Báo cáo khoa học: Structural and functional consequences of single amino acid substitutions in the pyrimidine base binding pocket of Escherichia coli CMP kinase pdf

Structural and functional consequences of single
amino acid substitutions in the pyrimidine base binding
pocket of Escherichia coli CMP kinase
Augustin Ofiteru
1
, Nadia Bucurenci
1
, Emil Alexov
2
, Thomas Bertrand
3,
*, Pierre Briozzo
3
,
He
´
le
`
ne Munier-Lehmann
4
and Anne-Marie Gilles
5
1 Laboratory of Enzymology and Applied Microbiology, Cantacuzino Institute, Bucharest, Romania
2 Department of Physics and Astronomy, Clemson University, SC, USA
3 UMR INRA-AgroParisTech 206 de Chimie Biologique, Institut National Agronomique Paris-Grignon, Thiverval-Grignon, France
4 Unite
´
de Chimie Organique, Institut Pasteur, Paris, France
5 Unite
´
de Ge

´
ne
´
tique des Ge
´
nomes Bacte
´
riens, Institut Pasteur, Paris, France
NMP kinases are key enzymes in the biosynthesis and
regeneration of ribo- and deoxyribonucleoside triphos-
phates [1]. They also participate in the activation of
prodrugs such as AZT or acyclovir which are mainly
used to treat cancer or viral infection [2]. They
catalyse reversible transfer of the c-phosphoryl group
from a nucleoside triphosphate, generally ATP, to a
particular nucleoside monophosphate according to
the scheme: Mg.ATP + NMP « Mg.ADP + NDP.
Although NMP kinases from different species are well
conserved in terms of both sequence and 3D structure,
variations in their substrate specificity [3–5] or quater-
nary structure [6–11] are frequently observed. In
eukaryotes, phosphorylation of UMP and CMP is
Keywords
CMP kinase; nucleobase specificity; protein
stability; site-directed mutagenesis; X-ray
crystallography
Correspondence
A M. Gilles, Unite
´
de Ge

´
ne
´
tique des
Ge
´
nomes Bacte
´
riens, Institut Pasteur, 28,
rue du docteur Roux, 75724 Paris, France
Fax: +33 1 45 68 89 48
Tel: +33 1 45 68 89 68
E-mail:
*Present address
Sanofi-Aventis Chemical Sciences, Vitry-sur-
Seine, France
(Received 5 February 2007, revised 2 May
2007, accepted 7 May 2007)
doi:10.1111/j.1742-4658.2007.05870.x
Bacterial CMP kinases are specific for CMP and dCMP, whereas the rela-
ted eukaryotic NMP kinase phosphorylates CMP and UMP with similar
efficiency. To explain these differences in structural terms, we investigated
the contribution of four key amino acids interacting with the pyrimidine
ring of CMP (Ser36, Asp132, Arg110 and Arg188) to the stability, catalysis
and substrate specificity of Escherichia coli CMP kinase. In contrast to euk-
aryotic UMP ⁄ CMP kinases, which interact with the nucleobase via one or
two water molecules, bacterial CMP kinase has a narrower NMP-binding
pocket and a hydrogen-bonding network involving the pyrimidine moiety
specific for the cytosine nucleobase. The side chains of Arg110 and Ser36
cannot establish hydrogen bonds with UMP, and their substitution by

hydrophobic amino acids simultaneously affects the K
m
of CMP ⁄ dCMP
and the k
cat
value. Substitution of Ser for Asp132 results in a moderate
decrease in stability without significant changes in K
m
value for CMP and
dCMP. Replacement of Arg188 with Met does not affect enzyme stability
but dramatically decreases the k
cat
⁄ K
m
ratio compared with wild-type
enzyme. This effect might be explained by opening of the enzyme ⁄ nucleo-
tide complex, so that the sugar no longer interacts with Asp185. The reac-
tion rate for different modified CMP kinases with ATP as a variable
substrate indicated that none of changes induced by these amino acid sub-
stitutions was ‘propagated’ to the ATP subsite. This ‘modular’ behavior of
E. coli CMP kinase is unique in comparison with other NMP kinases.
Abbreviations
AK, adenylate kinase; AK1, muscle cytosolic adenylate kinase; MCCE, multi-conformation continuum electrostatic.
FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS 3363
accomplished by a single enzyme [12–14]. In prokaryo-
tes, there are distinct NMP kinases for each pyrimidine
nucleotide: CMP ⁄ dCMP, UMP and TMP.
Bacterial CMP kinases (EC 2.7.4.14) conserve the
three-domain overall fold of eukaryotic UMP ⁄ CMP
kinases (EC 2.7.4.14): the central parallel b-sheet

together with surrounding a-helices, defined as the
CORE domain, is conserved in NMP kinases. It is
used as a rigid platform around which the short a-heli-
cal LID domain, situated in the C-terminal moiety,
and the NMP-binding (NMP
bind
) domain move in an
induced-fit mechanism, closing upon binding of the
phosphate donor and acceptor nucleotides, respectively
[15].
The crystal structure of Escherichia coli CMP kinase,
either alone or in complex with the reaction product
CDP or various NMPs (CMP, dCMP, AraCMP and
ddCMP), underlined the residues involved in recogni-
tion of the nucleobase, pentose moiety and phosphate
group(s) [15], and site-directed mutagenesis experi-
ments have further confirmed the role of Ser101,
Arg181 and Asp185 in pentose recognition [16]. How-
ever, the main difference between eukaryotic and
bacterial NMP kinases concerns the recognition of
pyrimidine nucleotides. The structure of E. coli CMP
kinase in complex with CMP or dCMP showed that
discrimination between CMP and UMP is achieved by
Ser36, Arg110 and Asp132, which form hydrogen
bonds with the amino group and the N3 atom of the
cytosine (Fig. 1). This study uses site-directed muta-
genesis to further explore the contribution of these
amino acids interacting with the pyrimidine ring to the
catalysis of E. coli CMP kinase. Substitution of the
side chain from a well-structured protein can have two

types of consequence: (a) a purely localized effect of
binding due to removal of a specific interaction
between the enzyme and its substrate; (b) a more glo-
bal effect due to subtle or gross changes in enzyme
conformation. Therefore, our study was completed
using numerical calculations to better emphasize the
role of each of these residues on protein stability. The
results highlight the importance of the hydrogen-
CMP UMP
N
O
OH
N
O
HO
NH
2
O
OH
P
HO
O
N
O
OH
O
HO
O
HO
P

HO
O
NH
O
D129
S3
CMP
R188
O
N4
N3O2
R110
D132
D185
OG
O2’
O3’
Fig. 1. Comparative structures of CMP and
UMP, and interactions between the cyto-
sine moiety of nucleotide and various side
chains of wild-type CMP kinase. (Upper)
Chemical differences between CMP and
UMP are indicated in red. (Lower) Hydrogen
bonds are indicated with green dots, carbon
atoms being indicated in grey (enzyme) or
yellow (nucleotide). D185, a residue close to
R188 and involved in ribose binding is also
shown. Drawn using
PYMOL [41].
Substrate specificity of E. coli CMP kinase A. Ofiteru et al.

