Abstract
We assessed the pharmacological activity of anabaseine, a toxin found in certain animal venoms, relative to nicotine and anabasine on a variety of vertebrate nicotinic receptors, using cultured cells, theXenopus oocyte expression system, contractility assays with skeletal and smooth muscle strips containing nicotinic receptors and in vivo rat prostration assay involving direct injection into the lateral ventricle of the brain. Anabaseine stimulated every subtype of nicotinic receptor that was tested. It was the most potent frog skeletal muscle nicotinic receptor agonist. At higher concentrations it also blocked the BC3H1 (adult mouse) muscle type receptor ion channel. The affinities of the three nicotinoid compounds for rat brain membranealpha-bungarotoxin binding sites and their potencies for stimulating Xenopus oocyte homomericalpha7 receptors, expressed in terms of their active monocation concentrations, displayed the same rank order, anabaseine>anabasine> nicotine. Although the maximum currents generated by anabaseine and anabasine at alpha7 receptors were equivalent to that of acetylcholine, the maximum response to nicotine was only about 65% of the acetylcholine response. At alpha4-beta2 receptors the affinities and apparent efficacies of anabaseine and anabasine were much less than that of nicotine. Anabaseine, nicotine and anabasine were nearly equipotent on sympathetic (PC12) receptors, although parasympathetic (myenteric plexus) receptors were much more sensitive to anabaseine and nicotine but less sensitive to anabasine. These differences suggest that there may be different subunit combinations in these two autonomic nicotinic receptors. The preferential interactions of anabaseine, anabasine and nicotine with different receptor subtypes provides molecular clues that should be helpful in the design of selective nicotinic agonists.
Neuronal nicotinic receptors have attracted much interest during the past few years, largely due to the discovery that the Alzheimer’s brain loses many of its nicotinic receptors by the time of death, whereas muscarinic receptors are much less affected (Kellar et al., 1987; Araujo et al., 1988). So far, therapeutic approaches directed toward cholinergic systems in the brain have focused on stimulation of postsyaptic muscarinic cholinergic receptors, either directly with muscarinic agonists or indirectly by cholinesterase inhibition. Unfortunately these two strategies have thus far yielded only modest improvements in the cognitive functions of Alzheimer’s patients. Stimulation of brain nicotinic receptors has been shown to enhance cognitive function in lower mammals (Woodruff-Pak et al., 1994; Arendash et al., 1995a, b; Decker et al., 1995; Bjugstad et al., in press), consistent with the idea that these nicotinic receptors may be potential targets treatment of Alzheimer’s and other dementias (Newhouse et al., 1993).
Molecular biological studies have revealed a plethora of nicotinic receptor subunits in the vertebrate brain (Papke, 1993; McGehee and Role, 1995; Lindstrom, 1996; Albuquerque et al., 1997). Although there is still little understanding of the functional consequences of this receptor multiplicity, several labs are investigating the pharmacological properties of the predominant nicotinic receptor subtypes occurring in the nervous system to provide a rational basis for the design of compounds selective for particular nicotinic receptor subtypes that influence particular mental or motor functions (Decker et al., 1995; de Fiebre et al., 1995). Flores et al. (1992) have shown that the major receptor subtype displaying high nicotine, cytisine and methylcarbamyl-choline affinity in the rat brain is the alpha4-beta2 combination. A major receptor subtype showing low affinity for nicotine but high affinity for BTX contains alpha7 subunits (Wonnacott, 1986; Luetje et al., 1990).Alpha7-containing receptors have been implicated in cognitive processes affected by hippocampal function, including sensory gating and spatial memory (Luntz-Leybman et al., 1992; Bjugstad et al., in press).
Pharmacological investigations of nicotinic receptors have been facilitated by the availability of many potent natural toxins, including curare, the erythrina alkaloids, the algal toxin anatoxin-a (Swanson and Albuquerque, 1992), the frog toxin epibatidine (Badio and Daly, 1994; Alkondon and Albuquerque, 1995), the flowering plant toxin methyllcaconitine (Ward et al., 1990), and of course, nicotine. The pharmacological properties of some other potent nicotinic toxins, including leptodactyline (Erspamer, 1959) and anabaseine, have not yet been reported in much detail.
Anabaseine (fig. 1) was initially isolated from a marine worm, but has subsequently been found in certain species of ants (Kem et al., 1971; Wheeler et al., 1981). It is as toxic as nicotine when injected in mice (Kemet al., 1976), stimulates acetylcholine release from rat brain cortical minces (Meyer et al., 1987) and elevates cortical ACh and norepinephrine levels in the intact rat (Summerset al., 1997). As with nicotine, anabaseine enhances passive avoidance behavior in nucleus basalis-lesioned rats (Meyer et al., 1994). At the molecular level anabaseine differs from nicotine and anabasine by having a tetrahydropyridyl ring whose imine double bond is electronically conjugated with the 3-pyridyl ring (fig.1). This causes its two rings to be approximately coplanar in their relative orientation, whereas the two rings in the tobacco alkaloids nicotine and anabasine are almost perpendicular to each other (Whidby and Seeman, 1976; Seeman, 1984).
Because some anabaseine derivatives have been shown to enhance a variety of cognitive behaviors (Meyer et al., 1994;Woodruff-Pak et al., 1994; Arendash et al., 1995b), we examined the pharmacological properties of anabaseine on a variety of vertebrate, mostly mammalian, nicotinic receptors. To quantitatively compare anabaseine with the tobacco alkaloids, we measured the potencies and binding affinities of all three compounds on the same receptors under identical experimental conditions. Several important pharmacological differences were found to exist between the three compounds. Each displays a unique spectrum of action upon the various nicotinic receptors. Our data indicate that these compounds will be useful molecular models to design agonists selective for particular nicotinic receptors. Portions of this study were previously reported in abstract form (Kem and Papke, 1992; Kem et al., 1994a).
Materials and Methods
Chemicals.
