Substrate Recognition by the Human Fatty-acid Synthase

Abstract

The human fatty-acid synthase (HFAS) is a potential target for anti-tumor drug discovery. As a prelude to the design of compounds that target the enoyl reductase (ER) component of HFAS, the recognition of NADPH and exogenous substrates by the ER active site has been investigated. Previous studies demonstrate that modification of Lys-1699 by pyridoxal 5′-phosphate results in a specific decrease in ER activity. For the overall HFAS reaction, the K1699A and K1699Q mutations reduced k_cat and k_cat/K_NADPH by 8- and 600- fold, respectively (where K_NADPH indicates the K_m value for NADPH). Thus, Lys-1699 contributes 4 kcal/mol to stabilization of the rate-limiting transition state following NADPH binding, while also stabilizing the most stable ground state after NADPH binding by 3 kcal/mol. A similar effect of the mutations on the ER partial reaction was observed, in agreement with the proposal that Lys-1699 is located in the ER NADPH-binding site. Most unexpectedly, however, both k_cat and k_cat/K_NADPH for the β-ketoacyl reductase (BKR) reaction were also impacted by the Lys-1699 mutations, raising the possibility that the ER and BKR activities share a single active site. However, based on previous data indicating that the two reductase activities utilize distinct cofactor binding sites, mutagenesis of Lys-1699 is hypothesized to modulate BKR activity via allosteric effects between the ER and BKR NADPH sites.

Fatty acid synthesis generates important intermediates for the construction of cell membranes and for energy storage (1). In humans the reactions resulting in the production of fatty acids are catalyzed by a multifunctional enzyme complex (HFAS),³ consisting of the following seven catalytic activities: acetyl/malonyl transacylase, β-ketoacyl synthase, β -ketoacyl reductase (BKR), β -hydroxyacyl dehydratase (DH), enoyl reductase (ER), and thioesterase (2–10). In addition, there is an acyl carrier protein (ACP) domain, which carries fatty acyl intermediates in the form of fatty acyl thioesters (5). Palmitate, a C₁₆-saturated fatty acid, is the major product produced (5, 11). Fatty acid synthesis, although important evolutionarily for survival during famines (1), was thought to have very little significance for today’s modern society (1, 9, 10, 12). This is because of the dietary intake of lipids, which causes HFAS to be down-regulated in normal cells (6, 7, 13, 14). However, HFAS overexpression in many cancers, such as in breast, prostate, and colon tumors, and also in many pre-malignant growths, has led to the hypothesis that HFAS is an important target for the study of tumor biology (7, 13, 14).

Our laboratory has a long standing interest in the development of antibacterial compounds that target prokaryotic fatty acid synthesis. Thus, an initial interest in HFAS was as a control system for validating the specificity of our bacterial FAS inhibitors. However, as described above, HFAS is now thought to be a bona fide target in its own right as a putative anti-tumor target. There have also been suggestions that HFAS inhibitors might provide some therapeutic benefit in the control of obesity (15–17). Initial reports on the anti-tumor activity of HFAS inhibitors focused on compounds that target the β-ketoacyl synthase activity, such as cerulenin and C75 (14). More recently, Anderson and co-workers (18) have reported that triclosan, an antibacterial additive in many personal care products, inhibited the enoyl reductase activity of HFAS with an IC₅₀ value of 50 µm and also prevented the growth of a breast cancer cell line at a similar concentration. In order to provide information relevant to the design of compounds that target the ER component of HFAS, we have performed a detailed kinetic analysis of the overall and partial reactions catalyzed by HFAS. These studies have been extended to include mutants of Lys-1699, a residue that is modified by pyridoxal 5′-phosphate (PLP). Lys-1699 is part of a characteristic SXXK motif found in PLP-dependent enzymes (19). Although mammalian FAS does not require PLP, the site-specific incorporation of PLP was used to show that Lys-1699 is located close to the ER active site in the goose and chicken FAS enzymes (2, 19–21). In order to explore the role of Lys-1699 in more detail, we used site-directed mutagenesis to alter Lys-1699, and we carried out detailed kinetic experiments to determine the effect of these mutations on the overall and partial reactions catalyzed by HFAS. For the overall HFAS reaction, these studies reveal that Lys-1699 stabilizes the rate-limiting transition state following NADPH binding by 4 kcal/mol and also stabilizes the ground state to which NADPH binds by 3 kcal/mol. Lys-1699 also makes similar contributions to ground and transition state stabilization in the ER reaction, as expected from the PLP labeling studies. However, unexpectedly the BKR reaction is also affected by mutation of Lys-1699. By analogy to the short chain dehydrogenase/reductase (SDR) family, a likely mechanism by which Lys-1699 acts is via direct interactions with NADPH in the ER cofactor binding site. A similar role can be envisaged for Lys-1699 in the BKR reaction, if both BKR and ER activities are catalyzed by a single active site. Alternatively, if the BKR and ER enzymes utilize separate cofactor binding sites, then modulation of the BKR reaction through interaction of NADPH with Lys-1699 must occur through allosteric effects between the two binding sites.

