Page 191 - Macromolecular Crystallography

180 MACROMOLECULAR CRYS TALLOGRAPHY

separator is used to ensure that only cold liquid observed reflections, including the intensity of spots
strikes the sample. Washing with liquid nitrogen is (I/σ), their shape, the percentage of the detected
used routinely at SGX-CAT during automated data spots that satisfy the indexing process and, once
collection, because the presence or absence of ice indexed, how well the spots refine within the chosen
cannot be known in advance. The process is per- point/space group. Such analyses yields a numer-
formed in parallel with other manipulations of the ical score that permits direct comparison between
sample, so that the time added to the mounting and crystals within a set. The estimated diffraction res-
positioning of a protein crystal is less than 30 sec. olution of the crystal is also determined, from
Constant use of this liquid nitrogen delivery sys- which the optimal sample-to-detector distance is
tem requires control of the environment within the calculated.
X-ray enclosure to eliminate condensation of water Software at SGX-CAT readily distinguishes crys-
on the phase separator located above the crystal tals of good quality from those that diffract poorly
sample. If this water were to reach the sample, the or display artefacts which compromise the data.
crystal would experience a temporary increase in Somewhat surprisingly, we have found that the use
temperature that can result in loss of crystallinity. of automatic scoring also improves the selection of
Dehumidification of the hutch eliminates this poten- crystals for data collection. Figure 12.4a presents an
tial problem. Removal of ambient humidity has the example of a crystal that most members of the struc-
added benefit of reducing the formation of ice on tural biology community would reject for data col-
all components that use liquid nitrogen, particularly lection. However, our evaluation system suggested
the dewars used to store crystals before and after that this crystal was likely to yield a useable dataset.
X-ray analysis. The electron density map of the active site, auto-
matically generated from diffraction data recorded
from the same crystal, is shown in Fig. 12.4b. Utiliza-
12.4 Evaluation of crystal quality
tion of the scoring system has significantly improved
A fully automated protein crystallography beam- data collection efficiency at SGX, because the entire
line at a third-generation synchrotron source can process can be conducted without staff oversight. In
screen several hundred crystals daily. Automatic addition, crystals that previously would have been
evaluation of the diffraction images to ascertain rejected upon visual inspection are routinely used
crystal quality is, therefore, a critical step for high- for structural studies of protein–ligand complexes.
throughput data acquisition. Evaluation of each This benefit of our system has the added advantage
image requires software that mimics the traditional of reducing the amount of time devoted to crystal-
visual assessment of crystal quality. lization, thereby improving throughput in structure-
Automaticdiffractionanalysisisavailablethrough based drug discovery. Our scoring system also mon-
the Diffraction Image Screening Tool and Library itors the number of overloaded reflections, which
(DISTL) (Zhang et al., 2006). DISTL incorporates permits identification of samples for which X-ray
automatic indexing and identification of extraneous beam attenuation and/or exposure time reduction
features such as ice rings, and provides an estimate may be required.
of the resolution of the diffraction data. Neural net- Within our crystal quality evaluation process, data
works have also been tested for the evaluation of collection strategy is determined. Because of the
crystal quality (Berntson et al., 2003). extended times for data collection with in-house
Given the nascent nature of these software tools, and earlier-generation synchrotron sources, opti-
SGX developed its own method to evaluate or mization of data acquisition remains a significant
score the diffraction quality. The SGX system consideration (Chapter 5). At third-generation syn-
is based on two established software programs, chrotron sources, experience has shown that brute
d*TREK (Pflugrath, 1999) and Mosflm (Leslie, 1992). force data collection (180 or 360 degrees, depend-
These programs index diffraction images to deter- ing on the Laue group) with standardized, short
mine the appropriate Laue group. In addition, exposure times usually suffices. In most cases, sam-
they provide an analysis of the properties of the ple integrity can be maintained throughout the