D ynaflow 1
м48, a microfluidic chip solution for increasing
throughput and data quality in patch-clamp-based drug
screening*
D an iel Gran feldt', M arie-Lou ise Johansson2, Jon S in cla ir1, Johan Pihl2, an d M attias Karls son 2
1 Departm ent o f Chem istry and Bioscience and M icrotechnology Centre, Chalm ers U niversity o f Technology, SE-412 96 Göteborg, Sweden,
2C ellectricon AB, Fabriksgatan 7, SE-412 50 G öteborg, Sweden
Abstract
Ion channels are transm em brane proteins, found in virtually all cell types throughout the human body. Ion channels underlie neural com m unication, mem ory, behavior, every movement and heartbeat, and are as such prone to cause disease if m alfunctioning. Therefore ion channels are very im portant targets in drug discovery. The gold standard technique for obtaining inform ation on ion channel function with high inform ation content and temporal resolution is patch-clamp. The technique measures the minute currents originating from the movement o f ions across the cellular membrane, and enables determ ination of the potency and efficacy o f a drug. However, patch-clam p suffers from serious throughput restrictions due to its laborious nature. To address the throughput problem s we have developed a microfluidic chip containing 48 m icrochannels for an extrem ely rapid, sequential delivery o f a large num ber o f completely controlled solution environm ents to a lifted, patch-clam ped cell. In this way, throughput is increased drastically compared to classical patch-clam p perfusion set-ups, with uncom prom ised data quality. The 48-m icrochannel chip has been used for the characterization o f drugs affecting ligand-gated ion channels including agonists, antagonists and positive m odulators with positive effects on both throughput and data quality.
K e y w o rd s: high throughput, ion channel screening, patch clam p, m icrofluidics, GA B Aa receptor
* W e ack now ledg e the su pp o rt o f profe sso r Ow e Orw ar. T his w ork w a s fou nded by the G öran G us tafss on Fo un d ation and the Sw edis h Fo und atio n for Strategic R esearc h (SSF).
Introduction
Ion channels are of central im portance in the body. They serve as conduits between the inside o f cells and the outside world, controlling basic bodily processes such as heartbeats, m etabolism and immunity as well as more complex processes such as em otions, learning and memory (1). Increasing evidence indicates that the direct or indirect m alfunctioning o f an ion channel can severely affect human health and many ion channels are thus considered to be im portant therapeutic targets (2). The standard technique for acquiring high-inform ation content functional ion channel data is the patch clam p m ethod (3). The technique relies on the electric clam ping o f the cellular m em brane potential o f a single cell, while m easuring the ionic current caused by ion channel activity. Due to the high temporal resolution and the sensitivity of the technique, it is possible to record the activity o f single ion channels with sub-m illisecond time resolution. However, acquiring high quality patch clam p data is a tedious process, requiring extensive work input from highly skilled operators. Due to these lim itations, the identification o f com pounds that affect the desired ion channel function is cum bersom e and time-consum ing. As the need for high quality ion channel data continues to increase, the patch clam p technique has becom e the bottleneck in the efforts to increase productivity in the process o f ion channel screening. Therefore, large efforts have been made in developing parallel patch clamp devices that aim to autom ate and m ultiplex the num ber o f patch-clam ped cells and thereby increase throughput in screening applications (4-10). A lthough the automated patch clam p devices indeed offer an increased rate o f com pound testing, this im provem ent unfortunately comes at a price, being reduced recording quality, time resolution and loss of flexibility and control during the experim ent (11). In particular, the automated patch clam p system s are not am enable for screening ligand-gated ion channel targets, mainly because these systems lack the ability for rapid and precise solution switching. Because activation and desensitization o f many types o f ligand-gated ion-channels occurs within a few m illiseconds (1,12-14), methods are required that can change a large num ber o f solutions containing different ligands around cells in the same tim e range in order to efficiently study such rapid receptor interactions. Exam ples o f pharm acologically important ion channels that display fast kinetics are nicotinic acethylcholine receptors, controlling m uscle function, and ligand- gated channels of fast chemical synapses such as GABA and A M PA receptors, involved in neuronal m odulation. Consequently, the need for developm ent of screening m ethods for ion channels that offer high throughput, low cost, fast, flexible and w ell-controlled solution exchange while still m aintaining the high
inform ation content offered by classic patch clam p screening is growing at a rapid pace.
