pubs.acs.org/NanoLett
Sealing of Submicrometer Wells by a Shear-Driven Lipid Bilayer Peter Jo¨nsson,* Magnus P. Jonsson, and Fredrik Ho¨o¨k* Department of Applied Physics, Chalmers University of Technology, SE-41296 Gothenburg, Sweden ABSTRACT A supported lipid bilayer (SLB) was formed in a microfluidic channel by vesicle fusion. The SLB, formed on a flat part of the surface, was driven by the shear forces of a bulk flow above the SLB to a part of the surface with embedded submicrometer wells. When using a bulk solution with a pH of 9.5 the advancing lipid bilayer sealed the wells, creating free-spanning membranes, whereas at a pH of 8.0 the SLB instead followed the contour of the wells. KEYWORDS Spreading lipid bilayers, black lipid membranes, microfluidics, molecule encapsulation, nanostructures
S
upported lipid bilayers (SLBs) mimic many of the properties of living cell membranes but are more stable and can be studied using a large number of surface-based techniques. Thanks to these qualities, SLBs have been extensively used as biophysical model systems for natural cell membranes.1,2 However, the close proximity of the lipid bilayer to the support may prevent the incorporation and function of biomolecules, such as transmembrane proteins, in the lipid bilayer.3,4 The interaction between the supporting substrate and the lipid bilayer can also drastically reduce the lateral mobility of molecules solubilized in the SLB.5 To overcome these limitations, while maintaining the stable, planar configuration of the SLB, different groups started to make lipid bilayers that span submicrometer-wide apertures on an otherwise flat surface.6-9 In this way, the free-spanning parts of the lipid bilayer can incorporate biomolecules that will not suffer from the close proximity of the support. At the same time, the small size of the apertures affords increased stability of the lipid bilayer compared with traditionally made black lipid membranes, where the aperture is typically 0.1-1 mm wide.10 A common approach in forming free-spanning lipid bilayers over both small and large apertures has been to dissolve the lipids that will form the bilayer in an organic solvent and then spread the dissolved lipids over an aperture in a hydrophobic material. After thinning, a single lipid bilayer spanning the aperture is formed.11,12 However, solvent remaining in the lipid bilayer may change the activity and the properties of the lipid bilayer and of the biomolecules embedded in the bilayer.4,13 Hence, solvent-free methods of creating free-spanning lipid bilayers would be advantageous in many instances. In this work, we present a new method of making solventfree lipid bilayers that span and seal submicrometer wells
fabricated on a hydrophilic SiO2 surface. The principle behind the technique is to use the shear forces arising from a bulk flow above the SLB to drive the lipid bilayer over the SiO2 surface. By optimizing the properties of the driving bulk solution, such as the salt concentration and the pH, a continuous lipid bilayer that spans and seals the wells can be made. We recently found that by applying a bulk flow inside a microfluidic channel it is possible to create and control the transport of an SLB on the floor of the channel.14 The shear force exerted on the SLB by the bulk flow in the channel causes the motion of the lipid bilayer. Since the shear-driven advancement of the SLB is created by the bulk flow of liquid above the lipid bilayer this provides considerable freedom in the choice of salt concentration and pH value of the bulk solution driving the SLB in contrast to self-spreading bilayers that are dependent on the interaction between the support and the lipid bilayer to spread.15,16 This offers an advantage compared to traditional methods of forming free-spanning bilayers, such as adsorption and rupture of lipid vesicles over a porous substrate,7,13 which only work under conditions where the interaction between the lipid bilayer and the surface is high enough to cause the vesicles to be adsorbed and rupture on the surface. In fact, previous studies of vesicle adsorption and rupture17 and self-spreading18 on structured SiO2 surfaces have shown that phosphatidylcholine bilayers tend to follow the profile of the surface when the dimension of the structures in the spreading direction is of similar size (∼100 nm) to those used in this work. Another advantage of the technique presented here is that all the steps required to drive the lipid bilayer over a surface with fabricated wells are performed in a microfluidic setup. This makes the technique an interesting alternative to traditional techniques of forming freespanning lipid bilayers, such as the Montal-Mu¨ller method19 and the Langmuir-Blodgett method,9,20 which require an open system that is difficult to automate and miniaturize for practical biotechnical applications.
