Controlling Desensitized States in Ligand ... - ACS Publications

The receptors display complex desensitization kinetics, occurring on vastly different time scales. This is not only important in biology and pharmacol...
1 downloads 0 Views 457KB Size
Download PDF
Anal. Chem. 2006, 78, 7947-7953

Accelerated Articles

Controlling Desensitized States in Ligand-Receptor Interaction Studies with Cyclic Scanning Patch-Clamp Protocols Daniel Granfeldt,† Jon Sinclair,‡ Maria Millingen,† Cecilia Farre,‡ Per Lincoln,† and Owe Orwar*,†

Department of Chemical and Biological Engineering, Physical Chemistry, Chalmers University of Technology, SE-412 96 Go¨teborg, Sweden, and Cellectricon AB, Fabriksgatan 7, SE-412 50 Go¨teborg, Sweden

Ligand-gated ion channels are important control elements in regulation of cellular activities, and increasing evidence demonstrates their role as therapeutic targets. The receptors display complex desensitization kinetics, occurring on vastly different time scales. This is not only important in biology and pharmacology but might also be of technological significance since populations of receptors under microfluidic control can function analogously to DRAM memory circuits. Using a novel microfluidic method, and computer modeling of the receptor state distributions, we here demonstrate that GABAA receptor populations can be controlled to display high or low EC50 values, depending on input function (i.e., the exact pattern of agonist application). The sensitivity of the receptors can be tuned up to 40-fold (β-alanine) by the particular agonist exposure pattern. By combining patch-clamp experiments with computer modeling of receptor state distributions, we can control the assembly of receptors in desensitized states. The technique described can be used as an analytical tool to study the effect of desensitization on the activity of ion channel effectors. We describe the differential blocking effect of the competitive antagonist bicuculline on the high- and low-EC50 GABAA receptor preparations and conclude that the inhibition is dramatically dependent on how the different desensitized states are populated. Furthermore, we show that both GABA and β-alanine, two agonists with different affinity but similar efficacy, induce the same type of desensitization behavior and memory effects in GABAA receptors.

Ligand-gated ion channels are essential in controlling signal transduction for cells in, for example, the central nervous system, 10.1021/ac060812z CCC: $33.50 Published on Web 10/27/2006

© 2006 American Chemical Society

the intestine, and muscles.1 For example, nicotinic acetylcholine receptors play important roles in synaptic signaling both at neuromuscular junctions and at neuronal synapses,2 while glutamate- and GABA receptors are involved in controlling synaptic excitability.3-5 A plethora of diseases have been connected to malfunctioning of ion channels, and the number is steadily growing.6 Ion channels are therefore important therapeutic targets7,8 and constitute ∼9% of all the targets under study by pharmaceutical companies today. Ligand-gated ion channels are activated by binding of agonists. This activation results in channel opening but also causes the ion channel to desensitize, that is, to reside in a ligand-bound yet shut and nonconducting state.9 Depending on the type of ion channel, the desensitized states are populated and depopulated on different time scales. The role of the desensitized states has gained increased attention as desensitization has proved an effective way for shaping and modulating current properties for several different ligand-gated ion channels.9-12 Due to the complex kinetics of ion * To whom correspondence should be addressed. E-mail: [email protected]. † Chalmers University of Technology. ‡ Cellectricon AB. (1) Hille, B. Ion Channels of Excitable Membranes, 3 ed.; Sinauer Associates: Sunderland, MA, 2001. (2) Changeux, J. P.; Bertrand, D.; Corringer, P. J.; Dehaene, S.; Edelstein, S.; Lena, C.; Le Novere, N.; Marubio, L.; Picciotto, M.; Zoli, M. Brain Res. Brain Res. Rev. 1998, 26, 198-216. (3) Madden, D. R. Nat. Rev. Neurosci. 2002, 3, 91-101. (4) Mayer, M. L. Curr. Opin. Neurobiol. 2005, 15, 282-288. (5) Farrant, M.; Nusser, Z. Nat. Rev. Neurosci. 2005, 6, 215-229. (6) Ashcroft, F. M. Ion Channels and Disease, 1st ed.; Academic Press: London, UK, 2000. (7) Xu, J.; Wang, X.; Ensign, B.; Li, M.; Wu, L.; Guia, A. Drug Discovery Today 2001, 6, 1278-1287. (8) LaVan, D. A.; Lynn, D. M.; Langer, R. Nat. Rev. Drug Discovery 2002, 1, 77-84. (9) Jones, M. V.; Westbrook, G. L. Trends Neurosci. 1996, 19, 96-101.

