Computational and experimental analyses converge to reveal a coherent yet malleable aptamer structure that controls chemical reactivity

Article abstract of DOI:10.1021/ja902719q

As short nucleic acids, aptamers in solution are believed to be structurally flexible. Consistent with this view, most aptamers examined for this property have been shown to bind their target molecules by mechanisms that can be described as "induced fit".

Full text of DOI:10.1021/ja902719q

Published on Web 09/24/2009
Computational and Experimental Analyses Converge to
Reveal a Coherent Yet Malleable Aptamer Structure That
Controls Chemical Reactivity
Tianjiao Wang,†,⊥ Julie A. Hoy,‡ Monica H. Lamm,§ and
Marit Nilsen-Hamilton*,†,⊥
Department of Biochemistry, Biophysics and Molecular Biology, Ames Laboratory,
Macromolecular X-ray Crystallography Facility, Office of Biotechnology, and Department of
Chemical and Biological Engineering, Iowa State UniVersity, Ames, Iowa 50011
Received April 9, 2009; E-mail: marit@iastate.edu
Abstract: As short nucleic acids, aptamers in solution are believed to be structurally flexible. Consistent
with this view, most aptamers examined for this property have been shown to bind their target molecules
by mechanisms that can be described as “induced fit”. But, it is not known to what extent this structural
flexibility affects the integrity of the target-aptamer interaction. Using the malachite green aptamer (MGA)
as a model system, we show that the MGA can protect its bound target, malachite green (MG), from oxidation
over several days. Protection is reversed by an oligonucleotide complementary to the MGA binding pocket.
Computational cavity analysis of the MGA-MG structure predicted that MG oxidation is protected because
a molecule as small as an OH- is sterically excluded from the C1 position of the bound MG. These results
suggest that, while the MGA-MG interface is sufficiently coherent to prevent OH- penetration, the bases
involved in the interaction are sufficiently mobile that they can exchange out of the MG binding interface
to hybridize with a complementary oligonucleotide. The computational predictions were confirmed
experimentally using variants of the MGA with single base changes in the binding pocket. This work
demonstrates the successful application of molecular dynamics simulations and cavity analysis in determining
the effects of sequence variations on the structure of a small single-stranded nucleic acid. It also shows
that a nucleic acid aptamer can control access to specific chemical groups on its target, which suggests
that aptamers might be applied for selectively protecting small molecules from modification.
Introduction
binding make modeling aptamer-target interactions more
difficult than for most protein-target interactions.
Aptamers are short, single-stranded nucleic acids that bind
to their target molecules tightly with high specificities and that
can be selected in vitro to recognize a large range of types and
sizes of targets.1 For protein structures, computational docking
can often correctly identify potential molecular interactions by
examining surface structural features. Also, certain protein
binding pockets have been shown to exclude small molecules
such as water from their bound targets.2-6 For nucleic acid
aptamers it has not been demonstrated that these more flexible
polymers can adopt sufficiently rigid structures when complexed
with their target molecules that they can exclude the entry of
small molecules into the binding pocket. Short nucleic acids
also sample more diverse structures in solution than do most
proteins.7-9 These structural changes upon target molecule
The malachite green aptamer (MGA) binds one of two
molecular targets, malachite green (MG) or tetramethylrosamine
(TMR). Structural analyses of the MGA-target complexes show
MG or TMR sitting in a binding pocket formed by the aptamer
bases with less than 20% of the target exposed.10,11 The aptamer
pocket interacts differently with MG and TMR being modified
in its structure by induced fit, which is a feature of the interaction
of many aptamers with their targets.11-19 Both aptamer and
target molecule change in their structures during binding, with
(5) Wogulis, M.; Wheelock, C. E.; Kamita, S. G.; Hinton, A. C.;
Whetstone, P. A.; Hammock, B. D.; Wilson, D. K. Biochemistry 2006,
45, 4045–4057.
(6) Schramm, A. M.; Mehra-Chaudhary, R.; Furdui, C. M.; Beamer, L. J.
Biochemistry 2008, 47, 9154–9162.
(7) Getz, M.; Sun, X.; Casiano-Negroni, A.; Zhang, Q.; Al-Hashimi, H. M.
Biopolymers 2007, 86, 384–402.
(8) Xia, T. Curr. Opin. Chem. Biol. 2008, 12, 604–611.
(9) Al-Hashimi, H. M.; Walter, N. G. Curr. Opin. Struct. Biol. 2008, 18,
321–329.
† Department of Biochemistry, Biophysics and Molecular Biology.
‡ Macromolecular X-ray Crystallography Facility, Office of Biotechnology.
§ Department of Chemical and Biological Engineering.
⊥ Ames Laboratory.
(1) Stoltenburg, R.; Reinemann, C.; Strehlitz, B. Biomol. Eng. 2007, 24,
381–403.
(10) Baugh, C.; Grate, D.; Wilson, C. J. Mol. Biol. 2000, 301, 117–128.
(11) Flinders, J.; DeFina, S. C.; Brackett, D. M.; Baugh, C.; Wilson, C.;
Dieckmann, T. ChemBioChem 2004, 5, 62–72.
(12) Dieckmann, T.; Butcher, S. E.; Sassanfar, M.; Szostak, J. W.; Feigon,
J. J. Mol. Biol. 1997, 273, 467–478.
(2) Winkler, F. K.; D’Arcy, A.; Hunziker, W. Nature 1990, 343, 771–
774.
(3) Kim, S. W.; Cha, S. S.; Cho, H. S.; Kim, J. S.; Ha, N. C.; Cho, M. J.;
Joo, S.; Kim, K. K.; Choi, K. Y.; Oh, B. H. Biochemistry 1997, 36,
14030–14036.
(13) Patel, D. J.; Suri, A. K.; Jiang, F.; Jiang, L.; Fan, P.; Kumar, R. A.;
Nonin, S. J. Mol. Biol. 1997, 272, 645–664.
(4) Sneeden, J. L.; Loeb, L. A. J. Biol. Chem. 2004, 279, 40723–40728.
(14) Williamson, J. R. Nat. Struct. Biol. 2000, 7, 834–837.
9
10.1021/ja902719q CCC: $40.75  2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 14747–14755 14747

A R T I C L E S
Wang et al.
precipitated product was centrifuged at 1500 × g for 10 min. The
pellet was collected and then dried at 74 °C for 6 h. All pHs reported
here were measured at room temperature (∼23 °C).
the result that the flatter MG in the aptamer pocket has a much
higher fluorescence output than unbound MG.20
The central carbon of MG is susceptible to attack by a base
such as a hydroxide ion (OH-), which results in the formation
of a colorless form (MG-OH). This reaction provides an
opportunity to examine the extent to which the aptamer-target
interaction is rigidly defined. If the aptamer-target complex
samples a large range of structures, some of these structures
might allow hydroxyl ions to slip into the core of the structure
to oxidize the target molecule.
