Benzylidene Analogs of Anabaseine Display Partial Agonist and Antagonist Properties at the Mouse 5-Hydroxytryptamine_3AReceptor

Materials and Methods

Analogs of 3-Benzylideneanabaseine.

Syntheses of DMXBA and three O-demethylated metabolites (Fig.1) and that of 2-OH and 3-OH benzylidene-anabaseines will be described in a future publication. The homogeneity of each compound was ascertained by nuclear magnetic resonance, mass spectroscopy, elemental composition, and melting point analyses. The dihydrochloride salts were dissolved in the appropriate physiological saline, and stock solutions were aliquoted and frozen. Because DMXBA is light-sensitive, the compounds were not exposed to strong light.

Isolation of Xenopus laevis Oocytes.

X. laevis frogs were kept in tanks of dechlorinated tap water on a 10-h light/14-h dark cycle at 19°C and fed a diet of AquaMax 500 grower (Purina Mills, St. Louis, MO) three times per week. Frogs were anesthetized by immersion in cold 0.12% 3-aminobenzoic acid ethyl ester for 20 min. After removal through a small incision in the frog's abdomen, ovarian lobes were placed in modified Barth's Solution (MBS) containing 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 10 mM HEPES, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, and 0.91 mM CaCl₂, pH 7.5.

Ovarian lobes were dissected manually into clumps of four to 10 oocytes and were then subjected to chemical separation and defolliculation. Clumps of oocytes were placed in medium containing 2 mg/ml collagenase type 2, 83 mM NaCl, 2 mM KCl, 1 mM MgCl₂, and 10 mM HEPES, pH 7.5, and rocked gently for 2 h. Oocytes were then removed to fresh collagenase medium and rocked gently for an additional 2 h. Lastly, oocytes were rinsed with MBS and stored in incubation medium composed of ND96 containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, and 5 mM HEPES, pH 7.5, plus 10 mg/l streptomycin, 50 mg/l gentamicin, 10,000 units/l penicillin, 96 mg/l sulfamethoxazole, 19 mg/l trimethoprim, 0.5 mM theophylline, and 2 mM sodium pyruvate.

Transcription of cDNA to cRNA.

The mouse wild-type 5-HT_3A receptor cDNA [provided by Dr. D. Julius (University of California, San Francisco)] was linearized withNotI, extracted with phenol-chloroform, precipitated with sodium acetate and ethanol, and resuspended in diethyl pyrocarbonate-treated water. The cDNA was then transcribed with T3 mMESSAGE mMACHINE (Ambion, Austin, TX).

Microinjection of Oocytes with 5-HT₃ Receptor cRNA.

An aliquot of cRNA was centrifuged at 15,000g, and the ethanol was removed with a tuberculin syringe. After air drying, the pellet was resuspended in a volume of diethyl pyrocarbonate water to yield a concentration of 5 to 30 ng of cRNA/50 nl. The cRNA was drawn up into a micropipette (10–20-μm tip size). In a volume of 50 nl, cRNA was injected into the animal/vegetal pole equator of each oocyte. Oocytes were stored in incubation medium in Corning cell well plates (Corning Glass Works, Corning, NY) at room temperature. Incubation medium was changed daily. Oocytes were recorded from days 2 through 7 after injection.

Electrophysiological Recordings.

Oocytes were perfused (2 ml/min) in a 100-μl volume chamber with MBS via a roller pump (Cole-Parmer Instrument, Co., Chicago, IL). Two glass electrodes (1.2 mm outside diameter and 1–10 megaohm resistance) filled with 3 M KCl were used to impale oocytes. A Warner Instruments model OC-725B or OC-725C oocyte clamp (Hamden, CT) was used to voltage clamp oocytes at −70 mV. In the voltage-dependent experiments, oocytes were clamped at −100, −70, −40, −10, and 10 mV. Clamping currents obtained at each individual voltage were plotted on a strip chart recorder (Cole Parmer Instrument, Co.).