3364 FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS
binding network surrounding the cytosine moiety in
the specificity of the enzyme for the acceptor nucleo-
tide. Because this specificity is characteristic of bacter-
ial CMP kinases, these enzymes represent possible
targets for antibacterial drugs [17].
Results
Overproduction and molecular characterization of
the modified variants of E. coli CMP kinase
The wild-type and various modified forms (S36A,
R110M, D132A, D132H, D132N, D132S and R188M)
of E. coli CMP kinase overproduced in strain
BL21(DE3) represented between 25 and 30% of sol-
uble E. coli proteins. Recombinant enzymes adsorbed
onto a Blue-Sepharose column equilibrated with
50 mm Tris ⁄ HCl pH 7.4, were then eluted with 1 m
NaCl. In the case of the D132S mutant, higher NaCl
concentrations (2 m) were required for complete elu-
tion of the protein. Gel permeation chromatography
on Ultrogel AcA54 yielded pure enzymes, which
according to appropriate markers corresponded to
monomers. After prolonged dialysis against ammo-
nium bicarbonate a small proportion of dimers were
formed, even in the case of the wild-type protein as
indicated by ESI-MS or SDS ⁄ PAGE in the absence of
reducing agents. The proportion of dimers increased
notably in D132A and D132S mutants.
Thermal denaturation experiments, summarized in
Table 1, indicated that S36A and R188M substitutions
did not affect protein stability, T

m
(melting tempera-
ture) values being identical or very close to that of the
native enzyme. Other substitutions (D132S, D132N
and D132H) led to a moderate decrease in stability,
lowering T
m
by 4–5 °C compared with the wild-type
protein. The last group includes amino acid substitu-
tions (D132A and R110M) that noticeably lowered the
stability of CMP kinase.
Limited proteolysis experiments did not detect signi-
ficant differences between wild-type CMP kinase and
its variants. The first-order rate constant of inactiva-
tion by TPCK-trypsin at 4 °C was found to be around
3 · 10
)3
Æs
)1
. Addition of ATP protected all modified
CMP kinases, decreasing the first-order rate constant
of inactivation by a factor of 5–10 (data not shown).
Effect of charge alterations on protein stability
Because all mutations altered the protein charge, we
evaluated their potential effect with numerical calcula-
tions using the CMP ⁄ CMP kinase model (PDB code
1kdo) for each of the sites selected for site-directed
mutagenesis (Table 1).
Numerical calculations showed that S36 is not
involved in significant interactions with the side chains

of its neighbouring protein residues. By contrast, S36
forms a strong hydrogen bond with the backbone of
D129. However, the favourable energy of the hydrogen
bond is almost completely cancelled out by the desol-
vation penalty of S36. Thus, its replacement by Ala is
not expected to change the protein stability.
In the wild-type protein, R188 is involved in salt-
bridge with D185 and in many other interactions with
neighbouring residues such as R110, D132. Despite
this complicated network of interactions, R188M sub-
stitution does not have a significant effect on the
experimentally measured protein stability. However,
calculations using the multi-conformation continuum
electrostatic (MCCE) method for the ionization states
of native and R188M-modified structures revealed a
major difference. In the absence of R188, D185 is cal-
culated to be neutral (protonated). Thus, by turning
off both charges (of R188 and D185), the protein
reduces the effect of amino acid substitution on
enzyme stability to almost zero. This a typical example
of charge rearrangement caused by an amino acid
substitution.
In the wild-type enzyme, D132 is also involved in
a complicated network of interactions, the strongest
being with R110. The energy balance in wild-type
CMP kinase shows that D132 contributes to the stabil-
ity by )37.7 kJ. Thus, replacing this residue with Ser,
Ala or His should have a significant effect on stability,
depending on the substituting residue. MCCE calcula-
tions showed that in all modified forms of CMP kinase

the R110 side chain reorients and becomes more
exposed to the solution. This reduces the energy cost
of the mutation. No change in the ionization states
was found to be induced by the mutation. However,
the D132H variant does not introduce charge reversal,
because His is calculated to be deprotonated (neutral)
Table 1. Thermal stability of E. coli CMP kinase variants (T
m
) and
calculated stability changes (DDG) upon amino acid substitution
with respect to the energy of the wild-type enzyme.
Enzyme T
m
(°C) DDG (kJ)
Wild-type 52 –
S36A 52 0
D132S 48 + 33.1
D132A 43 + 37.7
D132N 47 + 29.7
D132H 48 + 13.0
R110M 45 + 55.3
R188M 51 0
A. Ofiteru et al. Substrate specificity of E. coli CMP kinase
FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS 3365
in modified CMP kinase. Thus, the three variants
D132S, D132N and D132H result in replacement of a
negatively charged residue with a polar residue. Each
of the substituting residues is involved in favourable
interactions with its neighbours and thus further redu-
ces the effect of the mutation. This is why D132A sub-

