LTD underlies recognition memory

This is a summary of the work in which we show that LTD is the most likely mechanism to underlie recognition memory, published in Griffiths et al, 2008, Neuron 58, 186-194.

The expression of NMDA receptor-dependent LTD relies on the clathrin-mediated internalisation of AMPA receptors. A critical step in this process is the association of AP2, a clathrin adaptor protein, to the GluA2 subunit of the AMPA receptor (Lee et al 2002). Disruption of this interaction with a small competing peptide prevents AMPA receptor internalisation and blocks the expression of LTD in the hippocampus.  We have used this same peptide in perirhinal cortical neurons to try and block LTD and investigate the effects on visual recognition memory.

Competing peptides can be used to block selective protein-protein interactions

Scheme showing sequence of AP2 competing peptideScheme showing the peptide used to block the interaction between AP2 and the GluA2 subunit of the AMPA receptor.
The use of competing peptides to block specific protein-protein interactions has become a very useful tool in the study of the molecular processes that are involved in various neuronal plastic processes. Indeed, members of the MRC Centre have used this technique to investigate the role of GluA2 interactions with other proteins in LTP and LTD for many years ( Nishimune et al 1998, Noel et al, 1999, Luthi et al 1999, Daw et al 2000). Essentially, it involves the synthesis of a small peptide, 10-12 amino-acids long, that will compete for the binding site of one of the partners in the interaction over the full size protein. Flooding the cell with the competing peptide reduces, or abolishes completely, the interaction of the two protein partners. In this case, the peptide binds to AP2, preventing it binding to GluA2, thus blocking internalisation. Use of a control peptide, one that does not interfere with the GluA2-AP2 interaction, has no effect on LTD.
Disruption of the AP2-GluA2 interaction blocks LTDEffect of infusion of AP2 blocking peptide on NMDAR-dependent LTD in the perirhinal cortex in vitro. Mounted on the internet with the permission of Griffiths et al (2008) Neuron 58, 159-161 © Cell Press

Disruption of AP2-GluA2 interaction blocks LTD in vitro in perirhinal cortex

Perirhinal cortical slices were taken and whole-cell recordings made. Robust LTD was induced by pairing membrane depolarisation to -40mV with a conditioning stimulation of 200 stimuli at 1Hz. LTD was blocked by the application of AP5, a NMDA receptor antagonist, showing that the LTD was NMDAR-dependent.

Infusion of the competing peptide pep-ΔA849-Q853 through the recording pipette completely abolished NMDAR-dependent LTD in such slices. In contrast, infusion of a control peptide, pep-K844A, had no effect on LTD. This peptide has been shown to not affect the AP2-GluA2 interaction (Lee et al 2002) and the size of the LTD was no different in the presence of this peptide (50 ± 13%)  than under control conditions, with no peptide infusions (47 ± 6%).

Neither peptide had any effect on basal transmission.  In addition, we have shown that the ΔA849-Q853 does not affect LTP in these slices (data not shown).

In vivo transduction of AP2 competing peptide disrupts visual recognition memory

In order to determine if blocking LTD in the perirhinal cortex could affect the ability of an animal to discriminate novel and familiar objects, we used viral transduction to express  pep-ΔA849-Q853 in perirhinal neurons in vivo. Co-expression with GFP allowed us to demonstrate that transduction was limited to the perirhinal cortex and to determine the extent of transduction.

Effect of AP2 on Spontaneous Object RecognitionEffect of In vivo transduction of AP2 blocking peptide on familiarity discrimination. Mounted on the internet with the permission of Griffiths et al (2008) Neuron 58, 159-161 © Cell Press
Two behavioural tasks were used to investigate the role of NMDA receptor-dependent LTD in visual recognition memory - spontaneous object recognition and object-in-place tasks. The 'Discrimination Ratio' is calculated as the difference between the time spent investigating the novel and familiar objects divided by the total time spent investigating. If an animal spends more time at the novel object, the discrimination ratio is positive. If the animal spends an equal amount of time investigating the two objects, the result will be zero. Thus, an animal that can distinguish novel from familiar objects gives a positive result, and an animal that can't gives a result close to zero. Transduction of animals with the competing peptide completely abolishes the ability to recognise that it has seen an object before (DR close to zero) while transduction of the control peptide has the same effect as if there was no peptide present at all.
Effect of AP2 on Object in PlaceEffect of In vivo transduction of AP2 blocking peptide on object-in-place memory. Mounted on the internet with the permission of Griffiths et al (2008) Neuron 58, 159-161 © Cell Press

