AMPA receptors & synaptic plasticity Part 2

Following on from the introduction, I now discuss a number of recent studies which demonstrate that synaptic strengthening in different regions of the mammalian brain requires the incorporation of Ca2+-permeable GluR2-lacking AMPA receptors into the postsynaptic membrane of active or newly-potentiated synapses.

Neurotransmitter release regulates GluR1 levels at the postsynaptic membrane

Harms et al (2005) inhibited neurotransmitter release from specific subsets of cultured hippocampal neurons by transfecting individual cells with the tetanus toxin light chain tagged with cyan fluorescent protein, which effectively cleaves synaptobrevin and therefore prevents fusion of synaptic vesicles with the presynaptic membrane.

Antibody staining showed that there was a significant reduction in GluR1 levels in the postsynaptic membrane of silenced synapses compared to neighbouring active synapses. Postsynaptic density protein-95 (PSD-95), CaMKII-alpha and NMDA receptors were, however, present in normal amounts in the silenced synapses, as were the GluR2 and GluR3 AMPAR subunits.

When all neurotransmitter release was inhibited by bath application of tetanus toxin to the cell cultures, the differences in GluR1 levels abolished. This study therefore shows that AMPAR subunit composition is controlled locally by glutamate release, such that GluR1 levels at neighbouring synapses on the same dendrite differ according to the relative activity at each synapse.

Hippocampal LTP involves transient incorporation of GluR2-lacking AMPARs into the postsynaptic membrane

Plant et al (2006) induced NMDAR-mediated LTP in hippocampal slice cultures by simultaneous high frequency stimulation of Schaffer collaterals and CA1 pyramidal cells, and used whole-cell patch-clamp recordings to record AMPAR-mediated excitatory postsynaptic currents (EPSCs) in the CA1 neurons.

Induction of LTP produced potentiation of EPSCs recorded at a holding potential of -70mV, but not of those recorded at +40mV, indicating that LTP induction led to the incorporation of GluR2-lacking AMPARs into the strengthened synapses. The rectification observed did not occur when spermine was not included in the pipette solution, confirming that the EPSCs recorded were mediated by GluR2-lacking AMPARs.

Significantly, recruitment of GluR2-lacking AMPARs was found to be transient, as addition of the polyamine toxin philanthotoxin blocked potentiation when applied 3 minutes, but not 20 minutes, after LTP induction. Further, when recordings were taken from CA1 neurons more than 20 minutes after LTP induction, AMPAR-mediated currents were recorded, but no inward rectification was observed, suggesting that GluR2-lacking AMPARs had been replaced by heteromeric receptors.

Activity-dependent recomposition of AMPARs occurs in vivo

Clem and Barth (2006) examined the effects of sensory experience on AMPAR dynamics, using the single whisker experience, a protocol in which all but one of the whiskers are removed from mice. As a result, stimulation of the spared whisker causes an enhanced response in layer 4-layer 2/3 synapses in the barrel cortical column which receives sensory inputs from the spared whisker.

24 hours after stimulation of the remaining whisker, the animals' brains were dissected and electrophysiological recordings were taken from pyramidal neurons in layer 4-layer 2/3 of the barrel cortex. An increase in the amplitude of AMPAR-mediated EPSCs was observed in the barrel column which received sensory inputs from the spared whisker, but not in other parts of the barrel cortex, or in layer 4-layer 2/3 pyramidal synapses of control mice with all their whiskers intact.

EPSCs in the spared whisker also had faster decay times, suggesting that the synapses contained GluR1. The presence of a significant increase in inward rectification at layer 4-layer 2/3 synapses, and blockade of the AMPAR-mediated currents by Joro spider toxin, confirmed that the synapses contained Ca²⁺-permeable AMPARs; these were presumed to be GluR1 homomers.

Clem and Barth found that this experience-dependent plasticity caused input-specific changes in AMPAR distribution, as the distinct properties of GluR2-lacking AMPARs were observed in layer 4-layer 2/3 synapses, but not in layer 2/3 synapses of the same cells, which do not receive sensory inputs from the spared whisker.

Bellone and Lüscher (2006) used an ex vivo method to demonstrate that cocaine drives GluR2-lacking AMPARs into synapses of dopaminergic midbrain neurons. Following a single intraperitoneal injection of the drug into postnatal day 18-22 mice, electrophysiological recordings were made from dopaminergic neurons in acute slices of the ventral tegmental area (VTA).

It was found that a single dose of cocaine led to pronounced inward rectification, as confirmed by a plot of the I-V relationship of the AMPAR-mediated EPSCs, and by the sensitivity of the currents to the polyamine Joro spider toxin. This was not observed in slices of VTA from control mice which had been treated with saline.

When combined AMPAR/ NMDAR-mediated EPSCs were recorded at various holding potentials, a significant increase in the AMPAR/ NMDAR ratio was observed. This corresponded to a change in the rectification index (RI, the current at -70mV divided by the current at +40mV).

Bellone and Lüscher then interfered with the interaction between the GluR2 subunit and PICK1 (protein interacting with C kinase-1), a PDZ-containing protein known to mediate insertion and removal of GluR2-containing AMPARs. A peptide called Pep2-EVK1, which effectively blocks PICK1 function, was fused to an 11 amino acid sequence of the HIV protein Tat, and then delivered in vivo to dopaminergic VTA neurons. This was followed, 4 hours later, by a single dose of cocaine.

In cells from mice treated with cocaine and the peptide, the RI was the same as that measured in control animals. Cells from mice treated with an inactive control peptide, however, had a high RI. These data suggest that an interaction between GluR2 and PICK1 is required for the observed change in AMPAR/ NMDAR ratio. They further suggest that PICK1-mediated removal of GluR2-containing AMPARs precedes, and is necessary for, the cocaine-induced insertion of GluR1-lacking AMPARs into the postsynaptic membrane of VTA cells.

Together, these findings show that cocaine induces the insertion of GluR2-lacking AMPARs into the postsynaptic membranes of VTA neurons, and that this is preceded by the removal from the membrane of hetermoeric AMPARs.