Vibrio cholerae FabV Defines a New Class of Enoyl-Acyl Carrier Protein Reductase*

Enoyl-acyl carrier protein (ACP) reductase catalyzes the last step of the fatty acid elongation cycle. The paradigm enoyl-ACP reductase is the FabI protein of Escherichia coli that is the target of the antibacterial compound, triclosan. However, some Gram-positive bacteria are naturally resistant to triclosan due to the presence of the triclosan-resistant enoyl-ACP reductase isoforms, FabK and FabL. The genome of the Gram-negative bacterium, Vibrio cholerae lacks a gene encoding a homologue of any of the three known enoyl-ACP reductase isozymes suggesting that this organism encodes a novel fourth enoyl-ACP reductase isoform. We report that this is the case. The gene encoding the new isoform, called FabV, was isolated by complementation of a conditionally lethal E. coli fabI mutant strain and was shown to restore fatty acid synthesis to the mutant strain both in vivo and in vitro. Like FabI and FabL, FabV is a member of the short chain dehydrogenase reductase superfamily, although it is considerably larger (402 residues) than either FabI (262 residues) or FabL (250 residues). The FabV, FabI and FabL sequences can be aligned, but only poorly. Alignment requires many gaps and yields only 15% identical residues. Thus, FabV defines a new class of enoyl-ACP reductase. The native FabV protein has been purified to homogeneity and is active with both crotonyl-ACP and the model substrate, crotonyl-CoA. In contrast to FabI and FabL, FabV shows a very strong preference for NADH over NADPH. Expression of FabV in E. coli results in markedly increased resistance to triclosan and the purified enzyme is much more resistant to triclosan than is E. coli FabI.

between the FAS I and FAS II systems make the FAS II enzymes good targets for antibacterial inhibitors (3)(4)(5)(6). The paradigm FAS II system is that of Escherichia coli and this has provided an excellent model system. Most FAS II enzymes are relatively conserved among bacteria and some domains of the FAS I proteins are clearly derived from FAS II proteins (7). An exception is the last step of the elongation cycle ( Fig. 1) formation of a saturated acyl-ACP by an NAD(P)H-dependent reduction of the enoyl-ACP double bond. In E. coli this reaction is catalyzed by the product of the fabI gene, which was discovered as the target for the antibacterial action of a set of diazaborine compounds. FabI was later shown to also be the site of action of triclosan (8), an antibacterial compound used in hand soaps and a large variety of other everyday products. Although FabI homologues are widely distributed in bacteria and other FAS II-containing organisms, the existence of a number of bacterial species naturally resistant to triclosan was soon recognized. In these bacteria triclosan resistance was due to the presence of other enoyl-ACP reductase isozymes of varying resistance to triclosan. Bacillus subtilis contains two isozymes, a FabI homologue and a somewhat triclosan-resistant isozyme called FabL (9) that, like FabI, is a member of the short chain dehydrogenase reductase superfamily. Streptococcus pneumoniae contains a single enoyl-ACP reductase, FabK, that is refractory to triclosan and is a flavoprotein unrelated to the short chain dehydrogenase reductase isozymes (10).
Vibrio cholerae is a Gram-negative bacterium that causes cholera in humans. When we began our work only a single genome sequence was available for this organism (11) and this sequence argued that V. cholerae did not encode a convincing homologue of any of the known enoyl-ACP reductase isozymes (FabI, FabK, or FabL). This conclusion was recently confirmed by shotgun genome sequences of a large number of V. cholerae strains now available from the NCBI genomes data base. Moreover, the genome sequences of other Vibrio species also show the lack of known enoyl-ACP reductase homologues. The lack of a FabI homologue was especially surprising because otherwise the Vibrio FAS proteins are very similar to those of E. coli. Indeed, the major FAS II gene clusters of E. coli and V. cholerae have almost identical organizations. We report the isolation of the gene that encodes a fourth enoyl-ACP reductase isozyme that we have named FabV and some properties of the new enzyme.

