Langmuir 2006, 22, 5563-5565
5563
Formation of a Ripple Phase in Nanotubular Dimyristoylphosphatidylcholine Bilayers Confined Inside Nanoporous Aluminum Oxide Substrates Observed by DSC Ali M. Alaouie and Alex I. Smirnov* Department of Chemistry, North Carolina State UniVersity, 2620 Yarbrough DriVe, Box 8204, Raleigh, North Carolina 27695-8204 ReceiVed February 15, 2006. In Final Form: March 30, 2006 Phase properties of substrate-supported nanotubular dimyristoylphosphatidylcholine (DMPC) bilayers confined within nanoporous channels of anodic aluminum oxide were characterized by DSC and compared with unsupported vesicles. In addition to the main phase transition, all samples exhibited a pretransition with a characteristic midpoint hysteresis between heating and cooling scans. The pretransition indicates that nanotubular bilayers could exist in a ripple phase, whereas hysteresis points to a similarity in the phase transition mechanisms. Observance of the ripple phase in lipid nanotubes is an indication of fully hydrated and only slightly perturbed bilayer surface.
Substrate-supported phospholipid bilayers are known to mimic many properties of biological membranes1,2 and could be utilized to stabilize functional membrane proteins on solid surfaces. Recently, we have described a new type of substrate-supported nanotubular bilayers that are formed by self-assembling phospholipids inside nanoporous anodic aluminum oxide (AAO) substrates.3 This type of substrate-supported bilayer offers several advantages. One important feature is that the bilayers are packed inside a rigid structure that protects them from accidental mechanical disruption and/or surface contamination. Furthermore, the bilayer surface area is by a factor of 500-2000 greater than that of the planar bilayers for the same size of a planar chip. The increased quantity of deposited phospholipids enables studies of these nanopore-confined bilayers with DSC and other biophysical methods. Unsupported phospholipid bilayers are known to exist in several equilibrium phases that differ in lipid packing and bilayer properties: a gel phase (L′β), a ripple phase (P′β), and a fluid phase (LR). The L′β to P′β transition characterized by an equilibrium midpoint Tp is associated with structural transformation from a one-dimensional lamellar to a two-dimensional monoclinic lattice.4,5 It has been found that for zwitterionic dimyristoylphosphatidylcholine (DMPC) this phase transformation is extremely sensitive to experimental conditions. For example, although L′β to P′β transition for DMPC was observed in the course of DSC6 and EPR7 studies, Ono et al. did not observe a pretransition in some liposome preparations.8 Typically, an absence of pretransition indicates that at temperatures below Tm the lipid bilayer is entirely in the L′β phase. Several causes could contribute to the disappearance of the bilayer ripple phase.9-12 Detailed theoretical modeling of the
DMPC phase diagram based on continuum Landau theory indicated that the ripple phase would disappear at a relative humidity below 93%.13 This is in agreement with experimental studies demonstrating destabilization of the bilayer ripple phase and its consequent disappearance upon dehydration when water was replaced by DMSO.14,15 Clearly, the existence of the ripple phase greatly depends on the hydration level and therefore could be used as a very sensitive indicator of hydration and, perhaps, other surface perturbation effects that could play an important role in altering lipid packing and other properties of substratesupported bilayers. For example, the formation of the ripple phase for DMPC bilayers supported on silica is known to be prevented by lateral stress on the adsorbed bilayer.16,17 Stabilization of lipid bilayers on solid surfaces is thought to be governed by an interplay of electrostatic attraction forces and the chemical potentials for hydration of both surfaces, although long-range van der Waals forces may also play a significant role.18 If the hydration chemical potential of the lipids and/or transient hydrogen bonds formed between the lipid polar head and water is affected by the substrate, one would expect to observe changes in the DMPC phase diagram especially in the region of the ripple phase, which is very sensitive to hydration effects. Here we compare DSC data for two types of nanotubular bilayer packing inside nanoporous AAO substrates with those of unsupported DMPC bilayers and show that the phase properties are very similar. AAO membranes were purchased from Whatman, Ltd. (Middlesex, U.K.) and were specified to be 60 µm thick and have 20 and 200 nm pore diameter cutoffs. Examination of the first AAO substrate with a JEOL 6400F (JEOL, Tokyo, Japan) field emission scanning electron microscope revealed that the
* To whom correspondence should be addressed. Phone: 1-919-5134377. Fax: 1-919-513-7353. E-mail:
[email protected].
