Pyridoxal 5′-Phosphate Is a Slow Tight Binding Inhibitor of E. coli Pyridoxal Kinase

Pyridoxal 5′-phosphate (PLP) is a cofactor for dozens of B6 requiring enzymes. PLP reacts with apo-B6 enzymes by forming an aldimine linkage with the ε-amino group of an active site lysine residue, thus yielding the catalytically active holo-B6 enzyme. During protein turnover, the PLP is salvaged by first converting it to pyridoxal by a phosphatase and then back to PLP by pyridoxal kinase. Nonetheless, PLP poses a potential toxicity problem for the cell since its reactive 4′-aldehyde moiety forms covalent adducts with other compounds and non-B6 proteins containing thiol or amino groups. The regulation of PLP homeostasis in the cell is thus an important, yet unresolved issue. In this report, using site-directed mutagenesis, kinetic, spectroscopic and chromatographic studies we show that pyridoxal kinase from E. coli forms a complex with the product PLP to form an inactive enzyme complex. Evidence is presented that, in the inhibited complex, PLP has formed an aldimine bond with an active site lysine residue during catalytic turnover. The rate of dissociation of PLP from the complex is very slow, being only partially released after a 2-hour incubation with PLP phosphatase. Interestingly, the inactive pyridoxal kinase•PLP complex can be partially reactivated by transferring the tightly bound PLP to an apo-B6 enzyme. These results open new perspectives on the mechanism of regulation and role of pyridoxal kinase in the Escherichia coli cell.


Introduction
Pyridoxal 59-phosphate (PLP) is a cofactor for dozens of enzymes in the E. coli cell that are important in amino acid metabolism, as well as in several other pathways [1]. A de novo pathway for PLP biosynthesis occurs in E. coli, but a more important salvage pathway operates to recycle PLP during protein turnover [2]. The regulation of PLP homeostasis in E. coli and how each of the dozens of PLP requiring apo-B 6 enzymes competes for available PLP, to form the catalytically active holo-B 6 enzymes, are important unresolved problems.
Key reactions for maintaining PLP levels in the cell are shown in Figure 1. The biosynthetic pathway results in the formation of pyridoxine 59-phosphate (PNP, reaction 1), which is converted to PLP by the flavin enzyme PNP oxidase (reaction 2) [3]. PLP adds to newly synthesized apo-B 6 enzymes converting them to the catalytically active holo enzymes (reaction 3). PLP is released during protein turnover of B 6 enzymes (reaction 4) and is converted to pyridoxal (PL) by cellular phosphatases (reaction 5) [3][4][5]. The formed PL is phosphorylated back to PLP by pyridoxal kinase (PL kinase) (reaction 6) [3]. Pyridoxine and pyridoxamine are also substrates for PL kinase (reaction 6). PL kinase, PNP oxidase and PLP phosphatase constitute the salvage pathway.
PLP is a reactive compound because the 49-aldehyde forms aldimines with a-amines of amino acids and other compounds containing amino groups, the e-amino group of lysine residues on non-B 6 proteins, and thiazolidine adducts with sulfhydryl groups like cysteine. PL, which also contains the 49-aldehyde moiety is much less reactive than PLP, since its aldehyde group, in aqueous solution and at neutral pH, exists mostly in the hydrated form [6,7]. This reactivity of PLP poses two problems for the cell. First, to keep the cellular level of free PLP low so that it does not react with other nucleophiles, especially non B 6 -enzymes, and second, to supply enough PLP for the dozens of newly synthesized apo-B 6 enzymes to form catalytically active holo-B 6 enzymes. Some have proposed that the level of PLP is regulated and kept low by being an effective feedback inhibitor of both PL kinase [8] and PNP oxidase [9][10][11][12][13][14][15]. We report here on properties of E. coli PL kinase showing that PLP serves as a slow tight binding inhibitor of the enzyme.
The structure of PL kinase has been determined from several sources [11,[16][17][18][19]. In E. coli there are two PL kinases referred to as PL kinase1 and PL kinase2 [13,[20][21][22]. The activity of PL kinase2 is very low and there is a question if its function in the cell is to convert PL to PLP or if it is the enzyme for another unknown reaction. We have determined the structure and properties of both E. coli enzymes [11,13]. This study reports on the properties of only E. coli PL kinase1 and we refer to it as ePL kinase. During our kinetic measurements, we observed that the enzyme rapidly loses activity as it catalyzes the conversion of PL and ATP into PLP and ADP. The reason for this inhibition is the subject of this report.

