Slow Insertion Kinetics during Interaction of a ... - ACS Publications

Zubaidah Ningsih†, Mohammed Akhter Hossain†‡, John D. Wade†‡, Andrew H. A. Clayton*§, and Michelle L. Gee*†. School of Chemistry, Univers...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/Langmuir

Slow Insertion Kinetics during Interaction of a Model Antimicrobial Peptide with Unilamellar Phospholipid Vesicles Zubaidah Ningsih,† Mohammed Akhter Hossain,†,‡ John D. Wade,†,‡ Andrew H. A. Clayton,*,§ and Michelle L. Gee*,† †

School of Chemistry, University of Melbourne, Parkville, Victoria 3010, Australia Howard Florey Institute, University of Melbourne, Parkville, Victoria 3010, Australia § Centre for Micro-Photonics, Swinburne University of Technology, Hawthorn, Victoria 3122, Australia ‡

ABSTRACT: The mechanism of interaction between a model antimicrobial peptide and phospholipid unilamellar vesicle membranes was studied using fluorescence spectroscopy, fluorescence lifetime measurements, and light scattering. The peptide, a mellitin mutant, was labeled at position K14 with the polarity-sensitive probe AlexaFluor 430. The kinetics of the interaction of this derivative with various concentrations of 1,2-dipalmitoyl-sn-glycero3-phosphatidylcholine (DPPC) vesicles was examined. Our work unveiled two novel aspects of peptide−lipid interactions. First, the AB plot or phasor analysis of the fluorescence lifetime studies revealed at least three different peptide states, the population of which depended on the lipid to peptide (L:P) concentration ratio. Second, complex fluorescence kinetics were observed over extended time-scales from 30 s to 2 h. The extended kinetics was only observed at particular lipid concentrations (L:P ratios 20:1 and 10:1) and not at others (30, 40, 50 and 100:1 L:P ratio). Analysis of the complex kinetics revealed several intermediates. We assign these to distinct states of the peptide formed during helix insertion into the vesicle membrane that are intermediate to lytic pore formation.



INTRODUCTION The future development of new active therapeutic agents is of increasing importance with the emergence of bacterial strains resistant to existing conventional antibiotics.1 Antimicrobial peptides are promising therapeutic alternatives because of their ability to selectively kill bacterial cells but not cells of eukaryotic origin.2 Peptides that are found in many plants,1 insects,1,2 and human cells 3 are thought to act by disrupting the lipid bilayer 4−6 causing permanent physical damage 7 that leads to cell membrane lysis.8 The precise mechanism by which the lipid−peptide interaction ensues continues to be an area of intense research activity and debate. Biophysical studies aimed at understanding peptide secondary structure within membranes have revealed that many peptides are random coil in solution but form α-helices when physically interacting with phospholipid bilayers. Peptide helices can also adopt different orientations within lipid bilayers: parallel, perpendicular, or oblique with respect to the membrane plane. Previous studies have shown that peptide conformation and orientation9−11 depend on a number of factors. These include membrane composition,4,11−24 peptide physical and chemical properties,4,17,19,20,22,23 lipid to peptide (L:P) ratio,18,25 peptide−lipid intermolecular interaction (electrostatic or hydrophobic),11 temperature,26 and membrane cholesterol content.27,28 On the basis of these investigations, different structural models for the peptide−membrane interaction have been proposed. These models are illustrated schematically in Figure 1. They are based on either membrane lysis via the formation of membrane pores (barrel stave and torroidal-pore models), or detergent-like mechanisms involving structural © 2011 American Chemical Society

Figure 1. Models of peptide−lipid interactions. Peptides are represented by the cylinders. (A) Barrel stave pore model. Peptides are shown bound parallel to the surface of bilayers and in transmembrane oligomeric bundles. In the barrel stave model, the bilayer remains intact. (B) Torodial pore model. Same as A except the pore is formed from head-groups of bilayer and peptides in transmembrane orientation. (C) Carpet model. Peptides bind to surface but do not induce pore formation. Instead the peptides disintegrate the bilayer analogously to detergent solubilization.

