Acceleration of the Substrate Cα Deprotonation by an Analogue of the Second Substrate Palmitoyl-CoA in Serine Palmitoyltransferase*

Serine palmitoyltransferase (SPT) is a key enzyme of sphingolipid biosynthesis and catalyzes the pyridoxal 5′-phosphate (PLP)-dependent decarboxylative condensation reaction of l-serine with palmitoyl-CoA to generate 3-ketodihydrosphingosine. The binding of l-serine alone to SPT leads to the formation of the external aldimine but does not produce a detectable amount of the quinonoid intermediate. However, the further addition of S-(2-oxoheptadecyl)-CoA, a nonreactive analogue of palmitoyl-CoA, caused the apparent accumulation of the quinonoid. NMR studies showed that the hydrogen-deuterium exchange at Cα of l-serine is very slow in the SPT-l-serine external aldimine complex, but the rate is 100-fold increased by the addition of S-(2-oxoheptadecyl)-CoA, showing a remarkable substrate synergism in SPT. In addition, the observation that the nonreactive palmitoyl-CoA facilitated α-deprotonation indicates that the α-deprotonation takes place before the Claisen-type C–C bond formation, which is consistent with the accepted mechanism of the α-oxamine synthase subfamily enzymes. Structural modeling of both the SPT-l-serine external aldimine complex and SPT-l-serine–palmitoyl-CoA ternary complex suggests a mechanism in which the binding of palmitoyl-CoA to SPT induced a conformation change in the PLP-l-serine external aldimine so that the Cα–H bond of l-serine becomes perpendicular to the plane of the PLP-pyridine ring and is favorable for the α-deprotonation. The model also proposed that the two alternative hydrogen bonding interactions of His159 with l-serine and palmitoyl-CoA play an important role in the conformational change of the external aldimine. This is the unique mechanism of SPT that prevents the formation of the reactive intermediate before the binding of the second substrate.

Serine palmitoyltransferase (SPT) 3 is a pyridoxal 5Ј-phosphate (PLP)-dependent enzyme and catalyzes a decarboxylative condensation reaction of L-serine with palmitoyl-CoA to generate 3-ketodihydrosphingosine (KDS). This reaction is the first and rate-limiting step of the sphingolipid biosynthesis, and regulation of the SPT reaction directly influences cellular sphingolipid homeostasis (1). Therefore, the molecular mechanism of the SPT reaction has been attracting attention, since it has been shown that sphingolipid is a potent bioactive lipid mediator playing crucial roles in diverse aspects of cell structure and function (1)(2)(3)(4).
Eukaryotic SPTs have been known to exist as membranebound heterodimers composed of two subunits called SPTLC1 (LCB1) (long chain base 1), which does not have a PLP-binding motif, and SPTLC2 (LCB2), which carries a lysine residue that forms the Schiff base with PLP (5-10). The identification of a new third SPT subunit, SPTLC3, and an octameric SPT structure model have been proposed (11,12). However, further studies of the SPT reaction mechanism have not been carried out, because the instability and the hydrophobic nature of eukaryotic SPTs have precluded obtaining a sufficient amount of the enzyme for detailed observations of the reaction (13).
Recently, we isolated several SPT genes from the sphingolipid-containing bacteria, such as Sphingomonas paucimobilis, Sphingobacterium multivorum, Sphingobacterium spiritivorum, and Bdellovibrio stolpii (14 -16). These bacterial SPTs show about 30% identity in the amino acid sequence with eukaryotic SPT subunit proteins and conserve the amino acid residues of eukaryotic SPTs assumed to be involved in catalysis. All of the bacterial SPTs were successfully overproduced in Escherichia coli and purified as water-soluble active homodimers. Bacterial SPTs are considered to be a prototype of the eukaryotic membrane-bound enzymes, and this system was a major breakthrough that provided the first detailed mechanistic studies of the SPT reaction. The recent elucidation of a high resolution crystal structure of the S. paucimobilis SPT is expected to boost the structural/functional analysis of the catalytic reaction of this enzyme (17). Fig. 1 shows the proposed reaction mechanisms of SPT. The first step of the SPT reaction cycle is the binding of the * This work was supported by Grant-in-Aid for Encouragement of Young Scientists (B) 16770103 and Grant-in-Aid for Scientific Research (C) 18570114 (to H. I.) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and by Grant-in-Aid for Scientific Research (C) 16570125 (to H. H.) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence may be addressed. Tel.: 81-72-684-7291; Fax: 81-72-684-6516; E-mail: ikushiro@art.osaka-med.ac.jp. 2 To whom correspondence may be addressed. E-mail: hayashi@art. osaka-med.ac.jp.