Serotonin dissolved in MBS buffer was applied for 30 s to oocytes expressing the mouse 5-HT_3A receptor. To examine antagonist effects, BA analogs were coapplied with 5-HT. To characterize partial agonist actions, BA analogs were applied for 30 s. Applications of 5-HT, 5-HT plus BA analogs, or BA analogs were performed every 5 min.

Binding Assays.

Membranes were prepared from NCB-20 cells as described previously (Sun et al., 1999). The protein concentrations were determined with the bovine serum albumin protein assay reagents (Pierce, Rockford, IL). Radioligand binding was accomplished according to the method of Sun et al. (1999), with modifications as described in the following. Binding reactions consisting of crude cell membrane proteins (75 μg/tube), the radiolabeled 5-HT_3Areceptor antagonist, [³H]GR65630 (0.6 nM), and 3-(2,4-OCH3) BA (1 nM–0.5 μM) were incubated for 15 min in a final volume of 250 μl of HEPES buffer (50 mM, pH 7.4) at room temperature. Nonspecific binding was measured in the presence of 50 μM MDL-72222. Incubation was terminated by filtering the reaction mixtures through Whatman GF/B filters (presoaked for 30 min in 0.3% polyethyleneimine) with a Brandel M-24 Cell Harvester (Brandel Inc., Gaithersburg, MD). Filters were then washed four times with 10 ml of 50 mM HEPES buffer, pH 7.4, in the cell harvester. Radioactivity was counted in a Packard scintillation counter. The counting efficiency for tritium was approximately 48%. All analyses were performed in duplicate. Specific binding was determined by subtracting nonspecific binding from total binding. Specific binding values were used for analysis.

Data Analysis.

The values in the 5-HT concentration response curves for mouse 5-HT_3A receptors were expressed as a percentage of the respective maximal 5-HT (10 or 200 μM) responses. Unless otherwise noted, in all other experiments, data were expressed as a percent change from the control, baseline response. In all experiments, the control values were obtained by averaging the 5-HT-mediated response before and after the response to 5-HT, BA analogs, or 5-HT plus BA analogs. In experiments in which agonism was measured, the current measured from test drug stimulation was divided by the average response obtained with the maximal 5-HT concentration and multiplied by 100 to yield the percentage of maximal response. For antagonism, percent inhibition was calculated by subtracting the current obtained from the test drug plus 5-HT from the average current obtained with 5-HT alone; the difference was divided by the average 5-HT-mediated current, and the quotient was multiplied by 100.

Graphpad Prism (San Diego, CA) was used to calculate EC₅₀, IC₅₀, andK_i values and Hill coefficients. The equation used to calculate these parameters was: EC₅₀:I/I_control = 1/[1 + EC₅₀/Dⁿ]; IC₅₀:I/I_control =I/[I + (D/IC₅₀)ⁿ], whereI is current, I_control is the control current, D is the drug concentration, EC₅₀ is the concentration of drug that produces 50% of the maximal response, IC₅₀ is the concentration of drug that produces 50% inhibition of the response, and n is the Hill coefficient. Two-way analysis of variance was performed with Instat (San Diego, CA).

Results

The partial agonist actions of benzylidene-anabaseine compounds were examined in Xenopus oocytes expressing the mouse wild-type 5-HT_3A receptor (Fig.2A). Hydroxy and methoxy substitutions at the 2′ and 4′ positions were studied. All data were normalized as a percentage of the maximal, 10 μM 5-HT baseline response. The compounds 2,4-DiOH-BA and 2-OH-MBA demonstrated good partial agonist activity, with efficacies of ∼63% of the 10 μM 5-HT response. Their EC₅₀ values, 2.6 ± 0.3 and 2.0 ± 0.3 μM, respectively, were greater than that of 5-HT (0.9 ± 0.06 μM). The Hill slope of 2-OH-MBA (1.9) was less than that of 5-HT (3.2), whereas that for 2,4-DiOH-BA (3.1) was essentially the same. In contrast, 3-(2-methoxy,4-hydroxybenzylidene)-anabaseine displayed very poor partial agonist activity, with an efficacy of ∼30% of the maximal 10 μM 5-HT response. An EC₅₀ of 17.2 ± 1.0 μM and a Hill slope of 2.9 were calculated. DMXBA produced no agonist response at all. Collectively, these results suggest that a hydroxyl at the 2′ position is critical for partial agonist activity. In Fig. 2, B and C, representative tracings of 15 μM 2,4-DiOH-BA- and 2-OH-MBA-mediated currents are contrasted with that produced by 10 μM 5-HT. These tracings demonstrate that the hydroxy metabolites of DMXBA are good partial agonists at the 5-HT_3A receptor (Table1).

