Proc. Nati. Acad. Sci. USA Vol. 75, No. 2, pp.,784-788, February 1978
Biochemistry
Alteration of the physicochemical properties of triphosphoinositide by nicotinic ligands (neuromuscular blockers/hydrophobicity/phosphatidylinositol bisphosphate)
TAE MOOK CHO, JUNG SOOK CHO, AND HORACE H. LOH Langley Porter Neuropsychiatric Institute and Department of Pharmacology, University of California, San Francisco, California 94143
Communicated by Choh Hao Li, December 2,1977
from cerebral cortex by De Robertis et al. (7). Chemical determinations confirmed that P3inositide is present in this receptor fraction and suggested that P3inositide is the major component responsible for the binding of nicotinic ligands
ABSTRACT The concentrations of nicotinic drugs required to effect a 50% transfer of triphospho[3H]inositide from an aqueous phase to a nonaqueous phase closely approximated their concentrations for both in vivo neuromuscular blocking activity and binding to purified nicotinic receptors, and correlated well (correlation coefficient = 0.95) with their biological activities measured by other workers in an Electrophorus electroplax preparation. The triphospho[3Hjinositide transfer induced by nicotinic ligands was dependent on the lipid concentration and was potentiated by Ca +. The affinity constants of 45Ca2+ for triphosphoinositide were similar to those for the purified nicotinic receptor. These and other findings suggest the possibility that triphosphoinositide (phosphatidylinositol bisphosphate) is a binding site of the nicotinic receptor. Recently, Cavallito (1) and others (2, 3) have reported that electrostatic bond formation between the cationic portion of nicotinic ligands and the anionic site of the cholinergic receptor is essential for pharmacological activity. Cavallito further suggested that the highly charged phosphate macromolecules may be more closely related to the nicotinic receptor than are monovalent anions such as carboxylate, phosphate, and sulfate (4). From structural considerations of triphosphoinositide (Psinositide; phosphatidylinositol bisphosphate), a membrane acidic lipid with a maximum of five negative charges, it is theoretically possible that the molecule could interact with a variety of structurally unrelated nicotinic ligands. For example, the.cationic nitrogen and carbonyl oxygen of a drug such as carbachol could bind to an anionic oxygen atom in the phosphate group and to a hydroxyl group of the inositol ring, respectively. Similarly, the distance between the two cationic nitrogen atoms in the hexamethonium molecule is similar to the distance between two anionic oxygen atoms of the phosphates in the 4 and 5 positions of the inositol ring. Furthermore, the distance between the quaternary nitrogen atoms in cholinergic ligands such as decamethonium, succinylcholine, gallamine, and d-tubocurarine matches the distance between the two anionic oxygen atoms of the phosphate group in positions 1 and 4 or 1 and 5 of the inositol ring. These theoretical considerations are supported by experimental data showing that P3inositide binds nicotinic agonists and antagonists with high affinities, and that the dissociation constants obtained are comparable to those reported for the same agents binding to purified nicotinic receptor (5). Of the membrane acidic lipids examined, only P3inositide selectively binds di[14C]methyld-tubocurarine ([14CJDMTC), and the chromatographic behavior of the [14C]DMTC-PNinositide complex on a Sephadex LH-20 column (6) is identical to the behavior of the DMTC complex formed with the proteolipid receptor fraction isolated
The present study describes selective alterations in the physicochemical properties of P3inositide on interaction with various nicotinic ligands. The alterations have been shown to be related to the neuromuscular blocking activity of these ligands. The results suggest that P3inositide may function as a binding component of the nicotinic cholinergic receptor. MATERIALS AND METHODS Chemicals. P3inositide was isolated by the method of Michell et al. (8), further purified using triethylaminoethyl-cellulose column chromatography, and identified by the method of Gonzalez-Sastre and Folch-Pi (9), using thin-layer chromatography and visualization by iodine, which yielded a single spot with an RF value the same as the reported one. The P3inositide spot was removed and the phosphate content was determined by Bartlett's method (10). On the basis of the phosphate determination, the P3inositide was over 95% pure. P3[3H]inositide (specific activity 3 Ci/mmol) was prepared by C. T. Peng (School of Pharmacy, University of California, San Francisco) at the Lawrence Laboratories (Berkeley, CA). 45CaCl2 (20 Ci/g) was obtained from New England Nuclear (Boston, MA). The following chemicals were purchased from Sigma Chemical Co.: d-tubocurarine chloride, decamethonium chloride, trimethylbenzylammonium chloride, succinylcholine chloride, hexamethonium chloride, carbachol chloride, pilocarpine, methacholine chloride, (+)-muscarine chloride, 5hydroxytryptamine, dopamine, epinephrine, norepinephrine, and histamine. Nicotine, 1,1-dimethylphenylpiperazonium iodide, and lobeline were purchased from K & K Laboratories; gallamine triethiodide from Davis and Geck, American Cyanamid Co.; and mecamylamine from Merck, Sharp and Dohme Research Laboratories. a-Bungarotoxin was a gift from C. C. Chang of National Taiwan University (Taipei, Republic of China). Preparation of P3inositide Micelles. P3[3H]inositide was mixed with unlabeled P3inositide in a mixture of chloroform and methanol (2:1) and the solution was then taken down to dryness on a rotary vacuum evaporator, after which the dried P,3inositide powder was suspended in 5 ml of Tris-HCI buffer (2 mM, pH 7.4) and allowed to swell for an hour. The P3inositide was sonicated for 1 min and then diluted with the same
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 U. S. C. §1734 solely to indicate this fact.
Abbreviations: P3inositide, triphosphoinositide (phosphatidylinositol bisphosphate); DMTC, dimethyl-d-tubocurarine; IC50, concentration required to induce a 50% transfer of P3inositide from aqueous to nonaqueous phase.
(6).