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006 7947

channels, it is often difficult to study them and to obtain relevant data on efficacy and affinity.13 The standard technique for acquiring ion channel data is the patch-clamp method,14 having superior sensitivity and time resolution. A drawback with patch-clamp measurements, however, is that only conducting states, i.e., open states, are visible in the assay. Therefore, affinity and efficacy processes cannot be measured separately. Another issue that complicates current measurements is that populations of ion channels are dynamic nonlinear response entities that, depending on their state distributions, will undergo conformational changes and react differently to a stimulus. Thus, the current produced from a whole cell, by subsequent applications of agonist, will depend on the specific receptor state distributions, reflecting the activation history of the receptor population.15,16 It would therefore be beneficial to be able to control the distribution of states in the receptor population and study these different populations separately with regard to receptor properties and modulator functions. We have developed a microfluidic method17,18 that allows us to generate complex chemical waveforms and very rapidly change the solution environment around a patch-clamped cell.19 Using this method, in combination with a cyclic scanning patch-clamp (CSPC) protocol, we recently demonstrated that we could control the distribution of states in a population of receptors and construct a simple biohybrid memory circuit.16 We here extend these findings and demonstrate that the CSPC technique, in combination with computer modeling, can be used as an analytical tool aimed at studying the role of desensitization in the function of ion channel modulators. β-Alanine is a low-affinity GABAA receptor agonist with slower binding and more rapid unbinding (as compared to GABA) from the receptor.20-22 In contrast to the large difference in affinity, the efficacy of β-alanine is similar to that of GABA.20,21,23 Hence, β-alanine constitutes a useful tool to discriminate between the contributions from affinity and efficacy to the activity of the GABAAR-ligand complex. By using GABA and β-alanine as GABAA receptor agonists, we demonstrate that the CSPC technique can be used to titrate whole-cell receptor populations into different state distributions (10) Quick, M. W.; Lester, R. A. J. Neurobiol. 2002, 53, 457-478. (11) Krusek, J.; Dittert, I.; Hendrych, T.; Hnik, P.; Horak, M.; Petrovic, M.; Sedlacek, M.; Susankova, K.; Svobodova, L.; Tousova, K.; Ujec, E.; Vlachova, V.; Vyklicky, L.; Vyskocil, F.; Vyklicky, L., Jr. Physiol. Res. 2004, 53 (Suppl 1), S103-113. (12) Giniatullin, R.; Nistri, A.; Yakel, J. L. Trends Neurosci. 2005, 28, 371-378. (13) Colquhoun, D. Br. J. Pharmacol. 1998, 125, 924-947. (14) Hamill, O. P.; Marty, A.; Neher, E.; Sakmann, B.; Sigworth, F. J. Pflugers Arch. 1981, 391, 85-100. (15) Overstreet, L. S.; Jones, M. V.; Westbrook, G. L. J. Neurosci. 2000, 20, 7914-7921. (16) Sinclair, J.; Granfeldt, D.; Pihl, J.; Millingen, M.; Lincoln, P.; Farre, C.; Peterson, L.; Orwar, O. J. Am. Chem. Soc. 2006, 128, 5109-5113. (17) Sinclair, J.; Pihl, J.; Olofsson, J.; Karlsson, M.; Jardemark, K.; Chiu, D. T.; Orwar, O. Anal. Chem. 2002, 74, 6133-6138. (18) Olofsson, J.; Pihl, J.; Sinclair, J.; Sahlin, E.; Karlsson, M.; Orwar, O. Anal. Chem. 2004, 76, 4968-4976. (19) Olofsson, J.; Bridle, H.; Sinclair, J.; Granfeldt, D.; Sahlin, E.; Orwar, O. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 8097-8102. (20) Jones, M. V.; Westbrook, G. L. Neuron 1995, 15, 181-191. (21) Jones, M. V.; Sahara, Y.; Dzubay, J. A.; Westbrook, G. L. J. Neurosci. 1998, 18, 8590-8604. (22) Rabe, H.; Picard, R.; Uusi-Oukari, M.; Hevers, W.; Luddens, H.; Korpi, E. R. Eur. J. Pharmacol. 2000, 409, 233-242. (23) Wu, F. S.; Gibbs, T. T.; Farb, D. H. Eur. J. Pharmacol. 1993, 246, 239246.

