Molecular Neuroscience
Neurotransmission | Medicinal Chemistry | Receptors | Ion Channels | Interactions
Molecular neuroscience is concerned with the biochemical and molecular processes that underpin neuronal function. This includes investigations of mechanisms of neurotransmitter release, synaptic receptor function and regulation, ion channel function, and protein-protein interactions and their roles in neuronal excitability and plasticity. These processes are ultimately how neurons perform their myriad tasks in the brain.
Neurotransmitter release
The control of transmitter release is fundamental to neuronal function. The amount of neurotransmitter released can vary enormously, from the small amounts released in so-called 'kiss and run' mechanisms to full quantal release where all the transmitter from a vesicle is ejected into the synaptic cleft. The study of single vesicle fusion events in synapses is technically very challenging and few examples have ever been reported. Adrenal chromaffin cells are a popular model system for such work. Using capacitance current measurements in excised membrane patches, we showed that a single vesicle fusion event was preceded by the opening of a single co-localised Ca2+ channel (Powell & Marrion 2007).
Selective viral expression of EGFP in 5HT neurons. Adapted from Benzekhroufa et al 2009b © Benzekhoufra et alA curious fact is that the axonal varicosities that form the release sites of noradrenergic or serotonergic neurons often do not form synapses; they have neither synaptic form nor function (see Kasparov and Teschemacher 2009). Noradrenergic neurons are known to be closely associated with astrocytes and astroglial cells, thus the actions of these neurons may be, in part, via these cell populations.
Medicinal Chemistry
Bristol has a long history in the development of new ligands for glutamate receptors and such developments continue apace. A mainstay of work has been the development of increasingly potent and selective antagonists for the GluK1 subunit of kainate receptors (More et al 2004, Dolman et al 2005, Dolman et al 2007, Dargan et al 2009) and NMDA receptors (Morley et al 2005, Brothwell et al 2008, Costa et al 2010) Previous work had also developed ligands for mGlu receptors, including MCPG, CHPG and DCPG (Jane et al 1993, Doherty et al 1997, Thomas et al 2001, Miller et al 2003).
Compounds such as these are generated through a process of rational drug design. Computerised modelling of agonist/antagonist interactions with the binding domains (as determined by X-ray crystallography), is used to predict structural moieties that will bring the greatest level of sub-type selectivity and binding affinity. Thus, a novel kainate receptor antagonist, UBP310, have been used to probe the structure of the GluK1 binding domain (Atlason et al 2010). Modelling studies predicted the residues in the binding pocket that were likely to be important in determining selectivity for binding to this subunit. Point mutation of these residues showed this to be the case. Such information is used to generate higher affinity analogues, such as ACET (left; Dargan et al 2009).
Neuroreceptors
Investigations into the roles of synaptic receptors in neuronal function is central to neuroscience research in the School. There are extensive programmes investigating the roles of both ionotropic and metabotropic glutamate receptors and muscarinic receptors. These receptors is responsible for basal excitatory synaptic transmission as well as many forms of synaptic plasticity such as long-term potentiation (LTP) and long-term depression (LTD), mechanisms that are thought to underlie learning and memory. They are thus also potential targets for therapies for CNS disorders such as epilepsy and Alzheimer's diseases.
Panel A. Different μ opioid receptor agonists induce different desensitisation mechanisms. Adapted from Bailey 2009b © The British Pharmacological Society. Panel B. Reversal of tolerance is dependent on the desensitisation mechanism. Tolerance was induced by repeated injection of agonist (green circle - orange square) and reversed by injection of either GRK or PKC inhibitors (inverted blue triangle). DAMGO-induced tolerance is reversed by inhibition of GRK, while morphine-induced tolerance in only reversed by inhibition of PKC. Adapted from Hull et al 2010 © The American Society for Pharmacology and Experimental Therapeutics.Opioid receptors and morphine tolerance
Work on opioid receptors, a type A GPCR, has been a long-standing area of interest in the School, in particular the mechanisms of desensitisation by agonists and its role in morphine tolerance. Thus it was recently shown that a PKCα dependent mechanism was involved in morphine-induced desensitisation of μ-opioid receptors in mature locus coeruleus neurons in the brainstem (Bailey et al 2009a). However, it has also been shown that different mechanisms of desensitisation are induced by different agonists, so that while morphine induces a PKCα dependent mechanism, the peptide agonist DAMGO induces a GRK2-dependent mechanism (Bailey et al 2009b). Tolerance to morphine (and other low efficacy agonists) was reversed by the use of PKC inhibitors, while GRK inhibitors reversed only tolerance induced by the high efficacy agonist DAMGO (Hull et al 2010). Thus differential activation of the μ-opioid receptor results in the activation of different regulatory mechanisms and thus different tolerance mechanisms (McPherson et al 2010).
