INTRODUCTION

Fatty acyl-CoA synthetase (fatty acid:CoA ligase, AMP-forming; EC 6.2.1.3) plays a central role in intermediary metabolism by catalyzing the formation of fatty acyl-CoA from fatty acid, ATP, and CoA. Fatty acyl-CoAs represent bioactive compounds that are involved in protein transport (1, 2), enzyme activation (3, 4), protein acylation (5, 6), cell signaling (7), and transcriptional control (8, 9) in addition to serving as substrates for -oxidation and phospholipid biosynthesis. Given the multiple roles of fatty acyl-CoA, it is clear that fatty acyl-CoA synthetase occupies a pivotal role in cellular homeostasis, particularly in lipid metabolism. This enzyme catalyzes the formation of fatty acyl-CoA by a two-step process that proceeds through the hydrolysis of pyrophosphate (10). A fatty acyl-adenylate intermediate is formed during the first step of the reaction with the corresponding release of pyrophosphate. The second step of the reaction requires CoA for the enzymatic conversion of the fatty acyl-adenylate to fatty acyl-CoA with the concomitant release of AMP.

Our interests in fatty acyl-CoA synthetase stem from its role in the transport of exogenous long chain fatty acids (C12-C18) and in the regulation of the transcription factor FadR in the Gram-negative bacterium Escherichia coli. This enzyme activates exogenous long chain fatty acids to CoA thioesters concomitant with transport (i.e. vectorial esterification). As a component of the transport apparatus, fatty acyl-CoA synthetase may partition into the inner membrane and interact with a postulated fatty acid/H⁺ cotransporter (11-13). The net result of long chain fatty acid transport is a predicted increase in intracellular pools of long chain acyl-CoA, which in turn control the activity of the global transcription factor FadR. When long chain fatty acyl-CoAs are bound to FadR, there is transcriptional derepression of the genes involved in fatty acid transport, activation and degradation, and transcriptional inactivation of at least two genes involved in fatty acid biosynthesis (8, 9).¹ In this regard, fatty acyl-CoA synthetase plays a central role in maintaining the appropriate levels of long chain fatty acyl-CoA within the cell and thus the transcriptional activities of genes involved in fatty acid degradation and biosynthesis.

Exogenous long chain fatty acids represent an important class of hydrophobic compounds that can serve as a sole carbon and energy source to support the growth of E. coli. The acquisition of these compounds prior to -oxidation occurs by a highly specific process that minimally requires the outer membrane protein FadL and fatty acyl-CoA synthetase (9, 14-17). FadL represents a bona fide membrane-bound long chain fatty acid transport protein that we propose contains a specific hydrophobic channel for transport that becomes accessible upon ligand binding (9, 16-21). The process of long chain fatty acid transport responds to the energized state of the membrane and to the presence of the periplasmic protein Tsp (22). As noted above, exogenous long chain fatty acids are activated by fatty acyl-CoA synthetase concomitant with transport. The notion that the synthetase is a component of a fatty acid transport apparatus is in agreement with recent data showing that the transport and activation of exogenous long chain fatty acids in mouse adipocytes occurs via a fatty acid transport protein and associated fatty acyl-CoA synthetase (23).

The E. coli fatty acyl-CoA synthetase structural gene (fadD) has been cloned and sequenced and shown to encode a protein of 561 amino acid residues (26, 27). This enzyme shares amino acid sequence similarities to other fatty acyl-CoA synthetases and more broadly the family of adenylate-forming enzymes (26-28). The placement of this enzyme in the family of adenylate-forming enzymes is based on amino acid similarities within a proposed AMP/ATP binding domain (25). Of particular relevance to the present work was the identification of a second conserved segment that appears to be restricted to the family of fatty acyl-CoA synthetases (26). Within this latter region is a highly conserved 25-amino acid residue segment that we propose specifies a signature motif common to the family of fatty acyl-CoA synthetases. A series of site-directed alanine substitutions within the fadD gene corresponding to this signature motif were constructed and analyzed to provide information regarding the role of this region to the catalytic activity of the enzyme. Our present data confirmed the prediction that this region of the enzyme was essential for activity. Three distinct classes of mutations were identified on the basis of growth characteristics of a fadD strain harboring these alleles on a low copy plasmid, fatty acyl-CoA synthetase profiles using oleate and decanoate as substrates, and studies using purified mutant forms of the enzyme: 1) those that resulted in either wild-type or nearly wild-type fatty acyl-CoA synthetase activity profiles; 2) those that had little or no enzyme activity; and 3) those that resulted in altered fatty acid chain length specificity. Within this latter group, seven mutant fadD alleles have been generated that have altered activity with fatty acids of differing chain lengths when compared with the wild type. On the basis of the data presented in this work, we propose that this signature motif within fatty acyl-CoA synthetase functions in part to promote fatty acid chain length specificity and may compose part of the fatty acid binding site.