breakdown of the membrane (carpet model). Combinations of these models have also been proposed.19 Special Issue: Bioinspired Assemblies and Interfaces Received: September 27, 2011 Revised: November 22, 2011 Published: December 7, 2011 2217

dx.doi.org/10.1021/la203770j | Langmuir 2012, 28, 2217−2224

Langmuir

Article

were surprised to observe complex kinetics of the melittin derivative helix over time scales of minutes to hours, which were strongly dependent on lipid concentration. To our knowledge, this is the first report of long-time scale kinetics in melittin-lipid interactions and the first to reveal critical behavior in long-term kinetics. These observations suggest that the melittin derivative− phospholipid interaction kinetics cannot be described in terms of a simple co-operative process, as previously thought.

For the interaction of a peptide with a self-assembled phospholipid structure, such a vesicle often results in phase transition.29−31 Depending on the peptide, lipid and experimental conditions (i.e., temperature and concentration), vesicle aggregation, vesicle fusion, vesicle-to-bilayer disk transition, and micellization have been identified. These processes are often related to the structural model for the peptide−lipid interaction. For example, the detergent-like carpet model is related to the appearance of mixed phospholipid−peptide micelles and bilayer disks, whereas the pore states are believed to be involved in vesicle fusion. Kinetic information can provide richer mechanistic insight into peptide−membrane interactions. It enables deconvolution of different steps in the process, leading to an understanding of the number of different types of species in the system and their associated rates of formation. The aim of the present work is to monitor the long-time scale kinetics of peptide−membrane interactions using a model peptide−lipid system. Stopped-flow optical techniques have revealed that peptide binding to membranes generally occurs quite rapidly, on the time scales of milliseconds to seconds.32,33 Subsequent steps such as membrane insertion and helix−helix association have been identified in some cases.34 Binding and insertion are sometimes thought to be co-operative processes, involving peptide self-association.35 Recent studies have revealed kinetics over longer time scales, which probably reflect changes to peptide conformation and/or membrane structure subsequent to initial binding. One study, using linear dichroism, reported changes in peptide orientation after initial interaction over time scales of minutes to hours.36,37 Another study employed sumfrequency generation spectroscopy to demonstrate bilayer structural changes on the time scale of minutes. Taken together, these observations suggest that, in order to derive mechanistic information, it is desirable to obtain kinetic and structural data from both peptide and membrane during the peptide− membrane interaction over an extended time range. Here we monitor the long time scale kinetics of the interaction of a melittin derivative with 1,2-dipalmitoyl-sn-glycero3-phosphatidylcholine (DPPC) small unilamellar vesicles (SUVs). Melittin is a naturally occurring, prototypical α-helical peptide1 that has become a model system for studying antimicrobial peptides and protein−membrane interactions.30,38−42 Recently, we reported the synthesis of a new fluorescent derivative of melittin. It is based on parent melittin with a proline-14 to lysine substitution. The lysine 14, which is near the center of the melittin helix, contains a covalently attached AlexaFluor 430 polarity-sensitive fluorescent probe.43 We have shown that this derivative is α-helical in membranes and even more effective than its parent at membrane lysis.43We employed surface-selective time-resolved spectroscopic techniques to elucidate the molecular dynamics of the melittin derivative interacting with supported-lipid bilayers on the nanosecond time-scale. Our study demonstrated a polar, yet motionally restricted environment of the AlexaFluor 430 probe near the center of the melittin helix, consistent with its location within a torroidal pore.43 We monitor the kinetics of the melittin derivative as it interacts with DPPC SUVs using spectroscopic techniques, over a time scale of seconds to hours. Changes in the fluorescence intensity, modal wavelength position and excited-state lifetime were used to monitor changes in the environment about the AlexaFluor 430 probe at position 14 of the melittin helix. Dynamic light scattering (DLS) as well as right-angle light scattering was employed to monitor changes in vesicle structure. We