784
Biochemistry: Cho et al. buffer to give an appropriate final P3inositide concentration. P3[3H]inositide Transfer Studies. Five tenths milliliter of the P3[3H]inositide micelle preparation was added to 0.5 ml of Tris buffer containing drug and/or cation at room temperature. This aqueous solution was mixed with 1 ml of heptane on a Vortex mixer (speed setting at 5) for 1 min and the mixture was centrifuged at 1500 X g for 10 min. A 0.5-ml aliquot of the aqueous phase mixed with 5 ml of Scintiverse cocktail solution (Fisher Scientific Co.) was assayed in a Beckman liquid scintillation counter to determine P3[3H]inositide activity. The radioactivity in the nonaqueous phase (heptane plus interphase) was determined by subtracting the radioactivity in the aqueous phase from the total amount added. The percent of drug-induced P3[3H]inositide transferred from the aqueous phase to the nonaqueous phase was calculated by using the following equation: % of P3[3H]inositide transferred radioactivity without - radioactivity with drug in aqueous phase x 100 drug in aqueous phase radioactivity without drug in aqueous phase 45Ca2+ Binding to P3inositide. One hundred milliliters of 45Ca2+ (2 puM) was mixed with 100 ml of the prepared nonradioactive P3inositide micelles (20 jg/ml). An aliquot of the mixture (0.5 ml) was added to a 0.5-ml solution containing various concentrations of unlabeled calcium or neuromuscular blocking agents and the aqueous solution was mixed with 1 ml of heptane on a Vortex mixer (speed at 5) for 1 min. The mixture was centrifuged at 1500 X g for 10 min and the radioactivity of 45Ca2+ in the lower phase was determined by the same method described for the P3[3H]inositide transfer studies. RESULTS Distribution of P3[3H]inositide. The radioactive P3inositide in both the heptane and aqueous phases was determined in the presence and absence of decamethonium, and the amount of labeled P3inositide at the interphase was obtained by subtracting the measured radioactivities from the total added. As shown in Fig. 1, the P3[3H]inositide in the aqueous phase decreased with increasing concentration of the decamethonium added, while the radioactivity in the nonaqueous phases increased, particularly in the interphase. Effects of Calcium and Lanthanum Ions on P3[3H]inositide Transfer. Fig. 2 shows that both calcium and lanthanum, a calcium antagonist, promoted P3[3H]inositide transfer from the aqueous to the nonaqueous phase in a concentration-dependent manner. The concentration of lanthanum required to induce a 50% transfer from the aqueous to the nonaqueous phase (IC50) was approximately 250 times less than that of calcium. The P3[3H]inositide transfer was also determined with increasing concentrations of decamethonium and d-tubocurarine in the presence and absence of 100 AM calcium ion, which by itself did not significantly induce the transfer (Fig. 3). Calcium caused a 2-fold increase in the P3[3H]inositide transfer induced by both drugs and appeared to change the shape of the transfer-concentration curves from sigmoid to hyperbolic. To elucidate further the effect of calcium on P3[3H]inositide transfer induced by nicotinic ligands, 45Ca2+ binding to P3inositide was determined and [Ca2+hboUnd/ [Ca2+]free was plotted against [Ca2+]bound As shown in Fig. 4, calcium had at least three dissociation constants for P3inositide:
Proc. Natl. Acad. Sci. USA 75 (1978)
'or
785
Interface
9 8 0
7
x
E6 5 0.
0o
I
31 K\ie
Heptane phase
at---la
2F
° ~Oto.
0
4__hi Aqueous phase
I
2
1
I I I I I I 3 4 5 6 7 8 [Decamethoniuml, MM
I 9
I 10
FIG. 1. Distribution of P3[3H]inositide in the presence and absence of decamethonium. The influence of various concentrations of decamethonium on the distribution of P3[3H]inositide between the aqueous phase, the heptane phase, and the interface was determined by measurement of the radioactivity in the buffer and heptane components. The amount of P3[3H]inositide at the interface was indirectly
deduced by calculations, subtracting the measured radioactivities from the total.
K1 = 2.7 AtM, K2 = 25 MM, and K3 = 350 AM. These data suggest that in the presence of 100 AM calcium ion only the binding sites with the two lower dissociation constants (K1 and 00 r-
A'S
/O
of.
8C
~0)
4O (A
a) 60 c
La 3+
ICa 2t
0'
'a (a
*0 0
0/
40 I
I
-
4
l
20
0
0
L-
10-, 104 10 5 [Cation], M FIG. 2. Transfer of P3[3H]inositide induced by Ca2+ and La3+. 10
Inorganic ion-induced transfer of labeled P3inositide (0.2 gAg/ml) from the aqueous to the nonaqueous phases was determined from measurement of the change of radioactivity in the aqueous phase after addition of the ions as described in Materials and Methods.
Biochemistry: Cho et al.
786 100
0_
60
100
A _
80
0
:2 X.U
Proc. Nati. Acad. Sci. USA 75(1978) "I OS
PR' -- 8
Ci+ +Ca "
O/ /C2+
2+ +Ca-
4) 4)
I I
c! 'A, w 40
4
-'2. 0Ca o-/
0)
C
I
-
20
/g
2
4-
I
o'
I
0
4
8
6
FIG. 3. P3[3H]inositide transfer induced by decamethonium and d-tubocurarine: effects of 100 AM CaC12. Increasing concentrations of decamethonium and d-tubocurarine in the presence and absence of 100AM CaC12 were used to induce the transfer of 0.2 Ag/ml of tritiated P3inositide from the aqueous to the nonaqueous phase. (A) Decamethonium; (B) d-tubocurarine.
K2) may be saturated. Additional experiments were performed to study the effect of d-tubocurarine and gallamine on 45Ca2+ (0.5 ,uM) binding to (data not shown). At the concentrations used
in Fig. 3, neither drug inhibited the binding of 45Ca2+ to P3inositide, indicating that the calcium-binding sites with higher affinities (K1 and K2) may be different from the nicotinic binding sites.
Effects of P3inositide Concentration on P3[3H]inositide Transfer. The effect of P3inositide concentration on druginduced P3[3H]inositide transfer was also studied at two different P3inositide concentrations (0.2 and 5 ,ug/ml) in the presence of gallamine. As shown in Fig. 5, at the higher P31. inositideconcentration,the1IC5o of gallamine was 3 times higher than at the lower P3inositide concentration, and the transferconcentration curve was shifted from a hyperbolic to a more sigmoid shape. This effect of P3inositide concentration was also seen with other cholinergic ligands (Table 1). Potency, Specificity, and "Biological Correlation." Various compounds-including nicotinic and muscarinic cholinergic 0.20
IL
2 3 4 [d-TubocurarineJ, ,M
10
[Decamethonium], JM
r
0.18
3 4 2 [Gallaminel, }AM
5
FIG. 5. Gallamine-induced P3P3H]inositide transfer at different P3inositide levels. P3[3H]inositide at 0.2 and 5.0 Mg/ml was mixed with increasing concentrations of gallamine in order to induce transfer from the aqueous to the nonaqueous phases.
