- Email: [email protected]
A peptide derived from the shaker B K+ channel produces short and long blocks of reconstituted Ca2+-dependent K+ channels
Neuron.
Vol. 9, 229-236, August,
1992, Copyright
0 1992 by Cell Press
A Peptide Derived from the S/&er B K+ Channel Produces Short and long Blocks of Reconstituted Ca*+-Dependent K+ Channels Christine D. Foster,* Sungkwon Chung,*+ William N. Zagotta,’ Richard W. Aldrich,* and Irwin B. Levitan* *Graduate Department of Biochemistry and Center for Complex Systems Brandeis University Waltham, Massachusetts 02254 *Howard Hughes Medical Institute and Department of Molecular and Cellular Stanford University School of Medicine Stanford, California 94305
Physiology
Summary A 20 amino acid synthetic peptide, corresponding to the amino-terminal region of the Shaker B (ShB) K’ channel and responsible for its fast inactivation, can block large conductance Ca2+-dependent K+ channels from rat brain and muscle. The ShB inactivation peptide produces two kinetically distinct blocking events in these channels. At lower concentrations, it produces short blocks, and at higher concentrations long-lived blocks also appear. The L7E mutant peptide produces only infrequent short blocks(no long-lived blocks) at a much higher concentration. Internal tetraethylammonium competes with the peptide for the short block, which is also relieved by K+ influx. These results suggest that the peptide induces the short block by binding within the pore of Ca2+-dependent K’ channels. The long block is not affected by increased K’ influx, indicating that the binding site mediating this block may be different from that involved in the short block. The short block of Caz+-dependent K+ channels and the inactivation of Shaker exhibit similar characteristics with respect to blocking affinity and open pore blockade. This suggests a conserved binding region for the peptide in the pore regions of these very different classes of K+ channel. Introduction The Ca2+-dependent K+ channels form one subclass of a large superfamily of functionally distinct K+channels. Ca’+-dependent K+ channels are involved in the control of resting membrane potential, action potential repolarization, repetitive firing, and possibly higher integrative functions. These channels are widely distributed in neurons (Maue and Dionne, 1987; Reinhart et al., 1989; Lancaster et al., 1991). They are of particular interest because they provide a link between the membrane potential and Ca2+, a major intracellular messenger. Furthermore, Ca2+-dependent K+channels can often be modulated, via protein +Present address: Department of Pharmacology, School of Medicine, Yale University, P. 0. Box 333, New Haven, Connecticut 06510.
phosphorylation or guanyl nucleotide-binding proteins (e.g., Ewald et al., 1985; Lechleiter et al., 1988; Toro et al., 1990; Reinhart et al., 1991; Chung et al., 1991; White et al., 1991). Recently a component of a putative Ca*+-dependent K+ channel from Drosophila has been cloned and sequenced (Atkinson etal., 1991). The sequence shows that the predicted transmembrane profile of this channel resembles in some respects that of other K+ channel polypeptides, including members of the Shaker channel family. The initial cloning, sequencing, and heterologous expression of the Shaker B (ShB) K+ channel (Temple et al., 1987; lverson et al., 1988; Pongs et al., 1988) have made this channel the prototype with which the structures of all other K+ channels are compared. To explore possible structural homologies between Ca2+-dependent K+ channels and the Shaker K+ channel family, we took advantage of the wealth of structural and functional information available on the ShB K+ channel. The ShB channel is voltage dependent and rapidly inactivating. The inactivation, which is central to the channel’s function, is mediated by a 20 amino acid region at the amino terminus (Hoshi et al., 1990). A synthetic peptide corresponding to this region, which we refer to here as the ShB inactivation peptide, blocks K’ flux in Shaker mutants that do not exhibit rapid inactivation (Zagotta et al., 1990). These findings support the “ball and chain” model of channel inactivation, first proposed by Armstrong and Bezanilla (1977) for Na”channels. We have used the inactivation peptide as a probe for resemblances between Shaker and Ca*‘-dependent K’ channels. We show here that the ShB inactivation peptide interacts with Ca*+-dependent K’ channels from brain and muscle to produce short and long blocks, apparently by binding to two distinct receptor sites on the cytoplasmic face of Ca*‘-dependent K+ channels. One of these sites appears to lie within the conductive pathway. Toro et al. (1992) have obtained similar results for short blocks of a coronary artery Ca2’-dependent K+ channel by the ShB peptide. Results Two Kinetically Distinct Blocks of Ca’+-Dependent K’ Channels Previous experiments have shown that two kinetically and pharmacologically distinct types of large conductance, Ca*+-dependent K+channels are present in our rat brain plasma membrane preparation (Reinhart et al., 1989). Only the rapidly gating, charybdotoxin (CTX)-sensitive (type 1) channels were used for the experiments presented here, but similar results have been obtained for the slower gating (type 2) channels and for channels from muscle t-tubule membranes (data not shown). Rat brain or muscle Ca2+-dependent K+ channels were reconstituted into planar lipid bi-
NeUKJll 230
L7E Peptide
Wild Type Peptide
The left panel shows block of a type 1 CaJ’dependent K’ channel from rat brain by wild-type ShB peptide. A single channel was inserted into a planar bilayer, and recordings were made in symmetrical 150 m M KCI at a holding voltage of +40 mV. Channel openings are upward in this and subsequent figures. The top trace is a control record with no added peptlde. The following traces show that the peptide causes the induction of two different blocking events. At low peptide concentrations, short blocks occur, followed at higher concentrations of peptide (bottom trace) by the appearance of longer-lived blocked events. The right panel shows theeffects of apeptidewithasingleaminoacid mutation (L7E). The toD trace shows a control rccording, and the following traces depict the effects of increasing concentrations ot the mutant peptide. Note that the L7E mutant peptide is much less effective than the wild-type peptide at producing short blocks. No long blocks are evident at these concentrations of L7E peptide.
