| STUDY |
Lipid membranes can be used to make nanoscale fluidic devices that can control single nanoparticles and molecules, according to Swedish scientists. Orwar PHOTO BY JAN-OLOF YXELL 10µm CONNECT THE DOTS This network contains six vesicles, each about 4 µm in diameter. COURTESY OF OWE ORWAR "We believe that these systems are suitable for a number of applications, ranging from studies of chemical reactions in confined biomimetic compartments and single-polymer dynamics to construction of complex nanoscale fluidic devices with applications in bioanalysis, sensors, and computation," says Owe Orwar, professor of biophysical chemistry at Chalmers University of Technology, Göteborg. "Today, we use the networks to perform studies on enzyme kinetics as a function of dimensionality, geometry, and surface properties as well as to construct complex sensor devices based on coupled chemical reactions and chemical kinetics." Orwar and his colleagues at Chalmers and Göteborg University use liposomes to create fluidic networks consisting of vesicles connected by nanotubes [Anal. Chem., 75, 2529 (2003)]. The main challenges are finding the right membrane materials and suitable surface properties to immobilize the networks, according to Orwar. The networks are made by immobilizing liposomes made of soybean lipids on a surface and then using a pipette and micromanipulator to divide them into daughter liposomes connected by thin tubes less than 100 to 300 nm in diameter. The resulting liposomes can be made as small as 3 or 4 µm in diameter. The liposome reactors can each support different chemical reactions. The nanotubes allow mass transport between the different microreactors. The fluid flow is controlled by deforming the surface-adhered vesicles, which changes their surface-to-volume ratio and increases the surface tension, inducing a flow of the membrane that constitutes the walls of the system. Samples are introduced either by direct injection into a vesicle using microelectroinjection or by nanotube-assisted injection of sample-filled daughter vesicles. The researchers use the nanodevice to transport single 30-nm fluorescent beads, which they detect using laser-induced fluorescence. "We have constructed a fully operational and integrated nanofluidic device that essentially has reached the limit of how small you can go in terms of channel and device dimension if you are working with single, large-biopolymer systems in biochemical physics, analysis, or synthesis," Orwar says |
| UPDATE | 05.03 |
| AUTHOR |
Anders Karlsson, Mattias Karlsson, Roger Karlsson, Kristin Sott, Anders Lundqvist, Michal Tokarz, and Owe Orwar* Department of Physical Chemistry, and Microtechnology Centre, Chalmers University of Technology, SE-412 96 Göteborg, Sweden, and Department of Chemistry, Göteborg University, SE-412 96 Göteborg, Sweden |
| LITERATURE REF. |
Anal. Chem., ASAP Article 10.1021/ac0340206 S0003-2700(03)04020-4 Web Release Date: May 1, 2003 Copyright © 2003 American Chemical Society Nanofluidic Networks Based on Surfactant Membrane Technology Anders Karlsson, Mattias Karlsson, Roger Karlsson, Kristin Sott, Anders Lundqvist, Michal Tokarz, and Owe Orwar* Department of Physical Chemistry, and Microtechnology Centre, Chalmers University of Technology, SE-412 96 Göteborg, Sweden, and Department of Chemistry, Göteborg University, SE-412 96 Göteborg, Sweden Received for review January 9, 2003. Accepted April 1, 2003. Abstract: We explore possibilities to construct nanoscale analytical devices based on lipid membrane technology. As a step toward this goal, we present nanotube-vesicle networks with fluidic control, where the nanotube segments reside at, or very close (<2 m) to optically transparent surfaces. These nanofluidic systems allow controlled transport as well as LIF detection of single nanoparticles. In the weak-adhesion regime, immobilized vesicles can be approximated as perfect spheres with nanotubes attached at half the height of the vesicle in the axial (z) dimension. In the strong-adhesion regime (relative contact area, Sr* ~0.3), nanotubes can be adsorbed to the surface with a distance to the interior of the nanotubes defined by the membrane thickness of ~5 nm. Strong surface adsorption restricts nanotube self-organization, enabling networks of nanotubes with arbitrary geometries. We demonstrate LIF detection of single nanoparticles (30-nm-diameter fluorescent beads) inside single nanotubes. Transport of nanoparticles was induced by a surface tension differential applied across nanotubes using a hydrodynamic injection protocol. Controlled transport in nanotubes together with LIF detection enables construction of nanoscale fluidic devices with potential to operate with single molecules. This opens up possibilities to construct analytical platforms with characteristic length scales and volume orders of magnitudes smaller than employed in traditional microfluidic devices. -------------------------------------------------------------------------------- Following a strong advancement in miniaturization of fluidic devices to systems with characteristic length scales in the low-micrometer range, we now see an evolution toward nanoscale fluidic systems that have the potential capability to handle single molecules throughout the entire analysis/manipulation cycle. In such systems, mass goes toward unity, i.e., 10-23 mol, and concentration will correspondingly go toward infinity, when geometric confinement increases. To avoid that operations will be stochastic, control of location (position) and even orientation of single molecules within these devices will be of great importance. To achieve this goal, nanoscale fluidic devices offering unprecedented control over transport and manipulation as well as detection are required. Whereas the problem in detecting single molecules using, for example, confocal laser-induced fluorescence (LIF) microscopy has largely been solved for a range of usable probes,1 fully integrated nanofluidic systems that can perform a complex set of operations on single or small populations of molecules confined to volumes as small as 10-21 L still presents a daunting challenge. One of the challenges lies in the fact that nanoscale fluidic systems with controlled