Biophysical Jounal Volume 67 September 1994 1007-1014

The SaxitoxinlTetrodotoxin Binding Site on Cloned Rat Brain Ila NaChannels is in the Transmembrane Electric Field

Jonathan Satin, James T. Limberis, John W. Kyle, Richard B. Rogart, and Harry A. FozzardCardi Elcrhysioog Labs, Department of Medcine and th PhRmcooca and Physioli Scrieces and Cmmittee on

Cell Physioogy, The Universiy of Chicag, Chicag, linois 60637 USA

ABSTRACT The rat brain la (Brila) Na channel a-subunit and the brain 1 subunit were coexpressed in Xenopus oocytes,and peak whole-oocyte Na current (I) was measured at a test potential of -10 mV. Hyperpolarization of the holding potentialresulted in an increased affinity of STX and TTX rested-state block of Brila Na channels. The apparent half-block concentration(EDJ) for STX of Brila current decreased with hyperpolarizing holding potentials (Vhm). At Vhw of -100 mV, the ED50 was2.1 + 0.4 nM, and the affinity increased to a ED5,of 1.2 0.2 nM with Vhm of -140 mV. In the absence of toxin, the peakcurrentamplitude was the same for all potentals negative to -90 mV, deonsstrating that all of the channels were in a closed con-formation and maximally available to open in this range of holding potentals. The Woodhull model (1973) was used to describethe increase of the STX ED5 as a funcbon of holding potential. The equivalent elctrcl distance of blck (8) by STX was 0.18from the exracellular milieu when the valence of STX was fixed to +2. Analysis of the hoding potential dependence of TTXblck yiekled a similar 8 when the valence ofTTX was fixed to + 1. We conclude that the guanidinium toxin site is located partiallywithin the bansmembrane elkctric field. Previous site-directed mutagenesis shtdies demonstrated that an isoforn-specificphenylalanine in the BrIla channel is crical for high affinity toxin block. Therefore, we propose that amino acids at positionscorresponding to this Phe in the Brila channel, which lie in the outer vestibule of the channel adjacent to the pore entance,are partially in the transmembrane potential drop.

INTRODUCTION

Mutation studies of the Na channel a-subunit have identifiedresidues that are essential for guanidinium toxin block (Nodaet al., 1989). Mutations in the SS2 segment of the S5-S6loops (Guy and Conti, 1990) of the four Na channel repeatregions also alter single channel conductance (Terlau et al.,1991), confirming that the toxins block in or near the chan-nel's ion conduction path. Na channel isoforms also showmarked different sensitivity to toxin block resulting fromsequence differences in the toxin binding region of the ionconduction path (Satin et al., 1992a; Heinemann et al., 1992;Backx et al., 1992).

Divalent cations block within the open channel, and theyalso compete with toxin binding to the channel, with isoform-specific differences in the blocking affinity and competitiveefficacy (Schild and Moczydlowski, 1991; Doyle et al.,1993). Mutations affecting toxin affinity simultaneouslychange the divalent ion blocking affinity (Pusch et al, 1991;Heinemann et al., 1992; Backx et al., 1992), implying thatthe divalent ion-blocking site and the guanidinium-bindingsite overlap each other. Analysis of the voltage dependenceof Ca, Cd, and Zn block has revealed binding sites thatare 15-20o into the membrane field (Yamamoto et al.,

Receivedforpublication 6 December 1993 and infinalform 10 June 1994.Address reprint requests to Jonathan Satin, Ph.D., Cardiac Electophys-iology Labs., MC-6094, University of Chicago, 5841 S. Maryland Ave.,Chicago, IL 60637. Tel.: 312-702-4447; Fax: 312-702-6789, E-mail:satihs.bs.uchago.edu© 1994 by the Biophysical Society0006-3495,94A09/1007/08 $2.00

1984;Nilius, 1988; Sheets et al., 1987; Backx et al., 1992).If this divalent cation blocking site and the toxin site overlap,then the toxin binding site should be within the trans-membrane potential drop (Doyle et al., 1993; Backx et al.,1992).

Initial studies of guanidinium toxin block of BTX-treatedbrain Na channels in artificial lipid bilayers showed sub-stantial voltage dependence, as if the toxins were bound50-75% through the transmembrane potential gradient(French et al., 1984; Moczydlowski et al., 1984; Green et al.,1987a). TIhis voltage dependence has been attnbuted toconformation-dependent differences in binding affinityrather than to a binding site within the transmembranepotential drop, because the voltage dependence is the samefor the divalent STX and for the monovalent ITX (Worleyand Krueger, 1984; Rando and Strichartz, 1986). Theseconformation-dependent differences in toxin affinity (Cohenet al., 1981; Salgado et al., 1986; LUnnendonker, 1989) com-plicate the measurement of the equivalent electrical distanceof toxin block.We have reexamined the question of voltage-dependence

of TTX and STX block by studying the effects of hyperpo-larization on rested-state block in cloned rat brain and cardiacNa channels in a voltage range where conformationalchanges are limited. These channels expressed in Xenopusoocytes have a shift of kinetic parameters in the depolarizingdirection relative to native channels, so that the studies couldbe performed without significant binding to the open or theinactivated states of the channel. We found that blockingefficacies for both toxins were increased by hyperpolariza-tion. The voltage dependence of the divalent STX was

1 007

Volume 67 September 1994

greater than that of monovalent TTX, but when we adjustedfor the difference in valence, both toxins appeared to blockapproximately 15% into the trnsmembrane potentialgradient. This electrical distance is similar to that of divalention block, supporting the conclusion that both toxins anddivalent ions are affected by the transmembrane potentialdrop and block at overlapping sites.