amino acid substrate L-serine to the enzyme and the formation of the external aldimine (II). In a previous paper (18), we spectroscopically and kinetically analyzed this step using the purified SPT, L-serine, and its analogues. Through the study, it was found that the ␣-carboxyl group of the substrate L-serine is important for the recognition by the enzyme, and the Michaelis complex of SPT and L-serine readily undergoes transaldimination to form the external aldimine. As generally observed for the PLP-dependent enzymes, the formation of the quinonoid intermediate (carbanion) from the external aldimine is the critical step in the reaction cycle. In the early studies using a yeast cell free system, two contradictory mechanisms for the KDS formation have been proposed: formation of the quinonoid intermediate (A-III) by decarboxylation of the substrate L-serine, followed by acylation (A) and formation of the quinonoid intermediate (B-III) by ␣-deprotonation of L-serine, followed by acylation and subsequent decarboxylation (B) (19 -21). According to the report by Krisnangkura and Sweeley (21) that deuterium was introduced into the ␣-position of KDS during the reaction in D 2 O using rat liver microsomes, mechanism B has been thought to be more appropriate. However, considering the microscopic reversibility at the equilibrium between VI and B-V, the introduction of deuterium into the C2 position of KDS is possible even if the reaction proceeds via mechanism A. Therefore, the above-mentioned deuterium exchange experiment does not deny mechanism A, and it has been elusive which mechanism operates in the formation of KDS.
In this paper, we specifically studied the regulation of the ␣-deprotonation of the amino acid substrate L-serine and the involvement of the second substrate palmitoyl-CoA. We investigated using 1 H NMR to study the exchange of the ␣-proton of be formed in mechanism A through VI by the principle of microscopic reversibility; therefore, even if the SPT reaction proceeds via mechanism A, the proton derived from the solvent can be incorporated into the KDS. L-serine with the solvent in the presence and absence of S-(2oxoheptadecyl)-CoA, the structural analogue of palmitoyl-CoA. The results, together with the spectroscopic and kinetic studies, clearly demonstrated the presence of "substrate synergism," in which the ␣-proton of L-serine is activated by the binding of the second substrate palmitoyl-CoA. Molecular modeling of SPT complexed with L-serine and palmitoyl-CoA provided a plausible mechanism consistent with the experimental findings.

EXPERIMENTAL PROCEDURES
Chemicals-Palmitoyl-CoA was obtained from Funakoshi (Tokyo, Japan). Isopropyl 1-thio-␤-D-galactoside and coenzyme A were from Sigma. The PD-10 columns were from Amersham Biosciences/GE Healthcare. E. coli BL21(DE3) pLysS and plasmid pET21b were from Novagen. All other chemicals were of the highest commercially available grade.
Expression and Purification of SPT-By the PCR amplification using pET21b/SPT1 mod (18) as the template, the eight amino acid residues from Ala 2 to Pro 9 of the SPT gene were deleted, and new restriction sites, NdeI and HindIII, were introduced into the gene at the translation initiation and termination sites, respectively. The mutation of His 159 to phenylalanine (H159F) was introduced into the SPT gene by two-step PCR using the following mutagenic primers: 5Ј-GACTGTACGGAATTGCATATGACCGAAGCCGCCGC-TCAGCCCCACGC-3Ј (forward1), 5Ј-CCTCGACGCGACA-GCTTTGCGTCGATCTATGACG-3Ј (forward2), 5Ј-CGTCAT-AGATCGACGCAAAGCTGTCGGCGTCGAGG-3Ј (reverse1), 5Ј-GACTGTACGGTCGACTCAGCCGATGACGCCGACC-GCGCGG-3Ј (reverse2) (mutated bases are italicized). As the first step, the two DNA fragments encoding Met 1 -Asp 164 and Leu 154 -Gly 420 of SPT were PCR-amplified using the primer pairs of forward1-reverse1 and forward2-reverse2, respectively, against pET21b/SPT1 mod . Then the full-length SPT gene having the H159F mutation and the restriction site of NdeI or HindIII at each end was synthesized by the second PCR amplification against the mixture of former PCR products using the primer pair of forward1-reverse2. Each modified SPT gene was treated with both NdeI and HindIII, purified, and ligated into pET21b. Then these recombinant plasmids were used to transform the E. coli BL21 (DE3) pLysS cells. Expression of the protein was induced with 0.1 mM isopropyl 1-thio-␤-D-galactoside and continued for 6 h at 37°C. The recombinant enzyme was purified as previously described (14). The expression level of the N-terminal deleted construct was lower than that of the full-length SPT, but the activity and the physical characteristics of the purified enzyme were not significantly changed as summarized in Table 1.
