Interaction of Gramicidin Derivatives with Phospholipid Monolayers

Sep 10, 2004 - Center for Self-Organising Molecular Systems, School of Chemistry, University of Leeds, LS2 9JT, United Kingdom, Department of Chemical...
0 downloads 14 Views 226KB Size
Download PDF
Langmuir 2004, 20, 9291-9298

9291

Interaction of Gramicidin Derivatives with Phospholipid Monolayers C. Whitehouse,†,⊥ D. Gidalevitz,‡ M. Cahuzac,§ Roger E. Koeppe II,| and A. Nelson*,† Center for Self-Organising Molecular Systems, School of Chemistry, University of Leeds, LS2 9JT, United Kingdom, Department of Chemical and Environmental Engineering, Illinois Institute of Technology, Chicago, Illinois 60616, Department of Chemical Engineering, University of Leeds, Leeds, LS2 9JT, United Kingdom 103, and Department of Chemistry and Biochemistry, University of Arkansas, Fayetteville, Arkansas 72701 Received May 14, 2004. In Final Form: July 21, 2004 A study of the interaction of gramicidin A (gA), tert-butyloxycarbonyl-gramicidin (g-BOC), and desformyl gramicidin (g-des) with dioleoyl phosphatidylcholine (DOPC) and DOPC/phosphatidylserine (PS) mixed monolayers on a mercury electrode is reported in this paper. Experiments were carried out in electrolytes KCl (0.1 mol dm-3) and Mg(NO3)2 (0.05 mol dm-3). The channel-forming properties of the gramicidins were studied by following the reduction of Tl(I) to Tl(Hg). The frequency dependence of the complex impedance of coated electrode surfaces in the presence and absence of the gramicidins was estimated between 65 000 and 0.1 Hz at potentials of -0.4 V versus Ag/AgCl with 3.5 mol dm-3 KCl. Epifluorescence microscopy was used to qualitatively correlate the interaction of the gramicidin peptides with dipalmitoyl phosphatidylcholine (DPPC) and dipalmitoyl phosphatidylglycerol (DPPG) at the air-water interface. gA was shown to form Tl+ conducting channels in a DOPC monolayer, while g-BOC and g-des did not. In DOPC-30% PS (DOPC0.3PS) layers, there is a marked increase in channel activity of all three gramicidin derivatives. None of the peptides facilitate the permeability of the DOPC-0.3PS layer to Cd2+. All three peptides interact with the layer as shown by capacitance-potential curves and impedance spectroscopy indicated by penetration of the peptide into the dielectric, an increase in surface “roughness”, and an increased significance of low-frequency relaxations. The order of interaction is gA > g-des > g-BOC. The epifluorescence study of DPPC and DPPG layers at the air-water interface shows a selective action of the different gramicidins.

Introduction The interaction of biologically derived peptides with phospholipid monolayers and bilayers has been of interest for some time.1-12 The results have relevance to biological mechanisms since peptide and protein interactions with biological membranes are of great significance in many aspects of physiology,7 such as cell signaling and toxicology. Of particular interest are the antimicrobial pep* To whom correspondence should be addressed. Fax: 44 113 6452. Tel: 44 113 6447. E-mail: [email protected]. † Center for Self-Organising Molecular Systems, School of Chemistry, University of Leeds. ‡ Department of Chemical and Environmental Engineering, Illinois Institute of Technology. § Department of Chemical Engineering, University of Leeds. | Department of Chemistry and Biochemistry, University of Arkansas. ⊥ Present address: Research and Development Division, Fujirebio Inc., 51 Komiya-cho, Hachioji-shi, 192-0031 Tokyo, Japan. (1) La Rocca, P.; Biggin, P. C.; Tieleman, D. P.; Sansom, M. S. Biochim. Biophys. Acta 1999, 1462, 185-200. (2) Maget-Dana, R. Biochim. Biophys. Acta 1999, 1462, 109-40. (3) Papo, N,; Y. Shai, Y. Biochemistry 2003, 42, 458-66. (4) Dathe, M.; Meyer, J.; Beyermann, M.; Maul, B.; Hoischen, C.; Bienert, M. Biochim. Biophys. Acta 2002, 1558, 171-86. (5) van’t Hof, W.; Veerman, E. C.; Helmerhorst, E. J.; Amerongen, A. V. Biol. Chem. 2001, 382, 597-619. (6) Huang, H. W. Biochemistry 2000, 39, 8347-52. (7) Killian, J. A.; von Heijne, G. Trends Biochem. Sci. 2000, 9, 42934. (8) Cserhati, T.; Szogyi, M. Int. J. Biochem. 1994, 26, 1-18. (9) van Kan, E. J. M.; Breukink, D. E.; van der Bent, A.; de Kruijff, B. Biochemistry 2002, 41, 7529-39. (10) Zasloff, M. Nature 2002, 415, 389-95. (11) Shai, Y. Biochim. Biophys. Acta 1999, 1462, 55-70. (12) Chen, F. Y.; Lee, M. T.; Huang, H. W. Biophys. J. 2003, 84, 3751-8.

