LETTER pubs.acs.org/Langmuir
Anomalous Diffusion in Supported Lipid Bilayers Induced by Oxide Surface Nanostructures Ryugo Tero,*,†,|| Gen Sazaki,‡ Toru Ujihara,§ and Tsuneo Urisu† †
Division of Biomolecular Sensing, Institute for Molecular Science, Okazaki 444-8585, Japan The Institute of Low Temperature Science, Hokkaido University, Kita-ku, Sapporo 060-0819, Japan § Graduate School of Engineering, Nagoya University, Nagoya 464-8603, Japan ‡
bS Supporting Information ABSTRACT: Hierarchic structure and anomalous diffusion on submicrometer scale were introduced into an artificial cell membrane, and the spatiotemporal dependence of lipid diffusion was visualized on nanostructured oxide surfaces. We observed the lipid diffusion in supported lipid bilayers (SLBs) on step-and-terrace TiO2(100) and amorphous SiO2/Si surfaces by single molecule tracking (SMT) method. The SMT at the time resolution of 500 μs to 30 ms achieved observation of the lipid diffusion over the spatial and temporal ranges of 100 nm/millisecond to 1 μm/second. The temporal dependence of the diffusion coefficient in the SLB on TiO2(100) showed that the crossover from anomalous diffusion to random diffusion occurred around 10 ms. The surface fine architecture on substrates will be applicable to induce hierarchic structures on the order of 100 nm or less, which correspond to the microcompartment size in vivo.
L
ateral organization and diffusion of lipids and membrane proteins are crucial factors of biological reactions on cell membranes such as signal transduction and cell recognition.1 Various types and sizes of two-dimensional organizations, such as domains, clusters, and microcompartments, with scales from several nanometers to micrometers construct hierarchic structures in cell membranes.2 4 Lipids and membrane proteins undergo hop-diffusion because of microcompartments partitioned by transmembrane picket proteins, and thus, their apparent diffusion rates vary with spatiotemporal scales of observation.3,4 Such anomalous diffusion is a key mechanism to induce biological reactions locally and efficiently, and model systems mimicking the hierarchic structures on in vivo scales are valuable for understanding the size-dependent functions of lipid organizations2 4 and how the lateral molecular transportation at microscopic regions propagates to macroscale.5 Herein we introduced a hierarchic structure into a cell membrane model system using nanostructures on oxide surfaces as templates to realize a hierarchic structure with a comparable size in vivo. We visualized the anomalous diffusion behavior of lipids over the spatiotemporal scales over 100 nm/millisecond to micrometer / second by using single molecule tracking (SMT) method. Supported lipid bilayers (SLBs) have been studied as a cell membrane model and as a lipid membrane platform for membrane proteins.6 8 One of advantages of the SLBs is that functions of solid substrates can be applied to control membrane structures and properties. Several recent studies showed that lateral distribution of molecules or lateral organization in SLBs r 2011 American Chemical Society
on micrometer order can be controlled by applying fabrication and patterning techniques on solid surfaces.8 15 However, to introduce and observe submicrometer microcompartments comparable to those in vivo,3,4 surface fabrication techniques on submicrometer scale and observation methods with sufficient spatial and temporal resolutions are necessary. We prepared two oxide surfaces with different nanostructures (Figure 1) and observed the lipid diffusion in dioleoylphosphatidylcholine (DOPC)-SLB by SMT with the time resolution of 500 μs to 30 ms. A single crystal TiO2(100) surface consisting of single atomic steps and flat terraces (Figure 1A) was prepared by wet etching and oxygen annealing.16 Terraces with 550 nm width were divided by linear monatomic steps, and oval pits made of the atomic steps existed on the terraces. The height of the steps was 0.23 nm, which corresponded to the minimum step unit of TiO2(100).16 Thermally oxidized SiO2 on Si(100) wafer (Figure 1B) had random protrusion with the peak-to-valley roughness of ∼0.6 nm (Figure S1, Supporting Information). The DOPC-SLB containing a fluorescence-labeled lipid (lissamine rhodamine B-dipalmitoylphosphatidylethanolamine; Rb-DPPE) at the ratio of 4 10 9 6 10 8 was formed on the substrates by the vesicle fusion method,8,16 and the fluorescence image of each single Rb-DPPE molecule was obtained by a Received: April 21, 2011 Revised: June 29, 2011 Published: July 15, 2011 9662
dx.doi.org/10.1021/la201474h | Langmuir 2011, 27, 9662–9665
Langmuir
LETTER
Figure 1. AFM topographs of (A) step-and-terrace TiO2(100) and (B) amorphous SiO2/Si surfaces.
