Flight of a cytidine deaminase complex with an imperfect transition state analogue inhibitor: Mass spectrometric evidence for the presence of a trapped water molecule

Article abstract of DOI:10.1021/bi300516u

Cytidine deaminase (CDA) binds the inhibitor zebularine as its 3,4-hydrate (K_d ～ 10^-12 M), capturing all but ～5.6 kcal/mol of the free energy of binding expected of an ideal transition state analogue (K_tx ～ 10^-16 M). On the basis of its entropic origin, that shortfall was tentatively ascribed to the trapping of a water molecule in the enzyme-inhibitor complex, as had been observed earlier for product uridine [Snider, M. J., and Wolfenden, R. (2001) Biochemistry 40, 11364-11371]. Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) of CDA nebularized in the presence of saturating 5-fluorozebularine reveals peaks corresponding to the masses of E₂Zn₂W₂ (dimeric Zn-CDA with two water molecules), E₂Zn₂W₂Fz, and E₂Zn₂W₂Fz₂, where Fz represents the 3,4-hydrate of 5-fluorozebularine. In the absence of an inhibitor, E ₂Zn₂ is the only dimeric species detected, with no additional water molecules. Experiments conducted in H₂¹⁸O indicate that the added mass W represents a trapped water molecule rather than an isobaric ammonium ion. This appears to represent the first identification of an enzyme-bound water molecule at a subunit interface (active site) using FTICR-MS. The presence of a 5-fluoro group appears to retard the decomposition of the inhibitory complex kinetically in the vapor phase, as no additional dimeric complexes (other than E₂Zn₂) are observed when zebularine is used in place of 5-fluorozebularine. Substrate competition assays show that in solution zebularine is released from CDA (k_off > 0.14 s^-1) much more rapidly than is 5-fluorozebularine (k_off = 0.014 s^-1), despite the greater thermodynamic stability of the zebularine complex.

Full text of DOI:10.1021/bi300516u

Article
pubs.acs.org/biochemistry
Flight of a Cytidine Deaminase Complex with an Imperfect Transition
State Analogue Inhibitor: Mass Spectrometric Evidence for the
Presence of a Trapped Water Molecule
,
†,§
†
‡
†
,†
Gottfried K. Schroeder,* Li Zhou, Mark J. Snider, Xian Chen, and Richard Wolfenden*
†Department of Biochemistry and Biophysics, University of North Carolina, Chapel Hill, North Carolina 27599, United States
‡Department of Chemistry, College of Wooster, Wooster, Ohio 44691, United States
*
S Supporting Information
ABSTRACT: Cytidine deaminase (CDA) binds the inhibitor
−
12
zebularine as its 3,4-hydrate (K ∼ 10 M), capturing all but
d
∼
5.6 kcal/mol of the free energy of binding expected of an
−16
ideal transition state analogue (K ∼ 10 M). On the basis of
tx
its entropic origin, that shortfall was tentatively ascribed to the
trapping of a water molecule in the enzyme−inhibitor
complex, as had been observed earlier for product uridine
[
1
Snider, M. J., and Wolfenden, R. (2001) Biochemistry 40,
1364−11371]. Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) of CDA nebularized in the presence
of saturating 5-fluorozebularine reveals peaks corresponding to the masses of E Zn W (dimeric Zn-CDA with two water
2
2
2
molecules), E Zn W Fz, and E Zn W Fz , where Fz represents the 3,4-hydrate of 5-fluorozebularine. In the absence of an
2
2
2
2
2
2
2
1
8
inhibitor, E Zn is the only dimeric species detected, with no additional water molecules. Experiments conducted in H O
2
2
2
indicate that the added mass W represents a trapped water molecule rather than an isobaric ammonium ion. This appears to
represent the first identification of an enzyme-bound water molecule at a subunit interface (active site) using FTICR-MS. The
presence of a 5-fluoro group appears to retard the decomposition of the inhibitory complex kinetically in the vapor phase, as no
additional dimeric complexes (other than E Zn ) are observed when zebularine is used in place of 5-fluorozebularine. Substrate
2
2
−1
competition assays show that in solution zebularine is released from CDA (k > 0.14 s ) much more rapidly than is 5-
fluorozebularine (koff = 0.014 s ), despite the greater thermodynamic stability of the zebularine complex.
off
−1
8
he minimal affinity of an enzyme for the altered substrate
in the transition state (S ) can be estimated by comparing
the enzyme-catalyzed and uncatalyzed reaction rates and
larine (FZeb), show that both these inhibitors are bound as
⧧
T
covalently hydrated species (Figure 1B) that resemble a
tetrahedral intermediate approaching the presumed transition
state for the deamination of cytidine, with cytidine covalently
hydrated at the 3,4-double bond (Figure 1A). Figure 1B shows
that in these enzyme−inhibitor complexes the elements of
substrate water that were originally split between the zinc atom
and Glu-104 are incorporated into DHU.
establishes a benchmark against which the effectiveness of
1
−3
potential transition state analogue inhibitors can be tested.
4
That information has been reported for the complex formed
between the potent transition state analogue inhibitor 3,4-
dihydrouridine [DHU (Figure 1)] and Escherichia coli cytidine
deaminase (CDA, EC 3.5.4.5), a dimeric zinc enzyme that
catalyzes the conversion of cytidine to uridine and is
structurally and mechanistically related to the ApoB RNA-
Effects of temperature on the K value of DHU, and the
d
direct calorimetric titration of CDA with Zeb, indicate that the
binding of DHU is associated with a large and favorable change
in enthalpy (ΔH = −21.1 kcal/mol), closely matching the
5
editing family of enzymes. In CDA, two symmetrically situated
active sites are present at the dimer interface, each containing
the split elements of a substrate water molecule, with a
hydroxide ion bound by the zinc atom and a proton bound by
⧧
estimated enthalpy change associated with the binding of S
⧧
itself (ΔH = −20.2 kcal/mol). In contrast to the binding of S
6
(TΔS = 1.6 kcal/mol), the binding of DHU is achieved at an
entropic cost (TΔS) of −4.9 kcal/mol. That entropic penalty
the carboxylate group of Glu-104.
As illustrated in Figure 1, the structure of the CDA−DHU
(
−5 kcal/mol) is also observed in the formation of the
complex suggests that during deamination, the 3,4-double bond
+
enzyme−product complex with uridine, in which a trapped
of cytidine becomes covalently hydrated with addition of H at
−
water molecule is observed crystallographically (green sphere in
N3 and addition of OH at C4, forming a tetrahedra7l
9
Figure 1). The crystal structure of the enzyme−inhibitor
intermediate that approaches the transition state in structure.
Ammonia is then eliminated, with rate-determining C−N bond
6
cleavage, to yield the product uridine.
Received: April 20, 2012
Revised: July 6, 2012
Published: July 9, 2012
Crystal structures of inhibitory complexes of CDA with
7
zebularine (Zeb), or with the related inhibitor 5-fluorozebu-
©
2012 American Chemical Society
6476
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
4
Figure 1. Thermodynamic changes associated with binding of ligands on the reaction pathway (A) and inhibitors (B). Equilibrium constants and
thermodynamic parameters are shown for dissociation. Thermodynamic parameters for FDHU release were determined in this work. Crystal
10
structure figures were generated using PyMOL. The active site electrostatic surface potentials (in panel B) were generated with the Adaptive
11
Poisson−Boltzmann Solver (APBS) using the AMBER force field (PyMOL APBS Tools 2.1 plugin, M. G. Lerner) and the conversion program
1
2,13
PDB2PQR.
Red indicates negative charge, blue positive charge, and white hydrophobic areas.
Scheme 1. Proposed Sequence of Events within the Mass Spectrometer following Nebularization of the Complex of the Dimeric
a
Enzyme (blue and orange) with FDHU (yellow), Zinc (purple), and a Trapped Water Molecule (green) at the Active Site(s)
a
−13
The vacuum within the analysis chamber (∼10
atm) is sufficient to remove all solvating water molecules (red), yielding the dimer (1).
