Escherichia coli
cyclopropane fatty acid synthase
Mechanistic and site-directed mutagenetic studies
Fabienne Courtois, Christine Gue
´
rard, Xavier Thomas and Olivier Ploux
Laboratoire de Chimie Organique Biologique, UMR7613 CNRS, Universite
´
Pierre et Marie Curie, Paris, France
Escherichia coli fatty acid cyclopropane synthase (CFAS)
was overproduced and purified as a His
6
-tagged protein.
This recombinant enzyme is as active as the native enzyme
with a K
m
of 90 l
M
for S-AdoMet and a specific activity of
5 · 10
)2
lmolÆmin
)1
Æmg
)1
. T he enzyme is devoid of organic
or metal cofactors and is unable to catalyze the wash-out of
the m ethyl protons of S-AdoMet to the solvent, data that do
not support the ylide mechanism. Inactivation of the enzyme
by 5,5¢-dithiobis-(2-nitrobenzoic acid) (DTNB), a pseudo
first-order process with a rate constant of 1.2

lipids in bacteria [1], plants [2,3] and parasites [4]. Escheris-
hia coli cyclopropane fatty acid synthase (CFAS) [5–9] and
its closely related homologs from Mycobacterium tuber-
culosis [10] are the best known representatives of this class
of enzymes. In E. coli, cyclopropanation is thought to be
involved in long-term survival of nongrowing cells and is
often associated with enviromental stresses [1]. In M. tuber-
culosis, cyclopropanation has recently been associated
with virulence and persistance of the pathogen [11]. Hence,
cyclopropane s ynthases m ight be good targets for new
antituberculous d rugs. Indeed, t uberculosis remains a major
cause of death in the world and there is a real need for new
drugs to combat strains of M. tuberculosis that are resistant
to existing drugs [12]. We have been interested in studying
CFAS from E. coli as a model for M. tuberculosis cyclo-
propane synthases, for which an in vitro assay is still lac king.
Our goal is to co ntribute to the elucidation of this intrigu-
ing enzymatic reaction, but also to discover inhibitors of
cyclopropane synthases that might be good leads to
antituberculous drugs [13].
This enzymatic cyclopropanation reaction proceeds by
transfer of a methylene group from the activated methyl
group of S-adenosyl-
L
-methionine (S-AdoMet) to the
(Z)-double bond of an unsaturated fatty acid chain,
resulting in the formation of a cyclopropane ring on the
alkyl chain (Scheme 1). Early in vivo studies [14–16] showed
that two of the three methyl protons of S-AdoMet are
retained in the product, although s ome exchange w as

methionine:unsaturated-ph ospholipid methyltransferase (cyclizing),
cyclopropane-fatty-acyl-phospholipid synt hase, cycloprop ane
synthase (EC 2.1.1.79).
Note: A website is available at http://www.ccr.jussieu.fr/umr7613/
(Received 28 July 2004, revised 16 September 2004,
accepted 18 October 2004)
Eur. J. Biochem. 271, 4769–4778 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04441.x
Even though the mechanism involving a carbocation
intermediate is often cited in the literature [1,10,20], the
other reasonable alternatives deserve c onsideration and in
particular the metal-assisted ylide mechanism [21]. H owever,
recent crystallographic data [22], inhibition and mechanistic
studies [13,18,23,24], and data reported in this study argue
in favor of the carbocation mechanism. On the basis of
chemical modification experiments [9], the involvement of a
cysteine residue in the catalysis has been invoked. Indeed, the
thiolate side chain could be either the base that is required for
abstraction of t he methyl proton, or could stabilize the
carbocation, if that intermediate were formed, or even
participate in a covalent catalysis (Scheme 2). Interestingly,
the three dimensional structure of three cyclopropane
synthases from M. tuberculosis [22] showed the presence of
two cyseines at, or near, the active site: C139 and C354
(E. coli CFAS numbering). Futhermore, these residues, as
well as C176 (E. coli numbering), are strictly conserved in all
cyclopropane synthases discovered so far [1].
We report here the purification of a His
6
-tagged CFAS
and its characterization. We also report exchange experi-