Computational analysis of protein-target interactions can aid
experimental analysis when the results of amino acid residue
exchanges can be tested in silico. Although molecular dynamics
simulations have been used to examine the thermodynamic
parameters associated with target molecule binding, the ability
of computational analysis to accurately predict the structures
of aptamers with altered sequences is not known. The avail-
ability of experimentally determined structures of the MGA-
MG complex allowed us to test the ability of computational
analysis (cavity analysis and molecular dynamics simulations)
to accurately predict experimental results.
Malachite green is widely used as a dye in industry and as a
fungicide in aquaculture. However, it is toxic and can promote
cancer and embryonic defects.21-24 The mechanism of MG
toxicity is unknown but believed to involve its metabolism to
more toxic forms. The investigation described here of the MG-
MGA interaction shows for the first time that the reactivity of
this chemical can be controlled by an aptamer, a polymer with
the potential of in vivo function. This same concept might be
extended to other reactive chemicals.
Here we analyzed the NMR structures of the MGA-MG
complex by cavity analysis and molecular dynamics simulations
and extended this analysis to simulations of aptamers with
variations in sequence from the MGA. The computational
analyses accurately predicted the results of experimental
structural analyses of the MGA and its variants and their abilities
to protect MG from oxidation. These studies show that, although
it is a very flexible polymer, an aptamer can fold around its
target molecule in a stable, tightly constrained structure that
will not allow a molecule as small as a hydroxyl group to
penetrate the complex over a period of days.
Oligonucleotides were synthesized by Integrated DNA Technolo-
gies, Inc. (Coralville, IA), or by the DNA Synthesis and Sequencing
Facility at Iowa State University as listed: MGA,25 GGAUC-
CCGACUGGCGAGAGCCAGGUAACGAAUGGAUCC; MGA
(U25C), GGAUCCCGCAACGAAUGGAUC; MGA(A31C), GGA-
UCCCGACUGGCGAGCUGGAUCC; MGA (G8C, G24C, G29C),
GGAUCCCCACUGGCGAGAGCUAACCAAUGGAUCC; AS (an-
tisense oligonucleotide that targets the region of MGA from G19
to C37), GATCCAGTTACCTGGC; and Shuffled AS (with the
same bases but randomly shuffled with respect to the AS sequence),
TCTCTAGAGTCTCTGCACG.
In addition to chemical synthesis, in vitro transcription by T7
RNA polymerase was also used to generate a 100 nt random
sequence RNA pool: GGGAGACAAGAAUAAACGCUCAA(N53)-
UUCGACAGGAGGCUCACAACAGGC.
Mass Spectrometry. Electrospray ionization of 1 mM MG in
ddH2O was performed with a Finnigan TSQ700 triple quadrupole
mass spectrometer (Finnigan MAT, San Jose, CA) fitted with a
Finnigan ESI interface. The sample was introduced into the
electrospray interface through an untreated fused-silica capillary
with a 50 µm i.d. and 190 µm o.d. A mixture of 75 µg/mL horse
skeletal muscle myoglobin and 12 µg/mL Met-Arg-Phe-Ala (MRFA)
tetrapeptide in a 50:50 methanol/water solution was used for tuning
and routine calibration of the instrument. The tuning mixture in a
polypropylene vial was infused into the mass spectrometer at a rate
of 3 µL/min on a Harvard Apparatus (model 22, South Natick, MA)
syringe pump.
Electron impact ionization of MG-OH was performed on a
TSQ700 triple quadrupole mass spectrometer (Finnigan MAT, San
Jose, CA) fitted with a Finnigan EI/CI ion source. The sample was
introduced into the mass spectrometer using the solids probe that
was heated gradually from 100 to 400 °C. The instrument was used
as a single quadrupole and scanned from 35 to 650 Da.
NMR Spectrometry. Solid state 13C NMR spectra of MG and
MG-OH were recorded at room temperature at 150 MHz by a
Bruker AV-600 with CPTOSS as the pulse program.
1H NMR spectra of all samples were collected using a Bruker
Avance 700 spectrometer equipped with a 5 mm HCN-Z gradient
1
cryoprobe. All H NMR spectra were acquired using a WATER-
GATE pulse sequence with water flipback (Bruker sequence
p3919fpgp) to minimize solvent saturation transfer. Spectra were
averages of 512 transients. The spectral width was 18182 Hz (25
ppm). The samples were 100 µM MG, 50 µM MGA ( 100 µM
MG, 50 µM MGA(U25C) ( 200 µM MG, 50 µM MGA(A31C) (
100 µM MG, and 50 µM MGA(G8C-G24C-G29C) ( 500 µM MG.
All samples were in 10 mM KH2PO4, 5 mM MgCl2, 10 mM KCl,
pH 5.8, 5% D2O/95% H2O.
Affinities of MGA Variants for MG. Binding to the MGA
variants causes the λmax of MG absorption to shift from 618 nm to
between 629 and 632 nm depending on the variant.25 The binding
also enhances MG fluorescence.20 On the basis of these properties,
the affinities of the MGA variants were determined by UV-visible
spectroscopy either using a Biowave S2100 diode array spectro-
photometer (WPA, Cambridge, U.K.) or an ND-1000 spectropho-
tometer (NanoDrop, Wilmington, DE) or by a fluorescence
spectrophotometer (Cary Eclipse, Variant, Palo Alto, CA). An
antisense oligonucleotide that targets the region of MGA from G19
to C37 (AS) was used as a control for the effects of RNAs that
interact with MG nonspecifically.
Experimental Methods
Chemicals and Oligonucleotides. Malachite green oxalate (MG)
was purchased from Sigma (St. Louis, MO). MG carbinol base
(MG-OH) was prepared as follows: A 90 mL solution of 17 mg/
mL MG in 0.5 M HEPES, pH 7.4, was left at 23 °C for 24 h. The
(15) Merino, E. J.; Weeks, K. M. J. Am. Chem. Soc. 2003, 125, 12370–
12371.
(16) Katilius, E.; Katiliene, Z.; Woodbury, N. W. Anal. Chem. 2006, 78,
6484–6489.
(17) Lee, J. H.; Jucker, F.; Pardi, A. FEBS Lett. 2008, 582, 1835–1839.
(18) Lin, P. H.; Yen, S. L.; Lin, M. S.; Chang, Y.; Louis, S. R.; Higuchi,
A.; Chen, W. Y. J. Phys. Chem. B 2008, 112, 6665–73.
