Local Regional Stimulation of Single Isolated ... - ACS Publications

Jul 2, 2009 - To test this hypothesis, the clearance of fluorescently labeled 10 kDa dextran from the t-system was measured and found to be ∼0.5 s d...
0 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2009, 81, 6390–6398

Local Regional Stimulation of Single Isolated Ventricular Myocytes Using Microfluidics Norbert Klauke,*,† Godfrey Smith,‡ and Jonathan M. Cooper† Department of Electronics, University of Glasgow, Glasgow G12 8LT, and Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow G12 8QQ The regional manipulation of the microenvironment surrounding single adult cardiac myocytes in a microfluidic structure is described. The flow rates of laminar streams were adjusted such that the fluid interface between an injection flow and a perfusion flow was manipulated laterally to stimulate regions of the cell surface. Using this general principle, we were able to selectively expose defined regions of the cell to test solutions, with predefined pulse durations and frequencies. We demonstrate the transient depolarisation of the cardiomyocyte through the regional chemical stimulation of localized areas of the cell with elevated potassium concentrations (100 mM). The results show that chemical stimulation at frequencies e0.25 Hz evoked Ca2+ transients and cell shortening, comparable to those induced by electrical (field) stimulation. At higher frequencies the membrane potential failed to recover sufficiently from the depolarisation with the high K+ solution, possibly because of the slow clearance of the ion from the t-tubular system. To test this hypothesis, the clearance of fluorescently labeled 10 kDa dextran from the t-system was measured and found to be ∼0.5 s delayed compared to that of the bulk extracellular space, indicating the slow diffusion inside the confined space of the tubular membrane invaginations. INTRODUCTION Unique physical features, including those of viscous flow, short diffusion lengths, and high surface tension often dominate the behavior of fluids at the microscale. These have been exploited in a variety of microsystem technologies, e.g., microfluidic devices, to access, control, and study the microenvironment of single cells in which metabolites and gas are transported through diffusion.1,2 The flow conditions at low Reynolds numbers (where inertial forces are minimal) provide unprecedented spatial and temporal control over fluids entering and leaving an experimental system, e.g., a single cell within a microchannel.3 Previously, laminar flow has been used for the regional manipulation of the extracellular environment enabling the biophysical investigation of intracellular * Corresponding author. † Department of Electronics. ‡ Institute of Biomedical and Life Sciences. (1) Yu, H.; Meyvantsson, I.; Shkel, I. A.; Beebe, D. J. Lab Chip 2005, 5, 1089– 1095. (2) Folch, A.; Toner, M. Annu. Rev. Biomed. Eng. 2000, 2, 227–2256. (3) Sims, C. E.; Allbritton, N. L. Lab Chip 2007, 7, 423–440.

