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 M )1 Æs )1 , is not protected by substrates. Graphical analysis of t he inactiva- tion by DTNB revealed that only one cyst eine is responsible for the inactivation of the enzyme. The three strictly con- served Cys residues among cyclopropane synthases, C139, C176 and C354 of the E. coli enzyme, w ere mutated to serine. The relative catalytic efficiency of t he mutants were 16% for C139S, 150% for C176S and 63% for C354S. The three mutants were inactivated by DTNB at a rate com- parable to the rate of inactivation of the His 6 -tagged wild- type enzyme, indicating that the Cys responsible for the loss of activity is not one of the conserved residues. Therefore, none of the conserved Cys residues is essential for catalysis and cannot be in volved in covalen t catalysis o r general base catalysis. T he inactivation is p robably the result of steric hindrance, a phenomenon irrelevant to catalysis. It is very likely that E. coli CFAS operates via a carbocation mechanism, but the base and nucleophile remain to be identified. Keywords: cyclopropane fatty acid s ynthase; hydrogen isotope exchange; enzymatic reaction mechanism; site- directed mutagenesis; chemical modification. Cyclopropane synthases constitute an interesting class of enzymes that catalyze the cyclopropanation of unsaturated 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 observed under certain conditions [ 17,18], a nd that the vinylic and allylic protons of the substrate are also retained in the product. The stereochemistry is also retained; that is, the (Z)-double bond gives a cis-cyclopropane [5], although trans-cyclopropanes are also found in M. tuberculosis mycolic acids [10]. Chiral methyl analysis was also conduc- ted in v ivo using Lactobacillus plantarum cells, and showed retention of t he stereochemistry of the reaction [19]. This experimental observation is n ot in favor o f a carbenoid species (see below), which would probably racemize. Two types of reaction mechanism have b een proposed for this fascinating reaction: a carbocation type and an ylide type mechanism, schematically represented in Scheme 2. Correspondence to O. Ploux, Laboratoire de Chimie Organique Biologique – UMR CNRS 7613, Boıˆte 182, Tour 44–45, 4 P lace Jussieu, F-75252 Paris cedex 05, France. Fax: +33 1 44 27 71 50, Tel.: +33 1 44 27 55 11, E-m ail: ploux@ ccr.jussieu.fr Abbreviations: BSA, bovine serum albumin; CFAS, cyclopropane fatty acid synthase; DTNB, 5,5¢-dithiobis(2-nitrobenzoic acid); EDTA, ethylenediaminetetraacetic acid; NEM, N-ethylmaleimide; S-AdoMet, S-adenosyl- L -methionine; S-AdoHcy, S-adenosylhomo- cysteine. Enzymes: S-Adenosyl- L -homocysteine homocysteinylribohydrolase, adenosylhomocysteine nuc leosidase (EC 3.2.2.9); S-adenosyl- L - 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- ments that are not in favor of the ylide mechanism. The role of the conserved cysteines was studied using chemical modification and site-directed m utagenesis. It was found that the cysteines are not essential for catalysis. Experimental procedures General E. coli strains JM109 and BL21(DE3) were from Promega (Madison, WI, USA), and E. coli K12 was obtained from the Institut Pasteur Collection (CIP; Paris, France). Plasmid pET-24(+) was obtained from Novagen (Darmstadt, Germany). Synthetic oligonucleotides were products of Proligo (Paris, France) and were used without any fur- ther purification. Chemicals were purchased from Sigma- Aldrich (Saint Quentin, France) and were of the highest purity available. S-[Methyl- 14 C]adenosyl- L -methionine (60 m CiÆmmol )1 )andS-[methyl- 3 H]adenosyl- L -methionine (85 CiÆmmol )1 or 15 CiÆmmol )1 )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). Plasmid construction and site-directed mutagenesis The w ild-type histidine-tagged CFAS recombinant g ene was obtained using PCR amplification of the cfa gene from E. coli K12 genomic DNA. Briefly, E. coli K12 genomic DNA was