino Acid Sequences of the Native and Modified Peptides. The gray boxes highlightmutations from the original sequence to the variant.PeptideCecropin ASequenceKWKLF KKIEK VGQNI RDGII KAGPA VAVVG QATQI AK-amideCE2KWKLL KKLEK AGAAL KEGLL KAGPA LALLG AAAAL AK-amideMagainin 2GIGKF LHSAK KFGKA FVGEI MNSMG2GLGKL LHAAK KLGKA WLGEL LAA5

The means by which the antimicrobial peptides perturb a membrane can vary, but can beclassified as either all-or-none or graded. More popular mechanisms include the barrel-stavemodel, the toroidal pore model, the carpet model, the sinking raft model, and other modelsdescribing less structured pores (13). In the all-or-none model, individual vesicles eithercompletely release their contents or they release nothing at all. In the graded model, all vesiclesrelease the same portion of their contents. A simplified diagram representing the two extremesof the all-or-none and graded mechanisms of release is shown in Figure 2.The all-or-none release mechanism is normally attributed to the barrel-stave model, thecarpet model, and in some instances, the toroidal pore. In this mechanism, the Gibbs free energyof insertion of the peptide from the membrane interface to the hydrophobic core of the bilayer ishypothesized to be larger than 20 kcal/mol (10). This mechanism consists of four differentstates: 1) unstructured, unbound peptide free in solutions; 2) peptide bound as an α-helix to fullvesicles; 3) peptide inserted into the lipid bilayer in a pore-state; and 4) peptide bound to emptyvesicles (14).The following are several models which display the necessary criteria for the all-or-nonemechanism. In the barrel-stave model, several peptides insert into the membrane perpendicularto the bilayer forming a pore—the peptides, in contact with each other, form the ―staves‖ and theoverall structure provides the ―barrel‖ shape (15). The hydrophobic portions of the peptidespoint toward the acyl chains of the lipids and the hydrophilic regions line the solution side of thepore, allowing cytoplasmic contents to easily cross to the extracellular region and also causingpotential loss of electrochemical gradients. The carpet model involves peptides orientingthemselves parallel to the lipid bilayer, and upon reaching a critical threshold, they causepermeabilization (6,9). They do this by coating the membrane like a carpet, which requires6

Figure 2: The all-or-none and graded release mechanisms. All-or-none release of 50% of the vesicles causes half ofthe vesicles to release all of their contents and the other half to remain fully intact. Graded release of 50% of thevesicles contents causes all vesicles to lose half of their contents. The details of the different mechanisms wereomitted from this schematic drawing, but are discussed in detail by Yandek et al. (14).7

relatively high concentrations of peptides compared to other models. Since the concentration ofpeptides in this model is high, the action of the peptides is detergent-like and can causemicellization. The toroidal pore model (16,17) involves several peptides insertingperpendicularly into the membrane, however, unlike the barrel-stave model, they do notnecessarily have to be in contact with each other. Portions of the membrane can fold in and fillthe space between the peptides, creating peptide and lipid lined pores.Graded release is attributed to the toroidal pore and sinking raft mechanisms ofantibacterial peptide action. However, for toroidal pores to invoke graded release, the pore statemust be short-lived. In the sinking raft model, a stochastic structure of peptides complex andform the pore (18-20). The structure formed in this model can have the peptides insert into themembrane either perpendicular, parallel, or both. The mechanism of release is determined by thetype of peptide as well as the free energy of insertion. It is our working hypothesis that if thefree energy of insertion is greater than 20 kcal/mol, then the release will be all-or-none, and if itis below this threshold, it will be graded.Cecropin A is a 37 amino acid peptide derived from the giant silk moth, Hyalophoracecropia (21,22). Upon binding, cecropin A adopts a secondary structure with two alpha-helicalregions, one from Phe5 to Lys21 and the other from Pro24 to Gln37, linked by a Gly-Pro break(22,23). These two regions can be seen in a helical wheel projection in Figure 3. In thisprojection it is easy to see the amphipathicity that the peptide can adopt as a helix. Also visiblein comparing the native and minimalist WAL versions of the peptide is the conservation of theresidue properties at each location. The only noted change in individual amino acid properties isseen at position 11: the mutation Ala11 Val11, going from an uncharged polar residue to anonpolar residue.8

ACBDFigure 3: Helical wheel projections of cecropin A and CE2 peptides at neutral pH. A and B are projections ofcecropin A. C and D are projections of CE2. A and C represent the first helical segments of the peptide, composedof residues 5 to 21. B and D represent the second helical segments, residues 24 to 37. Blue symbols are basic,positively charged residues. Red symbols are acidic, negatively charged residues. White symbols are polar butuncharged residues. Gray symbols are nonpolar residues.9

