,
16 tweets,
8 min read
Read on Twitter
Congratulations Yu Mu @muyuuyum, Davis Bennett @davisvbennett, Mika Rubinov @rubinovlab, Sujatha Narayan @SujNarayan, and collaborators. It has been a wonderful, exciting scientific journey.
Here is a summary of the work:
Perseverance can lead to success, but some challenges are best met by high-activity bursts separated by behavioral quiescence.
How do animals collect observations that actions have been unsuccessful, and how do they abandon futile behavior?
Perseverance can lead to success, but some challenges are best met by high-activity bursts separated by behavioral quiescence.
How do animals collect observations that actions have been unsuccessful, and how do they abandon futile behavior?
We studied 'giving up' in zebrafish (or 'futility-induced passivity').
For zebrafish in a VR environment, where they normally move forward through the environment when they swim, we made it impossible for them to travel, as if they are stuck. Do they give up, like we would?
For zebrafish in a VR environment, where they normally move forward through the environment when they swim, we made it impossible for them to travel, as if they are stuck. Do they give up, like we would?
They do – after about 20 swim attempts, they suddenly stop trying. After a period of behavioral quiescence, they start swimming again. It’s as if they sense that their actions are futile, give up for some time, and then try again. 
We used light-sheet microscopy to image calcium in neurons and a glial cell type called 'radial astrocytes' across the entire brains of behaving zebrafish, and noticed that the radial astrocytes (bottom row) were very active during the passive state.
The most consistent signals were in the lateral medulla oblongata (L-MO), as you can see here in a later part of the experiment where the glial signals are smaller but still consistently present in L-MO...
... as established by statistical analysis of the calcium signals,
Activity in these glial processes in L-MO ramped up before the fish gave up, peaked after, and then decayed slowly. Could these cells be the causal driver of the behavior?
Using six different perturbation methods, including opto- and chemogenetics, ablation, and pharamacology, we confirmed they were indeed driving the passive behavioral state.
We next discovered that the glia were activated by a noradrenergic brainstem nucleus called NE-MO. This nucleus encodes the mismatch between 'expected' and perceived visual feedback during swimming. In other words, it encodes swim failures.
Here is activity in NE-MO cells and radial astrocytes during behavior in a mixed closed/open loop protocol. In a related experiment, we optogenetically activated NE-MO, which caused similar activation in L-MO radial astrocytes.
In stochastic environments, where swims are probabilistically successful or unsuccessful, the L-MO projecting astrocytes represent the evidence that actions are futile.
To conclude, this work establishes radial astrocytes as essential computational elements in experience-dependent brain-state switching.
One more video, a top projection (sped up 7x) with glia on the left, neurons on the right. Glia are most vigorously active after the first closed->open loop transition; later on in open loop activity becomes more localized to L-MO when fish give up (see earlier video).
It was a wonderful team effort that included Yu Mu @muyuuyum, Davis Bennett @davisvbennett, Mikail Rubinov @rubinovlab, Sujatha Narayan @SujNarayan, Brett Mensh @spiralmensh, Loren Looger. Here is also a video abstract:
