Kinetic Mechanism and the Rate-limiting Step of Plasmodium vivax Serine Hydroxymethyltransferase*

Background: Plasmodium vivax serine hydroxymethyltransferase (PvSHMT) catalyzes formation of glycine from l-serine and tetrahydrofolate. Results: Results indicate that PvSHMT can bind to either substrate first. The rate constant of glycine formation is similar to kcat. Conclusion: PvSHMT reaction occurs via a random-order mechanism and glycine formation is the rate-limiting step. Significance: The data are useful for future investigation on inhibition of SHMT for antimalarial drug development. Serine hydroxymethyltransferase (SHMT) is a pyridoxal 5′-phosphate (PLP)-dependent enzyme that catalyzes a hydroxymethyl group transfer from l-serine to tetrahydrofolate (H4folate) to yield glycine and 5,10-methylenetetrahydrofolate (CH2-H4folate). SHMT is crucial for deoxythymidylate biosynthesis and a target for antimalarial drug development. Our previous studies indicate that PvSHMT catalyzes the reaction via a ternary complex mechanism. To define the kinetic mechanism of this catalysis, we explored the PvSHMT reaction by employing various methodologies including ligand binding, transient, and steady-state kinetics as well as product analysis by rapid-quench and HPLC/MS techniques. The results indicate that PvSHMT can bind first to either l-serine or H4folate. The dissociation constants for the enzyme·l-serine and enzyme·H4folate complexes were determined as 0.18 ± 0.08 and 0.35 ± 0.06 mm, respectively. The amounts of glycine formed after single turnovers of different preformed binary complexes were similar, indicating that the reaction proceeds via a random-order binding mechanism. In addition, the rate constant of glycine formation measured by rapid-quench and HPLC/MS analysis is similar to the kcat value (1.09 ± 0.05 s−1) obtained from the steady-state kinetics, indicating that glycine formation is the rate-limiting step of SHMT catalysis. This information will serve as a basis for future investigation on species-specific inhibition of SHMT for antimalarial drug development.

Plasmodium SHMT has been shown to be essential for parasite growth and development, making it a prime target for antimalarial drug chemotherapy development (11). The recombinant expression and purification of Plasmodium falciparum (PfSHMT) and Plasmodium vivax SHMT (PvSHMT) as well as their biochemical characterizations have been reported (5,(12)(13)(14). In general, both of these enzymes are similar in their enzymatic properties. The crystal structure of PfSHMT has been recently solved, revealing the unique function of a cysteine pair at the folate binding pocket in regulating enzyme function through variation of the redox states of these cysteines. This regulation is different from the mammalian and bacterial SHMTs (15). Moreover, it has been shown that Plasmodium SHMTs display several properties that are distinct from the human enzyme, including the inactivation kinetics toward thiosemicarbazide (16). This implies that it is possible to design species-specific inhibitors for anti-malarial drug development. An in-depth understanding of Plasmodium SHMT kinetics and mechanism is therefore needed so that the differences in reaction details among SHMTs can be understood and used for the development of specific inhibitors.
Although the reaction of SHMT from various species has been investigated, many of these studies focused on the nonphysiological reaction, or the reverse reaction of glycine and CH 2 -H 4 folate (2,7,8,17,18). The pre-steady-state kinetics of the H 4 folate-dependent SHMT reaction has never been investigated. Double-reciprocal plots of bi-substrate kinetics of SHMT from rabbit liver cytosol, P. vivax, and P. falciparum display intersecting lines, suggesting that the reaction occurs via a ternary complex mechanism in which the formation of an enzyme⅐serine⅐H 4 folate complex is required for catalysis (5,12,17). However, it was unclear whether the substrates bind in a compulsory or random order. Although the binding of L-serine or other amino acids to SHMT can be clearly monitored by spectroscopic detection based on the formation of an external aldimine, until now, evidence supporting direct binding between SHMT and H 4 folate has not been demonstrated.
In this study, we employed various methodologies including ligand binding measurements, as well as transient and steady-state kinetics to investigate the PvSHMT reaction. Upon anaerobic titration of H 4 folate into the enzyme solution, a spectroscopic signal resulting from the direct binding of PvSHMT and H 4 folate was detected. Results from rapid-quench and HPLC/MS analysis and steady-state kinetics have clearly established glycine formation as the rate-limiting step for the overall reaction of PvSHMT.

