Many thanks to my wonderful Ph.D. advisor Eugene Shakhnovich for supporting me through this work, to Marina Rodnina and her lab for an inspiring collaboration from which I learned so much, and to my co-first author Siyu Wang
for her dedicated and brilliant work on the experiments presented here

This study, combining simulation and experiment, generates a unique atomistically-resolved picture of co-translational folding and the pipeline represents a technical advance that may inform future mechanistic folding studies on complex proteins, including those implicated in misfolding disease

The co-translational chaperone trigger factor is found to also destabilize this intermediate, consistent with previous findings suggesting this chaperone may prevent premature nasent chain collapse and favor post-translatioanl folding in certain proteins

In contrast, mutations to other residues were predicted minimally affect the intermediate while strongly impacting the final native state stability, pointing to a hierarchy of interactions in both states. Remarkably, experiments by the Rodnina lab using PET and FRET validated this exact hierarchy.

Interestingly, our simulations predict this non-native intermediate is stabilized by interactions between hydrophobic residues that are solvent exposed in the native state. Thus, our simulations predicted that by mutating these specific residues, the co-trans intermediate would be disrupted

To address this, we ran atomistic simulations of small HemK terminal domain, previously shown by the Rodnina lab to adopt a non-native co-translational intermediate. Using DBFOLD, a technique I developed in my PhD, we could predict detailed atomistic structures of this intermediate

Many proteins are known to start folding as they are being translated on the ribosome, and this non-eqilibirum process may be critical for ensuring correct folding. Yet it is extremely challenging to gain a detailed atomistic picture of translational folding with any single existing technique

Although deep learning tools like AlphaFold have transformed our ability to predict protein structures, we still do not fully understand how these native states are mechanistically attained in vivo, a question with key implications in misfolding disease and improving protein design

Hi, this link is broken fyi

congrats Julia!!

heard it as a rumble more than I felt it in Berkeley