MATERIALS AND METHODSXemopus oocyes were injeced with cRNA and clamped with the two-eletrode method as previously descrbed (Satin et aL, 1992b). Oocytesweremjected with 10-50 ng of cRNA. One to ten days after injection, oocyteswere impaled and whole-oocyte currents were recoded in a flowing bathsoluion cntaining (in mM): 90 NaCl, 25 KC1, 1 CaCl2, 1 MgQC10 HEPES (pH 7.6). The bath volume was less than 500 p1L Oocytes ex-pressing >5 ;A of4, were generally not useful because V. was not stableat the time of the peak current. A two electrode voltage damp/bath clamp(Dagan model CA-1) was used to measure Na currents in oocytes. To im-prove clamp speed, electrode resistances were usually less than 1 Mfl (rangefrom 0 2 to 1 MU). Capacitive transients were not subtracted because thelarge membrane area ofthe oocyte (diameter -1 mm) resulted in a tansientthat saturated the r system. Leak currents were not subtated be-cause at -10 mV the leak was negligible compared with the expressedcurrent We never detected an endogenous oocyte Na currenL

Both V. and current were recorded. V was measured as the differencebetween a voltage-recording electrode in the bath and the voltage-sensingelectrode impaled in the oocyte. Only oocytes clamped to the commandpotential at the time of peak current were used for data analysis. Initialtoxin-free and washout controls were performed on most oocytes for thedose response curves. Only stable oocyte impalements were used fordata analysis. Between the initial toxin-free recording and final washoutthe holding current at V,., = -100 mV increased by an average of -13+ 42 nA (n = 11). Peak current levels in washout controls for a Vof -10 mV ranged from 91 to 104% of the initial controls. Dose re-sponse curves for individual cells were fitted, and the ED50 values arereported as the mean + SD. Significant differences were tested usinga two-sample t-test.

TIhe brain 131 subumit was cloned from a rat brain cDNA library usingPCR. A 28 base forward rime was synthesized correspondig to base pair201-228 (Isom et al, 1992; 5'-ac _ _g -3'). Furterdetails are presented in Satin et aL (1994). The DNA sequence exactlycorresponded to that reported by Isom et aL (1992).

RESULTS

Fig. 1 shows Na currents elicited by a voltage step from aholding potential (V,.,) of -140 and -100 mV to a testpotential of -10 mV in oocytes expressing BrIla + ,B1.Changing V, from -100 to -140 mV had no effect on theamplitude of the peak current in the absence of toxin (Fig.1, A and B). The final toxin washout current traces are su-perimposed on the initial toxin-free currents. Throughout thisstudy, we observed less than a 10% change in peak currentamplitude between the initial toxin-free control and the finaltoxin washout.At pH 7.6, STX has a net valence of +2. If the block of

Na channels by STX occurs within the transmembrane-voltage drop, then block should occur with higher affinity athyperpolarized potentials. In the presence of3 nM STX, peakIN, during the test pulse is reduced after a V,, of -140 mVcompared with a V,., of-100 mV (Fig. 1, C and D). T-he

A Vh -l10nmV B Vh -140knV

ntra

4 ms

C D

3nMSIX

-10mV

-1O0mV-140mV

FIGURE 1 Effect of holding potential on SIX block of Na cuents ex-pressed firom Bria + 31 mXenopus oocytes, Na curent (JN.)were recordeduring steps to -10 mV after holding potentials (V,) of either -100 mV(A, C) or -140mV (B, D). (A, B) Currenttraces for initial toxin-free controland washout of tomn are superimposed. In the absence of toxi, the peakanmpltude of current was the same for all V. values m range from -140to -100 mV. (C, D) 3 nM STX blocks less current after a V., of -100mV (C; 67% block) than after -140 mV (D, 81% block). V., was main-rained for at least 90 s. The dashed line is the zero current leveL Cellz30322B.

block by STX shown in Fig. 1 represents tonic block, definedas that which occurs after infrequent stimulation. In theseexperiments, V,., was maintained for at least 90 s in thepresence of STX. Dose response curves for Brlla + 131and RHI from V},, -140, -120, and -100 mV are pre-sented in Fig. 2, A and B. Peak current after 30 s at Vbodwere normalized to the peak current in the absence ofSTX. The fitted ED50 from a single site dose responsecurve is reduced as V,., is made more negative for bothBrIla + (31 and RHI.The dependence of EDM on Vbow can be descnrbed by a

transmembrane voltage effect on STX block. Fig. 3 is thefitted ED50 plotted as a function of Vi,. Tne line is a fit tothe Woodhull model (1973) for voltage-dependent block ofthe form:

ED50(V) = ED_o(0) * exp[8 * z * Vr,Id * (e/k)], (1)

where ED50(0) is the extapolated half-block concentration at0 mV, 8 is the equivalent electrical distance (from the out-side), z is the valence ofthe blocker, and e, k, and Thave theirusual meanings. The 8 values calculated for STX block ofBrIla + 131 (0.18 + 0.02, n = 5) and RIM (0.14 ± 0.05,n = 3) assuming a valence of +2 for STX are only slightlydifferent (p = 0.09). This suggests that the STX site for BrIIaand RHI channels is located in a similar position with respectto the electrical field. The ED -o(100 mV) for BrIla + 1(2.1 + 0.4 nM) is 51 times more sensitive to STX than thatfor RHI (107 + 9 nM).The simplest model for voltage-dependent block is that

blocking efficacy is affected because the charged blocker isinfluenced by the transmembrane potential drop. At the bulksolution pH of 7.6, STX is primarily +2 and should have

1 008 Bi~ Joia

STX Block of Cloned Na Chanels

0.5

BrHla+fl 1 -

0.5 -

0

0.1 1 10

[STq (nM)

RHI

0 J10 100 1000

[ST)q (r*A)FIGURE 2 STX blocks with a higher affinity at more hyperpoarized holding potentials. Dose response curves at Vjd-100, -120, and -140 mV forBrIla + 31 (A) and RHI (B). Indvidual oocytes were fitted to single site dose response curve of the form

I,4sDz mm7RoL = (1 + ED4STX)-1 (5)

where ED4,5 represents the half-bock a of I,, by SIX The solid line is the average of the fitted ED, vawues from idvdual cells. The fitted

ED., is reduced by hyperpolarizatio For BruIa + 11, the ED50 values are: 21 0.4, 1.4 0.2, and 12 ±02 nM (n = 5 oocytes) at -100, -120, and-140 mV, respectively; for RHI, the ED50 values are: 107 9, 91 8, and 69 9 nM (n = 3 oocytes) at -100, -120, and -140 mV, respectively.Squares, triangles, and circles represent V... = -100, -120, and -140 mV, respectively.

100

10

w

I~~~1-140 -131) -1Q0

RIH SIX8=0.14+0.05

BrIIa+B1, TITX8=0.16+0.06

BrIIa+B1, SIX8=0.18+0.02

Vhold (mV)

FIGURE 3 Analysis of voltage dependence of toxin block according tothe Woodhull method (-). Fted STX ED,50 (from Fig. 2) for BrUI + 1and RHI plotted as a function of V,. TTX ED^5 is also ploted as a functionof V., for BrIIa + 31. The vaknce of STX was fixed to +2 and TIX +1for the fit to the data Trnmembrane voltage-dependent block quires that

the calculated elctric distance (8) for + 1 TITX is the same as +2 SIX Thefitted 8 for BrUIa + f31 block by SIX was 0.18 0.02 and by TTX was

0.20 0.07. For BrIIa + ,B1 the 0 mV ED:50 was 8.9 ± 3.0 OM for SIXand 24 ± 8 nM for TIX For RH! = 0.14 ± 0.05 and ED4(0) = 333 ±103 nM The solid line is the mean of the fitted parameters from indiviul

oocytesq

more of an effect than TTX, which is +1 at this pH. Ifthe toxin block site is within the transmembrane potentialgradient, and if the increased block with the more nega-

tive V... is not caused by a conformational change, thenthe calculated equivalent electrical distance, 8, should besimilar for '1TX and STX. The 8 calculated for TTXblock, assuming a valence of +1, was 0.16 + 0.06 (n =6); this is not significantly different (t-test, p > 0.1) fromthe 8 ofSTX block (z = +2) of BrIla. This result suggeststhat the site for rested state block is within the trans-membrane potential drop.

The assumption that no significant conformationalchanges occur in the voltage range from Vh1d of -100 to-140 mV was examined by steady-state availability curves.

Tlhe availability curve for BrIla + f31 (Fig. 4) demonstratesthat after a 10 s conditioning potential step (V,) at -90mV channels are maximally available to open upon depo-larization in the absence of toxin (Fig. 4, squares). The mid-points of the curves in the presence of toxin are shifted about-5 mV in the hyperpolarizing direction. We can attnrbutethis shift in the voltage range of -90 to -40 mV to a main-tained inactivated state block of Na channels (Carmeliet,1987; Eickhorn et al., 1990) and we did not consider thiseffect further. Peak currents are normalized to V. = -90mV to illustrate this shift (Fig. 4 B). In the range of the fullyavailable limb of the curves in Fig. 4 (-140 to -90 mV),additional toxin block is obtained as the conditioning po-

tential is made more negative. The steady-state block curves

for control (squares), 3 nM STX (circles), and 30 nM TTX(triangles) are superimposed. 3 nM STX and 30 nM TFXcause greater than 50% block of current (Fig. 4 A). Negativeto -90 mV, channels are maxmally available to open, andthe fraction of blocked channels in this range can be de-scnrbed by

(2)

where y(V) is the fraction of blocked channels as a func-tion of potential, and ED,(V) is the concentration of toxinfor half-block of INa (Eq. 1). The product of valence andelectrical distance (zS, the effective valence) was twiceas large for STX (0.28 + 0.05; n = 7) as for TTX (0.13

0.04; n = 4), and the 8 values are not significantlydifferent. Assuming a valence of +2 for STX and + 1 forTTX, both toxins block at the same equivalent electricaldistance.