EXPERIMENTAL PROCEDURES
Reagents-All chemicals and reagents used were analytical grade and of the highest purity commercially available as described previously (5). An expression plasmid for His 6tagged FAD-dependent NAD(P)H:5,10-methylenetetrahydrofolate oxidoreductase (His 6 -tagged MTHFR) was kindly provided by Dr. Elizabeth E. Trimmer, Grinnell College, IA. Concentrations of (6S)-tetrahydrofolate (H 4 folate) were determined based on its reaction with formic acid according to the protocol described previously (5,19,20). In brief, a solution of H 4 folate was added to formic acid (99%) containing 0.5 mM EDTA and 1 mM DTT and the reaction mixture was incubated in boiling water for 5 min. The solution was allowed to cool, and the absorption of the resulting stable compound, 5,10methenyltetrahydrofolate (CH ϩ -H 4 folate; ⑀ 350 ϭ 26 mM Ϫ1 cm Ϫ1 ) was measured. Based on the assumption that all of the H 4 folate is converted to CH ϩ -H 4 folate, the determined concentration of CH ϩ -H 4 folate is equivalent to the concentration of H 4 folate. Concentrations of PvSHMT and NADH were determined using their molar absorption coefficients: NADH, ⑀ 340 ϭ 6.22 mM Ϫ1 cm Ϫ1 ; PvSHMT, ⑀ 422 ϭ 6.37 mM Ϫ1 cm Ϫ1 (5).
PvSHMT Preparation and Activity Assay-The expression and purification of recombinant PvSHMT were performed as previously reported (5,14). In brief, the expression plasmid containing pET17b-pvshmt was transformed into Escherichia coli BL21(DE3) and induced by the autoinduction method to express native PvSHMT without tag at 16°C in ZYP-5052-rich medium (5 mM Na 2 SO 4 , 2 mM MgSO 4 , 1ϫ NPS (25 mM Na 2 HPO 4 , 25 mM KH 2 PO 4 , 50 mM NH 4 Cl), 1ϫ 5052 (0.5%(w/v) glycerol, 0.05%(w/v) D-glucose, and 0.2%(w/v) ␣-lactose) containing 50 g/ml of ampicillin for overnight (ϳ16 -18 h). Cells were disrupted by ultrasonication and then centrifuged to obtain crude extract. The enzyme was purified to homogeneity using polyethyleneimine precipitation, DEAE-Sepharose chromatography, and SP-Sepharose chromatography. The activity of purified PvSHMT was assayed at 25°C under anaerobic conditions by coupling its reaction with the reaction of His 6 -tagged MTHFR (5,21,22). In brief, a mixture of enzyme solution containing PvSHMT (1 M) and His 6tagged MTHFR (3 M) in 50 mM HEPES, pH 7.0, containing 0.5 mM EDTA, and 1 mM DTT was mixed with a substrate solution containing NADH (100 M), L-serine (2 mM), and H 4 folate (400 M) at 25°C under anaerobic conditions by using a stoppedflow spectrophotometer (TgK Scientific instruments, models SF-61DX2 or SF-61SX). A stopped-flow machine was used for steady-state assays to maintain anaerobic conditions of the experiments. The anaerobic condition was achieved by rinsing the flow circuits with O 2 -scrubbing solution (dithionite). Before experiments, the flow system was washed with anaerobic buffer. All reagents were prepared in an anaerobic glovebox (Ͻ5 ppm O 2 ). The reaction was measured by following the consumption of NADH, for which the absorbance at 375 nm was monitored to avoid interference from absorbance of H 4 folate. Slopes of the kinetic traces were analyzed by Program A (developed by Chun-Jen Chiu, Rong Chang, Joel Dinverno, and David P. Ballou from the University of Michigan, Ann Arbor, MI) and the initial velocities of PvSHMT were calculated using a molar extinction coefficient of 1.92 mM Ϫ1 cm Ϫ1 at 375 nm. One unit of PvSHMT was defined as the amount of the enzyme used for oxidizing 1 mol of NADH per minute at 25°C under pH 7.0.
Steady-state Kinetics of the PvSHMT Reaction at pH 8.0 -Two-substrate steady-state kinetics was performed and analyzed according to a protocol similar to that described in a previous report (5), except the reactions were investigated at pH 8.0 instead of pH 7.0. The study was carried out at pH 8.0 because PvSHMT was found to have the highest activity at this pH (5).
