Electrophoretic Transport in Surfactant Nanotube Networks Wired on

Publication Date (Web): July 4, 2006 ... We present a simple method to rapidly obtain compact surfactant nanotube networks of controlled geometry and ...
0 downloads 0 Views 600KB Size
Download PDF
Anal. Chem. 2006, 78, 5281-5288

Accelerated Articles

Electrophoretic Transport in Surfactant Nanotube Networks Wired on Microfabricated Substrates Johan Hurtig, Bodil Gustafsson, Michal Tokarz, and Owe Orwar*

Department of Chemical and Biological Engineering, and Department of Microtechnology and Nanoscience, Chalmers University of Technology, SE 412 96, Go¨teborg, Sweden

Nanofluidic devices are rapidly emerging as tools uniquely suited to transport and interrogate single molecules. We present a simple method to rapidly obtain compact surfactant nanotube networks of controlled geometry and length. The nanotubes, 100-300 nm in diameter, are pulled from lipid vesicles using a micropipet technique, with multilamellar vesicles serving as reservoirs of surfactant material. In a second step, the nanotubes are wired around microfabricated SU-8 pillars. In contrast to unrestrained surfactant networks that minimize their surface free energy by minimizing nanotube path length, the technique presented here can produce nanotube networks of arbitrary geometries. For example, nanotubes can be mounted directly on support pillars, and long stretches of nanotubes can be arranged in zigzag patterns with turn angles of 180°. The system is demonstrated to support electrophoretic transport of colloidal particles contained in the nanotubes down to the limit of single particles. We show that electrophoretic migration velocity is linearly dependent on the applied field strength and that a local narrowing of the nanotube diameter results from adhesion and bending around SU-8 pillars. The method presented here can aid in the fabrication of fully integrated and multiplexed nanofluidic devices that can operate with single molecules.

The fields of nanofluidics and single-molecule transport are rapidly receiving growing attention.1-3 The prospect of devices * To whom the correspondence should be addressed. E-mail: orwar@ chalmers.se. Phone: + 46-(0)31-772 3060. Fax: + 46-(0)31-772 3858. (1) Han, J.; Craighead, H. G. J. Vacuum Sci. Technol., A 1999, 17, 21422147. (2) Kuo, T.-C.; Cannon, D. M., Jr.; Shannon, M. A.; Bohn, P. W.; Sweedler, J. V. Sens. Actuators, A 2003, 102, 223-233. (3) Whitesides, G. M. Nat. Biotechnol. 2003, 21, 1161-1165. 10.1021/ac060229i CCC: $33.50 Published on Web 07/04/2006

© 2006 American Chemical Society

so miniscule in size that they are capable of sequencing, synthesizing, and characterizing single or very few molecules is tantalizing. Today, nanofluidic devices can be obtained in a wide variety of materials using a number of different fabrication protocols and procedures.4-8 Some of these systems have reached the limit of how far miniaturization can proceed, as the size of the molecules under study is close to the characteristic length scale of the nanofluidic channels used.8,9 The quantitative benefits of miniaturization are small reagent volumes, reduction in analysis time, parallel analysis, lower cost fabrication, and experiments that can be run at dimensions more relevant for small-scale biological systems, i.e., single cells and single organelles. Other, qualitative, benefits include the possibilities to manipulate single specimens of interest such as proteins and nucleic acids, as well as performing high-resolution analysis using local light sources and local detectors. To achieve a successful design of nanofluidic devices for the manipulation of single biopolymers such as single DNA and single proteins, an understanding is required of how the physics of the devices scale with decreasing dimensions.10,11 With micro- and nanofabricated structures, distinct characteristics such as rapid mixing by diffusion, rapid heat transfer, and the ability to handle volumes in the 10-9-10-18 L range are available, depending on the choice of fabrication technology.12-14 (4) Craighead, H. G. Science 2000, 290, 1532-1535. (5) Pearson, J. L.; Cumming, D. R. S. Microelectron. Eng. 2005, 78-79, 343348. (6) Quake, S. R.; Scherer, A. Science 2000, 290, 1536-1540. (7) Tegenfeldt, J. O.; Prinz, C.; Cao, H.; Huang, R. L.; Austin, R. H.; Chou, S. Y.; Cox, E. C.; Sturm, J. C. Anal. Bioanal. Chem. 2004, 378, 1678-1692. (8) Tokarz, M.; A° kerman, B.; Olofsson, J.; Joanny, J.-F.; Dommersnes, P.; Orwar, O. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 9127-9132. (9) Martin, F.; Walczak, R.; Boiarski, A.; Cohen, M.; West, T.; Cosentino, C.; Ferrari, M. J. Controlled Release 2005, 102, 123-133. (10) Madou, M. J.; Cubicciotti, R. Proc. IEEE 2003, 91, 830-838. (11) Reisner, W.; Morton, K. J.; Riehn, R.; Wang, Y. M.; Yu, Z.; Rosen, M.; Sturm, J. C.; Chou, S. Y.; Frey, E.; Austin, R. H. Phys. Rev. Lett. 2003, 94, 196101. (12) Arata, H. F.; Rondelez, Y.; Noji, H.; Fujita, H. Anal. Chem. 2005, 77, 48104814.

