Mechanisms of plasticity

Much of the research at the MRC Centre is directed towards understanding the mechanisms that underlie the induction and expression of synaptic plasticity. Such work involves many approaches, from the recording of individual channel currents to the dynamic visualisation of synapses undergoing plastic changes. While there are many forms of short term plasticity, such as paired pulse facilitation/depression, these pages are intended to detail some of the work of Centre PIs in the area of long-term plasticity, namely long-term potentiation and long-term depression.

Long-term potentiation (LTP)

Discovery of LTPDiscovery of LTP. Long term potentiation was first reported in the perforant path (entorhinal cortex - dentate gyrus in the hippocampal formation). Mounted on the internet by permission of Bliss & Lomo (1973) J. Physiol. 232; 357-374 © The Physiological Society

Long-term potentiation, the long lasting enhancement of synaptic transmission first reported by Bliss and Lomo over 30 years ago, has been the focus of an enormous amount of investigation. It has long been regarded, along with it's counterpart LTD, as a potential mechanism for memory formation and learning .

Since it's discovery in the perforant path of the hippocampal formation, the great majority of the work related to LTP has been through electrophysiological investigations. During this time, a wealth of evidence has been accrued demonstrating that LTP is not a single process, but rather an umbrella term that encompasses multiple molecular mechanisms which may be expressed in different neuronal populations in different brain regions and at different stages of development. Given the complexity of the field, we will focus on glutamatergic LTP in the most widely studied region of the brain - the rat hippocampus.

Hippocampal LTP - CA3-CA1 synapse

Induction of LTP is developmentally regulated

The most studied synapse in the rat hippocampus is that formed by projections from the CA3 region to the CA1 in the schaffer collateral commusural pathway. Using non-stationary fluctuation analysis to investigate plasticity at this synapse (Benke et al, 1998), we have shown that multiple forms of LTP are expressed during development (Palmer et al 2004). Early in development, LTP is expressed as an increase in the probability of glutamate release (Pr) from the pre-synaptic bouton (LTPa). Interestingly, the trigger for this form of LTP appears to be the loss of a high affinity, pre-synaptic kainate receptor (Lauri et al 2006).

At this age, the AMPA receptors in the post synaptic spine have a high conductance (γ), that is they allow a high level of current to pass through on activation. A few days later in development, the total amount of current that is passed increases, but the average conductance of the receptor population is reduced. This can be explained as the result of the addition of low conductance AMPA receptors into the post-synaptic spine (LTPb).

Previous work had already shown that in P14 rats, two thirds of LTP events is are the result of the replacement of low conductance AMPA receptors for high conductance receptors (LTPc). This may be either by the modification of existing receptors (Benke et al, 1998) or a swap of low conductance receptors for high conductance ones (Terashima et al 2004, Plant et al 2006). The remaining third are due to increased numbers of such receptors  (LTPd; Benke et al, 1998). Such developmental  changes in AMPA receptor subunit composition were also demonstrated using confocal imaging (Pickard et al 2000). Thus, at least four different mechanisms of LTP are expressed in rat brain within a 14 day period.

LTP induction in CA1 is NMDAR-dependent.

Most LTP events at the CA3-CA1 synapse are dependent on NMDA receptor activation following either a high frequency train of stimuli (tetanus) or paired-pulse facilitation (PP). Depolarisation of the post-synaptic membrane following such stimulation relieves a voltage-dependent Mg2+ block of the NMDA receptor, allowing an influx of Ca2+ ions. This influx in turn activates mutiple signalling cascades that results in both the rapid phosphorylation of existing AMPA receptors, increasing their single channel conductance (LTPc), and the insertion of new AMPA receptors with either low or high conductance (LTPc and d).