Binding of H 4 folate to PvSHMT-To determine whether PvSHMT can directly bind H 4 folate without requiring prior binding of L-serine, we investigated the binding of PvSHMT and H 4 folate under anaerobic conditions to avoid interference signals from H 4 folate oxidation. An anaerobic solution (1 ml) of PvSHMT (ϳ47 M; A 422 ϳ0.3) in 50 mM HEPES, pH 8.0, containing 0.5 mM EDTA and 1 mM DTT prepared inside the anaerobic glove box (Ͻ5ppm oxygen; Belle Technology, UK) was first scanned to record an enzyme spectrum before adding various concentrations of H 4 folate (0.050 -2.954 mM). After adding each concentration of H 4 folate, the mixture was thoroughly mixed and left for 5 min before recording a spectrum again. The spectrum of free enzyme was subtracted from all spectra obtained from each titration to determine the absorbance changes (⌬A). The ⌬A values were plotted against the H 4 folate concentration and the K d for the binding of H 4 folate to PvSHMT was determined according to Equation 3, where ⌬A represents the absorbance change, ⌬A max is the maximum absorbance change, [L] free is a concentration of free ligand, and K d is a dissociation constant of the enzyme⅐ligand complex. The analysis was done using Marquardt-Lavenberg algorithms in the KaleidaGraph program version 4.0.
Kinetics for the Binding of PvSHMT with L-Serine-A solution of 31 M PvSHMT was mixed with various concentrations of L-serine (0.4 -12.8 mM) in 50 mM HEPES, pH 8.0, containing 0.5 mM EDTA and 1 mM DTT at 25°C using a single-mixing mode stopped-flow spectrophotometer. The binding reaction was monitored at 435 nm where it gives a large absorbance change due to the formation of the external aldimine of the serine⅐PLP complex. All kinetic traces were analyzed using Program A. Observed rate constants (k obs ) associated with each kinetic phase were plotted against the concentrations of L-serine and analyzed according to Equations 4 and 5. k f and k r are the forward and reverse rate constants for the free enzyme isomerization prior to the ligand binding according to the conformational selection model (23), k max is the maximum observed rate constant, k intercept is the rate constant at an ordinate intercept, K S represents a concentration of substrate that is a half-value of k max . Detection of glycine and serine was carried out either via phenylisothiocyanate amino acid derivatization and monitored by HPLC/UV (Agilent 1100) or directly detected by HPLC/MS analysis (Agilent 1260 equipped with Agilent 6120 single quadrupole mass spectrometer). Samples from rapid-quench flow experiments were transferred into a Microcon ultrafiltration unit and then centrifuged at 12,290 ϫ g at 4°C for 20 min to separate the small molecules from the macromolecules. The derivatization method and HPLC/UV analysis were performed according to the protocol described previously with slight modifications (24). For HPLC analysis, the dried samples were redissolved in 50 l of solvent A (0.14 M sodium acetate, pH 6.4, containing 0.05% TEA) and centrifuged at 27,670 ϫ g at room temperature for 30 min to clarify the samples. A clear supernatant was injected onto a C18 column (Waters; 3.9 ϫ 150 mm) that was pre-equilibrated with 100% solvent A with a flow rate of 1 ml/min. The column was washed with 100% solvent A for 15 min, and then a gradient of 0 -53% of solvent B (60% acetonitrile/H 2 O) was applied, in which B was gradually increased over the course of 10 min and maintained at the highest concentration for 5 more minutes. A gradient of 53-100% solvent B was applied over the next 5 min and held for another 5 min before re-equilibrating the column in 100% solvent A. Under these conditions, derivatized serine and glycine can be clearly separated and showed retention times of 7.9 and 9.1 min, respectively. Similar amino acid derivatization as described was applied to standard glycine and L-serine solutions (1-30 M) and a standard curve was generated by plotting the averaged peak areas as a function of glycine or L-serine concentrations.
For the direct sample analysis using HPLC/MS, samples collected from rapid-quench flow experiments were filtered through a Microcon ultrafiltration unit by centrifugation at 12,290 ϫ g for 20 min. The filtrate was injected onto a Zorbax-NH 2 column (Agilent Technology; 4.6 ϫ 250 mm, 5 m) that was pre-equilibrated with 80% solvent C (acetonitrile ϩ 0.1% (v/v) formic acid) with a flow rate of 0.5 ml/min for 60 min. The column was then washed with 80% solvent C for 2 min before the eluent was gradually changed to 60% solvent C over the course of 3 min and further held at this condition for 15 min. The percentage of solvent C was then gradually increased to 80% over 22 min and held at this condition for 10 more minutes to re-equilibrate the column for new injections. HPLC chromatograms of glycine and serine showed retention times of 14.2 and 15.2 min, respectively. The eluent from the HPLC column was continuously flowed into the mass spectrometer equipped with an API-electrospray ionizer. Conditions for mass spectrometric detection were as follows. Nebulizer pressure was set at 35 p.s.i. The capillary source voltage was set at 3 kV. The drying gas temperature was maintained at 350°C. Drying gas flow was 12 liters min Ϫ1 . Scan mode was from 50 to 150 m/z. The polarity was set in a positively selected ion mode in which glycine and serine were detected separately at 76 and 106 m/z, respectively. The peak area for glycine and serine was quantified based on glycine and serine standard curves, for which the amino acid was treated with the same condition as those from the rapid quenched flow experiments. The amounts of glycine product formed at different time points were analyzed using a single exponential equation as shown in Equation 6, where A is the amount of glycine at any time, A 0 is the total amount of glycine that can be produced from the reaction, k obs is the observed rate constant for the glycine formation and t is the reaction time.