Anabaseine was synthesized according to Bloom (1990). The fully ionized, synthetic ammonium-ketone dihydrochloride salt (MW 251) was used in all of our experiments. DMAB-anabaseine dihydrochloride was synthesized as previously described (Kem, 1971;Zoltewicz et al., 1989). Stock solutions of anabaseine, anabasine, nicotine and DMAB-anabaseine were kept in the dark at 5°C for a maximum of 1 wk to avoid deterioration of the alkaloids. (S)-Anabasine free base and reagents used to synthesize anabaseine were obtained from Aldrich (Nukwayjee, WI); (S)-nicotine free base, mecamylamine and other experimental drugs from Sigma Chemical Co. (St. Louis, MO); BTX and TTX from Boehringer-Mannheim (Indianapolis, IN). Radioisotopically labeled compounds, 3H-MCC,125ICl and 86RbCl were purchased from Du Pont-New England Nuclear (Boston, MA).
Frog skeletal muscle contractility.
The two symmetrical rectus abdominis muscles of each frog (Rana pipiens, purchased from Nasco, Ft. Atkinson, WI) were used so that anabaseine potency relative to nicotine or anabasine could be measured on muscles from the same animal. Each muscle was mounted in a 10-ml tissue bath containing frog saline (115 mM NaCl, 5.0 mM KCl, 7.0 mM CaCl2 and 2.0 mM sodium phosphate buffer, pH 7.2) which was continuously bubbled with oxygen at room temperature. The resting tension was initially adjusted to 1.0 g. After 30 min the muscles were briefly (20 sec) contracted with isotonic KCl saline (NaCl replaced with KCl) to measure the maximum isometric force of contracture. After complete recovery, each muscle was challenged with a sequence of 9 or 10 increasing concentrations of agonist until a maximum contractural force was observed. After each application the muscles were washed twice with normal saline and allowed to recover at least 30 min before the next contracture, due to the relatively slow relaxation after exposure to the three alkaloids. After the various agonist concentrations were tested, the final contractility of each muscle was again measured with isotonic KCl saline. A concentration-response curve for each muscle was constructed and expressed as a percentage of the average KCl-induced contracture force. The concentration-response data for each compound was fitted to the Hill equation using SigmaPlot and the EC50 and its S.E. were calculated from the computer-fitted curve.
Patch clamp experiments with neuromuscular type nicotinic receptors.
BC3H-1 cells were cultured according to Covarrubiaset al. (1989). During single channel recordings they were bathed in a saline containing 140 mM NaCl, 5.4 KCl, 10 mM NaHEPES, 1.8 mM CaCl2 and 2.0 mM MgCl2titrated to pH 7.4. Single channels were recorded from cell-attached patches. The pipette saline containing agonist was otherwise identical to the bath saline. In most cases, cells were incubated for 5 to 12 min with 48 nM BTX before recordings to reduce the number of available channels in a patch. Single channel records were stored on videotape using a digital audioprocesser (20 kHz bandwidth). For computer analysis of single channel records, recordings were replayed and digitized at 50 kHz with analog filtering to yield a bandwidth of 5 kHz. Single channel events were analyzed with standard half amplitude threshold crossing methods (Auerbach and Lingle, 1987).
AChR activity was typically examined with agonist concentrations of 5 μM or higher. At such concentrations, channel openings and closings occur in groups that predominantly represent the behavior of single nicotinic receptors as they exit from relatively long-lived desensitized states (Sakmann et al., 1980; Sine and Steinbach, 1984; Auerbach and Lingle, 1987) and that provide information about the true EC50 for half activation of current and the microscopic agonist efficacy, without the complications of desensitization (Ogden and Colquhoun, 1985; Marshallet al., 1991; Lingle et al., 1992). From continuous records of channel activity, lists of channel transitions were subjected to a first pass sifting of groups of openings using an arbitrary group terminator, typically about 30 msec, to generate log-binned histograms (Sigworth and Sine, 1987). Based on the properties of closed intervals distributions, groups were then reselected with a new group terminator value. Group terminators were at least three to five times the closed interval identified as an activation closure (see “Results”), except at 5 μM where the group terminator was only 2-fold the activation closure. Group terminator values were 20 msec or longer for 50, 100 or 200 μM anabaseine, 50 or 100 msec for 20 μM anabaseine, and 100 msec for 5 or 10 μM anabaseine. For analysis of channel blockade by anabaseine, a simple and approximate two-state missed events procedure (Blatz and Magleby, 1986) was selected that only uses the fast component of closures and the longer open interval durations.
The probability of being open within a cluster was determined directly from the closed and open interval durations and the numbers of each component in the histograms (Ogden and Colquhoun, 1985; Marshallet al., 1991). Closures longer than the primary activation closure were considered to separate clusters and the putative blocking gap was also excluded from the po determination. All transmembrane potentials were calculated from the single channel current and the measured single channel conductance of 39 pA, assuming a reversal potential of 0 mV.
Xenopus oocyte expression and functional analysis of rat brain nicotinic receptors.
Preparation of in vitrosynthesized cRNA transcripts and oocyte injection have been described previously (de Fiebre et al., 1995). Recordings were made 2 to 7 days after injections. Oocytes were placed in a Lucite recording chamber with a 0.6 ml total volume and perfused at room temperature with frog saline (115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl2, 10 mM HEPES, pH 7.3) containing 1 μM atropine to block potential muscarinic responses. Calcium-activated chloride channels were not suppressed in our experiments, because their functional presence does not affect the agonist concentration-response relation (Papke et al., 1977a). Drugs were diluted in perfusion solution and then applied after preloading of a 2.0 ml length of tubing at the terminus of the perfusion system. A Mariotte flask filled with Ringer saline was used to maintain a constant hydrostatic pressure for drug deliveries and washes. The rate (6 ml/min) of drug delivery was constant for all compound concentrations and receptor subtypes. Current responses to drug administration were measured using a two electrode voltage clamp with a holding potential of -50 mV. Recordings were made using a Warner Instruments oocyte amplifier interfaced with National Instruments LabView software.
