Introduction

Top Abstract Introduction Methods Results Discussion References

A menagerie of neuronal AChR subunit cDNAs termed 2- 9 and 2-4 have been identified (Lindstrom et al., 1996; Papke, 1993; Role and Berg, 1996; Sargent, 1993). Sequence analysis revealed that they are members of a larger gene superfamily of ligand-gated ion channels whose members include the GABA_A, glycine and 5-HT₃ receptors (Barnard, 1992; Betz, 1990). In the AChR family, 7, 8 and 9 subunits form functional homomers when expressed in Xenopus oocytes (Couturier et al., 1990; Elgoyhen et al., 1994; Gerzanich et al., 1994; Schoepfer et al., 1990). Other AChR subtypes require coexpression of at least pairs of  and  subunits to form functional AChRs. Numerous investigators have exploited the inherent structural homogeneity of homomeric 7 AChRs to understand the molecular determinants of AChR properties such as desensitization (Revah et al., 1991), ligand affinity (Corringer et al., 1995; Galzi et al., 1991), ion selectivity (Galzi et al., 1992) and modulation by extracellular Ca⁺⁺ binding (Galzi et al., 1996), as well as to demonstrate the extensive structural similarities between members of the gene superfamily of ligand-gated ion channels (Eisele et al., 1993).

We have taken advantage of the structural homogeneity of the closely related homomeric 7 and 8 AChRs, in conjunction with their pharmacological differences, to identify regions of the N-terminal extracellular domain that affect activation of these AChRs. Previously, we have shown that ACh exhibits significantly lower potency (EC₅₀) and DMPP exhibits dramatically lower efficacy for activation of 7 AChRs compared with activation of 8 AChRs (Gerzanich et al., 1994). In this report, we establish the kinetic basis for the partial efficacy of DMPP on 7 AChRs. Single-channel analysis of 7 AChR activity shows that the partial efficacy of DMPP is due to a slower rate of activation by DMPP compared with ACh. In contrast, the rates of desensitization, mean open times and conductances are not significantly different for 7 channels activated by these two agonists. The functional properties of AChRs expressed from chimeras of subunits 7 and 8 have been used to identify regions that are responsible for the pharmacological differences observed between their cognate AChRs. The region 1-179 is responsible for the differences in efficacy of DMPP, whereas the region 180-208 is responsible for the differences in EC₅₀ values for activation by ACh and DMPP. DMPP competes with ACh for activation of 7 AChRs. The efficacy of DMPP is voltage dependent due to voltage dependence of the channel activation rate and not due to channel block. Collectively, our results demonstrate that regions within the N-terminal extracellular domain of 7 and 8 AChRs (and possibly other ligand-gated channels) govern not only the affinity for agonists but also the transition rates between closed and open states.

	Methods

Top Abstract Introduction Methods Results Discussion References

Mutants and chimeras. All site-directed mutagenesis was performed using the Altered Sites II in vitro Mutagenesis System (Promega, Madison, WI). The 7(1-115)/8 chimera was constructed by introducing a BamHI site in the 7 cDNA at a position analogous to the BamHI site in the 8 cDNA. Then, a HindIII/BamHI DNA fragment corresponding to residues 1-115 of the 7 subunit was used to replace the HindIII/BamHI fragment corresponding to residues 1-115 of the 8 subunit. The chimera 7(1-179)/8 was constructed by first introducing an EcoRI site in the 8 cDNA at a position analogous to the EcoRI site in the 7 cDNA. Then, a HindIII/EcoRI fragment corresponding to residues 1-179 of the 7 subunit was used to replace the HindIII/EcoRI fragment corresponding to residues 1-179 of the 8 subunit. The chimera 7(1-208)/8 was constructed by a three-way ligation involving a HindIII/EcoRV fragment corresponding to residues 1-197 of the 7 subunit, an EcoRV/HgaI double-stranded synthetic DNA cassette encoding residues 198-208 and a HgaI/BstEII fragment corresponding to residues 209-481 of the 8 subunit. The chimera 8(1-114)/7(115-179)/8(180-208)/8 was constructed by a three-way ligation involving a HindIII/BamHI fragment corresponding to residues 1-114 of the 8 subunit, a BamHI/EcoRI fragment corresponding to residues 115-179 of the 7 subunit and an EcoRI/BstEII fragment corresponding to residues 180-481 of the 8 subunit. The DNA sequence integrity of all mutant and chimera constructs was verified by standard dideoxy sequencing.

