Regulation of dCTP deaminase from Escherichia coli by
nonallosteric dTTP binding to an inactive form of the
enzyme
Eva Johansson
1,2
, Majbritt Thymark
1
, Julie H. Bynck
1
, Mathias Fanø
3
, Sine Larsen
1,2
and Martin Willemoe
¨
s
3
1 Centre for Crystallographic Studies, Department of Chemistry, University of Copenhagen, Denmark
2 European Synchrotron Radiation Facility, Grenoble, France
3 Department of Molecular Biology, University of Copenhagen, Denmark
Synthesis of dTMP by thymidylate synthase proceeds
by the reductive methylation of dUMP, which is
obtained via one of two parallel pathways. One path-
way, considered to be a minor supplier of dTTP [1–3],
involves the reduction of UDP (UTP) by the action of
ribonucleotide reductase. Subsequently, dUDP is phos-
phorylated to dUTP and cleaved to dUMP. The main
supply of dUMP, however, involves the deamination
Keywords
deoxynucleotide metabolism; dUTP; enzyme
regulation; hysteresis; deamination

70–80% of the total dUMP as a precursor for dTTP. Accordingly, dCTP
deaminase is regulated by dTTP, which increases the substrate concentra-
tion for half-maximal activity and the cooperativity of dCTP saturation.
Likewise, increasing concentrations of dCTP increase the cooperativity of
dTTP inhibition. Previous structural studies showed that the complexes of
inactive mutant protein, E138A, with dUTP or dCTP bound, and wild-type
enzyme with dUTP bound were all highly similar and characterized by hav-
ing an ordered C-terminal. When comparing with a new structure in which
dTTP is bound to the active site of E138A, the region between Val120 and
His125 was found to be in a new conformation. This and the previous con-
formation were mutually exclusive within the trimer. Also, the dCTP com-
plex of the inactive H121A was found to have residues 120–125 in this new
conformation, indicating that it renders the enzyme inactive. The C-ter-
minal fold was found to be disordered for both new complexes. We suggest
that the cooperative kinetics are imposed by a dTTP-dependent lag of
product formation observed in presteady-state kinetics. This lag may be
derived from a slow equilibration between an inactive and an active confor-
mation of dCTP deaminase represented by the dTTP complex and the
dUTP ⁄ dCTP complex, respectively. The dCTP deaminase then resembles a
simple concerted system subjected to effector binding, but without the use
of an allosteric site.
Abbreviations
E138A, mutant dCTP deaminase with a Glu138 to Ala substitution; H121A, mutant dCTP deaminase with a His121 to Ala substitution;
V122G, mutant dCTP deaminase with a Val122 to Gly substitution.
4188 FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS
of a deoxycytidine nucleotide [1–3]. In eukaryotes and
most of the well-studied Gram-positive bacteria (e.g.
Bacillus subtilis) a dCMP deaminase supplies dUMP
directly by deamination of dCMP [4–6]. dCMP deami-
nase is structurally related to cytosine- and cytidine

extremely important removal of the toxic intermedi-
ate, dUTP [21–23].
Kinetic analysis of dCTP deaminase from S. ent-
erica serovar Typhimurium showed competitive
inhibition of dCTP binding by dTTP. However, the
presence of dTTP in the assay incubation also
increased the apparent cooperativity of dCTP bind-
ing, which indicates that the mechanism of dTTP
inhibition is not caused only by a trivial competition
between substrate and inhibitor for binding to the
same site [12]. We have previously determined the
structures of wild-type dCTP deaminase in complex
with dUTP and the inactive E138A mutant protein
in complex with dUTP and dCTP. In all cases, we
observed an ordered C-terminal that was closed over
the active site, but in a different conformation to
that observed for dUTPase. For both the mono- and
bifunctional dCTP deaminases, as we show here, and
the dUTPase [24], the C-terminal fold is important
for the formation of a catalytically competent com-
plex by closing the active site, but not for binding
of the substrates. In this study, we present results
from structural and mechanistic studies on dTTP
inhibition of E. coli dCTP deaminase. Coordinated
closure of the active site and rearrangement of the
main chain and side chains in the active site appear
as key players in a slow transformation from an
inactive to an active enzyme. dTTP inhibition may
then be achieved by stabilizing the inactive form of
presumably both the mono- and bifunctional dCTP

tron-density map (Fig. 1). Furthermore, the c-phos-
phate of dTTP is not visible in the electron-density
map and a magnesium ion is only seen bound to the
phosphates of dTTP in one of the two subunits. Large
movement of a helix 2 (residues 55–65) [18] is also
observed. If the C-terminal residues had been folded
over the active site, as shown in previous structures of
E. coli dCTP deaminase, these residues would have
coincided with a helix 2 in this new position (Fig. 1B).
A loop containing active-site residues in the interior of
the enzyme (residues 120–125) is also totally different
compared with previously determined E. coli dCTP
E. Johansson et al. dTTP inhibition of dCTP deaminase
FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS 4189
deaminase structures (Figs 1B, 2 and 3). Interestingly,
the crystal structure of the other inactive mutant
enzyme H121A in complex with dCTP was very
similar to the structure of E138A in complex with
dTTP. In the H121A complex we also observed a dis-
ordered C-terminus and rearrangement of active-site
residues 120–125 that was almost identical to the
E138A complex (Fig. 2D,E)). Wild-type E. coli dCTP
deaminase in complex with dTTP crystallized in the
same form and under similar conditions as for the
E138A:dTTP and H121A:dCTP complexes. However,
resolution of the diffraction data for the wild-type
enzyme in complex with dTTP was poor (3.5 A
˚
). As a
result, details of the active site of the wild-type:dTTP

(yellow, cyan and magenta). (B) Stereo-
view of a superposition of one of the sub-
units of the E138A variant of E. coli dCTP
deaminase in complex with dTTP (yellow;
chain B), dUTP (cyan; Protein Data Bank
entry 1XS4, chain A), or dCTP (magenta;
Protein Data Bank entry 1XS6, chain A). The
nucleotides are shown in ball and stick rep-
resentations and the magnesium ions as
spheres. The N-terminus (N) and the extent
to which the C-termini were resolved in the
dTTP complex (C1) and the dCTP ⁄ dUTP
complexes (C2) are indicated. The solid
arrow points to the region of a helix 2 and
b strand 5 that moved towards the active
site in the absence of an ordered C-terminal
fold. The dotted arrow points to the region
in the active site constituted by residues
120–125 that deviated in position between
the dCTP ⁄ dUTP complexes and the dTTP
complex. The figure was created using
PYMOL (DeLano Scientific, San Carlos, CA).
dTTP inhibition of dCTP deaminase E. Johansson et al.
4190 FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS
Analysis of the presteady-state kinetic behaviour of
dCTP deaminase using rapid quench-flow experiments
showed a lag in the progress of product formation
(Fig. 4C). This lag, which indicates slow activation of
the enzyme upon substrate binding prior to the forma-
tion of a catalytic complex, increased in the presence

state velocity, V
ss
, greatly increased the errors of the
calculated constants in Eqn (4). Therefore, V
ini
was
fixed at 0 when performing the calculations. The late
data points obtained in the absence of dTTP showed a
deviation from linearity caused by beginning substrate
depletion (Fig. 4C) and were omitted from the calcula-
tions. Unfortunately, we were not able to perform
presteady-state experiments at subsaturating substrate
concentrations to fully characterize the kinetics of the
slow transition from inactive to active enzyme [25].
Attempts to do so were hampered by the experimental
requirement for high enzyme concentrations both in
terms of estimating the true free-ligand concentration
in the experiments and by rapid substrate depletion
resulting in an underestimation of V
ss
.
dTTP binding to dCTP deaminase was also
investigated by equilibrium binding. This revealed a
hyperbolic binding curve (Fig. 4D) with a stochiometry
of 1 : 1 of dTTP bound per subunit of dCTP
deaminase.
Mutational analysis of amino acid residues
involved in dTTP regulation of dCTP deaminase
The design of the mutant enzymes H121A and V122G
was inspired by the results from analysis of crystal

enzyme in complex with dUTP in cyan (Pro-
tein Data Bank entry 1XS1, chain A). The
superposition demonstrates the likely struc-
tural incompatibility between the two con-
formers due to a clash of the side chains of
Val122 and Thr123 as indicate by the
arrows. The figure was created using
PYMOL
(DeLano Scientific).
dTTP inhibition of dCTP deaminase E. Johansson et al.
4192 FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS
similar with only small changes in the arrangement of
water molecules around the 4 position of the pyrimid-
ine ring [18]. In this study, we compared the structures
of dCTP deaminase, represented by E138A, in complex
with all three nucleotides that bind to the enzyme.
A superposition of the trimer of E138A with the
nucleotides dCTP or dTTP bound is shown in Fig. 1A.
Whereas the previously determined structures of
E138A in complex with dCTP or dUTP overall are
virtually identical [18], the new dTTP complex revealed
a disordered C-terminus. This difference between the
two types of complex is more easily reconciled in the
comparison of a single subunit of E138A in complex
with dUTP, dCTP or dTTP (Fig. 1B). In the dTTP
complex, the entrance to the active site had partly col-
lapsed caused by a movement of the lip formed by
a helix 2 and b strand 5 [18]. Apparently, movement of
the active site lip prevented binding of the C-terminal
residues over the active site, or the absence of the

of the active site lip and closure of the C-terminal end
over the active site. However, it is reasonable to expect
these two events to be associated but the structural
change that mediates the communication between the
two regions appears to be very subtle.
Based on the observations described above, mutant
alleles encoding the enzymes H121A and V122G were
constructed to analyse the roles of His121 and Val122
in catalysis and regulation of dCTP deaminase.
Removal of the imidazole ring in H121A was anticipa-
ted to relieve or reduce inhibition by dTTP by prevent-
ing expulsion of the water molecule, wat5, as described
above (Fig. 2C). Replacing the Val122 side chain in
V122G aimed to relieve the suggested concerted struc-
tural transition of the trimer and perhaps reduce the
inhibition by dTTP. As mentioned in the results, the
AB
C
D
Fig. 4. Initial rate and presteady-state kinetics of dTTP inhibition
and dTTP binding to dCTP deaminase. Assays were performed as
described in Experimental procedures. (A) The concentration of
dCTP varied as indicated in the absence (closed circles) or pres-
ence (open circles) of 100 l
M dTTP. The kinetic constants calcula-
ted using Eqn (1) were (closed circles) k
cat
¼ 1.24 ± 0.09 s
)1
,

) and (open cir-
cles) rate
ss
¼ 0.16 ± 0.02 s
)1
with s ¼ 2.3 ± 0.5 s (k ¼ 0.43 s
)1
).
For comparison the straight lines represent the calculated steady-
state rate in the absence of a lag. (D) dTTP binding to dCTP
deaminase. Binding experiments were performed as described in
Experimental procedures. The nucleotide concentration varied as
indicated. The binding constants calculated using Eqn (3) were
N
max
¼ 1.01 ± 0.02 and K
d
¼ 35 ± 3 lM.
E. Johansson et al. dTTP inhibition of dCTP deaminase
FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS 4193
mutant proteins were both inactive and unfortunately
no suitable crystals for the structural analysis of
V122G could be obtained. However, the conformation
of the main chain in the region of residues 120–125 in
the H121A:dCTP complex was found to strongly devi-
ate from that of wild-type enzyme in complex with
dUTP and almost superimpose with the same region in
the structure of the E138A:dTTP complex (Fig. 2E,D).
In addition, the H121A:dCTP complex, like the
E138A:dTTP complex, had a disordered C-terminal

for dTTP with increasing dCTP concentrations
(Fig. 4B). From the equilibrium binding experiment
presented in Fig. 4D it can be seen that dTTP binds to
only one type of site with no cooperativity.
The lag observed in presteady-state kinetics shown
in Fig. 4C is a clear indication that the mechanism of
regulation of dCTP deaminase is not a simple rapid
equilibrium mechanism. The observed increase in
cooperativity of dTTP inhibition at increasing dCTP
concentrations (Fig. 4B), but complete absence of
cooperativity in equilibrium binding of dTTP
(Fig. 4D), indicates that the cooperativity effect of
dTTP inhibition is a kinetic phenomenon. Given the
right circumstances, a lag in the progress of product
formation is known to produce what is termed kinetic
cooperativity and several enzyme systems have been
shown to possess such properties [25–28]. The k for
activation of dCTP deaminase is of the same order of
magnitude as the k
cat
(Fig. 4A,), a condition that qual-
ifies for causing kinetic cooperativity, and very import-
ant, the lag is increased in the presence of dTTP. The
increase in cooperativity of dCTP saturation in the
presence of dTTP may therefore be explained by a
mechanism in which dTTP stabilizes an inactive form
that dominates the population of free enzyme, recall
that V
ini
(or rate

and the sigmoidity of the satura-
tion curve for dCTP in response to the binding of
dTTP to the enzyme.
Experimental procedures
Materials
All buffers, nucleotides and salts were obtained from Sig-
ma-Aldrich (Darmstadt, Germany). Radioactive nucleotides
were obtained as ammonium salts from Amersham Bio-
sciences (Hillerød, Denmark). TLC was performed with
poly(ethylene-imine)-coated cellulose plates from Merck
(Darmstadt, Germany).
Molecular biology and protein methods
Construction of mutant alleles of the dcd gene encoding the
dCTP deaminases H121A and V122G was achieved by
performing the QuikChange method (Stratagene, La Jolla,
CA) using the oligo-deoxynucleotides, where underlined
letters indicate the site of mutagenesis: H121A5–3,
dTTP inhibition of dCTP deaminase E. Johansson et al.
4194 FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS
GGGCTGATGGTGGCCGTCACCGCGCAC; H121A3–5,
GTGCGCGGTGACG
GCCACCATCAGCCC; V122G5–3,
GATGGTGCACG
GCACCGCGCACC; V122G3-5, GGT
GCGCGGTG
CCGTGCACCATC. The plasmid pETDCD
described previously [18] was used as a template for muta-
genesis. The pETDCD plasmid contains the reading frame
of the E. coli dcd gene under control of the late T7 promo-
ter in the vector pET11a (Novagen, Darmstadt, Germany).

formic acid at the time points given under results. Subse-
quently, the samples representing each time point were sub-
jected to TLC and analysed for the distribution of
radioactivity in spots of [5-
3
H] dCTP and [5-
3
H] dUTP as
above for steady-state kinetic samples.
In equilibrium binding experiments, the incubations con-
tained dCTP deaminase (50–100 lm), 50 mm Hepes,
pH 6.8, 2 m m MgCl
2
and between 0 and 320 lm [methyl-
3
H]
dTTP. Free nucleotide was separated from bound using
Amicon Ultrafree-MC 30.000 NMWL centrifugal filter
devices, as described previously [31,32]. Samples represent-
ing free and total radioactive nucleotide were washed by
TLC in 1 m acetic acid, cut out and quantified by liquid
scintillation as above for samples from kinetic experiments.
Data from presteady-state and steady-state kinetic and
equilibrium binding experiments were analysed using the
computer program ultrafit from biosoft (v. 3.0). The
equations used were: the Hill equation, Eqn (1), for sigmoid
saturation curves
rate ¼ k
cat
½S

0.5
is the concen-
tration of inhibitor for half-maximal inhibition. Equation (3)
was used for hyperbolic binding of ligands to the enzyme
N ¼ N
max
½L=ðK
d
þ½LÞ ð3Þ
where N is the degree of binding with the dissociation con-
stant K
d
of ligand L to the enzyme with a maximal number
of binding sites N
max
. Equations (4–6) were used to analyse
the data recorded for presteady-state kinetics
P ¼ V
ss
t ÀðV
ss
À V
ini
Þð1 À e
Àt=s
Þs ð4Þ
rate
ss
¼ V
ss

|| ⁄S
work
F
obs
. R
free
¼
S
test
||F
obs
| ) k|F
calc
|| ⁄S
test
F
obs
, where F
obs
and F
calc
are observed
and calculated structure factors, respectively, k is the scale factor,
and the sums are over all reflections in the working set and test
set, respectively. rmsd, root mean square deviation.
Diffraction data statistics
Protein:ligand E138A:dTTP H121A:dCTP
Space group P6
3
22 P6

free
(%) 30.7 (32.0) 27.2 (31.4)
Average B
factor
(A
˚
2
)29 29
Bond length rms
from ideal (A
˚
)
0.016 0.014
Bond angle rmsd
from ideal (deg)
1.7 1.6
E. Johansson et al. dTTP inhibition of dCTP deaminase
FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS 4195
rates) prior to and after the transition of the enzyme to a
more active form, respectively, t is the time and s is the lag-
time. The rate constant, k, for the activation of the enzyme is
obtained as 1 ⁄ s.
Crystallization
Crystals were grown in hanging drops as described previ-
ously [18] using the vapour-diffusion technique with hang-
ing drops. Protein solutions contained 3.7 or 5.1 mgÆmL
)1
protein for H121A and E138A mutant enzymes, respect-
ively, as well as 5 mm dCTP (H121A) or dTTP (E138A)
and 20 mm magnesium chloride in 50 mm Hepes, pH 6.8.

tron-density maps revealed density for the nucleotide
which was built using the program o [36]. However, the
final model only showed electron density for a magnesium
ion in one of the molecules of the asymmetric unit (chain
B) and there was no electron density visible for the
c-phosphate of dTTP. Therefore, the structure was
modelled with dTDP in the active sites. There was no
electron density for the C-terminal 20 amino acid residues
that were omitted from the model. Initially, the stretch of
amino acid residues from 121 to 125 was also unclear and
was excluded from the model. Cycles of refinement using
noncrystallographic symmetry restraints with refmac5 [37]
and model building with o [36] were performed and now
enabled model building of the 121–125 stretch and a new
position of a helix 2, containing residues 55–65 in one of
the molecules in the asymmetric unit (chain B). Further-
more, the model contains residues 1–174 in chain A, resi-
dues 1–171 in chain B and three water molecules in each
chain. The structure of the H121A mutant enzyme cocrys-
tallized with dCTP was determined using difference Fou-
rier techniques with the model of the E138A protein
crystallized in the same space group (P6
3
22). Refinement
and model building proceeded as for E138A cocrystallized
with dTTP. The final model includes residues 1–171 in
chain A, residues 1–174 in chain B and two magnesium
ions, both coordinated to the modelled dCDP, because
there was no electron density for the c-phosphate of
dCTP. Data and refinement statistics are shown in

Salmonella typhimurium. J Bacteriol 105, 657–665.
4 Nucci R, Raia CA, Vaccaro C, Rossi M & Whitehead
EP (1991) Allosteric modifier and substrate binding of
dTTP inhibition of dCTP deaminase E. Johansson et al.
4196 FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS
donkey deoxycytidylate aminohydrolase (EC 3.5.4.12).
Arch Biochem Biophys 289, 19–25.
5 Ellims PH, Kao AY & Chabner BA (1983) Kinetic
behaviour and allosteric regulation of human deoxycyti-
dylate deaminase derived from leukemic cells. Mol Cell
Biochem 57, 185–190.
6 McIntosh EM & Haynes RH (1984) Isolation of a
Saccharomyces cerevisiae mutant strain deficient in
deoxycytidylate deaminase activity and partial charac-
terization of the enzyme. J Bacteriol 158, 644–649.
7 Almog R, Maley F, Maley GF, Maccoll R & Van Roey
P (2004) Three-dimensional structure of the R115E
mutant of T4-bacteriophage 2¢-deoxycytidylate deami-
nase. Biochemistry 43, 13715–13723.
8 Weiner KX, Weiner RS, Maley F & Maley GF (1993)
Primary structure of human deoxycytidylate deaminase
and overexpression of its functional protein in Escheri-
chia coli. J Biol Chem 268, 12983–12989.
9 Porter DJ & Austin EA (1993) Cytosine deaminase. The
roles of divalent metal ions in catalysis. J Biol Chem
268, 24005–24011.
10 Ko TP, Lin JJ, Hu CY, Hsu YH, Wang AH & Liaw
SH (2003) Crystal structure of yeast cytosine deaminase.
Insights into enzyme mechanism and evolution. J Biol
Chem 278, 19111–19117.

18 Johansson E, Fano M, Bynck JH, Neuhard J, Larsen S,
Sigurskjold BW, Christensen U & Willemoes M (2005)
Structures of dCTP deaminase from Escherichia coli
with bound substrate and product: reaction mechanism
and determinants of mono- and bifunctionality for a
family of enzymes. J Biol Chem 280, 3051–3059.
19 Climie S, Lutz T, Radul J, Sumner-Smith M, Vanden-
berg E & McIntosh E (1994) Expression of trimeric
human dUTP pyrophosphatase in Escherichia coli and
purification of the enzyme. Protein Expr Purif 5, 252–
258.
20 Nord J, Larsson G, Kvassman JO, Rosengren AM &
Nyman PO (1997) dUTPase from the retrovirus equine
infectious anemia virus: specificity, turnover and inhibi-
tion. FEBS Lett 414, 271–274.
21 Guillet M, Van Der Kemp PA & Boiteux S (2006)
dUTPase activity is critical to maintain genetic stability
in Saccharomyces cerevisiae. Nucleic Acids Res 34,
2056–2066.
22 Galperin MY, Moroz OV, Wilson KS & Murzin AG
(2006) House cleaning, a part of good housekeeping.
Mol Microbiol 59, 5–19.
23 Kouzminova EA & Kuzminov A (2004) Chromosomal
fragmentation in dUTPase-deficient mutants of Escheri-
chia coli and its recombinational repair. Mol Microbiol
51, 1279–1295.
24 Nord J, Kiefer M, Adolph HW, Zeppezauer MM &
Nyman PO (2000) Transient kinetics of ligand binding
and role of the C-terminus in the dUTPase from
equine infectious anemia virus. FEBS Lett 472,

33 Leslie AGW (1992) Recent changes to the MOSFLM
package for processing film and image plate data. Joint
CCP4 + ESF-EAMCB Newsletter on Protein Crystal-
lography, p. 26.
34 Evans PR (1993) Data reduction. In Proceedings of
CCP4 study weekend on data collection and processing
(Sawyer L, Isaacs N & Bailey S, eds) pp. 114–122.
Daresbury Laboratory, Warrington, UK.
35 Navaza J (1994) AMoRe: an automated package for
molecular replacement. Acta Crystallogr A50, 157–163.
36 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991)
Improved methods for building protein models in elec-
tron density maps and the location of errors in these
models. Acta Crystallogr Sect A 47, 110–119.
37 Murshudov GN, Vagin AA & Dodson EJ (1997) Refine-
ment of macromolecular structures by the maximum-
likelihood method. Acta Crystallogr Sect D 53, 240–255.
38 Laskowski RA, Macarthur MW, Moss DS & Thornton
JM (1993) Procheck – a program to check the stereo-
chemical quality of protein structures. J Appl Crystal-
logr 26, 283–291.
39 Vriend G (1990) WHAT IF: a molecular modeling and
drug design program. J Mol Graph 8 (52–56), 29.
dTTP inhibition of dCTP deaminase E. Johansson et al.
4198 FEBS Journal 274 (2007) 4188–4198 ª 2007 The Authors Journal compilation ª 2007 FEBS