Presumably, the bound molecules “leak” out during the analysis time (2−3 s), yielding the various species shown (1−5), corresponding to the
indicated losses in mass. The dimeric species (3) corresponds to the most prevalent resultant mass peak observed previously, as indicated by the top
panel histogram (data taken from ref 14).
7
complex indicates the presence of a gap in the polar leaving
group site (Figure 1B, electrostatic surface potential im-
sequence of events in the mass spectrometer, beginning with
the nebularization of the dimeric enzyme complex with FDHU
from aqueous solution. Under nondenaturing conditions in the
absence of an inhibitor, the most significant mass corresponded
to that of the dimeric enzyme with two zinc atoms and no
water molecules (Scheme 1, stage 5). Evidently, all unbound
water molecules were stripped from the protein during
nebularization. In the presence of the fluorinated inhibitor
FZeb, three dimeric protein masses were evident (Scheme 1,
stages 1−3). The most significant of these corresponded to the
mass of the dimeric enzyme with two zinc atoms and two water
1
0−13
ages)
expected to be occupied by the ammonia leaving
group during deamination. It was suggested that the trapping of
an “extra” water molecule in the enzyme−inhibitor complex
might account for the entropic penalty observed and for the
enzyme’s less-than-perfect binding of DHU (Figure 1, chart).
In earlier work, we sought to test the possibility that a
water molecule in the enzyme−inhibitor complex might be
detectable using Fourier transform ion cyclotron resonance
mass spectrometry (FTICR-MS). Scheme 1 outlines a plausible
14
6
477
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
molecules (Scheme 1, stage 3). The less prominent second and
third peaks were augmented by the mass of covalently hydrated
FZeb (FDHU). The histogram in the top panel of Scheme 1
indicates the relative magnitudes of these peaks. Because those
preliminary experiments were conducted in the presence of low
concentrations of ammonium acetate buffer, the origin of the
additional mass of ∼35 Da attributed to two water molecules
transition state for its deamination. Accordingly, we measured
its rate of nonenzymatic hydrolysis and the values of kcat and
KM for the enzyme-catalyzed reaction. The effects of a single
hydrogen-to-fluorine substitution on the kinetic and thermody-
namic characteristics of inhibitor binding (Zeb vs Feb) were
also investigated.
(
Scheme 1, stage 3) could not be identified unambiguously.
MATERIALS AND METHODS
High-performance liquid chromatography grade water and
■
Earlier kinetic experiments had shown that E. coli CDA displays
+
15
a weak affinity for NH as an inhibitor. However, the
4
glacial acetic acid were purchased from Fisher Scientific
+
molecular mass of NH4 is essentially indistinguishable from
18
18
(Pittsburgh, PA). H2 O (97 O at. %) was obtained from
ISOTEC-Sigma (St. Louis, MO). Zeb and FZeb were gifts from
V. Marquez (National Cancer Institute, National Institutes of
Health, Bathesda, MD). Cation exchange resin (50W-X8, 100−
that of water (18 Da), leaving open the possibility that an
ammonium ion rather than a water molecule might be
responsible for the observed increment in mass. Interestingly,
preliminary attempts to repeat this experiment in the presence
of Zeb failed to show any noncovalent association of the
enzyme with either water or an inhibitor (Scheme 1, stage 5).
The experiments were undertaken to address the identity of the
trapped molecule in the dimeric enzyme−inhibitor complexes
and the surprising failure of Zeb to stabilize such complexes in
the vapor phase.
2
00 mesh, hydrogen form) was obtained from Bio-Rad
(
Hercules, CA). Other chemicals were obtained from TCI
America (Portland, OR) or Sigma-Aldrich Co. Recombinant
CDA from E. coli was overexpressed in E. coli and purified as
18,19
previously reported.
FTICR-MS Analysis. Enzyme samples (3 × 10− M in
monomer) were dialyzed overnight into ammonium acetate
buffer (5 × 10− M) at pH 5.6, where CDA maintains
5
3
In this work, CDA was found to be mainly monomeric in the
16
vapor phase in the absence of inhibitor, as described
essentially full activity. For experiments conducted under
nondenaturing conditions, enzyme samples were then mixed
1
4
previously. In the presence of the powerful inhibitor FZeb,
we observed masses corresponding to the dimeric enzyme−
inhibitor complexes described previously, including those
with (A) H216O, (B) FZeb (2.3 × 10 M) in H2 O, or (C)
−4
16
FZeb (2.3 × 10− M) in H2 O. Immediately before MS
analysis, all samples were acidified slightly with glacial acetic
acid (final concentration of 0.5% by volume), without the
addition of other organic solvents, to assist nebulization. For
experiments conducted under denaturing conditions, enzyme
samples were prepared in a 50% methanol and 2% acetic acid
solution immediately before nebularization.
4
18
16
containing the added masses of two molecules of H2 O (36
+
Da) or NH4 (36 Da) (Scheme 1, stages 1−4). To determine
whether those increments in mass were due to water or the
ammonium ion, we conducted the same experiment in the
1
8
18
presence of H2 O. If one or two molecules of H2 O were
+
bound at the active site, as opposed to NH , then an additional
4
shift of 2 or 4 Da (distinguishable in principle with the
ultrahigh mass accuracy of <1 ppm provided by an instrument
with a 12 T magnet) would be expected (Scheme 1, stages 3−
MS spectra were acquired using a hybrid Qe-Fourier
transform ion cyclotron resonance mass spectrometer,
equipped with a 12.0 T actively shielded magnet (Apex Qe-
FTICR-MS, 12.0 T AS, Bruker Daltonics, Billerica, MA), and
an Apollo II microelectrospray ionization (μESI) source. The
voltages on the μESI sprayer, interface plate, heated capillary
exit, deflector, ion funnel, and skimmer were set at 4.3 kV, 3.9
kV, 300 V, 250 V, 175 V, and 80 V, respectively. The
temperature of the μESI source was maintained at 180 °C.
Desolvation was conducted using both nebulization gas flow
(2.0 bar) and countercurrent drying gas flow (4.0 L/s). All
samples were directly infused using a syringe pump (Harvard
Apparatus, Holliston, MA) equipped with a 250 μL syringe
(Hamilton, Reno, NV), at an infusion flow rate of 90 μL/h.
Before transfer, ion packets accumulated inside the collision cell
for a fixed duration of 1−3 s. For each spectrum, 100 MS scans
were acquired in the ICR cell with a resolution of 580000 at an
m/z of 400 Da.
5). In all experiments with FZeb, the peak with two bound
water molecules (Scheme 1, stage 3) consistently exhibited the
highest signal-to-noise ratio. To look for the potential
1
8
incorporation of H2 O, it therefore seemed reasonable to
focus our analysis on that peak.
In preliminary experiments, dimeric enzyme complexes with
+
4
the added mass of water (or NH ) were not observed in the
presence of Zeb. This observation is surprising in view of the
4
fact that that DHU is bound ∼10-fold more tightly than
1
6
actual
i
17
FDHU (for K
values, see Figure 1B). Earlier work
showed that DHU is formed rapidly and reversibly from Zeb at
the enzyme’s active site. Surprisingly, preliminary experiments
suggested that the same might not be true of FZeb, and that a
slightly larger 5-fluorine substituent might have a substantial
retarding effect on the binding and release of FDHU, in
addition to its effect on the thermodynamics of binding. Such
an effect, if present, would be expected to be manifest in either
the vapor phase or the liquid phase.
Pre-Steady State Kinetics for the Release of FZeb or
Zeb. The off rate constant (k ) for the transition state
off
analogue inhibitor FZeb (or Zeb) was determined as follows.