Expression in oocytes. cRNA from linearized cDNA templates was synthesized in vitro using SP6 RNA polymerase in conjunction with reagents from the mMessage mMachine kit (Ambion, Austin, TX). Oocytes were prepared for injection as previously described (Wang et al., 1996). Oocytes were injected with 50 to 100 ng of RNA per oocyte and incubated at 18°C in saline solution with (in mM): NaCl, 96; KCl, 2; MgCl₂, 1; CaCl₂, 1.8; HEPES, 5; pH 7.6.

Oocyte electrophysiology. Currents were measured using a standard two-microelectrode voltage-clamp amplifier (oocyte clamp OC-725; Warner Instrument, Hamden, CT) as previously described (Gerzanich et al., 1995). Electrodes were filled with 3 M KCl and had resistances of 0.5 to 1.0 M for the current electrode and 1 to 2 M for the voltage electrode. All records were digitized (MacLab/2e interface; AD Instruments) and stored on an Apple Macintosh IIcx computer and were analyzed using SCOPE (AD Instruments, Castle Hill, Australia) and Axograph software (Axon Instruments). Data were analyzed using KALEIDAGRAPH (Synergy Software, Reading, PA). The recording chamber was continually perfused at a flow rate of 10 ml/min with saline solution. Application of drugs was performed using a set of eight glass tubes (internal diameter, 2 mm) ending in the bath 3 mm from the oocyte and connected to 10-ml syringes, which were placed 10 cm above the recording chamber. Manual clamping and unclamping of the flexible tube connecting the syringe to the perfusion glass tube directed at the oocytes were used to control the flow of drugs. The delay between the application of the drugs and the appearance of a response was typically 250 to 300 msec, corresponding to the time needed for the drugs to reach the oocyte. At negative holding potentials, the Ca⁺⁺ influx due to the activation of 7 AChRs and 8 AChRs by cholinergic agonists typically leads to the activation of endogenous Ca⁺⁺-dependent Cl channels, which may be responsible for some of the shoulder currents observed in our recordings. In all the experiments described in this report, we did not prevent the activation of these Ca⁺⁺-dependent Cl channels because we have previously shown that agonist EC₅₀ values for activation of 7 and 8 AChRs do not significantly change under conditions in which the currents from Ca⁺⁺ dependent Cl channels are minimized (Devillers-Thiery et al., 1992; Gerzanich et al., 1994) (i.e., in the presence of BAPTA AM or at holding potentials of ~20 mV, which is close to the reversal potentials for the Cl channels in oocytes).

Data analysis. Concentration-response curves were obtained by measuring the response to increasing concentrations of agonists. Only oocytes that gave reproducible responses to a fixed concentration of 1 mM ACh before and after the test doses were used for further analysis. The expression levels of the AChRs were found to vary over the year and even between oocytes obtained from the same animal. Hence, the holding potentials were varied from 30 to 70 mV to obtain responses suitable for analysis. However, the agonist EC₅₀ values for any given AChR did not change significantly as a function of the holding potential at which the experiment was done. The currents elicited by DMPP were always normalized to currents elicited by saturating concentrations of ACh on the same oocyte. The concentration-response curves were fitted to the Hill equation. All concentration-response curves shown are the mean of values obtained from at least three different oocytes.