As a control memory test, we looked at the effect of blocking the GluA2-AP2 interaction on short-term object-in-place memory. This form of memory is dependent on kainate receptors rather than NMDA receptors (Barker & Warburton 2008) and so disrupting NMDA receptor-dependent LTD should have no effect. Indeed, this was the case. Animals were just as good at the task when transduced with the competing peptide than with the control peptide or no peptide at all. It should be noted that the aquisition of long-term object-in-place memory is NMDAR-dependent (Barker & Warburton 2008), but this was not tested in these animals. More recently, cholinergic transmission has also been implicated in this form of memory (Barker & Warburton 2009) and Ii should also be noted that an intact perirhinal cortex is necessary for object-in-place memory (Barker et al 2007).

LTD is blocked in slices from animals transduced with competing peptide

Disruption of the AP2-GluA2 interaction blocks LTDEffect of transduction of AP2 blocking peptide on NMDAR-dependent LTD in the perirhinal cortex in vitro. Mounted on the internet with the permission of Griffiths et al (2008) Neuron 58, 159-161 © Cell Press
In order to confirm that viral transduction with pep-ΔA849-Q853 did affect LTD in the perirhinal cortex, brain slices were tested with the same LTD induction protocols used in the initial in vitro experiments but with not peptide infusions. We could not induce LTD in slices taken from animals trandsuced with the AP2 competing peptide whilst LTD was readily induced in slices from animals transduced with the control peptide. LTP, the inverse form of plasticity, was also present in those animals transduced with the competing peptide, showing that the loss of LTD was not due to a generalised loss of plastic capacity.

These results offer compelling evidence that NMDA receptor-dependent LTD is the mechanism of synaptic plasticity that underlies visual recognition memory. Not only does disruption of the interaction between GluA2 and AP2 disrupt this form of LTD, but it also disrupts the ability of animals to perform appropriate recognition memory tasks. Crucially, in these animals, NMDA receptor-dependent LTD is also blocked, but LTP is still present. Thus, the effect on behaviour is not through a general loss of neuronal performance, but as the result of the loss of a specific molecular mechanism.

Multiple mechanisms of LTD may be involved in visual recognition memory

Taken together, these results demonstrate a direct link between a mechanism of synaptic plasticity and a specific memory system. Given the widespread expression of LTD in many regions of the brain, this is likely to be of widespread significance in relation to other memory mechanisms. This is given further weight by the finding that LTD and depotentiation (essentially LTD of synapses that have previously undergone LTP induction) is absent in the perirhinal cortex of animals viewing familiar stimuli in a paired viewing procedure (Massey et al 2008). This result is interpreted as being due to saturation of the LTD-like mechanism during the paired viewing so that LTD cannot be induced in vitro in slices prepared from those animals. Such saturation of synaptic plasticity is a very well known phenomenon, where multiple rounds of induction of LTP or LTD result in the inability to induce plasticity further. Interestingly, in this study, the loss of LTD following paired viewing was blocked by administration of scopolamine, a cholinergic muscarinic antagonist, prior to trial. Thus the 'LTD memory mechanism' is dependent on cholinergic muscarinic receptors (mAChR) and not glutamate receptors. mAChR-dependent LTD has previously been demonstrated in the perirhinal cortex both chemically (Massey et al 2001) and synaptically (Warburton et al 2003). We also know that there are multiple glutamatergic mechanisms involved in various aspects of visual recognition memory, depending on the timescale of the memory under investigation,  as well as a neural network involving multiple brain regions and possibly multiple neuronal mechanisms.