Interestingly, the regional subunit composition of the NMDA receptor seems to play a role in what forms of plasticity are induced following activation. For instance, blockade of either GluN2A- and GluN2B- containing receptors blocked LTP in hippocampal CA1 neurones, while LTD was blocked only by blockade of GluN2A-containing NMDA receptors (Bartlett et al 2007). This follows on from similar work in the perirhinal cortex, where LTP was shown to be dependent on GluN2A-containing receptors while LTD was dependent on GluN2B-containing receptors (Massey et al 2004).

parallel kinase pathways in LTPMolecular mechanisms in NMDAR-dependent LTP

Multiple signalling cascades are induced by NMDA receptor activation that are involved in the induction of LTP. CaMKII-dependent phosphorylation of the GluA1 subunit at residue ser-831 is central to the increase in single channel conductance seen in 2/3 of LTP at CA1 synapses (Derkach et al 1999). While in adult animals, blockade of CaMKII is sufficient to block LTP (Bortolotto & Collingridge 1998), a combined block of both CaMKII and either PKC or PKA is necessary in younger animals (Wikstrom et al 2003). Thus there is a developmental change in the kinase cascades responsible for functional changes in existing receptors.

Further mechanisms involve the movement of receptors either to increase the receptor numbers expressed in the synapse or to change the composition of the receptor complement. The idea that post-synaptic AMPA receptors can move rapidly was first mooted in response to work showing that LTP involves the rapid unsilencing of 'silent synapses' (Isaac et al, 1995). It was postulated that the conversion of apparently AMPA receptor-lacking synapses into those with a full complement of AMPARs was due to the insertion of receptors from a pre-existing pool. Such a effect was directly visualised using antibodies directed against the exposed N-terminus of AMPA receptor subunits in cultured neurones (Pickard et al 2001) after it had been shown that AMPA receptors themselves are rapidly turned-over in the post-synaptic membrane (Nishimune et al 1998).

Over-expression of PICK1 in dissociated neurons results in increased rectification of AMPAR EPSCs suggesting a reduced proportion of GluA2 subunits expressed at these synapses. Mounted on the internet with the permission of Terashima  (2004) J.Neurosci. 24; 5381-5390 © Society for Neuroscience

The mechanisms that control the insertion of AMPA receptors into the synaptic membrane have been widely studied, but the roles of specific protein-protein interactions remain unclear (reviewed in Hanley 2010); NSF, GRIP and PICK1 interactions with GluA2 have all been implicated. Over-expression of PICK1 in dissociated hippocampal neurons results in a PKC/CaMKII-dependent increase in AMPA receptor mediated EPSCs ( Terashima et al 2004, 2008), that is associated with a change in the rectification parameters suggesting that the proportion of GluA2-containing receptors is reduced. A similar change is seen following the induction of LTP ( Plant et al 2006). FRAP studies with pHluorin-labelled GluA subunits have demonstrated reduced mobility of GluA1 receptor subunits following chemical LTP (Makino & Malinow 2009). This study suggested that excocytosis of AMPA receptors occurs in dendrites, while the new receptor are incorporated into synapses by lateral diffusion. Recent modelling work suggests that such lateral diffusion can easily account for the elctrophysiological changes associated with fast onset LTP ( Tolle & Le Novère 2010). This is reminiscent of work showing that pHluorin-labelled GluA2 containing receptors were lost from dendritic regions prior to spines themselves following chemically-induced LTD ( Ashby et al, 2004), suggesting that endocytosis occurs outside the synapse.

The changes in AMPA receptor composition that give rise to LTP can be very short lived. Thus, maintenance of the changes in synaptic function requires protein synthesis and changes in gene expression  - the so-called late phase of LTP (reviewed in Reymann & Frey 2007).

Hippocampal LTP - DG-CA3 synapse (Mossy Fibre)

The pathways between the dentate gyrus of the hippocampus and the CA3 region is known as the mossy fibre pathway. The mechanisms that underlie LTP in this pathway are different to those at the CA3-CA1 synapse and have proved controversial over many years.