Single-channel recordings. Xenopus oocytes were manually stripped of the vitelline membrane after osmotic shrinking with a 200 mM potassium aspartate solution (Methfessel et al., 1986). Outside-out configuration patches (Hamill et al., 1981) were formed from stripped oocytes expressing recombinant chicken 7 AChRs from cRNAs that were injected cytosolically at least 4 days earlier. Single-channel currents were activated in patches by a 200-msec application of ACh flowing from one barrel of a two-barrel glass capillary tubing attached to a piezoelectric bimorph application system that was activated by an isolated pulse stimulator (either an A-M Systems model 2100 or Grass Medical Instruments models SD9 or S48-B). Before application of agonist, the patch was isolated in a continuously flowing control solution from the other barrel. Agonists were selected among by a six-way valve. The speed of solution exchange was determined by clamping an open recording electrode at 0 mV and then measuring the change in the solution junction potential when moving from a normal recording solution to one that had been diluted 5-fold with deionized water. The time constant for the change in clamping current during the solution exchange was ~1 msec. To test the system for variability of solution flow rate, the delay between applicator activation and onset of the junction potential change was measured. The onset of the change occurred within 65 µsec of the same time point for each reservoir. Single-channel recording solutions were made in the same saline solution used for the oocyte recordings and the patch pipette contained a solution consisting of (in mM): CsF, 80; CsCl, 20; Cs-EGTA, 10; HEPES, 10; MgATP, 3; pH 7.2. The pH was adjusted with CsOH. Electrodes were formed from borosilicate glass tubing and had resistances typically of 7 to 15 M. Recordings were obtained with an Axopatch 1-D amplifier (Axon Instruments, Foster City, CA) and sampled with an VR-10A digital data recorder (Instrutech Corp., Great Neck, NY) onto video medium (Sharp Corp., Osaka, Japan; VHS model VC-A206C) for later analysis. Signals were sampled off-line at 50 kHz (Axoscope 2.0, Axon Inst.) and filtered at 7 kHz (Frequency Devices, Haverhill, MA; Model 902; eight-pole Bessel, 3 dB) for analysis. The effective cutoff frequency of the analysis was ~6.8 kHz after filtering and sampling. The system, therefore, would not detect events of duration <26 µsec. The 50% threshold analysis would not accurately resolve transitions of durations <50 µsec (Colquhoun and Sigworth, 1983). All single-channel analysis and fitting were performed with pClamp 6.0.3 (Axon Instruments). For channel open time determinations, events list files were formed by visual inspection of the data and manually accepting or rejecting putative events. The data in the events list files were then log-binned into histograms using eight or nine bins per decade and fitted with single- or double- (one patch) exponential functions using maximum likelihood optimization of a Simplex algorithm. The number of components that best described the fits was evaluated by the likelihood ratio. Open-level amplitude histograms were formed by conventional binning of accepted events that were longer in duration than 2.5 times the filter rise time (T_r = 0.3321/f_c) and the resulting distributions were fitted with a Gaussian function of the appropriate number of components. For the ensemble averages (see fig. 2, the data were acquired using fixed-length event driven acquisition in pClamp (Fetchan) using the rising phase of the stimulator artifact at a threshold of 20 pA. The traces (episodes) were then averaged in Axograph 3.55 (Axon Instruments). For figures, histograms with fits were then exported to Origin (Version 4.1; Microcal Software, Northampton, MA). Representative single-channel traces were constructed by opening data files in Axograph and exporting data to Canvas 5.0 (Daneba Software, Miami, FL).

Molecular modeling. Nonconserved residues near the ligand binding site probably help distinguish between ACh and DMPP on 7 and 8 AChRs. Candidate residues were identified with the help of a published molecular model of the extracellular domain of the mouse muscle AChR (Tsigelny et al., 1997) and the crystal structure of DMPP (Chothia and Pauling, 1978), based on the presumed similarity in secondary and tertiary structure between muscle AChRs and 7 and 8 AChRs. The AChR model lacked residues 1-30, which were not included in our analysis. At the 1/ interface, DMPP was placed between the 1 and  subunits and aligned lengthwise with its tertiary nitrogen at the midpoint of the vector between the -carbon atoms of 1 C192 (7 C190) and  D174 (7 D164) and with the quaternary nitrogen closer to  D174 (Karlin and Akabas, 1995). DMPP was positioned similarly at the 1/ interface using  D180 (7 D164) as a reference point on the  subunit. All residues at the 1/ and 1/ interfaces whose -carbons were within 20 Å of the tertiary nitrogen of the bound DMPP ligands were considered to be potential contributors to the discrimination between ACh and DMPP. These residues then were correlated with the equivalent residues of 7 and 8 by primary sequence similarity, and the nonconserved residues between 7 and 8 were identified. Sequence similarities were analyzed with the Pileup utility of the Wisconsin Sequence Analysis Package (Genetics Computer Group, Madison, WI). The model was displayed using InsightII (Molecular Simulations, San Diego, CA).

Simulations of concentration-response curves. Relative current as a function of time was simulated for the activation reaction shown in scheme 1. 

(Scheme 1)

The current was assumed to be proportional to the fraction of the AChR population in the open state A₂O. This fraction was calculated using the set of coupled differential pharmacokinetic equations 1-5, where A represents agonist; and C, AC and A₂C represent unliganded, monoliganded, and diliganded closed AChRs; A₂O represents diliganded open AChR; and D represents desensitized AChR. The association rate constants for agonist binding are k₁ and k₂; the dissociation constants are k_-1 and k_-2; the channel opening rate is ; the channel closing rate is ; and the rates for desensitization and recovery from desensitization are k₄ and k_-4. The solutions were derived with the condition that the amount of bound ligand was negligible relative to free ligand. This set of equations was solved analytically in symbolic form using the method of Laplace transformation (Zhang et al., 1989) and the symbolic computation software Maple V (Release 3, Waterloo Maple, 450 Phillip Street, Waterloo, Ontario, Canada N2L 5J2).

(1)

(2)

(3)

(4)

(5)

After assignment of numerical values for the rate constants, the concentration-response curves were generated by determining the peak relative current at discrete agonist concentrations. The EC₅₀ values and efficacies (I_max) were determined by fitting the simulated data points with a Hill-type equation (equation 6).