ligands, noncholinergic neurotransmitters, and inorganic cations-were tested for their ability to affect the Ps[3H]inositide transfer at two different concentrations of P3inositide in the presence and absence of calcium ion. Fig. 6 shows the concentration-response curves for the nicotinic ligands determined from the data obtained with P3inositide at 0.2 ,gg/ml in the presence of 100( ,M calcium ion. The slopes were steep and the curves were parallel. The IC50 values were determined from the curves by extrapolation and the results are summarized in Table 1. Of the experimental conditions tested, 100 ,uM of Ca2+ and 0.2 tig/ml of P3inositide required the lowest concentration of ligands to induce the P3[3H]inositide transfer. Under these conditions, nicotinic ligands were much more potent than either the muscarinic ligands or the noncholinergic neurotransmitters, and their ICso values were about one-third as high as their corresponding in vitro biological activities and apparent affinities as reported by Cohen and Changeux (11). The IC,0 values of the nicotinic compounds appeared to correlate more with their neuromuscular blocking activities, such as those represented by the Electrophorus receptor, than with their ganglionic blocking activities. For example, potent neuromuscular blockers such as a-bungarotoxin, pancuronium, alcuronium, d-tubocurarine, and gallamine were very effective in transferring P3[3H]inositide, while ganglionic blockers such
0.16 11
o0.14
2 3
1
-
4
5
6 7 8 9
80
12
/ 0.12
-
ax
U
n
70_
n
K=2.7M
0.10
C6
-
L._
0
+
0.08
\
0~~~0
. 50 30
0.06
-
0.04
-
40
I40 K2
25uMM II
10-7 0.02
FIG.
4.
10-6
lo-,
10-4
[Drug], M
I
I
0.2
0.4
II
I
1.0 1.2 0.6 0.8 Calcium bound, nmol
Scatchard plot of 45Ca2+
binding. P3inositide
1.4 at
5Mug/mi
and various concentrations of CaCl2 were used. It should be mentioned that although most of free P3inositide occurred in the aqueous phase under these conditions, calcium binding was observed mostly at the interface due to the small amount of P3inositide.
FIG. 6. Concentration-response curves for P3[3H]inositide transfer induced by nicotinic ligands. Dose-response curves for a variety of nicotinic ligands capable of inducing the transfer of tritiated P3inositide (0.2 Mg/ml) from the aqueous to the nonaqueous phases were determined. The data points are the mean of triplicate samples in two separate experiments. 1, a-Bungarotoxin; 2, d-tubocurarine; 3, gallamine; 4, decamethonium; 5, mecamylamine; 6, lobeline; 7, dimethylphenylpiperazonium; 8, trimethylbenzylammonium; 9, nicotine; 10, succinylcholine; 11, hexamethonium; 12, carbachol.
Biochemistry:
Cho et al.
Proc. Natl. Acad. Sci. USA 75 (1978)
787
Table 1. Concentration of various compounds required to induce P3[3H]inositide transfer by 50% (IC50) Carbachol
I
IC50, AsM
P3inositide (0.2 mg/ml)
Compound
-Ca2+
Nicotinic ligands a-Bungarotoxin
0.5
Pancuronium Alcuronium Dimethyl-d-tubocurarine d-Tubocurarine Gallamine Decamethonium Mecamylamine Dimethylphenylpiperazonium Trimethylbenzylammonium Nicotine Lobeline
0.4 0.35
P3inositide
(5.0 jg/ml)
+Ca2+
-Ca2+
+Ca2+
0.18 0.25 0.25
1.2
0.6
-
* Gallamine
0.60 1.1 1.6 3.0 6.0
0.30 0.50 0.80 2.1 4.6
1.2 1.8 3.0 7.4
3.0 4.5 7.6 15
15
35
25
26 50 10 120 130 210
80 160
46 90
40
20
Succinylcholine Hexamethonium Carbachol Muscarinic ligands Pilocarpine Methacholine
42 92 19 84 180 250
150 300 550
210
110 150
100
200
150
150
(W)-Muscarine
250
210
300 3000
200 2000
Cations CaCl2 LaCl3
MX lo-6
0 Tubocurarine
22
Neurotransmitters 5-Hydroxytryptamine Dopamine Epinephrine Norepinephrine Histamine
10-
-
-
350 0.9
130 230 500 800 1100 150 1.1
130 350
-
-
IC50 values were obtained from plots of percent of P3[3H]inositide transferred vs. the log concentrations of the various compounds by extrapolation at two P3inositide concentrations and in the presence or absence of 100 ,uM CaCl2. Data points are the mean of triplicate determinations in two separate experiments.
as hexamethonium and mecamylamine were relatively weak. However, it should be noted that mecamylamine, with an IC50 twice that of decamethonium, was only half as potent as decamethonium in the phrenic nerve diaphragm preparation (12). The IC50 values of the nicotinic ligands were plotted against the biological potencies obtained from the isolated Electrophorus electroplax (Fig. 7). According to the method of linear regression analysis, the data were fitted to the line (correlation coefficient = 0.95). DISCUSSION From the molecular considerations as well as experimental findings presented previously, we have suggested that P3inositide, a membrane acidic lipid, may be a binding component in the nicotinic chloinergic receptor. Using the drug-induced transfer of P3[3H]inositide from an aqueous to a nonaqueous phase, we have determined the IC50 of various nicotinic and muscarinic ligands, noncholinergic neurotransmitters, and cations, and we have found that the IC50 values of the nicotinic
10 6
10 5
10-4
IJ
10-3
IC50, M
FIG. 7. Correlation of the IC50 values of P3[3H]inositide transfer with the in vitro potency of various ligands. The IC50 values of these nicotinic drugs were taken from Table 1 and the values of Kapp (apparent dissociation constant) were obtained from the isolated monocellular electroplaque by Cohen and Changeux (11).