Synthesis of S-(2-Oxoheptadecyl)-CoA-1-Chloroheptadecan-2-one was synthesized from tetradecyl bromide and 1-chloro-2-methoxy-ethane. S-(2-Oxoheptadecyl)-CoA ( Fig.  2) was synthesized from 1-chloroheptadecan-2-one and coenzyme A according to the method described by Rudnick et al. (22). The rough product (70% purity) was purified using a Megabond Elute C18 cartridge (Varian), and then the eluate from the column was evaporated and freeze-dried. A final product of 410 mg was obtained from 800 mg of 1-chloroheptadecan-2one (38.7% yield). The purity was determined to be 96% by high pressure liquid chromatography using a SUMIPAX ODS A-212 (5-mm) column (3.9-mm inner diameter ϫ 150 mm; Sumika Chemical Analysis Service, Osaka   Spectrometric Measurements-The absorption spectra of SPT were recorded by a Hitachi U-3300 spectrophotometer at 25°C. The stopped-flow spectrophotometry was performed using an Applied Photophysics SX.18MV system (Leatherhead, UK) at 25°C. The dead time was 2.3 ms under a gas pressure of 500 kilopascals. The time-resolved spectra were collected by the SX.18MV system equipped with a photodiode array accessory and the XScan (version 1.08) control software. The absorption changes were analyzed using the program ProK-II (Applied Photophysics). The buffer solution for the spectrometric measurements contained 50 mM HEPES-NaOH (pH 7.5), 150 mM KCl, and 0.1 mM EDTA. SPT was equilibrated with this buffer by gel filtration using a PD-10 (Sephadex G-25) column prior to the measurements.
NMR Samples and Spectral Measurements-For the NMR spectroscopy, all exchangeable hydrogens of the enzyme, the ligands, and the buffer base were replaced with deuterium. Known amounts of the buffer base were dissolved in D 2 O and freeze-dried. After repeating this treatment twice, the buffer solutions were reconstituted in D 2 O. L-Serine, D-serine, and S-(2-oxoheptadecyl)-CoA were also subjected to the same pretreatment. They were each dissolved in 50 mM potassium pyrophosphate buffer (D 2 O, pH 7.5). SPT was equilibrated twice with deuterium-substituted buffer using a PD-10 column.
The 1 H NMR spectra were measured in Wilmad 5-mm NMR tubes kept at 23°C by a Varian UNITY 400 spectrometer operating at 400 MHz. A flip angle of 20°with a relaxation delay time of 5 s was used. To obtain a better signal-tonoise ratio, an exponential window function for 0.2-or 0.5-Hz line-broadening factor was applied to the free induction decays of NMR measurements. Chemical shifts are expressed as ppm relative to an external standard of 3-(trimethylsilyl)propionic acid-d 4 .
Structural Modeling-The modeling of S. paucimobilis SPT on the crystal structure of Rhodobacter capsulatus 5-aminolevulinic acid synthase (ALAS) complexed with succinyl-CoA (Protein Data Bank code 1BWO) was carried out using MOE (version 2005.06; Chemical Computing Group, Montréal, Canada) according to the manufacturer's instructions. The amino acid sequence of SPT (GenBank TM number AB055142) was aligned to that of R. capsulatus ALAS (GenBank TM P18709) using the BLOSUM62 substitution matrix with the gap start and gap extend penalties set to 7 and 1, respectively. The SPT sequence is slightly longer than that of ALAS, and these two sequences have 34% identity. After the alignment, the residues of SPT were fitted to the coordinates of subunits D and E of ALAS. Ten intermediates were generated using a coarse minimization, and the best intermediate was chosen for subsequent studies. The model consists of the residues Arg 22 -Ala 415 of S. paucimobilis SPT. The N-terminal 21 and C-terminal 5 residues were not modeled. The cis peptide bonds between Gly 149 and Glu 150 of chain 2 and that between Arg 319 and Glu 320 of chain 1, which were considered to be artifacts during the modeling, were changed to trans and were locally energy-minimized. The crystal structure of ALAS (subunits D and E) was superimposed on the model, and by subtracting the ALAS protein, PLP and succinyl-CoA were incorporated into the SPT model. These ligand structures were used to model the PLP-L-serine aldimine and palmitoyl-CoA, respectively. The C3-C4 -C4Ј-N and C4Ј-N␣-C␣-H dihedral angles of the PLP-L-serine aldimine were set to 180 and 90°, respectively, so that the conformation resembles those of the external aldimine of most PLP enzymes that activates the ␣-proton. The aliphatic group of palmitoyl-CoA was extended from C3 of succinyl-CoA using the Molecular Builder of MOE. Two structures, one with the PLP-L-serine aldimine and the other with the PLP-L-serine aldimine plus palmitoyl-CoA, were energy-minimized using the MMFF94x force field with all of the atoms of SPT and the ligands allowed to move.
Other Methods-The SPT activity was measured according to the previously described methods with minor modification (14). Aliquots of the reaction mixture were spotted on silica gel 60 HPTLC plate (Merck), and [3-14 C]KDS was separated from L-[3-14 C]serine with a solvent system of chloroform, methanol, 2 N NH 4 OH (40:10:1, v/v/v). The radioactivity of each KDS spot was quantified by a BAS2500 image analyzer (Fuji Film, Tokyo, Japan), using the intensity of the authentic L-[3-14 C]serine. The protein concentration during the purification procedure was determined with a BCA protein assay kit (Pierce) using bovine serum albumin as the standard. The protein concentration of the purified SPT was spectrophotometrically determined using the molar extinction coefficient of 2.83 ϫ 10 4 M Ϫ1 cm Ϫ1 at 280 nm for the PLP form of the enzyme (14). SDS-PAGE was carried out using the SDS-Tris system with a 10% polyacrylamide gel according to the procedure described by Laemmli (23).