(6)

	Results

Top Abstract Introduction Methods Results Discussion References

Kinetic basis of partial efficacy of DMPP. Previously, we showed that 7 AChRs differ significantly from 8 AChRs in their pharmacological properties (Gerzanich et al., 1994). ACh had 100% efficacy on both AChRs but lower apparent affinity for 7 AChRs (EC₅₀ ~110 µM) than for 8 AChRs (EC₅₀ ~2 µM). DMPP had only 9% efficacy on 7 AChRs but 100% on 8 AChRs. DMPP had lower apparent affinity for 7 AChRs (EC₅₀ ~30 µM) than for 8 AChRs (EC₅₀ ~6 µM).

Single-channel properties of 7 AChRs. To distinguish among these possible kinetic mechanisms, we analyzed single-channel currents activated by ACh and DMPP in patches isolated from oocytes expressing 7 AChRs. Applications of ACh or DMPP for 200 msec to outside-out patches evoked a rapid succession of extremely brief single-channel currents (fig. 1). Channel activity was extremely rare when the whole-cell currents were <7 µA (at 50 mV with 300 µM ACh) but were found in most membrane patches when the macroscopic responses were >15 µA. Channel openings began after a short delay (10 msec) after a stimulus artifact. The delay was due to the placement of the recording electrode away from the interface between control and agonist solution (fig. 1). Channel activity ceased either prior to the termination of agonist application (as a result of desensitization) or within 5 to 10 msec after terminating the application, depending on agonist concentration. At higher agonist concentrations, more simultaneous multiple-channel openings occurred and desensitization happened more rapidly. By using ACh or DMPP at 60 µM, channel activity usually was sustained throughout the duration of the application and could be reactivated after 4 sec in control solution. Channel "rundown" was evident during the course of repetitive applications. Enough data to perform kinetic analysis was obtained by repetitively evoking channel activity by these brief agonist applications. When possible, both ACh and DMPP were tested on the same patch (n = 3).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   7 AChR single-channel currents activated by ACh or DMPP have identical mean open times. Top, rapid application of ACh or DMPP for 200 msec to outside-out patches from oocytes expressing 7 AChRs evoked channel activity having extremely rapid open kinetics. Segments of each response to agonist application are also shown on an expanded time scale to illustrate the rapid kinetics. The solid bar above each trace represents the activation period of the application system. The delay between activation and the initial channel activity represents the distance traveled through the control solution into the agonist solution. The large current spikes are stimulator artifacts. Traces showing the entire application period were filtered at 3 kHz, while the expanded traces are shown at 7 kHz because this was the filter frequency for kinetic analysis. Bottom, representative channel open time histograms showing the distributions of channel open times activated by ACh or DMPP. Fits of the distributions to single-component exponential functions show that the mean channel open times for 7 AChRs activated by either agonist are the same. All primary data shown are from the same patch while the mean open time values reflect the fits from 4 or 5 patches. In all cases, the holding potential was 80 mV.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Saturating concentrations of agonists show partial efficacy of DMPP on 7 AChRs in outside-out patches. Left, from top to bottom are sequential applications of saturating concentrations of ACh or DMPP to the same outside-out patch from an oocyte expressing 7 AChRs. The gray bar above each trace represents the time during which the agonist applicator is activated, with the delay to first opening representing the movement of the control solution over the tip of the recording electrode into the agonist solution. The large current spikes on either side of the bar are stimulator artifacts. The time between each application was 2 min. The peak amplitude of simultaneous channel opening was greater for ACh compared with DMPP, reflecting the partial efficacy of DMPP in activating 7 AChRs in outside-out patches. The holding potential was 80 mV. Right, for a different outside-out patch than is shown in the left, ensemble averages of multiple applications of saturating concentrations of ACh or DMPP were constructed. The average traces represent 10 applications of ACh and 11 applications of DMPP. The peak amplitude of the average ACh response was larger than that of DMPP, whereas the onset of the current for DMPP had a slower rise to peak and was delayed. The holding potential was 80 mV.

Analysis of 7/8 subunit chimeras: determinants of efficacy and potency. To determine which regions of 7 and 8 subunits account for the observed differences in efficacy and EC₅₀ between ACh and DMPP, a series of chimeric 7/8 subunits were expressed as homomers in Xenopus oocytes. Their functional properties with respect to activation by ACh and DMPP were compared with those of the 7 and 8 AChRs (table 1).