stitution has the largest effect on stability. The side
chain of Ala, located in a very hydrophilic environ-
ment does not have favourable energy and further
destabilizes the mutant.
R110 is involved in many interactions but, as shown
previously, its major partner is D132. Removal of
R110 leaves D132 without the favourable pairwise
energy but D132 is calculated to be still ionized. Thus,
in contrast to R188M where D185 plays a compensa-
tory role, D132 does not and this results in a dramatic
decrease in protein stability upon R110M mutation.
However, the calculated energy change for the R110M
mutation is quite similar to that for D132A (Table 1).
Kinetic properties of the modified variants of
E. coli CMP kinase with CMP, dCMP and UMP
as variable substrates
Substitution by hydrophobic side chains of the four
amino acids demonstrated by crystallography as inter-
acting with the cytosine moiety of CMP and dCMP,
always affected the kinetic parameters of bacterial
CMP kinase (Table 2). The S36A substitution mainly
changed the K
m
value for the two natural substrates,
which increased by a factor of 70 (CMP) and 37
(dCMP) compared with the parent molecule. The
decrease in k
cat
of only 1.6-fold (CMP) and 7.4-fold
(dCMP) with respect to the wild-type enzyme sugges-

ted that the major role of S36 is related to
CMP ⁄ dCMP binding to the active site. The S36A sub-
stitution did not significantly affect phosphorylation of
UMP. We note that S36, which is common to CMP
kinases from Gram-negative organisms, alternates in
CMP kinases from Gram-positive bacteria with a Thr
residue, but never with Ala as in the case of Dictyoste-
lium discoideum UMP ⁄ CMP kinase. By contrast, the
A37T substitution in the slime mold enzyme (A. M.
Gilles, P. Glaser and L. L. Ylisastigui-Pons, unpub-
lished data) or the T39A substitution in the pig muscle
cytosolic adenylate kinase [18] had no consequence on
binding or phosphorylation of the corresponding
NMPs (A37 and T39 in these enzymes are equivalent
to S36 in E. coli CMP kinase).
Substitution of R110 with Met in E. coli CMP kin-
ase affected both k
cat
and K
m
. The k
cat
⁄ K
m
ratio with
CMP and dCMP decreased by a factor of > 10
5
in the
modified protein compared with wild-type enzyme.
The k

cat
⁄ K
m
ratio with UMP as substrate decreased by
a factor of only 200 compared with wild-type CMP
kinase. The loss in stability of R110M mutant might
be responsible at least in part for the modified kinetic
properties. This is not the case for the R188M variant,
whose thermal stability is similar to that of the wild-
type enzyme; however, the k
cat
⁄ K
m
ratio with CMP
and dCMP decreased by a factor > 10
4
compared with
wild-type enzyme.
To explain these effects, the R188M variant was
crystallized either alone or in complex with dCMP.
Information on data collection, processing, refinement
and model statistics are given in Table 3. For the free
enzyme, the structure of this mutant could be success-
fully refined. It was found to be identical to that of the
wild-type CMP kinase. By contrast, the data for this
variant in complex with dCMP were of poor quality
due to anisotropy of the crystals. As a consequence,
the model was refined to a R
cryst
of 25.9% and a R

free
of 33.9%. The R
free
⁄ R
cryst
ratio is 1.31, which is not
Table 2. Kinetic parameters of E. coli CMP kinase variants with
three NMP substrates at a single fixed concentration of ATP
(1 m
M). Curve-fit was performed using the nonlinear least-squares
fitting analysis of
KALEIDAGRAPH software. K
m
is the Michaelis–Men-
ten constant; k
cat
was calculated assuming a molecular mass of
E. coli CMP kinase of 24.7 kDa. Values are means of two to four
independent measurements.
Enzyme Nucleotide
K
m
(mM)
k
cat
(s
)1
)
k
cat

⁄ K
m
(s
)1
ÆmM
)1
)
Wild-type CMP 0.035 103 2940
dCMP 0.094 108 1150
UMP 0.93 0.82 0.88
S36A CMP 2.5 63 25.2
dCMP 3.5 14.5 4.1
UMP 1.9 0.57 0.30
D132S CMP 0.038 22.4 589
dCMP 0.090 21.1 230
UMP 8.0 8.3 1.04
D132A CMP 2.9 4.1 1.41
dCMP 1.8 0.73 0.40
UMP 7.9 9.9 1.25
D132N CMP 2.6 1.4 0.54
dCMP 0.08 0.15 1.88
UMP 5.4 0.45 0.083
D132H CMP 1.3 0.069 0.053
dCMP 0.055 0.06 1.1
UMP 3.9 0.013 0.0033
R110M CMP 20.2 0.23 0.0114
dCMP 7.3 0.05 0.0068
UMP 11.3 0.054 0.0048
R188M CMP 1.0 0.12 0.12
dCMP 0.77 0.04 0.052

UMP Not detectable
Substrate specificity of E. coli CMP kinase A. Ofiteru et al.
3366 FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS
unusual for a 2.8 A
˚
resolution structure [19]. However,
the electron-density map of the substrate-binding
region was unambiguous. As shown in Fig. 2, the
structure is in an ‘open’ form in which deoxyribose
does not establish H bonds with enzyme residues. In
the structure of the wild-type CMP kinase in complex
with dCMP [16], there are two, quite different, mole-
cules in the asymmetric unit (rmsd ¼ 0.89 A
˚
). In the A
molecule, the 3¢OH from deoxyribose forms hydrogen
bonds with D185 and R181 residues as for CMP bind-
ing. The R188M variant in complex with dCMP is in
many respects comparable with the B molecule of
wild-type CMP kinase complexed to dCMP, in which
the deoxyribose does not interact with R181 or D185.
Moreover, the hydrogen bond, which connected R188
and the carbonyl from cytosine, is lost. It seems there-
fore that the role of R188 is to maintain a closed
structure of the protein by direct interaction with the
substrate (Fig. 1).
Replacement of D132 with Ala had the most dra-
matic consequences on the protein stability as T
m
decreased by 9 °C in comparison with the wild-type

protein. The k
cat
⁄ K
m
ratio for CMP and dCMP
decreased by more than three orders of magnitude in
comparison with the wild-type enzyme, indicating a
loss of 19.2 and 20.0 kJ per mole in the stability of the
transition state complex. At the same time, the k
cat
⁄ K
m
ratio with UMP as substrate remained unchanged.
Moreover, the k
cat
value with UMP increased by one
order of magnitude with respect to the wild-type
enzyme, at the expense of the K
m
value. The overall
effect of this structural change is a twofold increase in
the k
cat
⁄ K
m
ratio with UMP as substrate over the
k
cat
⁄ K
m