To further examine the importance of the hydroxyl at the 2′ position, several monosubstituted benzylidene-anabaseine compounds were synthesized and tested (Fig. 1). The actions of 2-OH-BA, 3-OH-BA, and 4-OH-BA were compared with that of 2,4-DiOH-BA and 5-HT (Fig.3). All data were normalized as a percentage of the maximal, 10 μM 5-HT baseline response. 2-OH-BA had the same efficacy as 2,4-DiOH-BA, but was less potent (EC₅₀ of 20.3 ± 2.6 μM), probably because less of it was ionized. A Hill slope of 2.2 was obtained for 2-OH-BA. 4-OH-BA had weak partial agonist activity, as evidenced by its efficacy of ∼14% of the 10 μM 5-HT-mediated response. In addition, the 4-OH-BA compound was less potent than either the 2,4-DiOH-BA or 2-OH-BA compound. The EC₅₀ and the Hill slope were 32.3 ± 5.9 μM and 4.8, respectively, for the 4-OH-BA compound. Relative to 2-OH-BA, the 3-OH-BA analog had no partial agonist activity. Taken together, these results suggest that the 2′ OH group is necessary and sufficient for partial agonist activity of substituted benzylidene-anabaseine compounds at the mouse 5-HT_3A receptor. However, the additional occupation of the 4′ position with either a hydroxy or methoxy group enhances potency.

The two benzylidene-anabaseine analogs, DMXBA and 3-OH-BA, which showed no partial agonist activity at the mouse 5-HT_3A receptor, were also tested for antagonist effects (Fig. 4A). Oocytes expressing the mouse 5-HT_3A receptor were stimulated with 0.5 μM 5-HT in the absence and presence of either compound (1–150 μM). Both compounds produced a concentration-dependent inhibition of 5-HT-mediated currents. The IC₅₀ and Hill coefficient were 15.7 ± 0.9 μM and 1.3, respectively, for DMXBA. 3-OH-BA was less potent, with an IC₅₀ of 27.5 ± 4.7 μM and a Hill coefficient of 1.2. In addition, inhibition produced by both compounds did not reach 100%. The maximal inhibitions achieved with 150 μM DMXBA and 3-OH-BA were 77.7 ± 4 and 85.7 ± 1.6%, respectively.

The mechanism(s) of inhibition of the mouse 5-HT_3A receptor was further examined with DMXBA, because it was the more potent antagonist. Serotonin concentration response curves were performed in the absence and presence of increasing concentrations of DMXBA (Fig. 4B). Data were normalized to the 200 μM 5-HT baseline response. In the presence of 20 and 40 μM DMXBA, the 5-HT EC₅₀ values 1.7 ± 0.05 and 3.6 ± 0.6 μM, respectively, were shifted to the right of that in the 5-HT response curve generated in the absence of drug (0.9 ± 0.06 μM). In the presence of 5-HT (100 μM) and 20 or 40 μM DMXBA, the evoked currents were greater than 90% of the maximal 5-HT response, suggesting competition between the drug and 5-HT at the neurotransmitter binding site. The Hill slopes were 2.6 and 1.3, respectively, in the presence of 20 and 40 μM DMXBA. In contrast, at 100 μM DMXBA, the inhibitory effects were noncompetitive; in the presence of 200 μM 5-HT plus 100 μM DMXBA, evoked currents were only ∼61% of the 200 μM 5-HT baseline response. In the presence of 100 μM DMXBA, the EC₅₀ for 5-HT was 2.8 ± 0.4 μM, and the Hill slope was 4.4.