Generation of the Quinonoid Intermediate from the External Aldimine by S-(2-Oxoheptadecyl)-CoA-Sphingomonas SPT
has absorption bands at 338 and 426 nm, which arise from the enolimine and ketoenamine tautomers, respectively, of the PLP-Lys 265 Schiff base. The addition of L-serine to SPT increased the 426 nm peak and concomitantly decreased the 338 nm peak, reflecting the transaldimination process from the PLP-Lys 265 Schiff base to the PLP-L-serine Schiff base (i.e. the external aldimine) (Fig. 3, A  The 493-nm absorption band was stable; its intensity did not change for prolonged periods. In addition, no detectable amounts of CoA or condensation products were found in the solution (data not shown). This is consistent with the chem-istry of the palmitoyl-CoA analogue. Thus, the acyl carbonyl group and the S atom, which form the thioester bond in palmitoyl-CoA, are separated by a methylene group in S-(2oxoheptadecyl)-CoA, and, therefore, the analogue cannot undergo the Claisen-type condensation reaction with the carbanionic C␣ of the quinonoid intermediate.
Kinetic Analysis of the Spectral Transition of the SPT-L-Serine Complex upon Reaction with S-(2-Oxoheptadecyl)-CoA-In order to elucidate the process of the quinonoid formation in more detail, a transient kinetic analysis was carried out on the reaction described above. SPT (42 M) was saturated in advance with 30 mM L-serine and reacted with S-(2-oxoheptadecyl)-CoA, and the spectral changes were followed in a stopped-flow spectrophotometer. A time-dependent increase in the 493 nm absorbance was observed (Fig. 4, A and B). The reaction was followed at the S-(2-oxoheptadecyl)-CoA concentrations of 10, 20, 40, 80, 160, and 320 M. At each concentration, the spectral change was a nearly single exponential process. However, since the S-(2-oxoheptadecyl)-CoA concentrations were of the same order as the enzyme concentration, a numerical analysis using the Pro-KII program was necessary. The spectra generated according to the following scheme were well fit by a global analysis to the experimentally obtained time-resolved spectra, where E represents SPT, S 1 is L-serine, ES 1 is the SPT-L-serine external aldimine complex, S 2 is S-(2-oxoheptadecyl)-CoA, ES 1 S 2 is the ternary complex, and EQ 1 S 2 is the SPT-L-serine quinonoid intermediate complexed with S-(2-oxoheptadecyl)-CoA. The kinetic parameters were obtained as follows: K d analog ϭ 31.7 Ϯ 1.1 M, k ϩ2 ϭ 2.6 Ϯ 4.8 s Ϫ1 , and k Ϫ2 ϭ 64.4 Ϯ 4.2 s Ϫ1 . That the S.D. value of k ϩ2 exceeds the mean value is the result of the Levenberg-Marquardt minimization and not that of the experimental errors. The calculated absorption spectrum of ES 1 S 2 was very similar to the spectrum of ES 1 and had no absorption at 493 nm (Fig. 4C, lines 1 and 2). On the other hand, the calculated absorption spectrum of EQ 1 S 2 showed an intense absorption at around 493 nm with a shoulder at around 460 nm. The apparent dissociation constant (K d app ) obtained from the static spectroscopic analysis (Fig. 3C) is related to the above kinetic parameters as follows, The calculated value of K d app using the transient kinetic parameters obtained as above was 30 M, which was very close to the experimental value of 35 M, thus supporting the validity of the above kinetic analysis. The apparent equilibrium constant for the ES 1 S 2^E Q 1 S 2 reaction, k ϩ2 /k Ϫ2 , is 0.04. This small value explains why the quinonoid peak in Fig. 3 is small. S-(2-Oxoheptadecyl)-CoA Enhances the Exchange Rate of the ␣-Proton of L-Serine-1 H NMR was used to determine the extent of the SPT-catalyzed hydrogen-deuterium exchange at C␣ of L-serine. In D 2 O buffer at pH 7.5, L-serine gives four resonance peaks centered at 3.85 ppm and eight resonance peaks around 3.98 ppm, corresponding to three protons with the ABX spin system. The former peaks come from the ␣-proton, and the latter come from the ␤-methylene protons (Fig. 5A).