tides1,2,5-6,9-12 and the membrane-active peptides,3 which act by disrupting biological membrane structure and function. The ability of antimicrobial peptides to kill bacteria while not disrupting native cells is attracting a great deal of attention, especially since traditional antibiotics are becoming increasingly difficult to produce and because of increased bacterial mutation. The mechanism of action of antimicrobial peptides is not entirely understood, but it is clear that they interact not with membrane proteins but with the lipid matrix itself and therefore leave little or no possibility for mutation which could affect their performance. Three models of membrane-rupturing mechanisms have been proposed to date: barrel-stave,13-17 carpet,18-20 and toroidal.21-23 According to the barrel-stave model, peptides bound to the membrane recognize each other and oligomerize. Upon oligomerization, antimicrobial peptides (13) Juvvadi, P.; Vunnam, S.; Merrifield, R. B. J. Am. Chem. Soc. 1996, 118, 8989-97. (14) Ehrenstein, G.; Lecar, H. Q. Rev. Biophys. 1977, 10, 1-34. (15) Christensen, B.; Fink, J.; Merrifield, R. B.; Mauzerall, D. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 5072-6. (16) Matsuzaki, K.; Harada, M.; Handa, T.; Funakoshi, S.; Fujii, N.; Yajima, H.; Miyajima, K. Biochim. Biophys. Acta 1989, 981, 130-4. (17) He, K.; Ludtke, S. J.; Worcester, D. L.; Huang, H. W. Biophys. J. 1996, 70, 2659-66. (18) Pouny, Y.; Rapaport, D.; Mor, A.; Nicolas, P.; Shai, Y. Biochemistry 1992, 31, 12416-23. (19) Shai, Y. Biochim. Biophys. Acta 1999, 1462, 55-70. (20) Wu, M.; Maier, E.; Benz, R.; Hancock, R. E. Biochemistry 1999, 38, 7235-42. (21) Mor, A.; Nicolas, P. J. Biol. Chem. 1994, 269, 1934-9. (22) Matsuzaki, K.; Murase, O.; Fujii, N.; Miyajima, K. Biochemistry 1996, 35, 11361-8. (23) Ludtke, S. J.; He, K.; Heller, W. T.; Harroun, T. A.; Yang, L.; Huang, H. W. Biochemistry 1996, 35, 13723-8.

10.1021/la048797l CCC: $27.50 © 2004 American Chemical Society Published on Web 09/10/2004

9292

Langmuir, Vol. 20, No. 21, 2004

orient themselves, allowing the hydrophobic surface to interact with the hydrophobic core of the membrane and the hydrophilic surface to point inward to create a hydrophilic transmembrane pore. The carpet model suggests that antimicrobial peptides initially bind to and cover the surface of the target membrane. The electrostatic interaction between the peptide and the lipid headgroup imposes strain in the membrane, and membrane permeation is induced only at sites where the local peptide concentration is higher than certain threshold values. In the toroidal model, peptides similarly bind and interact with lipid headgroups, imposing a positive curvature strain on the membrane (e.g., magainin 2) and producing channels where the polar headgroup region expands to form ‘‘toroidal” pores. Gramicidin is a well-known membrane-active peptide consisting of 15 amino acid residues. It forms channels in phospholipid bilayers of two β6.3 helices joined together through the hydrogen bonding of the terminal formyl groups within the hydrocarbon environment of the bilayer.24 Due to these properties, gramicidin is the most used channel model.25 In phospholipid monolayers, gramicidin forms monomolecular conducting channels presumed to be β6.3 helices oriented with the four tryptophan residues next to the polar groups.26 In many ways, the activity of gramicidin is unique in that the inside of the gramicidin β6.3 helix acts as the channel lumen. The phospholipid environment greatly affects the gramicidin secondary structure within it. Also, the lateral lipid organization and domain size distribution are significantly influenced by the incorporation of the polypeptide into the lipid bilayer system via a molecular sorting mechanism.27 Gramicidin A (gA) interactions with phospholipids have been conducted in monolayers at the air-water interface, and horizontal aggregation of the gramicidin molecules was shown to occur even at very low gramicidin/ phospholipid ratios.28 There has been much interest in different derivatives of gramicidin and their channel-forming activity.29-31 Particular gramicidin derivatives of interest are gramicidin-BOC,29 where the formyl group has been replaced by the tert-butyloxycarbonyl group, and desformyl gramicidin,30,31 where the formyl group has been removed altogether leaving a positively charged nitrogen atom. It has been shown that the derivatives gramicidin-BOC29 (g-BOC) and desformyl gramicidin30,31 (g-des) have little cation conducting channel activity in phospholipid bilayers. This can be expected since these species have no terminal formyl group, which holds the two channels together to form the bimolecular channel. g-des transports water molecules in bilayers, and a double helical structure for it has been proposed with the positively charged nitrogen next to the polar groups.30,31 This model could explain its poor cation conducting ability. It has been proposed that g-BOC forms the β6.3 helical monomer in each leaflet of a phospholipid bilayer.29 In light of the properties of gramicidin derivatives in bilayers, this study (24) Killian, J. A. Biochim. Biophys. Acta 1992, 1113, 391-425. (25) Hille, B. Ionic Channels of Excitable Membranes; Sinauer Associates: Sunderland, MA, 1992. (26) Nelson, A. J. Electroanal. Chem. 1991, 303, 221-36. (27) Fahsel, S.; Pospiech, E.-M.; Zein, M.; Hazlet, T.L.; Gratton, E.; Winter, R. Biophys. J. 2002, 83, 334-44. (28) Diociaiuti, M.; Bordi, F.; Motta, A.; Carosi, A.; Molinari, A.; Arancia, G.; Coluzza, C. Biophys. J. 2002, 82, 3198-206. (29) He, K.; Ludtke, S. J.; Wu, Y.; Huang, H. W.; Andersen, O. S.; Greathouse, D.; Koeppe, R. E., II Biophys. Chem. 1994, 49, 83-9. (30) Saparov; S. M.; Antonenko, Y. N.; Koeppe, R. E., II; Pohl, P. Biophys. J. 2000, 79, 2526-34. (31) de Groot, B. L.; Tieleman, D. P.; Pohl, P.; Grubmuller, H. Biophys. J. 2002, 82, 2934-42.

