Neuron,
Vol. 2, 87-95,
January,
1990, Copyright
0 1990 by Cell Press
An Open-Channel Blocker Interacts with Adjacent Turns of a-Helices in the Nicotinic Acetylcholine Receptor Pierre Charnet: Cesar labarca: Reid J. Leonard:+ Nancy J. Vogelaar,*S Linda Czyzyk: Annie Couin: Norman Davidson: and Henry A. Lester* *Division of Biology *Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena, California 91125
Summary The binding site for an open-channel blocker, QX-222, at mouse muscle nicotinic acetylcholine receptors was probed using site-directed mutagenesis, oocyte expression, and electrophysiological analysis. The proposed cytoplasmic end of the M2 transmembrane helix is termed position 1’. At position lO’(aS252, bT263, yA261, 6A266), Ala residues yield stronger and longer binding of QX-222 than Ser or Thr residues. These effects are opposite and roughly equal (30%~50% per mutation) to previously reported effects at position 6’. The polar end of an anesthetic molecule seems to bind to the position 6’ OH groups, which provide a water-like region; the nonpolar moiety is near position 10’ and binds more strongly in a nonpolar environment. Interactions with adjacent OH-rich turns of an amphiphilic helix may explain the widespread blocking effects of local anesthetics at the conduction pore of ion channels. Introduction The first synthetic local anesthetic, procaine, was introduced by Einhorn in 1905 as an alternative to cocaine, which has additional psychotropic effects. These and newer local anesthetics share two structural features: an ionizable amino group and an aromatic group (Ritchie and Greengard, 1966; Hille, 1984; Ritchie and Greene, 1985); the two moieties are separated by 5-6 A. This distance corresponds well with the repeat distance between adjacent turns of an a-helix, about 5.4 A (Pauling et al., 1951). One hypothesis is therefore that local anesthetics interact with adjacent turns of a-helices and that such interactions help to account for the widespread blockade of ion channels by local anesthetics and their analogs. This hypothesis is now subject to decisive tests with the nicotinic acetylcholine receptor (AChR). Increasing evidence suggests that the M2 membrane-spanning region from each of the five subunits (a&8) lines the ion channel pore (reviewed by Miller, 1989; Dani, 1989a). First, experiments with photoaffinity
+ Present address: Department of Membrane Biochemistry and Biophysics, Merck Sharp and Dohme Research Laboratories, Rahway, New Jersey 07065. §Present address: Department of Chemistry, Princeton University, Princeton, New Jersey 08554.
labeling showed that chlorpromazine binds to amino acid residues in the M2 membrane-spanning region of the a, B, and 6 subunits (Ciraudat et al., 1986, 1987; Hucho et al., 1986). Second, initial experiments with chimeric mouse-calf S subunits showed that channel conductance was influenced by a region including the M2 region and the M2/M3 hinge (Imoto et al., 1986). Third, additional experiments with site-directed mutagenesis revealed that channel conductance is controlled by rings of negative charge, contributed from each of the four subunits, at each end of the M2 helix (Imoto et al., 1988). Fourth, mutations at a polar (Ser-rich) site within the M2 helix affected the residence time and equilibrium binding affinity for the prototype open-channel blocker, QX-222, and caused rectification in the channel conductance (Leonard et al., 1988). This paper describes evidence supporting the view that QX-222 (Figure IA), a quaternary ammonium derivative of lidocaine, does indeed interact with adjacent turns of the M2 helix of mouse muscle AChR. We have continued to study the binding of QX-222 to the receptor by employing site-directed mutagenesis, oocyte expression, and electrophysiological analysis. The region we have studied is shown in Figure IB. Our terminology, following that of Miller (1989), assigns the number 1’ to the N-terminus of the M2 region. Position 6’ corresponds to the “inner polar site” (IPS) of Leonard et al. (1988). At this site, Leonard and coworkers (1988) reported that the binding of QX-222 became weaker as polar amino acids (Ser) were replaced by nonpolar residues (Ala). We now show that similar mutations at position 10’ have an opposite effect: replacement of polar (Ser or Thr) residues by nonpolar residues actually increases the binding of QX-222. The straightforward interpretation of these data is that the charged, amino moiety of QX-222 is binding at position 6’ and that the aromatic, nonpolar end is binding at position IO’. We also present less extensive data on the effect of mutations at position 2’and at several other locations. These results support the general view that the M2 domain of each subunit forms an a-helix with a polar stripe consisting of residues 6’ and 10’ (and possibly also 2’), that these polar stripes line the channel, and that this region in the channel also forms the binding site for a class of open-channel blockers. Results Table 1 summarizes the equilibrium and kinetic properties of the ACh-induced currents for both normal and mutated AChRs. Most of the position 6’ mutant clones utilized in the present study were the same as those reported previously by Leonard et al. (1988); we have retested these and repeated the previous macroscopic measurements at 0.5 PM ACh, a concentration
QX-222
A
Position 10’ Mutations Do Not Affect Single-Channel Conductance There were no changes in single-channel conductance for any of the position IO’ mutations, even the as10.ABT,0.AY8 hybrid, in which 3 OH-containing residues were replaced by Ala. This result contrasts with the position 6’ series in which replacement of 3 Ser by Ala residues (aS6.Aby8Sb.A) produced a pronounced rectification of the channel currents (Leonard et al., 1988). Our study of the position 2’ mutants was limited by the fact that not all combinations yielded functional expression. In particular, the combinations with the most Ser (or Thr) to Ala substitutions, an&+,&i, aT2.Afi&2A, and aR’A(jYT2.A8s2.A, produced no agonistinduced currents. Each of the mutant subunits in the position 2’ series did display agonist-induced currents in combination with normal subunits (Table I), indicating that there were no major structural problems with these mutants. Interestingly, the single-channel conductance for the a&&Y8 hybrid was 20% lower than normal. This result was reproducible and significant.
CH3
CH3
I
NH
dH2
CH3-
rii+-m
CH3
CH3
B
QX-222 Residence Time Increases with of Ser Residues at Position 10
6’
Figurel. Structure of QX-222 and AChR Subunits in the M2 Region Position (1988).
6’was
termed
the Inner
polar
18’
14’
10'
Sequences
of
the
site (IPS) by Leonard
Mouse
Removal
The kinetic analysis of the single-channel data are exemplified by Figure 2 and summarized in Table 2. The position 10’ mutations affected the kinetics of QX-222 blockade. The fast closed times increase for mutations in which Ser or Thr residues are changed to Ala residues, and they decrease for a mutation in which an existing Ala residue on they subunit is replaced by Ser. The model we use for the interaction between QX-222 and the AChR is a simplified single binding scheme:
et al.
R r~owi-Ro~en
WQI +=
Wo,wn
Q)bkxked
(1)
F
that gave optimal results for the position 10’ mutations. The results were quite similar to those reported previously. We shall therefore concentrate on the description of the position 10’ mutations. All of the mutant hybrids that yielded functional receptors also displayed Hill coefficients approaching 2 in the dose-response studies. A Hill coefficient near 2 characterizes all known native AChRs and suggests that the open state of the channel is much more likely to be associated with the presence of two bound agonist molecules than with a single bound agonist molecule. Nearly all of the mutations had observable effects on the single-channel and macroscopic kinetics of the response to AChR. A later publication will describe and systematize these results; the present paper deals only with the effects of QX-222 on these kinetics.
in which R = AChR and Q = blocker (Adams, 1977; Neher and Steinbach, 1978). In this scheme, the channel is blocked when QX-222 is bound within it and the average fast closed time is simply equal to l/F. Figure 3 presents a graphic display of these trends by plotting F as a function of the number of Ser/Thr residues replaced by Ala. The linear fit in Figure 3 represents a change of 29% for each residue replaced at position IO’, with an opposite, but roughly as great (35%), change for the position 6’ mutations. Scheme 1 can also be used to analyze the secondorder bimolecular rate constant, G, for the binding of QX-222 to the AChR. A full analysis would call for measurements at varying QX-222 concentrations; we have not pursued such a study, but instead have confined our measurements to a single concentration, [QX-2223 = 20 PM. We use the relation
G = A(lit)l[QX-2221
(2)
Channel 89
Table
Blockers
at Nicotlnic
1. Characterization
Receptor
of the Single
N 4YS
0
Normal
and
Mutant
Hybrid
AChRs
Expressed
in Oocytes
Channel
Conductance
(pS)
Inward
Outward
Open
Macroscopic
36 k 1 (4)
24 + 1 (4)
15.3 * 1.2 (8)
1010 k 121 (23)
51.9
+ 2 (17)
1.6 i 0.2 (4)
* k + f +
(5) (3) (3) (6) (3)
403 1073 1630 1029 1087
* * + k +
75 (11) 267 (9) 314 (12) 381 (9) 322 (IO)
44.5 46.5 41.7 54.5 59.3
* k + * +
3.3 1.2 2.9 3.0 3.8
(11) (9) (9) (9) (IO)
1.9 1.7 2.3 1.7 1.7
* 2 + * (I)
0.1 0.1 0.1 0.02
(3) (5) (2) (4)
0.3 0.1 0.1 0.1
(5) (14) (7) (7)
Time
(ms)
Current
(nA)
Hill
T (ms)
Coefficient
Position 10 &*,,.d ahAYS a5dvA103 a5dY~ a\,nxPrKrAY~
+I -1 -1 -2 - 3
39 35 33 34 37
f * + k *
2 1 3 1 1
(6) (7) (4) (8) (7)
22 27 26 27 27
* + (I) f k
2 (4) 1 (7) 1 (8) 1 (7)
7.9 17.6 33.8 15.3 17.9
Position 6 alh,sy~ aWShA adM aYJA e YLA
+I -1 -2 -3
33 32 36 33
+ f f *
1 1 1 2
(6) (6) (6) (7)
22 21 24 13
+ + + f
1 1 2 1
(6) (6) (6) (5)
5.0 28.4 14.1 20.6
* 0.7 (3) (1) k 4.2 (3) -t 3.0 (3)
350 954 770 904
f +_ k +
62 (13) 152 (23) 144 (16) 199 (15)
16.6 66.9 42.2 46.9
* + k i
1.2 2.7 2.8 3.5
(9) (23) (16) (13)
1.6 1.7 1.7 1.7
fi + * +
Position 2 c&&j ah* AS aB-& A ar2 dkrt~G aT2 APY~ dYn A&Z A
+I -1 -1 -1 -2 -2
32 36 33 30 ND 36
k + + f
1 1 1 1
(14) (4) (9) (9)
22 + 1 23*1(4) 20 f 1 20 f 1 ND 24 + 1
(13)
15.1 ND 28.9 22.3 ND 27.9
+ 4.4 (2)
587 738 926 553 468 657
f * f k t *
87 (2) 309 (2) 313 (IO) 230 (5)" 139 (3) 54 (2)
39.9 87.6 73.3 55.9 65.7 101.3
+ f * k + k
0.9 10 3.1 3.1 2.7 5.2
(2) (3) (IO) (5) (3) (3)
1.5 1.8 1.7 ND ND 1.9
* 0.1 (2) * 0.1 (2) i 0.1 (8)
f 1 (4)
(6) (9) (4)
0.9 1.1 6.7 1.9 5.2
k 9.0 (2) (I) (1)
* 0.2 (2)
The parameter N represents the number of Ser or Thr residues relative to the normal ae$ receptor. Note that only the a~6.AP&,A receptors display a selective decrease in conductance for outward currents. Single-channel open times and agonist-Induced currents are given for -150 mV. Time constants T of voltage-jump relaxations are given for jumps from +50 mV to - 150 mV at 0.5 PM ACh. Data are mean k SEM (number of observations). d 1 PM ACh.
