Torsion angles to map and visualize the conformational space of a protein

Abstract

Present understanding of protein structure dynamics trails behind that of static structures. A torsion‐angle‐based approach, called the representation of protein entities, derives an interpretable conformational space that correlates with data collection temperature, resolution, and reaction coordinate. For more complex systems, atomic coordinates fail to separate functional conformational states, which are still preserved by torsion angle‐derived space. This indicates that torsion angles are often a more sensitive and biologically relevant descriptor for protein conformational dynamics than atomic coordinates.

This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/8237350.

Summary:

In this work, the author asks how protein structures change based on analyzing the torsion angles. Through examples they show that the distribution of points in this representation correlates with resolution and data collection temperature of the structures. They also construct the RoPE space of a protein using time-resolved experiment datasets and show that minor changes in the linear coordinate space are clearly observed in the RoPE space. This work demonstrates the utility of a non-linear representation of the conformational space in visualizing changes throughout the structure which are originally considered subtle. This work is very interesting and can have significant impact on ensemble studies on protein structures and in crystallization/cryo-EM and fragment screening efforts by showing the impact of temperature and resolution. The manuscript is very concise (perhaps too concise?) and well written.

Major points:

In Page 3, para 2, the author states differences associated with data collection temperature is preserved across space groups for trypsin and lysozyme but Figure 1(a) and 1(b) marks different space groups only for lysozyme and not for trypsin
The section on carboxymyoglobin has some unclear statements:
1. "The RoPE space of these structures showed that, over the first three picoseconds, two torsion angle modes are sufficient to represent a clear trajectory during release of carbon monoxide". Fig 1(e) does show a trajectory from -0.1ps to 3.0 ps but it is not clear how two torsion modes are sufficient to build the trajectory.
2. "The last three timepoints, 10 ps, 50 ps and 150 ps, are therefore beyond the biologically relevant timescales for CO dissociation in myoglobin and in-line with this, they did not strongly correlate with any other timepoints in RoPE space". We are confused about which figure/data supports this non-correlation. Is it to be interpreted from Fig 1(e)? If yes, then the author should describe what is correlation and non-correlation in the context of this figure.
3. The section on "mapping motion back onto structure" in the methods makes it unclear why the scaling is normalized to 1degree and how that might bias the magnitude of motion observed in Figure 2a (+/- 0.3 A)
We tried running some analysis on the RoPE website but it was either unclear how to go about submitting a job or the website became unresponsive after clicking on "view conformational space". The author can provide a run-through of the website usage with some examples.
It is unclear how important the vagabond refinement performed here is in the clustering. How would figure 1a, b look, for example, if the PDB or PDB-REDO models were subjected to ROPE without further refinement?
At the end of the SVD, it should be possible to project the contributions for each SV back onto the torsion angles most responsible for the differences. It would be interesting to plot that for BPTI and lysozyme to identify the torsions/areas leading to the greatest differences across temperatures.

Minor points:

There are some gray colored points in Figure 1(a) and 1(b) which are not accompanied by a legend and their significance not explained.
To highlight the advantage of RoPE space, the author can show clustering of the same protein chains when clustered based on RMSD. The crowding of points when using RMSD vs. the separation of points when using torsion angles can make the utility of RoPE space obvious to the reader.

Reviewed by Ashraya Ravikumar and James Fraser, UCSF

Competing interests

The author declares that they have no competing interests.

Read the original source

Torsion angles to map and visualize the conformational space of a protein

This article has been Reviewed by the following groups

Listed in

Abstract

Article activity feed