Inactivation of ePL Kinase during Catalytic Turnover
During our assay of ePL kinase to determine kinetic constants we observed a rapid loss of activity making kinetic studies from initial rates difficult. Figure 2 shows a typical assay for production of PLP. The rate of PLP formation is determined from the absorbance change at 388 nm. At pH 7.5, used in the assay, both PL and PLP have an absorption maximum at 388 nm, however, the extinction coefficient of PLP is much higher. From standard solutions, we calculated 5020 M 21 cm 21 to be the difference between the 388 nm molar extinction coefficients of the produced PLP and the consumed PL. The assay solution contained 1 mM MgATP, 1 mM PL and 0.9 mM ePL kinase. The rate of PLP formation decreased exponentially and after about 2 min the activity was near zero, with only 22 mM PLP being formed. The apparent rate of activity loss was around 1.5 min 21 . Since a large excess of MgATP and PL was present in the assay, and the reaction is essentially irreversible, the termination of activity should not be the result of exhaustion of either substrate. However, it could be the result of product inhibition by ADP and PLP. To test if this was the reason for loss of activity, we added a second and equal amount of ePL kinase to the cuvette. The increase in absorbance at 388 nm resumed at the same rate as observed with the first addition of ePL kinase, showing that product inhibition was not the reason for the loss of activity, since at this point PLP is present in the reaction mixture. Furthermore, this second measurable activity also exponentially decreased at a rate of 1.5 min 21 , so that after about 2 min there was no further increase in PLP production (Fig. 2). A third addition resulted in the same behavior (data not shown). The total absorbance change at 388 nm was the same for each addition of ePL kinase, showing that the amount of PLP produced was the same. We calculated the amount of PLP formed before the enzyme lost activity to be 22 mM, determined by dividing the absorbance change at 388 nm by 5020 M 21 cm 21 . Since we also knew the concentration of the enzyme we could determine that about 25 catalytic turnovers were made before the enzyme was inactivated.
This experiment suggests that the loss of activity is the result of a partition between product release and an inactivation reaction.

Properties of Inactive ePL Kinase
In order to check if the observed inhibition was reversible, we incubated ePL kinase (300 mM) with PL (400 mM) and MgATP (1 mM) for one hour at 37uC. The protein was then separated from excess substrates and products by size exclusion chromatography on a Sephadex G-50 column. As shown in Figure 3, the fractions containing the protein showed an absorbance at 388 nm (beside the 280 nm absorption peak), suggesting that PLP was bound to ePL kinase. Addition of NaOH to a final concentration of 0.2 M to the protein samples resulted in a spectrum identical to the spectrum of PLP in alkaline solution (data not shown). From the observed absorbance maxima at 388 nm and 278 nm for PLP and protein, respectively, we determined that the stoichiometry of PLP to ePL kinase subunits was 1:1.
What is of interest in the elution profile shown in Figure 3 is the lack of a tailing edge in the 388 nm absorption profile of the chromophore that co-elutes with the protein in the size-exclusion chromatography. This suggests that the complex between protein and PLP is very tight and does not readily dissociate upon the slow migration of the ePL kinase down the sizing column. The spectrum of a typical fraction containing protein-bound PLP is shown in the inset of Figure 3 and exhibits two distinct absorbance maxima at 336 nm and 420 nm. The band with maximum at 420 nm is characteristic of PLP bound as an aldimine [23]. The aldimine linkage (or Schiff base) between PLP and a lysine residue is a reversible covalent bond. It is characteristic of PLP bound to an eamino group of an active site lysine residue in holo-B 6 enzymes [24].