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006 5281

These types of devices, coupled with sensitive, high data acquisition rate detectors, can be used for single-molecule experiments. Typical target molecules are biopolymers such as DNA and proteins, but also other colloidal particles and polymers. Singlemolecule experiments differ distinctly from ensemble studies15 since they yield information on the instantaneous transformations of individual molecules. For example, the dynamics of enzymatic reactions,16 receptor binding, and conformational changes in polymers17,18 have all been performed on the single-molecule level. We have previously presented techniques to produce nanofluidic soft-matter networks based on self-assembly, self-organization, and forced shape transformations.19,20 Using micromanipulation methods, networks consisting of surface-immobilized unilamellar vesicles (∼5-25 µm in diameter), conjugated with suspended nanotubes 100-300 nm in diameter, can be produced with controlled geometry and topology.21,22 Patterned substrates for manufacturing and control of lipid nanotube networks have been used in a number of applications, for instance, in micropipet writing23 and in construction of three-dimensional networks.24 Transport of molecules and polymers between nodes within these networks can be controlled by membrane tension differences,25,26 electrophoresis,8 and diffusion.27 In this work, we demonstrate how tailored micropatterned structures improve the geometrical degrees of freedom in creating such networks. The nanotubes are pulled from lipid vesicles using a micromanipulation technique, with multilamellar vesicles serving as reservoirs of surfactant material, and in a second step, wired around microfabricated SU-8 pillars. In contrast to previous methods, the technique presented here can produce compact nanotube networks with arbitrary turn angles as well as single nanotubes directly conjugated to pillars in both ends. We furthermore show that the system supports electrophoretic transport of colloidal particles contained in the nanotubes. The system maintains structural integrity, and the electrophoretic migration velocity is linearly dependent on the applied field strength allowing controlled handling of particles and molecules. The method presented here can aid in the fabrication of fully (13) Jensen, K. F. Chem. Eng. Sci. 2001, 56, 293-303. (14) Rondelez, Y.; Tresset, G.; Tabata, K. V.; Arata, H.; Fujita, H.; Takeuchi, S.; Noji, H. Nat. Biotechnol. 2005, 23, 361-365. (15) Xie, X. S.; Trautman, J. K. Annu. Rev. Phys. Chem. 1998, 49, 441-480. (16) Lu, H. P.; Xun, L.; Xie, X. S. Science 1998, 282, 1877-1882. (17) Basche, T.; Nie, S.; Fernandez, J. M. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 10527-10528. (18) Lipman, E. A.; Schuler, B.; Bakajin, O.; Eaton, W. A. Science 2003, 301, 1233-1235. (19) Karlsson, M.; Sott, K.; Cans, A.-S.; Karlsson, A.; Karlsson, R.; Orwar, O. Langmuir 2001, 17, 6754-6758. (20) Karlsson, A.; Karlsson, M.; Karlsson, R.; Sott, K.; Lundqvist, A.; Tokarz, M.; Orwar, O. Anal. Chem. 2003, 75, 2529-2537. (21) Karlsson, M.; Sott, K.; Davidson, M.; Cans, A.-S.; Linderholm, P.; Chiu, D.; Orwar, O. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 11573-11578. (22) Lobovkina, T.; Dommersnes, P.; Joanny, J.-F.; Bassereau, P.; Karlsson, M.; Orwar, O. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 7949-7953. (23) Sott, K.; Karlsson, M.; Pihl, J.; Hurtig, J.; Lobovkina, T.; Orwar, O. Langmuir 2003, 19, 3904-3910. (24) Hurtig, J.; Karlsson, M.; Orwar, O. Langmuir 2004, 20, 5637-5641. (25) Karlsson, R.; Karlsson, M.; Karlsson, A.; Cans, A.-S.; Bergenholtz, J.; A° kerman, B.; Ewing, A. G.; Voinova, M.; Orwar, O. Langmuir 2002, 18, 4186-4190. (26) Dommersnes, P. G.; Orwar, O.; Brochard-Wyart, F.; Joanny, J. F. Europhys. Lett. 2005, 70, 271-277. (27) Sott, K.; Lobovkina, T.; Lizana, L.; Tokarz, M.; Bauer, B.; Konkoli, Z.; Orwar, O. Nano Lett. 2006, 6, 209-214.

