Modeling transmembrane and intracellular domain

[Anonymous]

Hi, everyone,

I am working on a transmembrane protein which is a homologus trimer.Its crystal structure contains only the extracellular portion but not the transmembrane and intracellular portion.  I need to model the missing transmembrane and intracellular domains. Does anyone know a pipeline in ROSETTA to do this job? I only know that the Rosetta application “helix_from_sequence” can create the structural model of a single transmembrane helix, but how to build the intracellular domain and assemble the modeled transmembrane region and the intracellular domain to the extracellular domain? How to ensure the correct contact among the three  modeled transmembrane helixes and the correct orientation when doing the assembly?

 

Is there anyone know how to solve these problems? I will be very much grateful if you can help. 

Thank you in advance.

[Anonymous]

The best protocol to use depends heavily on the type of protein you want to model (helical vs beta-barrel) and the sequence homology to a known structure. If it is high (>30%), you can use the RosettaCM protocol. If the sequence homology is low (I would say about 10-30%), you can try Sergey Ovchinnikov’s Gremlin protocol to predict evolutionary constraints from the sequence – just be aware that you need a deep sequence alignment for that (many sequences, I am not quite sure how many they typically take – it might be 5x sequence length). Also, things are a bit more difficult for a trimer as it’s hard to predict the interfaces. If the sequence similarity is too low and/or the sequence alignment is not deep enough to get enough sequences for a ‘good’ contact prediction from evolutionary constraints, ab initio structure prediction might be the only way to go. This protocol is the oldest for membrane protein structure prediction – it was published by Yarov-Yarovoy in 2006 and Patrick Barth in 2007. 

If you provide more detail about the protein, I might be able to help you more. Like, what type of protein is it? Which family does it belong to? Do you know exactly how many TM spans it has? Which TM span prediction methods did you use? [Not all of them are equally good and databases like Uniprot sometimes use methods that are not necessarily state-of-the-art.]

[Author]

Posts