Transition state structure of E. coli tRNA-specific adenosine deaminase

Article abstract of DOI:10.1021/ja078008x

Bacterial tRNA-specific adenosine deaminase (TadA) catalyzes the essential deamination of adenosine to inosine at the wobble position of tRNAs and is necessary to permit a single TRNA species to recognize multiple codons. The transition state structure of Escherichia coli TadA was characterized by kinetic isotope effects (KIEs) and quantum chemical calculations. A stem loop of E. coli tRNA^Arg2 was used as a minimized TadA substrate, and its adenylate editing site was isotopically labeled as [1′-³H], [5′-³H₂], [1′-¹⁴C], [6- ¹³C], [6-¹⁵N], [6-¹³C, 6-¹⁵N] and [1-¹⁵N]. The intrinsic KIEs of 1.014, 1.022, 0.994, 1.014 and 0.963 were obtained for [6-¹³C]-, [6-¹⁵N]-, [1-¹⁵N]-, [1′-³H]-, [5′-³H₂]-labeled substrates, respectively. The suite of KIEs are consistent with a late S _NAr transition state with a complete, pro-S-face hydroxyl attack and nearly complete N1 protonation. A significant N6-C6 dissociation at the transition state of TadA is indicated by the large [6-¹⁵N] KIE of 1.022 and corresponds to an N6-C6 distance of 2.0 A in the transition state structure. Another remarkable feature of the E. coli TadA transition state structure is the Glu70-mediated, partial proton transfer from the hydroxyl nucleophile to the N6 leaving group. KIEs correspond to H-O and H-N distances of 2.02 and 1.60 A, respectively. The large inverse [5′-³H] KIE of -3.7% and modest normal [1′-³H] KIE of 1.4% indicate that significant ribosyl 5′-reconfiguration and purine rotation occur on the path to the transition state. The late S_NAr transition-state established here for E. coli TadA is similar to the late transition state reported for cytidine deaminase. It differs from the early S_NAr transition states described recently for the adenosine deaminases from human, bovine, and Plasmodium falciparum sources. The ecTadA transition state structure reveals the detailed architecture for enzymatic catalysis. This approach should be readily transferable for transition state characterization of other RNA editing enzymes.

Full text of DOI:10.1021/ja078008x

Published on Web 02/02/2008
Transition State Structure of E. coli tRNA-Specific Adenosine
Deaminase
Minkui Luo and Vern L. Schramm*
Department of Biochemistry, Albert Einstein College of Medicine, Bronx, New York 10461
Received October 18, 2007; E-mail: vern@aecom.yu.edu.
Abstract: Bacterial tRNA-specific adenosine deaminase (TadA) catalyzes the essential deamination of
adenosine to inosine at the wobble position of tRNAs and is necessary to permit a single tRNA species to
recognize multiple codons. The transition state structure of Escherichia coli TadA was characterized by
kinetic isotope effects (KIEs) and quantum chemical calculations. A stem loop of E. coli tRNAArg2 was used
as a minimized TadA substrate, and its adenylate editing site was isotopically labeled as [1′-3H], [5′-3H2],
[1′-14C], [6-13C], [6-15N], [6-13C, 6-15N] and [1-15N]. The intrinsic KIEs of 1.014, 1.022, 0.994, 1.014 and
0.963 were obtained for [6-13C]-, [6-15N]-, [1-15N]-, [1′-3H]-, [5′-3H2]-labeled substrates, respectively. The
suite of KIEs are consistent with a late SNAr transition state with a complete, pro-S-face hydroxyl attack
and nearly complete N1 protonation. A significant N6-C6 dissociation at the transition state of TadA is
indicated by the large [6-15N] KIE of 1.022 and corresponds to an N6-C6 distance of 2.0 Å in the transition
state structure. Another remarkable feature of the E. coli TadA transition state structure is the Glu70-
mediated, partial proton transfer from the hydroxyl nucleophile to the N6 leaving group. KIEs correspond
to H-O and H-N distances of 2.02 and 1.60 Å, respectively. The large inverse [5′-3H] KIE of -3.7% and
modest normal [1′-3H] KIE of 1.4% indicate that significant ribosyl 5′-reconfiguration and purine rotation
occur on the path to the transition state. The late SNAr transition-state established here for E. coli TadA is
similar to the late transition state reported for cytidine deaminase. It differs from the early SNAr transition
states described recently for the adenosine deaminases from human, bovine, and Plasmodium falciparum
sources. The ecTadA transition state structure reveals the detailed architecture for enzymatic catalysis.
This approach should be readily transferable for transition state characterization of other RNA editing
enzymes.
