Conduction Properties of the Cloned Shaker K+ Channel

Biophysial Journal Volume 65 November 1993 89-96 Condution Properties of the Cloned Shaker K+ Channel 89 Lise Heginbotham and Roderik MaKinnon Department of Neurobiology, Harvard Medial Shool, Boston, Massahusetts 115 USA ABSTRACT The ondution properties of the loned Shaker K+ hannel were studied using eletrophysiologial tehniques. Single hannel ondutane inreases in a sublinear manner with symmetri inreases in K+ ativity, reahing saturation by.6 M K+. The Shaker K+ hannel is highly seletive among monovalent ations; under bi-ioni onditions, its seletivity sequene is K+ > Rb+ > NH+ > Cs+ > Na+, whereas, by relative ondutane in symmetri solutions, it is K+ > NH+ > Rb+ > Cs+. In Cs+ solutions, single hannel urrents were too small to be measured diretly, so nonstationary flutuation analysis was used to determine the unitary Cs+ ondutane. The single hannel ondutane displays an anomalous molefration effet in symmetri mixtures of K+ and NH, suggesting that the onduting pore is oupied by multiple ions simultaneously. INTRODUCTION Potassium hannels all share a set of highly harateristi open hannel properties. They are extremely seletive among monovalent ations. Although learly permeable to Rb+, NH+, and Tl+, they effetively exlude the smaller Na+ ion (Hille, 1973; Yellen, 1984). In addition, several independent lines of evidene suggest that K+ hannels have multi-ion pores. First, flux-ratio exponents greater than unity have been measured in a number of K+ hannels (Hodgkin and Keynes, 1955; Begenisih and DeWeer, 198; Vestergaard-Bogind et al., 1985). Seond, the affinities of several different K+ binding sites have been measured diretly in the largeondutane Ca"-ativated K+ hannel (Neyton and Miller, 1988a; Neyton and Miller, 1988b). And third, K+ hannels demonstrate the anomalous mole-fration effet typially assoiated with multiple oupany (Hagiwara et al., 1977; Eisenman et al., 1986). Although K+ hannels display very diverse gating mehanisms, these two permeation properties, high seletivity, and multiple oupany, appear nearly universal. The following experiments desribe several ion ondution properties of a loned Shaker K+ hannel. This hannel is highly seletive for K+ over other alkali metal ations: Na+ is not measurably permeant. In ontrast, the hannel learly onduts Cs+. In fat, beause of the high expression levels ahieved using heterologous expression, Cs+ permeation ould be studied in some detail. The Shaker hannel also displays a harateristi feature of multi-ion hannels: an anomalous mole-fration effet. These results illustrate that the ondution properties of the loned Shaker K+ hannel are harateristi of those traditionally found in other K+ hannels. Reeived for publiation 3 June 1993 and in final form 1 August 1993. Address reprint requests to Lise Heginbotham at the Department of Neurobiology, Harvard Medial Shool, Longwood Avenue, Boston, MA 115. C 1993 by the Biophysial Soiety 6-3495/93/11/89/8 $. METHODS Channel onstrut and expression The Shaker H4 hannel, with amino-terminal inativation removed, was used throughout these experiments. The onstrution of this hannel, in the pbluesript KS' vetor (Stratagene, La Jolla, CA), has been desribed previously (Hoshi et al., 199; Yellen et al., 1991). DNA was linearized with a HindIll (New England Biolabs, Beverly, MA) digest, and RNA transribed using T7 RNA polymerase (Promega Corp., Madison, WI). Xenopus ooytes (Xenopus One, Ann Arbor, MI) were injeted with varying amounts of RNA, and inubated at 18 C for a period ranging from 1 to 1 days prior to reording. Solutions A standard set of solutions was used in these experiments. The extraellular (pipette) solution ontained (in millimolar): 95 XCI (X = Na+, K+, Rb+, Cs+, NH+), 1.8 CaCl, 1. MgCl, and 1 HEPES, ph 7.1. The internal solution ontained 15 XCI (X= Na+, K+, Rb+, Cs+, NH ), 1. MgCl, 5. EGTA, and 1 4-(-hydroxyethyl)l-piperazineethanesulfoni aid (HEPES), ph 7.1. In eah ase, ph was adjusted to 7.1 with the hydroxide of the monovalent ation. In the ondutane-onentration and anomalous molefration experiments, KCI onentrations were inreased symmetrially. Other than the onentration of monovalent ation, the external solution was the same as above. In addition to KCI, the internal solution ontained (in millimolar) 1 N-methyl-D-gluamine, 5 EGTA, 1 MgCl, and 1 HEPES, ph 7.1. The anomalous mole-fration effet was measured at the single hannel level in symmetri solutions that ontained a total of 3 mm K+ and NH. Other than the onentration of monovalent ation, the solutions were the same as those used in ondutaneonentration experiments. All hemials were purhased from Sigma Chemial Co. (St. Louis, MO) and Fisher Sientifi (Pittsburgh, PA), with