EXPERIMENTAL PROCEDURES

The bacterial strains used in this study were BMH 71-18 (thi supE (lac proAB) mutS::Tn10 [F proAB lacI^qZM15]); JM109 (endA1 relA1 gyrA96 thi hsdR17 (r_K, m_K+) relA supE44 (lacproAB) [F traD36 proAB lacI^qZM15]); LS1187 (C600 fadR); LS6928 (fadR fadD88 zea::Tn10); and BL21 (DE3)/pLysS (8, 29-31). Bacterial cultures were grown at 37 °C in a Lab Line gyratory shaker in Luria broth (LB)² or tryptone broth (TB). When minimal media were required, medium E supplemented with vitamin B₁ was used (32). Carbon sources, sterilized separately, were added to final concentrations of 25 mM glucose, 25 mM potassium acetate, 5 mM decanoate, or 5 mM oleate. When oleate or decanoate was used as a carbon source, polyoxyethylene 20 cetyl ether (Brij 58) was added to a final concentration of 0.5%. As required, amino acids were added to a final concentration of 0.01%. When required to maintain plasmids, antibiotics were added to 100 µg/ml ampicillin, 40 µg/ml kanamycin, 10 µg/ml tetracycline, and 40 µg/ml chloramphenicol. Growth of bacterial cultures was routinely monitored using a Klett-Summerson^TM colorimeter equipped with a blue filter.

The plasmids used in this study are listed in Tables I and II. Details concerning the construction of the plasmids harboring the fadD mutants are described below and under "Results."

Table I. Site-directed fadD mutations, corresponding mutagenic oligonucleotides, and pACYC177 derivatives fadD allele Mutagenic oligonucleotide sequencea pACYC177 derivative Wild type pN300 fadDN431A ATCATCAAATGGCTGGTT pN3610 fadDG432A TCAAAAATGCTGGTTAC pN3586 fadDW433A AAAATGGCGTTACACAC pN3588 fadDL434A ATGGCTGGACACACCG pN3590 fadDH435A GCTGGTTACACCGGCG pN3592 fadDT436A GGTTACACCCGGCGAC pN3594 fadDG437A ACACACCGCGACATCG pN3596 fadDD438A CACACCGGCGCATCGCGGTA pN3559 fadDI439A ACCGGCGACCGCGGTAAT pN3614 fadDG446A ATGAAGAAGATTCCTGCG pN3616 fadDL448A GAAGGATTCGCGCATTGT pN3618 fadDR449A GGATTCCTGCATTGTCGA pN3612 fadDI450A TTCCTGCGCTGTCGATCG pN3620 fadDV451A CTGCGCATTGCCGATCGTAA pN3622 fadDD452A GCATTGTCGTCGTAAAAAAG pN3602 fadDR453A CATTGTCGATTAAAAAAGAC pN3561 fadDK454A TGTCGATCGTAAAAGACATG pN3608 fadDK455A CGATCGTAAAAGACATGATT pN3563 a  Specific changes in the fadD sequence are underlined.

Table II. fadD wild-type and mutant fatty acyl-CoA synthetase expression plasmids fadD allele Expression plasmid Vector control pRSET-Ba Wild type (His6)b pN3576 fadDW433A (His6) pN3516 fadDD438A (His6) pN3612 fadDD452A (His6) pN3614 fadDK455A (His6) pN3613 a  Obtained from Invitrogen Corp. b  His6 indicates the presence of a hexamer histidine tag on the amino termini of these proteins.

Protein sequence comparisons were performed using BioSCAN³ with sequences that are conserved among the fatty acyl-CoA synthetases. These comparisons were directed against the SWISS-PROT data base using a 0.01 probability threshold with complexity filtering. The score table employed was blosum62.

Restriction, ligation, and transformation procedures have been previously described (33). Oligonucleotides (17-24-mer, for mutagenesis and sequencing) were synthesized on a Pharmacia Biotech Inc. Gene Assembler Plus and purified by phenol-chloroform extraction and ethanol precipitation. The final concentrations of oligonucleotides were estimated by determination of optical density at 260 nm. When required, oligonucleotides (100 pmol) were phosphorylated using 5 units of T4 polynucleotide kinase in 50 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 5 mM dithiothreitol, 0.1 mM spermidine, 1 mM ATP. Small scale plasmid isolation was achieved by using alkaline lysis (33) or by using QiaPrep columns as specified by the vendor (Qiagen). Single-stranded phagemid DNA and plasmid DNA were isolated from the different mutants and sequenced using the chain termination method of Sanger et al. (34) using [-³⁵S]dATP and Sequenase (version 2.0; U.S. Biochemical Corp.). Synthetic oligonucleotides (17- and 21-mer) complementary to various regions of the fadD gene were used as primers (26).