6390

Analytical Chemistry, Vol. 81, No. 15, August 1, 2009

ion diffusion in freshly isolated cardiomyocytes.4,5 In a complementary study, rapid solution switching around separated regions of the cell has been used to evoke nonpropagating and propagating Ca2+ release through local chemical stimulation with caffeine and elevated [K+]e, respectively.6 These microfluidically controlled manipulations of the extracellular solution have also enabled the indirect measurement of diffusion coefficients of various molecular species at close proximity of the cell surface and indeed inside the cell.7-9 Within the heart cell, the complex structural architecture rather than the viscosity alone contribute to the mass transfer of molecules, e.g., insulin10 or myoglobin.11 This is illustrated by the effect of the process of convection, enhancing diffusion during the contraction cycle of the stimulated muscle, and revealing increased intracellular transport rates compared to quiescent muscle.12 The intracellular space (sarcoplasm) of the adult ventricular myocyte is densely packed with contractile filaments (myofilamental lattice) and with cell organelles including mitochondria, cell nuclei and the sarcoplasmic reticulum (SR). Additionally, the cell membrane (sarcolemma) invaginates into the sarcoplasm to form a network of transvers and longitudinally oriented tubules (t-tubular system) which is in continuum with the extracellular space.13 The complex architecture of this highly structured intracellular matrix exhibits rhythmic dynamics due to the contractile activity which affects mass transfer.14 The lower diffusion rates in the sarcoplasm compared to the aqueous solution are reflected in the size-restrained aqueous diffusion pathways (e11 nm) of the sarcoplasm, but are augmented by convective flow from cell contraction.15 For example, the effect on excitationcontraction coupling of the different diffusion coefficients of ion (4) Swietach, P.; Leem, C. H.; Spitzer, K. W.; Vaughan-Jones, R. D. Biophys. J. 2005, 88, 3018–3037. (5) Swietach, P.; Spitzer, K. W.; Vaughan-Jones, R. D. Biophys. J. 2008, 95, 1412–1427. (6) Klauke, N.; Smith, G. L.; Cooper, J. M. Anal. Chem. 2007, 79, 4581–4587. (7) Yao, A.; Spitzer, K. W.; Ito, N.; Zaniboni, M.; Lorell, B. H.; Barry, W. H. Cell Calcium 1997, 22, 431–438. (8) Takayama, S.; Ostuni, E.; LeDuc, P.; Naruse, K.; Ingber, D. E.; Whitesides, G. M. Nature 2001, 411, 1016–1016. (9) O’Neill, S. C.; Donso, P.; Eisner, D. A. J. Physiol. 1990, 425, 55–70. (10) Lauritzen, H. P. M. M.; Ploug, T.; Prats, C.; Tavare, J. M.; Galbo, H. Diabetes 2006, 55, 1300–1306. (11) Papadopoulos, S.; Endeward, V.; Revesz-Walker, B.; Juergens, K. D.; Gros, G. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 5904–5909. (12) Wu, P. I.; Edelman, E. R. J. Biomech. 2008, 41, 2884–2891. (13) Soeller, C.; Cannell, M. B. Circ. Res. 1999, 84, 266–275. (14) Papadopoulos, S.; Jurgens, K. D.; Gros, G. Biophys. J. 2000, 79, 2084– 2094. (15) Parfenov, A. S.; Salnikov, V.; Lederer, W. J.; Lukyanenko, V. Biophys. J. 2006, 90, 1107–1119. 10.1021/ac9008429 CCC: $40.75  2009 American Chemical Society Published on Web 07/02/2009

species, e.g., K+ or Ca2+ in different parts of the t-tubular system and the SR with their different tortuosities has been recently investigated on a cellular level to explore the contributions from diffusion and convection.16-19 This paper describes a microfluidic device for the regional superfusion and chemical stimulation of arrays of adult ventricular myocytes (with a flow arrangement that resembles the macroscale H-cell used for fast solution switching in traditional patch clamp experiments).20,21 The architecture of our microscale cell investigation chamber enables easy cell loading and local solution changing, using a micropipet. We show how hydrostatic pressure can be used to establish the laminar flow of buffer and test solutions, perpendicular to the longitudinal axis of the cardiomyocyte. The position of the interface between these two input flows (in each of the input arms of the H junction) can be adjusted by pressure driven flow to expose selected, separate local regions of the cell surface to different test fluids. The implementation of a microfluidic device provides precise control over the distribution of fluids around the cell. The inputs and outputs of the microfluidic cell were integrated onto the footprint of a 96 well plate. The effect of the pressure changes on the position of the laminar flow interface was corroborated by a simple finite element model of a modified H-cell matching the geometries of the physical device, using similar simulation techniques to those already published for model systems.22 Experimentally, we use the Ca2+ transient of the stimulated cardiomyocyte, and the loading of the cell with fluorescent dyes, to explore the effect of the regional chemical manipulation. The potential of this new method of cell excitation for the study of drug effects is demonstrated. In summary, our paper presents data showing the regional suffusion of single isolated ventricular myocytes using microfluidic chambers with integrated superfusion channels. The microfluidic flow cell combines the continuous superfusion of the cardiomyocyte with the periodical (fmax ∼0.2 Hz) focused chemical stimulation using elevated K+, ([K+]e ) 100 mM). A pipet-based regional injection system was also implemented to investigate the diffusion of small dye molecules in and out of the t-tubular system of the adult ventricular myocytes. Molecules smaller than ∼1 kDa were observed to move in and out of the transverse t-tubules in less than 0.5 s. Hydrostatic pressure injection was integrated and used for chemical stimulation producing the same results compared to a macroscale micropipet injection system. The ease of operation and robustness of the device, together with the simple fabrication method make this disposable flow cell an attractive superfusion system for single isolated cardiomyocytes. EXPERIMENTAL SECTION Cell Isolation. Experiments were performed using enzymatically isolated rabbit ventricular myocytes. Hearts were removed (16) Shepherd, N.; McDonough, H. B. Am. J. Physiol. 1998, 275, H852–H860. (17) Swift, F.; Stromme, T. A.; Amundsen, B.; Sejersted, O. M.; Sjaastad, I. J. Appl. Physiol. 2006, 101, 1170–1176. (18) Zima, A. V.; Picht, E.; Bers, D. M.; Blatter, L. A. Circ. Res. 2008, 103, e105–e115. (19) Edwards, J. N.; Launikonis, B. S. J. Physiol. 2008, 586, 5077–5089. (20) Krishtal, O. A.; Pidoplichko, V. I. Neuroscience 1980, 5, 2325–2327. (21) Fenwick, E. M.; Marty, A.; Neher, E. J. Physiol. 1982, 331, 577–597. (22) Brody, J. P.; Yager, P. Sens. Actuators, A 1997, 58, 13–18.

