Nicotinic Acetylcholine Receptor Labeled with a Tritiated

Bioconjugate Chem. 1997, 8, 472−480

472

Nicotinic Acetylcholine Receptor Labeled with a Tritiated, Photoactivatable Agonist: A New Tool for Investigating the Functional, Activated State† Florence Kotzyba-Hibert,*,‡ Pascal Kessler,§ Vincent Zerbib,‡ Thomas Grutter,‡ Christian Bogen,| Kenneth Takeda,| Akli Hammadi,§ Laurent Knerr,§ and Maurice Goeldner‡ Laboratoire de Chimie Bioorganique-URA 1386 CNRS and Laboratoire de Pharmacologie et Physiopathologie Cellulaires-URA 600 CNRS, Universite´ Louis Pasteur Strasbourg, B.P. 24, 67401 Illkirch, France, and De´partement d’Inge´nierie et d’Etudes des Prote´ines, CEA/Saclay, 91191 Gif-sur-Yvette, France. Received December 6, 1996X

Upon agonist activation, the nicotinic acetylcholine receptor undergoes allosteric transitions leading to channel opening and sodium ion influx. The molecular structure of the agonist binding site has been mapped previously by photoaffinity labeling, but most photosensitive probes used for this purpose interact only with closed receptor states (resting or desensitized). We have synthesized two novel photoactivatable 4-diazocyclohexa-2,5-dienone derivatives as cholinergic agonist candidates, with the objective of identifying structural changes at the acetylcholine binding site associated with receptor activation. One of these ligands, 9b, is a functional agonist at muscle acetylcholine receptors in human TE 671 cells. In photolabeling experiments with 9b, up to 35% inactivation of agonist binding sites was observed at Torpedo acetylcholine receptors. Tritiated 9b was synthesized, and photolabeling was found to occur mainly on the R-subunit in a partially protectable manner. This novel radiolabeled photoprobe appears to be suitable for future investigation of the molecular dynamics of allosteric transitions occurring at the active acetylcholine receptor binding site.

INTRODUCTION

Using a fluorescent agonist (Dns-C6-Cho),1 it was shown that the activated nicotinic acetylcholine receptor (AChR) cycles between at least four discrete interconvertible conformational states [resting, R; active, A; intermediate, I; and desensitized, D (Heidmann and Changeux, 1979; Heidmann et al., 1983)]. Channel opening rapidly follows agonist binding (R to A state, microsecond to millisecond time scale) with slower successive transitions toward the intermediate (millisecond to second time scale) and the desensitized states (second to minute time scale). The molecular structure of the acetylcholine (ACh) binding site has been probed by site-directed labels and by mutagenesis. For instance, residues contributing to the ACh binding site of Torpedo AChR were topographi† This work is dedicated to the memory of Prof. Christian Hirth. ‡ Laboratoire de Chimie Bioorganique-URA 1386 CNRS, Universite´ Louis Pasteur Strasbourg. § CEA/Saclay. | Laboratoire de Pharmacologie et Physiopathologie Cellulaires-URA 600 CNRS, Universite´ Louis Pasteur Strasbourg. * To whom correspondence should be addressed at Laboratoire de Chimie Bioorganique-URA CNRS 1386, B.P. 24, 74 route du Rhin, F-67401 Illkirch, France. Telephone: (33) 3 88 67 68 38. Fax: (33) 3 88 67 88 91. E-mail: kotzyba@aspirine. u-strasbg.fr. X Abstract published in Advance ACS Abstracts, June 15, 1997. 1 Abbreviations: Dns-C -Cho, dansyl-C -choline; AChR, nico6 6 tinic acetylcholine receptor; ACh, acetylcholine; DDF, [p-(N,Ndimethylamino)benzene]diazonium fluoroborate; NCB, noncompetitive blocker; R-BuTX, R-bungarotoxin; PCP, phencyclidine; Carb, carbamylcholine; Rf, retention frontal; tR, retention time; PBS, phosphate-buffered saline; t1/2, half-life; λmax, maximum absorbance wavelength; max, molar extinction coefficient; Mp, melting point; Kp, protection constant; Ki, inhibition constant; d-Tubo, d-tubocurarine; ET, energy transfer; EtOAc, ethyl acetate.