View this table: [in this window] [in a new window]   TABLE 1 Agonist potencies and efficacies for 7, 8, and chimeric AChRs

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Comparison of agonist efficacies and potencies for the various 7 and 8 subunit chimeras. Shown on top are representations of 7 AChRs, chimeric 7(1-208)/8 AChRs, 7(1-115)/8 AChRs, 7(1-179)/8 AChR, 8(1-114)/ 7(115-179)/8 AChRs and 8 AChRs. Open boxes represent regions of the 8 subunit and hatched boxes represent regions of the 7 subunit. Black boxes represent the transmembrane domains. Numbering refers to residues in the mature 7 and 8 subunit. Typical responses to 2 sec applications (depicted by gray bars) of various concentrations of ACh or DMPP are shown below each representation for that AChR expressed in oocytes. The responses shown for ACh and DMPP are from different oocytes, but the DMPP response in each case is normalized to the maximal ACh response elicited by a single application of saturating concentration of ACh to that oocyte. Each point on the concentration-response relationship represents the normalized mean value from at least three different oocytes and is accompanied by the fit to a Hill equation. The error bars represent the standard error of the mean. The holding potentials ranged between 30 (7 only) and 70 mV depending on peak current amplitudes.

DMPP competitively inhibits ACh-activated currents. To confirm that DMPP was acting only through the agonist binding site, we examined whether its interaction with ACh on 7 AChRs was competitive. Responses to various concentrations of ACh applied alone and coapplied with different concentrations of DMPP were measured as shown in figure 4. In the presence of both its EC₅₀ (30 µM) and at saturating concentrations of DMPP (300 µM), the activation of 7 AChRs by ACh was inhibited in a competitive manner. These results are consistent with DMPP acting solely through competition with ACh for the agonist binding site because any allosteric interaction including channel block by DMPP would be revealed by a submaximal response at the higher end of the concentration-response curve.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   DMPP and ACh interact competitively with 7 AChRs. The peak current elicited by ACh coapplied with either a saturating concentration of 300 µM DMPP or an EC50 concentration of 30 µM DMPP was plotted as a function of the ACh concentration. The peak response in each case is normalized to the peak response at 1 mM ACh in the absence of DMPP. The concentration-response curve for ACh is shifted to the right when going from 30 µM DMPP to 300 µM DMPP and ACh completely displaces DMPP to achieve its maximal response, compatible with competitive inhibition. Each point on the concentration-response relationship represents the normalized mean value from at least three different oocytes and is accompanied by the fit to a Hill equation. The error bars represent the standard error of the mean. The holding potential was 50 mV.

Voltage dependence of the efficacy of DMPP. We established that the partial efficacy of DMPP with 7 AChRs was due to a slower rate of activation ( in scheme 1) by DMPP compared with ACh. Because that step in muscle AChR activation depends on transmembrane potential (Chabala, 1992; Sine and Steinbach, 1986), we examined whether the efficacy of DMPP on 7 AChRs also was voltage dependent. The voltage dependence initially was revealed experimentally by the strong deviation in the macroscopic current-voltage relationship for DMPP from the ohmic relationship observed for ACh at negative holding potentials (data not shown). To address this issue directly, the full current-voltage relationships were recorded for 7 AChRs in response to 1 mM ACh (I_{max, ACh}) or 3 mM DMPP (I_{max, DMPP}) between 90 and +90 mV. The ratio of these values (I_{max, DMPP}/I_{max, ACh}) showed a strong voltage dependence (e-fold/35.5 mV) at negative holding potentials, but little voltage dependence at positive holding potentials (fig. 5). In addition, the efficacy of DMPP dramatically increased to >80% of ACh at positive potentials. In a different series of experiments, 300 µM DMPP was used to inhibit 7 AChR responses elicited by 1 mM ACh at holding potentials from 90 mV to +90 mV. Normalizing to the maximal response elicited by 1 mM ACh alone from the same oocyte showed that the inhibition by DMPP was strongly voltage dependent, being greater at more negative holding potentials (data not shown). However, it should be reiterated that the single-channel and chimera data described above have completely eliminated the possibility of a channel block mechanism for DMPP. Specifically, the mean open times and amplitudes for DMPP-activated channels were identical to those activated by ACh at a concentration of DMPP where its efficacy had peaked. Most persuasively, DMPP is a partial agonist on chimeras having an 8 channel domain, whereas DMPP with the full-length 8 AChR is a full agonist.

View larger version (22K): [in this window] [in a new window]   Fig. 5.   The efficacy of DMPP is voltage dependent. The ratio of the maximal response at a saturating concentration of 3 mM DMPP (Imax, DMPP) to the maximal response a saturating concentration of 1 mM ACh (Imax, ACh) elicited on 7 AChRs () and chimeric 7(1-179)/8 AChRs () is plotted as a function of the holding potential from 90 mV to +90 mV. The data points from responses at negative holding potentials are fitted to an exponential function with the slope of the fit representing the voltage dependence of DMPP efficacy.