9 Biophysial Journal Volume 65 November 1993 the exeption of CsCl, whih was from Fluka Chemial Co. (Ronkonkoma, NY). Eletrophysiologial reording Exept where speifially noted, all reordings were from exised, inside-out pathes. In most ases, the eletrode offset was trimmed to zero in the test solution prior to obtaining a seal. However, it was diffiult to obtain seals in the presene of high bath [K+] (above 3 mm), sine the ooytes shrank rather extensively. For these experiments only, most eletrodes were set to zero in bath solutions ontaining 3 mm KCl, and seals were obtained. High [K+] (: 5 mm) solutions were then washed into the bath after the path was exised. Beause of the large single hannel ondutanes of the Shaker hannel under these onditions, the reversal potential ould still be fairly aurately estimated. However, to serve as a ontrol at eah ion onentration, a single path was studied in whih the eletrode had been zeroed in the test solution. Even in ooytes expressing large amounts of urrent at normal K+ onentrations, there is a low probability of having a path with any gating Shaker hannels when the solutions ontained high [K+]. The probability of observing an ative hannel dereased as the K+ onentration inreased; by 1 M K+, only one path in approximately 4 had any open Shaker hannels. Also, it was diffiult to obtain more than one seal from ooytes whih had been exposed to K+ onentrations above 3 mm; therefore, in these experiments eah path was from a separate ooyte. Single hannel reordings were filtered at 1 khz and sampled at rates of.5 khz. Unless otherwise noted, marosopi urrent reordings were filtered at khz and sampled at 14-5 khz. In experiments using flutuation analysis, urrent reordings were filtered at 5 khz and sampled at 14 khz. Juntion potentials at the bath-eletrode surfae assoiated with solution hanges were insignifiant, sine the hloride onentration was onstant in the different bath solutions. All measurements were made at C. Data analysis When onstruting marosopii-vurves, a linear leak subtration was used prior to determining reversal potentials. The leak urrent measured between -1 and -11 mv was saled and subtrated from eah trae. In the urrent trae in Fig. 6 A, both stationary leak and apaitive urrent, measured during a pulse from -8 to -9 mv, were subtrated. Nonstationary flutuation analysis was performed muh as originally desribed by Sigworth (198). To alulate the variane at eah timepoint, eah trae was ompared to a mean of three sweeps (itself and two surrounding it), thereby minimizing the effets of drift in the marosopi urrent during the timeourse of the experiment. A single pulse protool was repeated 6 times, and the variane was determined on 58 of these sweeps. In the experiments with Cs', the RMS noise (due to both hannel gating and system noise) had a maximum of 4.6 pa, approahing the size of the A-D onverting bit unit (.48 pa for measurements at -1 and -15 mv). The slight error that this level of granularity will introdue is within the satter of our data, and will not affet our estimations of i and N. RESULTS Conentration dependene of single hannel ondutane This study begins by fousing on a very basi question: how is the hannel ondutane affeted by the K+ onentration? Single hannel urrents were reorded as K+ onentrations were varied symmetrially on both sides of the membrane (Fig. 1 A). These reordings show that, at a given voltage, the single hannel amplitude inreases as the ativity A. 43 mm K B. C. 6 mm K+ 65 mm K 1.15 M K._ u) Q a V (. - -4 4 1-5 4 3-1. -9-6-3 3 6 9..3.6.9 1. K + ativity, M 5li "I s 5 mse 1.15 M 65 mm 35 mm 6 mm 73 mm 43 mm FIGURE 1 The effet of K+ ativity on single hannel ondutane. (A) Single hannel reordings were made at +8 mv in symmetri K+ solutions. (B) Single hannel i-v relationships, in symmetri solutions of the indiated K+ ativities. Eah point represents the mean ± SE of three to six measurements from independent pathes. Curves were fit to the funtion i = a + bv + V (of no theoretial signifiane), to provide an estimate of the zero-voltage ondutane. (C) Zero-voltage ondutanes, determined from the urves shown in B, as a funtion of K+ ativity.