ligands correlate closely with their in vivo neuromuscular blocking activities. The precise nature of the interaction between cholinergic ligands and P3inositide in micellar form is not clear, but complex formation appears to alter the physicochemical properties of the lipid molecule to result in increased hydrophobicity (here defined as tendency to leave the aqueous phase). The IC50 values determined are not solely dependent on the affinity (dissociation constant) of the ligand for P3inositide, but also reflect the apparent change in physicochemical properties of the complex formed. Thus, for any given ligand, the ratio of KD to ICWo is a measure of the relative hydrophobicity of the ligand-P3inositide complex. In the case of the stabilizing blocking agent d-tubocurarine, with a KD of 0.2 ,uM (unpublished data) this ratio is approximately 0.4. However, with succinylcholine, a depolarizing blocking agent (KD 1.6 ,uM), the ratio is much lower (0.013), suggesting that at ligand concentrations close to the dissociation constant the succinylcholine-P3inositide complex is less hydrophobic than the complex of d-tubocurarine with P3inositide. At concentrations of succinylcholine closer to the IC50 value, the icomplex with P3inositide becomes more hydrophobic. Suchbehavior is consistent with ion-pair theory, which states that the formation of intimate ion pairs is dependent on the concentration of the cation in a given concentration of the anion (13, 14). The formation of two different types of succinylcholine-P3inositide complex may have relevance to our understanding of the biphasic effects of depolarizing neuromuscular blocking agents and of discrimination between stabilizing and depolarizing actions of the nicotinic blocking agents. It is well known that calcium ion increases the binding of the cholinergic ligands to their receptors (15, 16). We also observed calcium potentiation of P3[3H]inositide transfer induced by such ligands (Fig. 3). Because 100 ,uM CaCl2 was by itself not sufficient to transfer the Pa[3H]inositide from the aqueous to the nonaqueous phase, the calcium potentiation appeared due to the enhancement of the affinity of the ligand to P3inositide. This has been confirmed and, as in the case of the nicotinic receptor, the binding of ['4C]DMTC and [3H]decamethonium to P3inositide was also enhanced by various concentrations of calcium when the experiments were performed in KrebsRinger solution. However, in Tris.HCI buffer (pH 7.4), calcium enhanced the binding of [14C]DMTC to P3inositide, whereas the binding of [3H]decamethonium was inhibited. Thus, cal-
788
Biochemistry: Cho et al.
cium discriminates the stabilizer from the depolarizer in the drug binding to P3inositide (T. M. Cho and H. H. Loh, unpublished data). Several lines of evidence appear to support a possible role for P3inositide as a binding component of the nicotinic receptor: (i) The concentrations of nicotinic ligands required to induce 50% P3[3H]inositide transfer from the aqueous phase to the nonaqueous phase (IC5o) not only were close to the concentrations required for the ligands' in vio neuromuscular blocking activities and their binding to the purified nicotinic receptor, but also correlated well with their in vitro pharmacological activities obtained from the isolated eel electroplax and reported by Cohen and Changeux (11). (ii) Calcium enhanced the ligand-induced transfer of P3[3H]inositide, and this cation has been shown to increase binding of nicotinic ligands to the cholinergic receptor (15, 16) and to increase binding of [14C]DMTC and [3H]decamethonium to Psinositide.* The dissociation constants for the Ca2+-Painositide complex were almost identical to those for the purified nicotinic receptorCa2+ complex (15). (iii) De Jobertis et al. have shown binding of [14C]DMTC, a nicotinic antagonist, to the purified proteolipid nicotinic receptor isolated from both cat brain (7) and electric eel (17). Such proteolipid receptors appear similar to detergent-extracted cholinergic protein receptors except for their immunological properties (18, 19). Using the method of De Robertis et al. to purify nicotinic cholinergic receptors, we have reported the presence of P3inositide in this purified receptor fraction and shown that the binding of [14C]DMTC is mainly, if not solely, due to its interaction with P3inositide in this fraction (6). Furthermore, the chromatographic behavior of the [14C]DMTC-P3inositide complex was identical to that of the [14C]DMTC complex as reported by De Robertis et al.
(7).
On the basis of the present findings, it is tempting to speculate that neuromuscular blocking activity of the antagonist type may be elicited via the formation of hydrophobic ligand-receptor complexes, thereby stabilizing the membrane with blockade of the ionic conductance. Furthermore, our findings indicate
Proc. Natl. Acad. Sci. USA 75 (1978)
that physicochemical properties (i.e., hydrophobicity and hydrophilicity) of the ligand-receptor complex may be important in the discrimination of the neuromuscular blocking agents from the stimulating agents. 1. Cavallito, C. J. (1962) in Curare and Curare-like Agents, ed. De Reuck, A. V. S. (Little Brown & Co., Boston, MA), pp. 55-70. 2. Bovet, D. (1972) in Neuromuscular Blocking and Stimulating Agents, ed. Cheymol, J. (Pergamon Press, Oxford), Vol. 1, pp. 243-294. 3. Cheymol, J. & Bourillet, F. (1972) in Neuromuscular Blocking and Stimulating Agents, ed. Cheymol, J. (Pergamon Press, Oxford), Vol. 1, pp. 297-356. 4. Cavallito, C. J. (1967) Ann. N.Y. Acad. Sci. 144,900-912. 5. Cho, T. M. & Loh, H. H. (1975) Pharmacologist 17,254. 6. Wu, Y. C., Cho, T. M. & Loh, H. H. (1977) J. Neurochem. 29, 589-592. 7. De Robertis, E., Fiszer, S., Pasquini, J. M. & Soto, E. F. (1969) J. Neurobiol. 1, 41-52. 8. Michell, R. H., Hawthorne, J. N., Coleman, R. & Karnovsky, M. L. (1970) Biochim. Biophys. Acta 210,86-91. 9. Gonzalez-Sastre, F. & Folch-Pi, J. (1968) J. Lipid Res. 9,532533. 10. Bartlett, G. R. (1959) J. Biol. Chem. 234,466-468. 11. Cohen, J. B. & Changeux, J.-P. (1975) Annu. Rev. Pharmacol.
15,83-109. 12. Burn, J. H. & Seltzer, J. (1965) J. Physiol. (London) 179, 569576. 13. Gordon, J. E. (1975) in The Organic Chemistry of Electrolyte Solutions, ed. Gordon, J. E. (Wiley, New York), pp. 371-520. 14. Szwarc, M. (1972) in Ions and Ion Pairs in Organic Reactions, ed. Szwarc, M. (Wiley, New York), Vol. 1, pp. 1-24. 15. Chang, H. W. & Neumann, E. (1976) Proc. Natl. Acad. Sci. USA 73,3364-3368. 16. Cohen, J. B., Weber, M. & Changeux, J.-P. (1974) Mol. Pharmacol. 10, 904-932. 17. De Robertis, E. & De Plazas, S. F. (1970) Biochim. Biophys. Acta 219,388-397. 18. Komentani, T., Ideda, Y. & Kasai, Y. (1975) Biochim. Biophys. Acta, 413, 415-424. 19. De Robertis, E., De Plazas, S. F. & De Carlin, M. C. L. (1976) Nature 259,605.