Magainin 2 is a 23 residue peptide originally extracted from the skin of the Africanclawed tree frog, Xenopus laevis. When discovered, it was aptly named from the Hebrew word,―magain‖, meaning ―shield‖ (24). Magainin 2 adopts a single helix conformation upon binding.The helical wheel projections of Magainin 2 and MG2 can be seen in Figure 4. Experimentsshow that overall, magainin 2 invokes all-or-none release with data that suggests the mechanismis either toroidal pore formation or the sinking raft model (25).Large unilamellar vesicles (LUVs) were used to model bacterial cell membranes in theseexperiments. For some experiments, incorporation of a fluorophore was necessary. Theconcentrations of diacylphosphatidylglycerol (POPG) and diacylphosphatidylcholine (POPC)lipids were varied based on previous work with the respective peptides to test the similarity ofeffectiveness of the WAL mutants and the native peptides. These model the polar head groupsof lipids found in microorganisms – phosphatidylglycerol being anionic and phosphatidylcholinebeing neutral. Varying the compositions of the lipid vesicles allows production of LUVs whichhave cell membrane characteristics, simulating native peptide binding conditions.The kinetics of peptide binding will be calculated for the WAL mutants and compared tothe native peptides using fluorescence energy transfer experiments. Each peptide being studiedhas a Trp residue which is used to transfer energy upon excitation to a fluorophore, 7methoxycoumarin (7MC), which is incorporated into the lipid vesicles by attachment to thephosphate headgroup of the lipid diacylphosphatidylethanolamine (POPE). The 7MC is a probethat indicates the proximity of the peptide to the vesicle. The neutral residue mutations that weare investigating should impart little or no change in the binding constants previously determinedfor the native peptides.10

ABFigure 4: Helical wheel projections of magainin 2 (A) and MG2 (B) at neutral pH. The helical portion of thesepeptides includes the entire sequence. Blue symbols are basic, positively charged residues. Red symbols are acidic,negatively charged residues. White symbols are polar but uncharged residues. Gray symbols are nonpolar residues.11

MG2 has a calculated Gibbs free energy of insertion of 25 kcal/mol, based on anestimated helicity of 70%. We expected that this peptide should release vesicle contents by anall-or-none mechanism. Since CE2 has calculated insertion energy of 36 kcal/mol with anassumed helicity of 70%, all-or-none release was also expected with this peptide. One caveat toconsider is the helicity of the peptide in free solution and the helicity of the peptide in the boundstate. In the original hypothesis proposal, calculations were performed using an assumed helicityof 70% in the bound state and minimal helicity free in solution. However, the experimentallydetermined helicities are different than this and affect the values obtained for the free energy ofinsertion and the thermodynamic values calculated previously.Incorporation of a fluorescent dye within the LUVs facilitates studies of the effluxkinetics of the vesicle in the presence of peptide. Initially at self-quenching concentrations(50mM), carboxyfluoroscein dye cannot fluoresce while still encapsulated in the LUV. Once themembrane is perturbed, dye released into external buffer is diluted and can fluoresce. Thisallows kinetics and total release levels to be calculated.12

MATERIALS AND phosphocholine (POPC), -glycerol)(sodium salt) (POPG), and ne (POPE) were all purchased in chloroform solution from Avanti PolarLipids, Inc. (Alabaster, AL) (see Figure 5). 7-methoxycoumarin-3-carboxylic acid, succinimidylester (7MC), 5-(and -6)-carboxyfluorescein (CF), 8-aminonaphthalene-1,3,6-trisulfonic acid,(ANTS), and p-xylene-bis-pyridinium bromide (DPX) were purchased from MolecularProbes/invitrogen (Eugene, OR). 3-morpholinopropane-1-sulfonic acid (MOPS), ULTROL grade, was purchased from EMD Chemicals (Gibbstown, NJ). Potassium chloride (KCl),potassium hydroxide (KOH), ethylenediaminetetraacetic acid (EDTA), and sodium azide (NaN3)were all purchased from BDH (West Chester, PA). Ethanol (EtOH), 200 proof, was purchasedfrom AAPER Alcohol and Chemical (Shelbyville, KY). Dichloromethane (DCM), methanol(MeOH), and all other organic solvents were HPLC or ACS grade and purchased from Burdick& Jackson (Muskegon, MI). Ammonium molybdate, ACS grade, was purchased from ThermoFischer Scientific (Fairlawn, NJ). Dimethylformamide (DMF), ascorbic acid (USP grade) andperchloric acid (70%, ACS grade), were purchased from Mallinckrodt Chemicals (Phillipsburg,NJ). Water, filtered to 18.2 MΩ cm purity using a Milli-Q Direct Water Purification System byMillipore (Billerica, MA), was obtained on site and stored in a 20 L Nalgene carboy.13