Current electrodes were filled with 250 mM CsCl, 250 mM CsF and 100 mM EGTA, pH 7.3 and had resistances of 0.5 to 2.0 megaohms. Voltage electrodes were filled with 3 M KCl and had resistances of 1 to 3 megaohms. Only oocytes with resting membrane potentials more negative than -30 mV were used. Responses were normalized for the level of channel expression in each individual cell by measuring the response of the oocyte to an initial control ACh application 5 min before presentation of experimental solutions. The control ACh concentrations were 10 μM for alpha4-beta2 receptors and 500 μM for alpha7 receptors. Receptors expressed fromalpha-7 cDNA often display an increased responsiveness after an initial application of agonist that subsequently stabilizes (de Fiebre et al., 1995). Therefore, all alpha7 expressing oocytes received two control applications of ACh separated by 5 min at the start of recording. The second ACh response was then used to normalize the experimental response. For each experimental concentration the mean current response and S.E. were calculated from the normalized responses of at least four oocytes.
After the application of experimental drug solutions, the cells were washed with control Ringer’s solution for 5 min and then evaluated for potential inhibition and response stability by measuring the oocyte response to another application of the ACh (control) solution. These second control responses were normalized to the initial ACh control responses measured 10 min earlier. If the second ACh response showed a difference of > 25% of the initial ACh control response the oocyte was not used for further analysis. If the postexperimental ACh control was within 75% of the preexperiment control ACh response, the second ACh application was allowed to serve as the control response to normalize the response of any subsequent experimental application.
Radioligand binding to nicotinic receptors.
The steady-state binding of the three nicotinic compounds to neuromuscular type receptors was measured indirectly by assessing their ability to inhibit the rate of 125I-BTX binding to Torpedo californica membranes prepared according to Eldefrawi et al. (1980).
The ability of anabaseine to displace the specific binding of 5 nM3H-MCC to rat cerebral cortex synaptosomal membranes was used to assess the ability of anabaseine and related nicotinoid compounds to bind to high nicotine affinity neuronal receptors, which in the rat are primarily of thealpha4-beta2 subtype (Flores et al., 1992). The methods of Boksa and Quirion (1987) were followed with only minor modifications. Binding of 125I-BTX (1 nM, 119 Ci/mmol) to rat brain membranes was performed in a total volume of 0.25 ml, essentially as described by Marks and Collins (1982). Membranes (0.4 mg protein per sample) were incubated with the radioligand at 37°C for 3 hr in the buffer indicated above, which also contained 2 mg/ml BSA. Displacement curves were analyzed by EBDA (Ligand) and Ki values were calculated using the Cheng-Prusoff (1973) equation.
Measurement of ganglionic nicotinic receptor activation.
Rat pheochromocytoma (PC 12) cells grown in the absence of nerve growth factor on polylysine-coated plastic culture dishes were loaded overnight with 86Rb preceding the efflux assays, which were carried out essentially as described by Lukas and Cullen (1988) at pH 7.4. After washing away extracellular rubidium three times, the agonist (in saline containing 10 μM atropine sulfate to inhibit muscarinic receptors and 0.5 mM ouabain to prevent rubidium reuptake by active transport) was added and 1 min later the released rubidium was removed for gamma counting. The rubidium efflux during agonist stimulation was expressed as a percent of total cellular rubidium released by 1 mM carbachol during the same time. All efflux estimates were corrected for spontaneous efflux in the absence of agonist. The amount of 86Rb remaining after the 1 min test period in each cell sample was determined after exposure to 1.0 M NaOH for at least 1 hr. All measurements were done in quadruplicate.
A rat colon (Romano, 1981) preparation was used to assess the agonist activity of anabaseine. Longitudinal muscle strips with intact myenteric plexus were suspended in Tyrode saline (pH 7.4) and aerated with a continuous stream of oxygen bubbles. Each muscle was initially adjusted to a length which maintained a resting tension of 0.5g. Isometric contractions were recorded with a Grass FT.03 force displacement transducer connected to a Grass model 7 polygraph.
i.c.v. administration of anabaseine.
When injected i.c.v. with a nicotinic agonist, rats rapidly become immobile with extended legs (Abood et al., 1981). This prostration response was used to compare the central activity of anabaseine relative to nicotine. After implantation of a metal cannula into the third ventricle, 5 days were allowed for recovery from surgery. The rat received injections with 2 or 4 μl of the experimental compound dissolved in sterile 0.9% NaCl solution, pH 6.5. Prostration was judged to have occurred when all four legs of the Sprague-Dawley male rat (250–300 g) remained laterally extended for at least 10 sec. To detect prostration each rat was observed for at least 5 min after injection. All rats that were not prostrated after injection were killed after an additional i.c.v. injection of Evan’s blue dye to ascertain that the cannula was operational; the result was not used if the dye was absent from the lateral ventricular space.
Results
Activation of frog neuromuscular nicotinic receptors by anabaseine.
Anabaseine acted as a potent nicotinic agonist on frog rectus abdominis muscle. A wide variety of natural toxins and synthetic compounds have previously been tested on this preparation, which facilitated quantitative comparison of anabaseine with these other substances (table 1). When the median effective concentrations of the active, monocationic forms of each compound were compared, anabaseine was only 14- and 3.7-fold less potent, respectively, than epibatidine and anatoxin-a, which are the most potent nicotinic agonists thus far reported. Nicotine was 6.7-fold and cytisine 23-fold less potent than anabaseine.
The stimulatory action of anabaseine was competitively antagonized by the reversible nicotinic antagonist d-tubocurarine (fig.2A) but noncompetitively antagonized by BTX, which due its extremely tight binding, acts essentially irreversibly on skeletal muscle type nicotinic receptors (fig. 2B). At higher concentrations both antagonists completely inhibited the effects of micromolar concentrations of anabaseine, which is consistent with the hypothesis that anabaseine action on the muscle membrane is mediated entirely through nicotinic receptors. The concentration-response curves for anabaseine, anabasine and nicotine are shown in figure 3. The contractures generated by all three of these weakly basic compounds were slow in onset as well as in reversal compared with carbamylcholine (results not shown).