S1043-1802(97)00084-0 CCC: $14.00

cally mapped (Dennis et al., 1988; Galzi et al., 1990) to three different loops (loop A, Trp-86 and Tyr-93; loop B, Trp-149 and Tyr-151; and loop C, Tyr-190, Cys-192, Cys193, and Tyr-198) on the R-subunit NH2-terminal domain using the photosensitive antagonist [3H]DDF [[p-(N,Ndimethylamino)benzene]diazonium fluoroborate]. DDF and other photosensitive antagonist labels have also provided information on the ACh binding site in closed states of AChR (Galzi et al., 1991; Kotzyba-Hibert et al., 1995). As the transition from the resting (R) to the active state (A) takes place on a millisecond time scale, photosensitive agonists that have high quantum yields and generate reactive species with short lifetimes are required in order to label specifically the functional state (A) of the AChR. The previously described dynamic mapping of the ACh binding site by the agonist [3H]nicotine (Middleton and Cohen, 1991) suffers from extremely low labeling efficiency (around 1%) following an undefined photocoupling process. Therefore, we developed a new set of photoactivatable ligands displaying agonist activity at the AChR that allowed very efficient labeling of Torpedo marmorata AChR (Chatrenet et al., 1992). These aryldiazonium salts photogenerated a highly reactive species, the aryl cation, having a t1/2 of 1 day, pH 7.2) (Kessler et al., 1990) and show spectral characteristics that are appropriate for energy transfer photoactivation (Goeldner and Hirth, 1980). Binding Properties of 4-Diazocyclohexa-2,5-dienones. Ligands 9a and 9b have micromolar affinities for the ACh binding site in the D state (after proadifen preincubation) with a 10-fold decrease in affinity for the native form, as expected for cholinergic ligands (Table 1). These affinities are similar to those reported previously for 1 (Kp was 6 µM for both 1 and 9b; see Table 1). The probes are highly selective for the ACh binding site compared to the NCB site (Table 1), showing a difference in their respective affinities of ≈2 orders of magnitude (D state). It was not possible to test the affinity of the hydroquinone 10 as oxidation to 11 occurs rapidly, especially in diluted solutions, even at low temperatures. However, 11 was stable and shared an affinity for the ACh binding site, similar to that of 9a and 9b (the Kp value for 11 was 6.5 µM). From these data, it appears that the quaternary ammonium and the quinonoid-like ring constitute the pharmacophore necessary for cholinergic binding. Photolabeling of the ACh Binding Site with [3H]-9b. The diazo photoprobes are stable in the absence of light and generate upon irradiation extremely reactive carbenic species that should be able to react efficiently with the nonactivated C-H bond of the ACh binding site, as schematically represented in Figure 1. We have previously described this interaction of carbenes

478 Bioconjugate Chem., Vol. 8, No. 4, 1997

Figure 3. Photoincorporation of [3H]-9b into the different subunits (R, β, γ, and δ) of the desensitized Torpedo AChR. Receptor-rich membranes (220 pmol) were irradiated at 290 nm for 45 min, in the presence of 1 µM [3H]-9b (3.7 Ci/mmol) and 15 µM proadifen (to obtain the desensitized AChR state). Sixteen picomoles of solubilized photoalkylated AChR was loaded per well and analyzed on 10% SDS-PAGE. Distribution of the radioactivity in the four subunits was measured after gel slicing, digestion, and counting. Photoincorporation was determined without and with a protecting ligand (10 µM d-tubocurarine).

