A preview of this full-text is provided by IUCr.
Content available from Journal of Applied Crystallography
This content is subject to copyright.
computer programs
1884 doi:10.1107/S0021889813025387 J. Appl. Cryst. (2013). 46, 1884–1888
Journal of
Applied
Crystallography
ISSN 0021-8898
Received 30 May 2013
Accepted 12 September 2013
#2013 International Union of Crystallography
Printed in Singapore – all rights reserved
Incorporation of a hydration layer in the ‘dummy
atom’ ab initio structural modelling of biological
macromolecules
Alexandros Koutsioubas
a,b
and Javier Pe
´rez
a
*
a
Synchrotron Soleil, Beamline SWING, Saint Aubin BP48, F-91192 Gif sur Yvette Cedex, France, and
b
Ju
¨lich
Centre for Neutron Science (JCNS), Forschungszentrum Ju
¨lich GmbH, Outstation at MLZ, Lichtenbergstrasse 1,
D-85747 Garching, Germany. Correspondence e-mail: perez@synchrotron-soleil.fr
Ab initio algorithms for the restoration of biomacromolecular structure from
small-angle scattering data have gained popularity in the past 15 years. In
particular, ‘dummy atom’ models that require minimal information about the
system under study have been proven capable of recovering the low-resolution
shape of proteins and nucleic acids in many published works. However,
consideration of solvated biological molecules as particles of uniform electron
density contrast relative to the solvent neglects the presence of a hydration layer
around their surface, leading to an overall apparent swelling of the obtained
models and to a large overestimation of the volume of the particle. Here this
problem is addressed by the introduction of an additional type of ‘dummy atom’,
representing the hydration layer. Successful applications of this new approach
are illustrated for several proteins, and related results are compared with those
from the program DAMMIN [Svergun (1999). Biophys. J. 76, 2879–2886].
1. Introduction
Small-angle X-ray scattering (SAXS) of macromolecular systems in
solution is a powerful complement to high-resolution structural
studies by X-ray crystallography and NMR (Putnam et al., 2007; Koch
et al., 2003). Using SAXS, structural hypotheses can be directly tested
against experimental data in solution and conformational changes or
complex formation can be monitored, which helps the understanding
of structure–function relationships. A big advantage of the technique
is related to the fact that measurements can be performed close to
physiological conditions with relatively simple sample preparation
requirements. SAXS is essentially a contrast method where the
scattering signal comes from the difference in the average electron
density, , of solute molecules of interest (0.44 e A
˚
3
for proteins)
and the bulk solvent (0.33 e A
˚
3
for water).
In the early days of the use of SAXS for the study of biomolecules,
it was believed that, owing to the spatially averaged signal and the
associated inherent loss of phase information in a solution scattering
experiment, only the one-dimensional pair-distance distribution PðrÞ
and a small set of parameters related to the scattering invariants
could be deduced. These include the radius of gyration Rg, the
volume Vand the molecule’s surface area S.
With the advent of high-intensity synchrotron radiation sources the
reliability of the data related to biological molecules has been greatly
improved (Pe
´rez & Nishino, 2012). These instrument-wise advances,
coupled with the increasing power of computation, have driven the
development of various structure refinement methods aiming at the
recovery of three-dimensional information, which have had a great
impact on the wider use of SAXS by the structural biology commu-
nity.
Such approaches were pioneered by Svergun & Stuhrmann (1991),
who showed that, by incorporation of a priori constraints concerning
the electron density of the particle, a nonlinear reconstruction
procedure formulated in terms of the spherical harmonics expansion
of the scattering form factors FlmðqÞcan provide reliable low-reso-
lution envelopes, representative of the molecule’s shape. Limitations
of this method related to the description of complicated molecular
shapes containing internal cavities have led to the development of
more general ‘ab initio’ methods based on the representation of the
biological molecules as a set of ‘dummy atoms’ or beads of a certain
electron density contrast, in a finite volume.
This concept, first introduced by Chacon et al. (1998), involves the
implementation of algorithms that rearrange the configuration of
beads in an attempt to minimize the discrepancy between the
experimental curve and the scattering of the model. Several varia-
tions of this method have been proposed to date (Heller et al., 2003;
Walther et al., 2000), of which the simulated annealing algorithm by
Svergun (1999), which gives interconnected and compact final
models, is the most popular, mainly because of its robustness and
speed of calculation (Franke & Svergun, 2009).
A common feature of all ‘dummy-atom’ ab initio modelling algo-
rithms is that the molecule under study is considered as having a
uniform electron density.
1
However, an important characteristic of
solvated biological molecules is that they are surrounded by a thin
low-contrast hydration layer, where water molecules assume on
average a more compact configuration, giving rise to an electron
density that is locally higher than that of the bulk solvent (Svergun
et al., 1998). The resulting electronic density contrast necessarily
contributes to the scattering intensity and has therefore to be
accounted for. When comparisons of the calculated scattering curves
of crystallographic structures versus experimental data are attempted
without consideration of the hydration layer contribution, large
systematic deviations are observed (Svergun et al., 1995). In the case
1
Here we should exclude the multiphase ab initio analysis program MONSA
(Svergun, 1999), which is designed for the treatment of systems containing
domains with different scattering contrasts.