* To whom correspondence should be addressed. E-mail: (P.J.)
[email protected]; (F.H.)
[email protected]. Received for review: 03/04/2010 Published on Web: 04/20/2010 © 2010 American Chemical Society
1900
DOI: 10.1021/nl100779k | Nano Lett. 2010, 10, 1900–1906
by us14 and others.23 After rinsing excess vesicles from the channel and closing the valves on arms 2 and 3, the SLB was driven by the shear force of a bulk flow between 1 and 4 toward the region with the embedded wells. The bulk solution consisted of 5 mM NaCl and 2 mM tris[hydroxymethyl]aminomethane (TRIS) unless otherwise stated. To investigate the spreading behavior of the lipid bilayer over the well region, the fluorescence intensity, the drift velocity, and the diffusivity of the fluorescently labeled molecules in the lipid bilayer were measured. Experiments were also performed using the fluorescent molecule 5(6)carboxyfluorescein (CF) in the bulk solution to further demonstrate that under certain bulk conditions the advancing lipid bilayer spanned and sealed the wells. Results. When driving the SLB by the flow from the bulk solution at a pH of 8.0, the fluorescence intensity was increased by 84% when the lipid bilayer front reached the region of the surface with the fabricated wells (see Figure 2a, which shows the entire width of the channel). In contrast, when repeating the experiment with a bulk solution at a pH of 9.5, the fluorescence intensity was approximately the same on the part containing the wells as on the flat part of the surface (see Figure 2b, which shows the same region as in Figure 2a). The small imperfections in the lipid bilayer, over both the flat part of the surface and the well region, are surface defects that accumulate gradually each time the sample is cleaned and reused (see Supporting Information for details on the cleaning procedure). Note also that in Figure 2b the lipid bilayer did not advance over the well region close to the walls of the channel, where the driving force from the bulk flow goes to zero. A line profile of the fluorescence intensities in Figure 2a,b, taken from left to right in the center of the channel, is shown in Figure 2c. The data in Figure 2 were normalized to the fluorescence intensity on the flat part of the surface. The drift velocity of the lipid bilayer front was also investigated on both the flat part of the surface and in the region containing the wells. When the pH of the bulk solution was increased from 8.0 to 9.5 the drift velocity of the lipid bilayer on the flat part of the surface increased slightly from 0.19 ( 0.01 to 0.22 ( 0.01 µm/s with a bulk flow of 300 µL/min (see Figure 3). As described previously,24 the increased drift velocity indicates that either the coupling between the lipid bilayer and the support is slightly weakened, or that the frictional force between the two monolayers of the bilayer is slightly lower at the higher pH. When the lipid bilayer front reached and advanced into the region with embedded wells, the velocity of the moving bilayer front was observed to decrease. The reduction in velocity was around 40% when the lipid bilayer was driven by the bulk solution at a pH of 8.0. When the pH of the bulk solution was 9.5, only a 5% decrease in the drift velocity was seen in the well region, compared to that over the flat part of the surface (see Figure 3).
FIGURE 1. Schematic illustration of the substrate and the microfluidic setup used to drive the SLB toward the part of the surface containing embedded wells.
