[ccp4bb]: Summary for crystal packing

[Lan Xu <xulan@wistar.upenn.edu>]

Title:

Hi, there,
I really appreciated all your suggestions and comments, thank you so much.
 
I would like to summarize all the answers I received so far.  It is kind
of long, thanks for your patience.
 
My question was:

I have solved the structure of a receptor-binding protein in two  different
space groups. Form I is P2221, Form II is C2221, both have one  molecule
in an asymmetric unit. Recently, we used molecular replacement  to solve
the structure for the third crystal form (Form III). It belongs  to C2221,
but has two molecules in one asymmetric unit. 

When we examine the crystal packing of these three crystal forms, the  same
"dimer" is present in all three crystal forms. However, the protein is shown
to be a monomer in solution by rate zonal centrifugation, gel filtration,
cross-linking, sedimentation equilibrium and low angle X-ray scattering. 
Dynamic light scattering measurements indicated a monodisperse solution with
an apparent molecular weight of 79 kDa, which is in reasonable agreement
with that for the monomer (63.9 kDa).  The protein is rather long and thin,
146 x 46 x 34 Å.

A buried surface per subunit in the dimer is ~1206 Å2, which represents  ~5.5%
of the surface area of the monomer.  The contact area is separated to four
patches by cavities.

Native protein is in 20 mM Tris-HCL, pH 8.5, and 50 mM sodium chloride.  The 
crystal conditions for these three crystal forms are somewhat similiar.   
Form I crystals were crystallized in  30% MPD, 0.1 M sodium acetate at pH 
4.6, 20 mM CaCl2.  Form II was in 50% MPD, 0.1
M sodium acetate at pH 5.0, 10 mM CaCl2.  Form
III:  63% MPD, 0.19 M sodium acetate at pH 4.53, 10 mM CaCl2, 
with 0.1-0.2% final concentration of glutaraldehyde.  
 
How do we distinguish between a biological dimer or a  dimer because of crystal
packing? Whether a buried surface area of only  5.5% of the total surface
area is stable for a dimeric interaction? Is  there any reference that I
can refer to? 

Dr. Timm Maier (tmaier@chemie.fu-berlin.de):
In my opinion, your different crystal forms as well as the size of buried
surface are good indicators for a dimer in solution. However, you might
need more criteria:
on the one hand, there are some more characteristics of the potential
dimer interface you might check: residue conservation in related proteins,
hydrophobicity, surface complementarity ... , but on the other hand a
physical measurement of the oligomerization state in solution might be
even more interesting: You might try classical lab methods such as
size exclusion chromatography or analytical ultracentrifugation, but most
interesting and exact is probably measuring small angle X-ray scattering
(SAXS) of your protein solution. Sample requirements are pretty close to
what you probably need for crystallization and the determination of the
oligomerisation state should be straight forward with this method and is
less error prone then the indirect measures from sedimentation behavior
or gel filtration
Dr. Brent W. Segelke (segelke1@llnl.gov) :

It is very difficult to convince reviewers that your
oligomer is meaningful based solely on the crystallography results. I
worked on n.n.n Pla2 as a graduate student. We found a trimer in each of
3 very different crystals forms (tetrahedral, orthorhombic, and cubic).
There was ~20% of the surface area buried and still reviewers wanted an
independent verification that the oligomer was "real". If I recall
correctly, there was some rule of thumb that there should be >10% of the
surface area buried, but I don't have a reference.

Dr. Hannes Fishcher (hannes@if.usp.br):
We had the same problem and solved it with SAXS, in this article:
J Mol Biol 2001 Nov 9;313(5):1059-72
we describe how we solved the problem you had.

Dr. Rachelle Gaudet (gaudet@mcb.harvard.edu):
One way that I have used, to distinguish a functional
interaction from a crystal packing interaction, is to design a mutation
that should disrupt the interaction and test the protein using
functional assays.  For example:
B. E. Gewurz*, R. Gaudet*, D. Tortorella, E. W. Wang, H. L. Ploegh and
D. C. Wiley (2001) Antigen presentation subverted: Structure of the
human cytomegalovirus protein US2 bound to the class I molecule HLA-A2.
Proc. Natl Acad. Sci. USA 98, 6794-6799.

For another example, you can also look at Matsumoto, N, et al (2001) J.
Exp. Med. 193, 147-158
Dr. Alfred Antson (fred@ysbl.york.ac.uk):
it looks like your dimer forms only at high protein concentration.
It might still be biological, if "local" concentration of monomers in vivo
is high enough, for example if two monomers bind with their
non-dimerisation domains to a polymeric molecule like DNA
bringing two dimerisation surfaces close to each other.
~1206Å2 seems to be a reasonable surface area.
Reference on protein-protein interacting surfaces: S.Jones&
JM Thornton PNAS 1996, 93:13-20.

Dr. Louis Renault (renault@lebs.cnrs-qif.fr)

Could it be that your dimer was hidden in solution because of a too low affinity
(requiring anchoring on membrane in cells
for ex.) and that you used too low protein concentration/quantity when you
tested your oligomerization state in solution, for ex. in gel filtration?
or
Did you test your oligomerization state in solution by gel filtration and
other methods
close to the crystallization conditions (pH,salt+MPD) or only at physiological
conditions (pH 7, no MPD,...) ?
Your crystallization conditions at pH 4-5 + MPD could favor a dimer whereas
pH 7 without MPD
could rather favor monomer. Then what are the biological conditions 
that your protein encounters...

Dr. Shee Mei (sheemei@imcb.a-star.edu.sg):
Whether the crystal structure are true representatives 
of the protein in their physiological condition is debatable. This is because 
the protein is at their lowest energy (meaning static which does not happen 
in nature) and the most crystallization conditions are far from 
physiological.
Dr Andrew M. Gulick (gulick@hwi.buffalo.edu):
I was just wondering if you did your gel filtration, DLS, etc...
experiments at pH 5.0.  Is it possible that your protein
is a monomer at pH 8.5 yet forms dimers at the lower pH
where you observe all three crystal forms.

I had a similar case where I thought that addition of
substrates was causing a monomer to dimer transition
which turned out not to be the case when I did a gel
filtration experiment when the column was equilibrated
in substrate.
Dr. Deena Oren (oren@cabm.rutgers.edu)

You could try cross-linking with Glutaraldehyde in dilute solutions. You
need enough to visualize the MW on a SDS-PAGE gel. Crosslinking conditions
have been discussed on this bulletin board before. I recommend 0.1%
glutaraldehyde in various concentrations of your protein in question. It
must be dilute enough that the cross-linking will occur only within an
oligomer and not between them. This should be tested under physiological
conditions.

Thank you very much.

Have a good day.

Lan Xu