For analysis of the products resulting from the different types of mixing setups previously described, similar reactions as those performed in the stopped-flow experiments were carried out in a rapid quench flow machine, and the reactions were quenched at 10 s to measure glycine formation and to quantify the remaining L-serine. Measurement of L-serine and glycine was carried out as described above.

RESULTS
Bi-substrate Steady-state Kinetics of PvSHMT Reaction at pH 8.0 -As PvSHMT activity is the highest at pH 8.0 (5), we decided to re-investigate steady-state kinetics of PvSHMT at this pH. A solution of PvSHMT and His 6 -tagged MTHFR was mixed with various concentrations of L-serine and H 4 folate using the stopped-flow spectrophotometer under anaerobic conditions at pH 8.0 ("Experimental Procedures"). A doublereciprocal plot of e/v versus 1/L-serine shows a series of convergent lines at various concentrations of H 4 folate, indicating that the reaction proceeds via a ternary complex mechanism as the reaction that takes place at pH 7.0 ( Fig. 1). Steady-state kinetic parameters for the PvSHMT reaction at pH 8.0 are similar to those at pH 7.0 (5): K m of L-serine (pH 8.0 ϭ 0.26 Ϯ 0.04 mM, pH 7.0 ϭ 0.18 Ϯ 0.03 mM), K m of H 4 folate (pH 8.0 ϭ 0.11 Ϯ 0.01 mM, pH 7.0 ϭ 0.14 Ϯ 0.02 mM), and k cat (pH 8.0 ϭ 1.09 Ϯ 0.05 s Ϫ1 , pH 7.0 ϭ 0.98 Ϯ 0.06 s Ϫ1 ).
Thermodynamics and Kinetics of Binding of PvSHMT with L-Serine-Various concentrations of L-serine (0.16 -145 mM) were added to a solution of PvSHMT (31 M), pH 8.0, and the absorption changes were monitored. Results indicated that the enzyme binds L-serine and forms an external aldimine as an increase in absorbance at 435 nm could be observed (Fig. 2). The data analysis yielded a thermodynamic dissociation constant (K d ) of 2.2 Ϯ 0.5 mM for the PvSHMT⅐L-serine complex (inset of Fig. 2).
To gain insights into the binding mechanisms, stopped-flow spectrophotometry was used for investigating the kinetics of binding of L-serine to the enzyme. Kinetic traces of the absorption changes at 435 nm (Fig. 3) resulting from formation of the PvSHMT and L-serine complex showed three phases. The traces were analyzed to obtain the observed rate constants (k obs ) of each phase. k obs of the first phase (0.001-10 s) was hyperbolically dependent on the L-serine concentration (inset  Fig. 3). k obs of the second phase (10 -100 s) was independent of the L-serine concentration, with a rate constant of ϳ0.015 s Ϫ1 . For the third phase (100 -2500 s), k obs values hyperbolically decreased upon increasing the L-serine concentration, with the apparent lowest rate constant of 0.0020 Ϯ 0.0001 s Ϫ1 and the L-serine concentration that gives a half-minimum value of 2 Ϯ 1 mM (data not shown). Altogether, the data suggest that binding of PvSHMT and L-serine proceeds through three phases with the second and third phases not involved in the catalytic reaction because they are much slower than the k cat value (1.09 Ϯ 0.05 s Ϫ1 , Fig.  1). The first phase is likely the binding of L-serine to PvSHMT to form a PvSHMT⅐L-serine complex (ES), which then isomerizes to form the ES* complex with a rate constant of 0.30 Ϯ 0.04 s Ϫ1 and an overall K d value for the ES* complex formation was determined as 0.18 Ϯ 0.08 mM (Fig. 8A). This K d value is different from the value obtained from the plot in inset of Fig. 2 (2.2  mM), which was obtained from absorption signal changes after 15 min. The observed discrepancy between the K d values obtained from the two techniques implies that the equilibrium K d value (2.2 mM, inset of Fig. 2) does not represent the binding event that is involved in the productive pathway. It is likely that in the absence of H 4 folate, after the first step of binary complex formation the PvSHMT⅐L-serine complex proceeds to the unproductive path or promiscuous activity to give a large absorbance signal change (second and third phases of Fig. 3).