Biochemistry
Alteration of the physicochemical properties of triphosphoinositide by nicotinic ligands (neuromuscular blockers/hydrophobicity/phosphatidylinositol bisphosphate)
TAE MOOK CHO, JUNG SOOK CHO, AND HORACE H. LOH Langley Porter Neuropsychiatric Institute and Department of Pharmacology, University of California, San Francisco, California 94143
Communicated by Choh Hao Li, December 2,1977
from cerebral cortex by De Robertis et al. (7). Chemical determinations confirmed that P3inositide is present in this receptor fraction and suggested that P3inositide is the major component responsible for the binding of nicotinic ligands
ABSTRACT The concentrations of nicotinic drugs required to effect a 50% transfer of triphospho[3H]inositide from an aqueous phase to a nonaqueous phase closely approximated their concentrations for both in vivo neuromuscular blocking activity and binding to purified nicotinic receptors, and correlated well (correlation coefficient = 0.95) with their biological activities measured by other workers in an Electrophorus electroplax preparation. The triphospho[3Hjinositide transfer induced by nicotinic ligands was dependent on the lipid concentration and was potentiated by Ca +. The affinity constants of 45Ca2+ for triphosphoinositide were similar to those for the purified nicotinic receptor. These and other findings suggest the possibility that triphosphoinositide (phosphatidylinositol bisphosphate) is a binding site of the nicotinic receptor. Recently, Cavallito (1) and others (2, 3) have reported that electrostatic bond formation between the cationic portion of nicotinic ligands and the anionic site of the cholinergic receptor is essential for pharmacological activity. Cavallito further suggested that the highly charged phosphate macromolecules may be more closely related to the nicotinic receptor than are monovalent anions such as carboxylate, phosphate, and sulfate (4). From structural considerations of triphosphoinositide (Psinositide; phosphatidylinositol bisphosphate), a membrane acidic lipid with a maximum of five negative charges, it is theoretically possible that the molecule could interact with a variety of structurally unrelated nicotinic ligands. For example, the.cationic nitrogen and carbonyl oxygen of a drug such as carbachol could bind to an anionic oxygen atom in the phosphate group and to a hydroxyl group of the inositol ring, respectively. Similarly, the distance between the two cationic nitrogen atoms in the hexamethonium molecule is similar to the distance between two anionic oxygen atoms of the phosphates in the 4 and 5 positions of the inositol ring. Furthermore, the distance between the quaternary nitrogen atoms in cholinergic ligands such as decamethonium, succinylcholine, gallamine, and d-tubocurarine matches the distance between the two anionic oxygen atoms of the phosphate group in positions 1 and 4 or 1 and 5 of the inositol ring. These theoretical considerations are supported by experimental data showing that P3inositide binds nicotinic agonists and antagonists with high affinities, and that the dissociation constants obtained are comparable to those reported for the same agents binding to purified nicotinic receptor (5). Of the membrane acidic lipids examined, only P3inositide selectively binds di[14C]methyld-tubocurarine ([14CJDMTC), and the chromatographic behavior of the [14C]DMTC-PNinositide complex on a Sephadex LH-20 column (6) is identical to the behavior of the DMTC complex formed with the proteolipid receptor fraction isolated
The present study describes selective alterations in the physicochemical properties of P3inositide on interaction with various nicotinic ligands. The alterations have been shown to be related to the neuromuscular blocking activity of these ligands. The results suggest that P3inositide may function as a binding component of the nicotinic cholinergic receptor. MATERIALS AND METHODS Chemicals. P3inositide was isolated by the method of Michell et al. (8), further purified using triethylaminoethyl-cellulose column chromatography, and identified by the method of Gonzalez-Sastre and Folch-Pi (9), using thin-layer chromatography and visualization by iodine, which yielded a single spot with an RF value the same as the reported one. The P3inositide spot was removed and the phosphate content was determined by Bartlett's method (10). On the basis of the phosphate determination, the P3inositide was over 95% pure. P3[3H]inositide (specific activity 3 Ci/mmol) was prepared by C. T. Peng (School of Pharmacy, University of California, San Francisco) at the Lawrence Laboratories (Berkeley, CA). 45CaCl2 (20 Ci/g) was obtained from New England Nuclear (Boston, MA). The following chemicals were purchased from Sigma Chemical Co.: d-tubocurarine chloride, decamethonium chloride, trimethylbenzylammonium chloride, succinylcholine chloride, hexamethonium chloride, carbachol chloride, pilocarpine, methacholine chloride, (+)-muscarine chloride, 5hydroxytryptamine, dopamine, epinephrine, norepinephrine, and histamine. Nicotine, 1,1-dimethylphenylpiperazonium iodide, and lobeline were purchased from K & K Laboratories; gallamine triethiodide from Davis and Geck, American Cyanamid Co.; and mecamylamine from Merck, Sharp and Dohme Research Laboratories. a-Bungarotoxin was a gift from C. C. Chang of National Taiwan University (Taipei, Republic of China). Preparation of P3inositide Micelles. P3[3H]inositide was mixed with unlabeled P3inositide in a mixture of chloroform and methanol (2:1) and the solution was then taken down to dryness on a rotary vacuum evaporator, after which the dried P,3inositide powder was suspended in 5 ml of Tris-HCI buffer (2 mM, pH 7.4) and allowed to swell for an hour. The P3inositide was sonicated for 1 min and then diluted with the same
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 U. S. C. §1734 solely to indicate this fact.
Abbreviations: P3inositide, triphosphoinositide (phosphatidylinositol bisphosphate); DMTC, dimethyl-d-tubocurarine; IC50, concentration required to induce a 50% transfer of P3inositide from aqueous to nonaqueous phase.
(6).