)werefromNewEngland
Nuclear (Boston, MA, USA). Restriction enzymes, Taq
polymerase, T4 DNA ligase and molecular biology kits
were either from Promega or f rom Roche (Meylan, France).
Culture medium components were purchased from Difco
Laboratories (Detroit, MI, USA). Chromatographic equip-
ment (GradiFrac) and column phases were from Amersham
Biosciences (Orsay, France). UV-visible spectra were
obtained on an Uvikon-930 Kontron spectrophotometer
(Munchen, Germany) or a Lambda-40 Perkin Elmer
apparatus (Norwalk, CT, USA). Scintillation counting
was run on a 1214 Rackbeta LKB Wallac radioactivity
counter (Per kin Elmer). Sonication was performed on a
VibraCell sonicator from Bioblock (Illkirch, France). SDS/
PAGE was run on a Bio-Rad Protean II system (Hercules,
CA, USA), using the conditions described by the manufac-
turer, and DNA electrophoresis on a Mupid apparatu s
(Eurogentec, Seraing, Belgium), in 40 m
M
Tris/acetate
buffer, pH 7.5, 1 m
M
EDTA. Centrifugations were run o n
a Sorval RF5plus centrifuge (DuPont, K endro, Cortaboeuf,
France).
1
Hand
13
C-NMR spectra were obtained on an AC
400 M Hz Bruker apparatus (Rheinstetten, Germany).

S
R
2
CH
3
CH
3
CH
3
Enz-Nu
H
H
C
H
H
H
R
1
S
R
2
H
3
C
Ylide
R
1
S
R
2

S
NH
3
OOC
CH
3
O
A
OH
HO
S
NH
3
OOC
CFAS
S-Adenosyl-L-methionine
S-Adenosyl-L-homocysteine
+ H
Scheme 1. Reaction catalyzed by the cyclopropane synthases. In E. coli
CFAS the lipid substrate is an unsaturated ph ospholipid, while in
M. tuberculosis the unsat urated alkyl chain is probably bound to an
acyl carrier protein.
4770 F. Courtois et al.(Eur. J. Biochem. 271) Ó FEBS 2004
been engineered) of E. coli CFAS were constructed
using the QuikChange Site-Directed Mutagenesis Kit from
Stratagene (La Jolla, C A, USA). The following sets of
mutated primers (mutations are underlined) w ere u sed:
C139S: 5¢-CATGCAATATTCC
AGCGCTTACTGGAA
AG-3¢ and 5¢-CTTTCCAGTAAGCGC

potassium
phosphate buffer, pH 7.4 at the desired concentration
(% 20 mgÆmL
)1
). Phospholipids were assayed using the
ferric hydroxamate method, as described previously [27],
and using tripalmitin standards for calibration. Phospho-
lipid solutions were sonicated f or 30 s for dispersion prior to
use as substrates. Cyclopropanated phospholipids were
extracted, using t he same protocol, from isopropyl thio-b-
D
-galactoside (IPTG)-induced E. coli BL21(DE3)/pET-
24H6cfa cells.
CFAS purification
An overnight preculture [10 mL Luria–Bertani (LB)
medium, 50 lgÆmL
)1
kanamycin] of E. coli BL21(DE3)/
pET-24H6cfawasusedtoinoculate800mLofLBmedium
supplemented with 5 0 lgÆmL
)1
kanamycin. The culture was
shaken (180 r.p.m., 37 °C), and when the absorbance a t
600 nm reached a v alue of 0.7, IPTG was added at a final
concentration o f 100 l
M
. T he culture was then shaken
overnight at 37 °C. The cells were collecte d by centrifuga-
tion (4000 g, 15 min), washed (0.1
M

)1
and 7 mL fractions were collected. The pres-
ence of proteins was detected using the Bradford assay and
the purity of i ndividual fractions was analyzed by SDS/
PAGE. Fractions containing pure CFAS were pooled and
desalted on PD-10 columns (Amersham Bioscience) equili-
brated with buffer A. Highly concentrated enzyme solu-
tions were obtained by ammonium sulfate precipitation as
follows. Solid ammonium sulfate was added at 0 °Ctothe
enzyme solution, up to 40% saturation, and the precipitated
protein was recovered by centrifugation (10 min at
12 000 g). The pellet was then dissolved in the minimum
volume of 20 m
M
potassium pho sphate buffer , pH 7.4,
50% (v/v) glycerol, and the enzyme solution was stored at
)20 °C. The mutant proteins, which all carry an N -terminal
His
6
-tag, were purified as des cribed for the His
6
-tagged
wild-type enzyme.
Biochemical characterization
N-terminal protein sequencing of the enzyme, t ransfered
onto a polyvinyliden e fluoride membrane, was obtained at
the Plateau Technique d’An alyse et de Microsequenc¸ age
des Prote
`
ines (Institut Pasteur). For the determination of