from terminally anaesthetised rabbits (1 mg kg-1 euthatol). Myocytes were isolated from the left ventricle by perfusion with collagenase solution and kept in Base Krebs Solution, containing 120 mM NaCl, 20 mM sodium N-hydroxyethylpiperazineN’-2-ethane sulfonic acid (HEPES), 5.4 mM KCl, 0.52 mM NaH2PO4, 3.5 mM MgCl2. 6H2O, 20 mM taurine, 10 mM creatine, 11.1 mM glucose, 0.1% BSA, 0.1 mM CaCl2, pH adjusted to 7.4 with 100 mM NaOH.23 For local chemical stimulation with elevated potassium the Base Krebs Solution was modified to include 100 mM KCl. The increased osmolality of the buffer solution did not affect the cell volume, e.g., no regional shrinkage of the cardiomyocyte was observed after the regional stimulation with the high potassium solution. All solutions were filtered (200 nm pore size) to remove particles, which would otherwise block the small channels. Fabrication. The multilayered microfluidic device was fabricated through replica molding of the silicon rubber, poly(dimethylsiloxane), PDMS (Sylgard 184 Silicone, Dow Corning, VWR International, Lutterworth, UK) using a sacrificial master of AZ resist, as previously described.24 In detail, a ∼50 µm thick film of positive photoresist (AZ4562, Clariant, Glattbrugg, Switzerland) was spin-coated at 500 rpm for 30 s onto a clean wafer of thin glass (170 µm thick, H. V. Skan Ltd., Shirley, UK), oven-baked at 90 °C, and then patterned through multiple exposure at different doses to produce features of different heights.25 Before UV-light exposure the wafer was trimmed with a diamond knife to remove the bead and then diced into rectangular pieces (∼2 × 3 cm). The first pattern (channel lid, Figure 1A) was transferred (∼10 µm depth) into the photoresist with the chrome mask in hard contact with the photoresist (dose 7.2 W/cm2 for 10 s). The second pattern (channel, Figure 1B) was aligned to the first pattern using backside mask alignment (trans-illumination on MA6, SUSS MicroTec Ltd., Metheringham, UK) and then transferred into the entire thickness of the film (7.2 mW/cm2 for 60 s). The substrate was then brought into contact with the chrome mask (inverted orientation of the substrate) and the third pattern (gap, Figure 1c) was aligned and then transferred ∼10 µm deep into the film through the transparent glass substrate at a lower exposure dose (proximity exposure, proximity gap ∼170 µm, 7.2 W/cm2 for 5 s). The diffraction at the edges of the gap pattern generated a half dome relief structure in the photoresist film. After the three exposure steps, the pattern was developed (AZ400K for ∼7 min, Figure 1D) and the height of the different features was measured using a Dektak for surface scanning (Veeco) or a confocal microscope (Zeiss LSM 510). The patterned film was then used as a “sacrifical” master for the molding of the silicon rubber (PDMS, diluted 1:4 in Hexane). Precured PDMS was applied to selectively cover the lower level features (channel) but not the higher level features (inlet, outlet, cell chamber) with a micropipet mounted on a three axis positioner (Micromanipulator Leica, Wetzlar, Germany). The silicon rubber was oven-cured at 90 °C for 2 min and then soaked in acetone to remove the photoresist (“sacrificial” master, Figure 1E). Any residual photoresist was removed using brief periods (∼30 s) of agitation in an (23) Eisner, D. A.; Nichols, C. G.; O’Neill, S. C.; Smith, G. L.; Valdeomillos, M. J. Physiol. 1989, 411, 393–418. (24) Klauke, N.; Smith, G. L.; Cooper, J. Biophys. J. 2009, submitted. (25) Daly, D.; Stevens, R. F.; Hutley, M. C.; Davies, N. Meas. Sci. Technol. 1990, 1, 759–766.