ratio with CMP as substrate. The relative
increase in the specificity of the D132A variant for
UMP over CMP and dCMP is somehow unexpected.
It reflects an increase in local flexibility of the polypep-
tide chain with loss of discrimination between the three
nucleotides. Because of its size, polarity and charge
D132 plays a unique role in both protein stability and
kinetic properties. Consequently, several other variants
were explored in which each of these properties of the
aspartate side chain was altered individually. Because
the D132S variant essentially conserved the properties
of the wild-type enzyme, it appears that the major fac-
tor in CMP ⁄ dCMP recognition by D132 is its hydro-
gen-bonding capacity. The kinetic parameters of the
D132H variant were modified in the expected sense
because the charge of this residue was removed. The
D132N substitution was designed to ‘conserve’ the size
of the original molecule and part of its hydrogen-
Table 3. Structural data.
Data set R188M R188M–dCMP
Data collection
Wavelength (A
˚
) 1.5418 0.9490
Space group P6
3
P4
1
2
1

2
Unit cell (A
˚
, °)
a ¼ b 82.17 72.95
c 60.73 76.97
a ¼ b 90 90
c 120 90
Resolution (A
˚
) 1.9 2.8
Observed reflections 35412 61754
Unique reflections 18054 5424
Completeness (%) 97.9 (86.1)
a
92.3 (84.6)
a
I ⁄ r(I) 15.0 (2.9)
a
7.8 (2.6)
a
R
sym
b
(%) 3.9 (32.4)
a
9.2 (37.3)
a
Refinement statistics
R

cryst
c
(%) 23.1 1 25.9
R
free
d
(%) 24.2 1 33.9
rmsd
Bond lengths (A
˚
) 0.007 0.009
Bond angles (°) 1.27 1.56
a
Numbers in parentheses represent values in the highest resolu-
tion shell (last of 20 shells).
b
R
sym
¼ S
h
S
i
|I(h,i) ) < I(h) > | ⁄S
h
S
i
I(h,i) where I(h,i) is the intensity value of the i-th measurement of h
and < I(h) > is the corresponding mean value of I(h) for all i meas-
urements.
c

R
cryst
¼ S ||F
obs
| ) |F
calc
|| ⁄S |F
obs
|, where |F
obs
| and
|F
calc
| are the observed and calculated structure factor amplitudes,
respectively.
d
R
free
is the same as R
cryst
but calculated with a 10%
subset of all reflections that was never used in crystallographic
refinement.
D129
S36
dCMP
M188
R110
D132
D185

N4
Fig. 2. Electron-density map of the dCMP-binding region for the
R188M CMP kinase variant. The F
o
) F
c
omit map calculated for
the dCMP–R188M CMP kinase complex in the absence of the
dCMP model is green. The contour level is at 2 r. The same neigh-
bouring enzyme residues as those of Fig. 1 are shown, in the same
orientation, with their 2F
o
) F
c
map in magenta (contour level 1 r).
The only hydrogen bond still observed for dCMP–R188M CMP
kinase complex is indicated by green dots.
A. Ofiteru et al. Substrate specificity of E. coli CMP kinase
FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS 3367
bonding ability. However, it strongly affected the
stability and catalytic properties of the protein in com-
parison with the wild-type enzyme. This suggests that
both hydrogen bonds (Fig. 1) received by the side
chain of D132, from the H bond donors R110 (side
chain extremity) and N4 from cytosine, are important
for the enzyme. As shown in Table 2, introduction of
a H-bond donor via Asn substitution of D132 is
not compatible with enzyme binding and substrate
catalysis.
Kinetic and nucleotide-binding properties of

E. coli CMP kinase variants with ATP as variable
substrate
Determination of the reaction rates of different modified
CMP kinases with ATP as the variable substrate at fixed
concentrations of NMP (around the corresponding K
m
values) yielded apparent K
m
values for ATP between
0.04 and 0.08 mm, irrespective of the chemical nature of
NMP or the substituted residue. This was also con-
firmed by fluorescence experiments using Ant-dATP as
a reporter molecule [20]. The K
d
value for the complex
of various proteins with the fluorescent derivative was
between 4 and 10 lm, whereas K
d
values for complexes
with ATP varied between 14 and 25 lm. This means that
structural modifications affecting the NMP subsite of
the catalytic centre of bacterial CMP kinases are not
‘propagated’ to the ATP subsite. In this respect, E. coli
CMP kinase is unique in comparison with other NMP
kinases, in particular with adenylate kinases [21].
Discussion
A common property of various NMP kinases, except
for bacterial UMP kinases, is an overall fold consisting
of three domains, the CORE, the LID and the
NMP

bind
[1]. A characteristic of bacterial CMP kinases
is an extension of the NMP
bind
domain by 40 amino
acid residues forming a three-stranded antiparallel
b sheet and two a helices. This large NMP
bind
insert
undergoes rearrangement during the binding of cyto-
sine nucleotides, its b sheet moving away from the
substrate and the a helices coming closer to it [15].
Sequence comparison of E. coli or many other bacter-
ial CMP kinases indicated that the basic residues inter-
acting with the phosphate group of CMP or dCMP
(R41, R131 and R181) are conserved in NMP kinases
irrespective of the chemical nature of the acceptor sub-
strate (Fig. 3). Thus, R41 is conserved as R42 in
D. discoideum UMP ⁄ CMP kinase and as R44 in pig
muscle cytosolic adenylate kinase (AK1). The R44M
substitution in pig muscle AK1 decreases over two
orders of magnitude the k
cat
⁄ K
AMP
m
ratio in comparison
with the wild-type enzyme [18]. Similarly, R131 in
E. coli CMP kinase is conserved as R93 in D. discoideum
UMP ⁄ CMP kinase, R96 in human UMP ⁄ CMP kinase