MATERIALS AND METHODS

Materials. The lipid used in the unilamellar vesicles preparation was DPPC from Avanti Polar Lipids (Alabaster, AL, USA). This is a lipid containing two unsaturated 16-carbon chains and has a phase transition temperature of 41 °C. Other materials used were chloroform (spectroscopy grade) and sodium chloride (99.5% purity) purchased from Merck (Darmstadt, Germany), 4-(2-hydroxyethyl) piperazine-1ethanesulfonic acid (HEPES, 99% purity) purchased from Acros Organics (Geel, Belgium), and nitrogen gas (high purity) purchased from BOC Gases (Preston, VIC). Chemicals used in the phosphorus content determination experiment were ammonium phosphate (H9N2O4P, for atomic absorption spectroscopy) purchased from Sigma-Aldrich (St. Louis, MO, USA), ammonium molybdate (VI) tetrahydrate (analytical reagent), L-ascorbic acid (99% purity), and sulphuric acid (analytical reagent) purchased from Ajax chemicals (Sydney, NSW), and hydrogen peroxide (analytical reagent) purchased from Merck. All materials were used without further purification. For all experiments, ultra pure Milli-Q water (Millipore Corp, Bedford, MA, USA) with a resistivity of 18.2 Ω was used. Design of the Melittin Derivative, Alexa 430-Melittin K 14. Peptide synthesis and purification was described in a previous publication.43 The melittin was labeled with AlexaFluor 430 at the 14th position since it is situated in the intermediate region of the amphipathic helix. AlexaFluor 430 is a polarity-sensitive fluorescent probe. Its fluorescence behavior is linked to change in the microenvironment, particularly micropolarity, of the fluorophore (i.e., polarity change). When located at the 14th position along the melittin helix, it can sample both hydrophobic and hydrophilic regions in the bilayer during peptide−membrane interaction. However, proline-14 is an amino acid that does not have a side chain amine. Therefore, it was replaced by lysine, which provides a side chain amine to enable amine coupling of the AlexaFluor 430 through a succinimidyl ester unit. The sequence of melittin from GIGAVLKVLTTGLPALISWIKRKRQQ-CONH2 was changed to GIGAVLKVLTTGLKALISWIKRKRQQ-CONH2, where AlexaFluor 430 then coupled with lysine to form a fluorescently labeled melittin derivative. DPPC SUV Preparation and Characterization. SUVs were prepared by sonication as described previously.43 Briefly, a thin lipid film was made by dissolving the lipids in chloroform and then drying under a stream of nitrogen, followed by vacuum desiccation for a minimum of 2 h. The resulting lipid film was suspended in HEPES buffer solution with vortexing at a few degrees above the lipid phase transition temperature. This step produced multilamellar vesicles in 10 mM HEPES buffer with 150 mM NaCl at pH 7.4 with 1 mg/mL DPPC concentration. To facilitate water intrusion between the multilamellar lipids, the suspension was rotary-evaporated at 100 rpm at 50 °C. An opaque solution that indicates the existence of multilamellar vesicles was yielded. DPPC multilamellar vesicles were then homogenized through five cycles of freezing using liquid nitrogen and thawing slowly at room temperature followed by sonication in an ice bath for 1 h using a Branson (Danbury, USA) 450 digital tip sonifier operating in 50% duty cycle (30 min total sonication time) at 30% amplitude. The resulting SUV suspension was then centrifuged in a Hermle (Gosheim, Germany) Z233 M-2 microcentrifuge for 10 min at 15 000 rpm to remove titanium particles released from the sonifier tip as well as any residual multilamellar vesicles then diluted to 0.1 mg/mL and stored at 4 °C under nitrogen for use within 4 days of preparation. To characterize the amount of vesicles transformed during the process, phosphorus content determination was conducted, which is equivalent to the lipid concentration since the lipid has one 2218