Analytical Chemistry, Vol. 81, No. 15, August 1, 2009

6391

Figure 1. Sketch illustrating three consecutive exposures of the photoresist to generate a multilayered pattern. Note that the pattern of the third mask (Figure 1C) was transferred into the photoresist from the back, e.g., transillumination through the transparent substrate. The increased distance (170 µm, thickness of the glass substrate) between the photoresist and the chrome layer of the mask generated rounded edges due to the increased diffraction error during the proximity exposure.

ultrasonic bath in acetone. Dams of adhesive sealant (3145 RTV, Dow Corning Ltd., Allesley, UK) were formed around the microfluidic structures (inlet/outlet and the chamber) to prevent the flow of the precured PDMS (used for bonding to multititer plate) into the channels. The wafer was then aligned and sealed to the underside of a 96 well multititer plate (without base) with a thin layer of PDMS applied to the bottom side of the plate. Operating the Flow Cell. Three wells (inner diameter 6.8 mm) of a 96 well microtiter plate (Nunc GmbH & Co. KG, Langenselbold, Germany) were aligned to the inlets, the outlets and the cell chamber, respectively26 (Figure 2A). The experimental results presented focus on studies at one (proximal) end of the cell, which is addressed by half of the H cell (i.e., with one inlet and one outlet). The two wells above the inlet and the outlet were used to accommodate the soft elastic tube adapter (machined from thick walled silicon tubing, outer diameter 7.3 mm, inner diameter 3.1 mm; Cole-Parmer Instrument Co. Ltd., London, UK) for reversible sealing of a length of hard inelastic tubing (thick walled PTFT tubing, outer diameter 3.2, inner diameter 1.6; Bio-Chem Fluidics, Cambridge, UK) to the inside of each well. A three-way valve connected either a syringe pump or a reservoir to both the inlet and the outlet of the flow channel. The height of the inlet reservoir was vertically adjusted between 0 and 30 cm above the microscope stage to generate positive hydrostatic pressure and (26) Khine, M.; Ionescu-Zanetti, C.; Blatz, A.; Wang, L.-P.; Lee, L. P. Lab Chip 2007, 7, 457–462.