Figure 3. (A, B) MSD-τ plots of Rb-DPPE on (A) TiO2(100) (Δt = 996 μs, n = 110) and (B) SiO2/Si (Δt = 497 μs, n = 179). MSD of each Rb-DPPE is drawn as a light blue trace, and the averaged MSDs are in circles (n = 110 in A, and n = 179 in B). The dotted lines represent extrapolated MSD with the gradient of 4D(τ = 2Δt). (C, D) MSD-τ plots of Rb-DPPE on (C) TiO2(100) (Δt = 30 ms, n = 252) and (D) SiO2/Si (Δt = 30 ms, n = 209). The dotted lines in (A) and (B) are also extrapolated in (C) and (D), respectively.
Figure 2. (A) Schematic of the diagonal illumination for single molecule tracking of a SLB on an opaque substrate. (B, C) Fluorescence intensity distributions and (D, E) extracted 400 steps trajectories of RbDPPE in DOPC-SLBs on TiO2(100) (B, D) and SiO2/Si (C, E). Pixel sizes are (B) 194.64 nm and (C) 97.32 nm, and time resolutions are (D) 996 μs (1004 fps) and (E) 497 μs (2011 fps).
diagonal illumination setup (Figure 2A). In conventional SMT experiments, substrate materials are restricted to glass or quartz, because the excitation light is introduced from the substrate backside at the total internal reflection (TIR) condition and the fluorescence-tagged samples are illuminated by evanescent light.17 In the present study, we set the oxide substrates upside-down above a cover glass after the SLB formation and illuminated the sample surface diagonally by introducing the excitation light at the incident angle slightly lower than the TIR condition (Figure 2A).18 20 This diagonal illumination setup achieved the SMT measurement of SLB without the restriction on the substrate transparency and refractive index. Figure 2B and C shows the fluorescence intensity distributions of RbDPPE in DOPC-SLBs on the TiO2(100) and SiO2/Si surfaces, respectively. Sufficient signal-to-noise ratio was obtained for the high-speed SMT at the time resolution (Δt) of 996 μs (1004 fps) on TiO2(100), and at Δt = 497 μs (2011 fps) on SiO2/Si. The location accuracies21 were 37 nm on TiO2(100) at 1004 fps, and 21 nm on SiO2/Si at 2011 fps, which were sufficient to track the lipid molecules at the region of 100 nm. Extracted 400 steps of the Rb-DPPE trajectories on the TiO2(100) and SiO2/Si are shown in Figure 2D and E. The frame rate of 1000 fps was
Figure 4. (A, B) Dependence of D(τ) and mean diffusion distance √ ( MSD) on time interval τ calculated from the MSDs obtained on (A) TiO2(100) at 1004 fps (circles), 501 fps (squares), and 33 fps (crosses), and on (B) SiO2/Si at 2011 fps (circles), 1001 fps (squares), and 67 fps (crosses).