generated from 4-diazocyclohexa-2,5-dienones (Alcaraz et al., 1996). In the dark, 9b is very stable, while upon irradiation, it is very photosensitive, leading to efficient photodecomposition as described previously for similar compounds (Alcaraz et al., 1996). Photolysis of [3H]-9b with Torpedo AChR under ET conditions allowed us to quantify the radioactivity incorporated specifically into the different subunits. As for DDF (Langenbuch-Cachat et al., 1988), the R-subunits were predominantly labeled (Figure 3), with minor contributions of the other subunits (less than 8% for β, γ, or δ). The yield of photoincorporation into the R-subunits using 1 µM [3H]-9b was 6.6% of the total amount of ACh binding sites [6.6%; i.e. 1 pmol alkylated (8700 dpm) per 16 pmol of ACh binding sites loaded on SDS-PAGE per lane]. This 6.6% value is in fact largely underestimated if one takes into account the actual binding site occupancy (14%) in the used experimental conditions (Bayley, 1983). A series of controls were performed to ensure that the observed radioactivity pattern was due to the photolabeling of [3H]-9b; no affinity labeling (identical experimental conditions but in the absence of irradiation) was detected on gels with [3H]-9b, and photoaffinity labeling was negligible with the quinone [3H]-11, showing that covalent labeling is due to diazo coupling and not radical photoreaction from the quinone. Partial protection was observed with 10 µM d-tubocurarine (50% with 1 µM [3H]-9b, n ) 2, Figure 3). A maximum of 34.5% of the ACh binding sites involved in the photocoupling were labeled specifically with 20 µM [3H]-9b (Figure 4). Higher probe concentrations lead to increased nonspecific labeling. The limited photoincorporation might be due to a protector effect of quinone 11 which is formed during irradiation experiments (by oxidation of 10) and which shows a good affinity for the ACh binding site (Kp values of 6.5 and 6 µM for 11 and 9b, respectively). Electrophysiology. ACh-activated single-channel and whole-cell currents in TE 671 cells have been recently described by Kotzyba-Hibert et al. (1996), and our data are in agreement with previous reports (Sine, 1988; Luther et al., 1989). Briefly, single-channel currents observed with 50-500 nM ACh had a mean elementary conductance γ of 31.3 ( 1.9 pS (n ) 8) and a

Kotzyba-Hibert et al.

Figure 4. Concentration dependence of [3H]-9b photoincorporation (0.1-20 µM) into the R-subunits without and with 10 µM d-tubocurarine expressed in disintegrations per minute per mole of AChR loaded on the gel.

mean open time τ of 4.8 ( 0.4 ms (n ) 4). The interpolated zero-current potential was close to 0 mV, as expected for a nonselective cationic channel (Grassi et al., 1993). For desensitizing ACh concentrations (>1 µM), characteristic clusters of burst-like single-channel openings separated by long-duration closures (Sakmann et al., 1980) were observed, and whole-cell inward currents showed rapid decay (Feltz and Trautmann, 1980) during maintained agonist exposure (not shown). 9b Is a Functional Cholinergic Agonist. Singlechannel currents having characteristics essentially similar to those found for ACh were recorded from cellattached patches on TE 671 cells when 9b was included in the pipette at both 6.5 µM (n ) 7, Figure 5A) and 65 µM (n ) 5, Figure 5B). A full dose-response relationship was not established, but clearly, the threshold 9b concentration for noticeable channel activity was higher (g1 µM) compared to that for ACh. No channel activity was observed in the absence of agonist. The current-voltage relationship for 9b was linear (Figure 5C), with an average slope conductance of 30.5 ( 1.7 pS (n ) 5), again with an interpolated reversal potential near 0 mV. In the whole-cell recording configuration, macroscopic inward currents were obtained for 15 s applications of 65 µM 9b (n ) 5, Figure 6A). Such inward currents were maintained throughout the application of 65 µM 9b, showing little or no signs of desensitization, unlike the rapid decay observed during exposure to high ACh concentrations (Kotzyba-Hibert et al., 1996). A complete block of macroscopic currents to 65 µM 9b was produced within 1-2 min following bath application of 50 µM d-tubocurarine, with recovery after a 5 min washout (n ) 5, Figure 6B). The R-substituted compound 9a was also tested at 50 µM in TE 671 cells. No single-channel activity was observed in cell-attached patches (n ) 10, not shown). Similarly, macroscopic, inward whole-cell currents were not elicited by 15 s applications of 50 µM 9a (n ) 10, not shown). Thus, the position of the ammonium-methylene side chain relative to the carbonyl moiety, going from the R-position in 9a to the β-position in 9b, appears to be determinant for agonist activity, with only the β-substituted ligand 9b being a functional agonist. This was also observed with the first cyclohexadienone series (KotzybaHibert et al., 1996). Several characteristics of 9b-activated channels (linear current-voltage, reversal potential, and single-channel