purified using the Wizard Genomic kit from Promega, and the cfa gene was amplified using Taq DNA polymerase (Promega) and the following two primers: 5¢-CGCGAATTCAGGAGGATTTTATGCACCACCA CCACCACCACAGTTCATCGTGTATAGAAGAA-3¢ containing an EcoRI site, a ribosome binding site and a His 6 -tag sequence, and 5¢-CGCAAGCTTTTAGCGAGC CACTCGAAG-3¢ containing a Hin dIII site. The DNA fragments was purified (PCR Preps, Promega), digested by EcoRI and Hi ndIII, purified on agarose gel and ligated into pET-24(+) previously cut by the same restriction enzymes. After transformation in E. coli JM109, positive clones were selected and the plasmid extracted and purified (Wizard Plus Minipreps, Promega) for DNA sequencing ( ECSG, Evry, France). Plasmid pET-24H6cfa, thus obtained, was used for t ransformation in E. coli BL2 1(DE3) and t his construction afforded efficient expression of the enzyme. The mutated cfa genes, cfaC139S, cfaC176S and cfaC354S (numbering corresponds to the wild-type s equence, that is without counting the N-terminal His 6 -tag that has R 1 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 CH 2 OR Enz–Base Enz–BaseH Enz–Nu Enz–Nu A B Enz Metal Metal CH 2 Enz–Nu Enz–Base Enz–Base Carbenoid H H Scheme 2. Plausible reaction mechanisms for the catalyzed cyclopro- panation. (A) The carbocation mechanism; (B) the ylide mechanism. This mech anism would most probably require a metalloenzyme and transfer of a carbenoid to the metal (details are not shown for clarity). H H HH O A OH HO 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 TGGAATATTG CATG-3¢; C176S: 5¢-GGATATTGGC AGCGGCTGGG GCGGACTGGC-3¢ and 5¢-GCCAGTCCGCCCCAGCC GC TGCCAATATCC-3¢; C354S: 5¢-CTGAATGCCTCT GCAGGTGCTTTCCGCGCC-3¢ and 5¢-GGCGCGCGG AAAGCACCTGCA GAGGCATTCAG-3¢. Plasmid pET- 24H6cfa was used as the template. Transformants were selected and the plas mids were extracted, purified (Wizard P lus miniprep kit from Promega) and se quenced (Eurogentec) to ens ure integrity and the presence of the desired mutation. In the case of the C354S mutant, the mutation was confirmed by digestion with PstIasthe mutation introduces a new restriction site. Each mutated plasmid was then transformed into competent E. coli BL21(DE3) for protein expression. Protein assay Protein concentrations were determined using the colori- metric assay described by Bradford [25] and supplied by Bio-Rad. Phospholipids preparation Unsaturated E. coli K12 phospholipids were prepared according to Cronan [ 6] and as originally described by Ames [26]. Once purified the phospholipids were stored as a chloroform solution at )20 °C. Aqueous solution s of phospholipids were prepared by evaporating the chloroform and resuspending the phospholipids in 20 m M 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 potassium phosphate buffer, pH 7.4), centrifuged (4000 g, 15 min) and kept at )20 °C until use. The cell paste was resuspended in 40 m L of 20 m M potassium phosphate buffer, p H 7.4, and the suspension was sonicated on ice (5 min, with 1 min cooling period every minute). After centrifugation (15 000 g, 20 min), the supernatent was loaded directly on a nickel affinity column (Chelating Sepharose, Amersham Bio- science; 1.6 cm i.d., 5 cm long, 10 mL) prepared as recom- mended by the manufacturer and equilibrated with buffer A (20 m M potassium phosphate buffer, pH 7.4, 0.5 M NaCl). The column was successively washed with 30 mL of buffer A and 30 mL of buffer A containing 5% (v/v) of buffer B (20 m M potassium phosphate buffer, pH 7.4, 0.5 M NaCl, 1.0 M imidazole). T he proteins were eluted by a linear gradient starting from 5% (v/v) of buffer B to 40% (v/v) of buffer B, in buffer A . The column was run at a flow rate of 1mLÆmin )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 the metal content, several samples (18 nmol each) of purified CFAS were lyophilized and s ubjected to metal analysis (Zn, Ni, Co, Fe, Cu) using ICP-AES methodology (Service Central d’Analyse; CNRS, Vernaison, France). The