(19) Hermann, T.; Patel, D. J. Science 2000, 287, 820–825.
(20) Babendure, J. R.; Adams, S. R.; Tsien, R. Y. J. Am. Chem. Soc. 2003,
125, 14716–14717.
(21) Kraus, G. A.; Jeon, I.; Nilsen-Hamilton, M.; Awad, A. M.; Banerjee,
J.; Parvin, B. Molecules 2008, 13, 986–994.
The UV-visible spectra were obtained of duplicated samples
of MG and each MGA variant in buffers containing 100 mM Tris,
100 mM KCl, 5 mM MgCl2, pH 9.0. The concentrations of the
MGA and variants were adjusted to ensure saturation of MG in
(22) Mittelstaedt, R. A.; Mei, N.; Webb, P. J.; Shaddock, J. G.; Dobro-
volsky, V. N.; McGarrity, L. J.; Morris, S. M.; Chen, T.; Beland, F. A.;
Greenlees, K. J.; Heflich, R. H. Mutat. Res. 2004, 561, 127–138.
(23) Culp, S. J.; Mellick, P. W.; Trotter, R. W.; Greenlees, K. J.; Kodell,
R. L.; Beland, F. A. Food Chem. Toxicol. 2006, 44, 1204–1212.
(24) Jang, G. H.; Park, I. S.; Lee, S. H.; Huh, T. L.; Lee, Y. M. Biochem.
Biophys. Res. Commun. 2009, 382, 486–491.
(25) Grate, D.; Wilson, C. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 6131–
6136.
9
14748 J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009

Coherent yet Malleable Aptamer Structure
A R T I C L E S
each case with at least six different concentrations of the RNA.
The λmax for each scan was recorded and plotted against the
concentration of the appropriate MGA variant or control RNA (AS).
The plots were fitted to λ ) λ0 + ∆λ[M]/(Kd + [M]) using a
nonlinear regression function in the Costat program (CoHort
Software, Monterey, CA), where λ is the maximal absorption
wavelength of MG in the presence of different concentrations of
the RNA, λ0 is the maximal absorption wavelength of free MG,
∆λ is the difference of maximal absorption wavelengths between
free MG and MG saturated with each RNA, Kd is the dissociation
constant, and [M] is the concentration of the RNA.
To obtain Kd-values of the interaction of MG with MGA or its
variants, the fluorescence emission spectra for MG were determined
in 100 mM Tris, 100 mM KCl, 5 mM MgCl2, and pH 9.0 alone
and in the presence of each of 11 different concentrations of each
RNA. The samples were excited at 630 nm and the emission spectra
were recorded from 640-700 nm with 650 nm as the maximal
emission wavelength. The data were fit to F650/Fmax ) [M]/(Kd +
[M]) using the Costat program where F650 is the fluorescence
intensity at 650 nm of samples, Fmax is the F650 when MG is
completely saturated, [M] is the concentrations of the MGA variants,
and Kd is the dissociation constant for MG and the MGA or its
variants.
Kinetics of MG-OH Formation. The UV-visible spectra of
MG, with or without each RNA, in buffers containing either 10
mM HEPES, 100 mM KCl, 5 mM MgCl2, pH 7.2-7.4 (Figure
1) or 100 mM Tris, 100 mM KCl, 5 mM MgCl2, pH 9.0 (Figure
3), were determined by using either a Biowave S2100 diode array
spectrophotometer (WPA, Cambridge, U.K.) or a ND-1000
spectrophotometer (NanoDrop, Wilmington, DE). At least three
independent experiments were done of each described condition.
The rates of production of MG-OH were monitored for up to
46 h by recording A(λmax). The λmax ranged from 618 to 632 nm
depending on the concentration and the MGA variants. Then
A(λmax) was plotted against time and fitted by a nonlinear
regression function in Costat to the equations A ) A0 exp(-kt)
for MG alone at pH > 8 and A ) A0/(1 + ktC0) for all others,
where A0 is the A(λmax) at t ) 0, A is the A(λmax) at each time
point, k is the reaction rate, t is the time, and C0 is the initial
concentration of MG. The half-life of MG-OH formation was
calculated with the best fit as t1/2 ) 0.693/k (MG alone at pH >
8) and t1/2 ) 1/kC0 (MG alone at pH < 8 and in the presence of
MGA and its variants). The statistical analysis of the means of
the t1/2 of MG from Supporting Information Table S1 were done
in R, first by F-test and then by Tukey multiple comparisons of
mean values (Supporting Information Table S2).26,27
Molecular Dynamics (MD) Simulations of MGA-MG
Variants. MD simulation of the MGA-MG variant complexes
were carried out with NAMD2.6 that was developed by the
Theoretical and Computational Biophysics Group in the Beck-
man Institute for Advanced Science and Technology at the
University of Illinois, Urbana-Champaign (http://www.ks.uiuc.edu/
Research/namd/, accessed May 24, 2009).28 The setup for
simulation and analysis of the simulation results were done with
VMD1.8.6 (http://www.ks.uiuc.edu/Research/vmd/, accessed
May24,2009).29TheCHARMMtopologyfile(top_all27_prot_na.rtf.)
and the parameter file (par_all27_prot_na.prm)30 were modified
to include MG for setting up and running the simulation. For
Figure 1. Malachite green aptamer protects MG from oxidation. A, B:
The kinetics of MG oxidation in the presence of MGA or control RNA.
Two micromolar MG was incubated in the absence (3) or presence (]) of
10 µM 100 nt random RNA pool or in the presence of various concentrations
of the MGA: 0.5 µM (b), 1 µM (×), 2 µM (*), 4 µM (+), 6 µM (O)
MGA. C, D: The kinetics of MG oxidation in the presence of the MGA
and varying concentrations of AS. Two µM MG was incubated for 2 min
alone (]) or with 10 µM MGA (4). To this mixture was then added 500
µM shuffled AS (*) or AS at concentrations of 20 µM (b) or 100 µM (O).
A, C: The changes in MG color were tracked as a function of time. B, D:
The constants k and t1/2 were determined from the plots in A and C. The
figure shows the results of one of three independent experiments, each of
which gave consistent results.