Whitehouse et al.

was initiated to compare the cation conducting ability of g-des and g-BOC with that of gA in phospholipid monolayers. Preliminary experiments using electrochemical impedance spectroscopy have shown that gA interacts with dioleoyl phosphatidylcholine (DOPC) monolayers, increasing the surface “roughness” and introducing an extra capacitative element.32 It seemed pertinent therefore to extend these studies by examining whether the different gramicidins gave different impedance responses when interacting with the layer and to assess whether their abilities to form channels in the layers could be correlated with their interactions with the layer. In addition to the electrochemical studies, epifluorescence microscopy was used to qualitatively correlate the interactions of the gramicidin peptides with dipalmitoyl phosphatidylcholine (DPPC) and dipalmitoyl phosphatidylglycerol (DPPG) monolayers at the air-water interface. For the epifluorescence microscopy, phospholipids with saturated chains were used because of their ability to close-pack more efficiently than unsaturated lipids, hence allowing one to observe changes in lipid monolayer morphology caused by their interaction with peptides. Experimental Section (a) Electrochemical Techniques. (i) Apparatus and Materials. Two distinct measurements were carried out using the electrochemical apparatus. The first series of measurements investigated the transport of Tl+ and Cd2+ ions in which a faradaic process is involved. The second series focused on impedance measurements in which no faradaic process is involved. The latter ones concentrated on capacitive elements. The results of both measurements are considered in terms of the properties of the phospholipid monolayer. An Autolab system, FRA and PGSTAT 30 interface (Ecochemie, Utrecht, The Netherlands), controlled with Autolab software, was used in all the electrochemical experiments. The experiments were performed in a standard three-electrode cell which was temperature controlled at 25 °C. A Maclab acquisition board and software (AD Instruments Ltd.) interfaced to the PGSTAT 30 were used to measure the reduction of Tl+. An Ag/AgCl, 3.5 mol dm-3 KCl reference electrode, with a porous sintered glass frit separating the 3.5 mol dm-3 KCl solution from the electrolyte, and a platinum bar served as reference and counter electrodes located on either side of the working electrode, respectively. A solution resistance of around 280-300 Ω was recorded for the cell. Diagnostic plots of the impedance data showed it to be that of an RC series circuit as before.32 There was a distinct absence of instability at high frequencies, and for this reason, the use of a fourth pseudoreference electrode was not considered necessary at this stage. The electrochemical cell and screened cables were contained in an aluminum faraday cage. The electrolytes, KCl (0.1 mol dm-3) and Mg(NO3)2 (0.05 mol dm-3), were prepared from Analar KCl (Fisher Chemicals Ltd.) calcined at 600 °C and Mg(NO3)2 (Merck Chemicals) using 18.2 MΩ MilliQ water. A blanket of argon gas was maintained above the fully deaerated electrolyte during all experiments. Monolayers of DOPC and DOPC + 30% phosphatidylserine (DOPC-0.3PS) were prepared as described earlier33-35 by spreading 13 µdm3 of a 2 mg cm-3 solution of DOPC and DOPC/PS mixture in pentane (HPLC grade, Fisher Scientific Chemicals Ltd.) at the argonelectrolyte interface in the electrochemical cell.33-35 The working solution of DOPC was obtained by dilution of a 50 mg cm-3 stock solution (Avanti Lipids). A working solution of PS was obtained by dilution of a 20 mg cm-3 stock solution (Lipid Products). A fresh mercury drop (area A ) 0.0088 cm2) was coated with the phospholipid33-35 from the argon-electrolyte interface prior to (32) Whitehouse, C.; O’Flanagan, R.; Lindholm-Sethson, B.; Movaghar, B.; Nelson, A. Langmuir 2004, 20, 136-44. (33) Nelson, A.; Benton, A. J. Electroanal. Chem. 1986, 202, 253-70. (34) Nelson, A.; Auffret, N. J. Electroanal. Chem. 1988, 244, 99. (35) Bizzotto, D.; Nelson, A. Langmuir 1998, 14, 6269-73.

