Single inward rectifier potassium channels in guinea pig ventricular myocytes

Transcription

1 Single inward rectifier ptassium channels in guinea pig ventricular mycytes Effects f quinidine J. R. Balser, D. M. Rden, and P. B. Bennett Departments f Pharmaclgy and Medicine, Vanderbilt University Schl f Medicine, Nashville, Tennessee USA ABSTRACT The effects f quinidine n single inward rectifier K channels were investigated in cell-attached patches with 4.5 mm pipette ptassium cncentratins. Under these cnditins, the single-channel slpe cnductance f the predminant cnductance level fthe inward rectifier channels was 3.9 ± 0.3 ps at membrane ptentials between - 75 and -150 my. Quinidine reversibly decreased the likelihd f channel pening t the main cnductance level withut reducing the single-channel cnductance, and als reduced the prbability f channel pening t subcnducting levels. Quinidine had n significant effects n the channel pen times, and the inhibitin f channel pening was nly slightly vltage dependent ver the range f membrane ptentials investigated. Quinidine induced a cmplete cessatin f channel penings fr brief perids (up t 2 min), suggesting that quinidine prmted ccupancy f a state frm which pening was less likely. Occasinal lng perids (up t an hur) with an absence f channel activity were als bserved but quinidine did nt appear t prmte this behavir. The data suggest that quinidine decreases the ability f the channel t enter bth main and subcnducting states. By binding t a particular clsed cnfrmatin f the channel, quinidine culd reduce the likelihd f channel pening. The main features f these bservatins culd be accunted fr using the three-state kinetic mdel prpsed by Sakmann, B. and G. Trube (1984b. J. Physi!. [Lnd.]. 347: ) with quinidine binding t the middle clsed state. INTRODUCTION Quinidine, ne f the ldest antiarrhythmic agents in widespread usage, is a naturally ccurring cinchna alkalid that suppresses a number f distinct cardiac ptassium currents. These include delayed rectifier ptassium currents (I K r Ix; Clatsky, 1982; Hiraka et ai., 1986; Rden et ai., 1988; Furukawa et ai., 1989), transient utward currents (Imaizumi and Giles, 1987), and inward rectifier (time-independent) K currents (Hiraka et ai., 1986; Salata and Wasserstrm, 1988). Hwever, the mechanisms which underlie blck f cardiac ptassium channels by quinidine have nt been characterized at the single-channel level. Unlike delayed rectifier and transiently pening K channels in cardiac cells, inward rectifier K channel activity can be recrded fr lng perids in cell attached patches; hence, its single channel prperties have been well characterized (Sakman and Trube, 1984a; Trube and Hescheler, 1984; Matsuda, 1988). We have investigated the effects f quinidine n single inward rectifier ptassium channels in islated guinea pig ventricular cells. In many cases in channels in patches f membrane excised frm cells have altered gating behavir, and in particular inward rectifier channel activity smetimes ceases after patch excisin (Trube and Hescheler, 1984). Address Crrespndence t Dr. Paul B. Bennett, CC-2209 Medical Center Nrth, Vanderbilt University Schl f Medicine, Nashville, TN We have therefre attempted t preserve the intracellular milieu f the channel by using the cell-attached patch cnfiguratin f the patch-clamp methd (Hamill et ai., 1981). Recently, Salata and Wasserstrm (1988) described suppressin f the time~independentptassium current (IKl) in dg ventricular mycytes. Their wrk shwed irreversible reductin f the steady-state utward and inward current (steady-state current abve and belw E K ). A number f studies have indicated that n detectable utward current flws thrugh single inward rectifier K channels in intact cells (Sakmann and Trube, 1984a; Jsephsn and Brwn, 1986; Mazzanti and DeFelice, 1988). In fact, utward currents have nly been bserved at the single-channel level when the intracellular side f the patch was expsed t artificially lw levels f Mg++ (Vandenberg, 1987; Matsuda, 1988). Hence, the mechanism respnsible fr suppressin f the utward steady-state current by quinidine remains in questin. Hwever, the macrscpic inward current suppressed by quinidine at membrane ptentials negative t E K mst likely results frm blck f inward rectifier K channels; we have attempted t clarify this by investigating the effects f quinidine n single inward rectifier K channel penings at membrane ptentials belw E K Previus studies f single inward rectifier K channels have utilized high extracellular ptassium cncentratins ( mm), presumably t increase the ability t reslve single K channel penings (Sakmann /91/01/150/12 $2.00 Biphys. J. BiphySIcal Sciety Vlume 59 January

2 and Trube, 1984b; Trube and Hescheler, 1984; Matsuda, 1988). Hwever it is knwn, at least fr Na channels, that the cncentratin f permeant in can alter the blck caused by sme drugs (Cahalan and Almers, 1979). It is als knwn that the cnductance and kinetics f the inward rectifier ptassium channel are altered by changes in external ptassium cncentratin (Sakmann and Trube, 1984b; Chen et ai., 1989); alteratin f these channel prperties may secndarily affect drugchannel interactins. Therefre, we have studied the effects f quinidine n these single K channels under cnditins f nrmal extracellular ptassium cncentratin. METHODS Cell preparatin Acutely islated guinea pig ventricular mycytes were prepared as previusly described (Farmer, 1983; Bennett et ai., 1987). Hearts were perfused via the ascending arta with a minimum essential medium cntaining cllagenase (200 U/ml; Wrthingtn type II; Cper Bimedical, Malvern, PA). After perfusin fr 15 min the ventricles were minced, washed and filtered thrugh 200 ILm tefln mesh. The cells were then placed in Medium 199 (Sigma Chemical C., St. Luis, MO) with 1 mm calcium and stred in an incubatr at 37 C until used within 12h. Slutins Inic currents thrugh ther channels and transprt systems were suppressed by inic substitutin and blckers. Patch pipettes usually cntained (in millimlar): KCI 4.5; N-methyl-d-glucamine-CI 150; MgCl, 2.0; CaCl, 0.1; CdCl, 0.1; LaCI, 0.1; Hepes 20; glucse 12. N-methyl-d-glucamine was used as an inic substitute fr Na+. Channel blckers (Cd++, La+++) and a reduced extracellular calcium cncentratin were used t eliminate calcium-dependent currents. External calcium was lwered t 0.1 mm, but was nt eliminated cmpletely, t avid permeatin f Ca channels by ther ins (Hess and Tsien, 1984). The bath usually cntained (in millimlar): NaCI 145; Hepes 10; MgCl, 1.0; CaCl, 1.8; KCI 4.5; glucse 10. These slutins maximized islatin f ptassium currents while minimizing cntributins frm ther inic currents which in turn will imprve vltage clamp cntrl and prevent inadvertent uncntrlled changes in membrane ptential. Exceptins t these slutins are nted in the figure legends and text. All experiments were perfrmed at C. Patch clamp recrding Patch electrdes were made frm capillary tubing (Mdel 1812; Radnti Glass Technlgy, Inc., Mnrvia, CA), and after heat plishing had resistances f 9 ± 4 Mil when filled with the slutin listed abve. Electrdes were cated near the tip with Sylgard (Cmpund 184; Dw-Crning Crp., Midland, MI) t reduce capacitance and nise. This allwed reslutin f K channel penings at membrane ptentials within 15 mv f E K T reslve the small unitary currents (0.05 pa) required fr this study, it was necessary t btain seal resistances in excess f 100 Gil (Hamill et ai., 1981). Thus, we reprt here nly the results frm patches meeting these criteria (780 ± 510 Gil; N = 9). In mst experiments, current recrds were lw pass filtered at 100 Hz (-3dB; fur-ple Bessel filter) befre sampling at 500 Hz (five times the filter cut-ff frequency) by a 12-bit analg-t-digital cnverter cntrlled by a PC/AT micrcmputer (IBM Instruments, Inc., IBM Crp., Danbury, CT); exceptins are nted in the figure legends. The dead time f this filter is ms when the crner frequency (f,) was set t 100 Hz, and the data were sampled at an intelval near the filter dead time. Under these cnditins using a 50% threshld criterin fr event detectin, events with duratins shrter than the filter dead time were never detected. Events lnger than ms culd be reliably detected, but events with duratins between the filter dead time and 2.3 rns culd be detected nly part f the time and their duratins will be underestimated. We have nt crrected fr this sampling errr, as it apparently had little effect n ur analysis because missin f events up t 18 ms did nt distrt the prbability distributins. In sme cases, a deplarizing slutin cntaining 150 mm KCI and 10 Hepes was used t zer the cell resting ptential and allw accurate measurement fthe single-channel slpe cnductance ("y) and reversal ptential (V R.,,) under cnditins where the trans-patch ptential was accurately knwn. The patch pipette cntained 4.5 mm ptassium which determined the bselved reversal ptential. We then used the measured cnductance value as a standard t derive the membrane ptential (V M ) frm the measured single-channel current (i) when the cells were nt deplarized. In five experiments where the bath and pipette ptassium cncentratins were 4.5 mm, the average resting ptential btained in this fashin was ± 4.5 mv (23 C). In six additinal cells in this external slutin where the patch was brken, the measured resting ptential was ± 6.4 my. Therefre, membrane ptentials are reprted as an abslute trans-patch ptential. After btaining a Gil seal, cntrl data were cllected fr up t 30 min, then quinidine (10-50 ILM) was superfused at 1 ml per min thrugh a O.5-ml bath vlume and the channel activity was mnitred by cmputer. Whenever pssible, the drug was washed ut f the bath t btain a pstdrug cntrl. This usually required min f reexpsure t cntrl slutins, and the drug effects were nly partially reversible. Because f the criteria that we established fr making measurements, we have reprted results frm 9 ut f the 130 membrane patches btained in this series f experiments. The mst cmmn reasn fr rejecting a patch fr further analysis was that the seal resistance was < 100 Gil, r that the Gil seal was lst befre cmpletin f an experiment (70%). Hwever, a significant number f patches culd nt be used because f the presence f t many (> 6) channels (4% f patches), because inward rectifier channels were nt detected (15% f patches), r because f disappearance f the K channel activity during the cntrl perid (4%). In additin, sme patches were nt included in the analysis due t the presence f channels ther than the inward rectifier (7%). Data analysis T assess the steady-state prbability f channel pening at different membrane ptentials, patches were repeatedly clamped t test ptentials 40, 70, r 100 mv negative t the resting ptential fr perids frm 4-8 s. Because the prbability f channel pening is nnstatinary during the first m, the first 200 sample pints ( m) after the vltage step were mitted frm the steady-state analysis. Thus, the channel pening prbabilities reprted reflect nly the steady-state cnditins and nt the initial peak pening prbabilities. During the cntrl and drug expsure perids, between 16 and 64 test pulses were applied at each membrane ptential tested. Data frm (1) Balseret al. Inward Rectifier K Channels and Quinidine 151

3 multiple experiments are reprted as means plus r minus ne standard errr f the mean (mean ± SEM). Differences between cntrl and drug measurements were analyzed using a paired Student's t-test. Data were analyzed with custm sftware written in BASIC (Micrsft, QuickBASIC) using an event detectin scheme based n a half-amplitude criterin as described by Clquhun and Sigwrth (1983). Single channel recrdings were first analyzed by generatin f all pints amplitude histgrams which allwed identificatin fcurrent levels assciated with clsed, substate, and full amplitude penings. These mean current levels were then used in the fllwing event detectin scheme. An event was registered if tw cnsecutive digitized samples fell within a predetermined windw f current levels. As described by Sakmann and Trube (1984a), at least fur substate cnductance levels exist fr inward rectifier penings; hence, t evaluate the prbability f at least ne channel pening t any level (main r substate penings, the threshld fr event detectin was set t ne half the amplitude f the lwest amplitude substate current level. T determine the prbability f substate penings alne, the same lwer threshld was used and an upper limit fr event detectin was impsed t eliminate current levels abve the highest substate level (thereby eliminating main-state penings). Fr patches with mre than ne channel, we als calculated the prbability f single main state penings by setting the detectin windw abve and belw the main current level t eliminate bth substate penings and multiple simultaneus channel penings; such single mainstate pening prbabilities culd then be cmpared with the prbability f substate penings belw the mainstate. Windw limits were ptimized by visual inspectin. Estimates