Substitution of Alanine for Serine 250 in the Murine Fatty Acid Transport Protein Inhibits Long Chain Fatty Acid Transport*

The murine fatty acid transport protein (FATP) was identified on the basis of its ability to facilitate uptake of long chain fatty acids (LCFAs) when expressed in mammalian cells. To delineate FATP domains important for transport function, we cloned the human heart FATP ortholog. Comparison of the human, murine, and yeast amino acid sequences identified a highly conserved motif, IYTSGTTGXPK, also found in a number of proteins that form adenylated intermediates. We demonstrate that depletion of intracellular ATP dramatically reduces FATP-mediated LCFA uptake. Furthermore, wild-type FATP specifically binds [α-32P]azido-ATP. Introduction of a serine to alanine substitution (S250A) in the IYTSGTTGXPK motif produces an appropriately expressed and metabolized mutant FATP that demonstrates diminished LCFA transport function and decreased [α-32P]azido-ATP binding. These results are consistent with a mechanism of action for FATP involving ATP binding that is dependent on serine 250 of the IYTSGTTGXPK motif.

The murine fatty acid transport protein (FATP) was identified on the basis of its ability to facilitate uptake of long chain fatty acids (LCFAs) when expressed in mammalian cells. To delineate FATP domains important for transport function, we cloned the human heart FATP ortholog. Comparison of the human, murine, and yeast amino acid sequences identified a highly conserved motif, IYTSGTTGXPK, also found in a number of proteins that form adenylated intermediates. We demonstrate that depletion of intracellular ATP dramatically reduces FATP-mediated LCFA uptake. Furthermore, wildtype FATP specifically binds [␣-32 P]azido-ATP. Introduction of a serine to alanine substitution (S250A) in the IYTSGTTGXPK motif produces an appropriately expressed and metabolized mutant FATP that demonstrates diminished LCFA transport function and decreased [␣-32 P]azido-ATP binding. These results are consistent with a mechanism of action for FATP involving ATP binding that is dependent on serine 250 of the IYTSGTTGXPK motif.
The precise mechanism of long chain fatty acid (LCFA) 1 transport is not well understood. In mammalian cells such as myocytes and adipocytes, LCFA uptake is efficient and highly regulated. Experiments demonstrating specific, saturable LCFA uptake that is inhibited by prior protease treatment of the cell surface suggested that LCFAs are transported by a protein-mediated mechanism (1)(2)(3)(4)(5). CD36 and mitochondrial aspartate aminotransferase were initially proposed to facilitate LCFA transport because they are capable of binding LCFAs (6 -10). More recently, we identified the fatty acid transport protein (FATP) on the basis of its function in LCFA uptake. We isolated the cDNA encoding the murine FATP using an expression cloning strategy to screen a 3T3-L1 adipocyte cDNA library for cDNAs that increase LCFA uptake when expressed in mammalian cells (11). FATP may function as an LCFA transporter that facilitates bi-directional LCFA movement across the plasma membrane (12,13).
FATP is a 63-kDa integral plasma membrane protein. Stable overexpression of FATP confers a 4-fold increase in initial rates of LCFA uptake with a K m of 0.2 M for oleic acid, comparable to the K m for oleic acid uptake by 3T3-L1 adipocytes (11). FATP facilitates uptake of saturated and mono-enoic LCFAs with 14 -22 carbons, suggesting that it has broad specificity with respect to fatty acid chain length and degree of saturation (11). 2 FATP expression in cultured adipocytes is inhibited by insulin (13). In mice, FATP expression is induced during fasting in adipose and heart tissue, suggesting that FATP may be important not only for uptake of LCFAs into tissues with a metabolic requirement for this substrate, but also for efflux of LCFAs from adipocytes during lipolysis. Disruption of the gene encoding the Saccharomyces cerevisiae (yeast) ortholog, Fat1p, results in a 2-3-fold decrease in the rate of oleate uptake and impaired growth of yeast in which de novo fatty acid synthesis is inhibited (14).
The mechanism of action of FATP is unknown. Hydropathy analysis of the amino acid sequence of FATP predicts a protein with six transmembrane domains. The primary sequence and predicted number of membrane-spanning domains for FATP are distinct from other transporters; therefore, analogies to other transport mechanisms may not be appropriate. Unlike models in which classical transporters promote movement of their substrates through a polypeptide pore in the plasma membrane, FATP might function as a flippase, facilitating transport of its amphipathic LCFA substrate in direct contact with plasma membrane phospholipids.
