Langmuir 2007, 23, 1403-1409
1403
Characterization of Pore Formation by Streptolysin O on Supported Lipid Membranes by Impedance Spectroscopy and Surface Plasmon Resonance Spectroscopy Thomas Wilkop, Danke Xu, and Quan Cheng* Department of Chemistry, UniVersity of California, RiVerside, California 92521 ReceiVed August 30, 2006. In Final Form: October 20, 2006 We report the study of the interactions of bacterial toxin streptolysin O (SLO) and cholesterol-containing membranes using electrochemical impedance and surface plasmon resonance (SPR) spectroscopy at low hemolytic units on a novel supported membrane interface. The detailed understanding of the process will aid significantly the construction of nanoscale transport channels for biosensing applications. Cholesterol-containing egg PC vesicles, pristine and incubated with SLO toxin, were fused onto a hexyl thioctate (HT)-modified gold substrate. The charge-transfer resistance of the resulting lipid membrane, which is related to the formation of the transmembrane pores, is measured with the aid of an electroactive probe. Impedance spectra were collected over a range of 0.1-100 kHz, and the obtained complex resistance was fit to an equivalent circuit. The charge-transfer resistance decreases for increasing SLO concentration, following a first-order exponential decay. The two-part membrane interface was further characterized with SPR spectroscopy. For the hexyl thioctate support layer, an equivalent monolayer thickness of 1.3 nm was determined. This value suggests a loosely packed structure of the monolayer on gold, presenting an ideal platform for permeability studies. A comparative study on the fusion behavior of vesicles with and without SLO induced pores revealed no substantial difference for the two systems, indicating that the pore formation has no adverse effect on the integrity of the vesicles. The resulting lipid membrane thickness from pre-perforated lipids was found to be 3.2 nm, suggesting that one leaflet is knocked off during the fusion process and a hybrid membrane is formed. A slightly higher thickness value of 3.4 nm was obtained for membranes from non-perforated vesicles. Deposition of lipids and subsequent incubation with SLO, as monitored by SPR, shows that the HT surface chemistry allows partial insertion of the toxin into the membrane, indicating unique properties as compared to the previously explored long-chain alkylthiols.
Introduction Streptolysine O (SLO) is a pore-forming, β-hemolysin produced by the Gram-positive bacterium Streptococcus pyogenes. This bacterium is the causative agent for a variety of prevalent diseases, such as streptococcal toxic shock syndrome, streptococcal skin infection, and scarlet fever.1 Its virulence is exerted by insertion of SLO into the target cells’ lipid membrane, after binding to cholesterol and the formation of large pores, which lead to a loss of intracellular content and eventual cell death.2 The intricate biological mechanism by which the pores are formed is still a matter of ongoing research and debate.3-5 However, this phenomenon has been utilized for drug delivery6 and sensing.7-9 Pore formation, or the creation of large-scale molecular diffusion channels, was demonstrated in the detection of pore-forming agents by encapsulating reporting molecules inside a vesicle and observing the triggered release. This approach was employed for the detection of Eastern Diamondback rattlesnake venom ,7 SLO,8 and Listeriolysin O.9 Throughout * To whom correspondence should be addressed. E-mail: quan.cheng@ ucr.edu. Tel: (951) 827-2702. (1) Ryan, K. J. C. G. R. Sherris Medical Microbiology; McGraw-Hill Medical: New York, 2003. (2) Alouf, J. E.; Freer, J. H., Eds. ComprehensiVe Sourcebook of Bacterial Protein Toxins, 2nd ed.; Academic Press: London, 1999; p 718. (3) Bhakdi, S.; Tranumjensen, J. AdV. Exp. Med. Biol. 1985, 184, 3-21. (4) Hugo, F.; Reichwein, J.; Arvand, M.; Kramer, S.; Bhakdi, S. Infect. Immun. 1986, 54 (3), 641-645. (5) Sekiya, K.; Danbara, H.; Yase, K.; Futaesaku, Y. J. Bacteriol. 1996, 178 (23), 6998-7002. (6) Giles, R. V.; Grzybowski, J.; Spiller, D. G.; Tidd, D. M. Nucleosides Nucleotides 1997, 16 (7-9), 1155-1163. (7) Toby, A.; Jenkins, A.; Olds, J. A. Chem. Commun. 2004, 18, 2106-2107. (8) Xu, D.; Cheng, Q. J. Am. Chem. Soc. 2002, 124 (48), 14314-14315. (9) Zhao, J.; Jedlicka, S. S.; Lannu, J. D.; Bhunia, A. K.; Rickus, J. L. Biotechnol. Prog. 2006, 22 (1), 32-37.
