blockade of cardiac sodium channels by amitriptyline and

677

Blockade of Cardiac Sodium Channels byAmitriptyline and DiphenylhydantoinEvidence for Two Use-Dependent Binding Sites

Michael J. Barber, C. Frank Starmer, and Augustus 0. Grant

Cardiac toxicity is a frequent manifestation in amitriptyline overdose and is felt to be due, in part,to sodium channel blockade by the drug. Another agent with sodium channel blocking properties,diphenylhydantoin, has been used clinically to reverse cardiac conduction abnormalities induced byamitriptyline. This reversal of toxicity is believed to occur secondary to competition for the sodiumchannel binding site. We evaluated individually and in combination the effects of amitriptyline (0.4,uM) and diphenylhydantoin (10-80 ,uM) on the sodium current in isolated rabbit atrial andventricular myocytes at 17°C. Using the whole-cell variant of the patch-clamp technique, we foundthat both amitriptyline and diphenylhydantoin reduced the sodium current in a use-dependentfashion. The time constant of recovery (Tr) from block by amitriptyline at -130 mV was very slow(13.6±3.2 seconds), whereas rr during diphenylhydantoin exposure was fast (0.71±0.21 seconds,p<0.0001 compared with amitriptyline). During exposure of cells to a mixture of the two drugs, rwas found to be 6.6±1.8 seconds, but no evidence of direct competition between amitriptyline anddiphenylhydantoin was seen. Attempts to fit the recovery data of the mixture to two exponentialsresulted in no significant improvement in the fit when compared with that using a singleexponential. Use of the sodium channel blocking agent lidocaine (similar kinetics to diphenylhy-dantoin) in competition with amitriptyline resulted in findings consistent with direct competitionof these two drugs for a single binding site. These observations prompted us to evaluate thepossibility that diphenylhydantoin was not acting at (and therefore not competing for) the samechannel binding site as amitriptyline. Experiments altering pHi and pH. revealed dramaticdifferences between amitriptyline and diphenylhydantoin. When pH. was increased from 7.4 to 8.0,Tr was reduced approximately threefold (from 13.6±3.2 to 4.2±0.1 seconds, p<0.0001) duringexposure to amitriptyline, but no effect was seen on rr after exposure to diphenylhydantoin.Conversely, when pHI was increased from 7.3 to 8.0, Tr after amitriptyline was unaffected, but rafter diphenylhydantoin markedly increased (from 0.71±0.21 to 2.60±1.30 seconds, p<0.001).Additionally, diphenylhydantoin block demonstrated profound voltage dependence across therange of -130 to -90 mV, whereas amitriptyline block appeared less voltage sensitive. Single-channel studies using patch-clamp techniques in isolated ventricular myocytes supported thesedata. Superfusion of cells with diphenylhydantoin did not change mean channel open time butincreased the probability of failure of the channel to open. When diphenylhydantoin was placed inthe micropipette, no effect on sodium channel kinetics could be demonstrated. Our experimentsshow that amitriptyline blocks sodium channels via a drug-receptor complex site sensitive tochanges in pHO, whereas diphenylhydantoin blocks from a pH,-insensitive site that is susceptibleto changes in pHi. The decrease in amitriptyline-induced rr by diphenylhydantoin may result fromallosteric modulation and suggests the presence of a separate (intracellular) binding site fordiphenylhydantoin action. (Circulation Research 1991;69:677-696)

From the Departments of Medicine and Computer Science, Institute, a Grant-in-Aid and an Established Investigatorship fromDuke University Medical Center, Durham, N.C., and the Depart- the American Heart Association (A.O.G.), a Grant-in-Aid (VA-ment of Medicine, Division of Cardiology, University of Virginia 90-G25) from the American Heart Association (Virginia Affili-Health Sciences Center, Charlottesville, Va. ate), and contract 44 14 808 from the Office of Naval Research.

Supported in part by grants HL-32994 and HL-32708 from the Address for correspondence: Dr. Augustus 0. Grant, Box 3504,National Institutes of Health, National Research Service Award Duke University Medical Center, Durham, NC 27710.5T32 HL-07101 from the National Heart, Lung, and Blood Received December 20, 1990; accepted May 6, 1991.

by guest on February 13, 2018http://circres.ahajournals.org/

Dow

nloaded from

http://circres.ahajournals.org/

678 Circulation Research Vol 69, No 3 September 1991

Amitriptyline, a frequently prescribed tricyclicantidepressant, is often associated with drugoverdose.'2 Cardiac toxicity in the form of

impaired conduction,34 decreased contractility, andlife-threatening arrhythmias5-7 is seen. The mecha-nism(s) by which amitriptyline produces its toxic effectsis unknown, but some of these actions may be due to itssodium channel blocking properties.1"8 9 Hagerman andHanashirol and Boehnert and Lovejoy9 demonstratedimprovement in cardiac conduction as measured bynarrowing or normalization of PR interval and QRSduration abnormalities when patients who had takenan overdose of amitriptyline were treated with diphe-nylhydantoin. Some authors'011 even recommend ad-ministration of diphenylhydantoin prophylactically intricyclic antidepressant overdose. Callaham'2 and Da-vis,13 on the other hand, could demonstrate no benefi-cial effect of diphenylhydantoin after amitriptylineoverdose. Experimental data in whole animals havebeen equally confusing when amitriptyline toxicity istreated with diphenylhydantoin or other type IBagents.'4-16

Recent theoretical and experimental studies17-22suggest that under certain circumstances the combi-nation of two sodium channel blocking agents withmarkedly different binding kinetics may producesmaller levels of sodium channel block than thatproduced by the agent with the slow kinetics alone. Adrug with rapid association and dissociation kineticscould theoretically compete for sodium channel bind-ing sites and displace a drug with slower bindingkinetics. This could lead to a reversal of cardiotoxicmanifestations that result from sodium channelblockade. We recently have demonstrated such aphenomenon between the opiate analgesic propoxy-phene and lidocaine.22

Prior voltage-clamp experiments have shown thatdiphenylhydantoin has rapid binding and unbindingkinetics to the cardiac sodium channel23-25 and assuch would in principle displace amitriptyline fromits binding site. We set out to determine whethersuch a competitive blockade of the cardiac sodiumchannel could be demonstrated in vitro. First, weanalyzed the kinetics of development of and recoveryfrom block of the sodium current recorded fromisolated rabbit atrial myocytes under voltage-clampconditions during exposure to amitriptyline. Then,we confirmed that diphenylhydantoin had rapidblocking and unblocking kinetics under our studyconditions. Although we could demonstrate modula-tion of amitriptyline by diphenylhydantoin, we couldnot demonstrate a true reversal of amitriptyline-induced blockade of the cardiac sodium channel bydiphenylhydantoin. This suggested that the two drugsmay be acting at different sites.We evaluated the effects of internal and external pH

changes on the kinetics of recovery from block byamitriptyline (pK 9.4) and diphenylhydantoin (pK 8.3).The kinetics of recovery from block by amitriptylinewere slowed by a reduction of external pH only. Incontrast, the kinetics of recovery from diphenylhydan-

toin-induced block were accelerated by a decrease ininternal pH. These data suggest that diphenylhydantoinblocks the cardiac sodium channel from a site (intra-cellular?) insensitive to external protons. We were ableto show that lidocaine, a drug with fast binding andunbinding kinetics to a site sensitive to external pro-tons, may produce partial reversal of amitriptyline-induced sodium channel blockade.

Materials and MethodsCell PreparationWe performed the whole-cell voltage-clamp exper-

iments on myocytes isolated from the atria of adultrabbits. The methods of cell isolation and culturehave been described in detail in previous publicationsfrom our laboratory.26 Briefly, the heart was isolatedand perfused for 5 minutes with Ca2-free Krebs-Henseleit (K-H) solution (see below for solutioncomposition). Sterile technique was used, and allsolutions were maintained at 37°C. After 5 minutes,the K-H Ca2-free perfusate was changed to K-Hsolution containing 180 units/ml collagenase (Wor-thington Biochemical Corp., Freehold, N.J.) and 0.1mg/ml hyaluronidase (Sigma Chemical Co., St. Louis,Mo.), and the hearts were perfused for 25-45 min-utes. Enzyme activity was terminated with K-H solu-tion and 10% fetal bovine serum. Ventricular tissuewas separated from the atria. The atria were mincedinto small segments, transferred to K-H mediumcontaining elastase (0.5 mg/ml, Sigma), and gentlyagitated. On completion of the dissociation proce-dure, single myocytes were separated from tissuechunks by filtration through a 200-am nylon mesh.The isolated myocytes were washed twice with K-Hsolution and resuspended in a medium consisting ofF-12 Ham/Dulbecco's modified Eagle's Medium/an-tibiotics. Cells were then plated onto 18x18-mmlaminin-coated coverslips and stored in a CO2 incu-bator (Queue Systems, Inc., Parkersburg, W.Va.) at37°C. Under these conditions, rodlike atrial cellsassume a spherical shape after 24-48 hours in cul-ture. Atrial cells were used after they had assumed aspherical configuration, because theoretical analysishas indicated that the spherical shape is most favor-able for obtaining a homogeneous potential undervoltage-clamp conditions.26'27 Ventricular cells wereused 2-4 hours after isolation.

SolutionsThe K-H solution used in the cell isolation proce-

dure had the following composition (mM): NaCl118.2, CaCl2 2.7, KCl 4.7, MgSO4 . 7H2O 1.2,NaHCO3 25, NaH2P04 2, and glucose 10. Calcium-free K-H solution had no added calcium. K-H solu-tion was gassed with a 95% 02-5% CO2 mixture.For whole-cell recording, the standard intracellu-

lar solution had the following composition (mM):CsF 120, MgCl2 5, K2ATP 5, KH2P04 1, EGTA 5,glucose 5, and HEPES 10. The pH of the internalsolution was adjusted to 7.3 or 8.0 on the day of the


Dow

nloaded from


Barber et al Evidence for Two Use-Dependent Sodium Channel Binding Sites 679

experiment using CsOH (1 M). The internal solutionwas filtered with a 0.3-,m filter at time of use. Theexternal solution contained the following composi-tion (mM): NaCl 75, CsCI 75, MgCl2 1, KCI 5, CaCI21.5, glucose 5, and HEPES 10. The pH of the externalsolution was adjusted to the desired value (7.4 or 8.0)on the day of the experiment with NaOH (1 M).For cell-attached patch recordings, the standard

micropipette solution had the following composition(mM): NaCl 182, MgCl2 5.0, CaCl2 0.2, CsCl 1.0, andHEPES 5. The pH was adjusted to 7.4 with NaOH.The isolated myocytes were superfused in theseexperiments with a high external potassium solutionof the following composition (mM): KCl 70, potas-sium aspartate 80, NaCl 5, MgCl2 3, K2EGTA 0.05,HEPES 5, NaH2PO4 1.2, and glucose 10. The pH wasadjusted to 7.4 with KOH. The high external potas-sium solution was used to depolarize the cells to -0mV and to enable us to express membrane potentialsas absolute voltage.