3 uM
6 UM
6 PM
12 PM
12 PM
50 uM
50 PM
layers and clamped at positive potentials (+40 mV) in the presence of saturating levels (100 PM) of internal Ca*+. Under these conditions, they open and close on the millisecond time scale and maintain an open probability of approximately 0.95. A typical record of channel gating under these conditions is shown in the top traces of Figure 1. Note that the channel spends most of its time in the open state with the occasional brief closure. Application of the wild-type ShB peptide (Zagotta et al., 1990) to the intracellular side of Ca*+-dependent K+ channels causes the appearance of two kinetically distinct blocked events (Figure 1, left panel). At low concentrations of the peptide (3-12 PM), short-lived blocks are observed; at higher concentrations (50 PM in Figure I), the short blocks are still present and the appearance of longer-lived nonconducting intervals can be recognized clearly. These long-lived blocks occur at a much lower frequency than the short blocks caused by the peptide. A mutant peptide, in which the leucine residue in position 7 is changed to aglutamate (L7E mutant), produces blocking events only at much higher concentration (50 PM; Figure 1, right panel). Infrequent short blocks and no long-lived blocks are observed at the highest concentrations tested. This mutant peptide also fails to restore inactivation in a mutant K+channel in which inactivation has been disrupted (Zagotta et al., 1990) and fails to block the coronary artery Ca*+-dependent K’ channel (Toro et al., 1992). These findings suggest that the blocks induced by wild-type .ShB peptide involve its binding to a specific site or sites on Ca*+-dependent K’ channels. Kinetic Analysis of the Short and long Blocks Under there conditions of high open probability
Figure 1. ShB Peptide Produces Short and Long Blocks of Ca”-Dependent K+ Channels
and
in the absence of .ShB peptide, channel closed time distributions can be fitted well by two exponential timeconstantsof mean durations less than 5 ms, indicative of at least two distinct closed states (Figure 2A). Although there are some longer closings, they are so infrequent that no longer time constant is required to fit the closed time distribution under these control conditions. Following addition of 50 PM ShB peptide, two additional nonconducting states with time constants of 16 ms and 158 ms are required to fit the distribution of nonconducting times (Figure 2A). These correspond to the short and long blocks, respectively, observed in the single-channel records (Figure 1). Toro et al. (1992) report a mean blocked time of 26 ms for short blocks of the coronary artery channel, but have not analyzed longer blocked events. To analyze the long blocks in isolation, we excluded all nonconducting events shorter than 200 ms; an example of this is shown in Figure 2B. After exclusion of all events shorter than 200 ms, the distribution of the remaining blocked times can be fitted by a single-exponential time constant of duration 160 ms, identical to the long time constant derived from the complete data set (Figure 2A). We constructed a dose-response relationship showing the percent decrease in the open probability produced bytheShB peptide(Figure3A; long blockswere excluded in thisanalysis).A50% inhibitionoftheopen probability is achieved at a peptide concentration of 5 PM. Transformation of these data to a Hill plot results in a straight line with a slope of 0.98, suggesting that the short block produced by the peptide results from a bimolecular reaction between peptide and channel. This was confirmed by an examination of the rates for peptide association with and dissociation
ShB Peptide Blocks Cal+-Dependent
K’ Channels
231
0
Time Figure 2. Cumulative
100
600
400
200
200
300
(ms)
Dwell-Time
400
500
Time
Distributions
for Nonconducting
600
700
RflO
900
(ms)
States in the Absence
and Presence
of ShB Peptide
dwell times of the channel in nonconducting states are exponentially distributed. The control data (open circles) are best fitted with two exponentials (both shorter than 5 ms), indicative of two distinct closed states. In the presence of 50 W M ShB peptide (closed triangles), two additional nonconducting states with time constants of 16 ms (short blocks) and 158 ms (long bloc:ks) are observed. The control distribution is constructed from more than 2000 events, and the distribution in the presence of 50 PM peptide is constructed from more than 100,000 nonconducting events. Approximately 85% of these events were intrinsic closings, and the remainder were short (14%) and long (1%) blocks. P(t) is the probability that a given dwell time is greater than or equal to time t. (B) To focus on the dwell-time distribution for the long blocks, all nonconducting events shorter than 200 ms were dIscarded. The remaining events can be fitted with a single exponential (T = 160 ms) that is identical to the long time constant In (A). This distribution is constructed from a total of 420 events. The inset shows that transformation of the data to a semi-logarithmic plot gives a straight line with a slope equal to the time constant. (A) The
bimolecular reaction. The apparent dissociation constant (KJ for the short block, calculated from these rate constants, is 8 PM, similar to the value derived from the dose-response relationship. This is also similar to the affinity of the peptide for the ShB channel
from
the channel. As shown in Figure 38, the associareaction is linearlydependenton thepeptideconcentration (on rate of 0.00875 rns-l PM-‘), whereas the dissociation reaction is independent of peptide concentration (off rate of 0.07 ms-‘), as expected for a
tion
0
10
20
30
40
50
[Peptide] Figure 3. Characteristics
60
70
80
90
100
0
Block
40
[Peptide]
(#VI)
of the Short
70
Produced
60
60
1°C)
(/r.M)
by ShB Peptide
(A) The percent decrease in the channel open probability (P,) is expressed as a function of peptide concentration under the same conditions as those described in Figure 1. Each point represents the mean * SEM of 5 separate determinations on 5 different channels. Transformation of these data to a Hill plot gives a Hill coefficient of 0.98. (B) The rate for the association reaction (l/r,, open circles) of the short block increases linearly with peptide concentration, whereas the rate for the dissociation reaction (l/re, closed circles) is independent of peptide concentration. The apparent Kd of the peptide for the short block, calculated from these rate constants, is 8.2 PM.
Neuron 232
.
1 Peptide Figure 4. Kinetics
1 (j&l)
and Voltage
Dependence
llolding of the Long Blocks
Produced
potential,
tl)V
by ShB Peptide
Peptide on rates (open circles) and off rates (closed circles) for the long blocks were measured from single-channel blocked and burst intervals. Only events longer than 200 ms were included in this analysis. (A) The on rate is a linear function of peptide concentration, whereas the off rate is independent of peptide concentration. Each point represents the mean + SEM of 4 separate determinations on 4different channels. (B)Voltagedependence of the long block. A channel was observed with 150 m M symmetrical KCI in the presence of 50 PM ShB peptide, at the indicated applied voltages.
A
Control
50 FM Peptide
C
[TEA],
mM
50 PM Peptide and 60 mM TEA
Figure 5. Effect of Internal
TEA on the Short and Long Blocks
(A) A single channel was inserted into a lipid bilayer, and records were taken at +40 mV holding potential. The top trace shows a control record with no added blockers. The middle trace shows the effect of adding 50 PM ShB peptide to the internal side of the membrane. Note both long and short blocks. The bottom trace shows that 60 m M TEA, added to the internal chamber in the continued presence of peptide, competes with the short blocks. The quantitative underpinnings for this conclusion are given in (B). (6) Effects of TEA on mean unblocked time for the short blocks and on single-channel current. A single channel was observed in the presence of 50 NM peptide at the indicated concentrations of internally applied TEA. The ratio of the mean unblocked time measured in the presence of TEA to the mean unblocked time measured in the absence of TEA (open circles) and the ratio of the single-channel currents in the absence and presence of TEA (closed circles) are plotted. The fact that they exhibit an identical dependence on TEA concentration indicates that ShEi peptide competes with TEA for the site that mediates short blocks. (C) Long peptide blocks are still observed in the presence of TEA (see bottom trace of [A]). The mean duration of the long blocks, calculated by excluding all events shorter than 200 ms, is the same in the absence (closed circles) or presence (open circles) of TEA.
ShB Peptide Blocks Ca”-Dependent 233
K’ Channels
A
5 mM external K’ and 50 pM Peptide
0.01 I-, 1c.l 200 300 ,Llo 500 bW 100
50 tnM external K+ and 50 pM Peptide
Time
Figure 6. External
K’ Does
Not Influence
the Duration
of the Long Blocks
Induced
(ms)
by ShB Peptide
(A) A single channel was observed at a holding voltage of +30 mV. The solution on the internal side of the channel contained 150 m M K’, and that on the external side contained 5 m M K’. The top trace shows control behavior under these conditions. The middle trace shows the effect of addition of 50 PM ShB peptide to the internal side of the membrane. Note both long and short blocks. In the bottom trace the concentration of external K+ is increased from 5 m M to 50 m M in the continued presence of the peptide. Kinetic analysis as described in Figure 2 demonstrates that the time constant for the short blocks induced by the peptide is reduced by this treatment from 16 ms to I3 ms. (B) The time constant for the long blocks, measured as described in Figure 2, is the same at 20 m M (closed circles), 40 m M (open circles), and 60 m M (open triangles) external K+. The solid line is taken from the inset of Figure 2B (150 m M external K+i.
itself (R. D. Murrell-Lagnadoand R. W.Aldrich, unpublished data). The longer-lived nonconducting intervals are reminiscent of the type of block produced by CTX at the external mouth of Ca*+-dependent K+channels (Miller et al., 1985). In the case of CTX, Anderson et al. (1988) demonstrated that these blocked events represent the binding of a single CTX molecule with a channel in a simple bimolecular reaction. To evaluatewhether the ShB peptide produces long blocks in a similar fashion at the inside mouth of the channel, we measured the association and dissociation rates of the peptide. These were determined from a statistical analysis of the long-lived blocked times (nonconducting intervals longer than 200 ms) and burst times (combined open, closed, and short-blocked intervals). Figure 4A shows that the association rate for the peptide is linearly dependent on its concentration and that the dissociation rate remains constant, regardless of peptide concentration. This demonstrates that the long block produced by the peptide also involves its interaction with Ca*‘-dependent K+ channels in a bimolecular fashion. Thus each long-lived blocked event represents the binding of one molecule of peptide to one channel. The Kd of the peptide for the long-lived blocked events, calculated from these measured association and dissociation rates, is 275 PM in
this experiment, more than an order of magnitude higher than that for the short blocks. In other experiments the Kd for the long blocks ranged from 140 PM to 275 PM. Both the association and dissociation rates vary with voltage (Figure 4B), such that the long blocks are more frequent and of longer duration at depolarized voltages. The rate constants for the short blocks exhibit a similar dependence on voltage (data not shown; see Toro et al., 1992).