Binding of PvSHMT and H 4 folate-Although the binding of L-serine and SHMT to form an external aldimine complex (as demonstrated in the previous section) is clearly evident, the interaction of SHMT with H 4 folate has never been documented. We therefore investigated the binding of PvSHMT and H 4 folate under anaerobic conditions by titrating PvSHMT (47 M) with various concentrations of H 4 folate (0.050 -2.954 mM) at pH 8.0. The results indicated that the enzyme can also bind to H 4 folate to form a binary complex of PvSHMT⅐H 4 folate because spectra (Fig. 4) of solutions of PvSHMT plus H 4 folate and free PvSHMT are clearly different, especially around 477 nm (inset A of Fig. 4). The data were analyzed as described under "Experimental Procedures" to yield a K d value of 0.35 Ϯ 0.06 mM (inset B of Fig. 4).
The binding experiment was also carried out using stoppedflow spectrophotometry under anaerobic conditions to monitor the kinetics of PvSHMT and H 4 folate binding. Due to the very small change in signal, reliable kinetic data could not be obtained (data not shown). Nonetheless, the final spectral changes and the K d value of H 4 folate binding (0.25 Ϯ 0.13 mM) obtained from the stopped-flow experiment are similar to that from the static titration experiments (inset B of Fig. 4). These data clearly indicate that L-serine and H 4 folate can independently bind to PvSHMT and form a PvSHMT⅐L-serine or PvSHMT⅐H 4 folate binary complex, suggesting that the kinetic mechanism of PvSHMT is a random-order type.
Single Turnover Reaction of PvSHMT⅐L-Serine Binary Complex and H 4 folate-When a pre-equilibrated complex of PvSHMT⅐L-serine (45/30 M) was mixed with various concentrations of H 4 folate (0.2-6.4 mM) and the absorbance was monitored at 496 nm, the reactions showed five kinetic phases (Fig.  5A). The first phase (0.004 -0.07 s) was the phase with a large   increase in absorbance at 496 nm, whereas the second phase (0.07-0.2 s) showed a small absorbance decrease, indicating the formation and decay of a quinonoid species, respectively. The third phase (0.2-0.6 s) was a small lag phase that had no significant absorbance change. The fourth phase (0.6 -4.0 s) was the phase where the re-formation of a quinonoid species occurred. Interestingly, two different phenomena were observed for the fifth phase depending on the concentration of H 4 folate. For the reaction carried out in the presence of 0.2-0.8 mM H 4 folate, the fifth phase (4.0 -300 s) was a slow decay of the quinonoid species, whereas for the reaction with 1.6 -6.4 mM H 4 folate, the intermediate was stabilized until 1000 s. The final spectra for the various concentrations of H 4 folate were also recorded (data not shown).
All kinetic traces were analyzed as described under "Experimental Procedures" and the observed rate constants (k obs ) of each phase were plotted against H 4 folate concentrations. The plot in Fig. 5B showed that the k obs of the first phase decreased upon increasing the concentration of H 4 folate. This result may be explained according to a model in which an isomerization of the enzyme⅐L-serine complex exists and only one form of the complex reacts with H 4 folate (23,25). This explanation agrees well with the results found in the previous section, showing that a binary complex of PvSHMT⅐L-serine can be isomerized into several forms of complexes. According to Equation 5, the forward and reverse rate constants for the enzyme isomerization were determined to be 17 Ϯ 1 and 258 Ϯ 180 s Ϫ1 , respectively (Fig. 5B). A decreasing trend upon increasing the H 4 folate concentration gives a half-saturation value of 0.033 Ϯ 0.025 mM (Fig. 5B). Observed rate constants of the second phase (ϳ11 s Ϫ1 ) and the third phase (ϳ1 s Ϫ1 ) were independent of H 4 folate concentration (Fig. 5A). The third phase was later assigned as the formation of glycine (a dotted line with filled circles of Fig.  5A). The last two phases showed a decrease in k obs upon increasing the H 4 folate concentration. Based on the currently available data, the identities of the intermediates formed during the fourth and fifth phases cannot be unequivocally assigned. Because their rate constants (Ͻ0.3 s Ϫ1 ) are significantly less than the k cat value (1.09 Ϯ 0.05 s Ϫ1 ), these phases clearly are not relevant to the catalytic turnovers. They may belong to the binding of H 4 folate to free PvSHMT or PvSHMT⅐glycine after product formation.