EXPERIMENTAL PROCEDURES
Construction of Bacterial Strains and Plasmids-E. coli strain JP1111, which carried the temperature-sensitive (Ts) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1-S4 and fabI392 allele, was obtained from the Coli Genetic Stock Center, Yale University. The mutation, which will be called fabI(Ts), allows growth at 30°C but blocks growth at 42°C. The temperature-sensitive activity of the mutant enzyme has been directly demonstrated (12) and is due to a single point mutation (13). Strain EPI300 (genotype recA1 endA1 araD139 ⌬(ara,leu)7697 trfA, with the trfA gene being under arabinose regulation) (14) was made temporarily proficient in homologous recombination by introducing plasmid pEAK2, which carries a functional recA gene on an unstable plasmid (15). To move the fabI(Ts) mutation from JP1111 into this strain a closely linked Tn10 transposon insertion was introduced into the JP1111 chromosome by transduction to tetracycline resistance at 30°C with a phage P1vir stock grown on the trpC::Tn10 strain CAG18455 (16,17) followed by screening for recombinants that remained temperature sensitive. A phage P1vir stock grown on one of these recombinants was then used to transduce the EPI300 (pEAK2) strain to tetracycline resistance at 30°C with screening for temperature sensitivity and loss of the pEAK2 antibiotic resistance determinant to give strain PMT03. For plasmid constructions the primers listed in supplemental Table 1S were used to amplify the fabV gene from the complementing cosmid using Pfu Turbo DNA polymerase (Stratagene). The PCR products were directly cloned into the expression vectors as given in supplemental Table 1S using the EcoRI and PstI restriction sites included in the primer sequences. Cloning into pET101 was performed using the Champion pET101 Directional TOPO expression kit (Invitrogen). Plasmids expressing the native form of FabV were obtained by use of primers CtagVcfor and NtagVcREV for insertion into pET101 and primers EcoVCfor and PstVCREV for insertion into pBAD322. The plasmids expressing the C-terminal His 6 -tagged form of FabV were obtained by use of primers CtagVcfor and NtagVcREV for insertion into pET101 and primers EcoVCfor and pBADCtagVc rev for insertion into pBAD322. The pBAD322-derived plasmid that expressed the N-terminal His 6 -tagged form of FabV were obtained by use of primers pBADNtagVc and PstVCREV. The plasmid constructions were verified by DNA sequencing performed by the Keck Genomics Center of this institution.
Cosmid Library Production and Subcloning-The CopyControl Fosmid Library Production Kit from Epicentre (Madison, WI) was used to generate an unbiased V. cholerae genomic library. The pCC1FOS vector carries two origins of replication, a single copy origin (ori2) and an inducible high copy origin (oriV) (14). Genomic DNA from cells of the V. cholerae biotype albensis strain ATCC 14547 was extracted using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI). The DNA fragments were obtained by mechanically shearing genomic DNA by 20 passages through a 200-l pipette tip. The resulting DNA fragments were end-repaired to 5Ј-phosphorylated blunt ends and then separated on a 0.8% low melting pointagarose gel run overnight at 30 V. Sheared DNA fragments of about 40 kb were size selected, gel-purified, ligated to the dephosphorylated pCC1FOS vector, and then packaged into phage particles. This was used to transduce the fabI(Ts) E. coli recipient strain PMT03 to chloramphenicol resistance at 30°C under conditions where the entering plasmids would replicate at single copy. The resulting transductants were then tested for the ability to grow at 42°C, the nonpermissive temperature of the fabI(Ts) allele. A complementing clone that grew well at 42°C was obtained. The purified complementing cosmid was again transformed into the fabI(Ts) strain PMT03 to confirm the complementation. This strain was then induced with arabinose to induce the cosmid to high copy number and large quantities of the complementing cosmid DNA were obtained using the Qiagen Large-construct Kit (Valencia, CA). The cosmid was then partially sequenced using vectorspecific primers to localize the insert DNA on the V. cholerae genome. To obtain smaller complementing clones 50 g of the 40-kb cosmid was sonicated for 5 s, and the DNA fragments were end-repaired before ethanol precipitation in the presence of glycogen. Fragments larger than 3 kb were gelpurified, cloned into the pCC1FOS vector. The resulting plasmids were then transformed into mutant strain PMT03 and screened for complementation at 42°C and for triclosan resistance at 30°C. Each complementing subclone was completely sequenced and its genomic location identified by a BLAST search against the V. cholerae El Tor N16961 genome (11) using the TIGR Comprehensive Microbial Resource website (cmr.tigr.org).
