Scanning Electroporation of Selected Areas of Adherent Cell Cultures

between different electroporation agents, the method is suitable for design of cell cultures with complex composi- tion. Finite element method simulat...
0 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2007, 79, 4410-4418

Scanning Electroporation of Selected Areas of Adherent Cell Cultures Jessica Olofsson,† Mikael Levin,‡ Anette Stro 1 mberg,‡ Stephen G. Weber,§ Frida Ryttse´n,‡ and ,† Owe Orwar*

Department of Physical Chemistry, Chalmers University of Technology, Kemiva¨gen 10, SE-412 96 Gothenburg, Sweden, Cellectricon AB, Fabriksgatan 7, SE-412 50 Gothenburg, Sweden, and Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260

We present a computer-controlled scanning electroporation method. Adherent cells are electroporated using an electrolyte-filled capillary in contact with an electrode. The capillary can be scanned over a cell culture and locally deliver both an electric field and an electroporation agent to the target area without affecting surrounding cells. The instantaneous size of the targeted area is determined by the dimensions of the capillary. The size and shape of the total electroporated area are defined by these dimensions in combination with the scanning pattern. For example, striped and serpentine patterns of electroporated cells in confluent cultures can be formed. As it is easy to switch between different electroporation agents, the method is suitable for design of cell cultures with complex composition. Finite element method simulations were used to study the spatial distributions of the electric field and the concentration of an electroporation agent, as well as the fluid dynamics related to scanning and flow of electroporation agent from the capillary. The method was validated for transfection by introduction of a 9-base-pair-long randomized oligonucleotide into PC12 cells and a pmaxGFP plasmid coding for green fluorescent protein into CHO and WSS cells. A wealth of methods for accessing, screening, and manipulating the genome and the genetic machinery has evolved during the last 50 years. Through these methods, the entire genomes of several species have been mapped. Today, the introduction of a foreign gene or silencing the expression of the genome is routine in laboratories all over the world. The most commonly used methods for transfection today are based on chemical methods such as lipofection, cationic molecule-mediated transfection, or viral vectors or physical methods such as electroporation. All these methods traditionally target entire cultures of cells. The size of the culture can, however, vary from, for example, the surface area in a 384-well plate to large test tubes. Electroporation is a physical method where cells are exposed to a transient electric field to create pores in the cell membrane. * Corresponding author. E-mail: [email protected]. Fax: +46 31 7723858. † Chalmers University of Technology. ‡ Cellectricon AB. § University of Pittsburgh.

4410 Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

These pores offer routes for introducing many types of molecules, such as plasmids, oligonucleotides, antisense agents, siRNA, proteins, small dyes, and drugs. Traditionally, electroporation is performed on cultures in suspension by plate electrodes at fixed distances. In such electroporation, cells need to be detached by, for example, trypsination, centrifuged and resuspended before and after experiments, leading to poor efficiency and viability. There is a growing interest in the development of electroporation methods for adherent cells, including methods for addressing entire cultures as well as methods for addressing subpopulations or even single cells. The possibility to address subcultures of cells is often crucial when, for example, cell-to-cell variations are investigated. Obviously, chemical transfection methods are less suitable for subculture and single-cell targeting compared to electroporation as it is easier to achieve focused electric fields than solute distibutions. The first single-cell electroporation method targeting adherent cells was reported by Lundqvist et al., using micrometer-sized carbon fiber electrodes positioned by micromanipulators.1 Haas et al. and Rae and Levis have reported on single-cell electroporation methods based on a modified patchclamp technique, where a cell is attached to a micropipet and a substance and an electric field are coadministered to the cell through the pipet.2,3 Nolkrantz et al. used 30-cm-long electrolytefilled capillaries with an inner diameter of some tens of micrometers to codeliver substance and electric field to single or small groups of cells.4 Electroporation methods for targeting larger groups of adherent cells include methods where cells are grown on conducting substrates or on porous membranes and where the electric field is applied normal to the substrate/membrane surface.5-8 In other electroporation methods, electrodes attached to the substrate surface or devices with fixed electrodes are used (1) Lundqvist, J. A.; Sahlin, F.; Åberg, M. A. I.; Stro ¨mberg, A.; Eriksson, P. S.; Orwar, O. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 10356-10360. (2) Haas, K.; Sin, W. C; Javaherian, A.; Li, Z.; Cline, H. T. Neuron 2001, 29, 583-591. (3) Rae, J. L.; Levis, R. A. Pflu ¨ gers Arch. 2002, 443, 664-670. (4) Nolkrantz, K.; Farre, C.; Brederlau, A.; Karlsson, R. I.; Brennan, C.; Eriksson, P. S.; Weber, S. G.; Sandberg, M.; Orwar, O. Anal. Chem. 2001, 73, 44694477. (5) Tomai, E.; Vultur, A.; Balboa, V.; Hsu, T.; Brownell, H. L.; Firth, K. L.; Raptis, L. DNA Cell Biol. 2003, 22, 339-346. (6) Yang, T. A.; Heiser, W. C.; Sedivy, J. M. Nucleic Acids Res. 1995, 23, 28032810. (7) Muller, K. J.; Horbaschek, M.; Lucas, K.; Zimmermann, U.; Sukhorukov, V. L. Exp. Cell Res. 2003, 288, 344-353. (8) Yamauchi, F.; Kato, K.; Iwata, H. Nucleic Acids Res. 2004, 32, e187. 10.1021/ac062140i CCC: $37.00