Site-directed mutations in the fadD gene were generated using the Altered Sites^TM mutagenesis system of Promega. The EcoRI-HinDIII fragment from pN324 (26) was gel-purified and ligated into the phagemid pSELECT-1^TM to generate pN351. This phagemid contained the entire coding sequences of the fadD and fadR genes in opposite orientations. fadR was included in these constructs to prevent toxicity of fadD expression in this high copy plasmid. Phosphorylated mutagenic oligonucleotides harboring the required mutation (1-2 pmol; see Table I) and the ampicillin repair oligonucleotide (0.25 pmol) were annealed to single-stranded pN351 phagemid DNA (100-250 ng) in 20 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 50 mM NaCl by heating at 70 °C for 5 min followed by slowly cooling to room temperature. Following annealing, the mutagenic oligonucleotides were extended with T4 DNA polymerase (10 units) and ligated with T4 DNA ligase (2 units) in 10 mM Tris-HCl, pH 7.5, 1 mM ATP, 2 mM dithiothreitol, and 0.5 mM dNTPs (dATP, dGTP, dTTP, and dCTP). The double-stranded DNA was transformed into a repair-defective strain of E. coli (BMH 71-18) using established methods. The pool of transformed cells was grown overnight in LB containing 100 µg/ml ampicillin followed by phagemid DNA isolation. Phagemid DNA pools were transformed into strain JM109 with selection on LB plates containing 100 µg/ml ampicillin. Phagemid DNA was isolated from individual colonies as recommended by Promega and sequenced using fadD-specific or T7 promoter-specific oligonucleotides to confirm the mutation. Once the mutation was confirmed, BamHI fragments containing the different fadD alleles mutation were gel-purified and ligated into pACYC177. These final plasmid constructions are listed in Table I, and upon transformation into the appropriate host strains were analyzed as detailed below.

An EcoRI site was generated at the +1 site of translation of the E. coli fatty acyl-CoA synthetase in the phagemid pN351 to generate pN3571. The EcoRI-HinDIII fragment from pN3571 containing the coding region of the fadD gene was gel-purified and ligated into pRSET-B (Invitrogen) to generate pN3576. The expression plasmid pN3576 contains six histidine residues and an enterokinase cleavage site preceding the coding sequence of the fadD gene. Expression of His-tagged fatty acyl-CoA synthetase is driven off a T7 RNA polymerase responsive promoter. Expression plasmids containing specific fadD mutations were generated by ligating gel-purified SalI-HinDIII fragments from the phagemid constructs listed in Table I into pN3576 digested with SalI and HinDIII. The final designations of the expression plasmids are listed in Table II.

The plasmids noted in Table II were transformed into strain BL21 (DE3)(pLysS), and wild-type or mutant forms of the enzyme were overexpressed following induction of midlogarithmic grown cultures in either LB or minimal medium E supplemented with glucose with isopropyl-1-thio--D-galactopyranoside for 90 min (31). Following induction, cells were harvested and prepared as described below for enzyme purification or resuspended in SDS sample buffer and boiled, and proteins were resolved on a 9 or 12% SDS-polyacrylamide gel using the Laemmli buffer system (35). Cells that were grown in minimal media were treated with rifampicin for the last 60-min period of induction as described previously (24) and labeled with [³⁵S]methionine. In this case, samples were prepared as described above, and following electrophoresis, the gels were dried and subjected to autoradiography for 2-24 h.

The bacterial strain BL21 (DE3)/pLysS containing the different expression plasmids listed in Table II were grown to midlog phase (6 × 10⁸ cells/ml) in 30-50-ml cultures of LB containing ampicillin and chloramphenicol. Following induction of T7 RNA polymerase with isopropyl-1-thio--D-galactopyranoside, cells were harvested, washed with minimal medium E, resuspended in 3-5 ml of 50 mM sodium phosphate, pH 8.0, 300 mM NaCl (buffer A), and sonicated on ice using 3 × 30-s pulses (200 watts). The cell extract was clarified by centrifugation at 50,000 rpm in a TLS-50 rotor using a Beckman TL-100 ultracentrifuge. Four hundred microliters of a 50% slurry of Ni²⁺-NTA-agarose (Qiagen) was added per 3-5-ml cell extract that had been previously equilibrated with buffer A. The cell extract-Ni²⁺-NTA-agarose slurry was incubated for 2-12 h at 4 °C with end-over-end rotation and finally applied to a 0.5-cm (inner diameter) column. The column was washed with 30 ml of buffer A; 30 ml of 50 mM sodium phosphate, pH 6.0, 300 mM NaCl, 10% glycerol (buffer B) containing 50 mM imidazole; and 5 ml of buffer B containing 100 mM imidazole. The His-tagged enzyme was eluted from the column using a imidazole step gradient (750 µl of 150 mM, 200 mM, 250 mM, 300 mM, 400 mM, 500 mM, and 600 mM imidazole in buffer B). Purification was monitored on 9% SDS-polyacrylamide gels using the Laemmli buffer system (35). The His-tagged fatty acyl-CoA synthetase generally eluted between 200 and 300 mM imidazole from the Ni²⁺-NTA column and was greater than 95% pure. The purified His-tagged enzyme was dialyzed against buffer B, ATP was added to a final concentration of 50 mM, and the sample was frozen at 80 °C in 100-500-µl aliquots. Fatty acyl-CoA synthetase activity was stable in the frozen enzyme over a period of at least 4 months. The mutant forms of the E. coli fatty acyl-CoA synthetase were purified using the same system with identical or similar elution profiles from the Ni²⁺-NTA-agarose column.