Enzyme Purification-Primers were designed for cloning the candidate gene into a pET101 expression vector from the Invitrogen Champion pET101 Directional TOPO Expression Kit (Carlsbad, CA). Six histidine codons were added at the 3Ј-end of the gene to produce a His tag at the C-terminal end of the protein. The gene was PCR-amplified using Pfu Turbo DNA polymerase from Stratagene (La Jolla, CA). The PCR fragment was cloned into pET101, sequenced, then transformed into BL21 Star (DE3) chemically competent cells from Invitrogen. The expression of the cloned fragment was induced with 150 M isopropyl ␤-D-thiogalactopyranoside. The crude extract was then applied to a nickel-nitrilotriacetic acid spin column from Qiagen (Valencia, CA), washed FIGURE 1. The elongation cycle of fatty acid biosynthesis. Each elongation cycle is initiated by the condensation of malonyl-ACP with acyl-ACP carried out by one of the 3-ketoacyl-ACP synthases. The second step is the reduction of 3-ketoester by 3-ketoacyl-ACP reductase followed by the dehydratation of the resulting 3-D-hydroxyacyl-ACP to the trans-2 unsaturated acyl-ACP. This substrate that at the C4 stage is called crotonyl-ACP is reduced by enoyl-ACP reductase such as FabV, FabI, FabL, or FabK to generate an acyl-ACP two carbons longer than that the original acyl-ACP substrate.
with the lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 20 mM imidazole), and the reductase eluted with 250 mM imidazole in the same buffer. Similarly, the native protein (lacking a histidine-tag) was also cloned and overexpressed from the pET101 vector. Purification was first performed through a Vivapure D maxi H spin column from Vivascience (Sartorius). The protein was washed with 4 mM Tris-HCl, 1 mM EDTA (pH 8) lysis buffer, discontinuously eluted with a 50 mM Tris-HCl (pH 7.5) buffer containing 100, 200, 300, 400, 500, 700, or 1000 mM NaCl. Most of the reductase was eluted with 400 mM NaCl. The fractions eluted at 400 and 500 mM NaCl were pooled, then desalted by overnight dialysis before passing through a Blue Sepharose column from using the ÅKTA purification apparatus (GE Healthcare). Elution from the affinity column was achieved using a 2 M LiCl buffer solution after washing with 15 column volumes of 20 mM sodium phosphate buffer (pH 7). Finally the protein was concentrated by ultracentrifugation using a 30,000 MWCO Amicon Ultra centrifugal filter device from Millipore.