geometry and topology are not easily constructed using traditional microfabrication methods and materials. An alternative route might involve the use of molecules that aggregate or self-organize through noncovalent interactions2 to form useful structures on the supramolecular scale. Examples include microemulsions, surfactant membranes, and some polymer systems that can self-assemble or self-organize into interesting aggregate structures, e.g., nanotubes or molecularly thin sheets that can be further processed chemically or mechanically into mesoscale structures appropriate for handling and processing single molecules. In principle, biology that displays the only workable solution toward such single-molecule devices has solved the problem on the level of a single cell by using surfactant membranes that play a key role in transport, packaging, sorting, and providing a mesoscale reaction environment for single or a few molecules. With a large portion of inspiration from biological systems, we explore the possibilities of using surfactant structures such as lipid membranes for construction of nano- and microdevices for transportation and mixing of extremely small volumes of liquids (10-12-10-21 L). The networks consist of one or several sheets of a two-dimensional liquid (L, lamellar liquid crystal state). So far, we have demonstrated methods for formation of complex nanotube-vesicle networks (NVNs), having multiple containers and nanotube connections, through mechanical fission3 and a micropipet-based technique.4 We have also demonstrated how the topology of a network can be controlled by using an electrofusion protocol5 and how the membranes as well as the interior compartments can be functionalized with, for example, proteins to control influx and efflux of analytes and to perform reactions confined to certain containers within the network.6 Finally, we have demonstrated transport of molecules and nanoparticles across nanotubes to a given node within a network using a membrane-tension-driven flow.7 The simplest network consists of two surface-immobilized vesicles with hemispherical to near-spherical shape of equatorial plane radius Ra and Rb (Ra ~ Rb) conjugated by a cylindrical tube of radius Rt and length Lt. If the system is in the weak-adhesion regime, we can approximate the vesicles to have perfect spherical geometry connected by a tube that has perfect cylindrical geometry. The volume, V0 of the system is then equal to For a system with Ra = Rb = 5 m, Rt = 100 nm, and Lt = 10 m, then V0 = 1.0475 pL. The volume of the tube segment is 0.314 fL, which represent less than 0.03% of the volume of the whole system. Importantly, the equilibrium radius, Rt0, of the lipid nanotube is given by a force balance between curvature (bending modulus c) and lateral membrane tension, m:8 Consequently, the diameter of the nanotubes can to some extent be changed on-line by adjusting the membrane tension in a network by the use of, for example, micropipet aspiration protocols. Tube diameters as small as 25 nm have been estimated in similar systems.9 Thus, given a membrane thickness of 5 nm, the prospect of building and regulating fluidic channels with inner diameters only slightly larger than individual biomolecules is possible using this technology. Thus, in principle, the limit for how small the channels that can be employed in handling biopolymers such as RNA, DNA, and proteins has been reached. In our networks, a typical tube radius is ~100 nm, giving a cross-sectional area of 3,14 × 10-14 m2. This should be compared with the size of a single hemoglobin molecule, which is 5.5 nm in diameter, with a corresponding cross-sectional area of 1.59 × 10-17 m2. It is interesting to note that the total system volume of these networks is less than 1/106 of the injection volume used with traditional microseparation methods such as CE, and the nanotube volume is less than 1/109 over a traditional CE capillary. To construct NVNs of useful geometries and topologies, the material properties of fluid bilayer membranes are of decisive importance. Fluid bilayer membranes obey the shape/energy equations of thin elastic shells where the shape of a network is dictated by the membrane material mechanics, arranging in such a way that the elastic energy of the system is minimized. Following the reasoning of Helfrich,10,11 the elastic energy per unit area of a thin elastic shell can be described by The first term represents the area dilation (in-plane stretching) contribution to the surface energy and Ka is the area dilation modulus (~240 mN/m), A is the total membrane area, and A0 is the equilibrium starting area. The second term represents the local bending contribution to surface energy, where c is the local bending modulus (~0.9 × 10-19J, or ~40kbT, at 300 K) and c1 and c2 are the principle curvatures (1/R1 and 1/R2), respectively. Basically this means that lipid bilayer membranes are very easy to bend but difficult to stretch. The last term represents the nonlocal bending contribution to free energy where r is the nonlocal bending modulus (~3-4c) and h is the membrane thickness (~5 nm). The area difference, A in the nonlocal bending term is a function of the principal curvatures and membrane thickness integrated over the surface: The equilibrium shape of an unperturbed membrane structure in solution can be found from minimizing the elastic energy; From this reasoning, it should be evident that the tubular segments in our networks are residing in an elastically excited state due to the extreme curvature. If approximating the tubular segments as perfect cylinders, the minimum curvature energy of a tubular segment corresponds to Thus, as a valid approximation for long tubes (tube lengths several times the tube diameter), tube energy scales linearly with tube length and is inversely proportional to tube radius. Consequently, the nanotubes exerts attractive forces on the vesicle containers in a network in order to shorten the tubes, thus minimizing the elastic energy of the system. Therefore, these structures cannot exist free in solution since the tubes would force all containers to coalesce into a single vesicle in the case of a genus = 0 network. The vesicle containers in a network must therefore be