(9) Heerklotz, H.; Seekig, J. Biophys. J. 2002, 82, 1445. (10) Malcolmson, R. J.; Higinbotham, J.; Beswick, P. H.; Privat, P. O.; Saunier, L. J. Membr. Sci. 1997, 123, 243. (11) Prenner, E. J.; Lewis, R. N. A. H.; Kondejewski, L. H.; Hodges, R. S.; McElhaney, R. N. Biochim. Biophys. Acta 1999, 1417, 211. (12) Lambros, M. P.; Rahman, Y. E. Chem. Phys. Lipids 2004, 131, 63. (13) Chen, C.-M.; Lubensky, T. C.; MacKintosh, F. C. Phys. ReV. Lett. 1995, 51, 504. (14) Tristran-Nagle, S.; Moore, T.; Petrache, H. I.; Nagle, J. F. Biochim. Biophys. Acta 1998, 1369, 19. (15) Yu, Z.; Quinn, P. J. Biophys. J. 1995, 69, 1456. (16) Johnson, S. J.; Bayerl, T. M.; McDermott, D. C.; Adam, G. W.; Rennie, A. R.; Thomas, R. K.; Sackmann, E. Biophys. J. 1991, 59, 289. (17) Naumann, C.; Brumm, T.; Bayerl, T. M. Biophys. J. 1992, 63, 1314. (18) Tamm, L. K.; McConnell, H. M. Biophys. J. 1985, 47, 105.
(1) Brian, A. A.; McConnell, H. M. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 6159. (2) McConnell, H. M.; Tamm, L. K.; Weis, R. M. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 3249. (3) Smirnov, A. I.; Poluektov, O. G. J. Am. Chem. Soc. 2003, 125, 8434. (4) Janiak, M. J.; Small, D. M.; Shipley G. G. Biochemistry 1976, 15, 4575. (5) Sun, W.-J.; Tristram-Nagle, S.; Suter, R. M.; Nagle, J. F. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 7008. (6) Mabrey, S.; Sturtevant, J. M. Proc. Natl. Acad. Sci. U.S.A. 1976, 73, 3862. (7) Hoffmann, P.; Sandhoff, K.; Marsh, D. Biochim. Biophys. Acta 2000, 1468, 359. (8) Ono, A.; Takeuchi, K.; Sukenari, A.; Suzuki, T.; Adachi, I.; Ueno, M. Biol. Pharm. Bull. 2002, 25, 97.
10.1021/la060448z CCC: $33.50 © 2006 American Chemical Society Published on Web 05/23/2006
5564 Langmuir, Vol. 22, No. 13, 2006
average pore diameter was 177 ( 20 nm for one side and 29 ( 7 nm for a thin, 1-3 µm, filtration layer on the opposite side (data not shown). Because of the small thickness of the filtration layer, it is expected that only a negligible amount of lipid will be confined there. The second substrate was found to have pores of 208 ( 35 nm in diameter. Lipids were deposited into AAO substrates at T ≈ 35 °C > Tm ≈ 23.5 °C using two methods. In the manual deposition method, one side of the AAO disk was exposed to a multilamellar (MLV) aqueous dispersion of DMPC (Avanti Polar Lipids, Alabaster, AL) prepared as described previously.3 Such an exposure resulted in an immediate wetting of the nanoporous AAO disk, which became semitransparent. In the second method, an AAO disk was sandwiched between nylon filter supporting meshes, soaked in a buffer, and placed inside an ER-3 liposome extruder (Eastern Scientific LLC, Rockville, MD) that was attached to a 0.5 mL Hamilton syringe (Hamilton Company, Reno, NV) on each end. The aqueous MLV DMPC dispersion was loaded into one of the syringes and pushed back and forth between the syringes 15 times. After either deposition procedure, the exterior surface of AAO disks was cleaned with Kimwipes EX-L. The amount of deposited lipids was assessed by elemental chemical analysis. Assuming that lipid packing in AAO-confined bilayers is similar to that of unsupported bilayers, we have estimated that the first deposition method resulted in up to four nanotubular bilayers nestled one inside another to be formed per nanopore throughout the entire available surface area of the nanoporous substrate, while in the second method all internal lipid tubes were flushed out except the surface bilayer thus yielding a single nanotubular bilayer per nanopore. DSC measurements were performed with a Q100 differential scanning calorimeter using high volume steel pans (both from TA Instruments, New Castle, DE). Each sample contained excess buffer and approximately 1 mg of DMPC. Heating and cooling curves were obtained at a scanning rate of 0.5 °C/min in the following sequence: heating from 0 to 30 °C (scan 1), cooling from 30 to 0 °C (scan 1), heating from 0 to 30 °C (scan 2), cooling from 30 to 0 °C (scan 2), heating scan from 0 to 20 °C (i.e., below Tm), and then cooling from 20 to 0 °C (scan 3). Representative successive DSC heating and cooling scans for unsupported (control) DMPC MLV and nanoconfined-supported bilayers are shown in Figure 1. All DSC curves were corrected for baseline and normalized by the amplitude of the main phase transition during the first heating scan. The heating curves for the control sample show two phase transitions: a pretransition at Tp ) 10.57 ( 0.05 °C (scan 1) and 10.72 ( 0.05 °C (scan 2) and a main transition at Tm ) 23.24 ( 0.05 °C (scan 1) and 23.22 ( 0.05 °C (scan 2). These temperatures agree well with literature data.6,7 Likewise, the two transitions were also observed during cooling scans; however, although Tm remained unchanged, Tp decreased by approximately 3 °C when compared to the heating scans (Table 1). Similar hysteresis behavior for Tp was previously observed in calorimetry19 and scanning fluorescence depolarization20 experiments and is indicative of different transition mechanisms. Cho and co-workers suggested that during heating scans the L′β to P′β phase transformation proceeds directly according to an activated two-state model, whereas the reverse transition is more complex and is likely to occur via metastable states.19 Furthermore, in a temperature-jump small-angle X-ray diffraction study of DMPC MLV, Akiyama and co-workers observed that the L′β to P′β (19) Cho, K. C.; Choy, C. L.; Young, K. Biochim. Biophys. Acta 1981, 663, 14. (20) Lentz, B. R.; Freire, E.; Biltonen, R. L. Biochemistry 1978, 17, 4475.
Letters
Figure 1. Experimental DSC curves for DMPC bilayers for control multilamellar (MLV) sample (A) and lipid nanotubes confined inside 177 ( 20 nm (B) and 208 ( 35 nm (C) anodic aluminum oxides pores. The sample in (B) contained on average a single nanotubular bilayer per pore, whereas the sample in (C) contained up to four such bilayers nestled one inside another. Heating and cooling curves were obtained at 0.5 °C/min scanning rate in the following sequence: heating from 0 to 30 °C (a), cooling from 30 to 0 °C (c), heating from 0 to 30 °C (b), cooling from 30 to 0 °C (d), followed by a heating scan from 0 to 20 °C (i.e., below Tm) and then cooling from 20 to 0 °C (e). All curves are baseline corrected and normalized by the amplitude of the first heating scan. The pretransitions are 5-fold magnified and shown in bold. Table 1. Midpoint Phase Transition Temperatures of AAO-Supported and Unsupported DMPC Bilayers AAO, 177 ( 20 nm pores, single DMPC bilayer/pore
AAO, 208 ( 35 nm pores, ca. four DMPC bilayers/pore
Gel L′β to Ripple Phase P′β (Pretransition), Tp ) °C 10.57 ( 0.05 14.80 ( 0.05 7.70 ( 0.05 9.82 ( 0.05 10.72 ( 0.05 14.71 ( 0.05 7.67 ( 0.05 9.57 ( 0.05 7.74 ( 0.05 9.80 ( 0.05
14.89 ( 0.05 10.50 ( 0.05 14.60 ( 0.05 10.20 ( 0.05 10.50 ( 0.05
DMPC MLV (control)
heating scan 1 cooling scan 1 heating scan 2 cooling scan 2 cooling below Tm
heating scan 1 cooling scan 1 heating scan 2 cooling scan 2
Ripple P′β to Fluid Bilayer Phase LR (Main Phase Transition), Tm, °C 23.24 ( 0.05 24.16 ( 0.05 23.29 ( 0.05 23.79 ( 0.05 23.22 ( 0.05 24.06 ( 0.05 23.43 ( 0.05 23.71 ( 0.05
23.90 ( 0.05 23.79 ( 0.05 24.06 ( 0.05 23.90 ( 0.05
transformation was complete in less than 5 min, whereas for the opposite transformation, traces of the P′β phase were still present after 17 min.21 Thus, it appears that our observations of heating/ cooling hysteresis in Tp for DMPC MLV is an intrinsic property of lipid bilayers and is indicative of different transformation pathways. DSC heating and cooling scans for unsupported DMPC bilayers were compared with those for nanopore-confined bilayers of different preparations (Figure 1 and Table 1). For all AAOsupported samples, essentially no hysteresis of the main phase transition was observed regardless of the pore diameter or the sample preparation method (Table 1). Although the midpoint of the main phase transition in AAO-supported samples was higher by 0.2-0.8 °C vs the control sample, this shift is likely to be associated with rate-dependent distortions of the DSC curves and slow kinetics of the lipid bilayer phase transition in the nanopores.22 (21) Akiyama, M.; Terayama, Y.; Matsushima, N. Biochim. Biophys. Acta 1982, 687, 337. (22) Alaouie, A. M.; Smirnov, A. I. Biophys. J. 2005, 88, L11.