M
potassium
phosphate buffer, pH 7.4, in a final volume of 100 lL. The
reaction was initiated by addition of the enzyme and
incubated at 37 °C for 20 min. The reaction was stopped by
adding 1 mL 10% (v/v) trichloroacetic acid, a nd the solu-
tion was filtered over glass fi ber filters (Whatman GF/c,
Middlesex, USA; 25 mm). The filters, adapted on a
filtration device (Millipore, Billerica, M A, USA; 1225
model), were washed three times with 1 mL 10% (v/v)
trichloroacetic acid, three times with 1 mL H
2
O, oven-dried
(60 °C, 20 min) and finally counted for radioactivity in
5 mL o f s cintillation cocktail (Optiphase, Wallac). The
activity measured under these conditions was linear with
Ó FEBS 2004 Cyclopropane synthase reaction mechanism (Eur. J. Biochem. 271) 4771
time over a period of 2 0 min and linear with enzyme
concentration up to 0.1 mgÆmL
)1
of protein (data not
shown). One unit of CFAS is defined as the amount of
enzyme that transforms 1 lmol of substrate per min. The
kinetic parameters of His
6
-tagged wild-type and mutant
CFAS were determined by measuring the activity
(as described above for the e nzyme a ssay) at different
concentrations of S-AdoMet. D ata were a nalyzed using
nonlinear regression analysis run on

Exchange experiments
A sample consisting of 2 lg (45 pmol) CFAS, 2 m
M
dithiotheithol, 0.5 mgÆmL
)1
BSA, 10% (v/v) glycerol,
20 m
M
potassium ph osphate buffer, pH 7.4, and 1 m
M
[methyl-
3
H]S-AdoMet (13 mCiÆmmol
)1
) was incubated at
37 °C for 3 h. A control sample that did not contain the
enzyme was run at the same time. The reaction was stopped
by dilution with 1 mL water and immediate f reezing in
liquid nitrogen. Water was then lyophilized, re covered and
counted for r adioactivity in 4 mL of scintillation liquid. For
the incorporation of deuterium from D
2
O, the experiment
was run directly in the NMR tube (500 lL, total volume).
The sample c onsised of 2 lg (45 pmol) CFAS, 2 m
M
dithiotheithol, 0.5 mgÆmL
)1
BSA, 10% (v/v) glycerol,
20 m

Thiol titration by 5,5¢-dithiobis-(2-nitrobenzoic acid)
(DTNB)
Thiol titrations were run a s described by Riddles et al.[29].
Breifly, for titration i n d enaturing c onditions, 1.9 nmol
(84 lg, 2.4 l
M
final concentration) of purified His
6
-tagged
wild-type CFAS w ere added t o a solution (800 lL final
volume) c ontaining 6.0
M
guanidine hydrochloride,
0.31 m
M
5,5¢-dithiobis-(2-nitrobenzo ic acid) (DTNB),
0.1
M
potassium phosphate, pH 7.3, 1 m
M
EDTA at
20 °C. The exposed thiols were titrated by measuring the
change in absorbance at 412 nm (e ¼ 13 700 cm
)1
Æ
M
)1
).
For titration under n ative conditions the same protocol was
applied except that the guanidine hydrochloride was not

,5m
M
reduced glutathione to
quench t he inactivation, in 20 m
M
potassium phosphate
buffer, pH 7.4. The mixture was incu bated at 37 °Cfor
15 min. The reaction was stopped by adding 1 m L 10%
(w/v) trichloroacetic acid and treated as described above
for radioactivity counting.
Protection from inactivation by DTNB
His
6
-tagged wi ld-type C FAS was incubated with 2 m
M
DTNB in presence of 1 mg ÆmL
)1
phospholipids or
380 l
M
S-AdoMet, in 0.1
M
potassium phosphate,
pH 7.3, at 20 °C. Residual activity was measured as
described for the inactivation experiments (see above).
Tsou plot
Two identical samples w ere p repared as f ollows. His
6
-
tagged wild-type CFAS (1.9 nmol; 84 lg, 2.4 l

