2 December 2011. In neuronal transmission, as in real estate, location is everything. When N-methyl D-aspartate (NMDA) glutamate receptors sit in the synapse, they behave like good citizens, activating signaling pathways that promote memory formation and neuron survival. However, when the receptors loiter in the boondocks, they set off toxic signals that damage synapses and kill neurons. Extrasynaptic receptors fire particularly strongly in conditions like ischemia, Huntington’s disease (HD), and Alzheimer’s disease (AD), making them a promising therapeutic target (see Hardingham and Bading, 2010). In a symposium at the Society for Neuroscience annual meeting, held 12-16 November 2011 in Washington, DC, scientists discussed some of the latest research in this field. Session chair Stuart Lipton, from the Sanford-Burnham Medical Research Institute, La Jolla, California, pointed out that synaptic and extrasynaptic NMDA receptors vary in several ways, including their downstream targets, their subunit composition, and their firing patterns. Synaptic receptors fire in bursts, while extrasynaptic ones are continuously active. Speakers discussed these features in detail, as well as how extrasynaptic signaling might be targeted in HD and AD.
Synaptic and extrasynaptic NMDA receptors trigger antagonistic signaling pathways, emphasized Hilmar Bading at the University of Heidelberg, Germany. Synaptic signaling increases calcium levels in the nucleus, activates the transcription factor CREB, and leads to long-term learning and neuroprotection. By contrast, extrasynaptic signals shut off CREB, initiate apoptotic pathways, and cause breakdown of mitochondrial membranes. Bading’s group previously showed that the two types of NMDA signaling turn on distinct programs of gene expression (see ARF related news story on Zhang et al., 2007). Nuclear calcium may act as the master switch that controls adaptive responses such as neuronal plasticity and survival, Bading suggested. Of the nearly 200 genes regulated by nuclear calcium, about 10 of them make up the core neuroprotection program, Bading reported, and might be useful therapeutically. One of these, the transcription factor Atf3, reduces ischemic cell death by half in mice, Bading said.
Extrasynaptic NMDA receptors also vary from synaptic ones in their composition. All NMDA receptors consist of two GluN1 and two GluN2 subunits, but GluN2 subunits come in different forms. In the mature brain, GluN2A subunits predominate in synapses, while GluN2B preferentially populates extrasynaptic sites. This partitioning suggests that GluN2B plays the major role in excitotoxicity. The subunits also possess distinct C-terminal, cytoplasmic tails, allowing them to interact with different intracellular proteins. Interestingly, in early development, when GluN2B dominates at all NMDA receptor sites, synaptic signaling protects neurons from cell death, indicating that the GluN2B subunit does not necessarily activate cell death pathways (see Martel et al., 2009). Nonetheless, Giles Hardingham at the University of Edinburgh, U.K., wondered if subunit composition could help explain the differing toxicities of synaptic and extrasynaptic receptors.
To look more closely at the issue, Hardingham and colleagues made chimeric constructs, replacing the tail of GluN2B with that of 2A. Cultured neurons containing the chimeric receptor resisted excitotoxicity better than wild-type cells, Hardingham reported. Signaling by the GluN2B tail does play a role in harming cells, Hardingham concluded, and this difference is most noticeable in the context of a mild excitotoxic insult. Why is synaptic GluN2B less toxic than extrasynaptic? Hardingham noted that GluN2B toxicity only becomes apparent during chronic activation, such as that found in extrasynaptic sites. Looking for the downstream mechanism behind GluN2B excitotoxicity, Hardingham found that GluN2B signaling represses CREB, an important neuroprotective factor.
The difference between GluN2A and GluN2B excitotoxicity might have consequences for disease. Lynn Raymond at the University of British Columbia, Vancouver, pointed out that GluN2B is enriched in GABAergic medium spiny neurons of the striatum, which selectively degenerate in HD (see Raymond et al., 2011). Degeneration probably occurs through an excitotoxic mechanism, Raymond noted, as injecting glutamate and NMDA into the striatum can reproduce the symptoms of HD in mice. Raymond’s group found that HD mice have more extrasynaptic receptors than do wild-type mice (see ARF related news story on Milnerwood et al., 2010). In addition, overexpressing GluN2B in an HD mouse causes more atrophy, Raymond said.