784
Biochemistry: Cho et al. buffer to give an appropriate final P3inositide concentration. P3[3H]inositide Transfer Studies. Five tenths milliliter of the P3[3H]inositide micelle preparation was added to 0.5 ml of Tris buffer containing drug and/or cation at room temperature. This aqueous solution was mixed with 1 ml of heptane on a Vortex mixer (speed setting at 5) for 1 min and the mixture was centrifuged at 1500 X g for 10 min. A 0.5-ml aliquot of the aqueous phase mixed with 5 ml of Scintiverse cocktail solution (Fisher Scientific Co.) was assayed in a Beckman liquid scintillation counter to determine P3[3H]inositide activity. The radioactivity in the nonaqueous phase (heptane plus interphase) was determined by subtracting the radioactivity in the aqueous phase from the total amount added. The percent of drug-induced P3[3H]inositide transferred from the aqueous phase to the nonaqueous phase was calculated by using the following equation: % of P3[3H]inositide transferred radioactivity without - radioactivity with drug in aqueous phase x 100 drug in aqueous phase radioactivity without drug in aqueous phase 45Ca2+ Binding to P3inositide. One hundred milliliters of 45Ca2+ (2 puM) was mixed with 100 ml of the prepared nonradioactive P3inositide micelles (20 jg/ml). An aliquot of the mixture (0.5 ml) was added to a 0.5-ml solution containing various concentrations of unlabeled calcium or neuromuscular blocking agents and the aqueous solution was mixed with 1 ml of heptane on a Vortex mixer (speed at 5) for 1 min. The mixture was centrifuged at 1500 X g for 10 min and the radioactivity of 45Ca2+ in the lower phase was determined by the same method described for the P3[3H]inositide transfer studies. RESULTS Distribution of P3[3H]inositide. The radioactive P3inositide in both the heptane and aqueous phases was determined in the presence and absence of decamethonium, and the amount of labeled P3inositide at the interphase was obtained by subtracting the measured radioactivities from the total added. As shown in Fig. 1, the P3[3H]inositide in the aqueous phase decreased with increasing concentration of the decamethonium added, while the radioactivity in the nonaqueous phases increased, particularly in the interphase. Effects of Calcium and Lanthanum Ions on P3[3H]inositide Transfer. Fig. 2 shows that both calcium and lanthanum, a calcium antagonist, promoted P3[3H]inositide transfer from the aqueous to the nonaqueous phase in a concentration-dependent manner. The concentration of lanthanum required to induce a 50% transfer from the aqueous to the nonaqueous phase (IC50) was approximately 250 times less than that of calcium. The P3[3H]inositide transfer was also determined with increasing concentrations of decamethonium and d-tubocurarine in the presence and absence of 100 AM calcium ion, which by itself did not significantly induce the transfer (Fig. 3). Calcium caused a 2-fold increase in the P3[3H]inositide transfer induced by both drugs and appeared to change the shape of the transfer-concentration curves from sigmoid to hyperbolic. To elucidate further the effect of calcium on P3[3H]inositide transfer induced by nicotinic ligands, 45Ca2+ binding to P3inositide was determined and [Ca2+hboUnd/ [Ca2+]free was plotted against [Ca2+]bound As shown in Fig. 4, calcium had at least three dissociation constants for P3inositide:
Proc. Natl. Acad. Sci. USA 75 (1978)
'or
785
Interface
9 8 0
7
x
E6 5 0.
0o
I
31 K\ie
Heptane phase
at---la
2F
° ~Oto.
0
4__hi Aqueous phase
I
2
1
I I I I I I 3 4 5 6 7 8 [Decamethoniuml, MM
I 9
I 10
FIG. 1. Distribution of P3[3H]inositide in the presence and absence of decamethonium. The influence of various concentrations of decamethonium on the distribution of P3[3H]inositide between the aqueous phase, the heptane phase, and the interface was determined by measurement of the radioactivity in the buffer and heptane components. The amount of P3[3H]inositide at the interface was indirectly
deduced by calculations, subtracting the measured radioactivities from the total.
K1 = 2.7 AtM, K2 = 25 MM, and K3 = 350 AM. These data suggest that in the presence of 100 AM calcium ion only the binding sites with the two lower dissociation constants (K1 and 00 r-
A'S
/O
of.
8C
~0)
4O (A
a) 60 c
La 3+
ICa 2t
0'
'a (a
*0 0
0/
40 I
I
-
4
l
20
0
0
L-
10-, 104 10 5 [Cation], M FIG. 2. Transfer of P3[3H]inositide induced by Ca2+ and La3+. 10
Inorganic ion-induced transfer of labeled P3inositide (0.2 gAg/ml) from the aqueous to the nonaqueous phases was determined from measurement of the change of radioactivity in the aqueous phase after addition of the ions as described in Materials and Methods.
Biochemistry: Cho et al.
786 100
0_
60
100
A _
80
0
:2 X.U
Proc. Nati. Acad. Sci. USA 75(1978) "I OS
PR' -- 8
Ci+ +Ca "
O/ /C2+
2+ +Ca-
4) 4)
I I
c! 'A, w 40
4
-'2. 0Ca o-/
0)
C
I
-
20
/g
2
4-
I
o'
I
0
4
8
6
FIG. 3. P3[3H]inositide transfer induced by decamethonium and d-tubocurarine: effects of 100 AM CaC12. Increasing concentrations of decamethonium and d-tubocurarine in the presence and absence of 100AM CaC12 were used to induce the transfer of 0.2 Ag/ml of tritiated P3inositide from the aqueous to the nonaqueous phase. (A) Decamethonium; (B) d-tubocurarine.
K2) may be saturated. Additional experiments were performed to study the effect of d-tubocurarine and gallamine on 45Ca2+ (0.5 ,uM) binding to (data not shown). At the concentrations used
in Fig. 3, neither drug inhibited the binding of 45Ca2+ to P3inositide, indicating that the calcium-binding sites with higher affinities (K1 and K2) may be different from the nicotinic binding sites.
Effects of P3inositide Concentration on P3[3H]inositide Transfer. The effect of P3inositide concentration on druginduced P3[3H]inositide transfer was also studied at two different P3inositide concentrations (0.2 and 5 ,ug/ml) in the presence of gallamine. As shown in Fig. 5, at the higher P31. inositideconcentration,the1IC5o of gallamine was 3 times higher than at the lower P3inositide concentration, and the transferconcentration curve was shifted from a hyperbolic to a more sigmoid shape. This effect of P3inositide concentration was also seen with other cholinergic ligands (Table 1). Potency, Specificity, and "Biological Correlation." Various compounds-including nicotinic and muscarinic cholinergic 0.20
IL
2 3 4 [d-TubocurarineJ, ,M
10
[Decamethonium], JM
r
0.18
3 4 2 [Gallaminel, }AM
5
FIG. 5. Gallamine-induced P3P3H]inositide transfer at different P3inositide levels. P3[3H]inositide at 0.2 and 5.0 Mg/ml was mixed with increasing concentrations of gallamine in order to induce transfer from the aqueous to the nonaqueous phases.
