Thursday, October 19, 2006

Fast and slow spine dynamix   posted by Coffee Mug @ 10/19/2006 08:50:00 PM

Shame I gotta do white labels to keep my life stable. - Common

I checked out a couple sets of posters dealing with spine dynamics at SFN. One bundle from the Hayashi lab showed that the beta subunit of CaMKII serves an actin bundling role in dendritic spines. Get used to hearing about CaMKII. It makes up two percent of brain protein and has several interesting properties that make it intriguing as a memory-related signaling molecule. CaMKII at the synapse takes the form of 12-subunit donut-like structures (two interlocked 6-subunit structures). There is a family of CaMKII subunits. I don't know all of their properites, but CaMKII alpha is the one everyone usually concentrates on because it is synthesized following activity and has these nifty self-activating features that might make it good for maintaining a synaptic biochemical state. CaMKII beta is the boring wallflower subunit, but Hayashi may now have given it a chance to shine. They showed that actin polymerization (forming into long chains or scaffolds that provide structure for the spine) causes more beta-CaMKII to show up in dendritic spines. Dendritic spines are little knobs that stick off of dendrites where most excitatory synapses occur (here's some pics). The beta subunit slows down the actin dynamics responsible for spines wobbling around. Wobbling around is immature behavior for a spine, indicative of a weak synapse or perhaps searching for a new synaptic connection. Experimental reduction of beta CaMKII makes spines look more like filopodia, long, thin, wobbly growths off the dendrite (presumed to be the intial stages of spine formation). Expressing the beta-CaMKII binding domain can rescue this phenotype and make the spines look more stable and mature. Hayashi suggest a model in which initially actin is bundled in a fairly weak spine. Activity can destablize this bundling and allow the spine to wobble some and perhaps rearrange the post-synaptic protein formation. This might be mediated by increasing the alpha subunit content. I'm forming an alpha=unstable, beta=stable dichotomy in my mind right now. After the activity-induced destabilization, more beta-CaMKII can come in and re-stabilize/re-bundle the actin filaments in a new (maybe stronger) configuration.

Sutton and Schuman had some posters continuing their examination of synaptic homeostasis and mini EPSPs (excitatory post-synaptic potentials). Refresher: minis are caused by spontaneous neurotransmitter release even when there is no activity driving action potentials. We know from previous work that blocking minis can cause synapses to start inserting new receptors and grow in strength. In one poster they used a GFP flanked by the regulatory elements of the alpha-CaMKII mRNA to assay a certain signaling pathway's role in mini-mediated dendritic protein synthesis regulation. The experimental techniques get convoluted quickly because every manipulation involves inhibition of two processes to discover the effect of one. I'll skip those, you are welcome. In general they found that minis tend to increase phosphorylation of eEF2 and that action potentials tend to decrease the same. eEF2 phosphorylation is thought to decrease global translation during the elongation step. I mentioned it a couple weeks ago in reference to alpha-CaMKII synthesis as one of those counterintuitive regulatory paths in which turning down translation increases certain genes. Sutton and Schuman didn't find anything like that. Phospho-eEF2 (and thus minis) was associated with suppression of the reporter signal. Inhibiting the kinase that phosphorylates eEF2 (something that action potentials, non-spontaneous activity, might do) led to more protein synthesis and more reporter signal. Sutton didn't seem overly concerned that the direction of regulation for alpha CaMKII was opposite to that reported in Scheetz et al. They focused more on the idea that it is relatively uncommon to think of translation being regulated at the elongation step instead of the initiation step. I'll give'em that, but I thought the Scheetz story was cool, so I wish they would examine it a little more closely.

The other poster from Sutton and Schuman examined fast and slow spine dynamics in relation to minis and action potentials. We are now getting a dichotomy between minis and action potentials as stabilizing and destabilizing forces respectively. Fast dynamics was defined as movement of the spine during a 5 minute monitoring window. Slow dynamics usually involved spine growth (in the more stable widening of the head and neck manner) over about an hour or two. Minis tended to reduce fast dynamics and increase spine growth (slow dynamics) and action potentials did the opposite. Blocking minis or action potentials over an extended period of time produced effects on not just the size and shape of individual spines but also on the number of spines period. Initially letting minis run wild led to an increase in spine number, but over days this dropped off below baseline. The effect of action potentials appers to be to initially drop off the number of spines and then stabilize, maybe as neurons are adapting to the new higher activity level. Everything I have written has to be taken with the caveat that you can't actually isolate the effect of action potentials. All of the effects attributed to action potentials are really taken from experiments blocking the effect of minis AND action potentials at the same time, basically blocking any neural signals at all. The effects I attributed to minis are more directly related to actually looking at the different response between neurons with minis and without.

So with just a few posters we can build up a couple categories: stabilizing forces versus destabilizing forces.
  • Stabilizing: miniEPSPs, beta CaMKII, actin polymerization, eEF2 phosphorylation, inhibition of translation at the elongation step, inhibition of global protein synthesis.
  • Destabilizing: activity, action potentials, alpha CAMKII, protein synthesis upregulation
I'm thinking that memory may be encoded by passing through a destabilizing phase to a new, stronger stabilized phase. This would allow an interesting explanation for the phenomenon of reconsolidation. Reconsolidaton is complicated and a matter of debate, so keep in mind that the story is more complicated that I will present it now. Reconsolidation is demonstrated by allowing an animal to retrieve an established memory and then performing some manipulation to screw with their memory consolidation capabilities. For instance, you can give a strong electric shock or certain drugs. The animal then seems to have forgotten that particular reactivated memory in subsequent testing. Maybe the initial destabilizing force of synaptic activity for those memories allows a window of time when, if the pattern of activity becomes erratic, the memory will not be put back together right during the stabilizing phase.