To determine whether the rates of association of FZeb with
and dissociation of FZeb from the enzyme are slow in solution,
we used stopped-flow fluorescence and competition assays. The
relative contributions of enthalpy and entropy to FDHU−CDA
complex dissociation were estimated by isothermal titration
calorimetry (ITC) experiments using FZeb, and van’t Hoff
analysis of the temperature dependence of the equilibrium
constant for FZeb hydration using model compounds. To
complete the thermodynamic analysis, it was also necessary to
establish the affinity of the enzyme for 5-fluorocytidine in the
CDA was preincubated with FZeb (or Zeb) at a concentration
observed
in 10-fold excess of the Ki
value. After a preincubation
period of 5, 20, or 35 min, inhibited CDA was diluted 100-fold
into a concentrated solution of substrate cytidine (10-fold
excess of K ). The conversion of cytidine to uridine was then
M
monitored by the change in UV absorbance at 282 nm (Δε =
−1
−1
−3600 M cm ) on a Hewlett-Packard diode array 8452a
spectrophotometer for 10 min. The steady state rate was
extrapolated from the final 200 s of the reaction (linear
portion). A “lag” in the approach to the extrapolated steady
6
478
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
state turnover rate was used to determine the off rate constant
for the inhibitor.
= 282 nm (pH 7, water), δ (D O 500 MHz) 7.70 (1H, d, J =
H
2
5.8 Hz) and 3.32 [3H, s, N(1)CH ] (see Figure S1 of the
3
Pre-Steady State Kinetics for the Binding of FZeb. The
Supporting Information). Spectral properties were in agree-
ment with those reported previously.
2
3,24
intrinsic fluorescence of the inhibitor FZeb is quenched upon
16
binding to CDA. A rapid mixing device (RX.1000, Applied
Photophysics Ltd., Leatherhead, U.K.) and a spectrofluorom-
eter (Fluorolog-3, Horiba Jobin Yvon Inc., Edison, NJ) were
used to monitor the decay in FZeb fluorescence as a function of
time with increasing enzyme concentration. Bandpass filter
widths were set at 2 mm, and the fluorescence excitation and
Uncatalyzed Deamination of 1-N-Methyl-5-fluorocy-
tosine. In view of the known thermal instability of the product
25
5-fluorouridine, the uncatalyzed deamination rate of 5-
fluorocytidine was estimated using the model compound 1-N-
methyl-5-fluorocytosine. Aqueous solutions of 1-N-methyl-5-
fluorocytosine (0.01 M) in potassium phosphate buffer (0.1 M,
pH 6.8) were sealed in quartz tubes under vacuum and
incubated at elevated temperatures (90−140 °C) for varying
16
emission wavelengths were 322 and 390 nm, respectively. In a
typical experiment (25 °C), one reservoir of the rapid mixing
device was filled with a solution containing FZeb (0.1 μM, after
mixing), while the second contained a CDA solution of varying
concentrations (from ∼0.1 to ∼0.8 μM, after mixing).
Fluorescence acquisition was initiated when the two solutions
were mixed at a 1:1 ratio, with a total dead time of ∼0.4 s. The
fluorescence data were interpreted according to the kinetic
scheme shown in eq 1
periods of time. The contents were then diluted in D O and
2
1
analyzed by H NMR (Inova 500 MHz instrument, Varian,
Palo Alto, CA) using pyrazine as an added integration standard
−3
(10 M). The results were compared to those of cytidine and
1-methylcytosine, which have been shown to be essentially
26
identical.
Effect of Fluorine on Hydrophobicity. To test whether
observed
the slow off rate or the lower Ki
of FZeb (compared with
k
on
E + I XooY EI
that of Zeb) might be related to a simple difference in
hydrophobicity, we sought to determine the effect of a single
hydrogen-to-fluorine substitution on the ΔG of transfer of a
related molecule from water to cyclohexane. The distribution
coefficients of the tetrahydrofuryl derivative of 5-fluorouracil
k
(1)
off
where E is the enzyme, I the inhibitor, and EI the enzyme−
inhibitor complex. All kinetic traces represent an average of a
minimum of five experimental runs (2 min each) and were fit
by nonlinear regression to eq 2
[
1-(2-tetrahydrofuranyl)-5-fluorouracil] between water and
27
cyclohexane were measured by a double-extraction procedure
I] = Ae−
kobs
t
+ B
and compared with that of the tetrahydrofuryl derivative of
[
(2)
28
uracil.
where t is the time, kobs = kobserved, and A and B are constants.
Synthesis of 5-Fluoro-1-methyl-2-oxopyrimidine (2).
7-Chloro-7-fluoro-2,5-dioxabicyclo[4.1.0]heptane was synthe-
sized from 1,4-dioxene and dichlorofluoromethane according to
The resultant apparent first-order rate constants (k ) showed
obs
a linear relationship to enzyme concentration (eq 3)
29
the procedure of Molines et al., with a similar yield (∼90%).
The crude material was used to generate fluoromalonaldehyde
bis(diethyl acetal) (1) by refluxing in ethanol containing
kobs = k [E] + k
(3)
on
off
and an initial estimate for the on rate constant (k ) was
on
29
sulfuric acid, as described by Molines et al. Distillation (oil
determined from a plot of kobs versus enzyme concentration.
This value and the value for koff determined above were used as
starting parameters for fitting by simulation (KinTek Explorer,
bath) afforded yields (∼80%) equivalent to those reported
29
previously. 1 (200 mg, 0.8 mmol) was added to N-methylurea
20
[60 mg (0.84 mmol), recrystallized from toluene prior to use]
KinTek Corp., Austin, TX), using the model shown in eq 1.
and cation exchange resin (2 g, AG 50W-X8, hydrogen form) in
ethanol (1 mL) and concentrated HCl (1 mL) and heated (70
Steady State Kinetics of Deamination. Enzyme assays
2.5 nM in dimer) were conducted in potassium phosphate
(
°
C) while being stirred under argon. After ∼2 h, the reaction
buffer (0.1 M, pH 6.8) at 25 °C, following initial rates. Cytidine
1
−1
mixture was filtered and the resin was washed with water until
the filtrate no longer contained significant UV-absorbing
material (predominantly a peak at 290 nm). The resin was
then washed with ammonium hydroxide (0.25 M), and
(
(
ε290 = 2200 M− cm ) deamination was monitored at 290 nm
−1
−1
Δε = −1750 M cm ), and 5-fluorocytidine deamination
−1
−1
was monitored at 292 nm (Δε = −3500 M cm ). Extinction
−1
−1
coefficients of 5-fluorocytidine (ε = 5950 M cm ) and 5-
2
92
−
1
−1
collected fractions were monitored for product by UV (λ
fluorouridine (ε292 = 2450 M cm ) were experimentally
max
16,30
determined on the basis of the reported spectral properties of
= 336 nm)
in 1 M HCl. Fractions containing the product
21
these compounds.
were pooled, and the ammonium hydroxide was removed
under reduced pressure. The resultant white solid (18.3 mg,
20% yield) was triturated with cold ethanol to remove a trace
impurity. Alternatively, the compound was purified by flash
chromatography (silica) using ethyl acetate containing 5%
methanol (v/v), with similar results. The product was also
visualized on TLC plates by spraying with basic permanganate.
Synthesis of 1-N-Methyl-5-fluorocytosine. The syn-
thetic procedure was adapted from a procedure described
2
2
previously for the methylation of cytosine. 5-Fluorocytosine
(
(
0.32 mmol) was dissolved in water (3 mL) containing NaOH
0.16 M) and allowed to react with trimethylphosphate (0.9
mL, 25 equiv) for 24 h at room temperature. Basic reaction
conditions (pH 13) were maintained during the course of the
reaction by addition of NaOH (5 M). Residual trimethylphos-
phate was removed by evaporation (Kugelrohr), and the major
product of the reaction (∼95%) was the dimethylphosphate salt
of the desired compound. The product was desalted using an
anion exchange resin (100−200 mesh AG 1-X8, Bio-Rad,
Hercules, CA) and recrystallized from a minimal amount of
water at 4 °C, yielding fine crystals of the title compound: λmax
1
19
H and F NMR values [25 mM potassium phosphate buffer
(pH 7.0) containing 20% D O] for 2 were identical to the
2
16
reported values under the same conditions.