fr the pen-channel prbability, including the individual prbabilities f main and substate penings, were calculated as the fractin f time a channel spent at each current level. Althugh several substate levels exist, it was nt pssible t reliably quantify differential drug effects n the individual substate levels due t their lw prbability f ccurrence. This prblem was exacerbated in sme cases by the limited duratin ver which data culd be cllected. Hence, in all cases the "substate prbability f pening" refers t the lumped prbability f all substate penings belw the main amplitude level. In patches with mre than ne channel, the event detectin scheme culd prvide the verall prbability that at least ne channel was pen by keeping track f the fractin f time the current was abve the threshld fr detectin f an pen channel. T derive the prbability f pening f a single channel, we assumed that the pening f different channels were nt mutually exclusive events and therefre beyed the summatin law fr the prbability f the unin fn events (Rss, 1972) where N is the number f channels and an event is a channel pening as fllws. P(at least 1 channel pen when N channels are present) = I (prbability fn channels pening taken 1 at a time) - I jprbability fn channels pening taken 2 at a time] +I jprbability fn channels pening taken 3 at a time] I (prbability fn channels pening taken N at a timel. (2) We assumed that channels were functinally independent and had equal pening prbabilities (Clquhun and Sigwrth, 1983). Hence, the prbability (P) f at least ne channel pening when fur are present is as fllws (p is the prbability f ne channel pening and the sum fn events taken r at the time is n!/[(n - r)!r!]): Pial least ne channel penl = 4p - 6p 2 + 4p 3 - p4. (3) Because P was knwn, it was pssible t slve numerically fr the rts f the plynmial equatins t prvide the value fp (between 0 and 1). In each case nly ne such slutin was fund. The value f N fr each experiment was estimated by cunting the maximum number f channels pen simultaneusly during the drug-free perid (usually at the beginning f each hyperplarizing step). Others have fund that this simple cunting methd wrks as well as mre cmplicated statistical determinatins f channel number (Aldrich et ai., 1983; Sakmann and Trube, 1984b). Examinatin f the amplitude histgrams revealed that in many cases the peaks were prly separated frm each ther causing difficulty in calculating the area under the peaks fr analysis f prbabilities. Therefre, we als analyzed the data using a secnd methd which utilizes a mean-variance analysis recently described by Patlak (1988). The technique invlves calculating a mving average current and variance in a windw f 5-20 data samples. The windw is advanced ne sample pint and a new mean and variance are calculated. The sample variance is relatively lw when the channel is pen r when it is clsed, but during transitins frm clsed t pen (r vice versa) the variance is high. A graph f this variance as a functin f the mean current (Fig 4 a) results in a family f parablas that arc between regins f lw variance; these lw-variance pints typically indicate legitimate pen and clsed channel-current levels, while high-variance pints represent either transitins between levels r very brief events. Whereas this technique clarifies the separatin between current levels, the relative prbabilities f the varius cnductance states may be distrted because samples are excluded frm the analysis and thus the areas under the amplitude histgrams may be decreased. In cntrast, althugh there are uncertainties assciated with the chice f levels in the "levels detectin" methd the relative prbabilities f ccupying particular levels are les~ distrted by this technique (althugh picking ut the levels can be mre difficult). In patches with mre than ne channel, mean pen times were determined by using a first-pen, first-clsed scheme. This methd f making clsing assignments, althugh in principle a biased estimatr, was used because it gave results similar t methds using randm assignment f a clsing t an pening and was simpler. Because the prbability f channel pening was relative lw ( < 0.3) when using 4.5 mm external ptassium, and because the substate penings were relatively rare in many cases, the pen times were analyzed fr all the pen states f the channel as a grup t prvide sufficient numbers f events. Hence, ccasinal penings t substate levels were cnsidered penings and were nt distinguished frm main-state penings fr purpses fpen time analysis (similar t Sakmann and Trube, 1984b). Events which pened t a substate level and then t the main-state level were analyzed as ne event; a new event ccurred either when full clsure ccurred r when the current level exceeded the threshld fr a secnd pening. Because the pen-time distributins were well apprximated by a single expnential (see Results), then either the substate penings in these cases were nt frequent enugh t cause a detectable secnd expnential (als nted by Sakmann and Trube, 1984b), r the mean pen times f the substate events were similar t that f the main state. Because the penings were described by a single expnential, the maximum likelihd estimate (arithmetic mean) f the mean pen time was als calculated. This methd f estimating the mean pen times was preferable t expnential fitting because it avided sme f the pitfalls f binning the bservatins (Clquhun and Sigwrth, 1983). Data were fitted using methds described in detail elsewhere (Balser et ai., 1990). A mdified SIMPLEX algrithm (Neider and 152 Biphysical Jurnal Vlurne 59 January 1991

