| Jun
1, 2005 |
| By:
Andrew
May |
| Pharmaceutical
Discovery |
|
Figure 1. A TOPAZ 4.96 Screening
Chip for protein crystallization. The microfluidic reaction
chambers are encapsulated in a silicone elastomer (center) and
housed in a microtiter plate format carrier (SBS standard).
|
This note describes an innovative model for protein structure
determination that significantly reduces the time, experimental burden and
upstream cost associated with growing diffraction-quality crystals. With
this model, laboratories use the TOPAZ™ System to screen microscale
quantities of protein, in parallel, to determine which targets are
suitable for further crystallization study. Only those constructs showing
a propensity to crystallize are subsequently scaled up and optimized to
produce crystals of sufficient size and quality for X-ray diffraction
(Figure 1).
Figure 2. Relative sample
requirements for liquid-handling technologies on a per experiment
basis.
|
Introduction In 2002, Fluidigm (South
San Francisco, CA, USA) introduced the TOPAZ System for protein
crystallization, which consists of microfluidic screening chips,
preparative and analytical instrumentation and crystallization reagents.
The system imposes an extraordinarily low sample burden compared to
dispensing platforms – a 96-reagent screen requires only 1 mL of sample,
or approximately 10 nL per reaction (Figure 2). Fluids are metered via a
defined chip architecture – a distribution network of uniform reaction
chambers – which serves to eliminate the variability typically
associated with differences in reagent viscosities. The chip geometry, in
turn, supports both high-fidelity imaging and accurate auto-scoring of
results by virtue of consistencies in the size, shape and location of
crystal imaging regions. Existing versions of the TOPAZ Screening Chips
accept one, four or eight protein samples, which are automatically arrayed
onboard into 96-, 384- or 768 experiments, respectively. This parallel
throughput, along with the ultra low sample volume requirement, is
enabling a shift in the fundamental approach to protein crystallization.
Figure 3. Rate of attrition for
protein structure determination. Percentages represent averages of
reported attrition along the protein crystallization pipeline at
seven PSI/NIGMS sites (MCSG, NYSGC, BSGC, SECSG, CESG, TB and JCSG)
as of March 2005.
|
Experimental Conditions Protein
crystallization traditionally has involved a series of activities, which
may be broadly categorized as cloning, expression, purification, scale-up
(large-scale expression and purification) and crystallization. Until
recently, laboratories have pursued a serial path along this series of
low-probability experiments (Figure 3), and relied on feedback loops to
inform their progress. The propensity of a particular protein sample to
crystallize is only indirectly addressed through iterative quality-control
(QC) steps, and target selection is biased toward those samples that can
be scaled up rather than crystallized. The exhaustive screening process
that follows may or may not succeed in identifying the chemical conditions
that generate high-quality crystals. The costs incurred in ultimately
solving a structure are directly related to the accumulated probability of
each upstream step in the sequence and the predictive ability of the same
in leading to diffraction quality crystals.
Figure 4. A diagram showing serial
versus parallel screening methodologies. A significantly reduced
sample requirement for screening allows laboratories to obtain QC
data on crystallization early in the process as well as in
parallel, ensuring that downstream efforts are applied to viable
leads.
|
The emergence of ultra low-volume
screening techniques, including the free interface diffusion method (FID)
employed by the TOPAZ system, has allowed a retooling of the pipeline so
that crystallization screening can be placed as a QC step earlier in the
process (Figure 4). Small analytical batches of protein are processed in
parallel on the TOPAZ System using a 4.96 or 8.96 Screening Chip. Only
those targets that demonstrate crystallization are produced in the larger
volumes needed to generate crystals for X-ray diffraction. Culling out
poor crystallizers spares wasted downstream time and effort, thereby
streamlining the entire process. In this scenario, the laboratory
increases its capacity to pursue more targets, which is the goal of
protein structure initiative (PSI) sites like the University of Madison,
Wisconsin, Center for Eukaryotic Structural Genomics (CESG) (Madison, WI,
USA). Alternatively, researchers may attempt more constructs and
homologues among a few proteins of interest. In either case, the aim is to
generate more diffraction-quality crystals from the pipeline in a given
period of time and at a reduced cost.
This parallel processing method, with
QC crystallization as an intermediate step, currently is being used by the
CESG. Under the direction of Craig Bingman, the Crystallography Section
screens and optimizes protein crystals for structure determination from
Arabidopsis and other eukaryotes. Since the project's inception,
approximately 6000 targets have been identified and fed into the pipeline,
53 of which have been solved. This high attrition rate is typical of a PSI
site focused on eukaryotic (as opposed to prokaryotic) target proteins and
is further evidence of the fundamental problems in traditional
crystallization efforts.
Bingman and his colleagues are
interested in improving the efficiency with which they undertake their
crystallization projects and overcoming the bottlenecks in preparing
sufficient high-quality eukaryotic protein for biophysical studies. The
laboratory is exploring various methods for cloning, expressing and
purifying multiple targets in small amounts. Only those targets showing a
strong indication of crystallization are scaled up and carried forward to
large-scale screening with vapor diffusion. Although TOPAZ Screening Chips
represent additional costs in the pipeline, Bingman expects early
knowledge of "crystallizability" to save unnecessary passes
through large-scale protein production, cell growth and
purification-activities, which cost a few thousand dollars for each pass
and a month of time from start to finish.
