Friday, June 16, 2006

BDNF and Arc regulation: NMDARs vs AMPARs   posted by Coffee Mug @ 6/16/2006 11:00:00 AM

Brain-derived neurotrophic factor (BDNF) is very busy around the nervous system affecting things like cell survival, synaptic transmission, immune responses, and plasticity. It's been discussed around here before in a number of contexts. New articles come out fairly frequently concerning the effects of the Val66Met polymorphism in human BDNF on risk for various psychiatric conditions and also cognitive function in general. I will leave discussion of those to someone more qualified. It is becoming clearer how these more global effects occur at a basic level. Here's one, for instance, in Nature Neuroscience where experience-dependent plasticity in humans is affected by the polymorphism. I'm going to go down even further into the nitty-grit with this article (Rao et al. 2006) that came out at about the same time in the same journal concerning the mechanisms by which BDNF affects levels of a protein called activity-regulated cytoskeletal protein (Arc). The article provides a thorough examination of the interaction between BDNF, synaptic transmission, and gene regulation and provides a novel role for what is normally considered just a reliable workhorse of excitatory transmission (the AMPA-type glutamate receptor).

Arc is a favorite among memory researchers because it has all these nice properties that make it look like a good final effector for synaptic plasticity. The RNA and protein both increases in response to plasticity-inducing stimuli and experience in novel contexts (possibly the plasticity involved in creating a cognitive map). The RNA is localized in dendrites where it could be rapidly translated to make changes to post-synaptic properties. One of the most compelling findings is that even within the same dendritic field the protein is translated in an input-specific manner. Imagine a long dendrite with synaptic inputs up and down it's length. Tetanic stimulation of a pathway that only inputs to the distal regions of the dendrites will cause increases in protein just out there and not in portions of the dendrites more proximal to the cell body. This is important because learning is presumed to involve very fine-tuned changes at specific synapses rather than wholesale whole-cell changes. It is kind of funny given all this evidence that no one is really sure what Arc is doing once it gets synthesized. It is associated with the cytoskeleton... maybe it affects actin polymerization, but no one has really conclusively shown what its function is.

There is a model in the making that BDNF is released in response to tetanic stimulation and may act on the pre-synaptic side to increase neurotransmitter release or on the post-synaptic side to make changes to synaptic signaling pathways or receptor content. If you block BDNF, synaptic plasticity is impaired. If you apply BDNF to neurons, Arc levels will increase. One of the first questions addressed in this paper is kind of a chicken and egg problem. Does synaptic activity increase BDNF and thus lead to Arc expression or is it that BDNF increases synaptic activity and that is the Arc synthesis cue? The data points to the latter scenario since sequestering BDNF with a fake BNDF receptor molecule didn't affect activity-dependent Arc increases while blocking neural activity with an ion-channel blocker inhibited the Arc increases elicited by direct BDNF application. So at least in this in vitro system BDNF leads to synaptic activity which then leads to Arc RNA and protein increases.

So which sort of synaptic activity are we talking about here? Excitatory transmission is primarily a function of ionotropic (ion-channel) glutamate receptors. There are three major subtypes differentiated by their susceptibility to certain synthetic agonists: AMPA, NMDA, and kainate. Everyone ignores kainate which I'm sure is gonna come back to haunt us in the future, but for now we'll focus on AMPA and NMDA receptors. NMDA receptors are much loved by memory researchers because they provide a nice molecular analogue of the requirements for memory. They are usually blocked by a magnesium plug, but when a the post-synaptic membrane is depolarized (excited) the plug pops out. Normally glutamate binding to NMDA receptors won't do much of anything, but when the magnesium plug is gone calcium can pass into the cell and start initiating this or that signaling cascade and doing things that we think are important for changing synaptic strength. Since these changes are supposed only to occur when there is a coincidence between post-synaptic depolarization and pre-synaptic neurotransmitter release, the NMDA receptor has been labeled a "coincidence detector". While there are exceptions, one common requirement for classical conditioning is that the conditioned stimulus occur close in time to (coincident with) the reinforcing (unconditioned) stimulus. See the analogy? You can get these two levels of analysis a lot closer in a reduced system like the sea slug and really hammer home the link between learning and NMDA receptor activation.

The AMPA receptor on the other hand is usually thought of as less of a mover in synaptic plasticity and gene regulation and more of a movee. The final outcome in synaptic strength changes is often presumed to be trafficking of AMPA receptors into the post-synaptic membrane where they have the effect of producing greater neural activation in response to the same amount of neurotransmitter. So it was rather surprising to find AMPA receptors playing an active role in supressing gene expression in this paper. As expected, BDNF-induced Arc increases were inhibited to some degree by NMDA receptor antagonists. However, when the authors tested out AMPA receptor antagonists, they found a really striking increase in BDNF-induced Arc levels. I usually just think of AMPA receptors as allowing some sodium into the cell everytime they get glutamate. Sodium isn't supposed to be any big signaling second-messenger or anything, it is just supposed to change the membrane potential. It makes one really curious about the way that AMPA receptors could control the level of a given gene.

One relatively simple way out would be to blame the regulation on calcium-permeable AMPA receptors. Certain configurations of AMPA receptor subunits allow calcium in and that could be initiating some sort of regulation, but the authors eliminated that possibility early on using a drug specific to these subunits. Instead it appears that AMPA receptors are having their effect through associated G proteins. I'm guessing that it isn't so much the sodium influx as it is the conformational change associated with ion-channel opening that may trigger activation of these G proteins that go on to inhibit the cyclic AMP signaling pathway.

By altering these intracellular signaling pathways, AMPARs could be affecting all sorts of cellular processes. Which one's really matter for Arc regulation? The authors found that rather than directly antagonizing the immediate downstream effects of BDNF (certain kinases phosphorylated and whatnot), AMPARs must be working less specifically against BDNF and more directly on Arc expression processes. They also examined at where in the process of Arc synthesis this regulation is occuring. Protein and RNA levels increase in response to BDNF, but this could be because of reduced degradation of either one, increased synthesis of RNA and followed by protein increases, or independent increases in RNA and protein. The verdict is that AMPAR antagonists are no big deal when it comes to protein synthesis or degradation pathways. Rather, the majority of AMPAR impact is on Arc transcription.

This data is fairly consistent with a model of plasticity that relies heavily on nuclear signaling to stabilize long-term changes in synaptic strength. Especially interesting that G proteins and cAMP have popped up again. Everyone's favorite learning and memory related transcription factor is called CREB (cAMP Response Element Binding protein). Since AMPAR transmission activates a G protein that should inhibit the production of cAMP, one could imagine a simple scenario in which CREB is just held at a lower activation level and is less capable of positively regulating Arc transcription. This puts NMDA receptors at odds with AMPA receptors in terms of their downstream effects. The way I'm thinking of it for now is that AMPA receptors may act as negative feedback. I'm envisioning something like NMDA receptors initially increasing Arc production and plasticity which entails trafficking of AMPARs into the synapse. These would allow the synapse to be more easily depolarized and could potentially lead down a positive feedback track. But instead, along with their depolarizing properties AMPARs also limit the amount of plasticity-related proteins that can be created at once and therefore keep the cell from building synapses bigger all willy-nilly.