© 2007 American Chemical Society Published on Web 05/19/2007

Figure 1. (A) Photograph of the experimental setup. With the aid of a micromanipulator, the electroporation device is positioned in a Petri dish containing an adherent cell culture. The cell culture is placed on a computer-controlled motorized scanning stage to enable well-controlled scanning of the capillary over the cell culture. The steel plate and the counter electrodes are connected to a voltage source for pulse application. The substance-filled capillary in the electroporation device is connected to a pump with a tube for control of flow through the capillary. (B) Schematic of the electroporation device. (C) Closeup of the rim of a capillary positioned close to a nonconducting surface with an adherent cell culture. The gap height, i.e., the distance between the capillary rim and the nonconducting bottom plate, is typically 75 µm. Note that the dimensions are not to scale. When an electric field is applied, pores are created in the cells situated under the capillary wall and substance (black dots) enters these cells. The substance is delivered to the electroporation area through creating a flow through the capillary, either before or during electroporation. The flow is symbolized by the black bent arrows.

to apply an electric field tangentially to the surface.9-12 All these methods that target larger groups of cells either require that the cells are cultured on specific substrates/devices or they do not allow for flexible targeting of well-defined subgroups of cells. For high spatial resolution, the targeted cells need to be positioned at specific coordinates on the surface. We here report on a flexible method that can electroporate small circular areas of adherent cells in its stationary mode and define arbitrary patterns of electroporated cells in its scanning mode. The cells do not need to be cultured on specific substrates, and any cell region in a culture can be addressed without affecting surrounding cells. The method takes advantage of the possibility to accomplish local delivery of an electric field as well as an electroporation buffer containing an agent to be internalized with a hollow dielectric structure, previously used by Haas et al., Rae and Levis, and by Nolkrantz et al. to target single or small groups of cells. Local delivery of electroporation buffer and agents is advantageous in reducing sample volumes, and in sparing surrounding cells from buffer exchange and exposure to the agents. Here, we use a capillary equipped with an interior metal electrode. The dimensions of the capillary can be varied but are typically on the order of millimeters. The field-focusing effect of these capillaries close to dielectric surfaces was recently characterized.13 The interior of the capillary is loaded with an electroporation agent, and the rim of the capillary is positioned close to a nonconducting surface that cells are grown on (the bottom of a well in a 384-well plate, a Petri dish bottom, a glass coverslip, etc.). A minute flow of substance from the capillary interior is created by pressure application, and an electrical potential between the interior metal (9) Zheng, Q.; Chang, C. C. Biochim. Biophys. Acta 1991, 1088, 104-110. (10) Lin, Y. C.; Li, M.; Fan, C. S.; Wu, L. W. Sens. Actuators, A 2003, 108, 1219. (11) Teruel, M. N.; Meyer, T. Biophys. J. 1997, 73, 1785-1796. (12) http://www.btxonline.com/. (13) Olofsson, J.; Levin, M.; Stro ¨mberg, A.; Weber, S. G.; Ryttse´n, F.; Orwar, O. Anal. Chem. 2005, 77, 4667-4672.

electrode and an outer counter electrode is applied. As the current passing between the two electrodes has to pass through the small gap between the nonconducting bottom plate and the nonconducting capillary end, a locally enhanced electric field will be achieved under the capillary wall.13 The flow ensures local delivery of substance to the cells in the area exposed to the focused electric field. In stationary mode, the dimensions of the capillary and the voltage settings can be changed to affect the size of the treated cell area in a culture.13 The focusing effect enables targeting of several neighboring areas in one culture with the same or different electroporation agents, either sequentially or in parallel. Further, the possibility to scan the capillary continuously during electroporation allows for targeting variably shaped as well as variably large areas of adherent cells in a culture. EXPERIMENTAL SECTION Cell Culturing. Cells (PC12, NG108-15, WSS, CHO) were cultured in 35-mm cell culture dishes (Petri dishes) coated with poly(L-lysine) (BD Biosciences, Bedford, MA), according to standard procedures in 37 °C, 5% CO2. Cells were cultured for at least 2 days to achieve a dense monolayer of cells. Electroporation Setup and Procedures. Figure 1A shows a photo of the experimental setup and Figure 1B a schematic of the electroporation device comprising a capillary of glass (Polymicro Technologies, Phoenix, AZ). The top end of the capillary is mounted in a well volume that is in contact with a steel plate electrode, and a tube for regulation of substance flow through the capillary. An aluminum cylinder placed around the capillary, ending 3 mm above the outlet of the capillary constitutes a counter electrode (see Figure 1B). The capillary serves both to focus an electric field created between the steel plate and the counter electrode and to deliver electroporation agents. The dimensions of the capillaries that we used varied. In experiments where the capillary was not moved, we used capillaries with an inner diameter of 1 mm and an outer diameter of either 2 or 2.5 mm. Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