Internal Tetraethylammonium Short Blocks
Competes with the
Internal tetraethylammonium (TEA) is known to block the internal mouth of Ca*+-dependent K+channels, as well as other K+channels, by entering the K+conduction pathway and thus preventing current flow (Vergara et al., 1984; Villarroel et al., 1988). The blocking events produced by TEA are too brief to be resolved as individual closings, so TEA appears to reduce the amplitude of the single-channel current. This effect can be seen in Figure 5A by comparing the middle trace (no TEA present) with the bottom one (60 m M TEA). In addition, TEA added together with the ShB peptide reduces the frequency of the short blocks produced by the peptide. As shown in Figure 5B, the reduction in frequency of the short peptide blocks and the reduction in single-channel current show
NWKHl 234
identical dependences on TEA concentration, indicating that the two blockers bind in a mutually exclusive way to the channel. Although we did not have sufficient data to analyze the effects of TEA on the frequency of the long blocks, note that their durations remain unaffected by TEA (Figure 5C). Increased External K’ Destabilizes the Short but Not the Long Blocks Increasing the K+ concentration of the external soiution decreases the duration of the short block produced by 50 P M peptide, but does not affect the long block. Figure 6A shows single-channel records from a membrane with 150 m M K+and 50 P M peptideon the cytoplasmic side, at two external K+concentrations. At high extracellular K’, the short blocks produced by peptide are shorter, whereas the frequency and durations of the long blocks are unchanged. An analysis of the distribution of short blocked times shows that the mean blocked time induced by 50 t.rM peptide at low extracellular K+, 16 m s in this experiment, is reduced to 8 m s at high extracellular K’ (data not shown). In contrast, the distribution of the durations of long blocks is not affected by extracellular K’ over a wide range of concentrations (Figure 6B). Discussion The ShB peptide can produce two kinetically distinct blocking events in brain and muscle CaL+-dependent K+channels. Although both blocking events are bimolecular, the Kd for the long block is more than an order of magnitude higher than the Kd for the short block. The two blocking events can also be distinguished on the basis of competition with external K+. Taken together these data suggest that the long and short blocks may result from the binding of ShB peptide to two different receptor sites on Cal+-dependent K+ channels. In ShB it is known that internal TEA competes with the inactivation ball (Choi et al., 1991). Recently Demo and Yellen (1991) havedemonstrated that the inactivation ball can also bedisplaced by K+from theopposite side. Thus inactivation in Shaker has been proposed to occur by open channel pore blockade. W e tested whether the long and short blocks of Ca2+-dependent K’ channels by the inactivation peptide might occur by pore blockade. Internal TEA competes with the short block, consistent with the idea that the short blocks involve pore blockade. However, the fact that the two blockers compete does not necessarily demand that the two compounds bind to the same site. Rather they demonstrate that in the presence of internal TEA the peptide can no longer bind to the receptor producing the short block. The receptor site for TEA and the receptor site for the peptide could be physically close, and so binding of one automatically precludes binding of the other. Since the peptide has a net positive charge, it is also possible that the binding of TEA results in electrostatic repulsion that prevents
the peptide binding to its receptor. Finally, the binding of TEA could also induce some allosteric change in the channel protein that renders the channel incapable of binding peptide to produce the short blocks. W e found that increasing external K+ decreases the duration of the short blocks but leaves the long blocks unaffected. The easiest interpretation of this result is that K+ flowing through the channel displaces the peptide from the receptor normally producing the short blocks. The exact mechanism by which increasing the concentration of a permeant ion speeds the exit rate of a blocking particle on the other side is unknown. The mechanism proposed by MacKinnon and Miller (1988) for relief of external block of CTX by K’ could also be inferred here. It is possible that a K+-binding site located in the pore is positioned close to a positively charged residue on the peptide. When peptide is bound to the site producing short blocks, an electrostatic repulsion between K’and the peptide ensues, leading to an increase in the exit rate of the blocker. In any event, the short block demonstrates two characteristic features of pore blockade, competition with a known pore blocker and relief of block by raising the concentration of a permeant ion on the opposite side of the membrane. In contrast the long block is not relieved by the permeant ion. This may be because the long blocks involve peptide binding to the pore in an orientation such that it cannot be displaced by ion flux. Alternatively, the peptide may produce long blocks by binding to some other site that does not reside in the pore. Toro et al. (1992) have shown that the ShB peptrde produces short blocks of Cal+-dependent K’ channels from coronary artery. It is interesting that the conc-entration of peptide required to produce short blocks of these channels is 6- to IO-fold higher than that required for the brain and skeletal muscle channels we studied, suggesting that there may be variations among Cal’-dependent K’ channels in the structure of the binding site involved in short peptide block. The binding of the peptide to its receptor in at least some Shaker-like channels is thought to be an inherently voltage-dependent process (Ruppersberg et al., 1991). Such results are consistent with pore blockade. In our hands, both the short and long blocks of brain and muscle CaLi-dependent K+ channels are voltage dependent, as are the short blocks of the coronary artery channel (Toro et al., 1992). Accordingly, the receptor sites mediating both blocks must reside within the membrane, as they sense the potential field across the membrane. Thus it is somewhat surprising that, when the peptide binds to the site producing the long block, it cannot be displaced by permeant ion. It is possible that the receptor site producing the long block is remote from the conduction pathway, yet still within the membrane. Binding to this site could produce the long blocking events via an allosteric effect on the channel protein. According to the model of inactivation put forward by Aldrich and coworkers (Hoshi et al., 1990; Zagotta
ShB Peptide Blocks Ca”-Dependent 235
K+ Channels
et al., 1990), the ShB channel possesses a specific binding site for the cytoplasmic peptide domain that induces inactivation. Site-directed mutagenesis experiments have been used to probe for the location of the receptor site within the Shakerchannel, and the S4-S5 cytoplasmic linker, which may lie near the channel’s permeation pathway, has been identified as a region that might be part of a receptor for the inactivation gate (Isacoff et al., 1991). It is interesting to compare the sequence of this region with that of the S4-S5 linker in the recently identified component of a Ca2+dependent K+ channel from Drosophila (Atkinson et al., 1991). Although the overall sequence of the latter channel differsextensivelyfrom thoseof other cloned K+ channels, 5 out of 10 amino acids are conserved in the S4-S5 linker when compared with the sequence from ShB. Thus it is possible that this region may be involved in the actions of the inactivation peptide on Ca2’-dependent K+ channels. Experimental
Procedures