Introduction
Posttranscriptional RNA editing is the modification of RNA
sequence that is distinct from RNA splicing, capping or 5′ end
processing. Both tRNA-specific adenosine deaminase (TadA)
and adenosine deaminase that acts on dsRNA (ADAR) operate
on complex RNA substrates (Figure 1).1-4 Posttranscriptional
tRNA editing involves TadA-catalyzed adenosine to inosine
conversion (A-to-I) at the wobble position.5,6 The TadA activity
is essential for a single tRNA species to recognize multiple
codons since inosine is recognized as guanosine in base
pairing.3,4 ADARs carry out A-to-I conversion on dsRNA, and
this editing is prevalent in mammals and occurs in up to 10%
of all mRNAs containing inosine.7 This editing event can alter
the primary sequence codon, splice sites and structures of edited
RNA because inosine is recognized as guanosine in most cellular
processes. ADAR activities are necessary for neuron receptor
Figure 1. Structures of ectRNAarg2 and its truncated stem loop. The
structure of full-length ectRNAarg2 is shown in the left. The anticodon region
is highlighted with red and blue, and the TadA-editing site is red. The
truncated ectRNAarg2 stem loop is shown to the right with the same scheme.
ecTadA targets the A34 sites of these constructs with similar catalytic activity.
Adenosine A34 is converted to inosine (I34) by the deamination reaction.3,5
(1) Elias, Y.; Huang, R. H. Biochemistry 2005, 44, 12057-12065.
(2) Kuratani, M.; Ishii, R.; Bessho, Y.; Fukunaga, R.; Sengoku, T.; Shirouzu,
M.; Sekine, S.; Yokoyama, S. J. Biol. Chem. 2005, 280, 16002-16008.
(3) Kim, J.; Malashkevich, V.; Roday, S.; Lisbin, M.; Schramm, V. L.; Almo,
S. C. Biochemistry 2006, 45, 6407-6416.
(4) Losey, H. C.; Ruthenburg, A. J.; Verdine, G. L. Nat. Struct. Mol. Biol.
2006, 13, 153-159.
(5) Wolf, J.; Gerber, A. P.; Keller, W. EMBO J. 2002, 21, 3841-3851.
(6) Gerber, A. P.; Keller, W. Science 1999, 286, 1146-1149.
(7) Bass, B. L. Annu. ReV. Biochem. 2002, 71, 817-846.
diversification8,9 and hepatitis delta virus replication.10,11 For
the latter, A-to-I editing switches an amber stop codon (UAG)
to a tryptophan (UIG) for expressing fully functioned proteins.
9
10.1021/ja078008x CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 2649-2655
2649

A R T I C L E S
Luo and Schramm
Figure 2. TadA-mediated adenosine deamination. The initial step proposed
for TadA-catalyzed demination is formation of Zn2+-hydroxyl. The nearby
Glu residue serves as the general base to remove proton from Zn-bound
water. The structures of guanosine and the product inosine are shown.
Inosine in edited RNA is recognized as guanosine.
Figure 3. Measurement of forward commitment factor of ecTadA.18,19 The
isotope trapping experiment for measuring the forward commitment was
performed under rapid-mixing pre-steady-state conditions. Enzyme was
bound to the labeled substrate in a 2 ms mix and was then diluted into a
large excess of unlabeled cold substrate. The ratio of IMP formed to enzyme-
bound substrate was plotted vs time. The forward commitment factor (Cf)
was calculated from the ordinate intercept, fit to a linear equation with Cf
) 0.15.
Recent studies on human ADAR substrates further led to
identifying a new A-to-I editing target BC10 that has been linked
to bladder cancer and renal cancer proliferation. hADAR is also
characterized as one of few unambiguously up-regulated genes
in solid tumors and liver cancer.12-14 These pieces of evidence
strongly relate ADAR upregulation to cancer progression.
ADARs also process microRNAs and therefore alter expression
or target specificity of the microRNAs.15,16 Our interest in
transition-state structure led us to solve the transition state
structure of TadA, an RNA editing enzyme which targets a
structurally complex and biologically relevant tRNAs from
Escherichia coli (Figures 1 and 2).
KIEs are one of the few tools to characterize enzyme
transition states in structural detail.17-20 The elucidation of
enzyme transition state structure is valuable in the development
of potent inhibitors. Molecules that mimic enzymatic transition
states are expected to show Kd values of 10-18-10-21 M,
proportional to the enzyme-imposed rate enhancement of
catalysis.21 We have applied this approach to achieve inhibitors
with pM to fM affinities for N-ribosyltransferase enzymes.21
Here we apply similar approaches to probe the more complex
transition state structure of TadA. The results are of potential
utility for the design of transition state analogue inhibitors in
the family of RNA editing enzymes.
1.014, 1.022, and 0.993 were obtained for primary [6-13C]-,
[6-15N], and secondary [1-15N] positions, respectively. This suite
of KIEs was used as constraints to model a transition state for
ecTadA. ecTadA adopts a late SNAr transition state with a
complete, pro-S-face hydroxyl addition, nearly complete N1
protonation and significant N6-C6 dissociation. The rate-
limiting step of the ecTadA-catalyzed deamination is subsequent
to the formation of the tetrahedral Meisenheimer intermediate.