In the present study, we sought to identify evolutionarily conserved residues in the primary sequence of FATP that may have functional importance. We cloned and sequenced FATP from a human heart cDNA library. Comparison with the murine and yeast orthologs delineated several conserved regions that may be essential for FATP function. We focused on the 11-amino acid motif IYTSGTTGXPK (in which X is any amino acid) in a hydrophilic segment of the protein that is predicted to be intracellular. This motif is present in a number of proteins that form adenylated intermediates and may be involved in an interaction with ATP. In the present study, we show that FATP function is dependent on cellular ATP levels, that FATP binds azido-ATP, and that mutation of serine 250 to alanine (S250A) in the IYTSGTTGXPK motif impairs LCFA transport function and nucleotide binding.

EXPERIMENTAL PROCEDURES
cDNA Cloning-The 1.8-kb BsaAI-XbaI restriction fragment from the coding region of murine FATP was used to screen an oligo(dT) and random-primed normal adult human heart phage cDNA library (Stratagene). Standard molecular biology techniques were used for plaque lifts. Filters were hybridized with random-primed 32 P-labeled * 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 (15). Sequence comparisons were made using DNASTAR software. Cells-NIH 3T3 cells (ATCC) were grown in Dulbecco's modified Eagle's medium (DMEM) with 10% calf serum supplemented with 2 mM L-glutamine, 50 units/ml penicillin G sodium, and 50 units/ml streptomycin sulfate. 293GPG cells (16) were grown in DMEM with 10% inactivated fetal bovine serum supplemented with 2 mM L-glutamine, 50 units/ml penicillin G sodium, 50 units/ml streptomycin sulfate, 0.3 mg/ml G418, 1 g/ml puromycin, and 1 g/ml tetracycline. Transient transfections of 293GPG cells and harvest of viral supernatants were performed as described (16).
Stable Cell Lines-Murine FATP was subcloned into the ⌬U3 retroviral expression vector (⌬U3FATP) and transiently transfected into 293GPG packaging cells to produce high titer VSV-G pseudotyped retrovirus as described previously (16). 10 5 NIH 3T3 cells were infected in a 35-mm well by two successive 7-h exposures to 1 ml of the ⌬U3FATP retrovirus. Retrovirally transduced populations of NIH 3T3 cells were plated at limiting dilution to allow isolation of independent clonal cell lines. Individual cell lines were screened for FATP expression by Western blot analysis of microsomal proteins. Similar methods were employed to generate cell lines that overexpress the S250A mutant.
Protein Preparations-For total post-nuclear membrane (TM) preparations, confluent monolayers of cells were scraped and homogenized in 255 mM sucrose, 20 mM Tris, pH 7.4, 1 mM EDTA with 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 g/ml pepstatin A, and 1 g/ml leupeptin (SHB) using a glass-Teflon tissue grinder. Nuclei were removed by centrifugation at 1000 ϫ g for 10 min. The remaining cellular membranes and their associated proteins were pelleted by centrifugation at 356,000 ϫ g for 30 min. For plasma membrane (PM) isolation, whole cell homogenates were initially centrifuged at 16,000 ϫ g for 20 min to pellet nuclei, mitochondria, and plasma membranes. This pellet was resuspended in SHB and layered on a 1.12 M sucrose cushion. After centrifugation at 99,000 ϫ g for 20 min, plasma membranes were isolated at the interface and pelleted at 48,000 ϫ g for 45 min.
Western Blot Analysis-TMs and PMs were resuspended in 1% Triton X-100, 50 mM Tris, pH 7.4, 2 mM EDTA, 150 mM NaCl, 1 mM PMSF, 1 g/ml pepstatin, and 1 g/ml leupeptin, and protein was quantified by BCA assay (Pierce). Equivalent amounts of protein from various cell lines were separated by 7.5% or 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membranes (0.2-m pore). Ponceau-S staining was used to confirm even transfer. For Western analysis of FATP or mutant expression, we used a rabbit polyclonal antiserum directed against murine FATP residues 455-470 at a dilution of 1:2000. For detection of the G protein ␤ subunit, we used a rabbit polyclonal antiserum (SC-378, Santa Cruz Biotechnologies) at a dilution of 1:1000. For detection of the GLUT1 glucose transporter, we used a rabbit polyclonal antiserum at a dilution of 1:500 (gift of M. Mueckler). For detection of the medium chain acyl-CoA dehydrogenase enzyme (MCAD), we used a rabbit polyclonal antiserum at a dilution of 1:4000 (gift of A. Strauss). For detection of the 70-kDa peroxisomal membrane protein (PMP70), we used a rabbit polyclonal antiserum at a dilution of 1:200 (gift of D. Vallee). A horseradish peroxidasecoupled secondary donkey anti-rabbit antibody and chemiluminescence were used for detection. An LKB laser densitometer was used for quantitation.
