BBAPAP-39521; No. of pages: 7; 4C: Biochimica et Biophysica Acta xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′-phosphate☆ Martino Luigi di Salvo a,1, Isabel Nogués b,1, Alessia Parroni a, Angela Tramonti c,a, Teresa Milano a, Stefano Pascarella a, Roberto Contestabile a,⁎ a b c

Dipartimento di Scienze Biochimiche “A. Rossi Fanelli”, “Sapienza” Università di Roma, Piazzale Aldo Moro 5, 00185 Roma, Italy Istituto di Biologia Ambientale e Forestale, Consiglio Nazionale delle Ricerche, Via Salaria Km 29.300, 00015 Monterotondo Scalo, Roma, Italy Istituto di Biologia e Patologia Molecolari, Consiglio Nazionale delle Ricerche, Piazzale Aldo Moro 5, 00185 Roma, Italy

a r t i c l e

i n f o

Article history: Received 15 November 2014 Received in revised form 23 January 2015 Accepted 24 January 2015 Available online xxxx Keywords: Pyridoxal phosphate Vitamin B6 recycling Enzyme regulation Catalytic mechanism Inhibition mechanism Carbinolamine intermediate

a b s t r a c t Pyridoxal 5′-phosphate (PLP), the catalytically active form of vitamin B6, plays a crucial role in several cellular processes. In most organisms, PLP is recycled from nutrients and degraded B6-enzymes in a salvage pathway that involves pyridoxal kinase (PLK), pyridoxine phosphate oxidase and phosphatase activities. Regulation of the salvage pathway is poorly understood. Escherichia coli possesses two distinct pyridoxal kinases, PLK1, which is the focus of the present work, and PLK2. From previous studies dating back to thirty years ago, pyridoxal (PL) was shown to inhibit E. coli PLK1 forming a covalent link with the enzyme. This inhibition was proposed to play a regulative role in vitamin B6 metabolism, although its details had never been clarified. Recently, we have shown that also PLP produced during PLK1 catalytic cycle acts as an inhibitor, forming a Schiff base with Lys229, without being released in the solvent. The question arises as to which is the actual inhibition mechanism by PL and PLP. In the present work, we demonstrated that also PL binds to Lys229 as a Schiff base. However, the isolated covalent PLK1–PL complex is not inactive but, in the presence of ATP, is able to catalyse the single turnover production of PLP, which binds tightly to the enzyme and is ultimately responsible for its inactivation. The inactivation mechanism mediated by Lys229 may play a physiological role in controlling cellular levels of PLP. This article is part of a Special Issue entitled: Cofactor-dependent proteins: evolution, chemical diversity and bio-applications. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Pyridoxal 5′-phosphate (PLP), the catalytically active form of vitamin B6, is the cofactor of a great number of enzymes involved in central metabolic pathways [1–3]. In Escherichia coli, PLP is synthesised following the so-called deoxyxylulose 5-phosphate (DXP)-dependent biosynthetic pathway [4]. The final product of this route is pyridoxine 5′-phosphate (PNP), which is then oxidized to PLP by PNP oxidase

Abbreviations: PLP, pyridoxal 5′-phosphate; PNP, pyridoxine 5′-phosphate; PN, pyridoxine; PL, pyridoxal; ePLK1, E. coli pyridoxal kinase encoded by pdxK gene; ePLK2, E. coli pyridoxal kinase encoded by pdxY gene; saPLK, Staphylococcus aureus pyridoxal kinase ☆ This article is part of a Special Issue entitled: Cofactor-dependent proteins: evolution, chemical diversity and bio-applications. ⁎ Corresponding author at: Dipartimento di Scienze Biochimiche, “Sapienza” Università di Roma, Piazzale Aldo Moro 5, 00185, Roma, Italy. Tel.: + 39 0649917575; fax: + 39 0649917566. E-mail address: [email protected] (R. Contestabile). URL:E-mail addresses:E-mail address: http://w3.uniroma1.it/bio_chem/sito_biochimica/EN/index.html (R. Contestabile). 1 These authors contributed equally to this work.

[5,6]. In addition, Escherichia coli, as many other prokaryotic and eukaryotic organisms, utilizes a salvage pathway in which all B6 vitamers coming from growing medium and protein turnover can be converted into PLP by the concerted action of PNP oxidase, phosphatases and kinases [7–9]. E. coli possesses two different kinases that are able to phosphorylate pyridoxal (PL) [10]. E. coli PLK1 (ePLK1), the object of this study, is encoded by the pdxK gene and is also able to phosphorylate pyridoxine (PN), pyridoxamine (PM) and hydroxymethylpyrimidine [11], although E. coli has a specific kinase for this latter substrate [12]. E. coli PLK2 (ePLK2), encoded by the pdxY gene, is specific for PL [13]. About thirty years ago, it had been proposed that pyridoxal is capable to inhibit E. coli PLK1 binding to it as Schiff base to a lysine residue [14,15]. However, this lysine residue had not been identified. PL inhibition took place in the absence of ATP and therefore was not related to the enzyme catalytic turnover. We have recently observed that the enzyme also undergoes inhibition while catalysing the ATPdependent phosphorylation of pyridoxal, forming a tight complex with PLP through an aldimine linkage with residue Lys229 [16]. In this case, PLP that binds to ePLK1 and is responsible for its inactivation must be produced at the active site of the enzyme as a consequence of