All experiments were carried out on single, high conductance Cal+-dependent K’ channels from either rat skeletal muscle or rat brain plasma membrane vesicles, which were inserted into planar phospholipid bilayers. Membrane preparations were exactly as described previously (Anderson et al., 1988; Reinhart et al., 1989). Planar Bilayers and Membrane Vesicle Fusion Planar lipid bilayers were formed from a mixture of I-palmitoyl, 2-oleoyl-phosphatidyl-ethanolamine and I-palmitoyl, 2-oleoylphosphatidylserine (3:l) in n-decane, as described previously (Reinhart et al., 1989). Membranes were routinely added to the cis chamber. The solution on this side consisted of 150 m M KCI, 1.05 m M CaC&, 1 m M ECTA, 1 m M M&I,, 7 m M KOH, and 10 m M HEPES (pH 7.2). The composition of the trans solution was 0.1 m M EGTA, 7 m M KOH, and 10 m M HEPES (pH 7.2). Ca2’dependent K’channels inserted in one direction, that iswith the Cal+-sensing side facing the c/s solution. Thus, this is defined as the cytoplasmic side of the channel. Following incorporation of a channel, further fusion was suppressed by making thetrans K’ concentration the same as the cis K’ concentration. This was the case for all experiments except those in which the effect of extracellular K’ on peptide block was determined. Voltages are expressed as the voltage of the cis side with respect to the trans stde of the bilayer. Electrical Recordings and Analysis Single-channel currents were amplified using a current-tovoltage converter with a 10 CD feedback resistor, filtered at l-4 kHz using an 8-pole Bessel filter, and recorded on a B-videocassette recorder after l&bit digitization at 44 kHz with a pulse code modulator (Sony PCM-701ES). Data were analyzed off-line using an MS-DOS-compatible computer (Everex 386, 20 MHz) equipped with a Tecmar labmaster AID board coupled to an lndec IBX Instrumentation Interface and utilizing an lndec BASIC-Fastlab software library. Using operator-assisted computer programs, current records filtered at 1 kHz were sampled at 5 kHz and searched for transitions away from the closed or open level.Thethreshold fordetectingopeningand closingtransitions was set to 50% of the open level current for each individual event. In this way current levels corresponding to each closed/open transition were reduced to three values correspondSuch reduced ing to the closed time, open time, and amplitude. lists were then further transformed to yield the open probability, mean amplitude, and open and closed timedistributions. Cumulative dwell-time distributions were fitted with exponentials using the [-evenberg-Marquardt curve fitting algorithm, and the
goodness of fit was evaluated by chi-square analysis. P(t) is the probability that a given dwell time is greater than or equal to time t. All experiments were repeated at least three times. ShB Peptide A peptide corresponding to the first 20 amino acids of ShB and another in which the leucine at position 7 was changed to glutamate (L7E mutant) were synthesized by the Protein and Nucleic Acid Facility at Stanford University. An amtde was placed at the carboxyl terminustoprevent itschargefrom influencing peptide activity, and the synthetic peptides were purified by reversed phase, high-pressure liquid chromatography prior to use (see Zagotta et al., 1990, for details). Acknowledgments This work was supported by NIH grants NS17910 to (I. B. L.) and NS23294 to (R. W. A.). R. W. A. is an Investigator of the Howard Hughes Medical Institute, and W. N. 2. IS a Warner Lambert Fellow of the Life Sciences Research Foundation. We aregrateful to Drs. Shimon Marom and Chris Miller for their critical comments on the manuscript and to Dr. Manuel Esguerra for help with the statistical analysis. We also thank Drs. L. Toro, R. Latorre, and E. Stefani for communicating results prior to publication. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC Section 1734 solely to indicate this fact. Received
March
12, 1992; revised
May 15, 1992
References Anderson, C. S., Mackinnon, R., Smith, C., and Miller, C. (1988). Charybdotoxin blockof singleCa’+-activated K’channels. Effects of channel gating, voltage, and ionic strength. J. Gen. Physiol. 97, 317-333. Armstrong, C. M., and Bezanilla, sodium channel. II. Cating current 70, 567-590.
F. (1977). Inactivation of the experiments. J. Gen. Physiol.
Atkinson, N. S., Robertson, G. A., and Canetzky, B. (1991). A component of calcium-activated potassium channels encoded by the Drosophilas/o locus. Science 253, 551-555. Choi, K. L., Aldrich, R. W., and Yellen, G. (1991). Tetraethylammonium blockade distinguishes two inactivation mechanisms in voltage-activated K’ channels. Proc. Natl. Acad. Sci. USA 88, 5092-5095. Chung, S., Reinhart, P. H., Martin, B., Brautigan, D., and Levitan, I. B. (1991). Protein kinase activitycloselyassoc-iated with a reconstituted calcium-activated potassium channel. Science 253, 560-562. Demo, S. D., and Yellen, G. (1991). The inactivation gate of the Shaker K+ channel behaves like an open-channel blocker. Neuron 7, 743-753. Ewald, D. A., Williams, A., and Levitan, I. B. 11985). Modulation nt K’channel activity by protein phosof single Ca”-depende phorylation. Nature 375, 503-506. Hoshi, T., Zagotta, W. N., and Aldrich, R. W. (1990). Biophysical and molecular mechanisms of Shaker potassium channel inacttvation. Science 250, 533-538. Isacoff, E. Y., Jan, Y. N., and Jan, L. Y. (199’1) Putative the cytoplasmic gate in the Shaker K’ channel. 86-90.
receptor for Nature 353,
Iverson, L. E., Tanouye, M. A., Lester, H. A., Davidson, N., and Rudy, B. (1988). Expression of A-type potasstum channels from Shaker cDNAs. Proc. Natl. Acad. Sci. USA 85, 5723.-5727. Lancaster, B., Nicoll, R. A., and Perkel, D. J. (1991). Calcium vates two types of potassium channels in rat hippocampal rons in culture. J. Neurosci. 11, 23-30. Lechleiter,
J. D., Dartt,
D. A., and Brehm,
acttneu-
P (1988) Vasoactive
NWOl
236
intestinal peptide activates Cal+-dependent K’channels through a CAMP pathway in mouse lacrimal cells. Neuron 1, 227-235. MacKinnon, R., and Miller, C. (1988). Mechanism of charybdotoxin block of the high-conductance, C&+-activated K’ channel. J. Cen. Physiol. 91, 335-349. Maue, R. A., and Dionne, V. E. (1987). Patch clamp studies of isolated mouse olfactory receptor neurons. J. Gen. Physiol. 90, 95-125. Miller, C., Moczydlowski, Charybdotoxin a protein channels from mammalian
E., Latorre, R., and Phillips, M. (1985). inhibitor of single Ca2+-activated K’ skeletal muscle. Nature 373,316-318.
Pongs, O., Kecskemethy, N., Muller, R., Krah-Jentgens, I., Baumann, A., Kiltz, H. H., Canal, I., Llamazares, S., and Ferrus, A. (1988). Shaker encodes a family of putative potassium channel proteins in the nervous system of Drosophila. EMBO, J. 7,10871096. Reinhart, P. H., Chung, S., and Levitan, I. B. (1989). A family of calciumdependent potassium channels from rat brain. Neuron 2, 1031-1041. Reinhart, P. H., Chung, S., Martin, B. L., Brautigan, D. L., and Levitan, I. B. (1991). Modulation of calcium-activated potassium channels from rat brain by protein kinaseAand phosphatase2A. J. Neurosci. II, 1627-1635. Tempel, B. L., Papazian, D. M., Schwarz, T. L., Jan, Y. N., and Jan, L. Y. (1987). Sequence of a probable potassium channel component encoded at Shaker locus of Drosophila. Science 237, 770775. Toro, L., Ramos-France, J., and Stefani, E. (1990). GTP-dependent regulation of myometrial Kc. channels incorporated into lipid bilayers. J. Cen. Physiol. 96, 373-394. Toro, L., Stefani, E., and Latorre, R. (1992). Internal blockade Cal+-activated K’channel by ShakerB inactivating”ball”peptide. Neuron, this issue.
of a
Vergara, C., Moczydlowski, E., and Latorre, R. (1984). Conduction incorporated blockade and gating in a Ca 2+-activated K’channel into planar lipid bilayers. Biophys. J. 45, 73-76. Villarroel, A., Alvarez, O., Oberhauser, A., and Latorre, R. (1988). Probing a Ca*+-activated K+ channel with quaternary ammonium ions. Pfltigers Arch. 473, 118-126. White, R. E., Schonbrunn, A., and Armstrong, D. L. (1991). Somatostatin stimulates Ca2+-activated K’ channels through protein dephosphorylation. Nature 357, 570-573. Zagotta, W. N., Hoshi, T., and Aldrich, R. W. (1990). Restoration of inactivation in mutants of Shaker potassium channels by a peptide derived from Shaker. Science 250, 568571.