Measurement of Cellular ATP-After treatment with metabolic inhibitors, cellular ATP was measured in 10 3 cells using a bioluminescent assay for ATP (Boehringer Mannheim). Absolute concentrations were determined using an ATP standard.
Site-directed Mutagenesis of FATP-An alanine substitution was introduced at position 250 using an overlapping PCR strategy with Pfu polymerase (Stratagene). Primers A (5Ј-GCC GGT GTG GTG GCT GC-3Ј) and D (5Ј-CGC TGC TCC ACG TCG CG-3Ј) were chosen to flank the site of mutagenesis on the coding and non-coding strands, respectively. The following internal primers were chosen for the S250A mutation: B (5Ј-GGT CCC CGC GGT ATA GAT G-3Ј) and C (5Ј-CAT CTA TAC CGC GGG GAC C-3Ј). Initial 0.3-kb PCR products were generated with the following primer pairs: A and C, B and D. The two initial PCR products were used as template in a second PCR reaction with primers A and D to generate a 0.6-kb PCR product that incorporated the S250A mutation. This product was digested with BglII and ApaLI to generate a 167-base pair fragment that was subcloned into ⌬U3FATP. All PCRderived sequences were confirmed by double-stranded dideoxy chain termination sequencing to verify introduction of the desired mutation and fidelity of amplification.
Azido-ATP Labeling-Cross-linking experiments were performed using PM preparations from the stable cell lines described above and [␣-32 P]azido-ATP. This photoaffinity probe contains an azido group substitution at position 8 in the base ring that is inert until photoactivated by short wave ultraviolet light (254 nm) to form a highly reactive nitrene. A covalent linkage is then formed between the nitrene and neighboring polypeptides. Since the nitrene group formed with irradiation is located at the purine ring, it is expected to react with adenine binding sites in proteins. This photoaffinity probe has been used to identify sequences lining the nucleotide binding sites of several proteins (17,18).
PMs were resuspended in MES buffer (20 mM MES, pH 6.75, 150 mM NaCl, 1 mM PMSF, 1 g/ml pepstatin A, and 1 g/ml leupeptin), and proteins levels quantified by BCA. 25 g of protein was incubated with [␣-32 P]azido-ATP on ice and exposed to ultraviolet light (254 nm) for 5 min. ATP␥S was included as a competitor. The buffer was adjusted to a final 1% Triton X-100, 50 mM Tris, pH 7.4, 2 mM EDTA, 150 mM NaCl, 1 mM PMSF, 1 g/ml pepstatin, and 1 g/ml leupeptin. Wild-type and mutant FATP molecules were immunoprecipitated as described (19) using a rabbit polyclonal anti-peptide antiserum directed against FATP residues 628 -640 at 1:100 dilution. In control immunoprecipitations, we included excess immunizing peptide against which the antiserum was raised. Immunoprecipitates were separated by 7.5% SDS-PAGE, and dried gels were submitted to autoradiography and analysis using densitometry. For calculation of the amount of azido-ATP bound per g plasma membrane protein in the initial labeling reaction, the relevant band was cut from the gel and radioactivity counted.

RESULTS
Cloning of hFATP from a Human Heart cDNA Library-We used a 1.8-kb fragment of the murine FATP coding sequence to screen an adult human heart cDNA library under moderate stringency. We identified and sequenced several overlapping cDNA clones that contained the human heart FATP sequence (hFATP). hFATP contained a 1932 base pair open reading frame (data not shown) and encoded a 643-amino acid protein (Fig. 1). The hFATP sequence shows 61% identity with the murine FATP sequence (11) and 34% identity with the yeast FATP ortholog (14,20). Sequence similarity is highest in regions defined by murine residues 247-257, 316 -323, 335-348, 363-371, 383-391, and 503-525. Evolutionary conservation of these regions of primary sequence is consistent with an essential role of such residues in FATP function. The amino terminus is the most divergent.