The neuromuscular agonist potency of anabaseine on frog rectus abdominis muscles is compared with those of other naturally occurring nicotinic agonists in table 1. The estimated potency of the monocationic form of each agonist was calculated assuming that its nicotinic stimulatory potency was entirely due to its monocationic form. Much data support this assumption for nicotine (Barlow and Hamilton, 1962; Bartels and Podleski, 1964), anabaseine (Kem et al., 1994b; Kem et al., in preparation) and other nicotinic agonists. The cyclic iminium form of anabaseine was calculated to be approximately 7X more potent than monocationic nicotine.
Interaction of nicotinoid compounds with electric fish muscle nicotinic receptors.
The relative abilities of the three nicotinoid compounds to inhibit 125I-BTX binding to Torpedo electric organ membranes are shown in figure4. Anabaseine displayed the highest affinity of the three compounds although anabasine displayed the lowest affinity (table 2). The relativeKd for the electric fish muscle were very similar with the frog rectus muscle potency (EC50) estimates shown in table3.
Activation of mammalian neuromuscular nicotinic receptors by anabaseine.
The ability of anabaseine to activate mouse embryonic neuromuscular type nicotinic receptors was examined using cell-attached single channel recordings from the clonal BC3H-1 cell line. The single channel conductance in the presence of anabaseine was indistinguishable from that obtained using ACh. As reported for other nicotinic agonists (Colquhoun and Sakmann, 1985; Sine and Steinbach, 1986; Papke et al., 1988), a low anabaseine concentration (40 nM) caused two types of open channel activity: short duration (<500 μsec) bursts resulting primarily from single brief openings and long duration (>3 msec) bursts that were often interrupted by brief closures (results not shown). Histograms of burst durations revealed two components. The average duration of the longer component was somewhat shorter than for bursts activated by ACh, although the voltage-dependence of the burst durations was similar for both agonists. In some cases three components better described the distribution of burst durations, a characteristic also noted for bursts activated by ACh (Colquhoun and Sakmann, 1985).
To better compare the relative effectiveness of anabaseine and ACh as agonists, the behavior of nicotinic receptors activated by 5 μM or higher anabaseine was then examined. The appearance of groupings of channel openings and closings activated by either ACh or anabaseine is shown in figure 5 for three concentrations of each agonist. For both agonists, as the concentration of the agonist increased, the average duration of closures within the periods of activity became shorter and the average time the channel is open within clusters increased. Typically, open interval histograms were best fit with two exponential components, although at the highest anabaseine concentrations only a single open interval component was observed. Closed interval histograms were best described by four or five components. Examination of closed interval distributions revealed one component that occurred with a frequency of about 0.08 to 0.11 per msec open time. This component became shorter in duration with increases in anabaseine concentration and is therefore a strong candidate for a closure reflecting reopenings of a channel from closed states directly leading to channel opening. We term this an activation closure. Support for this idea arises from the fact that the average time between such closures is comparable to the mean burst duration observed at low agonist concentrations. The time constant of the activation closure observed in these histograms was as follows (at -80 to -100 mV): 5 μM: 5.9 ± 1.2 msec (mean S.D.),n = 4 patches; 10 μM: 27.8 ± 3.1 msec,n = 5; 20 μM: 1.2 ± 3.6 msec, n= 8; 50 μM: 3.9 ± 1.9 msec, n = 7; 100 μM 2.4 ± .9 msec, n = 6; 200 μM: 0.8 ± 0.5 msec, n = 5. Closed histograms in many cases contained components at particular agonist concentrations that would be likely to overlap or contaminate these activation closures. However, such additional components in all cases occurred at frequencies considerably less than 0.05 per msec open time. Thus, although these additional closures may bias the properties of the component arising primarily from activation closures, they only minimally alter our estimates of duration and frequency of the activation closures.
A notable characteristic of the openings activated by anabaseine is that, as anabaseine concentration is increased, there is a shortening of the apparent time a channel stays open before closing and an increase in frequency of the short lived closure. This short-lived closure of about 80 to 120 μsec duration was the primary component in the closed interval histograms particularly at 20 μM anabaseine and higher. A short-lived closure of similar duration was also observed at 40 nM anabaseine, but with a frequency of only 0.07 ± 0.04 per msec open time. From the total open time in a record and the number of detected short gaps, a blockage frequency plot (see Ogden and Colquhoun, 1985) was generated (fig. 6A), which possessed a slope of 1.2 × 107M−1 sec−1. This linear dependence of the short gap frequency on anabaseine concentration is consistent with simple channel block behavior, as described for numerous channel blockers (Ogden and Colquhoun, 1985;Marshall et al., 1991). Figure 6B shows the dependence of open interval duration and duration of the fast gap on anabaseine concentration. Values in figure 6B were corrected for missed events. The fit of a simple blocking model to the open interval durations yielded a drug blocking rate of 1.8 × 107M−1 sec−1. Considering the correction for missed events in the latter case, this blocking rate is comparable to that from analysis of the blocking gap frequency. The duration (fig. 6C) of the corrected blocking gap increased with membrane hyperpolarization (e-fold per 66 mV), although the frequency per unit open time of the fast gap (fig. 6D) exhibited only a slight increase (e-fold per 150 mV) with hyperpolarization. Analysis of the voltage-dependence of open interval durations also indicates that the forward rate of blockade is less voltage-dependent than the unblocking rate.
The relative effectiveness of ACh and anabaseine as agonists was examined by comparing cluster po values over a similar concentration range. Cluster pos were determined as described in “Materials and Methods,” based on identification of the activation closure described above. After correction for the occurrence of channel block, 20 μM anabaseine resulted in a po of 0.59 ± 0.12 (eight patches). In comparison, 20 μM ACh resulted in a cluster po of 0.90 ± 0.02 (eight patches). Estimates of po for multiple patches over a range of anabaseine concentrations are plotted in figure7. A fit using the modified Hill equation (see fig. 7 legend) to the po values suggests that the limiting po for anabaseine is about 0.75 ± 0.08 (± 90% confidence limit) with half-activation at 9.6 ± 2.3 μM. The EC50 with ACh was somewhat less than 5 μM (not shown). The limiting po value qualitatively indicates that the efficacy of anabaseine in opening this channel is somewhat less than for ACh. Taking into consideration that only 29% of the anabaseine molecules are in the active cyclic iminium form at pH 7.4, the potency (EC50) of the anabaseine cyclic iminium for activating the receptor was about twice that of ACh.
