Page 110 - Macromolecular Crystallography

MOLECULAR REPLACEMENT TECHNIQUES 99

Protocol 7.1 Checking your data

• Check for completeness and redundancy at the by calculating a native Patterson function and
desired resolution: usually a complete data set at sorting its peaks;
10–4 Ångstroms is fine. – check for two-fold axes, three-fold axes... using
• Check for possible twinning; read carefully the output of self-rotation function;
TRUNCATE in CCP4 (Collaborative Computational Project, – calculate the percentage of solvent in the crystal, for
1994) or submit your data to the scrutiny of Todd Yeates different hypothesis as to the number nmol of molecules
web site (http://www.doe-mbi.ucla.edu/Services/Twinning).
in the asymmetric unit:
• Check your space group (!):
– for instance, if P2(1)2(1)2(1), check extinctions 100 ∗ (1. − nmol ∗ (MW ∗ Vm)/Va.u.)
carefully;
– for a polar space group, remember to try both where MW is the molecular weight of a monomer,Vm is
3
enantiomorphic possibilities in the translation search, the density of a protein (∼0.73 cm /g), and Va.u. is the
because they cannot be distinguished through Patterson volume of the asymmetric unit. The percentage of solvent
methods used in the rotation function: P4(1) and P4(3), should be in the range 20–80%.
P6(1), and P6(5). Note: It might be of interest to read in detail a recent ‘tour
• Check for possible non-crystallographic symmetry de force’ success story using MR in a very difficult case, with
(NCS): both high NCS (12 copies in the asymmetric unit) and
– check for a possible pure translational twinning of the data (as discovered quite late in the process
non-crystallographic symmetry (pseudo-symmetry) of structure solution) (Lee et al., 2003).

There are, essentially, two issues in the application 7.2 Test data used throughout this study
of MR automatic protocols to structural genomics: with the different MR packages
1. Can one push the limits of the method so as to
The same data set at 2.5 Ångstrom resolution, col-
use models with less and less sequence identity with
lected at the European Synchrotron Research Facility
the target, i.e. models obtained by threading meth-
(ID14-EH2), for a T. brucei 6-phosphoglucono-
ods, with sequence identity levels between 15 and
lactonase (6PGL) (Delarue et al., 2007) was used
25%? The answer seems to be yes (Jones, 2001), with
throughout this study. The cell parameters are 70.3,
newer methods capable of solving problems with
80.8, 90.3 in P2(1)2(1)2(1). There are two molecules
sequence identity around 20% (Claude et al., 2004;
in the asymmetric unit, related by a pseudotrans-
G. Labesse, personal communication; Abstract in the
lation, discovered by sorting the native Patterson
GTBio Meeting, June 2004, Lyon, France; Keegan peaks (Protocol 7.1). There are two possible models:
and Wynn, 2007). 6PGL from T. maritima (1PBT) (about 40% sequence
2. How far should one pursue efforts to solve the
identity) and the glucosamine-6-phosphate deami-
molecular replacement problem once the automatic
nase (1DEA), which was detected by BLAST (25%
protocols have failed; in other words, how much
sequence identity).
time should one spend in trying to solve difficult
cases before deciding to go back to the bench and
grow SeMet crystals, or use the SAD method with 7.3 The standard molecular replacement
crystals soaked with one anomalous scatterer? This method
is actually a difficult question that depends on a
7.3.1 Historical background: Patterson
lot of different issues, such as the expertise already
methods
present in the lab. in MR techniques, the solubility
of the protein(s), which might or might not change The possibility and feasibility of molecular replace-
upon selenomethionylation. This question cannot be ment was demonstrated by Rossmann and col-
answered in general but this is clearly an issue that leagues in the 1960s, as part of an effort to use
one should keep in mind in defining the strategy for non-crystallographic symmetry to solve the phase
structure solution. problem for macromolecules (Rossmann, 1990).