[phenixbb] 3Fe4S clusters in real-space refinement

[Edward A. Berry]

Hi Rhys,
At some point the atom positions of F3S were changed, or maybe standardized, to match those of the SF4 Fe4S4 cluster. That is, if you take SF4 and remove Fe2 then you have F3S. This is good but it complicates refining some old structures, and using them to solve new structures.
If you get the modern F3S (Fetch Monomer in coot) and orient it with S3 toward you, the other three S atoms S1, S2, S4 make a triangle going counterclockwise, and the Fe atoms FE1, FE3, FE4 make a smaller triangle going clockwise. If you load an older entry like 2wdq or 1yq3 both triangles go in the opposite direction.

You can fix this easily in coot, although you will have to do each cluster separately. With the tool suggested by Nigel it appears they can all be fixed at once (if you are familiar with using phil options).
Load the pdb you posted into coot and center on S3 of X 402 (for example) Now do regularize zone" on just residue X402 to get something like the right shape for the ligand.  rotate the S3 atom toward you, and you will have something like the screenshot below. Notice the Fe atoms go counterclockwise like in the old pdbs - presumably coot optimizes bond angles starting from what you gave it, not worrying about chirality.

There is a problem: the links to Cys 242 and 263 are crossed. There are two ways to fix this, and the density should tell which is right.
(1) rename atoms Fe1 and Fe4 (swap the names) and don't change the link records. You should also swap S1 and S4 to keep the connectivity, although  that may fix itself during refinement. In addition to uncrossing the links, this gives you the chirality of the new model.
(2) rotate the F3S residue 180* abuot the vertical. This will uncross the links, but it will also put S3 behind the triangles. The density should say if this is right. This will still leave you with the old-style cluster and may get inverted in refinement, so if this second solution is correct you still need to swap Fe1/Fe4 and S1/S4; AND redefine the links for the new atom names. (NOT the pdb LINK records, but the "geometry_restraints.edits" or whatever the gui uses to define links. Or maybe auto-linking will take care of it, if the starting model is approximately right.

Or, comparing the coot-regularized model with what you posted, it looks like you had Fe1 and Fe2 correct and S1 and S4 were wrong, coot moved/switched Fe1 and Fe4 to correct the connectivity but got the wrong chirality. Going back to your starting model you may only need to switch S1 and S4.

Hope that helps,
or, if you've already got it fixed, that it may help someone else with the same problem.
Ed
-------------- next part --------------
A non-text attachment was scrubbed...
Name: F3S.png
Type: image/png
Size: 53564 bytes
Desc: not available
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20220427/61ba9c7c/attachment-0001.png>