Glutamate activates two types of receptors, ligand gated ion channels and G protein-coupled metabotropic glutamate receptors (mGluRs). At least eight mGluR subtypes have been cloned to date, and these receptors can be placed into three major groups on the basis of their pharmacology, second messenger coupling and sequence homology. Group III mGluRs consist of four different subtypes (mGluR4, 6, 7 and 8) and are activated by the selective agonist L (+)-2-amino-4-phosphonobutyrate (L-AP4). The localization of these receptors within presynaptic active zone is consistent with their role as autoreceptors mediating the feedback inhibition of glutamate release. By using biochemical and immunocytochemical techniques as well as Ca²⁺ imaging from single nerve terminals we have determined the contribution of the different signalling pathways (inhibition of Ca²⁺ channels activity, decrease in cAMP levels) to glutamate release inhibition in a preparation of cerebrocortical nerve terminals. In addition, we have studied the distribution of the distinct group III mGluRs in subpopulations of nerve terminals depending upon the type or combination Ca²⁺ channels present.

In nerve terminals from adult rats (2 months) we find that L-AP4 (1 mM) inhibited the Ca²⁺-dependent evoked release of glutamate by 25 %. This inhibition of release was mediated primarily by activation of mGluR7, which exhibited low affinity for the agonist L-AP4. This inhibitory effect was largely prevented by pertussis toxin but was insensitive to the inhibitor of protein kinase C (PKC), bisindolylmaleimide, and protein kinase A (PKA), H-89. Furthermore, this inhibition was associated with reduction in N-type Ca²⁺ channel activity in the absence of any detectable change in cAMP levels. In the presence of forskolin, however, L-AP4 decreased the levels of cAMP. The activation of this additional signalling pathway was very efficient in counteracting the facilitation of glutamate release induced either by forskolin or by the β-adrenergic receptor agonist isoprenaline. The specific inhibition by mGluR7 of the release component associated with the activity of N-type channels was unexpected because these Ca²⁺ channels support glutamate release to a lesser extent (29.9 %) than P/Q-type Ca²⁺ channels (72.7 %). However, imaging experiments to measure Ca²⁺ dynamics in single nerve terminals have revealed a specific co-localization of mGlu7 receptors and N-type Ca²⁺ channels. Ca²⁺ channels were distributed in a heterogeneous manner in individual nerve terminals as they contained N-type (31.1% conotoxin GVIA sensitive) P/Q-type (64.3% agatoxin-IVA sensitive) terminals or terminals insensitive to these two toxins (4.6 %). Interestingly, the great majority of the responses to L-AP4 (95.4 %) were located in N-type channels containing nerve terminals. This specific co-localization of mGlu receptor 7 and N-type Ca²⁺ channels could explain the failure of the receptor to inhibit the release component associated with P/Q-channels and also reveals the existence of a specific mechanism of targeting to place the two proteins in the same subset of nerve terminals.

In cerebrocortical nerve terminals from young rats (3 weeks), two mGluRs with high and low affinity for L-AP4 were identified by immunocytochemistry as mGluR4 and mGluR7, respectively. Ca²⁺ imaging experiments showed that voltage-dependent Ca²⁺ channels are distributed in a more heterogeneous manner than in adult animals. Thus, presynaptic terminals contained only N-type (47.5% conotoxin GVIA-sensitive), P/Q-type (3.9% agatoxin-IVA-sensitive), or both N- and P/Q-types of Ca²⁺ (42.6 %) channels, although the remainder of the terminals (6.1 %) were insensitive to these two toxins. Interestingly, mGluR4 was largely (73.7 %) located in nerve terminals expressing both N- and P/Q-type Ca²⁺ channels (N-P/Q terminals), whereas mGluR7 was predominantly located (69.9 %) in N-terminals. This specific co-expression of different group III mGluRs and Ca²⁺ channels may endow synaptic terminals with distinct release properties and reveals the existence of a high degree of presynaptic heterogeneity.

This work was supported by grants from DGESIC (PB97-0321), MCYT (BFI2001-1436) and CAM (08.5/0016/1998 and 08.5/0075.1/2000).