4411

In experiments where patterns were created through moving the capillary over the cell culture, a capillary with an inner diameter of 0.4 mm and an outer diameter of 0.8 mm was used. The reason for choosing a smaller capillary in these experiments was to increase resolution as only comparatively small patterns could be visualized using the microscope. The steel plate and the counter electrodes were connected to a Square Wave Electroporation System (ECM 830, BTX, Holliston, MA) for electrical pulse application. The tube was connected to a syringe and an inflate/withdrawal pump (KD Scientific, Holliston, MA) for control of flow through the capillary (see Figure 1A). To reduce lag times in the flow control, the syringe and the tube was filled with water. A small air bubble separated this water from the solution in the capillary. The capillary was always filled with electroporation buffer (Eppendorf, Hamburg, Germany) having a conductivity of 3.5 mS/ cm at 25 °C, supplemented with the substance to be introduced into the cells: 100 µM fluorescein diphosphate (FDP; Molecular Probes, Leiden, The Netherlands), 10 µM Panomer 9 (a randomsequence 9-bp oligonucleotide covalently labeled 5′ with Alexa 546; Molecular Probes), or 2.5 ng/µL DNA plasmid encoding green fluorescent protein (pmaxGFP; Qiagen GmbH, Hilden, Germany). The simplest way to fill the capillary with electroporation agent is to place a droplet of solution on a cover glass and to aspirate it using the attached syringe. Using this procedure, solution switching can easily be accomplished by ejecting the capillary contents and aspirating from a new droplet. For the present setup, the total volume of solution that is required for an experiment is dominated by the volume needed to fill the capillary interior to make electrical contact with the steel plate electrode, ∼80 µL. The amount of solution that is ejected during an experiment is typically much smaller and was no more than 2 µL in any of the experiments described. The electroporation device was mounted on a 3D micromanipulator (Narishigi, MWH-3 or MMN1, Tokyo, Japan) enabling its well-controlled positioning. The capillary orifice was positioned close to a surface containing an adherent cell culture in a Petri dish, so that a thin gap typically some tens of micrometers high was created between the capillary wall and the dish bottom (see Figure 1C). The height of the gap was estimated by changing the focus of the microscope from the cells to the capillary orifice while reading off the corresponding height difference from the marks on the knob for focus adjustements. In order to maintain the same gap height without adjustments during a scanning experiment, the surface that the cells are grown on has to be flat and horizontally arranged. For the experiments presented here, we never experienced a problem with surface variations of the Petri dishes. As a preventative measure to maintain a horizontal surface, the scanning stage where the Petri dish was positioned was kept free of particles. Because the cross-sectional area supporting current flow is the smallest in the narrow gap under the capillary, the local current density, and thus the electric field strength, is maximized there. As a result, a strong electric field will occur under the rim of the capillary, and cells situated there will be electroporated, as illustrated in Figure 1C. In the first series of experiments, the cell culturing medium was replaced by electroporation buffer, as is a standard procedure. 4412

Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

After a while we found that medium exchange in the whole cell culture is not necessary as electroporation buffer is delivered locally through the capillary. Thus, later experiments were performed without this unnecessary procedure. All experiments were performed with the cell culture placed on an inverted microscope (Leica DMIRB, Wetzlar, Germany) equipped with a Hg lamp and a CCD camera (Hamamatsu, Kista, Sweden). The contrast of fluorescence micrographs was enhanced for improved visualization of electroporation results using Adobe Photoshop Software. The microscope was further equipped with a programmable motorized scanning stage (Scan IM 120 × 100, Martzhauser Wetzlar GmbH & Co. KG, Wetzlar-Steindorf, Germany) controlled by a computer. The scanning stage has a travel range of 120 × 100 mm and moves with micrometer precision. Well-controlled scanning of the capillary over the cell area was accomplished through delivering prewritten moving sequences of coordinates and velocities to the scanning stage from the program HyperTerminal (Windows, Microsoft). Coordinates for desired trajectories were achieved by use of the function discretize in the program Origin (OriginLab, Northampton, MA). The inner and outer diameter of the capillary used determines the dimensions of the target area in stationary mode and, in combination with the scanning pattern, the size and shape of the total target area in scanning mode. Limitations on capillary design are set by fabrication methods as well as requirements on the homogeneity of the electric field strength (for details on this issue see ref 13). Further, if the capillary size is decreased to increase spatial resolution, it has to be placed closer to the cells in order to achieve the focusing effect. Therefore, especially in scanning mode experiments, the spatial resolution is limited by the flatness of the surface that the cells are grown on, as well as by the height and flatness of the cell layer. Method evaluation and transfection. Stationary Mode Electroporation. Initially, we used three different cell lines (PC12, NG108-15, WSS) and FDP as electroporation agent. The capillary was positioned at various heights, 50, 75, and 100 µm, above the cells. Pulses with an amplitude between 180 and 250 V and pulse lengths of 1, 10, 25, 50, 75, or 100 ms were tested in the stationary mode (i.e., when the capillary is not moved during electroporation). The results were evaluated optically with bright-field and fluorescence microscopy. In order to keep the number of experimental parameters down, the number of pulses and the resting time between pulses were not altered at this time. In all experiments, 30 pulses with 100-ms interpulse resting time were applied. One hundred milliseconds was the shortest interpulse time possible using our pulse generator. The electroporation agent was delivered to the gap region by applying a flow of 20 µL/min through the capillary for 5 s followed by a 5-s rest before pulse application. To validate the stationary mode of the method for transfection, we introduced the 9-base-pair-long oligonucleotide Panomer 9 into PC12 cells and pmaxGFP plasmid encoding for green fluorescent protein into WSS cells. As in previous experiments, the electroporation agent was delivered to the gap region by applying a flow of 20 µL/min for 5 s followed by a 5-s rest before pulse application. The capillary was positioned 50 µm above the cells, and a pulse train of 30 100-ms pulses of 210 V with 100-ms interpulse time was applied.

Figure 2. (A) Five adjacent circles of cells electroporated with FDP in a confluent cell culture of WSS cells. A stationary mode protocol was run sequentially at five different places with a capillary (0.4-mm inner diameter and 0.8-mm outer diameter) to show that adjacent areas can be independently electroporated. (B) A fluorescent “e” of WSS cells “written” in a confluent cell culture through electroporation with FDP. (C) “Snake pattern” of FDP electroporated into WSS cells. (D) Closeup of part of the “snake pattern” shown in (C). (E) Bright-field image of the area shown in (D). (A-C) are montages of several pictures.

Scanning Mode Electroporation. In scanning mode (i.e., when the capillary is scanned during electroporation), the electroporation agent was continuously fed to the target area at a constant flow rate. In all experiments, the capillary was placed 75 µm above the cell culture. To introduce FDP into WSS cells, we tried a protocol of 50-ms-long, 160-V pulses with 100-ms interpulse time combined with a scan velocity of 1 mm/s and a capillary flow rate of 10 µL/min. With this protocol, different scan patterns were tested. As the protocol turned out to work, no further optimization was performed. To find a suitable protocol for transfection of WSS cells with GFP plasmid, we made a rough optimization. In total, eight straight scans were made in two Petri dishes. The pulse length used was 20 ms with 100-ms interpulse time. The pulse height was either 180 or 200 V. With each pulse height, we tested a capillary scan velocity of 0.1 mm/s combined with a flow rate of 5 µL/min and a scan velocity of 0.5 mm/s combined with a flow rate of 5, 10, or 20 µL/min. These experiments were performed without exchanging culture medium for electroporation buffer. Finite Element Method Simulations. Distributions of electric field, velocity field, and concentration at the outlet of the capillary during a typical stationary mode elctroporation experiment were simulated using the commercial finite element method program FEMLAB 3.0 (Comsol, Stockolm, Sweden). Electric field, velocity field, and concentration simulations for a stationary capillary were performed for a capillary with inner diameter 1 mm, outer diameter 2.5 mm, and distance to the surface of 75 µm. In the simulated experiment, the applied voltage was 200 V and the electroporation agent fed to the gap region through a 20 µL/min flow for 5 s followed by a 5-s rest before electroporation. These simulations were done in cylindrical coordinates utilizing the axial symmetry of the experimental setup. The 2-D geometry used for