and R97 in pig muscle AK1. R97M substitution in the
latter enzyme decreases, by three orders of magnitude,
the k
cat
⁄ K
AMP
m
ratio in comparison with wild-type pro-
tein [21]. Finally, R181 in E. coli CMP kinase is con-
served as R149 in pig muscle AK1. Substitution of
these residues by the hydrophobic side-chain of methi-
onine decreases in these two enzymes both the K
m
for
NMP and the k
cat
compared with the parent molecules
[16,18]. In conclusion, the amino acids interacting with
the phosphate group of NMP and conserved in
eukaryotic UMP ⁄ CMP kinases, bacterial CMP kinases
and eukaryotic or bacterial adenylate kinases, most
probably have identical roles during catalysis.
The situation is different when comparing the
nucleobase recognition by eukaryotic UMP⁄ CMP
kinases and bacterial CMP kinases. Thus, despite
opposing hydrogen-bonding properties at positions 3
and 4 of the pyrimidine ring, UMP and CMP are
phosphorylated with similar efficiency by D. discoideum
UMP ⁄ CMP kinase [13]. This might be explained by
the fact that the base located in a hydrophobic pocket

of D. discoideum enzyme interacts with the protein
indirectly, via one (with CMP) or two (with UMP)
close water molecules connected to the carboxamide
group of N97. The side chain of N97, like the water
molecule, can switch to either hydrogen bond accep-
tor or donor depending on its orientation, and pro-
vides a flexible way to accommodate either CMP or
UMP in the NMP-binding site [22]. The residues
forming the hydrophobic pocket in D. discoideum
UMP ⁄ CMP kinase are conserved in the equivalent
human or yeast enzymes. This scenario is not compat-
ible with bacterial CMP kinases in which UMP is a
very poor substrate compared with CMP. The side
chain of R110 is a hydrogen bond donor to the N3
atom of cytosine. As the main chain carbonyl of
D129, a H-bond acceptor, interacts with the terminal
oxygen of S36 side chain, the latter can only behave
as a H-bond acceptor with the nucleobase as is the
case with the four-amino group of cytosine. These
hydrogen bonds involving side chains from R110 and
S36 could not be established with UMP.
Each of these residues could, in principle, also be
important for stability of the protein. Partial coupling
between the structural and functional roles can be
observed for D132. Substitution by Ser results in a
moderate decrease in stability without significant chan-
ges in K
m
for CMP and dCMP. However, replacement
of D132 with Asn or His results in a completely

Substrate specificity of E. coli CMP kinase A. Ofiteru et al.
3368 FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS
different K
m
for CMP, but only slightly affects the K
m
for dCMP. Finally, replacement of D132 with Ala has
a dramatic effect on both stability and activity. This
indicates that the residue at position 132 is structurally
important, and also plays a role in the reaction, most
probably as hydrogen acceptor to CMP ⁄ dCMP. Much
more prominent is the coupling effect for the Arg resi-
due at position 110. Mutation of Arg to Met causes a
dramatic decrease in both the stability and reaction
rate for all substrates. In contrast, residues 36 and 188
are ‘pure’ functional ones. Substitution of Ser36 to Ala
does not change the stability but has great effect on
the activity. Thus, Ser36 is not important energetically
but serves as a hydrogen bond acceptor for the sub-
strate. R188M substitution does not affect the stability,
but this may be related to the compensatory role of
Asp185. Thus, the salt bridge R188–D185 has a negli-
gible contribution to the stability of the protein and its
substitution does not cause a change in T
m
. However,
suppression of the R188–D185 ‘bridge’ has a dramatic
effect on the reaction.
Analysis of the effect of the mutations on protein
stability revealed three distinctive mechanisms of

relaxation, and in the case of S36A, no relaxation at
all. The first type of relaxation (structural relaxation),
which involves only proton and side chain motions, is
seen in D132 and R110 mutants. Substitution of either
D132 or R110 disrupts the salt bridge D132–R110
and causes the partner side chain to adopt a different
conformation, thereby reducing the effect of the muta-
tion. The other two types of relaxation are mainly
charge relaxation. Replacing D132 with His is sup-
posed to reverse the charge at position 132 and should
have a dramatic effect on stability. However, the sub-
stituting residue is calculated to be neutral and thus
has a zero net charge. Unfavourable interactions with
R110 make the pK
a
value for H132 far below the phy-
siological pH and thus turn off the His charge.
Because the pK
a
of isolated His is 6.5, the difference
in ionization energy of His in the denaturated state
(where His is presumably ionized) and in the protein
is very small and does not affect the results [23]. The
Fig. 3. Sequence alignment of E. coli CMP kinase with human, D. discoideum and yeast UMP ⁄ CMP kinases and with pig muscle cytosolic
adenylate kinase (AK1), respectively. Residues common to all proteins are indicated in black, residues common to the last four enzymes are
in grey. Asterisks and triangles on the top of sequences indicate conserved residues involved in the interaction with the phosphate moieties
of various NMPs (R41, R131 and R181 in E. coli CMP kinase) and the four modified residues of E. coli CMP kinase specifically interacting
with the cytosine moiety (S36, R110, D132 and R188), respectively.
A. Ofiteru et al. Substrate specificity of E. coli CMP kinase
FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS 3369

third case is a charge relaxation involving neighbour-
ing group. Substitution of R188 to Met does not
affect stability because the removal of R188 causes de-
protonation of its partner D185 and thus the net
effect is almost zero. A similar effect was suggested to
occur in the reaction centre when particular residues
were mutated [24]. This effect can be explained in
a different manner considering the salt bridge
R188–D185 as a dipole from the distal point of the
rest of the protein. Such a dipole will have weak inter-
actions with the rest of the protein. Thus, if turned off
(by removal of Arg and protonation of Asp185), the
protein energy should not change by much, as found
experimentally.
Experimental procedures
Chemicals
Nucleotides, restriction enzymes, T4 DNA ligase, T4 DNA
polymerase and coupling enzymes were from Roche
Applied Sciences (Indianapolis, IN). T7 DNA polymerase
was from Amersham-Biosciences (Piscataway, NJ). Affi-Gel
Blue was from Bio-Rad Laboratories (Hercules, CA).
CMP, dCMP, AraCMP and UMP were purchased from
Sigma (St Louis, MO). NDP kinase from D. discoideum
(2000 U mg
)1
of protein) was kindly provided by M. Ve
´
ron
(Institut Pasteur, Paris).
Bacterial strains, plasmids, growth conditions