6392

Analytical Chemistry, Vol. 81, No. 15, August 1, 2009

the height of the outlet reservoir between 0 and 30 cm below the microscope stage to generate negative hydrostatic pressure using custom build sliders (Figure 1A). Prior to cell loading, a syringe pump was used to wet the channel and the open cell chamber from both ends with the buffer solution. The open cell chamber was then sealed with an overlayer of perfluorocarbon oil (PP2 from Flutec, UK) to prevent the escape of the buffer solution into the well. Single cells were loaded by micropipetting into the microchannel, relying upon self-alignment to complete the process.27 A drop of buffer solution including a single ventricular myocyte was placed on top of the open chamber array and once the cardiomyocyte settled into an individual chamber the excess solution was removed. The flat cardiomyocytes were partially drawn into the outlet channel under negative pressure (Figures 3 and 4), against fabricated dams to hold the cell in position.This arrangement allowed for the unrestricted access from above through the fluidic lid, to pipet a cardiomyocyte into the capped chamber and to add different fluids to chosen regions of the cell through micropipets inserted into the chamber through the oil. After wetting the flow channels and the chamber, the valves at the inlet and the outlet were switched from the syringes to the reservoirs generating the hydrostatic pressure. The fluorescent dye fluorescein was injected to measure the flow velocity on confocal line scan images of the fluorescence transients inside the channel (Figure 2B). Injection was performed as previously described.6 Briefly, a micropipet (tip diameter ∼3-10 µm) was mounted on a micromanipulator and pressurized using a syringe pump. The pressure was adjusted to a level sufficient for the ejection of the physiological buffer solution into the microchannel. The micropipet tip was then vertically stepped through the oil/ aqueous interface into the buffer solution surrounding the cardiomyocyte using a one axis piezo translator (Physik Instrumente Darmstadt, Germany). A linear relation between the pressure head (height of column of fluid) and velocity and was measured over a velocity range from 0-3.5 mm/s and a pressure head of 0-30 cm (Figure 2C). During calibration both the positive pressure at the inlet and the negative pressure at the outlet were changed with equal amounts. A maximum flow velocity of ∼2 mm/s was used which was sufficient to wash the entire chamber within 580) and dextran-coupled Fluo-3 (10 kDa, ex 488/em >515) inside the t-tubular system of single ventricular myocytes was imaged on confocal sections

and line scan images of the cell border (510 Meta, Zeiss, Germany). RESULTS Device Design. Previously, we have fabricated a two-layer open architecture microfluidic device for the electrical stimulation of cardiac myocytes using replica molding against a two-layered SU-8 master.28 In this paper we now report upon the fabrication of a sacrificial multi layered master, adapted from our previous work but based on the multiple exposure of a thick photoresist25,29 (Figure 1). The replica-molded double-layered ∼50 µm thick PDMS film consisted of arrays of 10 open-air rectangular fluidic chambers (40 × 200 µm, 50 µm high, 70 µm pitch) arranged in groups of three and 9000 µm apart from each other (on the pitch of a 96-well plate). The two outer arrays were for the inlet/outlet connection and the central one for the cell investigation chambers. Connecting flow channels (40 × 9000 µm, ∼40 µm high with a ∼10 µm thick lid) were designed. As stated, this pattern matched (28) Klauke, N.; Smith, G. L.; Cooper, J. IEEE Trans. Biomed. Eng. 2005, 52, 531–538. (29) Hirai, Y.; Inamoto, Y.; Sugano, K.; Tsuchiya, T.; Tabata, O. J. Micromech. Microeng. 2007, 17, 199–206.

Analytical Chemistry, Vol. 81, No. 15, August 1, 2009

6393

Figure 3. Ca2+ transients stimulated through the transient elevation of the [K+]e. A Fluo-3 loaded cardiomyocyte was continuously superfused with Tyrode solution at ∼2 mm/s and periodically stimulated with 100 mM [K+]e. The dipping micropipet was positioned upstream at the edge of the perfusion chamber as indicated (* in Figure 3B). Ca2+ imaging revealed the synchronized Ca2+ release throughout the entire myoplasm and subsequent cell shortening, indicated through the moving edge of the right cell end (dotted lines in Figure 3B). The plot of Fluo-3 fluorescence versus time of the Fluo-3 emission recorded from the cell center revealed the prolongation of the Ca2+ transients and a gradual raise of the diastolic [Ca2+]i (Figure 3A).