these studies, different reporting molecules and detection schemes have been utilized, ranging from fluorescence9 to electrochemical techniques such as cyclic voltammetry8 and chronoamperometery.7 It is important to note that the analyte-triggered release of vesicular material constitutes an amplification step since binding of a small number of pore-forming toxin molecules can release a large number of reporting molecules. The degree of amplification (i.e., the ratio of molecules released to the number of toxin monomers) is partially limited by the amount of the encapsulated species. Transmembrane pore formation in vesicles is only one specific application of transport across artificial membranes. The collective applications were recently exhaustively reviewed by Janshoff.10 Despite the proven applications, details about the poreformation process and the vesicle characteristics after perforation remain unknown. A better understanding will significantly improve the effective design of sensing devices using this scheme. In this work, we demonstrate a new surface for hosting biomimetic membranes and the study of the pore-formation process by impedance spectroscopy (IS) and surface plasmon resonance (SPR) spectroscopy. Apart from the potential analytical significance of the SLO detection, the work underscores the complementary nature of IS and SPR for probing membrane integrity and other properties. Molecular transport channels are generated in vesicles through incubation with SLO toxin, and the vesicles are then fused onto a hexyl thioctate (HT)-modified gold substrate. The charge-transfer resistance of the resulting membrane, which is closely related to the permeability, is then measured with an electroactive probe using impedance spectroscopy. This approach has two advantages over existing methods: (a) the signal is time independent (i.e., diffusion of the (10) Janshoff, A.; Steinem, C. Anal. Bioanal. Chem. 2006, 385 (3), 433-451.
10.1021/la0625502 CCC: $37.00 © 2007 American Chemical Society Published on Web 12/10/2006
1404 Langmuir, Vol. 23, No. 3, 2007
encapsulated species does not constitute an additional variable) and (b) the concentration of the electroactive probe can be increased to levels beyond those obtainable by encapsulation. A key element for successful operation in this measuring scheme is the well-defined electrode behavior. To take advantage of Faradaic impedance spectroscopy, the electrode must allow facile electron-transfer while simultaneously facilitate fusion of the vesicles into a complete and dense layer. Vesicles are known to form ideal bilayers through spontaneous fusion onto silicate surfaces.11-13 Unfortunately, glassy surfaces are insulators and hence unsuitable for electrochemical measurments. On unmodified gold surfaces, vesicles remain intact.12 Even for the densest packing arrangement, much of the surface remains uncovered. Alkanethiols with hydrophilic endgroups have facilitated vesicle fusion into membranes,14 while hydrophobic headgroups yielded hybrid membranes.14,15 The inherently dense packing of these types of SAM layers prohibits an effective charge-transfer.16,17 To obtain a suitable SAM with a medium degree of hydrophobicity that simultaneously maintains its permeability for the electroactive species, we have synthesized a disulfide-terminated hexyl thioctate. Its unique molecular structure, with the two terminating sulfur units, ensures loose packing. The high sensitivity and noninvasive nature of impedance spectroscopy makes it an ideal tool for the characterization of electrode coatings and thin layer membrane systems.18 It has been extensively used for studying the various transport mechanisms of proteins across biological membranes,19 synthetic membranes,20 and fused lipid layers.21,22 Low-frequency IS can obtain thickness resolutions of 0.1 nm.23 Impedance measurements are made by applying a small AC voltage of dispersing frequency across the system and measuring simultaneously the current amplitude and the phase shift between the voltage and the current. In a multi-element system, it is possible to identify the individual elements since each has its characteristic frequency response. Detailed theoretical treatments and extensive application samples can be found in the literature.24-26 SPR is an optical technique that is sensitive to surface refractive index changes. At identical frequencies, the dielectric constant (11) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75 (3), 1397-1402. (12) Reimhult, E.; Zach, M.; Hook, F.; Kasemo, B. Langmuir 2006, 22 (7), 3313-3319. (13) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103 (13), 2554-2559. (14) Silin, V. I.; Wieder, H.; Woodward, J. T.; Valincius, G.; Offenhausser, A.; Plant, A. L. J. Am. Chem. Soc. 2002, 124 (49), 14676-14683. (15) Plant, A. L.; Gueguetchkeri, M.; Yap, W. Biophys. J. 1994, 67 (3), 11261133. (16) Campuzano, S.; Pedrero, M.; Montemayor, C.; Fatas, E.; Pingarron, J. M. J. Electroanal. Chem. 2006, 586 (1), 112-121. (17) Ding, S. J.; Chang, B. W.; Wu, C. C.; Lai, M. F.; Chang, H. C. Anal. Chim. Acta 2005, 554 (1-2), 43-51. (18) Coster, H. G. L.; Chilcott, T. C.; Coster, A. C. F. Bioelectrochem. Bioenerg. 1996, 40 (2), 79-98. (19) Chilcott, T. C.; Shartzer, S. F.; Iverson, M. W.; Garvin, D. F.; Kochian, L. V.; Lucas, W. J. Compt. Rend. 1995, 318 (7), 761-771. (20) Coster, H. G. L.; Chilcott, T. C. Surfact. Sci. Ser. 1999, 79 (Surface Chemistry and Electrochemistry of Membranes), 749-792. (21) Stelzle, M.; Weissmueller, G.; Sackmann, E. J. Phys. Chem. 1993, 97 (12), 2974-81. (22) Jenkins, A. T. A.; Bushby, R. J.; Evans, S. D.; Knoll, W.; Offenhaeusser, A.; Ogier, S. D. Langmuir 2002, 18 (8), 3176-3180. (23) Coster, H. G. L.; Smith, J. R. Biochim. Biophys. Acta 1974, 373 (2), 151-164. (24) Evgenij Barsoukov, J. R. M. Impedance Spectroscopy: Theory, Experiment, and Applications; Wiley-Interscience: New York, 2005. (25) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications; Wiley: New York, 2000. (26) Orazem, M. E.; Tribollet, B. Electrochemical Impedance Spectroscopy; Wiley-Interscience: New York, 2006. (27) Mullett, W. M.; Lai, E. P. C.; Yeung, J. M. Methods 2000, 22 (1), 77-91.
