Dear Naveed,
If your final model fits the final 1.7 Å electron maps density extremely well, without residual positive or negative difference density peaks, it almost certainly correctly describes your data. In the case Dale sent us, the fit was very bad. That the initial density was not that great happens more often, especially if the binding site reorganizes upon ligand binding.
However, depending on the impact factor of the journal you want to publish in, you may face some questions from the referees concerning the biological relevance of your findings. Especially the fact that adding your inhibitor would break the covalent bond between protein and cofactor. Here you would probably need additional biochemical evidence (gelfitration, gelelectrophoresis, masspec, other?) to prove that the same happens on biochemically relevant timescales.
Best, Herman.
PS: if you correctly label your Alt-conformations, Refmac will refine it correctly.
-----Original Message-----
From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Naveed A Nadvi
Sent: Wednesday, April 25, 2012 5:25 AM
To: [log in to unmask]
Subject: Re: [ccp4bb] Criteria for Ligand fitting
Dear Crystallographers,
Thank you all for responding! I will try to respond to the suggestions collectively. I did however have some questions for some of these suggestions...
@Herman Schreuder: I have performed refinement with alt conform as you suggested. I am not sure how to do group-occupancy of the inhibitor. Is that possible in Refmac?
@Colin Groom: Your suggestion for using maps from different dataset is very interesting and I am excited to chase this up! But I am not very competent with the exact terms and this procedure in general. Should I do this using FFT? I have the usual master MTZ from SCALA but not the other labels yet. How can I create them? I would appreciate if you could suggest to me which CCP4 MODULES (and their documentation) I should look at!
@David Mueller: apologies I wasn't sure what details of the hydrophobic crown you asked from me... There were about 7 hydrophobic residues with Trp, Tyr and Phe.
Regarding the covalent modification, apo crystals are present with a protein residue chemically bonded to the co-factor. Our inhibitor (NOT derived from substrate) was expected to displace the protein residue and chemically bind itself to the co-factor. We had observed this somewhat in the density map as a proof of inhibitor binding. There was a significant red density (COOT) where the residue and co-factor forms the bond. However, when trying to bond the co-factor and the inhibitor, that too was unfavourable. So we have left the three components un-bonded at the moment.
I have tried alternate conformations for all the three components (protein residue, co-factor, inhibitor). At the moment I am playing with 50% occupancy to keep things simple. I defined the 'A' conformers when inhibitor is present and 'B' for the apo structure. (Question1: Can Refmac refine separately based on these two possibilities?). After incorporating the inhibitor in the structure and refinement of all components at 50% occupancy, we have not seen any negative density. The 2Fo-Fc (blue) map fits extremely well after refinement. But I am a little paranoid if it is due to phase bias introduced from the current model with inhibitor. Q2: How can I deduce the percentage occupancy of alt conformers? Should I look at the bfactors as a guide somehow?
I am ultimately concerned whether it is acceptable to publish this enzyme-inhibitor complex given the lower occupancy and poor initial density. We have tried a large number of experiments to optimize co-crystallization by varying all common strategies: seeding, soaking, co-crystallizing and crystallization of inhibited protein. The complex crystals were finally obtained from co-crystallization drops that were seeded with apo crystal nuclei. So these crystals were more like co-crystallization rather than pure soaking. The final inhibitor concentration in the co-crystallization drop was >500-fold more concentrated than the IC50. The cryo buffer also contained same concentration of inh as mother liquor.
Thank you again for your comments (I love CCP4BB)!
Naveed
|