Computational analysis of the reaction supports a transition state
with a proton shuttle between the hydroxyl nucleophile and the
leaving group N6, mediated by nearby Glu70. Significant
distortion of the 5′-sugar and rotation of the ribosidic purine
on the path to transition state formation were indicated from
the large inverse [5′-3H2] KIE of -4% and significant [1′-3H]
KIE of 1.4%. This is the first time that KIEs have been
correlated with purine base rotation at an RNA-based transition
state. We propose that this KIE approach will be useful for
transition state and base rotation characterization of other RNA
editing enzymes. The ecTadA transition state structure reveals
the detailed architecture for catalytic efficiency and provides a
blueprint for designing tight-binding inhibitors.
Kinetic isotope effects are reported for the deamination of
RNA by E. coli TadA using substrates whose editing sites were
labeled with [6-13C], [6-15N], [1-15N], [5′-3H2], and [1′-3H]. The
transition state structure of ecTadA was explored with KIE-
constrained quantum chemical calculations. Intrinsic KIEs of
Experimental Section
(8) Macbeth, M. R.; Schubert, H. L.; VanDemark, A. P.; Lingam, A. T.; Hill,
C. P.; Bass, B. L. Science 2005, 309, 1534-1539.
Expression and Purification of ecTadA. A 13aa-truncated ecTadA
was expressed and purified as described previously with some
modification.3 Here the ecTadA was subject to two additional MonoQ
columns (GE Science) and a Superdex75 purification (GE Science) to
lower RNase activities.
Synthesis of Isotopically Labeled Substrates. [1′-3H]-, [5′-3H2]-,
[1′-14C]-, [1′-14C, 6-13C], [1′-14C, 6-15N], [1′-14C, 6-13C15N], and [1′-
14C, 1-15N]-labeled ATPs were prepared enzymatically as described
previously.18,19 The isotopically labeled stem loops were synthesized
by incorporating labeled ATPs via the T7 RNA polymerase reaction
(MEGAshort script T7 kit, Ambion). See Supporting Information for
detailed information.
(9) Tonkin, L. A.; Saccomanno, L.; Morse, D. P.; Brodigan, T.; Krause, M.;
Bass, B. L. EMBO J. 2002, 21, 6025-6035.
(10) Wong, S. K.; Lazinski, D. W. Proc. Natl. Acad. Sci. U.S.A. 2002, 99,
15118-15123.
(11) Sato, S.; Wong, S. K.; Lazinski, D. W. J. Virol. 2001, 75, 8547.
(12) Pilarsky, C.; Wenzig, M.; Specht, T.; Saeger, H. D.; Grutzmann, R.
Neoplasia 2004, 6, 744-750.
(13) Midorikawa, Y.; Tsutsumi, S.; Taniguchi, H.; Ishii, M.; Kobune, Y.;
Kodama, T.; Makuuchi, M.; Aburatani, H. Jpn. J. Cancer Res. 2002, 93,
636-643.
(14) (a) Clutterbuck, D. R.; Leroy, A.; O’Connell, M. A.; Semple, C. A.
Bioinformatics 2005, 21, 2590-2595. (b) Gromova, I.; Gromov, P.; Celis,
J. E. Int. J. Cancer 2002, 98, 539-546.
(15) Kawahara, Y.; Zinshteyn, B.; Sethupathy, P.; Iizasa, H.; Hatzigeorgiou,
A. G.; Nishikura, K. Science 2007, 315, 1137-1140.
(16) Yang, W. D.; Chendrimada, T. P.; Wang, Q. D.; Higuchi, M.; Seeburg, P.
H.; Shiekhattar, R.; Nishikura, K. Nat. Struct. Mol. Biol. 2006, 13, 13-21.
(17) Cleland, W. W. Arch. Biochem. Biophys. 2005, 433, 2-12.
(18) Luo, M.; Singh, V.; Taylor, E. A.; Schramm, V. L. J. Am. Chem. Soc.
2007, 129, 8008-8017.
(19) Singh, V.; Schramm, V. L. J. Am. Chem. Soc. 2007, 129, 2783-2795.
(20) Schramm, V. L. Curr. Opin. Struct. Biol. 2005, 15, 604-613.
(21) Schramm, V. L. Arch. Biochem. Biophys. 2005, 433, 13-26.
Measurement of Commitment Factor. Forward commitment for
ecTadA-catalyzed deamination was determined by an isotope trapping
method under rapid-mixing presteady-state conditions (Figure 3).18,19
Briefly, 25.1 µL of 30 µM ecTadA stock was rapidly mixed with 20.9
µL of 36.5 µM [1′-3H]-labeled ectRNAarg2 stem loop substrate (total 2
× 105 cpm) for 2 ms using a quench flow apparatus (PQF-3, KinTek).
9
2650 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

E. coli tRNA-Specific Adenosine Deaminase
A R T I C L E S
The reaction mixture was then rapidly injected into a chase solution
containing 40 µM unlabeled substrate to give the final volume of 1
mL. Subsequently, 100 µL aliquots were collected every 30 s for 2.5
min and quenched with 50 µL of 1 N HCl for 3 min, followed by
neutralization with 50 µL of 1 N KOH. The subsequent chase solutions
were subject to pH adjustment, P1 digestion, and HPLC purification
(Supporting Information), to resolve IMP as the reaction products. The
total cpm of IMP generated after the chase step was corrected for
background counting, obtained from a control in the absence of enzyme,
and the cpm of IMP generated during mixing, calculated on the basis
of kcat ) 13 min-1, the enzyme concentration and the mixing time.3
The amount of enzyme-bound substrate prior to the chase step was
calculated on the basis of Km ) 0.83 µM3 and the enzyme concentration.