Wilkop et al.
of a medium is related to the refractive index n by eq 1.
) n2
(1)
It is imperative to realize that the dielectric constant disperses over a large frequency range and that the dielectric constant used for SPR measurements conducted at fractional terahertz frequencies differs from that measured in the hertz to megahertz range by IS. SPR spectroscopy and its related imaging form have been extensively applied in biomolecular interaction analysis.27 Several studies have used SPR to characterize lipid membrane systems.28-30 In SPR spectroscopy, p-polarized light is directed through a prism onto a semi-transparent metal film, under total internal reflection conditions. Through the conservation of energy and momentum, energy from the incident light is then transferred into surface-bound propagating electron oscillations. When the wave vector of the incoming photon matches that of the propagating plasmon, the energy coupling manifests itself in a characteristic dip in the reflectivity spectrum. The popularity of SPR is rooted in its high sensitivity and the fact that analytes can be detected in real time without labels. In this work, we report the characterization of the HT/PC hybrid membranes and their interaction with the pore-forming streptolysin O toxin by SPR. Emphasis is placed on the comparison of the lipid membranes originating from perforated and pristine vesicles. The binding kinetics is assessed to provide information on possible structural changes in the vesicles during pore formation. Information obtained in the course of our study can aid the exploration of new surface architectures for future sensor development. Experimental Procedures Materials. L-R-Phosphatidylcholine (egg PC) was purchased from Avanti Polar Lipids (Alabaster, AL). Cholesterol, ferrocene carboxylic acid (FCA), dicyclohexyl carbodiimide, thioctic acid, 4-dimethylaminopyridine, triethylene anhydrous glycol, and 1-hexanol were obtained from Aldrich. Streptolysin O (from S. pyogenes), bovine serum albumin, and Trizma were bought from Sigma. Synthesis of Hexyl Thioctate (HT). A total of 2.0 mmol of 1-hexanol was added to a stirred solution of thioctic acid (2.0 mmol) in 10 mL of methylene chloride. The mixture was stirred in an ice/water bath at 0 °C for 15 min under N2. Dicyclohexyl carbodiimide (4 mmol) and 4-dimethylaminopyridine (0.6 mmol) in cold CH2Cl2 (10 mL) were added to the solution, and the mixture was stirred for an additional 15 min at 0 °C. The reaction was then brought to room temperature and allowed to react for 20 h. The resulting solution was filtered through a glass funnel with a fine fritted disc and washed with water. After separation, the organic layer was dried with MgSO4, filtered, and evaporated. The raw yellow oily product was further purified with a flash column (silica gel, 230-400 mesh) and eluted with CHCl3/methanol (97.5/2.5). The product was confirmed by mass spectrometry and NMR. NMR spectral data: 1H NMR (300 MHz, CDCl3). δ ) 0.73-0.97 (t,3H); 0.97-1.37 (m,6H); 1.381.79 (m,8H); 1.79-2.00 (m,1H); 2.16-2.34 (t,2H); 2.34-2.54 (m,1H); 2.96-3.28 (m,2H); 3.43-3.64 (m,1H); 3.90-4.15 (t,2H). Preparation of Vesicles and Hybrid Membranes on Gold Substrates. Vesicles were prepared by probe sonication (Model 250 Sonifier, Baranson Ultrasonics). Egg PC and cholesterol in chloroform were mixed inside an amber vial in a 1:1 molar ratio. The organic solvent was then removed with a N2 stream to form a thin lipid layer. Appropriate amounts of a 10 mM Tris buffer (pH 7.5 containing 150 mM NaCl) were added, yielding a final lipid concentration of 1.0 mg/mL. The solution was then sonicated at low (28) Naumann, R.; Schiller, S. M.; Giess, F.; Grohe, B.; Hartman, K. B.; Karcher, I.; Koper, I.; Lubben, J.; Vasilev, K.; Knoll, W. Langmuir 2003, 19 (13), 5435-5443. (29) Tawa, K.; Morigaki, K. Biophys. J. 2005, 89 (4), 2750-2758. (30) Ma, C.; Srinivasan, M. P.; Waring, A. J.; Lehrer, R. I.; Longo, M. L.; Stroeve, P. Colloids Surf., B 2003, 28 (4), 319-329.