Interaction of Gramicidins with Phospholipids each series of experiments. The DOPC-0.3PS mixed layer was used in these experiments as a suitable model for a negatively charged phospholipid monolayer since it is impermeable to Tl+. Monolayers of pure PS are permeable to Tl+ 36 and cannot be employed successfully for measuring gramicidin channel activity. gA (Calbiochem Ltd.) and g-des and g-BOC (synthesized as described in ref 37) were dissolved in methanol as working solutions (2.1 × 10-3 mol dm-3). Working solutions were stored at -20 °C. To look at the channel activity and interaction of gA, g-des, and g-BOC with the phospholipid layer, aliquots of the respective working solutions were injected below the layer into the electrolyte. A final electrolyte concentration of 13 nmol dm-3 peptide for the Tl+ and Cd2+ reduction measurements and 130 nmol dm-3 peptide for the impedance measurements of phospholipid-peptide interaction was used. The solution was then gently stirred for 5 min. The phospholipid monolayer with the incorporated gramicidin derivative was deposited on the electrode. In the experiments studying the Tl(I)/Tl(Hg) and Cd(II)/ Cd(Hg) reduction, TlNO3 (Sigma Products) and Cd(NO3)2 (BDH Chemicals Ltd.), respectively, were employed to prepare the stock solution (0.1 mol dm-3) from which aliquots were added to the electrolyte. At the end of the study, the working solutions of all three gramicidin derivatives were checked for purity by timeof-flight (TOF) mass spectroscopy where the spectra were acquired in the positive (ES+) mode. (ii) Electrochemistry of Tl+/Tl(Hg) and Cd2+/Cd(Hg) Redox Process. The following procedure38,39 was taken to measure gramicidin channel activity to the specific ion. After deaeration of the electrolyte, 10-4 mol dm-3 Tl(I) or Cd(II) was added from the stock solution. The phospholipid layer was then spread on the electrolyte and transferred to the electrode. A cyclic voltammogram was recorded to check the impermeability of the deposited layer. Following incorporation of gramicidin into the layer, the gramicidin-modified phospholipid layer was deposited on the electrode surface. A series of voltage pulses from -0.2 V to potentials from -0.3 to -0.7 V and back were initiated, and the current transients were recorded. The pulses were 40 ms long, and the currents were sampled at 40 kHz with a 20 kHz low pass filter. A delay period of 15 s between each pulse enabled the establishment of initial concentration conditions. After the pulsing program had been performed, an ac out-of-phase voltammogram was recorded to ensure that the phospholipid layer had not degraded during the experiment. Pulse transients were analyzed by sampling the current transient after a time interval of 2.5 ms from the beginning of the pulse and plotting this current value against potential as a sampled current voltammogram. (iii) Electrochemical Impedance. Measurements of capacity versus potential for the coated electrode were carried out by measuring the imaginary impedance (Z′′) at potentials between -0.2 and -1.05 V at a frequency (f) of 75 Hz with 0.005 V rms. These measurements were performed at 0.025 V intervals at potentials between -0.2 and -0.8 V and at 0.005 V intervals at potentials between -0.8 and -1.05 V. Capacitance (Cd) was calculated from the Z′′ value using the equation Cd ) (1/Z′′ω) where ω is the angular frequency ()2πf) assuming RC series behavior of the cell. Measurements of the impedance (Z) versus frequency of the electrode systems using frequencies logarithmically distributed from 65 000 to 0.1 Hz, 0.005 V rms at potentials of -0.4 V were carried out on the coated electrode systems. The experimental conditions for the measurement of impedance are listed in the following. For one measurement, one cycle was used except when the cycle was less than 1 s, in which case the measurement time was 1 s. To reach steady state, 10 cycles were used except when 10 cycles lasted more than 3 s, in which case 3 s was used. Each frequency scan took 5 min with the potential continually applied commencing with the highest frequency. These time intervals are a compromise in providing sufficient time to carry out the measurement and reaching steady state, while still enabling all the experiments to be done within a specified time period on one phospholipid layer without altering the structure of the layer. (36) Nelson, A. J. Chem. Soc., Faraday Trans. 1993, 89, 3081-90. (37) Greathouse, D. V.; Koeppe, R. E., II; Providence, L. L.; Shobana, S.; Andersen, O. S. Methods Enzymol. 1999, 294, 525-50. (38) Nelson, A.; Bizzotto, D. Langmuir 1999, 15, 7031-9. (39) Nelson, A. Biophys. J. 2001, 80, 2694-703.

Langmuir, Vol. 20, No. 21, 2004 9293 No significant difference in the spectra was noted when longer equilibration periods were used before each experiment. The impedance data were transformed to the complex capacitance plane, and the complex capacitance axes were expressed as Re Yω-1 and Im Yω-1, respectively. This was done using an Excel (Microsoft) spreadsheet. Curve fitting of the data was carried out using IGOR (Wavemetrics) in the same way as described previously.32 Due to the absence of any electroactive component, the simplest equivalent circuit model is the uncompensated solution resistance (Ru) of the cell and the capacitance (C) of the working electrode in series.40 Ru can be determined by extrapolating the Im Z versus Re Z plot to the Re Z axis.41 In the complex capacitance plane, values of Re Yω-1 were plotted against Im Yω-1 for all values of frequency.41-43 For a series RC circuit, the Re Yω-1 versus Im Yω-1 plot gives a single semicircle for the RC element, where the capacitor has no frequency dispersion. The extrapolation of this semicircle to the Im Yω-1 axis at low frequency gives the zero frequency capacitance (C)41-43 of the RC circuit which is therefore an empirical quantity. When applied to the phospholipid-coated electrode, any additional elements to the RC semicircle at lower frequencies will correspond to properties of the phospholipid layer. Further, if the semicircle representing the RC element is not perfect,44 the nonideality of the capacitor is indicated. This can be due to dielectric relaxations coupled to the RC charging process and to additional circuit elements at the interface between the capacitor and the solution resistance.45 All the impedance data were fitted to eq 1 below as done previously.32