LTP induction in CA3 is NMDAR-independent

LTP at the mossy fibre terminals in the hippocampus is different to that in the CA1 region. Mossy fibre LTP (MF-LTP) is induced without the need for NMDA receptor activation. Instead, it appears to require the activation of pre-synaptic kainate receptors. A positive feedback mechanism which leads to enhanced activation of KARs and glutamate release is apparent under high frequency stimulation (frequency facilitation). This is the basis of the LTP induction mechanism at the synapses.

Kainate receptors involved in MF-LTP contain GluK1 subunits .... or GluK2

Over the years there has been much controversy over whether the kainate receptors that induce MF-LTP contain GluK1 or GluK2 subunits (GluR5 or 6 subunits in the old nomenclature). There is conflicting evidence form pharmacological and genetic knockout work that suggests that either subunit, or indeed both, may play roles in this form on plasticity. For instance, it has been shown that in GluK2 knockout mice, MF-LTP is substantially reduced (Contractor et al 2001). However, GluK1 selective antagonists such as LY382884 and UBP296 block MF-LTP (Bortolotto et al 1999, More et al 2004 though see also Breustedt & Schmitz 2004). Further support for this view has come from direct visualisation of individual mossy fibre terminals in acute hippocampal slices using multi-photon imaging. We have shown that ACET, the most potent and selective GluK1 antagonist yet developed, regulates Ca2+ signalling in giant mossy fibre boutons (Dargan et al 2009).

The reasons for these discrepancies are still as yet unknown.

Long-term Depression (LTD)

Visualisation of AMPA receptor internalisation in 'LTD'.The N-terminus GluA2 subunit was tagged with ecliptic-pHluorin and used to visualise cell surface expression in cultured hippocampal neurones. Blue circles represent diffuse fluorescence indicative of extra-synaptic AMPA receptors; red circles represent punctate fluorescence, indicative of synaptic AMPA receptors. On NMDA application, there is a transient loss of extra-synaptic receptors followed by a loss of synaptic receptors. Mounted on the internet with permission of Ashby et al, 2004 J. Neurosci. 24; 5172-5176 © Society for Neuroscience

Long-term depression is the inverse of LTP, a long lasting reduction in synaptic transmission. LTD also covers multiple induction and possibly expression mechanisms.

NMDA receptor-dependent LTD

LTD can be induced by delivering low frequency stimulation to neurons (LFS). 900 impulses delivered to CA1 hippocampal neurons at 1 Hz (one impulse per second) will result in a form of LTD that is dependent on the activation of NMDA receptors. We have previously used non-stationary fluctuation analysis to show that this form of is not the result of changes in the conductance of existing AMPA receptors, but is more likely due to the loss of those receptors from those synapses (Luthi et al 1999).

One of the questions that has remained has been the molecular nature of the trigger that induces LTD. It has long been known that LTD is Ca2+ dependent, and it has now been shown that Hippocalcin, a member of the EF-hand containing neuronal calcium sensor (NCS) family, acts as the calcium sensor in a cascade that couples NMDA receptor activation to AMPA receptor internalisation (Palmer et al, 2005). Activation of hippocalcin via Ca2+ influx leads to the formation of a complex with AP-2, part of the clathrin-mediatd endocytic machinery. This complex binds to the GluA2 subunit of the AMPA receptor, displacing NSF and recruiting clathrin, thus initiating internalisation.

Work carried out here at the MRC Centre for Synaptic Plasticity using a pHluorin-tagged GluA2 subunit has directly visualised the removal of cell-surface AMPA receptors during a form of LTD induced by exposure to NMDA in cultured hippocampal neurones (Ashby et al, 2004). In this work, it was shown that there is a transient loss of extra-synaptic receptors that precedes the loss of synaptically localised receptors, suggesting that AMPA receptors are not internalised directly from the synapse, but from the surrounding membrane. Synaptic receptors could then be translocated out from the synapse. These two processes could combine to produce the long-lasting loss of synaptic AMPA receptors that is thought to be the molecular basis of LTD.

Such lateral diffusion has since been shown to be an integral part of the mechanisms by which the complement of AMPA receptors expressed at the synaptic surface is controlled (Ashby et al 2006).

