Hello,
On Wed, Feb 24, 2010 at 9:55 AM, Simon Becker
<simon.becker.uni-konstanz.de> wrote:
> Hi all,
>
> I was asked to assisst in a protein crystallography problem. Apparently a
> protein would crystallise with his binding partner but not without. So I was
> asked wether I could design a few peptides that help stabilisz the Protein
> and could be used for co-crystallisation.
> We figured that the loss of the partner could lead to an increse in
> flexibility in the binding area, which then prevents crystal formation. The
> first logical step was then to start with fragments of 10 residues of the
> partner that were in vdw range of the protein and evaluate their binding
> energy. Pretty fast I identified a number of peptides that bound tight to
> the protein. Not a surprise really, as I just cut them out of a pdb file and
> the peptides where already in binding conformation.
> Two questions arise however:
>
> A) Is this the only place the peptide would bind? Can alternative sites be
> found?
>
> B) Will the peptide really lead to a decrease in flexibility?
>
> Question B can be answered quickly. Just have a long run, to allow for loop
> movement, with the protein alone and in complex and have ptraj calculate
> flucuations. If a significant decrease can be observed in the binding area:
> success.
> For A it is different. I could place many peptides in close proximity to the
> protein's surface (not just the binding site) and see what happends during
The problem with this is the fact that the ligand can almost be
guaranteed not to find its proper binding site. Tractable MD
simulation timescales are nowhere near long enough to observe this
kind of behavior without applying biassing potentials to force the
ligand to where it needs to be. Unfortunately, doing this requires
previous knowledge of where the ligand binds :).
> MD. The use of a docking program was also suggested. Are there programs
> capable of flexible docking peptides to whole protein?
This is probably your best bet, though if you have *no* idea where the
ligand binds (i.e. you want to see if it binds *somewhere* else,
anywhere else), you may find these programs to be of limited utility.
Docking programs are usually decent when it comes to finding a binding
conformation when the location of the binding pocket itself is fairly
well-known (I think some docking programs might be decent at narrowing
down the location, but I think you still need a general region). The
reason for this is that you'd need to apply a coarse grid if you
wanted to cover the entire protein (and not take forever on the
calculations), and results for coarse grids are substantially poorer
(since not as many combinations are tried). Docking programs are also
poor, I believe, if the binding pocket changes topology substantially,
since it does not take this into account.
Long story short, this would be a great place to be able to get to,
but I don't think the field is quite there yet... Others may share
their ideas as well.
Good luck!
Jason
--
---------------------------------------
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Feb 24 2010 - 09:30:04 PST