in which t is the mean open time in the absence or presence of QX-222 (Neher and Steinbach, 1978). Values for G are given in Table 3. The important point is that the values for all of the mutants cluster within
*30% of the average value (1.9 x 10’ M-lsC at -100 mV and 2.6 x IO7 M-‘s-l at -150 mV). This indicates that the mutations we have performed change the equilibrium dissociation constant F/G only because Figure AChR
2. Single-Channel Mutants
Analysis
of the
Left column, typical single-channel traces in the presence of ACh alone (1 PM) at -150 mV. Middle column, traces in the presence of ACh plus QX-222 (20 PM). Right column, frequency distribution histograms for the fastest detectable closed times in the presence of ACh plus QX-222. The time constant of these distributions corresponds to the average residence time for QX-222 and equals l/F. The top row gives the data for the positlon 10 mutant hybrid, c~,~~,$~,~,~~; the middle row gives the data for the wild-type receptor; and the bottom row gives the data for the position 6’ mutant hybrid, a&y&,,,. Calibration bars: 3 PA; 25 ms for ACh alone; 12 ms for ACh plus QX-222.
Closed
Time (ms)
Table 2. Evaluation the Various Mutant
of the Dissociation Rate Constant, Hybrids at Two Different Voltages
F, for
Table 3. Evaluation of the Association the Interaction between QX-222 and Different Voltages
F(s-‘) - 100 mV
C(10’
150 mV
1400 f 100 (7)
390 k 13 (6)
1890 781 980 705 391
532 348 366 281 176
+ f (I) k +
9 (4) 24 (2)
Position a~yAd
aPTlrAYfi 8 (5) 3 (5)
k + + k
184 (3) 123 (4) 68 (4) 208 (4)
aSl~AhA,d a\lwAh6
951 ND 1557 1207 ND 1745
+ 74 (2)
298 625 760 838
-f + f k
24 (3) 1 (2) 46 (2) 12 (3)
294 ND 621 452 ND 502
+ 20 (2)
Position
Data
are mean
(1) (1)
@-b&A aSbAbY
(1)
f SEM (number
k 69 (2)
of observations).
Data
they change surements.
F, at least
within
the
limits
of our
mea-
Macroscopic Data Also Show Stronger Binding with Removal of Ser Residues at Position 10 The rationale for the analysis of macroscopic voltage-jump relaxations has been given by Leonard et al. (1988). Such relaxations are most readily interpreted when the agonist concentration is restricted to a range giving only small fractional activation of receptors. Under such conditions, a pure open-channel
0
\
Position
h
Position 2 ulhtY~ ah A6 &b A adGL ,YS a&W abT~A~~~ A
+ 68 (2) (1)
6’
are
1.99 f 0.08 (3)
2.76
f 0.16 (3)
2.55 1.95 ND 1.88 1.90
i 0.19 (2) (I)
f 0.42 (3) + 0.13 (2)
+ 0.13 (3) * 0.54 (3)
2.63 2.72 ND 2.80 2.28
2.29 1.96 1.41 1.52
f ? f +
1.21 t 0.45 (3) ND ND 1.75 (I)
+ 0.38 (3) + 0.20 (3)
6
alkb5y6
as6 AP&
- 150 mV
IO
aSl~AhvA~6
924 1839 1993 2896
‘s ‘1
100 mV aby6
(I) (1) (1) * 100 (3) (I)
M
Rate Constant, C, for the Receptor, at Two
mean
0.11 0.56 0.10 0.36
(3) (2) (2) (2)
1.77 + 0.16 (2) ND 1.92 (I) 1.29 (I) ND ND + SEM (number
2.21 ND 2.18 2.17 ND s.28
(I) i 0.19 (2) (1) (I)
of determinations).
blocker is expected to produce little diminution of the equilibrium agonist-induced currents. We satisfied this criterion by using 0.5 PM ACh; the blockade by QX-222 at equilibrium amounted to at most 17% at -150 mV or 10% at -110 mV and was usually much less. Voltage-jump relaxations (Figure 4A) generally have a single exponential component in the presence of agonist alone; but in the additional presence of an open-channel blocker, one expects two exponential components. The faster “inverse relaxation” (Adams, 1977) is too rapid to be resolved in oocytes with the two-microelectrode voltage clamp. The slower component was, however, easily resolved in our experiments. Its time constant is increased in the presence of QX-222:
1000
TO = ~(1 + [QX-222]/Ko) 7 z L
100
-4
-3
-2
-1
N = S/T Figure 3. Dissociation QX-222 and the AChR, 6’ and IO
0
1
2
3
4
removed/added
Rate Constant of the Complex between versus the Number of SerIThr at Positions
The abscissa is N, the number of Ser/Thr Ala (negative numbers) or substituted for numbers). The ordinate is the dissociation mined from analyses like those of Figure [ACh], 1 fiM.
residues changed to Ala or Phe (positive rate constant F, deter2. Voltage, -150 mV;
(3)
in which TV and T are the time constants in the presence and absence of QX-222, respectively, and Ko = F/G. The increased time constant may be explained intuitively by pointing out that this time constant essentially equals the average duration of the burst of openings associated with each activation of the channel. When an open-channel blocker is present, this burst is lengthened to include the blocked times as well (Neher and Steinbach, 1978; Neher, 1983). As predicted by Equation 3, voltage-jump relaxations were prolonged in the presence of QX-222 (Figure 4); furthermore, the prolongation increased with the number of Ser or Thr residues mutated to Ala at position 10’. Ko was calculated from the voltage-jump data (Figure 5). The effect of mutations on Ko was nearly
Channel
Blockers
at Nicotinic
Receptor
91
N = S/T
0 Position
6’
0 Position
10’
removed/added
Figure 5. Effects of the Mutations on the Equilibrium Constant for the Interaction between QX-222 and Channel
the
Binding AChR
The ordinate is Ko, determined from analyses like those of Figure 4 at a concentration of 0.5 NM ACh and -150 mV. The abscissa is N, the number of Ser/Thr residues changed to Ala (negative numbers) or substituted for Ala or Phe (positive numbers).
- A-*
position 6’ mutations (54% and 41% per mutation for the data at 0.5 and 1 FM, respectively). The mutations did not substantially affect the voltage dependence of Ko (Figure 46). Like our previous measurements for position 6’, the data suggest that the binding site senses 65%-75% of the membrane field. Our data do not permit a complete analysis of the interaction between QX-222 and position 2’, since the
B 22 .3 0 Y
a72’Ah%A, and eZdhdS2.A hybrids produced no agonist-induced currents. The available single-channel data (Table 2; Table 3) show that F is increased by SeriThr to Ala mutations and decreased by Gly to Ser mutations. This resembles the pattern for position 6’. However, the voltage-jump analysis (Table 4) reveals little or no change in Ko among members of the position 2’ series. aTTAhA6,
5-t
:
I -150
:
: -130
:
: -110
Voltage Figure 4. Voltage-Jump between QX-222 and
Relaxation the AChR
:
: -90
I
I
:
-70
:
:
I
-50
(mV) Analysis
of the
Interaction
(A) Comparison of a position 10’ mutant hybrid, a,7wAPlloA~S (top), the normal apyS (middle), and a position 6’ mutant hybrid, CI~~,,~&~~ (bottom). The traces depict voltage-clamp currents induced by ACh (05 PM) for a jump from +50 mV to -130 mV; passive and capacitive currents have been eliminated by subtracting the records in the absence of ACh. Voltage-clamp episodes are superimposed in the presence (* 1 and absence of QX-222, 20 PM. Calibrations, 650 nA and 50 ms. (B) Equilibrium constant Ko for QX-222 binding at various voltages, determined from the voltage-jump relaxation data.