The ratios of [1′-3H]IMP product to ecTadA-bound substrate were
plotted vs time t. The forward commitment factor Cf was obtained from
the ordinate intercept upon extrapolation to zero time (Figure 3, forward
commitment) intercept/(1 - intercept)). The ecTadA-catalyzed deami-
nation is irreversible and therefore its reverse commitment factor is
zero. Consequently, eq 1 was used for commitment factor correction,
in which X(V/K) is the observed heavy atom KIE, Xk is the intrinsic
KIE, XK is the equilibrium isotope effect between substrate and product,
Cf and Cr stand as the forward and reverse commitment, respectively.
Cr is expected to be zero in this case.
Figure 4. Isotopic reporters for transition state structure. The magnitude
of the [6-13C], [6-15N], [1-15N], [5′-3H2], and [1′-3H] KIEs reflect the degree
of C6 hybridization, N6 dissociation, N1 protonation, 5′-ribosyl distortion,
and ribosidic purine ring rotation at the TadA transition state, respectively.
3
Color codes for isotopic atoms: 13C, yellow; 15N, blue; H, red.
Xk + Cf + Cr × Keq
X
X(V/K) )
(1)
1 + Cf + Cr
KIE Determination. All ecTadA-catalyzed deamination reactions
were carried out at 25 °C in a buffer containing 10 mM Tris-HCl, 10
mM KCl, 5% glycerol (v/v), 25 µM of the stem loop substrate (labeled
and unlabeled), 100 unit/mL RNasin Plus RNase inhibitor (Promega),
and 100 nM ecTadA. Apparent KIEs were determined by the competi-
tive radiolabeled method (Supporting Information).18,19
Figure 5. KIEs for TadA-catalyzed deamination. All reported KIEs are
intrinsic following commitment factor correction. Details of the KIE and
forward commitment factor measurements are provided in the Supporting
Information. The calculated [6-13C, 15N] KIE is the product of [6-13C] and
[6-15N] KIEs and is consistent with the intrinsic [6-13C, 15N] KIE obtained
experimentally. Color codes for isotopic atoms are the same as described
in Figure 4.
Computational Modeling. With the intrinsic KIEs as constraints,
the ecTadA transition state structure was determined in Vacuo using
hybrid density functional methods with B3LYP/6-31G (d,p) level of
theory implemented in Gaussian 98 (Supporting Information). We used
9-methyladenine as a truncated substrate and a hydroxyl anion as the
nucelophile for calculating the [6-13C], [6-15N], and [1-15N] KIE values.18
The transition state structures were optimized from a fully formed
Meisenheimer intermediate by varying C6-Ohydroxyl, C6-N6 bond
distances as well as the degree of Ohydroxyl, N1 and NH2-6 protonation.
Bond frequencies for the substrate and transition states were calculated
and used as the inputs to the ISOEFF98 program for KIE determina-
tion.22 The remote isotope effects were calculated using 3′-phosphate-
5′-methylphosphate adenosine and its C6 hydroxyl adduct for the [5′-
3H2] and [1′-3H] KIE values. Their initial geometry (reactant) was
extracted from the NMR structure reported for ectRNAPhe G34 residue
(PDB, 1KKA).23 Structural candidates for the transition state were
generated on the basis of TadA-bound RNAarg2 A34 analogue (PDB,
2B3J).4 with its purine base replaced with 6-hydroxyl adenine (9-
methyladenine mimic of the transition state described above). Structural
candidates were generated by imposing constraints or varying the
dihedral angles that influence the [1′-3H] and [5′-3H2] KIEs. The
structures were optimized at the B3LYP/6-31G (d,p) level of theory.
Bond vibrational frequencies and KIEs were calculated as described
above. The structure of the TadA-bound Michaelis complex was
modeled by releasing all constraints except the C8-N9-C1′-O4
dihedral angle of 73.5° based on the structure of the residue 34 of TadA-
bound RNAarg2.4 The structure of the Meisenheimer intermediate was
optimized by restricting the C8-N9-C1′-O4 dihedral angle to 45.0°
and the N1-H distance to 1.10 Å.
Results and Discussion
Synthesis of ecTadA Substrate. To determine KIEs of
TadA-catalyzed deamination, we prepared the stem loop of E.
coli tRNAarg2 (ectRNAarg2) as a minisubstrate and its editing
site was isotopically labeled with [1′-3H], [5′-3H2], [1′-14C],
[6-13C], [6-15N], [6-13C, 6-15N], and [1-15N] (Figures 1 and S1).
The truncated ectRNAarg2 was shown to be as active as full-
length ectRNAarg2 for TadA editing.1,3,5 Here the KIEs at C6,
N6, and N1 positions report on C6 hybridization, extent of N6
dissociation and the degree of N1 protonation at the transition
state. The [1′-3H] and [5′-3H2] KIEs are expected to be sensitive
to the adenine-ribose glycosyl torsion angle (ring rotation) and
any ribosyl distortion between free reactant and the TadA
transition state (Figure 4).