In the course of kinetic experiments to determine k_cat and K_m values for the overall HFAS reaction as well as each individual component, we noted that k_cat for the reduction of crotonyl-CoA was lower than the value obtained for the overall reaction. In order to explore the role of the intrinsic ACP moiety in the interaction of exogenously added substrates with HFAS, we mutated Ser-2151, the point of pantetheine attachment, to HFAS (22, 23), and we investigated the chain length specificity of the enzyme toward exogenously added enoyl reductase substrates.

EXPERIMENTAL PROCEDURES

Materials

A pFastBac1 clone containing the cDNA coding for the entire HFAS protein with a C-terminal His tag was a generous gift from GlaxoSmithKline. Oligonucleotides were from Integrated DNA Technologies (Coralville, IA). Pfu polymerase, the BAC-to-BAC baculovirus expression system, Escherichia coli maximum efficiency DH10 Bac competent cells, and Spodoptera frugiperda (Sf9) insect cells were obtained from Invitrogen. pBluescript and E. coli XL-1 Blue competent cells were from Stratagene (La Jolla, CA), and the AatII, MluI, HindIII, EcoRI, and BamHI restriction enzymes were from New England Biolabs (Beverly, MA). ABI Prism Big Dye Terminator Cycle Sequencing was carried out on an MJ Research PTC-100 Programmable Thermal Controller from Global Medical Instrumentation, Inc. (Ramsey, MN). The DNA Sequencing Facility at Stony Brook University provided the ABI BigDye reaction mix. The NADPH, CoA lithium salt, ethyl chloroformate, triethylamine, 2-octenoic acid (predominantly trans), and crotonyl-CoA were purchased from Sigma, and the His-Bind resin was from Novagen (Madison, WI). Amersham Biosciences provided the Sephadex G-25 (fine) resin and the chromatography columns, and the 10% Tris-glycine gels were purchased from Bio-Rad. Dialysis membrane, all buffer reagents, and solvents were purchased from Fisher. 2-trans-Dodecenoic acid was purchased from TCI Chemicals.

Construction of the K1699A, K1699Q, and S2151A Mutants

Mutagenesis on the original HFAS insert and vector was problematic because of the large size of this construct (13 kb). Consequently, site-directed mutagenesis was performed on a portion of the HFAS cDNA containing the target site for mutagenesis that had been subcloned into pBluescript (3 kb). For the K1699A and K1699Q mutants, EcoRI and HindIII restriction sites were initially used to remove a portion of the HFAS cDNA from pFastBac1 and to insert this segment into pBlue-script KS. Following site-directed mutagenesis, the MluI and AatII restriction sites were used to remove a segment of the HFAS cDNA encompassing the site of mutation from pBluescript KS. The pFastBacI vector containing the mutant HFAS cDNA was then generated by three-way ligation. This involved ligating the MluI/AatIIDNAfragment to pFastBac1 that had been digested with MluI and HindIII, together with a second DNA fragment generated from HFAS-pFastBacI using the AatII and HindIII sites. The S2151A mutant was generated using a similar protocol using BamHI and HindIII restriction sites to transfer the relevant HFAS cDNA between pFastBacI and pBluescript. The primers used for the mutagenesis are given in TABLE ONE.

TABLE ONE.

Primers used for mutagenesis

Mutant	Primera
K1699A-Forward (F)	5′-GTGGGGTCGGCTGAGGCGCGGGCGTACCTC CAG-3′
K1699A-Reverse (R)	5′-CTGGAGGTACGCCCGCGCCTCAGCCGACCCCAC-3′
K1699Q-F	5′-GTGGGGTCGGCTGAGCAGCGGGCGTACCTC CAG-3′
K1699Q-R	5′-CTGGAGGTACGCCCGCTGCTCAGCCGACCC-CAC-3′
S2151A-F	5′-GACCTGGGCCTGGACGCGCTCATGAGCGCGCCG-3′
S2151A-R	5′-CGGCGCGCTCATGAGCGCGTCCAGGCCCAGGTC-3′

^aLocation of mutation is shown in boldface.

Expression and Purification of Wild-type and Mutant HFAS Proteins

The pFastBac1 constructs were utilized to produce recombinant baculoviral stocks by the transposition procedure of the BAC-to-BAC baculovirus expression system. Sf9 cells were transfected with recombinant bacmid DNA, and virus was harvested from cell culture medium at 72 h post-transfection. Each viral stock was subjected to two rounds of amplification after which the viral titer was determined. Sf9 cells were infected with recombinant baculovirus, and a cell pellet was collected 72 h post-infection. The pellet was resuspended in 15–30 ml of His-bind buffer (5 mM imidazole, 0.5 M NaCl, and 20 mM Tris-HCl, pH 7.9) and lysed by French press (5 times at 1000 p.s.i.). The cell lysate was then centrifuged at 33,000 rpm for 90 min, and the supernatant was filtered with a 0.45-µm Millipore filter and applied to a His-bind resin column (5-ml bed volume). The His-bind column was washed with His-bind buffer and then with wash buffer (60 mm imidazole, 0.5 m NaCl, and 20 mm Tris-HCl, pH 7.9). Protein elution was performed using a gradient of 5 mm to 1 m imidazole in 20 mm Tris-HCl and 0.5 m NaCl. Fractions containing wild-type or mutant HFAS proteins were then immediately chromatographed on a Sephadex G-25 column, equilibrated with 0.8 m KH₂PO₄, 1 mM EDTA, pH 7.4. The protein was dialyzed into 0.8 m KH₂PO₄, pH 7.4, buffer containing 1mm EDTA, and 10mm dithiothreitol using molecular porous membrane tubing. Subsequently, 100–200-µl aliquots of protein were snap-frozen in liquid nitrogen with 20% glycerol for storage at −80 °C. Protein expression and purification were analyzed by SDS-PAGE on 10% Tris-glycine gels. Protein concentration was determined by the absorbance at 280 nm using an extinction coefficient (ϵ₂₈₀) of 279650 m⁻¹ cm⁻¹ calculated at Expasy (www.expasy.ch) by assuming one active site per monomer. Protein yield was assessed using the calculated molecular mass of 273 kDa (www.expasy.ch).