Other
Mutations
As noted above, several position 2’ mutants failed to yield functional expression. Two other mutant subunits, YN6.A and 8AIVs, failed to yield detectable responses in any hybrid tested.
Table 4. Values at Position 2
the same as that on the blocked duration l/F, amounting to a Cfold difference over the entire series of mutations. This agrees with the observations, from the single-channel data, that the position IO’ mutations are affecting primarily the residence time l/F, although we cannot rule out small effects on the bimolecular binding rate G. Ko decreased by about 35% or 32% (data at 0.5 and 1 PM, respectively) for each Ser or Thr mutated to Ala. Figure 5 demonstrates that these effects for the position 10’ mutations are opposite but roughly equal in magnitude to those observed for the
of K4 for
Some
of the
Mutant
Hybrids
K&M) -110 76.5
mV
- 150 mV
k 7.8 (9)
20.0
+ 0.9 (7)
21.6 21.4 29.1 25.2 22.1 28.4
+ f + + i +
77.0 f 59.4 f 81.3 & 77.8 + 63.2 + 111.5 +
ACh
concentration
5.7 (IO) 8.9 (4) 7.9 (11) 4.5 (5) 4.6 (9) 25.7 (2)
was 1 WM.
1.5 (10) 1.6 (5) 2.1 (11) 1.12 (5) 1.3 (9) 4.5 (3)
Outside
i
Figure h. Molecular Model 01 the action between QX-222 and the Channel
InterOpen
The ML reggons ot each ot the five subunits (a$$) are presumed to interact equally with the QX-222 molecule, but the interact tlon is shown in two separate panels for claritv in a two-dimensional drawing.
Discussion The major observation reported here concerns the electrophysiological consequences of mutations that change polar amino acid residues (Ser and Thr) to nonpolar residues in the M2 region, which is thought to line the ion channel pore of the nicotinic AChR. The mutated receptors manifest roughly normal levels of conductance, normal Hill slopes, and (with one highly instructive exception reported by Leonard et al. [1988]) normal single-channel conductances-the hallmarks of correctly assembled vertebrate muscle AChRs. Our major observation is that these mutations at positions 6’ and IO’ have opposite effects on the blockade by QX-222: a more polar environment favors the QX-222 interaction at site 6’, but a less polar environment favors this interaction at site IO’.
Implications
for AChR Structure
Our tentative molecular model that explains these data is shown in Figure 6: Positions 6’ and 10’ are at nearly identical positions on adjacent turns of the M2 helix; the amino acid side chains project into the ion conducting pore; and these side chains interact respectively with the amino (polar) and aromatic (nonpolar) portions of the QX-222 molecule. This model presently has no basis in any threedimensional structural data at atomic resolution. It does, however, provide a self-consistent interpretation of our data, and it suggests several additional points for discussion: First, none of our receptor hybrids mutated at position 10’ showed an altered conductance, although removal of 3 Ser residues at position 6’ does produce a decreased conductance for outward currents (Leonard et al., 1988). Therefore, our model shows the channel tapered slightly to indicate that the lumen is narrower at position 6’ than at position 10’; several other authors suggest that such a taper arises, possibly because the helices are twisted (Dani, 1989b; Unwin et al., 1988). In summary, position 6’ could be the narrowest point of the channel, where “permeant ions pause at the local anesthetic binding site while
crossing the membrane” (Adams et al., 1981), as described from electrophysiological measurements on reversal potentials and streaming potentials (Dani, 198913) and from structural measurements (Toyoshima and Unwin, 1988). Second does the channel narrow still further at position 2’, as suggested in Figure 6? Because some of the crucial position 2’ hybrids failed to yield currents, the data are not available to decide this point. The slightly reduced conductance with the apA(3~iTSy6 hybrid calls for further study, perhaps with different permeant ions. The data are also inconclusive on whether QX-222 interacts with position 2’. However, for the entire series of position 2’ mutations that gave functional expression, there was barely a 2-fold change in F (Table 2) and an even smaller range in the Ko determined with macroscopic voltage-jump relaxations (Table 4). This may imply that the channel tapers too much to allow the QX-222 molecule access to the 2’ residues. Third, each mutation at either position IO’ or 6’ has roughly the same effect-a 30%-50% change in the binding of QX-222, corresponding to a free energy change of about 0.2 kcal/mol. This would correspond to a rather small binding energy, compared, for instance, with that of a hydrogen bond (-6 kcal/mol). The QX-222-receptor interaction depends in part on the total number of these low-energy polar or nonpolar interactions rather than on the identity of individual side chains at any one position; perhaps it is more accurate to say that at each turn the channel behaves like an annulus whose properties are averaged from those of the individual side chains. That the position 10’ mutations all have approximately equal effects argues against the idea that the aromatic moiety of QX-222 is intercalating specifically into the (presumably nonpolar) spaces between any two adjacent helices. It remains possible, of course, that other drugs, such as the histrionicotoxins (Spivak et al., 1982) or chlorpromazine (Giraudat et al., 1986, 1987), do interact with larger nonpolar domains on the protein. It should also be noted that our model leaves open the possibility that more lipid-soluble
anesthetics (such as the uncharged tertiary ammonium compounds) are in fact reaching this binding site through the lipid phase (Hille, 1977). Fourth, the M2 region has 2 additional positions, 4 and 12’, with substantial numbers of OH-containing amino acid residues. In our model, these side chains would point away from the channel and presumably contact other helices. What is their function? In preliminary experiments with the ynL,A mutant expressed with various other subunits, we noted no effects on channel conductance or QX-222 binding. However, systematic studies of more mutations at these positions should be undertaken. Energy of the Interaction at Position 10’ As recently pointed out by Miller (1989) in connection with the position 6’ data, the systematic changes in the dissociation rate constant, F, without changes in the forward rate, G, can be reasonably interpreted to mean that the mutations are directly affecting the energy of the complex between QX-222 and the receptor rather than an activation energy for the binding. This comment now appears relevant to the position IO’ mutations as well. This study shows that the aromatic group of the blocker interacts more strongly with the nonpolar group of Ala than with the polar groups of the naturally occurring Ser/Thr. It is natural to ask whether the OH groups of the Ser/Thr residues provide any stabilization energy at all. Although the earlier literature notes that many drugs contain polar groups separated by 5-6 A that could hydrogen bond to OH groups on adjacent turns of an a helix (Gero and Reese, 1956; Korolkovas, 1970), it has not been suggested that the aromatic group of an anesthetic molecule could make such contacts. There are, however, many examples in which an aromatic group accepts a hydrogen bond from an amino group at right angles to the plane of the ring (Burley and Petsko, 1986); the calculated energy of this interaction, 3 kcal/mol, is roughly one-half that of the more common hydrogen bonds (Levitt and Perutz, 1988). The OH groups of the Ser and Thr residues in position IO should therefore also be considered as possible hydrogen bond donors. Alternatively, it would be possible for both the 0 and H atoms of the OH group to interact with the aromatic ring. The energy would be similar to that calculated for the interaction between water and a benzene ring, about -2.5 kcal/mol (Thomas et al., 1982). The stabilizing energies considered here are more than enough to explain the 0.2 kcal/mol associated with each of our mutations; indeed, they are too high for a model in which an individual OH group contacts the aromatic ring. One is again led to the view that the blockerchannel interaction energy at each annulus is a timeaveraged value of the interactions with all five side chains. The relative efficacy of other open-channel blockers may be explained by the existence of moderate stabilizing energy from the interaction between the
OH-containing side chains at position IO’and the aromatic moiety of a blocker. For instance, QX-222 blocks the open channel with a dissociation constant of about 20 PM, whereas the tetraethylammonium ion, which presumably would lack completely the interaction at position IO’, blocks about IO-fold more weakly (Adler et al., 1979). On the other hand, one would also expect position IO to interact more strongly with a highly polar moiety. Yet decamethonium, with two quaternary ammonium groups, also binds more weakly to the channel blocking site than does QX-222 (Adams and Sakmann, 1978). Probably the charges are too far apart, and the decamethonium molecule too flexible, for favorable simultaneous interactions at positions 6’ and 10’. The unchanged voltage sensitivity of the interaction between QX-222 and the AChR (Figure 4B) implies that the mutations affect only the affinity of the binding and not its location. Generality of Local Anesthetic-Helix Contacts The available data do not allow a complete description of the interaction between local anesthetics and all membrane channels. In addition to the pure openchannel blockade that we have characterized here, many such drugs also interact with the closed and inactivated states of sodium channels (Hille, 1977; Hondeghem and Katzung, 1984). We made no attempt to address the data showing that local anesthetic potency is influenced both by the strength of a polarizable dipole (usually a carbonyl oxygen, as in QX-222) that lies between the aromatic group and the ionizable amino group and by substituents on the benzene ring (Ritchie and Creengard, 1966). No modern data are available to indicate whether these two effects are directly related to the interaction between the local anesthetics and the open channel. One can only speculate on the generality of the blocker-channel interaction that we have described. At present, the nicotinic receptor is the only channel for which site-directed mutagenesis has revealed the location of the residues lining the ion channel pore. There are likely to be differences in the precise nature of the blocker-channel interaction among different channels, due, for instance, to the fact that only four subunit helices may line the pore of the channel for voltage-dependent sodium channels, rather than five, as is the case for the AChR channel. The experiments that we and lmoto et al. (1988) have performed are in principle applicable to any cloned ion channel and are underway in several laboratories. Thus it may soon be known whether the conducting pore of other ion channels, both ligand-gated and voltage-gated, is also partially lined by OH side chains such as Ser, Thr, and Tyr. If so, interaction with adjacent turns of an a-helix may be a general phenomenon that underlies the blockade of cation channels by local anesthetics. Experimental