ligands, noncholinergic neurotransmitters, and inorganic cations-were tested for their ability to affect the Ps[3H]inositide transfer at two different concentrations of P3inositide in the presence and absence of calcium ion. Fig. 6 shows the concentration-response curves for the nicotinic ligands determined from the data obtained with P3inositide at 0.2 ,gg/ml in the presence of 100( ,M calcium ion. The slopes were steep and the curves were parallel. The IC50 values were determined from the curves by extrapolation and the results are summarized in Table 1. Of the experimental conditions tested, 100 ,uM of Ca2+ and 0.2 tig/ml of P3inositide required the lowest concentration of ligands to induce the P3[3H]inositide transfer. Under these conditions, nicotinic ligands were much more potent than either the muscarinic ligands or the noncholinergic neurotransmitters, and their ICso values were about one-third as high as their corresponding in vitro biological activities and apparent affinities as reported by Cohen and Changeux (11). The IC,0 values of the nicotinic compounds appeared to correlate more with their neuromuscular blocking activities, such as those represented by the Electrophorus receptor, than with their ganglionic blocking activities. For example, potent neuromuscular blockers such as a-bungarotoxin, pancuronium, alcuronium, d-tubocurarine, and gallamine were very effective in transferring P3[3H]inositide, while ganglionic blockers such
0.16 11
o0.14
2 3
1
-
4
5
6 7 8 9
80
12
/ 0.12
-
ax
U
n
70_
n
K=2.7M
0.10
C6
-
L._
0
+
0.08
\
0~~~0
. 50 30
0.06
-
0.04
-
40
I40 K2
25uMM II
10-7 0.02
FIG.
4.
10-6
lo-,
10-4
[Drug], M
I
I
0.2
0.4
II
I
1.0 1.2 0.6 0.8 Calcium bound, nmol
Scatchard plot of 45Ca2+
binding. P3inositide
1.4 at
5Mug/mi
and various concentrations of CaCl2 were used. It should be mentioned that although most of free P3inositide occurred in the aqueous phase under these conditions, calcium binding was observed mostly at the interface due to the small amount of P3inositide.
FIG. 6. Concentration-response curves for P3[3H]inositide transfer induced by nicotinic ligands. Dose-response curves for a variety of nicotinic ligands capable of inducing the transfer of tritiated P3inositide (0.2 Mg/ml) from the aqueous to the nonaqueous phases were determined. The data points are the mean of triplicate samples in two separate experiments. 1, a-Bungarotoxin; 2, d-tubocurarine; 3, gallamine; 4, decamethonium; 5, mecamylamine; 6, lobeline; 7, dimethylphenylpiperazonium; 8, trimethylbenzylammonium; 9, nicotine; 10, succinylcholine; 11, hexamethonium; 12, carbachol.
Biochemistry:
Cho et al.
Proc. Natl. Acad. Sci. USA 75 (1978)
787
Table 1. Concentration of various compounds required to induce P3[3H]inositide transfer by 50% (IC50) Carbachol
I
IC50, AsM
P3inositide (0.2 mg/ml)
Compound
-Ca2+
Nicotinic ligands a-Bungarotoxin
0.5
Pancuronium Alcuronium Dimethyl-d-tubocurarine d-Tubocurarine Gallamine Decamethonium Mecamylamine Dimethylphenylpiperazonium Trimethylbenzylammonium Nicotine Lobeline
0.4 0.35
P3inositide
(5.0 jg/ml)
+Ca2+
-Ca2+
+Ca2+
0.18 0.25 0.25
1.2
0.6
-
* Gallamine
0.60 1.1 1.6 3.0 6.0
0.30 0.50 0.80 2.1 4.6
1.2 1.8 3.0 7.4
3.0 4.5 7.6 15
15
35
25
26 50 10 120 130 210
80 160
46 90
40
20
Succinylcholine Hexamethonium Carbachol Muscarinic ligands Pilocarpine Methacholine
42 92 19 84 180 250
150 300 550
210
110 150
100
200
150
150
(W)-Muscarine
250
210
300 3000
200 2000
Cations CaCl2 LaCl3
MX lo-6
0 Tubocurarine
22
Neurotransmitters 5-Hydroxytryptamine Dopamine Epinephrine Norepinephrine Histamine
10-
-
-
350 0.9
130 230 500 800 1100 150 1.1
130 350
-
-
IC50 values were obtained from plots of percent of P3[3H]inositide transferred vs. the log concentrations of the various compounds by extrapolation at two P3inositide concentrations and in the presence or absence of 100 ,uM CaCl2. Data points are the mean of triplicate determinations in two separate experiments.
as hexamethonium and mecamylamine were relatively weak. However, it should be noted that mecamylamine, with an IC50 twice that of decamethonium, was only half as potent as decamethonium in the phrenic nerve diaphragm preparation (12). The IC50 values of the nicotinic ligands were plotted against the biological potencies obtained from the isolated Electrophorus electroplax (Fig. 7). According to the method of linear regression analysis, the data were fitted to the line (correlation coefficient = 0.95). DISCUSSION From the molecular considerations as well as experimental findings presented previously, we have suggested that P3inositide, a membrane acidic lipid, may be a binding component in the nicotinic chloinergic receptor. Using the drug-induced transfer of P3[3H]inositide from an aqueous to a nonaqueous phase, we have determined the IC50 of various nicotinic and muscarinic ligands, noncholinergic neurotransmitters, and cations, and we have found that the IC50 values of the nicotinic
10 6
10 5
10-4
IJ
10-3
IC50, M
FIG. 7. Correlation of the IC50 values of P3[3H]inositide transfer with the in vitro potency of various ligands. The IC50 values of these nicotinic drugs were taken from Table 1 and the values of Kapp (apparent dissociation constant) were obtained from the isolated monocellular electroplaque by Cohen and Changeux (11).