	Discussion

Top Abstract Introduction Methods Results Discussion References

We demonstrated that region 1-179 within the 7 AChR N-terminal extracellular domain affects efficacy of DMPP. This region, when swapped into 8 AChRs, converts the efficacy of DMPP from full to partial that is characteristic of its action on 7 AChRs. Partial efficacy is due to a slower opening rate . By comparison, region 180-208 affects EC₅₀ values. These two regions appear to be modular because their effects segregate independently in chimeras. We also conclude that the opening rate  is voltage dependent, suggesting that extracellular region 1-179 in addition to transmembrane regions that contain gating charges affect .

Partial efficacy of DMPP was not due to channel blockade or a faster channel closing rate  for DMPP. DMPP caused a parallel rightward shift in ACh concentration-response curves for 7 AChRs without reducing maximal response. Furthermore, the 7 extracellular domain confers partial efficacy to chimeras having the 8 channel domain, and DMPP is a full agonist on 8. These results eliminated allosteric sites for DMPP including channel blockade as possible mechanisms, leaving only differences in microscopic rates as potential mechanisms. We recorded single-channel currents to determine which rate constants for activation of 7 AChRs could account for partial efficacy of DMPP. Open times of 7 AChRs were similar for the two agonists, ruling out faster  and channel block because these would have caused shorter open times. Furthermore, unitary current amplitudes for DMPP-activated channels were not different from those for ACh-activated channels.

Possible mechanisms, therefore, were narrowed to either a faster desensitization rate out of the closed state (but not the open state) or a decreased rate of activation by DMPP compared with ACh. These possibilities were distinguished by decay in channel opening frequency during agonist application. 7 AChR channels activated at lower agonist concentrations (60 µM) decayed in frequency with time constants of 20 msec for ACh and 18 msec for DMPP. Differences in channel desensitization rates were, therefore, not significant.

Partial efficacy of DMPP was best explained by a slower activation rate . Onsets of opening (first latency) on agonist applications (150 µM ACh or DMPP) were different. Because binding steps were saturated by high DMPP concentration,  for DMPP compared with ACh accounted best for differences in efficacy. Slower  disfavors formation of the open state by DMPP, reducing its efficacy. Efficacy of DMPP should be increased by manipulations that favor formation of the open state, thus offsetting a reduced . Interestingly, ivermectin, an anthelmintic, almost completely restores the efficacy of DMPP by stabilizing the open state of 7 AChRs (Krause et al., 1998). These kinetics classify DMPP as a true partial agonist on chicken 7 AChRs, distinguishing it from partial agonists arising from other mechanisms. A similar conclusion was reached for activation of muscle AChRs by decamethonium, although channel block complicated the interpretation (Liu and Dilger, 1993).

High-sequence identity between 7 and 8 subunits (Schoepfer et al., 1990) facilitated mapping of regions determining differences in efficacy for DMPP, as well as differences in apparent affinity for ACh and DMPP. Only 47 residues are not conserved in the N-terminal extracellular domain of 7 and 8 subunits. They have identical residues in M2 and differ in M1 by 1 residue, in M3 by 3 residues, in the large cytoplasmic domain by 83 residues, in M4 by 6 residues, and in the C-terminus by two residues.

Region 1-179 was a sufficient determinant of efficacy of DMPP. Because proximity to the agonist binding site would seem required for distinguishing ACh from DMPP, it is reassuring that photoaffinity labeling and mutagenesis using Torpedo AChRs identified at least three homologous residues within this region that lie near the binding site: 1 W149,  W57 and  D174 (Changeux et al., 1992). Interestingly, no difference between efficacy of ACh and DMPP was observed in the chimera 8(1-114)/7(115-179)/8, suggesting that nonconserved amino acids within region 115-179 are not sole determinants of partial efficacy of DMPP. Instead, amino acids within region 1-115 of 7 also are required. To account for partial efficacy of DMPP, two or more of the 41 nonconserved amino acid residues within region 1-179 must interact together with DMPP to interfere with structural transitions for activation.

Based on the presumed similarity in secondary and tertiary structure between muscle AChRs and 7/8 AChRs, we used a model of mouse muscle AChR extracellular domain (Tsigelny et al., 1997) to suggest a subset of 14 nonconserved residues between 7 and 8 that may be nearest the binding site at the interface between subunits (fig. 6). These nonconserved residues of 7/8 within 20 Å of bound DMPP in the model include Y/H151, G/S152, S/L155 and L/I156 on the 1-equivalent subunit and Y/E32, F/L33, T/Q34, M/L38, T/I51, I/A54, T/V61, H/I63, H/S115 and G/N167 on the - and -equivalent subunits. The pi-pi interactions between DMPP and aromatic residues of the AChR at positions 32, 33 and 151 may be important in slowing conformational changes triggered by DMPP binding, resulting in a slower .