dx.doi.org/10.1021/la203770j | Langmuir 2012, 28, 2217−2224

Langmuir

Article

Steady-State and Kinetics Experiments. Steady-state measurements (fluorescence, light scattering, and fluorescence lifetime described) were carried out 2 h after preparing the peptide−lipid complexes. To monitor the kinetics of peptide−membrane interaction, measurements were performed after an initial mixing of melittin derivative with lipid (30 s). For the optical spectroscopy experiments, the kinetics function in the Varian Eclipse software was used to obtain the time-lapse data of fluorescence intensity. Scattering (excitation 540 nm/emission 542 nm) and fluorescence intensity (excitation 430 nm/emisson 510 nm and excitation 430 nm/emission 540 nm) were recorded in parallel. The time-lapse fluorescence lifetime imaging function in the Lambert Instruments software was used to obtain phase fluorescence lifetimes as a function of time. Fitting Kinetics Data to a Three-State Mechanism of Lipid− Membrane Interaction. To extract information on the mechanism of lipid−membrane interaction, time-lapse spectroscopic data was fitted to a simple model of peptide−membrane interaction involving the formation of three distinct peptide states. This model has been used previously to interpret slow peptide−membrane insertion kinetics36 in model membrane systems. We select this model since we identify three fluorescence states, but acknowledge that it may well be an oversimplification of the actual kinetics. Nonetheless, it allows an interpretation of our kinetic data. In this model, the net spectroscopic signal (e.g., fluorescence intensity) at time t, Stot(t), is given by

phosphorus atom. Ammonium molybdate and L-ascorbic acid were added to form the phosphorus−molybdate complex44 and absorbance read using a Cary UV−visible absorption spectrophotometer at 600− 900 nm. The hydrodynamic diameter of the SUVs was measured using DLS. The DLS instrument used was a Malvern Instrument (Lo-C Autosizer, Series 7032 multi-8 correlator) with an Ar ion laser light source of wavelength 488 nm. All measurements were performed at 25 °C at a detection angle of 90° and the data analyzed using the algorithm CONTIN. The mean hydrodynamic diameter obtained for freshly prepared SUVs was 37 nm with a narrow size distribution. This result is similar to another DLS study of DPPC vesicles21 and falls in the expected diameter range for SUVs in general (20−75 nm). SUV size and size distribution remained stable for at least 96 h. Preparation of Melittin Derivative-Membrane Complexes. Peptide-phospholipid complexes of defined lipid to peptide (L:P) ratio (range: 0:1,10:1, 20:1, 30:1, 40:1, 50:1 and 100:1 mols lipid/mol peptide) were prepared by adding melittin derivative to suspensions of SUVs. The concentration of melittin derivative was fixed at 2.5 μM. Optical Spectroscopy. Fluorescence spectra of AlexaFluor 430, the melittin derivative, and melittin derivative−membrane complexes were measured using a Cary Eclipse Fluorescence Spectrophotometer over the wavelength range of 450−800 nm at a scan speed of 120 nm/ min. The excitation wavelength was 430 nm. Right-angle light scattering was measured using the same instrumentation as for the fluorescence measurements, except the excitation wavelength was set to 540 nm and the emission wavelength was set to 542 nm. Fluorescence Lifetime Measurement Using Fluorescence Lifetime Imaging Microscopy (FLIM). Fluorescence lifetime experiments were carried out using a lifetime imaging attachment (Lifetime Imaging Fluorescence Attachment, Lambert Instruments, Leutingwolde, The Netherlands) mounted on an inverted microscope (TE2000U, Nikon Inc., Japan). A fluorescence cuvette containing a solution of melittin-AlexaFluor 430 derivative and complexes with DPPC was excited using epi-illumination with a sinusoidally modulated 470 nm light-emitting diode (LED) at 40 MHz. Fluorescence was observed with a 20× NA 0.5 air objective (Nikon Plan-Fluor, Nikon Inc., Japan) through a filter set that captures emission from the AlexaFluor 430 chromophore (Nikon FITC, DM 505, EM 515−555 nm). The phase and modulation lifetimes were determined from a series of images taken at 12 phase settings using software provided by the manufacturer and corrected for photobleaching using the pseudorandom phase approach as described by van Munster and Gadella.45 Rhodamine 6G (lifetime 4.1 ns) was used as a reference.46 The Alexa 430Fluor lifetime was measured at each L:P ratio for 1 h. The fluorescence lifetime measurements were represented in two different ways. First, to get an indication of trends, the average fluorescence lifetime was taken. This is the average of the phase and modulation lifetimes. The second approach utilized the AB plot47 (also referred to as a phasor48 or polar plot49) to graphically display the lifetime experiments. This plot represents an experiment by a point in two-dimensional (2D) space defined by x = m cos ϕ and y = m sin ϕ (where ϕ is the phase and m is the modulation of the fluorescence signal). A series of lifetime experiments plotted as a series of points and joined by lines is called a trajectory. This graphical approach has the advantage that the type of fluorescence decay (simple, complex, excited-state reaction/solvent relaxation) and the complexity of the system trajectory (binary or more complex50) can be deduced visually without further analysis. The phasor of single exponential-decaying fluorophores lies on a semicircle described by m = cos ϕ which intersects with points (0,0), (0.5, 0.5), and (1,0). Phasors from heterogeneous fluorescence decays lie in the region within the semicircle and follow the inequality m < cos ϕ. Excited-state reactions including solvent relaxation have phasors that lie outside the semicircle and obey the inequality m > cos ϕ.51 The linear combination of two phasors is described by a linear trajectory in AB space, whereas the mixing between three or more species is nonlinear.50,51