4 Mead, 1965) fr minimizing the residual sum f squares was used. All analysis f the effects f quinidine were btained frm paired bselvatins. RESULTS Because mst f these experiments were cnducted in cells that were nt expsed t elevated ptassium, we were cncerned abut the uncertainty f the resting membrane ptential in nndeplarized cells. Therefre, as described abve, vltage ramps were used t determine the single-channel cnductance and then Eq. 1 was used t estimate the resting membrane ptential. Fig. 1 (tp) shws current recrded during vltage ramps frm my. The cell was deplarized in 150 mm KCI, therefre, the clamp ptential shuld equal the transpatch membrane ptential and thus allw an accurate determinatin f the single-channel cnductance. The currents elicited frm tw f the vltage ramps are shwn. The channel was primarily pen belw -80 mv in ne case, with nly briefclsures, but was fully clsed Clsed Open Difference current 4 ps / sw 9, i(pa) -0.5 (pa) FIGURE I Single-channel current during vltage ramps frm -15()"'{) mv. The cell was deplarized in 150 mm KCI which allwed accurate measurement f the single-channel slpe cnductance ('Y) and reversal ptential (V Rev )' The patch pipette cntained 4.5 mm KCI. (Tp) Tw different recrds, ne recrded when the channel was clsed, and anther recrded when the channel was pen at -150 mv and remained pen thrughut the ramp (except fr brief clsures). (Bttm) The leak crrected single-channel K current. The averaged backgrund current was subtracted frm the recrd with the channel pen. The slpe cnductance ('Y) was btained by fitting a line t the current elicited at membrane ptentials between 130 and -90 mv. The zer current ptential (x-intercept r V Re.) was -77 mv. thrughut the vltage ramp in the ther case. Slpe cnductances were estimated frm the linear least squares regressin n the difference current (Fig. 1, bttm) measured between -90 and -130 my. Althugh the pen-channel current-vltage relatinship is nt perfectly linear ver an extended range fptentials, deviatins frm linearity were extremely small ver the range we have used. The reversal ptential was estimated frm the x-intercept f the fitted line. The single-channel cnductance was 3.9 ± 0.3 ps (n = 3) and the average reversal ptential was ± 2 my. As reprted by thers (Sakmann and Trube, 1984a; Vandenberg, 1987; Matsuda et ai., 1987) we bserved strng inward rectificatin f this channel in the cellattached mde (the difference current was negligible at membrane ptentials mre psitive than -80 my). Fig. 2a shws representative channel activity recrded frm an n-cell patch n a nndeplarized cell; bth the pipette and bath slutins cntained 4.5 mm KCI. The patch was vltage clamped t -174 mv frm the resting ptential f -74 my. There were single penings t the main cnductance level (1 pen), duble penings (2 pen), and sme additinal less frequent penings t lwer cnductance levels (a substate, b substate, and c substate). These did nt result frm a separate class fk channels, but reflect penings f the inward rectifier K channel int a lwer cnductance substate cnfrmatin and have been bserved by thers (Sakmann and Trube, 1984a; Jsephsn and Brwn, 1986). These lwer cnductance penings were 1/4, 1/2, and 3/4 f the main current level. We have bserved in patches with nly ne channel that these current levels never exceeded the main (3.9 ps) cnductance level as they wuld fr separate channels if the lwer amplitude penings were superimpsed upn the main cnductance. Fr a single channel with substates, ccupancy f the varius pen states is mutually exclusive and verlap des nt ccur. Fr example, in an experiment where substate penings were especially prminent and nly ne apparent channel was bserved, the prbability f pening was fr the 3.9 ps events whereas the prbability f all lwer amplitude events was If these events represented separate channels, the likelihd f bserving simultaneus penings during the perid recrded (mre than 40 min) was> Because n simultaneus penings were ever bserved, it was cncluded that the events did nt result frm separate channels. This is cnsistent with the findings f Sakmann and Trube (1984a). As shwn in Fig. 2 b, 50 J.LM quinidine caused a decrease in channel activity that was partially reversible when quinidine was remved. The effect f quinidine was t substantially reduce the prbability f channel pening, with n effect n the magnitude f the unitary current. In seven experiments (see Table 1), the mean Balseretal. Inward Rectifier K Channels and Quinidine 153

5 -74 ~lv7~l c n_t_r 1 1 "..._~ ~._~ ~./1,..,,,1. PAL ~ ~~~. b substate, pen 2 pen C substate 1 pen a substate 1 pen 1 sec n u Quinidine 1 pen r I In ~ '1 1 pal. 1" 1 sec I./lle8 c35 Washut ~.CJ IIftnrnn, I~ 1 pen V 2 pen lffi[ IU WV wyrl PAL 1 sec._._.._.._._.._.._._._..ll1j"t4rul.!!!!!!! u!..._ _ _. -"-r...! I_~!... _..-,...e-._.._.._._.._... r:_. _.._.._... substst. 1 pen FIGURE 2 Reversible inhibitin f single-channel activity by quinidine. (a) Five representative recrds btained during 8-s steps t -174 mv in cntrl. Pipette [K] = 4.5 mm, bath [K] = 4.5 mm. There are fur channels in the patch. Sample rate 250 Hz, filter 50 Hz. The main unitary current level assciated with single penings is indicated by the line labeled: 1 pen. Als shwn are the current levels assciated with duble penings (2 pen), and three levels f subcnductance penings (a substate, b substate and c substate). (b) The number f penings was reduced after 15 min f superfusin with 50 ~M quinidine. The subcnductance penings were eliminated. (c) The frequency f penings increased again after 20 min f drug wash-ut and subcnductance penings were seen again. rati f the single-channel current during drug expsure t that befre drug expsure was 1.00 ±.024; in cntrast, the prbability f pening was reduced in six ut f seven experiments. Nte that fr the prbabilities f channel pening given in Table 1, the prbabilities f pening t the main cnductance state and lwer cnductance substates were cmbined. The prbability f single channel pening fr multichannel patches was calculated using Eqs. 2 as described abve. The effect f quinidine shwn in Fig. 2 b ccurred after an expsure f - 10 min; nte that single and duble penings were present, but there were n substate penings. The drug effect was partially reversible 20 min after beginning t wash quinidine frm the bath (Fig. 2 c): bth the substate penings (arrw), as well as duble penings were again bserved. If quinidine acts as an pen-channel blcker (Armstrng, 1966), then it shuld shrten the time the channel spends in the pen state by terminating the penings when blck ccurs. We tested whether quinidine was an pen-channel blcker by directly bserving the lifetimes f the pen channel. The distributin f 154 Biphysical Jurnal Vlume 59 January 1991