Rat brain neuronal nicotinic receptors: Xenopus oocyte experiments.
At alpha7 receptors anabaseine and anabasine displayed very similar efficacies, although nicotine was significantly less efficacious (fig. 8A). Anabaseine and anabasine displayed the highest potencies for homomericalpha7 receptor. The concentration dependence of recovery in responsiveness to ACh to each compound displayed the same concentration dependence as its agonist activity (fig. 8B).
Anabaseine and anabasine were only weak partial agonists upon thealpha4-beta2 receptor, displaying 8 and 4%, respectively, of the maximal current elicited by ACh (fig.9). Nicotine displayed a much higher efficacy at this receptor. The EC50s of the three nicotinoid compounds, corrected for their differing degrees of ionization, are presented in table 3. It was previously reported that anabaseine-activated currents in the oocyte were prolonged in comparison with ACh, in their rates of activation and desensitization (de Fiebre et al., 1995). The anabasine and nicotine responses observed in this study also were prolonged relative to those generated by ACh.
Rat brain neuronal nicotinic receptors: radioligand binding experiments.
Next we examined the ability of anabaseine, anabasine and nicotine to bind to rat brain neuronal receptors using two radioligand binding assays, involving displacement of125I-BTX and 3H-MCC binding. Displacement of the first radioligand predominantly measures interaction with alpha7-containing receptors although displacement of the second one measuresalpha4-beta2 receptor binding (Flores et al., 1992).
The relative abilities of the three compounds to displace125I-BTX binding from brain receptors are shown in figure 10. When theKi of anabaseine was expressed in terms of its cyclic iminium form the affinity of anabaseine for these sites was nearly twice that of anabasine and seven times that of nicotine (table2). The displacement data were well fitted by the EBDA single binding site model.
Anabaseine previously was reported to displace the binding of tritiated cytisine to rat brain membranes (Meyer et al., 1994). However, the nature of this inhibition was not determined. We investigated the effect of anabaseine on specific binding of3H-MCC to rat brain high nicotine affinity receptors. Scatchard analysis of 3H-MCC binding in the absence and presence of anabaseine indicated that competitive inhibition occurred, as the slope decreased in the presence of anabaseine while the Bmax was not affected (fig.11).
Our next goal was to assess the affinity of anabaseine, relative to the two tobacco alkaloids. The displacement curves are shown in figure12 and theKi s for the monocationic forms are expressed in table 2. The Ki for nicotine displacement of 3H-MCC from thealpha4-beta2 receptor subtype was about 100X less than for the neuronal BTX binding site, although the affinities of anabaseine and anabasine for these two major brain receptors were very similar.
Anabaseine stimulation of autonomic nicotinic receptors.
All three nicotinic agonists acted as high efficacy nicotinic receptor agonists on PC12 cells (table 3). The major ganglionic nicotinic receptor (Rogers et al., 1992) in this transformed cell line consists of alpha3 and beta4 subunits, and was previously shown to possess relatively low affinity for nicotine and ACh in functional assays (Lukas and Cullen, 1988; Lukas, 1989; Wonget al., 1995). Because maximum nicotinic stimulation of the cells by the three alkaloids and carbamylcholine caused the release of only a small (usually less than 10%) fraction of the internal rubidium, our efflux measurements with this ion should reflect the average extent of receptor activation over the time (1 min) of the measurement. All three compounds acted as full agonists relative to carbachol (concentration-response curves not shown). When concentration was expressed in terms of the active cationic form of each compound, the potencies of the three nicotinoid compounds were very similar (table 3).
In terms of monocationic concentration, the rat colon relaxing potency of anabaseine was very similar to that of nicotine, but significantly greater than that of anabasine (table 3). The action of anabaseine on this preparation was non-competitively inhibited by tetrodotoxin, as would be expected if it acts by exciting myenteric plexus neurons (fig.13).
Intracerebroventricular administration of anabaseine: rat prostration experiments.
On a mole basis, anabaseine was approximately 2.7-fold less potent than nicotine in causing prostration when administered into the lateral ventricle of the unanesthetized rat (fig. 14). If only the monocationic forms of the two compounds are active, this equipotent mole ratio would become nearly one. Nicotine was less potent in our experiments than was previously reported by Abood et al. (1981). This may have been due to our use of a slightly more stringent behavioral endpoint for assessing prostration, as described in the Methods section. Mecamylamine and DHBE both inhibited the prostrating action of anabaseine. A large dose (80 μg) of DMAB-anabaseine failed to prostrate rats but did partially inhibit the prostrating action of nicotine.
Discussion
Anabaseine selectively stimulates nicotinic receptors.
Although anabaseine stimulated all of the nicotinic receptor preparations that we investigated, its high potency upon neuromuscular and alpha7 nicotinic receptors is particularly noteworthy. On the skeletal muscle membrane, anabaseine appears to work entirely on nicotinic receptors, because its action could be completely blocked by the insurmountable antagonist BTX. Nerve action potentials were unaffected by this compound, even at millimolar concentrations (Kem, 1971). However, an effect at nerve terminals cannot yet be ruled out, as some mammalian motoneuron terminals seem to possess nicotinic receptors. Anabaseine affected neither rat brain muscarinic receptors nor plasma cholinesterase, except at very high concentrations (>100 μM) where nonspecific membrane effects often occur (Kem et al., 1994c). Stimulation of the rat brain 5-HT3 receptor, which possesses a subunit sequence homologous with nicotinic receptor subunits, was only inhibited by 24% in the presence of 100 μM anabaseine (Machuet al., 1996). Thus, anabaseine is expected to selectively act on nicotinic receptors at concentrations of less than 100 μM.