Fatty Acid Synthesis Assays-The ability of the new gene to restore in vivo fatty acid synthesis was tested by transforming the fabI(Ts) mutant strain, JP1111 with a plasmid in which the gene encoding the putative V. cholerae enoyl-ACP reductase had been inserted into the arabinose inducible pBAD322 vector (18). The cultures were grown at permissive temperature, induced with arabinose, shifted to 42°C, and then [1-14 C]acetate (specific activity, 55 mCi/mmol) from American Radiolabeled Chemicals, Inc. (St Louis, MO) was added as described in Lai and Cronan (19). Labeled phospholipids were extracted, run on a thin layer chromatography plate, and visualized after overnight exposure to x-ray film (19). The in vitro fatty acid synthesis assay was performed using a mixture of 0.1 M sodium phosphate buffer (pH 7), 0.1 M LiCl, 50 M acetyl-CoA, 25 M holo-ACP, 150 M NADH, 1 mM 2-mercaptoethanol, 200 g of ammonium sulfate fraction of a cell extract (7), and 5 g of pure C-terminal His 6 -tagged FabV protein (when added). Following addition of 25 M [2-14 C]malonyl-CoA (specific activity, 51 mCi/mmol), the 200-l reactions were incubated at 37°C for 45 min (19). Radiolabeled malonyl-CoA was purchased from Moravia Biochemicals, Inc. (Brea, CA). Free fatty acids were extracted by saponification followed by acidification and petroleum ether extraction as reported by Gelmann and Cronan (20).
Cell-free Extract Preparation-Cell cultures of wild type E. coli and the fabI mutant strain JP1111 were centrifuged and resuspended in lysis buffer (0.1 M sodium phosphate, pH 7.5, mM 2-mercaptoethanol, 1 mM EDTA) before lysis through a French pressure cell. Partial purification of fatty acid synthetic enzymes was performed by gradual protein precipitation of each lysate with ammonium sulfate. The same procedure was used to prepare V. cholerae extracts. In the case of strain JP1111 the cultures were shifted from 30 to 42°C for 15 min before harvesting to inactivate the mutant FabI protein.
Preparation of Holo-ACP, Apo-ACP, and Crotonyl-ACP-Holo-ACP was purified following the procedure described by Thomas and Cronan (21). The purification of apo-ACP was identical to that of holo-ACP, except that AcpH, rather than AcpS, was overexpressed along with AcpP to convert the holo-ACP species to apo-ACP. 3 Crotonyl-ACP was enzymatically synthesized from crotonyl-CoA purchased from Sigma using B. subtilus Sfp phosphopantetheinyl transferase (22) expressed from pNRD136. In a 30-ml reaction volume, 25 mg of crotonyl-CoA, 1 mM dithiothreitol, 30 mg of apo-ACP, and 2 mg of Sfp enzyme were mixed in a buffer containing 50 mM Tris-HCl (pH 8.8), 10 mM MgCl 2 and then incubated for 4 h at 37°C. The reaction mixture was then precipitated with 5% trichloroacetic acid and 0.02% deoxycholate resuspended, dialyzed, and concentrated with an Amicon ultracentrifugation filter device from Millipore (5,000 MWCO) (21). Crotonyl-ACP synthesis was verified on a 20% native gel containing 2.5 M urea and by electrospray mass spectrometry (observed mass 8915.39; calculated mass 8915.09).
NADH Oxidation Assay-Enoyl-ACP reductase activity was monitored spectrophotometrically by decrease in absorbance at 340 nm using an NADH extinction coefficient of 6220 M Ϫ1 . Each 100-l reaction was performed in disposable UV-transparent microcuvettes obtained from Brand-Tech Scientific, Inc. The activity assays contained varying concentrations of NADH, 1 pg of the purified native FabV, varying substrate concentrations (crotonyl-CoA or crotonyl-ACP), and 0.1 M LiCl in a 0.1 M sodium phosphate buffer (pH 7). Kinetic constants were determined using GraphPad PRISM version 4 software. The K m values for NADH and NADPH were determined at a crotonyl-ACP concentration of 80 M. The K m values for crotonyl-ACP and crotonyl-CoA was determined using 150 M NADH. Triclosan was purchased from KIC Chemicals, Inc. (Armonk, NY).