6 TABLE 1 Effects f quinidine n single K channels Cell N* [K]b"h [K]p'p'lt, vtrest V m -\ -\ I cni I quin iquin/illcnt P ent P quin [Quin] mm mm mv mv pa pa ~ 4/ / O.OlD 50 5/6a /6b / / / *Number f channels in the patch; Imean V"" (frm cells 2--6) = ± 4.9 my; N = 5; Imain cnductance level penings nly; "mean iq,;n/i",n' = 1.0 ± 0.02; N = 7. pen times was well described by a single expnential befre and during quinidine expsure. Fig. 3 summarizes the pen times btained in several paired experiments, quinidine did nt significantly alter the pen lifetimes at the membrane ptentials tested (-96 t -177 mv). The ratis f the paired pen times are pltted in the upper graph and the mean rati was nt significantly different frm unity. The gruped cntrl pen times, averaged ver a narrw vltage range, were (J.L ± SE): 249 ± 60 ms (-96 t -117mY; n = 4),118 ± 23 ms (-126 t -147; n = 5), and 82 ± 1 ms (-174 t -177; n = 2). The crrespnding gruped mean pen times frm paired experiments in quinidine were 198 ± 48 ms, 119 ± 53 ms, and 54 ± 2 ms. As thers have bserved (Sakmann and Trube, 1984b) in elevated pipette ptassium cncentratins, the mean pen time decreased as the membrane ptential became mre negative. We estimated the effects fquinidine n the prbabil- ~ 300 E'" ::l 200 E i=., c::: 100 a. Cntrl I Quinidine Membrane ptential (mv) FIGURE 3 Effects f quinidine n pen channel lifetimes. Summary f results frm five experiments measuring pen times at varius membrane ptentials. Open bars indicate cntrl pen times, clsed bars indicate the presence f quinidine. The tp panel shws the ratis f the mean pen times in quinidine t the mean pen times in cntrl. A rati f 1.0 indicates n effect. The mean and standard deviatin f all ratis is shwn as the pen circle pltted in the middle f the vltage range. ItIes f channel pening using tw techniques (see Methds). Fig. 4 shws an analysis f the cnductance levels fr cell 4 (Tables 1 and 2) at -131 my; this patch cntained nly ne channel. Fig. 4 a shws the variance (rdinate) as a functin f the mean current (abscissa) in each f 4 different vltage-clamp steps. These histgrams (Fig. 4 band c) included nly thse pints having a variance less than r equal t the backgrund variance. The lw-variance pints indicate levels fcurrent assciated with clsed, substate penings, and main state penings; the high-variance pints are assciated with sample pints during transitins between cnductance levels r very brief penings and clsures. In this experiment nly events > 18 ms had lw enugh (backgrund) variance t be included in the analysis. Fig. 4 b shws the amplitude histgram when all sample pints frm 64 sweeps are included (dtted line) and when sample pints having a variance exceeding the backgrund variance are excluded. Nte that excluding the high-variance pints clarifies the divisins between the cnductance levels. Fig. 4 c shws the amplitude histgram fr the same cell after 15 min f quinidine (50 J.LM) expsure. Quinidine reduced the prbability (area under the curve) f pening t the main cnductance level by 20%, whereas the prbability f substate penings was reduced by 98%. Fig. 5 summarizes results frm five patches (cells 1-5). In the lwer panel, the prbability f single-channel pening (bth substates and the main pen state) is shwn as a functin f membrane ptential befre (pen bars) and during quinidine superfusin (slid bars). Fr each patch, the effects f quinidine were assessed at up t three different membrane ptentials (as patch lngevity wuld permit); each pair f bars (pen, slid) in Fig. 5 thus represents a paired bservatin f the effect f quinidine at a particular membrane ptential. Quinidine reduced P pen in all cases except at -96 my, and in ne additinal membrane patch where the verall prbability f channel pening was increased due t an increase in subcnductance penings (data nt shwn in Fig 3, Balseret al Inward Rectifier K Channels and Quinidine 155

7 a. 7.5 b. 'j" « ;:- c. 'in c Q> 0 E ~ a: 'j" « ~ 1.0 '"C Q> ~ : ,.0 a: Quinidine (50 ~M) Substates '<: Full a.!! M I 0 pen 5.0! ~ Q> I.) 2.5 c.,.~ > Single Channel Current (pal FIGURE 4 Effects f quinidine n single-channel cnductance and substates. (a) Mean-variance plt f fur cntrl sweeps at -131 my. The plt shws at least fur levels assciated with lw variance, including clsed channels, fully pen channels, and tw substate levels. Transitins between levels are shwn as pints with higher variance. A mving windw f five sample pints was utilized. (b) Amplitude histgrams fr 64 cntrl sweeps frm the same experiment at -131 my. The dtted line indicates the histgram when all pints are utilized; the slid line indicates the amplitude histgram when nly sample pints with a variance less than r equal t the backgrund variance were included. The relative number f samples per bin (prbability density) is pltted n a lgarithmic rdinate as a functin f current level f each sample. A peak in the histgram indicates a predminant current level. The peak centered at zer picamperes indicates the clsed channel current level. Other peaks indicate the amplitude (n thex-axis) fcurrent thrugh an pen state f the channel. Ntably the substate peaks are mre easily distinguished when high-variance pints are excluded. Bin width = pa. (c) Single-channel current amplitude histgram at -131 mv in the presence f 50 ~M quinidine. Again, sample pints with variance greater than backgrund variance were excluded. In this case, quinidine reduced the prbability f full penings by 20.1 % and substate penings by 98.2% (Table 2). cell 6 in Table 1). The tp panel f Fig. 5 shws the rati f P pen in quinidine relative t cntrl. A rati f unity indicates n effect. The results frm bth the levels analysis technique (slid circles) and frm areas in amplitude histgrams after mean-variance analysis (pen circles) are shwn. Quinidine reducedp pen and there was a vltage-dependent trend tward mre blck at mre negative membrane ptentials. The slid line represents a least squares linear regressin t the data represented by the slid symbls. The fitted slpe was significantly different frm zer (P < 0.05). The dashed line indicates a similar regressin line thrugh the pen circles with a slpe als significantly different frm zer (P < 0.01). Thus, bth methds f btaining P pen gave a similar result. In 4/4 experiments where substate penings were prminent in cntrl, the prbability f substate penings was cnsistently reduced mre than prbability f main state penings (Table 2). An exceptin t this trend was fund in ne experiment (cell 6 in Tables 1 and 2). In this experiment, n substate penings were bserved befre drug expsure; hwever, lwer amplitude penings were bserved after drug expsure. Althugh the prbability f mainstate pening was reduced, the verall prbability f channel pening was in fact increased because f the appearance f these additinal penings (hence, the apparent increase in the prbability f channel pening fr cell 6 in Table 1). Whereas we cannt readily accunt fr this difference, it appears cnsistent with a recent brief reprt frm ther investigatrs (Sat et ai., 1989). In ur experiments, hwever, in all cases the single channel current amplitude histgrams indicated that