Although the I,, was fully available at potentials of -90mV and more negative, there are likely to be alterations inthe equilibrium between different closed states, the so called

0

0

-a

SaX..Nn et a]. 1 009

I

AV) = i + EDM(V) -1

[toxin]

Voume 67 Septnber 1994

-120 -100 -80

VaCd (mV)

FIGURE 4 SIX and TIX block at the same equivalent eectrical distance. Steady-state availabilty curves for control (U, ), 30 nM TIX (A, ---),and 3 nM STX (0, -) in the same cell are (A) Peak current amplitues are plotted as a function of the conditioning potntial (V.,).Negative to -90 mV channls are fully available to open in the absence of toxn. In the presence of toxm the lines negative to V.,, = -90 mV are thedata fit to Eq. 3 (see text) For this cell 8 = 0.14 and 0.13 for STX and TTX, and ED,(0 mV) = 4 and 16 nM for SIX and TITX, respectively. Positiveto -90 mV the solid line is aBo nn fit to an availabity curve of the form:

I,. = (1 + exp[(Vcondn-VvkD)1, (6)

where V, is the midpoint and k is the slope. V,2 was -64 mV in controL-66 in 3 nM SIX, and -68 in the presence of 30 nM 1TX k(=5) was unchanged

in the presence of toxin. (B) Peak cuent amplitue were normalized to the peak current at V.,& = -90 mV. The conditioning pulse duration was 10s. cell z31103c.

Cole-Moore shift (Cole and Moore, 1960; Taylor andBezanilla 1983). Therefore, we performed an additional testthat is necesa to prove that SIX block in the fully avail-able voltage range is caused by a tnsmembrane potentialeffect The SIX association rate is two orders of magnitudeslower than a diffsion limited process, and is on the orderof 107 M-1 s-' (Ulbricht, 1986; Guo et aL, 1987). For con-

centrations used for BrIla + B1, we predic that we need tomaintain a constant voltage for several seconds to reach a

new equilibrium for STX block. The raw current trs su-

perimposed in the left panel of Fig. 5 correspond to currentelicited by depolarizations from a lOs long V. of either-140 (A and D) or -100 mV (B and C). The 10 s V.-d-dependent block by 10 nM SIX is shown in taces C and D.The dashed line in the right panel of Fig. 5 (open tringles)

25pA2 ms

represents the current measured in the presence of 10 nMSTX after conditioning pulses of only 100 ms. Presumably,a conditioning potential step (Vcd) for 100 ms did not allowvolage-dependent changes in block to occur, whereas a 10s Vcd in the same cell resulted in additional block with a

8 = 0.17 (compare traces C and D). This suggests that ifa conformational change is responsible for changes intoxin block in the hyperpolarized range of potentials, itmust develop with time constants of several seconds. Thetime course of this additional block is more appropriateto the expected toxin binding rate than to state-dependentchanges.

Because toxin takes seconds to reach voltage-dependentsteady-state block levels the fraction of V1-dAependent blockshould be independent of V... We measred the firction of

-

c

0

." 0.5E0

zAJB0 J

-140O -120 -100 -80 -60 -40

VokageFIGURE 5 Steady-stat availabilit of BrIla + 131 in the absence and preence of 10 am STX (flannels are maximally available to open after a 10 s

conditoning step to potentials negative to -90 mV in the absence of toxin (l). (left) Curen traces in absence of toxm show no effect on Va,_ of -100or -140 mV (traces A and B, respectively). Traces C and D are recorded in the presence of 10 nM STX More current was present after 10 s at -100 (traceC) than at -140 mV (trace D) (right) For hyperpolarized conditioning potentials (<-90 mV) of 10-s duration, the peak current was progre ly lessin the presence of 10 nM SIX (0). This volage-dependent block of current was dependent on the duration of the conditioning potential (V.). The dashedline is drawn through the normalized current after a 100-ms V.._. Only a slight dminution ofpeak current is observed after a 100-ms V. at hyperpolaizedpotental In the presence of 10 nM SIX the slope, k was the same, but the VLa was shifted -6 mV relative to control. Peak current amplitude werenormalized to the peak curent at Vd = -90 mV. Cell z30119A.