Vol. 9, 229-236, August,
1992, Copyright
0 1992 by Cell Press
A Peptide Derived from the S/&er B K+ Channel Produces Short and long Blocks of Reconstituted Ca*+-Dependent K+ Channels Christine D. Foster,* Sungkwon Chung,*+ William N. Zagotta,’ Richard W. Aldrich,* and Irwin B. Levitan* *Graduate Department of Biochemistry and Center for Complex Systems Brandeis University Waltham, Massachusetts 02254 *Howard Hughes Medical Institute and Department of Molecular and Cellular Stanford University School of Medicine Stanford, California 94305
Physiology
Summary A 20 amino acid synthetic peptide, corresponding to the amino-terminal region of the Shaker B (ShB) K’ channel and responsible for its fast inactivation, can block large conductance Ca2+-dependent K+ channels from rat brain and muscle. The ShB inactivation peptide produces two kinetically distinct blocking events in these channels. At lower concentrations, it produces short blocks, and at higher concentrations long-lived blocks also appear. The L7E mutant peptide produces only infrequent short blocks(no long-lived blocks) at a much higher concentration. Internal tetraethylammonium competes with the peptide for the short block, which is also relieved by K+ influx. These results suggest that the peptide induces the short block by binding within the pore of Ca2+-dependent K’ channels. The long block is not affected by increased K’ influx, indicating that the binding site mediating this block may be different from that involved in the short block. The short block of Caz+-dependent K+ channels and the inactivation of Shaker exhibit similar characteristics with respect to blocking affinity and open pore blockade. This suggests a conserved binding region for the peptide in the pore regions of these very different classes of K+ channel. Introduction The Ca2+-dependent K+ channels form one subclass of a large superfamily of functionally distinct K+channels. Ca’+-dependent K+ channels are involved in the control of resting membrane potential, action potential repolarization, repetitive firing, and possibly higher integrative functions. These channels are widely distributed in neurons (Maue and Dionne, 1987; Reinhart et al., 1989; Lancaster et al., 1991). They are of particular interest because they provide a link between the membrane potential and Ca2+, a major intracellular messenger. Furthermore, Ca2+-dependent K+channels can often be modulated, via protein +Present address: Department of Pharmacology, School of Medicine, Yale University, P. 0. Box 333, New Haven, Connecticut 06510.
phosphorylation or guanyl nucleotide-binding proteins (e.g., Ewald et al., 1985; Lechleiter et al., 1988; Toro et al., 1990; Reinhart et al., 1991; Chung et al., 1991; White et al., 1991). Recently a component of a putative Ca*+-dependent K+ channel from Drosophila has been cloned and sequenced (Atkinson etal., 1991). The sequence shows that the predicted transmembrane profile of this channel resembles in some respects that of other K+ channel polypeptides, including members of the Shaker channel family. The initial cloning, sequencing, and heterologous expression of the Shaker B (ShB) K+ channel (Temple et al., 1987; lverson et al., 1988; Pongs et al., 1988) have made this channel the prototype with which the structures of all other K+ channels are compared. To explore possible structural homologies between Ca2+-dependent K+ channels and the Shaker K+ channel family, we took advantage of the wealth of structural and functional information available on the ShB K+ channel. The ShB channel is voltage dependent and rapidly inactivating. The inactivation, which is central to the channel’s function, is mediated by a 20 amino acid region at the amino terminus (Hoshi et al., 1990). A synthetic peptide corresponding to this region, which we refer to here as the ShB inactivation peptide, blocks K’ flux in Shaker mutants that do not exhibit rapid inactivation (Zagotta et al., 1990). These findings support the “ball and chain” model of channel inactivation, first proposed by Armstrong and Bezanilla (1977) for Na”channels. We have used the inactivation peptide as a probe for resemblances between Shaker and Ca*‘-dependent K’ channels. We show here that the ShB inactivation peptide interacts with Ca*+-dependent K’ channels from brain and muscle to produce short and long blocks, apparently by binding to two distinct receptor sites on the cytoplasmic face of Ca*‘-dependent K+ channels. One of these sites appears to lie within the conductive pathway. Toro et al. (1992) have obtained similar results for short blocks of a coronary artery Ca2’-dependent K+ channel by the ShB peptide. Results Two Kinetically Distinct Blocks of Ca’+-Dependent K’ Channels Previous experiments have shown that two kinetically and pharmacologically distinct types of large conductance, Ca*+-dependent K+channels are present in our rat brain plasma membrane preparation (Reinhart et al., 1989). Only the rapidly gating, charybdotoxin (CTX)-sensitive (type 1) channels were used for the experiments presented here, but similar results have been obtained for the slower gating (type 2) channels and for channels from muscle t-tubule membranes (data not shown). Rat brain or muscle Ca2+-dependent K+ channels were reconstituted into planar lipid bi-
NeUKJll 230
L7E Peptide
Wild Type Peptide
The left panel shows block of a type 1 CaJ’dependent K’ channel from rat brain by wild-type ShB peptide. A single channel was inserted into a planar bilayer, and recordings were made in symmetrical 150 m M KCI at a holding voltage of +40 mV. Channel openings are upward in this and subsequent figures. The top trace is a control record with no added peptlde. The following traces show that the peptide causes the induction of two different blocking events. At low peptide concentrations, short blocks occur, followed at higher concentrations of peptide (bottom trace) by the appearance of longer-lived blocked events. The right panel shows theeffects of apeptidewithasingleaminoacid mutation (L7E). The toD trace shows a control rccording, and the following traces depict the effects of increasing concentrations ot the mutant peptide. Note that the L7E mutant peptide is much less effective than the wild-type peptide at producing short blocks. No long blocks are evident at these concentrations of L7E peptide.
