Crystallization by SER

1
Crystallization by SER David R. Cooper1,3,
Tomasz Boczek1,3, Katarzyna Grelewska1,3,
Malgorzata Pinkowska1,3, Michal Zawadzki1,3,
Lukasz Goldschmidt2,3, David Eisenberg2,3,
Zygmunt Derewenda1,3. 1Department of Molecular
Physiology and Biological Physics, University of
Virginia. Charlottesville, VA. 2Department of
Chemistry and Biochemistry, UCLA. Los Angeles,
CA. 3The ISFI (The Integrated Center for
Structure and Function Innovation)
ISFI
Abstract
The ISFI
Results
The fact that crystallization remains a
bottleneck for the structure determination of
even well-behaved, soluble proteins is leading
more laboratories to use protein engineering to
enhance crystallization rates. The SER approach
(Surface Entropy Reduction) involves replacing
large, highly entropic residues on the surface of
the protein with low side chain entropy residues,
e.g. alanines. This, among other things, reduces
the entropic cost of immobilizing large side
chains at the intermolecular contact regions and
creates patches capable of mediating crystal
contacts. This method has proven to be an
effective salvage pathway for proteins that are
difficult to crystallize. We have now completed
a systematic study to evaluate 1) the efficacy of
using other residues as the replacement residue
and 2) the use of alternative reservoir solutions
as an augmentation to traditional screening. We
replaced nine Lys/Glu-rich patches in a model
protein, RhoGDI, with alanines and four other
target residues His, Ser, Thr, and Tyr. Our
results reaffirm that alanine is a particularly
good choice for a replacement residue and
identifies tyrosines and threonines as additional
candidates with considerable potential to mediate
crystal contacts. The mutated residues often
participate in crystal contacts. The alternative
screening procedure proved to be quite fruitful.
This screening method, recently reported by Janet
Newman, replaces the normal crystallization
solution in the reservoir with an alternative
reservoir solution. More than half of the mutants
in this study yielding more crystals when 1.5 M
NaCl was used as the reservoir solution than with
the traditional reservoir solution. Together,
this suggests a crystallization strategy for
proteins for which crystallization is a major
bottleneck. First and foremost, the wild type
protein should be screened against alternative
reservoirs. If this proves fruitless, creating
mutants by replacing a patch of glutamates and
lysines with two or more target residues and
performing screens with normal and alternate
reservoir solutions greatly increases the chances
of obtaining diffraction quality crystals. We
have also created a web-based server designed to
identify mutations that may facilitate
crystallization (http//nihserver.mbi.ucla.edu/SER
). The server incorporates the three primary (and
several secondary) criteria that need to be
considered when selecting sites for mutagenesis.
These are 1) the entropy of the residues within a
continuous stretch, which can be represented as
an entropy profile graphed by residue number, 2)
the predicted secondary structure, and 3) the
sequence conservation.
Lawrence Berkeley National Laboratory
Lawrence Livermore National Laboratory
University of Chicago
University of Virginia
University of California Los Angeles
Los Alamos National Lab
The SERp Server (The SER Prediction
Server) http//nihserver.mbi.ucla.edu/SER We
have developed a web server that takes a
proteins sequence and predicts the mutations
that will increase its crystallization potential.