6
-tagged
recombinant protein, in order to simplify the purification
protocol [9]. The recombinant gene, containing an engin-
eered ribosome binding site [31] and a His
6
-tag was
constructed using PCR-based recombinant technology
and cloned into a pET-24(+) vector. A C-terminal tagged
protein was also constructed but the protein was expressed
as an insoluble and inactive polypeptide. The N-terminal
His
6
-tagged construct, pET-24H6cfa, whose DNA sequence
was verified, was used th roughout this study. Overexpres-
sion in E. coli BL21 (DE3)/pET-24H6cfa was optimized by
varying the usual parameters, that is IPTG concentration
(from 40 l
M
to 1 m
M
), t emperature ( 20 °C, 30 °Cand
37 °C), and incubation time after induction (from 3 h to
15 h). Our best results were obtained using the following
conditions: 100 l
M
IPTG, 37 °C and overnight incubation.
The His
6
-taggedCFASwaspurifiedintwosteps(Fig.1).

thol and 10% (v/v) g lycerol substantially stabilized the
enzyme activity during the assay. Typically, after 60 min
incubation the activity of a sample containing the additives
was twice over that of a c ontrol sample. Addition of
S-AdoHcy nucleosidase as suggested previously [7] to
hydrolyze the product, a competitive inhibitor [6,13], was
not necessary in our assay b ecause the concentration of
S-AdoHcy reached was too low to cause inhibition. The
effect of ph ospholipid concentration was also checked and
we found a biphasic curve as already observed, with a
saturation at 1 mgÆmL
)1
phospholipid [6]. Using this assay
we measured a K
m
of 90 ± 5 l
M
for S-AdoMet and a k
cat
of 2.2 ± 0.1 min
)1
, values in close agreement t o t hose
reported for the native enzyme [7]. Therefore, the presence
of the His
6
-tag does not perturb the catalytic activity.
N-terminal sequencing showed no contaminants and was
in agreement with the predicted seq uence. The UV-visible
spectrum of the protein did not show any absorption over
300 nm and thus no organic cofactor could be detected.

Fig. 2. pH profile for CFAS activity. His
6
-tagged wild-type CFAS
activity was measured at different pH values, using a series of buffers
(see Experiental procedures). Each data point represents the average of
two independent experiments with less than 5% deviation to the mean.
The data points were fitted to Eqn (1), for estimation of the two pK
a
.
Ordinates are plotted on a log scale.
Fig. 1. SDS/PAGE analysis of the purification of His
6
-tagged wild-type
and E. coli CFAS mutants. From left to right: lane 1, C139S; lane 2 ,
C176S; lane 3, C354S; lane 4, wild-typeCFAS;lane5,molecularmass
markers (from top to bottom, 66 kDa, 45 kDa, 36 kDa, 29 kDa,
24 k Da, 20.1 kDa).
Ó FEBS 2004 Cyclopropane synthase reaction mechanism (Eur. J. Biochem. 271) 4773
the enzyme in the presence of [methyl-
3
H]S-AdoMet but
without unsaturated phospholipids, for 3 h gave no more
counts in the water fraction than a control sample
containing no enzyme. The detection limit of this experi-
ment was e stimated at 0.2% exchange (i.e. that an exchange
of 0.2% or more would have been easily detected). Addition
of cyclopropanated phospholipids in the reaction mixture
that could trigger a conformational change upon binding
did not enhance this exchange reaction. The reverse
experiment, incorporation o f solvent protons into the