Experimental Procedure. The microfluidic setup illustrated in Figure 1 was used in the experiments to form and drive the lipid bilayer toward the area with embedded wells. A SiO2-coated glass slide constituted the floor of the microfluidic channel. Using colloidal lithography21 and subsequent etching,22 half of the SiO2 surface was modified so as to contain short-range ordered 80 nm wide and 123 nm deep wells, whereas the other half of the surface was flat. The walls and the ceiling of the microfluidic channel were made of polydimethylsiloxane (PDMS). The PDMS part of the channel was bonded to the glass slide, constituting the floor of the channel, after oxygen plasma treatment, such that the edge between the flat part of the surface and the wells was in arm 4 of the channel (see Figure 1). Details concerning the fabrication of the substrate with the wells and the microfluidic channel are presented in the Supporting Information. An SLB consisting of 1-palmitoyl-2-oleoyl-sn-glycero-3phosphocholine (POPC) with a fraction of fluorescently labeled lipids, either 1 wt % of 2-(12-(7-nitrobenz-2-oxa-1,3diazol-4-yl)amino)dodecanoyl-1-hexadecanoyl-sn-glycero-3phosphocholine (NBD C12-HPC) or 0.2 wt % Lissamine rhodamine B 1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (R-DHPE), was formed by injecting vesicles into arm 1 and a buffer solution into arm 4, while the valves on arms 2 and 3 were open. The laminar flow regime prevents the vesicle solution from mixing with the buffer solution from arm 4, thus restricting the vesicles to the left part of the channel, where an SLB is formed after the vesicles adsorb onto the surface and rupture, as shown previously © 2010 American Chemical Society
1901
DOI: 10.1021/nl100779k | Nano Lett. 2010, 10, 1900-–1906
FIGURE 2. The fluorescence intensity from the SLB when the lipid bilayer is driven from the flat part of the surface to the region with wells by a driving bulk solution with a pH of (a) 8.0 and (b) 9.5. (c) Line profile of the intensity in the middle of the channel in a and b, from left to right in the images.
FIGURE 3. Representative values of the position of the lipid bilayer front as a function of time (symbols), together with linear fits (solid lines), when driving the lipid bilayer with a bulk solution at a pH of 8.0 and 9.5 on the flat part of the surface and over the region with wells. The statistical spread in the data is given in Table 1.
FIGURE 4. Typical values of the fluorescence recovery (symbols) and exponential curve fits (solid lines; F(k,t) ) F(k,0)exp(-4π2k2Dt)) using the Hankel transform method to determine the diffusion coefficient of the labeled lipids on different parts of the surface.25 The statistical spread in the data is given in Table 1.
regions of the surface was 2.00 ( 0.05 µm2/s. This value is of the same magnitude as diffusivities measured previously in similar systems.14,25-27 The diffusivity measured in the well region after the lipid bilayer had been driven by the bulk solution at a pH of 9.5 was approximately the same as on the flat part of the surface (1% lower diffusivity over the region containing the wells). However, the diffusivity was 40% lower in the well region compared to that on the flat part of the surface when the lipid bilayer had been driven by a bulk solution at a pH of 8.0 (see Figure 4). The results obtained from the fluorescence intensity measurements, the drift velocity of the lipid bilayer and the diffusivity of the labeled lipids are summarized in Table 1 for a lipid bilayer labeled with 1 wt % NBD C12-HPC. Significant differences in these parameters can be observed between the flat part of the surface and the region with the wells when using a bulk solution with a pH of 8.0, whereas the influence of the wells is significantly reduced at a pH of 9.5. When repeating the experiments using 0.2 wt % R-DHPE as the fluorescent marker, the fluorescence inten-
TABLE 1. Comparison of the Fluorescence Intensity, I, the Drift Velocity, v, and the Diffusivity, D, Obtained in the Well Region and the Flat Part of the Surfacea Iwells/Iflat vwells/vflatb Dwells/Dflatc
pH ) 8.0
pH ) 9.5
1.84 ( 0.05 0.60 ( 0.03 0.60 ( 0.02
1.02 ( 0.05 0.95 ( 0.01 0.99 ( 0.04
a All values are given as the mean value ( one standard deviation obtained from at least three measurements. b vflat(pH ) 8.0) ) 0.19 µm/s and vflat(pH ) 9.5) ) 0.22 µm/s. c Dflat ) 2.00 µm2/s.