Anabaseine actions upon single neuromuscular channels.
The results show that the characteristics of openings and groups of openings activated by anabaseine at both low and high concentrations share many similarities to properties of openings activated by ACh. Based on the total concentration (∼10 μM) of anabaseine at which a half maximal channel open probability is achieved, anabaseine activates the mouse embryonic neuromuscular nicotinic receptor with an apparent affinity slightly less than that of ACh (2–10 μM; Sine and Steinbach, 1987; C. Lingle, unpublished results). However, because the cyclic minimum concentration of anabaseine is only 29% of its total concentration at pH 7.4, this active form of anabaseine is probably slightly more potent than ACh on this mammalian receptor, as at the amphibian neuromuscular receptor (table 1). In our experiments the intrinsic activity or true efficacy (after correcting for its channel-blocking action) of anabaseine displayed a limiting open probability of less than 0.8 compared with values in excess of 0.9 for ACh (Sine and Steinbach, 1987; Zhang et al., 1995). A higher limiting po value for the compound might have been achieved if higher anabaseine concentrations had been tested.
Anabaseine was a more effective open channel blocker than ACh (Ogden and Colquhoun, 1985; Sine and Steinbach, 1984). This conclusion is largely based on the longer duration of the blocking interval produced by anabaseine. Qualitatively, the voltage-dependence of block and the linear dependence of blocking event frequency on anabaseine concentration are both consistent with the idea that anabaseine blocks the nicotinic receptor pore by a simple channel blocking mechanism. For a simple block model, the microscopic affinity of anabaseine for its blocking site is describable by Kd (V) =Kd (0) * expA*V whereKd (0) is the 0-voltage affinity and A describes the voltage-dependence of the block. From the values derived from fitting the blocking and unblocking rates shown in figure 6C and D, Kd (0) = 7 mM with A = 0.024/mV, which corresponds with the movement of a single charged species a little more than halfway through the electric field. However, despite the somewhat stronger channel blocking effect of anabaseine, the affinity of anabaseine for block of the open channel was still much lower than the concentrations effective at activating the receptor. Over the physiological range of membrane potentials, the effectiveKd s for channel block exceeded 500 μM anabaseine, so even at concentrations in excess of about 20 μM the reduction of macroscopic current by channel block would be rather minor.
Electrophysiological comparison of the three alkaloids on oocyte-expressed neuronal nicotinic receptors.
Anabaseine (Papkeet al., 1994) and anabasine (table 3) both displayed low efficacies on the Xenopus oocyte expressedalpha4-beta2 receptor. A submaximal efficacy on this receptor has previously been observed with other potent nicotinic agonists, including cytisine, anatoxin-a, epibatidine, nicotine and the synthetic nicotinic agonist ABT-418 (Papke and Heinemann, 1994; Alkondon and Albuquerque, 1995; Papke et al., in press). Apparently the ligand molecular requirements for activating this receptor subtype are even more stringent than those for high affinity binding. Because most of the agonists that display high affinity are larger, less flexible molecules, high efficacy may be related to an ability to bind within a relatively restricted space on this receptor. An alternative interpretation would be that thealpha4-beta2 receptor channel is more readily blocked by receptor agonists, which would be reflected in a smaller maximum response or efficacy (Papke et al., 1997b). Patch-clamp analyses of the actions of these agonists are clearly needed to determine the basis for the reduced apparent efficacy of these compounds on alpha4-beta2 andalpha7 nicotinic receptors.
We observed a similar rank order, anabaseine> anabasine>nicotine for Xenopus oocyte alpha7 receptor potency (table3) as for rat brain alpha7 receptor affinity, as measured by BTX binding displacement (table 2). In both experiments the apparent affinities of anabaseine and anabasine for this nicotinic receptor were significantly higher than that of nicotine. Due to the rapid desensitization of alpha7 receptors, concentration-response curves for this receptor are quite dependent on the rate of agonist application (Papke et al., 1997). Because differences in experimental methods for agonist application between laboratories prevented us from quantitatively comparing our presentalpha7 data on anabaseine and anabasine with previously published data for nicotine, we again determined the concentration-response relation for nicotine under the same conditions. We found that the apparent efficacy (maximum current) of nicotine for stimulation of the homomeric alpha7 receptor was much less than that of anabaseine, anabasine or ACh (fig. 8).
Comparing the efficacies of the three alkaloids on thealpha7 and alpha4-beta2 receptors, it is readily apparent that anabaseine displays the greatest efficacy and affinity at alpha7 receptors, although atalpha4-beta2 receptors anabaseine and anabasine display a much lower efficacy and affinity than nicotine. This combination of properties predicts that the in vivo actions of anabaseine and anabasine on the brain are mostly mediated throughalpha7 receptors, although the actions of nicotine are largely mediated through alpha4-beta2 and possibly other high nicotine affinity receptors sharing a similar pharmacological profile in receptor binding and efficacy.
Anabaseine interaction with rat brain membrane nicotinic receptors.
Our binding data with the naturally expressed BTX-binding nicotinic receptor is in agreement with our functional data on the oocyte-expressed homomeric alpha7 receptor. Quiket al. (1996) have reported an excellent correspondence between the ligand binding properties of the rat brainalpha7-containing receptor and those of the homomericalpha7 receptor expressed in a transfected cell line, which is consistent with the notion that the alpha7 receptor in the rat brain may also be homomeric. However, Anand et al.(1993) have observed some pharmacological differences between the artificially expressed homomeric chick alpha7 receptor and brain receptors containing the alpha7 subunit, so at least in the chick brain the receptors are not the same.
We have shown that anabaseine acts as a competitive antagonist of3H-MCC, altering the apparent affinity of this radioligand without significantly affecting the receptor concentration available for binding (fig. 11). Compounds that act allosterically at sites other than the ACh recognition site would affect the Scatchard plot for MCC binding in a different manner (Takayama et al., 1989). Although our results indicate that anabaseine primarily interacts with the ACh recognition sites of the receptor, it is possible that anabaseine might also bind to one or more allosteric sites that would not be detected by displacement of tritiated MCC. Indeed, our finding that anabaseine has a channel blocking action above 20 μM on BC3H-1 cell nicotinic channels implies the existence of at least one allosteric site that might be detectable in electric organ membranes by measuring its ability to displace radiolabeled compounds which specifically bind to the nicotinic receptor ion channel (Eldefrawi et al., 1980).