Letters
Figure 2. Experimental X-band (9.0 GHz) rigid-limit (T ) -120 °C) EPR spectra of DMPC bilayers labeled with 1 mol % of 2,2,6,6teramethyl-piperidin-1-4-yl octadecanoate: (A) lipid bilayers confined inside ca. 208 nm anodic aluminum oxide pores and immersed in water; (B) control multilamellar DMPC sample (20% lipid by weight); (C) air-dried sample (A).
All substrate-supported samples exhibited a pretransition with midpoints ranging from Tp ) 9.57 ( 0.05 to 14.89 ( 0.05 °C depending on sample preparation and direction of the DSC scan (Table 1). It should be noted that Tps for all of the AAO-supported samples were approximately 4 °C higher than the control MLV sample during heating and 2-3 °C higher during cooling scans (Table 1). The pretransition peak in the cooling scans for both control and substrate-supported samples was much smaller than in the heating scans. A similar decrease in the pretransition peak has been previously observed for unsupported bilayers.23 From these DSC data, we can conclude that confinement of DMPC in ca. 177 or 208 nm pores does not perturb the main phase transition temperature nor abolish the ripple phase. However, it appears that confinement increases the magnitude of hysteresis effects in DMPC. The observance of the ripple phase in lipid nanotubes is an indication of a fully hydrated and/or only slightly perturbed bilayer (23) Tsong, T. Y.; Kanehisa, M. I. Biochemistry 1977, 16, 2674.
Langmuir, Vol. 22, No. 13, 2006 5565
surface. To further investigate bilayer hydration effects, we employed spin-labeling EPR method with 2,2,6,6-teramethylpiperidin-1-4-yl octadecanoate as a hydration probe. In lipid bilayers, this probe positions its nitroxide moiety in the lipid hydration layer. The nitrogen hyperfine splitting of this spin probe (2Az for rigid limit EPR spectra) serves as an indicator of water concentration.24 Figure 2 illustrates that, although 2Az ) 70.9 ( 0.2 G for spin-labeled DMPC in hydrated AAO is essentially identical to that of multilamellar vesicles (70.6 ( 0.1 G), an air-dried AAO-DMPC sample exhibits a significantly smaller 2Az ) 67.9 ( 0.2 G. The experimental procedure and further details are provided in the Supporting Information. In summary, we provide calorimetric evidence that substratesupported nanotubular bilayers exhibit a pretransition suggesting formation of a ripple phase. To the best of our knowledge, this is the first observation of the ripple phase in substrate-supported bilayers by DSC. Second, even though Tp in AAO-supported bilayers was shifted to higher values and the magnitude of the hysteresis between heating and cooling scans was somewhat larger than those observed for the control MLV sample, the hysteresis effect itself was preserved. This clearly indicates the similarity in phase transition mechanisms. Overall, we conclude that the phase properties of DMPC bilayers supported inside nanoporous substrates are similar to that of unsupported bilayers. Acknowledgment. This work was supported by DOE Contract DE-FG02-02ER15354 to A.I.S. Supporting Information Available: EPR experimental procedures and discussion of EPR data. This material is available free of charge via the Internet at http://pubs.acs.org. LA060448Z (24) Alves, M.; Peric, M. Biophys. Chem. 2006, 122, 66.