MG, the topology and parameter files (see Supporting Informa-
tion) were generated manually according to previously reported
quantum-chemical calculations.31
The procedure is described briefly as follows. The MGA-MG
complex (PDB ID: 1Q8N, model 5)11 and the MGA(A31C)-MG
and MGA(U25C)-MG derived from variations in the sequence of
the MGA by using the Coot package32 were solvated in a water
box (76 Å × 52 Å × 50 Å) containing 9261 TIP3P H2O. Ions (19
Mg2+ and 1 Cl-) were added to the solvated systems using the
Meadionize plugin 1.1 developed by Ilya Balabin, Duke University
(http://www.chem.duke.edu/∼ilya/Software/Meadionize/ docs/meadi-
onize.html, accessed May 24, 2009). The solvated and ionized
systems were then run on NAMD with the CHARMM force field.
For the simulation parameters, the electrostatic and van der Waals
interaction distance cutoff was set to 12 Å. The full-system periodic
electrostatics was calculated with a particle mesh Ewald summation
of ∼1 Å grid space. The time step was set to 2 fs. After 5000 steps
of minimization, the unconstrained simulation was run under a
periodic boundary condition at a constant temperature (298 K) and
pressure (1.01325 bar) for 5.4 ns for MGA-MG, 5.6 ns for MGA-
MG(A31C), and 7.9 ns for MGA-MG(U25C).
(26) Multiple comparisons, selection, and applications in biometry: a
festschrift in honor of Charles W. Dunnett; Hoppe, F. M., Ed.; Dekker:
New York, NY, 1993.
(27) R_Development_Core_Team; R Foundation for Statistical Computing:
Vienna, Austria, 2009.
(28) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.;
Villa, E.; Chipot, C.; Skeel, R. D.; Kale, L.; Schulten, K. J. Comput.
Chem. 2005, 26, 1781–1802.
(29) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996, 14, 33–
38.
(31) Nguyen, D. H.; DeFina, S. C.; Fink, W. H.; Dieckmann, T. J. Am.
Chem. Soc. 2002, 124, 15081–15084.
(32) Emsley, P.; Cowtan, K. Acta Crystallogr., Sect. D 2004, 60, 2126–
2132.
(30) MacKerell, A. D., Jr.; Banavali, N.; Foloppe, N. Biopolymers 2000,
56, 257–265.
9
J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009 14749

A R T I C L E S
Wang et al.
Cavity Analysis of Free MG and the MGA-MG Variants.
Analysis of the 25 NMR structures of the MGA-MG complex
(1Q8N.pdb) by LSQMAN33 indicates that the central chain is model
5 and the average rmsd between chains ) 0.709 Å (using C4* P
C1* C2* C3* O2* O3* O4* atoms to align chains). Model 15 has
the highest deviation from model 5 with the rmsd between chains
) 0.870 Å; thus both models 5 and 15 were analyzed for cavities.
Because no important differences were found between these two
models, only the results for model 5 are reported. Analysis of the
simulated structures of the MGA-MG variants was done by VMD.
Simulated structures with high RMSDs from the 1Q8N.pdb model
5 starting structure (especially for residues 8, 28, and 39) were
chosen for cavity analysis. The reason for these choices was to
determine if the variant molecules sampled structures that were
different enough from the original MGA structure to result in the
creation of a channel for OH- entry. Cavity analysis of the MGA
variants’ binding pockets and free MG were performed using
SURFNET.34 A radius of 1.32 Å was chosen as the smallest
possible radius for an hydroxide ion35 while the maximum radius
was set to 1.9 Å. The figure was created with SPDBV,36 PyMol
(Delano Scientific LLC, http://www.pymol.org, accessed May 24,
2009) and POV-Ray (Persistence of Vision Raytracer Pty. Ltd.).
Structural Probing of the MGA Variants. RNase I footprinting
was used to study the structural alterations of the binding pocket
of the MGA due to sequence variations (A31C, U25C, and G8C-
G24C-G29C). RNase I can cleave RNAs in single-stranded but not
in double-stranded regions. The MGA variants, including MGA,
MGA(A31C), MGA(U25C), and MGA(G8C-G24C-G29C), were
first labeled with γ-32P-ATP (MP Biomedicals, Solon, OH) by T4
polynucleotide kinase (Promega, Madison, WI) at their 5′-ends.
Twenty microliters of labeling reaction mixtures containing 5 µM
RNA, 1.5 µCi/µL γ-32P-ATP, 1 U/µL T4 polynucleotide kinase,
70 mM Tris-HCl (pH 7.6), 10 mM MgCl2, and 5 mM DTT were
incubated for 1.5 h at 37 °C. The labeled MGA variants were
purified by electrophoresis through 10% polyacrylamide gels in the
presence of 7 M urea and then incubated with RNase I in the
presence or absence of MG to probe the RNA structure. Reaction
mixtures of 10 µL containing 0.125 µM 32P-labeled RNA, 0.5 ×
10-4 U/µL RNase I, 100 mM KCl, 5 mM MgCl2, 10 mM HEPES,
pH 7.4, were incubated at 23 °C for 10 min in the absence or
presence of MG. Partial alkaline hydrolysis of the labeled RNAs
was done in 50 µM Na2CO3, pH 9.0 at 95 °C for 5 min. RNase T1
digestion of the labeled RNAs was carried out in 10 µL reaction
mixture containing 0.5-1 U/µL RNase T1, 5 M urea, 350 mM
sodium citrate, 0.7 mM EDTA, pH 5.0, at 50 °C for 4 min. The
RNA hydrolysis and cleavage reactions were stopped by bringing
the reaction mixtures to 47.5% formamide, 0.05% bromophenol
blue, and 0.05% xylene cyanol FF. The samples were resolved by
electrophoresis through 10% polyacrylamide gels in the presence
of 7 M urea. The radioactivity was recorded from the dried gels
using a phosphor screen and imaged with a Typhoon 8600 Variable
Model Imager (GE Healthcare, Piscataway, NJ).
MG had an expected peak at m/z 329, and the MG-OH
spectrum contained an additional m/z 346 peak as expected of
hydroxylated MG (Supporting Information Figure S2A). In the
NMR spectra of MG and MG-OH, the C1 peak was shifted
from the 175.91 ppm in MG to 82.59 ppm in MG-OH. The
C11 of MG at 165.115 ppm was shifted to 149.08 ppm in
MG-OH, and the double peaks of N(CH3)2 and +N(CH3)2
(40.91 ppm and 39.95 ppm) in MG became a single peak of
N(CH3)2 (40.39 ppm) in MG-OH (Supporting Information
Figure S2B). These results are consistent with the conclusion
that MG-OH is hydroxylated at C1 and that the bleached MG
is in the form of a carbinol base (Supporting Information Figure
S2C).
MGA Binding Pocket Protects MG from Oxidation. Because
the MGA binding pocket is formed by bases that lie above and
below MG, it seemed possible that the MGA might protect MG
from oxidation by sterically hindering the entrance of OH- into
the binding pocket. We tested this hypothesis by incubating MG
with various concentrations of the MGA and found that the
MGA inhibited MG oxidation with a dependence on MGA
concentration (Figure 1A,B). At 6 µM MGA, a concentration
that is about 6 times the Kd for the MGA-MG interaction, the
rate of MG oxidation was almost zero. By contrast, 10 µM 100
nt random sequence RNA pool had no effect on the rate of MG
to MG-OH oxidation (Figure 1A,B).
If the MGA protects the MG in its binding pocket from
hydroxyl attack, then a distortion of the binding pocket should
result in loss of its effect on MG oxidation. This result was
observed in experiments in which the reaction mixture included
an oligonucleotide (AS) with a sequence complementary to
bases 19 to 37 in the MGA sequence that constitutes one-half
of the MG-binding pocket. Addition of AS (but not its shuffled
sequence version) to a mixture of MGA and MG reversed the
protection by MGA of MG oxidation with ∼30% reversal at
2AS/MGA and almost complete reversal at 10AS/MGA (Figure
1C,D).
Cavity Analysis of the MGA-MG Complex Shows No Gap
for OH- Entry. Our results suggested that the ability of the MGA
to inhibit oxidation of MG is because the MGA-MG complex
sterically hinders OH- access to the C1 of the MG located in the
binding pocket. As a result of the very small size of OH-, this
condition requires a very snug fit of MG in the binding pocket of
the MGA. Cavity analysis was used to test the hypothesis. With
the route of OH- attack on MG being limited by the equatorial
phenyl rings, anion attack can only proceed by polar routes (Figure
2B,C). However, cavity analysis showed that, while sitting in the
MGA binding pocket, MG was protected from polar attack by G8-
C28 and C7-G29-G24-A31, which does not allow a gap of the
minimum size of an OH- group to reach C1 from the solvent
surface of the MGA (Figure 2D).
Results
Oxidation of MG. MG loses color by a mechanism that is
pH-dependent, and the data for MG at pHs above 8 fit best to
a first order kinetic equation, whereas the data for MG at pHs
less than 8 fit best to a second order equation. The results are
consistent with the bleaching kinetics at above pH 8 being
pseudo-first order (Supporting Information Figure S1). The
chemical form of the bleached MG was characterized by mass
spectrometry and NMR spectroscopy. The mass spectrum of
Changes in the MGA Binding Pocket Sequence
Compromise the Ability of the MGA To Protect MG
Oxidation. To test the hypothesis that steric hindrance in the
MGA binding pocket prevents hydroxyl attack, variants of the
MGA with base changes around the binding pocket were made
including MGA(A31C), MGA(U25C), and MGA(G8C-G24C-
G29C). These changes were chosen based on their locations in
the binding pocket (Figure 2A) and the results of cavity analysis.
The Kd of each MGA variant for MG was measured (Table
1), and a concentration of each MGA variant equivalent to
10-15 times of its Kd was chosen at which MG would be
saturated by the RNA. With these concentrations the rate of
MG oxidation was negligible in the presence of the MGA,
(33) Kleywegt, G. J. Acta Crystallogr., Sect. D 1996, 52, 842–857.
(34) Laskowski, R. A. J. Mol. Graph. 1995, 13, 323–330.
(35) Fox, M.; McIntyre, R.; Hayon, E. Faraday Discuss. Chem. Soc. 1977,
64, 167–172.
(36) Guex, N.; Peitsch, M. C. Electrophoresis 1997, 18, 2714–2723.
9
14750 J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009

Coherent yet Malleable Aptamer Structure
A R T I C L E S
Figure 2. Cavity analysis by Surfnet of MG alone and in the MGA binding pocket. (A) MG (magenta) is flanked in the binding pocket of MGA (1q8n.pdb,
model 5) by C7, G24, G29, and A31 (orange) on one side and G8 and C28 (yellow) on the other. Stem I (red), the base triples (green), stem II (blue), and
the tetraloop (purple) are also shown. (B) MG contains three aromatic rings around the central carbon (green). Hydroxide molecules can attack from above
or below the plane, but (C) the rings block approach within the plane of the ring. (D) Stereoview of the MG binding pocket of MGA, colored as above, with
cyan mesh indicating the area accessible to an hydroxide of minimum radius. The space above and below the plane of MG is blocked by the bases of the
binding pocket and is therefore not accessible, as indicated by the lack of mesh. The foreground of the image has been slabbed away to assist in viewing
the interior of the binding pocket.
Table 1. Affinities and λmax of the Interaction of MG with the MGA and Its Variantsa
RNA
MG
Kd (µM)
λmax (nm)
t1/2 of MG oxidation (h)
MGA
MGA(A31C)
MGA(U25C)
MGA(G8C-G24C-G29C)
AS
0.1-15 µM
0.2 µM
0.2 µM
0.4, 2 µM
20 µM
1.03 ( 0.53
1.93 ( 0.97
13.8 ( 3.40
274 ( 65
632
632
629
629
629
618
UD
24
24
5.5
ND
0.82
0.1-15 µM
1-400 µM
0.025-1.7 mM
0.025-1.2 mM
20 µM
172 ( 13
none
2, 20 µM
a The MGA, MGA variants, and AS were tested for their abilities to bind MG. The Kd values were determined using the concentrations of MG and
RNAs shown in the table. The t1/2 of MG oxidation was measured in the presence of concentrations equivalent to 10-15 × Kd of the MGA variants.
The data for t1/2 are derived from the fitting curves of the averages of four to nine independent experiments (Figure 3). The equations that gave the best
fit were used in each case. For MG alone (none) the equation was first order, and for all others the best fitting equation was second order. The average
residuals for the fits of each plot were <7%. In each independent experiment the relative order of t1/2 values [MGA . MGA(A31C), MGA(U25C) >
MGA(G8C-G24C-G29C), none] was the same as for the average values shown here (Supporting Information Figure S3). The p values for the Tukey
comparisons between the treatments were <10-8, except 0.327 for the comparison between MGA(A31C) and MGA(U25C) and 0.175 for the comparison
between MGA(G8C-G24C-G29C) and none (Supporting Information Tables S1 and S2). ND, not determined. UD, undetectable.
1H NMR Spectra Are Consistent with Proper Folding of
intermediate for the MGA(A31C) and MGA(U25C) and large
the MGA Variants and Binding of MG to Each MGA
for the MGA(G8C-G24C-G29C) (Figure 3 and Supporting
Information Figure S3 and Table 1 and Supporting Information
Tables S1 and S2).