Synthesis of 2-Octenoyl-CoA and 2-Dodecenoyl-CoA

2-Octenoyl-CoA (C₈-CoA) and 2-dodecenoyl-CoA (C₁₂-CoA) were synthesized from the respective acids using the mixed anhydride method (24, 25). 300 mg (1.5 mmol) of trans-2-dodecenoic acid was dissolved slowly in 2 ml of anhydrous tetrahydrofuran (THF) and then placed in a flask that had been purged with nitrogen gas. 210 mg (2.1 mmol) of anhydrous triethylamine in 1 ml of THF was added slowly to the flask followed by 386 mg (3.6 mmol) of ethylchloroformate in 1 ml of THF, which resulted in the formation of a white precipitate. An additional 2 ml of THF was added, and then the reaction mixture was stirred overnight at room temperature. 750 µl of the mixed anhydride was filtered with glass wool in a Pasteur pipette, and 500 µl of the filtrate was added dropwise with stirring to 30 mg of CoA dissolved in 1 ml of 50 mm Na₂CO₃ at room temperature. The reaction was monitored by evaluating the amount of free thiol in solution using 5,5′-dithiobis(2-nitrobenzoic acid) (26). When no more free thiol was present, the reaction mixture was immediately purified by high pressure liquid chromatography (Shimadzu), using a Phenomenex C₁₈-preparative column. Purification was performed with 20 mm ammonium acetate, 1.75% acetonitrile (Buffer A) with a 0–100% gradient of 95% acetonitrile, 5% H₂O (Buffer B) for 90 min and a flow rate of 4ml/min. Chromatography was monitored at 260 and 280 nm with a Shimadzu SPD-10A UV-visible detector, and the retention time for C₁₂-CoA was 35–40 min. Fractions containing C₁₂-CoA were combined, lyophilized, re-dissolved in 500 µl of H₂O, and re-lyophilized to remove all ammonium acetate. The C₁₂-CoA was then dissolved in 500 µl of H₂O and snap-frozen with liquid nitrogen in 50-µl aliquots for storage at −80 °C. The mass spectrometry (atmospheric pressure ionization-electrospray) calculated for C₃₃H₅₆N₇O₁₇P₃S was 947.27, and found was 948.20 [M + H]⁺. 2-Octenoyl-CoA was synthesized using a similar method. The product eluted from the C₁₈ column at 25 min. The mass spectrometry (atmospheric pressure ionization-electrospray) calculated for C₂₉H₄₈N₇O₁₇P₃S was 891.20 and found was 892.51 [M + H]⁺.

Kinetic Analysis of the Overall HFAS Reaction and HFAS Partial Activities

Specific activities for the overall HFAS reaction were determined at 25 °C in 500 µl of assay buffer (0.1 m KH₂PO₄, 1 mm EDTA, pH 7.0) using 100 µm NADPH, 20 nm wild-type HFAS or 80–160 nm mutant HFAS, 25 µm acetyl-CoA, and 100 µm malonyl-CoA. The rate of the reaction was determined by monitoring the decrease in NADPH absorbance at 340 nm, and initial velocities were calculated using an extinction coefficient (ϵ₃₄₀) of 6,300 m⁻¹ cm⁻¹. Partial activities were determined using a similar protocol in the same assay buffer. ER activity was determined using 100 µm NADPH, 80 nm HFAS or 160 nm mutant HFAS, and 100 µm crotonyl-CoA, whereas BKR activity was determined using 100 µm NADPH, 20 nm wild-type HFAS or 160 nm mutant HFAS, and 100 µm acetoacetyl-CoA. DH activity was determined by monitoring the hydration of crotonyl-CoA at 280 nm using 80 nm HFAS and 100 µm substrate. An extinction coefficient (ϵ₂₈₀) of 3600 m⁻¹ cm⁻¹ was used for the conversion of crotonyl-CoA into 3-hydroxybutyryl-CoA.

