how to generate a ensemble representation using Xplor

[q***@ustc.edu.cn]

Dear Charles:

I am a NMR-based structural biologist from university of science and technology of china. We met before in NMR meetings and I am a friend of Dr. Chun Tang.
I am recently using Xplor-NIH to refine one of our NMR structure based on some global restraints such as RDC and SAXS. Our system is a protein containing two rigid domains connected by a long flexible linker. We detected no NOE between two domains, and we had many evidence that that these two domains don’t have a fixed relative orientation. We had our structure submitted for publication recently and received the suggestion from the reviewer that we should refine our overall structure using RDC and SAXS to generate a ensemble representation. We then collected RDC and SAXS, applied them for ensemble determination using Xplor-NIH. However, due to the RDC restraints, the relative orientation of two domains in the all ensemble structures are basically same, which doesn't seem to reflect a real situation. I am wondering if there are some options in Xplor, which allow us to generate multiple structures, and only their average will meet the RDCs. Your early reply will be highly appreciated.

Best regards,
Qingguo Gong, Ph.D
School of Life Science
University of Science and Technology of China

[Charles Schwieters]

Hello Qingguo--

Post by q***@ustc.edu.cn
I am recently using Xplor-NIH to refine one of our NMR structure
based on some global restraints such as RDC and SAXS. Our system is
a protein containing two rigid domains connected by a long flexible
linker. We detected no NOE between two domains, and we had many
evidence that that these two domains don’t have a fixed relative
orientation. We had our structure submitted for publication recently
and received the suggestion from the reviewer that we should refine
our overall structure using RDC and SAXS to generate a ensemble
representation. We then collected RDC and SAXS, applied them for
ensemble determination using Xplor-NIH. However, due to the RDC
restraints, the relative orientation of two domains in the all
ensemble structures are basically same, which doesn't seem to
reflect a real situation. I am wondering if there are some options
in Xplor, which allow us to generate multiple structures, and only
their average will meet the RDCs. Your early reply will be highly
appreciated.

Have you tried using the ensembleSimulation functionality? Do you have
RDCs in a steric aligning medium (like bicelles)? If you you might see
the example in the eginput/sardc directory of the Xplor-NIH
distribution. More similar to your problem is the work published in

L. Deshmukh, C.D. Schwieters, A. Grishaev, R. Ghirlando,
J.L. Baber, and G.M. Clore, ``Structure and Dynamics of
Full-Length HIV-1 Capsid Protein in Solution,''
J. Am. Chem. Soc. 135, 16133–16147 (2013).

I don't yet have the script for this in the distribution, but if you
think they'd be useful, I'll send you a copy.

This approach does not work so well for RDCs measured in charged
media, as the alignment tensors are simply fit parameters in those
cases. If your domains align separately, then there must be different
alignment tensors for each domain, and there would be no reason for
the two domains to have a constant relative orientation. So I don't
understand that result.

One final point: if the SAXS data fit a single conformation, they will
be of little help in determining an ensemble.

I hope this helps--
Charles