The effects of two adamantane derivatives, 1‐trimethylammonio‐5‐(l‐adamantane‐methylammoniopentane dibromide) (IEM‐1460) and 1‐ammonio‐5‐(l‐adamantane‐methylammoniopentane dibromide) (IEM‐1754) on kainate‐induced currents were studied in Xenopus oocytes expressing recombinant ionotropic glutamate receptors and in freshly isolated neurones from rat hippocampal slices. The adamantane derivatives caused use‐and voltage‐dependent block of open channels of recombinant AMPA receptors. This antagonism was dependent on receptor subunit composition; channels gated by recombinant, homomeric GluRl and GluR3 receptors exhibited a higher sensitivity to block than those gated by receptors containing edited GluR2 subunits. In the former cases, IEM‐1460 had an IC50 of 1.6 μm at a holding potential (Vh) of −80 mV and IEM‐1754 was 3.8 times less potent than IEM‐1460. In contrast, 100 μm IEM‐1460 inhibited responses to 100 μm kainate of receptors containing edited GluR2 subunits by only 7.8 ± 2.4% (n= 5 oocytes) at a Vh of −80 mV. Native AMPA/kainate receptors in isolated hippocampal cells were inhibited by adamantane derivatives in a use‐ and voltage‐dependent manner. This antagonism was dependent on cell type: pyramidal neurones were less sensitive to IEM‐1460 (IC50= 1617 μm at Vh=−80mV) than interneurones (IC50= 1.6 μm at Vh=−80 mV). IEM‐1460 and IEM‐1754 were equipotent when applied to pyramidal neurones, but IEM‐1754 was less potent (∼3 times) than IEM‐1460 when applied to interneurones. It is concluded that the presence of the edited GluR2 subunit in recombinant AMPA receptors and native AMPA/kainate receptors inhibits channel block by organic cations and that adamantane derivatives are potentially valuable tools for identifying classes of AMPA/kainate receptors and their roles in synaptic transmission.

Open channel blockers of NMDA receptors interact with the channel gate in different ways. Compounds like MK-801 and phencyclidine exhibit pronounced trapping block, whereas 9-aminoacridine and tetrapentylammonium cannot be trapped. Some blockers such as memantine and amantadine exhibit intermediate properties, so called 'partial trapping'. To analyze the determinants of trapping we have synthesized a series of mono-and dicationic derivatives of phenylcyclohexyl. The blocking action of these compounds as well as that of amantadine has been studied on native NMDA receptors of hippocampal pyramidal neurons. Use-dependence and kinetics of the blockade have been analyzed to estimate the degree of trapping. Dimensions of the blocking molecules apparently do not correlate with their trapping. However, the degree of trapping is voltage-dependent and correlates with the kinetics of unblock. For instance, amantadine behaved as non-trapping blocker at positive voltages, but demonstrated significant trapping at negative voltages. The data may be explained by the model in which the NMDA receptor channel has two binding sites: the shallow and deep ones. Binding to the deep but not to the shallow site allows trapping of the blockers.

Molecular models of the M2 segments of the GluR1 channel have been elaborated using a molecular mechanics approach. The models are based on the homology between pore-lining segments of AMPA receptor channels and the KcsA K+ channel and on cyclic H bonds at the Q/R site of the AMPA receptor channel. The N-terminal region of an M2 segment of the channel is assumed, like that of the K+ channel, to adopt a helical conformation. Due to a deletion, the C-terminal end of the M2 segment of the AMPA receptor is more stretched than that of the K+ channel. As a result, only a single oxygen ring may be exposed to the AMPA receptor channel pore. Data on the block of AMPA receptor channels by dicationic adamantane derivatives have been used to select the most relevant model. The model with the oxygen of a Gly residue (position +2 from the Q/R site) exposed to the pore best fits the experimental data. This model also fits experimental data for another class of AMPA receptor antagonists, the polyamine amides. According to the model, the side-chains of the C-terminal residues are involved in intra-receptor interactions that stabilize the structure of the channel rather than in interactions with ions in the pore.

SUMMARY1. The membrane currents evoked by glutamate agonists on isolated and identified neurones of molluscan pedal ganglia were investigated using the voltage clamp technique.2. The fast chloride current (Er (reversal potential) =-41 mV) evoked in a Ped-9 neurone by application of glutamate, quisqualate and ibotenic acid could be blocked by furosemide (0-1 mM). The slow potassium current (Er = -85 mV) evoked in Ped-8 and Ped-9 neurones by glutamate, quisqualate and kainate could be blocked by tetraethylammonium (50 /M).3. N-Methyl-D-aspartate (NMDA) and a-amino-3-hydroxy-5-methyl-4-isoxazoleproprionic acid (AMPA) failed to induce a response in neurones studies.4. The spider venoms argiopine and argiopinine III (50-500 nM) selectively inhibited quisqualate-induced potassium current, but had no influence on glutamate-, ibotenate-or quisqualate-induced chloride and kainate-induced potassium currents. Glutamate-induced potassium current was partially inhibited by argiopine and argiopinine III.5. The existence of several types of distinct glutamate receptors was confirmed in cross-desensitization experiments, and a lack of interaction was observed between quisqualate and kainate.6. Potassium currents induced both by quisqualate and kainate strongly depended on temperature and could be blocked by pertussis toxin. Intracellular injection of the calcium chelator, EGTA, did not affect quisqualate and kainate responses.7. In neurones loaded with non-hydrolysable GTP analogues, GTP-y-S (guanosine-5'-O-(3-thio)triphosphate) or GppNHp (5'-guanylylimidodiphosphate), the potassium current was gradually induced in the absence of agonists. As this current progressed, the magnitude of the glutamate-or kainate-evoked current transients became smaller and finally negligible. The GTP-y-S-induced current was not inhibited by argiopine.

1The kinetics of open channel block of GluR2-containing and GluR2-lacking AMPA receptors (AMPAR) by dicationic compounds (IEM-1460, IEM-1754, and IEM-1925 have been studied in rat hippocampal neurones using whole-cell patch clamp recording and concentration-jump techniques. Neurones were isolated from hippocampal slices by vibrodissociation. 2 The dicationic compounds were approximately 100 ± 200 times more potent as blockers of GluR2-lacking AMPAR than as blockers of GluR2-containing AMPAR. The subunit speci®city of channel block is determined by the blocking rate constant of a dicationic compound, whereas di erences in unblocking rate constants account for di erences in potency. 3 Hyperpolarization may decrease the block produced by IEM-1460 and IEM-1754 block due to the voltage-dependence of the unblocking rate constants for these compounds. This suggests that dicationic compounds permeate the AMPAR channel at negative membrane potentials. The e ect was particularly apparent for GluR2-lacking AMPAR. These ®ndings indicate that the presence of GluR2-subunit(s) in AMPAR hinders the binding of the cationic compounds and their permeation through the channel.4 The most potent compound tested was IEM-1925. The presence of a phenylcyclohexyl moiety instead of an adamantane moiety, as in IEM-1460 and IEM1754, is probably responsible for the higher potency of IEM-1925. Dicationic compounds are important not only as pharmacological tools, but also as templates for the synthesis of new selective AMPAR blockers which may be potential therapeutic agents.