Rate of Formation of ePL KinaseNPLP Complex
The considerable separation between the ePL kinaseNPLP complex and free PLP, shown in Figure 3, provides a method to rapidly separate the enzyme from unbound substrates and products by using small spin columns (see Experimental Procedure). To determine the rate at which the complex was formed, ePL kinase (90 mM) was incubated with 400 mM MgATP and 150 mM PL at 37uC from 2 to 25 min. At several time intervals, a sample of the reaction mixture was removed and added to a small spin Sephadex G-50 column, kept at 4uC, to rapidly separate enzyme-bound PLP from unbound small molecules. The absorbance of the eluted enzyme at 420 nm and 336 nm was measured to determine the amount of bound PLP and the percentage of saturation per enzyme subunit. As shown in Figure 4 (open circle) the amount of bound PLP reached about 100% of saturation in about 10 minutes, with 50% saturation occurring in less than 2 min. However, if ePL kinase was incubated with 150 mM PLP, in place of PL, and 400 mM MgATP the amount of bound PLP after 10 minutes was about 60% and reached only 80% of saturation in 25 min. In this case, it took about 7.5 min to reach 50% saturation (Fig. 4, solid circle). This shows that PLP generated by enzyme turnovers leads to faster formation of the complex compared to when PLP is added as an external ligand.
The ePL kinase concentration in the previous experiment with PL and MgATP is high and the amount of time required to process the reaction would convert much of the PL to PLP and an equivalent amount of the MgATP to MgADP, suggesting that the complex may contain both PLP and MgADP. The experiment was repeated with PLP and MgADP, and the rate of formation of the complex was similar to the rate observed with PL and MgATP (Table 1). On the other hand, when PLP and Mg 2+ were used in the absence of any nucleotide the rate was slower and comparable to the rate observed with PLP and MgATP, shown in Figure 4 ( Table 1). This suggests that a tight complex is formed with PLP and MgADP. The spectrum of the ePL kinaseNPLP complexes in these experiments was the same as shown in the inset of Figure 3.
Keeping in mind that ePL kinase also phosphorylates PN and PM, we repeated the same experiments described above, to determine if PNP and PMP also formed a tight complex with ePL  kinase. Our results showed that a 25 min incubation at 37uC resulted in no detectable complex formation with these two vitamers (data not shown).

Nucleotide Ligands in the ePL KinaseNPLP Complex
The experiments described in the previous section had either MgATP or MgADP present at all times during the chromatographic separation of free PLP from the kinaseNPLP complex. To determine how important MgATP, MgADP and MgCl 2 were in forming the complex, and whether the nucleotides were also bound in the complex, we performed the incubation of ePL kinase with several different combinations of these compounds, and carried out the final separation of the enzyme from the small molecules in the absence of purine nucleotides. After the separation, we analyzed the complexes for the presence of PLP, MgATP and MgADP, after complete denaturation by 0.2 M NaOH (see Experimental Procedures). The results are reported in Table 2. PLP was always found to be present in the complex, while ATP was not detected in any of the analyzed complexes. On the other hand, ADP was present in the complexes, but not in a stoichiometric concentration with respect to enzyme subunits. Apparently, ADP is not bound as tightly as PLP to ePL kinase. PLP, in the absence of nucleotides or MgCl 2 , binds tightly to ePL kinase, although its binding is further promoted by the inclusion of exogenous MgCl 2 , suggesting that Mg 2+ plays a role in forming the tight complex (Table 2). These results show that the tight complex in the presence of nucleotides is actually ePL kinaseNPLPNMgADP. However, in the following experiments we focused our attention on the PLP moiety of the complex and therefore will refer to the ePL kinaseNPLP complex.