Variant. To determine whether the MGA variants still folded
1
properly and bound MG, H NMR spectra were acquired for
the MGA variants in the presence and absence of MG. The
spectrum of MG alone was also acquired as a control, which
showed sharp peaks in the 6.8-6.9, 7.25-7.35, 7.45-7.50, and
7.60-7.65 ppm regions (Figure 5). These peaks were similarly
broadened in the presence of the MGA and each MGA variant,
indicating a similar binding mode of MG to the MGA variants
as to MGA.
Structural Probing Shows Structurally Altered MG
Binding Pockets in the MGA Variants. To determine whether
there are structural changes in the binding pockets, RNase I
footprinting was performed of the MGA variants in the presence
of various concentrations of MG. The data showed that MG
binding to the MGA protected the binding pocket and desta-
bilized the tetraloop (Figure 4A). The MGA(A31C) showed a
similar cleavage pattern to the MGA except there was increased
cleavage of U32 and C31 (Figure 4B). A similar cleavage pattern
was observed of the MGA(U25C), as for the MGA (Figure 4C).
By contrast, the footprinting pattern of MGA(G8C-G24C-G29C)
showed all regions protected except C29 in three of four
experiments (Figure 4D).
The number of peaks from the MGA variants in the 10-15
ppm region, which references the base-pairings in the MGA
structure, was 9-10 for the MGA and all its variants in the
absence of MG. In the presence of MG, there were 13-14 peaks
for MGA, MGA(A31C), and MGA(U25C) and 7 for MGA(G8C-
G24C-G29C). These results are consistent with the conclusion
9
J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009 14751

A R T I C L E S
Wang et al.
Figure 3. Variations around the MGA binding pocket compromise its
ability to protect MG. MG oxidation in the presence of the MGA variants
at concentrations equivalent to 10-15 times Kd was measured spectropho-
tometrically. The concentrations of interacting components were 3 or 30
µM MG (orange O), 3 µM MG, 10 µM MGA (green 9), 3 µM MG, 20-30
µM MGA(A31C) (blue b), 30 µM MG, 140-200 µM MGA(U25C) (red
[), 30 µM MG, 3 mM MGA(G8C, G24C, G29C) (purple 2). The figure
shows the averages of the results of four to nine independent experiments
with standard deviations for the MGA and each MGA variant. The fitting
curves for MGA(A31C) and MGA(U25C) are identical. All experiments
gave consistent results of the t1/2s following the relationship of MGA.
MGA(A31C), MGA(U25C) > MGA(G8C-G24C-G29C), No RNA.
that, except for MGA(G8C-G24C-G29C), the MGA variants
fold properly into structures similar to MGA that can bind MG
in a defined binding pocket.
In the presence of MG the NMR spectra of the MGA and
MGA variants differed, especially in the 6.5 to 9.0 ppm range
where the proton exchange rates are not affected by pH, salt,
and other buffer conditions. We interpret these results to show
that MGA and its variants fold into similar, but nonidentical,
structures, each of which binds MG.
Figure 4. Footprinting analysis of the binding pockets of the MGA variants.
RNase I footprinting of (A) the MGA with and without increasing MG
concentrations (0.05 to 4 µM), (B) the MGA(A31C) with and without 20
µM MG that was equivalent to 10 times Kd, (C) the MGA(U25C) with and
without 140 µM MG that was equivalent to 10 times Kd, and (D) the
MGA(G8C-G24C-G29C) with and without 3 mM MG that was equivalent
to 10 times Kd. Each experiment was repeated at least twice with similar
results.
Cavity Analysis of Simulated MGA-MG Variants Shows
Routes for OH- Access to the C1 of MG in the Binding
Pockets of the MGA Variants. To determine if MD simulation
of the MGA variants could predict the observed accessibility
of the bound MG to oxidation, we performed MD simulation
followed by cavity analysis of the variants. The MD simulations
reached equilibrium after 0.5-1.5 ns as judged by constant
temperature (298 K), minimal energy, and stable root-mean-
square deviation (rmsd; data not shown). The coordinates of
the MGA-MG variants were allowed to further evolve for
4.2-6.4 ns. The equilibrated structures with high rmsd (espe-
cially the positions of residues 8 and 28 relative to MG) from
the starting NMR structures were chosen for cavity analysis.
Cavity analysis of the simulated MGA-MG showed no gap
for OH- access to C1 of MG inside the MGA binding pocket
as already found for the NMR structure of MGA-MG (Figure
2).Bycontrast,thesimulatedMGA(A31C)-MGandMGA(U25C)-
MG complexes showed routes for OH- access to C1 of MG in
the pocket (Figure 6).
ture, in which the smaller size of C compared to A allows a
shift of C7 and G29 toward C31, exposing the two nitrogenated
phenols of MG. The G24 of MGA(A31C)-MG is not part of
the quadruple and instead is in a position perpendicular to the
two nitrogenated phenols of MG (Figure 6B). This assists in
holding MG in the pocket but allows solvent entry on both sides
of the plane of MG. On the C7/G29 side, C1 is not quite exposed
enough for OH- attack in MGA(A31C)-MG. However, there
is a cavity in this variant in the pocket on the G8/C28 side.
In the MGA pocket, G8 and C28 stack directly over the
nitrogenated phenols of MG preventing access to C1 of MG
(Figure 6A). However, in the simulated MGA(A31C)-MG
structures in which a gap is found by cavity analysis, the shift
in position between MG and the quadruple is associated with a
shift of MG relative to G8-C28 where stacking of G8 and C28
occurs with one nitrogenated phenol and the un-nitrogenated
phenol, leaving the second nitrogenated phenol and C1 exposed
to OH- attack on the G8-C28 side (Figure 6B).
In the NMR and simulated MGA-MG structures, A31 and
G24 interact as a quadruple with base-paired C7 and G29. While
C7, G29, and G24 interact with the plane of MG, base A31
maintains the orientation of the other three members of the
quadruple but does not stack with MG (Figure 6A). This
arrangement changes in the simulated MGA(A31C)-MG struc-
In the simulated MGA(U25C)-MG structure in which a gap
was found, the G29 on the quadruple side of MGA(U25C) is
9
14752 J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009

Coherent yet Malleable Aptamer Structure
A R T I C L E S
1
Figure 5. 1H NMR spectra showed proper folding of the MGA variants and evidence of MG binding. The H NMR spectra includes two regions: 11-15
ppm and 6.5-9.0 ppm.
most involved with MG stacking, although C7, G29, and A31
maintain their quadruple orientations to each other and C1
remains protected on this side. G24 is not a part of the quadruple
in MGA(U25C)-MG; therefore, the quadruple has become a
triple in this MGA variant. On the G8/C28 side, U25 lines a
side of the pocket, perpendicular to MG, without forming
H-bonds in MGA-MG (Figure 6A). However, because of the
charge difference between U and C, the C25 in the simulated
MGA(U25C)-MG is shifted to the G8/C28 side of MG, angled
by ∼45° to MG, enabling hydrogen bonding to A27 and C28.