Fluorescence recovery after photobleaching (FRAP) was used to characterize the diffusivity of the fluorescently labeled lipids in the bilayer, both on the flat part of the surface and on the part containing the wells. The Hankel transform method, previously developed by us,25 was used to evaluate the experimental data. All diffusivity measurements were made with the bulk flow turned off. The pH of the bulk solution was 8.0 in all FRAP experiments, also when measuring over the areas of the bilayer that had been created when driving the lipid bilayer with a bulk solution at a pH of 9.5. The diffusivity of NBD C12-HPC over the flat © 2010 American Chemical Society
1902
DOI: 10.1021/nl100779k | Nano Lett. 2010, 10, 1900-–1906
Discussion. Figure 6 shows three possible scenarios after the lipid bilayer has been driven over a surface with embedded wells. In Figure 6a, the lipid bilayer follows the contour of the wells, in Figure 6b the lipid bilayer spans and seals the wells, whereas in Figure 6c the lipid bilayer circumvents the wells. All three scenarios presented in Figure 6 will influence the measured fluorescence intensity, the drift velocity of the lipid bilayer front, and the diffusivity of the lipids in the bilayer in different, characteristic ways. Comparison with the values obtained on the flat part of the surface allows the most probable conformation of the lipid bilayer to be identified at the two pH values used in these experiments. In addition, of the three scenarios in Figure 6, only the spanning-lipid bilayer in Figure 6b can encapsulate CF molecules inside the wells, resulting in a separate experiment to investigate whether spanning lipid bilayers can be made that seal the wells. Fluorescence Intensity. Since the diameter of the fabricated wells, and the average distance between them, is smaller than the lateral resolution of the optical microscope, the fluorescence intensity from individual wells cannot be discerned. What is observed is instead the average intensity in the plane of the channel. For the three cases illustrated in Figure 6, the average intensity will be higher in Figure 6a, the same in Figure 6b, and lower in Figure 6c, compared to the intensity on the flat region. For the case shown in Figure 6a, the relative change in fluorescence intensity over the wells, Iwells/Iflat, is given by
FIGURE 5. Fluorescence intensity over the region with wells arising from (a) R-DHPE in the lipid bilayer and (b) CF molecules. The lipid bilayer was driven using (1) a bulk solution without CF (pH ) 8.0), (2) a bulk solution with 0.5 mM CF (pH ) 9.5), and (3) a bulk solution without CF (pH ) 8.0). The image in (b) was acquired 15 min after step (3) started.
sity, the drift velocity, and the diffusivity ratio between the well region and the flat part of the surface were similar within one standard deviation to those obtained for NBD C12HPC (data not shown), making it unlikely that the observed trends are the effect of the fluorescent marker chosen to label the lipid bilayer. It should also be noted that the properties of the lipid bilayer already driven over the wells remained the same when the pH of the bulk solution above the bilayer was changed (from pH ) 9.5 to 8.0 or vice versa). Thus, the conformation of the lipid bilayer on different parts of the well region seems to be determined by the pH of the bulk solution used when the bilayer front is advancing over that specific part of the surface. This was also the case when the bulk solution was replaced with a buffer solution containing 100 mM NaCl. As a final experiment, an R-DHPE-labeled lipid bilayer was driven toward the well region by a bulk solution containing 0.5 mM of the fluorescent molecule CF. Since CF is fluorescent at shorter wavelengths than R-DHPE, this allows the light from the lipid bilayer and from the CF molecules to be visualized independently with different filter cubes (see Supporting Information for details). Figure 5 shows the results obtained from a measurement where the lipid bilayer was driven over the well region using a bulk solution without CF (pH ) 8.0) (1), then a bulk solution with 0.5 mM CF (pH ) 9.5) (2), followed by rinsing with a bulk solution without CF (pH ) 8.0) (3). A clear fluorescent signal from CF molecules was observed, also after rinsing, at region (2) in Figure 5b. Note also that the fluorescence intensity from R-DHPE in the lipid bilayer is lower over this area (see Figure 5a). This observation is in accordance with the results in Figure 2 where the fluorescence intensity over the well region is higher when driving the lipid bilayer at a pH of 8.0 than at a pH of 9.5. When the experiment illustrated in Figure 5 was repeated with the solution in step (2) at a pH of 8.0 instead of 9.5, no visible signal from the CF molecules was observed in the well region after rinsing. Neither could any signal from the CF molecules be seen after rinsing when the lipid bilayer was driven over the flat part of the surface with a bulk solution containing 0.5 mM CF molecules (pH ) 9.5). © 2010 American Chemical Society