Synthesis of 5-Fluoro-1,3-dimethyl-2-oxopyrimidine
(3). 1 (200 mg, 0.8 mmol) was added to N,N-dimethylurea [70
mg (0.8 mmol), recrystallized from toluene prior to use] in
formic acid (90%, 2 mL) and heated (60−65 °C) while being
stirred under argon. Product formation was monitored by UV
6
479
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
Figure 2. FTICR-MS analysis of CDA (nondenaturing conditions) in the presence of FZeb in H216O. Following deconvolution of the multiply
charged ion signals, the most significant peak corresponded to the mass of the enzyme dimer with two zinc atoms and two water molecules bound
(
shaded region). Additional peaks corresponding to the added mass of either one or two molecules of the hydrated inhibitor FDHU were also
observed.
1
6
(
337 nm) in 1 M HCl, with concomitant loss of a peak at 256
use, both solutions were degassed and solutions of CDA (14.1
μM) were titrated with 25 injections (4 μL) of FZeb (287 μM)
while being stirred (400 rpm). The amount of heat released
following each injection was determined by integration of the
calorimetric signal. The heat arising from binding was
determined from the difference between the observed heat of
interaction and the corresponding heat of dilution. Origin
nm. After 5 h, the formic acid was removed by rotary
evaporation, yielding a yellow oil. The material was dissolved in
a 1:1 ethyl acetate/hexane mixture (v/v), loaded onto a plug of
silica, and eluted, first with 100 mL of the same solvent mixture
and then with 100 mL of ethyl acetate. Rotary evaporation
yielded a yellow oil, which was dissolved in a 1:1 ethanol/HCl
mixture (1 M). A rotary evaporation of the solution yielded a
yellowish solid, which was triturated with ethanol, yielding a
fine white solid. The residual ethanol (colored yellow) was then
decanted from the crystals that were subsequently washed with
ethanol. The decanted ethanol solution was subjected to the
same procedure, yielding additional white crystals. The white
crystals (18.6 mg, 18% yield) migrated as a single spot on TLC.
3
2,33
(version 5.0, MicroCal, Inc.)
was used for signal integration
and data fitting (nonlinear regression, single-binding site
model). Reported thermodynamic data represent an average
of two independent titrations (25 °C).
RESULTS
■
1
19
FTICR-MS Analysis. The observed monoisotopic mass of
the CDA monomer under denaturing conditions was
31520.855 Da (Figure S2A of the Supporting Information),
which is within 0.25 ppm (0.008 Da) of the theoretically
calculated monoisotopic mass. Under nondenaturing con-
ditions, the spectrum shows isotopic resolution with 31539.0
Da as the most abundant isotopic mass of the monomer
H and F NMR values [25 mM potassium phosphate buffer
(
pH 7.0) containing 20% D O] for 3 were identical to the
2
16
reported values under the same conditions.
Thermodynamics of FDHU Formation in Solution.
Earlier, the equilibrium constant for FZeb hydration (FDHU
formation) in free solution was estimated by comparing the
equilibrium constants for the ionization of water (K ), for
protonation of 5-fluoro-1-methyl-2-oxopyrimidine (2, K ), and
w
(
Figure S2B of the Supporting Information), matching the
1
expected average mass of the monomeric enzyme derived from
the elemental composition (31538.96 Da). Notably, the same
value for the most abundant isotopic mass (31539.02 Da) was
observed previously for the monomer under denaturing
for addition of hydroxide ion to the 5-fluoro-1,3-dimethyl-2-
oxopyrimidinium ion (3, K′ ) according to the equation
2
4
,16
(K
) shown in Figure 7.
In this work, the dependence
hydration
of each equilibrium constant on temperature was determined to
estimate the relative contributions of enthalpy and entropy to
FZeb hydration in solution. The temperature dependence (15−
14
conditions. In the absence of Fzeb, CDA was found to be
predominantly monomeric in the gas phase under non-
denaturing conditions, but one additional major ion series
corresponding to the CDA dimer with two zinc atoms bound
(most abundant mass being 63191.8 Da) was also observed, as
6
5 °C) for the ionization constant of water was obtained from
31
the tables of Mesmer and Herting. The dependencies of K1
and K′ on temperature were determined from the dissociation
constant of the conjugate acids of 2 and 3, respectively. Thus, a
.05 M solution of 2 or 3 (0.5 M ionic strength, KCl) was
titrated to its pK value at 25 °C, and the pH of the solution
was then measured over a temperature range from 10 to 60 °C.
Isothermal Titration Calorimetry. ITC experiments were
conducted with a Microcal (Northampton, MA) MSC
calorimeter. Enzyme (CDA) solutions were dialyzed in
potassium phosphate buffer (0.10 M, pH 7), and FZeb
solutions were made using the dialysate. Immediately prior to
2
14
reported previously. However, in the presence of FZeb, the
dimeric form of the enzyme was greatly stabilized and the
0
14
complexes identified previously were observed (Figure 2).
The most significant of those peaks (63226.8 Da, shaded region
in Figure 2) corresponds to the mass of the dimeric enzyme,
plus two zinc atoms and two water molecules (or two
ammonium ions). The next most significant peak in this
region exhibited an increase in mass of 263.8 Da, the mass of
the covalent hydrate of FZeb (FDHU). There was also a small
a
6
480
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
peak that corresponded to the additional mass of a second
molecule of FDHU (CDA complex with two inhibitors bound).
To test whether water rather than ammonium ion was
actually bound, the FTICR-MS experiment with CDA and
FZeb was repeated in O-labeled water. The results of those
experiments, focusing on the most statistically significant peak
to this scheme. Figure 3B shows the deconvoluted mass
spectrum of CDA nebulized from H216O in the absence of an
inhibitor. This single peak corresponds to the mass of the
enzyme dimer with two zinc atoms bound. The next panel
(Figure 3C) shows that the most abundant dimeric CDA
species in the mass spectrum following preincubation with the
inhibitor FZeb exhibited a mass increment of 35.0 Da,
1
8
(
shaded region in Figure 2), are summarized in Figure 3. The
1
6
consistent with the binding of two molecules of H2 O. A
small additional mass peak was also observed that might
correspond to the loss of a single water molecule from this
most abundant mass peak (decrease in mass of 17.9 Da). The
final panel (Figure 3D) shows the results of the same
experiment repeated in the presence of H218O. Three peaks
are clearly evident. The mass of the least abundant peak
matches that of the native dimer (Figure 3B). The next two
peaks are shifted by an additional mass of either 19.8 or 39.5 Da
top panel (Figure 3A) shows the various dimeric species
observed in this mass region during MS analysis, labeled
according to the convention used in Scheme 1. Mass peaks in
the subsequent panels (Figure 3B−D) are annotated according
(
most abundant), the mass of either one or two molecules of
18
H2 O, respectively. That difference would not be expected if
NH4+ were bound instead. The fact that all three species are
observed furnishes additional evidence that water rather than
NH4+ (or K , 39 Da) is the species bound (as outlined in
Scheme 1, stages 3−5). Notably, no mass shifts related to the
loss (or gain) of water molecules were observed in the
experiments conducted with the enzyme in the absence of
inhibitor (Figure 3B), or in the presence of Zeb (data not
shown).