Figure 4. Probing the access to the t-tubules. A ventricular myocyte was continuously superfused at 2 mm/s and one-half of the cell was exposed to a fluorescent dye (SNARF) to examine the diffusion into the t-tubules. Moving the dye-loaded micropipet moved the border between the fluorescent/nonfluorescent solution (Figure 4A, 18.6 s and 45.1 s). Note that the dye entered the t-tubules only distal (downstream) to the pipet tip dipped into the suffusion chamber. When dipping the dye-loaded micropipet upstream into the corner of the suffusion chamber the entire cell was surrounded by dye solution (Figure 4B). After removal of the dye-loaded pipet the fluorescent dye immediately cleared from most of the t-tubules but sometimes remained trapped in a longitudinal section for ∼3-5 s (Figure 4B and C).

the footprint of a 96 well or 384 well multititer plate, so that after bonding the mold to the base of the plate, each of the three openair structures was sealed to an individual well (Figure 2A). The two outer wells served as inlet/outlet support structure and the central well as container for the cell suspension of the isolated cardiomyocytes from which individual cell were selected to be transferred into the open cell investigation chamber on the glass floor. Hydrostatic flow (∼2 mm/s, Figure 2B and C) was then 6394

Analytical Chemistry, Vol. 81, No. 15, August 1, 2009

generated, as described, to flush the chamber (Figures 2A and 3B). The arrangement was built as an array of chambers on a microscopic footprint with 70 µm pitch (SI Figure 1) enabling higher throughput measurements with a maximum of six cells being imaged on the CCD chip at a sufficient spatial resolution using a 20× objective lens. An overlayer of oil enhanced the hydrophobicity of the 30 µm wide PDMS wall of the chambers

Figure 5. Lateral diffusion within the t-tubules. The t-tubules of a ventricular myocyte were loaded with dextran coupled Fluo-4 (10 kDa) and the Fluo-4-dextran fluorescence was monitored on confocal line scan images to investigate the radial diffusion after removal of the extracellular dye. The dye cleared from the t-tubules (ROI 1 in Figure 5B) with a ∼0.5 s delay compared to the bulk extracellular space (ROI 2 in Figure 5B). The fluorescence intensity of ROI 1 and ROI 2 was plotted against time in Figure 5C (scan line positioned ∼10 µm below the cell surface as indicated on the sketch in Figure 5A).

thus preventing fluidic crosstalk between individual chambers. The lack of mixing together with the small footprint of the chamber array enabled the recording of the control responses (e.g., with mock solutions) and the experimental stimulation response (e.g., drug) of neighboring cells without moving the device on the microscope stage. As a proof of concept Ca2+ transients were stimulated with 100 mM K+ (Figure 3) and a time constant of the loading and clearing of the t-tubular system of ∼ 0.5 s was measured using fluorescein labeled 10 kDa dextran (Figures 4 and 5). A perfusion chamber (Figure 6) was designed with a pair of inlet/outlet channels at each end of the chamber adapted from the microfabricated H-filter.22 The H shaped microfluidic channels were used to generate laminar flow. (SI video 1). A T-junction was also implemented into the inlet channel at a distance of 300

µm away from the end of the cell investigation chamber to generate the laminar flow of the buffer solution and the injection solution inside the 12 µm wide inlet channel. The two heterogeneous streams maintained their flow characteristics (Re ) 2.4 × 10-2) over the 300 µm distance toward the chamber, through a 180° bend and back across the chamber width to the outlet channel without mixing with the bulk solution inside the chamber (Figure 6A and B, SI videos 1 and 2). The cardiomyocyte inside the chamber was locally stimulated by manipulating the position of the laminar interface with respect to the cell surface (Figures 5 and 6). The position of this laminar interface was moved by changing the relative flow velocities of the two streams (i.e., elevated [K+]e solution and buffer solution). The injection solution was removed from the bulk of the chamber through the perfusion across the length of the Analytical Chemistry, Vol. 81, No. 15, August 1, 2009