In whole-cell experiments, the cells were perfusedwith the external solution alone (control) or withexternal solution containing 0.4 ,uM amitriptyline(Sigma), 10-160 ,uM diphenylhydantoin (Sigma), or amixture of 0.4 ,uM amitriptyline and 10-80 ,uMdiphenylhydantoin dissolved in the superfusate. Ad-ditionally, in a series of experiments, lidocaine (80,uM) was added and superfused in combination withamitriptyline (0.4 ,um). On the day of each experi-ment, fresh stock solutions of amitriptyline (0.4 mM),diphenylhydantoin (1 mM), and/or lidocaine (80mM) were made up by dissolving the drug in theexternal solution. Diphenylhydantoin goes into solu-tion poorly and dissolves only at high pH (>11.0).This required the stock solution to be very alkaline.Stock solutions then were added to buffered externalsolution in appropriate amounts to give the desiredfinal concentration of drug(s). Solutions containingdiphenylhydantoin were rechecked to ensure that nosignificant alteration in pH occurred.

Whole-Cell Recording TechniquesMicropipettes were pulled from 1.5-mm o.d. borosil-

icate glass (N-51A, Drummond Scientific Co.,Broomall, Pa.) using a horizontal puller (model P80/PC Flaming Brown, Sutter Instrument Co., Novato,Calif.). Micropipettes were coated with Sylgard 184(Dow Corning, Midland, Mich.) to lower microelec-trode capacitance and then fire-polished on a micro-forge (model MF-83, Narishige, Tokyo). For whole-cellcurrent recordings, 400-1,000-kQ microelectrodeswere used. Each microelectrode was coupled to theinput of the patch-clamp amplifier (model EPC 7, ListElectronics, Darmstadt, FRG, or model 3900, DaganCorp., Minneapolis, Minn.) by an Ag/AgCl wire coatedwith Teflon up to its tip (In Vivo Metric Systems,Healdsburg, Calif.). A similar wire was used to make abath reference, which was inserted in an agar bridge(3% agar dissolved in internal solution). With thisbridge, voltage offset was typically 1-5 mV. These were

Whole-cell currents were filtered at a corner fre-quency of 5 kHz with an eight-pole Bessel filter(model 902, LPF Frequency Devices, Haverhill,Mass.) and digitized using a Compaq 386/20 micro-computer at 40 kHz. Data were stored on magnetictape and analyzed off-line (see data analysis).

Whole-Cell Experimental ProtocolsOn the day of the study, an 18x18-mm coverslip

containing the atrial cells was fixed to the base of therecording chamber, which sat on a thermostaticallycontrolled Peltier device (TS-2 thermal microscopestage, Sensortek, Clifton, N.J.) on the stage of aninverted microscope (Nikon Diaphot, Nikon Instru-ments, Garden City, N.Y.). The temperature was set at170C, and the bath was perfused at 1 ml/min withcooled external solution. The inflow to the bath wascontrolled by a series of solenoid valves, which permit-ted the rapid switch of perfusion solutions. To record inthe whole-cell configuration, a gigohm seal was ob-tained as outlined by Hamill et al.28 The capacitance ofthe microelectrode and amplifier input were nulled.The membrane patch then was ruptured by a briefpulse of suction, and the cell capacitance and seriesresistance were nulled. Finally, series resistance com-pensation was applied to a level just below that whichproduced ringing (usually 70-90%). The holding po-tential was fixed at -130 mV and a current-voltagecurve was obtained as shown in Figure 1. Voltage error,peak current, and series resistance were determinedusing readings from the patch-clamp amplifier andfrom the equation

/V=Ip1 -a)Rwhere AV is voltage error, Ip is peak current, a ispercent compensation, and R is series resistance. IfAV was >3 mV, the experiment was abandoned.

Before obtaining other information, the current-voltage relation was determined using 50-msecpulses of increasing amplitude applied at 1,500-msecintervals. The pulse was incremented in 5-mV stepsfrom a potential of -120 to +70 mV. We proceededwith the experimental protocol if the currents of thenegative limb of the current-voltage curve showed a

gradual increase with progressively larger depolariza-tions (onset to peak sodium currents spanning 30-40mV, see Figure 1).A steady-state inactivation curve was determined

using a 1-second prepulse to potentials from -170 to-50 mV, followed immediately by a 10-msec testpulse to -30 mV. The prepulse potential was incre-mented by 5 mV in each subsequent test. A repre-sentative steady-state inactivation curve is shown inFigure 1C with the data plotted in Figure 1D.

In the first series of experiments, trains of 40 pulses(pulse duration, 50 msec; holding potential, -130mV; step, to -30 mV) with a frequency of 5 Hz wereapplied from the holding potential to determine theamount of steady-state block in the absence or pres-ence of drug(s) (see below). In these studies, current

nulled before obtaining recordings. from the first pulse and last pulse of the train was


Dow

nloaded from



C _]

-130 m

lna

2 ms

D

C)._

en

0

b.

0

3

2

1'

-11

-30 mv-50 mV

-160 mV

-120 -80 -40

Test Potential (mV) Conditioning Potential (mnV)FIGURE 1. Typical characteristics of sodium current in rabbit atrial myocytes. Panel A: Currents recorded under voltage-clampconditions during test potentials from -75 mVto -35 mV Currents were obtained from a holding potential of -130 mV PanelB: The current-voltage curve for allpoints. The continuous line represents the fit ofthe data to a Hodgkin-Huxley m3h model, wherem is the activation variable and h is the inactivation variable. Panel C: Recordings depicting characteristics of sodium currentsduring the inactivation protocol. From a holding potential of -130 mV, 1 second prepulses to various potentials were followed bya 20-msec testpulse to -30 mV There is no crossover of the currents, because their size changes with holdingprepotential and thepeak ofeach current occurs at approximately the same time. Panel D: Steady-state inactivation curvefor allpoints. The continuousline depicts the predicted ideal fit to a relation of the form h =1/[exp(V- Vh)/s] (where h is the inactivation variable, Vis membranepotential, VK is the potential for half-maximal inactivation, and s is the slope factor) for the data. For this experiment, Vh=94.5mVand s=6.27.

measured, and steady-state block was expressed as a

percent of the first pulse current. To determinerecovery from steady-state block, pulse trains were

followed by a variable interval of recovery (2-50,000msec) to the holding potential, and a subsequent20-msec test pulse to -30 mV was applied to assessavailable sodium current. Measurements were madein the absence of drugs (control) as well as inmixtures containing amitriptyline alone (0.4 ,um),diphenylhydantoin alone (10-80 ,lM), amitriptyline(0.4 ,M) plus diphenylhydantoin (20-80 ,M), andamitriptyline (0.4 ,uM) plus lidocaine (80 ,M). Dur-ing control conditions and in the presence of diphe-nylhydantoin or lidocaine alone there was a 15-second rest interval between each train. During anyprotocol in which amitriptyline was used as a compo-nent of the superfusate, a 50-second rest interval wasused between each test pulse and the succeedingtrain or prepulse.

The development of frequency-dependent block wasdetermined by the application of trains of 50-msecpulses to -30 mV from the holding potential of -130mV with stimulus frequencies ranging from 0.1 to 5 Hz.There were 40 pulses in trains with stimulus frequen-cies of 5, 4, and 2 Hz, 30 pulses in trains with stimulusfrequencies of 1 and 0.5 Hz, and 20 pulses in trains withstimulus frequencies of less than 0.5 Hz. This was donebecause steady-state block was achieved with a smallernumber of pulses at slower stimulus frequencies.We analyzed the dependence of development of

steady-state block on membrane potential in theabsence of drug and in the presence of amitriptylineor diphenylhydantoin. By using trains of forty 50-msec pulses stepped to -30 mV at a frequency of 5Hz, the holding potential was varied from -130 to-90 mV in 10-mV increments during each successivepulse train series. In control or in the presence ofdiphenylhydantoin, a rest interval of 15 seconds was

A

B

0L.

C)0A-

-

Cl)

(n


Dow

nloaded from



used between trains, whereas in the presence ofamitriptyline, the rest interval was 50 seconds.The rate of development of block also was deter-

mined with a two-pulse protocol.26 We evaluatedchannel availability using conditioning prepulses ofincreasing duration (holding potential, -130 mV;prepulse potential, -30 mV; prepulse duration,2-10,000 msec), followed with a return to the holdingpotential for 500 msec and a subsequent test pulse of20 msec to -30 mV.The rate of recovery from block in the absence or

presence of drug also was determined with a two-pulse protocol. Using conditioning prepulses of 2seconds (maximal drug uptake without evidence ofslow inactivation of the sodium channel as shown bythe uptake protocol) to -30 mV, a variable period ofrecovery to the holding potential was used. Therecovery interval ranged from 100 to 40,000 msec andwas followed by a 20-msec test pulse to -30 mV toevaluate sodium current availability.

Single-Channel Recording TechniquesThe results of the whole-cell voltage-clamp experi-

ments suggested that diphenylhydantoin may produceblock from an internal membrane site. We did a seriesof single-channel experiments to determine the natureof diphenylhydantoin block and whether it was produc-ing block from a drug pool immediately accessible tothe internal or external membrane surface. We per-formed these experiments in cell-attached patches ofventricular myocytes, because this configuration af-forded the most stable single-channel kinetics.29-32Experiments were performed in the absence of diphe-nylhydantoin, with drug-free microelectrodes attachedto cells exposed to drug in the superfusate, or withdrug-containing microelectrodes attached to cells thathad not been exposed to drug.

Single-Channel Experimental ProtocolsOur protocol consisted of the application of a

1-second conditioning pulse to -20 mV, a recoveryinterval of 500 msec, and a 40-msec test pulse to -20mV. The 500-msec recovery period permitted therecovery of drug-free channels from inactivation (see"Results"). A rest period of 5 seconds separatedeach test pulse and the subsequent conditioningpulse. This allowed recovery of the channels fromblock by the drug. Whole-cell voltage-clamp experi-ments with diphenylhydantoin suggested rapid recov-ery from block at a holding potential of -120 mV.We performed preliminary experiments with theholding potential set at -120 and -90 mV. Block ofsingle channels could be demonstrated with theholding potential equal to -90 mV but was not welldetected at -120 mV. The remaining single-channelexperiments were performed at -90 mV.We obtained an ensemble of 50 conditioning and

test pulses and accepted the data only if two suchensembles were obtained under a given condition.Further, in all experiments we waited a minimum of

diphenylhydantoin-containing microelectrodes) aftergigohm seal formation before data were acquired.This circumvented the period of rapid spontaneouschanges in single-channel kinetics during which la-tency to first opening decreased and mean open timeincreased. An experiment required a stable seal forat least 15 minutes.

Data AnalysisAs noted above, filtered currents from whole cells

were digitized and subsequently transferred to aSUN 4/280 microcomputer (SUN Microsystems, Inc.,Mountainview, Calif.), where the peak values ofindividual current tracings were determined by usingcustom software developed in our laboratory andwritten in c programming language. These peakswere plotted, and exponentials were fitted to the datadescribing the recovery from block and the develop-ment of block and inactivation using the Marquardtroutine or Gauss-Newton method.33 From the resid-ual error an F statistic was calculated and considereda good fit at a significance level <0.05.