7948

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

with different EC50 values and examine the effects of competitive inhibition on these state distributions. We also show that the activation-dependent memory effect of the receptor population16 is not due to the affinity of the ligand used to activate the receptor but is instead an intrinsic property of the receptor population, i.e., the ability to undergo desensitization on long time scales. By combining the experimental data and computer modeling, it is possible to use this microfluidic method to make accurate predictions about the pharmacological properties of a ligand. EXPERIMENTAL SECTION Microfluidic System. All electrophysiology experiments were performed using a commercially available microfluidic device for patch-clamp (Dynaflow 16, Cellectricon AB, Go¨teborg, Sweden). The device used in this article has 16 channels of dimensions 50 × 57 µm (w × h) separated by 22-µm-thick walls at the point of exit into the open volume. The volume of the 16 sample reservoirs is 80 µL each, and the dimensions of the open volume are 30 × 35 mm. Prior to experiments, the device was loaded with different solutions using a micropipet. A 2-mm-thick polycarbonate lid was attached over the sample reservoirs using double adhesive tape (3M, Stockholm, Sweden) in order to create a closed system. The lid was connected to a syringe with PE tubing, and a syringe pump (CMA/100, microinjection pump, Carnegie Medicine, Cambridge, UK) was used to compress the air enclosed by the lid to initiate a well-defined, pressure-driven flow in the channels, forming a laminar patterned flow in the open volume with individual welldefined solution compartments, accessible for controlled delivery to a patch-clamped cell.17,18 For the experimental conditions used, the solution exchange time was 10-12 ms, as determined by solution-change-induced kinetics of hERG channel currents. Cell Culture. Adherent WSS-1 cells expressing the rat GABAA receptor (subunit composition R1β3γ2,24) were cultivated in Petri dishes for 4-8 days in (DMEM/F12) medium supplemented with antibiotics and antimycotin (0.2%), fetal calf serum (10%), and L-glutamine. Before the experiments, cells were washed and detached in a HEPES-saline buffer (HBS) containing (in mM) the following: 10 HEPES, 140 NaCl, 5 KCl, 1 CaCl2, 1 MgCl2, and 10 D-glucose (pH 7.4). All chemicals used in the cell culturing were from Sigma-Aldrich (Sigma-Aldrich Sweden AB, Stockholm, Sweden). Electrophysiology. Patch-clamp experiments were carried out in the whole-cell configuration (holding potential -40 mV, sampling frequency 5 kHz, filter frequency 1 kHz). The patchclamp electrode solution contained (in mM) 100 KCl, 2 MgCl2, 1 CaCl2, 11 EGTA, and 10 HEPES; pH was adjusted to 7.2 with KOH. All experiments were performed at room temperature (18-22 °C). CSPC experiments were performed as described elsewhere16 and are based on the scanning of a patch-clamped cell back and forth through fixed concentration gradients of receptor effectors with control of each cycle with respect to texp, twash, and between-scan time (trest) (Figure 1a). CSPC experiments were performed using different texp, ranging from 30 ms to 10 s (n ) 3-8 cells per texp scan). In all scans, texp was equal to twash. Except when noted, the between-scan time (trest) was always at least 3 min to accommodate a complete return of the receptor population to the ground state. (24) Davies, P. A.; Hoffmann, E. B.; Carlisle, H. J.; Tyndale, R. F.; Hales, T. G. Neuropharmacology 2000, 39, 611-620.

