https://www.science.org/doi/10.1126/sciadv.adg2067
Abstract: Extracellular vesicles (EVs) are heterogeneous in size, composition, and function. We show that the six-transmembrane protein glycerophosphodiester phosphodiesterase 3 (GDE3) regulates actin remodeling, a global EV biogenic pathway, to release an EV subtype with distinct functions.
GDE3 is necessary and sufficient for releasing EVs containing annexin A1 and GDE3 from the plasma membrane via Wiskott-Aldrich syndrome protein family member 3 (WAVE3), a major regulator of actin dynamics. GDE3 is expressed in astrocytes but not neurons, yet mice lacking GDE3 [Gde3 knockout (KO)] have decreased miniature excitatory postsynaptic current (mEPSC) amplitudes in hippocampal CA1 neurons.
EVs from cultured wild-type astrocytes restore mEPSC amplitudes in Gde3 KOs, while EVs from Gde3 KO astrocytes or astrocytes inhibited for WAVE3 actin branching activity do not. Thus, GDE3-WAVE3 is a nonredundant astrocytic pathway that remodels actin to release a functionally distinct EV subtype, supporting the concept that independent regulation of global EV release pathways differentially regulates EV signaling within the cellular EV landscape.
Commentary: So if the model is correct, that it is astrocytes which remodel themselves in response to stimuli and transfer that remodeling to neurons for “long term encoding” via RNA expression changes, we should be able to see a few things: 1) that each local group an astrocyte is synapsed with should have a discrete morphology which represents the response data being stored, and 2) and a mechanic through which an astrocyte can monitor, suppress, and modify a signal through a synapse.
This paper adds support for the model by demonstrating parts of the physical remodeling process, demonstrating that the process is initiated by and requires astrocytes, and that the each vesicle contains a discrete morphology which modifies signalling on the neuronal side when a good match is made.
Interestingly, this suggests to me that astrocytes “never forget”, and have response data hardcoded into them even after the neuronal remyelination and pruning has occurred. Neurons themselves can be remodeled, and even have certain synaptic end points destroyed (“forgetting”), but astrocytes themselves are cumulative. Modifying astrocytic vesicles can have downstream effects on how other vesicle functional groupings are calculated.
One of the more interesting (to me) MOA theories for drugs like ketamine is that they work as a cholesterol sponge, sopping up excess cholesterol (and probably amyloid species build up) intracellularly and allowing astrocytes to signal “loud enough” to re-open previously obstructed pathways.
This type of work, which demonstrates the global and local effect of information processing, IMO lends weight to this concept by describing how ketamine can “reset” circuits without being destructive to data.
Just all around a really fascinating piece of work that happens to align pretty closely to what the model predicts.