http://dx.doi.org/10.1016/j.bbapap.2015.01.013 1570-9639/© 2015 Elsevier B.V. All rights reserved.

Please cite this article as: M.L. di Salvo, et al., On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′phosphate, Biochim. Biophys. Acta (2015), http://dx.doi.org/10.1016/j.bbapap.2015.01.013

2

M.L. di Salvo et al. / Biochimica et Biophysica Acta xxx (2015) xxx–xxx

its catalytic turnover. Although PLP is tightly bound to the inactivated enzyme and cannot easily be removed, it can be transferred to a PLPdependent apoenzyme with consequent reactivation of ePLK1 [16]. Given the observations on ePLK1 inhibition by either substrate or product, the question arises as to whether PL and PLP actually inhibit the enzyme following two different mechanisms. Whether the regulation of enzyme activity, and therefore the availability of PLP through the salvage pathway, is based on these observed inhibition processes is still a main point to be elucidated. In this work, we present a series of experiments carried out to clarify the mechanism of inhibition of ePLK1 by PL and PLP and also the role of Lys229 in enzyme inhibition and catalysis. 2. Materials and methods

binding equilibria were then calculated from saturation curves obtained measuring the protein fluorescence emission intensity as a function of increasing ligand concentration. Either PL or PLP (from 1 to 100 μM) was added to enzyme samples (0.5 μM subunit concentration) at 37 °C in 20 mM potassium HEPES at pH 7.5, containing 0.5 mM MgCl2. Preliminary experiments demonstrated that the binding equilibrium was established within 10 min. Fluorescence emission measurements were carried out at 37 °C with a FluoroMax-3 Jobin Yvon Horiba spectrofluorimeter using a 1-cm path length quartz cuvette. Fluorescence emission spectra were recorded from 300 nm to 450 nm with the excitation wavelength set at 280 nm. Excitation and emission slits were set at 5 nm. Fluorescence emission spectra of samples containing the same concentration of ligand (PL or PLP) but without ePLK were also recorded and subtracted from the respective spectra with ePLK at each titration point. Data were analysed according to Eq. (3).

2.1. Materials 2.5. Data analysis Ingredients for bacterial growth were from Fluka. Ni-NTA Agarose for purification of 6xHis-tagged proteins was from Qiagen Inc. (Valencia, CA, USA). All other chemicals were from Sigma-Aldrich (St. Louis, MO, USA) and Carlo Erba (Milano, Italy). Wild type and K229Q ePLK1 were purified as previously described [16]. E. coli bacterial strains BW25113 (wild-type), JW1628 (ΔpdxY) and JW2411 (ΔpdxK) used for cell extracts assays were obtained from the E. coli Genetic Stock center (Yale University, New Haven, CT, USA). 2.2. Activity assays Activity assays and all other experiments were carried out in 20 mM potassium HEPES, pH 7.5, containing 0.5 mM MgCl2. The catalytic activity of ePLK1 was determined spectrophotometrically at 37 °C by observing the initial increase in absorbance at 388 nm during the conversion of PL to PLP [13]. The concentration of PLP produced by the enzyme in potassium HEPES buffer at pH 7.5 was measured using an extinction coefficient at 388 nm of 5020 M− 1 cm− 1. This extinction coefficient was calculated as the difference between the extinction coefficients of PLP and PL, determined independently from solutions whose concentrations were previously measured in NaOH [17]. Kinetic measurements in the activity assays were performed on a HewlettPackard 8453 diode-array spectrophotometer. Inhibition kinetics were analysed using Eq. (1).