R ather than resorting to a parallel patch-clam p concept, we have developed a system where conventional single-cell-based patch clam p technique is com bined with chip-based m icrofluidics for rapid sequential delivery of ion channel agonists or antagonists onto patch-clam ped cells (15,16). Here, we extend the previous findings and dem onstrate a 48 channel chip facilitating advanced investigations o f ion channel function by utilization o f com plex substance application patterns, possibly involving a range o f different substances to be applied to the cell sequentially. To dem onstrate the functionality o f this device we perform ed patch clamp recordings o f several different dose response curves from a single cell on one chip. Furtherm ore, we dem onstrate that the excellent experim ental control, and solution switching properties, enables direct extraction o f kinetic param eters from the obtained data sets.
M aterials and Methods
Cell cu ltu re. The WSS-1 cell line (HEK 293 transfected with G AB A a
receptor) was used. Cells were grown adherent for 2-4 days on plastic Petri dishes in DM EM supplem ented with antibiotics, L-glutam ine, glucose (4.5g/L), Na-carbonate and fetal ca lf serum (10% ).
S o lu tio n s a n d drug s. The extracellular solution used in all experim ents
contained, in mM; 140 NaCl, 5 KC1, 1 MgCl2, 1 C aCl2, 10 Hepes, 10 D- Glucose, pH 7,4. The electrodes were filled with intracellular solution, in mM; 100 KC1, 2 M gCb, 1 CaCl2, 11 EGTA, 10 Hepes, pH 7.2. GA BA was obtained from Calbiochem (La Jolla, CA) and ß-alanine and bicuculline m ethiodide, were from Sigma (St. Louis, MO).
Electrophysiology. W hole cell patch clam p m ethods were used in all
experiments. Glass electrodes were fabricated from thick wall borosilicate glass and the resistances o f the pulled patch pipettes were 3-5 M U. All data was recorded using an Axopatch2B (Axon Instruments) patch clam p amplifier. The cells were clam ped at -40mV. Current signals were recorded at a sampling frequency o f 5 kHz, low pass filtered (1 kHz, 4-pole Bessel filter) and analyzed using the Clam pfit8 software. Tw o different GABA receptor agonists (GABA and ß-alanine) and one antagonist (Bicuculline) were used. Every other well was loaded with buffer for w ashing interdigitated with the various concentrations of agonists and antagonist. GABA concentrations used was 1, 5, 10, 20, 50, 100, and 500 цМ, ß-alanine concentrations used was 1, 5, 10, 20, 50, 100, and 500 mM, and bicuculline concentrations of 0.01, 0.1, 1, 5, 10, and 100 /xM was со-applied with 100 д М GABA.
System operation. The fabrication of the devices has been described in
detail elsew here (16). The DF-48 chip is com prised o f 48 sam ple wells connected to an open volum e through 50x60 ц т (wxh) m icrochannels. At the exit into the open volume, the channels are tightly packed, separated by 30 [im wide walls. The chip was placed on a software controlled, m otorized scanning stage m ounted on an inverted microscope. A cell was patch-clam ped in the open volume, lifted up approxim ately 30 /xm, and translocated to the channel outlets. Pressure was applied to initiate a flow corresponding to 3 min/s, and the cell was translocated between different solution environm ents and exposed to ion channel effectors according to a predefined experim ental protocol.
Results and discussion
We present a m icrofluidic device designed to increase throughput in ion channel screening. The system is com posed o f a 48-w ell m icrofluidic chip, a motorized scanning stage, a pump for creating flow, and software to control the scanning actions o f the m otorized scanning stage. The system is easily integrated onto the m icroscope o f an existing patch clam p setup (Figure 1). Due to the small size o f the m icrochannels, the fluids behave in ways that differ in a fundam ental way from m acroscopic systems. The flow in the m icrofluidic device is in the low Reynolds num ber regime, characterized by a completely laminar flow, with m ixing occurring only by diffusion (17). As a result o f this, the individual channel com positions is m aintained well out into the open volume (16). Thus, virtual m icrocontainers of different drug solutions are created in an open volum e that can be accessed by a biological cell detector only a few microns large, e.g. a patch clamped cell. By translating a lifted, patch clamped cell in front o f the channel outlets, solution switch times o f 10-15 ms is well achievable. Apart from the advantages in ion channel screening, several other benefits are achieved by placing the patch clam ped cell in the lam inar flow. The force that the lam inar fluid flow exerts on the cell-pipette system stabilizes the seal and produces higher seal resistances (without affecting the access resistances), longer recording times, and lower noise levels (18).