reacting thiols (less accessible) and three buried cysteines
that react extremely slowly. The upward curvature of the
trace in Fig. 3, after 40 min (a reproducible phenomenon),
is probably due to a partial unfolding o f the protein,
exposing the buried cysteines, which consequently react
faster. It i s not clear i f the protein unfolds because of
multiple chemical modifications or if it is simply due to the
long incubation time.
CFAS inactivation by DTNB
As already reported by Cronan and coworkers [9], we found
that CFAS could be inactivated by thio l-directed reagents
such as DTNB and N-ethylmaleimide (NEM). Kinetic
analysis of the inactivation process by DTNB is shown i n
Fig. 4. The inactivation follows a pseudo first-order kinetics
with no saturation and w ith a second-order r ate constant of
1.2
M
)1
Æs
)1
, a low but not unprecedented value [32]. Similar
analysis using NEM showed that the inactivation occurred
similarly with a rate constant of 2.4
M
)1
Æs
)1
(not shown).
The inactivation process was not significantly slowed down
in the presence of a saturating concentration of S-AdoMet

Note that the i class belongs to the p class. The graph shown
in Fig. 5 c onfirms the presence o f three classes of free
cysteines in the enzyme. First, three cysteines react quickly,
with no loss of activiy, then two more cysteines react with
concomitant loss of e nzyme activity, and finally the buried
cysteines react. The portion of the graph where the activity
is lost perfectly fits to a straight line when i ¼ 1(the
correlation coefficient i s 0 .99). Wh en the same data are
plotted with i ¼ 2ori¼ 3 the data points clearly deviate
from linearity. Therefore the data of Fig. 5 are most
consistent with one cysteine, chemical modifi cation of which
leads to inactivation. Futhermore the plot a llows the
estimation of s and p, as the o rdinate in tercept is (p+s)/
p ¼ 2.1, the abscissa intercept is p+s ¼ 5 and the slope is
)1/p ¼ –0.44. Thus p ¼ 2.3 and s ¼ 2.7, values in close
agreement with the numbers deduced from Fig. 3 (where
p ¼ 2ands¼ 3).
0
1
2
3
4
5
6
7
0 102030405060
Number of reacting thiol
per monomer (mol/mol)
Time (min)
Fig. 3. Kinetics of DTNB titration of E. coli His

obs
(min
-1
)
[DTNB] (mM)
10
100
0 2 4 6 8 10 12 14 16
Residual activity (%)
Time (min)
Fig. 4. Inactivation of His
6
-tagged wild-type E. coli CFAS by DTNB.
Top: His
6
-tagged wild-type CFAS was incubated in the presence of
DTNB, 0 m
M
(d), 0.5 m
M
(h), 1 m
M
(r), 1.5 m
M
(s), 2 m
M
(j), at
20 °Cin0.1
M
potassium phosphate buffer, pH 7.3, containing 1 m

-tagged
wild-type CFAS
No protection 0.13
0.38 m
M
S-AdoMet 0.10
1mgÆmL
)1
Phospholipids 0.15
C139S CFAS No protection 0.14
C176S CFAS No protection 0.11
C354S CFAS No protection 0.12
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
012345678
(Fraction of resudual activity)
1/i
Number of titrated thiol per monomer
Fig. 5. Tsou plot for the inactivation of His
6
-tagged wild-type CFAS by
DTNB. CFAS (1.9 nmol, 2.4 l
M
) was incubat ed with 0.31 m

(l
M
)
k
cat
(min
)1
)
k
cat
/K
m
(min
)1
Æm
M
)1
)
Relative catalytic
efficiency (%)
Wild type 90 2.2 24.8 100
C139S 88 0.3 3.9 16
C176S 73 2.7 37.9 150
C354S 105 1.6 15.7 63
Ó FEBS 2004 Cyclopropane synthase reaction mechanism (Eur. J. Biochem. 271) 4775
than the t hiol group. All mutant genes were obtained by
PCR amplification using two sets of mutated primers, and
the Stratagene technology. The desired mutations were
verified by DNA sequencing and the mutated proteins were
expressed and purified as described for the His