fast enough to follow the lipid diffusion in 100 nm order regions. Mean square displacements (MSDs) of the single molecule trajectories were calculated and plotted against time interval (τ). Figure 3A and B shows the MSD-τ plots of the Rb-DPPE diffusion in the DOPC-SLBs on the step-and-terrace TiO2(100) recorded at 1004 fps and on the SiO2/Si at 2011 fps, respectively. The dependence of the MSD on the time interval is expressed as Ær(τ)2æ = 4D0τR = 4D(τ)τ, where R (e1) is an anomalous diffusion exponent and D(τ) = D0τR 1 is time-dependent diffusion coefficient.22,23 The MSD-τ plot is linear and D(τ) is independent of t for random diffusion (R = 1), whereas if the anomalous diffusion (R < 1) is caused by corrals or obstacles, D(τ) decreases with time depending on the corral size and its barrier height.22 Nonlinear MSD-τ plots in Figure 3A and B showed that anomalous diffusion occurred at the time interval below 20 ms. On the other hand, the lipid diffusions at the 9663
dx.doi.org/10.1021/la201474h |Langmuir 2011, 27, 9662–9665
Langmuir
Previous studies on lipid bilayer fluidity showed that SLBs have smaller diffusion coefficients than free-standing lipid bilayers such as giant liposomes,30,31 but it is not clear what kind of interaction between SLBs and substrates reduces the lateral lipid diffusion even through the 1 2 nm thick aqueous layer. The values of D(τ = 1 s) obtained from the linear fitting of MSD-τ plots in Figure 3C and D were 3.02 and 2.53 μm2/s on TiO2(100) and SiO2/Si, respectively, and these values corresponded to the previously reported diffusion coefficients in SLBs.30 32 The results in Figure 4 indicate that the distortion in SLBs caused the barrier against lateral lipid diffusion, and that the spatiotemporal tendency of the reduction in D was affected by the substrate nanoscale morphology. The lateral sizes of the surface structures in Figure 1 were similar to the compartment sizes in plasma membranes (30 250 nm).3,4 The inner two-dimensional structures in the SLB induced by the inorganic substrate structure were immobile, but the shape and size of the compartments in plasma membranes are also stable at least at the observation time of several seconds though they may change at the time scale of hours to days.3,4 Fine architecture of nanostructures on solid substrates will lead to the control of the lateral structures and molecular transportation on the orders over 10 nm to μm, and of their hierarchic structures. In conclusion, we introduced a hierarchic structure on submicrometer scale into SLBs using surface nanostructures on oxide surfaces, and observed the spatiotemporal dependence of lipid diffusion by SMT. The SMT with diagonal illumination setup at the time resolution of Δt = 497 μs to Δt = 30 ms showed the crossover from anomalous diffusion to random diffusion in the submicrometer region. These methods will be applicable to realize microcompartments and hierarchic structures of cell membranes in artificial bilayer membrane systems.
’ ASSOCIATED CONTENT
bS
Supporting Information. Experimental details, AFM topograph analysis, and movies of SMT. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Present Addresses
)