Single-channel currents were filtered at a cornerfrequency of 2.5 kHz and digitized at 25 kHz. Toreduce the size of the files, we digitized only the first40 msec of the conditioning pulse. At -20 mV, theaveraged currents relaxed to zero in .20 msec.34 Onthe rare occasions in which bursts were observed ineither the conditioning or the test pulse, we excludedthat pulse and its associated conditioning or testpulse from the analysis.The single-channel currents also were analyzed on

a SUN 4/280 microcomputer using methods outlinedby Grant et al.34 Briefly, residual capacitive andleakage currents were reduced by subtracting theaverage of null sweeps from each sweep. We used anautomatic event detection algorithm with the thresh-old at 0.5. The performance of the algorithm waschecked routinely by direct comparison of the origi-nal and idealized records.We compared single-channel closing kinetics by

using mean open times. Particularly in the test pulsesin the experiments using diphenylhydantoin, thenumber of events was sometimes too small to con-struct a histogram. The apparent probability of fail-ure of a channel to open was obtained by taking theNth root of the number of null sweeps, where N is theapparent number of functioning channels in thepatch. N was estimated from the maximum numberof overlapping events at a strongly depolarized po-tential (-30 mV). Because the holding potential wasusually set at -90 mV, the estimate of N wasconservative. The conservative estimate does notchange the major conclusions of the paper.

In reporting results, all data are expressed asmean+ SD. The statistical significance of the differ-ences was determined using the appropriate Stu-dent's t test for paired or unpaired data. Whenmultiple group comparisons were made, a one-way ortwo-way analysis of variance was performed. Whenan analysis of variance demonstrated a significance5 minutes (7 minutes with the experiments with


Dow

nloaded from



between some data group(s), a t test or Scheffe's testwas used to determine significance between individ-ual means. Significance was considered to be presentat p<0.05. Statistical techniques were taken fromGilbert35 and Snedecor and Cochran.36

ResultsCharacteristics of the Cultured Rabbit Atrial MyocytesFor reasons reported earlier,26'27 experiments in

this study were performed on spherical cells only. Forall of the experiments (total n = 141 cells), the cellcapacitance averaged 26±1 pF, which was in excel-lent agreement with our previous report.26 We wereable to compensate .70% of the series resistance inall experiments. At all test potentials used in thisstudy, the capacitive current was clearly separatedfrom the sodium current. The threshold potential ofour cell population was - 66 +5 mV with peak sodiumcurrent of the current-voltage relation occurring at-35±4 mV. All steady-state inactivation curves werefitted with a Boltzmann function and demonstrated ahalf maximal inactivation value of - 90+5 mV. Be-cause of the absence of sodium ions in the pipettesolution, no current reversal was seen. There was nocrossover of the current tracing during steady-stateinactivation determination or threshold phenomenaon the negative limb of the current-voltage curve(see Figure 1). These characteristics suggest thatadequate voltage-clamp conditions existed duringour experiments despite the comparatively high con-centration of sodium in the external solution.26

Description of Kinetics of Sodium Current Blockby AmitriptylineAs illustrated in Figure 2, amitriptyline blocked

the sodium current in a use-dependent manner. Inthe absence of drug (panel A), sodium current de-clined by 11% during trains of 40 pulses with astimulation frequency of 5 Hz. During exposure to0.4 ,M amitriptyline (panel B), sodium current de-clined progressively until reaching a steady-stateblock of 74%. The steady-state block was achievedafter 21 pulses in the 40-pulse train (panel C). Usingmultiple frequencies of stimulation from 0.1 to 5 Hz,we observed a progressive increase in the amount ofsteady-state block achieved. Summary of the level ofsteady-state block as a function of stimulus frequencyseen in 15 cells is shown in panel D. Amitriptylineblock during pulse train stimulation was not depen-dent on the holding potential (panel E).The progressive decline in sodium current during

pulse train stimulation resulted from enhanced bind-ing of amitriptyline to a sodium channel receptorduring the depolarizing pulses. This enhanced bind-ing may be due to differential affinity of amitriptylinewith various states of the channel (i.e., open, inacti-vated, and resting) and/or differential access to"guarded" binding sites of fixed affinity.37,38 Betweendepolarizing pulses, some block dissipates because ofdrug dissociation from the channel binding site. The

rate of drug dissociation (recovery from block) undergiven experimental conditions is a constant. There-fore, if the stimulus frequency is greater than fivetimes the recovery time constant, block accumulatesuntil steady-state is reached; that is, the amount ofblock gained per pulse is equal to the relief of block.As shown in Figure 2D, accumulation of block, evenat slow stimulus frequencies, is significant with ami-triptyline and suggests slow dissociation of drug fromthe sodium channel at 17°C.We examined the kinetics of recovery from steady-

state block using trains of pulses (40 pulses; duration,50 msec; test, to -30 mV) in the absence (Figure 3A)and presence (Figure 3B) of 0.4 ,uM amitriptyline.After each pulse train, a variable recovery period(100-50,000 msec) was used before application of arecovery test pulse. Currents from the test pulse werecompared with the peak (first pulse) sodium currentat 10-15 recovery intervals after the pulse train, andthese data were fitted to a single exponential equa-tion to determine the time constant of recovery (7Tr)(Figure 3C). In the absence of drug (control), only asmall amount (<10%) of steady-state block devel-oped, and the current recovered rapidly with a mean,r of 30+8 msec in our cell population (Table 1). Thisvalue agreed with previously published -r in myocytesunder similar voltage-clamp conditions.22,23,39 Duringexposure to amitriptyline, significant steady-stateblock developed (63 + 10%), and rr was increased to13,600±3,200 msec, demonstrating slowed kinetics ofrecovery from steady-state block.

Hille40 has suggested that for tertiary amines, theneutral species may access its binding site when thechannel is open or closed, whereas the chargedspecies may access the binding site only when thechannel is open. The charged and uncharged speciesare thought to have different dissociation rates fromthe channel. To evaluate the role played by thecharged versus neutral species of amitriptyline, therate of recovery from block in the absence (Figure3D) and presence (Figure 3E) of amitriptyline wasevaluated in the same cell using a two-pulse protocol.Because charged species access open sodium chan-nels, pulse train data reveal kinetics predominantlyof the charged species, whereas holding the cell at-30 mV (prepulse) inactivates sodium channels andallows access to channel binding sites predominantlyby the neutral form of the drug.By using a fixed conditioning prepulse (duration,

2,000 msec) followed, after a variable period (100-50,000 msec), by a test pulse of 20 msec to -30 mV,sodium current recovery was measured. In the ab-sence of drug, no significant decrease in peak sodiumcurrent was seen between consecutive conditioningpulses (no slow inactivation).41 Perfusion with exter-nal solution containing amitriptyline showed devel-opment of significant block. The data are showngraphically in Figure 3F and demonstrate a -r similarto that seen with pulse train stimulation. Table 1compares results from the pulse train and two-pulseprotocols. In six experiments, the r. value using the


Dow

nloaded from



Amitriptylinea

Pulse Number

E 806

as0ak

40O

0i9

A-

-3 0

-130 mV-

-130

Holding Pote

FIGURE 2. Blocking properties ofamitriptyline. Panel A: Sodium cur-rent recorded from an atrial myocyteduring the first, third, fifth, seventh,ninth, and 40th pulses in a 40-pulsetrain (5 Hz; holding potential, -130mV; pulse duration, 50 msec; step, to-30 mV). Panel B: Sodium currents

N 440 in the same cell after 15-minute ex-posure to 0.4 pM amitriptyline. The

I nA first, third, fifth, 10th, 30th, and 40thpulses are shown. The current andtime calibration are the samefor both

2ms tracings. Panel C: Plot showing thedecline of the inward sodium currentas a function ofpulse number in thetrain in the absence (control, 0) andpresence (n) of 0.4 pM amitriptyline.Panel D: Plot characterizing the use-dependent block with amitriptyline.Data represent the mean+±SD of thepercent ofthe steady-state block com-pared with the first pulse in 15 cellsduring superfusion with amitriptylineplotted as a function of stimulus fre-quency. In the absence of drug, thepercent steady-state block during thesame protocols ranged from 0% (0.1Hz) to 9% (5 Hz). For stimulusfrequencies of 5, 4, and 2 Hz, 40pulses were used; for 1 and 0.5 Hz, 30pulses were applied; and for rates of0.3 and 0.1 Hz, 20 pulses wereneeded to reach steady state. PanelE: Steady-state block (mean±SDcalculated as a fiaction of the firstpulse) compared with the first pulse

-110 0 in six cells during application ofpulseential (mV) trains (40 pulses; pulse duration, 50

msec; step, to -30 mV) plotted as afunction of the holding potentiaL

two-pulse protocol was 45 ± 10 msec in the absence ofdrug (p=NS for r, in absence of drug with pulse trainstimulation). When the cells were perfused withamitriptyline, significant block developed (83±8%).This value was statistically different from that seenduring pulse train stimulation protocol (83±8%versus 63±10%, p<0.001) in spite of equal totaltime of depolarization in both protocols. Thesmaller amount of block with pulse train stimula-tion may result from partial recovery from blockbetween pulses. Using the two-pulse protocol, the r-for amitriptyline was 14,800+4,100 msec. This valuewas not significantly different from the value of Tr, inthe absence of drug.The kinetics of decline of the sodium current

during pulse train stimulation is the result of a

combination of uptake of the drug at the binding siteduring the pulse and dissociation of drug from the

binding site between pulses. Recovery data evaluatedissociation from the binding site. We also examineddrug uptake in our model.By using a two-pulse protocol to evaluate uptake

kinetics, prepulses of varying duration (2-10,000msec) were followed 500 msec later by a 20-msec testpulse to -30 mV to evaluate sodium current. Arepresentative experiment is shown in Figure 4. Inthe absence of drug (panel A), no change was seen insodium current measured during the test pulses withprepulse durations of <2,000 msec. With prepulsesof >3,000 msec, the current seen during the testpulse began to decline slightly, secondary to slowinactivation of the sodium channel.41 In the presenceof amitriptyline (panel B), progressively longer pre-pulses caused a decline in the sodium current. Theresults of a single study are shown in panel C. Thedecline in sodium current in the presence of amitrip-

AControl

c

Cc

c

c

0

0

Eaa

D ID

0U)

U0

011

u Z 4Frequency of Stimulation (Hz)


Dow

nloaded from



Control

c ccc

c

0

Etn

Am itriptyl inea

-13 0 -30

N = 40

ImA

2 ms

-1

Amitriptyline (Tr = 13.3 s)

Control (rr = .04 s)0 20 40

Recovery Interval (s)

lnA

D _m E-n

c *a. -2 Amitriptyline (tr s)

= -2

E

o Control (cr = .03 s)

0 20 40

Recovery Interval (s)

tyline could be fitted with a single exponential withan uptake time constant of 1,410 msec in this exper-iment. The mean value for all experiments usingamitriptyline (n=6) was 1,350+60 msec.The Tr measured for amitriptyline was considered to

be quite long, on the same order of magnitude as the irreported by Whitcomb et a122 for propoxyphene. Asthese authors demonstrated, the long r, during expo-sure to a drug was consistent with the large amount ofcumulative block observed even at slow stimulus rates.Whereas the r, of amitriptyline is shorter than that ofpropoxyphene (13.6 versus 24.6 seconds), the rr of thetype IB agents (i.e., lidocaine and diphenylhydantoin) isshorter, < 1 second. If the slow binding agent and theagent with more rapid kinetics shared a common bind-ing site, they would "compete" for available bindingsites during each depolarizing pulse. During thesepulses, the drug with the more rapid kinetics mightcompete more effectively for the binding site and

FIGURE 3. Recovery from amitrtptyline-in-duced block Panel A: Sodium current re-

cordedfrom an atrial myocyte in the absenceofdrug afterpulse train stimulation (5 Hz; 40pulses; holding potential, -130 mV; step, to-30 mV) and at recovery intervals of 20msec, 40 msec, 50 msec, 100 msec, and 1second (superimposed on 100-msec tracing).Panel B: Sodium currents recorded in thesame cell after 15-minute exposure to 0.4 1iMamitriptyline. Tracings show the finalpulse ofa blocking train and recovery intervals of 5,10, 15, 20, 25, 40, and 50 seconds (superim-posed on 40-second tracing) afterpulse trainstimulation. The current and time calibrationsare the same for panels A and B. Panel C:Recovery curves for control (0) and amitrip-tyline (m) trials in this cell plotted with asso-

ciated values for the time constant ofrecovery(,r). The continuous lines are the single expo-nential fit for the data. Panel D: Sodiumcurrents from another cell in which inhibitionwas induced by a single pulse (pulse duration,2 seconds; holding potential, -130 mV; step,to -30 mV) in the absence of drug. Tracingswere obtained 2 msec (smallest sodium cur-

rent), 10 msec, 20 msec, 40 msec, 100 msec,and 1 second (superimposed on 100-msectracing) after termination of the pulse. PanelE: Sodium currents in the same cell after20-minute cxposure to 0.4 pM amitriptyline.Tracings are at 2 msec (smallest sodiumcurrent) and at 5, 10, 15, 20, 25, 40, and 50seconds (superimposed on 40-second tracing)after the prepulse. The current and time cali-brations are the same for panels D and E.Panel F: Recovery curves for control (a) andamitriptyline (m) plotted with associated rrvalues. The continuous lines are the singleeponential fit for the data.