Procedures
Generation and sequencing cDNAs, oocyte injectton, and
ot the mutant elcctrophyslologicdl
AChR subuntt testing were
nearly identical to the procedures described by Leonard et al. (1988). Mutations were generated by oligonucleotide-prlmed synthesis on single-stranded uracil-containing (Kunkel et al., 1987) template DNA from M13mp19 or pBluescript (Stratagene, San Diego, CA) vectors by means of the Mutagene kit (Bio-Rad, Richmond, CA). Mismatched oligonucleotides were 15-20 nucleotides in length and contained 1 or 2 mismatches per oligomer. Mutations were screened by dideoxy sequencing. Mutated inserts that were in M13mp19 were recloned into pGEM2 or pBluescript for in vitro RNA synthesis, performed with SP6 or T7 RNA polymerase. Mixtures of RNAs encoding either normal or mutated a, B, y, or 6 subunits (2-3 ng per subunit RNA) were injected into Xenopus oocytes in various combinations. Electrophysiological measurements were performed 36-72 hr later. ACh was present at concentrations of 0.1-1.0 bM; the temperature was 12°C. Single-channel measurements were performed on both cell-attached and excised outside-out patches. Both the pipette and the bath contained 100 mM KCI, 2 mM MgCI,, 10 mM HEPES, and 10 mM EGTA (pH 7.5). Conductance was measured with voltage ramps that swept from -150 mV to +I50 mV in 400 ms; straight lines were fit separately to the limbs at positive and negative voltages. Data were sampled continuously at 44 kHz by a pulse-code modulator and stored on video tape. Data were recorded with a Dagan (Minneapolis, MN) 8900 amplifier (headstage resistor, 10 a), played back through an 8-pole Bessel filter (-3 dB point at 2-3 kHz), digitized by a Labmaster interface (Scientific Solutions, Solon, OH), and analyzed by programs in the pCLAMP series (Axon Instruments, Foster City, CA). Single-channel open times were measured from the time constants of exponential decays fit to duration distributions of channel openings. For measurements of fast closed times in the presence of QX-222, all resolvable sojourns at the baseline level (>~I00 ps) were treated as bona fide closings. Macroscopic measurements were conducted in a bath solution containing 96 mM NaCI, 2 mM KCI, 1 mM MgCI,, and 5 mM HEPES; there was no added calcium to avoid activation of the endogenous chloride channels. We utilized a two-electrode, voltage-clamp circuit (AXOCLAMP-2A, Axon Instruments) under the control of pCLAMP programs. For voltage-jump relaxations, the membrane potentialwas stepped from a holding level of -30 mV to +50 mV for 50 ms, followed by a step to a test potential between -150 and -50 mV. Time constants were fit with leastsquare routines in the pCLAMP series. Acknowledgments We thank Dr. Bertil Takmann of Astra Pharmaceuticals for gifts of QX-222; Drs. Takmann and R. A. North for discussion; and W. J. Goddard for the use of computer modeling facilities. This work was supported by grants from the National Institutes of Health (NS-11756) and from the Muscular Dystrophy Association and by postdoctoral fellowships to R. J. L. (NS-8083) and P. C. (Bourse Lavoiser and Fondation pour la Recherche Medicale). Received
August
17, 1989;
revised
August
29, 1989
W., Dwyer, T. M., and Hille, B. (1981). by permeant cations in frog skeletal 78, 593-615.
Adams, P. R. (1977). Voltage jump analysis frog end-plate. J. Physiol. 268, 291-318.
of procaine
Adams, P. R., and Sakmann, B. (1978). Decamethonium opens and blocks endplate channels. Proc. Natl. Acad. 75, 2994-2998. Adler, M., Oliviera, A. C., Albuquerque, E. X., Mansour, and Eldefrawi,A. T. (1979). Reaction of tetraethylammonium the open and closed conformations of the acetylcholine tor ionic channel complex. J. Gen. Physiol. 74, 129-152.
Block mus-
action
at
interactions
Dani, J. A. (1989a). Site-directed mutagenesis and single-channel currents define the ionic channel of the nicotinic acetylcholinereceptor. Trends Neurosci. 72, 125-128. Dani, J. A. (198913). Open channel structure and ion binding sites of the nicotinic acetylcholine receptor. 1. Neurosci. 9, 884-892. Gero, A., and Reese, V. J. (1956). Science 123, 100.
Chemical
model
of drug
action.
Giraudat, J., Dennis, M., Heidmann, T.. Chang, J. Y., and Changeux, 1. P (1986). Structure of the high-affinity binding site for noncompetitive blockers of the acetylcholine receptor: serine262 of the delta subunit is labeled by [‘HIchlorpromazine. Proc. Natl. Acad. Sci. USA 83, 2719-2723. Ciraudat, I., Dennis, M., Heidmann, T., Haumont, P.. Lederer, F., and Changeux, I. P. (1987). Structure of the high-affinity binding site for noncompetitive blockers of the acetylcholine receptor: [3H]chlorpromazine labels homologous residues in the s and d chains. Biochemistry 26, 2410-2418. Hille, B. (1977). Local anesthetics: pathways for the drug-receptor 497-515. Hille, land,
B. (1984). Ionic Channels MA: Sinauer Associates).
hydrophilic reaction. In Excttable
and hydrophobic J. Cen. Physiol. 69, Membranes
(Sunder-
Hondeghem, L. M., and Katzung, B. G. (1984). Antiarrhythmic agents: the modulated receptor mechanism of action of sodium and calcium channel-blocking drugs. Annu. Rev. Pharmacol. Toxicol. 24, 387-423. Hucho, F., Oberthur. W., and Lottspeich, F. (1986). The ion channel of the nicotinic acetylcholine receptor is formed by the homologous helices M II of the receptor subunits. FEBS Lett. 205, 137-142. Imoto, K., Methfessel, C., Sakmann. B., Mishina, M., Mori, Y., Konno, T., Fukuda, K., Kurasaki, M., Bujo, H., Fujita, Y., and Numa, S. (1986). Location of a delta-subunit region determinlng ion-transport through the acetylcholine-receptor channel. Nature 324, 670-674. Imoto, K., Busch, C., Sakmann, B., Mishina, M., Konno, T., Nakai, J., Bujo, H., Mori, Y., Fukuda, K., and Numa, S. (1988). Rings of negatively charged amino acids determine the acetylcholine receptor channel conductance. Nature 335, 645-648. Korolkovas, Background
A. (1970). Essentials of Molecular Pharmacology: for Drug Design (New York: Wiley-Interscience).
Kunkel, T. A., Roberts, efficient site-specific tion. Meth. Enzymol.
I. D., and Zakour, R. A. (1987). Rapid and mutagenesis without phenotypic selec754, 367-383.
Leonard, R. J., Labarca, C., Charnet, F’., Davidson, N., and Lester, H. A. (1988). Evidence that the M2 membrane-spanning region lines the ion channel pore of the nicotinic receptor. Science242. 1578-l 581. Levitt, M., and Perutz, M. F. (1988). Aromatic rings gen bond acceptors. J. Mol. Biol. 207, 751-754. Miller, C. (1989). Genetic manipulation approach to structure and mechanism.
References Adams, D. J., Nonner, of endplate channels cle. J. Gen. Physiol.
Burley, S. K., and Petsko, G. (1986). Amino-aromatic in proteins. FEBS Lett. 203, 139-142.
act as hydro-
of ion channels: a new Neuron 2. 1195-1205.
Neher, E. (1983). The charge carried by single-channel currents of rat cultured muscle cells In the presence of local anesthetic\. J. Physiol. 339, 663-678. Neher, E., and Steinbach, J. H. (1978). Local anaesthetics siently block currents through single acetylcholine-receptor channels. J. Physiol. 277, 153-176.
tran-
both Sci. USA
Pauling, L., Corey, R. B., and Bransom, H. R. (1951). The structure of proteins: two hydrogen-bonded helical configurations of the polypeptide chain. Proc. Natl. Acad. SCI. USA 3i: 205-211.
N. A., with recep-
Ritchie, 1. M., and Greene, N. M. (1985). Local anesthetics. In The Pharmacological Basis of Therapeutics, A. C. Gilman, L. 5. Gilman, T. W. Rail, and F. Murad, eds. (New York: Macmillan), pp. 302-321.
C-hannel 95
Blockers
at Nrcotrnrc
Rrtchie, of local
J. M., and anesthetics.
Greengard, Annu.
Spivak,
C. E., Maleque,
Receptor
P (1966). On the mode of actron Rev. Pharmacol. 6, 405-430.
M. A., Oliviera,
A. C., Masukawa,
L. M.,
Tokuyama, T., Daly, J. W., and Albuquerque, E. X. (1982). Actrons of the histrionicotoxins at the ion channel of the nicotinrc acetylcholine receptor and at the voltage-sensitive ion channels of muscle membranes. Mol. Pharmacol. 21, 351-361. Thomas, K. A., Smrth, G. M., Thomas, T. B., and Feldman, (1982). Electronic distributions with protein phenylalanine matrc rings are reflected by the three-dimensional oxygen environments. Proc Natl. Acad. Sci. USA 79, 4843-4847.
R. J. aroring
Toyoshima, C., and Unwin, N. (1988). Ion channel of acetylcholine-receptor reconstructed from images of postsynaptic membranes. Nature 336, 247-250. Unwin, N., Toyoshima, C., and Kubalek, E. (1988). Arrangement of the acetylcholine-receptor subunits in the restrng and desensitized states, determined by cryoelectron microscopy of c.rystallized Torpedo postsynaptrc membranes. J, Cell Biol. 707. 11231138.