The system was used together with WSS-1 cells expressing the GAB Ад- receptor, chloride selective ion channels that becom e activated after binding a wide variety o f agonists (19). Investigation o f the receptor function was perform ed using a device containing 48 separate channels enabling the sequential exposure o f a single cell to 48 different solutions, allowing the determ ination o f several distinct dose response curves from the same cell using different agonists and antagonists with varying properties for the activation
üCCttBKa
Fig. 1. C o m po nen ts o f the D F48 system . The m ic rofluidic ch ip (b o tto m ) is c on trolled by a versatile softw are (to p right) and integrate s easily into an existin g m icrosc ope se t-up (to p left).
kinetics o f the receptor (Figure 2). The ion channels were activated by sequential exposure to seven different concentrations o f GABA and ß-alanine, respectively, followed by six different concentrations o f the antagonist bicuculline, со -applied with 100 дМ GABA. Due to the greater num ber of channels available it is possible to record com plete dose response curves, for several different agonists, from the same cell. In this way, cell to cell variation is elim inated, producing better data quality along with the possibility to com pare the effects of different agonists and antagonist more clearly, since the same cell is used throughout the recording. At the same time, the applied flow results in longer recording tim es and im proved recording param eters.
b-alanine GABA/biciiculline to IBicuculllne] (mM) i 10 юо [GABA] (mM) — 00 1UOO 1 10 100 Ib-alanlnel (mM) ECso = 16.9 + /- 0.75 mM EQ>0 = 7.18 +■/■ 0.06 mM IC5 0 = 9.30 +/- 0.54 mM
F ig . 2. M u ltip le a go n is t a pp lic atio n to th e sam e cell. T hree d iffere nt d o se re s po ns e c urves re cord ed from the sam e cell in o ne co ntinu ou s trace. A lifted patch c lam pe d W SS-1 cell e xpressin g the G A B Aa re ce p to r was ex po se d to th ree diffe ren t sets o f antag onist a n d /o r ag on is t and the resulting cu rre nt w as recorded. L ow er p art show s do se re sponse c urves o btain e d by p lo ttin g peak cu rre nt versus c on ce ntratio n o f ag o nis t o r a nta gon ist (n=3).
Because o f the rapid solution switching and com plete control over solution environm ents various properties o f different agonists can be resolved. To illustrate this, close-ups o f the whole-cell currents induced by b rief (100 ms) applications o f saturating concentrations o f GABA and ß-alanine respectively are shown (Figure 3). The onset of current is the same for both agonists whereas the more rapid unbinding o f ß-alanine (20,21) is clearly seen during the return to baseline phase, where the GABA induced current returns much slower as com pared to the ß-alanine induced current. Thus, using the system described here to probe ion channel function, it is possible to resolve differences in unbinding rates o f different ligands. The system has also previously been used successfully for extrem ely rapid ion channel system s such as AM PA/kainate, NM D A, and nicotinergic a7 ion channels, dem onstrating the versatility o f the system with regards to experim ental systems requiring high tem poral resolution.
In conclusion, the D y naflow 1M 48 m icrofluidic system makes it possible to record a large num ber o f dose response curves quickly, with the added benefit o f producing longer recording times and lower noise levels. The described microfluidic chip system offers robust and reliable detection with m ajor
F ig . 3. D ifferen ces in u nb ind ing rates are clearly resolved. C lose up o f w ho le cell kinetics from G A B Aa rec eptors, in duce d by rap id app licatio n o f G A B A (black trace) o r ß -a lanin e (grey trace). The p ron o un ced differe nce in off-rate betw ee n the two ag onis ts is cle arly seen.
increases in throughput for conventional, high-quality patch-clam p screening of ion channel functions. M oreover, the 48 channel layout allows for several different agonists, antagonists or com binations thereof to be applied to the same cell, elim inating cell to cell variation. Data quality is im proved by the fact that entire dose responses are easily obtained from the same patch-clam ped cell and with increased signal-to-noise ratio due to the flow stabilization o f the seal. In the long term, the D y naflo w IM 48 system offers the possibility o f ion channel screening with m aintained information content, increased throughput and decreased cost.
References
1. Hille, B. 2001. Ion C han nels o f E xcitable M em branes. Sina u er A sso ciates , Inc., Sund erland, MA, USA.
2. A shcroft, F.M . 2000. Ion C h a nn els and Disease. A cadem ic Pre ss , L o ndo n, UK.
3. Ham ill, O.P., A. M an y, E. N eher, B. Sakm ann and F.J. Sigw orth . 1981. Im prov ed patch- clam p tec hn iq ues fo r hig h-res olu tion curren t recording from cells an d cell-free m em brane patches . Pflug ers A rc h 597 :85 -100 .
4. X u, J., X. W ang, B. E nsign, M. Li, L. W u and A. Guia. 20 01. Io n-cha nne l ass ay tec hnolog ies: qu o va dis? D ru g D isc ov T oday 6:12 78-1 287 .