+
Pre-Steady State Kinetics for the Release of FZeb or
Zeb. A plot of the time-dependent release of FZeb from CDA
(
Figure 4A) showed a clearly defined lag phase prior to steady
state turnover (solid line, extrapolated from the final 200 s of
the time course, gray). The difference (vertical distance)
between the kinetic data trace and the extrapolated line (gray)
is proportional to the concentration of the inhibited enzyme. A
semilogarithmic plot of this distance versus time was used to
determine the off rate constant (Figure 4B) of FZeb. The t1/2
for release (vertical distance decreased by half) was ∼60 s (k
off
−1
=
0.011 s ). The time of preincubation with the inhibitor (5−
3
5 min) showed no effect on the observed off rate. Results with
Zeb were identical to those of the uninhibited enzyme (data
not shown).
Pre-Steady State Kinetics for the Binding of FZeb.
Stopped-flow traces for the decrease in FZeb fluorescence with
binding, in the presence of increasing concentrations of CDA
are shown in Figure 5. The kinetic curves fit satisfactorily to a
2
single exponential [R = 0.99 (data not shown)], and the
resultant apparent first-order rate constants (k ) obtained are
obs
plotted as a function of enzyme concentration in Figure 5
(
inset). Fitting the data to eq 3 yields the following initial
5
−1
estimates for the binding rate constants: k = 3.7 × 10 M
on
−
1
−1
s , and k = 0.01 s . Using these parameters as initial
off
estimates, the full time course data were fit by simulation
2
0
(
KinTek Explorer) to a single-step kinetic model for
Figure 3. Deconvoluted FTICR mass spectra of dimeric CDA
(
nondenaturing conditions), with a scheme (A) illustrating the various
reversible inhibitor binding (eq 1), yielding the solid lines in
5
−1
species observed in this mass region (see Figure 2, shaded area). The
components of the most significant peak in each spectrum (in daltons)
are listed in each panel (right side). (B) CDA in H2 O in the absence
of an inhibitor. (C) CDA preincubated with the inhibitor FZeb in
H2 O. Note the mass shifts of 17.1 or 35 Da from the apo form of the
enzyme shown in panel B correspond to one or two H2 O (or
possibly one or two NH ) molecules, respectively. (D) CDA
preincubated with the inhibitor FZeb in H2 O. The least abundant
peak corresponds to the apo form of CDA as seen in panel B (gray
dotted lines). Note the additional mass shifts of either 19.8 or 39.5 Da,
corresponding to one or two molecules of bound H218O, respectively.
Figure 5 and the following values: k = (4.1 ± 0.2) × 10 M
on
1
−
−1
s , and k = 0.016 ± 0.002 s . Error limits were estimated
off
16
34
using FitSpace and reflect a 10% deviation from the minimal
sum squared error of the best fit, which was well constrained.
Taking an average of the off rate constant determined by
competition assay (vida supra) with the value determined by
16
16
+
4
18
this analysis, we determined a final value for k of 0.014 ±
off
1
−
0
.003 s . The dissociation constant, determined as the ratio of
k /k [(3.4 ± 0.5) × 10− M], is within experimental error of
the value reported previously, 3.5 × 10 M, for the Ki
8
off on
−8
observed
of
6
481
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
Figure 4. Competition assay (with excess cytidine) to determine the rate of release of CDA from Fzeb inhibition. (A) Time-dependent release from
inhibition following preincubation (20 min) of CDA with Fzeb. (B) Semilog plot of the vertical distance between the extrapolated line (gray) and
kinetic trace vs time. The t for release from inhibition is calculated as the time for this distance to decrease by half (63 s). Accordingly, koff was
1
/2
−1
calculated to be 0.011 s [koff = ln(2)/t1/2].
Figure 6. Arrhenius plot of the apparent first-order rate constants for
the decomposition of 1-N-methyl-5-fluorocytosine at neutral pH. The
kobs values for the solid points (●) were determined by nonlinear
Figure 5. Stopped-flow fluorescence changes of FZeb (0.1 μM, after
mixing) upon binding to CDA. Each trace represents an average of five
experiments. The solid lines represent simulated fits to the data
according to the model for reversible binding of an inhibitor to an
regression analysis of time course data (see Figure S4A,B of the
2
Supporting Information). The solid line is a linear fit to the data (R =
0
.99), and the dashed line represents the Arrhenius plot for 1-
enzyme in a single step (eq 1), with k and koff rates of (4.1 ± 0.2) ×
on
5
−1 −1
−1
methylcytosine deamination, included for reference. Extrapolation
1
0 M
s
and 0.016 ± 0.002 s , respectively. The inset corresponds
(
solid line) to 25 °C yields a nonenzymatic rate of deamination (knon)
to a plot of kobs (single-exponential fit) vs enzyme concentration. The
−11
−1
2
of 7.5 × 10
s
(t1/2 ∼ 300 years).
solid line corresponds to a linear fit to the data (R = 0.99) and yielded
initial estimates for kon and koff (where kobs = k [E] + k ).
on
off
dashed line). The data points at 100 and 140 °C reflect
nonlinear regression fits of time course data to a single
exponential (see Figure S4A,B of the Supporting Information).
1
7
FZeb. The latter value will be used in the discussion that
follows.
2
Steady State Kinetics of Deamination. Initial rates of the
enzyme-catalyzed deamination of both 5-fluorocytidine and
cytidine are plotted versus substrate concentration in panels A
and B of Figure S3 of the Supporting Information. Both
substrates followed Michaelis−Menten kinetics, and the fitted
lines represent a nonlinear regression fit (R = 0.99) to the
equation v = k [S]/(K + [S]), where v is the initial rate and
S] is the substrate concentration. The kinetic parameters
In both cases, fits were satisfactory (R = 0.99) and indicate that
the reaction is first-order. The observed free energy of
⧧
activation (ΔG ) of the reaction was 31.2 kcal/mol, with a
⧧
⧧
ΔH of 26 kcal/mol and a TΔS of −5.2 kcal/mol. By
extrapolation, the rate constant (k ) for 1-N-methyl-5-
non
2
−11 −1
fluorocytosine deamination at 25 °C is 7.5 × 10
s
(t1/2
∼ 300 years). CDA therefore generates a rate enhancement
cat
M
[
(k /k ) of 3 × 1012 for the deamination of 5-fluorocytidine
2
4
cat non
1
observed for 5-fluorocytidine were as follows: k = 240 ± 10
that is somewhat greater than the value for cytidine (1 × 10 ).
cat
−1
s , and K = 80 ± 5 μM. Those for cytidine were as follows:
Effect of Fluorine on Hydrophobicity. The tetrahy-
drofuryl derivative of 5-fluorouracil had a ΔG of transfer from
water to cyclohexane of 4.8 ± 0.2 kcal/mol, which can be
M
−1
kcat = 320 ± 10 s , and K = 60 ± 5 μM.
M
Uncatalyzed Deamination of 1-N-Methyl-5-fluorocy-
tosine. Apparent first-order rate constants for the neutral
deamination of 1-N-methyl-5-fluorocytosine are shown as a
linear Arrhenius plot in Figure 6 (solid line). The Arrhenius
28
compared to the value of 5 ± 0.2 kcal/mol for the
tetrahydrofuryl derivative of uracil. It is therefore unlikely that
observed
the 10-fold lower Ki
of FZeb compared to that of Zeb
plot for the deamination of 1-N-methylcytosine (k = 2 ×
arises from a simple difference in the hydrophobicity of these
molecules.
non
−
10 −1
26
10
s
at 25 °C) is also included for reference (Figure 6,
6
482
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
Thermodynamics of FDHU Formation in Solution. The
thermodynamic changes associated with FZeb hydration in free
solution were estimated by measuring the temperature
dependence for protonation of 5-fluoro-1-methyl-2-oxopyrimi-
dine (K ), for hydroxide addition to 5-fluoro-1,3-dimethyl-2-
1
oxopyrimidinium (K′ ), and for the ionization of water (K ), as
2
w
shown in Figure 7. The temperature dependence of K and K′
1
2
Figure 8. Thermodynamic cycle used to estimate the equilibrium
actual
constant (Ki ) for dissociation of the enzyme−FDHU complex
along with the relative contributions of enthalpy and entropy. The
observed
value of Ki
for FZeb was taken from ref 17.