Synaptic receptors and plasticity
Research into the roles of glutamate receptors in the induction and the expression of neuronal plasticity has been ongoing in Bristol for many years. The modulation of AMPA receptor function and composition following NMDA receptor activation underlies widespread mechanisms of both long-term potentiation (LTP) and long-term depression (LTD). Both pre-synaptic and post-synaptic kainate receptors are involved in NMDA receptor-independent forms of LTP, while metabotropic glutamate receptors are involved in various forms of LTD.
Recently, attention has also turned to muscarinic receptors that are involved in forms of LTD of AMPA (Dickinson et al 2009) and NMDA (Jo et al 2010) receptor-mediated responses as well as the facilitation of LTP (Buchanan et al 2010). This is interesting as muscarinic receptors are known to be involved in visual recognition memory (Warburton et al 2003) and while NMDAR-dependent LTD was recently shown to block spontaneous object recognition memory in the perirhinal cortex, we also know that there are multiple mechanisms operating in different neuronal populations. For instance, we have shown previously that long-term visual recognition memory is dependent on NMDA receptor activation while short term memory is dependent on kainate receptors (Barker et al 2006).
Signalling mechanisms
The molecular mechanisms that underlie LTP and LTD involve multiple signalling cascades including the p38 MAP kinase and tyrosine kinase pathways in mGlu receptor-dependent LTD (Moult et al 2008, Gladding et al 2009), Ca2+ binding proteins in LTD (Palmer et al 2005; Jo et al 2008), and PDZ-binding proteins (Terashima et al 2004, 2006; Dickinson et al 2009; Jo et al 2010). We have also shown that GSK3β is the sole ser/thr kinase that is involved in NMDA receptor-dependent LTD (Peineau et al 2009) after previously showing that activation of the PI3K-Akt-GSK3β pathway by following LTP induction inhibits the induction of LTD for about 30 minutes (Peineau et al 2007). This form of meta-plasticity could be an important mechanism in memory consolidation, as LTD-induction protocols induce a form of plasticity known as depotentiation in synapses that have undergone LTP, which can effectively reset the synapse back to baseline levels of transmission.
A novel involvement of the JAK/STAT signalling pathway in NMDA receptor-mediated LTD was unexpected (Nicolas et al 1012). Pharmacological blockade of Janus Kinase 2 (JAK2) specifically blocked the induction of de-novo NMDAR-LTD with no effect on depotentiation, while shRNA-mediated knockdown of JAK2 abolished NMDAR-LTD. It was further shown that inhibition of STAT3, a known down-stream effector of JAK2 in many non-neuronal processes, also blocked NMDAR-LTD. Unlike in many other cellular processes in which STAT3 is involved, nuclear translocation was not required. These results are important as they further demonstrate the complexity of the signalling processes involved in plasticity. It is interesting that inhibtion of JAK2 has been shown to affect spatial learning and memory (Chiba et al 2009) especially given the involvement of LTD in recognition memory (Griffiths et al 2008).
Structure, assembly and localisation
Surface crosslinking by biotinylation shows that far more unedited GluK2 subunit reaches the cell surface than edited. Adapted from Ball et al 2010 © Ball et al & The International Society for NeuroscienceWe have shown that the assembly of kainate receptors and their export from the endoplasmic reticulum is determined by editing of the Q/R site in the channel pore loop (Ball et al 2010). This study also showed that the level of expression of the GluK5 subunit is dramatically reduced in GluK2 knockout mice. This builds on earlier work, which showed that the GluK5 subunit (formerly known as KA-2) is necessary for the surface expression of GluK2 (formerly known as GluR6 - Gallyas et al 2003). These studies show the tight inter-relationship between the expression of different subunits of the same protein complex.