the main cnductance peaks were nt shifted by quinidine. We frequently bserved brief perids during which channel penings were absent even withut quinidine (hereafter referred t as null episdes) which typically lasted less than 2 min. These null episdes suggested that the channel ccasinally entered a state frm which pening was unlikely. We tested whether the effect f quinidine was t increase the number f null episdes, and thereby prmte ccupancy f such a state. In paired bservatins frm three patches cntaining nly ne channel, the fractin f episdes withut penings was increased frm 0.29 ± 0.2 in cntrl t 0.43 ± 0.1 in quinidine (p < 0.05). In additin, quinidine (50 f-lm) als reduced the cnditinal prbability f channel pening in episdes that cntained penings by 14 ± 7% (p < 0.05). Hence, an increase in null episdes accunted fr a prprtin f quinidine's effect t reduce the prbability f channel pening; hwever, the cnditinal prbability f pening in episdes with penings als decreased. DISCUSSION We have investigated the inhibitin f current thrugh single inward rectifier K channels by quinidine in guinea pig ventricular mycytes. Channel permeatin, as shwn 156 Biphysical Jurnal Vlurne 59 January 1991

8 TABLE 2 Effects f quinidine n the fractinal pen time f the main- and sub-cnductance levels Cntrl Quinidine % decrease Cell V M F main F sub F main F sub F main F sub 1. 4/ / /6a /6b / F m in = fractin f time that ne channel was pen at the main cnductance level; F,ub = fractin f time that ne channel was pen at nnzer current levels belw the main cnductance level. Patch 4 cntained nly ne channel, therefre these values refer t the prbability f channel pening t the main r subcnductance levels, respectively. Fr the ther multichannel patches, these values refer t the cnditinal prbability that ne channel is pen t the main r subcnductance levels given that there were N channels present in the patch. Cell 1, 10 IJ.M quinidine, [Klbn1h = 150 mm; cells 2-4, 6, 50 IJ.M quinidine, [K]bnth = 4.5 mm; cells 5, 7, n substate penings in cntrl r during quinidine; ++ substate penings were bserved in quinidine but nt in cntrl. by Armstrng (1971) fr TEA blck f delayed rectifier K channels in nerve, may mdulate the characteristics f channel blck (see als Cahalan and Almers, 1979). Therefre, we used n-cell patches and expsed the external side f the channel t 4.5 mm ptassium, instead fthe elevated ptassium cncentratins used in c z 0 w()... a:> z ~ a. 1.0] ~-"-~ 0.5.~ ~W)-l W) UCntrl I Quinidine Membrane ptential (mv) FIGURE 5 Effects f quinidine f K channel pening prbability. (Bttm) The prbability f single-channel pening (bth substates and the main pen state) is shwn as a functin f membrane ptential frm five experiments at ne r mre membrane ptentials. Each pair f bars (pen, slid) in the bttm panel represents a paired bservatin f Ppen befre and during quinidine at a particular membrane ptential. Fur f the experiments shwn used 50 IJ.M quinidine and ne was in 10 IJ.M quinidine. Because multiple vltages were examined in sme experiments, the fllwing symbls indicate thecell number in Table 1: squares, cell 1; circles, cell 2; triangles, cell 3; inverted triangles, cell 4; stars, cells. (Tp) Ratis fp pen in quinidine t Ppen in cntrl. A rati f 1.0 indicates n effect. Slid circles shw the data frm Fig 3 a; the mean and standard deviatin f all ratis is shwn as the pen circle pltted in the middle f the vltage range. The slid line thrugh the data represents a least squares linear regressin with a slpe that was significantly different frm zer (p < 0.05). The larger pen circles shw the results when the prbabilities were calculated frm histgrams derived frm mean-variance analysis (see Methds). The dtted line shws the least squares linear regressin f these data pints; this slpe was als significantly different frm zer (p < 0.01). earlier studies (Sakmann and Trube, 1984a and b; Trube and Hescheler, 1984; Kurachi, 1985; Matsuda, 1988). Sakmann and Trube (1984a) bserved a pwer law relatinship between single K channel cnductance ('Y) and ptassium cncentratin ver the mm cncentratin range they studied. By extraplatin f this relatinship ('Y = 1.3[K+]~ 62 ps), the single-channel cnductance shuld be ps in 4.5 mm external ptassium. Using 10 mm K in the patch pipette, Mazzanti and DeFelice (1989) measured a cnductance f -14 ps (frm their Fig. 10), and extraplated their cnductance data ( mm K) t 2-3 ps fr mm K. Hence, ur measurement f 3.9 ± 0.3 ps in 4 mm [K] agrees fairly well with the predictins fther investigatrs. The main effect f quinidine bserved n the channel activity was reductin f the prbability f channel pening. These results are similar t thse reprted by Jsephsn (1988) fr lidcaine effects n inward rectifier K channels; in that study, lidcaine reduced the prbability f channel pening with n effect n the single-channel cnductance. Quinidine had n effect n the unitary current amplitude r n the duratin f channel penings. Openchannel blck as described fr TEA+effects n delayed rectifier K channels by Armstrng (1966; squid axns) and Spruce et al. (1987; frg skeletal muscle), is usually expected t either reduce the mean pen time f the channel r reduce the apparent single-channel cnductance (fast-flicker blck). Rather, quinidine appears t act by reducing the likelihd that channels will pen. Because we have used n-cell patches t preserve the natural internal envirnment f the channel, ur experiments cannt reslve whether this is a direct effect f quinidine n the channel prtein. The effects f quinidine were nt immediate, smetimes taking as lng as 15 min after expsure t the bath slutins. Because the external surface f a channel faces the electrde slu- Balser et al. Inward Rectifier K Channels and Quinidine 157