Explanation of Results Table The mutant
designation (A-I) is shown on the left. Coloring
within the table is primarily for readability.
The yellow shaded cells below the main table
contain the totals for each target residue
series. The yellow cells to the right of the
main table contain the totals for each mutation
cluster. The conditions column within each
target residue series is the number of different
conditions that produced hits when both screens
were performed. The term Unique in the
Totals section is the number of different
crystallization conditions producing hits for the
mutation cluster.
Experimental Design
Wild Type RhoGDI Results No hits in the
traditional screen. 1 hit in the salt screen.
Get Pretty Results then get crystals.
Nine high-entropy clusters of RhoGDI, were
replaced with five different target residues
Ala, His, Ser, Thr, and Tyr Wild Type RhoGDI
and SER mutants were screened using 2
procedures, a traditional vapor diffusion
set-up and an alternate reservoir set-up.
Structures from this Study
HSP33 Structure
Server Vetting We have submitted all of the
reported, successful SER structures to the SERp
server and compared the default output with
reported experimental data. In the vast majority
of cases the manually chosen mutation sites of
the crystallized variants were easily identified
by the SERp server. In 11 out of 13 cases
representing the canonical use of the SER
approach, the SERp server identifies the crystal
yielding mutation as one of the top
recommendations. We are currently in the
process of testing these predictions using a pool
of targets from the JCSG, MCSG and TBsgc that are
recalcitrant to crystallization in their wild
type form.
Crystallization Solutions The Super Screen was
based on data mining experiments from the Stevens
and Edwards labs (see refs). The JCSG (Qiagen)
is very similar to the Super Screen and is what
we currently use. Standard Screens Drops are
mixed 11 with protein solution and Super Screen
reagent. Reservoir is 60 µl of Super Screen
reagent. Salt Screens See Newman
Reference Drops mixed as above. Reservoir of
every well is 60 µl of 1.5 M NaCl. Whats a
Hit? The drop was only considered to be a hit
if any well defined crystals, no matter how
small, could be detected by microscopic
examination. Mutant Nomenclature Throughout
this poster, specific cluster/residue
combinations will be referred to by a two letter
designation. The first letter corresponds to
the mutation cluster. The second letter is the
target residues single letter code. Thus, the
CY mutant contains the mutations K135Y, K138Y,
and K141Y, whereas the IA mutant contains the
mutations K98A and K99A. Mutation Cluster Key

• Comparing the Screens
• Overall, the salt screens produced almost 33
  more hits 242 vs. 183.
• For 27 of the 38 mutants that produced hits, the
  salt screen yielded as many or more hits than the
  standard screen.
• In 18 of the 40 mutants, it produced more hits.
  For 5 of the 8 mutation series, the salt screen
  produced more hits.
• The two screens yielded crystals in different
  conditions. There are few conditions that
  produce hits in both screens. This greatly
  increases the number of conditions that can
  produce hits when both screens are used.
• There is also a strong preference for the salt
  screen for particular mutants.
• In a number of cases the salt screen produces
  many more hits than the
• traditional screen. There is only one case
  where the standard screen
• produced many more hits than the salt screen
  (HY).

• Additional Server Features
• Although designed with reasonable defaults, there
  are many parameters that may be accessed and
  changed.
• Modifying job parameters can be done on the fly
  when viewing results.
• Mini-Meta-Server Function Sequences are
  submitted to ProLinks and Blocks to identify
  possible interactions and functional motifs. For
  sequences with similarity to a previously
  determined structure, a link to the PDB is
  provided and DSSP is run to help evaluate the
  surface accessibility of proposed mutations.
• New Security Features
• Its fast.

• References
• Zygmunt Derewenda. Rational Protein
  Crystallization By Mutational Surface
  Engineering. Structure. 12529-35.
• Matthew Kimber, et al. Data Mining
  Crystallization Databases Knowledge-based
  Approaches To Optimize Protein Crystal Screens.
  Proteins. 51562-8.
• Janet Newman. Expanding Screening Space Through
  The Use Of Alternative Reservoirs In
  Vapor-diffusion Experiments. Acta Cryst.
  D61490-3.
• Rebecca Page and Raymond Stevens. Crystallization
  Data Mining In Structural Genomics Using
  Positive And Negative Results To Optimize Protein
  Crystallization Screens. Methods. 34373-89.
• Stephen D. Pickett and Michael J. E. Sternberg.
  Empirical Scale Of Side-chain Conformational
  Entropy In Protein Folding. J Mol Biol.
  231825-39.

SER structures in the PDB 1htj 1mn3 1p3q
1p4o 1r6f 1s99 1sbr 1uxo 1vzy
1y7m 1yuw 1zhm 1zhp 1zhr 2bch
2bjo 2bra 2bxw 2cgj 2cgk 2cgl
2fy2 2fy3 2fy4 2fy5 2g8c 2hdv
2hdx 2jhs 2jht 2jhu 2jhv 2jhw
2jhx 2jhy 2jhz 2ji0 Yours
could be next.
Funding The Integrated Center for Structure and
Function Innovation is funded by NIH U54
GM074946. Use of the Advanced Photon Source was
supported by the U. S. Department of Energy,
Office of Science, Office of Basic Energy
Sciences, under Contract No. W-31-109-Eng-38.
2
(No Transcript)