Hi everyone,
I have a situation that I'd love to get some experienced opinions
about. I recently solved structures of two single point mutants of a
multi-domain protein by molecular replacement, which crystallized in the
same crystal form (crystal form #1). I subsequently was able to grow
crystals of the native protein (without mutations) under the same
conditions by streak seeding with one of the mutants, but
(interestingly) the native crystals belonged to a different space
group. The native structure was then also solved by MR. One domain
(~50 residues out of 450) has significantly higher B-factors in the two
mutant structures (70 compared with average of 30 in one mutant, 70
compared with average of 25 in the other), but has continuous density
and could be modeled without too much difficulty. In the native
structure, which has the same conformation and superimposes very well
with the mutants, this 50-residue domain has very poor, broken density
and the B-factors for the domain refine to 120 compared with average of
~35 when the domain is included in refinement. If I omit the domain, I
clearly see some positive difference density (although very broken) that
matches well with the location of that domain (as it was built in the
mutant structures). Refinement with the domain omitted increases R by
0.7% and decreases Rfree by 0.2% relative to refinement with the domain
included.
I don't want to deposit a structure containing this domain that is not
strongly supported by my data, but I know that a scientist combing the
PDB for this protein will download this structure and not those of the
mutants because it is the native sequence. My feeling is to omit this
domain in the native model and try to make it clear in the paper that
the mutant structures are nearly identical and include the domain. Any
thoughts?
Thanks,
ERIC
|