Inhibition kinetics were analysed using Eq. (1) in which [PLP] is the observed PLP concentration, [PLP]max is PLP concentration produced after complete enzyme inactivation and ki is the observed inhibition rate constant.   −k t ½PLP  ¼ ½PLP max  1−e i

1

The observed inhibition rate constant shown in Fig. 2 (lower panel) were analysed according to Eq. (2), in which kobs is the observed inhibition constant, kmax is the maximum inhibition constant, [PL] is pyridoxal concentration and Kd is an apparent dissociation constant. kobs ¼ kmax 

½PL ½PL þ K d

2

Fluorescence measurements obtained from binding equilibria experiments were transformed in fractional variation with respect to the extrapolated maximum change at infinite ligand concentration and then analysed according to Eq. (3) in which f is the observed fluorescence fractional variation, [L] is the ligand and Kd the dissociation constant. f ¼

½L  L þ Kd

3

2.3. Isolation and stoichiometry of the ePLK1–PL complex ePLK1 (150 μM) was mixed with 0.5 mM PL and incubated at 37 °C for one hour. The sample was then loaded onto a Sephadex G-50 column (1 cm Ø × 25 cm) equilibrated and then eluted with potassium HEPES buffer collecting 1 mL fractions. The first 20 fractions were analysed by measuring their absorption spectra between 250 and 500 nm to detect the presence of protein and pyridoxal. The stoichiometry of protein and PL in the ePLK1–PL complex was determined by adding NaOH to a final concentration of 0.2 M to denature the protein and release bound PL. The absorbance at 388 nm was used to determine the concentration of the cofactor, that exhibits a molar extinction coefficient of 1700 M− 1 cm−1 [17]. The concentration of protein 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−1 cm−1, which was determined from the absorption spectrum of a protein sample whose concentration was previously determined in buffer (extinction coefficient at 278 nm of 27,910 M−1 cm−1) [13]. 2.4. Analysis of PL and PLP binding equilibria Analyses took advantage of the protein intrinsic fluorescence quenching observed upon binding event. Dissociation constants of

All data were analysed using the software Prism (GraphPad Software Inc., San Diego, CA, USA). 2.6. Activity measurement on cell extracts Starter cultures of BW25113 (wild-type), JW1628 (ΔpdxY) and JW2411 (ΔpdxK) (E. coli Genetic Stock center) were grown overnight at 37 °C in LB medium. Overnight cultures (2 mL) were used to inoculate 200 mL of fresh LB containing kanamycin (for the knock out mutant strains). The cultures were grown with shaking at 37 °C until OD600 reached 1.2 and were then harvested by centrifugation at 9000 rpm for 15 min at 4 °C. Cells were suspended in 20 mL of 20 mM potassium HEPES buffer at pH 7.5, containing 0.5 MgCl2 and centrifuged again. This washing step was repeated twice to remove any residual of growth medium, ending with cells suspended in 10 mL of buffer. Cells were disrupted by sonication and centrifuged at 12.000 rpm for 15 min at 4 °C. Supernatants were assayed for PL kinase activities. PL kinase activity was measured by the fluorimetric assay as described in [10,18]. Reaction mixtures (2 mL) contained 1 mM MgATP, 0.5 mM MgCl2, 0.2 mM PL and 200 μL of crude extract in 20 mM potassium HEPES buffer. Reactions were started by addition of crude extract and incubated at 37 °C. At time intervals, the sample was quickly transferred into a thermostated cuvette at 37 °C under continuous stirring.

Please cite this article as: M.L. di Salvo, et al., On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′phosphate, Biochim. Biophys. Acta (2015), http://dx.doi.org/10.1016/j.bbapap.2015.01.013

M.L. di Salvo et al. / Biochimica et Biophysica Acta xxx (2015) xxx–xxx

Hydroxylamine was then added to a final concentration of 1 mM, and fluorescence intensity (excitation wavelength 380 nm; emission wavelength, 450 nm; excitation and emission slits were set at 5 nm) was determined 150 s later to maximize the signal to background ratio of hydroxylamine adducts of PLP over PL. The combined fluorescence intensities of control reaction mixtures lacking either crude extract or PL were subtracted from fluorescence measured for each sample. Protein concentration in cell extracts was determined using the Bradford's method [19]. Proportionality between initial PLK activity and protein concentration was checked over a wide range. A calibration curve was constructed using samples of pure PLP so that fluorescence measurements obtained from cell extracts could be transformed in enzyme activity expressed as PLP produced over time. 3. Results 3.1. Kinetics of PL binding to ePLK1 and enzyme inhibition We ran a series of experiments to compare the rate of enzyme inhibition with the kinetics of PL binding to the enzyme as Schiff base. The rate of enzyme inhibition upon incubation of ePLK1 (0.25 μM) with 0.5 mM PL in 20 mM potassium HEPES buffer pH 7.5, containing 0.5 mM MgCl2, was measured at 37 °C. After the addition of PL as last component, the activity of the enzyme was assayed at time intervals by taking aliquots of the mixture and adding 1 mM MgATP to start the phosphorylation reaction. The initial velocity of PLP formation was obtained by measuring the absorbance change at 390 nm and using an extinction coefficient of 5020 M−1 cm−1 [16] (closed circles in Fig. 1). In a separate experiment, the Schiff base formation between PL and ePLK1 was followed at 420 nm as 10 μM enzyme was mixed with 0.5 mM PL in the same buffer and at the same temperature (continuous line in Fig. 1). The superimposition of the kinetics obtained from the two experiments shows that the rates of enzyme inhibition and Schiff base formation coincide. Data were globally analysed using an exponential equation (Eq. (1) in the Materials and methods section), obtaining an observed inhibition constant of about 0.4 min−1. When the ePLK1 K229Q mutant was used in place of wild type ePLK1, no inactivation was observed (Fig. 1, triangles). It is clear that the enzyme gets inactivated as PL binds covalently to it forming a Schiff base with Lys229. It is important to notice that the inactivation by PL takes place before ATP is added to assay enzyme activity. However, we know that the enzyme also undergoes inhibition while catalysing the ATP-dependent phosphorylation of PL. In fact, when 0.5 mM PL was added to a solution