Heginbotham and MaKinnon Shaker Condution Properties 91 of K+ is raised up to.6 M K+ (Fig. 1 A). However, the single hannel amplitude is the same in.6 and 1. M K+. The single hannel i-v relationships show that there is a qualitatively similar result over the entire voltage range studied (Fig. 1 B). At eah K+ onentration, the ondutane was measured from the slope of the i-v urve at mv. Fig. 1 C shows how the ondutane varies as a funtion of K+ ativity (Fig. 1 C). Condutane inreases in a sublinear fashion over the lower K+ onentrations, as if the pore is beoming saturated. In fat, saturation is reahed by high K+ ativities: there is no appreiable differene between ondutanes measured in.6 and 1. M K+. The finding of ondutane saturation is onsistent with the idea that there is a saturable step in ion ondution. The shape of the ondutane-onentration urve suggests that this proess has an "affinity" of approximately 3 mm. This result is similar to what has been reported in a high ondutane Ca+-ativated K+ hannel (Eisenman et al., 1986). At low K+ onentrations (Fig. 1 B), the Shaker hannel displays a slight inward retifiation. Here we see that as the K+ onentration is raised, the i-v relationship beomes more linear. In some K+ hannels, inward retifiation results from blokade of the outward urrent by internal Mg+. A previous study has reported that this is not the ause of retifiation in the Shaker hannel over the voltage range studied here; the hannel still retifies in the absene of Mg+ (MaKinnon and Yellen, 199). Ioni seletivity This setion investigates the ability of the Shaker K+ hannel to disriminate among monovalent ations. Traditionally, two kinds of measurements have been used to haraterize ioni seletivity. The first is the permeability ratio, determined from the reversal potential when a test ion is plaed on one side of the membrane and a omparison ion (K+ in the experiments desribed here) on the other side. The seond measure of seletivity is the relative ondutane: omparing ondutane of different ions measured under symmetri onditions. Permeability ratios were measured using inside-out pathes. Channels were ativated by an initial pulse to mv, after whih the path was stepped to different test potentials (Fig. A). In the example shown, NH' (95 mm) was present outside the path, and K+ (15 mm) was present on the inside. Isohronal urrent-voltage (I-V) urves were generated by measuring the urrent as a funtion of the test potential (Fig. B). The reversal potential of -57 mv indiates that K+ is onsiderably more permeant than NH' under these onditions. Fig. 3 shows isohronal I-V relationships measured with K+ on the outside of the path and the indiated test ions on the inside. Outward urrents are learly visible with either K+, Rb+, NH, or Cs'; the hannel is permeable to eah of these ations. In ontrast, we were unable to measure sig- A. B. U1) LU -1 C: f* mv -1 mv _ Q1) :3-1 -5 1 mse FIGURE Example of marosopi reversal potential measurements. (A) Marosopi urrent reording from an inside-out path, with 15 K+ in the bath (internal), and 95 mm NH' in the pipette (external). Membrane urrent was reorded during the illustrated voltage protool. After an initial depolarizing pulse to mv, the membrane was lamped at different test potentials. (B) An isohronal I-Vurve was generated from the data shown ina. Current was measured at a time after the apaitive urrent had deayed (illustrated by the arrow in A), and was orreted for any linear membrane leak, as desribed in Methods. The reversal potential was determined as the point at whih straight lines between the measured values rossed the absissa. nifiant Na+ ondution. The small outward urrent visible in the Na+ I-V at extreme depolarized potentials is indistinguishable from bakground ooyte urrent. Based on the reversal potentials, the Shaker hannel displays a relative permeability sequene of K+ > Rb+ > NH' > Cs' >> Na+. This same permeability sequene has been found for several other K+ hannels, inluding those in the frog node of Ranvier (Hille, 1973), lymphoytes (Shapiro and Deoursey, 1991), and a Ca"-ativated K+ hannel (Blatz and Magleby, 1984). The relative permeability (PA/PB) of the two monovalent ations A and B is operationally defined by the relationship obtained from Nernst-Plank eletrodiffusion PA = [B] PB [A]; -(FV./RT) where A and B are on the inside and outside, respetively, (1)