The spectrum was essentially unchanged when only SPT was added to the solution. However, the further addition of S-(2oxoheptadecyl)-CoA to the solution caused a time-dependent decrease in the ␣-proton signal without altering the intensity of the ␤-protons (Fig. 5, B-E). This clearly shows that the binding of S-(2-oxoheptadecyl)-CoA to the external aldimine enhances ␣-deprotonation of L-serine and is consistent with the observation that S-(2-oxoheptadecyl)-CoA causes accumulation of the quinonoid intermediate ( Fig. 3 and 4).  The rate of the hydrogen-deuterium exchange was quantitatively analyzed as follows. The relative intensities of the ␣-proton signal to the ␤-proton signals under various conditions were plotted versus time in Fig. 6. The time-dependent decay of the ␣-proton signals was well fit to a single exponential function, where A is the initial value of the ␣/␤ ratio (y), usually around 0.485, and k app is the apparent rate constant. The values of k app together with the half-time (t1 ⁄ 2 ) are summarized in Table 2. No decrease in the ␣/␤ ratio was observed in the control experiments (without SPT), confirming that the ␣-deprotonation is catalyzed by the enzyme. When L-serine was incubated with only SPT, a very slow exchange of the ␣-proton of L-serine with t1 ⁄ 2 ϭ 103 Ϯ 6 h was observed. The exchange rate in the presence of CoA was also very slow, with t1 ⁄ 2 ϭ 79.1 Ϯ 14.0 h. On the other hand, when S-(2-oxoheptadecyl)-CoA was added to the reaction solution, t1 ⁄ 2 decreased to 0.75 Ϯ 0.03 h, much shorter than the above cases.
A minimal kinetic scheme for the SPT-catalyzed hydrogendeuterium exchange at the C␣ of L-serine is represented as follows, where E represents SPT, S 1 is L-serine, ES 1 is the SPT-L-serine external aldimine complex, and EQ is the quinonoid intermediate. Assuming that the rate of disappearance of the ␣-proton of L-serine is equal to the rate of the C␣-hydrogen abstraction from ES 1 , the following differential rate equation is obtained, where f denotes the fraction of the C␣-hydrogen species in the mixture of the C␣-hydrogen and C␣-deuterium species. Solving Equation 3, we obtain the following.
Using the values [E] t ϭ 5 M, [S 1 ] ϭ 10 mM, and k app ϭ 2.56 ϫ 10 Ϫ4 s Ϫ1 , k ϩ2 is calculated to be 0.51 s Ϫ1 from Equation 4. This value roughly coincides with the k ϩ2 value of 2.2 s Ϫ1 obtained from the analysis of the time-resolved spectra on the SPT ternary complex formation, showing the consistency of the spectroscopic and NMR analyses described above. Similar experiments were carried out with D-serine. The exchange of the ␣-proton of D-serine with the solvent was slow either when incubated with SPT only (t1 ⁄ 2 ϭ 150 Ϯ 5 h), SPT and palmitoyl-CoA (t1 ⁄ 2 ϭ 456 Ϯ 99 h), or SPT and S-(2-oxoheptadecyl)-CoA (t1 ⁄ 2 ϭ 1948 Ϯ 211 h).
Reaction with the True Substrate-The reaction of the SPT-L-serine external aldimine complex with the true substrate palmitoyl-CoA was followed by 1 H NMR in the same way as the analogue. In the presence of 10 mM L-serine and 5 M SPT, the addition of 1.5 mM palmitoyl-CoA resulted in stoichiometric decreases in the signal intensity of the ␤-methylene proton of L-serine (data not shown), showing that the reaction completely proceeded. The hydrogen-deuterium exchange at C␣, as judged from the ␣/␤ ratio, was 2% for 1.5 mM palmitoyl-CoA. These low levels of exchange indicate that the rate of the C-C bond formation from the quinonoid intermediate is significantly higher than the rate of reprotonation of the quinonoid intermediate (see "Discussion").
Models for the SPT-L-Serine and the SPT-L-Serine-Palmitoyl-CoA Complexes-In order to structurally explain the activation of the ␣-proton of the external aldimine, crystallographic analyses of the enzyme-substrate binary and ternary complexes are required. The crystal structure of the substrate-free form of S. paucimobilis SPT at 1.3 Å has been reported (17). However, the crystals of the complex of this enzyme with substrates (or ana-    (25), ALAS-glycine complex (26), and ALAS-succinyl-CoA complex (26). Therefore, based on the crystal structure of R. capsulatus ALAS complexed with succinyl-CoA, structural models for the SPT-L-serine external aldimine complex and the ternary complex of the SPT-L-serine external aldimine and palmitoyl-CoA were constructed. The active site structure of the SPT model was very similar to that of the crystal structure of ALAS; residues conserved between SPT and ALAS were essentially superimposable with each other (data not shown). These residues include His 159 located at the re face of the PLP-Lys internal aldimine and stacking with the PLP pyridine ring, Asp 231 interacting with PLP N1, and His 234 interacting with PLP O3Ј.
In the initial model for the binary (external aldimine) complex, the C␣-H bond of L-serine was perpendicular to the PLP pyridine ring (the dihedral angle C4Ј-N␣-C␣-H being 90°). After energy minimization, however, the carboxylate group of L-serine approached His 159 and formed a strong hydrogen bond/ionic interaction with N⑀2 of His 159 (Fig. 7). As a result, the dihedral angle of C4Ј-N␣-C␣-H became 159°(149°; hereafter, the values in parenthesis are those of subunit 2). Thus, the C␣-H bond of L-serine deviated by 60 -70°from the theoretically ideal direction for ␣-deprotonation.