5. L aV a n, D .A ., D .M . Lynn and R. Langer. 2002. M oving sm alle r in drug disco very and delivery. N at R ev D rug D iscov /:7 7 -8 4 .
6. W orley, J.F., 3rd an d M .J. M ain. 2 002. An in dustrial pe rsp ectiv e on u tiliz in g func tional ion c hann el assays for high throu gh pu t screening. R e ceptors C ha nn els £ :269 -2 82 .
7. W illum sen, N.J., M. Bcch, S.P. O lcsen, B.S. Jensen, M .P. K orsg aard and P. C h risto phersen . 2 003. H igh thro u gh pu t electroph ysio log y: new perspec tives for ion chan nel drug discovery. R ec epto rs C ha nn els 9:3-12.
8. W ang, X. and M. Li. 2 003. A utom ated electrophysiology: h ig h th ro u gh p u t o f art. A ssay D rug Dev T echno] / : 695 -708.
9. W ood, С ., С. W illiam s an d G.J. W aldron. 2004. Patch c lam pin g by num bers. D rug Discov T o da y 9:434 -441.
10. Z h eng , W ., R.H. S pe nc er and L. Kiss. 2004. High thro ugh put assay tech nologie s for ion ch annel d ru g d isc overy. A ssay D rug Dev T e chnol 2:5 43-552 .
11. Sorota , S., X .S. Z h ang, M. M argulis, K. T ucke r and T. Priestley. 2005. C ha rac teriza tio n o f a hE R G screen using the Ion W orks HT: com pa rison to a hE R G rub id ium efflux screen. A ssay D rug Dev T ec hn o l 3:41-51.
12. Jon es, M .V . a nd G .L . W estb rook. 1996. T he im p act o f re ce p to r d es e nsitiza tio n on fast syna ptic tran sm issio n. T rend s N e urosci 19:96-101.
13. B reitinger, H .G. 2 001. Fa st k inetic analysis o f liga nd-gated ion c han ne ls. N eu ro sc ien tis t 7:95- 103.
14. K rusek, J., I. D ittert, T. H endrych, P. Hnik, M. H orak, M. Pe trovic , M. Sedlac ek, K. S usan kova, L. S vo bod ova , K. T o uso va, E. U jec, V. V la ch ova, L. V yklicky, F. V ys kocil and L. V yklicky, Jr. 2004. A ctivatio n and m odula tion o f liga nd-gate d ion ch anne ls. Physiol Res 53 S up p l /:S 103-113.
15. Sinclair, J., J. Pih l, J. O lofs son , M. K arlsson, K. Jardem a rk, D.T. C h iu and O. O rw ar. 2002. A ce ll-ba sed ba r c od e rea de r for high-thro u gh pu t sc reenin g o f ion ch a nn e l-lig a nd interactions. A nal C he m 74 :61 33-613 8.
16. O lo fsso n, J., J. P ih l, J. S inc lair, E. Sa hlin, M. K arlsson a nd O. O rw ar. 2 004. A m icrofluidics a pp roac h to th e p ro blem o f creatin g sepa rate solution e nv iron m e nts a cce ssib le from m ac ros cop ic vo lum es. A nal C hem 76:496 8-4 976 .
17. N gu yen, N. a nd S.T. W ereley. 2002. Fu ndam en tals and A pp lic atio ns o f M icroflu idics. A rtech H ouse, Inc.
18. Sincla ir, J., J. O lofs son , J. Pihl and O. O rw ar. 2003. S ta biliz atio n o f H igh-R e sista nce Seals in Pa tch -C lam p R ecordin gs by L am in ar Flow. Anal C h em 75:671 8-6 722 .
19. Jo hn s ton , G .A. 1996. G A B AA rece pto r ph arm acology. Pharm acol T h e r 6 9:173 -19 8.
20. Jones, M .V . an d G .L. W estbro ok. 1995. D es ensitized states p rolo n g G A B A A channe l res pon ses to b rie f a go nis t pu lse s. N euron /5 :18 1 -1 9 1 .
21. Jones, M .V ., Y. Saha ra, J.A. D zubay and G.L. W estbrook. 1998. D e finin g affinity w ith the G A B A A re cepto r. J N euro sci /5 :8 5 90 -8 6 0 4.
C o rre sp on d in g a u th or: D a n iel G r a n fe ld t D e partm ent o f C he m istry and B iosc ie nce a nd M ic ro tec hn olog y C entre C h alm ers U nive rsity o f T e chnolo gy, K em ivägen 10, SE -4 1 2 96 G ö teb org , Sweden T el: (4 6 31) 772 6112 e-m ail: da nie l.g ra nfe ld t@ ch em bio.cha lm ers .s e
