Back

N-terminal signals in the SNX-BAR paralogs Vps5 and Vin1 guide coat complex formation

Shortill, S. P.; Frier, M. S.; Davey, M.; Conibear, E.

2024-01-25 cell biology
10.1101/2024.01.24.576755 bioRxiv
Show abstract

Endosomal coat complexes assemble by incorporating membrane-binding subunits such as those of the sorting nexin (SNX) family. The S. cerevisiae SNX-BAR paralogs Vin1 and Vps5 are respective subunits of the endosomal VINE and retromer complexes that arose from a fungal whole genome duplication. Interactions mediated by the Vin1 and Vps5 BAR domains are required for protein complex assembly and membrane association. However, a degree of promiscuity is predicted for yeast BAR-BAR pairings, suggesting that another mechanism guides the formation of specific endosomal coat complexes. Previous work by our group and others has implicated the unstructured N-terminal domains of Vin1 and Vps5 in complex assembly. Here, we map N-terminal signals in both SNX-BAR paralogs that contribute to the formation and function of two distinct endosomal coats in vivo. Whereas Vin1 leverages a polybasic region and adjacent hydrophobic motif to bind Vrl1 and form VINE, the N-terminus of Vps5 interacts with the retromer subunit Vps29 at two separate sites. We show that one of these Vps5 motifs binds to a conserved hydrophobic pocket in Vps29 that is shared with other accessory proteins and targeted by a bacterial virulence factor in humans. Lastly, we examined the sole isoform of Vps5 from the milk yeast K. lactis and found that ancestral yeasts may have used a nested N-terminal signal to form both VINE and retromer. Our results suggest that the specific assembly of Vps5-family SNX-BAR coats depends on inputs from unique N-terminal sequence features in addition to BAR domain coupling, expanding our understanding of endosomal coat assembly mechanisms.

Matching journals

The top 5 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.