5282

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

integrated and multiplexed nanofluidic devices that can operate with single molecules. MATERIALS AND METHODS Chemicals. Liposome buffer composition was at pH 8.1 (10 mM K3PO4, 10 mM KH2PO4, 5 mM Trizma base, 90 mM KCl, 1 mM MgSO4, 0.5 mM EDTA); the chemicals were of analytical grade and purchased from Sigma (St. Louis, MO). Soybean Polar Lipid Extract was purchased from Avanti Polar Lipids Inc. (Alabaster AL). FM 1-43 (N-(3-triethylammoniumpropyl)-4-(4(dibutylamino)styryl)pyridinium dibromide), 3,3′-dioctadecyloxacarbocyanine perchlorate, and FluoSpheres (carboxylated, diameter 30 and 200 nm, 505/515) were from Molecular Probes (Eugene, OR). SU-8 25 negative photoresist was from MicroChem Corp. (Newton, MA), XP SU-8 developer was from Microresist Technology (GmbH, Berlin, Germany), and UV-5 positive photoresist and MF24A developer were from Shipley (Marlborough, MA). The SC-1 wash was of the composition 1:1:5 of NH4 (25%)/ H2O2 (30%)/DI H2O with pro analysi chemicals from VWR International AB (Stockholm, Sweden) as well as hexamethyldisilazane (HMDS). Acrylamide and bisacrylamid were from BioRad Laboratories (Hercules, CA). Materials. Four-inch soda lime wafers coated with 1000 Å of low reflective chrome were from Nanofilm (Westlake Village, CA). Borosilicate coverslips were from Menzel Gla¨ser (Braunschweig, Germany), with thickness of 0.08-0.12 mm (No. 0), and borosilicate capillaries, 1.0 mm o.d. × 0.78 mm i.d., with inner filament were from Harvard Apparatus (Edenbridge, U.K.). Microstructured Substrates. Substrate fabrication methods were used as described in Hurtig et al.24 Briefly, patterns specifically developed for these experiments were drawn in AutoCAD format and transferred to a chrome mask with e-beam lithography and wet etching. Borosilicate substrates were SC-1 cleaned, oven dehydrated, and spin coated with the ultrathick negative photoresist. The pattern was transferred to the negative resist and developed, forming a positive relief with a UV lithography technique.28 Finally, the processed coverslips were rinsed in 2-propanol. Single-Molecule Confocal Microscopy. The 488-nm line from an air-cooled Ar+ laser (Spectra-Physics 163C) was sent through a neutral density filter (Newport ND 10, Irvine, CA), a color glass filter (GG475, Melles Griot Inc., Carlsbad, CA), and a band-pass filter (10LF10-488, Newport), before the beam entered an inverted microscope (Leica DM IRB, Wetzlar, Germany). The light was reflected by a dichroic mirror (Q495LP, Omega Optical, Inc., Brattleboro, VT) through a microscope objective (Leica, HCX PL APO 63×, NA 1.32, oil imm,). The fluorescence was collected back from the objective and a dichroic mirror, through a 50-µm pinhole (Melles Griot), which serves to spatially reject the outof-plane light. Light from the image plane was sent through a bandpass filter (HQ 525/50, Chroma Technology Corp., Brattleboro, VT) to reject the Raman and Rayleigh scattering, and last, a planoconvex lens (Newport) focused the light onto an avalanche photodiode (SPCM-AQR-16, EG&G Canada Ltd., Vaudreuil, Canada). The pinhole, the spherical lens, and the detector were mounted on separated three-axis translation stages (Newport) for (28) MicroChem Corp. Nano SU-8 Negative Tone Photoresists Formulation 2-25 http://www.microchem.com/products/pdf/SU8_2-25.pdf, 2005.