40030

a0a,

4.5

0

-1

-2

40

t00

-400 a4)0

00

1O-a 'Aj0t: 0

00

N 03 -

0z

0-140

Voidn (mV)

10nVcd

-60

1010 Boyca Jotua

STX Block of Cloned Na Channels

block as a function of V,., using Vs,. ranging from -20 to+ 10 mV in the linear phase of the current-voltage curve. 90s was maintained between voltage steps to allow steady-stateconditions to develop. Varying V, had no effect on the de-gree of V,,,-dependent block by STX

Estimaton of the rates of dev and rehefof voltage-dependent block

Hyperpolariztion of the holding potential from -100 to-140 mV for 10 s (Fig. 4) or 90 s (Figs. 1 and 2) increasedthe affinity of tonic STX andTIX block. To estimate the rateof development of this extra block induced by negativepotentials, we held the membrane potential at -140 mVfor variable durations before stepping to a test pulse. Thisprotocol has no effect on the amplitude of toxin free current(Fig. 6 A). Fig. 6 B shows the time course of developmentof field effect block of BrIIa + (31 by 1 and 3 nM STXDevelopment of field effect block is fit with a singleexponential. The rate of development (from the singleexponential fit) of field effect block by STX increaseswith dose. If the rate of development ofblock is a functionof STX concentration, then it should be fit with a straightline (Fig. 6 C). Ifwe assume a first order reaction schemesuch that

STX + Channed ± STX Channel (3)

then

lTon = Koff + Koo * [STX]- (4)

The fit to this equation in Fig. 6 C yields a K. of 63-06M` s-t and a Kff of 0.24 s-' at -140 mV for field effectblock by STX (n = 15 cells pooled). These values aresimilar to previously reported values for STX associationand dissociation rates for neuronal Na channels (Ulbrichtet al., 1986; Satin et al., 1994).

DISCUSSION

The main conclusion of this study is that STX and TTXappear to block the Na channel at a site that is partially withinthe transmembrane potential gradient We favor this inter-pretation over the alternative of state-dependent changes inaffinity for the following reasons. 1) The voltage-dependentchange in blocking efficacy occurs over a voltage rangewhere the channel is fully available, avoiding the well dem-onstrated changes in affinity to the inactivated state (e.g.,Cohen et al., 1981). 2) The time course of development andrelief of this voltage-dependent toxin block is slow and issimilar to rate constants measured with radiolabeled toxins,rather than those more rapid rates expected from gatingchanges. 3) The effective valence (zS) ofSIX is about twiceas large as that of TTX 4) The extent of voltage dependenceis similar to that of cations that are thought to block withinthe channel's transmembrane potential gradient and thatcompete with radiolabeled toxin for binding.

Voltage- versus state dependence of toxin block

The effect of membrane electric potential drop on guani-dinium toxin block has been difficult to demonstrate becausevoltage-dependent gating states are associated with differentapparent affinities. These state related types ofblock include:1) rested state, or tonic block; 2) maintained inactivated stateblock (possibly more than one conformation); and 3) acti-vated state block, also referred to as post-repolarization pha-sic block (Maielski et al., 1993). In this study, we measuredvoltage-dependent block of rested, or closed state channels.We emphasize that the voltage-dependent block measured inthis study is different from use-dependent block (indirectlyvoltage-dependent) measured after depolarizations that aresufficiently long to accumulate inactivated state blockedchannels (Carmeliet, 1987; Cohen et al., 1981; Vassilevet al., 1986). The third type of block, which is use-dependentbut does not involve inactivated channels, was more recently

Bhk (s)

0 10 20 30n.,

AJ

O.CZAL2ms

u

-100

- -200c -30 -400

=, -5000-700-700

-140mVintwv(s)

IlflM STX

C

0

:

12-1-

0.8.0.6- a

0.40.2 '.

0 2 4 6 8 10[STXJ (nM)

FIGURE 6 The rate of devek)pment of STX block at -140 mV. (A) Current traces for toxin-free control from V^,, -100, stepped to -140 mV for avariable duratio, and to a Vr.. of -10 mV. Largest amplitude currents show no dependence on the duration at -140 mV in the absence of toxin Currenttraces recorded after a 100 ms and 30 s long hyperpolarizatio to -140 mV. In the presence of 1 nM STX, more current is measured after a 100-mshyperpolarization to -140 mV ( * ) than for a 30 s long hyperpolarization (polygon). (B) Representative peak current amplitdes plotted as a function ofthe time interval avablable for transmembrane voltage drop block to develop at -140 mV. Solid lines are single exponential fits for this cell (cell z30428D).(C) The rate of development of block (fitted paameter from panel A) vs, [STXI is fit with a staight where the slope represents the assoc n rate constant(KR), and they intercept is the dissociation rate (Kff). The pooled fitted data yielded a K. = 63 X 10' M-' s-' and Kff = 0.24 s-' (19 values obainedfrom 15 cells).

Satin et a]. 1011

Vduffne 67 September 1994

discovered in crayfish axons (Salgado et al., 1986) and waslater descnrbed in native cardiac (MakieLsid et al., 1993) andneuronal preparations (Lonnendonker, 1989), and in BrIla(Patton and Goldin, 1991) and RHI expressed in oocytes(Satin et al., 1994). ibis latter type of block is elicited afterbrief depolarizations and is observed after relatively longsojouns in a tanmiently available state that occurs duringrecovery from inactivation.