metal content, in each case, represented less than 1% of what was expected for 1 mol of metal per mol of enzyme. CFAS assay CFAS activity was assayed as described previously [7] with slight modifications. The assay consisted in 1.0 mgÆmL )1 phopholipids, 0.5 mgÆmL )1 bovine serum albumin (BSA), 10% (v/v) glycerol, 2 m M dithiothreitol (dithiotheithol), 0.75 m M S-AdoMet, either 14 C-labelled (specific radioac- tivity of 5.0 mC iÆmmol )1 )or 3 H-labelled (specific radioac- tivity of 1.5 mCiÆmmol )1 ), 2 lgCFAS,in20m 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 KALEIDAGRAPH soft- ware, to fit to M ichaelis–Menten kinetics. S-Adenosyl- homocysteine nucleosidase was purified from an overproducing strain (E. coli BL21(DE3)/pEXH6MTAN), generously given by K. Cornell and M. Riscoe (VAMC, Portland, OR, USA), and assayed as described previously [28]. pH profile CFAS activity was determined as described above in the following buffers: 4-morpolinoethanesulfonic acid ( pH 5.5– 7.0), 2 -(4-(2-hydoxyethyl)-1-piperazine) ethanesulfonic a cid (pH 7.0–8.5) and 3-(tris(hydroxymethyl)methylmino)-1- proanesulfonicacid (pH 7.7–9.5) all at a concentration of 150 m M . The pH was a djusted by a dding aqueous HCl or aqueous NaOH. The activity vs. pH p rofile was bell-shaped and the data points were fitted to Eqn (1): V ¼ V max =ð1 þ 10 pHÀpKa1 þ 10 pKa2ÀpH Þ Eqn ð1Þ using a nonlinear regression analysis supported by KALEIDAGRAPH software (Synergy Software, Reading, PA, USA). 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 M potassium phosphate buffer, pD 7.4 (corrected), and 1 m M S-AdoMet. In order to minimize the H 2 O concentration the buffer was exchanged in D 2 Oand lyophilized prior t o u se. A control sample that did not contain the enzyme was run at the same time. The samples were incubated f or 4 h and were a nalyzed by 1 H-NMR. The signal at 3.11 p.p.m., w hich corresponds to the methyl group of the natural diastereoisomer of S-AdoMet [(S,S) configuration], was quantified and compared to an authen- ticsampleofcommercialS-AdoMet. The methyl group of para-toluenesulfonate, present in the commercial sample of S-AdoMet, was used as an internal standard. No modifi- cation of the signal was ob served. A 13 C-NMR ( 1 H decoupled) s pectrum was also recorded to see if a ny exchange on the methyl g roup had occured, becau se deuterium incorporation would shift the signal and would give a scalar coupling. 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 added. Thiols were titrated by measuring the change in absorbance at 412 nm (e ¼ 14 150 cm )1 Æ M )1 ). Inactivation by DTNB His 6 -tagged wild-type and mutant CFAS pr oteins were treated with various concentration of DTNB in 0.1 M potassium phosphate, pH 7.3, 1 m M EDTA, at 20 °C. Aliquots of 30 lL were transferred, at different time points, to a CFAS assay mixture (final volume of 100 lL) containing, 1.0 mgÆmL )1 phopholipids, 0.5 mgÆmL )1 BSA, 0.76 m M [methyl- 3 H]S-AdoMet at a final specific radio- activity of 25 mCiÆmmol )1 ,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 M final concentration) was added to a solution (800 lL final volume) containing 0.1 M potassium phosphate, pH 7.3, 1m M EDTA, 0.31 m M DTNB. Addition o f the enzyme was performed at the same time in both samples and the absorbance at 412 nm (e ¼ 14 150 cm )1 Æ M )1 ) was followed against time, using one sample. Th e residual activity was followed against time using the second samp le. Both incubations were run at 20 °C. Determination of residual activity was carried out as described a bove for the DTNB inactivation experiments. Data obtained at the same time points, that is the number of titrated thiols and the residual activity, were used in the graphical representation described by Tsou [30]. 4772 F. Courtois et al.(Eur. J. Biochem. 