Figure 1. Schematic design of the CSPC experiments and dose-response experiments recorded at different texp. (A) Closeup of the channel exits into the open volume. The channels are loaded with increasing concentrations of agonist (C1-C7), to which the cell is exposed for a specified time texp, interdigitated with buffer (denoted as B), for washing for a specified time twash. The cell is scanned in ascending or descending order. (B-D) Current traces recorded from the same cell using different texp ) twash and trest ) 3 min for β-alanine. Note the difference in the horizontal scale bar while the vertical scale bar is the same for all traces. For comparison, the inset in (B) shows a recording with texp ) twash ) 100 ms for GABA. (E-G) Representative dose-response curves, recorded from the same cell, for texp ) 100 ms (E), 1 s (F), and 10 s (G). Peak currents shown are from doses applied in descending order (black squares) and ascending order (red circles). (H) The difference in current response to the maximum dose for ascending- and descending-dose applications shown for GABA (black) and β-alanine (red) fitted to a single-exponential function.

Agonists. The cells were exposed to series of agonist exposures, interdigitated with buffer solution, in either ascending or descending order. For GABA, the concentration gradient used was 1, 5, 10, 20, 50, 100, and 500 µM, whereas for β-alanine the concentration gradient used was 0.5, 1, 2, 7.5, 20, 50, and 100 mM. Modeling of Receptor Kinetics. For modeling of GABA currents, we used the kinetic model of Jones and Westbrook20 with the rate constants of Lindquist et al.,25 which we have previously modified.16 For modeling of β-alanine currents, we used the same kinetic model but replaced the binding and unbinding rates with values for this agonist.21 The distribution of the individual states included in the model was determined numerically using the Q-matrix method.26 Assuming a certain fixed GABA concentration, the rate laws describing the interconversion of the receptor states are all of (pseudo) first order. The calculations (25) Lindquist, C. E.; Laver, D. R.; Birnir, B. J Neurochem. 2005, 94, 491-501. (26) Colquhoun, D.; Hawkes, A. G. In Single channel recording, 2nd ed.; Sakmann, B., Neher, E., Eds.; Plenum Press: New York, 1995; pp 589-633.

were performed with the MATLAB software package (The MathWorks, Natick, MA). Statistics. All error bars are SD. For hypothesis testing, Student’s t-test was used with p < 0.05 considered as significant. Except when noted, peak currents were normalized to an undesensitized response to the highest agonist concentration used. RESULTS AND DISCUSSION We performed CSPC experiments in a 16-channel microfluidic device, operating in the low Reynolds number regime.17 We titrated patch-clamped cells to series of applications of GABA or β-alanine, at different exposure times (texp), interdigitated with buffer solution for clearance (twash) after each agonist exposure. Titration protocols were performed in either ascending (from low to high concentrations) or descending order (high to low concentrations) (Figure 1A). As can be seen in Figure 1B and E, brief-period exposures (texp in the range of 30-100 ms) to β-alanine yielded similar current responses in ascending and descending Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

7949

Figure 2. Order of application and texp tunes the sensitivity of the GABAA. (A, B) Normalized and averaged dose-response curves, obtained at different texp (n ) 3-8 cells per texp), with β-alanine as agonist, in descending (A) and ascending (B) order. (C, D) Normalized and averaged dose-response curves, obtained at different texp (n ) 3-8 cells per texp), with GABA as agonist, in descending (C) and ascending (D) order. (E) EC50, values for different texp, application order and ligand used. Note the different scales on the y-axis, µM for GABA and mM for β-alanine. (F) Hill slope as a function of texp, mode of application, and ligand used. Legend from (E) applies also in (F).

scans. However, at increasing values of texp, the response to the two different orders of application becomes more and more different (Figure 1C, D, F, and G). This behavior is identical to that obtained with GABA16 although the more rapid unbinding of β-alanine is clearly observed as a faster return to the baseline during twash when comparing the current traces for the two agonists acquired at texp ) 100 ms (Figure 1B, showing β-alanine, and inset, showing GABA). Both ascending and descending doseresponse curves for short texp are similar and display a sigmoid behavior reaching saturation at the higher concentrations of agonist (Figure 1E). As texp is increased above 100 ms, the order in which the doses are applied results in differences between the curves. Dose-response curves obtained in ascending order are still sigmoidal and reach saturation at lower concentrations of agonist than for shorter texp (Figure 1F and G). The doseresponse curves obtained for the doses applied in descending order are, however, close to linear at the higher doses and span the whole concentration range with a dynamic range of 3 orders 7950 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