Ifthe guanidinium moiety ofTIX orSTX binds within thepore, then it was reasoned that the 1TX/STX binding sitemight be within the transmembrane electric field (Hille,1975; reviewed by Hille, 1992). Subsequent studies ofbatrachotoxin-treated Na channels showed, however, that as-sociation and dissociation rates of guanidinium toxin blockwas not correlated with the valence of toxin (French et al,1984; Moczydlowski et al., 1984). This led to the conclusionthat the voltage dependence of block was indirect, and wasmediated by a voltage dependent conformational change ofthe channel. In the range ofpotentials tested in these previousstudies, we too observe an apparent increase in block affinity.Specifically, for mained depolarized potentials we ob-serve an appant parallel shift of the availability curve to thehyperpol g direction in the presence of toxin (Figs. 4and 5). We interpret this to be caused by a higher affinity,inactivated state block of the channel, or a slowing of thekinetics of recovery from this inactivated state (Eickhornet al, 1990; Carmeliet, 1987). By the use of maintained hy-perpolarized potentials (with only brief, infquent testdepolarizations) and coexpression of the (31 bunit withBrIla, we can minimize entrance of channels into a main-tained inactivated state. Peak Na current in whole-oocyteclamp occurs in about 2 nis, so that the degree of blockin these experiments reflects only block from toxin bindingat V,ad and is unrelated to the potential used to elicit Nacurrent.

This ehanced blocking efficacy with hyperpolarization hasbeen described before. Eickhorn and colleagues (1990) alsoobserved TTX block induced by extreme hyperpolarized po-tentials in cardiac myocytes, but they were unable to quantifytheir data in this range because of inability to control thevoltage in cells at very negative potentials. Channels ex-pressed in oocytes provided advantages over native cellpreparations for studying toxin block. In comparison withnative channels, the general property of the depolarized shiftof the voltage dependence of cloned channels expressed inoocytes (Satin et al., 1992b) allowed us to maintain holdingpotentials as much as 100mV negative to the midpoint of theavailability curve. Given the relatively slow rates of changeof block, the increased block at more negative V., valuesappears to be only a function of V... At such negative po-tentials, we assume that no conformational tansitions of thechannel occur that can affect toxin block. Therefore, we con-clude that the ieased affinity for STX and '1TX block withhyperpolarization is most consistent with a blocking sitewithin the transmembrane potential drop.

Implications of toxin voltage depen fordwi e st nctr

The well known competition between protons and divalentions and guanidinium toxins for binding to the Na channelsuggests overlapping sites of interaction (Doyle et al., 1993).Because the proton block (Woodhull, 1973; Daumas andAndsen, 1993) and divalent ion block (Krueger et al., 1986;Green et al., 1987b; Ravindran et aL, 1991; Sheets andHanck, 1992) are about 20% into the tansmembrane voltagedrop of the open channel, it was reasonable to think that thetoxin biding site might also be within the transimembraneelectric fiekL Point mutations of the SS2 region of the S5-S6loop of rpeat I ultasly pduced changes in toxinbinding and in single channel conductance, identifying afour-amino-acid stretch as the toxin binding site and also aspart of the ion conduction path (Teflau et al., 1991). Isoformdifferences in that four-amino-acid stech influence thetoxin affinity and sultansly the affinity for voltage-dependent divalent ion block. Specifically, the cardiac iso-form has a cysteine in position 385 (using the BrIla num-bering system) and it has a low affinity for toxin but a highaffinity for Cd2' block. Ihe brain Ila isoform has a phen-ylalanine in position 385, and it has a high affinity for toxinand a low affinity for Cd2+ block. Swap of these residuesconverts these isoform characteristics (Satin et al., 1992a).The comparable experiment on the skeletal muscle isoformalso indicates that the toxin and divalent ion blocking sitesinclude a residue at the same position (Backx et al., 1992).If the divalent ion blocking site is 20% into the transmem-brane potential gradient, then the toxin site should also bewithin the potential gradient to a similar extent Exact agree-ment of the fraction of elecrical distance would not be ex-pected because the divalent cation and toxin sites are notidentical and because the large toxin molecule might not beconsidered equivalent to a point charge. Nevertheless, theseexperiments indicate that toxin binding is influenced byabout 15% of the electric field of the closed channel, pro-vided that the measurements are done in such a manner thatthe channels remain in the closed configuration during toxinequihibrium. A second caveat to assigning an equivalent elec-trical disance to a paticular residue is that we do not knowto what extent, if any, permeant ions can affect our electricaldistance ofblock measurements. Ifguanidinium toxins blockNa channels in a "pore-directed binding event" analogous tocharybdotoxin block of Shaker K channels (Goldstein andMiller, 1993), then we must also consider whether permeantions influence our measurement of S. If the Na and toxin sitesoverlap and Na competes with toxin binding, then hyper-polarization would also drive more Na ions into the chan-nel and we would expect a decreased apparent affinity oftoxin. This scenario would result in a small underestimateof the calculated electrical distance ofblock. The simplestexplanation, in the context of the mutagenesis data, is thatour electrical distance measurements approximate theextent of the transmembrane potential drop in the channel

1012 Bofts-4. Jmxnal

Satin et al. STX Block of Cloned Na Chamels 1013

at the location of the cysteine of RHi, or the phenylala-nine in the analogous position (in IS5-S6) of the Br2achannel.