271) Ó FEBS 2004 Results Cloning, expression and purification of CFAS E. coli CFAS was expre ssed as an N-terminal His 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). An affinity nickel chromatography was followed by a necessary desalting s tep by g el filtration because high imidazole concentration inhibits the enzyme activity. Start- ing from 0.8 L of culture (100 mg of protein in the crude extract with a specific activity of 0.9 · 10 )2 UÆmg )1 ), 5 m g of pure protein (Fig. 1) was obtained with a specific activity of 5.0 · 10 )2 UÆmg )1 , a value comparable to previously reported data [9,24]. The yield of this purification is 28%, and the purification fold is 5.5. This simple protocol is fast enough (a few hours in total) to keep this labile enzyme active. The enzyme was stored best in 20 m M phosphate buffer, pH 7.4 containing 50% (v/v) glycerol, at )20 °C. Assay and characterization The recombinant CFAS was assayed as described by Cronan and coworkers w ith slight modifications [7]. We found that addition of 0.5 mgÆmL )1 BSA, 2 m M dithiothei- 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. Search for usual metals found in proteins (Zn, Ni, Cu, Co, Fe) was unsuccessful. We therefore concluded that CFAS has no cofactor, a result in agreement with the three dimensional structure obtained for the M. tuberculosis cyclopropane synthases [22]. The effect of pH on the activity of CFAS, under saturating conditions, is shown in Fig. 2. The profile is b ell-shaped with a maximum around pH 7.5. Fitting the data to a simple model using Eqn (1) (with two ionisable groups involved in catalysis) gave a pK a1 of 6.8 and a pK a2 of 8.7. Exchange experiments Exchange of the methyl p roton of S-AdoMet catalyzed by CFAS was tested by measuring the wash-out of tritium from the methyl group to the solvent water. Incubation of 1 10 100 5678910 V (mU/mg) pH pK a1 =6.8 pK a2 =8.7 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 substrate, which should be faster than the wash-out as no intramolecular kinetic isotope effect on th e abstraction should occur, was tested using the same conditions but in the presence of unlabeled S-AdoMet and deuteriated buffer. The reaction was followed by 1 Hand 13 C-NMR, and again no incorporation of deuterium could be detected. Therefore, under our conditions, CFAS is unable t o c atalyze the exchange of the methyl protons of S-AdoMet, a result that does not support the ylide mechanism. Thiol titration by DTNB The amino acid sequence of wild-type E. coli CFAS predicts eight cysteines [9]. The total thiol content of the purified enzyme was spectrophotometrically titrated using DTNB, in denaturing conditions using standard protocols [29]. A ratio of 7 .5 ± 0.3 mol of free thiols p er mol of CFAS monomer was f ound, consistent with eight free cysteines in the E. coli wild-type CFAS. In native condi- tions (Fig. 3), six thiols per monomer were titrated in one hour with triphasic kinetics. Three cysteines reacted within 4 min, two more cysteines reacted more slowly within 40min,andonecysteinereactedinathirdveryslowphase. If the enzyme was left longer under these conditions (for two further hours), the absorbance at 412 nm finally reached a value compatible with eight free cysteines. Therefore E. coli CFAS contains three classes of free cysteines, three fast reacting thiols (exposed), two slowly 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 (0.38 m M ) o r i n the presence of 1 mgÆmL )1 unsaturated phospholipids ( Table 1). This suggests t hat the cys teine residue responsible for the inactivation is not located in the active site. For the graphical analysis of t he inactivation, His 6 -tagged wild-type CFAS (2.4 l M ) was treated with excess DTNB (0.31 m M ) at pH 7.3 and the residual activity together with the number of modified sulfhydryls per CFAS were determined at the same time points. The data were analyzed graphically as described by Tsou (Fig. 5) using t he following Eqn (2) [30]: (a) 1=i ¼ðp þ s À mÞ=p ð2Þ where m is the number of modified cysteines per monomeric CFAS, s t he number o f fast r eacting cysteines that are nonessential, a the fraction of remaining activity when m residues have been modified, p is the number of cysteines reacting slower than the s group, and i is the number of essential r esidues for catalytic a ctivity, as defined by Tsou. 