and DNA manipulation
Site-directed mutagenesis was performed according to Kun-
kel et al. [25] with single-stranded DNA of pHS210 [20]
grown in E. coli strain CJ236 in the presence of the helper
phage M13K07. The primers used to create the point
mutations, where the changed codons are underlined, were:
S36A: AATTGCACC
TGCGTCCAGCAGATG; R110M:
TAATGCTTC
CATAACGCGTGGGAA; D132A: CGTTC
CCAT
TGCGCGGCCATCGGC; D132H: TACCACCGT
TCCCAT
ATGGCGGCCATCGGCAAT; D132N: TACCAC
CGTTCCCAT
ATTGCGGCCATCGGCAAT; D132S:
TAC CACGTTCCCAT
GGAGCGGCCATCGGCAAT;
R188M: CGCTACCGC
CATGTTACGATCGCG. For
each mutagenesis, the whole sequence of the cmk gene was
checked for the absence of any other mutation [26]. Plasmid
pHS210 and derivatives were introduced into the E. coli
strain BL21(DE3) ⁄ pDIA17 [27]. Overproduction was car-
ried out by growing bacteria at 37 °C in 2YT medium [28]
supplemented with ampicillin (100 lgÆmL
)1
) and chloram-
phenicol (30 lgÆmL
)1

). When A
600
¼ 1.5, isopropyl thio
b-d-galactoside (1 mm final concentration) was added to
the medium. Bacteria were harvested by centrifugation 3 h
after induction at 5000 g for 15 min at 4 °C (Sorvall RC 5B).
Purification of the enzymes, activity assays and
other analytical procedures
Overproduced wild-type and modified variants of E. coli
CMP kinase were purified as described previously [20] and
checked by MS (a quadrupole API-365 mass spectrometer
from Perkin-Elmer, Norwalk, NJ) equipped with an ion
spray (nebulizer-assisted electrospray) source. Protein con-
centration was measured according to Bradford [29].
SDS ⁄ PAGE was performed as described by Laemmli [30].
Enzyme activity was determined at 30 °C and 340 nm using
a coupled spectrophotometric assay in 0.5 mL final volume
on an Eppendorf ECOM 6122 photometer [31]. The reac-
tion medium contained 50 mm Tris ⁄ HCl (pH 7.4), 50 mm
KCl, 2 mm MgCl
2
,1mm phosphoenolpyruvate, 0.2 mm
NADH, different concentrations of ATP and NMPs, and
2 units each of pyruvate kinase, lactate dehydrogenase and
NDP kinase (forward reaction). The rate was calculated
assuming that two ADP are generated during the reaction.
One unit of CMP kinase corresponds to 1 lmol of product
formed per minute. The thermal stability of CMP kinase
variants was tested by incubating the purified enzymes
(1 mgÆmL

)1
)in50mm Tris ⁄ HCl (pH 7.4) containing 0.1 m
NaCl at temperatures between 30 and 60 °C for 10 min.
The results, expressed as the percentage of residual activity
compared with unincubated controls, were used to calculate
the temperature of half inactivation (T
m
) of each variant.
Proteolysis of bacterial CMP kinase (1 mgÆmL
)1
in 50 mm
Tris ⁄ HCl, pH 7.4) was followed at 4 °C in the presence of
2 lgÆmL
)1
of TPCK-trypsin. At different time intervals,
aliquots were withdrawn and diluted in buffer containing
10 lgÆmL
)1
of soybean trypsin inhibitor. The first-order
rate constant of inactivation of CMP kinase by TPCK-tryp-
sin was calculated from the log
10
of residual activity versus
the time. Binding of nucleotides to E. coli CMP kinase was
measured from the fluorescence of Ant-dATP (k
exc
¼
330 nm, k
em
¼ 420 nm) on a Jasco spectrofluorimeter

FP 750, thermostated at 25 °C using a UV-grade quartz
cuvette [20,32].
Numerical calculations
The MCCE [33–35] method was used to calculate the ion-
ization states, polar hydrogen positions and possible side
chain rearrangements in the native structure and the corres-
ponding variants. In all calculations, we used the PDB file
(code: 1-KDO) of E. coli CMP kinase in complex with its
major natural substrate CMP. The effect of a mutation on
the stability of CMP kinase is calculated as:
DG
i
ðmutÞ¼ÀDG
self
i
ðWTÞÀ
X
N
j ¼1;
j 6¼ i
DG
pairwise
i;j
ðWTÞ
þ
X
k
DDG
self
k

ðmutÞþ
X
k
X
N
j ¼1;
DDG
pairwise
k;j
ðmutÞð1Þ
Substrate specificity of E. coli CMP kinase A. Ofiteru et al.
3370 FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS
where ÀDG
self
i
ðWTÞ is the loss of the self energy in the
wild-type enzyme of the residue that was mutated,
DG
pairwise
i;j
ðWTÞ are the pair wise energies of the original resi-
due ‘i’ in the native protein with the rest of the residues,
DDG
self
k
ðmutÞ are the changes of the self energies of the resi-
dues ‘k’ that change either their ionization or conformation
upon the mutation and DDG
pairwise
k;j

ðmutÞ are the pair-wise
energies changes caused by the mutation. All energies are
calculated with respect to hypothetical unfolded state of
extended polypeptide (noninteracting residues assumption).
This is an obvious oversimplification, but because we are
interested in the difference in stability of wild-type versus
modified proteins, the vast part of the possible error will
cancel out – most probably both denaturated states (wild-
type and the mutant) will be very similar. Thus, the first
two terms account for the loss of the protein energy of the
residue that is mutated and the last two terms account for
the change of protein energy due to ionization or confor-
mation changes in the mutant protein.
Preparation of the structures used in the
calculations
Calculations on the wild-type CMP kinase were performed
on the 1-KDO file (CMP ⁄ CMP kinase complex) using
molecule A of the asymmetric unit. The rmsd between mole-
cules A and B of the asymmetric unit is only 0.45 A
˚
and
thus this choice is not critical for the calculations. Side
chain mutations were performed with scap [36]. The pro-
tons were generated with MCCE.
Crystallography of the R188M variant
Two types of crystals were studied: enzyme alone (R188M)
and in complex with nucleotide (R188M–dCMP). They
were grown at 20 °C using the vapour-diffusion method, in
a50mm Tris ⁄ HCl buffer pH 7.4, with a hanging droplet
(6 lL) containing 10 mgÆmL

