hard to believe changes in Rfree
Dr. John Badger
jb at MSI.COM
Fri Jul 17 12:31:44 EST 1998
Regarding Na+ in proteins. We looked at this a while back
in cubic insulin crystals. It's quite likely that there will
be some sodiums bound in many proteins. In our case, the hallmark of
sodium binding sites that contain well-ordered sodium was
several backbone carbonyls pointing to the water/sodium atom (and,
after all, this is what ion channels and carriers have).
At 1.7A we also had similer clues that the water was really
sodium (B-factor abnormally low, position refined too close
to ligating groups in the protein). Experimentally we nailed
it by replacing the sodium with other monovalent ions (Li, K,
Rb, Tl) and computing the difference map - you only need low
resolution data to do this.
Regarding the differential in free R of 0.2%:
If you use the equation on page 161 of Blundell and Johnson
to estimate the average change in intensity (i.e. assume
that water>Na adds about 1/5 of an atom to the 900 atom model) you
get intensity changes of about 2%. How this will translate
into F's and differences in free-R isn't so clear but the
effect might not be completely negligable.
I don't think it is at all impossible that the small change
in free R is due to the change in scattering factor and
the structural changes that result from the water>sodium swap.
Another calculation you could do is the simple point replacement
of water with sodium to see how big a change you get just
from change in scattering factor.
Having said all that, I would also be relying on stereochemical
arguments to decide what to do.
Product Development Manager, Macromolecular Crystallography
More information about the Xtal-log