Characterization of Pore Formation by Streptolysin O
Langmuir, Vol. 23, No. 3, 2007 1405
Figure 1. (a) Structure of the self-assembled monolayer of the disulfide-terminated hexyl thioctate on the Au surface. (b) Cartoon illustration of the two approaches used in the study. The top pathway represents the vesicle fusion into a hybrid supported bilayer with subsequent SLO insertion, while the bottom part represents the incubation of lipid vesicles with SLO toxin and its subsequent fusion on the HT surface. amplitude at 0 °C for 30 min. After sonication, the resulting solution was filtered and stored at 4 °C for 1 h and used within 36 h after preparation. The substrates for IS were prepared by thermal evaporation of a 200 nm thin film of gold on a microscope glass slide. The substrates for the SPR analysis (47 nm of gold with a Cr adhesion layer) were purchased from Microanalytical Systems (Sinzing, Germany). Prior to monolayer modification, the Au substrates were cleaned with freshly prepared piranha solution (Caution!), followed by extensive rinsing with deionized water and drying in a N2 stream. The substrates were then immersed in 1.0 mM HT in ethanol for 2 h to form a monolayer through self-assembly (SAM). The HT SAM layer was rinsed with ethanol and deionized water before vesicle fusion. Electrochemical Impedance Measurements. The electrochemical impedance measurements were carried out with a three-electrode system using a CHI 650 electrochemical work station (CH Instruments, Austin, TX). The Au substrates were assembled in the electrochemical cell (defined by an O-ring with an i.d. of 2 mm), modified with the hybrid layer, and used as the working electrode. A Ag/AgCl electrode was used as reference, and a platinum wire was used as the counter electrode. Impedance spectra were recorded over a range of 0.1-100 kHz, using 1.0 mM ferrocenecarboxylic acid (FCA) as probe in 10 mM Tris buffer (pH 7.5) with 150 mM NaCl. The spectral analysis was performed with Zview (Scribner Associates, Southern Pines). The DC bias for the Faradic reaction of FCA was set to 0.3 V, corresponding to the oxidation potential of the ferrocenecarboxylic acid, as determined by cyclic voltammetry. Surface Plasmon Resonance Spectroscopy. Spectroscopic SPR experiments were conducted with a Biosuplar 2 instrument from Microanalytical Systems (Sinzing, Germany). The instrument scans the reflectivity of the substrates mounted in the Kretschmann configuration at 670 nm. The experiments were conducted at elevated room temperature (26 °C). Fluids were injected with a syringe pump, via a chromatography valve into a single flow cell (volume 20 µL).
SPR measurements, performed after the baseline stabilized, were used to monitor the HT immobilization and the lipid deposition process in situ. Measurements were carried out in two modes. Angular spectra were recorded for systems under equilibrium conditions, and fixed angel reflectivity measurements were used for kinetic measurements.
Results and Discussion SPR Characterization of the HT Monolayer and Its Associated Membranes. The use of alkylthiol SAMs as sublayers for vesicle fusion has been previously investigated.14 It is established that the packing of the SAM has an enormous impact on the membrane structure formed on it. We initially set out to characterize the HT monolayer formation by SPR. The molecular structure of the HT and the membrane-forming scheme are shown in Figure 1. The self-assembly process of the HT onto the gold electrode surface is very rapid. For a concentration of 1 mg/mL, it is essentially complete after 3 min. The HT monolayer shifts the resonance angle by 0.16°. On the basis of a refractive index of 1.47 for the alkyl functional group, the resulting equivalent layer thickness can be calculated as 1.3 nm. Considering the oxygen atoms in the molecular structure, the result indicates that the film is not densely packed. Figure 2 shows the cyclic voltammograms for FCA on the gold electrode (solid curve) and the HT-modified electrode (dash curve). Although the current magnitude is slightly reduced on the HT surface, the electrokinetics, demonstrated by the peak separation in the CV, is essentially the same. This indicates that the HT layer does not substantially block the electron-transfer, as other alkylthiols such as decanethiol15 do in similar systems.
1406 Langmuir, Vol. 23, No. 3, 2007
Figure 2. Cyclic voltammograms for 1 mM FCA on a bare (solid curve) and HT-modified electrode (dashed curve). The electrolyte solution contained 10 mM Tris buffer (pH 7.5) and 150 mM NaCl. The scan rate was 100 mV/s.
