Affinity labeling fatty acyl-CoA synthetase with 9-p-azidophenoxy nonanoic acid and the identification of the fatty acid-binding site.

Fatty acyl-CoA synthetase (FACS, fatty acid:CoA ligase, AMP-forming, EC ) catalyzes the esterification of fatty acids to CoA thioesters for further metabolism and is hypothesized to play a pivotal role in the coupled transport and activation of exogenous long-chain fatty acids in Escherichia coli. Previous work on the bacterial enzyme identified a highly conserved region (FACS signature motif) common to long- and medium-chain acyl-CoA synthetases, which appears to contribute to the fatty acid binding pocket. In an effort to further define the fatty acid-binding domain within this enzyme, we employed the affinity labeled long-chain fatty acid [(3)H]9-p-azidophenoxy nonanoic acid (APNA) to specifically modify the E. coli FACS. [(3)H]APNA labeling of the purified enzyme was saturable and specific for long-chain fatty acids as shown by the inhibition of modification with increasing concentrations of palmitate. The site of APNA modification was identified by digestion of [(3)H]APNA cross-linked FACS with trypsin and separation and purification of the resultant peptides using reverse phase high performance liquid chromatography. One specific (3)H-labeled peptide, T33, was identified and following purification subjected to NH(2)-terminal sequence analysis. This approach yielded the peptide sequence PDATDEIIK, which corresponded to residues 422 to 430 of FACS. This peptide is immediately adjacent to the region of the enzyme that contains the FACS signature motif (residues 431-455). This work represents the first direct identification of the carboxyl-containing substrate-binding domain within the adenylate-forming family of enzymes. The structural model for the E. coli FACS predicts this motif lies within a cleft separating two distinct domains of the enzyme and is adjacent to a region that contains the AMP/ATP signature motif, which together are likely to represent the catalytic core of the enzyme.

Over the past 15 years, there has been increasing interest in the roles played by bioactive lipids in cellular homeostasis, including fatty acids and fatty acid derivatives. It is now widely accepted that these compounds influence a wide variety of cellular processes including phospholipid synthesis (1,2), pro-tein export (3), protein modification (4 -6), enzyme activation or deactivation (7)(8)(9)(10), cell signaling (11,12), membrane permeability (13), membrane fusion (14), and transcriptional control (15)(16)(17)(18)(19). Our interest in bioactive lipids stems from investigations directed at elucidating the mechanisms governing how exogenous long-chain fatty acids traverse the membrane, become activated to CoA thioesters and influence downstream events using both bacterial and yeast model systems (20,21). It is recognized that fatty acids traverse the cell envelope of Escherichia coli by a protein-mediated process involving both FadL (an outer membrane-bound long-chain fatty acid transport protein) and fatty acyl-CoA synthetase (FACS 1 ; fatty acid: CoA ligase, AMP forming; EC 6.2.1.3). The formation of the fatty acyl-CoA thioester by FACS renders this process unidirectional and thus the role this enzyme plays in transport has been described as vectorial esterification (22). Similarly in yeast at least three fatty acyl-CoA synthetases have been implicated as components of the fatty acid import apparatus (23,24). 2 Fatty acyl-CoA synthetase plays a central role in intermediary metabolism by catalyzing the formation of fatty acyl-CoA by a two-step process proceeding through an adenylated intermediate. In E. coli, the specificity of this enzyme is primarily directed toward long-chain fatty acids (C14:0-C18:1), but it is clear this enzyme can also activate medium-chain fatty acids (C8:0-C12:0) (25). In previous work (26,27), we characterized the FACS structural gene, fadD, and using alanine-scanning mutagenesis, identified a region of the enzyme hypothesized to contribute to the fatty acid-binding site (FACS signature motif). The FACS signature motif is common to both eukaryotic and prokaryotic fatty acyl-CoA synthetases indicating that it is important to the structure and/or catalytic efficiency of these enzymes. This sequence motif is present in acyl-CoA synthetases with medium-to long-chain specificity and not found in the very long-chain acyl-CoA synthetases or acetyl-CoA synthetases suggesting it is fairly restrictive. Our studies identified specific mutations within fadD corresponding to the FACS signature motif that resulted in altering the chain length specificity of the enzyme. More specifically, we were able to generate altered forms of FACS that preferred medium-chain fatty acids as substrates as opposed to long-chain fatty acids (27). In this regard, we hypothesized this region of FACS is involved in fatty acid binding and provides a molecular ruler to specify chain length (27).
