W0351
Automated CRYSTOOL Crystallization Screening at the TB
Structural Genomics Crystallization Facility. Brent Segelke, Timothy Lekin,
Dominque Toppani, Johana Schafer, Bernhard Rupp. Macromolecular
Crystallography, Biology and Biotechnology Program, Lawrence Livermore National
Laboratory, L448, POB 808, Livermore, CA 94551.
The M. tuberculosis Structural Genomics Consortium
Crystallization facility is being developed at Lawrence Livermore National Lab.
Our efforts focus on adapting automated design of crystallization screens using
CRYSTOOL to automated setup, tracking and analysis. By considering crystal
screening as a sampling problem, we have previously demonstrated, by probability
theory, the inherent efficiency of CRYSTOOL screening. Using CRYSTOOL we are
able to generate any number of random combinations of crystallization conditions
from a large set of starting stock solutions and have interfaced CRYSTOOL to an
automated liquid-handling system (Packard, MPII-HT). 1 ml each of three 96
condition CRYSTOOL-Screens can be mixed in under 3h. Sitting-drop experiments
are set up in 96-well Intelli-Plates using a Hydra Plus One (Apogent inc.) in ~2
min per plate. Intelli-Plates, have been designed for robotic handling and ease
of crystal harvesting. Plates are imaged with a robotic imaging system,
VersaScan developed with Velocity11, and images are processed with prototype
crystal detection software. Control software and a LIM system integrate the
VersaS-can with a plate mover, the liquid-handling robot with a sealer yet to be
added for complete automation. Our current throughput is estimated at 12
proteins (288 conditions each) per day. To date, we have processed >100
protein samples from consortium facilities or member labs and ~35% of proteins
provided yield lead conditions from initial screens. Optimization remains a
significant bottleneck. Mining the database of crystallization experiments
amassed thus far provides quantitative success rate comparisons for reagents
used in crystallization. With continued application of automated CRYSTOOL we
will be able to narrow initial screens to “hot spots” in
crystallization parameter space and increase our success rate and throughput
while reducing cost.
This work was performed under U.S. DOE auspices by the
University of California, Lawrence Livermore National Laboratory, Contract
W-7405-Eng-48.