Many competitive antagonists have additional channel blocking actions at higher concentrations in the related and prototypic nAch receptor (Arias, 1996). Therefore, the possibility that DMXBA noncompetitively inhibits the mouse 5-HT_3A receptor by blocking the ion channel was explored (Figs. 5, A–C). After application of 5-HT plus an ion channel blocking drug, washout of the drug from the pore can elicit a second small “tail” current (Lovinger and Zhou, 1993). This tail current was evident when higher concentrations of the DMXBA analog were used, as indicated in Fig. 5A. Another feature of pore blockers of ligand-gated ion channels is use-dependent inhibition. In the continued presence of the blocking drug, increasing inhibition of currents is observed with sequential applications of ligand. Typically, a very low concentration of drug is used to demonstrate the accumulated inhibitory effect. Because DMXBA appears to have competitive actions at lower concentrations, an intermediate concentration of 50 μM was used with 2 μM 5-HT, approximately an EC₈₅. A typical recording demonstrating use dependence of DMXBA is illustrated in Fig. 5B. In the absence of pretreatment, blockade by DMXBA was 60.4 ± 4.62% (Fig. 5C). After the initial application of DMXBA plus 5-HT, oocytes were perfused continuously with DMBXA for a total of 15 min. During the next three applications of DMXBA plus 5-HT, inhibition increased from ∼93 to 98%. DMXBA application was then terminated, and responses to 5-HT alone were measured every 5 min. Partial recovery from DMBXA blockade was observed. At 30 min post DMXBA application, 5-HT-mediated responses were ∼40% less than the initial baseline control.

Another feature of many, but not all, charged pore blocking compounds is that inhibition is voltage-dependent (Hille, 1992). DMXBA is basic and contains two nitrogens that can be ionized. At the pH 7.5 used in the present study, the pyridine ring nitrogen (pK_a ∼ 3.0–3.5) is uncharged. The tetrahydropyridyl ring (pK_a = 7.62) is ∼57% ionized and 43% un-ionized. The voltage dependence of inhibition by DMXBA of 5-HT-mediated currents was examined (Fig.6). The inhibition of 5-HT-mediated currents by DMXBA was tested at −100, −70, −40, −10, and 10 mV. Inhibition of 0.5 μM 5-HT-mediated currents by 25 and 50 μM DMXBA were measured, whereas 2 μM 5-HT-evoked currents were inhibited with 100 and 200 μM DMXBA. Inhibition statistically significantly increased as a function of increasing DMXBA concentration and less negative voltage; however, the inhibition was not eliminated at more positive potentials. A positively charged channel-blocking drug would be predicted to be expelled from the channel at more positive membrane potentials. Although these results do not rule out the possibility that DMXBA is a pore blocker, they suggest that if indeed this charged compound is one, it is atypical.

To examine the apparent competitive antagonist component of the actions of DMXBA, displacement studies with the 5-HT_3Areceptor ligand, [³H]GR65630, which binds with high affinity to the 5-HT recognition site, were performed in membranes of NCB-20 cells (Fig. 7). Membranes were incubated in the presence of 0.6 nM [³H]GR65630 and DMXBA (1 nM–0.5 mM). Complete displacement of [³H]GR65630 binding was observed, and an IC₅₀ of 1.3 ± 0.2 μM was calculated. The previously determined K_d for [³H]GR65630 was 0.4 nM (Coultrap et al., 1999); a K_i of 0.53 ± 0.9 μM for DMXBA was calculated. Displacement curves with concentrations of [³H]GR65630 higher than 0.6 nM were not examined in any detail due to the observation that nonspecific binding was affected by DMXBA concentrations in excess of 500 μM.