The CESG has surveyed a large number
of high-throughput Escherichia coli expression strategies to find one that
gives satisfactory results, that is, in generating outcomes at the 10–20
µL scale that may be reliably replicated at the volumes required for
screening with vapor diffusion. The laboratory recently has developed a
new "autoinduction" media that works for both 96-well master
blocks and in large-scale cultures. This breakthrough is important for
high-throughput operations because it allows processing of targets in
parallel with less labor. The expression of recombinant proteins from E.
coli typically is at levels suitable for a first-pass crystallization
screen. Samples from expression-positive wells are purified, also in
high-throughput mode, desalted, concentrated and loaded into a TOPAZ
Screening Chip.
Figure 5. Crystal images collected
on the TOPAZ AutoInspeX Workstation from protein samples
subsequently selected for scale-up.
|
The laboratory currently is using the
4.96 Chip, but expects to migrate to the eight-sample version once it
becomes widely available. Each target is screened against 192 unique
conditions using Fluidigm's OptiMix™ Reagents and imaged on the
AutoInspeX™ Workstation. Crystallization data for each sample–reagent
pairing then is automatically scored and displayed by Crystal Vision
software, which sorts the auto-calls in a prioritized ranking. Typically,
only those images in the top third of the rankings are reviewed and
assigned a user-based call. Ultimately, the laboratory employs a scale-up
strategy to vapor diffusion on those samples that produce one large
crystal in a single reagent condition or small crystals seen across three
reagent conditions (Figure 5).
Table I. Recent comparative trial
data from the CESG for Vapor Diffusion and TOPAZ using either
large-scale or micro sample preparations.
|
Results
Bingman reports that the predictive power of microscale parallel screening
is advancing rapidly, with an increasing correlation between micro and
macroscale results (Table I). The laboratory has run micro samples for
several novel proteins on TOPAZ Screening Chips. The results are
consistent with the success rate for first-pass novel targets screened by
macro methods. Qualified leads currently are being forwarded to larger
scale production and subsequent vapor diffusion trials. In addition, based
on the results of these initial microscale studies, the CESG now is also
reviving targets where yields had been insufficient or quantities too
scarce to complete their analysis by conventional means.
Conclusions
The trajectory of structure discovery
historically has followed a serial path, and the level of attrition
inherent in the process from clone to crystal is a likely byproduct. By
dramatically reducing the amount of sample required for crystallization
study, researchers can take advantage of new opportunities. First, by
reducing the experimental burden associated in producing samples in
sufficient volume, more targets can be screened for a given level of
laboratory expenditure. Parallelization at the various stages in the
pipeline will in turn lead to higher throughout and faster times to
structure. Second, by moving crystallization upstream in the process,
researchers can now make a more direct inference on a particular target's
ability to yield diffraction quality crystals. Time and effort will be
conserved by strategically selecting crystals with the most promise, once
again reducing costs and accelerating discovery.
The described model not only has the
potential to dramatically increase the number of structures solved at
laboratories funded under the PSI directive, but is equally applicable to
any structural biology researcher wishing to increase throughput. By
eliminating non-viable crystallization candidates earlier in the process,
this methodology also is suitable for evaluating multiple constructs,
target–ligand co-crystallizations and expression/purification schemes.
Crystallization should no longer be considered the rate limiting step in
structure determination, but rather can be viewed as an enabling tool in
creating a more efficient discovery pipeline.
Bibliography
Bingman, Craig. Comparative Crystallomics at the Center for Eukaryotic
Structural Genomics (CESG), Nigms.nih.gov
directory 24, CraigBingman.ppt, 3 Feb. 2005 http://pub.nigms.nih.gov/ppcw2005/PPCW2005PowerPoints/Session_3/
Bingman, Craig et al. Comparative
Crystallomics at the Center for Eukaryotic Structural Genomics (CESG)http://www.uwstructuralgenomics.org/presentations/
PPCW_2005_bingman_final.pdf
Hansen, C. L. et al. "A robust
and scalable microfluidic metering method that allows protein crystal
growth by free interface diffusion." PNAS 99(26), 16531-16536 (2002)
Segelke, B. "Macromolecular
crystallization with microfluidic free-interface diffusion." Expert
Rev. Proteomics 2(2), 165-172 (2005)
Xiao, T. et al. "Structural
basis for allostery in integrins and binding to fibrinogen-mimetic
therapeutics." Nature 432, 59-67 (2004)
Acknowledgement
Fluidigm Corporation wishes to thank Dr. Craig Bingman for his
contribution to this article.
Fluidigm Corporation
7100 Shoreline Court, South
San Francisco, CA 94080
Tel. 650-266-6000; Fax 650-871-7152
www.fluidigm.com.
|