Raymond compared HD mice (YAC128), which express mutant huntingtin protein containing 128 CAG repeats, with control mice that have normal huntingtin with 18 repeats. She reported that the YAC128 mice have more GluN2B at the cell surface than do controls, and that most of it is extrasynaptic. Looking for the mechanism, she found that the scaffold protein PSD-95, which stabilizes and maintains GluN2B in synaptic sites, shifts to extrasynaptic locales in HD mice. PSD-95 binds directly to huntingtin protein, and binds more tightly to the mutant form, Raymond noted, which may be linked to the relocation. In addition, striatal-enriched tyrosine phosphatase (STEP) is more active in HD mice than in wild-type. STEP dephosphorylates GluN2B, causing it to abandon the synapse and potentially freeing it up for incorporation into extrasynaptic receptors. Inhibiting STEP increases synaptic GluN2B, Raymond said. Downstream of GluN2B, Raymond found that the mitogen-activated protein kinase p38 was more active in the HD mice, and inhibiting it protected cultured HD neurons from death, suggesting that this protein could be a therapeutic target.
One sign of the importance of extrasynaptic signaling in HD is that memantine, an approved AD drug that selectively silences extrasynaptic signaling (see ARF related news story on Xia et al., 2010), reverses motor learning deficits in HD mice (see Okamoto et al., 2009). A UBC group led by Blair Leavitt is currently testing memantine in HD patients in a Phase 2 trial.
Memantine relieves symptoms in moderate AD, probably by dampening extrasynaptic signaling. In his talk, Lipton turned to the role that this type of signaling plays in AD. Recent work showed that Aβ can inhibit synaptic glutamate reuptake, causing the neurotransmitter to spill over and activate extrasynaptic sites (see ARF related news story on Li et al., 2011). Lipton reported that oligomeric Aβ also induces cultured astrocytes to spit out more glutamate. Lipton noted that this finding dovetails with data presented by Annalisa Scimemi at the National Institute of Neurological Disorders and Stroke, Bethesda, Maryland, in a separate SfN session. The most abundant glutamate transporter in adult brain, GLT-1, is found mostly in astrocytes, and is responsible for mopping up extracellular glutamate. Scimemi reported that adding synthetic Aβ42 (oligomeric and monomeric) to hippocampal slices increased deposits of insoluble GLT-1 and doubled the time course of glutamate clearance by the transporter. This provides yet another mechanism behind elevated glutamate in AD brains.
Lipton, an author on worldwide memantine patents, said that the drug improves AD symptoms by acting as a low-affinity, transient NMDA channel blocker. This helps “turn down the volume” on NMDA transmission without silencing it, Lipton said. In addition, memantine blocks only open channels, allowing the drug to selectively block extrasynaptic receptors, which are chronically active, while largely sparing synaptic sites. However, memantine is not effective enough, Lipton said. He is developing a new version, nitro-memantine, that is even more selective for extrasynaptic sites. In addition to having the same effect as memantine in the NMDA receptor ion channel, nitro-memantine also nitrosylates the NMDA receptor, i.e., it transfers a nitric oxide group to a cysteine residue on the receptor, which desensitizes the channel. Lipton tested the drug in hippocampal slices from transgenic APP mice (J20), which show chronic extrasynaptic activity. In these mice, synaptic spine density is half that in wild-type mice. While memantine protects some spines, nitro-memantine restored spine density to virtually normal levels. Moving to an in-vivo system, Lipton reported that when 3xTg mice were treated with nitro-memantine for three months, levels of synaptophysin, a marker of synapses, returned to normal, and the mice improved in an object exploration test. In contrast, when treated with memantine, the mice showed no behavioral improvement. Lipton hopes the greater efficacy of nitro-memantine will eventually translate to people, although clinical trials are not yet scheduled.
Synaptic NMDA signaling not only protects neurons from insults, but also enables the encoding of long-term memories. In his talk, Bading noted that nuclear calcium may play a crucial role in learning, perhaps through the action of CREB, a key memory protein (see, e.g., ARF related news story). Using transgenic flies that express a neuronal calcium sensor, Bading found that calcium floods into the nucleus during learning. When he activated CaMBP4, an inhibitor of nuclear calcium signaling, the flies’ ability to remember an association 24 hours after learning dropped dramatically. Another gene stimulated by calcium signaling is vascular endothelial growth factor D (VEGF-D). Bading reported that VEGF-D maintains dendrite length and complexity, and is necessary for long-term memory in mice (see ARF related news story on Mauceri et al., 2011). Nuclear calcium signaling is not beneficial in all contexts, however. Blocking nuclear calcium signaling in the spinal cord dampens chronic inflammatory pain, Bading noted.—Madolyn Bowman Rogers.