Rate of Dissociation of the ePL KinaseNPLP Complex
To determine the rate of dissociation of PLP from the ePL kinaseNPLP complex, we took advantage of the property that free PLP does not elicit an optical signal in a CD spectrum because it is a symmetrical molecule. But, when PLP is bound to B 6 enzymes, the asymmetric environment at the active site induces a CD signal in PLP that can be either negative or positive [25]. A ePL kinaseNPLP complex was obtained by incubating 130 mM ePL kinase with 0.5 mM PL and 1 mM MgATP, and separated on a sizing column in the absence of purine nucleotides. The CD spectrum of the obtained ePL kinaseNPLP complex exhibits negative ellipticity bands at about 330 nm and 415 nm (Fig. 5A, curve (a)). To determine the rate of PLP dissociation from the complex we followed the rate of decrease of the CD signal at   415 nm when the complex was incubated with a specific PLP phosphatase (10 mM) [26]. As PLP dissociates from the complex, the phosphatase rapidly converts it to PL and this results in a decrease of the CD signal. The observed kinetics (Fig. 5B) shows that the rate of dissociation is very slow (0.012 min 21 ) and fits very well to a first-order decay curve. The final CD spectrum after 120 min shows that only a fraction of the bound PLP was released (Fig. 5A, curve (b)). Measurement of the residual PLP in the sample showed that about 60% of the initially bound PLP was present even after a 2-hour experiment. Very similar results were obtained using alkaline phosphatase in place of PLP phosphatase. In another experiment, an identical ePL kinaseNPLP complex was reduced with NaBH 4 . This potent reductant is often used to reduce the PLP aldimine double bond (absorbing at 420 nm) to a single bond. The reduction is accompanied by a large spectral change. Also in this case, the reduction of only a small fraction of the enzyme-bound PLP was observed (as a decrease of the 420 nm absorbing band and a concomitant appearance of a band at around 325 nm). Evidently, there is a fraction of tightly bound PLP that does not dissociate from ePL kinase and is not easily accessible from the solvent by NaBH 4 .

Mechanism of Formation of ePL KinaseNPLP Complex
The spectrum of the ePL kinaseNPLP complex shown in Figure 3 shows absorption peaks at 336 nm and 420 nm. These absorbance maxima are also represented in the CD spectra as observed in Figure 5A. The peak at 420 nm is characteristic of a protonated PLP aldimine and therefore clearly indicates that PLP is covalently bound, albeit reversibly to a lysine residue of the enzyme [23]. The absorbance of the peak at 336 nm could also be attributed to a covalently bound PLP, most likely the carbinolamine intermediate which occurs in the aldimine linkage formation (Fig. 6A). One has to keep in mind that an aldimine linkage is a readily reversible covalent bond and therefore PLP is still able to dissociate from the enzyme. The structure of ePL kinase shows the presence of a lysine residue, namely Lys229 (K229), near the active site. In the published crystal structure of ePL kinase with bound PL, the side chain of K229 is about 4.5 Å from PL, but its side chain can rotate to place the e-amino group ,1.6 Å from the aldehyde group of PL. (Fig. 6B) [11]. This suggests a possible conformational change of the lysine side-chain position to form the Schiff-base linkage with the PLP. We have not been able to obtain crystals of the ePL kinaseNPLP complex to ascertain if PLP is actually bound to this lysine residue. Therefore, we decided to change this residue to a glutamine to determine if K229 was critical for catalytic activity, enzyme inactivation, tight binding of PLP and the spectral properties shown in Figures 3 and 5A.
Previous studies may have given erroneous kinetic constants for ePL kinase because they were not appreciative of the rapid loss of activity due to PLP inhibition, as shown in Figure 2. We have more carefully determined the kinetic constants for the wild type enzyme by determining the initial rate in the first few seconds before inhibition becomes a factor. The results are shown in Table 3. The K229Q mutant enzyme was expressed and purified as described for the wild type enzyme. Kinetic constants were determined and are recorded in Table 3. These results show that K229 plays a role in both binding of substrates and catalysis but is not essential for activity. Note that the affinity for MgATP in K229Q enzyme is greater compared to wild type ePL kinase. When the mutant enzyme is incubated with PL and MgATP and passed down a sizing column, no PLP is found tightly bound as shown in Figure 3 for wild type ePL kinase. Furthermore, unlike wild type ePL kinase, incubation of K229Q with PL and MgATP does not result in a rapid loss of activity (Fig. 7).