optimal alignment. The detector is coupled to a data acquisition card (NI PCI-6024E, National Instruments, Austin TX) on personal computer running an in-house-built LabView 7.1 data collection software counting number of incident photons with bin size of 10 ms. Video Microscopy. Excitation of dyes in the imaging experiments was performed with an Hg lamp (100 W; Leica), and the excitation wavelength was selected with an I-3 filter block (Leica) or with the 488-nm line from a water-cooled Ar laser (SpectraPhysics Stabilite 2017), focused onto the coverslips using a 63× objective (Leica, HCX PL APO 63x, NA 1.32, oil immersion) mounted in an inverted microscope (Leica DMIRB). Images were acquired with a three-chip color CCD camera (C6157-01, Hamamatsu, Kista, Sweden), recorded on digital video (DVCAM, DSR-11, Sony), digitized with Adobe Premier or with an EMCDD (iXon DV877-BI Andor Technology, Belfast, Northern Ireland), and analyzed in Adobe Photoshop and Andor Bioimaging Tracker. Micropipet-Assisted Formation of Unilamellar Networks. Fabrication of unilamellar networks of nanotube-interconnected liposomes, using a micropipet-assisted technique on SBL giant unilamellar-multilamellar vesicles (GUV-MLV), was performed as described previously.19 High graduation micromanipulators (MWH-3, Narishige, Tokyo, Japan) allowed precise control of vesicle immobilization in the x-y-z position. Borosilicate micropipets were pulled on a CO2 laser puller instrument (model P-2000, Sutter Instrument Co., Novato, CA). Two types of pipets were used: one thin and flexible with a orifice diameter of 200 nm, for network formation, and one with a 500-nm opening for injection of FluoSphers. A microinjection system (Eppendorf, CellTram Vario) and a pulse generator (Digitimer Stimulator DS9A, Welwyn Garden City, U.K.) were used to control the electroinjections and the electroporation for insertion of micropipets. As the starting material for creation of networks, we used giant vesicles made from soybean polar extract using a dehydration/rehydration protocol.29 Microelectrode Preparation. Microelectrodes used for electrophoretic transport were made as described by Tokarz et al.8 Briefly, a silver wire was electrodeposited with AgCl(s) and inserted into a borosilicate micropipet filled with buffer solution containing chloride ions, thus making an Ag/AgCl electrode. The interior of the pipet was treated with HMDS in order to reduce the surface charge density and thus suppress electroosmotic flow. Furthermore, the pipet tip was plugged by polymerizing acrylamide and bisacrylamide; the cross-linked gel acts as a salt bridge, preventing bulk solvent flow, but allowing electric current to pass. Following formation of desired network geometry and filling of the selected vesicle with a probe (FluoSpheres), the gel-plugged microelectrodes were electroinserted into the filled GUV. One microelectrode was withdrawn and translated across the patterned substrate so that the trailing nanotube was bent around a pillar. Potentials of 300-700 mV were applied in order to induce electrophoretic transport of FluoSpheres. The ζ potential for the soybean lipid mixture has been measured, ζmemb ∼ -46 mV, in microelectrophoresis experiments as well as the value for the carboxylated FluoSpheres, ζbead ∼ -59

mV.30 The mechanism of electrophoretic transport in lipid nanotubes is modeled by Tokarz et al.8 and described below.

(29) Karlsson, M.; Nolkrantz, K.; Davidson, M. J.; Stro ¨mberg, A.; Ryttse´n, F.; A° kerman, B.; Orwar, O. Anal. Chem. 2000, 72, 5857-5862.

(30) Supporting Information. (31) Seifert, U. Adv. Phys. 1997, 46, 13-138.