Detection of Glycine Product Formation Using Rapid Acidquenched Flow/HPLC-UV or HPLC-MS Techniques-Although
the stopped-flow experiments in the previous section showed the kinetics of quinonoid species formation, the data could not identify the step involved with glycine formation. Therefore, a rapidquench experiment was employed to elucidate the step that is involved in the C ␣ -C ␤ bond cleavage in L-serine to form glycine. A solution of PvSHMT⅐L-serine (45/30 M) binary complex was mixed with a low or high concentration of H 4 folate (0.4 or 3.2 mM) using rapid acid-quenched flow-HPLC/UV or HPLC/MS analysis as described under "Experimental Procedures." Glycine (retention time of 14.2 min) and serine (retention time of 15.2 min) were clearly identified and well separated in the HPLC/MS chromatograms (inset of Fig. 6). Glycine concentrations measured from reactions quenched at different time points were plotted against age times (Fig. 6) and analyzed according to a single-exponential equation (Equation 6) as described under "Experimental Procedures." The observed rate constants (k obs ) for glycine formation derived from reactions carried out at different H 4 folate concentrations are similar, 1.02 Ϯ 0.09 s Ϫ1 and 1.24 Ϯ 0.14 s Ϫ1 for 0.4 and 3.2 mM H 4 folate, respectively. It was noted that the total amount of glycine product formed in the reaction with 3.2 mM H 4 folate and detected by  phenylisothiocyanate derivatization was less than that produced from the reaction with 0.4 mM H 4 folate (data not shown). This might be due to low efficiency of glycine derivatization when H 4 folate is present in a high amount.
To confirm if the rapid-quench experiment detected all glycine formed either in the free form or in the E⅐Gly⅐H 4 folate complex, we measured the amount of glycine detected under the rapid-quench conditions in both forms (data not shown). The results indicate that H 4 folate does not have a significant influence on the amount of glycine detected and that under the rapid-quench experiment detected all glycine formed.
When the rapid-quench results were compared with the kinetics of the stopped-flow experiments, the data indicated that the glycine formation phase can be correlated with a small lag period detected as the third phase (ϳ0.2-0.6 s) in Fig. 5A. Furthermore, these k obs values are similar to the catalytic rate constant (k cat ) of the PvSHMT reaction (1.09 Ϯ 0.05 s Ϫ1 ) measured from steady-state kinetics (Fig. 1). These results have therefore, for the first time, unambiguously identified the glycine product formation as the rate-limiting step of the H 4 folatedependent SHMT reaction.

Reaction of PvSHMT with L-Serine and H 4 folate under Different Mixing Conditions-The binding studies revealed that
PvSHMT can independently bind L-serine or H 4 folate (Figs. [2][3][4], implying that the enzyme binds both substrates in a random ordered fashion. However, whether the random ordered binding mechanism is relevant to the PvSHMT catalysis or whether the enzyme requires a specific pathway to form the product was not known. Three different types of stopped-flow mixing setups (see "Experimental Procedures") were employed to pre-form different binary complexes and the absorbance was monitored at 496 nm to follow the formation of a quinonoid species. The appearance of quinonoid absorption was used as an indicative signal for detecting the ternary complex that was committed to catalysis (Figs. 5 and 7) because the quinonoid species has been proposed to be part of the SHMT catalytic cycle (1,2). The absorbance change at 435 nm was also monitored, but the signal change was too small to obtain reliable kinetic data.