of magnitude (Figure 1F and G). β-alanine and GABA induce identical changes to the shape of the dose-response curves, and on identical time scales. Moreover, the normalized difference between the responses to the maximum dose obtained for ascending- and descending-order scans, ∆I[Cmax] ) (I[Cmax]Desc - I[Cmax]Asc)/I[Cmax]Desc, where I[Cmax]Desc and I[Cmax]Asc are the current responses for the highest agonist concentration in descending and ascending order scans, respectively, are also very similar for both agonists (Figure 1H). When comparing the two agonists it is clear that, apart from the effect of affinity, no major difference between the agonists can be discerned. This indicates that the observed effects are not dependent on the affinities of the two different agonists but instead due to intrinsic properties of the receptor ensemble. It is likely that GABA induces somewhat higher levels of desensitization at short texp (due to higher affinity), but we were not able to resolve this difference. Tuning of the Receptor Population. The EC50 value of a receptor population will depend on the time of exposure to agonist

Figure 3. The duration of the memory effect is in the range of seconds to minutes. The effect of different trest on current depression was studied by exposing cells to gradients of agonist, followed by a test pulse (after trest) and a normalization pulse (after at least 3 min). The normalized difference between the two pulses (n ) 5-10 cells) was significantly different from zero for trest ) 3, 15, 30, 60, and 120 s for texp ) 3 s but only for trest ) 100 ms for texp ) 100 ms. Accordingly, the duration of the memory effect is less than 15 s with texp ) 100 ms (red trace) and over 2 min with texp ) 3 s (black trace). No difference in memory duration was found between ascending and descending gradients of agonist.

as well as the order by which a concentration ramp is applied. To examine this property in further detail, descending and ascending dose-response curves (n ) 3-8 per texp and direction of scan) for both agonists were averaged and normalized (Figure 2A-D). When agonist was applied in descending order, increased texp shifts the dose-response curves to the right, toward higher concentrations (Figure 2A and C), whereas an opposite effect, i.e., a leftward shift toward lower concentrations, is observed when agonist application is performed in ascending order (Figure 2B and D). The dose-response curves were fitted to the Hill equation yielding EC50 values and Hill slopes, respectively (Figure 2E and F). More specifically for β-alanine, in ascending order scans, increasing texp lowers the EC50 value from 17 (texp ) 30 ms) to 3.2 mM (texp ) 10 s) (Figure 2E). This shift is accompanied by an increase in the Hill slope from 1.5 to 2.5 (Figure 2F). Corresponding values for descending order scans show an increase in EC50 from 15 to ∼120 mM (Figure 2E) and a decrease in Hill slope from 1.2 to 0.8 (Figure 2F). In experiments with GABA, the EC50 value decreases from 35 (texp ) 30 ms) to 7.6 µM (texp ) 10 s) while the Hill slope increases from 1.7 to 2.0 for ascending scans. For doses applied in descending order, the EC50 value increases from 40 to 130 µM, having a Hill slope of ∼1. With increasing texp, there is a growing divergence between the two modes of application, for both GABA and β-alanine. Thus, as texp increases, the receptor population is gradually tuned toward one of two different EC50 values. Notably, the EC50 values can differ by as much as a factor of 17 for GABA and nearly a factor of 40 for β-alanine. It is obvious that it is not possible to tune the properties of an individual receptor, but using the CSPC technique, we are able to induce state distributions that remove certain amounts of receptors from the responsive fraction and allows us to control the appearance of the dose-response curves obtained from all receptors in a cell. For efficient tuning and separation of whole-cell receptor populations toward high or

Figure 4. Bicuculline affects the high and low EC50 preparations differently. (A, B) Dose-response curves (n ) 3 cells) for GABA and GABA coadministrated with 1 µM bicuculline, applied in descending or ascending order with texp ) 100 ms (A) or 3 s (B). The doseresponse curves for GABA together with antagonist are shifted to the right. The difference in dose-response function and dynamic range between ascending and descending order of application becomes less distinct when antagonist is present.