RESULTS AND DISCUSSION Formation of Nanotube Networks Wired on Microstructured Substrates. We have previously demonstrated24 that the surface properties of SU-8 are suitable for immobilization of soybean lipid (SBL) vesicles. In this paper, we use micropillars made from the same substrate material for point-attaching lipid nanotubes extracted from unilamellar vesicles using micropipets.19 By attaching lipid nanotubes to microstructured substrates, it is possible to create networks with control over connectivity, tube lengths, turn angles, and exact location of network vertexes. These geometrical parameters are governed by the design of the microstructured substrate. Thus, the upper size limit of the network is only limited by the accessible surface area (mm-cm range) and the supply of lipid bilayer material. The lower limit is governed by the aspect ratio of the resist, and an aspect ratio of 10/1 gives a smallest footprint area of ∼10 µm2 for each structure. The minimal spacing between pillars that still allows for translating the micropipets is ∼2 µm. The initial formation and pulling of the nanotubes follows the same procedures as previously described for formation of nanotube vesicle networks (NVNs).19 However, the vertexes in NVNs are defined by the positions of surface-adhered vesicle, whereas with the current method they are defined by SU-8 pillars. Specifically, using micromanipulators that can be translated in the x-y-z directions, a nanotube pulled from a mother vesicle is (i) manually guided and threaded around the SU-8 pillars (Figure 1 a, b); thus, nanotube patterns can be wired around pillars in a flat-weaving fashion. Or alternatively, (ii) directly attached to desired contact points on SU-8 pillars (Figure 1h, j) to create specific patterns. In both instances, extremely long (mm-cm range), continuous nanotube networks can be created on a small surface area. Upon contact between a SBL nanotube and a SU-8 pillar, the nanotube adheres along the area of contact (Figure 1b, c). After each point of contact, the tube can be further extended, where membrane material for the tube elongation is supplied from the mother vesicle (Figure 1d). Figure 1e displays a fluorescence micrograph of a meandering nanotube pattern of 1.1-mm length on a micropatterned area of 0.01 mm2 made using the procedure schematically shown in Figures 1a-d. Lipid membrane material flows smoothly across the contact point during translation. This indicates that adhesion of nanotubes to SU-8 pillars does not cause a structural collapse of the membrane tube. Given the elastic properties of the membrane, it is reasonable to assume that it will become somewhat flattened at the contact point.31 Thus, the adhesion forces will locally change the cross section of the nanotube (Figure 1f). This constriction will create a local hinder to passage of large particles as well as a locally increased field strength in electrophoresis experiments as will be discussed in greater detail below. As stated above, it is also possible to create networks where nanotube ends are directly attached to SU-8 pillars. In this procedure, a nanotube pulled from a unilamellar vesicle is guided

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5283

Figure 1. Formation of lipid nanotube networks utilizing tube adhesion on microstructured surfaces. (a) A micropipet is electroporated into a unilamellar vesicle connected to a multilamellar vesicle-giant unilamellar vesicle (MLV-GUV) complex. (b) As the tube is manually guided and threaded around or, alternatively, directly attached to desired contact points on SU-8, it adheres to the structures. (c, d) The tube is extended further with supply of membrane material flowing over the adhesion point from the MLV-GUV complex. (e) Micrograph of a tapered micropipet forming a lipid nanotube connection between the mother liposome and the pipet tip; the MLV-GUV complex is not visible. Nanotube length and exact dimensions are controlled by point attachment of the tube to the microstructured substrate; total length of this tube is 1.12 mm, a length impossible to achieve in a controlled fashion in visible workspace using freely suspended nanotubes. (f) Three-dimensional sketch of the experiment with an inset illustrating the proposed constriction of the nanotube at contact points. NB: the tube is not to scale with the pillars. (g, h) Micrograph and line graph of nanotube networks supported by microstructured substrates where the lower tubes are terminated at support structures without collapsing toward the vesicle. (i, j) Micrograph and line graph of a tube network disconnected from all vesicles, i.e., a network consisting of nanotubes solely.