1

Y) R+

[

(iω)βω01-β

1 Cs - Cinf

1 + (iωτ)R

]

(1)

+ Cinf

In eq 1, Y is the admittance, R is equivalent to the uncompensated solution resistance (Ru), Cinf is equivalent to the zero frequency capacitance (C) of the monolayer, Cs - Cinf is the additional lowfrequency capacitative element with relaxation time constant (τ), R is the coefficient which represents the distribution of time constants around a most probable value, and β is the coefficient which characterizes nonidealities at the interface between R and C and is equivalent to a surface roughness.44 (b) Epifluorescence Measurements. All epifluorescence data were collected by using a custom-built two-barrier Teflon Langmuir trough equipped with a Wilhelmy plate as described previously.46 Dulbecco’s phosphate-buffered saline (Invitrogen Life Technologies) without Ca2+ and Mg2+ was used as the subphase. The temperature of the subphase was maintained at 30 °C ( 0.5 °C for insertion experiments, and a resistively heated indium tin oxide coated glass plate was placed over the trough to minimize interference by contamination, air currents, and evaporative losses and also to prevent condensation of water on the microscope objective. Excitation between 530 and 590 nm and emission between 610 and 690 nm were gathered through the use of a HYQ Texas red filter cube; 0.5 mol % of lipid-linked Texas red (TR-DHPE) dye from Molecular Probes was incorporated into the spreading phospholipid solutions. Due to steric hindrance, the dye partitions between the disordered and the ordered phases, rendering them bright and dark, respectively. Images from the fluorescence microscope were collected at a video rate of 30 frames s-1 using a silicon intensified target (SIT) camera and recorded on Super-VHS formatted videotape with a recorder. (40) Wiegand, G.; Arribas-Layton, N.; Hillebrandt, H.; Sackmann, E.; Wagner, P. J. Phys. Chem. B 2002, 106, 4245-54. (41) Janek, R. P.; Fawcett, W. R.; Ulman, A. J. Phys. Chem. B 1997, 101, 8550-8. (42) Lindholm-Sethson, B. Langmuir 1996, 12, 3305-14. (43) Strasˇa´k, L.; Dvorˇa´k, J.; Hason, S.; Vetterl, V. Bioelectrochemistry 2002, 56, 37-41. (44) Peng Diao, P.; Jiang, D.; Cui, X.; Gu, D.; Tong, R.; Zhong, B. J. Electroanal. Chem. 1999, 464, 61-7. (45) Lingler, S.; Rubinstein, I.; Knoll, W.; Offenhausser, A. Langmuir 1997, 13, 7085-91. (46) Lipp, M. M.; Lee, K.Y. C.; Zasadzinski, J. A.; Waring, A. J. Rev. Sci. Instrum. 1997, 68, 2574-82.

9294

Langmuir, Vol. 20, No. 21, 2004

Whitehouse et al.

This assembly permits the monolayer morphology to be observed over a large lateral area while isotherm data are obtained concurrently. The entire apparatus was set on a vibration isolation table. Monolayers of DPPC and DPPG were spread from aliquots of chloroform solution and then compressed to 20 mN m-1. DPPG was used as a negatively charged phospholipid consistent with previous experiments.47,48 The peptides were injected underneath the monolayers at the constant pressure conditions. The peptide concentrations for epifluorescence measurements were the same as used in the electrochemical experiments for the reduction of Tl+ and Cd2+ (13 nmol dm-3). These experiments were a first attempt to image the interactions of the different gramicidin peptides with the phospholipid monolayer and correlate the results with those of electrochemical measurements. However by using a labeled phospholipid rather than a labeled peptide, the evidence for the interaction is indirect. Future studies will involve more detailed work where the different forms of gramicidin will be labeled to see if these might partition into the monolayer differently.

Results (a) Effect of Gramicidins on the Tl(I)/Tl(Hg) and Cd(II)/Cd(Hg) Redox Process. Sampled current voltammograms (Figure 1) show that in contrast to gA, g-des and g-BOC do not significantly facilitate permeability of Tl+ in DOPC. However, all gramicidins enable the permeability of DOPC-0.3PS layers to Tl+. Nonetheless, in DOPC-0.3PS layers, g-des and g-BOC increase the permeability of the layer to Tl+ to a smaller extent than gA and the facilitated permeability by g-des to Tl+ is potential dependent. All three gramicidin peptides do not increase the permeability of DOPC-0.3PS layers to Cd2+; in fact, they decrease the permeability of the phospholipid layer to Cd2+ in the order g-des > g-BOC > gA. These results were obtained with 13 nmol dm-3 gramicidin peptide added to the electrolyte. (b) Effect of Gramicidin Derivatives on the Impedance Properties of Phospholipid Monolayers. Figure 2 shows the effect of the gramicidin-DOPC interaction on the capacitance versus potential plots. These results were obtained from measurements carried out immediately following the electrolyte stirring period. Interaction of the gramicidin peptides with the phospholipid layers depressed the two DOPC capacitance peaks. This depression followed the order gA > g-des > g-BOC. In 0.05 mol dm-3 MgNO3 electrolyte, the suppression of the two peaks due to g-des interaction was not so great as in 0.1 mol dm-3 KCl. The capacitance minimum value showed the same trend, whereby the gramicidin interaction which effected the greatest suppression of the capacitance peaks caused the greatest increase in the capacitance minimum value. Figure 3 displays the effect of the gramicidin-DOPC interaction on the impedance data plotted in the complex capacitance plane. These results were obtained from measurements carried out immediately following the electrolyte stirring period. The influence of the gramicidin on the impedance data of the monolayer is related to the effect on the capacitance-potential plots and is manifest as an increased significance of low-frequency relaxations, a depression of the RC semicircle, and an increase in the zero frequency capacitance. The order of these effects caused by the gramicidin derivatives is gA > g-des > g-BOC, although each derivative had a qualitatively (47) Gidalevitz, D.; Ishitsuka, Y. J.; Muresan, A. S.; Konovalov, O.; Waring, A. J.; Lehrer, R. I.; Lee, K. Y. C. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6302-7. (48) Neville, F.; Cahuzac, M.; Nelson, A.; Gidalevitz, D. J. Phys.: Condens. Matter 2004, 16, S2413-20.

