A R T I C L E S
Ruscio et al.
rable catalytic antibody by ∼2 to 3 orders of magnitude.4,5 The
kinetics of RA22 were reported to be complex, showing a burst
phase, which was taken as evidence that product inhibition was
a limiting feature of the intended design, given that an X-ray
crystal structure on a virtually identical RA22 sequence was
within ∼1 Å rmsd of the intended active site geometry.2
In this work, we focus on a completely different design
criterion than static considerations of proximity alone, by asking
whether the geometric criteria of the RA22 active site and its
variants are satisfied temporally during a molecular dynamics
simulation. We want to assay the influence of protein structural
fluctuations executed at physiological temperatures and under
aqueous solvent conditions on realizing the active site archi-
tecture and its intended substrate and intermediate interactions.
Since our own experiments on RA22 fail to reproduce the burst
phase reported by Jiang and co-workers,2 it appears that product
inhibition is not limiting the catalytic efficiency. Instead we find
that catalysis is inhibited by the primary retrol-aldol step of
proton abstraction from the covalently bound substrate, due to
dynamical fluctuations that fail to meet the geometric criteria
of substrate interactions with the histidine-aspartate dyad on a
temporal basis.
tanone (AsisChem, Cambridge, MA), was carried out in 2.7%
acetonitrile, 25 mM HEPES, 100 mM NaCl (pH 7.5) at 28 °C;
180 µL buffer containing 16.44 µM protein and 5 µL substrate
(1.25, 2.5, 5, 10, 20, and 40 mM) in acetonitrile. The progress of
the reaction was followed using a Spectramax M2 plate reader
(Molecular Devices, Sunnyvale, CA) in Costar 96 well black round
bottom plystyrene non-binding surface microplates (Cornnig, Low-
ell, MA). Reactions were monitored over 5 h, with λex of 330 nm
and λem of 452 nm and a cutoff filter at 435 nm. Measurements of
known product concentrations were used to quantify the reaction
fluorescence curves.
Reaction rates for RA22 and Ala-RA22 were determined using
a linear fit to the fluorescence data. The fluorescence intensity at
452 nm was converted to product concentration values using a linear
calibration equation, which was based on data collection at known
product concentrations. The reaction rate fit was applied over 300
min of data for each of seven initial substrate concentrations,
ignoring the initial 10 min to omit a lag phase we observed for
both proteins at most concentrations. The reaction rate data at
different substrate concentrations were fit to the Michaelis-Menten
equation21 to obtain the Km and Kcat values for the reaction.
Theoretical Methods
Computational Models. The initial structures for the apo-
enzymes were obtained from the RCSB for Ala-RA22 (3B5V.pdb)
and the Supporting Information from Jiang and co-workers.2 The
protein was modeled with the AMBER ff99SB protein force field,22
and the TIP4P-Ew model23 was used to describe the molecular
solvent. Charges and parameters for the substrate were generated
with the Antechamber module of the AMBER package (version
10)24 using the AM1-BCC charge model and the general AMBER
Force Field (GAFF).25 Charges and parameters for the COV2
complex were generated with the Antechamber module using the
AM1-BCC charge module, and the lysine portion of the molecule
is described with standard lysine parameters while the substrate
portion of the molecule is described with GAFF-derived parameters.
Additionally, we reparameterized the force constant for the dihedral
angle of the iminium bond, as the force constant assigned by
Antechamber was not strong enough to maintain the double bond
character.
We used the sander module in the AMBER package for all of
the molecular dynamics simulations.24 The enzymes were solvated
in a box of water and neutralized before undergoing four stages of
minimization: three stages with loosening restraints (5.0, 1.0, and
0.1 kcal/mol/Å2) on the solute and one stage with no restraints.
The system then underwent five steps of equilibration. In the first
step, under constant volume, the temperature of the system was
raised from 0 to 300 K using the Anderson thermostat with a 10.0
kcal/mol/Å2 harmonic restraint on the solute. The next three steps
were under NVT conditions while the harmonic restraints were
loosened (5.0 kcal/mol/Å2, 1.0 kcal/mol/Å2, and 0.5 kcal/mol/Å2).