Ion Channels
Ion channels are at the heart of neuronal function. Ligand-gated ion channel receptors like AMPA or GABAA receptors mediate neurotransmission while the voltage-gated Na+ and K+ channels are responsible for the generation and control of action potentials. These channels, in concert with Ca2+ channels, also modulate the basal membrane potential in both neuronal and non-neuronal cells, whilst Ca2+ channels themselves are vital regulatory units in the myriad Ca2+ dependent signalling cascades found in all cell types. All of the voltage-gated ion channels exist as large families of channels and research within the School investigates many aspects of their function, though here we will cover just those themes pertinent to neuroscience.
K+ Channels
K+ channels come in many guises and one of the main areas of research within the school for many years has been the Ca2+-activated K+ channels. These come as either large conductance (BK) or small conductance (SK) channels that allow passage of K+ ions following binding of Ca2+ ions and their activation underlies the falling phase of the action potential and the slow afterhyperpolarisation, respectively.
One of the current focuses of work is the mechanism of action of small conductance channels (for review, see Weatherall et al 2009). We have recently demonstrated the histidine residues responsible for the modulation of channel opening by protons (Goodchild et al 2009) and, utilising a combination of molecular modelling, site-directed mutagenesis and electrophysiology, shown that the diagnostic inhibition by the bee venom toxin, apamin, is via an allosteric mechanism rather than by directly blocking the channel pore (Lamy et al 2010).
SK channels also have potentially very important roles in cognition and memory storage due to their control of membrane potential - inhibition results in increased neuronal excitability and thus makes it easier to induce form of plasticity such as LTP and LTD. Indeed, we have recently shown that SK channels are directly modulated by M1 muscarinic receptors in a PKC-dependent manner (Buchanan et al 2010). Inhibition of SK channels, either directly or via pharmacological and synaptic activation of M1 receptors, promotes NMDA receptor activity and thus facilitates the induction of LTP.
Voltage-gated K+ channels also play a major role in the control of membrane potential and initiation of action potentials. They are thus well placed to encode information in the learning circuits. For instance, the neuronal M current, which is driven by voltage-gated K+ channels (Kv7.2/7.3 heteromers) is an important driver of intrinsic plasticity in CA3 hippocampal neurons (Brown & Randall 2009). In addition to this work, we have shown that the muscarinic receptor-dependent suppression of the neuronal M current is itself dependent on their colocalisation within lipid rafts (Oldfield et al 2009). These are gel-like membrane microdomains that limit diffusion. Thus different proteins can be held in close proximity to each other without being held by the sub-membrane cytoskeleton. This gives another level of control over the protein-protein interactions that are such a feature of cellular function. For example, previous work has shown that different Ca2+ channels are localised alongside BK and SK channels in neuronal membranes, ensuring a source of Ca2+ ions for activation (Marrion & Tavalin 1998, Bowden et al 2001, Loane et al, 2007).
Another area of research is the role of voltage-gated K+ channels Shaw and KCNQ in the generation of behaviour, using the fruit fly, Drosophila, as a model system. For instance, we have shown that Shaw channels have a central role in the coordinated and rhythmic output from Drosophila clock neurons, which maintain the 24 hours circardian cycle (Hodge & Stanewsky 2008).
Ca2+ Channels
Ca2+ channels also come in many guises. Ligand-gated Ca2+ channels include the IP3 and ryanodine receptors that are crucial in the control of Ca2+ release from intracellular stores. These receptors are a fundamental part of the signalling cascades of many GPCRs and other receptors that give rise to Ca2+ influx such as kainate receptors. For instance, Ca2+-induced Ca2+ release is crucial in the induction of NMDA receptor-independent LTP in the hippocampal mossy fibre pathway by enhancing neurotransmitter release (Lauri et al 2003). Indeed, activation of both ryanodine and voltage-gated Ca2+ channels is necessary.
