Metabotropic mechanisms of NMDARs, the classically defined ionotropic receptors
Reflections on revelations from the symposium ‘Unconventional NMDA Receptor Signaling’ or One example of when it is okay to teach the controversy
The history of science is riddled with controversies. The earth is round versus the earth is flat. You should wash your hands between medical procedures versus washing your hands between medical procedures is a giant waste of time. Ham and pineapple are acceptable toppings on a pizza versus sensibility. On Tuesday this symposium:
introduced us to a new controversy: NMDARs may, in addition to their ionotropic function, signal through metabotropic pathways.
NMDA receptors are classically taught as being ionotropic receptors that operate postsynaptically. Most lectures will use an image such as this one that I took from a class I had discussing NMDA:
Source for photo unknown
It shows an ionotropic receptor with multiple inhibitory sites, a binding site for its main agonists and a co-binding site for the co-agonists. Just as a quick reminder these receptors require both agonist binding and depolarization of the cell, which removes the endogenous channel blocker magnesium from the pore, in order to pass ions. When you look at this diagram, the intracellular domain of the receptor shows nothing; there is no information about interactions NMDARs may have with intracellular proteins that can affect signaling pathways in the absence of cation flux. That may change in the not-too-distant future.
One of the speakers for the symposium, Dr. Kim Dore, explored the possibility that conformational changes of NMDARs have downstream effects on postsynaptic plasticity. Using FRET, her lab showed that in the presence of glutamtergic site agonists, antagonists to the co-agonist site and open channel blockers, NMDARs undergo a conformational change. It makes sense that NMDARs will change conformation simply with binding of agonists as the only other necessity for NMDAR ionotropic function is cell depolarization. Also, in the presence of the same pharmacological environment, long-term depression (LTD) occurred, a process we typically associate with calcium influx through NMDARs. Next, with an NMDAR GluN1 subunit antibody, Dr. Dore was able to block this conformational change, again shown via FRET, and LTD disappeared. These experiments hinted at the necessity of NMDAR activation, but not ionotropic function, for LTD. The rest of her talk was centered around uncovering the downstream pathway (PP1 -> CaMKII -> AMPAR endocytosis) which can be seen in the image below.
Photo courtesy of http://www.jneurosci.org/content/37/45/10800
This diagram shows an additional metabotropic function of postsynaptic NMDARs, spine shrinkage via p38 activation. Work from the lab of Dr. Karen Zito addressed metabotropic NMDAR function by investigating spine shrinkage. Her group hypothesized the spine shrinkage is due to NMDAR-driven LTD given that:
- Activity patterns driving synaptic weakening (LTD) also shrink or eliminate dendritic spines and
- APV blocks shrinkage of spines (NMDAR dependent)
Point 1 was exemplified by stimulating axons via electrode. Dr. Zito examined shrinking in over 70% of her monitored spines following electrode stimulation with the stipulation that stimulation was occurring over a rather large area that is not representative of a physiologically relevant excitatory pattern. However, it led to the question of what is driving the loss of individual spines during this stimulation? Her group carried out single spine activation via low-frequency uncaging (LFU) of glutamate at a monitored spine, inducing an NMDAR-dependent LTD (this LTD was blocked in the presence of CPP, a competitive NMDAR antagonist). At high-frequency uncaging (HFU) they saw spine enlargement, a sign of long term potentiation (LTP) but with the NMDA EPSC and calcium ion transients blocked, that enlargement was converted to shrinkage. To address the role of AMPARs and mGluRs in this mechanism, they blocked those receptors pharmacologically and there was no effect on shrinkage. Again, through protein manipulation she was able to show the acting proteins that carry out the metabotropic signal (p38 -> MK2 -> spine shrinkage; likely driven by actin reorganization via RhoA).
The above diagram also includes presynaptic NMDARs functioning without ion flux. Dr. Jesper Sjostrom presented evidence that while evoked vesicular release appears to be dependent on ionotropic mechanisms of NMDARs, spontaneous release does not. This lecture was really challenging to a lot of previously held notions; however, Dr. Sjostrom used cumulative EPSC plotting to show a decreased vesicle replenishment rate (as it mirrors vesicle release) after wash-in of the NMDAR blocker Ro. Ro was used here as it specifically blocks NMDARs that contain the GluN2B subunit, which are only found presynaptically. Additionally, the Ro blockers are competitive inhibitors of the internal MK-801 site, meaning they are open-channel blockers and do not interfere with the agonist binding sites. The same results were seen in low calcium environments, which would imply they are reliant on calcium flux. This mechanism may only be effective for evoked vesicular release, though. During evoked release cells are constantly being depolarized. Glutamate released during an action potential will retroactively bind presynaptically and by the time they are bound the cell will again be depolarized, opening magnesium-blocked NMDARs and allowing brief periods of calcium influx. This triggers vesicle release and the process starts over again. This may not apply to spontaneous release if the short depolarization that allows for vesicle release subsides prior to the glutamatergic binding on these autoreceptors. Spontaneous release was shown to not be dependent on magnesium, implying a metabotropic pathway. The Sjostrom group then investigated downstream effectors and found that RIM1 was necessary for evoked release but not spontaneous release and JNK2 was necessary for spontaneous release but not evoked release, implying two completely different pathways for these actions.
I would be remiss to not mention Dr. Pablo Castillo’s lecture that began the symposium. Dr. Castillo showed evidence of NMDAR plasticity and even a potential paired ion channel (Pannexin-1) that can dynamically change the synaptic function of neurons. While his presentation did not add to the above diagram outlining metabotropic functionality of NMDARs, it did highlight the dynamic activity of these receptors which is extremely consequential to cellular signaling and function.
A quick aside: This brief review necessarily left out a lot of details for the sake of time. If you are interested in this topic, you should absolutely read the symposium review in the Journal of Neuroscience (found here) and follow up with previous work from the authors.
This impressive slate of talks began with a disclaimer: Some of the findings and conclusions they presented have been reproduced while other studies have shown evidence to the contrary. A question arose around the possible involvement of voltage-gated calcium channels and the panel acknowledged that their activation alongside NMDAR metabotropic signaling has not been studied, offering a possible calcium source and alternative explanation to these studies. I found the acknowledgment of the conflicts between this talk and the literature to be refreshing and I appreciate scientists that welcome open and honest dialogue around their work. Dr. Sjostrom stated that the purpose of the symposium was to acknowledge the need to expand our view on NMDA receptor function. We are trained to be skeptical and that skepticism should not be limited to the work of others. I know I will approach my work with NMDA (and all receptors) with a more open mind following this symposium and I hope you all will too.