temporal scale of 100 ms to 1 s gave linear MSD-τ plots on both the TiO2(100) and SiO2 surfaces (Figure 3C and D), therefore following the random diffusion. Diffusion coefficients at the time interval of τ = nΔt were obtained from the MSD recorded at various time resolutions and were plotted √ against time (Figure 4). The mean diffusion distances ( MSD) were plotted against τ as well as D(τ) τ in Figure 4; therefore, Figure 4 shows the spatiotemporal dependence of the lipid diffusion in the DOPC-SLBs. Figure 4A reveals that D(τ) of Rb-DPPE on the step-and-terrace TiO2(100) decreased at√the time interval from 2 to 10 ms, which corresponded to MSD of 210 390 nm; thus, anomalous diffusion occurred at these temporal and spatial ranges. The value of R was 0.75 below 6 ms, while R = 1.0 at 100 ms to 1.0 s. This is a characteristic behavior of the hop-diffusion in compartmented lipid bilayers; lipids in a compartment diffuse randomly, but the diffusion at longer time and space decelerates if the compartment wall works as the diffusion barrier, and the diffusion again seems random at sufficiently larger regions than the compartment size.3,4,12,24 Obstacles23,25,26 and biding sites27 are other possibilities causing anomalous diffusion, but they were not the case in this study. The DOPC-SLB did not have domains or defects working as obstacles, and trapping of lipid was not observed in the trajectories analyzed in Figure 3 at least at the time resolution in this study. The result in Figure 4A clearly shows the crossover from the anomalous diffusion at the submicrometer region to the random diffusion at the micrometer region, and indicates the existence of ∼200 nm compartments in the DOPC-SLB on the step-and-terrace TiO2(100) surface. This compartment size matched to the average interstep distance of 248 nm on the TiO2(100), which was evaluated from AFM images (Supporting Information). Hence, we attributed the origin of the anomalous diffusion on the TiO2(100) surface to the single atomic steps on the surface. Recently we reported that PC-SLB on the step-andterrace TiO2(100) precisely follows the atomic step structure (0.23 nm in height) of the substrate, and the SLB has atomic size distortion above the substrate steps.16 The SLB above the terrace region is uniform not only in the AFM topograph but also in the phase-shift image, which is a viscoelasticity mapping,28 and any artificial domain or boundary is not found. Therefore, we concluded that the distortion of the SLB induced by the substrate step was the only dominant factor introducing a diffusion barrier into the single-component SLB. AFM topographs show the width of the distorted region, or maybe a faultlike structure, is ∼15 nm, which is close to the cantilever tip size. This SLB distortion was converged on the narrow area and worked as a barrier for the lipid diffusion, causing anomalous diffusion, even though the distortion is small compared with the hydrophobic core thickness (3.6 nm for DOPC).29 A sharp decrease in D(τ) was not detected on SiO2/Si (Figure 4B), and D(τ = 1 ms) = 3.24 μm2/s was already close to D(τ = 100 ms) = 2.88 μm2/s. If the origin of the decrease in D on the SiO2 was the surface structure of the substrate similar to TiO2(100), a sharp decrease√in D like in Figure 4A would occur at a smaller region than MSD = 100 nm, because the pitch of the surface protrusions in Figure 1B was ∼40 nm and thus could not be detected at the time resolution of the current study. At much higher time resolution, plateaus in the D(τ) τ plots will appear in the submillisecond region due to random lipid diffusion within the compartments. The detailed estimation of the compartment sizes in the SLBs on the TiO2(100) and SiO2 surfaces is currently ongoing.