Voltage-gated Ca2+ channels, like all other ion channels, exist as a large family of closely related individual sub-types. Five types of Ca2+ channels have been identified by their electrophysiological and pharmacological characteristic - L-type, N-type, P/Q-type, R-type and T-type, all of which can be found in neurons. Cloning of individual subunits has revealed a complex family of subunits that form multimeric assemblies of four subunits to make a single channel complex. Not only are there multiple classes of Ca2+ channel, like many neuroreceptors, splice variation greatly increases the number of potential isoforms. Indeed, individual neurons express multiple splice variants of a single sub-type of channel as well as different types of channel (Kanumilli et al 2006). The complement of channels expressed also changes during maturation and channels with new properties are still being found (Tringham et al 2007, 2008) reflecting the molecular diversity of this group of proteins.
Protein-protein interactions
There are many protein-protein interactions that are important in neuronal function, be it in de/phosphorylation cycles, post-translational modifications or in receptor trafficking. Work within the School focuses on the role of SUMOylation in kainate receptor function and the roles of various proteins in the movement of receptors into and out of the post-synaptic membrane.
T287 phosphorylation is activity-dependent and synapse-specific. Alteration in the activity of synapses located in specific muscles in Drosophila that are innervated by the same neurons resulted in differing levels of autophosphorylation. Adapted from Hodge et al 2006 © Cell Press LtdCamKII and CASK
CaMKII is the main protein that is found in the hippocampal post-synaptic density, constituting 1-2% of total protein. It has been shown to be a critical kinase not only in the establishment of LTP and in LTD but also in learning and memory mechanisms in drosophila and rats (for example Mehren & Griffith 2004, Tinsley et al 2009). Once activated by increased Ca2+, CaMKII is able to cause a switch in its own activity (T287 autophosphorylation), which allows it to have increased activity independent of elevated Ca2+. This loss of Ca2+sensitivity has been dubbed the molecular memory switch. The same mechanisms have been shown to occur in drosophila, the model species used in these studies. A second autophosphorylation in the CaM binding domain of the drosophila CaMKII, on T306, is dependent on the MAGUK protein CASK (drosophila equivalent, camguk or dCASK - Lu et al 2003). This modification regulates T287 autophosphorylation, thus controlling the 'switch' and is activity-dependent as well as synapse-specific (Hodge et al 2006). It is unclear as yet whether phosphorylation of T306 requires a direct interaction of CamKII with CASK, or whether this is achieved via an intermediate step. Work is continuing here within the School to investigate the role of CASK and its regulation of CaMKII activity in synaptic plasticity underlying learning.
SUMOylation
SUMO, a Small Ubiquitin-like MOdifier, is a small protein that is becoming recognised as a major modulator of neuronal function. Work here in Bristol has shown that this protein modification is important in the endocytosis of GluK2 (formerly GluR6) subunit of the kainate receptor (Martin et al 2007). Work has continued into the role played by SUMO in synaptic function.
Receptor trafficking
Research at Bristol into the role of different proteins in trafficking of receptors has been ongoing for many years. For instance the mechanisms that control the insertion and removal of AMPA receptors into and from the synaptic membrane during LTP and LTD have been widely studied (reviewed in Hanley 2010). The interactions between the proteins NSF, GRIP and PICK1 with GluA2 have all been implicated in these mechanisms but the roles of some specific protein-protein interactions remains unclear.
NSF appears to stabilise GluA2-containing receptors at the synaptic surface - previous work here in Bristol has shown that blockade of this interaction results in the internalisation of these receptors (Nishimune et al 1998). AP2, a protein that is part of the clathrin-mediated internalisation mechanism, binds to a site that overlaps the NSF binding domain on the GluA2 subunit, and blockade of that interaction blocks NMDA receptor-dependent LTD (Griffiths et al 2008). AP2 is directed to the GluA2 subunit by the Ca2+ sensor protein hippocalcin, inhibition of which also blocks NMDA receptor-dependent LTD (Palmer et al 2005). The interaction between AP2 and hippocalcin has also recently been shown to be necessary for the muscarinic receptor-dependent LTD of NMDA EPSCs (Jo et al 2010).
Insertion of GluA1 is dependent on the interaction with Myo-VI and CaMKII. Adapted from Nash et al 2010 © Nash et al and Appleby et al 2011 © Appleby et al.