The 11-amino acid sequence between murine residues 247 and 257 is completely conserved in the human, murine, yeast, and rat (21) FATP. Similar sequences are found in a number of proteins, including luciferase, long chain fatty acyl-CoA synthetase, several ligases, and enzymes that catalyze the nonribosomal synthesis of cyclic peptide antibiotics (gramicidin S synthetase 2, aminoadipyl-cysteinyl-valine synthetase), suggesting that the IYTSGTTGXPK motif has an important function ( Fig. 2A). A common feature among these proteins is their ability to activate the carboxyl group of their respective substrates by transfer of adenosine from ATP to the substrate to form an adenylated intermediate, liberating pyrophosphate. The adenylated-substrate intermediate is then hydrolyzed with release of AMP and formation of an ester or amide derivative of the substrate. Hydropathy analysis of the amino acid sequence of FATP predicts there are six membrane-associated domains (Fig. 2B), and our prior immunofluorescence studies suggest the carboxyl terminus of FATP is intracellular (11). Based on this model, we predict the IYTSGTTGXPK motif lies on an intracellular loop of the protein.
Effects of FCCP and Cyanide on FATP-mediated LCFA Uptake-Because the IYTSGTTGXPK motif is present in proteins known or thought to form adenylated intermediates, this motif may facilitate an interaction with ATP. We postulated that FATP-mediated LCFA uptake may depend on cellular ATP levels. To test this hypothesis, we generated stable cell lines overexpressing murine FATP at high levels in NIH 3T3 fibroblasts that have low basal expression of FATP and low level LCFA uptake and metabolism (11). The murine FATP cDNA was subcloned into the ⌬U3 retroviral expression vector (16). High titer retrovirus was generated by transient transfection of the construct into the 293GPG packaging cell line, and used to transduce NIH 3T3 cells. Clonal cell lines with stable integration of the FATP-encoding provirus were isolated by plating cells at limiting dilution. We screened for cell lines with high level expression of FATP by Western blot analysis of proteins from individual cell lines. Western analysis data from a representative pair of cell lines is shown in Fig. 3A. An anti-peptide antiserum directed against FATP residues 455-470 specifically recognizes the 63-kDa FATP in the stable FATP-overexpressing cell line, whereas parental NIH 3T3 cells have low level FATP expression (not appreciated on the exposure shown). Moreover, compared with parental NIH 3T3 cells, LCFA uptake was increased 9-fold in the FATP-overexpressing cell line, as measured using a 1-min incubation with the fluorescent LCFA analog BODIPY3823 at 37°C, followed by flow cytometric analysis (Fig. 3, B and C). This increase in LCFA uptake is greater than that observed with previously published cell lines (11), likely due to higher levels of expression in the present cell lines (data not shown).
To examine whether FATP-mediated LCFA uptake is dependent on cellular ATP levels, we assayed LCFA uptake after cells were treated with metabolic inhibitors. Parental and FATP-overexpressing cells were incubated for 15 (cyanide) or 30 (carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP)) min with increasing concentrations of inhibitor, washed and tested as above for LCFA uptake. For flow cytometric analysis, 1 M propidium iodide was included in the media in order to gate out non-viable cells. LCFA uptake specifically attributable to FATP overexpression was calculated at each concentration of inhibitor by subtracting the LCFA uptake of parental cells from that of FATP-overexpressing cells. The inhibitors shown deplete cellular ATP by independent mechanisms. Cyanide inhibits re-oxidation of cytochrome a3 in the mitochondrial electron transport chain, whereas FCCP is a proton ionophore that dissipates proton gradients and uncou- ples ATP synthesis and mitochondrial electron transport.
Both inhibitors depleted cellular ATP in a dose-dependent manner, and this was correlated with a commensurate decrease in LCFA uptake in the FATP-overexpressing cells (Fig.  3, D and E). There was no significant effect of the inhibitors on the low levels of uptake by the parental NIH 3T3 cells (data not shown). In FATP-overexpressing cells, there was a significant reduction in cellular ATP concentrations at inhibitor concentrations that demonstrated an effect on LCFA transport. With FCCP we observed 50% inhibition at a concentration of 1 M (at ATP level 79% of control), and with cyanide we observed 50% inhibition at 1 mM (at ATP level 80% of control). These results suggest that LCFA uptake specifically attributable to FATP expression (the difference between FATP-overexpressing and parental cell lines) is impaired in the face of cellular ATP depletion. These experiments cannot distinguish whether this effect on FATP function is due to a direct interaction of FATP with ATP (i.e. primary active transport) or to the indirect effects of ATP depletion on the general maintenance of coupled electrochemical gradients (i.e. secondary active transport in which ATP is used to establish a gradient of a driving cotransported solute or symported species).