Fig. 1. Kinetics of ePLK1 inhibition. Enzyme activity was measured after mixing 0.25 μM ePLK1 with 0.5 mM PL. Percentage of inactivation was calculated from initial velocity measurements obtained with wild type (circles) and K229Q (triangles) ePLK1. The continuous line superimposed to experimental points represents the absorbance change at 420 nm measured when wild type ePLK1 (10 μM) was mixed with PL (0.5 mM). Inset. In a different experiment, PLP production was measured continuously upon addition of PL (0.5 mM) to a solution containing ePLK1 (0.25 μM) and MgATP (1 mM).

3

containing ePLK 1 (0.25 μM) and MgATP (1 mM), the rate of PLP formation decreased exponentially until the enzyme was completely inactivated and a final concentration of about 10 μM PLP was formed (Fig. 1, inset). Therefore, it can be calculated that complete inhibition occurred in about 40 catalytic turnovers. In this case the observed inhibition constant was about 0.7 min− 1, which is higher than the 0.4 min−1 inactivation rate observed when the enzyme was incubated with only PL. In another series of experiments, the amount of PLP produced after the enzyme gets completely inactivated was measured varying ePLK1 concentration between 0.125 and 1 μM, while PL and MgATP were kept fixed at saturating 0.5 mM and 1 mM concentrations. The amount of PLP produced was linear with respect to enzyme concentration and the enzyme:PLP proportion was 1:40, confirming that in these conditions the enzyme gets completely inactivated after 40 turnovers (Fig. 2). At all enzyme concentrations, the observed inhibition constant was 0.7 min−1 (Fig. 2, inset). The observed 1:40 proportionality failed in an additional series of experiments, in which PL concentration was varied while enzyme and MgATP were kept fixed at 0.25 μM and 1 mM, respectively. At low PL concentrations (from 5 to 50 μM) all PL was converted into PLP, as shown by the linear behaviour observed in the inset of Fig. 3 (upper panel). Therefore, at 50 μM PL, the enzyme must have turned over about 200 times. In these conditions, the enzyme was able to convert all substrate into product before being completely inactivated. As PL concentration was increased from 50 μM to 1 mM, the concentration of produced PLP decreased to a limiting value of about 10 μM (Fig. 3, upper panel). Evidently, as PL concentration increased, the rate of inactivation also increased until the enzyme was saturated with PL. Under these conditions, the enzyme turns over only 40 times before being completely inactivated. The lower panel of Fig. 3 shows the observed inactivation rate constants determined from the same series of experiments. It can be seen that inactivation rate increases hyperbolically as a function of PL concentration, reaching a maximum value of 0.77 ± 0.03 min−1. As mentioned above, inactivation was observed only when the concentration of PL was higher than 50 μM and it was therefore possible to fit the curves shown in the inset of Fig. 3, lower panel, by an exponential equation, obtaining appropriate rate constant values. 3.2. Isolation of the ePLK1–PL Schiff base complex PL (0.5 mM) and ePLK1 (150 μM) were incubated at 37 °C for 1 h in 20 mM potassium HEPES buffer pH 7.5, containing 0.5 mM MgCl2. This mixture was then separated on size exclusion chromatography.

Fig. 2. PLP production as a function of ePLK1 concentration. The amount of PLP produced upon complete enzyme inactivation was measured in a series of experiments in which 0.5 mM PL was mixed with different enzyme concentrations and 1 mM ATP. The inset shows the inactivation curves from which PLP concentrations were obtained.