)1
of the R188M CMPKeco
variant, and in the case of dCMP–R188M complex with a
large excess of nucleotide (200 mm). Drops were equili-
brated with a reservoir solution (1 mL) containing the pre-
cipitant ammonium sulphate (1.3 m in the case of enzyme
alone, and 1.7 m for the R188M–dCMP variant). Diffrac-
tion data were collected at room temperature on a Rigaku
rotating-anode RTP 300 RC X-ray generator for crystal of
the enzyme alone, and at 100°K (using glycerol as a cryo-
protectant) on the LURE synchrotron (beamline DW32) in
Orsay, France for R188M–dCMP crystal. Crystals of the
enzyme alone belong to the hexagonal space group P6
3
,
those with dCMP to the tetragonal space group P4
1
2
1
2. In
both cases there is one molecule per asymmetric unit. Dif-
fraction data were processed using denzo and scaled and
reduced with scalepack [37]. The structures were solved by
molecular replacement with amore [38], using the wild-type
enzyme as the search model. Models were built with o [39],
and refined with cns [40]. The first refinement steps used
simulated annealing. For the dCMP–R188M complex, the
dCMP density was unambiguous before this ligand was
included in the refinement. The Protein Data Bank codes
are 2FEM for the R188M enzyme alone and 2FEO for the

R188M–dCMP complex.
Acknowledgements
We thank O. Baˆ rzu for interest and continuous sup-
port, Y. Janin for carefully reading this manuscript
and constructive criticism, L. Tourneux for providing
some modified forms of CMP kinase. This work was
supported by grants from Institut Pasteur, the Centre
National de la Recherche Scientifique (URA2185,
URA2171, URA2128), the Institut National de la
Sante
´
et de la Recherche Me
´
dicale, and the Institut
National de la Recherche Agronomique (UMR 206).
References
1 Yan H & Tsai M-D (1999) Nucleoside monophosphate
kinases: structure, mechanism, and substrate specificity.
Adv Enzymol Related Areas Molec Biol 73, 103–133.
2 Mitzuya H, Weinhold KJ, Furman PA, St Clair MH,
Nusinoff-Lehrman S, Gallo RC, Bolognesi D, Barry
DW & Broder S (1985) 3¢-Azido-3¢-deoxythymidine
(BW A509U): an antiviral agent that inhibits the
infectivity and cytopathic effect of human T-lympho-
tropic virus type III ⁄ lymphadenopathy-associated virus
in vitro. Proc Natl Acad Sci USA 82, 7096–7100.
3 Chenal-Francisque V, Tourneux L, Carniel Christova P,
Li de la Sierra I, Baˆ rzu O & Gilles A-M (1999) The
highly similar TMP kinases of Yersinia pestis and
Escherichia coli differ markedly in their AZTMP phos-

phorylating activity. Eur J Biochem 265, 112–119.
4 Lavie A, Ostermann N, Brundiers R, Goody RS, Rein-
stein J, Konrad M & Schlichting I (1998b) Structural
basis for efficient phosphorylation of 3¢-azidothymidine
monophasphate by Escherichia coli thymidylate kinase.
Proc Natl Acad Sci USA 95, 14045–14050.
5 Munier-Lehmann H, Chaffotte A, Pochet S & Labesse
G (2001) Thymidylate kinase of Mycobacterium tuber-
culosis: a chimera sharing properties common to euk-
aryotic and bacterial enzymes. Protein Sci 10, 1195–
1205.
6 Gentry D, Bengra C, Ikehara K & Cashel M (1993)
Guanylate kinase of Escherichia coli K-12. J Biol Chem
268, 14316–14321.
7 Sekulic N, Shuvalova L, Spangenberg O, Konrad M &
Lavie A (2002) Structural characterization of the closed
conformation of mouse guanylate kinase. J Biol Chem
277, 30236–30243.
A. Ofiteru et al. Substrate specificity of E. coli CMP kinase
FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS 3371
8 Perrier V, Burlacu-Miron S, Boussac A, Meier A &
Gilles A-M (1998) Metal chelating properties of
adenylate kinase from Paracoccus denitrificans. Protein
Eng 11, 917–923.
9 Vonrhein C, Bo
¨
nish H, Scha
¨
fer G & Schulz GE (1998)
The structure of a trimeric archaeal adenylate kinase.

J Mol Biol 282, 167–179.
10 Hible G, Renault L, Schaeffer F, Christova P, Zoe
Radulescu A, Evrin C, Gilles AM & Cherfils J (2005)
Calorimetric and crystallographic analysis of the oligo-
meric structure of Escherichia coli GMP kinase. J Mol
Biol 352, 1044–1059.
11 Hible G, Christova P, Renault L, Seclaman E, Thomp-
son A, Girard E, Munier-Lehmann H & Cherfils J
(2006) Unique GMP-binding site in Mycobacterium
tuberculosis guanosine monophosphate kinase. Proteins
62, 489–500.
12 Mu
¨
ller-Dieckmann H-J & Schulz GE (1995) Substrate
specificity and assembly of the catalytic center derived
from two structures of ligated uridylate kinase. J Mol
Biol 246, 522–530.
13 Wiesmu
¨
ller L, Noegel AA, Baˆ rzu O, Gerish G & Schlei-
cher M (1990) cDNA-derived sequence of UMP–CMP
kinase from Dictyostelium discoideum and expression of
the enzyme in Escherichia coli. J Biol Chem 265 , 6339–
6345.
14 Pasti C, Gallois-Montbrun S, Munier-Lehmann H,
Veron M, Gilles AM & Deville-Bonne D (2003)
Reaction of human UMP–CMP kinase with natural and
analog substrates. Eur J Biochem 270, 1784–1790.
15 Briozzo P, Golinelli-Pimpaneau B, Gilles A-M, Gaucher
J-F, Burlacu-Miron S, Sakamoto H, Janin J & Baˆ rzu O