The E. coli FACS is a central component of the long-chain fatty acid transport apparatus. Once transported across the cell envelope and activated to CoA thioesters, long-chain fatty acids are primarily degraded via ␤-oxidation but can be incorporated into phospholipids. In addition, exogenously derived long-chain fatty acyl-CoA acts as potent bioactive lipids by specifically regulating the DNA binding activity of the transcription factor FadR (15,18). When the intracellular levels of long-chain fatty acyl-CoA are low, FadR is DNA bound and acts both to repress the transcription of genes involved in fatty acid transport, activation, and ␤-oxidation (fad) and activate the transcription of at least two genes required for fatty acid biosynthesis (fab) (18,28). FadR specifically binds long-chain fatty acyl-CoA as their intracellular level increase, which results in a loss of DNA binding activity. The net result is the derepression of the fad genes and a decrease in the expression of the fab genes. These observations support the notion that FadR responds to the intracellular pools of long-chain fatty acyl-CoA to coordinately regulate the differential expression of genes involved in fatty acid biosynthesis and degradation.
In the present work, we have employed a direct biochemical approach using affinity labeling to probe the fatty acid-binding domain within purified E. coli FACS. These experimental methods identified a peptide containing the affinity labeled fatty acid that begins at Pro 422 and contains the FACS signature motif supporting the proposal that this region of the enzyme represents the fatty acid-binding domain. These data, in combination with computer modeling to predict the structure of FACS and our past work using alanine-scanning mutagenesis, are in agreement with the hypothesis that the fatty acid and ATP-binding sites within the enzyme are localized within a cleft separating two distinct domains of the enzyme.

EXPERIMENTAL PROCEDURES
His 6 -FACS Overexpression and Purification-Strain BL21 (DE3)(pLysS) was transformed with plasmid pN3576 encoding His 6 -FACS. The expression and purification of His 6 -FACS has been previously described (27). Briefly, cells (600-ml cultures) were induced with isopropyl-1-thio-␤-D-galactopyranoside (0.5 mM) for 90 min and enzyme purified from a clarified cell extract using Ni 2ϩ -NTA-agarose (Qiagen) as described by Black et al. (27). The His-tagged enzyme elutes as a single peak between 150 and 200 mM imidazole. The purity of samples was monitored using SDS-polyacrylamide gel electrophoresis and reverse phase HPLC. This method routinely results in 3-5 mg of purified enzyme per 600 ml of culture.
Circular Dichroism Spectroscopy-The circular dichroism spectra of purified FACS was determined using a Jasco J720 spectropolarimeter and analyzed using SELCON (29). Protein was dissolved in 10 mM potassium phosphate, pH 7.5, to a final concentration of 18.5 M. The protein sample was scanned from 260 to 178 nm using a light path of 0.005 cm at 25°C. Far UV data were recorded at 0.2-nm increments at a scan speed of 20 nm/min, 1.0-nm band width, and an averaging time of 1 s. A baseline was taken using buffer only and was subtracted from each protein spectrum. To improve the signal to noise ratio, a Savisky-Golay filter was used and four scans in the far UV region were averaged.
Measurement of Fatty Acyl-CoA Synthetase Activity-FACS activity was monitored using a modification of a protocol defined by Kameda and Nunn (25). Briefly, purified FACS was added to a reaction mixtures containing 200 mM Tris HCl, pH 7.5, 50 M ATP, 8 mM MgCl 2 , 2 mM EDTA, 20 mM NaF, 0.1% Triton X-100, varying concentrations of [ 3 H]oleate, 0.5 mM coenzyme A and incubated 10 min at 37°C. Reactions were initiated by the addition of CoA and terminated by the addition of isopropyl, n-heptane, 1 M H 2 SO 4 (40:10:1). The aqueous phase, containing oleoyl-CoA formed during the reaction, was extracted three times with 2.5 ml of n-heptane and subjected to scintillation counting. All data presented represent the mean (Ϯ S.E. of the mean) for three independent experiments.