Iwells /Iflat ) Acell /πR2 ) 1 + 2hp/R0
(1)
where Acell is the total surface area of a unit cell containing a single cylindrical well, R is the radius of the unit cell in the optical plane (parallel to the flat part of the surface), R0 is the radius of the well, p ) πR02/πR2 is the surface coverage of wells, and h is the height of the well. Values of h ) 123 ( 7 nm, R0 ) 40 ( 3 nm, and p ) 12%, obtained from a scanning electron microscopy image (see Supporting Information for details) yield Iwells/Iflat ≈ 1.74 ( 0.07, which corresponds to a 74% increase in the fluorescence intensity. If the lipid bilayer spans the wells as depicted in Figure 6b, then Iwells/Iflat ) 1, and for the case in Figure 6c the average intensity would be Iwells/Iflat ) 1 - p ) 0.88. The increase in fluorescence intensity when the lipid bilayer was driven over the well region by a bulk solution at a pH of 8.0 was observed to be 84%. This value is comparable to, but slightly higher than, the predicted increase in light intensity for the bilayer conformation presented in Figure 6a, estimated to be 74%. This indicates a scenario where the lipid bilayer follows the contours of the wells when driven by a bulk solution at a pH of 8.0. The reason for the slight discrepancy between the experimental and predicted values may be that the wells interfere with the fluorescence from a single fluorescent molecule. Another possibility is that 1903
DOI: 10.1021/nl100779k | Nano Lett. 2010, 10, 1900-–1906
FIGURE 6. A cross section through the center of a well where the lipid bilayer either (a) follows the contour of the well, (b) spans the well, or (c) stops at the edge of the well forming a pore in the lipid bilayer (figures not drawn to scale).
the equilibrium concentration of fluorescent markers in the lipid bilayer is not the same in the region with wells as on the flat part of the surface. If the lipid density is affected by the well, for example, due to bending when following the surface profile, the fluorescent markers may have a different equilibrium concentration in the well region. A different scenario is observed when the lipid bilayer is driven by a bulk solution with a pH of 9.5. In this case, the fluorescence intensity is the same on the flat part of the surface as in the region containing the wells, within the experimental accuracy. This observation indicates a scenario where the lipid bilayer spans the wells, as shown in Figure 6b, when driven over the well region by a bulk solution at a pH of 9.5. Drift Velocity. Under the assumption that the lipid density within the bilayer has reached steady state, the number of lipids that pass each cross section of the channel per unit time should be constant. Thus, the bilayer front should move more slowly over the region with wells than on the flat part of the surface if the lipid bilayer follows the contours of the wells as shown in Figure 6a. The reason for this is that more lipid material is required to cover a certain distance when the contours of the wells are followed than on the flat surface, resulting in vwells/vflat ) πR2/Acell ) 0.58 ( 0.02. If, on the other hand, the lipid bilayer spans the wells, as illustrated in Figure 6b, the drift velocity should be constant, vwells/vflat ) 1, whereas it should increase by vwells/ vflat ) (1 - p)-1 ) 1.14 if the bilayer circumvents the wells, as shown in Figure 6c. The decrease in drift velocity of 40% observed when the lipid bilayer is driven over the wells by a bulk solution at a pH of 8.0 compares well with the predicted value for the situation shown in Figure 6a, where the lipid bilayer follows the contours of the wells. The similar drift velocities on the flat part of the surface and on the well region at a pH of 9.5, vwells/vflat ) 0.95 indicate that the majority of the wells are sealed by a lipid bilayer in this case. However, the small decrease observed can be interpreted as the presence of a fraction of wells where the lipid bilayer follows the contours of the wells. Diffusivity. Since the wells are much smaller than the length scale over which the fluorescent molecules diffuse in these experiments, the observed diffusivity is an effective value. The effective diffusivity in a three-dimensional system © 2010 American Chemical Society