9 Biophysial Journal Volume 65 November 1993-7 a) C -, 4 - -4.. -. -.4 a, ) 4 -- -4 + K -6-3 -3 3 6 Cs -.6 3 6 9 1 5 ~d 1 3 6 9 1 3 6 1-1. -. -.4 3 6 9 Na + 3 6 9 1 FIGURE 3 Isohronal I-V relationships measured under bi-ioni onditions. In eah ase, the pipette solution ontained 95 mm K+, while the bath solution ontained 15 mm of the indiated test ion. Reversal potentials under these onditions were: K+, - mv; Rb+, +5 mv; NH+, +5 mv; Cs', +59 mv; Na+, > +9 mv. Vrev is the reversal potential, and F, R, and T have their usual meanings (Goldman, 1943; Hodgkin and Katz, 1949). The permeability ratios of the different ions are ompared in Table 1. The Shaker hannel displays the same overall permeability sequene when the orientation of the ions is reversed, i.e., when the permeability of an external test ion is measured relative to internal K+. In fat, even the exat permeability ratios depend only slightly on orientation (Table 1). In order to ompare relative ondutanes, single hannel i-v relationships were measured in symmetri solutions ontaining a single permeant ion (Fig. 4 A). Both inward and outward urrents were learly disernible in either symmetri K+ or Rb+ (Fig. 4 B). However, in NH', hannel openings were extremely flikery at negative membrane potentials, making the amplitude of the open state impossible to determine. As illustrated in Table 1, the Shaker hannel shows a relative ondutane sequene of K+ > NH' > Rb+. We were not able to reord Cs' or Na+ urrent through single hannels under these onditions. The seletivity measured by ondutane ratios is idential to what has been reported previously in the Ca"-ativated K+ hannel (Eisenman et al., 1986). It is worth noting that seletivity of the hannel as determined by ondutane ratios is different than that determined by relative permeability. As these are two independent measures of seletivity, the differene is not unexpeted. Interestingly, the Ca"-ativated K+ hannel shows the same relationship between seletivity measured in these two ways (Eisenman et al., 1986) as does the Shaker K+ hannel. Cs+ ondution through a potassium hannel In Fig. 3 we saw that Cs' supported measurable marosopi urrent through a membrane path ontaining many Shaker A. +8 mv -8 mv B.. K NH4 Rb. 4 mse * K * NH4 v Rb 1 TABLE 1 Seletivity of the Shaker hannel PX/PK Ion (X) K+ K+gX/gK K+ 1. 1. 1. Rb+.66.88.5 NH+.9.16.75 Cs+.6.1.1 Na+ <.3 Reversal potentials were measured under bi-ioni onditions, with 95 and 15 mm monovalent ation in the external and internal solutions, respetively. The permeability of eah ion X+ relative to K+ was alulated aording to Eq. 1. Single hannel ondutanes (g) were determined from either diret single hannel reordings (K+, NH+, and Rb+, Fig. 4) or from flutuation analysis (Cs+, Fig. 6). - -9-6 -3 3 6 9 V. mv FIGURE 4 Single hannel ondutanes in symmetri solutions ontaining a single permeant ion. (A) Single hannel reordings made in either K+ or Rb+ showed lear openings and losings at both +8 and -8 mv. In ontrast, single hannel amplitude ould only be determined at positive potentials in NH+ solutions. (B) Single hannel i-v relationships for different ations. Eah point represents the mean ± SE for three to six measurements in separate pathes. Again, urves were fit to the funtion i = a + bv + V (of no theoretial signifiane) to provide an estimate of the zero-voltage ondutane. In eah ase, the pipette and bath solutions ontained 95 and 15 mm ation, respetively.

Heginbotham and MaKinnon Shaker Condution Properties 93 K+ hannels. The single hannel Cs' urrent is very small, but its amplitude an be determined by studying the noise produed by Cs' onduting hannels, using flutuation analysis of marosopi urrents. The marosopi urrent traes shown in Fig. 5 provide a qualitative indiation of how big the unitary Cs+ ondutane must be. Currents reorded in symmetri Cs' are ompared to those arried by K+ under similar onditions. The two pathes ontain roughly the same amount of total urrent. But there is one striking differene between the Cs' and K+ urrents: the traes display very different amounts of noise. In K+ solutions, random noise is learly visible on individual urrent sweeps, partiularly those reorded at the most extreme potentials. In ontrast, the reordings made in symmetri Cs+ have substantially less noise. Beause the marosopi urrents have similar amplitudes, the differene in noise suggests that the path in Cs+ ontains many more hannels than the path in K+, and that these hannels have a muh lower single hannel ondutane. Furthermore, pathes that displayed marosopi urrent in symmetri Cs+ ould only be obtained from ooytes that expressed immeasurably high levels of K+ urrent. To determine the ondutane of the Shaker hannel in symmetri Cs+, flutuation analysis was performed on marosopi urrents suh as those shown in Fig. 5. After depolarizing the membrane path to maximally ativate the hannels, the voltage was stepped to a test potential. Analysis was arried out on "tail" urrents during the hyperpolarizing pulse, sine the rate of deativation is slower than the rate of ativation. Thus it was possible to sample many points over the entire range of open probability, and at the same time make measurements at potentials far from the reversal potential. The data obtained from one suh experiment are illustrated in Fig. 6. In this ase, an ativating pulse to 3 mv was followed by a hyperpolarizing step to -1 mv. The average urrent, a mean of 58 individual traes, is shown (Fig. 6 A) along with the variane at eah time point (Fig. 6 B). Within the tail pulse, the variane is learly smallest at the end of the pulse, when the marosopi urrent is small (when P nears ). For a population of idential, independently gating A. Cs Iq 15 mse B. K ICZ-- hannels, the relationship between variane and mean urrent was given by Sigworth (198), a = i *I- () where I and an are the mean and variane of the marosopi urrent, N is the total number of hannels, and i is the single hannel urrent. Plotting variane as a funtion of mean urrent (Fig. 6 C), the single hannel urrent (i) is found to be 18 fa and the number of hannels (N) is about 16,. The high level of expression suggested by the above alulation (N = 16, hannels) is learly apparent upon areful inspetion of the urrent traes in Fig. 6A. The initial bump in the pulse to +3 mv is gating urrent; the linear apaitive urrent has been fully subtrated in this figure. The large ratio of gating urrent to ioni urrent illustrated here is ompatible with a previous estimate of gating harge in the Shaker hannel (Shoppa et al., 199). The estimate of a low ondutane, together with a large number of hannels in a single path, is onsistent with the lak of noise seen in the marosopi Cs+ urrent (Fig. 5). As a ontrol we also used flutuation analysis to estimate the single hannel ondutane in symmetri K+. Under these onditions, the variane of the marosopi urrent is muh greater than in symmetri Cs+ (Fig. 6 D). For this path, i is approximately 1.8 pa, with N = 38. This measurement gives a single hannel ondutane of 15 ps, similar to our diret measurement of single Shaker hannels in the same K+ solutions (see Fig. 1). Outward Cs+ ondution depends on the speies of external ion Cs' has long been used to blok K+ hannels in whole ell experiments (e.g., Sigworth, 198), yet we have seen that Cs+ arries urrent under the onditions desribed above. Does Cs' always arry urrent through the Shaker hannel, or is it impermeant under some onditions? We looked to see if the ioni omposition of the external solution affeted the outward Cs' urrents. The results of this experiment are illustrated in Fig. 7. Using outside-out pathes, urrent was eliited by a step to +5 mv (near the reversal potential), with subsequent tail potentials ranging from -4 to + 1 mv. With potassium in the external solution (Fig. 7A), both inward K+ urrents and outward Cs+ urrents are learly visible during the tail pulses. In the same path, outward Cs+ urrents are muh smaller when the external solution is swithed to Na+ (Fig. 7 B). This explains why Cs+ an be used to blok K+ urrents in whole ell experiments: in the presene of external Na+, there is little outward Cs' urrent. FIGURE 5 Marosopi urrent reordings from pathes in solutions ontaining either Cs' (A) or K+ (B). In both ases, urrents were generated in response to voltage steps from -5 to +4 mv, in 1-mV inrements, from an initial holding potential of -8 mv. The subsequent hyperpolarizing pulse is to -1 mv. These measurements were made with 15 mm monovalent ation in the bath, and 95 mm in the pipette. Anomalous mole fration effet One striking harateristi of some multi-ion hannels is the presene of an anomalous mole-fration effet. The Shaker hannel learly demonstrates this effet at the single hannel level. Single hannel urrents were reorded in symmetri