TABLE 1. Steady-State Block and Time Constants of Recovery forPulse Train and Two-Pulse Protocols With Amitriptyline and WithDiphenylhydantoin

Pulse trains Two-pulse protocol

Control Drug Control Drug

AmitripylineSSB (%) 6±3 63+10* 9±2 83+8*1r (msec) 30+8 13,600+3,200* 45±10 14,800+4,100*n 15 27 6 6

DiphenylhydantoinSSB (%) 9+3 14±2 6+3 31+6t,r, (msec) 35+8 710±210* 50+ 10 1,100±500*n 10 22 8 10

Values are mean±+SD. Control, absence of drug; Drug, in thepresence of amitriptyline or diphenylhydantoin; SSB, steady-stateblock expessed as a percent of first pulse current; r,, time constantof recovery; n, number of cells.

*p<0.0001 and tp<0.001 vs. corresponding control value.


Dow

nloaded from



Am itriptylineB _m

C -2-

-

._

3E

0

cl)

0

-30 - 30

500 msec

lnA

2 ms

/ Amitriptyline (tu = 1.41 s)

Control

2 4 6 8 10

Prepulse Duration (s)FIGURE 4. Sodium current as a function ofprepulse duration. Panel A: Sodium currents recorded from an atrial myocyte in theabsence ofdrug asprepulse duration was increased. The holdingpotential was -130 mVwith step to -30mV The recovery intervalbetween the prepulse and test pulse was 500 msec. Peak sodium current remained unchanged until prepulse duration was >2seconds. Tracings show sodium current for prepulses of <2 (three largest currents), 3, 5, 7, and 10 seconds. Panel B: Sodiumcurrents recorded from the same cell after 17-minute exposure to 0.4 gtM amitriptyline using the same protocol as above. Currenttracings from test pulses after prepulses of 10 and 50 (superimposed largest currents), 200, 400, and 700 msec and 1, 2, 4, and 6seconds (superimposed smallest currents) are shown. The current and time calibrations are the same for both pulses. Panel C:Uptake curves in this cell for control (c) and amitriptyline (n) plotted with the uptake time constant (Qu) for amitriptyline. Thecontinuous line in amitriptyline is the single exponential fit of the data, whereas the line in control connects the points.

therefore displace the slower drug. If this occurred, atappropriate rates of stimulation the amount of sodiumchannel blockade in the presence of two channel-blocking agents could actually be less than the amountseen in the presence of the single agent with slowkinetics. Theoretically, a drug such as diphenylhydan-toin or lidocaine might lessen the amount of sodiumcurrent block seen with amitriptyline when also presentin the perfusate.21

Description of Kinetics of Sodium Current Blockby Diphenylhydantoin

Diphenylhydantoin (10-160 ,uM) also acted as asodium channel blocking agent, demonstrating use-and frequency-dependent block but to a lesser extentthan amitriptyline. In the absence of drug (Figure5A), little steady-state block developed, and with theaddition of 20 ,uM diphenylhydantoin (Figure 5B),block increased only slightly. Steady-state block wasreached in 13 pulses (Figure 5C). By using several

frequencies of stimulation, trains of forty pulses (50msec; test, to -30 mV) were applied to obtainsteady-state block (Figure 5D). The level of steady-state block for 20 ,uM diphenylhydantoin increasedas the stimulus frequency increased, although muchless total block (even at 160 ,uM diphenylhydantoin)of the sodium current was seen than in the presenceof amitriptyline. Diphenylhydantoin, in contrast toamitriptyline, showed a steep dependence of steady-state block on the holding potential (Figure 5E).We examined the kinetics of recovery from steady-

state block during pulse train and two-pulse proto-cols in the absence (Figure 6A) and presence (Figure6B) of 20 gM diphenylhydantoin. In the absence ofdrug, very little block developed (<10%), and the rrmeasured at steady-state during pulse trains was 40msec. With the addition of diphenylhydantoin, the rrincreased to 810 msec in this cell. Determination of -rris shown in Figure 6C. As shown in Table 1, the rr fora population of 10 cells in the absence of drug was

Control

-4 l0 1

-31


Dow

nloaded from



FIGURE 5. Blocking properties ofdiphenylhydantoin. Panel A: Sodium

0

-3

Diphenylhydantoin

. T Control

20 40

Pulse NumberE

/0/ u~~~~~~~~~~~~~~~~~~~')

m

0

CD)

CD

U

0

2 4

60.

40*

20-

.]-130

current recorded from atrial myocytes3 0 nduring the first, third, fifth, seventh,

Z| L ~~~ninth, and 40th pulses in a pulse train-- N = 40 (5 Hz; pulse duration, 50 msec; holding

potential, -130 mV; step, to -30 mV).l nA Panel B: Sodium current in the same

cell during pulse train stimulation after20-minute exposure to 20 M diphenyl-

2 ms ...........hydantoin.The first, fifth, 10th, and20th, and 40th pulses are shown. Thecurrent and time calibrations are thesame for both tracings. Panel C: Plotshowing the decline in the inward so-dium current for all points in the pulsetrain in the absence (.) and presence(A) of diphenylhydantoin. Panel D:Use-dependent nature of block withdiphenylhydantoin. Data are presentedas mean±SD ofsteady-state block seenin 18 cells during superfusion with 201M diphenylhydantoin plotted as a

...... ,.2function of stimulus frequency. PanelE: Steady-state block in 20 gM diphe-

antoin nylhydantoin (mean±SD in eight cells)B) / during pulse train stimulation (40

pulses; pulse duration, 50 msec; step, to-30 mV) plotted as a function of theholdingpotential. The stippled line rep-resents similar data obtained with 0.4pM amitriptyline in another cell (see

-110 -90 Figure 2).

Frequency of Stimulation (Hz)

35+8 msec. In these same cells as well as in anadditional 12 cells, the r, in the presence of diphe-nylhydantoin increased to 710±210 msec. This valuewas in agreement with data from other voltage-clampstudies.23'24 The rr increased significantly comparedwith the rr in the absence of drug but was -20 timesless than the -rr in the presence of amitriptyline seenwith pulse train stimulation. Using a two-pulse pro-tocol (Table 1), the rr in the absence of drug was50+10 msec, whereas in the same cells subsequentlyexposed to diphenylhydantoin, the ir was 1,100+±500msec. As seen with amitriptyline, a single pulse of2,000 msec resulted in a greater decrease in thesodium current than was seen in the pulse trainprotocols (31±1-6% versus 14+8%, p<0.01).By using the two-pulse protocol as above, the

uptake time constant of diphenylhydantoin was mea-sured. Figures 6D and 6E show results in a single cell.With progressively longer prepulses, there was adecline in sodium current in the presence of diphe-nylhydantoin (Figure 6E), whereas in the absence ofdrug (Figure 6D), no decrease in sodium current wasseen until the prepulse duration increased to >2

Holding Potential

seconds. The results of this study are plotted inFigure 6F. The uptake time constant in six cells fordiphenylhydantoin was 1,150±380 msec.Because of the differences in r, between amitrip-

tyline and diphenylhydantoin (13,600± 3,200 versus710±210 msec,p<0.0001), we predicted that compe-tition between the two agents might occur. A series ofexperiments was designed to test this hypothesis.

Description of Kinetics of Sodium Current Block With aCombination ofAmitrptyline and DiphenylhydantoinWe examined whether addition of diphenylhydan-

toin altered the amount of steady-state block fromthat seen in the presence of amitriptyline alone. Byusing various stimulus frequencies, trains of 40 pulses(50 msec; holding potential -130 mV; test, to -30mV) were applied as above. As with the single drugperfusion experiments, the level of steady-state blockin the presence of both drugs increased as thestimulus frequency increased. This is demonstratedfor a single experiment in Figure 7A. In the presenceof the two sodium channel blocking agents, a higherlevel of steady-state block was seen at every stimulus

AControl

B

Diphenylhydantoin

c

c -2.3C

0. -2.7

-3.1c

:60Co

D 2006CD)

100

C 10

ax0i

0.0

~3.5 1

U 1 ' E -w


Dow

nloaded from



aDiphenyihydantoin

-30

N = 40

2 ms

a)

lnA

2 ms

= 1.3 s)

-2.0Diphenyihydantoin (?r = .81

o.~~ ~~J_ X a

Control (zr = .04 s)-2.21

0 5 10 15Recovery Interval (s)

- 1

1.5a . . . . . .

Dlphenylhydantoln (,u-2 *

-2.5 Control

* 3 *0 4 8

Prepulse Duration (a)

FIGURE 6. Recovety (panels A-C)and uptake (panels D-F) properties ofdiphenylhydantoin. Panel A: Sodiumcurrent recorded in an atrial myocyte inthe absence of drug at steady-state(smallest current) and at recovery inter-vals of 20 msec, 40 msec, 50 msec, 100msec, and 1 second (superimposed on100-msec tracing) afterpulse train stim-ulation (40 pulses; 5 Hz; holding poten-tial, -130 mV; step, to -30 mV). PanelB: Currents recorded in the same cellafter 30-minute exposure to 20 pMdiphenylhydantoin. Tracings shown areat steady-state (smallest sodium current)and at recovery intervals of 400 and 800msec and 1, 2, 2.5, and 3 seconds(superimposed on 2.5-second tracing)afterpulse train stimulation. The currentand time calibrations are the same forpanels A and B. Panel C: Recoverycurves for this cell in control (c) anddiphenylhydantoin (A) plotted with asso-ciated values for the time constant ofrecovery (ir). The continuous lines arethe single exponential fit for the data.Panel D: Sodium currents recordedfrom a different cell shown for the up-take protocol. The holdingpotential was-130 mV with a step to -30 mVPrepulse duration was serially increasedand followed 500 msec later by a 20-msec test pulse. In control, the sodiumcurrent remained constant until prepulseduration was >2 seconds. Currents areshown for test pulses after prepulses of.2 (two largest currents), 3, 5, and 10seconds. Panel E: Currents recorded inthe same cell after 25-minute exposure to20 pM diphenylhydantoin. Tracings arefrom test pulses afterprepulses of 10, 50,200, and 500 msec and 1, 2, and 4seconds. The current and time calibra-tions are the same for both panels. PanelF: Uptake curves in this cell for control(-) and diphenylhydantoin (A). ru, up-take time constant.

frequency when compared with the level of steady-state block seen with amitriptyline alone. UnlikeWhitcomb et al,22 who described a "crossover" of thepropoxyphene and propoxyphene-lidocaine curves atslower stimulus rates (<1 Hz), we were unable todemonstrate less block at any stimulus frequency inthe presence of a combination of amitriptyline anddiphenylhydantoin than with amitriptyline alone in10 cells.We evaluated whether ir was modified in the

presence of amitriptyline and diphenylhydantoin per-fusion. Results from a single experiment are shown in

Figure 7B. During exposure to a combination ofamitriptyline and diphenylhydantoin, the sodium cur-rent recovered faster (Tr, 4,800 msec) than duringexposure to amitriptyline alone (1rr, 12,400 msec) butdid not recover as rapidly during control conditionsor in the presence of diphenylhydantoin alone. The irfor 23 experiments in which cells were exposed to a

combination of amitriptyline and diphenylhydantoinwas 6,600±1,800 msec.