The above assay conditions were also used for the determination of k_cat and K_m values. For the overall HFAS reaction, K_NADPH was determined by using 100 µm malonyl-CoA and 25 µm acetyl-CoA and by varying the NADPH concentration from 5 to 100 µm for wild-type and from 80 to 800 µm for the mutant enzymes. Determination of K_NADPH for the mutant enzymes was performed by monitoring NADPH oxidation at 370 nm due to the high concentration of NADPH and by using an extinction coefficient of 2400 m⁻¹ cm⁻¹ (27). Similarly, K_MAL (where K_MAL is the K_m value for malonyl-CoA) was determined by varying the concentration of malonyl-CoA (5–100 µm for WT; 0.5–20 µm for K1699Q; 0.5–100 µm for K1699A) at fixed concentrations of NADPH (100 µm) and acetyl-CoA (25 µm). Finally, K_ACET (where K_ACET is the K_m value for acetyl-CoA) was determined by varying the concentration of acetyl-CoA (0.5–50 µm for WT; 5–80 µm for K1699A; 0.06–25 µm for K1699Q) at fixed concentrations of NADPH (100 µm) and malonyl-CoA (100 µm). For the BKR reaction, the K_NADPH was determined by using 100 µm AcAc-CoA, 20 nm WT, or 160 nm of mutant enzyme and by varying the NADPH concentration from 5 to 80 µm for WT, 10 to 800 µm for K1699Q, and 20 to 400 µm for K1699A. The K_ACAC (where K_ACAC is the K_m value for acetoacetyl-CoA) with BKR was determined by using 100 µm NADPH, 20 nm WT or 800 µm NADPH, 160 nm of mutant enzyme and by varying the AcAc-CoA concentration from 0.5 to 100 µm for WT and K1699A and 0.5 to 20 µm for K1699Q. For ER, K_NADPH was determined by using 100 µm crotonyl-CoA, 80 nm WT, or 160 nm of mutant enzyme and by varying the NADPH concentration from 0.5 to 200 µm for WT and 100 to 800 µm for the mutants. TheK_CRT (where K_CRT is the K_m value for crotonyl-CoA) for ER was evaluated by using 100 µm NADPH, 80 nm WT or 800 µm NADPH, 160 nm mutant enzyme and by varying the concentration of crotonyl-CoA from 1 to 100 µm for WT, 1 to 400 µm for K1699A, and 1 to 100 µm for K1699Q. The K_CRT for DH was measured by using 80 nm WT or 160 nm mutant enzymes and by varying the concentration of crotonyl-CoA from 5 to 40 µm for WT, 5 to 200 µm for K1699A, and 5 to 400 µm for K1699Q. For the DH reaction the decrease in absorbance at 280 nm was monitored. In each case k_cat and K_m values were calculated by fitting the observed initial velocities to the Michaelis-Menten equation using Grafit 4.0 (Erithacus Software Ltd.).

Steady State Kinetic Analysis of C₈-CoA and C₁₂-CoA with Enoyl Reductase

Kinetic parameters for the C₈ and C₁₂ substrates were determined as described above for crotonyl-CoA. The K_m value for C₁₂-CoA with ER was determined by using 100 µm NADPH and 20 nm WT HFAS and by varying the concentration of C₁₂-CoA from 6.25 to 200 µm. The K_NADPH was determined by using 25 µm C₁₂-CoA, 20 or 160 nm WT HFAS and by varying the concentration of NADPH from 100 to 800 µm where the oxidation of NADPH was monitored at 370 nm. k_cat and K_m values were calculated by fitting the observed initial velocities to the Michaelis-Menten equation using Grafit 4.0.

Interaction of NADPH with HFAS Monitored by Fluorescence Spectroscopy

NADPH binding to HFAS was analyzed at 25 °C using a Spex Fluorolog 3 spectrofluorimeter. NADPH fluorescence was excited at 340 nm (5 nm slit width) and monitored between 400 and 500 nm (1 nm slit width). 1-µl aliquots of 0.3 mm NADPH were added to a cuvette containing 0.5 µm wild-type or mutant HFAS enzymes in 0.1 m KH₂P0₄, 1 mm EDTA, pH 7 buffer. The NADPH concentration ranged from 0 to 20 µm.

RESULTS

Expression and Purification of Wild-type and Mutant HFAS Enzymes

500 ml of Sf9 insect cell culture yielded 5 mg of soluble wild-type HFAS that was >95% pure as judged by SDS-PAGE. Similar quantities of the K1699Q, K1699A, and S2151A mutant proteins were also obtained. Specific activities of the overall and partial reactions catalyzed by HFAS are given for the wild-type and mutant enzymes in TABLE TWO. Wild-type HFAS had a specific activity of 440 nmol/min/mg for the overall reaction. This value, together with those for the other partial activities catalyzed by the enzyme, were in good agreement with previously published data apart from the BKR activity that was ~7–10-fold lower (3). Replacement of Lys-1699 with Ala or Gln resulted in about a 10-fold decrease in specific activity for the overall reaction (TABLE TWO). In contrast, the S2151A mutant was completely inactive in all assays. To shed more light on the impact of the mutagenesis on the reactions catalyzed by HFAS, we determined k_cat and K_m values for the overall reaction as well as each partial reaction.

TABLE TWO.