Bacteria (wild-type and fadD strains containing the collection of fadD⁺ and fadD clones) were grown to midlog phase (6 × 10⁸ cells/ml) in TB or TB supplemented with 5 mM oleate and 0.5% Brij 58 (TBO) and with antibiotics as required. Cells were harvested by centrifugation, washed twice with medium E, resuspended to a density of 1.2 × 10⁹ cells/ml in 10 mM Tris-HCl, pH 7.5, and lysed by three cycles of sonication at 0 °C. Fatty acyl-CoA synthetase activities were determined in sonicated cell extracts or using the purified wild-type or mutant hexamic histidine-tagged enzymes as described by Kameda and Nunn (24). The reaction mixtures contained 200 mM Tris-HCl, pH 7.5, 2.5 mM ATP, 8 mM MgCl₂, 2 mM EDTA, 20 mM NaF, 0.1% Triton X-100, 10 µM [³H]oleate, [³H]myristate or [¹⁴C]decanoate, 0.5 mM coenzyme A, and cell extract in a total volume of 0.5 ml. The reactions were initiated with the addition of coenzyme A, incubated at 35 °C for 10 min, and terminated by the addition of 2.5 ml of isopropyl alcohol, n-heptane, 1 M H₂SO₄ (40:10:1). The radioactive fatty acid was removed by organic extraction using n-heptane. Fatty acyl-CoA formed during the reaction remained in the aqueous fraction and was quantified by scintillation counting. Protein concentrations in the enzyme extracts and purified enzyme samples were determined using the Bradford assay and bovine serum albumin as a standard (36). The values presented represent the average from at least three independent experiments. All experiments were analyzed using analysis of variance (StatView; Abacus Concepts, Inc.)

Fatty acid binding capacity of the wild-type and mutant forms of fatty acyl-CoA synthetase were determined using Lipidex (for C18:1) or LH-20 (C10:0) as described previously (17). For oleic acid binding, substrate concentration was varied from 10 nM to 2.5 µM while the purified enzyme concentration was held at 0.2 mg/ml in 200 mM Tris-HCl, pH 7.5. Samples were incubated for 20 min at 37 °C, cooled to 0 °C in a salt-ice bath for 5 min, centrifuged (30 s at 3,000 rpm) through Lipidex preequilibrated with buffer, and held at 0 °C. The amount of protein-bound fatty acid in the eluate was determined by scintillation counting. For decanoic acid binding, the substrate concentration was varied from 50 nM to 50 µM while the purified enzyme concentration was held at at 0.2 mg/ml. Samples were treated as above, but they were passed through LH-20 instead of Lipidex. The amount of protein-bound fatty acid in the eluate was determined by scintillation counting. Background counts from samples containing no protein were subtracted from those in the eluates containing the protein samples. The data were converted into nmol of fatty acid bound/mg of purified protein. All of the data presented represent the average from at least three independent experiments.

[³H]Oleic acid, [³H]myristate, [³⁵S]methionine, and [-³⁵S]dATP were purchased from DuPont NEN. [¹⁴C]decanoic acid was purchased from Sigma. Enzymes for DNA sequencing (Sequenase) were obtained from U.S. Biochemical Corp., and enzymes for routine DNA manipulations were obtained from Life Technologies, Inc., Pharmacia, U.S. Biochemical Corp., or New England Biolabs. Chemicals for the synthesis of fadD-specific oligonucleotides and the mutagenic oligonucleotides were purchased from ABN/Biogenex or Pharmacia. Antibiotics and other supplements for bacterial growth were obtained from Difco and Sigma. All other chemicals were obtained from standard suppliers and were of reagent grade.