the simulations is shown in Figure 5A. We name this geometry 1. The geometry corresponds to a fluid environment consisting of a 6-mm-high fluid column inside the capillary, the gap, and part of the solution bath outside the capillary. Velocity field and concentration simulations for a moving capillary were performed for a capillary with an inner diameter of 0.4 mm, an outer diameter of 0.8 mm, and a gap height of 75 µm. A scan rate of 1 mm/s and a constant capillary flow rate of 10 µL/min were assumed. The modeled experimental situation matches the one for electroporation of WSS cells with FDP in scanning mode. The 3-D simulation geometry is shown in Figure 5E. This geometry is named geometry 2. To make these simulations less memory demanding, symmetry was utilized and only half of the setup simulated. The geometry corresponds to half of the fluid column inside the capillary, half the gap region, and half of the nearby surrounding bath. Note that these simulations were done such that the coordinate system moved with the capillary. Thus, relative movement between the capillary and its surrounding was not accomplished through moving the capillary but through giving the surrounding fluid and the bottom plate with the adherent cell culture a speed of 1 mm/s. Potential and Electric Field. The equation ∇‚(σ∇V) ) 0, where σ is the conductivity and V the potential, was solved for a modified version of geometry 1 (see Figure 5A) where the capillary had been made 19 mm long (which corresponds to its real length), but otherwise identical. The conductivity was set to 3.5 mS/cm. At boundary 1 (see Figure 5A), the potential was set to 200 V. On boundaries 7 and 8 (see Figure 5A), the potential was set to zero. On all other boundaries, electric insulation/ symmetry was used. That the imposed boundaries did not affect the potential in the gap region was tested by varying the size of the “bath”. Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

4413

Figure 3. (A) PC12 cells transfected with the 9-base-pair-long oligonucleotide Panomer 9 covalently labeled with Alexa 546 shown in fluorescence. (B) Bright-field image of the area shown in (A). (C) WSS cells transfected with pmaxGFP plasmid encoding for GFP. The fluorescence micrograph showing positive expression of GFP was taken 24 h after transfection. (D) Bright-field image of the area shown in (C).

Figure 4. Scanning mode transfection of WSS cells with pmaxGFP plasmid encoding for GFP. All pictures are taken 24 h after electroporation. (A) Fluorescence micrograph of a streak of successfully transfected cells. (A) is a montage of several pictures. (B) Closeup of part of the area shown in (A). (C) Bright-field image of the area shown in (B). (D) Fluorescence micrograph of cells electroporated with a scanning mode protocol resulting in cell detachment. (E) Bright-field picture of the area shown in (D).

Velocity Field. To simulate the velocity field, the Navier-Stokes equation

-η∇2u j + F(u j ‚∇)u j + ∇p ) F was solved in combination with the continuity equation

∇‚u j)0 where η is the viscosity of the fluid, uj its velocity, F its density, p the pressure, and F body forces. The equations were solved for both geometry 1 and 2. In the simulations, the body force was 4414

Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

set to zero, the fluid density was set to 103 kg m-3, and the viscosity was set to 1 mPa s. For simulations of velocity profiles when applying a flow of 20 µL/min through a stationary capillary, geometry 1 (see Figure 5A) was used. An inflow of 420 µm/s (corresponds to 20 µL/min for the capillary cross section) was applied on boundary 1, on boundary 10 axial symmetry was applied, and on boundaries 7 and 8, the pressure was set to zero. On all other boundaries, a no-slip boundary condition was used. For simulations of a capillary moving with a velocity of 1 mm/s while a flow of 10 µL/min is applied through it, geometry 2 (see Figure 5E) was used. Note that the simulations were done in a coordinate system moving with the capillary. Due to computer

Figure 5. Finite element method simulations. Left side subfigures are for a stationary capillary with an inner diameter of 1 mm and an outer diameter of 2.5 mm. Right side subfigures are for a scanning capillary with an inner diameter of 0.4 mm and an outer diameter of 0.8 mm. (A) The 2-D geometry used for the simulations of the stationary capillary. We name this geometry 1. (B) Simulation of the electric field in the capillary and under the rim for geometry 1. The electric field is high under the rim, is low inside the capillary, and is virtually zero in the bath volume. (C) The speed profile obtained for a capillary flow of 20 µL/min for geometry 1 (stationary capillary). (D) The distribution of substance after 5 s of a capillary flow of 20 µL/min followed by 5 s of diffusion for geometry 1 (stationary capillary) (E) The 3-D geometry used in simulations of a moving capillary. We name this geometry 2. (F) Speed profiles for the planes y ) 5 µm and x ) 0 for a moving capillary (geometry 2). Note that speeds are plotted relative to the nonmoving bottom plate although the simulation was done in a coordinate system moving with the capillary. (G) The substance distribution in the plane z ) 0.5 µm for the case of a moving capillary (geometry 2).

memory limitations, the geometry could not be made larger. The following boundary conditions were used: On boundary 1, an inflow velocity of 1.33 mm/s (corresponds to a total flow of 10 µL/min through the capillary), and on boundaries 2, 4, 5, and 6 a flow velocity in the positive x direction of 1 mm/s. On boundary 7, a symmetry boundary condition was used, and on boundary 3, a neutral boundary condition setting transport by shear stress to zero across the boundary and placing no constraint on the velocity across it. On all other boundaries, the no-slip boundary condition was used. The boundary settings on boundaries 2, 3, 5, and 6, in combination with the limited size of the geometry, possibly result in a minor increase of the fluid flow in

the positive x direction in the gap region compared to the real case. Concentration Distribution. To obtain the concentration profile of FDP in the cell bath the equation,

dc/dt ) ∇‚(-D∇c + cu j) where c is the concentration and D is the diffusion coefficient, was solved utilizing the simulated velocity profiles. To obtain the concentration distribution at the moment of electroporation for a stationary capillary, the simulation was divided into two parts. In the first part, corresponding to 5 s of flow application, the velocity Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

