Pseudo-symmetric complex reconstruction


I am trying to solve a complex structure of a protein complex. Protein A is an octamer and the total MW is around 800 kDa. Protein B is a hexamer ATPase (~180 kDa in total) and binds to protein A when activated by ATP.

I collected a couple of datasets using different strategies to reconstruct the complex but got confused about where I did wrong…The target is to get like different intermediate states during ATPase turnover, but till now, only one dataset gave me resonable reconstruction.

In one of the datasets, I got like 30k particles in final reconstruction out of 1 million particles (6k micrographs). Some of the typical 2D classes after final refinement↓

In another dataset, 2D classification in a mask diameter of the whole complex seems to be unable to distinguish A+B from just adjacent A+A (unfortunately, particles seem to be clustered even I tried lower protein conc. ). So I just excluded trash classes in the mask diameter of A and move onto hetero-refine (like what I did in the first dataset). But, in hetero-refine in C1, the refinement seems to be centered on A and only smeared density around it was observed. I am not sure, but is this something driven by the symmetry of A and low population of complex particles?

I found several similar topics here, and tried like symmetry expansion→3D classification w/o alignment in cryoSPARC, but it didn’t work… I’m considering to try with Relion symmetry relaxation though.

Do you have any advice on case like this?
Many thanks!

smeared density in the second dataset

This seems to be a great test cast for 3D-Flex or cryo-DRGN if the conformational heterogeneity cannot be resolved by muti-class Ab-initio or 3D Calss.

I don’t think C2 expansion is helpful for this complex. You should treat as C1.

You should generate a generous mask around the flexible subunit and use local refinement to get high resolution, it looks like a good candidate where both sides sides of the complex can be solved to high resolution. You may not be able to resolve the whole complex as the two major parts might move independently.

After aligning particles to flexible part of your complex, use 3D variability (with the mask) to check for discrete conformations and then cluster, another way is 3D classification with various filtered resolutions. Both methods will be sufficient at separating and at least identifying if there is variability in your complex. Then refine each class with the mask and also without the mask, one side of your complex may dominate alignments and smear out the conformation each time, but you need to test this and see.