tions of unsaturated lipids, such as the formation of
a-methylketo- or a-methylhydoxy- fatty acids [35,36].
However, it quickly appeared that addition of a s ulfur
ylide, derived from S-AdoMet, to the double bond of the
fatty acid could b e another plausible alternate r eaction
mechanism [14,17,18,21]. The two mechanisms differ from
one another not only in the order of making and breaking
bonds, but also in the type of intermediate formed. Progre ss
has recently been achieved with cloning and purification
of the E. coli enzyme [9], as well as solving the three
dimensional structure of M. tuberculosis enzymes [22], and
reports of some mechanistic experiments [23,24].
We report here the purification and characterization of a
recombinant CFAS bearing an N-terminal His
6
-tag. The
use of nickel affinity chromatography allowed rapid
preparation o f pure enzyme in s ubstantial amounts. A
similar successful strategy was recently followed by Liu and
coworkers [24].
As the ylide mechansim would be most likely to i nvolve
carbenoid transfer to a metal [21], we searc hed for metals in
the enzyme. No cofactors, organic or metallic, were found, in
accordance with structural data obtained for the M. tuber-
culosis enzymes(forwhichnoin v itro catalytic a ctivity has
ever been reported). The reaction mechanism mu st therefore
rely solely on side chain functional groups, and thus only
acid-base or nucleophilic catalysis must operate.
The pH profile of the activity, in saturating conditions,
revealed two ionisable g roups important for catalysis: a first

hence trigger a conformational change, a strategy that was
successful in the citrate synthase case [37]. Therefore, our
data do not support the ylide mechanism. Of course, one
cannot exclude the possibility that the abstraction is
promoted by a monoprotic base that exchanges its proton
with the s olvent very slowly. However, Buist and coworkers
reported feeding experiments, using deuteriated methionine
and L. plantarum cells, which were then interpreted by
invoking an exchange of the methyl protons on the
carbocation intermediate ( Scheme 2A) but not on an ylide
species [17,18]. Such a fast exchange (33% exchange) would
probably require a polyprotic base and a reversible forma-
tion of th e cyclopropane ring. However, these experiments
were conducted using whole cells and were dependent on
growth conditions, and thus need to be confirmed on the
isolated enzyme.
We have also addressed the role, in c atalysis, of the
cysteines of the E. coli enzyme. It has been suggested in the
literature that a cysteine could b e important for catalysis [9].
Indeed, a thiolate could e ither abstract a proton on the
methyl group or stabilize the carbocation, or even form a
covalent adduct (the base or the nucleophile in Scheme 2).
The alignment of the seq uence of all cyclopropane synthases
known so far [1], shows that only three cysteines among the
eight cysteine residues of the E. coli enzyme are conserved:
C139, C176 and C354. In the three dimensional structure of
the homologous M. tuberculosis enzymes [22], C35 which
corresponds to C139 of E. col i CFAS shares a hydrogen
bond, by its N-H, to a carbonate in the active site. C269 of
the M. tuberculosis enzyme, which corresponds to C354 in

mutants of enzymes known to use the thiolate group as a
base (racemases) or as a nucleophile (methyl transferases)
[38–40]. The fact that inactivation by DTNB of the His
6
-
tagged wild-type enzyme was not protected by substrates,
and that the three C ysfiSer mutants prepared in this report
are inactivated by DTNB at the same rate as the His
6
-tagged
wild-type enzyme, shows that the cysteine responsible for the
inactivation cannot be C139, C176 or C354. There are five
other cysteine residues in the E. coli enzyme, and it is not
possible at t he moment to attribute the residue t hat is
responsible for the inactivation. Furthermore, this chemical
inactivation does not seem to be related to catalysis.
In conclusion, the findings reported here do not support
the ylide m echanistic proposal but further s upport the
carbocation mechanism. Furthermore, it is shown here that
the conserved cysteines of E. coli CFAS are not directly
involved in catalysis and that the inactivation observed after
chemical modification o f another cysteine probably comes
from steric hindrance, that is no t relevent to catalysis. The
base and the nu cleophile supposedly involved in the r eaction
mechanism are very likely other residues o r f unctional
groups, e.g. E239 or the active site carbonate. F uther
mutagenesis experiments are underway to explore these
hypotheses.
Acknowledgements
We wish to thank Thierry Drujon and Diane Delaroche for technical