Signalling mechanisms in LTD

We have found that a number of signalling cascades are important in NMDAR-LTD. These include pathways involving the Ser/Thr kinase GSK-3β (Peineau et al 2007) and the tyrosine kinase JAK2 (Nicolas et al 2012).

LTP inhibits LTD via regulation of GSK-3β

Glycogen synthase kinase 3 (GSK-3) is a multifunctional serine/threonine kinase that was originally identified as a regulator of glycogen metabolism and is ubiquitously expressed in eukaryotic cells. Of the two known isoforms, GSK-3β is highly expressed in the brain and has been implicated in a wide range of neurological disorders including Alzheimer's disease, schizophrenia and bipolar disorder. We showed recently that inhibition of GSK-3β blocked the induction of LTD in hippocampus

Scheme showing inhibition of LTD by LTP via GSKScheme showing the signalling mechanism via which NMDAR-LTP inhibits NMDAR-LTD. Activation of NMDA receptors activates PI3K, which activates Akt that in turn promotes the phosphorylation of ser9 on GSK-3β resulting in de-activation of this enzyme and inhibition of AMPA receptior internalisation. Activation of caspase-3 results in cleavage of Akt and phosphorylation of ser9. In this state, GSK-3β promotes internalisation of AMPA receptors and LTD. Scheme based in data from Peineau et al (2007) Neuron 53, 703-717 and Li et al (2010) Cell 141, 750-752

while the delivery of LFS to slices resulted in a dephosphorylation, and thus activation, of the enzyme ( Peineau et al 2007; indeed, subsequent studies have shown that GSK-3β is the only ser/thr kinase involved in LTD ( Peineau et al 2009)). Interestingly, we also found that GSK-3β was associated with AMPA receptors and that chemically-induced insertion of AMPA receptors using sucrose and glycine resulted in decreased AMPAR-associated GSK-3β activity. This suggests that LTP stimuli can block the induction of LTD - and indeed we observed that this is the case. Induction of LTP inhibits the induction of LTD for 30-60 minutes following tetanus. The sucrose/glycine-mediated insertion of AMPA receptors has previously been shown to be associated with an increase in PI3 kinase (PI3K) activity. PI3K is an upstream regulator of GSK-3β, acting via Akt to phopsphorylate ser9 and leading to inactivation. Inactivation of GSK-3β would prevent LTD induction. Indeed pharmacological inhibition of either PI3K or Akt blocked the ability of LTP to inhibit LTD. This may provide a mechanism by which information stored in synapses modified by LTP is held safe until consolidated.

Complementary studies have implicated caspases, a groups of proteases known to be involved in apoptosis, in NMDAR-LTP and -LTD. Collaborative work has shown inhibition of caspase-3 and caspase-9 by the use of either interfering peptides blocks the induction and expression of NMDAR-LTD in CA1 hippocampal neurones (Li et al, 2010) as well as the internalisation of AMPA receptors, the primary mechanism by which LTD is expressed.  Genetic knockout of caspase-3 (KO animals and RNAi methodologies) also resulted in a loss of AMPA receptor internalisation. Caspase-3 is also known to cleave Akt and over-expression of a cleavage-resistant mutant abolished LTD, suggesting that proteolysis of Akt is required for LTD induction/expression. Thus PI3K and caspase-3 play opposite roles in the regulation of Akt and hence NMDAR-LTD. Activation of PI3K leads to de-activation of GSK-3β and thus inhibition of NMDAR-LTD - conversely, activation of caspase-3 results in the activation of GSK-3β and the expression of NMDAR-LTD. Interestingly, we have also shown that amyloid-β1-42 (Aβ), thought to be a major mediator of cognitive defecits in Alzheimer's disease, blocks LTP and promotes LTD via activation of caspase-3 and GSK-3β (Jo et al 2011).

These results suggest that the PI3K-Akt-GSK3 axis plays a central role in the regulation of synaptic connectivity in both health and disease.