Surprisingly, single exponential fits of the sodiumcurrent recovery curve in the presence of the drugcombination provided as good a fit to the data as a

A Control

C

C

UE

09

CD)

-

C

OK

c

Z

0

E

0C,

A11.oQ_


Dow

nloaded from



1 2 3 4Frequency of Stimulation (Hz)

a 4.8 a)

Recovery Interval (a)

1 2 3 4

Frequency of Stimulation (Hz)

FIGURE 7. Graphs showing block and recovery with combi-nations of drugs. Panel A: Steady-state block in a single atrialmyocyte in thepresence of 0.4 MM amitriptyline alone (a) andduring superfusion with 0.4 WM amitriptyline in combinationwith 20 s.M diphenylhydantoin (+) as a function of stimulusfrequencies ranging from 0.1 to 5 Hz. Note that for allfrequencies examined, no evidence ofcrossover ofthe curves isseen. The continuous lines connect the data points. Panel B:Results from another cell showing recovery from block in thepresence of 0. 4 pM amitriptyline (m) and 0.4 pM amitriptylineand 20 plM diphenylhydantoin (+). rr, Time constant ofrecovery. The continuous lines represent the single exponentialfit of the data. Panel C: Data from atrial myocyte in thepresence of 0.4 XM amitriptyline (m)and 0.4 pMamitriptylineplus 80 pM lidocaine (.). The amount of steady-state blockseen with the mixture is greater than that seen with amitrip-tyline alone at higher stimulus frequencies. For stimulusfrequencies <1 Hz, the block observed at steady-state was lessin the presence of the mixture than with amitriptyline alone.Though these differences were small, this phenomenon was

reproduced in all eight cells with the crossoverpoint occurringat 0. 7--0.1 Hz (mean +SD).

double exponential. These results differed from theresults seen with competition between otherdrugs.22,42'43 Our data suggested that competitionfor a single channel binding site might not beoccurring. There appeared to be significant modifi-cation of the recovery from sodium channel block-ade in the presence of the drug combination with-out obvious evidence for displacement of the slowagent by the drug with the faster kinetics.We considered several possibilities to determine

why the competition between amitriptyline anddiphenylhydantoin did not seem the same as thatseen by Whitcomb et a122 with propoxyphene andlidocaine. Amitriptyline has characteristics similar topropoxyphene and other sodium channel blockingagents, but it was possible that in some way it differedin its effect at the level of the blocking site. Toevaluate this, we performed experiments examiningwhether amitriptyline could be displaced from thereceptor site by other agents (i.e., lidocaine). Theresults of one of these experiments are shown inFigure 7C. In the presence of amitriptyline, decreas-ing amounts of block of the sodium current were seenas the stimulus frequency decreased. With the addi-tion of lidocaine to the preparation, larger amountsof total block were seen at faster stimulus frequen-cies, but as the stimulation rate approached 1 Hz, thetwo curves merged and actually crossed in a mannersimilar to that described by Whitcomb et a122 forpropoxyphene and lidocaine. Although these differ-ences in the percent of block were small, the cross-over of the amitriptyline and the amitriptyline-lidocaine curves occurred consistently in all eightcells studied. This crossover reflected less block inthe presence of two sodium channel blocking agentsthan was seen in the presence of the single agent(amitriptyline) alone.From this information, it appeared that amitrip-

tyline was acting in a manner similar to propoxy-phene and could be competitively displaced from itsreceptor site by lidocaine. We considered thatlidocaine could be competing more effectively thandiphenylhydantoin for the sites occupied by amitrip-tyline within the sodium channel. The concept ofcompetition requires consideration of two pro-cesses-binding and unbinding of drugs at the levelof the channel receptor site. Examining the 'rr (un-binding) of the drugs lidocaine and diphenylhydan-toin would not explain the results found when thesedrugs were combined with amitriptyline. The rr ofdiphenylhydantoin as found in this and other stud-ies24,25 is -500-800 msec, whereas the reported ir forlidocaine under similar experimental conditions is1-2 seconds.2244 If lidocaine were to compete moreeffectively than diphenylhydantoin for the siteswithin the channel, one would predict that the un-binding of lidocaine would be faster, not slower, thanthe unbinding of diphenylhydantoin. Additionally,the "ratio" of the rr of the slow agents to the r, of thefast agents would appear to favor diphenylhydantoin

A

0

0

0

i*0

B

c

cc

Ea03

:G0

C

cat:

z

0m


Dow

nloaded from


Barber et al Evidence for Two Use-Dependent Sodium Channel Binding Sites

TABLE 2. Effects of Changing Internal and External pH on Time Constants of Recovery and on Uptake TimeConstants in the Absence of Drug and in the Presence of Amitriptyline and Diphenylhydantoin

No drug Amitriptyline DiphenylhydantoinMean±SD (msec) n Mean±SD (msec) n Mean±SD (msec) n

Ir

Control pH 30±8 6 13,600+3,200 27 710±210 23High external pH 42±10 6 4,200±100* 6 700±125 10High internal pH 33±11 6 13,300+1,700 6 2,600±1,300* 17

I-u

Control pH ... 1,351+61 6 1,152+380 6High external pH ... 1,221+202 6 1,010+ 162 6High internal pH ... 1,290+71 6 650±90t 6

n, Number of cells in each group; rr, time constant of recovery; ru, uptake time constant; control pH, pHi 7.3 and pHo7.4; high external pH, pHi 7.3 and pH1 8.0; high internal pH, pHi 8.0 and pHO 7.4.

*p<0.0001 and tp<0.001 compared with corresponding values obtained at control pH.

as the more effective competitive agent when com-pared with lidocaine.

Effects of Changes in Internal and ExternalpH onKinetics of the Sodium Current

Failure of diphenylhydantoin to reverse amitrip-tyline block was consistent with the presence of twoseparate binding sites capable of modulating sodiumchannel kinetics. There are reports in the literaturethat describe modulation of sodium channel functionthat appears to be the result of more than a singlebinding site within the sodium channel.17'8,2045As a test of this possibility, we measured the r, and

uptake rates as internal and external pH were varied(Table 2). Under control conditions, the nominalinternal pH was 7.3 and external pH was 7.4. The rrvalues in the absence of drug, presence of amitrip-tyline, and presence of diphenylhydantoin were 30,13,600, and 710 msec, respectively. In separate exper-iments, the r, was evaluated in the absence andpresence of drug under conditions of high externalpH (8.0). In the absence of drug (n = 6), the Tr, thoughnot significantly different (42+10 msec) from controlconditions (30+8 msec), tended to be longer. Similarresults under somewhat different voltage-clamp con-ditions have been seen when altering external pH.46

In cells perfused with pH 8.0 solution duringamitriptyline exposure, the rr was decreased from-14,000 to 4,200+100 msec (p<0.001). Becauseamitriptyline is a weak base (pK 9.4), increasing thepH of the perfusing solution containing amitriptylineshould increase the amount of neutral drug and makethe drug more effective in accessing and leaving thebinding site, thus lowering the Tr. In cells perfusedwith diphenylhydantoin (pK 8.3) in external solutionwith pH 7.4, the -rr was 710+210 msec. When the pHof the external perfusate was increased to 8.0, the rr(n= 10 cells) was 700+125 msec. This was not differ-ent from the value seen under conditions of normalpH. Additionally, changes in external pH from 7.3 to8.0 did not alter the uptake time constants measuredwith two-pulse protocols of amitriptyline (1,351 +61

versus 1 221 +202 msec,p=NS) or diphenylhydantoin(1,152±380 versus 1,010+161 msec,p=NS).

Similar experiments were performed keeping thepH of the external perfusate constant at 7.4 but nowaltering the pH of the pipette solution from 7.3 to 8.0.In the absence of drug (n=6), the -rr was not altered(mr, 33±11 msec) from that seen when the pH of thesolution in the pipette was 7.3. In cells perfused underthese same pH conditions with amitriptyline (n=6),the -Tr was not significantly changed from the r, atinternal pH 7.3 (13,300±1,700 versus 13,600+3,200msec, p=NS). On the other hand, in experiments inwhich the internal pH was increased and cells weresuperfused with diphenylhydantoin (n= 17), the rr wasincreased to 2,600+1,300 msec (p<0.0001 comparedwith the 'Jr of diphenylhydantoin at internal pH 7.3).Because diphenylhydantoin is an anion, a larger frac-tion of receptor bound drug should be charged at thehigher pH. Assuming that charged drug dissociatesmore slowly from the receptor, the results should beinternally consistent with the opposite effect of exter-nal pH on amitriptyline recovery. The uptake timeconstants measured with two-pulse protocols in thepresence of diphenylhydantoin with the internal solu-tion at pH 8.0 decreased from 1,152+380 to 650±90msec (p<0.0001); the uptake time constant for ami-triptyline was unchanged from control conditions(1,351±61 versus 1,290±71 msec,p=NS). These datasuggest that sodium channel blockade by diphenylhy-dantoin was modulated by changes in the internal pHonly, whereas that by amitriptyline was modulatedfrom a site sensitive to changes in the external pH.

Results of Single-Channel StudiesWe hoped to demonstrate block of single sodium

channels during diphenylhydantoin exposure by com-paring the current during the conditioning and thetest pulses. The recovery interval between the condi-tion and the test had to be sufficiently long to permitrecovery from inactivation but not to allow substan-tial recovery of diphenylhydantoin-induced block.Figure 8 shows single sodium channel current in theabsence of drug with a recovery interval of 500 msec

689


Dow

nloaded from



-90C 1 1 Vt

Av c o-s ----I

B -k.F #

-A'#R-gIWAbapqioAwbd0ki4A4~Vo

3/1 pA

20 ms

FIGURE 8. Effect of recovery interval between conditioningand test pulses. Tracings recorded from a single cell-attachedpatch in a ventricular myocyte in the absence of drug are

shown. Panel A: Single channel (first five tracings) andaveraged current (lower tracing) recordings obtained duringconditioning (V,) and test (V,) pulses separated by 500 msec.