low EC50 values, longer (>3-4 s) texp are required. The EC50 and Hill slope values are sometimes regarded as estimates of the apparent affinity and cooperativity of the receptor. Although no solid physical background exists, the Hill slope can indicate the number of binding sites involved in binding and activation. The Hill slopes of ∼1 and ∼2, obtained for descending and ascending scans, respectively, fit well with the number of binding sites available for titration during each type of scan. Lifetimes of the State Distributions. To determine the lifetime of the receptors kept in desensitized states, experiments were performed by exposing patch-clamped cells to gradients of β-alanine (applied in either ascending or descending order), followed by a test pulse of 100 mM β-alanine after a specified time (trest). The difference between the response to the test pulse and Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

7951

Figure 5. Simulations of β-alanine-induced activity of the GABAA-receptor showing that increasing texp induces accumulation of receptors in desensitized states and cause transformation of the dose-response curves. (A) The model used for the simulations. (B-D) Simulated state distributions induced by gradients of β-alanine using texp ) 100 ms (B), 1 s (C), and 10 s (D), respectively. Note that the x-axis is labeled with the concentrations of β-alanine applied during each pulse. Dslow and Dfast indicate slow and fast desensitized states while O1 + O2 show the sum of the open states, equaling the conducted current. Doses were applied in descending (left) or ascending order (right), respectively. (E) Simulated dose-response curves derived from the simulated state distributions. Peak currents shown are from doses applied in descending order (black squares) or ascending order (red circles). (F) Simulated state distributions of unbound, mono- and double-liganded receptor during a descending (left) and an ascending (right) β-alanine scan with texp ) 1 s. R indicates unbound receptor, while RL and RL2 indicates mono- and doubleliganded receptor, respectively.

to an undesensitized response recorded from the same cell was normalized to the undesensitized response, according to ∆I ) (Test - Norm)/Norm, where Test and Norm are the test and normalization pulses, respectively. We used two different texp values (100 ms and 3 s), to discriminate between rapid and slow events (Figure 3). Using a short texp of 100 ms, the effect of current depression is gone already after 15 s whereas a texp of 3 s induces a current depression that remains for over 120 s, identical to durations induced by GABA.16 The lifetime of the transient state preparations thus ranges from seconds to minutes and can be controlled by the exact pattern of agonist application. For example, longer texp results in longer lifetimes of the desensitized states. Effect of Competitive Inhibition on the Different State Distributions. We next wanted to investigate the effect of different application patterns in a competitive antagonist assay. We chose the competitive GABAA receptor antagonist bicuculline as a model substance. We used bicuculline at the following concentrations; 0.01, 0.1, 1, 5, 10, and 100 µM coadministered with 100 µM GABA. As in the experiments with agonists, every second channel was loaded with buffer solution for ligand clearance, and a resting time in buffer of at least three minutes was used before a return scan was made. Dose-response curves were obtained using texp of 100 ms or 3 s, respectively. When the antagonist was 7952 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

applied in ascending order, IC50 values of 5.17 ( 0.4 (texp ) 100 ms) and 8.46 ( 1.4 µM (texp ) 3 s) were obtained, respectively. In contrast, descending-order applications yielded IC50 values of 3.56 ( 0.9 (texp ) 100 ms) and 1.84 ( 0.4 µM (texp ) 3 s). Thus, there is nearly a factor of 5 difference in blocking efficacy at long texp when comparing ascending and descending scans. We also performed dose-response experiments with GABA and a fixed concentration of 1 µM bicuculline, roughly corresponding to the IC20 value. The dose-response curves obtained in this manner are displaced to the right, toward higher concentrations compared to pure GABA responses (Figure 4A and B). The displacement is generally somewhat stronger for longer values of texp. For ascending-order scans, the dose-response curve is displaced to a larger extent than for the descending-order scans. It is clear that addition of the antagonist eradicates some of the differences observed in dynamic range and tunability of the receptor population with regard to GABA responses. The determination of the potency of a specific inhibitor will thus be influenced by previous receptor activations. Importantly, the pharmacological effect of a drug appears to be strongly dependent on the state distribution of the receptor population upon which it acts, a finding that will influence pharmacology as well as other studies of protein function.

