Bob,
It seems you have got the correct solution or very close to. Though,
something you described are confusing.
1) Normally for normal protein crystals, you may not get fitting well
in both P222 and P212121 space groups, even if
you have pseudo-translational NCS. Based on your Rs, it seems you
have got the correct solution.
2) If SG is P212121, then it does not normally refine well in P2; it
should be P21. P21 is the maximal subgroup of P212121.
Though again, pseudo-translational NCS may confuse this.
3) Difference between Rs for two refinements may be from:
a) P2 may not be the correct S.G.
b) You might have just got a half content solved and the other half
is left in the layer you described. If the P222s were
the "correct" lattice and you index them into P2/P21, then the
asymmetric unit of P2/P21 is doubled as that of P222s. And,
you need solve doubled copies of molecules.
4) I would stick to P212121/P222 first, and use information from data
statistics, systematic absence, MR solution LLG and
contrast, etc. for a reasonable judgement.
Lijun
On Mar 15, 2010, at 7:12 AM, Bob lainer wrote:
Dear ccp4bbers,
I need some suggestions regarding structure solution of fused
protein complex.
After repeatedly failing to obtain crystals of a 45 kDa protein
(mostly a-helical) we fused the protein of interest to another
protein of known structure which is 30 kDa is size (mostly b-
strands). Many fusions were made and we selected a fusion-protein
(Ca. 75 kDa) that has a minimal linker between the two proteins for
crystallization trials. Both partners in the fusion are active and
function as expected of wild-type proteins.
Crystals of the fusion protein were obtained and couple of datasets
was collected. Running crystals of the fusion protein after X-ray
data collection on SDS-PAGE shows that we have a band that is
running approx. 60 kDa instead of 75 kDa indicating some form of
proteolytic processing during crystallization (there are no bands
that are equal to the 30 kDa fusion partner of known structure). One
of the crystals diffracted to 2.5Å resolution. We could index the
data in P2 (as well as P222 and P212121) space group, the highest
resolution shell was cut off at 2.5 Å using I/σ = 2.0 as a guide.
The overall Rmerge of 15% and completeness of 99%. The Matthews
coeff. is 2.32 with 43% of solvent content as calculated with the
molecular wt. of the fused protein complex (75kDa). We used Phaser
(AutoMR, Phenix) and Molrep and Amore (CCP4) for molecular
replacement using the known structure as the search model (the 30
kDa protein). In P222 and P212121, the statistics after one round of
refinement post-MR statistics are good (R-factor and R-Free of 21%
and 24%, respectively) and the known protein molecule fits and packs
well with no extra density for the protein of interest. In contrast,
in P2, we can see the extra density between planes formed by the
known structural model. The spacing between these planes is about
30Å. The statistics of one round of refinement post MR is very poor
(R-factor and R-Free of 41% and 47%, respectively).
The questions we are interested in are:-
1) How do we determine the correct MR solution (P2 or P212121)?
2) How should we proceed further in case like this?
Thanks in advance for the suggestions.
Bob
Lijun Liu
Cardiovascular Research Institute
University of California, San Francisco
1700 4th Street, Box 2532
San Francisco, CA 94158
Phone: (415)514-2836
