Post

so I just realized I forgot to #skeetprint my Contactomes paper... (is that what we're calling it?) www.biorxiv.org/content/10.6... @kordinglab.bsky.social @dacolon.bsky.social smithneurallab.weebly.com @jhuapl.bsky.social #academicsky #preprint #neuroscience

TLDR: We know neurons communicate at synapses. But they also talk in a bunch of other ways! diffusion, glial interactions, ephaptic coupling, neuropeptides (cc @lidia-ripolls.bsky.social!) The list is long. probably longer than we know :) So let's make a map of all physical contact between neurons!

I first got excited about this ~8y ago when I started seeing a pattern in #microns and other datasets at bossdb.org @alleninstitute.org @bcmhouston.bsky.social @princeton.edu @mit.edu Usually neuron cell bodies are spread out... But every so often, you get what I call "cuddle buddies". wtf?

More generally, if you knew every cell a neuron touches, would you know something new beyond its synaptic #contactome? How many other cells does a neuron touch, anyway? Here was our first exploration: a single pyramidal cell (cyan), followed by all its synaptic partners, and then all its contacts:

Our first result was super obvious: in mice, CONTACT networks (the #contactome) are 5-10 times as dense as the synaptic connectome.

(That also presents the engineering challenge and the reason these analyses aren't commonplace: the resulting network is 100s of GB, before you even _think_ about the ~petabytes of underlying data to generate that.)

It also turns out that different cell types have different "fingerprints" of their volume to surface-area relationships. Unsurprisingly, you can immediately tell who's a neuron and who's glia (green = glia, blue = neuron), and in fact different cell types even have their own "sub-fingerprint"!

SECOND: One of the weaknesses of connectomics today is we only get a single snapshot in time. If we instead watched the neurons form and kill synapses, how many new partners might be available? One way to learn this is to look at a "halo" around each cell: how many un-contacted neurons are nearby?

ok cool so let's take each neuron (blue), and draw increasing radii around it: 0.5µm (green), 1µm (yellow), 2µm... all the way up to 10µm. That's actually quite a lot of distance! ~1000x more volume than the neuron itself. So how many new partners are there in that HUGE halo we don't already touch?

The answer, INCREDIBLY, is... basically ZERO! *A neuron already touches virtually every neuron that comes within ~10µm of its cell membrane.* omg!

I mean... that's crazy. That means that synaptogenesis needs to move MANY microns away to even see a new potential partner. That has some pretty big implications for biological learning algorithms, which, like, I hear that's a big deal bsky.app/profile/sean... #neuroai

When we look at contactomes instead of connectomes, we also get an obvious bonus: GLIA!! Glia aren't commonly involved in synapses, but they are EVERYWHERE. Lots of chatter recently about astrocytes: www.thetransmitter.org/astrocytes/t... @thetransmitter.bsky.social #astrocytes #glia

So let's see if we can find any evidence of an astrocytic long-range network (a "syncytium," which is totally award-winning spelling for a word)... Let's look at our contactomes and get the network of astrocytes-contacting-astrocytes. Can we see that long-range network?

Seeing that astrocytic network was a jaw-drop moment for me. Thank you to Cody Smith at Notre Dame for driving this science: This changed the way I think about neuroscience entirely!! smithneurallab.weebly.com

For our next trick: Does that glial network touch synapses specifically? Another resounding YES: PRACTICALLY EVERY SYNAPSE has a glial cell within 1µm, and near the synaptic site, almost ALL of the surrounding tissue is glial. Glam-shot: a synapse's microenvironent is dominated by an astrocyte :)

And it's just... always true. EVERY synapse is like this. Glia are EVERYWHERE. Here's the same synapse in two views; the glia (green) surrounds the synapse, and could foreseeably tune/train/kill it. HOT TAKE: I think big chunks of learning & memory will turn out to be mainly driven by this :)