Reactivation of Inhibited ePL KinaseNPLP Complex
If the inactivation of ePL kinase takes place in an E. coli cell, how does the kinase continue to function in order to generate PLP? Is there a mechanism by which the tightly bound PLP is removed to reactivate the enzyme? Previous studies with PNP oxidase and PLP synthase have been shown to bind PLP tightly. For both enzymes this tightly bound PLP was transferred to an apo-B 6 enzyme, apo serine hydroxymethyltransferase (apo-SHMT) for PNP oxidase and aspartate aminotransferase for PLP synthase [27,28]. To test if an apo-B6 enzyme could also regenerate an active ePL kinase we added purified apo-E. coli SHMT (eSHMT) and determined if the tightly bound PLP on the kinase was transferred to the apo-eSHMT to form the active holo-eSHMT. eSHMT is a good B 6 enzyme to use since the active enzyme forms an abortive complex with product glycine and substrate tetrahydrofolate that exhibits an absorption peak at 495 nm with a large molar absorbance coefficient [27]. The rate of formation of this absorbing complex can be followed continuously with time.
As shown in Figure 8, when an equivalent subunit concentration of apo-eSHMT is added to ePL kinaseNPLP complex, there is a transfer of PLP to form holo-eSHMT. The rate of transfer far exceeds the rate of dissociation of PLP from the complex as determined by CD studies (Fig. 5). Of interest is the observation that only 50% of the tightly bound PLP is transferred, as suggested by the effect of the addition of an excess of free PLP to apo-eSHMT. The addition of a phosphatase did not reduce the amount of transfer of PLP suggesting PLP was being transferred directly without coming into contact with the phosphatases. Although speculative, it is possible the PLP is not even released into the solvent but directly channeled to apo-eSHMT. We observed that some catalytic activity of PL kinase returns upon transfer of the tightly bound PLP to apo-eSHMT but is much less than the expected 50% (data not shown). Future studies are focused on understanding in much more detail the mechanism of transfer of PLP and the reactivation of PL kinase.

Discussion
There are three known enzymes in living systems that catalyze the production of PLP: PNP oxidase [3,29], present in both prokaryotic and eukaryotic organisms; PL kinase, which is also widely distributed in nature [3,11]; and PLP synthase, which is found in plants and many microorganisms [30]. Both PNP oxidase and PLP synthase have been shown to bind PLP tightly and to transfer the tightly bound PLP to an apo-B 6 enzyme [28,31]. This report is the first study on the properties of the formation and dissociation of a tightly bound PLP in ePL kinase.   Three classes of enzyme competitive inhibitors have been described [32]: those which inhibit rapidly, those which inhibit rapidly followed by a slow conformational change, and those which inhibit slowly. Classical competitive inhibitors act rapidly and show a high affinity for the active site of the ground state enzyme, whereas slow binding inhibitors show a high affinity for an intermediate state of the enzyme. Slow binding inhibition is characterized by an initial weak binding to the ground state enzyme, followed by tighter binding to the transition state structure. In general, this type of inhibition is considered more physiologically relevant since upstream accumulation of the substrate cannot relieve the inhibition brought about by this form of inhibition [33].
The results presented here suggest that PLP is a slow tight binding inhibitor of ePL kinase. The mechanism of inhibition consists in the formation of a Schiff base between PLP and an active site lysine residue (Fig. 6A). The inactivation of the enzyme is faster when both PLP and MgADP are present, compared to when PLP is present alone or together with MgATP ( Fig. 4 and Table 1). Therefore, the inhibition occurs more rapidly during the catalytic turnover of the enzyme, in which the enzyme may go through an intermediate state whose conformation favors the covalent binding of PLP. It appears that during the catalytic cycle, or when both PLP and MgADP are bound, the active site of ePL kinase is in a conformation that places the e-amino group of K229 in a favorable position to form a covalent bond with C49 of PLP. The position of K229 in the active site structure of the unliganded ePL kinase is shown in Fig. 6B. Formation of an aldimine between PLP and the e-amino moiety of K229 is suggested by the absorption maximum at 420 nm (Fig. 3 inset) and by the failure of PLP to bind tightly to K229Q ePL kinase and to inhibit its activity (Fig. 7). The 336 nm absorbing band of the tightly bound PLP (Fig. 3, inset) can be accounted for by several possible structures. One of the most probable is a carbinolamine intermediate, which occurs during the formation of the aldimine (Fig. 6A). In the carbinolamine structure, the C49 carbon of PLP is tetrahedral because of the addition of the e-amino moiety of K229 across the double bond to oxygen. Another possible structure is the enolimine tautomer of the PLP protonated aldimine also found at the active site of PLP-dependent enzymes [25,34].
The rate of dissociation of PLP from the ePL kinaseNPLP complex is very slow, as shown by the CD studies in the presence of specific and non-specific PLP phosphatases (Fig. 5A and B). This slow rate cannot account for the order of magnitude faster rate of transfer of the tightly bound PLP to apo-eSHMT.
Our results raise questions about the role of ePL kinase in vivo. The observed inhibition mechanism and the transfer of PLP to apo-B 6 enzymes may be a strategy to tune ePL kinase activity on the actual requirements of the PLP cofactor. Moreover, since PLP is such a reactive compound, having it bound tightly to ePL kinase would afford protection against unwanted side reactions, in which it can be dephosphorylated or form aldimines with free amino acids or eamino groups on lysine residues in non-B 6 proteins. We observed that the tightly bound PLP is protected from dephosphorylation by either a specific PLP phosphatase or alkaline phosphatase. But if protecting PLP from the unproductive side reactions is the purpose of its tight binding, then there must be a mechanism by which PLP is released to activate the newly synthesized apo-B 6 enzymes, restoring the catalytic turnover of the kinase.