SPR Measurements for Pristine and Toxin Incubated Vesicles. The effect of pore formation by SLO on the vesicle fusion was investigated by SPR. The fusion of both types of vesicles, pristine and perforated with 1 HU/µL SLO, was monitored in a flow-injection format, and the results are compared in Figure 3. HU is the hemolytic unit for SLO toxin and is defined as the amount of protein that causes 50% lysis of a 2% red blood cell suspension in phosphate-buffered saline at pH 7.4. The 1 HU/µL of SLO corresponds roughly to a 12.5 nM concentration. After the baseline stabilized, vesicles were injected from a 500 µL sample loop. At a flow rate of 12 mL/h, both types of vesicles appear to fuse rapidly. Analysis of the rate of deposition inside the flow cell yields a value of 0.12 and 0.17 RU/s for the SLO incubated and non-incubated vesicles, respectively. This is a relatively small difference. The course of the deposition for the pristine vesicles is monotone; the signal increases constantly until stabilizing. For the incubated vesicles, the signal exhibits a shallow peak before it assumes a steady-state value. The total signal difference for both membranes is very similar. These results suggest that incubation with SLO did not disrupt the integrity of the vesicle structure, which is consistent with earlier findings by Xu31 and Sekiya.5 Under constant flow, the lipid layer requires an extensive washing period before a stable film thickness is obtained. A markedly different deposition kinetics was observed under zero flow. No plateau was obtained during the course of the experiment, suggesting a continuous deposition under this condition. This behavior is not unique for the HT surface but has also been observed for hydrophobic SAMs and binary SAMs. The addition of partial addlayers and non-fused vesicles could account for this kind of behavior.14 We determined the thickness of the resulting lipid layers from a series of vesicle depositions. By using the refractive index of 1.45 for phospholipids,32-34 the observed angular shifts (averaged from at least three independent experiments) correspond to a medium film thickness of 3.4 and 3.2 nm for pristine and (31) Xu, D. K.; Cheng, Q. J. Am. Chem. Soc. 2002, 124 (48), 14314-14315. (32) Terrettaz, S.; Ulrich, W. P.; Guerrini, R.; Verdini, A.; Vogel, H. Angew. Chem., Intl. Ed. 2001, 40 (9), 1740-1743. (33) Meuse, C. W.; Krueger, S.; Majkrzak, C. F.; Dura, J. A.; Fu, J.; Connor, J. T.; Plant, A. L. Biophys. J. 1998, 74 (3), 1388-1398. (34) Lang, H.; Duschl, C.; Gratzel, M.; Vogel, H. Thin Solid Films 1992, 210 (1-2), 818-821.
Wilkop et al.
Figure 3. SPR sensorgrams for the deposition and fusion processes of the lipid vesicles. The curve with full circles represents vesicle injection and incubation under zero-flow conditions. The other two trances show the constant flow deposition for SLO incubated (full squares) and non-incubated vesicles (empty squares).
perforated membranes, respectively. Given the hydrophobic nature of the HT surface, we conclude that one leaflet is knocked off during the fusion process and that a hybrid membrane is formed, similar to observations on octadecanethiol35 and decanethiol.15 As compared to the 3.4 nm thickness for pristine membranes, the slightly lower value obtained from the SLO incubated sample is indicative of the insertion of pores and removal of lipid molecules, resulting in a less complete film with a partition of solvent molecules. Another interesting property of the interface is if it can support in situ pore formation. Hence, the interaction of the SLO toxin with the fused lipid on the HT SAM was studied for toxin concentrations of 0.1 and 1 HU/µL. The toxin interaction process has two aspects: (a) binding and (b) pore formation. The process is complex since the two effects influence the SPR coupling angle in opposite directions. The SLO binding to the membrane leads to an effective surface mass increase, whereas the holeforming process, with its displacement of material, may decrease the signal. To ensure that the changes are accurately measured and not compromised by variations in the scans, we recorded each spectrum at least twice to ensure repeatability. Figure 4 shows SPR spectra for the membrane deposition and interaction with 0.1 HU/µL SLO toxin. A close up of the spectra around the maximum resonance angle is plotted for clarity. The fusion of vesicles results in a shift of 0.3°. Incubation of the membrane with 0.1 HU/µL SLO results only in a very small angular change. Noticeably, the degree of resonance decreases during each step. The process is apparently affected by two parameters: the angular position of the resonance maximum and the absolute reflectivity value. The change in the reflectivity can only be accounted for by an increase in the extinction coefficient k, a measure for the absorptivity of a medium. Since the materials are non-absorbing at 670 nm, we attribute the apparent increase in k to scattering effects. This is likely caused by the surfacebound SLO and possible insertion into the membrane. The Figure 4 insert shows the interaction of a tenfold higher concentration of SLO (1 HU/µL) with a fused membrane. (35) Krueger, S.; Meuse, C. W.; Majkrzak, C. F.; Dura, J. A.; Berk, N. F.; Tarek, M.; Plant, A. L. Langmuir 2001, 17 (2), 511-521.
Characterization of Pore Formation by Streptolysin O
Langmuir, Vol. 23, No. 3, 2007 1407
Figure 4. SPR angular reflectivity spectra for different stages of the membrane formation and incubation with 0.1 HU/µL SLO. Inset shows the spectra for 1 HU/µL SLO interacting with a separate pristine membrane.
Figure 6. (a) Frequency dispersive complex impedance for the lipid membrane, as deposited from vesicles incubated with different amounts of SLO toxin. (b) Corresponding phase change for the processes shown in panel a. Figure 5. SPR sensorgrams for non-specific protein adsorption on the different types of membranes studied.