Figure 5: Structure of POPC, POPG, and POPE. The only difference between these phospholipids appears in theheadgroup. Neutral, zwitterionic POPC is shown (top). Negatively charged POPG is shown as a sodium salt(middle). Neutral POPE, a reactant in the 7MC-POPE synthesis is also shown (bottom).14

PeptidesCE2 (KWKLLKKLEKAGAALKEGLLKAGPALALLGAAAALAK-amide) lot:B05973, 82% purity, was purchased from Bachem, Inc. (Torrance, CA). CE2, lot: pr1770, 98%purity, and MG2 (GLGKLLHAAKKLGKAWLGELLAA), lot: pr1202, 94% purity, werepurchased from New England Peptide, LLC (Gardner, MA). Lyophilized peptide was stored at-20 C. Stock peptide solutions were prepared by mixing lyophilized peptide in 1:1 (v/v)water/ethanol. Peptide concentration was determined using a Cary 1E UV-Visspectrophotometer, in matched 1.000 cm 6Q quartz cells, scanning from 500 to 250 nm using theabsorbance maximum of tryptophan at 280 nm with an extinction coefficient of 5,579 M-1 cm-1.Solutions were then aliquoted into small Eppendorf tubes and flash frozen using either liquidN2 or an acetone/dry ice bath. Peptide solutions were stored at -80 C and kept on ice duringexperiments.Buffer PreparationMOPS buffer was prepared in water with 20mM MOPS, 100 mM KCl, 0.1 mM EDTAand 0.02% NaN3, then brought to pH 7.50 using 1 M KOH and brought to final volume. CFbuffer was prepared by grinding CF powder with mortar and pestle, then mixing 50 mM CF, 20mM MOPS, 0.1 mM EDTA, and 0.02% NaN3, in water and brought to pH 7.50 with 1 M KOH,then brought to final volume. ANTS/DPX hydration buffer was prepared in water with 5mMANTS, 10mM DPX, 20mM MOPS, 80mM KCL, 1.0 mM EDTA, and 0.02% NaN3.ANTS/DPX titration buffer was prepared in water with 45 mM DPX, 20 mM MOPS, 30 mMKCl, 1.0 mM EDTA, and 0.02% NaN3.15

Large Unilamellar Vesicle PreparationAll glassware and syringes were cleaned by rinsing or vortexing with 1:1 DCM/MeOH.Large unilamellar vesicles (LUVs) were prepared by combining appropriate volumes of stocklipid solutions in a round bottom flask. The solvent was rapidly removed using a rotaryevaporator at 65 C (Buchi R-3000, Flawil, Switzerland). The lipid film was then dried for atleast 4 hours in a dessicator under vacuum with 100g Drierite in the base. The lipid film wasthen hydrated with the appropriate buffer and vortexed 1 minute to remove all lipids from theglass walls and suspend them. The lipid solution was then transferred to a test tube andsubjected to five freeze-thaw cycles using liquid N2 (or a dry ice/acetone bath), then roomtemperature water, then 40 C water. This was done to promote vesicle fusion, creating largemultilamellar vesicles, and assisting dye encapsulation (for CF and ANTS/DPX vesicles). Ahigh-pressure extruder (10 mL water-jacketed Lipex Extruder, Lipex Biomembranes, Vancouver,CAN) was assembled with two stacked Nucleopore 0.1 µm polycarbonate filters (Whatman,Florham, NJ), and rinsed once with 10 mL MOPS buffer, then ten times with 1.5 mL MOPSbuffer each. The lipid solution was then extruded ten times at 450 psi, to create homogeneous,100 nm unilamellar vesicles. CF and ANTS/DPX vesicles were subsequently passed through aSephadex G-25 column to remove unencapsulated dye.Lipid Concentration DeterminationA modified Bartlett assay was used to determine the phosphate concentration of the lipidsolutions and hence, the lipid concentration (28). Briefly, 1.053 mM phosphate standard waspipetted in 60 µL increments from 0 to 300 µL along with vesicle solution samples in glass testtubes. All were pipetted in triplicate with the balance as water to keep volumes equal. 700 µL of16 pa

the secondary structure of the antimicrobial peptides, and gauge the effects caused by conservatively modifying targeted individual residues. There are several structural classes of antimicrobial peptides. The major structure types include α-helical, β-stranded, extended coil, and loops (11). Examples of these can be seen in Figure 1.