Size Exclusion Chromatography-The native FabV was chromatographed on a Superdex 200 HR 10/30 column in a 0.15 M NaCl, 50 mM sodium phosphate buffer at pH 7 on the AKTA purifier (GE Healthcare). The calibration curve was prepared using chymotrypsinogen, ovalbumin, bovine serum albumin, aldolase, ferritin, and thyroglobulin as standards ranging from 25,000 to 669,000 Da. The FabV peak was detected by absorbance at 280 nm and the reductase activity in the fraction was confirmed by spectrophotometric assay and the size of the monomer was confirmed by 8% SDS-PAGE of the peak fractions.

Identification of a V. cholerae Gene Encoding a Novel
Enoyl-ACP Reductase-A cosmid was selected from the V. cholerae genomic library that consistently complemented growth of the fabI(Ts) strain PMT03 at 42°C (Fig. 2A). The large V. cholerae genomic fragment (about 40 kb) carried by the cosmid was fragmented and the fragments were cloned into the same vector to obtain smaller complementing plasmids. Four of these subclones were able to complement growth of the fabI(Ts) mutant strain at 42°C and only those clones also allowed growth on triclosan at 30°C (Fig. 2B). Upon partial sequencing, all four clones were found to contain a common genomic region that contained only three open reading frames (ORFs) by comparison with the genome sequence of V. cholerae O1 biovar El Tor. N16961, the only sequenced V. cholerae genome then available. One of the three ORFs (VC1737) was annotated as encoding protein synthesis initiation factor IF-1, whereas the other two ORFs, VC1738 and VC1739, were annotated as hypothetical proteins (supplemental Fig. 1S). The latter two ORFs had been predicted to be in a single transcriptional unit (www.BioCyc. org). To identify these ORFs, a BLAST search of each of the ORFs was performed against all of the bacterial genomes in the data base. Both VC1738 and VC1739 showed homology to a single ORF found in a number of bacterial genomes. These were (accession number follows the species name) Xanthomonas campestris (YP361851) Vibrio parahemeolyticus (NP797610), Vibrio vulnificus (NP761639), Vibrio fischeri (YP204271), Shewanella oneidensis (NP717408, Shewanella denitrificans (YP_563490), Aeromonas hydrophila (YP857138, Xylella fastidiosa (NP779243), and Photobacterium profundum (YP130609). That is, when one of the V. cholerae N16961 ORFs aligned with an ORF of one of these bacteria, the other V. cholerae N16961 ORF aligned with that same ORF without overlap. Hence, the original V. cholerae genomic sequence seemed to contain a frameshift introduced by a sequencing error that had cut a single ORF into two ORFs. We fully sequenced the insert of one of the complementing V. cholerae subclones and found that the VC1738 and VC1739 ORFs were indeed a single ORF consistent with a prediction by the Swiss-Prot data base. The error was omission of a single base within a run of six adenine bases on the coding strand (supplemental Fig. 1S). This was confirmed by the finding that the genome fragment containing the putative VC1738 and VC1739 ORFs expressed only one protein of the molecular mass (45 kDa) expected from elimination of the frameshift (see below). Our sequencing data were subsequently confirmed upon submission of shotgun genomic sequences of a large number of V. cholerae strains to the NCBI Whole Genome Shotgun (wgs) data base. All show that the original VC1738 and VC1739 ORFs lie within a single protein. The comparative analysis of the protein sequence encoded by our candidate gene extended the presence of homologues of our candidate protein to Xanthomonas oryzae (YP449055) Xanthomonas axonopodis (NP640497), Pseudomonas putida (NP746744), Pseudomonas entomophila (YP610084), Yersinia pestis (NP407525), Yersinia pseudotuberculosis (YP072425), Pseudomonas aeruginosa (NP251640), Saccharophagus degradans (YP528097), Methylobacillus flagellatus (YP545785), Burkholderia species (e.g. YP443187) Cytophaga hutchinsonii, Pseudoalteromonas species (e.g. NC008228), Clostridium species (NC008261), Idiomarina loihiensis (NZAAMX01000026), Treponema denticola (NC002967), and Streptomyces avermitilis (NC003155). These alignments range from 96.5% protein identity for the V. fischeri homologue, to 52.9% residue similarity for the Streptomyces avermitilis homologue. We have named the gene fabV and the encoded protein FabV because it is predominately found in Vibrio species and closely related organisms where it seems to be the sole enoyl-ACP reductase. However, due to the diverse reactions catalyzed by the enzymes of the short chain dehydrogenase reductase (SDR) superfamily some of the more poorly aligned proteins may not be enoyl-ACP reductases. Several of the above bacteria are human pathogens and others infect plants and marine organisms. There may be bacteria that have both FabV and FabK such as various of the  Clostridia such as Clostridium tetani. However, like FabV various putative FabK homologues may not posses enoyl-ACP reductase activity (23). FabV, like FabI and FabL, is a member of the SDR superfamily (24,25) although FabV aligns only weakly with FabI or FabL even when many gaps are allowed (Fig. 3). FabV does not contain the Tyr-(Xaa) 3 -Lys motif typical of most members of the SDR superfamily. FabI and FabL also lack this motif and instead use a Tyr-(Xaa) 6 -Lys active site motif. In FabV another variation is seen, a Tyr-(Xaa) 8 -Lys motif (Fig.  3). Given the conservation of residues that neighbor the key tyrosine and lysine residues this seems likely to be the FabV active site although this remains to be tested by mutagenesis. Despite the low homology scores FabV can be modeled on the crystal structures of known SDR enoyl-ACP reductases by the Geno3D server (geno3d-pbil.ibcp.fr), which takes into account known protein structures as well as predicted geometrical restraints (26).
The Enzymatic Activity of FabV-Expression of FabV in a fabI(Ts) mutant strain restored de novo fatty acid synthesis at the non-permissive temperature to wild type levels (Fig.  4). Moreover, addition of purified FabV (see below) to cell extracts of fabI(Ts) mutant strains gave wild type rates of in vitro fatty acid synthesis (supplemental Fig. 2S). These early in vitro experiments were done with a version of FabV carrying a His 6 purification tag and purified by Ni 2ϩ -cheleate chromatography (see below). Later we found that under conditions of low expression (where protein production is stochastic among cells (27,28)) the survival of an E. coli fabI(Ts) mutant strain expressing this construct was much lower than that of the same strain expressing the native form of the protein (Fig. 5). Because the N-terminal His 6 -tagged FabV  also gave inefficient complementation (albeit to a much lesser extent), we produced the native FabV at high levels in E. coli and purified it from that source by two chromatographic steps (Fig. 6). The most effective step in the purification of the native protein was chromatography on Blue Sepharose CL-6B where the immobilized Cibacron Blue F3G-A dye mimics adenylate-containing cofactors such as NAD (29). A similar step was used in the purification of E. coli FabI (30).
The native FabV was active on both crotonyl-CoA and crotonyl-ACP using the standard NADH oxidation spectrophotometric assay and thus commercially available crotonyl-CoA was often used to assay the enzyme (Figs. 7 and 8). The native FabV was much more active then the C-terminal His 6 -tagged form (Fig. 8A) and showed a very strong preference for NADH over NADPH when assayed with crotonyl-CoA (Fig. 8B). The K m for NADPH was 60-fold higher than that for NADH ( Table 1). The experimentally observed maximal velocities of FabV with crotonyl-CoA and crotonyl-ACP were similar with the rate with the latter substrate being about 50% higher. The K m for crotonyl-ACP (195 M) was 6-fold lower than that for crotonyl-CoA (1178 M) ( Table 1). It should be noted that the ACP used was that of E. coli rather than that of V. cholerae. However, the ACP of V. cholerae is 96% identical to that of Vibrio harveyi, which has been shown to functionally replace the ACP of E. coli in vivo (31) (the ACPs of V. cholerae and E. coli are 84% identical).