Vol. 2, 87-95,
January,
1990, Copyright
0 1990 by Cell Press
An Open-Channel Blocker Interacts with Adjacent Turns of a-Helices in the Nicotinic Acetylcholine Receptor Pierre Charnet: Cesar labarca: Reid J. Leonard:+ Nancy J. Vogelaar,*S Linda Czyzyk: Annie Couin: Norman Davidson: and Henry A. Lester* *Division of Biology *Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena, California 91125
Summary The binding site for an open-channel blocker, QX-222, at mouse muscle nicotinic acetylcholine receptors was probed using site-directed mutagenesis, oocyte expression, and electrophysiological analysis. The proposed cytoplasmic end of the M2 transmembrane helix is termed position 1’. At position lO’(aS252, bT263, yA261, 6A266), Ala residues yield stronger and longer binding of QX-222 than Ser or Thr residues. These effects are opposite and roughly equal (30%~50% per mutation) to previously reported effects at position 6’. The polar end of an anesthetic molecule seems to bind to the position 6’ OH groups, which provide a water-like region; the nonpolar moiety is near position 10’ and binds more strongly in a nonpolar environment. Interactions with adjacent OH-rich turns of an amphiphilic helix may explain the widespread blocking effects of local anesthetics at the conduction pore of ion channels. Introduction The first synthetic local anesthetic, procaine, was introduced by Einhorn in 1905 as an alternative to cocaine, which has additional psychotropic effects. These and newer local anesthetics share two structural features: an ionizable amino group and an aromatic group (Ritchie and Greengard, 1966; Hille, 1984; Ritchie and Greene, 1985); the two moieties are separated by 5-6 A. This distance corresponds well with the repeat distance between adjacent turns of an a-helix, about 5.4 A (Pauling et al., 1951). One hypothesis is therefore that local anesthetics interact with adjacent turns of a-helices and that such interactions help to account for the widespread blockade of ion channels by local anesthetics and their analogs. This hypothesis is now subject to decisive tests with the nicotinic acetylcholine receptor (AChR). Increasing evidence suggests that the M2 membrane-spanning region from each of the five subunits (a&8) lines the ion channel pore (reviewed by Miller, 1989; Dani, 1989a). First, experiments with photoaffinity
+ Present address: Department of Membrane Biochemistry and Biophysics, Merck Sharp and Dohme Research Laboratories, Rahway, New Jersey 07065. §Present address: Department of Chemistry, Princeton University, Princeton, New Jersey 08554.
labeling showed that chlorpromazine binds to amino acid residues in the M2 membrane-spanning region of the a, B, and 6 subunits (Ciraudat et al., 1986, 1987; Hucho et al., 1986). Second, initial experiments with chimeric mouse-calf S subunits showed that channel conductance was influenced by a region including the M2 region and the M2/M3 hinge (Imoto et al., 1986). Third, additional experiments with site-directed mutagenesis revealed that channel conductance is controlled by rings of negative charge, contributed from each of the four subunits, at each end of the M2 helix (Imoto et al., 1988). Fourth, mutations at a polar (Ser-rich) site within the M2 helix affected the residence time and equilibrium binding affinity for the prototype open-channel blocker, QX-222, and caused rectification in the channel conductance (Leonard et al., 1988). This paper describes evidence supporting the view that QX-222 (Figure IA), a quaternary ammonium derivative of lidocaine, does indeed interact with adjacent turns of the M2 helix of mouse muscle AChR. We have continued to study the binding of QX-222 to the receptor by employing site-directed mutagenesis, oocyte expression, and electrophysiological analysis. The region we have studied is shown in Figure IB. Our terminology, following that of Miller (1989), assigns the number 1’ to the N-terminus of the M2 region. Position 6’ corresponds to the “inner polar site” (IPS) of Leonard et al. (1988). At this site, Leonard and coworkers (1988) reported that the binding of QX-222 became weaker as polar amino acids (Ser) were replaced by nonpolar residues (Ala). We now show that similar mutations at position 10’ have an opposite effect: replacement of polar (Ser or Thr) residues by nonpolar residues actually increases the binding of QX-222. The straightforward interpretation of these data is that the charged, amino moiety of QX-222 is binding at position 6’ and that the aromatic, nonpolar end is binding at position IO’. We also present less extensive data on the effect of mutations at position 2’and at several other locations. These results support the general view that the M2 domain of each subunit forms an a-helix with a polar stripe consisting of residues 6’ and 10’ (and possibly also 2’), that these polar stripes line the channel, and that this region in the channel also forms the binding site for a class of open-channel blockers. Results Table 1 summarizes the equilibrium and kinetic properties of the ACh-induced currents for both normal and mutated AChRs. Most of the position 6’ mutant clones utilized in the present study were the same as those reported previously by Leonard et al. (1988); we have retested these and repeated the previous macroscopic measurements at 0.5 PM ACh, a concentration
QX-222
A
Position 10’ Mutations Do Not Affect Single-Channel Conductance There were no changes in single-channel conductance for any of the position IO’ mutations, even the as10.ABT,0.AY8 hybrid, in which 3 OH-containing residues were replaced by Ala. This result contrasts with the position 6’ series in which replacement of 3 Ser by Ala residues (aS6.Aby8Sb.A) produced a pronounced rectification of the channel currents (Leonard et al., 1988). Our study of the position 2’ mutants was limited by the fact that not all combinations yielded functional expression. In particular, the combinations with the most Ser (or Thr) to Ala substitutions, an&+,&i, aT2.Afi&2A, and aR’A(jYT2.A8s2.A, produced no agonistinduced currents. Each of the mutant subunits in the position 2’ series did display agonist-induced currents in combination with normal subunits (Table I), indicating that there were no major structural problems with these mutants. Interestingly, the single-channel conductance for the a&&Y8 hybrid was 20% lower than normal. This result was reproducible and significant.
CH3
CH3
I
NH
dH2
CH3-
rii+-m
CH3
CH3
B
QX-222 Residence Time Increases with of Ser Residues at Position 10
6’
Figurel. Structure of QX-222 and AChR Subunits in the M2 Region Position (1988).
6’was
termed
the Inner
polar
18’
14’
10'
Sequences
of
the
site (IPS) by Leonard
Mouse
Removal
The kinetic analysis of the single-channel data are exemplified by Figure 2 and summarized in Table 2. The position 10’ mutations affected the kinetics of QX-222 blockade. The fast closed times increase for mutations in which Ser or Thr residues are changed to Ala residues, and they decrease for a mutation in which an existing Ala residue on they subunit is replaced by Ser. The model we use for the interaction between QX-222 and the AChR is a simplified single binding scheme:
et al.
R r~owi-Ro~en
WQI +=
Wo,wn
Q)bkxked
(1)
F
that gave optimal results for the position 10’ mutations. The results were quite similar to those reported previously. We shall therefore concentrate on the description of the position 10’ mutations. All of the mutant hybrids that yielded functional receptors also displayed Hill coefficients approaching 2 in the dose-response studies. A Hill coefficient near 2 characterizes all known native AChRs and suggests that the open state of the channel is much more likely to be associated with the presence of two bound agonist molecules than with a single bound agonist molecule. Nearly all of the mutations had observable effects on the single-channel and macroscopic kinetics of the response to AChR. A later publication will describe and systematize these results; the present paper deals only with the effects of QX-222 on these kinetics.
in which R = AChR and Q = blocker (Adams, 1977; Neher and Steinbach, 1978). In this scheme, the channel is blocked when QX-222 is bound within it and the average fast closed time is simply equal to l/F. Figure 3 presents a graphic display of these trends by plotting F as a function of the number of Ser/Thr residues replaced by Ala. The linear fit in Figure 3 represents a change of 29% for each residue replaced at position IO’, with an opposite, but roughly as great (35%), change for the position 6’ mutations. Scheme 1 can also be used to analyze the secondorder bimolecular rate constant, G, for the binding of QX-222 to the AChR. A full analysis would call for measurements at varying QX-222 concentrations; we have not pursued such a study, but instead have confined our measurements to a single concentration, [QX-2223 = 20 PM. We use the relation
G = A(lit)l[QX-2221
(2)
Channel 89
Table
Blockers
at Nicotlnic
1. Characterization
Receptor
of the Single
N 4YS
0
Normal
and
Mutant
Hybrid
AChRs
Expressed
in Oocytes
Channel
Conductance
(pS)
Inward
Outward
Open
Macroscopic
36 k 1 (4)
24 + 1 (4)
15.3 * 1.2 (8)
1010 k 121 (23)
51.9
+ 2 (17)
1.6 i 0.2 (4)
* k + f +
(5) (3) (3) (6) (3)
403 1073 1630 1029 1087
* * + k +
75 (11) 267 (9) 314 (12) 381 (9) 322 (IO)
44.5 46.5 41.7 54.5 59.3
* k + * +
3.3 1.2 2.9 3.0 3.8
(11) (9) (9) (9) (IO)
1.9 1.7 2.3 1.7 1.7
* 2 + * (I)
0.1 0.1 0.1 0.02
(3) (5) (2) (4)
0.3 0.1 0.1 0.1
(5) (14) (7) (7)
Time
(ms)
Current
(nA)
Hill
T (ms)
Coefficient
Position 10 &*,,.d ahAYS a5dvA103 a5dY~ a\,nxPrKrAY~
+I -1 -1 -2 - 3
39 35 33 34 37
f * + k *
2 1 3 1 1
(6) (7) (4) (8) (7)
22 27 26 27 27
* + (I) f k
2 (4) 1 (7) 1 (8) 1 (7)
7.9 17.6 33.8 15.3 17.9
Position 6 alh,sy~ aWShA adM aYJA e YLA
+I -1 -2 -3
33 32 36 33
+ f f *
1 1 1 2
(6) (6) (6) (7)
22 21 24 13
+ + + f
1 1 2 1
(6) (6) (6) (5)
5.0 28.4 14.1 20.6
* 0.7 (3) (1) k 4.2 (3) -t 3.0 (3)
350 954 770 904
f +_ k +
62 (13) 152 (23) 144 (16) 199 (15)
16.6 66.9 42.2 46.9
* + k i
1.2 2.7 2.8 3.5
(9) (23) (16) (13)
1.6 1.7 1.7 1.7
fi + * +
Position 2 c&&j ah* AS aB-& A ar2 dkrt~G aT2 APY~ dYn A&Z A
+I -1 -1 -1 -2 -2
32 36 33 30 ND 36
k + + f
1 1 1 1
(14) (4) (9) (9)
22 + 1 23*1(4) 20 f 1 20 f 1 ND 24 + 1
(13)
15.1 ND 28.9 22.3 ND 27.9
+ 4.4 (2)
587 738 926 553 468 657
f * f k t *
87 (2) 309 (2) 313 (IO) 230 (5)" 139 (3) 54 (2)
39.9 87.6 73.3 55.9 65.7 101.3
+ f * k + k
0.9 10 3.1 3.1 2.7 5.2
(2) (3) (IO) (5) (3) (3)
1.5 1.8 1.7 ND ND 1.9
* 0.1 (2) * 0.1 (2) i 0.1 (8)
f 1 (4)
(6) (9) (4)
0.9 1.1 6.7 1.9 5.2
k 9.0 (2) (I) (1)
* 0.2 (2)
The parameter N represents the number of Ser or Thr residues relative to the normal ae$ receptor. Note that only the a~6.AP&,A receptors display a selective decrease in conductance for outward currents. Single-channel open times and agonist-Induced currents are given for -150 mV. Time constants T of voltage-jump relaxations are given for jumps from +50 mV to - 150 mV at 0.5 PM ACh. Data are mean k SEM (number of observations). d 1 PM ACh.