The Effect of S250A Mutation on LCFA Uptake-To evaluate the role of the IYTSGTTGXPK motif in FATP function, PCR was used to substitute alanine for serine at position 250 (S250A) of this motif. The mutant cDNA was cloned into the ⌬U3 retroviral expression vector. Multiple stable cell lines overexpressing the mutant FATP were isolated and screened by Western analysis (data not shown), and a representative cell line is shown in all subsequent experiments. Pulse-chase studies were used to evaluate the expression and metabolism of S250A as compared with wild-type FATP (Fig. 4). We observed comparable levels of expression of S250A and wild-type FATP molecules, with half-lives of 10 and 8 h for S250A and FATP, respectively. Thus, S250A is expressed and metabolized in a fashion indistinguishable from wild-type FATP.
To determine whether S250A was correctly targeted to the plasma membrane, plasma membrane fractions of S250A-and wild-type FATP-overexpressing cell lines were isolated by differential density centrifugation and the use of a discontinuous sucrose gradient. For Western blot analysis of plasma membranes from S250A and FATP cells, we used a polyclonal rabbit antiserum specifically directed against FATP residues 455-470 that are identical in the mutant and wild-type proteins (Fig. 5). Total cellular membranes are shown as a positive control for proteins relatively depleted from the plasma membrane. Western blots using antibodies to the G protein ␤ subunit (G␤, plasma membrane), the GLUT1 glucose transporter (unglycosylated and core glycosylated forms of GLUT1 in the endoplasmic reticulum (ER)/Golgi and fully glycosylated form in the plasma membrane and Golgi), MCAD (mitochondria), and the 70-kDa peroxisomal membrane protein (PMP70) demonstrate that the plasma membrane fractions were enriched in plasma membrane proteins and relatively depleted of ER, Golgi, mitochondrial and peroxisomal proteins. This analysis ensured that the S250A mutant was appropriately targeted to the plasma membrane. The level of expression of S250A is comparable to wild-type FATP. The conservative S250A substitution is thus unlikely to have grossly altered FATP structure or targeting.
The ability of mutant and wild-type FATP molecules to promote LCFA uptake was examined using the fluorescent assay described above. Compared with wild-type FATP, overexpression of S250A resulted in a 78% reduction in LCFA uptake (Fig.  6). These results show that S250A is minimally active in LCFA transport and that serine 250 is critical for FATP function. Our control studies show that this difference does not result from lower levels of expression of the S250A protein at the plasma membrane compared with wild-type FATP.
Azido-ATP Binding by Wild-type FATP and S250A-Given the conservation of the IYTSGTTGXPK motif and the evidence that a mutation of this sequence impairs FATP function, we examined the ability of wild-type FATP to bind the nucleotide analog [␣-32 P]8-azido-ATP. Equivalent amounts of plasma membrane protein isolated from different cell lines were incubated with [␣-32 P]8-azido-ATP, cross-linked by exposure to FIG. 3. Effects of metabolic inhibitors on FATP-mediated LCFA uptake. A, 20 g of protein from parental NIH 3T3 cells and stable NIH 3T3 cell lines that overexpress FATP were separated by SDS-PAGE (7.5% gel) and analyzed by Western blotting using either a rabbit polyclonal anti-peptide antiserum raised against FATP residues 455-470 or preimmune serum (1:2000 dilution). Bands were detected using a horseradish peroxidase-coupled anti-rabbit IgG and chemiluminescence. For controls, specific immunizing peptide or nonspecific peptide were included at 50 g/ml in the primary antibody incubations. B, cell lines described in A were tested for uptake of BODIPY3823 as described under "Experimental Procedures." Cells were resuspended in growth media with 1 M propidium iodide and analyzed by flow cytometry. Histograms display fluorescence on a log scale (FL-1) versus cell number (counts) for 10 5 cells. C, bar graph displays the median fluorescence of each cell line shown in B. Similar results were obtained in five independent experiments. D and E, NIH 3T3 cells and stable FATP-overexpressing cell lines were trypsinized, washed and incubated at 37°C with increasing concentrations of the metabolic inhibitors FCCP (D) or potassium cyanide (E) in PBS with 0.1% albumin for 30 or 15 min, respectively. Cells were washed with serum-free media, and BODIPY3823 uptake was measured. Median fluorescence of 10 5 propidium-iodide negative (live) cells was determined. Specific FATP-mediated LCFA uptake was calculated as the difference between parental NIH 3T3 cells and FATP-expressing cells and is plotted (q) over a range of inhibitor concentrations as a percentage of the uptake in the absence of inhibitor. At each concentration of inhibitor, intracellular ATP concentration was determined and plotted (f) as a percentage of ATP concentration in the absence of inhibitor. Similar results were obtained in three independent experiments. short wave ultraviolet light, and immunoprecipitated with a polyclonal anti-peptide antisera directed against FATP residues 628 -640, which are identical in the wild-type and mutant proteins (Fig. 7A). Immunoprecipitable FATP was labeled by [␣-32 P]8-azido-ATP in plasma membranes derived from FATPoverexpressing cells (lane 2), but not when immunoprecipitation was carried out in the presence of excess immunizing peptide (lane 3) or when plasma membrane proteins were derived from NIH 3T3 cells (lane 1). Experiments shown in Fig. 7 (B and C) were performed to assess the specificity of this labeling. Labeling of a fixed amount of plasma membrane protein with increasing concentrations of [␣-32 P]8-azido-ATP demonstrated saturation of label incorporation with half-maximal incorporation into 25 g of plasma membrane protein at 130 M [␣-32 P]8-azido-ATP (Fig. 7B). Approximately 3 fmol of [␣-32 P]8azido-ATP was incorporated per g of plasma membrane protein. Azido-ATP incorporation was effectively competed with 100-fold molar excess of ATP␥S, a non-hydrolyzable ATP analog (Fig. 7C). The saturability of labeling and the inhibition of labeling by a non-hydrolyzable ATP analog are features consistent with a specific interaction between FATP and ATP.