consisting of randomly arranged, spherically symmetric objects, has previously been described analytically by Jo¨nsson et al.28 The corresponding general formula for the case presented here, that is, diffusion over a flat surface with nonperiodic, cylindrical wells, is derived in the Supporting Information. From this formula, the following expression is obtained for the effective diffusivity, Dwells, of a molecule in a lipid bilayer that follows the contours of the wells, as shown in Figure 6a (see Supporting Information for a derivation of this special case)
Dwells ) DflatπR2 /Acell ) Dflat /(1 + 2hp/R0)
(2)
The last equality in eq 2 follows from eq 1. From the dimensions and coverage of the wells this yields πR2/Acell ) 0.58 ( 0.02 and Dwells ) 0.58Dflat. The diffusivity decreases since the molecules in the bilayer must diffuse a longer distance over the surface to have traveled a certain distance in the plane of the channel. An analytical expression can also be obtained for the two situations in Figure 6b,c (see Supporting Information for details)
Dwells ) Dflat
Dflat(1 - p) + Dfree(1 + p) Dflat(1 + p) + Dfree(1 - p)
(3)
where Dfree is the diffusivity of the lipids in the free-spanning parts over the wells. Using Dfree/Dflat ) 2.5, as obtained by Przybylo et al.,5 the effective diffusivity over the wells becomes Dwells ) 1.11Dflat. The effective diffusivity in the case where the lipid bilayer circumvents the wells (see Figure 6c) can be obtained from eq 3 by setting Dfree ) 0. This results in Dwells ) 0.79Dflat. The experimentally observed value of Dwells/Dflat ) 0.60 for the diffusivity in a lipid bilayer formed over the wells at a pH of 8.0 is in good agreement with the value of 0.58 calculated from eq 2. This is in line with the hypothesis that the lipid bilayer follows the contours of the wells when driven by a bulk solution at a pH of 8.0. However, the diffusivity in a lipid bilayer formed over the wells by a bulk solution at a pH of 9.5 was approximately the same as over the flat part 1904
DOI: 10.1021/nl100779k | Nano Lett. 2010, 10, 1900-–1906
of the surface (Dwells/Dflat ) 0.99). This value is slightly lower than the theoretical value of Dwells ) 1.11Dflat when the lipid bilayer spans the wells. The reason for this discrepancy could be that the lipid bilayer follows the contours of a fraction of the wells, but spans most of the others (c.f. the discussion of the drift velocity measurements). It may also be that the surface close to the wells has a rougher texture than the flat parts of the surface, resulting in higher frictional coupling of the lipid bilayer, thus reducing the average diffusivity in the bilayer. Encapsulation of Carboxyfluorescein. As an additional test of whether the lipid bilayer spans and seals the wells when driving the lipid bilayer with a bulk solution at a pH of 9.5, CF was added to the bulk solution. Because of its low permeability through a lipid bilayer, CF has previously been used as a fluorescent dye that can be encapsulated inside lipid vesicles.29,30 Thus, if the lipid bilayer seals the wells any CF molecules inside the wells will be trapped, whereas if the bilayer circumvents the wells or follows the contours of the wells, the CF molecules should be removed when rinsing with a bulk solution containing no fluorescent molecules. Figure 5b shows that after rinsing to remove the CF molecules, a clear fluorescent signal remained in the region where the lipid bilayer had been driven over the wells. This demonstrates that when the bulk solution has a pH of 9.5, the lipid bilayer seals the wells. No fluorescent signal was observed when the lipid bilayer was driven over the well region at a pH of 8.0, although the bulk solution contained CF. This is in agreement with the hypothesis that the lipid bilayer follows the profile of the surface when driven at a pH of 8.0. The intensity of the fluorescence from R-DHPE remains approximately constant for at least 15 h, whereas that from CF gradually decreases on a shorter time scale (∼1 h). This time of leakage is significantly faster than previously observed for CF encapsulated in lipid vesicles,30 which suggests that CF molecules gradually leaks out of the wells, either through pores formed in the lipid bilayer, or perhaps by diffusion underneath the lipid bilayer to defects in the bilayer. However, since the properties of the lipid bilayer driven over the well region remain approximately the same for over 15 h, this suggests that any pores formed in the freespanning parts of the lipid bilayer are self-healing. Effect of Salt Concentration. In all the experiments described above the bulk solution contained 5 mM NaCl and 2 mM TRIS, except when the SLB was formed from vesicles. In this case, the lipid vesicles were dissolved in a solution with a higher salt concentration, 100 mM NaCl, 10 mM TRIS, and 1 mM ethylenediaminetetraacetic acid disodium salt dehydrate (EDTA), otherwise the vesicles did not adsorb to the surface and ruptured. The reason for the lower salt concentration when driving the lipid bilayer was to prevent detached lipid material from the SLB from sticking to the surface in front of the bilayer. It was also observed that CF in the bulk solution prevented the advancement of the lipid © 2010 American Chemical Society