RESULTS

In our previous work we identified two regions within the E. coli fatty acyl-CoA synthetase that were similar to other adenylate-forming enzymes (24). The first region, corresponding to amino acids 200-273 within the E. coli enzyme, contains a putative AMP binding site. On the basis of this highly conserved sequence common to all adenylate-forming enzymes, the E. coli fatty acyl-CoA synthetase has been placed into the superfamily of AMP-binding proteins (27). This region contains a sequence that is similar to ATP-binding P-loops and may represent part of the ATP binding pocket within fatty acyl-CoA synthetase (37, 38). The second region that we noted having similarity to other adenylate-forming enzymes corresponds to amino acid residues 353-455 of the bacterial enzyme. Using BioSCAN, we identified a highly conserved 25-amino acid residue segment within this second region that is common to all fatty acyl-CoA synthetases. This segment within the bacterial enzyme, ⁴³¹NGWLHTGIAVMDEEGFLRIVDRKK⁴⁵⁵, contains amino acid residues that are 84% identical or highly conserved to other fatty acyl-CoA synthetases currently within the data base (Fig. 1). This finding has led us to propose that the 25-amino acid residue consensus sequence, DGWLHTGDIGXWXPXGXLKIIDRKK, is a signature motif common to the family of fatty acyl-CoA synthetases specifically (FACS signature motif). In addition to the fatty acyl-CoA synthetases, the more distantly related enzymes, 4-coumarate-CoA ligase (EC 6.2.1.12) and Photinus-luciferin 4-monooxygenase (EC 1.13.12.7; firefly luciferase), share similarities with this motif. It is noteworthy that the coumarate-CoA ligases and luciferases lack both the glycine at position 16 and the lysine at position 24 within the signature motif. In addition, both enzymes contain either a phenylalanine or a tyrosine at position 11 that is not conserved within the fatty acyl-CoA synthetases. These differences are addressed under "Discussion."

Using the FACS signature motif as a guide, a series of site-directed mutations within fadD were constructed to evaluate the contribution of this region to fatty acyl-CoA synthetase function. The different fadD mutations and the corresponding mutagenic oligonucleotides and plasmid constructs generated for this study are listed in Table I. Following in vitro mutagenesis of fadD⁺ and sequence confirmation of each of the desired mutations, BamHI fragments containing the respective fadD alleles were gel-purified and ligated into pACYC177 for further analysis. Consistent with our earlier work, we chose to analyze this collection of fadD mutations cloned into pACYC177 due to the low to moderate copy number of this plasmid (39). Initially, we monitored the growth of this collection of fadD mutant alleles in the fadD fadR strain LS6829 on minimal media plates containing acetate, decanoate, or oleate as sole carbon and energy sources. Positive and negative controls were strain LS6829 (fadD fadR) harboring the fadD⁺ plasmid pN300 or the plasmid vector pACYC177, respectively. All of these fadD mutants grew on acetate minimal plates within 24 h. Strain LS6928 harboring the fadD mutations N431A and H435A grew on oleate and decanoate minimal plates at rates equivalent to the wild type, while the remaining fadD site-directed mutants grew at a slower rate on oleate minimal plates when compared with the wild type, indicating that we had disrupted the normal enzymatic capacity or structural integrity of the enzyme. Within the latter group, we observed that the fadD mutations K454A and K455A had increased growth rates on decanoate when compared with the wild-type, suggesting that the specificity of the enzyme had been altered.

The altered growth characteristics of the fadD strain LS6829 harboring plasmids encoding the mutant fadD alleles on minimal media containing either oleate or decanoate suggested that specific amino acid residues within the FACS signature motif were required for the uptake and activation of fatty acids. To further test these observations, fatty acyl-CoA synthetase activities were determined on crude extracts using decanoate (C10:0), myristate (C14:0), and oleate (C18:1) as substrates in the fadD fadR strain LS6829 harboring the wild-type fadD gene (pN300), the collection of fadD alleles (see Table I), or the plasmid vector pACYC177. These data, presented in Fig. 2, demonstrated that the wild-type enzyme had fatty acyl-CoA synthetase activities using decanoate, myristate, or oleate as substrates that were consistent with previous observations (24). For this collection of fadD mutants, fatty acyl-CoA synthetase activities 1) were nearly wild-type, 2) were 30-50% of wild-type; 3) were less than 10% wild-type; 4) showed a preference for C18:1 substrates as opposed to C10:0 substrates; or 5) showed a preference for C10:0 substrates as opposed to C18:1 substrates.

Only N431A resulted in a fatty acyl-CoA synthetase activity profile that was comparable with that of the wild type. As noted in Fig. 1, the E. coli enzyme is the only fatty acyl-CoA synthetase within the data base that contains an asparagine residue at position 1; the remaining fatty acyl-CoA synthetases contain an aspartic acid. From the current data we predict that the carboxyl group of the aspartate does not contribute to the catalytic activity of this enzyme.