In three independent experiments, we compared the ability of [␣-32 P]8-azido-ATP to label plasma membranes from wildtype-and S250A-expressing cells. Compared with equivalent amounts of wild-type FATP, labeling of S250A was reduced by 61% (Fig. 8). Serine 250 thus appears to play an important role in nucleotide binding by the IYTSGTTGXPK motif. This decrease in labeling of S250A is consistent with the observed decrease in LCFA transport function by this mutant. DISCUSSION FATP is highly conserved among widely divergent species, consistent with an important role of this protein in the physiology of LCFA metabolism in eukaryotic cells. The IYTSGTT-GXPK motif common to all FATP sequences is homologous to domains in other proteins postulated or known to form adenylated intermediates. This observation suggested that FATP may facilitate uptake of LCFAs via an ATP-dependent mechanism. In our stable cell lines, FATP-mediated LCFA uptake is diminished significantly in the face of cellular ATP depletion, suggesting that ATP is essential for FATP function. Mutation of the central serine in the IYTSGTTGXPK motif dramatically reduces FATP function. Our [␣-32 P]8-azido-ATP cross-linking studies suggest that FATP interacts directly with ATP, and that serine 250 is important for this interaction. This residue is critical for LCFA transport function likely due to a role in ATP binding.
Prior studies have examined the effect of uncoupling agents (2, 4-dinitrophenol, sodium arsenate), proton ionophores (carbonyl cyanide m-chlorophenylhydrazone), and electron transport inhibitors (cyanide, azide) on LCFA uptake by mammalian cells. In studies by Abumrad and co-workers, 1 M dinitrophenol (2) and 25 mM sodium arsenate (22) did not affect basal oleate transport in adipocytes. However, insulin-mediated antagonism of epinephrine stimulation of LCFA transport was blocked by 1 mM dinitrophenol (23). In contrast, studies by Stremmel and co-workers (5) demonstrated inhibition of oleate uptake into isolated rat hepatocytes by 100 M carbonyl cyanide m-chlorophenylhydrazone, 4 mM 2,4 dinitrophenol, and 1 mM potassium cyanide. Our findings of significant inhibition of LCFA uptake with decreases in intracellular ATP induced by

FIG. 4. Metabolism of wild-type and mutant FATP molecules.
A, FATP-and S250A-overexpressing cell lines were labeled for 1 h with [ 35 S]cysteine-methionine and chased for 0 -17 h in media without radioactivity. Wild-type and mutant FATP were immunoprecipitated using a rabbit polyclonal anti-peptide antiserum directed against FATP residues 628 -640. In control immunoprecipitations, excess immunizing peptide was included at a concentration of 100 g/ml. Proteins were by separated by SDS-PAGE and analyzed by autoradiography. B, 35 S incorporation in A was quantified using an LKB laser densitometer. The bar graph displays wild-type or mutant FATP at each chase time point as a percentage of the amount of protein observed at 0 h of chase.

FIG. 5. Western analysis of wild-type and mutant FATP.