The model for the ternary complex (SPT-L-serine external aldimine plus palmitoyl-CoA) is shown in Fig. 8, A and B. Palmitoyl-CoA docks with SPT as a bent structure to insert its thioester into the active center of the enzyme. The adenosine 3Ј,5Ј-bisphosphate moiety of palmitoyl-CoA is seen near the entrance of the active site, and the acyl chain moiety of palmitoyl-CoA is suitably placed into the large pocket, which is rich in hydrophobic amino acid residues. Fig. 8C shows the active site view of the ternary complex. His 159 forms a hydrogen bond with the acyl carbonyl O (thioester O) of palmitoyl-CoA with a distance of 2.8 Å. The carboxyl group of L-serine loses its interaction with His 159 and, instead, forms two hydrogen bonds with the guanidino group of Arg 390 . The C␣-H bond takes a perpendicular orientation to the plane formed by the PLP pyridine ring and the imine of the Schiff base, which is suitable for the ␣-proton abstraction by Lys 265 .
Substitution of His 159 of SPT by Phe Resulted in Inactivation of the Enzyme-For the assessment of the role of His 159 in the catalysis of SPT, the residue was changed to Phe, which was chosen to conserve the aromaticity and the size of the side chain. H159F SPT was stably overexpressed and purified in the same way as the wild type enzyme. In the case of the wild-type enzyme, clear UV-visible and CD spectral changes were observed by the addition of the L-serine, indicating the external aldimine formation. For H159F SPT, although the 420 nm peak increased slightly, the 330 nm peak did not decrease, and the CD spectrum did not change. The apparent K d value for L-serine was 11.1 mM, compared with 1.

DISCUSSION
Acceleration of the ␣-Deprotonation of the External Aldimine by the Substrate Acyl-CoA Analogue-The results presented above clearly show that the addition of the palmitoyl-CoA analogue, S-(2-oxoheptadecyl)-CoA, to the SPT-L-serine complex causes accumulation of the quinonoid intermediate, which is not detectable in the SPT-L-serine complex (Figs. 3 and 4). 1 H NMR studies indicated that S-(2-oxoheptadecyl)-CoA accelerates the ␣-deprotonation of the L-serine moiety in the external aldimine by more than 100-fold (Fig. 6). Apparently, this enhancement in the deprotonation rate is considered to be the cause of the accumulation of the quinonoid intermediate upon the addition of S-(2-oxoheptadecyl)-CoA. Zhang and Ferreira (27) are the first who reported the acceleration of the substrate ␣-deprotonation by the acyl-CoA substrate; they found that the physiological substrate succinyl-CoA and its analogue acetoacetyl-CoA increased the rate of glycine proton removal ϳ250,000and 10-fold, respectively. The activation of the ␣-proton of the external aldimine by the binding of the second substrate acyl-CoA may therefore be a common feature among the ␣-oxamine synthase subfamily enzymes.

␣-Deprotonation as the Prerequisite
Step for the Claisen-type C-C Bond Formation-The catalytic reactions of all PLP-dependent enzymes acting on amino acids involve the formation of the quinonoid intermediate (28,29). For the ␣-oxamine synthase subfamily enzymes, it had been an issue whether the quinonoid intermediate is formed by (A) ␣-decarboxylation or (B) ␣-deprotonation of the substrate amino acid. An early proposal was (A), which was simply assumed because the products lose the carboxyl group present in the substrate amino acids. However, subsequent studies on eukaryotic SPT (21), AONS (30), ALAS (31), and KBL (32) showed that the solvent deuterium is incorporated into the position of the substrate C␣ and favored mechanism (B). The decarboxylation is considered to be catalyzed in the external aldimine formed just after the Claisen-type condensation reaction, in which the carboxyl group is expected to be activated not only by the PLP moiety but also by the carbonyl group derived from the substrate acyl-CoA and situated at the ␤ position to the carboxyl group. This, on the other hand, raises the possibility that the incorporation of the deuterium into C␣ can be a result of the hydrogen-deuterium exchange that occurred at the product-PLP aldimine (VI), since the proton should be activated like the carboxyl group. The present study, which uses an acyl-CoA analogue that stops the catalytic reaction at the step of the C-C bond formation, strongly suggests that the ␣-deprotonation is an event that occurs before the C-C bond formation and supports the notion that ␣-deprotonation is the prerequisite step for the Claisentype condensation reaction.