Interestingly, after toxin incubation, the coupling angle shifts by 0.15 degree to lower angles, indicating a substantial loss of surface-bound material. A possible reason for this is the expulsion of PC molecules during either the pore-formation process or the membrane disruption by the toxin. This observation appears to underscore the difference between the HT monolayer and the long-chain alkylthiol monolayer and SLO versus R-hemolysin.36 The strongly concentration-dependent behavior suggests that lipid removal may prevail at higher toxin concentrations. However, the detailed mechanism of the SLO interaction with these supported membranes remains to be understood. To investigate the effect of non-specific protein adsorption onto the membrane and determine if potentially non-specific binding contributes to the observed signal for the SLO perforation process with the membrane, control experiments were conducted with very high concentrations of BSA. For the bare HT monolayer, we observed a very strong binding of BSA (Figure 5). Even after extensive washing, about 75% of the material remained bound to the monolayer. However, with vesicles fused onto the HT monolayer, the non-specific binding was reduced to below detection. We compared the influence of 0.1 and 1 mg/mL BSA on the membranes. Except for a little bump, possibly due to the
change of background refractive index, the end results are essentially the same, showing that no noticeable permanent adsorption occurs on these surfaces. The observed strong protein adsorption resistance agrees well with that determined for lipid layers on silicate surfaces37 and further confirms that the membranes are free of defects and are continuous. We also tested the membranes from vesicles perforated by SLO, which also exhibited no noticeable adsorption. A slight downward drift in the signal is observed in the case of the fused and perforated membranes. This possibly indicates that the SLO insertion into an existing membrane is not identical to that into intact vesicles and that the insertion does disrupt the membrane structure substantially. It should be pointed out that the BSA concentrations used here are more than a thousand times higher than that of the SLO. From these results, we conclude that the lipid membranes
(36) Glazier, S. A.; Vanderah, D. J.; Plant, A. L.; Bayley, H.; Valincius, G.; Kasianowicz, J. J. Langmuir 2000, 16 (26), 10428-10435.
(37) Glasmastar, K.; Larsson, C.; Hook, F.; Kasemo, B. J. Colloid Interface Sci. 2002, 246 (1), 40-47.
Figure 7. Equivalent circuit model used for the impedance analysis. Rs is the solution resistance, RCT the Faradic charge-transfer resistance, and CPE the constant phase element, which represents the capacitive element of the membrane.
1408 Langmuir, Vol. 23, No. 3, 2007
Wilkop et al.
Table 1. Results of the Faradaic Impedance Analysis for Equivalent Circuit Elements with Residual Errors toxin concn (HU\µL)
Rs (Ω)
Q (F) CPE
n CPE
RCT (Ω)
residual error (%) Rs
residual error (%) Q
residual error (%) n
residual error (%) RCT
0 1 2 3 4 6 8
687 806 866 1194 951 1196 644
6.88 × 10-8 9.80 × 10-8 9.70 × 10-8 7.54 × 10-8 9.48 × 10-8 2.20 × 10-7 2.30 × 10-6
0.920 0.907 0.878 0.874 0.880 0.780 0.550
323 640 161 230 140 680 71 787 51 426 62 758 29 947
2 2 2.8 3.2 3.5 5.5 12.5
3.9 3.7 6.7 9.7 11.3 15.7 20.6
0.5 0.8 0.96 1.33 1.53 2.5 4.87
1.5 1.6 1.9 2 2.2 3.5 5.8
share the property of effectively suppressing protein fouling and that the observed signal for SLO binding is indeed a specific response. Impedance Analysis. SPR measurements are highly effective in probing binding properties on the membrane surface and determining the membrane thickness. However, the permeability property, which is a macroscopic property associated with the pore-forming process, can be more effectively assessed with Faradaic impedance spectroscopy. The inherent advantage of using Faradaic over non-Faradaic impedance spectroscopy is the vastly increased signal in the membrane resistivity. The SLO toxin interaction with the vesicular membrane was studied by using SLO incubation in bulk followed by membrane fusion on the HT monolayer. In this method, the solution was incubated for 30 min at room temperature and then transferred onto the HT SAM surface in the electrochemical cell. After another 30 min incubation, the cell was rinsed twice with deionized water and Tris buffer before the 1 mM FCA solution was introduced for impedance characterization. The electrochemical cell was never allowed to become dry during the rinsing or measurement. The complex impedance spectra obtained from perforated and non-perforated lipid membranes, resolved into their impedance magnitude Z and phase shift φ, are presented in Figure 6a,b. The impedance magnitude strongly disperses at low frequencies, consistently dropping in magnitude with increasing toxin concentration. At high frequencies, the capacitive element of the membrane leads to