The pulse paradigm is shown at the top of the figure, and onlythe first 40 msec of the 1 second V, is depicted. Ensemblecurrent averages in Vc and Vt were similar in all cells in theabsence of drug when V, and V, were separated by 500 msec.

Panel B: Currents recorded from the same patch 15 minutesafter those in panel A, except that in this experiment V,preceded V, by only 250 msec. As can be seen, there issignificant inactivation ofsodium channels by the conditioningpulse, and 250 msec was insufficient time to allowfull recoveryofthese inactivated channels. The time and current calibrationlines are the same for both panels. Current calibration was 3pA for single-channel recordings and 1 pA for ensemblecurrent recordings.

(panel A) or 250 msec (panel B). The holdingpotential was -90 mV, and the test potential was-20 mV. Single sodium channel openings with amean amplitude of 1.4 pA occurred after very brieflatency in the conditioning and test pulses. Meanopen times during the conditioning and test pulseswere 0.65 and 0.66 msec, respectively. The peakamplitude of the averaged current was the same inthe conditioning and test pulses. The current tracingsin panel B were obtained from the same patch as

those in panel A after 15 minutes had elapsed. The

recovery interval in this run was 250 msec. In com-paring the averaged current in the conditioningpulses in panels A and B, there was significantrundown of current over the 15-minute period. Thenumber of null sweeps was increased, and the aver-aged current for the test pulse was decreased. Thissuggests that significant inactivation occurs after arecovery interval of only 250 msec. Data from fiveexperiments with a recovery interval of 500 msec aresummarized in Table 3. The 500-msec recovery pe-riod was sufficient to permit recovery from inactiva-tion.

Figure 9 illustrates current in conditioning and testpulses in an experiment in which the cells had beenexposed to 80 gM diphenylhydantoin and the micro-electrode was drug free. The averaged current in thetest pulse was less than that during the conditioningpulse. The decrease in current in the test pulseresulted primarily from an increase in the fraction ofnull sweeps. In the sequences 2, 3, 4, 5, and 10,openings in the conditioning pulse were followed bynulls in the test pulse. In this experiment, the meanopen time was 0.73 msec during the conditioningpulse and 0.64 msec during the test pulse. The resultsof five experiments are summarized in Table 3. Theintegral of the current decreased by 46% during thetest pulse and the apparent probability of a channelto fail to open increased significantly from 0.52±+0.19during the conditioning pulse to 0.70±+0.11 duringthe test pulse. The mean open time was not signifi-cantly decreased between the conditioning and thetest pulses.

Similar experiments in the absence of drug (n=4)performed at a holding potential of -120 mV dem-onstrated that mean open time of the sodium channelwas not different between conditioning (0.65+0.11msec) and test (0.67±0.10 msec) pulses. These valuesalso were not different from those at -90 mV. Theprobability that a channel would fail to open in bothconditioning and test pulses was identical(0.21±0.10). When diphenylhydantoin was added tothe superfusate (n=4), no difference in mean chan-nel open time was seen between conditioning (0.70±0.18 msec) and test (0.68±0.13 msec) pulses. Addi-tionally, the probability of the channel to fail to openwas not different between conditioning (0.19±0.08)and test (0.21±0.11) pulses. These experiments showthat diphenylhydantoin can produce block of singlesodium channels when a membrane patch is exposedto drug-free solution and that this block is voltagedependent. It is assumed that during the prior incu-bation in diphenylhydantoin-containing solution thedrug was able to diffuse across the cell membraneand produce block from an intracellular or in-tramembrane site.We performed six experiments in which the micro-

electrode contained 80 ,uM diphenylhydantoin andthe cells were not exposed to drug in the solution.Although there was a trend for the drug to decreasethe current during the test pulse and to increase the

-0- - PM.0or

VA olooikm 114 0 h


Dow

nloaded from


Barber et al Evidence for Two Use-Dependent Sodium Channel Binding Sites

TABLE 3. Effects of Diphenylhydantoin on Single Sodium Channel Current

<t> (msec) pf fldt (pA. msec) fIvtdt/flvdtControl (n=5)

V, 0.71+0.11 0.51+0.11 115+36Vt 0.71±0.10 0.52±0.14 118±40 1.02

DiphenylhydantoinBath (n=5)

V, 0.66±0.06 0.52±0.19 139±77Vt 0.62±0.05 0.70±0.11* 75±35* 0.54

Pipette (n=6)VC 0.56±0.10 0.73±0.16 74±40Vt 0.50±0.07 0.77±0.12 50±18 0.68

Values are Mean±SD. <t>, Mean open time; pf, probability channel will fail to open; fldt, integral of single-channelcurrent; V,, conditioning pulse; Vt, test pulse.

*p<0.05 compared with corresponding value for V,.

probability of the channel to fail to open, the changeswere not statistically significant.

-201 vc -

-90

--- - - - - -- L *I-

_~~~~~~~~~ ---'fL A t .

, - --

jTV -1. ~ .

1f~~~~~--. V4'.k

311 pA

20 ms

FIGURE 9. Effect of diphenylhydantoin on single sodiumchannel kinetics. Shown are the single-channel current re-

sponses during conditioning (V,) and test (Vt) pulses for 10consecutive trials with 80 MM diphenylhydantoin present in thesuperfusate. The pulse paradigm is shown at the top of thefigure with only the first 40 msec of Vc depicted and V,separated from V, by 500 msec. Averaged current for theconditioning and testpulses are shown in the bottom tracings.In the presence of diphenylhydantoin, the averaged current inthe test pulse was significantly less than that seen in theconditioning pulses. Current and time calibration lines are as

shown. Current calibration was 3 pA for single-channel re-

cordings and 1 pA for ensemble current recordings.

DiscussionMajor Findings

Based on previous studies demonstrating the com-petitive relief of block, our studies have examined theindividual and combined blocking actions of amitrip-tyline and diphenylhydantoin on the sodium currentin cultured rabbit atrial myocytes. Our major findingsare as follows: 1) Both amitriptyline and diphenylhy-dantoin demonstrate use- and frequency-dependentblockade of the sodium channel. 2) Amitriptylinedemonstrates a longer r, (intracellular pH 7.3, extra-cellular pH 7.4) than diphenylhydantoin under simi-lar conditions. 3) Uptake time constants of the twodrugs are similar. 4) Increased external pH signifi-cantly decreased the rr of amitriptyline (pK 9.4)without affecting the rr of diphenylhydantoin (pK8.3), whereas increased internal pH increased the r,for diphenylhydantoin without affecting the r, ofamitriptyline. 5) Lidocaine attenuated the amitripty-line-induced block at low rates of stimulation. Unlikeprevious studies2243 using similar agents, the com-bined application of amitriptyline and diphenylhy-dantoin did not appear to reverse the blockade. Thisdifference in our results might have been caused bythe location of the binding site of the diphenylhydan-toin anion and the cationic form of tricyclic antide-pressants or local anesthetics.

Effects ofAmitriptyline on the Sodium CurrentLittle in vitro data exist concerning the effects of

amitriptyline as a sodium channel blocking agent.More data exist on the tricyclic antidepressant imip-ramine, which has similar structure to amitriptyline.Voltage-clamp studies in Myxicola giant axon47 foundimipramine to reversibly suppress sodium conduc-tance at low concentrations while not affecting theresting membrane potential. Imipramine did shift theconductance-voltage curve in the hyperpolarizing di-rection in a concentration-dependent fashion but didnot change the time to peak current or the time tohalf-maximal steady state.

691


Dow

nloaded from



In cardiac tissue, most of the effects of tricyclicantidepressants have been described using informa-tion obtained from Vmax data in Purkinje fibers.Rawling and Fozzard48 showed that imipramine re-duced the rate of rise of Vm.X and markedly slowedconduction velocity in canine Purkinje fibers. Similarstudies by Brennan,49 Weld and Bigger,50 and Muir eta151 in Purkinje fibers and by Weld and Bigger50 inventricular muscle corroborated the effects of tricy-clic antidepressants on cardiac tissue. Sasyniuk andJhamandas52 observed that, although amitriptyline(1.8 ,M) did not affect maximum diastolic potentialof the cells, it significantly depressed action potentialamplitude and duration as well as Vmax. This concen-tration of amitriptyline was two to four times thetherapeutic level of the drug but was well within therange of concentrations obtained in patients withoverdose. These studies all implied that the tricyclicantidepressants, including amitriptyline, acted as so-dium channel blocking agents.Ogata and Narahashi39 demonstrated that at very

negative holding potentials (-140 mV) imipraminecaused little resting block but that shifting the hold-ing potential to less negative values increased theamount of resting block seen. During repetitive stim-ulations, significant use-dependent block was seen.There also was voltage dependence, but no shift inthe current-voltage relation was seen when comparedwith the drug-free state. Steady-state inactivation wasnegatively shifted -18 mV. The recovery from drug-induced block was relatively slow (1.8+0.3 seconds).The results of our study with amitriptyline support

the concept of potent sodium channel blockade bythis group of drugs in cardiac tissue39,48-51 with someimportant differences. Although both amitriptylineand imipramine have use- and frequency-dependentblocking properties on the sodium current, amitrip-tyline appears to be more potent. At concentrationsof 0.4 ,M, amitriptyline caused marked reduction ofsodium current (78% block at 5 Hz; 60-70% block at2 Hz, see Figure 2), whereas under similar voltage-clamp conditions, 3-Hz stimulation in the presence of3 ,M imipramine resulted in -50% block (Figure 6from Reference 39). Neither drug appeared to sig-nificantly block resting sodium channels at hyperpo-larized potentials (holding potential, -130 to -140mV).

Imipramine and amitriptyline seemed to demon-strate relatively different properties of dissociationfrom the binding site under voltage-clamp conditions.Compared with lidocaine with a r, of 1-2 seconds ordiphenylhydantoin with a r, of 0.7 second, the r, ofimipramine (1.8 seconds) was similar, whereas the r,of amitriptyline was very slow (13.6 seconds).22Drugs with slow dissociation kinetics from cardiac

tissue may demonstrate greater degrees of cardiotox-icity than drugs with more rapid r,. Slowed timeconstants of recovery of sodium current48,52 anddecreased conduction velocity of cardiac tissues8 bytricyclic antidepressants most probably extrapolatedirectly to the clinical findings associated with toxic-

ity including increased atrioventricular conductiontime, decreased intraventricular conduction, and in-creased propensity for arrhythmias.8 914,16,52,53

Effects of Diphenylhydantoin on the Sodium CurrentInterest in examining whether diphenylhydantoin

and amitriptyline were interacting at the same site atthe level of the sodium channel was stimulated by theclinical controversy describing reversal of tricyclicantidepressant-induced cardiac conduction abnor-malities with diphenylhydantoin.1,9-15,22 We22 re-cently reported normalization of cardiac conductionabnormalities in a case of propoxyphene overdosewhen lidocaine, an agent with significantly differentrecovery kinetics from those of propoxyphene, wasadministered. Review of the literature yielded sup-port of the concept that amitriptyline had slowrecovery kinetics and diphenylhydantoin had fastrecovery kinetics at the level of the sodium channeland that, under appropriate conditions, competitiverelief of block from amitriptyline by diphenylhydan-toin might be demonstrable.The effects of diphenylhydantoin on the sodium

current have been examined in studies using severalmodels. In neuroblastoma cells under voltage-clampconditions, Willow et a124 found use-dependent blockof voltage-activated sodium currents with a rapidonset of block (steady-state block reached in seven to10 pulses) at concentrations of 30 btM diphenylhy-dantoin. Depolarization of the holding potential in-creased the amount of resting block as well as theamount of block seen at steady state. Half time ofrecovery from block was 1.36 seconds at a holdingpotential of -90 mV.