TMs and PMs were isolated as described under "Experimental Procedures" from NIH 3T3 and FATP-and S250A-overexpressing cell lines. For each sample, 12 g of protein was separated by SDS-PAGE and analyzed by Western blot. A polyclonal rabbit antiserum directed against FATP residues 455-470 was used to detect wild-type and mutant FATP. Control blots include G-protein ␤ subunit (PM), GLUT1 glucose transporter (unglycosylated and core glycosylated form in ER, glycosylated form in PM and Golgi), MCAD (mitochondrial), and the 70-kDa peroxisomal membrane protein (PMP70). Detection for each blot was performed with a horseradish peroxidase-coupled anti-rabbit antiserum and chemiluminescence. metabolic inhibitors are most consistent with the latter studies. These experiments, together with our observation that FATP specifically binds [␣-32 P]8-azido-ATP, are consistent with a model in which FATP facilitates ATP-dependent LCFA transport.
The IYTSGTTGXPK motif has been proposed to play a role in ATP binding by a number of proteins (24 -28). Many proteins with this motif catalyze formation of an amide or ester linkage to the carboxyl group of their substrates via adenylated intermediates. The IYTSGTTGXPK motif has been proposed to form a phosphate binding loop in which lysine binds the ␥-phosphate of the nucleoside triphosphate (25). The motif does have numerous amino acid residues with hydroxyl groups on their side chains that may participate in hydrogen bonding with distant residues to form a nucleotide binding fold. In luciferase, for which the crystal structure is known (29), the IYTSGTTGXPK motif lies in a loop connecting antiparallel strands 6 and 7 of ␤-sheet A, and the side chain of S198 is predicted to form a hydrogen bond with another residue located across the putative nucleotide binding cleft.
In all proteins containing the IYTSGTTGXPK motif, Ser-250 is invariant. In the present study, we chose to mutate FATP serine 250 in such a way as to remove a potential hydrogen bond donor residue that may be important in creating a nucleotide binding fold. Our studies provide the first demonstration of a mutation in any putative mammalian fatty acid transporter which impairs transport function and the first demonstration in any protein that mutation of the central serine in the IYTSGTTGXPK motif specifically impairs nucleotide binding. The configuration of amino acid residues surrounding the adenine ring in the ATP-bound state of FATP is likely to be significantly different from that found in the peptide synthetases, since peptides from tyrocidine synthetase I that are crosslinked to the photoaffinity probe 2-azido-ATP (G373-K384, W405-R416, and L483-K494) are not found in FATP (30). Furthermore, tyrocidine synthetase can only be labeled by 2-azidopurine derivatives and not by the 8-azidopurine derivative used in the present study to label FATP.
Other mutations in the IYTSGTTGXPK motif have been studied in tyrocidine synthetase and luciferase. Amino acid substitutions G180A, G183A, and P185V in the tyrocidine synthetase sequence do not affect the enzyme's phenylalanine-dependent ATP-PP i exchange activity; however, substitution of Arg or Tyr for Lys-186 resulted in marked decrease in enzyme activity (27). In luciferase, alanine substitutions at positions 198, 201, 202, and 203 resulted in enzymes with 10 -100% activity, whereas S199A (analogous to the FATP S250A mutation) and G200A mutations resulted in complete loss of enzyme activity (26). Luciferase mutation S198T produced a partially functional enzyme with an abnormal pH optima (shifted from 8.1 to 8.7), a decreased rate of bioluminescence decay, and an increase in avidity for binding Mg 2ϩ -ATP. These findings are consistent with a role for the IYTSGTTGXPK motif in ATP binding.
Although implicated in ATP binding in a number of proteins, the IYTSGTTGXPK motif differs in several important ways from classical ATP binding motifs. While it is glycine-rich and contains a positively charged lysine residue at its 3Ј end, the motif has little homology to previously identified nucleotidebinding consensus sequences, (G/A)X 4 (G/A)K(S/T) and (R/ K)X 1-4 GX 2-4 ⌽X⌽ 2 (D/E) (31,32), in which X is any amino acid and ⌽ is a hydrophobic amino acid. In peptide synthetases, but not in FATP, the IYTSGTTGXPK motif is associated with a TGD motif, which is implicated in Mg 2ϩ -ATP binding in ATPases (27,33,34).
Some, but not all, proteins containing the IYTSGTTGXPK motif have CoA synthetase activity. Notable exceptions include the peroxisomal peripheral membrane protein Pcs60p (35) and luciferase. Despite 40% sequence identity between rat FATP (21) and rat very long-chain acyl-CoA synthetase (36), FATP lacks the 25-amino acid signature motif, DGWLHTGDIGXWX-PXGXLKIIDRKK, previously identified in fatty acyl-CoA synthetases (37) and lacks the QVKIXGXRIEXXE motif found in many proteins that form adenylated intermediates (33,34). Our prior studies suggest that FATP facilitates cellular uptake of LCFAs without net enzymatic modification of the substrate (11). Moreover, membrane fractions from our FATP expressing cell lines do not have long chain acyl-CoA synthetase activity. 2 In addition, during starvation when lipolysis and net efflux of LCFAs from adipocytes are stimulated, expression of FATP increases (13). It is unlikely that esterification would increase in this physiologic circumstance.