Materials
All buffers, reagents and chromatography materials were of the purest grade available and are the same as previously reported [11,20]. ePL kinase was expressed, purified and stored as previously described [11]. Calf intestine alkaline phosphatase was purchased from Fermentas International Inc.

Methods
All experiments were carried out in 20 mM potassium HEPES, pH 7.5, containing 0.2 mM MgCl 2 except when otherwise stated. The catalytic activity of ePL kinase was determined spectophotometrically at 37uC by observing the initial increase in absorbance at 388 nm during the conversion of PL to PLP [20].

Purification of PLP Phosphatase
Human brain PLP phosphatase cDNA cloned in pET19b was kindly provided by Dr. Anette Bøe, University of Bergen, Norway [35]. The plasmid was transformed into Escherichia coli strain Rosetta (lDE3) pLysS. The recombinant cells were grown at 37uC in LB broth with ampicillin (50 mg/ml) and chloramphenicol (34 mg/ml) until OD 600 was about 1.0 and induced with 50 mM isopropyl-b-Dthiogalactopyranoside (IPTG). The cells were grown for additional 24 hours at 18uC and then harvested by centrifugation. The Histagged enzyme was released by homogenization in 50 mM sodium phosphate buffer, pH 8, containing 300 mM NaCl and 10 mM imidazole. After centrifugation the cell extract was added to a 10 ml Ni-NTA agarose column (Qiagen) and the column was washed sequentially with 20, 30, 50, 75, 150 mM imidazole buffers, pH 8, until the absorbance at 280 nm was less than 0.1. The PLP phosphatase was then eluted with 250 mM imidazole buffer. The purified enzyme was dialyzed against 160 mM NaCl and 4 mM MgCl 2 in 40 mM sodium BES, pH 7.0 for 4 hours. A second dialysis in the same buffer, except with 80 mM NaCl, was continued overnight. The enzyme was stored at 220uC. About 20 mg of 90% pure enzyme was obtained per liter of culture. Activity was determined by monitoring the decrease in absorbance at 388 nm during the conversion of PLP to PL [4].