nebularized after preincubation with FZeb, the relative
abundance of the dimeric form increases ∼10-fold but remains
∼
10-fold lower than that of the monomer. Under these
conditions, new masses are observed (Figure 2) that
correspond to the native dimer and an added mass of 35 Da,
along with other mass peaks further shifted by the mass of
either one or two molecules of the covalent hydrate of FZeb
(
FDHU). The experiments with 18O-labeled water and FZeb
presented here exhibited an added mass of 39.5 Da (vs 35 Da),
indicating that this added mass corresponds to two additional
water molecules (H218O) rather than ammonium ions (Figure
3
). In every experiment with FZeb, the most abundant dimeric
form of CDA contained the additional mass of two water
35
molecules. Apparently, the covalently hydrated inhibitor
molecules “leak” out over time, followed by the trapped
water molecules, yielding the additional peaks observed
Figure 7. Thermodynamic changes associated with the equilibrium
constant for hydration of 5-fluoro-1-methyl-2-oxopyrimidine (2),
based on the thermodynamic changes accompanying the ionization
(
Scheme 1).
As noted in our original publication, the entire complex
14
31
of water to generate a proton and hydroxide anion (Kw), addition of
remains intact in the vapor phase, at least for the 2−3 s required
for analysis. Thus, the enzyme appears to retain a somewhat
native-like conformation, at least in the region of the active site.
FDHU binds across the dimeric interface of CDA, and <0.1%
a proton to the CN bond (K ), and addition of a hydroxide anion to
1
the protonated CN bond (K ). The parameters associated with K
2
2
were estimated by modeling, utilizing addition of a hydroxide anion to
the quaternary amine formed by N-methylation (K′ , 3). Values for
2
8
the equilibrium constants at 25 °C were obtained as described in ref
of the bound inhibitor is exposed to bulk solvent. It seems
16.
likely that a water molecule is trapped in the product
ammonium ion binding pocket behind the pyrimidine ring of
the inhibitor [as observed when product is bound (see Figure
yielded linear van’t Hoff plots from 10 to 60 °C (Figure S5A,B
of the Supporting Information), corresponding to ΔH values of
1
A,B)], and that the presence of FDHU counteracts the
tendency of the dimer to dissociate by acting as a kind of
noncovalent cross-link” across the dimer interface. The
−
8.4 ± 0.2 and 5.3 ± 0.1 kcal/mol (25 °C), respectively.
Isothermal Titration Calorimetric Study of FZeb
“
Binding by Cytidine Deaminase. Binding of FZeb was
apparent stability of the enzyme−water complex may simply
represent the steady state concentration of this species in the
vapor phase.
exothermic with a ΔH of −13.6 ± 0.2 kcal/mol, after
cal
subtraction of the heat of dilution of the inhibitor, measured
after binding saturation (Figure S6 of the Supporting
Information). The Kd value determined at 25 °C from
36−38
A growing body of work
indicates that nebularization is
37
an effective sampling method, furnishing a snapshot of the
situation in solution. Accordingly, the kinetic stability of the
FZeb−CDA complex in solution was also investigated, with the
results shown in Figures 4 and 5. In solution, the kinetic barrier
nonlinear fitting of the titration data (single-binding site
observed
model) was within 2-fold of the K
value reported earlier
i
−8
17
for this complex (3.5 × 10 M). Using the relationship ΔG =
ΔH − TΔS, an entropy change of −3.5 ± 0.2 kcal/mol (25 °C)
was estimated for the binding of FZeb.
to dissociation of FZeb from CDA is substantial (k = 0.014
off
−
1
s ), equivalent to a t of ∼1 min, which could be related to
1/2
the stability of the complex in the vapor phase. In similar
experiments that aimed to measure the dissociation of Zeb
from CDA in solution, there was no evidence of slow release
DISCUSSION
■
When nebularized under nondenaturing conditions, CDA is
predominantly monomeric in the vapor phase, and zinc is
absent (this work and ref 14). The native dimer is also
observed, retaining both zinc atoms but devoid of water
molecules. In view of the high vacuum of the analysis chamber,
and experimental evidence that implies that the formation of
the zinc-bound hydroxyl (Figure 1) occurs after substrate
from inhibition: a lower limit for k can be estimated to be
off
0.14 s− (t1/2 ∼ 5 s) based on the dead time of the experiment.
On the basis of this lower limit, an on rate exceeding the
diffusion limit would be required to account for the observed
1
observed
−7
inhibition constant of Zeb (Ki
that DHU must be generated by hydration at the active site
(see also ref 17). This kinetic instability may help to explain
= 4.6 × 10 M), implying
6
binding, the absence of water is not surprising. When CDA is
6
483
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
Table 1. Comparison of Kinetic and Thermodynamic Parameters for Binding of Zeb and FZeb to CDA
a
b
c
d
Values taken from ref 4. Value reported in ref 17; k /k . Determined by ITC measurements with the inhibitor FZeb. Estimated using 2 and 3
off on
e
(
see Figure 7 and the corresponding text). K = k /(k /K ), with k estimated using the model compound N1-methyl-5-fluorocytosine (see
tx non cat m non
f g
the text). See the thermodynamic cycle in Figure 8 and the corresponding text. Note that the liberation of a trapped water molecule would be
h
K
tx‑K
i
actual
entropically favorable. ΔΔG
= −RT ln(K /Ki ).
tx
why experiments with Zeb failed to generate the dimeric
species observed by MS analysis with FZeb, despite the fact that
the covalent hydrate of Zeb (DHU) is bound 10-fold more
tightly than is FDHU.
inhibitor is bound. Such a change was previously postulated to
explain the slow release of THU.
39
In view of the ∼70-fold more favorable hydration of FZeb
versus Zeb in solution [which is presumably responsible (at
observed
16,17
It seems plausible that the stability of the FDHU complex in
the vapor phase reflects kinetic rather than thermodynamic
stability. The inhibitor 3,4,5,6-tetrahydrouridine (THU,
structurally identical to DHU, but lacking the 5,6 double
least in part) for the lower Ki
of FZeb],
it seemed
possible that the addition of a fluorine atom at C5 might affect
the uncatalyzed and enzymatic deamination rates to differing
extents, resulting in a difference in transition state affinities
between the two reactions. The uncatalyzed deamination of
5,6-dihydrocytidine (lacking the double bond) was previously
3
9
bond) exhibits a slow off rate (t ∼ 1 min), like that of
1
/2
−7
17
4
FZeb, but is bound (K ∼ 10 M) ∼10 -fold less tightly than
d
5
40
FDHU. Preliminary MS experiments with CDA and THU
yielded peaks that corresponded to the mass of (A) the native
dimer with both zinc atoms and two water molecules and (B)
the native dimer with both zinc atoms, two water molecules,
and two molecules of THU (data not shown). That result is
consistent with the view that the barrier to complex dissociation
in the gas phase is predominantly kinetic in origin. Notably, the
presence of a trapped water molecule in the presence of THU
was previously postulated on the basis of thermodynamic
found to proceed 10 -fold more rapidly than that of cytidine.
We therefore determined the relevant parameters for both the
uncatalyzed deamination of the model compound 1-N-methyl-
5-fluorocytosine (knon) and the enzymatic deamination of 5-
fluorocytidine (kcat and K ), and these were compared with the
M
corresponding parameters for the natural substrate cytidine.