and contacted to a pillar using the micropipet tip (Figure 1b). In a second step, the nanotube connection from the micropipet tip is cut with an electric dc pulse. This result in a retraction of excess membrane material from beyond the contact point toward the SU-8 pillar due to membrane tension in the system and a nanotube end conjugated to a pillar structure (Figure 1g, h). It is furthermore feasible to create a network consisting only of nanotubes, directly conjugated to SU-8 pillars, by also cutting the connection to the mother vesicle (Figure 1i, j). Electrophoretic Transport in Nanotube Networks. According to the microelectrophoretic measurements, presented in the online Supporting Information, the soy bean lipid membrane has a negative ζ potential with the salt solutions used in this work.30 The membrane is therefore pulled by the electric field toward the positive electrode. Velocity for spherical lipid vesicles in free 5284

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

solution at high ionic strengths can simply be modeled by the Smoluchowski formula νm ) 0Eζ /η, where E is the applied electric field, ζ is the zeta potential,  is the solvent dielectric constant, 0 is the permittivity of vacuum, and η is the viscosity. In our system, the field acts on a cylindrical tube, and the electric field is primarily applied inside the tube. Due to the low conductivity of the membrane (∼10-14 S/m) compared to the conductivity of the buffer (0.5 S/m), the field at the nanotube exterior surface is small. The electrophoretic equations therefore only need to be solved for a nanotube geometry with an electric field acting on the interior side of the nanotube membrane wall. As shown by Tokarz et al.,8 the electrostatic potential close to the membrane can be approximated by that of a flat surface. The electric field inside the tube is constant and parallel to the tube axis; outside the tube the field is small and the electric force on

Figure 2. Theoretical description of electrophoretic transport mechanisms in lipid nanotubes. (a) Schematic of tube structure around SU-8 pillars illustrating the hypothesis of a constricted geometry. (b) The theoretical flow profile in an undeformed tube. The essentially flat flow profile for the solvent inside the tube, with a velocity at the center line ∼10 times lower than the membrane electrophoretic velocity, resulting in an absence of any substantial liquid transport. Transport of objects in the tube is hence due to their own charge or adhesion to the membrane tube wall.

the exterior bilayer can be neglected. The cross-linked gel in the pipet connected to the nanotube has low water permeability; the net water flux in the tube is therefore set to zero. Solving the equations for nanotube geometry yields an electrophoretic membrane-wall velocity 10% smaller than predicted by the Smoluchowski formula, due to surface friction at the exterior leaflet.8 This results in an essentially flat flow profile for the solvent inside the tube, depicted in Figure 2b. There is an electroosmotic flow generated due to the force on the positively charged Debye layer. However, since the membrane wall is also moving, the effective solvent velocity close to the wall is much smaller than the velocity of the electroosmotic flow. The velocity at the center line is ∼10 times lower than the membrane electrophoretic velocity, resulting in an absence of any substantial liquid transport, in agreement with the fact that we do not observe any accumulation of fluid at the electrode tip.

The velocity of particles transported inside the nanotube will depend on several factors including particle charge, nanotube length, applied potential, and variations in the nanotube crosssectional geometry at contact points. Furthermore, large particles may interact with the membrane surface causing local shape deformations. In general, an applied potential of 100-700 mV is required to drive the transport in our setup. By employing an equivalent circuit model of the electrophoretic system with an electrode resistance, a seal resistance, and a nanotube resistance, it comes out that 62.5% of the applied potential falls over the nanotube.8 This number is used in the calculation of field strength in the experiments. The difference between the electrophoretic mobilities of the lipid membrane, and the latex particles is ∼20%.30 Electrophoretic Transport of 200-nm-Diameter Beads. We have earlier demonstrated electrophoretically driven transport of DNA fragments in lipid nanotubes,8 and this method was now utilized to translate latex particles in networks with point-attached nanotubes. Initially, transport experiments were performed using time-lapse fluorescence microscopy of electrophoretically transported 200-nm-diameter latex beads. These particles are roughly of the same diameter as the lipid nanotube and are easy to track during transport due to the high intensity of the fluorescence signal. Experiments were performed by first inflating a unilamellar vesicle with a bead suspension, after which two microelectrodes were electroinjected into the vesicle. A nanotube was then formed by retracting one of the microelectrodes, and the desired electrophoresis potential was applied. The first experiments were performed in straight nanotubes of equal length biased at increasing electrode potentials, i.e., field strengths. The setup is schematically illustrated in Figure 3a. These experiments demonstrate a linear relationship between the applied electric field and particle velocity (Figure 3b). The fact that the line does not pass through the origin can be attributed to membrane spreading on the glass electrode or to a nonlinear behavior in this region. In experiments where the field strength was altered in steps during transport, it was found that particles changed their velocity in less than 2 s after the voltage change (Figure 3c). This is an indication that charging times are short and that the field is established rapidly across the nanotube. There appear also not to be any long-time fluctuations in field strength, as the particles maintain a constant velocity over extended time periods, for instance, in Figure 3d, seven beads are translated with constant speed for a total transport time of 43 s. We then examined electrophoretic transport in lipid nanotubes bent around SU-8 pillars. These experiments demonstrate that the bent nanotube is intact and that it is possible to transport particles as large as 200 nm in diameter around adhesion points. Figure 4a illustrates the experimental setup. With the microelectrode pipet, a tube is extracted and adhered to a pillar and a voltage is applied over the tube to the counter electrode inside the unilamellar vesicle. At the point of contact, three behaviors are observed: (i) the beads move unhindered around the pillar as illustrated in the micrographs in Figure 4b-e and Figure 4f (particles 2-4 and 6-8). (ii) The beads pass the adhesion point with decreased velocity. This might be a result of the migrating object being a small aggregate of beads, too large to pass through the constricted part of the nanotube. The tube therefore needs to Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5285