interaction with phospholipid vesicles. Biochemistry 18, 3292 –
3300.
7. Taylor, F.R., Grogan, D.W. & Cronan. J.E. (1981) Cyclopropane
fatty acid synthase from Escherichia coli. Methods Enzymol. 71,
133–139.
8. Gr ogan, D.W. & Cronan. J.E. (1984) Cloning and manipulation
of the Escherichia coli cyclopropane fatty acid synthase gene:
physiological aspects of enzyme overproductio n. J. Bacteriol. 158,
286–295.
9. Wang, A., Grogan, D.W. & Cronan, J.E. (1992) Cyclopropane
fatty acid synthase of Escherichia coli: deduced a mino acid
sequence, purification, and studie s of the enzyme ac tive site.
Biochemistry 31, 11020–11028.
10. Barry, C.E., Lee, R .E., Mdluli, K., Sam pson, A .E., S chroeder,
B.G.,Slayden,R.A.&Yuan,Y.(1998)Mycolicacids:structure,
biosynthesis and physiological functions. Prog. Lipid Res. 37,143–
179.
11. Glickman, M.S., Cox, J.S. & Jacobs, W.R. (2000) A novel mycolic
acid cyclopropane synthetase is required for cording, persistance,
and virulence of Mycobacterium tuberculosis. Mol. Cell. 5, 717–
727.
12. Raviglione, M.C. (2003) The TB epidemic from 1992 t o 2002.
Tuberculosis 83, 4–14.
13. Gue
´
rard, C., Bre
´
ard, M., Courtois, F., Drujon, T. & Ploux, O .
(2004) Synthesis and evaluation of an alogues of S-adenosyl-1-
methionine, as inhibitors of the E. coli cyclopropane fatty acid

J. Am. Chem. Soc. 96, 5627–5628.
22. Huang,C.,Smith,V.,Glickman,M.S.,Jacobs,W.R.&Sacchet-
tini, J.C. (2002) Crystal structures of mycolic acid cyclopropane
synthases from Mycobacterium tuberculosis. J. Biol. Chem. 277,
11559–11569.
23. Buist, P.H. & Pon, R.A. (1990) An unexpected reversal of
fluorine substituent effects in the biomethylenation of two
Ó FEBS 2004 Cyclopropane synthase reaction mechanism (Eur. J. Biochem. 271) 4777
positional isomers: a serendipitous discovery. J. Org. Chem. 55 ,
6240–6241.
24. Molitor, E.J., Paschal, B.M. & Liu, H W. (2003) Cyclopropane
fatty acid synth ase from Es cherichia coli. Enzyme p urification and
inhibition by vinylfluorine and epoxide-containing substrate ana-
logues. Chembiochem. 4, 1352–1356.
25. Bradford, M .M. (1976) A rapid and sensitive method for the
quantitation of microgram q uantities o f protein utilizing the
principle of protein-dye binding. Anal Biochem. 72, 248–254.
26. Ames, G.F. (1968) Lipids of Salmonella typhimurium and
Escherichia coli: structure and metabolism. J. Bacteriol. 95, 833–
843.
27. Dittmer,J.C.&Wells,M.A.(1969) Quantitative and qualiative
analysis of lipids and lipid components. Methods Enzymol. 14,
482–530.
28. Lee, J.E., Cornell, K.A., Riscoe, M.K. & Howell, P.L. (2001)
Expression, purification, c rystallization and preliminary X -ray
analysis of Escherichia coli 5¢-methylthioaden osine/S-adenosyl-
homocysteine nucleosidase. Acta Crystallogr. D Biol. Crystallogr.
57, 150–152.
29. Riddles, P.W., Blakeley, R.L. & Zerner, B. (1983) Reassessment of
Ellman’s Reagent. Methods Enzymol. 91, 49–60.

38. Gabbara, S., Sheluho, D. & Bhagwat, A.S. (1995) Cytosine
methyltransferase f rom Escherichia coli in which active site
cysteine is replaced with serine is partially active. Biochemistry 34,
8914–8923.
39. Glavas, S. & Tanner, M.E. (1999) Catalytic acid/base residues of
glutamate racemase. Biochemistry 38, 4106–4113.
40. Ko o, C.W., Sutherland, A., Vederas, J.C. & Blanchard, J.S. (2000)
Identification of active site cysteine residues that function as gen-
eral bases: diaminopimelate epimerase. J. Am. Chem. Soc. 122,
6122–6123.
4778 F. Courtois et al.(Eur. J. Biochem. 271) Ó FEBS 2004