immobilized onto a substrate to sustain a given geometry. In this work, we present methods for construction of lipid bilayer nanofluidic networks where vesicles are adsorbed to surfaces under conditions of differential adsorption potentials. Specifically, we explored the possibilities to construct nanotube networks at, or very close (<2 m) to, transparent surfaces to enable implementation of highly sensitive optical probing techniques in detection of fluorescent probes transported inside the nanotubes. We demonstrate LIF detection of single nanoparticles (30-nm-diameter fluorescent latex spheres) inside the confines of single nanotubes adsorbed close to the surface of microscope coverslips. Transport of nanoparticles was induced by a surface tension differential applied across nanotubes using a hydrodynamic injection protocol acting on targeted vesicles. Materials and Methods Liposome Preparations. Preparation of multilamellar liposomes. Soybean lipids (SBL) were dissolved in chloroform to a suitable concentration, typically 100 mg/mL as stock solution. Lipid solution, chloroform, and buffer solution (intracellular buffer, 135 mM KCl, 5 mM NaCl, 20 mM HEPES, 1.5 mM MgCl2 and 10 mM D-glucose, pH 7.4) were placed in a round flask (total concentration of lipid in solution was ~0.25 mg/mL). The solution was rota-evaporated until the organic solvents were removed. This procedure formed a range of unilamellar and multilamellar liposomes as described by Moscho et al.12 To make unilamellar liposomes, a dehydration/rehydration method described by Criado and Keller13 was used with modifications.14 Briefly, 5 L of aqueous lipid dispersion (1 mg/mL) was placed on a coverslip glass and the drop was dehydrated in a vacuum desiccator at 20 C. The partially dehydrated lipid film was then carefully rehydrated with PBS buffer (Trizma base 5 mM, K3PO4 30 mM, KH2PO4 30 mM, MgSO4 1 mM, EDTA 0.5 mM, pH 7.8). After a few minutes, giant unilamellar liposomes started to form. Instrumentation. Two different microscope setups were used in the experiments. One setup was used for Nomarski and fluorescence imaging and one for semiconfocal LIF detection. Normarski and Fluorescence Imaging. An inverted microscope (Leica DM IRB, Wetzlar, Germany) equipped with a Leica PL Fluotar 40× and PL APO 63× objective was used for Nomarski and fluorescence imaging (Figure 1). The 488-nm line of an Ar+ laser (2025-05, Spectra-Physics Lasers Inc., Mountain View, CA) was used in the fluorescence measurements. The laser light passed through a spinning disk to break the coherence and scatter the laser light. A lens collected the light, and a filter cube (I3, Leica Microsystems AB, Sweden) sent it through the objective to excite the fluorophores. The same objective collected the fluorescence and a charge-coupled device (CCD) camera (C6157-01, Hamamatsu, Kista, Sweden), controlled by an Argus-20 image processor (Hamamatsu Photonics), was used to detect the image, which was recorded by a digital video (DVCAM, DSR-11, Sony, Japan). -------------------------------------------------------------------------------- Figure 1 Schematic showing the instrumental setup. All experiments were performed on an inverted microscope, and networks were produced on glass coverslips by mechanical fission or a micropipet-based technique. For mechanical fission, two carbon fibers (CF) were controlled by micromanipulators (MM). When the microelectroinjection technique was used, one of the carbon fibers was replaced by a microinjection pipet (MIP) connected to a pressurized-air microinjector (MI). Dc voltage pulses were applied with a voltage generator (VG). A microscope objective (MO), 40× or 63×, was used for Nomarski and fluorescence imaging of the networks. For fluorescence imaging, a 488-nm laser light emitted from an Ar+ laser was sent via a mirror (MR), a spinning disk (SD), a focusing lens, and a polychroic beam splitter through the objective. Nomarski and fluorescence images were collected by the same objective and detected by a CCD camera. The images were digitally enhanced by an Argus-20 unit and then recorded by a digital video (DV). A monitor was also used to show the output from the Argus-20. A similar setup was used when the single-nanoparticle detection experiments were performed with an avalanche photodiode detector replacing the CCD camera. In this case, a pinhole was used to filter out the fluorescence arising from other planes than the plane of focus. A lens was also used to focus the collected fluorescence onto the detector. -------------------------------------------------------------------------------- LIF Detection of Single Nanoparticles Inside a Single Nanotube Channel. The 488-nm line from an air-cooled Ar+ laser (Spectra-Physics 163C) was sent through a neutral density filter (Newport ND 20, Irvine, CA), a color glass filter (GG475, Melles Griot Inc., Irvine, 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 (HCX PL APO 63×, NA = 1.32, oil immersion, PH 3, working distance 0.17 mm, Leica). The fluorescence was collected back from the objective and a dichroic mirror, through a 50-m pinhole (Melles Griot), which serves to spatially reject the out-of-plane light. The fluorescence from the image plane was sent through a band-pass filter (HQ 525/50, Chroma Technology Corp., Brattleboro, VT) to reject the Raman and Rayleigh scattering, and last, a plano-convex lens (Newport) was used to focus the light onto an avalanche photodiode (SPCM-AQ-161, EG&G Canada Ltd., Vaudreuil, Canada). The pinhole, the spherical lens, and the detector were mounted on separated three-axis translation stages (Newport) for optimal alignment. The detector is coupled to a multichannel scaler (MCS-plus) from EG&G Ortec (Oak Ridge, TN) and a personal computer with Companion software (A68-BI, v.1.0) to facilitate data collection. The borosilicate coverslips (rectangular, no. 1) used for the experiments were cleaned by rinsing in ethanol and water or by RCA-1 cleaning (H2O/H2O2/NH3, 5:1:1, 70 C, 10 min) followed by much rinsing in water. Formation on Multilamellar Networks by Mechanical Fission. A diluted suspension of multilamellar liposomes was placed on a coverslip glass, and the liposomes were allowed to settle on the surface for a few