Functional Photosensitive Cholinergic Agonist

Bioconjugate Chem., Vol. 8, No. 4, 1997 479

Figure 6. Macroscopic 9b-activated inward currents from TE 671 cells obtained in the whole-cell recording configuration. (A) The current evoked by 65 µM 9b was maintained during a 15 s application, showing negligible signs of desensitization. 9b was locally microperfused using a pressurized puffer pipette. The holding potential was -70 mV. (B) Block of 9b-activated current by bath application of 50 µM d-tubocurarine and partial recovery after a 5 min washout. The holding potential was -70 mV. Figure 5. Single-channel currents activated by 9b at 6.5 µM (A) and 65 µM (B) recorded from cell-attached patches of TE 671 cells. The pipette potential was 70 mV. The dashed lines indicate current segments shown below on a faster time scale. Data were filtered at 1 kHz. Inward currents are downward. (C) Current-voltage relationship for 9b-activated single-channel currents from a cell-attached patch. The voltage axis is plotted as -Vpipette.

conductance) are quite close to those found for ACh. A notable difference was the apparent lack of desensitization for whole-cell currents during maintained (15 s) applications of 65 µM 9b. Nevertheless, the block of 9binduced whole-cell current by d-tubocurarine together with the single-channel data is highly consistent with 9b activating AChRs on TE 671 cells. CONCLUSION

From both the binding and photolabeling data obtained on Torpedo AChR and the electrophysiological studies on TE 671 cells, we conclude that [3H]-9b is a good candidate for exploring the active state of AChR at the molecular level. [3H]-9b acts as a functional agonist of AChR that irreversibly labels the ACh binding site with efficiency. In photolabeling experiments, over 60% of the radioactivity was located on the R-subunit, as previously found for [3H]DDF (Langenbuch-Cachat et al., 1988). The uncharged photosensitive part of 9b seems not to be critical for cholinergic recognition, unlike [3H]DDF (Langenbuch-Cachat et al., 1988; Dennis et al., 1988; Galzi et al., 1990). Thus, covalent labeling with 9b might occur in different molecular regions of the ACh binding site, thereby providing new insight into the structural basis underlying the functional, activated state of AChR. Along these lines, on the basis of mutations of the δAsp180 and γAsp-174 residues (which are thought to be at an appropriate distance from RCys-192/193 to interact electrostatically with the positively charged ammonium of cholinergic ligands during agonist binding), it was recently concluded that these residues are involved in

the conformational changes whereby receptor subunits move closer to bound agonist in the activated (A) state (Martin et al., 1996). Structural transitions associated with ACh binding have also been described using analysis of electron images of crystallized Torpedo AChR (Unwin, 1995). Upon binding, a cavity (thought to represent the ACh binding site) in the Rδ-subunit (R-subunit in contact with δ) disappears in the activated structure, while the β-subunit moves away from the Rδ-subunit toward the Rγ-subunit. This coordination of structural responses of the two R-subunits is proposed to be central to the cooperative mechanism responsible for channel opening (Unwin, 1995). Also, new information obtained with 9b should allow us to better understand how the transition between the active (A) and desensitized (D) states occurs at the molecular level. From labeling studies with DDF, it was shown that upon desensitization the contribution to the agonist binding site was increased for the δ-subunit and decreased for the γ-subunit (Galzi et al., 1991). Functionally, 9b activates currents through AChR as does ACh, with the exception that essentially no desensitization of whole-cell currents was observed at 65 µM 9b, similar to our recently reported results for 1 (KotzybaHibert et al., 1996). This should be useful for photolabeling the open channel A state of AChR with adapted flash-photolysis techniques. Because 9b seems to modify the allosteric pathway for AChR reactivation, and given that desensitization is kinetically limiting (minute time range), photocoupling of 9b to the A state of AChR may be greatly increased, and thus, very rapid photolabeling may be unnecessary. We intend to use [3H]-9b in photolabeling experiments to characterize the radiolabeled amino acids after purification and sequencing. This molecular investigation of the ACh binding site will first be carried out on the desensitized (D) state. Such characterization of the open channel A state will then be attempted by labeling the receptor using rapid mixing techniques. A more precise molecular understanding of conformational changes un-

480 Bioconjugate Chem., Vol. 8, No. 4, 1997

derlying the dynamic structural processes regulating cholinergic neurotransmission should be possible by comparing the topography of the ACh binding site before, during, and after AChR activation. ACKNOWLEDGMENT