bilayer, when using a buffer solution with the higher salt concentration to drive the bilayer. This could be due to the adsorption of some of the CF molecules to the surface, whereas at the lower salt concentration the adsorption of CF molecules to the surface is prevented. However, it was possible to drive the lipid bilayer using the same salt concentration as when forming the SLB from lipid vesicles, when no CF molecules were present in the bulk solution. A spanning lipid bilayer could be obtained when using a pH of 9.0. Using a higher pH value resulted in desorption of the lipid bilayer from the surface which, when using the lower salt concentration, occurred at pH g 10. Thus, the behavior of the lipid bilayer at a specific bulk pH is affected by the salt concentration. Furthermore, when using the higher salt concentration it was more difficult to obtain reproducible results when reusing the sample. For example, the lipid bilayer occasionally did not span the wells or it started to be torn away from the surface already at a pH of 9.0. This problem was reduced when using the lower salt concentration. However, note that it was possible to revert to a higher salt concentration after the free-spanning lipid bilayers have been formed over the wells without any change in the conformation of the bilayer over the wells during the time of the experiment. Conclusions. Flowing a bulk solution above an SLB makes it possible to drive the lipid bilayer in a rolling motion in the direction of the bulk flow. This technique was used in the present study to move the lipid bilayer from the flat part of a SiO2 surface toward a region of the same surface with embedded wells. Since this method relies on the shear forces arising from the bulk flow acting on the lipid bilayer, it allows considerable freedom in the choice of salt concentration and pH value of the bulk solution driving the SLB. This in turn makes it possible to manipulate the interaction between the lipid bilayer and the support by changing the properties of the driving bulk solution. In fact, measurements of the fluorescence intensity, the drift velocity, and the diffusivity of fluorescently labeled molecules in the lipid bilayer showed that free-spanning lipid bilayers were not formed over the 80 nm wide wells in the SiO2 surface when the SLB was driven by a bulk solution with a pH of 8.0; instead the bilayer followed the contours of the wells in this case. In contrast, the experiments showed that free-spanning lipid bilayers that sealed the wells were formed when the pH of the driving bulk solution was raised to 9.5. This was also confirmed by encapsulating the fluorescent molecule CF inside the sealed wells. It should be noted that it is only when the lipid bilayer is formed over a well that the pH of the bulk solution determines whether the lipid bilayer spans the well or follows the contours of the well. Once the lipid bilayer has been formed over the wells, it retains its properties for a long time (at least 15 h in the current experiments), even when the properties of the bulk solution are changed. This in turn allows different types of lipid bilayer structures to be made over the wells 1905
DOI: 10.1021/nl100779k | Nano Lett. 2010, 10, 1900-–1906
by driving the lipid bilayer alternately at a pH of 8.0 and 9.5. Since all the steps employed when driving the lipid bilayer over the surface with fabricated wells were carried out in a microfluidic setup, the technique would also be suitable for lab-on-a-chip applications, potentially resulting in a miniaturized and automated platform for the study of transmembrane proteins.
(8) (9) (10) (11) (12) (13)
Acknowledgment. This work was financially supported by the Swedish Research Council for Engineering Sciences, contract number 2005-3140, the INGVAR grant from the Swedish Strategic Research Foundation, and a research grant (Ma¨rta and Erik Holmberg’s donation) from the Royal Physiographic Society in Lund.
(14) (15) (16) (17)
Supporting Information Available. Two sections can be found in the Supporting Information, (i) an experimental section with detailed information on the experiments and the analysis of the data obtained, and (ii) a theoretical section where the derivation of the effective diffusivity over the flat surface with nonperiodic, cylindrical wells is presented. This material is available free of charge via the Internet at http:// pubs.acs.org.