We carried out 3H-MCC binding displacement experiments with all three compounds under identical conditions to facilitate quantitative comparisons between them. Although the rank order of binding affinities, nicotine≫ anabaseine> anabasine (table 2), were in excellent agreement with the rank order of potencies shown in table 3, the Ki and EC50 concentrations for each alkaloid were quite different. These differences arise from the fact that the steady-state binding assay measures the affinity of the desensitized receptor, whereas the functional assay measures the affinity of the activateable receptor. It is interesting that the ratio, EC50/Ki for anabaseine was only 131, compared to a ratio of 3410 for nicotine.
Anabaseine actions on PC12 cells and parasympathetic neurons.
On PC12 cell receptors anabaseine displayed a potency similar to nicotine and anabasine when the extent of ionization was taken into consideration, and the maximal responses (data not shown) were nearly identical with that of carbachol. The uncorrected EC50 value of 29 μM for nicotine stimulation of86Rb efflux is in excellent agreement with other reported nicotine EC50 values for these cells (29 μM, Kemp and Edge, 1987; 20 μM, Lukas, 1989). However, our observation that the maximal effect of nicotine on PC12 cells is comparable with the 1 mM carbamylcholine response differs from some previously reported data that indicated that nicotine’s maximal effect was significantly less than the effect of 1 mM carbamylcholine (Lukas and Cullen, 1988; Lukas, 1989). Several factors, such as differences in the composition or degree of expression of nicotinic receptors between different PC12 cultures, could possibly contribute to such a difference. PC12 cells express other nicotinic receptor subunits besides alpha4 and beta3 (Rogers et al., 1992). PC12 alpha7 receptors bind BTX, but probably have at most, only a small contribution to the rubidium fluxes we measured over a 1-min interval (Kemp and Edge, 1987; Rogers et al., 1991).
Anabaseine apparently relaxes the colon smooth muscle indirectly by stimulating nicotinic receptors on paraympathetic neurons of the myenteric plexus, because TTX blocked its effect. TTX blocks the stimulatory effect of nicotine on guinea pig ileum smooth muscle by depressing the electrical excitability of myenteric plexus neurons (Torocsik et al., 1991). Anabaseine and nicotine were much more potent than anabasine in relaxing the rat colon (table 3). Haefely (1974) also reported that anabasine was only about 4% as potent as nicotine in affecting the cat superior cervical ganglion preparation. The maximal relaxing effect of anabaseine was very similar to that of nicotine, as shown in figure 13. Both anabaseine and nicotine displayed much higher potencies (32- and 25-fold, respectively) in relaxing the rat colon muscle relative to their potency in stimulating86Rb efflux from PC12 cells. Our data suggest that the nicotinic receptors in these two autonomic preparations are probably different in their subunit compositions, and warrants further investigation. Also, Bencherif et al. (1996) reported quite different nicotinoid affinities for PC12 cells and guinea pig ileum nicoinic receptors and suggested that myenteric plexus receptors are composed of alpha3, beta2 and possibly a third subunit.
Whole animal actions of anabaseine.
At an initial stage of this investigation the rat prostration response to lateral ventricular injection of nicotinic agonists was selected as an in vivobioassay for demonstrating neuronal nicotinic agonist activity of anabaseine. The in vivo agonistic and antagonistic activities of a variety of nicotinic compounds, neurotransmitters and toxins had previously been demonstrated using this assay (Aboodet al., 1981). The prostration response displayed pronounced stereo-specificity for the (S)-form of nicotine, as had been observed in radioligand binding experiments with the brain high nicotine affinity binding site. Because displacement of BTX binding to low affinity receptors shows little stereospecificity (Wonnacott, 1986), the existing data suggest that alpha7 type receptors do not play a major role in causing this prostration response. Abood et al. (1981) reported that i.c.v. injection of hexamethonium or mecamylamine immediately preceding nicotine administration partially inhibited its prostrating action. They also reported that preadministration of BTX or TC failed to inhibit the action of prostrating action of nicotine, although TC injected alone caused seizures.
In our prostrations assays anabaseine was approximately equipotent with nicotine, when expressed in terms of the total μmol amount of cyclic iminium form of anabaseine injected. This suggests that the nicotinic receptor mediating prostration behavior is not thealpha4-beta2 type, because this receptor subtype displays much higher (>10X) affinity for nicotine than for anabaseine (table 2). Both DHBE and mecamylamine antagonized the prostrating action of i.c.v. anabaseine. DMAB-anabaseine and other 3-substituted anabaseine derivatives are alpha7 partial agonists but antagonists at alpha3-beta4 and other nicotinic receptors (Kem and Papke, 1992; Papke et al., 1994; de Fiebre et al., 1995). Although the large dose of DMAB-anabaseine failed to cause prostration, it did inhibit the prostrating action of nicotine (fig. 14). Thus, our results and those of Abood et al. (1981) suggest that the nicotinic cholinergic receptors mediating the nicotinic prostration are neither the alpha7 nor the alpha4-beta2 types. The alpha3-beta4 subtype is a candidate receptor for mediating prostration in the rat, because nicotine and anabaseine were found to be of very similar potency (table 3) in stimulating PC12 cell rubidium efflux through nicotinic receptor channels, which are generally considered to be predominantly thealpha3-beta4 combination. This autonomic receptor subtype also displays stereospecificity in its interactions with the two isomers of nicotine (Madhok and Sharp, 1992). Other receptor subunit combinations such as alpha3-beta2 are also possibly involved. Further experiments with compounds selective for particular nicotinic receptor subtypes may assist in the identification of the nicotinic receptors mediating this behavior.
Preferential actions of the three alkaloids upon particular nicotinic receptors.