We thank Prof. A. Me´nez (CEA/Saclay) for 3H-labeled N. nigricollis R-neurotoxin, S. Braun and P. Poindron (Laboratoire d’Immunologie, Universite´ Louis Pasteur Strasbourg) for providing the TE 671 cell line, and E. Krempp for technical support. Financial support was from CNRS (Centre National de la Recherche Scientifique), AFM (Association Franc¸ aise contre les Myopathies), and Naturalia and Biologia. LITERATURE CITED Alcaraz, M. L., Peng, L., Klotz, P., and Goeldner, M. (1996) Synthesis and properties of photoactivatable phospholipid derivatives designed to probe the membrane-associated domains of proteins. J. Org. Chem. 61, 192-201. Arnold, B. R., Sciaiano, J. C., Bucher, G. F., and Sander, W. W. (1992) Laser flash photolysis studies on 4-oxocyclohexa-2,5dienylidenes. J. Org. Chem. 57, 6469-6474. Bayley, H. (1983) in Photogenerated reagents in biochemistry and molecular biology (T. S. Work and R. H. Burdon, Eds.) p 97, Elsevier, Amsterdam. Bradford, M. M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248254. Chatrenet, B., Kotzyba-Hibert, F., Mulle, C., Changeux, J.-P., Goeldner, M. P., and Hirth, C. (1992) Photoactivatable agonist of the nicotinic acetylcholine receptor: potential probe to characterize the structural transitions of the acetylcholine binding site in different states of the receptor. Mol. Pharmacol. 41, 1100-1106. Dennis, M., Giraudat, J., Kotzyba-Hibert, F., Goeldner, M., Hirth, C., Chang, J. Y., Lazure, C., Chre´tien, M., and Changeux, J.-P. (1988) Amino acids of the Torpedo marmorata acetylcholine receptor R-subunit labeled by a photoaffinity ligand for the acetylcholine binding site. Biochemistry 27, 2346-2357. Eldefrawi, A. T., Miller, E. R., Murphy, D. L., and Eldefrawi, M. E. (1982) [3H]Phencyclidine interactions with the nicotinic acetylcholine receptor channel and its inhibition by psychotropic, antipsychotic, opiate, antidepressant, antibiotic, antiviral and antiarrhythmic drugs. Mol. Pharmacol. 22, 7281. Feltz, A., and Trautmann, A. (1982) Desensitization at the frog neuromuscular junction: a biphasic process. J. Physiol. (London) 331, 257-272. Galzi, J.-L., Revah, F., Black, D., Goeldner, M., Hirth, C., and Changeux, J.-P. (1990) Identification of a novel amino acid R-Tyr 93 within the cholinergic ligands-binding sites of the acetylcholine receptor by photoaffinity labeling: additional evidence for a three-loop model of the cholinergic ligandsbinding sites. J. Biol. Chem. 265, 10430-10437. Galzi, J.-L., Revah, F., Bouet, F., Me´nez, A., Goeldner, M., Hirth, C., and Changeux, J.-P. (1991) Allosteric transitions of the acetylcholine receptor probed at the amino acid level with a photolabile cholinergic ligand. Proc. Natl. Acad. Sci. U.S.A. 91, 5051-5056. Gasper, S. M., Devadoss, C., and Schuster, G. B. (1995) Photolysis of substituted benzenediazonium salts: spinselective reactivity of aryl cations. J. Am. Chem. Soc. 117, 5206-5211. Goeldner, M. P., and Hirth, C. G. (1980) Specific photoaffinity labeling induced by energy transfer: application to irreversible inhibition of acetylcholinesterase. Proc. Natl. Acad. Sci. U.S.A. 77, 6439-6442. Grassi, F., Giovannelli, A., Fucile, S., Mattei, E., and Eusebi, F. (1993) Cholinergic responses in cloned human TE 671/RD tumour cells. Pfluegers Arch. 425, 117-125. Heidmann, T., and Changeux, J.-P. (1979) Fast kinetic studies on the interaction of a fluorescent agonist with the membrane-