Discussion

DMXBA, which is currently in clinical trials for Alzheimer's disease, and its primary metabolites and related analogs were evaluated for their actions at the mouse 5-HT_3A receptor expressed in Xenopus oocytes. DMXBA antagonized 5-HT_3A receptor-mediated currents in a complex manner indicative of a “mixed” form of antagonism. Lower concentrations (<50 μM) appeared to be competitive, whereas higher concentrations produced a noncompetitive form of inhibition. These results are in agreement with a previous report that determined that the inhibitory action of 100 μM DMXBA on 10 μM 5-HT-mediated currents was noncompetitive (Gurley and Lanthorn, 1998). The IC₅₀ of 15.7 μM determined in the present study was nearly identical with the 14.5 μM value reported by this group. The IC₅₀ for DMXBA at the mouse 5-HT_3A receptor is greater than the plasma concentration attained (1–1.5 μM) in a rat given a standard oral dose of the compound. Therefore, the pharmacological significance of DMXBA at the 5-HT_3A receptor in vivo is likely minimal.

The most interesting finding of this study was that substitution of a hydroxy for the methoxy group at either the 2′ or 4′ position of the benzene ring in DMXBA changed the action of the drug at the 5-HT_3A receptor from antagonism to partial agonism. Substitution with a hydroxy group at the 2′ position produced 2-OH MBA, a compound with good partial agonist activity. Substitution of both the 2′ and 4′ positions with a hydroxy group resulted in a compound with similar efficacy and potency. The monosubstituted 2-OH-BA had the same efficacy as the 2-OH-MBA and 2,4-DiOH-BA compounds, but was less potent. While these results demonstrate that the hydroxy at the 2′ is critical for partial agonist activity, they also indicate that occupation of the 4′ position with either a hydroxy or methoxy group enhances affinity for the receptor. Furthermore, substitution of a hydroxy group for the 4-methoxy group of DMXBA enhances its affinity at the rat α7 receptor (Kem et al., 1996) and its efficacy at human α7 receptors expressed in Xenopus oocytes (Meyer et al., 1998a). It is possible that some of the memory enhancing actions of DMXBA may be caused by these two metabolites. However, it has not yet been demonstrated that these more polar compounds actually reach pharmacologically active concentrations (probably in the 0.1 to 5 μM range) in the brain. Peak plasma concentrations of 2-OHMBA, 4-OHMBA, and 2,4-DiOH-BA (∼500 ng/ml) are lower than those for DMXBA (W. R. Kem, V. M. Mahnir, L. Prokai, X. Cao, F. Soti, R. L. Papke, and K. Prokai-Tatrai, unpublished results; Kitagawa et al., 1998). Given the polarity of these compounds, it is likely that the brain concentrations achieved are significantly less than 500 ng/ml. However, the EC₅₀ concentration of 2-OHMBA for partial agonist activity at the mouse receptor is 2 μM. It is likely that 2-OHMBA has effects on 5-HT₃ mouse receptors in vivo, particularly those in the gut and area postrema, whose access is not limited by the blood-brain barrier.

Minor modifications in the structure of a drug may have profound effects on its ability to bind to its receptor. Few examples exist, however, of compounds in which the substitution or elimination of a functional group at a single position converts a competitive antagonist to a partial agonist or vice versa. For instance, Kooyman and coworkers (1994) have shown that 5-hydroxyindole, which lacks the amine group in the 5-HT molecule, is a low-potency competitive antagonist. In most cases, agonists and their agonist analogs or antagonists and their antagonist analogs, respectively, have been examined. The nAch and 5-HT₃ receptor antagonist,d-tubocurarine, and its various analogs have been investigated thoroughly at the 5-HT_3A receptor (Yan et al. 1998). Removal of the methyl group at the R1 position reduces affinity at the 5-HT₃ receptor by more than 200-fold. Substitution of a methoxy for the hydroxy at the R4 position decreases affinity by more than 50-fold. A number of halogen substitutions in phenylbiguanide, a 5-HT₃receptor agonist, alter affinity (Oxford et al., 1992). Likewise, substitution of single moieties in a series of piperazinopyridopyrrolopyrazine agonist compounds increases their affinity and selectivity for the 5-HT₃ receptor (Prunier et al., 1997).