Please cite this article as: M.L. di Salvo, et al., On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′phosphate, Biochim. Biophys. Acta (2015), http://dx.doi.org/10.1016/j.bbapap.2015.01.013

4

M.L. di Salvo et al. / Biochimica et Biophysica Acta xxx (2015) xxx–xxx

Fractions containing the protein eluted in the void volume were well separated from free PL. The absorption spectrum of the protein shows a prominent band at 420 nm (Fig. 4), corresponding to a protonated aldimine between enzyme and PL (Scheme 1) [20]. A band at 318 nm and a shoulder at 340 nm are also visible. Addition of 0.2 M NaOH to an aliquot of this protein sample gave the typical absorption spectrum of PL, from which a 1:0.6 stoichiometric ratio of protein to PL could be calculated [17]. This experiment shows that a fairly tight complex is formed between ePLK1 and PL since during the chromatography only a part of PL dissociated from the protein–PL complex. The 318 nm band that is visible in the absorption spectrum of the protein after chromatography may be attributed to free PL, while the shoulder at about 340 nm may correspond to the carbinolamine intermediate that occurs in the Schiff base formation (Scheme 1). 3.3. Catalytic activity of the isolated ePLK1–PL complex

Fig. 3. PLP production as a function of PL concentration (upper panel). The amount of PLP produced upon complete enzyme inactivation was measured when different PL concentrations were mixed with 0.25 μM ePLK1 and 1 mM MgATP. The inset shows that in the 0 to 50 μM PL concentration range the amount of PLP produced corresponds to initial PL concentration. Dependence of inactivation rate on PL concentration (lower panel). The observed inactivation rate constant (kobs), calculated from inactivation curves obtained in the 100 μM to 1 mM PL concentration range, depends hyperbolically on PL concentration (the continuous line was generated by the least square fitting of kobs values to Eq. (2), which yielded a maximum inhibition rate of 0.77 ± 0.03 min−1). The inset shows the inactivation curves obtained at 500 μM PL ( ), 300 μM PL (—) and at 100 μM PL (—).

An aliquot of the ePLK1–PL complex sample isolated by size exclusion chromatography was assayed for steady-state pyridoxal kinase activity at the concentration of 0.25 μM, in the presence of PL (0.5 mM), MgATP (1 mM) and MgCl2 (0.5 mM). About 50% of activity was measured with respect to an untreated free enzyme sample. Assuming that the enzyme–PL complex is inactive, this result is in good agreement with the calculated 1:0.6 stoichiometric ratio of protein to PL. An additional experiment was devised in order to check whether the ePLK1–PL complex, when provided with ATP, was able to catalyse a single turnover phosphorylation of the enzyme-bound PL, producing enzyme-bound PLP. An aliquot of the ePLK1–PL complex, in which the protein concentration was 60 μM (and therefore PL amounted to 36 μM), was mixed with 2 mM pyridoxine (PN) in the usual buffer at 37 °C. The reaction was then started by the addition of 1 mM MgATP. After 30 s, 0.2 M NaOH was added to the sample, in order to denature the enzyme and measure the PLP that had been formed (Fig. 5, continuous line). The reason of having a large excess of PN (a competitor of PL for binding to ePLK1 active site) in this experiment was to make any free PL unavailable to the enzyme, so that if formation of PLP was observed this must had come from enzyme-bound PL. The experiment showed that a concentration of about 40 μM PLP had formed, considering an extinction coefficient for PLP at 390 nm of 6550 M−1 cm−1 [17] Enz Enz

H2N +H

O HO

Enz O-

2N

+HN

+

HO

HO

HO

OH

OH

N

N

Free PL

H2O

N

Carbinolamine intermediate

Protonated Schiff base (aldimine)

Enz Enz

-S

O-

S

O

HO

HO

HO OH N

Free PL Fig. 4. Absorption spectrum of the ePLK1–PL complex. The figure shows the absorption spectra of the ePLK1–PL complex, which eluted in the void volume of the size exclusion chromatography column (—), and of free PL (—) that eluted afterward. Expanded view of the same spectra in which the spectrum of an ePLK1 K229Q sample that was incubated with PL and passed through a size exclusion chromatography column is shown for comparison ( ).

N

Hemithioacetal intermediate

Scheme 1. Upper scheme. Reaction mechanism for the formation of a Schiff base linkage between a lysine residue and PL. A carbinolamine intermediate is formed in which a negative charge has developed on the 4′-oxygen. Lower scheme shows the formation of a hemithioacetal intermediate between a cysteine residue and PL. Also in this case a negative charge develops on the 4′-oxygen.

Please cite this article as: M.L. di Salvo, et al., On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′phosphate, Biochim. Biophys. Acta (2015), http://dx.doi.org/10.1016/j.bbapap.2015.01.013

M.L. di Salvo et al. / Biochimica et Biophysica Acta xxx (2015) xxx–xxx

5

Schiff base formation previously observed with PL. Dissociation constants of PLP (3.2 ± 0.2 μM) and PL ( 18.5± 3.0 μM) binding equilibria were obtained measuring fluorescence changes observed at increasing pyridine ligand concentrations (Fig. 6). It is clear that PLP binds to ePLK1 with much higher affinity than PL. 4. Discussion

Fig. 5. Catalytic activity of the isolated ePLK1–PL complex. Absorption spectrum (—) obtained after addition of NaOH (0.2 M) to the solution in which MgATP was added to the ePLK1–PL complex. The observed 390 nm absorption band corresponds to about 40 μM PLP. The absorption spectrum obtained in the control experiment (—) corresponds to the PL concentration that was included in the experiment (36 μM).