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 6 -tagged wild-type CFAS in nondenaturing conditions. CFAS (2.4 l M ) was titrated with 0.31 m M DTNB at 2 0 °Cin0.1 M potassium phosphate bu ffer, pH 7.3, containing 1 m M EDTA. The absorbance at 412 nm was recorded against time. 4774 F. Courtois et al.(Eur. J. Biochem. 271) Ó FEBS 2004 Characterization of C139S, C176S and C354S mutant proteins Alignment o f the sequence of a ll cyclopropane syntha- ses known s o far shows t hat among the e ight cysteine residues of the E. coli CFAS only three are strictly conserved: C139, C176 and C 354 [1]. W e thus prepared three corresponding Cys fi Ser mutants for analysis. This particular re placement was chosen because it is isosteric, but the O H group is much less acidic and much less nucleophilic 0.00 0.05 0.10 0.15 0.0 0.5 1.0 1.5 2.0 2.5 k 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 M EDTA. Aliquots were withdrawn at different time points and the residual activity was m easured (se e Experim antal proced ures for details). Each data point represents the average of two independant experiments, with less than 5% deviation from the mean. Error bars are omitted for clarity. The observed first-order rate constants, k obs , were calculated by fitting the data points to simple exponential decays. Bottom: observed rate con stants, k obs , corrected for the slow inacti- vation in the absence of DTNB, were plotted aga inst DTNB concen - tration. Data were fitted to a straight line. Table 1. Summary of the inactivation r ate constant in the presence of DTNB. All experiments were run under the following conditions: 2m M DTNB in 0.1 M potassium p hosphate buffer, pH 7.3, 1 m M EDTA at 20 °C. A protectant was sometimes added as indicated. Enzyme Protectant k obs (min )1 ) His 6 -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 M DTNB in 0.1 M potassium buffer, pH 7.3, 1 m M EDTA, a t 20 °C f or 60 min. The absorbance at 412 nm (e ¼ 14 150 cm )1 Æ M )1 ) and the activity were monitored at the same time (see Experimental procedures). The fraction of residual activity, a 1/i ,i¼ 1(d), i ¼ 2(s), i ¼ 3(n)was plotted against the n umber of modified thiols. The data for i ¼ 1were fitted to a straight line. Each data point represents the average of tw o independent experiments. Error bars are not shown for clarity. See text for details of the analysis. Table 2. Kinetic parameters for His 6 -tagged wild-type and muta nt CFAS. Enzyme K m for S-AdoMet (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 6 -tagged wild- type enzyme (Fig. 1). Table 2 summarizes the kinetic parameters for individual mutants and His 6 -tagged wild- type enzyme. It is clear that all m utant are active, the slowest being C139S (16% relative catalytic efficiency). Therefore none of the conserved cysteines is essential for the activity. The three mutants w ere inactivated by DTNB at a rate similar to that m easured for the H is 6 -tagged wild-type enzyme under the same conditions (Table 1). Thus, the cysteine residue that is responsible for the inactivation is not one of the conserved cysteines, a result compatible with the fact that no protection was observed by the substrates. Discussion The cyclopropyl group is present in number of natural products and its unusual properties have always stimulated chemists and biochemists [33,34]. Biosynthesis of this structural element follows diverse schemes, but the direct methylenation of double bonds, catalyzed by cyclopropane synthases, is one of the most interesting. Most intriguing is the c hemical m echanism by which this class o f c losely related e nzymes effects the cyclopropanation. The carboca- tion mechanism, first described by Lederer [20], is chemic- ally sound and can also