Back

Cohesin erases genomic-proximity biases to drive stochastic Protocadherin expression for proper neural wiring

Kiefer, L.; Servito, G. I. F.; Rajkumar, S. M.; Langen, J.; Chiosso, A.; Buckley, A.; Cha, E. S.; Horta, A.; Mui, M. H.; Canzio, D.

2022-03-11 genomics
10.1101/2022.03.09.483674 bioRxiv
Show abstract

Clustered Protocadherin (Pcdh) proteins act as cell-surface recognition barcodes for neural circuit formation. Neurites expressing the same barcode repel each other, but this mechanism is deployed in two different ways. For instance, convergence of olfactory sensory neuron (OSN) projections requires stochastic expression of distinct Pcdh isoforms in individual cells, while tiling of neural arbors of serotonergic neurons (5-HTs) requires expression of the same isoform, Pcdhc2. Despite their essential role, however, the molecular mechanisms of cell-type specific Pcdh barcoding remain a mystery. Here, we uncover a new role of cohesin: that of regulating distance-independent enhancer-promoter interactions to enable random Pcdh isoform choice via DNA loop extrusion in OSNs. Remarkably, this step mediates DNA demethylation of Pcdh promoters and their CTCF binding sites, thus directing CTCF to the chosen promoter. In contrast, the uniform pattern of Pcdh expression in 5-HTs is achieved through conventional cohesin-independent, distance-dependent enhancer/promoter interactions, that favor choice of the nearest isoform. Thus, cell-type specific cohesin deployment converts a distance-dependent and deterministic regulatory logic into a distance-independent and stochastic one. We propose that this mechanism provides an elegant strategy to achieve distinct patterns of Pcdh expression that generate wiring instructions to meet the connectivity requirements of different neural classes.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.