94 Biophysial Journal Volume 65 November 1993 A FIGURE 6 Stationary flutuation analysis of marosopi Cs' urrents. Mean (A) and variane (B) of the urrent from 58 reordings from a single path. Current was eliited by an initial step to +3 mv, from a holding potential of -8 mv; the subsequent test (hyperpolarizing) potential was -1 mv. In A and B, the dotted lines orrespond to urrent and variane, respetively. (C) Variane as a funtion of marosopi urrent, from the same experiment shown in A and B. The ontinuous urve is alulated from the relationship o = i*i - I1/N (see text), with i =.18 pa, and N = 1.6 X 15 hannels. A onstant omponent of noise (of 7.1 pa, the mean of the varianes measured between 17 and 4 ms in the tail step) has been subtrated. (D) Same as C, but of a path in symmetri K+ solutions. In this ase, the urve is based on i = 1.8 pa, and N = 38. These measurements were made with 15 mm monovalent ation in the bath, and 95 mm in the pipette. B C 4.L_ a - 1 D N a).l_ I6) 5s 5 ms,-i -I..5 1. 1.5..5 Current, na Current, na A. K out/ Cs in Na out/ Cs in ------ B. 1 mse FIGURE 7 Cs' urrents in the presene of different external ations. With 15 mm Cs' in the pipette, a single outside-out maropath was exposed to either 95 mm K+ (A) or 95 mm Na+ (B) in the bath. In both ases, the membrane was depolarized by a step to +5 mv (near the reversal in external K+) and then stepped to different tail potentials, ranging from -4 to + 1, in 1-mV inrements. These are raw urrent reordings: the urrent whih remains with external Na+ may be apaitive, leak, rapidlyinativating urrent through the Shaker hannel, or a ombination of these. solutions ontaining either 3 mm NH' (Fig. 8A, top), 3 mm K+ (bottom), or a 1:1 mixture of the two (middle). Current amplitude is nearly the same in either K+ or NH4 solutions; however, it is notieably redued in the mixture. As the mole-fration ofnh + in the solution is gradually inreased, the single hannel urrent goes through a minimum (Fig. 8 B). This anomalous effet is observed at all positive voltages, but it is more dramati at low membrane potentials than during extreme depolarizations (Fig. 8 C). In some K+ hannels, the reversal potential, measured under asymmetri ioni onditions, also exhibits an anomalous mole-fration effet (Hagiwara et al., 1977). However, this is not seen in the Shaker hannel. As the frations of K+ and NH' are varied on one side of the membrane, the marosopi reversal potential hanges in a stritly monotoni fashion (Fig. 8 D). Analogous experiments, varying the mole-frations of Cs' and K+ in the bath solutions, showed a qualitatively similar outome. DISCUSSION This work desribes some of the basi permeation properties of the Shaker K+ hannel. Working with a loned hannel made it possible to study both single hannel and marosopi urrents under idential onditions. In addition, beause the hannel an be expressed at very high levels, we have been able to study Cs' ondution through this K+ hannel in some detail. The ondutane of the Shaker hannel saturates at high [K+], similar to what is desribed in other K+ hannels (Coronado et al., 198; Eisenman et al., 1986). It is interesting that saturation of the Shaker hannel ours over the same K+ range as it does in the high-ondutane Ca`ativated K+ hannel, despite the nearly -fold differene in absolute ondutane between the hannels. Although similar interations may underlie this measured "affinity" in

Heginbotham and MaKinnon Shaker Condution Properties 95 A B FIGURE 8 Anomalous mole-fration effet in mixtures of NH' and K+. In the experiments shown in A-C, single hannel reordings were made in symmetri solutions ontaining mixtures of NH' and K+, where [NH4'] + [K+] was 3 mm. (A) Single hannel reordings, at + 1 mv, in solutions ontaining K+, NH', or a 1:1 mixture of the two ations. (B) Effet of mole-fration on single hannel urrent measured at 1 mv. (C) Single hannel i-v relationships in solutions with different mole-frations of NH'. The illustrated urves are fits of the,.1, and.3 fration K+ data to the equation i = av + bv (no theoretial signifiane). (D) Effet of mole-fration on marosopi reversal potential. Reversal potentials were determined with 15 mm K+ on the inside of a path, and the illustrated solutions on the outside ([NH+] + [K+] = 95 mm). In this ase, the urve is the predition of the Goldman- Hodgkin-Katz equations with PNH4/PK =.9. C 1% NH4 5% NH4, 5% K.5. 1.5 m 1.._r.5. 