Figure 3. Electrophoretic transport of latex particles, 200-nm diameter, in a straight nanotube. (a) Schematic illustration of the setup for transport in straight nanotubes. A nanotube is extracted from a unilamellar vesicle containing a bead suspension, and an electric field of 100-700 mV is applied in four steps equivalent to ∼8-39 V/cm. (b) Data plot of beads transported with increasing electrode potentials performed as separate experiments, but with the same microelectrodes and tube lengths, showing the linear relationship between applied voltage and the electrophoretic velocity of the particles. The line is a linear fit to the data with an R2 value of 0.9899. (c) Plots of bead displacement during electrophoretic transport, where the applied voltage is increased during the experiment, demonstrating the ability of fast adaptation of particle velocity. Tube length is 103 µm (d) Plots of bead displacement during transport, demonstrating equal and constant velocities for all particles over an extended time scale.

expand before the aggregate can pass, which involves transport of lipid material to the contact point. This behavior is displayed by particles 1 and 9 in Figure 4f. Particle 5 in Figure 4f is released when approached from behind by another particle. This is a recurring behavior and has been observed many times, but the physical explanation remains an open question. (iii) In the last case, a particle stops at the pillar indefinitely, thus blocking the passage for further transport. This behavior is generally observed either with very large aggregates, i.e., much larger than the tube 5286 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

Figure 4. Electrophoretic transport of latex particles in a nanotube bent around a pillar. (a) Schematic illustration of the setup for transport in adhered nanotubes. A nanotube is extracted from a unilamellar vesicle, containing a suspension of beads with a diameter of 200 nm, and the tube is bent around the SU-8 pillar. (b) When tube adhesion is established, an electrophoretic potential is applied over the microelectrodes. Dotted lines indicate tube location, and the kink in the line is the location of the adhesion point. Arrows indicate the location of a transported particle. (c-e) A bead in the tube approaches the pillar and passes the point of contact as proof of an intact tube at the adhesion point. (f) Examples of bead behavior by plot of bead displacement acquired by particle tracking software from video recording. Particle 2-4 and 6-8 pass the point of contact (at ∼18 µm of displacement) without any loss of velocity. Particle 1 passes the pillar with reduced velocity, and particles 5 and 9 stop completely for some time at the pillar before they pass, resuming the same velocity as they had before the pillar. Particle 5 is dislodged by beads arriving at the contact point.