This interaction again shifts the G8/C28 stacking with the MG
to one nitrogenated phenol and the unmodified phenol, leaving
the other nitrogenated phenol and C1 exposed to OH- attack
on this side (Figure 6C). G24 shifts in position with C25,
decreasing its ability to block MG from leaving the pocket. This
latter observation is consistent with the experimental observation
of a lower affinity of MGA(U25C) for MG.
We characterized the MG oxidation by UV-visible spectro-
photometry, mass spectrometry, and NMR spectrometry. The
results of these studies suggest that the most likely form of
MG-OH contains a carbinol group at position C1. MG
oxidation is also pH-dependent and second order at pH’s below
8. Thus, it appears that the C1 of MG is attacked by an OH- to
form MG-OH. This reaction product was also predicted in a
study of MG oxidation in alkaline solution.37
The MGA inhibits MG oxidation in a concentration-depend-
ent manner with a stoichiometry that suggests steric hindrance
rather than a catalytic mechanism. The specificity of the effect
of the MGA on MG was demonstrated by the observation that
a high concentration of a random sequence RNA pool has little
effect on the rate of MG oxidation. The ability of an oligo-
nucleotide (AS), complementary to the binding pocket of the
MGA, to reverse the effect of the MGA on MG oxidation
showed that the MGA must bind MG to prevent oxidation.
Similar reversibility has been shown in other systems in which
the activities of aptamers are regulated.38,39
Several mechanisms were considered for how the MGA could
protect MG from OH- attack. First, the MGA could alter the
charge on C1 to make it less susceptible to OH- attack. The
MGA has been shown to change the charge distribution across
MG, but physical chemical calculations showed that C1 is more
positively charged in the MGA-MG complex compared to in
solution.31 A more positively charged MG would be more
susceptible to OH- attack rather than less as we have observed.
By this analysis, we excluded charge distribution as the
mechanism by which the MGA prevents MG oxidation. Second,
the aptamer could inhibit MG oxidation by kinetic hindrance.
However, there is no relation between the kinetic parameters
of the MGA variants and the MG hydroxylation rate. Whereas
MGA and MGA(31C) have the same affinities for MG but
different effects on MG oxidation, MGA(31C) and MGA(U25C)
have different affinities for MG but the same effect on MG
oxidation.
Thus, MD simulation followed by cavity analysis of the MG
variants A31C and U25C resulted in predictions, confirmed by
experimental results, that both variants could only partially
protect MG from oxidation and that the A31C variant should
have a higher affinity for MG than the U25C variant.
Discussion
MG is held tightly in the pocket of the MGA that is created
by a bulge between two short stems resulting in a structurally
modified MG.31 Our results show that this interaction is so snug
that it prevents access of a molecule as small as a hydroxyl ion
to the central carbon of MG. During the period that MG sits in
the MGA pocket the MD simulation shows that the ligand and
aptamer move in concert with the result that, despite movement
of the aptamer, the ligand appears continuously protected. When
the ligand is released by the aptamer it is expected to be
susceptible to oxidation. However, there was no evidence of
MG oxidation over a several day period when in the presence
of saturating concentrations of the MGA. The explanation is
likely the kinetics. The MGA-MG interaction is characterized
To test the hypothesis, that the MGA might protect MG from
oxidation by steric hindrance, we carried out a computational
structural analysis of the MGA-MG complex11 and free MG.
The results of these studies showed that the rings A, B, and C
by kon ) 2.32 ( 0.76 × 106 M-1 s-1, koff ) 0.39 ( 0.13 s-1
,
and dissociation half-life ) 1.8 ( 0.83 s (Wang, unpublished
data). By contrast, the MG oxidation reaction half-life is much
longer at 0.70 ( 0.07 h. Thus, although the MGA-MG
interaction involves a dynamic exchange of bound and free MG,
the kinetics of the MGA-MG interaction is too fast to allow
measurable MG modification of the free MG when in the
presence of saturating concentrations of the MGA.
(37) Chen, D.; Laidler, K. Can. J. Chem. 1959, 37, 599–611.
(38) Cong, X.; Nilsen-Hamilton, M. Biochemistry 2005, 44, 7945–7954.
(39) Rusconi, C. P.; Scardino, E.; Layzer, J.; Pitoc, G. A.; Ortel, T. L.;
Monroe, D.; Sullenger, B. A. Nature 2002, 419, 90–94.
9
J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009 14753

A R T I C L E S
Wang et al.
that sits over one side of the MG plane, and the U25C variation
changes the U-turn connecting the top and bottom portions of
the binding pocket. The G8C-G24C-G29C variation changes
the G8-C28 and C7-G29 base pairs between which MG
intercalates and was expected to completely alter the binding
pocket.
1
The H NMR spectra of the MGA and its variants were
consistent with their proper folding, suggesting that the pockets
remained intact albeit with alterations in structure. In the absence
of MG these RNAs showed 9-10 peaks in the 10-15 ppm
region, which corresponds to ∼9-10 base pairings in the MGA
variants. These results for the MGA are consistent with its
reported secondary structure.11 On binding MG, the number of
peaks in the 10-15 ppm region of the 1H NMR spectra
increased to 13-14 in the MGA, MGA(A31C), and MGA(U25C)
and decreased from 10 to 7 in the MGA(G8C-G24C-G29C).
1
Although these results from H NMR spectra do not provide
structural information regarding the interaction of MG inside
the binding pocket of the MGA variants, they are consistent
with the RNase I footprinting results and lead to the interpreta-
tion that the MG-binding pocket is formed in the MGA,
MGA(A31C), and MGA(U25C) but not in the MGA(G8C-
G24C-G29C) that appears to bind MG nonspecifically.
Structural changes in the binding pockets of the MGA variants
are indicated by the following evidence: (1) enhanced RNase I
cleavage of U32 and C31 in the MGA(A31C), (2) changed λmax
of the MG interaction with the MGA(U25C), and (3) a
significantly altered RNase I footprint of the MGA(G8C-G24C-
G29C).