minutes. Networks of vesicles and nanotubes were created by mechanical fission of surface-immobilized liposomes (5-30 m in diameter), accomplished by using flexible 5-m-diameter, 30-m-long carbon fibers (Dagan Corp., Minneapolis, MN) controlled by high-graduation micromanipulators (Narishige MWH-3, Tokyo, coarse manipulator, Narishige MC-35A, Tokyo). The carbon fiber was placed on the equator of a liposome for homofission, resulting in two equally sized daughter liposomes, or at some desired latitude for heterofission, resulting in two differently sized daughter liposomes. The fiber was translated in the z-direction until it touched the surface of the coverslip (Figure 2A,B). To separate the daughter liposomes after division, the micromanipulator-controlled carbon fiber was used to slowly (a few m/s) push one of the daughter liposomes to a target site on the surface (Figure 2C). Tubes could be made of controlled length up to several hundred micrometers by translating the daughter liposome, before it was allowed to adhere to the coverslip surface. The positional precision was generally better than ±5 m but is susceptible to changes in surface properties. The diameter of the daughter liposomes could be controlled with a precision of ~±1 m for a 10-m-diameter mother liposome when forming daughter liposomes in the range of 5-8 m in diameter. However, several trials might be needed on the same liposome before a daughter vesicle of desired size is produced. -------------------------------------------------------------------------------- Figure 2 Formation of nanotube-vesicle networks. Multilamellar lipid networks were formed by mechanical fission of surface-immobilized vesicles by the use of 5-m-diameter carbon fibers as cutting tools as illustrated in (A-C). (A, B) By pressing down the micromanipulator-controlled carbon fiber, the vesicle is ultimately divided into two "daughter" vesicles connected with a nanotube. (C) The daughter vesicles could then be separated by moving the carbon fiber horizontally along the surface. When the vesicles were separated with a desired distance and the vesicle had reached its target site on the surface, the translation was stopped and the vesicle was allowed to adhere to the surface. This fission and translation technique could be iterated to expand the network. For creation of unilamellar networks, a micropipet technique was used for pulling nanotubes and inflating daughter vesicles from unilamellar, or thin-walled, GUVs as illustrated in (D-G). (D, E) By the use of mechanical excitation and electrical pulses, the injection needle was easily inserted into the vesicle. (F, G) After the lipid adhered to the tip, a nanotube was pulled and buffer was injected into the nanotube, through the injection tip. A daughter vesicle started to form, and when the vesicle had reached a desired size and location on the surface, the vesicle was allowed to adhere to the surface and the injection needle was removed. The parameters describing the geometry of these networks are defined in Figure 1. The vesicle size ranges between 5 and 30 m in diameter, the separation distance, Lt, is usually between 10 and 100 m, and the nanotube diameter is approximately 100-300 nm. -------------------------------------------------------------------------------- To minimize the interaction between the lipids and the carbon fiber used for fission, the fibers were coated with bovine serum albumin (BSA) from a 1 mg/mL solution. Occasionally, the lipid nanotubes stuck to the carbon fibers. These problems were overcome by mechanical release of the nanotubes, simply by moving the fiber at a high velocity from the liposomes or by applying a short pulse of an electric field across two electrodes to repel the lipid membrane. The nanotubes then detached from the fiber, and the membrane material was re-integrated into the liposome. Micropipet-Assisted Formation of Unilamellar Networks. A carbon fiber microelectrode and a tapered micropipet, controlled by high-graduation micromanipulators, were used to create the unilamellar networks of nanotube-interconnected liposomes, using a micropipet-assisted technique from giant SBL unilamellar, or thin-walled, vesicles attached to multilamellar vesicles (GUV-MLV) as described previously.5 With this method, networks can be produced with controlled nanotube length, angle between nanotube extensions, and vesicle container diameter. In brief, a tapered borosilicate-glass micropipet with an outer-tip diameter of 0.5-1 m, back-filled with aqueous medium and mounted onto an electroinjection system,14 was pressed against the membrane of a surface-immobilized vesicle (Figure 2D). By applying dc voltage pulses of field strengths between 10 and 40 V/cm and duration of 1-4 ms over the micropipet, the lipid membrane was penetrated. The micropipet was then slowly pulled out and away from the mother vesicle, forming a lipid nanotube connection between the mother liposome and the pipet tip (Figure 2E,F). Aqueous medium was thereafter injected into the nanotube using a pressurized-air-driven microinjector, thus forming a small satellite vesicle at the outlet of the micropipet-tip. This newly created vesicle could then be released and immobilized onto the substrate surface at the desired coordinates (Figure 2G). The tapered injection micropipets were made from borosilicate capillaries (GC100TF-10, Clark Electromedical Instruments, Reading, U.K.) that were pulled on a CO2 laser puller instrument (model P-2000, Sutter Instrument Co., Novato, CA). A microinjection system (Eppendorf Femtojet or Celltram Vario, Hamburg, Germany) and a pulse generator (Digitimer Stimulator DS9A, Welwyn Garden City, U.K.) were used to control the electroinjections. Chemicals. Chloroform, EDTA (titriplex III), magnesium sulfate, potassium dihydrogen phosphate, potassium chloride, sodium chloride, magnesium chloride, and HEPES were from Merck (Darmstadt, Germany). Trizma base, potassium phosphate, D-glucose, and BSA were from Sigma (St. Louis, MO). Glycerol was from J. T. Baker, and deionized water from a Milli-Q system (Millipore Corp., Bedford, MA) was used. Soybean polar lipid extract (composed of 22 1% phosphatidylethanolamine, 45.7% phosphatidylcholine, 6.9% phosphatic acid, 