in which t is the mean open time in the absence or presence of QX-222 (Neher and Steinbach, 1978). Values for G are given in Table 3. The important point is that the values for all of the mutants cluster within
*30% of the average value (1.9 x 10’ M-lsC at -100 mV and 2.6 x IO7 M-‘s-l at -150 mV). This indicates that the mutations we have performed change the equilibrium dissociation constant F/G only because Figure AChR
2. Single-Channel Mutants
Analysis
of the
Left column, typical single-channel traces in the presence of ACh alone (1 PM) at -150 mV. Middle column, traces in the presence of ACh plus QX-222 (20 PM). Right column, frequency distribution histograms for the fastest detectable closed times in the presence of ACh plus QX-222. The time constant of these distributions corresponds to the average residence time for QX-222 and equals l/F. The top row gives the data for the positlon 10 mutant hybrid, c~,~~,$~,~,~~; the middle row gives the data for the wild-type receptor; and the bottom row gives the data for the position 6’ mutant hybrid, a&y&,,,. Calibration bars: 3 PA; 25 ms for ACh alone; 12 ms for ACh plus QX-222.
Closed
Time (ms)
Table 2. Evaluation the Various Mutant
of the Dissociation Rate Constant, Hybrids at Two Different Voltages
F, for
Table 3. Evaluation of the Association the Interaction between QX-222 and Different Voltages
F(s-‘) - 100 mV
C(10’
150 mV
1400 f 100 (7)
390 k 13 (6)
1890 781 980 705 391
532 348 366 281 176
+ f (I) k +
9 (4) 24 (2)
Position a~yAd
aPTlrAYfi 8 (5) 3 (5)
k + + k
184 (3) 123 (4) 68 (4) 208 (4)
aSl~AhA,d a\lwAh6
951 ND 1557 1207 ND 1745
+ 74 (2)
298 625 760 838
-f + f k
24 (3) 1 (2) 46 (2) 12 (3)
294 ND 621 452 ND 502
+ 20 (2)
Position
Data
are mean
(1) (1)
@-b&A aSbAbY
(1)
f SEM (number
k 69 (2)
of observations).
Data
they change surements.
F, at least
within
the
limits
of our
mea-
Macroscopic Data Also Show Stronger Binding with Removal of Ser Residues at Position 10 The rationale for the analysis of macroscopic voltage-jump relaxations has been given by Leonard et al. (1988). Such relaxations are most readily interpreted when the agonist concentration is restricted to a range giving only small fractional activation of receptors. Under such conditions, a pure open-channel
0
\
Position
h
Position 2 ulhtY~ ah A6 &b A adGL ,YS a&W abT~A~~~ A
+ 68 (2) (1)
6’
are
1.99 f 0.08 (3)
2.76
f 0.16 (3)
2.55 1.95 ND 1.88 1.90
i 0.19 (2) (I)
f 0.42 (3) + 0.13 (2)
+ 0.13 (3) * 0.54 (3)
2.63 2.72 ND 2.80 2.28
2.29 1.96 1.41 1.52
f ? f +
1.21 t 0.45 (3) ND ND 1.75 (I)
+ 0.38 (3) + 0.20 (3)
6
alkb5y6
as6 AP&
- 150 mV
IO
aSl~AhvA~6
924 1839 1993 2896
‘s ‘1
100 mV aby6
(I) (1) (1) * 100 (3) (I)
M
Rate Constant, C, for the Receptor, at Two
mean
0.11 0.56 0.10 0.36
(3) (2) (2) (2)
1.77 + 0.16 (2) ND 1.92 (I) 1.29 (I) ND ND + SEM (number
2.21 ND 2.18 2.17 ND s.28
(I) i 0.19 (2) (1) (I)
of determinations).
blocker is expected to produce little diminution of the equilibrium agonist-induced currents. We satisfied this criterion by using 0.5 PM ACh; the blockade by QX-222 at equilibrium amounted to at most 17% at -150 mV or 10% at -110 mV and was usually much less. Voltage-jump relaxations (Figure 4A) generally have a single exponential component in the presence of agonist alone; but in the additional presence of an open-channel blocker, one expects two exponential components. The faster “inverse relaxation” (Adams, 1977) is too rapid to be resolved in oocytes with the two-microelectrode voltage clamp. The slower component was, however, easily resolved in our experiments. Its time constant is increased in the presence of QX-222:
1000
TO = ~(1 + [QX-222]/Ko) 7 z L
100
-4
-3
-2
-1
N = S/T Figure 3. Dissociation QX-222 and the AChR, 6’ and IO
0
1
2
3
4
removed/added
Rate Constant of the Complex between versus the Number of SerIThr at Positions
The abscissa is N, the number of Ser/Thr Ala (negative numbers) or substituted for numbers). The ordinate is the dissociation mined from analyses like those of Figure [ACh], 1 fiM.
residues changed to Ala or Phe (positive rate constant F, deter2. Voltage, -150 mV;
(3)
in which TV and T are the time constants in the presence and absence of QX-222, respectively, and Ko = F/G. The increased time constant may be explained intuitively by pointing out that this time constant essentially equals the average duration of the burst of openings associated with each activation of the channel. When an open-channel blocker is present, this burst is lengthened to include the blocked times as well (Neher and Steinbach, 1978; Neher, 1983). As predicted by Equation 3, voltage-jump relaxations were prolonged in the presence of QX-222 (Figure 4); furthermore, the prolongation increased with the number of Ser or Thr residues mutated to Ala at position 10’. Ko was calculated from the voltage-jump data (Figure 5). The effect of mutations on Ko was nearly
Channel
Blockers
at Nicotinic
Receptor
91
N = S/T
0 Position
6’
0 Position
10’
removed/added
Figure 5. Effects of the Mutations on the Equilibrium Constant for the Interaction between QX-222 and Channel
the
Binding AChR
The ordinate is Ko, determined from analyses like those of Figure 4 at a concentration of 0.5 NM ACh and -150 mV. The abscissa is N, the number of Ser/Thr residues changed to Ala (negative numbers) or substituted for Ala or Phe (positive numbers).
- A-*
position 6’ mutations (54% and 41% per mutation for the data at 0.5 and 1 FM, respectively). The mutations did not substantially affect the voltage dependence of Ko (Figure 46). Like our previous measurements for position 6’, the data suggest that the binding site senses 65%-75% of the membrane field. Our data do not permit a complete analysis of the interaction between QX-222 and position 2’, since the
B 22 .3 0 Y
a72’Ah%A, and eZdhdS2.A hybrids produced no agonist-induced currents. The available single-channel data (Table 2; Table 3) show that F is increased by SeriThr to Ala mutations and decreased by Gly to Ser mutations. This resembles the pattern for position 6’. However, the voltage-jump analysis (Table 4) reveals little or no change in Ko among members of the position 2’ series. aTTAhA6,
5-t
:
I -150
:
: -130
:
: -110
Voltage Figure 4. Voltage-Jump between QX-222 and
Relaxation the AChR
:
: -90
I
I
:
-70
:
:
I
-50
(mV) Analysis
of the
Interaction
(A) Comparison of a position 10’ mutant hybrid, a,7wAPlloA~S (top), the normal apyS (middle), and a position 6’ mutant hybrid, CI~~,,~&~~ (bottom). The traces depict voltage-clamp currents induced by ACh (05 PM) for a jump from +50 mV to -130 mV; passive and capacitive currents have been eliminated by subtracting the records in the absence of ACh. Voltage-clamp episodes are superimposed in the presence (* 1 and absence of QX-222, 20 PM. Calibrations, 650 nA and 50 ms. (B) Equilibrium constant Ko for QX-222 binding at various voltages, determined from the voltage-jump relaxation data.