Four fadD alleles resulted in depressing fatty acyl-CoA synthetase to 30-50% of wild-type levels but had no change in fatty acid chain length specificity (fadDG432A, fadDH435A, fadDG437A, and fadDI450A). Included within this group are two of the highly conserved glycine residues within the FACS signature motif. In both of these cases, alanine substitutions were insufficient to completely disrupt activity. Three fadD alleles resulted in fatty acyl-CoA synthetase activities that were comparable with the fadD null (fadDW433A, fadDT436A, and fadDR453A). Three additional substitutions resulted in nearly abolishing both decanoyl-CoA and oleoyl-CoA synthetase activities but retaining detectable myristoyl-CoA synthetase activities (fadDL434A, fadDD438A, and fadDG446A). These data argue that either these residues are crucial for catalytic activity or these substitutions disrupted protein structure to the point that the mutant forms of the enzymes were either dysfunctional or nearly dysfunctional.

The alleles fadDI439A, fadDL448A, and fadDR449A resulted in lowering fatty acyl-CoA synthetase activity for all three substrates but had a change in specificity. These mutations resulted in a higher oleoyl-CoA synthetase activity relative to the decanoyl-CoA synthetase activity. Just downstream, the alleles fadDV451A, fadDD452A, fadDK454A, and fadDK455A had high decanoyl-CoA synthetase activities relative to oleoyl-CoA synthetase activities. As noted above, the fadDK454A and fadDK455A alleles resulted in an increased growth rate on decanoate when compared with the wild type. The finding that specific amino acid substitutions within the FACS signature motif change substrate specificity is consistent with the proposal that this region of the enzyme is directly involved in fatty acid binding.

On the basis of the fatty acyl-CoA synthetase activities presented above, we were interested in purifying several of these mutant forms of the enzyme and evaluating their enzymatic properties in vitro. Therefore, we initially added a tag of six histidine residues to the amino terminus of the E. coli fatty acyl-CoA synthetase so that the enzyme could be rapidly purified using a nickel chelate affinity chromatography following expression from a T7 RNA polymerase-responsive promoter. Using this construct, we were able to purify substantial quantities of His₆-tagged fatty acyl-CoA synthetase in a single-step purification protocol (Fig. 3A). The induction of the T7 RNA polymerase resulted in a 1800-fold increase in enzyme activity using [³H]oleate as a substrate. Using the purified His-tagged enzyme, we demonstrated that at different concentrations of enzyme, the production of fatty acyl-CoA was linear over a 30-min period. Furthermore, the apparent V_max and K_m using oleate as a substrate were estimated to be 292 nmol/min/mg protein and 2.2 µM, respectively. The K_m was slightly lower than that defined for the purified enzyme, while the apparent V_max was quite similar to that of the purified enzyme (24). Consistent with our data showing fatty acyl-CoA synthetase activities in the fadD fadR strain LS6829 harboring pN300 (fadD⁺), the purified enzyme had maximal specificity for myristate (C14:0), which is also in agreement with the purified native enzyme (24).

Expression plasmids were constructed that contained one of four specific mutations within fadD: W433A, D438A, D452A, and K455A. These four fadD alleles were chosen because 1) they all had reduced oleoyl-CoA synthetase activities and 2) D452A and K455A had high decanoyl-CoA synthetase activities. Following induction of these mutant forms of fatty acyl-CoA synthetase, they were purified over a Ni²⁺-NTA resin, and decanoyl-CoA and oleoyl-CoA synthetase activities were determined (Fig. 3B). As predicted, all four purified mutant forms of fatty acyl-CoA synthetase had no oleoyl-CoA synthetase activity (Fig. 3C). Both fadD452A and fadDK455A alleles had decanoyl-CoA synthetase activities that were nearly wild type, while the fadDW433A and fadDD438A alleles had no decanoyl-CoA synthetase activity. It is difficult to determine whether the fadDW433A and fadDD438A alleles result in an incorrectly folded protein or whether they are unable to bind one or both of the substrates tested for fatty acyl-CoA synthetase. In the case of the fadDD452A and fadDK455A alleles, these data support the proposal that specific amino acid residues within the FACS signature motif (including Asp⁴⁵² and Lys⁴⁵⁵) are involved in defining the fatty acid specificity of the enzyme.

To further evaluate the notion that this region of fatty acyl-CoA synthetase is involved in fatty acid binding, oleic and decanoic acid binding profiles were determined for purified wild-type enzyme as well as for the mutant enzyme forms containing the W433A and K455A substitutions. Fig. 4 shows the oleic acid binding profiles of these three enzymes using Lipidex. The wild-type enzyme was able to bind oleate, whereas the mutant forms only weakly bound (K455A) or did not bind at all (W433A). We attempted to evaluate binding using Lipidex but found that the medium chain fatty acid bound to the resin, making this approach unacceptable. The resin LH-20 was therefore employed to evaluate decanoic acid binding. In these experiments, a higher binding activity was noted for the mutant form of the enzyme containing the K455A substitution when compared with the wild type (Fig. 5). The mutant form of fatty acyl-CoA synthetase containing the W433A substitution had barely detectable binding. Collectively, these data agree with those noted above on activity profiles. Most notable in this regard was the finding that the K455A mutation resulted in a change in substrate specificity, providing additional evidence that this region of the enzyme is involved in fatty acid binding.