18.4% phosphatidylinositol, and 6.9% other) was purchased from Avanti (Alabaster, AL). Fluorescent carboxylate-modified microspheres with diameters of 27 nm (±3 nm) (FluoSpheres, F-8787) and 3,3'-dioctadecyloxacarbocyanine perchlorate (DiO) were from Molecular Probes (Eugene, OR). All solutions were carefully filtrated though a PVDF 0.2-m filter (Acrodisc LC 13). Results and Discussion Formation of Surface-Immobilized Two-Dimensional Nanotube-Vesicle Networks. Fluid-state lipid bilayer networks were constructed on inverted microscopes (Figure 1) by the use of two recently developed techniques, mechanical fission using 5-m-diameter flexible carbon fibers3 (Figure 2A-C) and a micropipet-based technique4 (Figure 2D-H). As starting material for the production of networks we used either multilamellar liposomes (MLVs), made by a rota-evaporative technique,12 or giant unilamellar vesicles (GUVs), formed by a dehydration/rehydration technique.14 From both preparations, 5-30-m-diameter liposomes made from SBL were used as starting material. Figure 2I shows the geometry of the simplest network consisting of two weakly adsorbed vesicles with near-spherical shape of equatorial plane radius Ra and Rb (Ra ~ Rb) conjugated by a cylindrical tube of radius Rt and length Lt. As discussed below, the degree of adsorption of vesicles to surfaces will affect their shape and tension and effectively change critical system parameters. One of the motivations behind this work was to enable detection of singular or low numbers of fluorescent species residing inside or on the membrane surface of nanotubes using optical probing techniques such as evanescent-wave and confocal LIF detection. With these probing techniques, the geometrical distance of the interrogated object from the surface is of importance. In particular, with evanescent-wave fluorescence detection, the penetration depth is limited to less than 200 nm.15 Thus, objects untouched by the evanescent field will not be detected. Therefore, we explored the possibilities to construct nanotube networks at, or very close (<2 m) to, transparent surfaces by modifying adsorption potentials or constructing nanotube detection zones between two small surface-adsorbed vesicles. Adhesion Effects on Immobilized Networks. When a vesicle container is immobilized onto a substrate, the contact area of the adhering vesicle will, because of Brownian motion of the membrane, grow progressively until equilibrium between adhesion energy and elastic energy is reached.16 Thus, vesicle adhesion affects the energy of the system, and an adhesion energy term has to be introduced into the elastic energy equation (eq 5). The surface forces can be approximated to act only at the liposome-substrate interface, due to electrostatic shielding. Gathering these forces into one contact potential we get where is the effective potential of adhesion and S* is the area of surface-membrane contact. Consequently, the adhesive forces will contribute to the shape of the bound vesicle containers and thus the membrane tension (and therefore also affect the tube radius) in the network according to eq 2. The degree of vesicle adhesion is therefore also essential for the stability of a network. If the membrane tension overcomes the threshold of pore expansion, a transient hole will form in the membrane allowing leak-out of a fraction of the internal fluid in order to decrease the membrane tension by release of hydrostatic pressure.17,18 In fact, if the adhesive force is very strong, the vesicle containers will undergo lysis (rupture) and spread on the surface.17 Typical lysis tensions for liposomes are in the range 4-10 mN/m.19 To investigate the effects of adhesion on network geometry, we developed a method based on fluorescence microscopy for monitoring the interactions between a vesicle container and a planar surface (Figure 3A). Normally, in epiluminescence microscopy, having coverslips mounted perpendicular to the propagation of light, the x-y plane is viewed, giving 2-D images with an axial view from below. We used instead a planar glass surface mounted parallel to the beam path of the microscope. This results in images of the objects in the z-x plane giving 2-D views of the objects from the side. We investigated liposome adhesion to borosilicate substrates treated in two different ways. First, No. 1 borosilicate coverslip glasses were rinsed in 95% ethanol followed by thorough rinsing in water and were blown dry in a stream of N2 gas. These surfaces displayed a ~45 contact angle with water droplets. Second, coverslips were subjected to the RCA-1 wash protocol, rinsed in water, and dried at 80 C for ~2 h before use. These surfaces are highly hydrophilic and displayed a contact angle with water droplets of <5. -------------------------------------------------------------------------------- Figure 3 Interactions of vesicle nanotube networks with the substrate surface at different adhesion forces. (A) Schematic of the viewing procedure. A glass surface was mounted parallel to the microscope beam path, making it possible to view a network from the side. (B) A fluorescence micrograph showing immobilization of SBL vesicles, stained with DiO, on ethanol-water-treated substrates. In these experiments, the SBL vesicles appeared as slightly truncated spheres with a relative contact area, Sr* of 0.07. We define Sr* as the ratio between contact area and total membrane area (we neglect the tube area). (C) An RCA-1 washed surface displays a much higher contact potential, which results in hemispherical vesicles with Sr* ~0.30 The interconnecting tube was hard to visualize. In this situation, the tube should be situated close to or at the surface. The scale bar is 10 m, and this scale is also applicable to the image in (B). (D) Schematic showing how Ls-t is getting progressively shorter as vesicle diameter decreases for vesicles ranging in diameter from 4 to 10 m. A Sr* of 0.07 was used in the calculations. Ls-t is given in micrometers and is slightly less than the radius of the adsorbed vesicle. (E) A Normarski image showing six sequentially made vesicles that were set to be 2 m in radius; the measured average radius ± SD for these six vesicles was 1.87 ± 0.2. The scale bar is 10 m. (F) Schematic