I am working on the structural analysis of a membrane protein of about 150 kDa in size, in complex with a drug.
So far, I have obtained a structure with a resolution of approximately 3.3 Å, but I have not yet been able to identify the drug-binding site. I am using GDN as the detergent and Quantifoil 300 mesh Cu grids (R1.2/1.3) for the grids.
Based on in vitro assays, I have narrowed down the potential drug-binding region to some extent, but when looking at the average images from 2D classification, this region appears quite blurry. I suspect that the drug binding might cause some dynamics, preventing the identification of the drug-binding site.
If I can stabilize this region, I believe structural identification will become possible. What methods might be considered for stabilizing this region?
Yes, I created a mask near the candidate drug binding sites and performed 3D classification. I set the O-EM batch size to 2000 and experimented with multiple conditions for filter resolution and class similarity, but it didn’t work as expected.
Well yes, based on another post it seems that 3D classification will only work with very good resolution (on the 2 angstroms)… unless you have more significant structural changes, like a change in shape or domain orientation. But just like you did, I’d also have tried anyways.
Now: do you mean the same region in the apo protein is not blurry? Or you don’t have that structure?
If the blurriness is caused by a mix of holo and apo forms in the sample: try increasing the concentration of compound to > 20X the Kd - more if possible.
If you don’t have other ligands to add to the sauce - cofactors, antibodies or protein partners - you can consider going back to chemistry and trying other compounds in the same chemical space.
Crosslinkers are also an option, but beware of what you can tell in your publication if you succeed - they will also cause artefacts.
Blurriness has been observed even in the apo state, so I believe it is not caused by the holo form or apo form. It is likely that the blurriness has been further intensified by the addition of the compound.
I had not thought of a crosslinker. Crosslinkers are used to prevent the dissociation of complexes, but to my knowledge, I have not seen any examples where they are used to stabilize dynamic parts. Are you aware of any literature on this?
Of course it’ll depend on how it moves, and sorry I don’t recall any specific publication… Just very deep in my brain there is something about people using Grafix for those cases - by the time Grafix came out, a long time ago. Our lab is precisely interested in flexibility, so we avoid cross-linkers as much as we can.