6395

Figure 6. Suffusion chamber with integrated microfluidic injection. A Fluo-3 loaded ventricular myocyte was positioned into the suffusion chamber and periodically stimulated with an elevated [K+]e. The fluidic setup and three situations (rest, subthreshold and threshold stimulation) of an infinite element simulation (Comsol) are illustrated in Figure 6A and B, respectively. The entire well (∼400 µL) above the suffusion chamber was filled with buffer solution and continuous suction (2 mm/ s) was applied away from the chamber toward the two parallel outlet channels. The inlet channel carrying the high [K+]e joined the suction channel 300 µm upstream from the suffusion chamber and was periodically pressurized to overcome the resistance in the suction channel and inject the high [K+]e into the cell chamber. The pressure change in the inlet channel was generated using a peristaltic pump. A fluorescent dye was added to the 100 mM KCl solution to indicate the cell surface area in contact with the stimulating solution. The extracellular fluorescence intensity was measured at both ends of the cell (ROI 1 and ROI 3) and inside the cell (ROI 2) as indicated in Figure 6C. The fluorescence images before (0.0 s), during (0.1 s), and after (2.0 s) the chemical stimulation of an intracellular Ca2+ transient are shown in Figure 6C. Note that the intracellular Ca2+ transient (ROI 2) peaked before the extracellular fluorescein-transient (ROI 1), indicating that once the cell was successfully stimulated the excitation propagated faster than the stimulating solution (Figure 6D). The stimulating solution never reached the opposite side of the cell chamber (ROI 3).

chamber (Figure 6), as described. The multiwell plate interconnect of the microfluidic device was comparable to the capillary interconnect fabricated from tapered quartz capillaries30 (SI Figure 2). Micropipet Probing. The open architecture of the array enabled the microfluidic manipulation of the content of the cell investigation chamber not only through the perfusion channels but also through manipulator-mounted micropipets lowered into the chamber (Figure 2 and 3). Fluidic manipulation of defined extracellular regions around the cardiomyocyte was achieved by positioning the pipet opening at defined positions along the longitudinal axis of the chamber (Figures 3 and 4). The flow velocity in the pipet was low compared to the chamber so that no added solution was detectable upstream of the pipet opening. Regional fluidic manipulation was used to find the minimal cell surface area depolarised with a high [K+]e solution sufficient for (30) Hartmann, D. M.; Nevill, J. T.; Pettigrew, K. I.; Votaw, G.; Kung, P.-J.; Crenshaw, H. C. Lab Chip 2008, 8, 609–616.

6396

Analytical Chemistry, Vol. 81, No. 15, August 1, 2009

Figure 7. Confocal line scan image of a Fluo-3 loaded ventricular myocyte stimulated with a high [K+]e using the microfluidic injection described in Figure 6. The K+ depolarisation during the first two injections was below the threshold for excitation as indicated through the absence of the Ca2+ transient. The third injection carried more potassium solution as the previous ones (ROI 1) and evoked the global transient rise in [Ca2+]i.

the takeoff of an action potential as measured fluorescently using the Ca2+ transient. The regional probing was visualized by adding a fluorescent dye to the stimulating solution (Figures 4, 6, and 7). It was found that the exposure of ∼30% of the cell surface with the high [K+]e was sufficient to trigger an action potential indicated through the synchronous cell shortening and concomitant Ca2+ transients (Figure 3). Compared to the monophasic decay of Ca2+ transients during electrical stimulation, the decay of Ca2+ transients in chemically stimulated cardiomyocytes had two phases, an initial fast decay from peak values followed by a second peak at a lower level indicating prolonged action potentials31 (Figure 3). At the same time the diastolic [Ca2+]i gradually rose during a ∼20s period of repetitive K+ stimulation at ∼ 0.25 Hz (Figure 3). Cardiomyocytes stimulated at higher frequencies (>0.25 Hz) did only respond to subrepeat stimulation. These results could be explained by the slow diffusion of K+ inside the t-tubular system with its detrimental effect on the repolarisation of the membrane potential.17 To test this hypothesis we monitored the loading and clearance of a fluorescently modified dextran in the t-tubular system. In a no-flow system, dye molecules of 70 kDa MW (dye conjugated dextran) have been shown to be able to move into the t-tubular system.13,24 Using the flow-system in this paper, with flow velocities of 2 mm/s, only smaller dye molecules (10 kDa MW) were able to load into the t-system (Figure 5). The high temporal and spatial resolution of the confocal imaging system revealed the fast exchange between the extracellular space and the transversal parts of the tubular invaginations (Figure 5) and enabled the observation of dye trapped in a longitudinal section of the t-system (Figure 4). Operation of Valveless Injection. The principle of the valveless injection using the laminar flow in a microfabricated (31) Roden, D. M. N. Engl. J. Med 2004, 350, 1013–1022.