ligands correlate closely with their in vivo neuromuscular blocking activities. The precise nature of the interaction between cholinergic ligands and P3inositide in micellar form is not clear, but complex formation appears to alter the physicochemical properties of the lipid molecule to result in increased hydrophobicity (here defined as tendency to leave the aqueous phase). The IC50 values determined are not solely dependent on the affinity (dissociation constant) of the ligand for P3inositide, but also reflect the apparent change in physicochemical properties of the complex formed. Thus, for any given ligand, the ratio of KD to ICWo is a measure of the relative hydrophobicity of the ligand-P3inositide complex. In the case of the stabilizing blocking agent d-tubocurarine, with a KD of 0.2 ,uM (unpublished data) this ratio is approximately 0.4. However, with succinylcholine, a depolarizing blocking agent (KD 1.6 ,uM), the ratio is much lower (0.013), suggesting that at ligand concentrations close to the dissociation constant the succinylcholine-P3inositide complex is less hydrophobic than the complex of d-tubocurarine with P3inositide. At concentrations of succinylcholine closer to the IC50 value, the icomplex with P3inositide becomes more hydrophobic. Suchbehavior is consistent with ion-pair theory, which states that the formation of intimate ion pairs is dependent on the concentration of the cation in a given concentration of the anion (13, 14). The formation of two different types of succinylcholine-P3inositide complex may have relevance to our understanding of the biphasic effects of depolarizing neuromuscular blocking agents and of discrimination between stabilizing and depolarizing actions of the nicotinic blocking agents. It is well known that calcium ion increases the binding of the cholinergic ligands to their receptors (15, 16). We also observed calcium potentiation of P3[3H]inositide transfer induced by such ligands (Fig. 3). Because 100 ,uM CaCl2 was by itself not sufficient to transfer the Pa[3H]inositide from the aqueous to the nonaqueous phase, the calcium potentiation appeared due to the enhancement of the affinity of the ligand to P3inositide. This has been confirmed and, as in the case of the nicotinic receptor, the binding of ['4C]DMTC and [3H]decamethonium to P3inositide was also enhanced by various concentrations of calcium when the experiments were performed in KrebsRinger solution. However, in Tris.HCI buffer (pH 7.4), calcium enhanced the binding of [14C]DMTC to P3inositide, whereas the binding of [3H]decamethonium was inhibited. Thus, cal-
788
Biochemistry: Cho et al.
cium discriminates the stabilizer from the depolarizer in the drug binding to P3inositide (T. M. Cho and H. H. Loh, unpublished data). Several lines of evidence appear to support a possible role for P3inositide as a binding component of the nicotinic receptor: (i) The concentrations of nicotinic ligands required to induce 50% P3[3H]inositide transfer from the aqueous phase to the nonaqueous phase (IC5o) not only were close to the concentrations required for the ligands' in vio neuromuscular blocking activities and their binding to the purified nicotinic receptor, but also correlated well with their in vitro pharmacological activities obtained from the isolated eel electroplax and reported by Cohen and Changeux (11). (ii) Calcium enhanced the ligand-induced transfer of P3[3H]inositide, and this cation has been shown to increase binding of nicotinic ligands to the cholinergic receptor (15, 16) and to increase binding of [14C]DMTC and [3H]decamethonium to Psinositide.* The dissociation constants for the Ca2+-Painositide complex were almost identical to those for the purified nicotinic receptorCa2+ complex (15). (iii) De Jobertis et al. have shown binding of [14C]DMTC, a nicotinic antagonist, to the purified proteolipid nicotinic receptor isolated from both cat brain (7) and electric eel (17). Such proteolipid receptors appear similar to detergent-extracted cholinergic protein receptors except for their immunological properties (18, 19). Using the method of De Robertis et al. to purify nicotinic cholinergic receptors, we have reported the presence of P3inositide in this purified receptor fraction and shown that the binding of [14C]DMTC is mainly, if not solely, due to its interaction with P3inositide in this fraction (6). Furthermore, the chromatographic behavior of the [14C]DMTC-P3inositide complex was identical to that of the [14C]DMTC complex as reported by De Robertis et al.
(7).
On the basis of the present findings, it is tempting to speculate that neuromuscular blocking activity of the antagonist type may be elicited via the formation of hydrophobic ligand-receptor complexes, thereby stabilizing the membrane with blockade of the ionic conductance. Furthermore, our findings indicate
Proc. Natl. Acad. Sci. USA 75 (1978)
that physicochemical properties (i.e., hydrophobicity and hydrophilicity) of the ligand-receptor complex may be important in the discrimination of the neuromuscular blocking agents from the stimulating agents. 1. Cavallito, C. J. (1962) in Curare and Curare-like Agents, ed. De Reuck, A. V. S. (Little Brown & Co., Boston, MA), pp. 55-70. 2. Bovet, D. (1972) in Neuromuscular Blocking and Stimulating Agents, ed. Cheymol, J. (Pergamon Press, Oxford), Vol. 1, pp. 243-294. 3. Cheymol, J. & Bourillet, F. (1972) in Neuromuscular Blocking and Stimulating Agents, ed. Cheymol, J. (Pergamon Press, Oxford), Vol. 1, pp. 297-356. 4. Cavallito, C. J. (1967) Ann. N.Y. Acad. Sci. 144,900-912. 5. Cho, T. M. & Loh, H. H. (1975) Pharmacologist 17,254. 6. Wu, Y. C., Cho, T. M. & Loh, H. H. (1977) J. Neurochem. 29, 589-592. 7. De Robertis, E., Fiszer, S., Pasquini, J. M. & Soto, E. F. (1969) J. Neurobiol. 1, 41-52. 8. Michell, R. H., Hawthorne, J. N., Coleman, R. & Karnovsky, M. L. (1970) Biochim. Biophys. Acta 210,86-91. 9. Gonzalez-Sastre, F. & Folch-Pi, J. (1968) J. Lipid Res. 9,532533. 10. Bartlett, G. R. (1959) J. Biol. Chem. 234,466-468. 11. Cohen, J. B. & Changeux, J.-P. (1975) Annu. Rev. Pharmacol.
15,83-109. 12. Burn, J. H. & Seltzer, J. (1965) J. Physiol. (London) 179, 569576. 13. Gordon, J. E. (1975) in The Organic Chemistry of Electrolyte Solutions, ed. Gordon, J. E. (Wiley, New York), pp. 371-520. 14. Szwarc, M. (1972) in Ions and Ion Pairs in Organic Reactions, ed. Szwarc, M. (Wiley, New York), Vol. 1, pp. 1-24. 15. Chang, H. W. & Neumann, E. (1976) Proc. Natl. Acad. Sci. USA 73,3364-3368. 16. Cohen, J. B., Weber, M. & Changeux, J.-P. (1974) Mol. Pharmacol. 10, 904-932. 17. De Robertis, E. & De Plazas, S. F. (1970) Biochim. Biophys. Acta 219,388-397. 18. Komentani, T., Ideda, Y. & Kasai, Y. (1975) Biochim. Biophys. Acta, 413, 415-424. 19. De Robertis, E., De Plazas, S. F. & De Carlin, M. C. L. (1976) Nature 259,605.