1% K 5 m3e -.5 4 6 8 1 fration K O 1. K *.8 K A.5 K O.3 K+ v. K* *.1 K+. K D._ 3 1 ' 1 ---I...4.6.8 1. > -3-64 fration K +...4.6.8 1. fration K + the two hannels, the determinants of absolute ondutane must differ. Despite other evidene that the Shaker K+ hannel ontains a multi-ion pore, the ondutane-onentration urve provides no suggestion of multiple oupany. The antiipated derease in ondutane at extremely high ion onentrations has been reported in a few hannels (Hladky and Hayden, 197; Villarroel et al., 1988): perhaps a similar derease would also be seen in the Shaker hannel at higher onentrations of K+. The Shaker hannel is highly seletive for K+ over other monovalent ations. Na+, with a radius of.95 A, does not arry measurable urrent through the hannel. Cations larger than K+ are permeable, although in general, the larger the ion, the less permeable it is. Cs' is the largest ion found to ondut urrent; Cs' urrent is visible at a marosopi level under both bi-ioni onditions and in symmetri Cs' solutions. The finding of Cs' permeation plaes a lower limit of 3.34 A (the diameter of a Cs' ion) on the diameter of the ondution pore. Our ability to measure Cs' urrents was a diret result of high levels of expression using heterologous expression. The Shaker hannel is not unique among K+ hannels in displaying Cs' permeability; other voltage-ativated K+ hannels have similar Cs' permeabilities under bi-ioni onditions (Shapiro and Deoursey, 1991; Reuter and Stevens, 198). In ontrast, Cs' permeation has not yet been reported in the high ondutane Ca+-ativated K+ hannel. In these hannels, Cs' urrents are not visible either under bi-ioni onditions (Yellen, 1984; Cehi et al., 1987) or in symmetri Cs' (Cehi et al., 1987) at a single hannel level; instead, Cs' appears to blok ondution. However, Cs' blokade is relieved at extreme membrane potentials, suggesting that Cs' does permeate these hannels, although poorly (Cehi et al., 1987). With the reent loning of the Ca"-ativated K+ hannels (Atkinson et al., 1991; Adelman et al., 199; Butler et al., 1993), it may soon be possible to measure Cs+ urrent through the high-ondutane Ca+ativated K+ hannels using marosopi urrent reordings. The Shaker hannel displays the permeability sequene K+ > Rb+ > NH+ > Cs+. The same seletivity is found in many other K+ hannels. Despite their differenes in gating properties, delayed retifiers (Hille, 1973; Wagoner and Oxford, 1987; Shapiro and Deoursey, 1991) and Ca+ativated K+ hannels (Yellen, 1984; Blatz and Magleby, 1984; Eisenman et al., 1986) are all similar to the Shaker hannel with respet to seletivity measured under bi-ioni onditions. It is interesting to ompare the seletivity of the Shaker hannel with that of the high ondutane Ca+-ativated K+ hannel (Table ). These two hannels are distantly related at a moleular level, and show extensive homology in one region that is thought to form part of the pore. These hannels are equally seletive when measured under bi-ioni onditions, and yet they have very different relative ondutanes grb+/gk+ and gnh4/gk+ (Table ). It is as if the determinants of bi-ioni seletivity, but not ondutane, have been onserved in these two hannels. The Shaker hannel displays a feature that is harateristi of multi-ion pores: an anomalous mole-fration effet. It also resembles many other K+ hannels in this regard.

96 Biophysial Journal Volume 65 November 1993 TABLE Comparison of the seletivities of the high-ondutane Ca+-ativated K+ hannel and the Shaker K+ hannel Ion (X) BK Shaker PXIPK T1+ 1.3 K+ 1. 1. Rb+.7.66 NH+.1.9 Cs+.6 gx/gk KS+ 1. 1. NH+.18.75 Rb+.7.5 Cs+.1 T1+.41 The seletivity of the Shaker K+, from Table 1, is ompared to that of the high-ondutane Ca"-ativated K+ hannel (BK: Yellen, 1984; Blatz and Magleby, 1984; Eisenman et al., 1986). Conentration-dependent permeability ratios (Wagoner and Oxford, 1987) have provided evidene that other voltageativated K+ hannels an have multi-ion pores. The most diret onfirmation of multiple oupany omes from determinations of the flux-ratio exponent in voltage-dependent K+ hannels from uttlefish (Hodgkin and Keynes, 1955) and squid giant axon (Begenisih and DeWeer, 198) and the Ca`-ativated K+ hannel from red ells (Vestergaard- Bogind et al., 1985). In addition, the affinities of multiple K+ binding sites have been measured diretly in the highondutane Ca`-ativated K+ hannel (Neyton and Miller, 1988a, 1988b). These experiments illustrate that the ondution properties of the Shaker K+ hannel are similar to those previously desribed in a wide variety of K+ hannels. The Shaker hannel is both highly seletive among monovalent ations and displays an anomolous mole-fration effet onsistent with it being a multi-ion hannel. We thank Tanya Abramson and Lili Barush for help with RNA preparation, Jaques Neyton for expert injetion of several ooytes, and Brue Bean and Fred Sigworth for helpful disussions on flutuation analysis. This work was supported by National Institutes of Health grant GM474. REFERENCES Adelman, J. P, K.