Results from the first type of mixing setup (mixing PvSHMT⅐L-serine with H 4 folate, solid lines with filled circles in Fig. 7, A and B) were similar to that shown in Fig. 5A discussed earlier, but different from the other two experiments. For the second experiment in which a solution of PvSHMT⅐H 4 folate binary complex (0.4 and 3.2 mM, solid lines with filled rectangles in Fig. 7, A and B, respectively) was mixed with L-serine, the reactions showed four phases in which all phases displayed an increase in absorbance. The first phase (0.005-0.2 s) of the reaction showed less amplitude change than that of the mixing of PvSHMT⅐L-serine with H 4 folate. k obs values of the reactions with 0.4 and 3.2 mM H 4 folate are 11 and 6 s Ϫ1 , respectively. Although the kinetics of this phase are slower than the first kinetic phase from the mixing of PvSHMT⅐L-serine with H 4 folate, these values are still significantly faster than the ratelimiting step of the reaction (ϳ1 s Ϫ1 , Fig. 5A). The following kinetic phase (0.2-0.6 s) showed an increase in absorbance at 496 nm and is likely to be the glycine formation step that is the rate-limiting step in the overall reaction. The latter part of the reaction (2-1000 s) occurs much slower than the turnover number (ϳ1 s Ϫ1 ), and thus is not relevant to catalysis. For the third experiment in which the free enzyme was mixed with a solution of L-serine and 0.4 and 3.2 mM H 4 folate (solid lines with filled triangles in Fig. 7, A and B), k obs values for quinonoid formation in the first phase (0.005-0.2 s) in the reactions of 0.4 and 3.2 mM H 4 folate were 14.7 and 8 s Ϫ1 , respectively. The second phase of the reaction (0.2-0.6 s) showed an increase of absorbance; this step likely corresponds to the glycine formation step. Other phases following this step (2-1000 s) are slower than the catalytic turnover.
Although k obs values for initial quinonoid formation in these three experiments are different, this step was faster than the glycine formation step (0.2-0.6 s) for all of the reaction setup conditions tested. Moreover, all reactions yielded similar amounts of product (see next section). Therefore, we conclude that all types of mixings can lead to a productive path for the SHMT reaction. As the mixing of free PvSHMT with a mixture of L-serine and H 4 folate resulted in kinetics and amplitude changes that are different from the reactions of either preformed complexes, the data imply that when providing the two substrates simultaneously, the free enzyme binds randomly to both substrates, agreeing with the data shown in Figs. 3 and 4. Based on this study, the overall reaction catalyzed by PvSHMT can be described by the scheme displayed in Fig. 8B.
Comparison of Glycine Generated from Different Reaction Setups-The amount of glycine generated from the three mixing reactions (Fig. 7) was analyzed using rapid-quench and LC-MS techniques. Samples at the reaction times of 10 s were collected and analyzed. Similar amounts of glycine product (ϳ22 M) were obtained from all types of mixing reactions (Table 1). Because the conditions employed only allow for a single turnover of the reactions to proceed, these data indicated that all mixing setups resulted in a similar amount of product after the glycine formation step, suggesting that all mixing setups led to a productive pathway and the PvSHMT reaction occurs via a random-order mechanism (Fig. 8B).
Kinetic Mechanism of PvSHMT-Based on currently available data, the reaction of PvSHMT can be summarized as shown in Fig. 8. PvSHMT binds to either L-serine or H 4 folate as a first substrate. The binary complex then binds to another substrate and proceeds to form glycine. In principle, glycine or CH 2 -H 4 folate should also be released in random order. All data also imply that under steady-state enzyme turnovers in which L-serine is in excess, the glycine release is quick and the rate-limiting step is the glycine formation. However, in the presence of excess H 4 folate such as in Figs. 5 and 7, once glycine forms, it is trapped in the form of the enzyme⅐Gly⅐H 4 folate complex as noted by high absorbance at 496 nm (Fig. 8B).

DISCUSSION
Our investigation based on transient kinetics, thermodynamics, and steady-state kinetics provides insight into the reaction mechanism of PvSHMT and has identified glycine formation as the rate-limiting step of this enzyme reaction. Direct binding experiments under anaerobic conditions has shown for the first time that the free enzyme can independently bind to H 4 folate and that the reaction of PvSHMT obeys a randomorder mechanism in which the enzyme can bind either to L-serine or H 4 folate as the first substrate (Fig. 8B).