(in a similar experiment, in which no NaOH was added but the sample was repeatedly concentrated and diluted through centrifugal filter units to remove all unbound small molecules from protein solution, the absorption spectrum of the sample was that of a typical ePLK1– PLP complex, with absorption bands at 336 and 420 nm [16]; data not shown). In a control experiment, 60 μM free enzyme and 36 μM free PL were used instead of the ePLK1–PL complex. Also in this case, 2mM PN was present. The obtained absorption band (Fig. 5, dashed line) cannot correspond to 40 μM PLP. The small absorption at 390 nm is rather attributable to free PL that was included in the experiment (36 μM), considering an extinction coefficient for PL of 1700 M−1 cm−1 [17]. 3.4. Binding equilibria of PL and PLP The intrinsic fluorescence emission of ePLK1 is quenched when either PL or PLP are mixed with the protein and this method can be used to analyse binding equilibria and measure dissociation constants. The kinetics of the observed fluorescence change was similar with both pyridine compounds and showed two exponential phases. The faster phase had very small amplitude and was completed in the mixing time. The rate of the slower phase was compatible with the kinetics of

The inactivation of ePLK observed when the enzyme is incubated with pyridoxal in the absence of ATP coincides with the formation of a Schiff base linkage between the PL aldehyde group and the enzyme Lys229 residue (Fig. 1). This inactivation process is relatively slow and at 0.5 mM PL concentration takes place at a rate of 0.4 min−1. It is important to notice that the formation of this Schiff base takes place at equilibrium conditions, when the enzyme is not catalysing any reaction. In these conditions, enzyme inactivation due to PL binding was followed in a discontinuous assay, in which the residual steady state activity of the enzyme was measured. The covalent enzyme–PL complex is not able to catalyse multiple turnovers. However, it is clear that the isolated enzyme–PL complex is able to catalyse a single turnover when ATP is added to it (Fig. 5). The PLP produced in a single turnover is also bound to the enzyme as Schiff base, as demonstrated by the fact that after elimination of small, unbound molecules, the enzyme shows a clear 420 nm absorption band. The enzyme–PLP complex is ultimately inactive. When enzyme activity was followed in a continuous assay, while the enzyme was turning over in the presence of both PL (0.5 mM) and ATP (1 mM) (Fig. 1, inset), nearly a twofold increase in the inactivation rate was observed (0.7 min−1) with respect to the inactivation rate measured in the presence of only PL (0.4 min− 1). This observation suggests that, in these conditions, an additional process other than simple Schiff base formation between enzyme and PL affects the rate of enzyme inactivation. When PL concentration is varied, the partition ratio of catalysis over inactivation is altered. In particular, the inactivation rate is a hyperbolic function of PL concentration and reaches a maximum when this is saturating (Fig. 3, lower panel). We believe that all our experimental results may be explained by the mechanism depicted in Scheme 2. According to this mechanism, PL would initially bind very rapidly and non-covalently to the enzyme. In the absence of ATP, the E•PL non-covalent complex is slowly converted into a covalent E = PL complex (through a carbinolamine intermediate, E–PL), in which a Schiff base linkage between PL and Lys229 has

ATP ADP

E + PL

E PL

1

E PLP

E + PLP

ATP ADP

E-PL H2O

2

E-PLP H2O

H2O

H2O

ATP ADP

E=PL

Fig. 6. Dissociation constants of ePLK1 for PL and PLP. Saturation curves obtained from fluorescence changes observed upon addition of either PL (open circles) or PLP (closed circles) to ePLK1. Fluorescence changes were transformed into fractional fluorescence change. Continuous lines through experimental points were obtained from least square fitting of data to Eq. (3) (see Materials and methods section), obtaining dissociation constants of 3.2 ± 0.2 μM for PLP and 18.5 ± 3.0 μM for PL.

3

E=PLP

Scheme 2. Reaction scheme for the inhibition of ePLK1 by either PL or PLP. E•PL, noncovalent complex between ePLK1 and PL; E•PLP, non-covalent complex between ePLK1 and PLP•E–PL, carbinolamine intermediate of PL; E–PLP, carbinolamine intermediate with PLP; E = PL, Schiff-base covalent complex between ePLK1 and PL; E = PLP, Schiff-base covalent complex between ePLK1 and PLP. This latter complex represents the actual inhibited form of the enzyme. PL may be phosphorylated in reactions 1, 2 and 3. Once the PL carbinolamine intermediate is formed (E–PL), in the presence of ATP this may be converted into a PLP carbinolamine intermediate (E–PLP). Both PL and PLP carbinolamines can eliminate a water molecule and generate a Schiff base. While it is expected that the carbinolamine intermediate forms very rapidly, the observed Schiff base formation is relatively slow and is probably the rate-limiting step in enzyme inactivation.