DISCUSSION

E. coli contains a single fatty acyl-CoA synthetase with broad chain length specificity for saturated, unsaturated, and polyunsaturated fatty acids (24, 25). This enzyme is essential for the activation of exogenous long chain fatty acids destined for -oxidation and plays an essential role in the regulation of the transcription factor FadR. We and others (26-28) have demonstrated that the E. coli fatty acyl-CoA synthetase shares considerable similarities with other fatty acyl-CoA synthetases and more broadly, the superfamily of adenylate-forming enzymes. The present study was undertaken to evaluate the functional role of a 103-amino acid residue segment within the E. coli fatty acyl-CoA synthetase that is common to other fatty acyl-CoA synthetases and several members of the AMP-binding protein superfamily. Using BioSCAN, we have identified a stretch of 25 amino acid residues within this 103-amino acid residue segment that appears to be restricted to the family of fatty acyl-CoA synthetases. We propose that the consensus sequence, DGWLHTGDIGXWXPXGXLKIIDRKK, is common to all fatty acyl-CoA synthetases and represents a FACS signature motif.

There are a number of features within the FACS signature motif that are notable. 1) This region contains two invariant glycine residues (at positions 2 and 7) and a highly conserved (11/12) glycine at position 16. Therefore, it is reasonable to predict that this region in all fatty acyl-CoA synthetases adopts a similar tertiary structure. 2) This region contains an additional six residues that are invariant in the family of fatty acyl-CoA synthetases: Trp at position 3, Thr at position 6, Asp at position 8, Asp at position 22, Arg at position 23, and Lys at position 25. 3) The consensus sequence predicts an aspartic acid residue at position 1; however, in the bacterial enzyme this is an asparagine, and conversion of the asparagine to alanine has no effect on enzyme activity, indicating that the presence of the carboxylate is not crucial for activity. 4) The residue in the fourth position is hydrophobic and is a leucine (6/12), a methionine (1/12), or phenylalanine (5/12). 5) This region of the enzyme contains hydrophobic residues (leucine, isoleucine, or valine) at positions 4, 9, 18, 20, and 21. We predict that these residues, in addition to tryptophan residues at position 3, may compose part of a fatty acid binding pocket. 6) There is a preference for basic residues at positions 19 and 24 in addition to those at positions 22 and 25.

Eighteen site-directed mutations within the fatty acyl-CoA synthetase structural gene (fadD) corresponding to the FACS signature were constructed to evaluate the contribution of the conserved amino acids of the enzyme to catalytic activity. On the basis of our mutagenesis data, the FACS signature motif contains specific amino acids that are essential for catalytic activity and appear to specify a fatty acid binding site within the enzyme. Three distinct classes of mutations were identified on the basis of growth characteristics, fatty acyl-CoA synthetase profiles using oleate and decanoate as substrates, and studies using purified wild-type and mutant forms of the enzyme: 1) only one substitution (N431A) resulted in wild-type fatty acyl-CoA synthetase activity profiles; 2) 10 substitutions abolished or greatly diminished enzyme activity; and 3) 7 substitutions resulted in altered fatty acid chain length specificity. Of these seven, five also had reduced activity compared with the wild-type enzyme.

Alanine substitution of Gly⁴³², Trp⁴³³, Leu⁴³⁴, His⁴³⁵, Thr⁴³⁶, Gly⁴³⁷, Asp⁴³⁸, Gly⁴⁴⁶, Ile⁴⁵⁰, and Arg⁴⁵³ lowered or eliminated fatty acyl-CoA synthetase activities using all three fatty acid substrates. Substitution of alanine for Ile⁴³⁹, Leu⁴⁴⁸, Arg⁴⁴⁹, Val⁴⁵¹, Asp⁴⁵², Lys⁴⁵⁴, and Lys⁴⁵⁵ appeared to alter fatty acid chain length specificity. Four mutant enzymes (V451A, D452A, K454A, and K455A) had a preference for C10:0 substrates as opposed to C18:1, while I439A, L448A, and R449A resulted in a preference for C18:1 as a fatty acid substrate as opposed to C10:0. This was unexpected, since Suzuki et al. (41) had proposed that the corresponding region within the rat liver fatty acyl-CoA synthetase is involved in ATP binding. In direct fatty acid binding assays, we were able to show that purified protein carrying K455A had an increased binding affinity for C10 and reduced binding affinity for C18:1. Therefore, we suggest that this region of fatty acyl-CoA synthetase is specifically required for fatty acid binding. This interpretation is consistent with the observation that this FACS signature motif is restricted to the adenylate-forming enzymes that also use fatty acids as substrates.