H-filter is shown in Figure 6A. The interface between the fast flowing stimulation solution and the quiescent buffer solution in the cell chamber was manipulated toward the edge of the cell until it covered >30% of the cell surface and triggered an action potential, indicated through the synchronous Ca2+ release and cell shortening (Figure 6C). The removal of the stimulation solution from the cell chamber took ∼4 s but this slow clearance did not affect the faster recovery of the [Ca2+]i (Figure 6D). The confocal line scan image depicts three consecutive attempts to stimulate a cardiomycoyte with increasing amounts of the 100 mM [K+]e solution (Figure 7). The stimulation solution never reached the distal end of the cell, indicating tight control of the position of the laminar interface through the pressure change in the injection channel. DISCUSSION The technical advance of designing and implementing devices to generate fluid patterns on the cellular scale has previously enabled the behavioral study of primary cells, including cardiomyocytes and neurons, in a controlled microfluidic environment using specialized pipet based superfusion systems for fast solution switching, e.g., microflow from single or multi barreled pipettes or from Y-tubes.16,32,33 Microengineering has now made it possible to easily and accurately build microfluidic structures to influence the cell’s environment by applying spatially discrete stimuli in order to quantitatively control spatial and temporal alterations of the bath solution in a microchamber.6,34-38 The convenience and flexibility of the various fabrication protocols for microfluidic devices has enabled the microfluidic control of aspects of the cellular microenvironment, including gas exchange, solution switching and drug delivery, while monitoring the cellular response.2 This paper, for the first time, describes the use of laminar flow in a microfluidic channel for the regional chemical stimulation, providing an understanding of focused chemical stimulation of single adult ventricular myocytes. The design was adopted from the method of drug superfusion with the Y-tube39 and the device resembles a greatly miniaturized version of the experimental setup used for the local stimulation of neurons.32 The small dimensions and flow characteristics (and subsequent low Re numbers) provide the flow system with unique fluidic properties allowing focal stimulation of the cell surface. Static versus Dynamic Systems. In the static system of a confined picoliter sized chamber, with no flow, the injection of fluid results in a volume change. For this reason in our previously described static system we used a fountain pen type micropipet consisting of two concentric pipettes, the inner for fluid dispension, the outer for fluid suction6 to avoid volume changes during (32) Murase, K.; Ryu, P. D.; Randic, M. Neurosci. Lett. 1989, 103, 56–63. (33) Spitzer, K. W.; Bridge, J. H. B. Am. J. Physiol. 1989, 256, C441–C447. (34) Cai, X.; Klauke, N.; Glidle, A.; Cobbold, P.; Smith, G.; Cooper, J. M. Anal. Chem. 2001, 74, 908–914. (35) Werdich, A. A.; Lima, E. A.; Ivanov, B.; Ges, I.; Anderson, M. E.; Wikswo, J. P.; Baudenbacher, F. J. Lab Chip 2004, 4, 357–362. (36) Peterman, M. C.; Noolandi, J.; Blumenkranz, M. S.; Fishman, H. A. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 9951–9954. (37) Li, X.; Li, P. C. H. Anal. Chem. 2005, 77, 4315–4322. (38) Ganitkevich, V.; Reil, S.; Schwethelm, B.; Schroeter, T.; Benndorf, K. Circ. Res. 2006, 99, 165–171. (39) Pidoplichko, V. I.; Dani, J. A. J. Neurosci. Methods 2005, 142, 55–66.

regional drug application. This paper presents a dynamic system where the cardiomyocyte is confined in a microfluidic chamber with capped inlet/outlet channels for continuous suffusion. This arrangement combines the advantage of unlimited micropipet access to the cell with microfluidic control of the extracellular space.40 The flow rate was adjusted to 2 mm/s as measured on confocal line scan images (Figure 2). This speed was sufficient to renew the extracellular solution in the 200 µm long chamber within