LETTER
Electronics-Inspired Interdisciplinary Research Institute, Toyohashi University of Technology, Toyohashi 441-8580, Japan.
’ ACKNOWLEDGMENT This work is supported by KAKENHI (#21107530, #21015027, #21685004) from MEXT of Japan, and the collaborative research program of ILTS. ’ REFERENCES (1) Lingwood, D.; Simons, K. Science 2010, 327, 46–50. (2) Jacobson, K.; Mouritsen, O. G.; Anderson, R. G. W. Nat. Cell Biol. 2007, 9, 7–14. (3) Kusumi, A.; Nakada, C.; Ritchie, K.; Murase, K.; Suzuki, K.; Murakoshi, H.; Kasai, R. S.; Kondo, J.; Fujiwara, T. Annu. Rev. Biophys. Biomol. Struct. 2005, 34, 351–378. 9664
dx.doi.org/10.1021/la201474h |Langmuir 2011, 27, 9662–9665
Langmuir
LETTER
(4) Murase, K.; Fujiwara, T.; Umemura, Y.; Suzuki, K.; Iino, R.; Yamashita, H.; Saito, M.; Murakoshi, H.; Ritchie, K.; Kusumi, A. Biophys. J. 2004, 86, 4075–4093. (5) Masuda, A.; Ushida, K.; Okamoto, T. J. Photochem. Photobiol., A 2006, 183, 304–308. (6) Goksu, E. I.; Vanegas, J. M.; Blanchette, C. D.; Lin, W.-C.; Longo, M. L. Biochim. Biophys. Acta 2009, 1788, 254–266. (7) Giocondi, M.-C.; Yamamoto, D.; Lesniewska, E.; Milhiet, P.-E.; Ando, T.; Le Grimellec, C. Biochim. Biophys. Acta 2010, 1798, 703–718. (8) Castellana, E. T.; Cremer, P. S. Surf. Sci. Rep 2006, 61, 429–444. (9) Kam, L.; Boxer, S. G. Langmuir 2003, 19, 1624–1631. (10) Yoon, T.-Y.; Jeong, C.; Lee, S.-W.; Kim, J. H.; Choi, M. C.; Kim, S.-J.; Kim, M. W.; Lee, S.-D. Nat. Mater. 2006, 5, 281–285. (11) Parthasarathy, R.; Yu, C.-h.; Groves, J. T. Langmuir 2006, 22, 5095–5099. (12) Takimoto, B.; Nabika, H.; Murakoshi, K. J. Phys. Chem. C 2009, 113, 3127–3132. (13) Okazaki, T.; Tatsu, Y.; Morigaki, K. Langmuir 2010, 26, 4126–4129. (14) Rossetti, F. F.; Bally, M.; Michel, R.; Textor, M.; Reviakine, I. Langmuir 2005, 21, 6443–6450. (15) Tero, R.; Watanabe, H.; Urisu, T. Phys. Chem. Chem. Phys. 2006, 8, 3885–3894. (16) Tero, R.; Ujihara, T.; Urisu, T. Langmuir 2008, 24, 11567– 11576. (17) Axelrod, D. Traffic 2001, 2, 764–774. (18) Sazaki, G.; Okada, M.; Matsui, T.; Watanabe, T.; Higuchi, H.; Tsukamoto, K.; Nakajima, K. Cryst. Growth Des. 2008, 8, 2024–2031. (19) Hanasaki, I.; Takahashi, H.; Sazaki, G.; Nakajima, K.; Kawano, S. J. Phys. D 2008, 41, 095301. (20) Tokunaga, M.; Imamoto, N.; Sakata-Sogawa, K. Nat. Methods 2008, 5, 159–161. (21) Sbalzarini, I. F.; Koumoutsakos, P. J. Struct. Biol. 2005, 151, 182–195. (22) Saxton, M. J.; Jacobson, K. Annu. Rev. Biophys. Biomol. Struct. 1997, 26, 373–399. (23) Ratto, T. V.; Longo, M. L. Langmuir 2003, 19, 1788–1793. (24) Saxton, M. J. Biophys. J. 1995, 69, 389–398. (25) Saxton, M. J. Biophys. J. 1994, 66, 394–401. (26) Ehrig, J.; Petrov, E. P.; Schwille, P. Biophys. J. 2011, 100, 80–89. (27) Saxton, M. J. Biophys. J. 1996, 70, 1250–1262. (28) Radmacher, M.; Tilmann, R. W.; Gaub, H. E. Biophys. J. 1993, 64, 735–742. (29) Nagle, J. F.; Tristram-Nagle, S. Biochim. Biophys. Acta 2000, 1469, 159–195. (30) Przybylo, M.; Sykora, J.; Humpolickova, J.; Benda, A.; Zan, A.; Hof, M. Langmuir 2006, 22, 9096–9099. (31) Guo, L.; Har, J. Y.; Sankaran, J.; Hong, Y.; Kannan, B.; Wohland, T. ChemPhysChem 2008, 9, 721–728. (32) Goksu, E. I.; Hoopes, M. I.; Nellis, B. A.; Xing, C.; Faller, R.; Frank, C. W.; Risbud, S. H.; Satcher, J. H., Jr; Longo, M. L. Biochim. Biophys. Acta 2010, 1798, 719–729.
9665
dx.doi.org/10.1021/la201474h |Langmuir 2011, 27, 9662–9665