T'he voltage dependence of block by protons and divalentions was measured when the channel was blocked in the openconformation. In contrast, the toxin voltage dependence wasthat of the closed channel. The similar frAction of the fieldsensed by the two types ofblocing reactions implies that thefield at this site is similar during channel gating from cosedto open configurations. An important implication ofthe simi-larity in elecial distance for both conducting and noncon-ducting states is that this site must be continuously accessiblefrom the outside in both closed and open channels. The com-mon toxin/divalent site is resticted only in the sense that apotential drop occurs between the site and the exraceliularmiieu. In conclusion, the contradiction of voltage depen-dence of divalent ion block and voltage independence oftoxin block is resolved by these studies, and the hypothsisthat the toxin blocking site is in the ion conducting path issupported.

We thank Drs, William reen Jonathan Makieli and Dorothy Hanck forhelpful discussion and reading of this manusai Dr. AL Goklin (UC-Irvine) kindly provided us with the Br2a clone.This work was supported by National hfitutes of Health grant P01-HL20592 and grants-n-aid form the American Heart Association Metro-pd- icago Affiliate.

REFERENCESBackx, P, D_ Yue, J. awrence, E. Maban, and G. Tomaselli 1992. Mo-

kcular kxalization of an ion-binding site within the pore of mammaliansodium channels Science. 257:248-251.

Cameliet, E. 1987. Voltage ndent block by tetrodotomn of the sodimchannel in rabbit cardiac Purkinje fies Biophyq J 51:109-114.

Cohen,C J, B. P. Bean, T. J. Colatsky, and R W. Tsen 1981. Tetrodotoxinblock of sodium chanels in rabb Purkije fibers: int is betweentoxin binding and channel gating J. Gex Physi)L 78383-411.

Cole, K. S, and J. W. Moore. 1960.Poum ion curnt in the squidgtaxon: dynamic chaacterisfcs BiopAys. J. 1:161-202.

Daumas, P, and 0. S. Andersen. 1993. Proton block of rat brain sodiumchannels: evidece for two proton binding sits and multipe occupancy.J. Gem PkysioL 10127-43.

Doyle, D. D., Y. G. Guo, S. L Lustig, J. Satin, R. B. Rogart, and H.A. Fozzard. 1993. Divalent cation competition with 3H-saxitoxinbinding to tetrodotoxin-resistant and -sensitive sodium channels. Atwo-site shuctural model of ion/toxin interaction. J. Gen- PhysioL101:153-182.

Eickon, R, J. Webich, D. Hornung, and 17. Antoni. 1990. Use dqendenceof sodim curt iibiti by tedotoxin in rat cardiac muscle: in-fluence of dcnnel state. PJ1gers Arch 416 39B-405.

French, Rl J, J. F. Workey, and B. K. Krueger. 1984. Voltage-dependentblock by saxton of sodium channels incorpoad into plnar lipidbilayers, Bphy&s J. 45.301-310.

Goldstein, S. A. N, and C. Miler. 1993. Mechanism ofdcybdotoxin blockof a votage-gated K channeL Biophys. J. 65:1613-1619.

Green, W. N., L B. Weiss, and 0. S. Andersen. 1987a. Barachotoxin-modified sodim channels m planar lipid bilayers: charactezaion ofsaxitoxin- and tetrodotoxin-induced channel closures J. Gen. PhysioL89:873-903.

Green, W. N., L B. Weiss, and 0. S. Andersen 1987b. Batachotoxin-modified sodium channels in planar lipid bilayers ion permatio andblock. J. Geim Physio 89:841-8Th

Guo, Z, A. Uchara, A. Ravindran, S. HL Bryant, S. Hall, and E.Moczydkwski 1987. Kinetic basis for insensitivity to tetrodooxin andsaxitoxinin sodium channels of canine heart and denervated rat skeletalmuscle. Biochenmisny 26:7346-7556

Guy, H. R, and F. Conti. 1990. PUIsuing the strucre d function ofvoltage-gated channels. TVVS. 13921-206.

Hei n, S. H, IL Terau, and K. Imdoo. 1992. Moblular basis forphuracological diffrecs,beween brin and cardiac sodium channels.P1ugers ArLch 422:91-92.

MHle, B. 1975. The receptor for tetrdontoin and saxitoxin. A strutnuralhypothesis. Bkqprys J. 15.615-619.

Hille, B. 1992. Ionic ianwls of Excitabe Membranes. Vols, 3-4. 2nd Ed.Siauer Assocates, Sunderand, MA. 607 pp.

Isomn, L L, K. S. Dejongh, D. E. Paton, B. F. X Reber, J. Offord, H.Chabonnau, K. Walsh,A L Gokdin, and W. A. CataIL 1992. Primnystructre and functional expresson of the betal-subunit of the rat brainsodium channeL Science 256:8394842.

Krueger, B. K, J. F. Wodey, and R. J. French. 1986. Block of sodiumchannels in planar lid bilayers by gI .1.. toxins and calium: arethe mechanism of voltage dependence the same? In Tetrodotoxin, Saxi-toxim, and the Moecular Bioogy of the Sodium Chanel, C Y. Kao andS. R. Levinson, editos New Yok Academy of Sciences, New Yodr257-268.