Determination of Intrinsic KIEs of TadA-Catalyzed
Deamination Reaction. Intrinsic [1′-3H], [5′-3H2], [6-13C],
[6-15N], and [1-15N] KIEs were measured for the TadA-catalyzed
deamination (Figure 5). Vmax/Km 3H KIE experiments were
carried out under competitive conditions18 with the [1′-14C]
labeled substrate as the silent competitive isotopic pair (see
Experimental Section). Apparent 13C and 15N KIEs were
measured using [1′-14C] as the isotopically silent remote label
for heavy isotopic substrates (13C or 15N) and [1′-3H] as their
(22) Anisimov, V.; Paneth, P. J. Math. Chem. 1999, 26, 75-86.
(23) Cabello-Villegas, J.; Winkler, M. E.; Nikonowicz, E. P. J. Mol. Biol. 2002,
319, 1015-1034.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2651

A R T I C L E S
Luo and Schramm
Table 1. Intrinsic KIEsa vs Computational KIEsb of ecTadA
intrinsic KIEs
vs computational KIEsd,e
intrinsic
values
computational
substrate pairsc
type of KIEs
values
[6-13C, 1′-14C]
vs [1′-3H]
R-primary
13C
1.014 ( 0.003
1.022 ( 0.002
0.994 ( 0.002
1.036 ( 0.004
1.036 ( 0.005
0.963 ( 0.002
1.014 ( 0.001
1.014
[6-15N, 1′-14C]
vs [1′-3H]
R-primary
1.023
0.994
1.037
1.037
0.966
1.012
15N
[1-15N, 1′-14C]
vs [1′-3H]
â-secondary
15N
[6-13C, 6-15N, 1′-14C]
vs [1′-3H]
R-primary
13C15N
13C15N calcdf
R-primary
13C15N
[1′-14C]
remote 3H
vs [5′-3H2]
[1′-14C]
remote 3H
vs [1′-3H]
Figure 7. Stereospecific C6 nucleophilic addition. The Zn2+-chelated
hydroxyl of ecTadA attacks the substrate C6 center from the pro-S face in
contrast to the stereospecific pro-R hydroxyl addition for Plasmodium
falciparum, human and bovine adenosine deaminases.18 Left, Meisenheimer
intermediates; right, MEP of transition states (9-methyladenine mimics used
here).
a Intrinsic KIEs were obtained upon correcting the remote [1′-3H] KIE
for13k,15k,13,15k, and the commitment factor. b Computational KIEs were
calculated from the frequencies of the substrate and the transition state pairs
using ISOEFF98.30 [6-13C], [6-15N], and [1-15N] KIEs were calculated with
9-methyladenine as a substrate mimic and its hydroxyl adduct as a transition
state mimic. 3′-Phosphate-5′-methylphosphate adenosine and its C6 hydroxyl
adduct were chosen for [5′-3H2] and [1′-3H] computation. c Isotopically
labeled positions are indicated. d At least 4 measurements were made for
each experimental KIE and standard errors are calculated from the variation
between experiments. e Intrinsic values are from experiencetal KIEs and
the “computational values” are the calculated KIE values. f Calculated from
the product of [6-13C] and [6-15N] KIEs.
Figure 8. Base rotation at the ecTadA transition state. 5′-Ribosyl distortion
and purine ring rotation occur on the path to formation of the ecTadA
Meisenheimer intermediate and the transition state. (Top) Comparison of
unbound substrate and Meisenheimer intermediate A34 conformations (C,
green; H, gray; O, red; N, blue; P, brown). (Bottom) Cartoon for the base
rotation process. The two cartoon structures represent the stem loop regions
of ectRNAPhe (PDB, 1KKA)23 and TadA-bound RNAarg2 (PDB, 2B3J),4
respectively. The structures were directly generated from the corresponding
PDB files.
Figure 6. ecTadA transition state structure. The ecTadA transition state
structure was solved using quantum chemical calculations at B3LYP/6-
31G(d, p) level of theory (Gaussian98) with the intrinsic KIEs as constraints.
[6-13C], [6-15N], and [1-15N] KIEs were adequate to give a unique structure
for the deamination chemistry at the transition state. Geometry for the 5′-
ribosyl and purine orientation of substrate and the ecTadA transition state
were selected to match the [5′-3H2]-, and [1′-3H] KIEs and also to be the
most similar to the structures of tRNAPhe and tRNAarg2 reported from NMR
and X-ray crystallography.4,23 The interaction distances and dihedral angle
variations reflect the altered geometries from free substrate and the ecTadA
transition state. Color codes for isotopic atoms are the same as described
in Figure 4.
1.036 ( 0.004, in good agreement with the product of the
corresponding primary [6-13C] and [6-15N] KIEs (Figure 5).
Computational Modeling of ecTadA Transition State
Structure. We applied the intrinsic KIEs as constraints to match
ecTadA transition state structures modeled by Gaussian98 at
the B3LYP/6-31G(d,p) level (see the Experimental Section).