View larger version (25K): [in this window] [in a new window]   Fig. 6.   Nonconserved residues of 7 that may contribute to the pharmacological differences of ACh and DMPP on 7 and 8 AChRs. A model of the extracellular domain (residues 31-200, based on mouse muscle AChR) of nicotinic AChRs (Tsigelny et al., 1997) was used to find nonconserved residues that may be within 20 Å of the agonist binding site. Top, the -carbon trace of the original muscle AChR model as viewed from the extracellular face with the placement of DMPP at the interface between the 1 and  subunits and on the vector between 1 C192 (7 C190) and  D174 (7 D164) (Karlin and Akabas, 1995). The muscle subunits are labeled , ,  and . A region of 20 Å radius centered on the tertiary nitrogen of DMPP is indicated by the circle. With residues renumbered by primary sequence similarity according to the 7 sequence, segments in the model with carbon atoms within 20 Å of DMPP included residues 110-111, 148-155 and 181-194 from the 1-equivalent subunit and residues 31-41, 45-48, 52-65 and 164-171 from -equivalent subunit. Similar, although not identical, segments at the --equivalent interface of the model were residues 136-137, 150-158 and 181-194 from the 1-equivalent subunit and residues 31-39, 44-46, 48-65, 109-111, 113-115 and 164-170 from the -equivalent subunit. Bottom, region around DMPP and displays the heavy atoms of the following 12 residues of 7 that are not conserved between 7 and 8: Y/H151, G/S152, S/L155 and L/I156 on the 1-equivalent subunits and Y/E32, F/L33, T/Q34, M/L38, I/A54, T/V61, H/I63 and G/N167 on the -equivalent subunit. The -carbons of residues C190 and D164 also are labeled as reference points. Numbering of the residues in the lower panel is based on the 7 sequence. Similar analysis with DMPP at the 1/ interface added T/I51 and H/S115 of the -equivalent subunit to form the subset of 14 nonconserved residues that are close to the agonist binding site in the model and, therefore, that may have a high likelihood of contributing to the pharmacological differences between ACh and DMPP on 7 and 8 AChRs.

A second region, 180-208, determines differences in activation EC₅₀ values of 7 and 8 AChRs between ACh and DMPP. This region contains residues that previously were identified as contributing to agonist binding including Y190, C192, C193 and Y198 (Bertrand et al., 1992; Changeux et al., 1992). A combination of two or more of the six nonconserved residues (184, 186, 187, 198, 200 and 202) within this region accounts for the differences in EC₅₀ values, since EC₅₀ values were not significantly altered when these residues were mutated individually.

Numerical modeling of concentration-response curves confirmed that our kinetic and structural interpretations were consistent with scheme 1 (fig. 7). Differences in opening rates for ACh and DMPP were sufficient to account for differences in efficacy. Moreover, simulations showed that opening rate and agonist binding properties that segregated with regions 1-179 and 180-208, respectively, accounted for observed efficacies and EC₅₀ values of DMPP on 7, 8 and chimeric 7(1-179)/8 AChRs. These results are consistent with region 180-208 affecting EC₅₀ through contributions to ligand binding.