Please cite this article as: M.L. di Salvo, et al., On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′phosphate, Biochim. Biophys. Acta (2015), http://dx.doi.org/10.1016/j.bbapap.2015.01.013

6

M.L. di Salvo et al. / Biochimica et Biophysica Acta xxx (2015) xxx–xxx

formed. At this point, if ATP is added, the enzyme is able to catalyse a single turnover conversion of PL to PLP, but no steady-state activity can be measured. This observation indicates that the covalently bound PL (in either the E–PL or the E = PL form) can be phosphorylated by ATP, forming the carbinolamine (E–PLP) and the Schiff base (E = PLP) complexes with PLP, respectively (irreversible steps 2 and 3 in Scheme 2). Because PLP cannot easily be released from these complexes, as shown by our previous results [16], the enzyme is inactivated. If ATP is already present when PL is added to the enzyme, the E•PL noncovalent complex may either react with ATP, forming the E•PLP product (reaction 1 in Scheme 2), which is slowly converted into the inactive E– PLP and E=PLP covalent complexes, or react with Lys229 to generate the E–PL and E=PL covalent complexes. On the other hand, also E–PL and E=PL may be phosphorylated by ATP and transformed into the inactive covalent E–PLP and E=PLP forms. Our equilibrium binding experiments (Fig. 6) show that PLP forms a tighter complex than PL with the enzyme. Therefore, it is expected that the more PLP be formed at the active site the more enzyme will be inhibited. Moreover, as PL phosphorylation proceeds, free PLP accumulates in the solvent, and this may further contribute to enzyme inhibition by increasing the concentration of the E•PLP form. However, we do not expect PLP in the solvent to play any significant role in inhibition, since the faster rate of inactivation corresponds to the lower amount of PLP present in the solvent (Fig. 3). In conclusion, at PL concentration below the Km for this substrate (which is about 100 μM, [13]), most of the enzyme would not contain bound PL and the inactivation rate would mostly reflect formation of a Schiff base of PLP before the product is slowly released. However, when PL becomes saturating, most of the enzyme will contain bound PL and therefore the inactivation process will also be affected by the rate of PL Schiff base formation. All these considerations, summarized in Scheme 2, explain the observed hyperbolic increase of the inactivation rate on PL concentration (Fig. 3B). We excluded the possibility of an accidental allosteric effect due to the reaction of PL with lysine residues other than Lys229, since we observed that the activity of the K229Q mutant as a function of PL concentration fully conforms to the Michaelis–Menten equation, with no signs of cooperativity (positive or negative) [16]. Scheme 2 is therefore compatible with the previously reported results leading to the conclusion that ePLK1 is inhibited by its PL substrate [14,15] and also by the PLP product formed at the active site during the catalytic cycle [16]. This work provides strong evidence that Lys229 is involved in the covalent binding of both substrate PL and product PLP and therefore in the enzyme inhibition process. Moreover, the observation that the ePLK1 K229Q mutant, although catalytically active, shows a ten-fold lower kcat value and a 6-fold higher Km for PL, giving a kcat/Km ratio which is 2% of wild type [16], suggests that Lys229 may also play a marginal role in catalysis. It has been recently proposed that in PLK from Staphylococcus aureus (saPLK) a hemithioacetal formed between the enzyme Cys110 residue and PL plays a crucial role in the catalytic mechanism [21]. Cys110 is located on the active site lid that characterizes all kinases having a ribokinase fold [22–24] and closes on the active site upon substrates binding to shield from the bulk solvent. According to the proposed hypothesis, in saPLK, the negative charge that develops on the 4′-oxygen atom of PL, as a result of hemithioacetal formation, acts to lower the pKa of PL 5′-alcohol (Scheme 1). A second cysteine residue (Cys214) then acts as a base to remove the proton from the alcohol group and activates it for the nucleophilic displacement of ATP γ-phosphate. Pyridoxine, which cannot form the hemithioacetal intermediate, is phosphorylated less efficiently than PL by saPLK. The carbinolamine intermediate that forms as PL reacts with Lys229 at the ePLK1 active site is similar to the hemithioacetal intermediate. Also in the carbinolamine, the 4′-oxygen is negatively charged (Scheme 1) and may therefore play the same role of lowering the pKa of the 5′-alcohol group of PL. The capability of ePLK1 to catalyse phosphorylation of PL bound as Schiff base at the active site may be related to this possible catalytic role of Lys229. In our isolated enzyme–PL complex, the Schiff