Sanchez-Chapula and Josephson23 evaluated theeffects of diphenylhydantoin in rat ventricular cellsduring voltage-clamp experiments. They found theblocking effect of diphenylhydantoin on the sodiumcurrent to be both voltage and use dependent. At aconcentration of 20 ,uM diphenylhydantoin and aholding potential of -90 mV, sodium current wasreduced 45% during steady-state stimulation. Usinga two-pulse protocol, the -r of six cells was 1,367+ 126msec. When the holding potential was reduced to-80 mV, 7r increased to 1,514 msec, and when thecells were hyperpolarized to -105 mV, rr decreasedto 718 msec. Their -r for cells in the absence of drugwas 70-80 msec. Xu et al,25 using rat ventricularmyocytes and 40 ,M diphenylhydantoin, reportedlow levels of steady-state block (31±4%) with 950-msec pulses. They also demonstrated a rapid -r(520±140 msec) when cells were held at -140 mV.Our results agree with those previously pub-

lished.23,25 Under voltage-clamp conditions we dem-onstrated low levels of steady-state block (14+2%)when the cells were held at -130 mV and stimulatedwith 50-msec pulses. Our preliminary experiments(n = 4) had shown that longer pulse durations re-sulted in greater amounts of steady-state block at thesame stimulus frequency. Our rr for diphenylhydan-toin correlated well with those seen in neuroblastoma


Dow

nloaded from



cells24 and in rat ventricular myocytes.23'25 Onset ofblock in our preparation was rapid and reachedsteady state in 10-15 pulses. The amount of steady-state block was steeply voltage dependent.Because of clinical information using diphenylhy-

dantoin to treat amitriptyline overdose1l9 and exper-imental demonstration of competitive relief of blockwith other drugs,22'43'44 we felt that investigation ofreversal of amitriptyline effects by diphenylhydantoinwas reasonable.

Effects of Combinations of Sodium Channel BlockingAgents on Sodium Channel Kinetics

Hille40 and Schwarz et a154 proposed a model oflocal anesthetic action based on a common receptorsite located within the sodium channel. This receptorsite could bind both charged and neutral drugs. Thiscommon receptor site has gained widespread supportin the literature through the years.Rimmel et al'8 examined the concept of the com-

mon receptor site by looking at the interaction of twodrugs - procaine and benzocaine - at the level of thesodium channel in the frog node of Ranvier. Usingkinetic analysis, these authors concluded that pro-caine (pK 8.9, cationic form at pH 7.0) appeared tohave limited access to the receptor site because of thecharged nature of the molecule over a wide range ofpHs, whereas benzocaine (pK 2.6, uncharged over awide range of pH) essentially had unlimited access tothe receptor site at all pHs tested. When the kineticsof the channel were analyzed in a mixture of procaineand benzocaine with equieffective concentrations ofthe two drugs, the membrane responded as it did inbenzocaine alone. The authors postulated that ben-zocaine and procaine acted at a common receptorsite and that the rate of sodium channel blockappeared to be limited by access to the receptor site.Benzocaine, the neutral drug, had easy entry into thechannel and, hence, access to the binding site,whereas entry into the channel by procaine, thecharged drug, was limited. In three experiments, theauthors saw an increase in Vmax during exposure tobenzocaine and procaine compared with procainealone. The "overall" binding kinetics of the mixturewere similar to those seen in the presence of ben-zocaine alone. Mathematical modeling (Figure 9 ofReference 18) of this reversal could not be fittedusing two separate and independent binding sites forthe drugs. Modeling for one binding site resulted inclose agreement with their experimental results.Modeling by Starmer21 supported the concept of asingle binding site for the drugs but stressed theimportance of varying frequency of stimulation as ameans to demonstrate apparent reversal of blockade.

Evidence exists for the presence of more than asingle receptor site that may affect the sodium chan-nel. Mrose and Ritchie17 observed that competitionfor blockade of the sodium channel existed betweenthe drugs benzocaine and lidocaine. From their data,they concluded that there appeared to be a secondsite of action for benzocaine in addition to the

sodium channel receptor site classically affected bylidocaine. Previously, Shrivastav et a120 had demon-strated that volatile compounds such as alcohols orgeneral anesthetics produced effects on the inactiva-tion curve of the sodium channel similar in magni-tude, but by a different mechanism, to effects pro-duced by benzocaine. This led to the postulate that,although a common receptor may reside within thesodium channel itself and predominate in the mod-ulation of sodium channel function,40'54 at least oneother site of drug action appeared to be involved inthe control of sodium channel kinetics and function.Alpert et a145 have postulated the presence of asecond local anesthetic binding site present outsideor near the outside of cardiac sodium channels.The concept of competitive reversal in the pres-

ence of two sodium channel blocking agents in car-diac tissue has been examined by several au-thors.22'4344'55 Whitcomb et a122 described that a drugwith rapid association (binding) and dissociation(unbinding) kinetics may, in some circumstances,displace a drug with slower binding/unbinding kinet-ics such that effects produced by the slower drug maybe altered or "reversed" with the addition of anagent with faster kinetics. We further showed thatthe degree of steady-state sodium channel blockadeby a drug with a slow unbinding rate could bereduced by adding a drug with faster recovery kinet-ics. In these studies, the "slow" drug, propoxyphene,exhibited a rr of -23 seconds, whereas the "fast"competitor, lidocaine, exhibited a rr of 1-2 seconds.In our present studies, we did not observe a reduc-tion in the steady-state blockade when amitriptyline(Tr, 13.6 seconds) competed with diphenylhydantoin(Trr, 0.7 second). Was this evidence of two separatebinding sites where a competitive effect would notoccur, or was this compatible with a single bindingsite?To explore this question, we looked at the dynam-

ics of drug mixture. In the presence of both amitrip-tyline and diphenylhydantoin, there was no evidenceof competitive displacement of amitriptyline at anystimulus frequency examined (no crossover as seenwith propoxyphene and lidocaine22 or when lidocainecompeted with amitriptyline in the present study).Though the ir of amitriptyline and diphenylhydantoinvaried by a factor > 10, we were unable to reproduceresults similar to those obtained with propoxypheneand lidocaine22 or amitriptyline and lidocaine. Weobserved that the binding rate for amitriptyline wasslightly larger than that of diphenylhydantoin. In-creasing the concentration of diphenylhydantoin by30 to reverse this relation also was ineffective inproducing a crossover. However, competition with 80,uM lidocaine did produce this phenomenon. Thesedata suggested to us that, whereas amitriptyline andlidocaine were binding to a common site, diphenyl-hydantoin was binding to a separate site. This stim-ulated us to explore alternative experimental ap-proaches to distinguish whether a common binding


Dow

nloaded from



site was involved or whether two binding sites neededto be postulated.We performed experiments varying internal (7.3)

and external (7.4) pH while other factors were heldconstant to assist in evaluating whether different bind-ing sites might be involved. We found pronounceddifferences in the effect of changing extracellular orintracellular pH on the recovery kinetics of amitrip-tyline and diphenylhydantoin. In the presence of in-creased extracellular pH, amitriptyline (pK 9.4) dem-onstrated a marked decrease in Tr. The rr fordiphenylhydantoin (pK 8.3) was not affected bychanges in extracellular pH. This suggested that ami-triptyline, but not diphenylhydantoin, bound to a sitethat was accessible to changes in external pH. Con-versely, when the nominal intracellular pH waschanged to 8.0 in the presence of amitriptyline ordiphenylhydantoin, no alteration in binding or unbind-ing kinetics of amitriptyline was seen, whereas thebinding and unbinding kinetics of diphenylhydantoinwere significantly modified. This was interpreted toreflect that diphenylhydantoin, but not amitriptyline,bound to a site that is susceptible to modulation bychanges in internal pH of the cell. Morello et a156 alsoexamined the blocking action of diphenylhydantoin onthe sodium current in squid giant axon as extracellularand intracellular pH values were varied, but they con-sidered first pulse effects only. Both extracellular andintracellular pH affected block, depending on the siteof drug application. They did not examine the kineticsof the development and recovery from phasic block.Because of differences in the stimulation protocol, theirresults cannot be compared with ours.

Effects of Diphenylhydantoin on SingleSodium ChannelsWe attempted to further define the blocking mech-

anism and site of action of diphenylhydantoin by per-forming single sodium channel measurements withdrug-free microelectrodes on cells previously exposedto drug or with drug-containing microelectrodes in cellsnot previously exposed to drug. Although the experi-mental paradigm was not ideal, it was practical underour recording conditions. A similar approach was usedby Carmeliet et a157 to examine block of single sodiumchannels by penticainide.We performed these experiments in cell-attached

patches only. Experiments with inside-out patcheswith drug applied to the internal membrane surfacewould appear attractive, but single sodium channelkinetics are not stable in excised patches.29-32 Therewould have been an advantage in obtaining drug-freemeasurements in a patch and then applying the drugto the cell or in exchanging drug-free microelectrodesolution with drug-containing solution. Such experi-ments were not practical in terms of the amount oftime (15 minutes) required to obtain the initial dataset and the time during which stable recording couldbe obtained from a seal. The emphasis of the com-parison in these experiments is the current in theconditioning and test pulses

In the absence of drug, a 500-msec rest period wasenough to permit recovery from inactivation devel-oped during the conditioning pulse. With drug-freemicroelectrodes and prior exposure of the cells todiphenylhydantoin, the averaged current during thetest pulse was decreased at a holding potential of-90 mV. The primary mechanism of the decreasewas a decrease in the number of channel openings.The mean open time was not significantly decreased.Using a similar protocol, McDonald et a158 concludedthat lidocaine blocked the sodium current during thetest pulse primarily by decreasing the number ofopenings. Unlike our results, they also noted a de-crease in channel open time in the test pulse. Be-cause of this observation, they suggested thatlidocaine-associated channels may conduct with adecreased open time. Their experiments were per-formed with lidocaine present in both the microelec-trode and the bath.The fact that we observed block in the experiments

without diphenylhydantoin in the microelectrode isconsistent with block occurring from an intracellularand/or intramembrane site. The fact that we did notobserve significant block when drug was present onlyin the microelectrode may be a concentration-depen-dent phenomenon. With drug in the microelectrodeonly, the cell acts as a large sink for drug and dilutionof the drug concentration may occur. Resolution ofthis uncertainty would not be made any easier bypatch excision, because the entire bath would thenact as a sink for the drug.The results of the whole-cell voltage-clamp exper-

iments, including those with variations of intracellu-lar and extracellular pH, and the single-channelrecordings can best be integrated by observing thatdiphenylhydantoin is able to cross the cell membraneand bind to some (internal?) site that is not accessi-ble to external protons and that amitriptyline binds toa more superficial site that is accessible to externalprotons. We and others21 have reported that a modelof binding to separate sites predicts enhanced blockduring exposure to the combination of drugs with fastand slow kinetics.

AcknowledgmentsWe would like to acknowledge the excellent tech-

nical support of Ms. Anne Stone and thank Ms. PatDean for preparing the manuscript.