a steep decrease in Z, converging more closely. The phase change shows a stronger concentration dependence. At low toxin concentrations (0 and 1 HU/µL), the phase shift levels off at the lower end of the spectra. For higher concentrations, the phase shift takes on a maximum, with the position shifting to higher frequencies. This strongly non-linear behavior deviates from that observed for ideal capacitive/resistive structures. To analyze the impedance more quantitatively, spectra were fitted to an equivalent circuit with the physical membrane properties assigned to the circuit elements. The electrode-lipid-solution structure can be represented by the equivalent circuit as shown in Figure 7. CPE represents a constant phase element, Rs the solution resistance, and RCT the charge-transfer resistance. The constant phase element was chosen to represent the capacitive nature of the lipid membrane, and its impedance is given by eq 2
ZCPE )
1 (2πfQ)n
(2)
where f is the frequency and Q is a measure for the capacitance. For n values close to 1.0, the CPE resembles a capacitor with a phase change of less than 90°. We selected a CPE since it describes real world systems better than a purely capacitive structure. It takes into account the frequency dispersion of the membrane properties caused by surface roughness and inhomogeneities of the membrane. As compared to a purely capacitive structure, the use of a CPE improved the quality of fit throughout the analysis. At very high toxin concentrations, substantial
residuals errors were observed. Therefore, the fitting analysis is limited to a dynamic range between 0 and 8 HU/µL. The CPE representing the non-perforated membrane has a Q value of 6.88 × 10-8 F with n ) 0.92. Approximating the Q of the CPE with the pure capacitance of a parallel plate capacitor (n ) 1),38 the total membrane thickness can be determined. On gold surfaces, lipid bilayers exhibit typical membrane capacitances of 0.5-1 µF/cm;2,39,40 our observed value of 1.4 µF/cm2 is a further indication for the existence of a hybrid bilayer, consisting of just one lipid leaflet and an HT membrane with water inclusion. The parallel plate capacitance is given by eq 3
C)
A0 d
(3)
where is the effective dielectric constant of the membrane structure, 0 the absolute permittivity, d the distance between the plates, and A the planar plate area. The dielectric constant of lipid membranes varies widely in the literature from 2.1,41 2.3,42 to 2.7,15 depending on composition, preparation, supporting monolayer, and inclusion of solvent molecules. The intricacies of dielectric permittivity profiles in a non-uniform system are well-recognized,43 and it was further noted that complicated underlying monolayers yield a more complicated impedance response,44 in particular, with an ionic reservoir existing between the gold surface and the membrane. For our given system, the membrane structure is composed of a monolayer of HT with water trapped between the alkane chains and a lipid layer on top of it. The loose packing and the resulting water inclusion is further indicated by the finite charge-transfer resistance. Water inclusion also generates a high uncertainty in the value of . Even a small degree of water inclusion significantly alters the effective of the film. Assuming a homogeneous, dense film with values from 2.1 to 2.7, a range of thickness between 1.35 and 1.73 nm is calculated. These are clearly only apparent thicknesses, since the HT layer thickness, as determined by SPR, is already 1.3 nm. Apparently, the use of these low dielectric constants for the description of the total membrane structure is inappropriate. When combining the monolayer thickness of the HT and the lipid membrane, a combined dielectric constant of 7.5 is obtained. This high value reflects the water inclusion in the HT membrane, and using a mixed model system, the relative water content can be determined as ca. 7%. Water inclusion has a more pronounced influence on electrochemical measurements as compared to optical measurements. The dielectric contrast between water and alkanes is small in the (38) Hsu, C. H.; Mansfeld, F. Corrosion 2001, 57 (9), 747-748. (39) Steinem, C.; Janshoff, A.; Ulrich, W. P.; Sieber, M.; Galla, H. J. Biochim. Biophys. Acta 1996, 1279 (2), 169-180. (40) Wiegand, G.; Arribas-Layton, N.; Hillebrandt, H.; Sackmann, E.; Wagner, P. J. Phys. Chem. B 2002, 106 (16), 4245-4254. (41) Plant, A. L. Langmuir 1993, 9 (11), 2764-2767. (42) Lingler, S.; Rubinstein, I.; Knoll, W.; Offenhaeusser, A. Langmuir 1997, 13 (26), 7085-7091. (43) Stern, H. A.; Feller, S. E. J. Chem. Phys. 2003, 118 (7), 3401-3412. (44) Raguse, B.; Braach-Maksvytis, V.; Cornell, B. A.; King, L. G.; Osman, P. D. J.; Pace, R. J.; Wieczorek, L. Langmuir 1998, 14 (3), 648-659.
Characterization of Pore Formation by Streptolysin O
Langmuir, Vol. 23, No. 3, 2007 1409
be best described by a first-order exponential decay given by eq 4.