To compare nicotinic agonists in molecular terms, it is necessary to quantitatively express potencies in terms of the concentration of the active form of each compound. This can be estimated with knowledge of the bulk pH of the saline and the pKa of the ionizable group. Fixed negative charges might alter the local pH at the ACh recognition site, so that it may differ from that of the bulk pH (Stauffer and Karlin, 1994). For instance, a slightly lower local pH at the ACh recognition site would enhance agonist ionization and increase the estimated potencies of secondary and tertiary amine compounds relative to a quaternary ammonium salt like carbamylcholine. Correction for the local pH effect would probably not greatly affect the potency comparisons of the non-quaternary compounds in table 1.
Our examination of the relative potencies and affinities of these closely related compounds provides useful insights for designing nicotinic compounds selective for a particular receptor subtype. Among the three compounds, the anabaseine structure seems optimal for strong neuronal alpha7 and neuromuscular agonist activities, although the nicotine structure seems optimal for designingalpha4-beta2 selective compounds. The anabasine structure, because of its low neuromuscular potency, would serve as a good model for designing alpha7-selective compounds.
Structural comparison of anabaseine with the two tobacco alkaloids.
Both nicotine and anabasine possess a tetrahedral chiral carbon at position 2 of the saturated ring, whereas the same carbon atom in anabaseine is part of a trigonal imine bond whose pi electrons are conjugated with those of the pyridyl ring. Conformational analyses predict that the saturated ring is twisted approximately 90 degrees out of the plane of the pyridyl ring in nicotine and anabasine (Whidby and Seeman, 1976; Seeman, 1984), although the two rings of anabaseine are coplanar (Prokai et al., in preparation).
The electropositive N-methyl group of nicotine will be in the same plane as the 3-pyridyl ring, which makes its presumed receptor-facing surface more similar to that of anabaseine. However, anabasine is predicted to lack a positive charged group in the same plane as its pyridyl ring. Analyses of anabaseine indicate that its two rings are co-planar. The neuromuscular nicotinic receptor ACh recognition sites (there are two, one for each alpha subunit) apparently interact most readily with a positively charged group that resides within the same plane as the pyridyl ring. The neuronalalpha7 ACh recognition sites apparently do not have this coplanarity requirement. Anabaseine and anabasine readily activate this receptor type relative to nicotine, perhaps because their cationic, unmethylated nitrogens are able to make more intimate contact with thealpha7 receptor ACh recognition sites.
Because anabasine and nicotine are thought to possess similar preferred conformations (ring twists) in solution, their differingalpha4-beta2 affinity is probably due to some other chemical differences between the two molecules. We suggest that optimal receptor binding to this receptor occurs when the ligand possesses an N-methyl substituent. Also, other experiments have shown that the greater size of the piperidine ring in anabasine relative to the pyrrolidine ring of nicotine also reduces its interaction with this receptor (Kem et al., unpublished results).
The ligand binding requirements we observed for the parasympathetic-type nicotinic receptors in the rat colon myenteric plexus most closely resembled the neuromuscular receptor requirements. Both receptors displayed relatively strong affinities for nicotine and anabaseine, but a much lower affinity for anabasine.
Comparison of anabaseine with 3-substituted anabaseines.
Our study now provides a foundation for understanding the pharmacological properties of the benzylidene and cinnamylidene derivatives of anabaseines, including DMAB-, DMXB- and DMAC-anabaseines, which preferentially stimulate neuronal nicotinic receptors containingalpha7 subunits (Kem et al., 1994c; Meyeret al., 1994; Papke et al., 1994; de Fiebreet al., 1995) and enhance cognitive behavior (Woodruff-Paket al., 1994; Meyer et al., 1994; Arendashet al., 1995b; Bjugstad et al., in press). As with anabaseine, these compounds display high affinity and efficacy onalpha7 receptors but low affinity and efficacy withalpha4-beta2 receptors. Thus, the 3-substitution of anabaseine merely increases further an alpha7vs. alpha4-beta2 preferential activity already present in anabaseine. 3-Substituted anabaseines also lack significant agonist activity on peripheral nicotinic receptors of the autonomic and neuromuscular types (Kem et al., 1994c). It is extremely interesting that the addition of a 3-substituent to anabaseine seems to diminish its peripheral nervous system andalpha4-beta2 stimulation without reducing centralalpha7 stimulation. This is probably the major pharmacodynamic advantage of the 3-substituted anabaseines over the parent toxin. One of these derivatives, DMXB-anabaseine (also known as GTS-21; Kem et al., 1996), is currently in clinical trials for possible treatment of Alzheimer’s dementia. Further studies are needed to fully understand the nature of the molecular differences between the 3-substituted derivatives of anabaseine and the natural toxin.
Acknowledgments
The authors thank Drs. Linda Bloom and Katalin Prokai-Tatrai for synthesizing anabaseine, Sima Jain and Jeff Thinschmidt for assistance in carrying out the Xenopus oocyte experiments, Baixi Lin for carrying out the intestinal smooth muscle experiments, Ma’an Raja for carrying out the frog rectus muscle experiments and many computer analyses, Linda Abraham for carrying out the rat prostration experiments, Dr. Scott Rogers for providing the PC12 cell line and Ms. Judy Adams for assistance in preparation of the manuscript.
Footnotes
-
Send reprint requests to: Dr. William R. Kem, Department of Pharmacology and Therapeutics, University of Florida College of Medicine, Gainesville, Florida 32610-0267.
-
↵1 This work was supported by Taiho Pharmaceutical Company, Ltd., Tokushima, Japan.
- Abbreviations:
- ACh
- acetylcholine
- BTX
- alpha-bungarotoxin
- DHBE
- dihydro-B-erythroidine
- i.c.v.
- intracerebroventricular
- 3H-MCC
- 3H-methylcarbamylcholine
- TC
- d-tubocurarine
- TXX
- tetrodotoxin
- Received January 7, 1997.
- Accepted June 19, 1997.
- The American Society for Pharmacology and Experimental Therapeutics