Other
Mutations
As noted above, several position 2’ mutants failed to yield functional expression. Two other mutant subunits, YN6.A and 8AIVs, failed to yield detectable responses in any hybrid tested.
Table 4. Values at Position 2
the same as that on the blocked duration l/F, amounting to a Cfold difference over the entire series of mutations. This agrees with the observations, from the single-channel data, that the position IO’ mutations are affecting primarily the residence time l/F, although we cannot rule out small effects on the bimolecular binding rate G. Ko decreased by about 35% or 32% (data at 0.5 and 1 PM, respectively) for each Ser or Thr mutated to Ala. Figure 5 demonstrates that these effects for the position 10’ mutations are opposite but roughly equal in magnitude to those observed for the
of K4 for
Some
of the
Mutant
Hybrids
K&M) -110 76.5
mV
- 150 mV
k 7.8 (9)
20.0
+ 0.9 (7)
21.6 21.4 29.1 25.2 22.1 28.4
+ f + + i +
77.0 f 59.4 f 81.3 & 77.8 + 63.2 + 111.5 +
ACh
concentration
5.7 (IO) 8.9 (4) 7.9 (11) 4.5 (5) 4.6 (9) 25.7 (2)
was 1 WM.
1.5 (10) 1.6 (5) 2.1 (11) 1.12 (5) 1.3 (9) 4.5 (3)
Outside
i
Figure h. Molecular Model 01 the action between QX-222 and the Channel
InterOpen
The ML reggons ot each ot the five subunits (a$$) are presumed to interact equally with the QX-222 molecule, but the interact tlon is shown in two separate panels for claritv in a two-dimensional drawing.
Discussion The major observation reported here concerns the electrophysiological consequences of mutations that change polar amino acid residues (Ser and Thr) to nonpolar residues in the M2 region, which is thought to line the ion channel pore of the nicotinic AChR. The mutated receptors manifest roughly normal levels of conductance, normal Hill slopes, and (with one highly instructive exception reported by Leonard et al. [1988]) normal single-channel conductances-the hallmarks of correctly assembled vertebrate muscle AChRs. Our major observation is that these mutations at positions 6’ and IO’ have opposite effects on the blockade by QX-222: a more polar environment favors the QX-222 interaction at site 6’, but a less polar environment favors this interaction at site IO’.
Implications
for AChR Structure
Our tentative molecular model that explains these data is shown in Figure 6: Positions 6’ and 10’ are at nearly identical positions on adjacent turns of the M2 helix; the amino acid side chains project into the ion conducting pore; and these side chains interact respectively with the amino (polar) and aromatic (nonpolar) portions of the QX-222 molecule. This model presently has no basis in any threedimensional structural data at atomic resolution. It does, however, provide a self-consistent interpretation of our data, and it suggests several additional points for discussion: First, none of our receptor hybrids mutated at position 10’ showed an altered conductance, although removal of 3 Ser residues at position 6’ does produce a decreased conductance for outward currents (Leonard et al., 1988). Therefore, our model shows the channel tapered slightly to indicate that the lumen is narrower at position 6’ than at position 10’; several other authors suggest that such a taper arises, possibly because the helices are twisted (Dani, 1989b; Unwin et al., 1988). In summary, position 6’ could be the narrowest point of the channel, where “permeant ions pause at the local anesthetic binding site while
crossing the membrane” (Adams et al., 1981), as described from electrophysiological measurements on reversal potentials and streaming potentials (Dani, 198913) and from structural measurements (Toyoshima and Unwin, 1988). Second does the channel narrow still further at position 2’, as suggested in Figure 6? Because some of the crucial position 2’ hybrids failed to yield currents, the data are not available to decide this point. The slightly reduced conductance with the apA(3~iTSy6 hybrid calls for further study, perhaps with different permeant ions. The data are also inconclusive on whether QX-222 interacts with position 2’. However, for the entire series of position 2’ mutations that gave functional expression, there was barely a 2-fold change in F (Table 2) and an even smaller range in the Ko determined with macroscopic voltage-jump relaxations (Table 4). This may imply that the channel tapers too much to allow the QX-222 molecule access to the 2’ residues. Third, each mutation at either position IO’ or 6’ has roughly the same effect-a 30%-50% change in the binding of QX-222, corresponding to a free energy change of about 0.2 kcal/mol. This would correspond to a rather small binding energy, compared, for instance, with that of a hydrogen bond (-6 kcal/mol). The QX-222-receptor interaction depends in part on the total number of these low-energy polar or nonpolar interactions rather than on the identity of individual side chains at any one position; perhaps it is more accurate to say that at each turn the channel behaves like an annulus whose properties are averaged from those of the individual side chains. That the position 10’ mutations all have approximately equal effects argues against the idea that the aromatic moiety of QX-222 is intercalating specifically into the (presumably nonpolar) spaces between any two adjacent helices. It remains possible, of course, that other drugs, such as the histrionicotoxins (Spivak et al., 1982) or chlorpromazine (Giraudat et al., 1986, 1987), do interact with larger nonpolar domains on the protein. It should also be noted that our model leaves open the possibility that more lipid-soluble
anesthetics (such as the uncharged tertiary ammonium compounds) are in fact reaching this binding site through the lipid phase (Hille, 1977). Fourth, the M2 region has 2 additional positions, 4 and 12’, with substantial numbers of OH-containing amino acid residues. In our model, these side chains would point away from the channel and presumably contact other helices. What is their function? In preliminary experiments with the ynL,A mutant expressed with various other subunits, we noted no effects on channel conductance or QX-222 binding. However, systematic studies of more mutations at these positions should be undertaken. Energy of the Interaction at Position 10’ As recently pointed out by Miller (1989) in connection with the position 6’ data, the systematic changes in the dissociation rate constant, F, without changes in the forward rate, G, can be reasonably interpreted to mean that the mutations are directly affecting the energy of the complex between QX-222 and the receptor rather than an activation energy for the binding. This comment now appears relevant to the position IO’ mutations as well. This study shows that the aromatic group of the blocker interacts more strongly with the nonpolar group of Ala than with the polar groups of the naturally occurring Ser/Thr. It is natural to ask whether the OH groups of the Ser/Thr residues provide any stabilization energy at all. Although the earlier literature notes that many drugs contain polar groups separated by 5-6 A that could hydrogen bond to OH groups on adjacent turns of an a helix (Gero and Reese, 1956; Korolkovas, 1970), it has not been suggested that the aromatic group of an anesthetic molecule could make such contacts. There are, however, many examples in which an aromatic group accepts a hydrogen bond from an amino group at right angles to the plane of the ring (Burley and Petsko, 1986); the calculated energy of this interaction, 3 kcal/mol, is roughly one-half that of the more common hydrogen bonds (Levitt and Perutz, 1988). The OH groups of the Ser and Thr residues in position IO should therefore also be considered as possible hydrogen bond donors. Alternatively, it would be possible for both the 0 and H atoms of the OH group to interact with the aromatic ring. The energy would be similar to that calculated for the interaction between water and a benzene ring, about -2.5 kcal/mol (Thomas et al., 1982). The stabilizing energies considered here are more than enough to explain the 0.2 kcal/mol associated with each of our mutations; indeed, they are too high for a model in which an individual OH group contacts the aromatic ring. One is again led to the view that the blockerchannel interaction energy at each annulus is a timeaveraged value of the interactions with all five side chains. The relative efficacy of other open-channel blockers may be explained by the existence of moderate stabilizing energy from the interaction between the
OH-containing side chains at position IO’and the aromatic moiety of a blocker. For instance, QX-222 blocks the open channel with a dissociation constant of about 20 PM, whereas the tetraethylammonium ion, which presumably would lack completely the interaction at position IO’, blocks about IO-fold more weakly (Adler et al., 1979). On the other hand, one would also expect position IO to interact more strongly with a highly polar moiety. Yet decamethonium, with two quaternary ammonium groups, also binds more weakly to the channel blocking site than does QX-222 (Adams and Sakmann, 1978). Probably the charges are too far apart, and the decamethonium molecule too flexible, for favorable simultaneous interactions at positions 6’ and 10’. The unchanged voltage sensitivity of the interaction between QX-222 and the AChR (Figure 4B) implies that the mutations affect only the affinity of the binding and not its location. Generality of Local Anesthetic-Helix Contacts The available data do not allow a complete description of the interaction between local anesthetics and all membrane channels. In addition to the pure openchannel blockade that we have characterized here, many such drugs also interact with the closed and inactivated states of sodium channels (Hille, 1977; Hondeghem and Katzung, 1984). We made no attempt to address the data showing that local anesthetic potency is influenced both by the strength of a polarizable dipole (usually a carbonyl oxygen, as in QX-222) that lies between the aromatic group and the ionizable amino group and by substituents on the benzene ring (Ritchie and Creengard, 1966). No modern data are available to indicate whether these two effects are directly related to the interaction between the local anesthetics and the open channel. One can only speculate on the generality of the blocker-channel interaction that we have described. At present, the nicotinic receptor is the only channel for which site-directed mutagenesis has revealed the location of the residues lining the ion channel pore. There are likely to be differences in the precise nature of the blocker-channel interaction among different channels, due, for instance, to the fact that only four subunit helices may line the pore of the channel for voltage-dependent sodium channels, rather than five, as is the case for the AChR channel. The experiments that we and lmoto et al. (1988) have performed are in principle applicable to any cloned ion channel and are underway in several laboratories. Thus it may soon be known whether the conducting pore of other ion channels, both ligand-gated and voltage-gated, is also partially lined by OH side chains such as Ser, Thr, and Tyr. If so, interaction with adjacent turns of an a-helix may be a general phenomenon that underlies the blockade of cation channels by local anesthetics. Experimental