FabV is a monomeric protein in solution as determined by size exclusion chromatography (Fig.  6). This is in contrast to FabI, which is a homotetramer. The solution structure of FabL has not been reported, although it has been crystallized (32).
Triclosan Resistance of FabV-Early in our work we found that expression of FabV in E. coli imparted resistance to triclosan (Fig. 2B). A similar picture was seen in an E. coli in vitro fatty acid synthesis system in which the inactivated mutant FabI was replaced with FabV (supplemental Fig. 3S) or when an in vitro fatty acid synthesis system prepared from V. cholerae cells was treated with triclosan (data not shown). Moreover, growth of the wild type V. cholerae strain ATCC 14547 was 20-fold more resistant to triclosan than was the wild type E. coli strain MG1655 (supplemental Fig. 4S). These observations argued that FabV might be significantly more resistant to triclosan than is E. coli FabI. Two modes of triclosan inhibition of enoyl-ACP reductases have been observed. Some triclosan-sensitive enoyl-ACP reductases, such as FabI,areirreversiblyinhibitedbyformationofadead-endreductase-NAD ϩ -triclosan ternary complex, whereas other reductases such as FabL show reversible inhibition without formation of a ternary complex. Moreover, under the usual assay conditions triclosan is a slow binding inhibitor of FabI-type enzymes because the reductase-NAD ϩ complex must be formed before the inhibitor can bind. Triclosan inhibition of FabV is very weak (Fig.  9). However, the degree of FabV inhibition increased as the reaction proceeds indicating that a product, presumably NAD ϩ , may be involved in the inhibition. Indeed, triclosan inhibition of FabV was potentiated by preincubation with NAD ϩ (Fig. 9). However, even in the presence of NAD ϩ , triclosan was not a potent inhibitor of FabV.

DISCUSSION
We have identified a new enoyl-ACP reductase isoform encoded by V. cholerae. The lack of sequence alignment with known enoyl-ACP reductases together with its atypical active site and much greater size indicates that the V. cholerae enzyme should be considered a member of a new class of enoyl-ACP reductases having homologues in both closely and distantly related species of bacteria. We have named this fourth enoyl-ACP reductase isoform, FabV for Vibrio, a species in which this is the only enoyl-ACP reductase now predictable by sequence alignment. Like its FabI and FabL functional homologues, it is a member of the SDR superfamily. This very large family of NAD(P)-dependent oxidoreductases shows a significantly conserved structure despite little sequence homology (15-30%) among members. The largely conserved SDR folding pattern allows specific sequence motifs to be assigned, with those for the coenzyme-binding and active site regions being the most definitive. A recent bioinformatic analysis of the SDR superfamily placed the enzymes into five families (25). One of these families (called divergent) is composed of the FabI-type enoyl-ACP reductases of bacteria and plants (25). FabV does not fall cleanly into this or any of the other four families. FabV is 60% larger than the typical SDR family member (which are generally about 250 residues in length) and the spacing between the putative FabV active site tyrosine and lysine residues is eight residues, two more than FabI and FabL and one more than the maximum reported for other SDR proteins. Conversely the coenzyme-binding site of FabV has the classical Rossman fold motif like that of FabL, whereas the FabI coenzyme-binding fold departs from that motif. Persson and co-workers (25) have reported that the coenzyme preference for the SDR enzymes can be predicted with Ͼ93% accuracy based on the presence or absence of key residues. The very strong preference of FabV for NADH over NADPH fits the predictive rules of Persson and co-workers (25). These rules also correctly predict the coenzyme preferences of B. subtilis FabL (NADPH strongly preferred) (9) and E. coli FabI (little or no preference) (12).
In vivo and in vitro assays demonstrate that expression of FabV restores the otherwise defective de novo fatty acid synthesis of the E. coli fabI392(Ts) mutant at the nonpermissive temperature to normal (Figs. 2B and 4). Because  The NADH control lacking crotonyl-CoA is also shown (X-X).