Modeling of Receptor State Distributions. With the aim of being able to understand and explain the experimental findings reported herein, we performed computer modeling of the experimental protocols used. A model that describes GABAA receptor functionality must contain all the microscopic rate constants for binding and unbinding of agonist as well as opening, closing, and desensitization. We used a model (Figure 5A) proposed by Jones and Westbrook,20 with rate constants modified by Lindquist et al.25 and by ourselves16 and modeled the state distributions resulting from the experimental protocols used (Figure 5B-F). This model accurately predicts the time-dependent transformation of the dose-response curves for both GABA16 and β-alanine (Figure 5E). At short texp (100 ms), the slowly and rapidly desensitized states are of about equal importance and accumulation of receptors in these states is low during the application of the agonist gradients (Figure 5B). As texp increases, receptor accumulation in slowly desensitized states is increased (Figure 5C and D), causing peak current depression and changes of the dose-response curves acquired in descending and ascending order, respectively. The modeling shows that, at longer texp, there is a substantial aggregation of receptors in a slowly desensitized state, responsible for the effect studied here. Furthermore, during exposure to lower concentrations of agonist, the unbound receptor dominates over the singly bound and is, therefore, the main target for binding of both agonists and antagonists, whereas the opposite situation occurs during exposure to highest concentrations of agonist (Figure 5F). Although we were unable to conduct extensive modeling of the experiments with bicuculline, the state distributions induced by agonist application can give clues to how the antagonist may interfere with the binding and activity of the ligand-receptor complex. Bicuculline causes the largest effect at lower concentration of agonist and when it is applied together with ascending doses of agonist. As illustrated by the simulations, unbound receptors dominate over singly liganded receptors during these conditions. Binding of agonists to receptor molecules proceeds by rates slower than those expected, assuming a diffusion-limited process.21,27,28 This is caused by the energy required to induce structural rearrangements at the binding site eventually leading to channel opening.28 Competitive antagonists do not induce these conformational changes and therefore may bind faster, without performing any work on the receptor molecule. Estimations of binding and unbinding rates for competitive antagonists of the GABAAR show faster binding and slower (27) Akk, G.; Auerbach, A. Biophys. J. 1996, 70, 2652-2658. (28) Jones, M. V.; Jonas, P.; Sahara, Y.; Westbrook, G. L. Biophys. J. 2001, 81, 2660-2670. (29) Baumann, S. W.; Baur, R.; Sigel, E. J. Neurosci. 2003, 23, 11158-11166. (30) Ueno, S.; Bracamontes, J.; Zorumski, C.; Weiss, D. S.; Steinbach, J. H. J. Neurosci. 1997, 17, 625-634.

unbinding rates as compared to corresponding rates of agonists.21,25,28 At the low initial ratios of GABA compared to bicuculline, the antagonist is, in general, able to bind first and prevent subsequent agonist binding causing the proportion of receptors able to open to decrease and thus shifting the dose-response curve to the right. Higher agonist concentration results in a lower antagonist-to-agonist ratio and a corresponding decrease in the inhibitory potency of the antagonist. There is no clear picture of the precise mechanism that controls binding of bicuculline with regard to binding site specificity, although a preference for the first binding site has been suggested,29 and bicuculline has also been suggested to work as an inverse agonist,30 stabilizing nonconductive states of the receptor and opposing the GABA effect for channel opening and desensitization. Successful modeling of ligand-receptor interactions can be used to predict the activity and functionality of ion channel effector molecules. It is tempting to speculate that the CSPC experiments in combination with computer modeling may serve to increase throughput in pharmacologic characterization of ion channel effectors by reducing the number of experiments required for thorough evaluation of compound properties and development of new therapeutic agents. CONCLUSIONS Increased understanding of ligand-gated ion channel pharmacology is important for development of therapeutic agents against a wide range of diseases. It is common practice in pharmacology to evaluate the activity of different compounds by comparing EC50 or IC50 values. However, depending on the design of the experiments, and the specific receptors and substances used, the acquired data may be substantially skewed by the process of receptor desensitization. In other words, the same effector molecule can have dramatically different effects depending on the circumstances dominating during the application. By using the CSPC technique described herein, we are able to control the desensitization process and study the function of ion channel effector substances on well-defined and controlled receptor states. ACKNOWLEDGMENT This work was funded by the Royal Swedish Academy of Science, the Swedish Research council (VR), The Go¨ran Gustafsson Foundation, Cellectricon AB, and the Swedish Foundation for Strategic Research (SSF) through a donation from the Wallenberg Foundation. Received for review May 2, 2006. Accepted August 18, 2006. AC060812Z

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

7953