be applied to other methyl transferases found in mycobacteria that are homologous with cyclopropane synthases, and which catalyze modifica- 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 pK a1 of 6.8 and a second pK a2 of 8.7. Interpretation of pH effects are most difficult, but it is interesting to note that a carbonate (pK a ¼ 6.4), bound in the a ctive s ite of the M. tu berculosis enzymes, has been suggested to be the base necessary to abstract the methyl proton [22]. It is then tempting to attribute the pK a1 ¼ 6.8, detected by kinetic means to the carbonate, that was proposed to be the base on structural grounds. Alternatively, this pK a could b e attrib- uted to a His r esidue, such as His266 (His 167 in the M. tu berculosis sequence), which lies in the active site and could participate in a proton relay. Futh er mutagenetic experiments are in progress to clarify this point. One of the best ways to prove the ylide mechanism would be to show that the enzyme is able to catalyze the exchange of the methyl protons of S-AdoMet in the absence of other substrates. Numbers of enzymatic r eaction mechanisms have been supported on this type of experimental grounds [37]. However, under our conditions we did not observe such an exchange, even in the presence of the cycloprop- anated product that could mimic the second substrate and 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 the E. coli enzyme, was found to be located near the active site. The third conserved c ysteine, C72 in the M. tuberculosis enzyme, corresponding to C176 in the E. coli enzyme, is located near the S-AdoMet binding site but far from the active site. T herefore, it was reasonable to postulate an important role in catalysis for these conserved residues. Cronan and coworkers reported very briefly in a re view [1] that mutation of Cys176 and Cys354 to Ala, in the E. coli enzyme, gave active mutants. No experimental data were reported and the third conserved cysteine, Cys139, was not mutated. We thus decided to re-examine this issue in detail. Our results show that, in the E. coli enzyme, all of the eight cysteines are free and that they can be classified into three classes: three fast reacting cysteines, two cysteines re acting at a moderate rate, and three cysteines reacting very slowly. 4776 F. Courtois et al.(Eur. J. Biochem. 271) Ó FEBS 2004 Futhermore, using the graphical analysis developed by Tsou [30], we s how here that only one cysteine of the t wo moderately reacting residues, is responsible for the inactiva- tion of CFAS by DTNB. Thus, five cysteines react with thiol-directed reagent in 40 min under our conditions, but only one modification leads to inactivation. The three mutated enzymes, C139S, C176S and C354S, w ere p repared by site-directed mutagenesis, and were found to be active, the slowest (16% active) being C139S, primarily affected on its catalytic constant. If any of these cysteines were invo lved in base catalysis or nucleophilic catalysis, the corresponding serine mutant should have been at least a hundred to a thousand times less efficient than the wild-type enzyme. A dramatic drop in activity is usually observed in CysfiSer 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 assistance in preparing E. coli pho pholipids a nd in constructing mutants of E. coli cyclop ropane fatty acid s ynthase. We are grateful to Sabin e Cornet for initial work on cloning the cfa ge ne. We are indebted to Dr Kenneth A. Cornell and Dr Michael Riscoe for the generous gift of plasmid p EXH6MTAN and for advice in preparing S-AdoHcy nucleosidase. This work was supported in part by the ACI program of the ÔMiniste ` re de lÕEducation Nationale de la Recherche et de la Technologie’, grant number 0693. References 1. Grogan, D.W. & Cronan, J.E. (1997) Cyclopropane ring forma- tion in membrane lipids of bacteria. Microbiol. Mol. Biol. Rev. 61, 429–441. 2. Bao,X.,Katz,S.,Pollard,M.