(18) (19) (20) (21) (22) (23) (24)
REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7)
McConnell, H. M.; Watts, T. H.; Weis, R. M.; Brian, A. A. Biochim. Biophys. Acta 1986, 864, 95–106. Sackmann, E. Science 1996, 271, 43–48. Castellana, E. T.; Cremer, P. S. Surf. Sci. Rep. 2006, 61, 429–444. Reimhult, E.; Kumar, K. Trends Biotechnol. 2008, 26, 82–89. Przybylo, M.; Sykora, J.; Humpolickova, J.; Benda, A.; Zan, A.; Hof, M. Langmuir 2006, 22, 9096–9099. Han, X. J.; Studer, A.; Sehr, H.; Geissbuhler, I.; Di Berardino, M.; Winkler, F. K.; Tiefenauer, L. X. Adv. Mater. 2007, 19, 4466–4470. Hennesthal, C.; Steinem, C. J. Am. Chem. Soc. 2000, 122, 8085– 8086.
© 2010 American Chemical Society
(25) (26) (27) (28) (29) (30)
1906
Schmidt, C.; Mayer, M.; Vogel, H. Angew. Chem., Int. Ed. 2000, 39, 3137–3140. Simon, A.; Girard-Egrot, A.; Sauter, F.; Pudda, C.; D’Hahan, N. P.; Blum, L.; Chatelain, F.; Fuchs, A. J. Colloid Interface Sci. 2007, 308, 337–343. Tiefenauer, L. X.; Studer, A. Biointerphases 2008, 3, Fa74–Fa79. Janshoff, A.; Steinem, C. Anal. Bioanal. Chem. 2006, 385, 433– 451. Mueller, P.; Wescott, W. C.; Rudin, D. O.; Tien, H. T. J. Phys. Chem. 1963, 67, 534–535. Schmitt, E. K.; Nurnabi, M.; Bushby, R. J.; Steinem, C. Soft Matter 2008, 4, 250–253. Jonsson, P.; Beech, J. P.; Tegenfeldt, J. O.; Hook, F. J. Am. Chem. Soc. 2009, 131, 5294–5297. Radler, J.; Strey, H.; Sackmann, E. Langmuir 1995, 11, 4539– 4548. Nissen, J.; Gritsch, S.; Wiegand, G.; Radler, J. O. Eur. Phys. J. B 1999, 10, 335–344. Jonsson, M. P.; Jonsson, P.; Dahlin, A. B.; Hook, F. Nano Lett. 2007, 7, 3462–3468. Furukawa, K.; Sumitomo, K.; Nakashima, H.; Kashimura, Y.; Torimitsu, K. Langmuir 2007, 23, 367–371. Montal, M.; Mueller, P. Proc. Natl. Acad. Sci. U.S.A. 1972, 69, 3561–3566. Osborn, T. D.; Yager, P. Langmuir 1995, 11, 8–12. Hanarp, P.; Sutherland, D. S.; Gold, J.; Kasemo, B. Colloids Surf. A 2003, 214, 23–36. Reimhult, E.; Kumar, K.; Knoll, W. Nanotechnology 2007, 18, 275303. Kam, L.; Boxer, S. G. J. Am. Chem. Soc. 2000, 122, 12901–12902. Jonsson, P.; Beech, J. P.; Tegenfeldt, J. O.; Hook, F. Langmuir 2009, 25, 6279–6286. Jonsson, P.; Jonsson, M. P.; Tegenfeldt, J. O.; Hook, F. Biophys. J. 2008, 95, 5334–5348. Gilmanshin, R.; Creutz, C. E.; Tamm, L. K. Biochemistry 1994, 33, 8225–8232. Tamm, L. K. Biochemistry 1988, 27, 1450–1457. Jonsson, B.; Wennerstrom, H.; Nilsson, P. G.; Linse, P. Colloid Polym. Sci. 1986, 264, 77–88. Schwarz, G.; Arbuzova, A. Biochim. Biophys. Acta 1995, 1239, 51– 57. Weinstein, J. N.; Yoshikami, S.; Henkart, P.; Blumenthal, R.; Hagins, W. A. Science 1977, 195, 489–492.
DOI: 10.1021/nl100779k | Nano Lett. 2010, 10, 1900-–1906