Specific activities of wild-type and K1699A, K1699Q, and S2151A HFAS

Reactiona	Specific activity (nmol/min/mg)
	WT-HFAS	K1699A	K1699Q	S2151A	HepG2FASb	MBP-FASb
Overall	440 ± 70	46 ±8	33 ± 4	Inactive	462	611
ER	110 ± 30	12 ± 4	15 ± 5	Inactive	89.3	34
BKR	240 ± 40	36 ± 5	43 ± 4	Inactive	2514	1921
DH	140 ± 40	47 ± 11	55 ± 3	Inactive	57	49

^aData are given for the specific activity using NADPH, malonyl-CoA, and acetyl-CoA as substrates (overall), NADPH and crotonyl-CoA (ER), NADPH and acetoacetyl-CoA (BKR), and crotonyl-CoA (DH). Note that the DH activity was assayed in the direction of hydration.

^bData are from Jayakumar et al.(3).

Kinetic Analysis of Wild-type and Mutant HFAS Enzymes

k_cat and K_m values for the overall reaction catalyzed by wild-type and mutant HFAS enzymes, as well as for each partial reaction, are given in TABLE THREE. The k_cat value for the overall HFAS reaction was 160 min⁻¹, whereas the K_m values for NADPH (K_NADPH), malonyl-CoA (K_MAL), and acetyl-CoA (K_ACET) were all in the low micromolar range, resulting in k_cat/K_m values of 20–30 µm⁻¹ min⁻¹. Replacement of Lys-1699 with either an Ala or Gln residue resulted in a modest (8-fold) decrease in k_cat and a large 600-fold decrease in k_cat/K_NADPH. In contrast, k_cat/K_MAL and k_cat/K_ACET were unaffected by the K1699Q mutation and decreased only slightly in the K1699A enzyme. One explanation for these data is that the interactions with Lys-1699 stabilize the most stable ground state following association of NADPH with the enzyme by around 3 kcal/mol and contribute an additional ~1 kcal/mol to stabilization of the rate-limiting transition state for a total transition state stabilization of ~4 kcal/mol.

TABLE THREE.

Kinetic evaluation of wild-type, K1699A, and K1699Q HFAS

Enzyme	Overall reaction
	kcat	KNADPH	KMAL	KACET	kcat/KNADPHa	kcat/KMALa	kcat/KACETa
	min−1	µm	µm	µm	µm−1 min−1	µm−1 min−1	µm−1 min−1
WT	160 ± 15	5 ± 1	6 ± 2	7 ± 3	32 ± 9	27 ± 11	23 ± 12
K1699A	19 ± 4	370 ± 90	3.0 ± 0.8	7 ± 2	0.05 ± 0.02 (0.2%)	6 ± 3 (22%)	3 ± 2 (13%)
K1699Q	18 ± 2	400 ± 80	1.0 ± 0.3	0.5 ± 0.2	0.05 ± 0.02 (0.2%)	18 ± 7 (67%)	36 ± 18 (150%)
Enzyme	β-Ketoacyl reductase
	kcat	KNADPH	KACAC		kcat/KNADPHa	kcat/KACACa µm−1 min−1
	min−1	µm	µm		µm−1 min−1
WT	240 ± 12	4 ± 2	10 ± 2		60 ± 33	24 ± 6
K1699A	20 ± 1	52 ± 9	2.0 ± 0.3		0.39 ± 0.09 (0.7%)	10 ± 2 (42%)
K1699Q	26 ± 1	123 ± 14	1.2 ± 0.3		0.21 ± 0.03 (0.4%)	22 ± 6 (83%)
Enzyme	β-Hydroxyacyl dehydratase
	kcat	KCRT	kcat/KCRTa
	min−1	µm	µm−1 min−1
WT	6 ± 1	7 ± 5	0.9 ± 0.8
K1699A	24 ± 3	40 ± 13	0.6 ± 0.3 (67%)
K1699Q	NDb	ND	ND
Enzyme	Enoyl reductase
	kcat	KNADPH	KCRT	kcat/KNADPHa	kcat/KCRTa
	min−1	µm	µm	µm−1 min−1	µm−1 min−1
WT	20 ± 2	3 ± 2	6 ± 2	7 ± 5	3 ± 1
K1699A	9 ± 1	400 ± 40	3.0 ± 0.7	0.02 ± 0.01 (0.3%)	3 ± 1 (100%)
K1699Q	13 ± 3	350 ± 180	3.0 ± 0.5	0.04 ± 0.03 (0.6%)	4 ± 2 (130%)

^aNumbers in parentheses are values relative to wild type (%).

^bND indicates not determined.