L6nnendonker, U. 1989. Use-dependent block of sodiumc els m frogmyelinated nerve by tetrodtoxin and saxitoxin at negative holding po-tentials. BiochimL Biophys Acta 985:153-160.

Makielski, J. C, J. Satin, and Z Fan. 1993. Pog-repolariction block ofardia sodinm chanels by o Biophy. J. 65:790-798.

Moczydowski, F, S. Hall, S. S. Garber, G. S. Stridcar, and C Mflle.1984. Volage-dependen blckade of muscle Na- dnmels by Guani-dinium toxins. J. Gen- PhysioL 84:687-704.

Nilius, B. 1988. Calm block of guinea- heart sodium channels withand without modifiato by the pprlindole DPI 201-106.J. PhysioL 399537-558.

Noda, M, H. Suzui, S. Numa, and W. Stfiuner. 1989. A sie pointmuttion confers tetodotoxin and saxitoxin on the sodiumchnlIL FEBS Let 259.213-216.

Paton, D. E, and A. L Goldin. 1991. Avo aepedent gaig aionindces use-dependent block by tecodoxin of rat IA sodium channelsexpressed in Xenopus oocytes Neuro. 7:637-647.

Rando, T. A, and G. R. Strir 1986. Sxoxin bloks batachotoxin-modified sodium channels in the node of Ranvier in a voltage-dependentmanner. Biophys& J. 49:785-794.

Ravindran, A, L Schld, and E. Moczydlowski 1991. Divalet cation se-lectivity for exrnal block ofvolage-dependent Na+ (mels pndogedby batrachoaoxin-n2+ induces discrete substates in cardiac Na+ chan-nels, J. Gen. PhysioL 97:89-115.

Salgado, V. L, J. Z. Yeh, and T. Narahashi. 1986. Use- and voltage-dependent block of the sodium channel by sa ox hi Tetrodotoxin,Saxitoxin and the Molecular Biology of the Sodium ChanneL VoL 479.C Y. Kao and S. R Lvinson, editors. The New York Academy ofSciences, New Yoark 84-95.

Satin, J, J. W. Kyle, Chen, P. Bell, L L Cribbs, H. A. Fozard, and R.B. Rogart 1992a. A mutant of TfX-resstant cardiac sodium channelswith TIX-sensitive propnte Scie. 256:1I2-1205.

Satin, J, J. W. Kyle, M Chen, R B. Rogart, and H. A. Fozzad. 1992b. Thecdoed cardiac sodium channel a-subunit expresse in Xenopus oocytesshow gating and blking pperties of native channels J. Membr. BioL130:11-22.

Satin, J. C., J. W. Kyle, R B. Rogart, H A. Fozzard, and J. C. Makielski.1994. Post-repolarization block of cloned sodium channels by saxi-toxin; the contribution of pore-region amino acids. Biophys. J. 66:1353-1363.

Scid, L, and E Moczydlowski 1991. Competitive binding interactionbween ZD2+ and saxitoxin in cardiac Na+ channels-evidence fora sulfiydryl group in the Zn2+ saxitoxin binding site. Biophys J.59:523-537.

Sheets, Nt F, and D. A. Hanck 1992. Mechanism ofexacelular divalentand trivalent block of the sodium current in canine cardiac Purkine cells,J. PhsL 454:299-32).

1014 Josnal Vokune 67 Seterer 1994

Sheets, P F, B. E. Scanley, D. A. Hanck, J. C. Makelski, and H. A.Fozard. 1987. Open sdium diane! prO of sin canine cardiaPurdnje eIs BDphy& J. 52:3-22.

Taylor, R. E, and F. Bezanilla 1983. Sodium and gating ent time shfftsresulting om danges in initia caditios J. Gmn PhysioL 81:773-784.

Tedau, H, S. H- Hein _ann, W. Stuhmer, PA Psch, F. Coti, K. Imoto,and S. Numa. 1991. Maping the site of block by od n saxi-toin of sodium dne1-IL FEBS Lem 293:93-9

URlidct, W, H. H. Wagnr, am J.SLyid r. 1986. Knetics of TIX-SIX block of sodiumn dcae hI Tetrodoloxin, x anm theMof-

ecular Biology of the Sodium iamnet NewYoikAcemyofSdmx&Vol 479. C. Y. Kao amd S. R. Levmon, editors. 68483.

Vasev, P. M, R W. Hadley, K. S Lce, and J. R Hme. 1986. Vola-dependent action of tctxKkfoxin in mammalian rdiNanyocytes ADLJ. PhysioL 251H4754-480.

Woodbull, A. Ko. 1973. Ionic blocage of sodium channels in nerve. J. GeLPhysioL 61:687-708.

Worey, J. F. L L L, and EL K. Kmeger. 1984. Volae-d dent block bysaxitoxin of sodium s i td intop lipid blayesB&ohys J. 45:301-310.

Yamamoko, D, J. Z Ych, and T. Naashi 1984. Voltage-depenent cal-bixy ofn43 4and et s e sodium dn

Biopys J. 45:337-344.