Stott (t ) =

Q1P1(t ) + Q 2P21(t ) + Q 3P3(t ) Ptot

(1)

where Q1, Q2, and Q3 are the spectroscopic signals corresponding to states 1, 2 and 3, respectively. These are weighted by the timedependent concentrations of peptide in these states: P1(t), P2(t), and P3(t), respectively. Ptot is the total peptide concentration. Assuming a simple mechanism in which peptide microenvironment changes sequentially to P1 with rate constant k1 after initial introduction to solution, from P1 to P2 with rate constant k2, and from P2 to P3 with rate constant, k3, we obtain36

P k (e−k1t − e−k2t ) P1(t ) = tot 1 k2 − k1

(2)

⎡ e−k1t P2(t ) = Ptotk1k2⎢ ⎢⎣ (k2 − k1)(k3 − k1) +

⎤ e−k3t ⎥ k3)(k2 − k3) ⎥⎦

(3)

⎤ k1k3e−k2t k1k2e−k3t ⎥ − (k1 − k2)(k3 − k2) (k1 − k3)(k2 − k3) ⎥⎦

(4)

e−k2t + (k1 − k2)(k3 − k2) (k1 −

⎡ k2k3e−k1t P3(t ) = Ptot⎢1 − ⎢⎣ (k2 − k1)(k3 − k1) −

Fluorescence intensity data as a function of time (20:1 and 10:1 L:P samples) were fit to eqs 1−4. Nonlinear least-squares fitting procedures were carried out using Microsoft Excel with the solver function. Use of eqs 2−4 may be considered an overparameterization of our experimental data because of the number of parameters involved, and we considered whether simpler models (involving less parameters) might equally well fit the data. Fits to models containing one baseline and one exponential term (three parameters) and one baseline and two exponential terms (five parameters) resulted in poor fits to the data (sum of least-squares = 0.03). However, inclusion of three exponential terms and a baseline (seven parameters) significantly improved the fit by better than an order of magnitude (sum of leastsquares = 0.002). We conclude that the data demands a kinetic scheme that must involve at least three kinetic steps. 2219

dx.doi.org/10.1021/la203770j | Langmuir 2012, 28, 2217−2224

Langmuir

Article

To obtain an estimate of the time-scale of dynamic changes in light scattering from the vesicle preparations, a simple kinetics model was used to fit the scattering data. For scattering signals that increased with time, the light scattering intensity as a function of time, I(t), was fitted by

I(t ) = 1 + B(1 − e−t/ τs) I(0)

(5)

For scattering signals that decreased with time, the light scattering intensity as a function of time, I(t), was fitted by

I(t ) = (1 + B)1 − e−t/ τs + B I(0)

(6)

In eqs 5 and 6, B is a constant baseline signal, and τs is a time constant. As t approaches infinity, eqs 5 and 6 approach B, the relative baseline signal. The value of B therefore provides a measure of the relative scattering signal at long times, compared to the initial measurement. It is important to note that B may be positive or negative, depending on whether scattering increased or decreased relative to the initial measurement.