Fluorine addition led to a ∼4-fold reduction in the rate of
−
11 −1
nonenzymatic deamination (k = 7.5 × 10 s ) compared
with that of cytidine (knon = 2.7 × 10
non
−
10 −1 42
s ). We also
4
0
41
evidence. It is of interest that a recent paper, using a
different MS approach (quadrupole time of flight), reports
findings consistent with the presence of noncovalently bound
water molecules trapped in a different enzyme−inhibitor
complex in the vapor phase.
observed a slight reduction (<2-fold) in the value of k /K (3
cat
M
6
−1 −1
× 10 M s ) compared with that of cytidine (this work). The
virtual dissociation constant of the enzyme−substrate complex
in the transition state for 5-fluorocytidine deamination (K )
tx
can therefore be estimated to be 2 × 10− M [knon/(k /K )],
17
cat
M
5
−1
The origin of the slow binding [kon = (4.1 ± 0.2) × 10 M
which differs by only ∼5-fold from the K value for cytidine
tx
−1
−16
s ] and release of FZeb, relative to that of Zeb, is not
immediately apparent. The region of the active site surrounding
the fluorine atom is predominantly hydrophobic, with three
nearby aromatic residues (Phe-71, Tyr-126, and Phe-165′) and
the ring “face” (with a partial negative charge) of Phe-165′
deamination [10 M (Figure 1)].
Transition state analogue inhibitors, because of their limited
geometric and electronic resemblance to the altered substrate in
the transition state, fall short of capturing the ideal binding
affinity. To more fully understand the functional basis of that
difference in binding affinity, it was of interest to measure and
compare the thermodynamic changes that accompany the
equilibrium binding of these inhibitors with those of the actual
transition state. Thermodynamic parameters associated with
FDHU formation in solution were estimated using model
compounds 2 and 3 (Figure 7). In conjunction with the
experimentally determined thermodynamic parameters for
FZeb binding (ITC), a thermodynamic cycle of equilibria can
8
oriented toward the fluorine atom. To test whether an increase
in hydrophobicity might be responsible for the observed kinetic
stability of the fluorine derivative, we measured the distribution
coefficient from water to cyclohexane of the fluorinated product
analogue, 1-(2-tetrahydrofuranyl)-5-fluorouracil. The free en-
ergy of transfer from water to cyclohexane (4.8 ± 0.2 kcal/mol)
was found to be the same, within experimental error, as that
observed earlier for 1-(2-tetrahydrofuranyl)uracil (5.0 ± 0.2
kcal/mol). An alternative possibility is that the kinetic barrier to
FDHU release, but not DHU release, is linked to a slow
conformational change of the protein when the fluorinated
actual
be assembled to estimate the equilibrium constant (Ki ) and
the corresponding contributions of enthalpy and entropy to
dissociation of the enzyme−FDHU complex (Figure 8).
6
484
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
Table 1 summarizes the results of this thermodynamic
analysis and shows the effects of fluorine on the kinetics and
thermodynamics of inhibitor binding. Perhaps most striking is
the increase in kinetic stability introduced by fluorine addition.
FDHU falls short of capturing the expected free energy of
binding of an ideal transition state analogue by 7.9 kcal/mol, as
indicated by the final entries in Table 1. That value closely
matches the entropic cost (TΔS = −8.3 kcal/mol) associated
with FDHU binding (shaded entries in Table 1). This result is
consistent with the entropic penalty observed previously for
REFERENCES
■
(
1) Wolfenden, R. (1972) Analog approaches to the structure of the
transition state in enzyme reactions. Acc. Chem. Res. 5, 10−18.
2) Lienhard, G. E. (1973) Enzymatic catalysis and transition-state
(
theory. Science 180, 149−154.
(3) Wolfenden, R., and Snider, M. J. (2001) The depth of chemical
time and the power of enzymes as catalysts. Acc. Chem. Res. 34, 938−
945.
(4) Snider, M. J., and Wolfenden, R. (2001) Site-bound water and
the shortcomings of a less than perfect transition state analogue.
Biochemistry 40, 11364−11371.
4
Zeb binding and seems understandable in terms of a trapped
water molecule at the active site.
(
5) Navaratnam, N., Fujino, T., Bayliss, J., Jarmuz, A., How, A.,
Richardson, N., Somasekaram, A., Bhattacharya, S., Carter, C., and
Scott, J. (1998) Escherichia coli cytidine deaminase provides a
molecular model for ApoB RNA editing and a mechanism for RNA
substrate recognition. J. Mol. Biol. 275, 695−714.
(6) Snider, M. J., Reinhardt, L., Wolfenden, R., and Cleland, W. W.
(2002) 15N kinetic isotope effects on uncatalyzed and enzymatic
deamination of cytidine. Biochemistry 41, 415−421.
7) Xiang, S., Short, S. A., Wolfenden, R., and Carter, C. W., Jr.
1995) Transition-state selectivity for a single hydroxyl group during
catalysis by cytidine deaminase. Biochemistry 34, 4516−4523.
8) Betts, L., Xiang, S., Short, S. A., Wolfenden, R., and Carter, C. W.,
CONCLUSION
■
This work confirms the entropically unfavorable trapping by
FDHU of noncovalently bound water by CDA, previously
inferred from thermodynamic evidence. This appears to
represent the first identification by FTICR-MS of an enzyme-
bound water molecule in an active site situated at a subunit
interface. Remarkably, this water molecule remains bound long
enough to allow detection at the high vacuum levels present in
the mass spectrometer. In principle, an analogue that displaced
this water molecule might be a far better inhibitor.
(
(
(
Jr. (1994) Cytidine deaminase. The 2.3 Å crystal structure of an
enzyme:transition-state analog complex. J. Mol. Biol. 235, 635−656.
9) Xiang, S., Short, S. A., Wolfenden, R., and Carter, C. W., Jr.
(
ASSOCIATED CONTENT
■
(1997) The structure of the cytidine deaminase-product complex
provides evidence for efficient proton transfer and ground-state
destabilization. Biochemistry 36, 4768−4774.
*
S
Supporting Information
1
Additional H NMR and MS spectra, along with Michaelis−
Menten plots, uncatalyzed kinetic time course data, van’t Hoff
plots, and ITC results. This material is available free of charge
via the Internet at http://pubs.acs.org.
(
10) DeLano, W. L. (2002) The PyMOL Molecular Graphics System,
DeLano Scientific, San Carlos, CA.
11) Baker, N. A., Sept, D., Joseph, S., Holst, M. J., and McCammon,
J. A. (2001) Electrostatics of nanosystems: Application to micro-
tubules and the ribosome. Proc. Natl. Acad. Sci. U.S.A. 98, 10037−
10041.
(12) Dolinsky, T. J., Nielsen, J. E., McCammon, J. A., and Baker, N.
A. (2004) PDB2PQR: An automated pipeline for the setup of Poisson-
Boltzmann electrostatics calculations. Nucleic Acids Res. 32, W665−
W667.
(
AUTHOR INFORMATION
■
*
Corresponding Author
G.K.S.: phone, (512) 471-8860; e-mail, gkschroeder@mail.
utexas.edu. R.W.: phone, (919) 966-1203; fax, (919) 966-2852;
e-mail, water1@ad.unc.edu.
(
13) Dolinsky Todd, J., Czodrowski, P., Li, H., Nielsen Jens, E.,
Present Address
Jensen Jan, H., Klebe, G., and Baker Nathan, A. (2007) PDB2PQR:
Expanding and upgrading automated preparation of biomolecular
structures for molecular simulations. Nucleic Acids Res. 35, W522−
W525.
§
Division of Medicinal Chemistry, College of Pharmacy,
University of Texas, Austin, TX 78712.
Author Contributions
G.K.S. and L.Z. contributed equally to this work.
(
14) Borchers, C. H., Marquez, V. E., Schroeder, G. K., Short, S. A.,
Snider, M. J., Speir, J. P., and Wolfenden, R. (2004) Fourier transform
ion cyclotron resonance MS reveals the presence of a water molecule
in an enzyme transition-state analogue complex. Proc. Natl. Acad. Sci.
U.S.A. 101, 15341−15345.