Estimation of the Effect of the True Substrate Palmitoyl-CoA-The true substrate palmitoyl-CoA reacts with the quinonoid intermediate, leading to the formation of the final product, KDS and CO 2 . Even a transient accumulation of the quinonoid intermediate was not observed when palmitoyl-CoA was added to the SPT-L-serine external aldimine. Therefore, it is difficult to carry out a kinetic analysis of the C␣ deprotonation in the presence of palmitoyl-CoA. However, by combining the observation of the hydrogen-deuterium exchange of L-serine in the presence of palmitoyl-CoA and the result of the kinetic analysis of the reaction with S-(2- The ␣-proton of L-serine (green) took a perpendicular orientation to the plane formed from the PLPpyridine ring and the double bond of the Schiff base, suitable for the ␣-proton abstraction by the Lys 265 ⑀-amino group located at a distance of 3.14 Å. Panels were generated using PyMOL version 0.99. oxoheptadecyl)-CoA, we can gain some understanding of the reaction of the SPT-L-serine external aldimine with palmitoyl-CoA.
For the hydrogen-deuterium exchange reaction of L-serine in the presence of palmitoyl-CoA, Scheme 2 is rearranged into Scheme 3, SCHEME 3 where SH and SD represent the hydrogen and deuterium forms of L-serine, respectively, P is palmitoyl-CoA, and K is KDS. Hydrogen removed as a proton from E⅐SH⅐P is considered to be rapidly diluted by the solvent D 2 O, and, for that reason, there would be no protonation of Q to regenerate E⅐SH⅐P. The K d Ser value (1.4 mM) is taken from the previous study (18). The other parameters were not experimentally obtained; therefore, we made the following assumptions. The K d PalCoA , k ϩ2 , and k Ϫ2 values were considered to be the same as the corresponding K d , k ϩ2 , and k Ϫ2 values obtained from the reaction with S-(2-oxoheptadecyl)-CoA. The kinetic deuterium isotope effect for the C␣ deprotonation of the external aldimine was assumed to be 7.3, based on the report by Onuffer and Kirsch (33). Accordingly, the K d and decreased the extent of the exchange. It was found that in order to explain the less than 2% exchange, k ϩ3 must have a value greater than 75 s Ϫ1 . This value is much greater than the k ϩ2 value (2.2 s Ϫ1 ) and exceeds the value of k Ϫ2 (62 s Ϫ1 ). This indicates that the quinonoid intermediate is rapidly converted into the product and is consistent with the finding that no apparent accumulation of the quinonoid intermediate was observed for the reaction of SPT with L-serine and palmitoyl-CoA. It also indicates that the C␣ deprotonation is the ratelimiting step of the overall reaction of SPT.
Structural Considerations Based on the Model-Although the crystal structure for the SPT-substrate complex has not been obtained so far, the high homology of SPT to other enzymes of the ␣-oxamine synthase subfamily enabled us to construct the models for the SPT-substrate binary and ternary complexes. The models shown in Figs. 7 and 8 provided a plausible mechanism that explains why the SPT-L-serine external aldimine is essentially inert to deprotonation, whereas the binding of the second substrate palmitoyl-CoA causes instantaneous ␣-deprotonation.
The external aldimines of PLP and substrate amino acids in the PLP enzymes generally have a nearly planar structure, in which the protonated imine N makes an intramolecular hydrogen bond with O3Ј of PLP and the imine bond lies in the same plane as the pyridine ring. However, the bond around N␣-C␣ can freely rotate, and the external aldimine can take various conformations at this bond. The PLP enzymes exploit this partial conformational flexibility of the external aldimine to exert their reaction specificity. Among the three bonds around C␣ (the bonds attached to C␣ except N␣-C␣), the bond that is most perpendicular to the imine-pyridine plane has the largest overlap of its HOMO with the LUMO of the -orbital of the imine-pyridine conjugate system and is expected to be preferentially cleaved. This hypothesis, first proposed by Dunathan (34), has been proven to apply to all of the catalytic reactions of the PLP enzymes whose crystal structures have been so far elucidated (29). In the model for the SPT-L-serine external aldimine, the C␣-H bond approaches the imine-pyridine plane (Fig. 7). This conformation is quite unfavorable for cleaving the C␣-H bond (schematically shown in Fig. 9A). The observations that the binding of L-serine alone does not cause accumulation of the quinonoid intermediate (Fig. 3) and that the external aldimine exchanges ␣-hydrogen with the solvent at a very slow rate can be understood by the low HOMO (C␣-H)-LUMO (PLP) interaction in this conformation of the external aldimine. The principal factor that makes the external aldimine take this FIGURE 9. Proposed conformations of the pyridoxal aldimines of L-serine and D-serine as viewed along the C␣-N␣ bond. The aldehyde carbon of PLP is shown as Cp, and the atoms of the pyridine ring are indicated by the rectangular box. In both cases of the SPT-L-serine external aldimine and the SPT-D-serine external aldimine, the C␣-H bond of serine is not perpendicular to the imine-pyridine conjugate system; therefore, the ␣-deprotonation is prevented. By the binding of the second substrate palmitoyl-CoA or its analogue S-(2-oxoheptadecyl)-CoA, the configuration of L-serine changes to the geometry appropriate for activation of the ␣-proton (top). On the other hand, in the SPT-D-serine-acyl-CoA ternary complex, the C␣-CH 2 OH bond, not the C␣-H bond, becomes perpendicular to the imine-pyridine plane. Therefore, the acyl-CoA binding does not promote ␣-deprotonation of D-serine (bottom).
conformation is the strong hydrogen bond/ionic interaction between the carboxyl group of L-serine and N⑀2 of His 159 (Figs. 7 and 9A).