Synthesis and Preparation of 9-p-Azidophenoxy [ , and 3000 pmol) in a total volume of 100 l in a 96-well microtiter plate. Following incubation, samples were illuminated with UV light (256 nm, 1 cm height) for 60 s to cross-link the [ 3 H]APNA to the enzyme. Samples were subsequently resolved on 12% SDS-polyacrylamide gels (30). The gels were stained with Comassie Blue, soaked in En 3 Hance (PerkinElmer Life Sciences), dried, and subjected to fluorography for 5-7 days at Ϫ80°C. For the experiments showing specificity of labeling, 400-pmol aliquots of FACS (using the buffering conditions defined above) were incubated with 1500 pmol of [ 3 H]APNA and increasing concentrations of lignocerate (C24:0), palmitate (C16:0), decanoate (C10:0), hexanoate (C6:0), or acetate (C2:0) (0, 250, 500, 750, 1000, 1250, 1500, 2500, and 5000 pmol) in the dark. Samples were illuminated with UV light to generate the crosslinks and resolved on SDS gels and prepared as described above. For preparative work, the reactions were scaled up 10 -20-fold maintaining a FACS:[ 3 H]APNA molar ratio of 1:3 as under these conditions, there was maximal labeling of the enzyme. We estimated the efficiency of labeling purified FACS with APNA to be between 25 and 30%.

Purification of the [ 3 H]APNA-labeled Peptide and Protein
Sequencing-Following UV illumination and confirmation that the enzyme was labeled with [ 3 H]APNA, the samples were digested with trypsin (100 g/ml final concentration) for 18 h at 37°C. The controls contained enzyme or [ 3 H]APNA alone. The resultant peptides were resolved on a Nucleosil column (250 ϫ 4.6 mm, C18, Phenominex) and developed with a linear gradient of acetonitrile with 0.1% trifluoroacetic acid (0 -100% in 60 min). Peptides were detected at 230 nm using a Beckman System Gold High Pressure Liquid Chromatography System. One-milliliter fractions were collected and aliquots assayed for radioactivity. Peptides containing a radioactive signal were identified and repurified by a second round of HPLC. Following purification, peptides were subjected to NH 2 -terminal peptide sequence analysis. Peptide sequencing was done at the Amino Acid Analysis and Peptide Sequencing Core Facility of the Wadsworth Center of the New York State Department of Public Health using an Applied Biosystems 477A Protein Sequencer.
Structural Predictions and Amino Acid Sequence Comparisons-The predicted structure for the E. coli FACS, generated using SWISS-MODEL and visualized using Swiss-Pdb Viewer has been previously described (15). Figures of the predicted structure of the E. coli FACS were generated using MolScript v2.1.2 on a Silicon Graphics workstation. Protein sequence comparisons were performed using BioSCAN (31) or MultAlign v5.3.3 (32), with sequences that are conserved among the fatty acyl-CoA synthetases or members of the adenylate forming superfamily. When using BioSCAN these comparisons were directed against the SWIS-PROT data base using a 0.01 probability threshold with complexity filtering. The score table employed was blosum62.
Other-[ 3 H]Oleic acid was purchased from PerkinElmer Life Sciences. Trypsin (ultrapure) was obtained from Sigma. 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
Purification and Kinetic Properties of FACS-Our previous work identified a conserved 25-amino acid residue segment within FACS, which we hypothesized is part of the fatty acidbinding domain within the enzyme (27). In order to test this hypothesis, the enzyme containing an amino-terminal hexameric histidine tag (His 6 -FACS) was purified using Ni 2ϩ chelation chromatography and kinetically characterized in the presence of 0.1% Triton X-100 prior to photolabeling experiments using an affinity labeled long-chain fatty acid (Fig. 1). The enzyme was not active in the absence of added nonionic detergent or phospholipid indicating that catalysis is likely to occur at the membrane. 3 The His-tagged enzyme displayed classical Michaelis-Menten kinetics at oleate concentrations between 0 and 5 M, which allowed apparent kinetic parameters to be defined. The apparent V max and the K m for purified His 6 -FACS were 77.7 nmol/min/mg protein and 1.1 M oleate, respectively. These values were consistent with published values for the purified native enzyme from E. coli (25).