9 tin, which is initially drug free, the drug must first diffuse int the cell t gain access t the channel. Mre than 90% f the quinidine mlecules are charged at physilgical ph, therefre entry int the cell is expected t be slw. After entering the cell, quinidine may bind t a prtin f the channel expsed t the cytplasm, r it culd diffuse back thrugh the patch f membrane int the electrde and then bind t the uter surface f the channel, althugh this seems less likely. Alternatively, nce in the cell, quinidine culd interfere with a cytplasmic factr imprtant fr channel activatin, r diffuse int the membrane and directly inhibit the channels by binding in the lipid phase. Any f these mechanisms culd explain the relatively slw nset f the drug effect. We bserved cnsiderable variability in the magnitudes f drug effects (Tables 1 and 2; Fig. 3). It appears that this variability relates t quinidine's mechanism factin n the K channel in intact cells. Ifchannel regulatin r blck are linked t membrane bund r cytplasmic factrs such as G-prteins r camp fr example, the variability in channel gating and blck culd be explained by the variability in these uncntrlled systems. Quinidine is knwn t reduce cyclic AMP levels inguineapigatrial cells (Mirr, 1981) which wuld be cnsistent with an indirect effect fquinidine if the channels were mdulated by cyclic AMP, althugh there is n direct evidence fr this at present. Additinal experimentatin will be required t discriminate these pssibilities. The effects f quinidine were nly partially reversible (Fig. 2). Our bservatins are prbably nt incnsistent with the apparent irreversible quinidine-induced suppressin f time-independent whle-cell inward current (Salata and Wasserstrm, 1988). Ntably, inward rectifier K channel activity may cease abruptly and indefinitely in excised patches. Such a lss f channel activity is difficult t separate frm a drugeffect, but it ccurs in the absence f a drug als. If a lss f channel activity ccurred in the dialyzed whle-cell experiments during expsure t quinidine, such an effect might be errneusly attributed t the drug. In urexperiments, suppressin f channel penings by quinidine were clearly reversible upn remval f the drug. Further, althugh inward rectifier channels may cease functining spntaneusly, until this ccurred the inward rectifier channel behavir appeared temprally stable. Three f the patches reprted (cells 4, 5, and 7 in Table 1) cntained nly ne channel and in all three cases, the likelihd f channel pening was reduced by quinidine. Hence, spntaneus lss f channel activity cannt explain the results frm these single-channel patches. Salata and Wasserstrm (1988) did nt bserve vltage r use-dependent suppressin f whle-cell I KI ; hwever, their data were acquired nly at ptentials psitive t -110 my. Our data exhibited a vltagedependent reductin f Ppe. by quinidine at membrane ptentials between -100 and -180 mv (see Fig. 5). In mst cases we bserved a preferential reductin in the likelihd f substate penings in the presence f the drug; althugh the mechanistic implicatins f this bservatin are nt precisely knwn, ne pssibility is that quinidine altered the free-energy barriers separating the pen cnfrmatins f the channel prtein. Thus, the channel was less likely t pen t the main cnductance level and even less likely t assume a substate cnfrmatin. In ne experiment during quinidine expsure an increase in the ccurrence f reduced amplitude penings was bserved. This may represent sme nnstatinary behavir f the prbability f substate penings that ccurs ver a time perid lnger than ur ability t recrd the events. Alternatively, it culd represent hetergeneity in the respnse t quinidine because similar effects have been bserved by thers (Sat et ai., 1989) when using 140 mm ptassium cncentratin in the patch pipette. We cannt distinguish these pssibilities at the present time. Trube and Hescheler (1984) reprted that inward rectifier K channel activity ften disappears after patch excisin. We have als bserved lng perids (minutes up t an hur) withut activity in cell-attached patches where a channel activity was previusly bserved; ccasinally channel activity ceases and des nt return. These bservatins suggest the pssibility f an alternate gating mde, and may indicate that cytplasmic r membrane-bund factrs play an imprtant rle in the regulatin fthis channel. Similar lng "nnactivatable" perids have als been nted fr n-cell patches f single-cardiac Na channels (Hrn et ai., 1984; Khlhardt et ai., 1988); this Na channel behavir has als been interpreted as a spntaneusly ccurring "nnactivatable" gating mde (Khlhardt et ai., 1988). Because the effect f quinidine n single channel patches included a reductin in the cnditinal prbability f channel pening in episdes (hyperplarizing steps) that cntained activity, it is clear that quinidine did nt act exclusively by placing the channel in such a "nnactivatable" state. Further, althugh in sme patches channel activity ccasinally disappeared fr lng perids f time (minutes up t ne hur), this was nt crrelated with the applicatin f quinidine. We have attempted t interpret ur results within the cntext f the mdel prpsed by Sakmann and Trube (1984b). This mdel culd accunt fr the brief clsures 158 Biphysical Jurnal Vlume 59 January 1991

10 ccurring between channel penings during hyperpolarizing steps as well as the lnger clsures assciated with null episdes. This descriptin is clearly an versimplificatin because clsed histgrams suggest that there are mre than tw clsed states (Sakmann and Trube, 1984b). Further, this mdel ignres the subcnductance states f the channel (nly ne pen state is utilized). Hwever, it is a tractable mdel that can reasnably accunt fr a majrity f the bserved kinetic features. We first used the bserved mean pen times t fix the rate cnstant (k 23 ). l/tpen(v) = k 23 (V) = k 23 (O). exp(z&vlkti, (4) where kjj(o) is the value f the rate cnstant in the absence f an electrical field, V is membrane ptential, z8 is the fractinal electrical field strength, e is the value f an elementary charge, k is the Bltzmann cnstant, and T is temperature. The value f kt/e is - 25 mv at 20 C. The fit t the cntrl mean pen times is shwn in Fig. 6 (upperpanel); the fitted value fr k 23 (O) was 1.81 S-I and z8 was This left three additinal rate cnstants t be estimated. The middle panel in Fig. 6 shws the fit f the c-c- mdel t the steady-state cntrl P pen data (shwn in Fig. 3, slid circles) using the expnential vltage dependence f the rate cnstants '- '''j..... (msee) 20~ -n.. - '00 J Cntrl POpen. POp n Quin / Cnt ::j~::::: iii Ii, Membrane Ptential {my} FIGURE 6 Kinetic mdel fitted t cntrl and quinidine data. (Tp) Least squares fit f vltage dependence f the mean pen times (Eq. 4). The fitted values fr k 23 (O) and z8 are given in the text. (Middle) Fit f the C-C-O mdel t the steady-state cntrl Ppen data using the vltage-dependence fr the rate cnstants predicted by Eyring thery (see text). The fitted values fr the rate cnstants are prvided in the text. (Bttm) Fits f mdels A and B (see text) t the data frm Fig. 4 (tp) shwing rati f the prbability f pening in quinidine t cntrl. The fit t mdel A is shwn by the slid line, mdel B by the dashed line. Nte the apparent vltage