To compute ecTadA transition state structures, 9-methyladenine
was used as a reactant mimic and a hydroxyl anion was used
competitive isotopic pair. Vmax/Km KIEs of [6-13C], [6-15N], and
[1-15N] were obtained by correcting the corresponding apparent
KIEs with the remote [1′-3H] KIE (Table 1). All KIEs were
further subject to correction for forward commitment, which
was determined to be Cf ) 0.15 under an isotope trapping
condition (Figure 3). The small value of Cf indicates that the
Michaelis complex of TadA converts to product at slower rate
than release and equilibration with unbound ectRNAarg2 stem
loop. Pre-equilibration kinetics are consistent with the modest
kcat ) 13 min-1 for this enzyme.18,19 The intrinsic KIEs of [1′-
3H], [5′-3H2], [6-13C], [6-15N], and [1-15N] are 1.014 ( 0.001,
0.963 ( 0.002, 1.014 ( 0.003, 1.022 ( 0.002, and 0.994 (
0.002, respectively (Figure 5). The accuracy of this set of KIEs
was further validated by comparing the [6-13C, 15N] KIE of
as the nucelophile.18 Upon systematically varying C6-Ohydroxyl
,
C6-N6 bond distances as well as the degree of Ohydroxyl, N1,
and NH2-6 protonation, a single transition state structure was
found to match the intrinsic [6-13C], [6-15N], and [1-15N] KIEs
of ecTadA (Table 1, Supporting Information) and therefore
represents the best estimate of transition state structure for the
TadA-catalyzed deamination (Figure 6). In contrast to the early
SNAr character of nucleotide adenosine deaminase (ADA),18
ecTadA adopts a late SNAr transition state with a complete, pro-
9
2652 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

E. coli tRNA-Specific Adenosine Deaminase
A R T I C L E S
Figure 9. Steric view of docking structures of the Meisenheimer intermediate (left) and the transition state (right) at enzyme active site. Lines, His68,
Cys98, and Cys101; sticks, Glu70, intermediate, and transition state; green, carbons; blue, nitrogens; red, oxygens; yellow, sulfurs; blue sphere, Zn; gray
sphere, the shuttled proton; dashed line, key interaction distances for enzyme catalysis labeled in angstroms. The structures were generated by docking the
solved structures of the ecTadA Meisenheimer intermediate and transition state structure into the enzyme active site (PDB, 1Z3A). ecTadA was first aligned
with S. aureus TadA (PDB, 2B3J) and then the nebularine ribosyl moiety was used as a guide to overlay with the ribosyl moieties of the ecTadA Meisenheimer
intermediate and the transition state. PyMOL (W. L. DeLano, PyMOL Molecular Graphics System) was used for illustration.
position.18,19 However, remote 3H substitution can be influenced
by local geometrical distortion at enzyme-bound transition
states.19,24
The 5′-ribosyl distortion at the ecTadA transition state was
explored by computational chemistry for altered 5′-geometries
that generate the 0.963 inverse [5′-3H2] KIE. Pairs of substrate
and transition state candidates for [5′-3H2] computation were
generated by altering the [5′-3H2] KIE-sensitive dihedral angles
at the corresponding ecTadA-editing site on the basis of tRNA
structures (see the Experimental Section).4,23 The structures that
match the experimental KIE show a characteristic O5-C5′-
C4′-O4 dihedral angle of -68° for unbound substrate and of
172° for the TadA-bound transition state. This variation corre-
sponds to a 120° rotation of the adenosine-A34 5′-hydroxyl
moiety at the transition state relative to the unbound substrate
(Figure 8). Therefore, ecTadA binding to RNA induces a
conformational change at the stem loop region of the substrate.
This interaction is essential for ecTadA-mediated A-to-I editing
because it makes the A34 residue, in particular its purine ring,
fully exposed and thus provides ready access into the enzyme
active site (Figure 8-10). For the 5′-ribosyl group, the KIE-
derived geometries of the substrate and ecTadA transition state
are similar to but not the same as in structures of unbound
tRNAPhe and TadA-bound tRNAarg2 analogue, which have been
solved by NMR and X-ray crystallography, respectively (Sup-
porting Information).4,23 The structural difference stems from
varied dihedral angles (less than 20 degrees) around the A34
5′-ribosyl region and are within the resolution limitation of NMR
and X-ray structures (Supporting Information).
