AMPA receptors are involved in mediating most forms of fast glutamatergic neurotransmission. There are four known subunits GluR1 to GluR4 - sometimes termed GluRA to GluRD - which are widely, but differentially, distributed throughout the CNS (Hollmann and Heinemann, 1994). The types of subunits forming these receptors determine their biophysical properties and pharmacological sensitivity. Two alternative splice variants of GluR1 to GluR4 subunits designated as ‘flip’ and ‘flop’ have been shown to differ in their expression throughout the brain and during development and to impart different pharmacological properties (Sommer et al., 1990; Monyer et al., 1991).
Other non-NMDA ionotropic receptor subunits are designated GluR5, GluR6, KA1 and KA2 and normally form receptor assemblies previously designated as high affinity kainate receptors. Kainate receptors were previously believed to be largely presynaptic, for example they are expressed in the dorsal root ganglia, and activation of these kainate receptors has been shown to facilitate transmitter release (Schmitz et al., 2001). LTP and short-term synaptic facilitation is reduced in knockout mice lacking the GluR6, but not the GluR5, kainate receptor subunit suggesting that kainate receptors act as presynaptic autoreceptors on mossy fibre terminals to facilitate synaptic transmission (Contractor et al., 2001). More recent evidence indicates that they are also postsynaptically involved in neurotransmission in some pathways (Abram and Olson, 1994; Wilding and Huettner, 1997; Vignes et al., 1997; Lerma et al., 1997).
Glutamate-induced currents show rapid and profound desensitization in all non-NMDA ionotropic receptors whereas kainate causes much more pronounced desensitization of GluR5, GluR6, GluR7, KA1 and KA2 containing receptors. AMPA is selective for GluR1 to GluR5 containing receptors and induces strong desensitization. SYM 2081, previously assumed to be a kainate receptor antagonist, is actually an agonist (Jones et al., 1997) which produces profound and rapid kainate receptor desensitization and thereby acts as a functional antagonist when continuously present (Wilding and Huettner, 1997). Cyclothiazide is a selective positive modulator of AMPA receptors whereas concanavalin-A is much more effective on kainate preferring receptors (Partin et al., 1993). 2,3-benzodiazepines such as GYKI 52466 are non-competitive AMPA receptor antagonists and are much less active at kainate receptors (Bleakman et al., 1996). Although the 2,3-benzodiazepines and cyclothiazide show strong allosteric interactions, it is now clear that these effects are mediated at different recognition sites (Rammes et al., 1996; Rammes et al., 1998). In general AMPA receptor flip isoforms show somewhat slower desensitization kinetics (Mosbacher et al., 1994) and are more sensitive to the positive modulatory effects of cyclothiazide (Partin et al., 1996).
The GluR2 subunit imparts particular properties to heteromeric AMPA receptors. Receptors containing this subunit show low Ca2+ permeability, linear current-voltage relationships and low sensitivity to block by polyamines and spider toxins. Receptors lacking this subunit show relatively high Ca2+ permeability (Burnashev, 1996), strong rectification i.e. non-linear current-voltage relationships (Verdoorn et al., 1991) mediated by channel blockade via intracellular polyamines such as spermine (Bowie and Mayer, 1995) and are sensitive to block by toxins such as Joro Spider toxin, Philanthotoxin-343 and Argiotoxin-636 (Herlitze et al., 1993; Brackley et al., 1993). The GluR2 subunit shows developmentally-distinct edited and unedited - posttranslational modified protein – forms (Burnashev, 1996) and it is the presence of a positively charged arginine (R) residue in the second membrane-inserted segment (MIS, position 586) of edited receptors that renders then Ca2+ impermeable. Unedited homomeric GluR2 receptors are also much more sensitive to the positive modulatory effects of cyclothiazide. Ca2+-permeable receptors, are most prominent at early stages of development, and show a much more limited distribution in the adult brain (Pellegrini-Giampietro et al., 1992)
There are some indications that Ca2+-permeable AMPA receptors are expressed at higher levels under certain pathological conditions such as global ischaemia (Goldberg et al., 1996). However, they also play an important physiological role on inhibitory g-aminobutyric acid (GABA) interneurons, and selective blockade could lead to excitotoxicity via disinhibition (Racca et al., 1996). Moreover, Ca2+-permeable AMPA receptors seem to have an important role for correct structural and functional relations between Bergman glia and glutamatergic synapses in the cerebellum such as the removal of synaptically released glutamate (Iino et al., 2001).
A very interesting finding is that AMPA receptor activation inhibits ADP-ribosylation and forskolin-stimulated activity of adenylate cyclase in rat cortical neurons (Wang et al., 1997). These effects were independent of Ca2+ and Na+ influx suggesting that the ionotropic AMPA receptor is also directly coupled to metabotropic processes. This is supported by the more recent report that AMPA receptors activates a G-protein (Kawai and Sterling, 1999).
Brief kainate exposure caused long-lasting inhibition of a postspike potassium current (I (sAHP)) in CA1 pyramidal cells and this inhibition did not require ionotropic action or network activity, but was blocked by an inhibitor of pertussis toxin-sensitive G proteins (ethylmaleimide), or the PKC inhibitor calphostin C (Melyan et al., 2002). Agonist dependent downregulation of recombinant NR1/2A receptors by tyrosine dephosphorylation independent of ion flux has also recently been reported (Vissel et al., 2001).
If these forms of ionotropic glutamate receptor activation have a different pharmacology, then they could represent very promising new therapeutic strategies.