3 uM
6 UM
6 PM
12 PM
12 PM
50 uM
50 PM
layers and clamped at positive potentials (+40 mV) in the presence of saturating levels (100 PM) of internal Ca*+. Under these conditions, they open and close on the millisecond time scale and maintain an open probability of approximately 0.95. A typical record of channel gating under these conditions is shown in the top traces of Figure 1. Note that the channel spends most of its time in the open state with the occasional brief closure. Application of the wild-type ShB peptide (Zagotta et al., 1990) to the intracellular side of Ca*+-dependent K+ channels causes the appearance of two kinetically distinct blocked events (Figure 1, left panel). At low concentrations of the peptide (3-12 PM), short-lived blocks are observed; at higher concentrations (50 PM in Figure I), the short blocks are still present and the appearance of longer-lived nonconducting intervals can be recognized clearly. These long-lived blocks occur at a much lower frequency than the short blocks caused by the peptide. A mutant peptide, in which the leucine residue in position 7 is changed to aglutamate (L7E mutant), produces blocking events only at much higher concentration (50 PM; Figure 1, right panel). Infrequent short blocks and no long-lived blocks are observed at the highest concentrations tested. This mutant peptide also fails to restore inactivation in a mutant K+channel in which inactivation has been disrupted (Zagotta et al., 1990) and fails to block the coronary artery Ca*+-dependent K’ channel (Toro et al., 1992). These findings suggest that the blocks induced by wild-type .ShB peptide involve its binding to a specific site or sites on Ca*+-dependent K’ channels. Kinetic Analysis of the Short and long Blocks Under there conditions of high open probability
Figure 1. ShB Peptide Produces Short and Long Blocks of Ca”-Dependent K+ Channels
and
in the absence of .ShB peptide, channel closed time distributions can be fitted well by two exponential timeconstantsof mean durations less than 5 ms, indicative of at least two distinct closed states (Figure 2A). Although there are some longer closings, they are so infrequent that no longer time constant is required to fit the closed time distribution under these control conditions. Following addition of 50 PM ShB peptide, two additional nonconducting states with time constants of 16 ms and 158 ms are required to fit the distribution of nonconducting times (Figure 2A). These correspond to the short and long blocks, respectively, observed in the single-channel records (Figure 1). Toro et al. (1992) report a mean blocked time of 26 ms for short blocks of the coronary artery channel, but have not analyzed longer blocked events. To analyze the long blocks in isolation, we excluded all nonconducting events shorter than 200 ms; an example of this is shown in Figure 2B. After exclusion of all events shorter than 200 ms, the distribution of the remaining blocked times can be fitted by a single-exponential time constant of duration 160 ms, identical to the long time constant derived from the complete data set (Figure 2A). We constructed a dose-response relationship showing the percent decrease in the open probability produced bytheShB peptide(Figure3A; long blockswere excluded in thisanalysis).A50% inhibitionoftheopen probability is achieved at a peptide concentration of 5 PM. Transformation of these data to a Hill plot results in a straight line with a slope of 0.98, suggesting that the short block produced by the peptide results from a bimolecular reaction between peptide and channel. This was confirmed by an examination of the rates for peptide association with and dissociation
ShB Peptide Blocks Cal+-Dependent
K’ Channels
231
0
Time Figure 2. Cumulative
100
600
400
200
200
300
(ms)
Dwell-Time
400
500
Time
Distributions
for Nonconducting
600
700
RflO
900
(ms)
States in the Absence
and Presence
of ShB Peptide
dwell times of the channel in nonconducting states are exponentially distributed. The control data (open circles) are best fitted with two exponentials (both shorter than 5 ms), indicative of two distinct closed states. In the presence of 50 W M ShB peptide (closed triangles), two additional nonconducting states with time constants of 16 ms (short blocks) and 158 ms (long bloc:ks) are observed. The control distribution is constructed from more than 2000 events, and the distribution in the presence of 50 PM peptide is constructed from more than 100,000 nonconducting events. Approximately 85% of these events were intrinsic closings, and the remainder were short (14%) and long (1%) blocks. P(t) is the probability that a given dwell time is greater than or equal to time t. (B) To focus on the dwell-time distribution for the long blocks, all nonconducting events shorter than 200 ms were dIscarded. The remaining events can be fitted with a single exponential (T = 160 ms) that is identical to the long time constant In (A). This distribution is constructed from a total of 420 events. The inset shows that transformation of the data to a semi-logarithmic plot gives a straight line with a slope equal to the time constant. (A) The
bimolecular reaction. The apparent dissociation constant (KJ for the short block, calculated from these rate constants, is 8 PM, similar to the value derived from the dose-response relationship. This is also similar to the affinity of the peptide for the ShB channel
from
the channel. As shown in Figure 38, the associareaction is linearlydependenton thepeptideconcentration (on rate of 0.00875 rns-l PM-‘), whereas the dissociation reaction is independent of peptide concentration (off rate of 0.07 ms-‘), as expected for a
tion
0
10
20
30
40
50
[Peptide] Figure 3. Characteristics
60
70
80
90
100
0
Block
40
[Peptide]
(#VI)
of the Short
70
Produced
60
60
1°C)
(/r.M)
by ShB Peptide
(A) The percent decrease in the channel open probability (P,) is expressed as a function of peptide concentration under the same conditions as those described in Figure 1. Each point represents the mean * SEM of 5 separate determinations on 5 different channels. Transformation of these data to a Hill plot gives a Hill coefficient of 0.98. (B) The rate for the association reaction (l/r,, open circles) of the short block increases linearly with peptide concentration, whereas the rate for the dissociation reaction (l/re, closed circles) is independent of peptide concentration. The apparent Kd of the peptide for the short block, calculated from these rate constants, is 8.2 PM.
Neuron 232
.
1 Peptide Figure 4. Kinetics
1 (j&l)
and Voltage
Dependence
llolding of the Long Blocks
Produced
potential,
tl)V
by ShB Peptide
Peptide on rates (open circles) and off rates (closed circles) for the long blocks were measured from single-channel blocked and burst intervals. Only events longer than 200 ms were included in this analysis. (A) The on rate is a linear function of peptide concentration, whereas the off rate is independent of peptide concentration. Each point represents the mean + SEM of 4 separate determinations on 4different channels. (B)Voltagedependence of the long block. A channel was observed with 150 m M symmetrical KCI in the presence of 50 PM ShB peptide, at the indicated applied voltages.
A
Control
50 FM Peptide
C
[TEA],
mM
50 PM Peptide and 60 mM TEA
Figure 5. Effect of Internal
TEA on the Short and Long Blocks
(A) A single channel was inserted into a lipid bilayer, and records were taken at +40 mV holding potential. The top trace shows a control record with no added blockers. The middle trace shows the effect of adding 50 PM ShB peptide to the internal side of the membrane. Note both long and short blocks. The bottom trace shows that 60 m M TEA, added to the internal chamber in the continued presence of peptide, competes with the short blocks. The quantitative underpinnings for this conclusion are given in (B). (6) Effects of TEA on mean unblocked time for the short blocks and on single-channel current. A single channel was observed in the presence of 50 NM peptide at the indicated concentrations of internally applied TEA. The ratio of the mean unblocked time measured in the presence of TEA to the mean unblocked time measured in the absence of TEA (open circles) and the ratio of the single-channel currents in the absence and presence of TEA (closed circles) are plotted. The fact that they exhibit an identical dependence on TEA concentration indicates that ShEi peptide competes with TEA for the site that mediates short blocks. (C) Long peptide blocks are still observed in the presence of TEA (see bottom trace of [A]). The mean duration of the long blocks, calculated by excluding all events shorter than 200 ms, is the same in the absence (closed circles) or presence (open circles) of TEA.
ShB Peptide Blocks Ca”-Dependent 233
K’ Channels
A
5 mM external K’ and 50 pM Peptide
0.01 I-, 1c.l 200 300 ,Llo 500 bW 100
50 tnM external K+ and 50 pM Peptide
Time
Figure 6. External
K’ Does
Not Influence
the Duration
of the Long Blocks
Induced
(ms)
by ShB Peptide
(A) A single channel was observed at a holding voltage of +30 mV. The solution on the internal side of the channel contained 150 m M K’, and that on the external side contained 5 m M K’. The top trace shows control behavior under these conditions. The middle trace shows the effect of addition of 50 PM ShB peptide to the internal side of the membrane. Note both long and short blocks. In the bottom trace the concentration of external K+ is increased from 5 m M to 50 m M in the continued presence of the peptide. Kinetic analysis as described in Figure 2 demonstrates that the time constant for the short blocks induced by the peptide is reduced by this treatment from 16 ms to I3 ms. (B) The time constant for the long blocks, measured as described in Figure 2, is the same at 20 m M (closed circles), 40 m M (open circles), and 60 m M (open triangles) external K+. The solid line is taken from the inset of Figure 2B (150 m M external K+i.
itself (R. D. Murrell-Lagnadoand R. W.Aldrich, unpublished data). The longer-lived nonconducting intervals are reminiscent of the type of block produced by CTX at the external mouth of Ca*+-dependent K+channels (Miller et al., 1985). In the case of CTX, Anderson et al. (1988) demonstrated that these blocked events represent the binding of a single CTX molecule with a channel in a simple bimolecular reaction. To evaluatewhether the ShB peptide produces long blocks in a similar fashion at the inside mouth of the channel, we measured the association and dissociation rates of the peptide. These were determined from a statistical analysis of the long-lived blocked times (nonconducting intervals longer than 200 ms) and burst times (combined open, closed, and short-blocked intervals). Figure 4A shows that the association rate for the peptide is linearly dependent on its concentration and that the dissociation rate remains constant, regardless of peptide concentration. This demonstrates that the long block produced by the peptide also involves its interaction with Ca*‘-dependent K+ channels in a bimolecular fashion. Thus each long-lived blocked event represents the binding of one molecule of peptide to one channel. The Kd of the peptide for the long-lived blocked events, calculated from these measured association and dissociation rates, is 275 PM in
this experiment, more than an order of magnitude higher than that for the short blocks. In other experiments the Kd for the long blocks ranged from 140 PM to 275 PM. Both the association and dissociation rates vary with voltage (Figure 4B), such that the long blocks are more frequent and of longer duration at depolarized voltages. The rate constants for the short blocks exhibit a similar dependence on voltage (data not shown; see Toro et al., 1992).