In the ternary complex model (Fig. 8), however, the thioester O of palmitoyl-CoA forms a stable hydrogen bond with the N⑀2 atom of His 159 , displacing the carboxyl group of L-serine. The L-serine carboxyl group, in turn, forms hydrogen bonds/ionic interaction with the guanidino group of Arg 390 . This interaction fixes the PLP-L-serine external aldimine to a new conformation, in which the C␣-H bond is almost perpendicular to the imine-pyridine plane and approaches the ⑀-amino group of Lys 265 (schematically shown in Fig. 9A). Both factors are preferable for the ␣-deprotonation to proceed, and this properly explains the acceleration of the ␣-deprotonation of the external aldimine by palmitoyl-CoA.
Based on these models, the phenomenon that the exchange rate of ␣-proton of D-serine in both binary and ternary complex was extremely slow is explained as well (Fig. 9B). Assuming that the carboxyl group of D-serine is recognized by His 159 in the SPT-D-serine binary complex and by Arg 390 in the SPT-D-serine-acyl-CoA ternary complex, the C␣-H bond is not perpendicular to the imine-pyridine plane in either of the two complexes (Fig. 9B). These conformations are clearly unfavorable for the ␣-deprotonation.
Importantly, Webster et al. (24) have already presented a mechanism similar to this for the reaction of AONS. They interpreted the crystal structure of the AONS-product external aldimine to imply that His 133 , which corresponds to His 159 of SPT, makes a hydrogen bond/salt bridge to the carboxylate group of the alanine moiety of the external aldimine and deviates the ␣-hydrogen from the ideal orientation, whereas the additional binding of the pimeloyl-CoA substrate would disrupt this interaction, allowing the C␣-H bond to be perpendicular to the imine-pyridine plane.
Comparison with Other ␣-Oxamine Synthase Subfamily Enzymes and the Role of the Active Site Residues-The above findings suggest the critical role of His 159 in the regulation of the ␣-deprotonation of L-serine during the course of the catalytic reaction of SPT. This amino acid residue is conserved among all members of the ␣-oxamine synthase subfamily: His 133 in AONS, His 136 in KBL, and His 142 in ALAS.
In most PLP enzymes belonging to the fold type I, which includes the ␣-oxamine synthase subfamily, the residue that exists at the re face of the PLP-Lys internal aldimine is an aromatic amino acid, such as Trp, Tyr, and Phe. The presence of His instead of these aromatic residues may be reasonable, considering the ability of His to form hydrogen bonds with the carboxyl group of the amino acid substrate and the acyl carbonyl O of the acyl-CoA substrate and its role in the precise control of the ␣-proton activation. Our finding that H159F SPT does not seem to form the external aldimine with L-serine and lose its catalytic activity supports the idea that His 159 is important as an anchoring residue of the ␣-carboxylate of the L-serine moiety in the external aldimine. However, Trp has an N atom (N⑀1) corresponding to N⑀2 of His and still is not found in the ␣-oxamine synthase subfamily enzymes. The advantage of His over Trp is that it can carry a dissociable proton at N⑀2. This proton can activate the thioester bond by simply migrating to the acyl carbonyl O during the nucleophilic attack of the carbanion of the quinonoid intermediate (Fig. 10). It can also stabilize the carboxyl group of the amino acid substrate by neutralizing the negative charge. Since the carboxylate group is almost perpendicular to the imine-pyridine plane and is expected to be activated in the external aldimine, this interaction may be important for avoiding decarboxylation of the substrate. In fact, it has recently been reported that, if the alanine moiety of the external aldimine of AONS is replaced by trifluoroalanine, which has more electron-withdrawing substituent the external aldimine undergoes decarboxylation, indicating the potential lability of the C␣-COO Ϫ bond (35). Most recently, on the mouse ALAS2 isozyme, Hunter et al. (36) reported the possibility that His 207 , which corresponds to His 142 of Rhodobacter ALAS, would act as an acid catalyst at the decarboxylation step of the ␣-amino-␤-ketoadipate intermediate to form 5-aminolevrinate by carrying out the single and multiple turnover rapid scanning stopped-flow experiments.
In summary, the present results clearly demonstrate that the ␣-deprotonation of the SPT-L-serine external aldimine does occur before the Claisen-type C-C bond formation and is strictly controlled by the presence/absence of the second substrate palmitoyl-CoA. The significance of this mechanism is that SPT escapes abortive transamination by minimizing the accumulation of the quinonoid intermediate, which, by accidental protonation at C4Ј, yields ketimine and subsequently pyridoxamine 5Ј-phosphate and hydroxypyruvate. Furthermore, the model building study strongly suggested that His 159 plays important roles in this control mechanism. The proposed mechanism will be tested in future crystallographic and mutagenesis studies.