Assuming that this FACS signature motif defines an essential part of the fatty acid binding domain of the fatty acyl-CoA synthetase, the essential and conserved hydrophobic amino acids including Trp⁴³³ and Leu⁴³⁴ may compose part of a fatty acid binding pocket. The potential roles of the essential charged residues are less clear, especially since substitutions of Arg⁴⁴⁹, Asp⁴⁵², Lys⁴⁵⁴, and Lys⁴⁵⁵ confer differential fatty acid chain length specificity. Perhaps these charged residues maintain a fatty acid binding pocket that is altered in the alanine substitutions that results in a "better fit" for medium chain fatty acids as opposed to long chain fatty acids. We noted that both the Pseudomonas oleovarans fatty acyl-CoA synthetase and the yeast Faa2p have specificity for medium chain fatty acids, and both lack the lysine residue at the position equivalent to Lys⁴⁵⁴ of the E. coli enzyme. In these two cases there are valine and alanine residues, respectively (42, 43). This strengthens our hypothesis that Lys⁴⁵⁴ in the E. coli enzyme contributes to chain length specificity, since substitution of Lys⁴⁵⁴ results in increased activity toward C10. The aspartic acid at position 438 and the arginine at 453 also appear to be crucial for function. While these residues may be part of the fatty acid binding domain, they might alternatively interact with the carboxylate of the fatty acid and thus be crucial for the formation of the adenylate intermediate. We cannot conclusively distinguish these possibilities at the present time.

The analyses of the His-tagged mutant enzyme forms confirmed the data generated in whole cell extracts. Both the fadDD452A and fadDK455A mutations had high decanoyl-CoA synthetase activity relative to oleoyl-CoA synthetase activity. In addition, both fadDW433A and fadDD438A had no fatty acyl-CoA synthetase activity, arguing that these Trp⁴³³ and Asp⁴³⁸ are essential for function. At this point, we cannot discern the precise step in the catalytic cycle that is disrupted in these two mutants. The fatty acid binding profiles obtained further support the conclusion that this region of fatty acyl-CoA synthetase is involved in fatty acid binding.

As mentioned under "Results," both the coumarate CoA ligases and firefly luciferases contain regions that share amino acid similarities to the FACS signature motif. The substrates for these enzymes, coumarate and firefly luciferin, are hydrophobic cyclic compounds that form adenylated intermediates as part of the reaction mechanism. In this regard, coumarate CoA ligase and firefly luciferase have catalytic mechanisms similar to the fatty acyl-CoA synthetases. These two enzymes deviate from the FACS signature in that each contains an aspartate or a lysine in lieu of a glycine at position 16 and a leucine instead of a lysine at position 24. Additionally, these two enzymes contain either a phenylalanine or a tyrosine at position 11 of the motif. The glycine residue at position 16 is present in all members of the fatty acyl-CoA synthetase family with the exception of the presumed fatty acyl-CoA synthetase from Hemophilus influenzae. Perhaps this glycine contributes to protein structure by maintaining a fatty acid binding pocket that cannot accommodate the cyclic coumarate or luciferin. As noted above, two fatty acyl-CoA synthetases with medium chain specificity have an aliphatic amino acid in place of a basic amino acid at position 24. Both the coumarate CoA ligases and luciferases contain a leucine at position 24, suggesting that these two enzymes are more similar to those fatty acyl-CoA synthetases with a medium chain length specificity. Coumarate CoA ligase and luciferase also contain either a phenylalanine or a tyrosine at position 11. Of the fatty acyl-CoA synthetases, only Faa2p from yeast contains a phenylalanine at this position. Faa2p has a preference for medium chain substrates. Thus, if the presumption is correct that the FACS signature motif is involved in the recognition of fatty acid substrates, small changes within this region are expected to have considerable changes on substrate recognition. This may result in preferences for hydrophobic cyclic substrates as opposed to long or medium chain alkyl substrates.

A number of enzymes that form adenylated intermediates were not detected using the FACS signature motif as a query sequence against the nonredundant data base. We and others have noted that there exists a second region of homology within the family of fatty acyl-CoA synthetases that also includes most adenylate-forming enzymes. Most notable in this regard is a region of the E. coli enzyme from amino acid residues 200-273. This region contains a signature that is presumed to specify AMP binding (26-28, 49). From the data described in the present work, it is tempting to speculate that adenylate-forming enzymes are of common ancestry. The family of the fatty acyl-CoA synthetases have maintained that AMP binding signature while evolving their own specific properties to include a presumed fatty acid binding domain that is typified by the FACS signature motif.