Preparation of K229Q ePL Kinase
In order to avoid any contamination from wild type ePL kinase from the host cells, the K229Q mutant was expressed as a Histagged protein and purified by affinity chromatography. The addition of the His-tag at the C-terminus of ePL kinase was obtained by subcloning the coding sequence of wild type ePL kinase [20] into pET28a plasmid (Novagen), between NdeI and EcoRI restriction sites.
The K229Q mutant was prepared by site-directed mutagenesis using the pET28-ePL kinase construct as template, by the Quick-Change TM method from Stratagene. Two oligonucleotides (synthesized by MWG-Biotech) containing the mutations were used as primers: 59-GGTTAAAACTGACCTGCAAGGGACTGGC-GACC-39 and its complementary oligonucleotide (the mutated bases are underlined). E. coli DH5a cells were transformed and used to amplify the mutated plasmid. Both strands of the coding region of the mutated gene were sequenced. The only differences with respect to wild type were those intended.
Enzyme expression of the His-tagged form of the enzyme, performed using the HMS174(lDE3) strain of E. coli, was induced by adding 0.2 mM IPTG. After incubation at 28uC with constant agitation for 20 h, cells were harvested and suspended in 50 mM sodium phosphate buffer, pH 8, containing 0.3 M NaCl. Cells were ruptured by sonication. Following the addition of DNase I (0.02 mg/ml), the lysate was centrifuged, and the supernatant was loaded on a 25 ml Ni-NTA agarose column, equilibrated with lysis buffer. The column was washed with the same buffer, containing 40 mM imidazole, and the enzyme was then eluted with a 400-ml linear gradient of imidazole from 40 mM to 0.5 M. Fractions containing the enzyme were detected by SDS-polyacrylamide gel electrophoresis, pooled and dialyzed against 20 mM potassium HEPES, pH 7.5. The His-tagged form of the wild type enzyme was also purified and characterized with respect to its catalytic properties, which were found to be identical to the original form of the enzyme.

Rate of Formation of ePL KinaseNPLP Complex in the Presence of Different Substrates and Products
The rates of formation of ePL kinaseNPLP complex were determined by small Sephadex G-50 spin columns as follows. One ml plastic syringes with a glass wool plug were filled to the 0.5 ml mark with Sephadex G-50 in 50 mM potassium HEPES, pH 7.5. In an Eppendorf tube, 90 mM ePL kinase was incubated for various times with 0.4 mM MgATP or MgADP, 0.150 mM PL or PLP, 0.2 mM MgCl 2 , in 0.1 ml reaction volume at 37uC. After incubation, each sample was placed on the G-50 column that was equilibrated with buffer, which contained either 1 mM MgATP or 1 mM MgADP and 0.2 mM MgCl 2 in 20 mM potassium HEPES buffer, pH 7.5, at 4uC. In the experiment in which only PLP was present, the nucleotides were omitted from the buffer. The sample was allowed to soak into the column. To each column, 400 ml of equilibration solution at 4uC was added and the plastic syringes (placed in a glass test tube) were inserted in a swinging bucket centrifuge and rotated at about 150 rpm. The buffer containing the protein elutes into the glass tube and the small molecules remain on the column. The 400 ml of eluted sample is then placed in a 1 cm cuvette and the spectrum recorded. The total time to separate ePL kinase from small molecules is about 3 min. Controls show that 8963.5% of ePL kinase eluted in the 400 ml and less than 2% of the excess PL and PLP were eluted.

Stoichiometry of ePL KinaseNPLP Complexes
The stoichiometry of the formation of the ePL kinaseNPLP complex was determined by adding NaOH to a final concentration of 0.2 M to denature the protein and release the tightly bound PLP. The absorbance at 388 nm was used to determine the concentration of the cofactor, that in 0.2 M NaOH exhibits a molar extinction coefficient of 6600 M 21 cm 21 [7]. The concentration of ePL kinase in the complex was also determined in 0.2 M NaOH, using an extinction coefficient at 292 nm (a wavelength at which PLP has negligible absorbance) of 35,416 M 21 cm 21 , which was determined from the absorption spectrum of a protein sample whose concentration was previously determined in buffer (extinction coefficient at 278 nm of 28,850 M 21 cm 21 ) [20]. The concentration of PLP in the buffer at pH 7.5 was calculated using an extinction coefficient at 388 nm of 5305 M 21 cm 21 . This coefficient was derived from the absorption spectrum of a PLP sample whose concentration was determined in NaOH.