Remote [1′-3H] KIEs. The ecTadA-mediated deamination
reaction also displays an unexpected but significant [1′-3H] KIE
of 1.4%. This remote KIE can arise from the C1′-N9 bond
rotation.19 Every 1-2% KIE corresponds to 10° rotation of the
purine ring from its geometry in free reactant to enzyme-bound
transition state.19 We interpreted the 1.4% [1′-3H] KIE as a
signal of the purine rotation around the ribosidic bond that
occurs prior to the pro-S-face hydroxyl attack (the same
argument as for [5′-3H2] KIE above). Computational modeling
results indicated that alteration of the C8-N9-C1′-O4 dihedral
angle from 39° for unbound substrate to 45° at the transition
state accounts for the observed [1′-3H] KIE (Figure 8-10). As
demonstrated by the docking models (Figure 9 and results
below), this C1′-N9 purine rotation places the Zn-bound
Figure 10. Molecular electrostatic potential (MEP) surfaces of stepwise
structures. MEP surfaces of substrate, Michaelis complex, Meisenheimer
intermediate, and the ecTadA transition state were generated using the
CUBE subprogram of Gaussian 98 and visualized using Molekel 4.0 at a
density of 0.15 electron/Å3. The [5′-3H2] and [1′-3H] KIEs are proposed to
be generated as the reaction processes from the unbound substrate to the
ecTadA-stabilized Michealis complex.
S-face hydroxyl attack and nearly complete N1 protonation
(Figure 7). The N1 protonation at the transition state is consistent
with a [1-15N] KIE of 0.994 and a N1-H distance of 1.10 Å.
A fully protonated N1 at the transition state corresponds to a
[1-15N] KIE of 0.989 and an N1-H distance of 1.05 Å.
Significant N6-C6 dissociation at the ecTadA transition state
is reported by the large [6-15N] KIE of 1.022 and corresponds
to a N6-C6 distance of 2.0 Å at the transition state. A unique
feature of the transition state is the partial proton transfer from
the attacking hydroxyl nucleophile to the 6-NH2 leaving group
with a H-O distance of 2.02 Å and a H-N distance of 1.60 Å,
respectively.
Remote [5′-3H2] KIEs. The remote [5′-3H2] KIE is a
remarkable inverse -3.7% (0.963 ( 0.002) for ecTadA (Figure
5). This large remote KIE arises as a consequence of 5′-ribosyl
distortion at the ecTadA editing site on the path to forming the
transition state. We propose that this 5′-ribosyl distortion occurs
prior to the pro-S-face hydroxyl attack because the purine base
rotation into TadA active site is a prerequisite step for the
subsequent editing (see discussion below) and KIEs report all
the steps between free reactants and the first chemically
irreversible step, in this case, the hydrolytic deamination
following transition state formation. The 5′-tritium of the
reactive adenosine is eight bonds from TadA reaction center
and is not expected to be sensitive to deamination at the C6
(24) Birck, M. R.; Schramm, V. L. J. Am. Chem. Soc. 2004, 126, 6882-6883.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2653

A R T I C L E S
Luo and Schramm
Figure 11. Stepwise mechanism proposed for the ecTadA-mediated A-to-I editing. The transition state is the highest energetic barrier along the reaction
coordinate. The overall energy profile is relative and the corresponding structures are displayed with the arrows for bond rotation or electron flow in each
step. The Zn2+-chelated hydroxyl and the proton of the catalytically essential Glu70 are highlighted.
hydroxyl adjacent to C6 of the edited A34 residue and therefore
facilitates the subsequent hydroxyl nucleophilic addition. The
KIE-derived C8-N9-C1′-O4 dihedral angle at the ecTadA
transition state (computed to be 45°) is around 30° smaller than
found for the chemically unreactive complex of the TadA-bound
tRNAarg2 analogue (73°).4 The TadA-bound tRNAarg2 analogue
has been proposed to mimic the enzyme-bound substrate
(Michaelis complex).4 However, variation of the C8-N9-C1′-
O4 dihedral angles from 39 to 45° are sufficient to explain
adenine exposure and the transition state geometry for ecTadA
(Figure 10).
0.8 Å at the ecTadA transition state. The 5′-ribosyl distortion
and C1′-N9 ribosidic purine rotation (results above) also
contribute to the position of the Zn2+-hydroxyl. In the Meisen-
heimer complex the O6hydroxyl is shifted by 40° away from the
Zn2+-coordinated water in the ecTadA crystal structure, reflect-
ing reorientation of the hydroxyl nucelophile to form the
intermediate.
Stepwise Reaction Mechanism of ecTadA-Catalyzed Deam-
ination Reaction. A stepwise reaction mechanism for ecTadA
is based on the structures of the TadA-bound substrate analogue
and the ecTadA transition state structure reported here (Figure
10, 11). Formation of the Michaelis complex is accompanied
by a 120° distortion at the A34 5′-region of the RNA and
generates an inverse 4% [5′-3H2] KIE. The purine rotates 34°
based on the C8-N9-C1′-O4 dihedral of 39° for the unbound
substrate relative to 73° in the Michaelis complex. The ecTadA-
ectRNAarg2 Michaelis complex shows an additional 28° purine
rotation (C8-N9-C1′-O4 dihedral angle ) 45°). Meanwhile,
attack of the Zn2+-chelated water at C6 of A34 is accompanied
by Glu70-mediated N1 protonation, to form the tetrahedral
Meisenheimer intermediate. These steps generate a [1′-3H] KIE
of 1.014 and a [1-15N] KIE of 0.994. The highest chemical
barrier on the reaction coordinate defines the transition state
and involves the Glu70-facilitated proton shuttle from the
Ohydroxyl to N6, reflected by [6-13C] and [6-15N] KIEs of 1.014
and 1.022, respectively. This is followed by N6 dissociation in
a step after the transition state formation.