Circular Dichroism Spectra of Purified FACS-We have recently proposed a three-dimensional model for FACS developed using SWISS-MODEL and visualized using Swiss-Pdb Viewer (15). This prediction suggests FACS adopts a three-dimensional architecture similar to that defined for firefly luciferase and phenylalanine activating subunit (PheA) of gramicidin synthetase 1 (33,34). The crystal structures for these adenylate-forming enzymes have been solved and demonstrate that both contain a large NH 2 -terminal domain and small COOHterminal domain separated flexible linker that results in a cleft. Specific amino acid residues exposed within this cleft as well as those in the flexible linker are hypothesized to represent the catalytic core of FACS. For firefly luciferase, the NH 2terminal domain consists of a ␤-barrel and several ␤-strands. These ␤-strands are flanked by ␣ helices to form an ␣Ϫ␤Ϫ␣ 5-layered structure. (33). The far UV circular dichroism spectrum of purified FACS was defined in order to assess the ␣ and ␤ secondary structure. The far UV (240 -178 nm) CD spectra for the purified enzyme was determined as detailed under "Experimental Procedures" and analyzed using SELCON (Fig.  2). These data indicated this enzyme contained 47% ␣, 17% ␤, 13% ␤ turn, and 23% other structure. The relative amounts of ␣ and ␤ structure were consistent with predicted the threedimensional model for the E. coli FACS (15) and similar to those for firefly luciferase and phenylalanine activating subunit (PheA) of gramicidin synthetase 1 (33,34).
Labeling  4A). As illustrated in Fig. 4B, a 50% reduction of [ 3 H]APNA labeling was achieved using 1200 -1400 pmol of palmitate. These data indicated that labeling of the enzyme by [ 3 H]APNA was, 1) saturable and 2) specific on the basis of competition by palmitate. The specificity of APNA labeling was further investigated by incubating the purified enzyme (400 pmol) with increasing concentrations of fatty acids with differing chain lengths (C24:0, C10:0, C6:0, and C2:0) (Fig. 5). Palmitate was the most effective fatty acid in competing out [ 3 H]APNA binding followed by decanoate and hexanoate. Lignocerate and acetate were unable compete for [ 3 H]APNA binding. These data argue that the site within the enzyme modified by [ 3 H]APNA is specific for long-chain fatty acids, but contains some specificity toward medium-chain fatty acids. These results are consistent with earlier kinetic studies of the enzyme (25). These findings demonstrated that [ 3 H]APNA could be used as an effective affinity label to probe that fatty acid-binding domain of FACS.

Identification of a [ 3 H]APNA-labeled Tryptic
Peptide of FACS-In order to define the region of FACS that had become modified, the experiments were scaled up 10 -20-fold and following UV illumination, the [ 3 H]APNA-labeled FACS was proteolyzed with trypsin as detailed under "Experimental Procedures." As noted above, the specific labeling of FACS with APNA was estimated to be between 25 and 30%. As a control, FACS alone was proteolyzed under identical conditions. The proteolyzed samples were resolved on a Nucleosil column and developed using a linear acetonitrile gradient (0 -100% acetonitrile containing 0.1% trifluoroacetic acid). Peptides were detected at 230 nm and 1.0-ml fractions collected. Aliquots from 3 J. D. Weimar and P. N. Black, unpublished observations. There was consistent labeling of T33 with [ 3 H]APNA in four independent experiments. [ 3 H]APNA was resolved in the absence of protein to 1) define the retention characteristics of this compound, 2) assess purity prior to photolabeling, and 3) define if there were fragmentation products. As shown in Fig. 6, APNA alone had an average retention time between 15 and 17 min on the Nucleosil column. The [ 3 H]APNA-labeled peptide (T33) was isolated and purified using reverse-phase HPLC and sequenced using automated Edman degradation. The NH 2terminal sequence of this peptide, PDATDEIIK, corresponds to residues 422-430 in FACS and is adjacent to the region of the enzyme that contains the FACS signature motif (residues 431-455) (Fig. 7). The APNA-labeled peptide, T33, is therefore likely to contain part or the entire region of the enzyme that includes the FACS signature motif. These data fully support our previous observations using alanine-scanning mutagenesis of the fadD gene that this region of the enzyme is involved in fatty acid binding (27).
Interpretation of the Fatty Acid-binding Site within the Predicted Structure for FACS-The E. coli FACS is a member of the AMP-binding protein superfamily and contains signature sequences predicted to specify ATP binding (ATP/AMP signature motif) (36 -39). Using the crystallographic information for two enzymes containing the ATP/AMP signature motif (firefly luciferase (33) and the phenylalanine activating subunit (PheA) of gramicidin synthetase 1 (34)), we proposed a threedimensional model for the E. coli FACS (Fig. 8A (15)). This model predicted the region we identified as the FACS signature (hypothesized to specify fatty acid binding) formed a ␤, ␤-turn-␤ structure, which was on the same face of the enzyme as elements that comprise the ATP/AMP signature motif (presumed to specify in ATP binding; boxed in Fig. 8A). The identification of an APNA-labeled peptide, beginning with Pro 422 , adjacent to and contiguous with the FACS signature (involved in fatty acid binding) confirmed our hypothesis regarding the fatty acid-binding domain making this model quite compelling. On the basis of the predicted structure of this enzyme, the region bound by ␤-1 and ␤-2 of the FACS signature motif (Fig.  7) contributes to a cavity that we predict is the fatty acidbinding site. Fig. 8B illustrates the ␣-carbon trace of the FACS signature motif (Asn 431 -Lys 455 ) in the same orientation as shown in Fig. 8A, which emphasizes the ␤, ␤-turn-␤ structure.
On the basis of this information, it seems likely that the region of FACS, which includes the cleft separating the two domains of the enzyme, represents the catalytic core of the enzyme. We suspect that upon ligand binding (ATP and fatty acid), the two domains close to facilitate the formation of the fatty acyl adenylate.

DISCUSSION
Fatty acyl-CoA synthetase represents a pivotal enzyme in lipid metabolism. The goal of our work is to define the biochemical mechanisms that underpin the role of FACS in the transport of exogenous fatty acids. We hypothesize fatty acid transport proceeds across the cell membrane through the vectorial esterification of exogenous fatty acids. In E. coli, the esterification of exogenous long-chain fatty acids with coenzyme A is tightly coupled to a regulatory network linking fatty acid import to the differential expression of genes involved in fatty acid degradation and biosynthesis (15,20,21). There is emerging evidence that in a number of eukaryotic systems, the fatty acid activation-coupled transport also occurs, indicating this is a . ␣ denotes the predicted ␣ helix upstream from the FACS signature motif; ␤-1, ␤-2, and ␤-3 denote the ␤, ␤-turn-␤ structure of the FACS signature; and L denotes the linker between the large NH 2 -terminal and small COOH-terminal domains of the enzyme. The boxed region denotes the ␤-loop-␤ structure that comprises the first sequence element of the ATP/AMP signature motif while the residue identified with an asterisk is Glu 361 (within the second sequence element of the ATP/AMP signature motif), which is conserved in all adenylate-forming enzymes. B, ␣-carbon tracing of the FACS signature motif (Asn 431 -Lys 455 ) highlighting the ␤, ␤-turn-␤ structure that is proposed to contribute to the fatty acid binding pocket within the enzyme. common mechanism promoting the movement of exogenous fatty acids across the membrane in a highly regulated manner (24, 39 -40). In the present report, we provide experimental evidence defining the regions within FACS, which are essential for activity and estimate structural characteristics of the protein.