-Z. Shen, M. P. Kavanaugh, R. A. Warren, Y.-N. Wu, A. Lagrutta, C. T. Bond, and R. A. North. 199. Calium-ativated potassium hannels expressed from loned omplementary DNAs. Neuron. 9:9-16. Atkinson, N. S., G. A. Robertson, and B. Ganetzky. 1991. A omponent of alium-ativated potassium hannels enoded by the Drosophila slo lous. Siene (Wash. DC). 53:551-555. Begenisih, T. B., and P. DeWeer. 198. Potassium flux ratio in voltagelamped squid giant axons. J. Gen. Physiol. 76:83-98. Blatz, A., and K. Magleby. 1984. Ion ondutane and seletivity of single alium-ativated potassium hannels in ultured rat musle. J. Gen. Physiol. 84:1-3. Butler, A., S. Tsunoda, D. P. MCobb, A. Wei, and L. Salkoff. 1993. mslo, a omplex mouse gene enoding "maxi" alium-ativated potassium hannels. Siene (Wash. DC). 61:1-4. Cehi, X., D. Wolff,. Alvarez, and R. Latorre. 1987. Mehanisms of Cs' blokade in a Ca+-ativated K+ hannel from smooth musle. Biophys. J. 5:77-716. Coronado, R., R. L. Rosenberg, and C. Miller. 198. Ioni seletivity, saturation, and blok in a K+-seletive hannel from saroplasmi retiulum. J. Gen. Physio. 76:45-453. Eisenman, G., R. Latorre, and C. Miller. 1986. Multi-ion ondution in the high-ondutane Ca+-ativated K+ hannel from skeletal musle. Biophys. J. 5:15-134. Goldman, D. E. 1943. Potential, impedane, and retifiation in membranes. J. Gen. Physio. 7:37-6. Hagiwara, S., S. Miyazaki, S. Krasne, and S. Ciani. 1977. Anomalous permeabilities of the egg ell membrane of a starfish in K+-Tl1 mixtures. J. Gen. Physiol. 7:69-81. Hille, B. 1973. Potassium hannels in myelinated nerve. Seletive permeability to small ations. J. Gen. Physiol. 61:669-686. Hladky, S. B., and D. A. Hayden. 197. Ion transfer aross lipid membranes in the presene of gramiidin A. I. Studies of the unit ondutane hannel. Biohim. Biophys. Ata. 74:94-31. Hodgkin, A. L., and B. Katz. 1949. The effet of Na ions on the eletrial ativity of the giant axon of the squid. J. Physiol. 18:37-77. Hodgkin, A. L., and R. D. Keynes. 1955. The potassium permeability of a giant nerve fibre. J. Physiol. 18:61-88. Hoshi, T., W. N. Zagotta, and R. W. Aldrih. 199. Biophysial and moleular mehanisms of Shaker potassium hannel inativation. Siene (Wash. DC). 5:533-538. MaKinnon, R., and G. Yellen. 199. Mutations affeting TEA blokade and ion permeation in voltage-ativated K+ hannels. Siene (Wash. DC). 5:76-79. Neyton, J., and C. Miller. 1988a. Potassium bloks barium permeation through alium-ativated potassium hannels. J. Gen. Physio. 9: 549-567. Neyton, J., and C. Miller. 1988b. Disrete Ba+ blok as a probe of ion oupany and pore struture in the high-ondutane Ca+-ativated K+ hannel. J. Gen. PhysioL 9:569-586. Reuter, H., and C. F. Stevens. 198. Ion ondutane and ion seletivity of potassium hannels in snail neurones. J. Membr. Bio. 57:13-118. Shoppa, N. E., K. MCormak, M. A. Tanouye, and F. J. Sigworth. 199. The size of gating harge in wild-type and mutant Shaker potassium hannels. Siene (Wash. DC). 55:171-1715. Shapiro, M. S., and T. E. Deoursey. 1991. Seletivity and gating of the type L potassium hannel in mouse lymphoytes. J. Gen. Physiol. 97: 17-15. Sigworth, F. J. 198. The variane of sodium urrent flutuations at the node of Ranvier. J. Physio!. 37:97-19. Vestergaard-Bogind, B., P. Stampe, and P. Christophersen. 1985. Single-file diffusion through the Ca+-ativated K' hannel of human red ells. J. Membr. Biol. 88:67-75. Villarroel, A.,. Alvarez, and G. Eisenman. 1988. A maximum in ondutane ours for the large Ca-ativated K hannel at high Rb onentration. Biophys. J. 53:59a. (Abstr.) Wagoner, K., and G. S. Oxford. 1987. Cation permeation through the voltage-dependent potassium hannel in the squid axon. J. Gen. Physio. 9:61-9. Yellen, G. 1984. Relief of Na+ blok of Ca+-ativated K' hannels by external ations. J. Gen. PhysioL 84:187-199. Yellen, G., M. Jurman, T. Abramson, and R. MaKinnon. 1991. Mutations affeting internal TEA blokade identify the probable pore-forming region of a K' hannel. Siene (Wash. DC). 51:939-941.