Kotzyba-Hibert et al. bound acetylcholine receptor from Torpedo marmorata. Eur. J. Biochem. 94, 255-279. Heidmann, T., Bernhardt, J., Newmann, E., and Changeux, J.P. (1983) Rapid kinetics of agonist binding and permeability response analyzed in parallel on acetylcholine receptor rich membranes from Torpedo marmorata. Biochemistry 22, 5452-5459. Kessler, P., Ehret-Sabatier, L., Goeldner, M., and Hirth, C. (1990) 4-Diazocyclohexa-2,5-dienones as photoaffinity reagents for proteins. Tetrahedron Lett. 31, 1275-1278. Kotzyba-Hibert, F., Kapfer, I., and Goeldner, M. (1995) Recent trends in photoaffinity labeling. Angew. Chem., Int. Ed. Engl. 34, 1296-1312. Kotzyba-Hibert, F., Kessler, P., Zerbib, V., Bogen, C., Snetkov, V., Takeda, K., Goeldner, M., and Hirth, C. (1996) Novel photoactivatable agonist of the nicotinic acetylcholine receptor of potential use for exploring the functional, activated state. J. Neurochem. 67, 2557-2565. Laemmli, U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 277, 680-685. Lange, G. L., and Gottardo, C. (1990) Facile conversion of primary and secondary alcohols to alkyl iodides. Synth. Commun., 1473-1479. Langenbuch-Cachat, J., Bon, C., Mulle, C., Goeldner, M., Hirth, C., and Changeux, J.-P. (1988) Photoaffinity labeling of the acetylcholine binding sites on the nicotinic receptor by an aryldiazonium derivative. Biochemistry 27, 2337-2345. Luther, M., Schoepfer, R., Whiting, P., Casey, B., Blatt, Y., Montal, M. S., Montal, M., and Lindstrom, J. (1989) A muscle acetylcholine receptor is expressed in the human cerebellar medulloblastoma cell line TE 671. J. Neurosci. 9, 1082-1096. Martin, M., Czajkowski, C., and Karlin, A. (1996) The contribution of aspartyl residues in the acetylcholine receptor γ and δ subunits to the binding of the agonists and competitive antagonists. J. Biol. Chem. 271, 13497-13503. Middleton, R. E., and Cohen, J. B. (1991) Mapping of the acetylcholine binding site of the nicotinic acetylcholine receptor: [3H]nicotine as an agonist photoaffinity label. Biochemistry 30, 6987-6997. Morgat, J.-L., Demares, J., and Cornu, M. (1975) Dispositif automatique de transfert de gaz (tritium, deute´rium, hydroge`ne) (Automatic gas transfer apparatus for tritium, deuterium, and hydrogen). J. Labelled Compd. 11, 257-264. Saitoh, T., and Changeux, J.-P. (1980) Phosphorylation in vitro of membrane fragments from Torpedo marmorata electric organ. Eur. J. Biochem. 105, 51-62. Sakmann, B., Patlak, J., and Neher, E. (1980) Single acetylcholine-activated channels show burst-kinetics in the presence of desensitizing concentrations of agonist. Nature 286, 71-73. Schmidt, T. J., and Raftery, M. A. (1973) A simple assay for the study of solubilized acetylcholine receptors. Anal. Biochem. 52, 349-354. Schoepfer, R., Luther, M., and Lindstrom, J. (1988) The human medulloblastoma cell line TE 671 expresses a muscle-like acetylcholine receptor: cloning of the R subunit cDNA. FEBS Lett. 226, 235-240. Sine, S. M. (1988) Functional properties of human skeletal muscle acetylcholine receptors expressed by the TE 671 cell line. J. Biol. Chem. 263, 18052-18062. Smith, H. (1965) Intermediates in steroid synthesis. Chem. Abstr. 62, 7692-7693. Unwin, N. (1995) Acetylcholine receptor channel imaged in the open state. Nature 373, 37-43. Weber, M., and Changeux, J.-P. (1974a) Binding of Naja nigricollis [3H]-R-toxin to membrane fragments from Electrophorus and Torpedo electric organs. Binding of the tritiated R-neurotoxin in the absence of effector. Mol. Pharmacol. 10, 1-14. Weber, M., and Changeux, J.-P. (1974b) Binding of Naja nigricollis [3H]-R-toxin to membrane fragments from Electrophorus and Torpedo electric organs. Effect of cholinergic agonists and antagonists on the binding of the tritiated R-neurotoxin. Mol. Pharmacol. 10, 15-34.

BC970084+