RESULTS Interaction of Fluorescently Labeled Melittin Derivative with DPPC SUVs. The interaction of the melittin derivative with DPPC SUVs was examined by epifluorescence spectroscopy and fluorescence lifetime measurements, as a function of lipid/peptide (L:P) ratio. These data are shown in Figure 2, specifically, the wavelength of fluorescence emission maximum (Figure 2A), fluorescence intensity at this maximum (Figure 2B), and fluorescence lifetime (Figure 2C). At this juncture we do not make any assumptions about the species formed during peptide−membrane interaction so refer to these generally as complexes of peptide and lipid. In the absence of lipid, the AlexaFluor 430 probe on the melittin derivative displayed a fluorescence emission maximum at 534 nm and a fluorescence lifetime of 3.4 ns. Addition of lipid was accompanied by a progressive blue-shift in the emission spectrum (to shorter wavelengths), an increase in fluorescence intensity, and a concomitant increase in fluorescence lifetime. The spectroscopic signals from the melittin derivative appeared to saturate somewhere between 50:1 and 100:1 lipid/ peptide (L:P) ratio. At 100:1 L:P ratio, the AlexaFluor 430 fluorescence from the melittin derivative had an emission maximum at 515 nm, was 2.5-fold more fluorescent than in the absence of lipid, and had a mean lifetime of 4.5 ns. These photophysical changes are indicative of a change in the microenvironment about the AlexaFluor 430 probe during peptide−membrane interaction. Given that our previous time-resolved fluorescence studies revealed an increased AlexaFluor 430 lifetime in solvents of decreasing polarity,43 the data presented here suggests that the AlexaFluor 430 probe is experiencing an environment with a lower polarity than in water as a result of peptide−lipid interaction. In Figure 3A, the fluorescence lifetime measurements are displayed graphically in the form of an AB-plot (phasor or polar plot). The trajectory of the phasor on the AB plot disclosed an interesting pattern. Instead of following a linear path, as expected for a simple two-state system in which, for example, adsorbed peptide ↔ inserted peptide equilibrium exists, the phasor trajectory tended to one direction as lipid concentration was increased (L:P ratios 10:1, 20:1, and 30:1) and then changed direction (at L:P ratio 40:1) with further increases in lipid concentration. This transition is also apparent from inspection of the plots of the phasor components as a function of L:P ratio (Figure 3B). For example, the m sin ϕ component

Figure 2. Interaction of melittin derivative with DPPC vesicles by fluorescence spectoscopy. (A) Plot of the wavelength position of the fluorescence spectrum maximum as a function of lipid concentration (wavelength value ±2 nm). The lipid concentration is represented as the ratio of lipid concentration to that of the peptide concentration (peptide concentration was fixed at 2.5 μM). (B) Plot of the relative fluorescence intensity (±3%) as a function of lipid concentration (in units of L:P ratio as for A). (C) Plot of the apparent average lifetime (lifetime value ±0.05 ns) as a function of the lipid concentration (in units of L:P ratio as for A).

of the phasor decreased initially with increasing lipid concentration, reaching a minimum at an L:P of 40:1 and then increased upon further increases in lipid concentration. This is not a dilution effect since peptide concentration is kept constant, as detailed in the Materials and Methods section. By contrast, a simple two-state system would show either a linear decrease or increase to a limiting value. The observed nonlinear/nonmonotonic behavior indicates the existence of more than two different peptide states in the peptide−lipid complexes (50, 51). These are distinguished by different timedecays of the AlexaFluor 430 probe near the center of the melittin helix for these two states. This result suggests the presence of complex kinetics during which at least three different peptide states are formed: surface adsorbed peptide ↔ inserted peptide state/s ↔ lytic pore state. We elaborate on this, together with the other experimental data, in the Discussion section. It should be noted that, in the presence of lipid, the points on the polar plot of Figure 3A go above 0.5 on the y-axis. This is 2220

