Tobias,
we had a somewhat similar case ("Structural Insights into Nonspecific Binding of DNA by TrmBL2, an Archaeal Chromatin Protein", http://dx.doi.org/10.1016/j.jmb.2015.08.012 ) which gave us trouble in modelling. We tried a number of possibilities and settled for the most convincing solution.
In your case, I don't quite understand what the evidence is to change the spacgroup to lower symmetry than I422 ? Why not I422, and different binding modes? Have you tried I4 and other subgroups, and does the problem go away when you have twice the number of molecules in the ASU?
best,
Kay
On Tue, 12 Apr 2016 04:05:56 +0000, Bock, Tobias <[log in to unmask]> wrote:
> Dear colleagues,
> we have recently crystallized a homodimeric protein in complex with a palindromic 16-bp DNA fragment. The spacegroup is I422 and data collection statistics look completely > unsuspicious. We could solve the structure by Molecular Replacement and find one DNA duplex and two homodimers of our protein in the asymmetric unit. Both homodimers > bind to the same DNA molecule, but seemingly at different positions (i.e. one to the palindrome as expected, the other one shifted and rotated by 4 bp such that is does not
> overlap with the “correct” homodimer).
> And here is where our trouble starts: we find additional electron density for 4 bp at one end of our DNA duplex. If we use this to position our DNA-16-mer, the second protein
> homodimer also is in register with the palindrome. We now wonder: is there an indexing problem, despite the fact that the data statistics look fine such that the symmetry is > lower and the ASU contains two separate copies of the protein/DNA-complex, which have somehow been folded on top of each other in I422? Or do we look at partial
> occupancies of both DNA and protein? Needless to say we could not get this to work in phenix.refine and could also do with some help here…
> I am attaching figures of the structure(s) to this message.
> Thank you in advance for your help,
> Tobias
|