Finally, we simulated the system under NPT conditions using a
weak barostat coupling at 1 bar for 1 ns. For the MD simulations,
we used a 2 fs time step, with the long-range electrostatic
interactions calculated using the Particle Mesh Ewald method and
a cutoff of 9.0 Å for real space electrostatics and LJ interactions.
Experimental Methods
RA22 Sequences. In our computational and experimental studies
we used two sequences reported by the Baker group on RA22. The
first was the intended design based on creation of 13 mutations
into the TIM barrel protein scaffold as reported,2 which we refer
to as RA22. In showing that the RA22 geometric design criteria in
the engineered active site were met with these 13 mutations, the
crystal structure of RA22 was reported in 3B5V, with the
understanding that an additional 14th mutation involving S210A
was to be ignored as that sequence did not have catalytic activity.
However, 3B5V contained an additional 15 alanine mutations of
scaffold residues lysine, asparagine, and glutamic acid on the surface
of the TIM protein; we call this sequence Ala-RA22.
Expression and Purification. Codon-optimized genes for RA22,
and its variant Ala-RA22, were synthesized by GenScript (Piscat-
away, NJ). The genes were directionally subcloned between NdeI
and XhoI sites in a pET29b+ vector (Novagen, Madison, WI)
encoding a carboxy-terminal 6xHis tag. The plasmids were
transformed into BL21(DE3) (Stratagene, Inc.) and cultured in LB-
broth at 37 °C. Protein expression was induced with isopropyl-ꢀ-
D-thiogalactopyranoside at A600 ) 0.6, and growth was permitted
to continue for an additional 8 h. The cells were then pelleted by
centrifugation for 10 min at 500 rpm and resuspended in 50 mM
potassium phosphate buffer (pH 7) and 200 mM NaCl (loading
buffer) with an additional 10 mM imidazole. The cells were then
lysed by French press and the lysate cleared by centrifugation at
15000 rpm for 1 h. The cleared lysates were loaded onto a Ni-
NTA agarose (Qiagen, Inc.) affinity column, washed with 10
column volumes of loading buffer with an additional 40 mM
imidazole, and eluted with a 40-250 mM imidazole gradient. The
purity of the protein, estimated at greater than 90%, was evaluated
by SDS-page. Protein concentration was determined by absorbance
at 280 nm using a Lambda 650 UV-vis spectrometer (PerkinElmer,
Fremont, CA). The extinction coefficient of 24410 M-1 cm-1 was
calculated from the amino acid sequence using the ProtParam Tool
(Swiss Institute of Bioimformatics).
Results
Experimental Results. We analyzed the kinetics of the retro-
aldol reaction for the RA22 and Ala-Ra22 enzymes as described
in methods section. We obtained Km and Kcat values which were
within the (large) error bars of the steady state rate reported by
Jiang and co-workers2 (Figure 2a), but not within error of the
Kcat that they obtained for the burst phase. In fact, for RA22
the Kcat/Km value we obtained, 0.017 ( 0.010 M-1 s-1, is very
close to the steady-state value reported by Jiang and co-workers,
0.018 ( 0.006 M-1 s-1. However, we did not observe a burst
phase in the first 20 min of our progress curve (Figure 2b). The
Enzyme Kinetics and Data Analysis. Evaluation of RA22 and
Ala-RA22 kinetics was performed as previously described.2 The
retro-aldol reaction of 4-hydroxy-4-(6-methoxy-2-anphthyl)-2-bu-
(4) List, B.; Barbas III, C. F.; Lerner, R. A. Proc. Natl. Acad. Sci. U.S.A.
1998, 95, 15351–15355.
(5) Sterner, R.; Merkl, R.; Raushel, F. M. Chem. Biol. 2008, 15, 421–
423.
9
14112 J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009