References1. Hagerman GA, Hanashiro PK: Reversal of tricyclic-

antidepressant-induced cardiac conduction abnormalities byphenytoin. Ann Emerg Med 1981;10:82-86

2. Spiker DG, Weiss AN, Chang SS, Ruwith JF Jr, Biggs J:Tricyclic antidepressant overdosage: Clinical presentation andplasma levels. Clin Phannacol Ther 1975;18:539-546

3. Dumovic P, Burrows GD, Hohra J, Davies B, Scoggins BA:The effect of tricyclic antidepressant drugs on the heart. ArchToxicol 1976;35:255-262

4. Sacks MH, Bonforte RI, Lasser RP, Dimich I: Cardiovascularcomplications of imipramine intoxication. JAMA 1968;205:116-118


Dow

nloaded from



5. Barnes RJ, Kong SM, Wu RWY: Electrocardiographicchanges in amitriptyline poisoning. Br Med J 1968;3:222-223

6. Serafimovski N, Thorball N, Amussen I, Lundig M: Tricyclicantidepressive poisoning with special reference to cardiaccomplications. Acta Anaesthesiol Scand Suppl 1975;57:55-63

7. Thorstrand C: Clinical features in poisoning by tricyclic anti-depressants with special reference to the ECG. Acta MedScand 1976;199:337-344

8. Bigger JT, Giardina EGV, Perel JM, Kantor ST, GlassmanAH: Cardiac antiarrhythmic effect of imipramine hydrochlo-ride. N Engl J Med 1977;296:206-208

9. Boehnert M, Lovejoy FH Jr: The effect of phenytoin oncardiac conduction and ventricular arrhythmias in acute tricy-clic antidepressant overdose (abstract). Vet Hum Toxicol 1985;28:297

10. Bryson PD: Comprehensive Review in Toxicology. Rockville,Md, Aspen Press, 1986, pp 119-140

11. Kulig K: Emergency case records: Prophylactic phenytoin inTCA overdose. J Emerg Med 1983;1:169-177

12. Callaham M: Tricyclic antidepressant overdose. J Am CollEmerg Phys 1979;8:413-425

13. Davis JM: Overdose of psychotropic drugs-tricyclic antide-pressants. PsychiatrAnn 1973;3:6-11

14. Callaham M, Schumaker H, Pentel P: Phenytoin prophylaxisof cardiotoxicity in experimental amitriptyline poisoning. JPharmacol Exp Ther 1988;245:216-220

15. Mayron R, Ruiz E: Phenytoin: Does it reverse tricyclic anti-depressant induced cardiac conduction abnormalities? AnnEmerg Med 1986;15:876-880

16. Nattel S, Mittelman M: Treatment of ventricular tachyarrhyth-mias resulting from amitriptyline toxicity in dogs. J PharmacolExp Ther 1984;231:430-435

17. Mrose HE, Ritchie JM: Local anesthetics: Do benzocaine andlidocaine act at the same single site? J Gen Physiol 1978;71:223-225

18. Rimmel C, Walle A, Kessler H, Ulhricht W: Rates of block byprocaine and benzocaine and the procaine-benzocaine inter-action at the node of Ranvier. Pflugers Arch 1978;376:105-118

19. Sanchez-Chapula J: Interaction of lidocaine and benzocaine indepressing Vmax of ventricular action potentials. J Mol CellCardiol 1985;17:495-503

20. Shrivastav BB, Narahashi T, Kitz RJ, Roberts JD: Mode ofaction of trichloroethylene on squid axon membranes. J Phar-macol Exp Ther 1976;199:179-188

21. Starmer CF: Theoretical characterization of ion channelblockade: Competitive binding to periodically accessiblereceptors. Biophys J 1987;52:405-412

22. Whitcomb DC, Gilliam FR III, Starmer CF, Grant AO:Marked QRS complex abnormalities and sodium channelblockade by propoxyphene reversed with lidocaine. J ClinInvest 1989;84:1629-1636

23. Sanchez-Chapula J, Josephson IR: Effect of phenytoin on thesodium current in isolated rat ventricular cells. J Mol CellCardiol 1983;15:515-522

24. Willow M, Gonoi T, Catterall WA: Voltage-clamp analysis ofthe inhibitory actions of diphenylhydantoin and carbamaze-pine on voltage-sensitive sodium channels in neuroblastomacells. J Mol Pharmacol 1985;27:549-558

25. Xu Y-Q, Pickoff AS, Clarkson CW: Evidence for differences inphenytoin block of Na channels in adult vs neonatal ratmyocytes (abstract). Circulation 1989;80(suppl II):II-606

26. Gilliam FR III, Starmer CF, Grant AO: Blockade of rabbitatrial sodium channels by lidocaine: Characterization of con-tinuous and frequency-dependent blocking. Circ Res 1989;65:723-739

27. Eisenberg RS, Johnson EA: Three dimensional electrical fieldproblems in physiology. Prog Biophys Mol Biol 1970;20:1-65

28. Hamill OP, Marty A, Neher E, Sakmann B, Sigworth EJ:Improved patch-clamp techniques for high-resolution currentrecording from cells and cell-free membrane patches. PflugersArch 1981;391:85-100

29. Aldrich RW, Stevens CF: Voltage-dependent gating of singlesodium channels from mammalian neuroblastoma cells. JNeurosci 1987;7:418-431

30. Horn R, Vandenberg CA: Inactivation of single sodium chan-nels, in Ritchie J, Keynes R (eds): Ionic Channels in NeuralMembranes. New York, Alan R Liss, Inc, 1986, pp 71-83

31. Kirsch GE: Modified single sodium channel gating in excisedmembrane patches from neonatal rat ventricular myocytes, inHondeghem LM (ed): MolecularMechanisms ofAntiarrhythmicAgents. Mt Kisco, NY, Futura Publishing Co, Inc, 1989, pp81-99

32. Nilius B: Modal gating behavior of cardiac sodium channels incell-free membrane patches. Biophys J 1988;53:857-862

33. Marquardt DW: An algorithm for least-squares estimation ofnon-linear parameters. J Soc Indust Appl Math 1963;11:431-441

34. Grant AO, Dietz MA, Gilliam FR III, Starmer CF: Blockadeof cardiac sodium channels by lidocaine: Single channel analy-sis. Circ Res 1989;65:1247-1262

35. Gilbert N: Statistics. Philadelphia, WB Saunders, Co, 197636. Snedecor GW, Cochran WG: Statistical Methods. Ames, Iowa,

Iowa State University Press, 197137. Hondeghem LM, Katzung BK: Time and voltage-dependent

interactions of antiarrhythmic drugs with cardiac sodiumchannels. Biochim Biophys Acta 1977;472:373-398

38. Starmer CF, Grant AO, Strauss HC: Mechanism of use-dependent block of sodium channels in excitable membranesby local anesthetics. Biophys J 1984;46:15-27

39. Ogata N, Narahashi T: Block of sodium channels by psycho-tropic drugs in single guinea-pig cardiac myocytes. Br J Phar-macol 1989;97:905-913

40. Hille B: Local anesthetics: Hydrophilic and hydrophobic path-ways for drug-receptor reaction. J Gen Physiol 1977;69:497-515

41. Khodorov B, Shishkova L, Peganov E, Revenko S: Inhibitionof sodium currents in frog Ranvier node treated with localanesthetics: Role of slow sodium inactivation. Biochim BiophysActa 1976;433:409-435

42. Bennett PB, Woosley RL, Hondeghem LM: Competitionbetween lidocaine and one of its metabolites, glycylxylide, forcardiac sodium channels. Circulation 1988;78:692-700

43. Kodama I, Toyama J, Yamada K: Competitive inhibition ofcardiac sodium channels by aprindine and lidocaine studiedusing a maximum upstroke velocity of action potential inguinea-pig ventricular muscle. JPharmacol Exp Ther 1987;241:1065-1071

44. Bean BP, Cohen CJ, Tsien RW: Lidocaine block of cardiacsodium channels. J Gen Physiol 1983;81:613-642.29

45. Alpert LA, Fozzard HA, Hanck DA, Makielski JC: Is there asecond external lidocaine binding site on mammalian cardiaccells. Am J Physiol 1989;26:H79-H84

46. Wendt D, Starmer F, Grant A: Recovery kinetics of sodiumcurrent are accelerated by decrease in external pH (abstract).Biophys J 1990;57:300a

47. Schauf CL, Davis FA, Kessler RL: Actions of the antidepres-sant drug imipramine on the voltage-clamped Myxicola giantaxon. J Pharmacol Exp Ther 1975;193:669-675

48. Rawling DA, Fozzard HA: Effects of imipramine on cellularelectrophysiological properties of cardiac Purkinje fibers. JPharmacol Exp Ther 1979;209:371-375

49. Brennan FJ: Electrophysiologic effects of imipramine anddoxepin on normal and depressed cardiac Purkinje fibers. AmJ Cardiol 1980;46:599-606

50. Weld FM, Bigger JT Jr: Electrophysiologic effects of imipra-mine on ovine cardiac Purkinje and ventricular muscle fiber.Circ Res 1980;46:167-175

51. Muir WW, Strauch SM, Schaal SF: Effects of tricyclic antide-pressant drugs on the electrophysiologic properties of dogPurkinje fibers. J Cardiovasc Pharmacol 1982;4:82-90

52. Sasyniuk BI, Jhamandas V: Mechanism of reversal of toxiceffects of amitriptyline on cardiac Purkinje fibers by sodiumbicarbonate. J Pharmacol Exp Ther 1984;231:387-394

53. Nattel S, Keable H, Sasyniuk BI: Experimental amitriptylineintoxication: Electrophysiologic manifestations and manage-ment. I Cardiovasc Pharmacol 1984;6:83-89


Dow

nloaded from



54. Schwarz W, Palade PT, Hille B: Local anesthetics: Effect of

pH on use-dependent block of sodium channels in frog muscle.Biophys J 1977;20:343-368

55. Clarkson CW, Hondeghem LM: Evidence for a specific recep-

tor site for lidocaine, quinidine and bupivacaine, associatedwith cardiac sodium channels in guinea-pig ventricular myo-cardium. Circ Res 1985;56:498-506

56. Morello RS, Begenisich T, Yeh J: Determination of the active

form of phenytoin. J Pharmacol Exp Ther 1984;230:156-161

57. Carmeliet E, Nilius B, Vereecke J: Properties of the block ofsingle Na channels in guinea-pig ventricular myocytes by the localanesthetic penticainide. J Physiol (Lond) 1989;409:241-262

58. McDonald TV, Courtney KR, Clusin WT: Use-dependentblock of single sodium channels by lidocaine in guinea-pigventricular myocytes. Biophys J 1989;55:1261-1266

KEY WORDS * lidocaine * amitriptyline * sodium channel .

diphenylhydantoin


Dow

nloaded from


M J Barber, C F Starmer and A O Grantfor two use-dependent binding sites.

Blockade of cardiac sodium channels by amitriptyline and diphenylhydantoin. Evidence

Print ISSN: 0009-7330. Online ISSN: 1524-4571 Copyright © 1991 American Heart Association, Inc. All rights reserved.is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation Research

doi: 10.1161/01.RES.69.3.6771991;69:677-696Circ Res.

http://circres.ahajournals.org/content/69/3/677World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://circres.ahajournals.org//subscriptions/

is online at: Circulation Research Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. Further informationEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculation Researchin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:


Dow

nloaded from

http://circres.ahajournals.org/content/69/3/677

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://circres.ahajournals.org//subscriptions/


blockade of cardiac sodium channels by amitriptyline and

Documents