dx.doi.org/10.1021/la203770j | Langmuir 2012, 28, 2217−2224

Langmuir

Article

Figure 3. AB (or phasor) plot analysis of the interaction of melittin derivative with DPPC vesicles. (A) Lifetime trajectory of melittin derivative alone and with increasing concentration of lipid. Squares represent individual time-resolved fluorescence experiments of the melittin derivative at each lipid concentration. Starting from right to left, arrows indicate direction of trajectory in order of increasing L:P ratio from 0, 10, 20, 30, 40, 50 to 100:1. (B) Components of the phasor as a function of L:P ratio. Filled squares represent the m cos ϕ, and open squares represent the m sin ϕ. Figure 4. Long-time kinetics during the interaction of melittin derivative with DPPC vesicles. Experiments were conducted at different L:P ratios. These are represented in the figure as follows: red squares (10:1), yellow squares (20:1), green squares (30:1), blue circles (40:1), black squares (50:1), purple squares (100:1), and orange squares (lipid only). (A) Plot of the relative fluorescence intensity (excitation 430 nm, emission 510 nm) of melittin as a function of time (in minutes). The relative fluorescence is defined as the fluorescence measured at time t, divided by the fluorescence measured at time t = 30 s after mixing the peptide with the vesicles. The solid lines are fits to a sequential kinetic model (eq 1 with parameters in Table 1) as discussed in text. Note that the 0, 30, 40, 50, and 100:1 L:P ratio samples reached equilibrium within the 30 s mixing time and so are overlaid in this figure. Only the 30:1 data are shown for clarity. (B) Plot of the relative change in phase lifetime as a function of time (in minutes). The relative phase lifetime is defined as the lifetime calculated from the phase measured at time t, divided by the phase lifetime observed at time t = 30 s after mixing the peptide with the vesicles. C. Relative change in 90◦ light scattering as a function of time. The relative light scattering is defined as the light scattering measured at time t, divided by the light scattering measured at time t = 30 s after mixing the peptide with the vesicles. Symbols are the experimental data points, and solid lines are fits to eq 5 with parameters collected in Table 1.

outside the universal circle (described above in the Materials and Methods section) and is hence indicative of an excited state process, such as solvent relaxation. Kinetics of the Interaction of Fluorescently Labeled Melittin Derivative with DPPC SUVs. To obtain mechanistic information on the peptide−membrane interaction of our model peptide interacting with DPPC SUVs, the spectroscopic signals from the melittin derivative and the lipid dispersion were recorded as a function of time, over an extended time window. The time-courses of AlexaFluor 430 fluorescence intensity, AlexaFluor 430 phase lifetime, and light scattering intensity from the melittin derivative−phospholipid complexes (after initial 30 s mixing) are displayed in Figure 4. We turn first to the fluorescence intensity (Figure 4A) and lifetime (Figure 4B) data. Fluorescence intensity is represented in Figure 4A as a ratio of intensity at time, t, to intensity at t = 30 s, the time at which mixing is first complete. It is apparent that for the range of L:P ratios 100:1−30:1, equilibrium was reached before or during the 30 s mixing time. However, at L:P ratios of 20:1 and 10:1, kinetics were much slower, and fluorescence intensity did not start to plateau until around 60 min. Using eqs 1−4 to analyze the kinetics at these two L:P ratios revealed three sequential steps in the evolution of these systems toward their respective equilibrium positions. These steps had time constants in the subminute range (0.25−0.55 min) and minutes to tens of minutes range (5−30 mins), respectively. We found that different combinations of parameters could fit the data equally well, and so have reported the fitting parameter range (see Table 1). Nevertheless, the qualitative features of the fluorescence intensity with time for the 20:1 and 10:1