Procedures
Generation and sequencing cDNAs, oocyte injectton, and
ot the mutant elcctrophyslologicdl
AChR subuntt testing were
nearly identical to the procedures described by Leonard et al. (1988). Mutations were generated by oligonucleotide-prlmed synthesis on single-stranded uracil-containing (Kunkel et al., 1987) template DNA from M13mp19 or pBluescript (Stratagene, San Diego, CA) vectors by means of the Mutagene kit (Bio-Rad, Richmond, CA). Mismatched oligonucleotides were 15-20 nucleotides in length and contained 1 or 2 mismatches per oligomer. Mutations were screened by dideoxy sequencing. Mutated inserts that were in M13mp19 were recloned into pGEM2 or pBluescript for in vitro RNA synthesis, performed with SP6 or T7 RNA polymerase. Mixtures of RNAs encoding either normal or mutated a, B, y, or 6 subunits (2-3 ng per subunit RNA) were injected into Xenopus oocytes in various combinations. Electrophysiological measurements were performed 36-72 hr later. ACh was present at concentrations of 0.1-1.0 bM; the temperature was 12°C. Single-channel measurements were performed on both cell-attached and excised outside-out patches. Both the pipette and the bath contained 100 mM KCI, 2 mM MgCI,, 10 mM HEPES, and 10 mM EGTA (pH 7.5). Conductance was measured with voltage ramps that swept from -150 mV to +I50 mV in 400 ms; straight lines were fit separately to the limbs at positive and negative voltages. Data were sampled continuously at 44 kHz by a pulse-code modulator and stored on video tape. Data were recorded with a Dagan (Minneapolis, MN) 8900 amplifier (headstage resistor, 10 a), played back through an 8-pole Bessel filter (-3 dB point at 2-3 kHz), digitized by a Labmaster interface (Scientific Solutions, Solon, OH), and analyzed by programs in the pCLAMP series (Axon Instruments, Foster City, CA). Single-channel open times were measured from the time constants of exponential decays fit to duration distributions of channel openings. For measurements of fast closed times in the presence of QX-222, all resolvable sojourns at the baseline level (>~I00 ps) were treated as bona fide closings. Macroscopic measurements were conducted in a bath solution containing 96 mM NaCI, 2 mM KCI, 1 mM MgCI,, and 5 mM HEPES; there was no added calcium to avoid activation of the endogenous chloride channels. We utilized a two-electrode, voltage-clamp circuit (AXOCLAMP-2A, Axon Instruments) under the control of pCLAMP programs. For voltage-jump relaxations, the membrane potentialwas stepped from a holding level of -30 mV to +50 mV for 50 ms, followed by a step to a test potential between -150 and -50 mV. Time constants were fit with leastsquare routines in the pCLAMP series. Acknowledgments We thank Dr. Bertil Takmann of Astra Pharmaceuticals for gifts of QX-222; Drs. Takmann and R. A. North for discussion; and W. J. Goddard for the use of computer modeling facilities. This work was supported by grants from the National Institutes of Health (NS-11756) and from the Muscular Dystrophy Association and by postdoctoral fellowships to R. J. L. (NS-8083) and P. C. (Bourse Lavoiser and Fondation pour la Recherche Medicale). Received
August
17, 1989;
revised
August
29, 1989
W., Dwyer, T. M., and Hille, B. (1981). by permeant cations in frog skeletal 78, 593-615.
Adams, P. R. (1977). Voltage jump analysis frog end-plate. J. Physiol. 268, 291-318.
of procaine
Adams, P. R., and Sakmann, B. (1978). Decamethonium opens and blocks endplate channels. Proc. Natl. Acad. 75, 2994-2998. Adler, M., Oliviera, A. C., Albuquerque, E. X., Mansour, and Eldefrawi,A. T. (1979). Reaction of tetraethylammonium the open and closed conformations of the acetylcholine tor ionic channel complex. J. Gen. Physiol. 74, 129-152.
Block mus-
action
at
interactions
Dani, J. A. (1989a). Site-directed mutagenesis and single-channel currents define the ionic channel of the nicotinic acetylcholinereceptor. Trends Neurosci. 72, 125-128. Dani, J. A. (198913). Open channel structure and ion binding sites of the nicotinic acetylcholine receptor. 1. Neurosci. 9, 884-892. Gero, A., and Reese, V. J. (1956). Science 123, 100.
Chemical
model
of drug
action.
Giraudat, J., Dennis, M., Heidmann, T.. Chang, J. Y., and Changeux, 1. P (1986). Structure of the high-affinity binding site for noncompetitive blockers of the acetylcholine receptor: serine262 of the delta subunit is labeled by [‘HIchlorpromazine. Proc. Natl. Acad. Sci. USA 83, 2719-2723. Ciraudat, I., Dennis, M., Heidmann, T., Haumont, P.. Lederer, F., and Changeux, I. P. (1987). Structure of the high-affinity binding site for noncompetitive blockers of the acetylcholine receptor: [3H]chlorpromazine labels homologous residues in the s and d chains. Biochemistry 26, 2410-2418. Hille, B. (1977). Local anesthetics: pathways for the drug-receptor 497-515. Hille, land,
B. (1984). Ionic Channels MA: Sinauer Associates).
hydrophilic reaction. In Excttable
and hydrophobic J. Cen. Physiol. 69, Membranes
(Sunder-
Hondeghem, L. M., and Katzung, B. G. (1984). Antiarrhythmic agents: the modulated receptor mechanism of action of sodium and calcium channel-blocking drugs. Annu. Rev. Pharmacol. Toxicol. 24, 387-423. Hucho, F., Oberthur. W., and Lottspeich, F. (1986). The ion channel of the nicotinic acetylcholine receptor is formed by the homologous helices M II of the receptor subunits. FEBS Lett. 205, 137-142. Imoto, K., Methfessel, C., Sakmann. B., Mishina, M., Mori, Y., Konno, T., Fukuda, K., Kurasaki, M., Bujo, H., Fujita, Y., and Numa, S. (1986). Location of a delta-subunit region determinlng ion-transport through the acetylcholine-receptor channel. Nature 324, 670-674. Imoto, K., Busch, C., Sakmann, B., Mishina, M., Konno, T., Nakai, J., Bujo, H., Mori, Y., Fukuda, K., and Numa, S. (1988). Rings of negatively charged amino acids determine the acetylcholine receptor channel conductance. Nature 335, 645-648. Korolkovas, Background
A. (1970). Essentials of Molecular Pharmacology: for Drug Design (New York: Wiley-Interscience).
Kunkel, T. A., Roberts, efficient site-specific tion. Meth. Enzymol.
I. D., and Zakour, R. A. (1987). Rapid and mutagenesis without phenotypic selec754, 367-383.
Leonard, R. J., Labarca, C., Charnet, F’., Davidson, N., and Lester, H. A. (1988). Evidence that the M2 membrane-spanning region lines the ion channel pore of the nicotinic receptor. Science242. 1578-l 581. Levitt, M., and Perutz, M. F. (1988). Aromatic rings gen bond acceptors. J. Mol. Biol. 207, 751-754. Miller, C. (1989). Genetic manipulation approach to structure and mechanism.
References Adams, D. J., Nonner, of endplate channels cle. J. Gen. Physiol.
Burley, S. K., and Petsko, G. (1986). Amino-aromatic in proteins. FEBS Lett. 203, 139-142.
act as hydro-
of ion channels: a new Neuron 2. 1195-1205.
Neher, E. (1983). The charge carried by single-channel currents of rat cultured muscle cells In the presence of local anesthetic\. J. Physiol. 339, 663-678. Neher, E., and Steinbach, J. H. (1978). Local anaesthetics siently block currents through single acetylcholine-receptor channels. J. Physiol. 277, 153-176.
tran-
both Sci. USA
Pauling, L., Corey, R. B., and Bransom, H. R. (1951). The structure of proteins: two hydrogen-bonded helical configurations of the polypeptide chain. Proc. Natl. Acad. SCI. USA 3i: 205-211.
N. A., with recep-
Ritchie, 1. M., and Greene, N. M. (1985). Local anesthetics. In The Pharmacological Basis of Therapeutics, A. C. Gilman, L. 5. Gilman, T. W. Rail, and F. Murad, eds. (New York: Macmillan), pp. 302-321.
C-hannel 95
Blockers
at Nrcotrnrc
Rrtchie, of local
J. M., and anesthetics.
Greengard, Annu.
Spivak,
C. E., Maleque,
Receptor
P (1966). On the mode of actron Rev. Pharmacol. 6, 405-430.
M. A., Oliviera,
A. C., Masukawa,
L. M.,
Tokuyama, T., Daly, J. W., and Albuquerque, E. X. (1982). Actrons of the histrionicotoxins at the ion channel of the nicotinrc acetylcholine receptor and at the voltage-sensitive ion channels of muscle membranes. Mol. Pharmacol. 21, 351-361. Thomas, K. A., Smrth, G. M., Thomas, T. B., and Feldman, (1982). Electronic distributions with protein phenylalanine matrc rings are reflected by the three-dimensional oxygen environments. Proc Natl. Acad. Sci. USA 79, 4843-4847.
R. J. aroring
Toyoshima, C., and Unwin, N. (1988). Ion channel of acetylcholine-receptor reconstructed from images of postsynaptic membranes. Nature 336, 247-250. Unwin, N., Toyoshima, C., and Kubalek, E. (1988). Arrangement of the acetylcholine-receptor subunits in the restrng and desensitized states, determined by cryoelectron microscopy of c.rystallized Torpedo postsynaptrc membranes. J, Cell Biol. 707. 11231138.