To gain further insight into the role of Lys-1699 in HFAS catalysis, we also determined k_cat and K_m values for each partial reaction. For the BKR reaction, the k_cat value for the reduction of acetoacetyl-CoA was 240 min⁻¹ for the wild-type enzyme. Similar to the overall HFAS reaction, mutagenesis of Lys-1699 reduced the k_cat value for the BKR reaction by 10–12-fold while reducing k_cat/K_NADPH value by 150–300-fold and leaving k_cat/K_ACAC value unaffected. For the dehydrase reaction, assayed by monitoring the hydration of crotonyl-CoA, k_cat was 6 min⁻¹ for wild-type and 24 min⁻¹ for the K1699A mutant, whereas k_cat/K_CRT value was unchanged. Finally, we also assessed the effect of mutating Lys-1699 on the kinetic parameters for the ER partial reaction. The k_cat value for the wild-type enzyme was 20 min⁻¹, 8-fold lower than that for the overall reaction. As observed for the overall and BKR reactions, again the major impact of the Lys-1699 replacements was a large 175–350-fold decrease in k_cat/K_NADPH values, whereas the k_cat/K_CRT value was unaffected. Analysis of these data on the partial HFAS reactions provides a similar conclusion to that reached above for the overall HFAS reaction. Lys-1699 stabilizes the transition state following binding of NADPH to the E·AcAc-CoA or E·Crt-CoA complexes by 3–3.5 kcal/mol and stabilizes the respective ground states by 2–3 kcal/mol.

Kinetic Analysis of Wild-type HFAS as a Function of Substrate Chain Length

A primary goal of the kinetic experiments was to establish assays for assessin the interaction of inhibitors with the different HFAS active sites. As noted above, k_cat for the reduction of crotonyl-CoA by HFAS was around 8-fold slower than the corresponding value for the overall reaction. To explore whether alteration in substrate chain length resulted in an ER substrate that more closely mirrored the reduction of substrates when covalently attached to HFAS, we synthesized octenoyl-CoA (C₈-CoA) and dodecenoyl-CoA (C₁₂-CoA). However, comparison of the kinetic parameters for C₈-CoA and C₁₂-CoA with those obtained for Crt-CoA (TABLE FOUR) clearly show that the increase in substrate chain length results in a less optimal substrate for analyzing the ER partial reaction.

TABLE FOUR.

Kinetic parameters for the enoyl reductase activity of wild-type HFAS as a function of chain length

Enoyl-CoA substrate	kcat	KNADPH	KACYL	kcat/KNADPHa	kcat/KACYLa
	min−1	µm	µm	µm−1/min−1	µm−1/min−1
Crotonyl-CoA (C4)	20 ± 2	3 ± 2	6 ± 2	7 ± 56	3.3 ± 1.3
Octenoyl-CoA (C8)	3 ± 1	100 ± 70	NDb	0.03 ± 0.03 (0.4%)	ND
Dodecenoyl-CoA (C12)	13 ± 5	540 ± 300	7 ± 1	0.02 ± 0.02 (0.3%)	2 ± 1 (61%)

^aNumbers in parentheses are values relative to wild type (%).

^bND indicates not determined.

NADPH Binding to Wild-type and Mutant HFAS Enzymes

In order to assess the direct impact of the Lys-1699 mutations on the affinity of NADPH for HFAS, we attempted to use the change in NADPH fluorescence on binding to determine K_d values for wild-type and mutant HFAS enzymes. NADPH fluorescence was excited at 340 nm and monitored between 400 and 500 nm. In each case the addition of NADPH to the enzyme solutions resulted in a linear increase in fluorescence up to the maximum concentration of NADPH used (20 µm) (data not shown). The fluorescence signal was measurably increased in the presence of the wild-type and mutant HFAS proteins, indicating that NADPH was binding to the enzymes. However, because binding did not saturate over the concentration range used, a K_d value could not be determined. All we can conclude is that the K_d value for NADPH is >20 µm for each enzyme. Note that although the S2151A mutant was completely inactive in all the kinetic assays, NADPH fluorescence did increase in the presence of this enzyme, which suggests that S2151A still has an intact NADPH-binding site(s).

DISCUSSION

It has been hypothesized that HFAS is a target for anti-cancer drug discovery. In order to direct inhibitor design, methods need to be available to quantitate the affinity of inhibitors for the enzyme. Because the enzyme could be inhibited by blocking any of the active sites in the multienzyme complex, we need to be able to assess inhibition of each individual partial activity. Although IC₅₀ measurements are useful for evaluating the potency of a series of inhibitors under identical conditions, K_i values are needed in order to provide information about the mechanism of inhibition. Because K_i values are determined by obtaining the apparent values of V_max and K_m at different inhibitor concentrations, we set out first to determine k_cat and K_m values for the overall HFAS reaction as well as each individual partial activity. More importantly, these studies were focused on the human fatty-acid synthase, a mammalian FAS that has been less extensively studied than the FAS enzymes from other sources such as chicken, rat, or goose. In addition, although there is no high resolution structural data for a mammalian FAS, our studies are also focused on supplementing published information on residues important for catalysis.