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Numerical modeling of AChR concentration-response curves for DMPP. Numerical simulations of concentration-response curves for DMPP with 7 AChRs, 8 AChRs and chimeric 7(1-179)/8 AChRs are consistent with the assignment of gating effects to region 1-179 and binding effects to region 180-208. The simulations according to linear activation scheme 1 (see Methods) were derived from computed relative peak currents as a function of agonist concentration. The peak DMPP currents were normalized to the ACh-evoked peak current at saturation, which was assumed to be the same for 7, 8 and chimeric 7(1-179)/8 AChRs. The intrinsic equilibrium dissociation constants of the two agonist binding sites were assumed to be equal within each AChR molecule. The closing rate was fixed at 10,000 sec1 in agreement with our single-channel data. Rate constants for the binding steps and the opening rate for ACh were approximated from single-channel analysis that was reported for similar steps of the muscle-type AChR expressed in fibroblasts (Sine et al., 1990). The desensitization rate k4 (100 sec1), recovery rate k-4 (0.02 sec1), association rate k1 (1.6 × 108 M1 sec1) and association rate k2 (8 × 107 M1 sec1) were the same for all three AChRs for both ACh and DMPP. ACh-specific rates for 7 AChR: k-1 = 14,000 sec1, k-2 = 28,000 sec1,  = 45,000 sec1; for 8 AChR: k-1 = 100 sec1, k-2 = 200 sec1,  = 45,000 sec1. DMPP-specific rates for 7 AChR: k-1 = 1800 sec1, k-2 = 3600 sec1,  = 1500 sec1; for 8 AChR: k-1 = 200 sec1, k-2 = 400 sec1,  = 45,000 sec1. The starting point for the numerical simulations was the concentration-response curve for DMPP with 8 AChR, which had full efficacy relative to ACh. The second step was simulation of the 7-like efficacy associated with the 1-179 region of the 7(1-179)/8 chimera while maintaining an 8-like EC50. This effect was obtained by decreasing the opening rate from 45,000 sec-1 to 750 sec1 (value appropriate for 7), which decreased the efficacy to 8% but did not significantly change the 8-like EC50. The third step was an attempt to simulate both the EC50 and efficacy of 7 AChRs. Increasing the dissociation steps k-1 from 200 to 1800 sec1 and k-2 from 400 to 3600 sec1 (values appropriate for 7) changed the EC50 from 6 µM to 50 µM while maintaining the efficacy at 8% ( = 750 sec1). These three steps demonstrate that 7-like or 8-like EC50 values and efficacy could be independently controlled at the agonist binding steps and at , respectively. Moreover, the experimental results from the chimeras together with these simulations are consistent with the conclusion that region 180-208 controls the agonist binding steps and region 1-179 independently controls . The solid lines are fits of the calculated points to the Hill equation (equation 6). The following Hill parameters were found for the simulations of DMPP with 7 AChR: EC50 = 50 µM, n = 1.2; for 8 AChR, EC50 = 5 µM, n = 1.1; for 7(1-179)/8 AChR, EC50 = 6 µM, n = 1.2.

Voltage sensitivity of efficacy of DMPP can be explained by scheme 1. We suggest that residues in region 1-179 decrease efficacy of DMPP by slowing  of 7 AChRs. Furthermore, we propose that  is voltage dependent, regardless of the agonist, because it involves gating components located within the transmembrane electric field. However,  is several-fold slower than  for DMPP and limits peak current amplitude. The current-voltage relationship for DMPP, therefore, is non-ohmic (data not shown), causing voltage-dependent efficacy. In contrast,  for ACh is several times faster than  and is not the rate-limiting step for peak current amplitude. Therefore, even though  is voltage dependent for ACh, the relationship between voltage and peak current remains ohmic at negative potentials. Failure of DMPP to approach the efficacy of ACh for the chimeric 7(1-179)/8 AChR regardless of the holding potential reinforces the idea that the magnitude of  is the major determinant of DMPP efficacy. Additionally, voltage dependence of  implies linkage between determinants of partial efficacy of DMPP and charges within the transmembrane electric field that move during gating. Voltage dependence of gating of muscle-type AChRs in BC3H-1 (Sine and Steinbach, 1986), rat muscle (Chabala, 1992) and mouse muscle AChRs in transfected HEK 293 cells (Auerbach et al., 1996) is well established. Molecular determinants within muscle AChR subunits conferring voltage dependence of gating were investigated in the study using transfected HEK 293 cells. Voltage dependence of channel closure was related directly to the number of agonist molecules bound by the AChR. They concluded that charge movement within the transmembrane electric field during channel closure confers voltage sensitivity of gating. This suggested linkage between the extracellular domain and charged residues located within the membrane that serve as voltage sensors much like what we are suggesting for 7 AChRs.

Mechanisms of other partial agonists on AChRs have been reported. Nicotine was a partial agonist on chicken 3 2 but not rat 3 2 AChRs (Hussy et al., 1994). Residue 198 of 3 played a crucial role in this species-specific difference. However, partial efficacy of nicotine was voltage independent and may involve binding at a site overlapping the agonist binding site of these AChRs. Binding at this adjacent site by nicotine would antagonize activation. Cytisine was a partial agonist on 3 2 AChRs but a full agonist on 3 4 AChRs based on competition for activation by ACh (Papke and Heinemann, 1994), demonstrating that 4 conferred full efficacy to cytisine. Decamethonium and succinylcholine are partial agonists and channel blockers on muscle AChRs (Adams and Sakmann, 1978; Marshall et al., 1990). True partial agonism of decamethonium on muscle AChRs results from an extremely slow  of 32 sec¹, although it also is a potent channel blocker (Liu and Dilger, 1993). Here, we have demonstrated the kinetic basis of true partial agonism of DMPP on chick 7 AChRs. Compared with ACh, interaction of DMPP with the agonist binding site is less efficiently coupled to gating due to interactions with residues 1-179 of the N-terminal extracellular domain, which is reflected in a slower channel opening rate.