illustration showing how differential adhesion affects Ls-t, and container geometry. The containers were approximated as truncated spheres with constant membrane area, and the area contribution from the nanotube was neglected. Five cases were plotted with Sr* ranging from 0 (perfect sphere) to 0.33 (perfect hemisphere). For Sr* = 0, the corresponding vesicle radius is 1.73 m. Ls-t is given in micrometers, and it can be seen that it goes from 1.73 to 0 m as Sr* increases. -------------------------------------------------------------------------------- The vesicle-substrate interaction was fundamentally different for these two surfaces (Figure 3B,C). SBL vesicle containers immobilized onto an ethanol/water-treated glass surface appear as weakly truncated spheres with a relative contact area Sr*, of ~0.07 (Figure 3B). We define Sr* as the ratio between contact area and total membrane area. We used the contact area obtained in experiments with ethanol-washed coverslips to calculate Ls-t, i.e. the distance from the coverslip surface to the nanotube, for vesicles ranging in diameter from 4 to 10 m (Figure 3D). Ls-t is slightly less than the radius of the adsorbed vesicle. Thus, to make tubes residing close to surfaces, it is crucial to prepare immobilized vesicles with small radii. The diameter of immobilized vesicles can be controlled down to ~3-4 m with the pipet-assisted micromanipulation technique, which implies that the smallest distance of the nanotube to the surface in this adsorption regime can be controlled to ~1.2-1.6 m. Figure 3E shows six sequentially made vesicles that were set to be 2 m in radius, the measured average radius ± SD for these six vesicles was 1.87 ± 0.2. This experiment shows that the precision in making such small vesicles is fairly good. However, making vesicles in a reproducible manner with a radius less than 1.5 m is difficult with this technique as well as with the fission technique. If nanotubes are to reside closer to the surface, vesicles have to be immobilized using stronger adsorption potentials to create hemispherical or close to hemispherical vesicles. Vesicle containers immobilized onto glasses treated with the RCA-1 wash protocol displayed very strong adhesion behavior. The vesicle containers now exhibited a hemispherical, or domelike, shape and displayed a relative contact area of ~0.3 (Figure 3C). In fact, the adhesion potential for this surface was high enough to promote pore formation and the vesicle containers typically released 10-12% of their internal volume before they stabilized on the substrate surface (data not shown). Furthermore, the nanotubes appeared to be thinner than observed with ethanol-washed surfaces, indicating that the membrane tension actually is higher in this system. Importantly, in this strong-adhesion regime, the nanotubes were emanating from the vesicle containers with virtually no separation distance between the tube and the glass substrate. Figure 3F shows vesicles immobilized at different relative contact areas and illustrate how Ls-t progressively decreases with Sr*. We found that, in the strong adhesion obtained with RCA-1-washed coverslips, also the nanotubes could be fully immobilized onto the substrate surface. Consequently, networks could be produced having tubular connections of arbitrary shape allowing even more complex design of these membrane devices. Importantly, nanotube-vesicle networks formed with such strong adsorption potentials leading to short Ls-t distances should allow the opportunity to employ evanescent-wave microscopy15 for detection of fluorescent species transported in the nanotubes. Figure 4A, shows a three-container network containing tubes attached onto the surface. When the tubes are "glued" to the substrate, the dynamical self-organizing behavior displayed by nonadhered tubular segments is lost.5 However, networks produced on surfaces with strong adsorption potentials appear to be significantly less stable than networks produced on substrates having weak adsorption potentials and could typically only be sustained for 30 min up to 1 h before they underwent stuctural collapse. Also, there is a tendency for formation of a vesicular structure at surface-immobilized nanotube junctions to relieve some of the stress that is implied on these types of structures (Figure 4A). Similar phenomena have been observed with optical tweezers acting on nanotubes leading to pearl-chain formation.20 For comparison, a three-container network produced on an ethanol-treated substrate is shown in Figure 4B. Here, the nanotubes are not adhering to the substrate and therefore connect between containers in a way describing the minimum pathway. -------------------------------------------------------------------------------- Figure 4 (A) Three-container network displaying strong adhesion to the substrate (RCA-1 washed borosilicate). The network was formed using the micropipet-assisted technique. Due to the strong adhesion, the vesicles obtain a hemispherical shape and the tubes are thus located close to or at the surface. The Normarski image shows that the tubes can connect to each other with 90 angles when they are adhering to the substrate. When the tubes are not adhering to the substrate, they connect in a way minimizing the amount of tube in the network. This pathway minimization results in 120 angles between the tubular segments as shown in (B), where a three-container network was constructed on an ethanol-treated borosilicate surface. Scale bar 10 m. -------------------------------------------------------------------------------- Probing Single Nanoparticle Transport in Nanofluidic Networks with LIF Detection. We designed optimized networks for sample injection, flow control, and continuous high-speed recording of fluorescent species transported in nanotubes using LIF detection. The generic design of the network developed is shown in Figure 5A, and the procedure for its production is described in Figure 2D-H. The vesicles were made by SBL lipids and formed on ethanol-washed surfaces with Sr* of ~0.07. The formation of the networks is based on the adhesion behavior of vesicles and the ability to control their diameters. The network consists of two larger vesicles, one that is used for