The present work represents the first direct biochemical investigation into the identity and location of the fatty acidbinding domain within FACS. We have demonstrated the affinity labeled long-chain fatty acid, APNA, precisely modifies a region of the enzyme, which includes the FACS signature motif. This work confirms our hypothesis, based on alanine-scanning mutagenesis of the FACS structural gene (fadD), that this region of the enzyme contributes to the fatty acid-binding site (27). From these data and previous analyses of fadD alleles with single amino acid substitutions, a number of conclusions can be drawn regarding the ligand-binding sites within the enzyme. First, the fatty acid-binding site, which includes the FACS signature motif, is predicted to be localized at the carboxyl end of the larger domain of the enzyme and extends into the linker separating the large NH 2 -terminal domain from the smaller COOH-terminal domain (corresponding to amino acid residues 431-455). This signature motif is predicted to be composed of a ␤ strand followed by two anti-parallel ␤ strands connected by a turn, which within the three-dimensional structure of the enzyme may contribute to a cavity that accommodates the acyl chain of the fatty acid. Second, the region of the enzyme containing the two sequence elements that comprise the ATP/AMP signature motif (hypothesized to be involved in ATP binding) are in juxtaposition with those that comprise the FACS signature motif. The first element of the ATP/AMP signature motif involved in ATP binding (residues 213-222) was found within two antiparallel ␤-strands connected by a disordered loop. The information gleaned from the firefly luciferase structure suggests this disordered loop contributes to a region of the enzyme that is flexible (33). And third, within the second sequence element of the ATP/AMP signature motif (residues 356 -361) is an invariant glutamate (residue 361 in FACS), which is found in all adenylate-forming enzymes and is essential for maintaining catalytic proficiency. 4 The predicted structure of FACS places this residue in a turn that is in the same proximity as the two antiparallel ␤-strands that comprise the first sequence element of the ATP/AMP signature motif. Both elements of the ATP/AMP signature motif may act to anchor ATP, or alternatively, may interact with and participate in the expulsion of the pyrophosphate leaving group (34). Collectively, these findings support the proposal that the fatty acid-and ATP-binding sites are adjacent to each other within the predicted three-dimensional structure of the enzyme. We suggest this region of the enzyme, indeed comprises a common structural fingerprint for adenylate-forming enzymes and that the specificity for the second carboxyl-containing substrate is dictated by specific residues corresponding to the region that in the fatty acyl-CoA synthetases contains the FACS signature motif (27).
On the basis of the predicted model for the E. coli FACS, the fatty acid-binding site extends into the cleft where we presume the acyl adenylate is formed. We have previously suggested that subtle changes in the region of the enzyme corresponding to the FACS signature motif results in marked changes with respect to substrate recognition (27). Both the Pseudomonas oleovarans fatty acyl-CoA synthetase (42) and the Saccharomyces cerevisiae fatty acyl-CoA synthetase, Faa2p (43), have specificity for medium-chain fatty acids and lack the lysyl residue at the position equivalent to Lys 454 of the E. coli enzyme. In these two cases there is an aliphatic residue at this position (Ala and Val, respectively). Substitution of Lys 454 within the E. coli enzyme with Ala results in a mutant form that has specificity toward medium-chain fatty acid substrates (27). These data support the notion that this residue contributes to a region of the enzyme that acts as a molecular ruler and specifies the length of the fatty acid substrate.
The role of FACS in fatty acid transport is somewhat controversial, but there is recent evidence supporting the capacity of this enzyme to function in a coupled import-activation mechanism. The seminal observations made by Overath and co-workers (22) demonstrated the E. coli FACS was localized in both cytosolic and particulate fractions within the cell. In fact, these early observations proposed that fatty acid import into the cell was driven by a vectorial esterification mechanism. More recent studies in both yeast and mammalian cells indicate that a similar process may also be operational. In yeast a fatty acyl-CoA synthetase (Faa1p or Faa4p) is required for fatty acid import and activation of exogenous long-chain fatty acids (23). An additional protein within yeast, Fat1p, is involved in fatty acid import and has an acyl-CoA synthetase activity, suggesting it may work in conjunction with either Faa1p or Faa4p to promote coupled fatty acid import activation (24,44). In mammalian cells, the emerging evidence suggests that fatty acyl-CoA synthetase is, likewise, involved in fatty acid import (39 -41). The challenges that we are presently faced with not only include detailed mechanistic and structural investigations on the FACS family of enzymes, but also to more precisely determine how these enzymes function in fatty acid import. Our laboratory is presently investigating these questions.