4415

field from the simulation using geometry 1 was used. In the second part, corresponding to the following 5 s of rest, the convection was set to zero and only diffusive mass transport considered. The following boundary conditions were used: On boundary 1, the concentration was set to 1, on boundary 7 and 8, a boundary condition implying mass transport only through convection (i.e., diffusive mass transport is neglected) was used, and on boundary 10, axial symmetry was used. On all other boundaries, an insulation boundary condition was used. The diffusion coefficient was set to 4.4 × 10-10 m2/s. As initial conditions (time 0), the concentration inside the capillary was set to 1 and elsewhere to 0. For the case of a moving capillary, a steady-state concentration distribution is achieved in the vicinity of the capillary if a constant scanning velocity is used. This is true as transport of agent by diffusion there is neglible compared to the convective transport for the used scanning velocities and flow rates. Thus, a steadystate concentration was modeled and dc/dt set to zero. Note that it is the concentration as viewed in a coordinate system following the moving capillary that is modeled. The following boundary conditions were used (geometry 2): On boundary 1, the concentration was set to 1, and on boundaries 2, 5, and 6, the concentration was set to 0. On boundaries 4 and 7, isolation/ symmetry was used, and on boundary 3, a boundary condition implying mass transport only through convection was used. As there might be a slightly increased directioning of the flow in the positive x direction in the gap region, these simulations gives the lower limit of the gap region area exposed to electroporation agent. For mathematical details, we refer to FEMLABs handbook or their homepage (www.comsol.com). RESULTS Electroporation experiments with FDP. Stationary Mode. Initially, experiments were performed with FDP as electroporation agent. FDP is only weakly fluorescent until cleaved by phosphatases in the cytoplasm to form fluorescein. Figure 2A shows a fluorescence micrograph of five adjacent circles of WSS cells that have internalized FDP. The circles were created by running an electroporation protocol at five different locations. The picture was taken ∼5 min after electroporation. The focusing effect of the electric field under the rim of the capillary causes the ringshaped area of electroporated cells (see below for further details). The smaller the gap between the capillary rim and the bottom surface is, the more focused the electric field becomes. During the experiments where different parameters (cell type, gap height, pulse amplitude, pulse length) were varied, we made the following observations: All tested pulse lengths, with the exception of 1-ms pulse lengths, resulted in electroporation. For 1-ms pulses, very low yields were obtained independently of other parameters. For gap heights of 50 and 75 µm, there were no notable qualitative difference. However, for our setup and small gap heights, the electric field strength under the capillary roughly scales inversely proportional with the height of the gap. Consequently, for otherwise identical settings, cells were more heavily electroporated when using a 50-µm gap height than a 75-µm gap height. For gap heights of 100 µm and larger, the ring-formed areas of electroporated cells were not well-defined, and the electroporation yield in general very low. However, by increasing the pulse amplitude, good electroporation results, 4416

Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

although with a less well-defined target area, can probably be obtained for these large gap heights. We found it advantageous to work with a 75-µm gap height compared to a 50-µm gap height for two reasons. First, the risk to bump into groups of cells floating just above the surface is smaller, and second, as the electric field strength in the gap roughly scales with the inverse of the gap height, the smaller the gap, the more sensitive the setup becomes to errors in capillary placement. In general, as the gap height decreases shorter pulse lengths and lower pulse amplitudes are required to obtain the same degree of electroporation. Narrow gaps in combination with long-duration, high-amplitude pulses resulted in morphological changes of cells situated under the capillary rim, in particular close to the center opening. On the other hand, using pulses that are too short and at a too low pulse amplitude for the chosen gap, only cells situated close to the center opening are electroporated. Using a gap height of 75 µm, we found that 200-225-V pulses of 10-25-ms duration gave good electroporation yields over the entire target area without resulting in morphological changes of the cells. Scanning Mode Electroporation. Ring-formed electroporation patterns as those shown in Figure 2A, are not always ideal. For example, it may be beneficial to create line patterns along cell laminae in, for example, brain tissue slices or homogeneous electroporation of subcolonies or larger areas of cultured cells. To achieve these goals, we developed a computer-controlled scanning electroporation method. By placing the cell culture dish on a computer-controlled scanning stage, the capillary could be moved over the cell culture during pulse application so that cells were electroporated “on the fly”. As the stage has sub-micrometer resolution in the x-y direction, and is controlled by a computer, electroporation patterns are well-defined and can be automatically executed. In principle, the capillary is used as a pencil to draw a pattern delivered by the computer in the cell culture. Figure 2B and C shows an “e” and a “snake” pattern drawn in confluent cell cultures in this way through electroporation with FDP. In Figure 2D and E, closeups of the snake pattern are shown in fluorescence and bright-field microscopy, respectively. In scanning mode, the number of pulses experienced by a cell passing under the capillary varies with the distance to the capillary center line and is further determined by the pulse frequency in combination with the scan velocity and the dimensions of the capillary. In the experiments where an “e” and a “snake” pattern were created, the maximum number of pulses experienced by any cell was five. To change the number of pulses experienced by each cell, one can either alter the pulse frequency or alter the scan speed and the capillary wall thickness. Transfection. Nucleotide transfection was first performed in the stationary mode by using a 9-base-pair-long randomized oligonucleotide named Panomer 9. Panomer 9 covalently labeled at the 5′-end with Alexa 546 was electroporated into PC12 cells. Figure 3A shows a fluorescence micrograph of a ring of cells that have internalized such oligonucleotides, and in Figure 3B, the corresponding bright-field microscopy image is shown. The images were taken 2 min after electroporation. Further, using the stationary mode, a pmaxGFP plasmid coding for green fluorescent protein was successfully transfected into WSS cells. Positive expression of GFP was detected 24 h after transfection, as shown

in Figure 3C. The corresponding bright-field image is shown in Figure 3D. For transfection in the scanning mode, we used the pmaxGFP plasmid. In total, eight different protocols were tested to find a suitable combination of pulse settings, capillary flow rate, and scan speed (for details, see Experimental Section). Both protocols (180and 200-V pulse amplitude) combining a scan speed of 0.5 mm/s and a flow rate of 0.5 µL/min worked well. In Figure 4A, a trace of WSS cells expressing green fluorescent protein obtained with the 180 V protocol is shown. The image was taken 24 h after transfection. The length of the trace (5 mm) was limited by the maximum number of 99 pulses that the pulse generator can deliver in series. In Figure 4B, a closeup of the electroporated trace is shown. Figure 4C shows the corresponding bright-field picture. It has been shown that the number of pulses in general affects transfection yields considerably and that, as a general rule of thumb, the highest transfection yields are obtained for many long pulses at relatively low field strengths.14 In the experiment presented in Figure 4A and B, the maximum number of pulses experienced by any cell was 12, which is fairly low in this context. Thus, it can be anticipated that the transfection yield can be considerably improved by optimizing the pulse/scan protocol so that the number of pulses is increased. Further, the need to expose cells to sufficient pulses imposes an upper limit on what scan speeds can be used. Protocols combining a scan speed of 0.1 mm/s, and a flow rate of 0.5 µL/min, as well as protocols combining a scan speed of 0.5 mm/s with flow rates of 10 and 20 µL/min resulted in too much stress on the cells. As a result, streaks where cells had detached from the surface were observed after 24 h. In Figure 4D and E, part of the electroporation trace resulting from a pulse amplitude of 200 V, and a combination of a scan speed of 0.5 mm/ s and a flow rate of 20 µL/min, is shown in fluorescence and bright-field microscopy, respectively. As mentioned, a scan rate of 0.5 mm/s in combination with a flow rate of 5 µL/min worked well. However, keeping a scan rate of 0.5 mm/s but increasing the flow rate challenged the cells and yielded poor results. This shows that the flow rate affects the optimal choice of the other parameters. It appears that shear stresses induced by the scanning movement of the capillary and the fluid flow through it results in a shift of optimal pulse protocols toward milder ones. This is in line with other studies reporting on facilitated electroporation when adding a mechanical stress to the cell membrane during electric field exposure.15,16 Finite Element Method Simulations. We simulated the electric field, the fluid flow, and the concentration of electroporation agent in the gap region using the computer program FEMLAB. Figure 5A shows the geometry used for simulations of stationary capillaries in cylindrical coordinates. Figure 5B shows a simulation of the electric field intensities created when a 200-V pulse is applied to a 19-mm-long capillary (1-mm inner diameter, 2.5-mm outer diameter) positioned 75 µm above the insulating surface of the cell dish. The electric field intensity is highest in the gap and virtually zero outside the capillary, which explains the ring-shaped electroporation patterns. A more extensive (14) Canatella, P. J.; Prausnitz, M. R. Gene Ther. 2001, 8, 1464-1469. (15) Zhelev, D. V.; Needham, D. Biochim. Biophys. Acta 1993, 1147, 89-104. (16) Barrau, C.; Tessie´, J.; Gabriel, B. Bioelectrochemistry 2004, 63, 327-332.