diameter, congesting the tube at the constriction or when the first particle transported trough the nanotube is unable to pass the contact point, indicating a defect in the nanotube at the pillar. Electrophoretic Transport 30-nm-Diameter Beads. Transport experiments with confocal fluorescent detection of electrophoretically transported 30-nm-diameter beads were performed to investigate whether single particles can be interrogated and tracked in the bent tubes and, furthermore, to investigate the transport behavior of particles smaller than the nanotube diameter. Experiments were performed in a manner similar to that for the 200-nm-diameter beads. A unilamellar vesicle was inflated with a bead suspension, microelectrodes were electroinjected into the vesicle, and a tube was extracted and bent around a pillar. Here, confocal fluorescence detection with a photoavalanche diode was performed in nanotubes to ensure detection of single beads. To investigate transport around pillars, the confocal detection spot was first positioned along the nanotube, before the contact point and then after the contact point, in two separate experiments. The confocal detection spot is larger than the nanotube, so all light from particles passing through the nanotube is ideally detected. A time trace of fluorescent bursts, where the detection spot is located before the contact point, is shown in Figure 5a, and the measurement performed after the contact point is shown in Figure 5b. Figure 5c shows the background noise level. The number of beads passing around the pillar is indicated by the peak frequency and the size of a passing object by the peak amplitude. We observed both a reduced peak frequency and reduced peak amplitudes after the pillar; compare panels a and b in Figure 5. The amplitude of the bursts was ∼3 times higher before the contact point, with mean and standard deviation, 204 ( 191 counts (n ) 3), than after the contact point, with 63 ( 60 counts (n ) 3). Comparing the distributions, it becomes apparent that a local sized-based separation of latex spheres has occurred. The distribution of peak amplitudes measured before the contact point represents a wide range of aggregate sizes, as well as single particles. The peak amplitude distribution measured after the contact point is much narrower, but still contains a range of aggregates as well as single particles; the larger species have been filtered and are thus not present in the distribution. Since single particles as well as small aggregates are expected to be present both before and after the contact point, the distributions naturally overlap. We ascribe this filtering effect to the constriction of the nanotube as it is bent around the pillar. Thus, similar to the behavior displayed for 200-nm-diameter beads, 30-nm-diameter beads are arrested at the contact point between a bent nanotube and a SU-8 pillar. Large aggregates that arrive at the contact point accumulate over time and form a plug. Single beads and very small aggregates are able to pass between the plug and the lipid tube by forcing the tube to expand locally. However, since this is not possible to resolve optically, we have no definitive proof of this mechanism. This behavior is not observed for straight nanotubes where transport occurs smoothly for both types of beads. CONCLUSIONS We demonstrate methods to construct nanotube vesicle networks by wiring the connecting nanotubes to microstructured substrates. This technique increases the compactness of lipid nanotube networks and stabilizes large-scale networks. Furthermore, the technique allows for direct suspension of single

Figure 5. Representative results of single-photon microscopy of electrophoretic transport of fluorescent latex beads through nanotubes bent around a pillar. Measurements were performed on carboxylated latex beads (diameter 30 nm, 0.02% w/w) at points along the nanotube before and after the adhesion point. (a) Time trace (40 s) of bead transport measured before the pillar, background ∼40 counts/ 10 ms. (b) Time trace (40 s) of bead transport measured after the pillar, background ∼25 counts/10 ms. A 300-mV potential was applied between the electrodes in all measurements. (c) Time trace (10 s), measurement of the solution background fluorescence in close vicinity (∼ 20 µm) to the inflated vesicle filled with 30-nm beads.

nanotubes anchored at both ends to SU-8 supports. Adhesion of lipid nanotubes to the photoresist polymer SU-8 is a fundamental key to this scheme, maintaining small pattern areas while still retaining very long nanotube lengths. We show that adhered nanotubes maintain their integrity on microfabricated SU-8 structures and support electrophoretic transport. The adhered tubes also display signs of a constricted geometry at adhesion points. This can potentially be used to introduce filtering functionalities into complex networks, perhaps making it possible to use a meandering tube, similar to the one shown in Figure 1e, to perform size exclusion chromatography and single-molecule analysis in nanotube vesicle networks. ACKNOWLEDGMENT The authors thank the Swedish Research Council (VR), the Swedish Foundation for Strategic Research (SSF, Nano-X, INGVAR), Chalmers Bio-Initiative for financial support, and Go¨ran Gustafsson foundation. Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5287

SUPPORTING INFORMATION AVAILABLE Zeta potential measurements for soybean lipid mixture and carboxylated FluoSpheres. This material is available free of charge via the Internet at http://pubs.acs.org.

5288

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

Received for review February 3, 2006. Accepted June 9, 2006. AC060229I