Kinetic Parameters for the Overall and Partial HFAS Activities

Methods for assaying the partial HFAS reactions involve the addition of exogenous substrates to the enzyme. Ideally, these reactions would closely mimic the reaction catalyzed by each active site during the overall chain elongation reaction. In this situation we would predict that k_cat value for the partial reaction would be similar to, or greater than, the k_cat value for the overall HFAS reaction of 160 min⁻¹. Perusal of TABLE THREE reveals that k_cat value for the reduction of acetoacetyl-CoA by HFAS, which is an assay for the BKR reaction, is indeed greater than 160 min⁻¹. However, the k_cat values for the DH and ER activities are significantly less than the k_cat value for the overall reaction. For the DH activity we are actually monitoring the DH reaction in reverse (i.e. substrate hydration and not dehydration). This reaction will compete with the ER active site for crotonyl-CoA, thereby reducing the ER k_cat value. Both the ER and DH activities have similar K_m values for crotonyl-CoA, and we note that the k_cat value for DH is 3-fold lower (6 min⁻¹) than that for ER (20 min⁻¹). By assuming that the binding of NADPH does not affect k_cat or K_m values for the DH reaction, a corrected value of 26 min⁻¹ can be calculated for the ER reaction in the absence of substrate hydration. This value is still substantially smaller than k_cat for the overall reaction (160 min⁻¹), suggesting that the exogenous CoA moiety prevents optimal positioning of the crotonyl acyl group in the ER active site compared with the endogenous ACP carrier or that a physical step, such as product release, is rate-limiting for the reduction of crotonyl-CoA by the enzyme.

The observation of a lower k_cat value for crotonyl-CoA reduction by HFAS compared with the k_cat value for the overall reaction prompted us to examine the chain length specificity of the ER reaction. However, as noted under “Results,” C₈-CoA and C₁₂-CoA substrates were reduced with k_cat values similar to or smaller than the value observed for Crt-CoA. In addition, we were also curious to examine whether the endogenous pantetheine group could interfere with binding of the exogenously added substrates (28). Consequently, we replaced Ser-2151, the point of attachment of the FAS pantetheine, with an alanine residue. A similar experiment had been reported previously by Smith and co-workers (22) in which the equivalent mutation in the rat FAS abolished the overall FAS reaction without affecting the BKR, DH, or ER partial reactions. However, in our case, although the purified S2151A HFAS enzyme was soluble and showed some ability to bind NADPH based on the fluorescence assay (see under “Results”), this enzyme showed no activity in any of the partial reactions. Consequently, in addition to acting as a substrate carrier for the overall FAS reaction, the pantetheine must also play a structural role in stabilizing the active form of the human enzyme.

Role of Lys-1699 in HFAS Catalysis

Early studies to identify catalytic residues in the reaction catalyzed by FAS utilized PLP to covalently modify lysine residues in the enzyme. These experiments demonstrated that modification of Lys-1699 selectively inhibited the ER reaction by interfering in the binding of NADPH to the ER enzyme (2, 19, 29, 30). However, although these data localize Lys-1699 to the ER NADPH-binding site, they provide no information on the precise role of Lys-1699 in catalysis. For example, although Lys-1699 could be involved in simply binding NADPH, this residue could also have a more direct role in catalysis as exhibited by a conserved lysine residue in the SDR enzyme family (25, 32). Consequently, we used site-directed mutagenesis to replace Lys-1699 with Ala and Gln residues, and we determined the impact of these mutations on the kinetic parameters for the FAS-catalyzed reactions.

As can be seen in TABLE THREE, mutagenesis of Lys-1699 results in a large decrease in k_cat/K_NADPH values for the overall reaction without affecting k_cat/K_MAL and k_cat/K_ACET values. This change in k_cat/K_NADPH is equivalent to a loss of 4 kcal/mol in stabilization of the rate-limiting transition state that comes after NADPH binding. In addition, because the k_cat value is only modestly affected in the two mutant enzymes, Lys-1699 must also stabilize the most stable ground states after NADPH binding by around 3 kcal/mol. Thus, in agreement with the PLP labeling experiments, mutagenesis of Lys-1699 has likely affected forms of the enzyme that interact with NADPH.

The impact of the Lys-1699 mutations on the overall HFAS reaction are mirrored by similar changes to the ER reaction, implicating Lys-1699 in both ground and transition state stabilization during substrate reduction. In the SDR family, the conserved active site lysine interacts with the NAD(P)H ribose adjacent to the nicotinamide ring and also, in some cases, hydrogen-bonds to an adjacent conserved tyrosine residue. Although Lys-1699 is not part of a consensus SDR Tyr/Lys motif, the effect of mutating Lys-1699 on the ER reaction is consistent with Lys-1699 interacting with substrates at the ER active site in a similar way to that observed for the SDR enzymes as shown in Fig. 1. Our data are consistent with Lys-1699 occupying a position in the ER active site where it is involved in NADPH binding and can also, directly or indirectly, provide some additional stabilization to the developing enolate intermediate formed during substrate reduction. In the SDR family, the conserved active site lysine hydrogen-bonds to the NAD(P)H ribose adjacent to the nicotinamide ring and also, in some cases, interacts with the conserved tyrosine to assist in transition state stabilization (32). We propose that Lys-1699 fulfills a similar function in HFAS, although the identity of the (putative) catalytic tyrosine has not been identified in this enzyme.

FIGURE 1. Proposed interactions of Lys-1699 with the NADPH cofactor and substrate in the ER active site.

This scheme is derived from the known role of a conserved lysine in the active site of the SDR enzyme family. Lys-1699 is shown interacting with the substrate via a tyrosine residue because the SDR enzymes also utilize a conserved tyrosine in the reactions they catalyze. In addition, the hydride transfer reaction is shown occurring from the pro-(4R) NADPH hydrogen in agreement with the known stereo-chemistry of the ER reaction (31).