Figure 1. Sampled current voltammograms of the reduction of Tl+ and Cd2+ in 0.1 mol dm-3 KCl with (a,b) added Tl(I) (10-4 mol dm-3) and (c) added Cd(II) (10-4 mol dm-3) at (a) DOPCcoated and (b,c) DOPC-0.3PS-coated electrodes. The gramicidin derivatives added to the electrolyte (13 nmol dm-3) with their representative symbols are as follows: gA (open squares), g-BOC (open triangles), g-des (filled circles), and no peptide addition (crosses).

different influence on the nature of the impedance versus frequency plot. The impedance spectra of the gA interaction with DOPC and DOPC-0.3PS layers showed the presence of a second low-frequency element outside of the RC semicircle, which was also evident following interaction of g-des with the DOPC layer in 0.05 Mg(NO3)2 solution. Higher concentrations of gA were necessary to effect these changes than used previously.32 This was because in the previous experiments the gA was added to the electrolyte with no electrolyte stirring. Figure 4 shows a representative fit to data of eq 1.32 Only values of β and Cinf whose variation is most representative of the extent of interaction will be discussed in this study. Figure 5 displays a plot of the extracted β values against the zero frequency capacitance (Cinf) derived from the fit of eq 1 to the impedance data. This plot was obtained from successive (g6) measurements carried out on the monolayers successively deposited on the mercury from the gas-electrolyte interface. Results showed that over time the interaction of the gramicidins with the monolayer increased (shown by a decrease in β and an increase in Cinf with time) except with monolayers in Mg(NO3)2 electrolyte where the opposite effect with time occurred. These effects were not so evident with lower concentrations of gramicidin in the electrolyte (not shown). For all systems, the plot between β and Cinf is generally

Interaction of Gramicidins with Phospholipids

Langmuir, Vol. 20, No. 21, 2004 9295

Figure 2. Differential capacity calculated as 1/(Z′′ω) measured at 75 Hz and with 0.005 V rms versus potential of DOPC-coated mercury in (a) 0.1 mol dm-3 KCl and (b) 0.05 mol dm-3 Mg(NO3)2. The gramicidin derivatives added to the electrolyte (130 nmol dm-3) with their representative symbols are as follows: gA (open squares), g-BOC (open triangles), g-des (filled circles), and no peptide addition (crosses).

Figure 4. Representative fit (solid line) of eq 1 to impedance data (filled circles) in the complex capacitance plane obtained from DOPC-coated mercury in 0.1 mol dm-3 KCl with 130 nmol dm-3 added gA. Values for parameters and their respective standard deviations are listed below the figure. The numbers on the plot indicate frequencies of adjacent data points expressed as and representing values in log(ω/rad s-1) as follows: 1, 5.61; 2, 4.42; 3, 3.24; and 5, 0.87.

Figure 3. Plots in the complex capacitance plane of impedance data derived from (a) DOPC in 0.1 mol dm-3 KCl, (b) DOPC0.3PS in 0.1 mol dm-3 KCl, and (c) DOPC in 0.05 mol dm-3 Mg(NO3)2, deposited on mercury at -0.4 V. The gramicidin derivatives added to the electrolyte (130 nmol dm-3) with their representative symbols are as follows: gA (open squares), g-BOC (open triangles), g-des (filled circles), and no peptide addition (crosses). Numbers on the plots indicate frequencies of adjacent data points expressed as and representing values in log(ω/rad s-1) as follows: 1, 5.61; 2, 4.42; 3, 3.24; and 5, 0.87.

linear and the order of effect of the gramicidin derivatives of gA > g-des > g-BOC on the monolayer is clearly seen. The slope of the β versus Cinf plot derived from data of experiments carried out in 0.05 Mg(NO3)2 is steeper (Figure 5c). For g-BOC and g-des interaction with DOPC0.3PS, the points on the β versus Cinf plot tend to fall

above those corresponding to the gA/DOPC-0.3PS interaction (Figure 5b). (c) Interaction of Gramicidin Derivatives with DPPC and DPPG Monolayers at the Air-Water Interface. Fluorescence micrographs are displayed in Figures 6 and 7. These images were obtained after a sufficient time had been allowed for interaction to take place which was about 10 min for the DPPC and 40 min for the DPPG interaction. There is little significance in the length of these times except that the images were chosen to be most representative of the particular interaction. The images indicate that insertion of the peptide disrupts the morphology of the phospholipid monolayers showing a more pronounced condensed phase leading to a change in size and shape of the dark gray “liquid-condensed” monolayer phase domains and of the brighter disordered peptide-induced “fluid” film phase. Fluorescence microscopy showed a distinct difference in the nature of the gA, g-des, and g-BOC interaction with

9296

Langmuir, Vol. 20, No. 21, 2004

Whitehouse et al.