Determination of ATP and ADP Concentration in ePL KinaseNPLP Complexes
The complexes were obtained by incubating 130 mM ePL kinase samples with different combinations of PLP, PL, MgATP, MgADP and MgCl 2 as detailed in Table 2, at 37uC for 1 hour in 20 mM potassium HEPES, pH 7.5. The samples were then passed through a G-50 Sephadex column (1.3 cm624 cm) that had been previously equilibrated with 20 mM potassium HEPES, pH 7.5, containing 0.5 mM MgCl 2 and eluted with the same buffer. Fractions containing ePL kinase were pooled. PLP and protein concentrations were measured as described above. In order to determine nucleotide concentration, 600 mL of sample was treated as follows. The protein was denatured by the addition of 0.2 M NaOH in order to allow the release of protein-bound molecules. Addition of 0.2 M HCl neutralized the solution followed by centrifugation to remove the protein precipitate. The supernatant (450 mL) was transferred in a clean tube and assayed for the presence of ATP and ADP. ADP concentration was measured, by coupling pyruvate kinase and lactate dehydrogenase reactions, from the absorbance change at 340 nm due to NADH oxidation. A typical assay, carried out at 37uC in 20 mM potassium HEPES, pH 7.5, in a final volume of 600 mL, contained 450 mL of the complex sample, 2 mM phosphoenolpyruvate, 0.25 mM NADH, 18 units of D-lactate dehydrogenase and 10 units of pyruvate kinase. ATP concentration was determined using human PL kinase activity, measuring the absorbance change at 388 nm due to PLP formation from PL. Assays, carried out at 37uC in 20 mM potassium HEPES, pH 7.5, in a final volume of 600 mL, contained 450 mL of the complex sample, 0.17 mM PL and 8 mM PL kinase. Both assays had been previously validated using commercial nucleotide samples of known concentrations, which ranged from 1.5 mM to 40 mM, obtaining two calibration curves for ATP and ADP measurements. The minimum detectable nucleotide concentration was estimated to be around 1.5 mM for both assays.

Rate of Dissociation of PLP from the ePL KinaseNPLP Complex
The CD spectrum of bound PLP in the ternary complex was used to determine the rate of dissociation of PLP. Bound PLP in the complex gives negative bands at about 336 nm and 415 nm, but free PLP does not exhibit a CD signal since it is a symmetrical molecule. A 60 mM solution of ePL kinaseNPLP complex was placed in a 1 cm cell and the signal at 415 nm followed with time at a bandwidth of 5 nm at 37uC. Included in the sample was a 10 mM solution of PLP phosphatase or alkaline phosphatase to rapidly convert any released PLP to PL.

Determination of PL Kinase Kinetic Constants for Wild Type and K229Q ePL Kinases
All assays were performed at 37uC in a 1 cm thermostated cuvet. The increase of absorbance at 388 nm, due to the difference between PLP production and PL consumption (e = 5020 M 21 cm 21 at pH 7.5), was monitored for about 60 s in an Agilent 8453 spectrophotometer. The very first initial velocity was used to calculate kinetic parameters in order to avoid enzyme inhibition. K m and k cat values for ATP and PL were determined as previously described [20].

Transfer of PLP from the ePL KinaseNPLP Complex
The complex was prepared as described in the section on the analysis of ADP and ATP. In a 1 ml cuvette in HEPES buffer, pH 7.5, 37uC, the following was added; glycine (50 mM), tetrahdyrofolate (100 mM); and apo-eSHMT (20 mM). The transfer was started by the addition of the 20 mM PL kinaseNPLP. Transfer of PLP from the PL kinase complex was followed by the formation of the holo-SHMTNGlyNtetrahdyrofolate abortive complex that absorbs at 495 nm [27]. The first experiment was to determine if any of the PLP in the complex was transferred by following the absorbance at 495 nm. To determine the fraction of holo-eSHMT formed a control was performed where the PL kinase complex was replaced with 25 mM PLP, which would fully saturate the apo-eSHMT giving the A495 nm for fully saturated holo-eSHMT. Each of these two experiments was duplicated in the presence of 3 mM PLP phosphatase that was added to the reaction solution prior to starting the transfer with either free PLP or PL kinaseNPLP.