TABLE 1 The kinetic parameters of FabV
The catalytic efficiency of each enzyme was determined under pseudo first-order conditions with NADH in excess and crotonyl-CoA (or crotonyl-ACP) concentrations under K m values (100 and 15 M, respectively). Under these conditions, the first-order rate constant k obs was determined over time and catalytic efficiency was determined as k obs /͓E T ͔ ϭ K cat /K m . The enzyme concentrations were 2.61 pM for native FabV and 25 nM for the C-terminal His 6 -tagged (C-tag) FabV. The C-terminal-tagged FabV was studied only with crotonyl-CoA and NADH. The large standard deviation seen with NADPH as the substrate was due to the very low activities observed.

K m
Catalytic efficiency k obs /͓E T ͔ ‫؍‬ k cat /K m , crotonyl-CoA/crotonyl-ACP FabI is the sole enoyl-ACP reductase of E. coli and is known to reduce enoyl-ACPs of acyl chain length from C2 to C18 (9,33), the complementation data indicate that FabV is active with this range of acyl chain length substrates. Moreover, expression of FabV confers triclosan resistance to E. coli (Figs. 2B and supplemental Fig. 4S) as previously seen for FabK and FabL. The resistance imparted by FabV is also observed in cell extracts and for the purified enzyme. Therefore, FabV is the first intrinsically triclosan-resistant enoyl-ACP reductase found in a Gram-negative bacterium. Indeed, FabV shows little inhibition at triclosan concentrations approaching the solubility limit of the compound. Moreover, expression of FabV from a single copy vector in E. coli made the host more resistant to triclosan than was the V. cholerae strain from which the encoding gene was derived. Two of the possible explanations for this unexpected result are that V. cholerae may have a second target for triclosan, either another enoyl-ACP reductase or a FAS II-independent target (34,35) or that FabV expression is negatively regulated in V. cholerae, but not in E. coli. Triclosan is a very weak inhibitor of FabV even when inhibition is potentiated by addition of NAD ϩ . For example, FabL shows 50% inhibition by 12 M triclosan (9), whereas the same extent of inhibition of FabV requires 136 M triclosan in the absence of added NAD ϩ (Fig. 9). Interpretation of the effects of NAD ϩ on triclosan inhibition is complicated by the finding that NAD ϩ was a much stronger inhibitor of FabV than triclosan (5 M NAD ϩ inhibited similarly to 168 M triclosan). This is not the case for FabI. NAD ϩ does not inhibit FabI and indeed FabI fails to bind NAD ϩ unless triclosan is present (9, 36 -38). Triclosan increases the affinity of FabI for NAD ϩ by 1000-fold such that the inhibitory ternary complex can be formed (9, 36 -38). FabV may also form an inhibitory ternary complex. If so, the triclosan-mediated increase in the affinity of FabV for NAD ϩ would be very modest relative to that seen with E. coli FabI and would require much higher levels of triclosan. In the presence of NAD ϩ E. coli FabI is completely inhibited by 2 M triclosan (36), whereas complete inhibition of FabV in the presence of NAD ϩ required about 300-fold higher levels of triclosan (Fig. 9). In any event triclosan is such a weak inhibitor of FabV that further investigation of the mode of inhibition seems of little practical use. In contrast to FabI and FabV, the reversible triclosan inhibition of FabL does not seem to involve its interaction with NAD ϩ (9).
The diversity of the bacterial enoyl-ACP reductases relative to the lack of structural and mechanistic diversity seen in the other enzymes of the FAS II elongation cycle argues that naturally occurring compounds exist that selectively inhibit one or another of these enzymes. This hypothesis has recently been confirmed by the discoveries of natural enoyl-ACP reductase inhibitors of fungal origin that specifically target FabI (Cephalochromin) (39) and FabK (Atromentin and Leucomelone) (40).