Internal Tetraethylammonium Short Blocks
Competes with the
Internal tetraethylammonium (TEA) is known to block the internal mouth of Ca*+-dependent K+channels, as well as other K+channels, by entering the K+conduction pathway and thus preventing current flow (Vergara et al., 1984; Villarroel et al., 1988). The blocking events produced by TEA are too brief to be resolved as individual closings, so TEA appears to reduce the amplitude of the single-channel current. This effect can be seen in Figure 5A by comparing the middle trace (no TEA present) with the bottom one (60 m M TEA). In addition, TEA added together with the ShB peptide reduces the frequency of the short blocks produced by the peptide. As shown in Figure 5B, the reduction in frequency of the short peptide blocks and the reduction in single-channel current show
NWKHl 234
identical dependences on TEA concentration, indicating that the two blockers bind in a mutually exclusive way to the channel. Although we did not have sufficient data to analyze the effects of TEA on the frequency of the long blocks, note that their durations remain unaffected by TEA (Figure 5C). Increased External K’ Destabilizes the Short but Not the Long Blocks Increasing the K+ concentration of the external soiution decreases the duration of the short block produced by 50 P M peptide, but does not affect the long block. Figure 6A shows single-channel records from a membrane with 150 m M K+and 50 P M peptideon the cytoplasmic side, at two external K+concentrations. At high extracellular K’, the short blocks produced by peptide are shorter, whereas the frequency and durations of the long blocks are unchanged. An analysis of the distribution of short blocked times shows that the mean blocked time induced by 50 t.rM peptide at low extracellular K+, 16 m s in this experiment, is reduced to 8 m s at high extracellular K’ (data not shown). In contrast, the distribution of the durations of long blocks is not affected by extracellular K’ over a wide range of concentrations (Figure 6B). Discussion The ShB peptide can produce two kinetically distinct blocking events in brain and muscle CaL+-dependent K+channels. Although both blocking events are bimolecular, the Kd for the long block is more than an order of magnitude higher than the Kd for the short block. The two blocking events can also be distinguished on the basis of competition with external K+. Taken together these data suggest that the long and short blocks may result from the binding of ShB peptide to two different receptor sites on Cal+-dependent K+ channels. In ShB it is known that internal TEA competes with the inactivation ball (Choi et al., 1991). Recently Demo and Yellen (1991) havedemonstrated that the inactivation ball can also bedisplaced by K+from theopposite side. Thus inactivation in Shaker has been proposed to occur by open channel pore blockade. W e tested whether the long and short blocks of Ca2+-dependent K’ channels by the inactivation peptide might occur by pore blockade. Internal TEA competes with the short block, consistent with the idea that the short blocks involve pore blockade. However, the fact that the two blockers compete does not necessarily demand that the two compounds bind to the same site. Rather they demonstrate that in the presence of internal TEA the peptide can no longer bind to the receptor producing the short block. The receptor site for TEA and the receptor site for the peptide could be physically close, and so binding of one automatically precludes binding of the other. Since the peptide has a net positive charge, it is also possible that the binding of TEA results in electrostatic repulsion that prevents
the peptide binding to its receptor. Finally, the binding of TEA could also induce some allosteric change in the channel protein that renders the channel incapable of binding peptide to produce the short blocks. W e found that increasing external K+ decreases the duration of the short blocks but leaves the long blocks unaffected. The easiest interpretation of this result is that K+ flowing through the channel displaces the peptide from the receptor normally producing the short blocks. The exact mechanism by which increasing the concentration of a permeant ion speeds the exit rate of a blocking particle on the other side is unknown. The mechanism proposed by MacKinnon and Miller (1988) for relief of external block of CTX by K’ could also be inferred here. It is possible that a K+-binding site located in the pore is positioned close to a positively charged residue on the peptide. When peptide is bound to the site producing short blocks, an electrostatic repulsion between K’and the peptide ensues, leading to an increase in the exit rate of the blocker. In any event, the short block demonstrates two characteristic features of pore blockade, competition with a known pore blocker and relief of block by raising the concentration of a permeant ion on the opposite side of the membrane. In contrast the long block is not relieved by the permeant ion. This may be because the long blocks involve peptide binding to the pore in an orientation such that it cannot be displaced by ion flux. Alternatively, the peptide may produce long blocks by binding to some other site that does not reside in the pore. Toro et al. (1992) have shown that the ShB peptrde produces short blocks of Cal+-dependent K’ channels from coronary artery. It is interesting that the conc-entration of peptide required to produce short blocks of these channels is 6- to IO-fold higher than that required for the brain and skeletal muscle channels we studied, suggesting that there may be variations among Cal’-dependent K’ channels in the structure of the binding site involved in short peptide block. The binding of the peptide to its receptor in at least some Shaker-like channels is thought to be an inherently voltage-dependent process (Ruppersberg et al., 1991). Such results are consistent with pore blockade. In our hands, both the short and long blocks of brain and muscle CaLi-dependent K+ channels are voltage dependent, as are the short blocks of the coronary artery channel (Toro et al., 1992). Accordingly, the receptor sites mediating both blocks must reside within the membrane, as they sense the potential field across the membrane. Thus it is somewhat surprising that, when the peptide binds to the site producing the long block, it cannot be displaced by permeant ion. It is possible that the receptor site producing the long block is remote from the conduction pathway, yet still within the membrane. Binding to this site could produce the long blocking events via an allosteric effect on the channel protein. According to the model of inactivation put forward by Aldrich and coworkers (Hoshi et al., 1990; Zagotta
ShB Peptide Blocks Ca”-Dependent 235
K+ Channels
et al., 1990), the ShB channel possesses a specific binding site for the cytoplasmic peptide domain that induces inactivation. Site-directed mutagenesis experiments have been used to probe for the location of the receptor site within the Shakerchannel, and the S4-S5 cytoplasmic linker, which may lie near the channel’s permeation pathway, has been identified as a region that might be part of a receptor for the inactivation gate (Isacoff et al., 1991). It is interesting to compare the sequence of this region with that of the S4-S5 linker in the recently identified component of a Ca2+dependent K+ channel from Drosophila (Atkinson et al., 1991). Although the overall sequence of the latter channel differsextensivelyfrom thoseof other cloned K+ channels, 5 out of 10 amino acids are conserved in the S4-S5 linker when compared with the sequence from ShB. Thus it is possible that this region may be involved in the actions of the inactivation peptide on Ca2’-dependent K+ channels. Experimental
Procedures