L:P samples are similar, inferring pseudounimolecular kinetics under these conditions. The fits to the fluorescence intensity data over time using eqs 1−4 yielded relative fluorescence intensities of the three different peptide states, Q1, Q2, and Q3, that were dependent on the L:P ratio and detection wavelength. At both 20:1 and 10:1 L:P ratios, analysis of the data at two emission wavelengths revealed that each successive kinetic step was accompanied by an increase in fluorescence intensity, i.e., Q3 > Q2 > Q1, and a blue shift to shorter emission wavelength. This is consistent with a 2221

dx.doi.org/10.1021/la203770j | Langmuir 2012, 28, 2217−2224

Langmuir

Article

revealed several peptide−lipid intermediates, which correspond to distinct lipid−peptide structures. There are several potential sources of critical behavior that have been observed for peptide−lipid interactions using different techniques. Huang and colleagues51−53 have demonstrated using oriented circular dichroism so that the orientation of a number of peptides in membranes displays critical behavior. In these systems they have shown that at high L:P ratios (low relative peptide concentrations), peptides typically adsorb onto the membrane, parallel to the membrane surface. As the L:P ratio is reduced (high relative peptide concentrations), peptide inserts into the membrane. The specific L:P ratio at which the adsorbed-to-inserted transition occurs depends on the type of lipid, peptide, and conditions such as degree of hydration and temperature. The change we observe in the direction of the phasor trajectory of the time-resolved fluorescence from AlexaFluor 430 near L:P 40:1 and the appearance of long time scale fluorescence kinetics at the critical concentration of L:P 20:1 and below might be due to the onset of helix insertion. This would place the AlexaFluor 430 probe in a less-polar microenvironment than it experiences on the surface of the bilayer. Critical behavior has also been reported in the context of morphological changes to phospholipid vesicles as a direct result of their interaction with peptides.29−31 It has been reported that the interaction of native melittin with phospholipid vesicles leads to the formation of bilayer discs at low L:P ratios and vesicle fusion at medium L:P ratios. At high L:P ratios, melittin appears to adsorb onto the bilayer surface. In the present study, we found evidence for some partial vesicle aggregation and/or vesicle fusion, as indicated by the increases in light scattering intensity. We did not find evidence for the formation of bilayer discs, which would have been detected through a significant decrease in light scattering intensity compared to the lipid-only control. It is important to add that bilayer discs are only observed in the gel phase of DPPC after the lipid has been taken through its main acyl chain melting transition. This might explain why we did not observe large decreases in vesicle scattering in the low lipid-to-peptide samples. The complex fluorescence kinetics observed at low L:P ratios for the melittin derivative were interpreted as arising from a sequence of events during the peptide−membrane interaction. At high L:P ratios, the spectroscopic signals reached a steady state within the 30 s mixing time. It is therefore reasonable to suggest that the peptide adsorption or binding event itself is very rapid. Indeed, the time-constant for the interaction of native melittin with gel-phase DPPC vesicles has been measured to be milliseconds35 (cf. vesicle fusion, which was recorded to be on the order of seconds to minutes54). For the 10:1 and 20:1 L:P samples, the short time-constants reported here for the melittin-derivative fluorescence (0.3 and 0.5 min) are therefore too slow to be due to initial surface binding and thus must represent subsequent peptide insertion and/or helix− helix assembly steps. The longer time constants (10−30 min) for melittin derivative fluorescence kinetics were unexpected, but they were associated with a decrease in environment micropolarity about the AlexaFluor 430 probe. It is important to note that the melittin derivative differs from native melittin due to a lysine substituted for proline at position 14. Studies of melittin derivative with an alanine substituted for proline at position 14 revealed differences in the rate of pore formation, channel formation (under an applied potential difference), and efflux kinetics compared to native melittin.55 Thus the unexpected kinetic behavior of the melittin derivative might be partially due to the proline-to-lysine substitution, which could potentially slow down some of the steps involved in the peptide−membrane

Table 1. Kinetic Parameters for Melittin Derivative−DPPC Complexes of Varying Lipid Concentrationa L:P ratio

scattering (τ, min)

scattering B

flu τ1 min

100:1 50:1 40:1 30:1 20:1 10:1 lipid only

7±1 140 ± 10 20 ± 5 26 ± 5 2 ± 0.5 22 ± 5 9±1

−0.07 ± 0.01 0.72 ± 0.04 0.66 ± 0.04 0.63 ± 0.04 0.13 ± 0.01 −0.14 ± 0.01 −0.11