Using rapid-quench and HPLC/UV/MS experiments, the rate constant of glycine formation was measured as ϳ1 s Ϫ1 , which is similar to the k cat value of the PvSHMT reaction obtained from steady-state kinetics under the same condition. The data indicate that the C ␣ -C ␤ bond cleavage of L-serine to result in glycine formation is the rate-limiting step for the physiological H 4 folate-dependent reaction of PvSHMT. Recently, quantum mechanics calculation of H 4 folate-independent retro-aldol cleavage of ␤-hydroxyamino acids via the ␤-hydroxyl proton abstraction based on the structure of Bacillus stearothermophilus (Bs) SHMT in complex with L-allo-threonine suggests that the rate-limiting step for the aldolase activity of BsSHMT reaction is also the C ␣ -C ␤ bond cleavage (26). Because formation of the E⅐L-serine⅐H 4 folate ternary complex (evidenced by formation of a quinonoid intermediate, Figs. 5 and 7) occurs before the glycine formation step (Fig. 7), it implies that the presence of H 4 folate is required for the C-C bond breakage. These results agree with findings of our previous investigation that free formaldehyde could not be detected in the absence of H 4 folate (5). The results here also support the mechanism suggested (2,18), which proposes that the binding of H 4 folate may cause a conformational change that facilitates the glycine product formation.  -d). B, a random-order model describing the overall reaction of PvSHMT mechanism. PvSHMT binds randomly either to L-serine or H 4 folate to form an active E⅐Ser or E⅐H 4 folate binary complex. The binding of enzyme and L-serine is a multistep process as described in A. Only a relevant form of the external aldimine complex can bind to H 4 folate to form the ternary complex of E⅐Ser⅐H 4 folate. Glycine is generated with a rate constant of 1 s Ϫ1 . Both products are randomly released. The current data suggest that in the presence of a high concentration of H 4 folate, the E⅐Gly complex can be trapped in the dead-end E⅐Gly⅐H 4 folate complex. Transient kinetics and direct binding experiments suggest that PvSHMT binds L-serine and H 4 folate via a random-order mechanism because the enzyme can bind to each ligand independently. The K d value for H 4 folate binding is 0.35 Ϯ 0.06 mM (Fig. 4), whereas a K d value for L-serine binding relevant to catalysis is 0.18 Ϯ 0.08 mM (Fig. 8A). A key experimental setup that allowed us to directly observe the binary complex of PvSHMT⅐H 4 folate for the first time was conducting the reaction under anaerobic conditions. As H 4 folate can be readily oxidized, resulting in an absorbance increase at the 330 -360 nm regions (5,27,28), it is impossible to aerobically monitor the interaction of H 4 folate and protein interaction without any background interference. Previously, most of the binding studies of SHMT and H 4 folate were only monitored in the presence of amino acid substrates, especially in the presence of glycine, for which an intense band of quinonoid signal could be observed (29 -31). Different mixing setups, in which either PvSHMT⅐L-serine or PvSHMT⅐H 4 folate was preformed, resulted in the same amount of glycine product at the end of a single turnover, supporting the conclusion that both paths of ligand binding lead to the productive pathway (Fig. 8B). A previous study of the rabbit cytosolic SHMT using equilibrium isotope exchange of L-serine and CH 2 -H 4 folate also showed that the rate of isotope exchange under equilibrium conditions was not affected by increasing glycine concentrations, implying that the ordered preference for dissociation of both products, glycine and CH 2 -H 4 folate, is not required (17). All results support that the kinetic mechanism of PvSHMT conforms to a random order model (Fig. 8B).
The binding of PvSHMT and L-serine is complicated and a multistep process in which the PvSHMT⅐L-serine complex is in equilibrium of several forms (Fig. 8A). After the initial binding to form a Michaelis complex of PvSHMT⅐L-serine, the complex must proceed to form the gem-diamine and then the external aldimine complex. When the preformed PvSHMT⅐L-serine was mixed with various concentrations of H 4 folate, the k obs was decreased. These data suggest that the isomerization step preceding the bimolecular reaction with H 4 folate exists. Based on our current data, we cannot distinguish if this isomerization step is caused by the enzyme conformational change or the equilibrium of gem-diamine and external aldimine complexes. In the presence of H 4 folate, it is possible that a PvSHMT⅐H 4 folate complex can readily bind and facilitate the formation of the external aldimine complex.
The random-order binding mechanism of PvSHMT implies that each ligand binds to the enzyme via separate binding pockets and that access to each pocket is independent. Currently, there are seven x-ray structures of SHMT ternary complexes with amino acid and folate analogues (PDB code number 1EQB (32), PDB code number 1DFO (33), PDB code number 1KL2 (34), PDB code number 2VGW (35), PDB code numbers 2W7H and 2W7M (36), and PDB code number and 2W7M (37)) including the recently published structure of PvSHMT in complex with D-serine and H 4 folate analogue (6R-5-CHO-H 4 folate) at 2.4-Å resolution (PDB code 4OYT) (38) (Fig. 9). The crystal structure reveals defined binding pockets for amino acid and folate that could support the independent binding of each substrate via a random-order mechanism as proposed (Fig. 8B).
In conclusion, we have shown for the first time direct evidence supporting that the kinetic mechanism of PvSHMT occurs via a random-order model and that glycine formation is the rate-limiting step of SHMT. This information will serve as a basis for future investigation for other SHMTs. The comparative studies of SHMTs from parasite and host will pave the way for the design of species-specific inhibitors for antimalarial drug development.