&Ohlrogges, J.B. (2002) Carbo- cyclic fatty-acids in plants: Biochemical and molecular genetic characterization of cyclop ropane fatty acid synthesis of Sterculia foetida. Proc. Natl Acad. Sci. USA 99, 7172–7177. 3. Bao, X., Thelen, J.J., Bonaventure, G. & Ohlrogges, J.B. (2003) Characterization of cyclopropane fatty-acid synthase from Ster- culia foetida. J. Biol. Chem. 278, 12846–12853. 4. Rahman, M.D., Ziering, D.L., Mannarelli, S.J., Swartz, K.L. & Huang, D . (1988) Effects of sulfur-containing analogues of stearic acid on growth and fatty acid biosynthesis in the protozoan Crithidia fasciculata. J. Med. Chem. 31, 1656–1659. 5. Cronan, J.E., Nunn, W.D. & Batchelor, J.G. (1974) Studies on the biosynthesis of cyclopropane fatty acids in Escherichia c oli. Biochimica Biophysica Acta 348, 63–75. 6. Taylor, F.R. & Cronan. J.E. (1979) Cyclopropane fatty acid synthase of Escherichia coli. Stabilization, purification, and 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 synthase. Bioorg Med. Chem. Lett. 14, 1661–1664. 14. Pohl, S., Law, J.H. & Ryhage, R. (1963) The path o f hydrogen in the formation of cyclopropane fatty acids. Biochem. Biophys. Acta 70, 583–585. 15. Polacheck,J.W.,Tropp,B.E.&Law,J.H.(1966)Biosynthesisof cyclopropane compounds. J. Biol. Chem. 241, 3362–3364. 16. Law, J.H. (1971) Biosynthesis of cyclopropane rings. Acc. Chem. Res. 4, 199–203. 17. Buist, P.H. & MacLean, D.B. (1980) The b iosynthesis of cyclo- propane faty acids. I. Feeding expe riments with oleic ac id-9,10-d 2 , oleic acid-8,8,11,11-d 4 , and 1-methionine-methyl-d 3 . Can. J. Chem. 59, 828–838. 18. Buist, P.H. & MacLean, D.B. (1980) The b iosynthesis of cyclo- propane f aty a cids. II. Mechanistic studies using methionine labelled with one, two, and three deuterium atoms in the methyl group. Can. J. Chem. 60, 371–378. 19. Arigoni, D. (1987) Stereochemische Untersuchungen von biolog- ischen Alkylierungsreactionen. Chimia 41, 188–189. 20. Le dere r, E. (1969) Some problems con taining biological c-alky- lation reactions and phytosterol biosynthesis. Q. Rev. Chem. Soc. 23, 453–481. 21. Cohen,T.,Herman,G.,Chapman,T.M.&Kuhn,D.(1974)A laboratory model for the biosynthesis of cyclopropane rings. Copper-catalyzed cyclopropanation of olefins by sulfur ylides. 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. 30. Tsou , C. (1962) Relation between modifica tion of func tional groups of proteins and their biological activity. Sci. Sin. 11, 1535– 1558. 31. MacFe rrin, K.D ., Terranova, M.P., S chreiber, S.L. & Verdine, G.L. (1990) Overproduction a nd dissection of proteins by the expression-cassette polyme rase c hain re action. Proc.NatlAcad. Sci. USA 87, 1937–1941. 32. Ploux, O., Lei, Y., Vatanen, K. & Liu, H W. (1995) Mechanistic studies on CDP-6-deoxy-D-3,4-glucose en re ductas e: The role of cysteine re sidues in catalysis as probed by chemical modification and site-directed mutagenesis. Biochemistry 34, 4159–4168. 33. Liu, H W. & Walsh, C.T. (1987) Biochemistry o f the c yclopropyl group. In The Chemistry of the Cyclopropyl Group (Rappoport, Z., ed.), pp. 959–1025. J. Wiley and Sons, NY. 34. De Meijere, A. (2003) Introduction: cyclopropanes and related rings. Chem. Rev. 103, 931–932. 35. Yuan, Y. & Barry, C.E. (1996) A common mechanism for the biosynthesis of methoxy and cyclopro pyl mycolic acids in Myco- bacterium tuberculosis. Proc. Natl Acad. Sci. USA 93, 12828– 12833. 36. Dubnau, E., Laneelle, M.A., Soares, S., Benichou , A., Vaz, T., Prome, D., Prome, J.C., Daffe, M. & Qu e ´ mard, A. (1997) Mycobacteriu m bovis BCG genes involved in the biosynthesis of cyclopropyl keto- and hydroxy-mycolic acids. M ol Microbiol. 23, 313–322. 37. Walsh, C.T. (1979) Enzyme-catalysed Claisen condensation. In Enzymatic Reaction Mechanisms, pp. 759–763. W.H. Freeman, New York. 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 . Escherichia coli cyclopropane fatty acid synthase Mechanistic and site-directed mutagenetic studies Fabienne Courtois, Christine. 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