[ccp4bb] Molecular Replacement of a protein complex

Main steps:

Special:

Basic tutorials:

Xtal community:

( 17) 0.889 0.900 00l: l=2n (zone 3)

( 18) 0.889 0.900 h00: h=2n, 00l:
l=2n (zones 1,3)
..........

( 16) 0.057 0.057

( 17) 0.057 0.057 h00: h=2n (zone 1)
..........

( 18) 0.039 0.040 0k0: k=2n, 00l:
l=2n (zones 2,3)

( 19) 0.039 0.040 h00: h=2n, 0k0:
k=2n, 00l: l=2n (zones 1,2,3)

I reindexed to both space groups and proceeded with Mathews Coeff. Assuming
a 1:1 complex had formed (MW = 39000) it predicts 1 complex per ASU. MolRep
using the larger protein (MW=27000) found the protein with a contrast of
3.33 (p2221) and 2.61 (p21221). However repeating MolRep for the smaller
protein in both spacegroups failed to locate it. Contrast was low, 1.11
(p2221) and 1.10 (p21221).

I reran Mathews Coeff in case the second protein was not their (though gels
of crystals show both proteins) and it predicted 2 proteins per ASU. MolRep
looking for 2 of the larger protein generated contrasts of 1.99 (p2221) and
2.27 (p21221).

Using Refmac5, I performed a rigid body refinement of all 4 cases with
details summarised below

p2221 1proteinper ASU - Proteins are arranged in layers separated by a
55Angstrom space. Some electron density is present but nothing that could be
generate a polypeptide. Perhaps two or three amino acids here and there.
Density fits the protein well. Rfac, 0.527, Rfree 0.523

p21221 1proteinper ASU - Proteins are again arranged in layers separated by
a 55Angstrom space. Again some electron density is present that could
accommodate two or three amino acids. Density also fits the protein well.
Rfac, 0.527, Rfree 0.512

p2221 2proteinper ASU - No layers are observed. Density fits the proteins
very poorly. Rfac, 0.541, Rfree 0.546

p21221 2proteinper ASU - No layers are observed. Density fits the proteins
very poorly. Rfac, 0.540, Rfree 0.547

My questions are this; In general what is the best procedure for molecular
replacement and refinement of protein-protein complex where the Kd is 1uM
and weaker and it is difficult to observe one of the proteins; In this case
I believe the second protein is their, though either the starting model is
poor (80% identical) or a gross conformational change of the protein has
taken place (which is unlikely based upon similar complexes that have been
solved). I am also concerned about Rfree being smaller than Rfac. Should I
continue with these two data set and try modeling alanine in the electron
density I can observe and see if further density is generated on further
refinement? Or should I go back and look at alternative space groups?

Thanks in advance for you opinions, suggestions and help

Daniel A. Bonsor,
Boston Biomedical Research Institute,
64 Grove Street,
Watertown,
MA 02472 USA