As synapses form between pre- and postsynaptic elements in the developing brain, neurotransmitter receptors are inserted into the postsynaptic membrane. Little is known about how this is initiated but there is strong evidence to suggest that, at excitatory glutamatergic synapses, NMDA receptors (NMDARs) are the first to appear and that AMPARs are then inserted in response to NMDAR activation. Synapses which contain NMDARs but lack AMPARs have been observed both electrophysiologically and anatomically in developing hippocampus and cortex (Isaac, 2003). This has led to the idea that the presence of NMDARs may be a prerequisite for AMPAR insertion at newly formed synapses. Early evidence in support of this view came from electrophysiological experiments in which electrical stimulation protocols that normally induce long-term potentiation resulted in the rapid appearance of an AMPAR-mediated current at synapses where previously synaptic currents were mediated purely by NMDARs. It was subsequently shown directly, using AMPARs tagged with green fluorescent protein, that AMPARs can indeed be inserted into synapses following NMDAR activation (Shi et al. 1999).
In this issue of The Journal of Physiology, Lachamp et al. (2005) revisit this issue in the cerebellum, where mature Purkinje cells (PCs) do not express either synaptic or extrasynaptic NMDA receptors (Perkel et al. 1990; Llano et al. 1991). Functional NMDARs formed from NR1 and NR2D subunits are, however, expressed by PCs in the first postnatal week (Momiyama et al. 1996), but their significance is not yet understood. Unlike NR2A- and NR2B-containing NMDARs expressed by most other neurones, those containing the NR2D subunit are found extrasynaptically, and have slow deactivation kinetics (Cull-Candy et al. 2001). PC NMDARs were initially suggested to be involved in the elimination of multiple climbing fibres (Rabacchi et al. 1992), but the NMDAR dependence of this process was later shown to occur after the phase of NMDAR expression by PCs and was suggested to involve local circuitry rather than being a direct affect of blocking receptors expressed by the PCs themselves (Kazikawa et al. 2000). Lachamp et al. (2005) ask whether the early expression of NMDARs in PCs may instead be required for the targeting of AMPA receptors to excitatory synapses. Using a compelling combination of a recently developed antigen retrieval technique, to enhance sensitivity of immunolabelling, and electrophysiology, to demonstrate synaptic function, they show that, even at embryonic stages, PCs already express AMPAR subunits on primordial dendrites, albeit at low density, and that their location tends to follow patterns of developing climbing fibre innervation. The appearance of AMPARs seems to be totally independent of NMDAR activation. Although NMDARs are present, they are randomly localized and not associated with synaptic contacts. Moreover, activity-dependent block of NMDARs in utero with the antagonist MK-801 had no effect on the appearance and clustering of AMPARs in PCs.
In PCs therefore a mechanism different from that in hippocampus and cortex must exist to drive AMPARs into synapses. It is known that the trafficking of the type of AMPARs expressed by PCs (GluR2/3) is strongly regulated by N-ethylmaleimide-sensitive factor (NSF; Lin & Sheng 1998), and this interaction has recently been shown to be required in PCs for the synaptic incorporation of AMPARs from extrasynaptic sites and their subsequent removal during long-term depression (Steinberg et al. 2004). However, since this is thought to occur at synapses that already have AMPARs (Shi et al. 2001), what triggers the initial insertion of AMPARs into the PC plasma membrane remains an open question.