Comparison of the Transition State Structures of ecTadA
and Other Deaminases. Two features distinguish the transition
state structure of ecTadA from those of adenosine deaminases
(ADA) but relate the ecTadA to cytidine deaminase (CDA).
First, both CDA and ecTadA adopt late SNAr transition states
following the formation of the Meisenheimer intermediate.25
This transition state is characterized by complete hydroxyl-C6
bond formation, nearly complete N1 protonation, and partial
N6 amino group dissociation. In contrast, the transition states
Docking the Transition State Structure Into the Active
Site of ecTadA. Crystal structures of bacterial TadAs reveal
RNA structural motifs in relation to the Zn2+ binding pocket
for the deamination reaction.1-4 Models of the ecTadA Meisen-
heimer intermediate and the transition state structure were
docked into the active site of substrate-analogue-bound Sta-
phylococcus aureus TadA to explore the architecture for ecTadA
catalysis (Figure 9). The docking model shows that ecTadA
Glu70 is within hydrogen bond distances to N1, N6 and Ohydroxyl
in the docking models (Figure 9). Consequently, the Glu70
residue can interact with both the N1 amide proton and serve
as a proton shuttle between the N6-Ohydroxyl. Docking models
with the transition state support the process. The possibility that
N6 proton was transferred from nearby water rather than the
Ohydroxyl was ruled out because the pro-S-face hydroxyl attack
makes N6 buried into a solvent-exclusive pocket (docking
structure not shown). The interaction between ecTadA Glu70
and the N1 amide bond also supports its function as a proton
donor for the formation of N1 amide bond at the step of Zn2+
-
activated hydroxyl nucleophilic addition (discussion below and
Figure 11).
The catalytically essential zinc ion coordinates the Ohydroxyl
with distances of 2.2 Å and 3.0 Å in the ecTadA Meisenheimer
intermediate and transition state models, respectively. The Zn2+
-
chelated O6hydroxyl of the Meisenheimer intermediate is posi-
tioned for a reversible nucleophilic attack at C6 of the adenine.
In contrast, the Zn2+-Ohydroxyl bond has weakened an additional
(25) Snider, M. J.; Reinhardt, L.; Wolfenden, R.; Cleland, W. W. Biochemistry
2002, 41, 415-421.
9
2654 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

E. coli tRNA-Specific Adenosine Deaminase
A R T I C L E S
of ADAs show significant early SNAr character with incomplete
N1 protonation and no C6-N6 dissociation.18 More remarkably,
the Zn2+-hydroxyls of ecTadA and CDA approach their targeted
bases from the pro-S-face in contrast to the pro-R-face attack
for ADAs (Figure 7). The stereospecific hydrolysis of these
enzymes is consistent with the structures of enzyme-bound
nonhydrolyzable substrate analogues2,26,27 and stereospecific
transition-state analogue inhibitors.18,28,29 In spite of chemistry
most similar to adenylate deamination reactions, the distinct late
SNAr transition state and stereospecificity of TadA argue that
this enzyme is mechanistically more related to CDA rather than
to ADAs.
ecTadA. Together with docking studies, the work supports a
stepwise mechanism for the tRNA editing process. The key
events are facilitated by the enzyme-promoted base flip and the
Glu70-mediated proton shuttle. Given the sequence similarity
at the catalytic domains of TadA and ADAR families, it is likely
that similar transition state structures have been adopted by the
ADARs. The structural features of this transition state, such as
the S-face hydroxyl, the amino group dissociation and the
interaction between the shuttled proton and the nearby Glu
residue, are informative for the development of tight-binding
inhibitors.
Acknowledgment. We thank J. Kim and S. C. Almo (Albert
Einstein College of Medicine) for the ecTadA plasmid and J.
Gao of the University of Minnesota Supercomputer Institute
for computer time. This work was supported by NIH grant
CA72444.
Conclusion. The A-to-I RNA editing enzymes have received
attention due to their complex RNA substrates and still-evolving
biological functions. Here, we applied KIEs and quantum
molecular modeling to solve the transition state structure of
(26) Ireton, G. C.; Black, M. E.; Stoddard, B. L. Structure 2003, 11, 961-972.
(27) Sharff, A. J.; Wilson, D. K.; Chang, Z.; Quiocho, F. A. J. Mol. Biol. 1992,
226, 917-921.
Supporting Information Available: Detailed methods, Figure
S1, and structures of optimized substrates, intermediates and
transitions. See Supporting Information online. This material
is available free of charge via the Internet at http://pubs.acs.org.
(28) Tyler, P. C.; Taylor, E. A.; Frohlich, R. F.; Schramm, V. L. J. Am. Chem.
Soc. 2007, 129, 6872-6879.
(29) Chung, S. J.; Fromme, J. C.; Verdine, G. L. J. Med. Chem. 2005, 48, 658-
660.
(30) Anisimov, V.; Paneth, P. J. Math. Chem. 1999, 26, 75-86.
JA078008X
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2655