Discussion

Figure 5. Plot of values of β versus Cinf derived from the fitting of eq 1 to impedance data from all measurements of (a) DOPC in 0.1 mol dm-3 KCl, (b) DOPC-0.3PS in 0.1 mol dm-3 KCl, and (c) DOPC in 0.05 mol dm-3 Mg(NO3)2 coated mercury. Gramicidin derivatives (130 nmol dm-3 in electrolyte) together with their symbols are shown as follows: gA (open squares); g-des (filled circles); g-BOC (open triangles). Errors are within symbol size.

both DPPC and DPPG monolayers. gA and g-des to a lesser extent both cause aggregation of the condensed phase of DPPC, whereas g-BOC increases the area of the fluid film phase (Figure 6). The results of peptide interaction with the negatively charged DPPG are somewhat different. Only g-des causes aggregation of the condensed phase. gA interaction leads to the near disappearance of the condensed phase, and g-BOC has little apparent effect (Figure 7).

(a) Electrochemical Data from DOPC and DOPC0.3PS at the Hg-Electrolyte Interface. The height of the Tl+ reduction currents can be interpreted as relating to gramicidin channel activity in the monolayers.38,39 It is interesting that g-des and g-BOC do not conduct Tl+ in DOPC monolayers. The effect of the negatively charged DOPC-0.3PS monolayer is also interesting in that it increases the channel activity of the three gramicidin derivatives. The influence of the gramicidins on the reduction of Cd2+ through the DOPC-0.3PS monolayer shows a blocking effect, which is more significant with g-des and g-BOC than gA and shows that these peptides are also associated with the monolayer. Impedance results are sensitive to modification of the monolayer structure when the gramicidin concentrations added to the electrolyte are above those necessary to facilitate significant conduction of the layers to ions. The gramicidin interaction, as shown previously,32 gives rise to low-frequency relaxations outside the RC charging in addition to increasing the surface roughness of the monolayer, as indicated by a decrease in the β coefficient. The physical meaning of the low-frequency element outside of RC is due to peptide present in the monolayer. It is also present when experiments are carried out in Mg(NO3)2 electrolyte and therefore cannot be due to ion movement since the Mg2+ ion is unable to enter the gramicidin channel.25 Such low-frequency relaxations have been shown to be characteristic of phospholipid monolayers on mercury where inhomogeneity in the dielectric is introduced as for example during a phase transition.32 The effect on the impedance parameters of peptide interaction can be generally correlated with the effect of the interaction on the DOPC phase transitions. A depression of the capacitance peaks has previously been attributed to a loss of order in the monolayer.49 The increase in the zero frequency capacitance of the monolayer is linked to a penetration of the peptide into the dielectric of the monolayer. This leads to the linear relationship between β and Cinf. gA shows the strongest and g-BOC shows the weakest interaction. gA forms monomolecular channels in phospholipid monolayers,38,39 and this is commensurate with the

Figure 6. Fluorescence micrographs of DPPC taken before and after injection of gramicidin derivatives at constant pressure of 20 mN/m: (A) pure DPPC, (B) after injection of gA, (C) after injection of g-des, and (D) after injection of g-BOC.

Interaction of Gramicidins with Phospholipids

Langmuir, Vol. 20, No. 21, 2004 9297

Figure 7. Fluorescence micrographs of DPPG taken before and after injection of gramicidin derivatives at constant pressure of 20 mN/m: (a) pure DPPG, (b) after injection of gA, (c) after injection of g-des, and (d) after injection of g-BOC.

increased capacitance and surface roughness. Deformation of the monolayer surface in the region of the channel mouth and lateral aggregation of monomer channels would account for the increase in monolayer roughness.28,50 The increasing effect of the gramicidins on the monolayer impedance properties with time could be correlated with such a change in the gramicidin secondary structure. This would be more significant with higher concentrations of gramicidin in the monolayer. At concentrations of 13 nmol dm-3 gA in the electrolyte, the low concentration of channels partitioned into the monolayer38,39 is not sufficient to significantly affect the impedance data. The generally stronger interaction of the gramicidin peptides with monolayers of DOPC-0.3PS than with monolayers of DOPC (see Figure 5a,b) can go some way to explaining the increased channel activity of all peptides in DOPC0.3PS. The decreased interaction of the BOC derivative with the monolayers can be related to the effect of the BOC group. It has been established that a terminal BOC group can significantly change the secondary structure of a small peptide51 and could inhibit the gramicidin derivative forming the β6.3 species, which penetrates the phospholipid layer and conducts Tl+. Although the channel activity of the gramicidins can be correlated with the extent to which they interact with the monolayer, there are differences in the permeability, which cannot be wholly explained by gramicidin-phospholipid interaction. A good example of this is the lack of channel activity of g-des in DOPC monolayers although the impedance data show a significant interaction of g-des with DOPC layers. Desformyl gramicidin has been reported as being mainly in the double helical conformation29,30 in phospholipid bilayers. The reason for forming this species is to prevent the positively charged nitrogen group from being embedded in the hydrocarbon matrix of the layer, which is energetically unfavorable. The cationic permeability of g-des in bilayers is 5 times lower than that due to gA due to the positive charge at the channel mouth. At the same (49) Nelson, A.; Auffret, N.; Borlakoglu, J. Biochim. Biophys. Acta 1990, 1021, 205-16. (50) Goforth, R. L.; Chi, A. K.; Greathouse, D. V.; Providence, L. L.; Koeppe, R. E., II; Andersen, O. S. J. Gen. Physiol. 2003, 121, 477-93. (51) Ganesh, S.; Jayakumar, R. J. Peptide Res. 2002, 59, 249-56.