characterization of the local electric field enhancement can be found in a previous article.13 Figure 5C shows the fluid speed profile during a 20 µL/min flow through the capillary. As can be seen, high fluid velocities are obtained just under the capillary as fluid is pushed through the gap. This is also the only region where high velocity gradients are found close to the bottom surface. The magnitude of shear stress felt by a surface exposed to fluid flow is proportional to the fluid velocity gradient normal to the surface.17 Thus, the simulation tells us that shear stresses induced on cells by fluid flow is high only for cells situated under the capillary rim. In Figure 5D, the simulated concentration distribution of an electroporation agent (diffusion coefficient chosen as for FDP, i.e., 4.4 × 10-10 m2/s) at the moment of electroporation (i.e., after 5-s application of a flow of 20 µL/min followed by a 5-s break) is shown. The simulation shows that there is a high concentration of agent inside the capillary and in the gap region. From the simulations, we draw the conclusion that the exposure to high electric fields as well as shear stress is limited to cells situated under the rim of the capillary. These cells are further, together with the cells situated under the center of the capillary, the only cells exposed to high concentrations of an electroporation agent. Thus, two reference areas to the electroporated area are automatically achieved. These are the area in the middle of the ring (where cells have been exposed to electroporation agent but neither to a high enough electric field to create pore formation nor high shear stress) and the area outside the ring where cells have not been exposed to the electroporation agent, the shear stress, or a permeating electric field. Figure 5E shows the geometry used for simulations of a scanning capillary. In scanning mode, a constant flow was applied during electroporation. We simulated the fluid movement resulting from a flow of 10 µL/min through the capillary and a simultaneous 1 mm/s scanning movement. In Figure 5F, the speed profile relative the bottom plate of the cell dish is shown for the planes y ) 5 µm and x ) 0. From this simulation, we could conclude that high-velocity gradients close to the bottom surface (z ) 0) are uniquely found in the gap region. We also simulated the steady-state concentration of electroporation agent (diffusion coefficient chosen as for FDP, i.e., 4.4 × 10-10 m2/s) around a capillary moving with a velocity of 1 mm/s while an agentcontaining solution is fed through it with a rate of 10 µL/ min. In Figure 5G, the steady-state concentration of electroporation agent in the gap region for the plane z ) 0.5 µm (i.e., 0.5 µm above the bottom surface) is shown. Only regions that are under the capillary rim or, alternatively, regions that historically spent some time under the capillary rim, experience high concentrations of FDP. Due to the scanning movement, a high concentration of FDP never reaches the front side of the capillary delivering the first electrical pulses to the cells. From the simulations, we conclude that in the scanning mode, as in the stationary mode, only cells passing under the capillary will experience high shear stress, high electroporation agent concentration, and high electric field strengths. Surrounding cells will be virtually unaffected by the electroporation procedure. (17) Vennard, J. K.; Street, R. L. Elementary Fluid Mechanics; John Wiley & Sons; New York, 1982.

Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

4417

CONCLUSIONS We have developed a scanning electroporation method for adherent cells. The method takes advantage of the possibility to deliver a locally strong electric field as well as an electroporation buffer, containing an agent, with the same capillary. As the cells outside the targeted area are not affected and the electroporation agent easily can be switched, several independent experiments can be performed in the same cell culture. By moving the capillary to different areas of a cell culture, multiple adjacent or distant spots can be electroporated independently. Further, by moving the capillary over the cell culture during pulse application, cells can be electroporated “on the fly”. In this way, areas of arbitrary shape can be targeted. Having these properties, the method should be suitable for design of nonhomogeneous adherent cell cultures. Further, the method has advantages of being easy to implement and having fairly low substance consumption (∼80 µL). In stationary mode, the method automatically provides three reference points within the same culture: cells exposed to neither electric field nor electroporation agent, cells exposed only to the electroporation agent, and cells exposed to both electroporation agent and electric field. We demonstrate introduction of FDP,

4418

Analytical Chemistry, Vol. 79, No. 12, June 15, 2007

transfection of a 9-base-pair-long randomized oligonucleotide (Panomer 9), and plasmids coding for pmaxGFP into cells. This electroporation tool should be suitable for introduction of any type of agent such as fluorescent dyes, oligonucleotides, siRNA, antisense, plasmids, and proteins into any type of adherent cells such as, for example, primary cells, tissue slices, or cells considered hard to transfect with common technology. ACKNOWLEDGMENT We express our gratitude to Irene Lindqvist for help with cell culturing and Daniel Fagerlund for help with images and illustrations. The work was supported by the Royal Swedish Academy of Sciences, the Swedish Research council (VR), the Swedish Foundation for Strategic Research (SSF) through a donation from the Wallenberg Foundation, the Go¨ran Gustafsson Foundation and the National Institutes of Health through R01 GM 66018.

Received for review November 13, 2006. Accepted March 19, 2007. AC062140I