Several potential mechanisms for FATP action involving ATP binding can be envisioned. FATP may be an active transporter that uses energy from ATP hydrolysis to drive LCFA transport across the plasma membrane. The energy from ATP hydrolysis may be used to effect a conformational change in FATP that facilitates flipping of its LCFA substrate from the exoplasmic to the cytosolic leaflet of the plasma membrane. Other potential mechanisms of FATP action involving ATP binding include covalent attachment of the LCFA substrate to FATP by way of an initial adenylated LCFA intermediate. The adenylate group might then be hydrolyzed, creating an LCFA-FATP covalent intermediate (ester linkage) and releasing AMP. This FATP-LCFA complex would need to be hydrolyzed once the LCFA was transported to the cytosolic face of the plasma membrane. Alternatively, ATP may bind as an allosteric modifier of FATP's function without significant hydrolysis, or ATP might be required in a regulatory phosphorylation cycle for FATP. There is, however, no evidence to date that FATP undergoes post-translational modification (data not shown).
It is unknown whether FATP mediates ATP hydrolysis. FATP-mediated transport is not inhibited by well characterized inhibitors of P-type (100 M vanadate (Ref. 38 In addition, we detect no difference in ATPase activity in plasma membranes from NIH 3T3 cells and from cells overexpressing FATP (data not shown). Use of permeabilizing agents to introduce [␣-32 P]8-azido-ATP or non-hydrolyzable ATP analogs into our fibroblast cell lines would render difficult subsequent measurements of LCFA transport to determine whether photo-labeling or non-hydrolyzable substrates inhibit FATP function. Definitive assessment of potential ATPase activity will require purification of FATP in its native conformation. Since FATP is a multiple membranespanning protein, such experiments are beyond the scope of the present study.
Our studies show that the serine residue in the IYTSGTT-GXPK motif, which is present in other ATP-dependent proteins, is critical for FATP function and nucleotide binding. Analysis of nucleotide binding sites in proteins for which crystal structures have been established indicate that at least two and as many as five non-contiguous peptide regions may be required to form a ligand-binding pocket. The spacing of such segments can be variable. Identification of additional residues involved in FATP's nucleotide binding domain through the isolation of photo-labeled polypeptides will be greatly facilitated by purification of FATP. Furthermore, the effects of additional mutations in the IYTSGTTGXPK motif will expand our understanding of the sequence requirements for this motif in ATP binding. These studies will provide further clues regarding FATP's structure and function.  2 and 3) was incubated at 4°C with 40 M [␣-32 P]8-azido-ATP and exposed to ultraviolet light at 254 nm for 5 min. FATP was immunoprecipitated using a polyclonal rabbit antiserum directed against FATP residues 628 -640. Control immunoprecipitation was carried out in the presence of 100 g/ml immunizing peptide (lane 3). After SDS-PAGE, incorporation of 32 P was detected by autoradiography. B, 25 g of plasma membrane protein from FATPoverexpressing cells was incubated with increasing concentrations of [␣-32 P]8-azido-ATP and exposed to ultraviolet light at 254 nm for 5 min. FATP was immunoprecipitated and analyzed by SDS-PAGE and autoradiography as described above. C, 25 g of plasma membrane protein from FATP-overexpressing cells was incubated with 40 M [␣-32 P]8azido-ATP and increasing concentrations of ATP␥S and was exposed to ultraviolet light at 254 nm for 5 min. FATP was immunoprecipitated and analyzed by SDS-PAGE and autoradiography as described above.
FIG . 8. [␣-32 P]8-Azido-ATP photo-incorporation into S250A mutant. 25 g of plasma membrane protein from FATP-or S250Aoverexpressing cell lines was incubated with 17 M [␣-32 P]8-azido-ATP and exposed to ultraviolet light at 254 nm for 5 min. Wild-type and mutant FATP were immunoprecipitated using a polyclonal rabbit antiserum directed against FATP residues 628 -640. After SDS-PAGE, gels were dried and incorporation of 32 P was detected by autoradiography. 32 P incorporation was quantified using an LKB laser densitometer. The bar graph displays 32 P incorporation in arbitrary units as a function of cell line. Error bars reflect standard deviation for assays performed in triplicate. These results were confirmed in three independent experiments.