In addition to effects on the overall and ER reactions, the Lys-1699 mutations also affected the BKR reaction, causing a 10-fold decrease in k_cat values and a 150–300-fold decrease in k_cat/K_NADPH values when acetoacetyl-CoA was the substrate. One explanation for this observation is that the BKR and ER activities share a single NADPH-binding site and that Lys-1699 fulfills a similar function in both the ER and BKR reactions. This proposal is plausible given the overall similarities in the chemistry of the ER and BKR reactions. Indeed, the structure of a type II 3-hydroxyacyl-CoA dehydrogenase provides insight into how a Tyr/Lys dyad, as suggested for the ER reaction (see above), can also function in reduction of a β-keto group (33).

However, a significant amount of evidence has accumulated indicating that the ER and BKR reactions have separate NADPH-binding sites. Indeed, the original assignment of Lys-1699 to the ER enzyme was based on the observation that PLP labeling selectively inhibited the ER activity while leaving the other partial reactions unaffected. In addition, Smith and co-workers (34, 35) mutated residues in the proposed NADPH-binding sites of the ER (Gly-1672 and Gly-1673) and BKR (Gly-1886 and Gly-1888) enzymes of rat FAS, and they demonstrated that these mutations had selective effects on the two partial activities. Given the extensive data for two NADPH-binding sites, our current hypothesis is that NADPH binding to one cofactor site is sensed by the second cofactor site and, thus, that the result of mutating Lys-1699 has an allosteric effect on the BKR reaction. To shed further light on the role of Lys-1699 in NADPH binding, we studied the binding of NADPH to the free enzyme by using fluorescence spectroscopy.

NADPH Binding Studied by Fluorescence Spectroscopy

In an attempt to determine whether Lys-1699 mutagenesis has a selective effect on the interaction of NADPH with the ER cofactor site, we used fluorescence spectroscopy to analyze the interaction of wild-type and mutant HFAS with NADPH. This method has been used previously to study the interaction of NADPH with the chicken and rat FAS complexes. For chicken FAS, K_d values in the low micromolar range have been reported for the two NADPH sites (11, 30, 36). Conversely, for rat FAS no saturation with NADPH was observed up to an NADPH concentration of 16 µm (35). In our case, no saturation was observed when NADPH was titrated into a solution of the wild-type HFAS enzyme up to an NADPH concentration of 20 µm NADPH. Thus, similar to rat FAS, NADPH binds to the free HFAS enzyme with a K_d value greater than 20 µm. In addition, similar NADPH binding data were obtained for the two mutants, and thus, these data are unable to provide further insight into the interaction of Lys-1699 with NADPH.

Although our fluorescence data were unable to provide further information into the existence of one or two NADPH-binding sites in HFAS, previous studies on FAS enzymes from other sources support the proposal that the ER and BKR NADPH sites communicate allosterically. Previously, Smith and co-workers (35) quantified the interaction of NADPH with the ER and BKR enzymes in rat FAS by introducing mutations that separately compromised the binding of NADPH to each site. This method gave K_d values of 4.2 and 15 µm for NADPH binding to the ER and BKR cofactor binding sites, respectively. However, the data obtained for NADPH binding to the wild-type enzyme did not show evidence of a saturable component with K_d ~4 µm or other indications of biphasic behavior up to 16 µm NADPH, suggesting to us that in the wild-type enzyme the ER and BKR sites are not independent of each other. In addition, Cardon and Hammes (30) observed that labeling of Lys-1699 by PLP in the chicken enzyme altered the affinity of the BKR enzyme for NADPH. Furthermore, Poulose and Kolattukudy (20) suggested that binding of NADPH to the ER enzyme was important for formation of the functional FAS dimer and that the inhibition of NADPH binding to this site by PLP labeling of Lys-1699 resulted in an effect on other activities in the FAS enzyme. Consequently, our preferred explanation for the effect of Lys-1699 mutagenesis on the ER and BKR activities reported here is that the two HFAS reductase reactions utilize distinct NADPH-binding sites, and that these two sites can communicate allosterically with each other. Because the FAS complexes are homodimers, it is possible that the ER and BKR NADPH-binding sites on separate polypeptide chains are physically close in space, providing an interface for allosteric communication. Answers to some of these questions will have to await high resolution structural data.

Conclusion

Site-directed mutagenesis has been used to investigate the role of Lys-1699 in the reaction catalyzed by HFAS. Kinetic analysis of the K1699A and K1699Q mutants reveals that Lys-1699 provides around 4 kcal/mol in transition state stabilization and 3 kcal/mol in ground state stabilization for the overall reaction catalyzed by HFAS. Similar effects on the ER partial reaction following mutation of Lys-1699 are consistent with the proposed location of this residue in the ER NADPH-binding site. However, these mutations also impact the BKR reaction. Given the large body of data supporting the presence of separate NADPH-binding sites for the ER and BKR partial activities, it is hypothesized that the two NADPH sites communicate allosterically with each other.