The Kd values of the MGA variants for MG differed widely
with the MGA(G8C-G24C-G29C) showing a significant loss
of affinity for MG. An intermediate change in the Kd was
observed for MGA(U25C), and a negligible change in the Kd
was observed for MGA(A31C). By contrast, the A31C base
exchange decreases the affinity of MGA to TMR significantly.11
This difference in the effect of the A31C base exchange on the
affinity of MGA to MG and TMR may be due to the structural
difference between MG and TMR. MG is less flat and more
flexible than TMR and thus may more easily adapt to an A31C
base exchange in the pocket. Although A31 makes no direct
contact with MG, the A31C exchange changed a component of
the base quadruple (C7-G29-G24-A31) that sits above MG in
the MGA binding pocket. The U25C variation changed the base
in U-turn that connects the top and bottom parts of the binding
pocket.
With MG saturating each RNA (at 10-15 times its Kd) the
MGA(A31C), MGA(U25C), and MGA(G8C-G24C-G29C) were
found to protect MG from oxidation in decreasing order of
effectiveness with the effect of the MGA(G8C-G24C-G29C)
being the same as for free MG. Whereas MGA and MGA(A31C)
had similar affinities for MG but different abilities to protect
MG from oxidation, MGA(A31C) and MGA(U25C) had dif-
ferent affinities for MG but similar abilities to protect MG from
oxidation (Table 1). These results suggest that structural changes
in the MGA variants compromise their abilities to protect MG
from oxidation. Consistent with these observations, the com-
bination of MD simulation followed by cavity analysis showed
that the decreased effectiveness of the MGA variants to protect
MG from oxidation can be explained by rearrangements in the
binding pockets of the MGA variants that allow OH- access to
the C1 of MG in the pockets of MGA variants.
Figure 6. Binding pocket cavity analysis of MGA-MG variants compared
with the original aptamer. (A) The binding pocket bases from the NMR
structure (1Q8N.pdb, model 5) of the MGA are color-coded and demonstrate
the protection that prevents hydroxide access to the C1 carbon (green) of
MG (magenta). In the A31C variant (B) and U25C variant (C), rearrange-
ment of the base quadruple (C7 and G29, orange; G24, blue; A or C31,
red), the C8-G28 (yellow) base pair, and U or C25 (purple) creates a gap
connecting solvent to the MG C1 carbon, which allows hydroxide access
as indicated by the mesh (cyan). Note the hydrogen bonding (green dashed
lines) that occurs in the U25C variant.
of MG could protect MG from OH- attack equatorially but
allow OH- attack from the poles at the top and bottom. Cavity
analysis of the MGA-MG complex showed that the MGA
binding pocket further protected MG from OH- attack from
the poles. Thus, it appeared that the mechanism by which the
MGA protects MG from oxidation is steric hindrance.
To evaluate the prediction made by cavity analysis, we tested
the effects of several MGA variants on MG oxidation. Base
replacements in the MGA binding pocket were chosen that
might alter the pocket sufficiently to allow OH- access to C1
of MG but not so much as to cause collapse of the pocket. The
A31C exchange affects the base quadruple (C7-G29-G24-A31)
In addition to structural rearrangements in the binding pocket,
the variant MGAs were predicted to be more mobile than the
9
14754 J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009

Coherent yet Malleable Aptamer Structure
A R T I C L E S
MGA. Our MD simulations showed that the NMR-derived
MGA interacted with MG with a rmsd (G8, C28, MG) of less
than 0.658 Å during the simulation trajectory at equilibrium,
which is not sufficiently mobile to create a gap for OH- access
to the central carbon of bound MG. However, the MGA(A31C)
and MGA(U25C) are more mobile than MGA. The RMSDs of
(G8, C28, MG) in MGA(A31C) and MGA(U25C) reached up
to 1.107 Å and 1.237 Å, respectively. Along with the structural
alteration, this additional mobility promotes the existence of a
gap in the G8/C28 side for OH- access to the central carbon of
bound MG. It should be noted that the extent of mobility in the
MD simulation of the MG-MGA aptamer depends on the
structure from which the MD simulation starts. Previous MD
simulations of the MGA-MG interaction showed less mobile
interactions of the MG rings and MGA bases during simulation
time courses that were started with the X-ray-derived MGA
structure than those that were started with the NMR-derived
MGA structure.40
In conclusion, in this model system, we have shown that,
when sequestered in the binding pocket of its cognate aptamer,
MG can be protected from attack by a molecule as small as
1.32-1.9 Å. The combined use of experimental and computa-
tional analysis provides a rational explanation of how the
structural features of the aptamer binding pocket prevent
hydroxyl access to the C1 of MG. Computational analysis by
MD simulation and cavity analysis was also able to predict
changes in the binding pockets of MGA variants that could result
in the experimentally observed increase in OH- access to the
C1 of MG. Thus, our results demonstrate that, although the
aptamer structure is flexible and binds its target molecule by
an induced fit process whereby both aptamer and target molecule
change in structure, the resulting complex is sufficiently rigid
to prevent the entry of as small a molecule as a hydroxyl ion.
The capability of an aptamer to wrap so snugly around its target
as to prevent access of even small ions over periods of days
has not been previously reported and might find application in
protecting small molecules from reactive species.
The additional ability to regulate the aptamer structure with
an oligonucleotide that is complementary to the aptamer binding
pocket provides a means of rapidly releasing the sequestered
small molecule. Aptamers can also function inside living cells.
Thus, this novel finding suggests that aptamers might be used
to manipulate molecular reactivity of small molecules both in
vitro and in vivo.
Acknowledgment. Work at the Ames Laboratory was supported
by the Department of Energy-Biological Systems Science Division,
Office of Energy Science Research under Contract No. DE-AC02-
07CH11358. We thank Mark Hargrove and Amy Andreotti for
useful advice and Gaya Amarasinghe for the same and also for
help in interpreting the NMR data for the MGA variants. We also
thank Bruce Fulton in the Biomolecular Structure Facility and
Kamel Harrata and Shu Xu in the Chemical Instrumentation Facility
at Iowa State University for their technical support and advice in
the NMR and mass spectrometry experiments. We thank Aimin
Yan for his help on statistical analysis. We also thank the journal’s
anonymous reviewers for their insightful comments.
Supporting Information Available: pH dependence of MG
oxidation; hydroxylation of MG bleaching; variations around
the MGA binding pocket compromise its ability to protect MG;
half lives (t1/2) of MG in the presence and absence of MGA
and its variants; P values of Tukey multiple comparisons of
half-lives (t1/2) of MG in the presence and absence of MGA
and its variants; topology and parameter files for malachite
green. This material is available free of charge via the Internet
at http://pubs.acs.org.
(40) Nguyen, D. H.; Dieckmann, T.; Colvin, M. E.; Fink, W. H. J. Phys.
Chem. B 2004, 108, 1279–1286.
JA902719Q
9
J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009 14755