Funding
This work was supported by National Institutes of Health
Grant GM-18325.
(
15) Cohen, R. M., and Wolfenden, R. (1971) Cytidine deaminase
Notes
from Escherichia coli. Purification, properties, and inhibition by the
potential transition state analog 3,4,5,6-tetrahydrouridine. J. Biol.
Chem. 246, 7561−7565.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
(
16) Carlow, D. C., Short, S. A., and Wolfenden, R. (1996) Role of
We thank Ashutosh Tripathy for his assistance with the
fluorescence data acquisition at the Macromolecular Inter-
actions Facility (University of North Carolina) and William H.
Johnson for his help and guidance with the synthesis of 2 and 3.
glutamate-104 in generating a transition state analog inhibitor at the
active site of cytidine deaminase. Biochemistry 35, 948−954.
(17) Frick, L., Yang, C., Marquez, V. E., and Wolfenden, R. V. (1989)
Binding of pyrimidin-2-one ribonucleoside by cytidine deaminase as
the transition-state analog 3,4-dihydrouridine and contribution of the
4
(
(
-hydroxyl group to its binding affinity. Biochemistry 28, 9423−9430.
18) Smith, A. A., Carlow, D. C., Wolfenden, R., and Short, S. A.
1994) Mutations affecting transition-state stabilization by residues
coordinating zinc at the active site of cytidine deaminase. Biochemistry
ABBREVIATIONS
■
Zeb, zebularine; CDA, cytidine deaminase; FZeb, 5-fluoroze-
bularine; DHU, 3,4-dihydrouridine; FTICR, Fourier transform
ion cyclotron resonance; MS, mass spectrometry; FDHU, 5-
fluoro-3,4-dihydrouridine; TLC, thin layer chromatography;
ITC, isothermal titration calorimetry; THU, 3,4,5,6-tetrahy-
drouridine.
3
3, 6468−6474.
(19) Yang, C., Carlow, D., Wolfenden, R., and Short, S. A. (1992)
Cloning and nucleotide sequence of the Escherichia coli cytidine
deaminase (ccd) gene. Biochemistry 31, 4168−4174.
6
485
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486

Biochemistry
Article
(
20) Johnson, K. A., Simpson, Z. B., and Blom, T. (2009) Global
(41) Wang, S., Lim, J., Thomas, K., Yan, F., Angeletti, R. H., and
Schramm, V. L. (2012) A complex of methylthioadenosine/S-
adenosylhomocysteine nucleosidase, transition state analogue, and
nucleophilic water identified by mass spectrometry. J. Am. Chem. Soc.
134, 1468−1470.
Kinetic Explorer: A new computer program for dynamic simulation
and fitting of kinetic data. Anal. Biochem. 387, 20−29.
(
21) Alderfer, J. L., Loomis, R. E., Sharma, M., and Hazel, G. (1985)
Halogenated nucleic acids: Base properties and conformation of
fluorinated derivatives. Prog. Clin. Biol. Res. 172, 249−261.
(42) Snider, M. J., Gaunitz, S., Ridgway, C., Short, S. A., and
Wolfenden, R. (2000) Temperature effects on the catalytic efficiency,
rate enhancement, and transition state affinity of cytidine deaminase,
and the thermodynamic consequences for catalysis of removing a
substrate “anchor”. Biochemistry 39, 9746−9753.
(
22) Yamauchi, K., Tanabe, T., and Kinoshita, M. (1976)
Methylation of nucleic acid-bases with trimethyl phosphate. J. Org.
Chem. 41, 3691−3696.
(
23) Durr, G. J. (1965) Some 5-fluoropyrimidines. J. Med. Chem. 8,
2
(
53−254.
24) Atkins, P. J., and Hall, C. D. (1983) Stereochemistry of the
formation of 4-alkoxyimino-5,6-dihydro-6-alkoxyaminopyrimidin-
(1H)-ones from cytosines and hydroxylamines. J. Chem. Soc., Perkin
Trans. 2,, 155−160.
25) Dorta, M. J., Munguia, O., Farina, J. B., Martin, V. S., and
2
(
Llabres, M. (1997) Stability indicating high performance liquid
chromatography methods for 5-fluorouridine in aqueous solution.
Arzneim. Forsch.,, 1388−1392.
(
26) Carlow, D., and Wolfenden, R. (1998) Substrate connectivity
effects in the transition state for cytidine deaminase. Biochemistry 37,
1873−11878.
27) Bone, R., Cullis, P., and Wolfenden, R. (1983) Solvent effects on
1
(
equilibria of addition of nucleophiles to acetaldehyde and the
hydrophilic character of diols. J. Am. Chem. Soc. 105, 1339−1343.
(
28) Shih, P., Pedersen, L. G., Gibbs, P. R., and Wolfenden, R.
(
1998) Hydrophobicities of the nucleic acid bases: Distribution
coefficients from water to cyclohexane. J. Mol. Biol. 280, 421−430.
29) Molines, H., and Wakselman, C. (1989) Fluoromalonaldehyde
(
bis(dialkyl acetals): Synthesis by carbene condensation and trans-
formation to dialkyl fluoromalonates and fluorinated heterocyclic
compounds. J. Org. Chem. 54, 5618−5620.
(
30) Cech, D., Beerbaum, H., and Holy, A. (1977) Nucleic acid
components and their analogs. CXCI. 5-Fluoro-2-pyrimidinone and its
N1-substituted derivatives. Collect. Czech. Chem. Commun. 42, 2694−
2700.
(
31) Mesmer, R. E., and Herting, D. L. (1978) Thermodynamics of
−
ionization of D O and D PO . J. Solution Chem. 7, 901−913.
2
2
4
(
32) Indyk, L., and Fisher, H. F. (1998) Theoretical aspects of
isothermal titration calorimetry. Methods Enzymol. 295, 350−364.
33) Wiseman, T., Williston, S., Brandts, J. F., and Lin, L. N. (1989)
(
Rapid measurement of binding constants and heats of binding using a
new titration calorimeter. Anal. Biochem. 179, 131−137.
(
34) Johnson, K. A., Simpson, Z. B., and Blom, T. (2009) FitSpace
Explorer: An algorithm to evaluate multidimensional parameter space
in fitting kinetic data. Anal. Biochem. 387, 30−41.
(
35) Another mass peak corresponding to the additional mass of one
18
FDHU molecule, albeit weak, was also observed in the H2
experiment (data not shown).
O
(
36) Gao, J., Wu, Q., Carbeck, J., Lei, Q. P., Smith, R. D., and
Whitesides, G. M. (1999) Probing the energetics of dissociation of
carbonic anhydrase-ligand complexes in the gas phase. Biophys. J. 76,
3
(
253−3260.
37) Li, Z., Sau, A. K., Shen, S., Whitehouse, C., Baasov, T., and
Anderson, K. S. (2003) A snapshot of enzyme catalysis using
electrospray ionization mass spectrometry. J. Am. Chem. Soc. 125,
9
(
938−9939.
38) Yu, Y., Kirkup, C. E., Pi, N., and Leary, J. A. (2004)
Characterization of noncovalent protein-ligand complexes and
associated enzyme intermediates of GlcNAc-6-O-sulfotransferase by
electrospray ionization FT-ICR mass spectrometry. J. Am. Soc. Mass
Spectrom. 15, 1400−1407.
(
39) Ashley, G. W., and Bartlett, P. A. (1984) Inhibition of
Escherichia coli cytidine deaminase by a phosphapyrimidine nucleoside.
J. Biol. Chem. 259, 13621−13627.
(
40) Snider, M. J., Lazarevic, D., and Wolfenden, R. (2002) Catalysis
by entropic effects: The action of cytidine deaminase on 5,6-
dihydrocytidine. Biochemistry 41, 3925−3930.
6
486
dx.doi.org/10.1021/bi300516u | Biochemistry 2012, 51, 6476−6486