sample loading (left-hand vesicle in Figure 5A) and one for controlling lipid flow by regulation of surface tension, shown to the right in Figure 5A. These two vesicles are ideally ~10 m in diameter because both sample loading5,14 and flow control7 are performed with micropipets a few micrometers in diameter. Sample loading can be performed by either direct injection into a vesicle using microelectroinjection protocols14 or nanotube-assisted injection of daughter vesicles filled with sample.5 Alternatively, the vesicles can be loaded during their formation in the rehydration step. Here, we loaded vesicles with 30-nm-diameter fluorescent beads using a microelectroinjection protocol. -------------------------------------------------------------------------------- Figure 5 (A) Network consisting of four vesicles interconnected by nanotubes that was used in combination with LIF detection. The two small vesicles were included to be able to position the nanotube close to the substrate surface. Fluorescent beads were injected into the system as flow markers. (B) When buffer (same as the surrounding medium) was injected into the vesicle to the right, a lipid flow is created over the system to eliminate the membrane tension difference that arises. The lipid flows from the vesicle to the left, over the two small vesicles and all interconnecting nanotubes. The lipid flow creates a flow of fluid and particles inside the tube through viscous coupling. (C) Normarski image of a four-container network similar to that described in (A) and (B). The injection needle continuously injects buffer into the attached vesicle, thereby creating a lipid flow in the system from the large/giant vesicle that is attached to a multilamellar vesicle. The multilamellar part donates lipid into the network on demand when, for example, the membrane tension is increased in a vesicle container in the system. The detection site was positioned between the two smaller vesicles (red ring), to be able to detect nanoparticles that flow inside the nanotube. Scale bar 10 m. (D) Schematic of part of the vesicle-nanotube system, the nanoparticles that flow inside the tube, and the laser focus used for fluorescence excitation of the fluorescent beads. Due to the restricted geometry of the nanotube, the particles are more or less aligned and pass through the detector probe volume when a lipid flow is created. -------------------------------------------------------------------------------- A continuous flow of nanoparticles was induced by hydrodynamic injection of buffer solution into a vesicle using a micropipet coupled to a pressure-driven microinjector. The nanobeads are transported by the lipid membrane flow induced by the tension difference between vesicle containers.7 By injecting buffer into the vesicle to the right (Figure 5B), a continuous lipid flow from the vesicle to the left, over the two smaller vesicles and the interconnecting nanotubes, was created. It has previously been shown that the lipid flow drags along the fluid and particles present in the lipid nanotube.7 Detection was achieved by focusing the laser light to a diffraction-limited spot, wherein the fluorescent nanoparticles was excited. The fluorescence collected by the objective was sent through a 50-m pinhole to reject out-of-focus light, leading to a highly sensitive detection system. The working distance of our objective is ~170 m, roughly corresponding to the thickness of the coverslips used. Thus, optimal detection is performed close to the surface. We therefore constructed a detection zone consisting of a nanotube conjugated to two small vesicles of equal size (diameter ~4 m) resulting in a nanotube residing ~1.6 m above and parallel to the surface between the two vesicles. The tube was made short in order to minimize the convective motion of the tube and to ensure that the nanotube resided in the detector probe volume at all times. Figure 5C shows a Nomarski image of a nanofluidic system having a LIF detection zone consisting of two smaller vesicles conjugated by nanotubes. The location of the laser focus and point of detection on the nanotube between the two smaller vesicles on the surface is indicated in the figure. In LIF detection, a probe volume as small as 1-5 fL is achieved as a result of the tight focusing. The ultrasmall scale of the lipid nanotubes can be perfectly matched in size, making it possible to register single fluorescent nanoparticles passing through the detector probe volume (Figure 5D). We detected 30-nm fluorescent beads at two different concentrations in transit through the probe volume. Figure 6 shows traces at the two different concentrations used. The low concentration (diluted 1000 times) shows well-separated narrow fluorescence bursts corresponding to single beads traversing the probe volume (Figure 6B). At higher concentrations, the fluorescence bursts from beads are more frequent as expected (Figure 6A). The burst times are also increased several milliseconds, indicating that cluster formation and simultaneous detection of several beads are possible at this high concentration. A control experiment was performed with a network containing no fluorescent beads shown in Figure 6C. -------------------------------------------------------------------------------- Figure 6 Photodiode detector traces (photon counts versus time) as the fluorescent particles pass through the detector probe volume. The experiment was performed as shown in Figure 5. (A) and (B) show two different concentrations of nanoparticles flowing through the tube. The solutions were diluted from a stock solution of 2% solids. (A) Diluted 100 times. (B) Diluted 1000 times. (C) Control experiment with a flow of buffer solution through the nanotube containing no beads. After sonication and dilution, the solutions were filtered through a 0.2-m PVDF filter to remove large clusters of beads, giving a final concentration of beads that is lower than indicated by the dilution factor. The sampling rate for all traces was 3 ms per channel at a laser power at the objective of 12 W and a detection length of ~2 m. The flow velocity varied between different networks but on average a velocity of 60 m/s was calculated from the