Local Refinement: Advice on Large Rigid Protein with Small Flexible Area

Hi all,

I’m working on reconstructing a very large, mostly rigid protein with D4 symmetry that contains small areas of flexibility. I’ve been able to achieve 2.37 Angstrom resolution overall, but the flexible areas (my regions of interest) have closer to 4 Angstrom resolution. Because of the protein’s symmetry, the flexible areas are repeated 8 times. I am mostly interested in a chain of only about 10 residues, but I know that this is too small to be reliably refined using Local Refinement. Therefore, I’ve been trying to mask out larger areas that contain the flexible region to limited success. If anyone could provide any help, that would be great! Here are my main questions:

  1. Is Local Refinement the right job?
  2. Is the 10-residue chain I’m trying to resolve too small for any reliable refinement?
  3. What are best practices for ensuring my mask is masking the correct region I want to refine? (I have performed Local Refinement with and without Particle Subtraction).
  4. Is there any way I can take advantage of the molecule’s symmetry to refine the region of interest?

Thank you for your help!

-Nick

1 Like

Local refinement is the right job and where to put the mask is something to optimize. You could perform D4 symmetry expansion from a D4 homo/NU refinement, run 3DVA with a global mask and inspect the movement. How is the region around your 10 residues moving? Put a mask on the domain(s) that move synchronously - this can become your focus mask for local refinement. Consider putting this mask on a smaller subregion but probably don’t make it smaller than 30-50kda worth of protein (try different types of mask). Perform local refinement using the symmetry expanded particles and the focus mask on domain(s) around your region of interest.
Also perform 3DVA using this focus mask to better understand how your region of interest is moving.
Use this focus mask for focused 3D classification to sort the heterogeneity, then put good classes to local or homogeneous/NU refinement.
Repeat 3DVA/3D classification with the focus mask on subsets to further inspect residual heterogeneity. It should be reduced in “good” subsets.

Other option is a multi-body refinement type of strategy where you signal subtract all regions other than the region to be local refined. I am not sure how effective this is compared to the local refinement strategy, but I think local focused refinement is pretty good, even to remove highly heterogeneous things like a micelle or nanodisc, where signal subtraction used to be a preferred option for difficult datasets.

1 Like