All experiments were carried out on single, high conductance Cal+-dependent K’ channels from either rat skeletal muscle or rat brain plasma membrane vesicles, which were inserted into planar phospholipid bilayers. Membrane preparations were exactly as described previously (Anderson et al., 1988; Reinhart et al., 1989). Planar Bilayers and Membrane Vesicle Fusion Planar lipid bilayers were formed from a mixture of I-palmitoyl, 2-oleoyl-phosphatidyl-ethanolamine and I-palmitoyl, 2-oleoylphosphatidylserine (3:l) in n-decane, as described previously (Reinhart et al., 1989). Membranes were routinely added to the cis chamber. The solution on this side consisted of 150 m M KCI, 1.05 m M CaC&, 1 m M ECTA, 1 m M M&I,, 7 m M KOH, and 10 m M HEPES (pH 7.2). The composition of the trans solution was 0.1 m M EGTA, 7 m M KOH, and 10 m M HEPES (pH 7.2). Ca2’dependent K’channels inserted in one direction, that iswith the Cal+-sensing side facing the c/s solution. Thus, this is defined as the cytoplasmic side of the channel. Following incorporation of a channel, further fusion was suppressed by making thetrans K’ concentration the same as the cis K’ concentration. This was the case for all experiments except those in which the effect of extracellular K’ on peptide block was determined. Voltages are expressed as the voltage of the cis side with respect to the trans stde of the bilayer. Electrical Recordings and Analysis Single-channel currents were amplified using a current-tovoltage converter with a 10 CD feedback resistor, filtered at l-4 kHz using an 8-pole Bessel filter, and recorded on a B-videocassette recorder after l&bit digitization at 44 kHz with a pulse code modulator (Sony PCM-701ES). Data were analyzed off-line using an MS-DOS-compatible computer (Everex 386, 20 MHz) equipped with a Tecmar labmaster AID board coupled to an lndec IBX Instrumentation Interface and utilizing an lndec BASIC-Fastlab software library. Using operator-assisted computer programs, current records filtered at 1 kHz were sampled at 5 kHz and searched for transitions away from the closed or open level.Thethreshold fordetectingopeningand closingtransitions was set to 50% of the open level current for each individual event. In this way current levels corresponding to each closed/open transition were reduced to three values correspondSuch reduced ing to the closed time, open time, and amplitude. lists were then further transformed to yield the open probability, mean amplitude, and open and closed timedistributions. Cumulative dwell-time distributions were fitted with exponentials using the [-evenberg-Marquardt curve fitting algorithm, and the
goodness of fit was evaluated by chi-square analysis. P(t) is the probability that a given dwell time is greater than or equal to time t. All experiments were repeated at least three times. ShB Peptide A peptide corresponding to the first 20 amino acids of ShB and another in which the leucine at position 7 was changed to glutamate (L7E mutant) were synthesized by the Protein and Nucleic Acid Facility at Stanford University. An amtde was placed at the carboxyl terminustoprevent itschargefrom influencing peptide activity, and the synthetic peptides were purified by reversed phase, high-pressure liquid chromatography prior to use (see Zagotta et al., 1990, for details). Acknowledgments This work was supported by NIH grants NS17910 to (I. B. L.) and NS23294 to (R. W. A.). R. W. A. is an Investigator of the Howard Hughes Medical Institute, and W. N. 2. IS a Warner Lambert Fellow of the Life Sciences Research Foundation. We aregrateful to Drs. Shimon Marom and Chris Miller for their critical comments on the manuscript and to Dr. Manuel Esguerra for help with the statistical analysis. We also thank Drs. L. Toro, R. Latorre, and E. Stefani for communicating results prior to publication. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC Section 1734 solely to indicate this fact. Received
March
12, 1992; revised
May 15, 1992
References Anderson, C. S., Mackinnon, R., Smith, C., and Miller, C. (1988). Charybdotoxin blockof singleCa’+-activated K’channels. Effects of channel gating, voltage, and ionic strength. J. Gen. Physiol. 97, 317-333. Armstrong, C. M., and Bezanilla, sodium channel. II. Cating current 70, 567-590.
F. (1977). Inactivation of the experiments. J. Gen. Physiol.
Atkinson, N. S., Robertson, G. A., and Canetzky, B. (1991). A component of calcium-activated potassium channels encoded by the Drosophilas/o locus. Science 253, 551-555. Choi, K. L., Aldrich, R. W., and Yellen, G. (1991). Tetraethylammonium blockade distinguishes two inactivation mechanisms in voltage-activated K’ channels. Proc. Natl. Acad. Sci. USA 88, 5092-5095. Chung, S., Reinhart, P. H., Martin, B., Brautigan, D., and Levitan, I. B. (1991). Protein kinase activitycloselyassoc-iated with a reconstituted calcium-activated potassium channel. Science 253, 560-562. Demo, S. D., and Yellen, G. (1991). The inactivation gate of the Shaker K+ channel behaves like an open-channel blocker. Neuron 7, 743-753. Ewald, D. A., Williams, A., and Levitan, I. B. 11985). Modulation nt K’channel activity by protein phosof single Ca”-depende phorylation. Nature 375, 503-506. Hoshi, T., Zagotta, W. N., and Aldrich, R. W. (1990). Biophysical and molecular mechanisms of Shaker potassium channel inacttvation. Science 250, 533-538. Isacoff, E. Y., Jan, Y. N., and Jan, L. Y. (199’1) Putative the cytoplasmic gate in the Shaker K’ channel. 86-90.
receptor for Nature 353,
Iverson, L. E., Tanouye, M. A., Lester, H. A., Davidson, N., and Rudy, B. (1988). Expression of A-type potasstum channels from Shaker cDNAs. Proc. Natl. Acad. Sci. USA 85, 5723.-5727. Lancaster, B., Nicoll, R. A., and Perkel, D. J. (1991). Calcium vates two types of potassium channels in rat hippocampal rons in culture. J. Neurosci. 11, 23-30. Lechleiter,
J. D., Dartt,
D. A., and Brehm,
acttneu-
P (1988) Vasoactive
NWOl
236
intestinal peptide activates Cal+-dependent K’channels through a CAMP pathway in mouse lacrimal cells. Neuron 1, 227-235. MacKinnon, R., and Miller, C. (1988). Mechanism of charybdotoxin block of the high-conductance, C&+-activated K’ channel. J. Cen. Physiol. 91, 335-349. Maue, R. A., and Dionne, V. E. (1987). Patch clamp studies of isolated mouse olfactory receptor neurons. J. Gen. Physiol. 90, 95-125. Miller, C., Moczydlowski, Charybdotoxin a protein channels from mammalian
E., Latorre, R., and Phillips, M. (1985). inhibitor of single Ca2+-activated K’ skeletal muscle. Nature 373,316-318.
Pongs, O., Kecskemethy, N., Muller, R., Krah-Jentgens, I., Baumann, A., Kiltz, H. H., Canal, I., Llamazares, S., and Ferrus, A. (1988). Shaker encodes a family of putative potassium channel proteins in the nervous system of Drosophila. EMBO, J. 7,10871096. Reinhart, P. H., Chung, S., and Levitan, I. B. (1989). A family of calciumdependent potassium channels from rat brain. Neuron 2, 1031-1041. Reinhart, P. H., Chung, S., Martin, B. L., Brautigan, D. L., and Levitan, I. B. (1991). Modulation of calcium-activated potassium channels from rat brain by protein kinaseAand phosphatase2A. J. Neurosci. II, 1627-1635. Tempel, B. L., Papazian, D. M., Schwarz, T. L., Jan, Y. N., and Jan, L. Y. (1987). Sequence of a probable potassium channel component encoded at Shaker locus of Drosophila. Science 237, 770775. Toro, L., Ramos-France, J., and Stefani, E. (1990). GTP-dependent regulation of myometrial Kc. channels incorporated into lipid bilayers. J. Cen. Physiol. 96, 373-394. Toro, L., Stefani, E., and Latorre, R. (1992). Internal blockade Cal+-activated K’channel by ShakerB inactivating”ball”peptide. Neuron, this issue.
of a
Vergara, C., Moczydlowski, E., and Latorre, R. (1984). Conduction incorporated blockade and gating in a Ca 2+-activated K’channel into planar lipid bilayers. Biophys. J. 45, 73-76. Villarroel, A., Alvarez, O., Oberhauser, A., and Latorre, R. (1988). Probing a Ca*+-activated K+ channel with quaternary ammonium ions. Pfltigers Arch. 473, 118-126. White, R. E., Schonbrunn, A., and Armstrong, D. L. (1991). Somatostatin stimulates Ca2+-activated K’ channels through protein dephosphorylation. Nature 357, 570-573. Zagotta, W. N., Hoshi, T., and Aldrich, R. W. (1990). Restoration of inactivation in mutants of Shaker potassium channels by a peptide derived from Shaker. Science 250, 568571.