Although several properties are theorized to render compounds active at the 5-HT₃ receptor, the minimum requirement is that each contains a positively charged nitrogen, which may be embedded in an aromatic or aliphatic heterocycle. Although anabaseine and its derivatives all possess a positively charged nitrogen, anabaseine is far less potent at the 5-HT_3A receptor than its benzylidene analogs (IC₅₀ ∼ 187 μM, data not shown). Clearly, although the addition of a benzylidene ring at a key position of the tetrahydropyridyl ring of anabaseine increases affinity at the 5-HT_3A receptor, it is the presence of a hydrogen bond donor group at the ortho- or para- positions of this ring that determines the efficacy of the compound.Rizzi and coworkers (1990) have proposed, in addition to a hydrogen bond acceptor, a hydrogen bond donor and a hydrophobic domain in the receptor for ligand recognition. It is reasonable to speculate that the entire benzylidene ring structure of DMXBA, its primary metabolites, and related analogs contributes to both spatial and hydrogen bond-accepting interactions with the 5-HT recognition site.

Several amino acids in the 5-HT_3A receptor have been implicated strongly in ligand recognition. In large part, their identification has been guided by studies on homologous nAch receptors, especially the muscle subtype. A number of loops (A–F) have been identified in the nAch receptor as playing a role in ligand recognition in this muscle subtype (for review, see Arias, 2000). Tryptophans in loops A, B, and D in the nAch receptor were identified as putative ligand binding residues through photoaffinity labeling (for review, seeArias, 2000). Corresponding tryptophans in the mouse 5-HT_3A receptor, Trp90 (loop D), Trp121 (loop A), and Tr183 (loop B), and all other tryptophans in the N terminus (Trp60, Trp95, Trp102, Trp195, and Trp214) were mutated to tyrosine and serine (Spier and Lummis, 2000). Mutations at Trp90, Trp183, and Trp195 had marked effects on ligand binding and function, suggesting a critical role in ligand recognition. In another study (Yan et al., 1999), mutations of Trp90, Arg92, and Tyr94 had differential effects on affinity of 5-HT and other 5-HT_3A receptor ligands, suggesting that these ligands have different points of contact with the receptor. 5-HT_3A receptors cloned from different species have differences in sensitivities tod-tubocurarine and 5-HT_3A receptor partial agonists. Several investigators have constructed interspecies chimeras and point mutant receptors to identify domains that are responsible for alterations in responsiveness to drugs (Lankiewicz et al., 1998; Hope et al., 1999; Mochizuki et al., 1999). All domains implicated in ligand recognition include certain amino acids that correspond to the loop C region in the nAch receptor (Arias, 2000). One hypothesis suggests that the aromatic ring (including the tryptophan of 5-HT compounds and the pyridyl ring of nicotinic compounds) and ionizable nitrogen moieties of nAch receptor ligands interact through cationic π interactions (Dougherty and Stauffer, 1990). By analogy, the tetrahydropyridyl ring in DMXBA and related analogs may interact with tryptophans in loops B or D in the 5-HT_3A receptor. Residues in other regions of the N terminus of the 5-HT_3A receptor, such as loop C, may interact with the ortho-hydroxy substituent of the benzylidene ring, given that the presence of a hydrogen-bond donating substituent at this position seems optimal for conferring ability to stimulate receptor function.

In summary, our findings demonstrate that DMXBA and its analogs have markedly different effects at the mouse 5-HT_3Areceptor. Hydroxylation at the ortho-position of the benzylidene ring, which occurs to some extent after oral administration to rats, creates a good partial agonist at the 5-HT_3A receptor. Further investigations combining studies of the structure/activity relationships of the benzylidene-anabaseine compounds and site-directed mutagenesis in the mouse 5-HT_3A receptor should assist in defining the 5-HT recognition site. In addition, such studies may provide even more potent 5-HT_3A receptor drug candidates.