Today after pitches from each of the discussion groups in https://capocaccia.cc/2018/workgroups/, we started with a discussion session on brain development organized by Gaby Michel, with Denis Jabaudon and Adam Kepacs.
Denis started off with aim to show 3 aspects of brain development: Cell diversity in cortex, connectivity of areas, progenitors, and activity dependent connectivity. He had to stop after the first due to lack of time.
From the neural tube, progenitor cells migrate to the cortical surface, maybe following chemical signal gradients. The brain is built over a week (in mouse) into 6 layers with many diverse cell types. Each layer is enriched with a particular type of cell. Each embryonic day, a new type of cell dominates birth and initial migration. Day E12, layer 6 going to E15, layer 2/3. You see bumps of production centered on these days. Cells are labeled, e.g. with radioactivity to see what cells are born when and where they go. It's pretty hard in mouse because it's so fast, but in primate it is easier because development is slower.
Nowadays, transcriptonics is the gold standard for classifying cells. But RNA doesn't say everything. It leaves out slicing of RNA or other things that are controlled. Also, you might be too sensitive, and find different clusters of cells located far apart that seem related but also different in terms of RNA, but are actually the same cells at different maturity. In response to question from Rodney, Denis stated that the RNA pattern differences are known between subcortical and local projecting cell types. A transcription factor is a protein that controls gene expression, i.e. manufacturing of proteins. For example the transcriptions factor for dopaminergic cells is known.
If you had to divide cortex, one possibility is deep and superficial layers. Related to glutaminergic cells (excitatory), and projection neurons, the deep layer neurons are corticalfugally projecting (to other areas), while superficial neurons project locally. This has implications for wiring, but it has an evolutionary origin: The thalamic nuclii getting primary sensor input (e.g. from retina) project to superficial layers. Other nuclii get input from internal brain areas, e.g. pulvinar. These project to deeper cortical layers. Each cortical area has a nucleus in thalamus, but these nuclii do not connect to each other at all in thalamus. This control of cortex via cortical thalamic cortical connections is universal across species.
Driven by the audience, the discussion veered away from development and back to the theme of brain area connectivity covered by Rodney on Monday morning.
The main point Denis of this discussion about thalamus was that thalamic nuclii evolved in reptiles to connect cortical areas since the thick scales prevented high resolution sensing. It was only later that evolution co-opted these nuclii to pass high resolution sensory input to cortex and to regulate connectivity between areas via diffuse modulatory connections.
Adam Kepacs started off before the coffee break, with a discussion of cell type architectronics, with one example of area VIP.
He started out discussing the question, "what if I'm wrong?" in relation to confidence, defined as P(correct|evidence, choice). It is a scalar, not distribution. It is subjective, but is subject to error since evidence is not always available. There are some expts that allow mapping human's subjective confidence to external truth. Behavioral expts are possible, but can we do it in cortex?
The discussion went on for some time but specificity that would anchor it was missing. Next report, after the coffee break.
Comments
Post a Comment