(1998) Structures of Escherichia coli CMP kinase alone
and in complex with CDP: a new fold of the nucleoside
monophosphate binding domain and insights into cyto-
sine nucleotide specificity. Structure 6, 1517–1527.
16 Bertrand T, Briozzo P, Assairi L, Ofiteru A, Bucurenci
N, Munier-Lehmann H, Golinelli-Pimpaneau B, Baˆ rzu
O & Gilles A-M (2002) Sugar specificity of bacterial
CMP kinases revealed by crystal structures and muta-
genesis of Escherichia coli enzyme. J Mol Biol 315 ,
1099–1110.
17 Yu L, Mack J, Hajduk PJ, Kakavas SJ, Saiki AY,
Lerner CG & Olejniczak ET (2003) Solution structure
and function of an essential CMP kinase of Streptococ-
cus pneumoniae. Protein Sci 12, 2613–2621.
18 Yan H, Dahnke T, Zhou B, Nakazawa A & Tsai M-D
(1990a) Mechanism of adenylate kinase. Critical evalua-
tion of the X-ray model and assignment of the AMP
site. Biochemistry 29, 10956–10964.
19 Tickle IJ, Laskowski RA & Moss DS (1998)
R
free
and
the r
free
ratio. I. Derivation of expected values of cross-
validation residuals used in macromolecular least-
squares refinement. Acta Crystallogr D Biol Crystallogr
54, 547–557.
20 Bucurenci N, Sakamoto H, Briozzo P, Palibroda N,
Serina L, Sarfati RS, Labesse G, Briand G, Danchin A,

Baˆ rzu O et al. (1996) CMP kinase from Escherichia coli
is structurally related to other nucleoside monophos-
phate kinases. J Biol Chem 271, 2856–2862.
21 Tsai M-D & Yan H (1991) Mechanism of adenylate
kinase: site-directed mutagenesis versus X-ray and
NMR. Biochemistry 30, 6806–6818.
22 Scheffzek K, Kliche W, Wiesmu
¨
ller L & Reinstein J
(1996) Crystal structure of the complex of UMP ⁄ CMP
kinase from Dictyostelium discoideum and the bisub-
strate inhibitor P
1
-(5¢-adenosyl) P
5
-(5¢-uridyl) penta-
phosphate (UP
5
A) and Mg
2+
at 2.2 A
˚
: implications for
water-mediated specificity. Biochemistry 35, 9716–9727.
23 Alexov E (2004) Calculating proton uptake ⁄ release and
binding free energy taking into account ionization and
conformation changes induced by protein–inhibitor
association: application to plasmepsin, cathepsin D and
endothiapepsin–pepstatin complexes. Proteins 56, 572–
584.

24 Alexov E, Miksovska J, Baciou L, Schiffer M, Hanson
DK, Sebban P & Gunner MR (2000) Modeling the
effects of mutations on the free energy of the first elec-
tron transfer from QA to QB in photosynthetic reaction
centers. Biochemistry 39, 5940–5952.
25 Kunkel TA, Roberts JD & Zakour RA (1985) Rapid
and efficient site-specific mutagenesis without pheno-
typic selection. Methods Enzymol 154, 367–382.
26 Sanger F, Nicklen S & Coulson AR (1977) DNA
sequencing with chain terminating inhibitors. Proc Natl
Acad Sci USA 74, 5463–5467.
27 Munier H, Gilles A-M, Glaser P, Krin E, Danchin A,
Sarfati RS & Baˆ rzu O (1991) Isolation and characteriza-
tion of catalytic and calmodulin-binding domains of
Bordetella pertussis adenylate cyclase. Eur J Biochem
196, 469–474.
28 Sambrook J, Fritsch EF & Maniatis T (1989) Molecular
Cloning: A Laboratory Manual, 2nd edn. Cold Spring
Harbor Laboratory Press, Cold Spring Harbor, NY.
29 Bradford MM (1976) A rapid and sensitive method for
the quantitation of microgram quantities of protein util-
izing the principle of protein–dye binding. Anal Biochem
72, 248–254.
30 Laemmli UK (1970) Cleavage of structural proteins
during the assembly of the head of bacteriophage T4.
Nature (Lond) 227, 680–685.
31 Blondin C, Serina L, Wiesmu
¨
ller L, Gilles A-M &
Baˆ rzu O (1994) Improved spectrophotometric assay of

nucleoside monophosphate kinase activity using the
pyruvate kinase ⁄ lactate dehydrogenase coupling system.
Anal Biochem 220, 219–221.
32 Sarfati RS, Kansal VK, Munier H, Glaser P, Gilles
A-M, Labruye
`
re E, Mock M, Danchin A & Baˆ rzu O
(1990) Binding of 3¢-anthraniloyl-2¢deoxy-ATP to
calmodulin-activated adenylate cyclase from Bordetella
Substrate specificity of E. coli CMP kinase A. Ofiteru et al.
3372 FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS
pertussis and Bacillus anthracis. J Biol Chem 265,
18902–18906.
33 Alexov E (2003) Role of the protein side-chain fluctua-
tions on the strength of pair-wise electrostatic interac-
tions: comparing experimental with computed pK
a
s.
Proteins 50, 94–103.
34 Alexov EG & Gunner MR (1997) Incorporating protein
conformational flexibility into the calculation of pH-
dependent protein properties. Biophys J 72, 2075–2093.
35 Georgescu RE, Alexov EG & Gunner MR (2002) Com-
bining conformational flexibility and continuum electro-
statics for calculating pK
a
s in proteins. Biophys J 83,
1731–1748.
36 Xiang Z & Honig B (2001) Extending the accuracy lim-
its of prediction for side-chain conformations. J Mol

Biol 311, 421–430.
37 Otwinowski Z & Minor W (1993) DENZO, A Film
Processing Program for Macromolecular Crystallogra-
phy. Yale University Press, New Haven, CT.
38 Navaza J (1994) AMoRe: an automated package for
molecular replacement. Acta Crystallogr A50, 157–163.
39 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 A47, 110–119.
40 Brunger AT, Adams PD, Clore GM, DeLano WL, Gros
P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges
M, Pannu NS et al. (1998) Crystallography and NMR
system: a new software suite for macromolecular struc-
ture determination. Acta Crystallogr D Biol Crystallogr
54, 905–921.
41 DeLano WL (2002) The Pymol Molecular Graphics
System, Version 0.97. DeLano Scientific, San Carlos,
CA.
A. Ofiteru et al. Substrate specificity of E. coli CMP kinase
FEBS Journal 274 (2007) 3363–3373 ª 2007 The Authors Journal compilation ª 2007 FEBS 3373