Figure 8. Circular dichroism spectrum of g-des in DMPC and 0.1 mol dm-3 phosphate buffer at pH 7.

time, g-des was shown to facilitate very high water conductivity. It is interesting and relevant to this study that a circular dichroism (CD) spectrum (see Figure 8) of g-des in dimyristoyl phosphatidylcholine (DMPC) shows a mixture of single-stranded and double-stranded β helical conformations. The sample was prepared for CD spectroscopy as described by Greathouse et al.52 In phospholipid monolayers, it is generally unfavorable to form the double helical species of g-des since the length of the double helix of 2.6 nm30,31 exceeds the thickness of the monolayer (∼1.3 nm).39 The results of this study show that g-des penetrates the phospholipid monolayer to almost the same extent as gA but exhibits a lower channel activity to Tl+. If the g-des is present as a β6.3 helix form, then its orientation in the monolayer is uncertain due to its positive nitrogen atom. A reasonable suggestion is that it forms the β6.3 helix but its positively charged nitrogen group is oriented toward the polar head region. This would account for its inability to conduct Tl+ in DOPC due to electrostatic repulsion at the channel mouth and its increased conductivity in layers of DOPC modified with the negatively charged PS. (b) Epifluorescent Data from DPPC and DPPG at the Air-Water Interface. Similar to other antimicrobial peptides, for example, protegrin (PG-1)47 and LL-3748 from the cathelicidin family, gA, g-des, and g-BOC are likely to insert into the disordered fluid phase of the monolayer and disrupt the ordered liquid-condensed phase. This leads to a pronounced change in shape and size of the condensed (52) Greathouse, D. V.; Hinton, J. F.; Kim, K. S.; Koeppe, R. E., II Biochemistry 1994, 33, 4291-9.

9298

Langmuir, Vol. 20, No. 21, 2004

phase domains, which is most significant for gA-DPPC interaction and least significant for g-BOC-DPPC interaction. This is a similar order of effect to the gramicidin-DOPC interaction where the phospholipid headgroup is identical. In the same manner, gramicidin peptides insert into the fluid phase of the DPPG monolayer with a subsequent increase in a bright domain area. This is not directly comparable to the electrochemical data since DPPG is different from the mixed layer DOPC-0.3PS and the effect of the g-des interaction could be associated with its positive charge interacting with a negatively charged phospholipid. Despite the qualitative nature of the results, it is important to note that morphological changes in DPPC and DPPG monolayers caused by their interaction with gA, g-des, and g-BOC peptides are quite dissimilar for each peptide. This indicates a difference in peptidephospholipid interactions, due to the structural modifications of the peptide. Conclusions gA and its derivatives g-BOC and g-des each have specific interactions with phospholipid monolayers. Tl+ conductance by gA is the highest and a factor of 2 greater in monolayers of DOPC-0.3PS than in monolayers of DOPC. g-des and g-BOC do not significantly increase the conductance of Tl+ in DOPC monolayers. The Tl+ conductance of g-BOC is half that due to gA in DOPC-0.3PS. The Tl+ conductance of g-des in DOPC-0.3PS monolayers is similar to that of g-BOC, but voltage dependent.

Whitehouse et al.

Impedance data show that the interaction of the gramicidins with the phospholipid monolayers is in the order gA > g-des > g-BOC, although there are small characteristic differences in interaction between the peptides. The low Tl+ conductance in g-BOC-modified phospholipid can be attributed to a weaker g-BOC interaction with DOPC. The lower Tl+ conductance in g-des-modified phospholipid appears to be due to the lower inherent channel conducting activity to Tl+ of the positively charged peptide. Results from epifluorescent microscopy of the gramicidin derivative interaction with DPPC and DPPG at the airwater interface show that the interactions with gA, g-des, and g-BOC peptides are quite dissimilar, indicating a difference in peptide-phospholipid interactions, due to their structural modifications. Acknowledgment. Funding for this work was provided by the Joint Grant Scheme (MoD-NERC) and the EPSRC Grant Ref. GR/R67439. We are grateful to Ka Yee C. Lee for her hospitality at the University of Chicago. Thanks to Alison Ashcroft of Leeds University for checking the purity of the gramicidin peptides at the end of the investigation. R.K. II acknowledges the financial support of the Arkansas Biosciences Institute. We thank Denise Greathouse for synthesizing the desformyl and BOC gramicidins and Anna Daily for recording the CD spectrum. LA048797L