dependence f the blck is accunted fr by mdel A even thugh the drug binding rate cnstants (k n and k ff ) were nt vltage dependent. Mdel B failed t accunt fr the apparent vltage dependence f blck. The sum-f-squared errr fr mdels A and B with the same number fdegrees f freedm were and , respectively. predicted frm reactin rate thery (Eyring et ai., 1980; Stevens, 1978). The fitted equatins fr the ther three rate cnstants were as fllws (k jj in secnds-i, V in millivlts): k 12 = expl+o.078evlkt) k 21 = expl-00486<vlkt) k 32 = exp(+o4q2"vlkt) The values f these rate cnstants at -110 mv differed frm thse previusly determined by Sakmann and Trube (1984b) using this mdel; hwever, their rate cnstants (see their Fig. 9) give a calculated steady-state prbability f channel pening that is higher than we bserve experimentally. This culd be due t differences in experimental cnditins between their study and urs, especially with regard t external ptassium cncentratins (Chen et ai., 1989). A majr effect f quinidine was t increase the number f null episdes during hyperplarizatin f the patch. The simplest mdificatin t the mdel that culd accunt fr these effects f quinidine n the prbability f channel pening was t add a drug-induced state t ne f the tw clsed states, thereby creating an additinal state with n direct cnnectin t the pen state. We excluded pen channel interactins because the drug had n significant effect n the channel pen times. If in fact quinidine acts by direct state-dependent binding t the channel, then the additinal druginduced state may be viewed as drug-assciated clsed ("blcked") state (Hndeghem and Katzung, 1977; Hille, 1977). We tested tw pssible mdels: k 21 k 32 C 1 ~~~03 k l2 k 23 kof! 1 ~ k n Catcked Mdel A k 21 k 32 C 1 ~ C2~ 0 3 k 12 k 23 kof! 1 ~ k n ~Icked Mdel B The rate cnstants k 12, k 21, k 23, and k 32 were determined frm the cntrl data (middle panel, Fig. 6); therefre, nly tw rate cnstants remained t be estimated fr each mdel (koff and k n ). We made the simplest assumptin that these rate cnstants were nt vltage dependent. Bth mdels accunted equally well fr the increase in the number f null episdes, hwever, nly mdel A gave the apprpriate vltage dependence fr P pen Fr mdel A, k Off = S-I and k n = 3.40 S-I (k a = 60,800 M- I S-I). Fr mdel B, k Off = S-I and k n = S-I (k a = 25,500 M- I S-I). The bttm panel ffig. 6 shws the ratis fthe prbability fpening in quinidine t cntrl with the least-squares fits t mdels (5) (6) (7) Balseretal. Inward Rectifier K Channels and Quinidine 159

11 A and B. Mdel B failed t accunt fr the apparent vltage dependence f blck and had a higher sum-fsquares errr than mdel A (see figure legend). Because neither mdel is a subset the ther (they are nnnested), simple expressins fr statistical discriminatin between the tw mdels d nt exist. It is pssible t discriminate the mdels based n the lg errr rati (LER; Akaike, 1974; Hrn, 1987), hwever, significance levels cannt be calculated with this apprach. The LER criterin indicated that mdel A was superir t mdel B. Statistical prf fr the superirity f mdel A lies in mre sphisticated methds fr discriminatin f nnnested mdels using Mnte Carl methds described by Hrn (1987); such an analysis warrants further study but is beynd the scpe f this paper. Therefre, the superirity f mdel A is nt prven; hwever, based n bth simple bservatin and n the LER criterin, mdel A seems mre cnsistent with ur data than mdel B. Fig. 7 cmpares single-channel data frm a patch befre and during expsure t quinidine (Fig. 7A) with recnstructed single-channel and ensemble-average behavir (Fig. 7 B) f the cntrl three-state mdel (7 B, left) and mdel A (7 B, right). If we assume a mdulated receptr mdel f direct state-selective drug binding t state 2, then a K D f - 10 J.LM is estimated fr mdel A. Hwever, this value is based n the bath cncentratin (50 J.LM) fquinidine and we d nt knw the cncentratin at the actual site f actin. Furthermre, we have nt assessed whether a clear cncentratin-dependence exists fr these quinidine effects. In summary, we have demnstrated that quinidine reduces current thrugh inward rectifier K channels by decreasing the likelihd that a given single channel will pen. The single-channel mechanism f blck des nt invlve a reductin f the pen-channel duratin, r a decrease in the single-channel cnductance. The results suggest that quinidine decreased the ability f the channel prtein t enter either the main- r the subcnducting cnfrmatins. This further suggests that quinidine may stabilize the channel in a clsed (r "blcked") state and thus reduce the chances f channel pening. These bservatins were accunted fr quantitatively by using the kinetic mdel prpsed by Sakmann and Trube (1984b) and mdified fr quinidine binding t the middle clsed state. We thank Drs. Luc Hndeghem and Dirk Snyders fr valuable cmments and suggestins. We als thank Dr. James Davis and Mrs. Hlly Gray fr technical assistance. Supprted by the United States Public Health Service (Natinal Institutes fhealth; HL32694, HL40608), the American Heart Assciatin (Tennessee Affiliate), and the Stahlman Cardivascular Research Endwment. J. R. Balser was supprted by the Medical Scientist Training Prgram (Natinal Institutes f Health; GM07347). A. -70 l I JI -~ ~ l:5 pa Cntrl Quinidine 500 ms rv-~ ~ ~ Ui r I If [1 pa ~ r--1. B. Cntrl Quinidine : Mdel A ~- -p~;;..-;..-~-..:;.;..;-: C';=- C.;= 0 ~.n C=C= II QC 500 ms FIGURE 7 Simulatin f ensemble and single channel gating behavir. (A) Left: Cntrl channel penings recrded during vltage clamp steps t -140 my. Right: similar steps recrded in quinidine. Ensemble averages are shwn abve each set f recrds. (B) Simulated single-channel penings during steps t -140 my (belw) and the crrespnding ensemble averatges f 200 such steps (abve). The simulatins use the rate cnstants derived frm fitting the C-C-O mdel t the data (left) and the quinidine mdel A (right). Because the simulatins required steps frm ptentials near E K where we had n data regarding Ppe" fr the C-C-O mdel P pen was set t be 0.62 at t = 0 (as determined in Sakmann and Trube, 1984b). The peak ensemble current in quinidine was 20% f the peak cntrl ensemble current; thus, the initial P pen fr the quinidine simulatin was (20% f 0.62) Received fr publicatin 26 January 1990 and in final fnn 28 August REFERENCES Akaike, H A new lk at the statistical mdel identificatin. IEEE (Inst. Electr. Electrn. Eng.) Trans. Autmatic Cntrl. ACI9: u r 160 Biphysical Jurnal Vlume 59 January 1991