RCT ) RCT0 + Ae-Bc
Figure 8. Charge-transfer resistance for the lipid membrane as a function of SLO toxin concentration.
optical regime (ca.1.77 vs 2.1), whereas at frequencies that allow for contributions from molecular orientations, it is very high (80 vs 2.1-2.7). Table 1 summarizes the results obtained from the impedance analysis. Concurrent with the decrease in the charge-transfer resistance, there is an increase in the Q value of the CPE and a consistent decrease in n. The use of a CPE, instead of a shear capacitor, is meant to account for contributions from surface roughness. For a rough surface, the fractal dimension (D) is between 2 and 3. This means that the surface fills between 2 (perfectly flat) and 3 (rough and branching) dimensions. Mulder proposed45 that for rough electrodes, the interfacial impedance is modified by an exponent, n ) 1/(D - 1). For a smooth surface, the fractal dimension of D ) 2 results in n ) 1. But for a highly porous/rough surface with D ) 3, one obtains n ) 0.5. Hence, an increasing degree of membrane perforation, which yields a highly irregular membrane structure, is expected to exhibit a marked decrease in n. This is clearly observed for n (CPE) in Table 1. It is further expected that Q increases for higher degrees of perforation, mainly due to two effects: (a) the decrease of the effective membrane thickness and (b) the inclusion of larger amounts of water into the membrane. This behavior is also observed in the data shown in the table. The observed change in the membranes’ non-Faradaic properties is very complex. The above explanation is based primarily on geometrical considerations. It is well-documented that the insertion of different species (i.e., a change in the membrane composition) results in changes in its impedance.46-48 However, no detailed universally applicable models exist that can account for the combined effect of perforation, insertion, and structural changes. The charge-transfer resistance is an independent measure of the membrane permeability as it stems solely from the Faradic current caused by the redox-active probe. Figure 8 shows the change in membrane resistance as a function of toxin concentration. Clearly, the charge-transfer resistance shows a strong, nonlinear dependence on the toxin concentration. The response can (45) Mulder, W. H.; Sluyters, J. H.; Pajkossy, T.; Nyikos, L. J. Electroanal. Chem. 1990, 285 (1-2), 103-115. (46) Perez, E. J. W. Eur. Biophys. J. 1988, 16, 23-29. (47) Schiller, S. M.; Naumann, R.; Lovejoy, K.; Kunz, H.; Knoll, W. Angew. Chem., Intl. Ed. 2003, 42 (2), 208-211. (48) Ashcroft, R. G.; Coster, H. G. L.; Laver, D. R.; Smith, J. R. Biochim. Biophys. Acta 1983, 730 (2), 231-238. (49) Deflorian, F.; Fedrizzi, L.; Rossi, S.; Bonora, P. L. J. Appl. Electrochem. 2002, 32 (8), 921-927.
(4)
RCT is the charge-transfer resistance, RCT0, A, and B are function specific constants, and c is the toxin concentration. For our system, the constants were determined as RCT0 ) 41.58 kΩ, A ) 277 260, and B ) 1.47, with c in HU/µL. Exponential decays in the chargetransfer resistance are frequently observed in the case of permeated or failing coatings and for increasing permeating pore sizes.49 For a perforating toxin system, this behavior is not completely understood. We speculate that the available cholesterol sites to which the SLO binds, prior to insertion and formation of pores, undergoes an exponential decay during pore formation, similar to the decay of available adsorption sites in a Langmuir adsorption isotherm.
Conclusion In this paper, we report the study of the pore-formation process by SLO toxin on an egg PC membrane surface by SPR and impedance spectroscopy. A short chain, hydrophobic disulfide derivative was synthesized to form a loosely packed sublayer that allows permeation of water molecules. This sublayer enables facile vesicle fusion and generates lipid films suitable for electrochemical impedance studies. The formation of the HT monolayer was monitored and characterized by SPR, and an equivalent layer thickness of 1.3 nm was obtained. Impedance analysis revealed that a certain degree of water inclusion had occurred in this film. No substantial difference in the fusion kinetics of pristine vesicles and those incubated with SLO was observed, indicating that perforation by SLO toxin does not disrupt the structure and that the vesicles remain intact. The average membrane thickness on the HT monolayer was determined to be 3.4 and 3.2 nm for pristine and perforated membranes. The interaction of the fused lipid layer on HT with the SLO toxin was studied for toxin concentrations of 0.1 and 1 HU/µL. The process is complex, yielding results that are highly concentrationdependent. Both types of membranes show a strong ability to resist non-specific protein adsorption. Faradaic impedance spectroscopy was carried out to investigate the permeability properties of the membrane upon toxin perforation. A CPE was used to represent the capacitive properties of the membrane and showed a consistent decrease in n with a concurrent increase in the Q value for increasing toxin concentrations. This clearly demonstrates changes in the 3-D structure of the membrane due to pore formation. The charge-transfer resistance, which is independent of the capacitive properties of the membrane, showed a first-order exponential decrease with respect to the toxin concentration. We put forward the hypothesis that the number of cholesterol molecules undergoes an exponential decay with pore formation, similar to the decay of available adsorption sites in a Langmuir adsorption isotherm. The work presented here highlights the unique property of the HT surface, extending the choice of supporting layers for generating quality lipid membranes and allowing Faradaic impedance spectroscopy to be performed for probing permeability and transport phenomena. LA0625502