traces. -------------------------------------------------------------------------------- The flow velocity was calculated using the burst time and the radius of the probe volume, which was estimated to be 1 m. Only burst times from single beads were counted, and an average velocity of 60 m/s was calculated, however, ranging from 35 to 110 m/s (n = 25). These calculations are in good agreement with calculations made earlier from lipid flow velocity measurements.7 Conclusion and outlook We investigated two different approaches to construct NVNs where the nanotubes reside close to a transparent surface. It was shown that, in the weak-adsorption regime, tubes are placed at the equatorial plane, and this distance gets progressively closer to the surface as adsorption increases. To make tubes reside close to surfaces in cases of low-to-medium adsorption, the size of the anchoring vesicles need to be made small. We show that, by using ethanol-washed coverslips with a relative contact area, Sr*, of 0.07, SBL vesicles can be made reproducibly ~3-4 m in diameter using formation of daughter vesicles from micropipets. The distance of nanotubes to the surface with these vesicles is 1.2-1.6 m. If the distance needs to be shorter, higher adsorption potentials are required and we show surface-adsorbed tubes at submicrometer distances using highly hydrophilic glass surfaces treated with the RCA-1 wash protocol. The main drawback with such high surface adsorption is that it reduces the stability of the networks. Finally, we designed optimized networks with a detection region for LIF detection constituted by a nanotube segment at a short distance between two small vesicles. The other vesicles in the network were made large enough to allow for manipulation, i.e., insertion of electrodes or pipets for sample injection, sample detection, and flow control. Using such networks under fluid control in combination with LIF detection, we demonstrated high-resolution detection of single nanoparticles flowing through lipid nanotubes. With further optimization, this setup should have the potential to detect, for example, single fluorescently labeled polynucleotides or single proteins inside the spatially restricted interior of lipid nanotubes. The extremely reduced dimensions of these systems compared to, for example, microfluidic channels, make them suitable model systems for studies of single-molecule behaviors,1 synchronized population behaviors of single enzymes in confined spaces,21 and diffusible behaviors of molecules in restricted geometries,22,23 These systems may also constitute a platform for performing nanoscale analytical chemistry. In particular, the generic network design described herein can be extended to include secluded compartments defined by one or several vesicles in the network6 where dedicated reactions/interactions are performed on an analyte. For example, enzyme catalysis, binding, or partitioning interactions with functionalized beads, as well as biosensors, including whole-cell bacterial or eukaryotic sensors, can be implemented in the networks. Such entities can be used for detection, counting, or sorting molecules as well as provide for their chemical or physical manipulation in creating, for example, retention mechanisms to separate out two or several dissimilar chemical species or for altering their chemical structure. Such work is currently in progress. Acknowledgment This work was supported by the Royal Swedish Academy of Sciences through a donation by the Wallenberg foundation, the Swedish Research Council (VR), and the Swedish Foundation for Strategic Research (SSF). Prof. E. Evans and Dr. Volkmar Heinrich are gratefully acknowledged for generously sharing their knowledge. * To whom correspondence should be addressed. E-mail: orwar@ phc.chalmers.se. Göteborg University. Chalmers University of Technology. 1. Xie, X. S.; Trautman, J. K. Annu. Rev. Phys. Chem. 1998, 49, 441-480. 2. Lehn, J.-M. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 4763-4768.[Medline] 3. Karlsson, A.; Karlsson, R.; Karlsson, M.; Cans, A.-S.; Stromberg, A.; Ryttsén, F.; Orwar, O. Nature 2001, 409, 150-152.[Medline] 4. Karlsson, M.; Sott, K.; Cans, A.-S.; Karlsson, A.; Karlsson, R.; Orwar, O. Langmuir 2001, 17, 6754-6758.[Full text - ACS] 5. 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. 6. Davidson, M.; Karlsson, M.; Sinclair, J.; Sott, K.; Orwar, O. J. Am. Chem. Soc. 2003, 125, 374-378. 7. Karlsson, R.; Karlsson, M.; Karlsson, A.; Cans, A.-S.; Bergenholtz, J.; Ĺkerman, B.; Ewing, A. G.; Voinova, M.; Orwar, O. Langmuir 2002, 18, 4186-4190. 8. Evans, E.; Bowman, H.; Leung, A.; Needham, D.; Tirell, D.Science 1996, 273, 933-935. 9. Bo, L.; Waugh, R. E. Biophys. J. 1989, 55, 509-517. 10. Helfrich, W. Z. Naturforsch. 1973, 28C, 693-703. 11. Deuling, H. J.; Helfrich, W. J. Phys. 1976, 37, 1335-1345. 12. Moscho, A.; Orwar, O.; Chiu, D. T.; Modi, B. P.; Zare, R. N. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 11443-11447. 13. Criado, M.; Keller; B. U. FEBS Lett. 1987, 224, 172-176.[Medline] 14. Karlsson, M.; Nolkrantz, K.; Davidson, M. J.; Stromberg, A.; Ryttsen, F.; Akerman, B.; Orwar, O. Anal. Chem. 2000, 72, 5857-5862.[Full text - ACS][Medline] 15. Xu, X. H. N.; Yeung, E. S. Science 1998, 281, 1650-1653.[Medline] 16. Evans, E. A. Biophys. J. 1980, 31, 425-432. 17. Bernard, A.-L.; Guedeau-Boudeville, M.-A.; Jullien, L.; di Meglio, J.-M. Langmuir 2000, 16, 6809-6820.[Full text - ACS] 18. Sandre, O.; Moreaux, L.; Brochard-Wyart, F. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 10591-10596. 19. Evans, E.; Needham, D. J. Phys. Chem. 1987, 91, 4219-4228. 20. Bar-Ziv, R.; Moses, E.; Nelson, P. Biophys. J. 1998, 75, 294-320. 21. Stange, P.; Zanette D.; Mikhailov A.; Hess B. Biophys. Chem. 1999, 79, 233-247. 22. Chiu, D. T.; Wilson, C. F.; Karlsson, A.; Danielsson, A.; Lundqvist, A.; Stromberg, A.; Ryttsen, F.; Davidson, M.; Nordholm, S.; Orwar, O.; Zare, R. N. Chem. Phys. 1999, 247, 133-139. 23. Chiu, D. T.; Wilson, C. F.; Ryttsen, F.; Stromberg, A.; Farre, C.; Karlsson, A.; Nordholm, S.; Gaggar, A.; Modi, B. P.; Moscho, A.; Garza-Lopez, R. A.; Orwar, O.; Zare, R. N. Science 1999, 283, 1892-1895.[Medline |
|
Want more information ? Interested in the hidden information ? Click here and do your request. |