base may be at equilibrium with the carbinolamine intermediate, as suggested by the absorption spectrum (Fig. 4). The possible catalytic role played by PL carbinolamine intermediate would explain why the enzyme-bound PL is converted to PLP upon addition of ATP. It is important to notice that ePLK1, as saPLK, is also able to phosphorylate pyridoxine, although with a lower efficiency. The fact that the K229Q ePLK1 mutant retains some activity may be explained by the presence of His59 and Asp233 at the active site of this enzyme, which interacting with the 5′-hydroxy group of PL may lower its pKa facilitating phosphorylation [21,25]. What is the meaning of ePLK1 inhibition by PL and PLP in E. coli metabolism? This is a very interesting and challenging question. As mentioned before, E. coli possesses two different kinases that are able to phosphorylate PL: PLK1 that also phosphorylates pyridoxine and pyridoxamine and PLK2 that is specific for PL [10]. Recombinant E. coli PLK2 (ePLK2, encoded by pdxY gene) has been purified and characterized and its crystal structure has been solved [26]. In analogy with saPLK, ePLK2 in the crystal appeared to bind PL through a covalent linkage with a Cys residue. Although this residue does not necessarily correspond to Cys110 of saPLK, it is located in the same active site lid [21]. We measured PL kinase activity in cell extracts of wild type, ΔpdxK and ΔpdxY E. coli strains (lacking ePLK1 and ePLK2 enzymes, respectively). Although we found activity in all three strains, we noticed a net decrease of activity over time only in the wild type and ΔpdxY (lacking ePLK2) strains. Such decrease of activity was not observed in the ΔpdxK strain that lacks ePLK1 (data not showed). These results suggest that ePLK2 does not undergo inactivation, while ePLK1 present in cell extracts is inactivated, as observed with the recombinant enzyme. This observation is in agreement with the fact that ePLK2 does not possess a lysine residue that is equivalent to ePLK1 Lys229. Therefore, ePLK2 appears to be a PL-specific, housekeeping kinase, whose activity guarantees recycling of PL into PLP. On the other hand, the less specific ePLK1, which is also capable to phosphorylate pyridoxine and pyridoxamine, would be inactivated when either PL or PLP (which may also come from pyridoxine 5′-phosphate and pyridoxamine 5′-phosphate through the action of PNP oxidase and phosphatases) reaches a critical cellular level. Measurements of B6 vitamers in E. coli grown to stationary phase in minimal medium indicated that free PL and PLP contained in the cells are about 10 μM and 120 μM, respectively [27]. In these conditions, ePLK1 is expected to be fully inactivated by PLP, while PL might not play a significant role. However, there might be different cellular conditions in which higher PL concentrations may be in turn responsible for ePLK1 inactivation. The complex interplay between pyridoxal kinases and the other enzymes involved in the PLP recycling pathway, together with the modulation of their activity, is of physiological importance to maintain the correct balance of B6 vitamers and keep PLP concentration at an appropriate level. It is also worth noting that PNP oxidase, catalysing in E. coli the last step of PLP biosynthesis shows PLP product inhibition and binds tightly PLP in a secondary site [28–30]. In this view, it should be emphasised that, as we have previously showed, the inactivated ePLK1–PLP complex can be partially reactivated by transferring PLP to PLP-dependent apoenzymes [16].

Transparency document The Transparency document associated with this article can be found, in online version.

Acknowledgements This work was supported by grants from Finanziamento Progetti di Ricerca 2011 (prot. C26A11KHY2) and 2013 (prot. C26A13BKKY) of Sapienza University of Rome.

Please cite this article as: M.L. di Salvo, et al., On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′phosphate, Biochim. Biophys. Acta (2015), http://dx.doi.org/10.1016/j.bbapap.2015.01.013

M.L. di Salvo et al. / Biochimica et Biophysica Acta xxx (2015) xxx–xxx

References [1] M.L. di Salvo, M.K. Safo, R. Contestabile, Biomedical aspects of pyridoxal 5′phosphate availability, Front. Biosci. (Elite Ed.) 4 (2012) 897–913. [2] R. Percudani, A. Peracchi, A genomic overview of pyridoxal-phosphate-dependent enzymes, EMBO Rep. 4 (2003) 850–854. [3] P.K. Mehta, P. Christen, The molecular evolution of pyridoxal-5′-phosphatedependent enzymes, Adv. Enzymol. Relat. Areas Mol. Biol. 74 (2000) 129–184. [4] T.B. Fitzpatrick, N. Amrhein, B. Kappes, P. Macheroux, I. Tews, T. Raschle, Two independent routes of de novo vitamin B6 biosynthesis: not that different after all, Biochem. J. 407 (2007) 1–13. [5] M. Di Salvo, E. Yang, G. Zhao, M.E. Winkler, V. Schirch, Expression, purification, and characterization of recombinant Escherichia coli pyridoxine 5′-phosphate oxidase, Protein Expr. Purif. 13 (1998) 349–356. [6] M.L. di Salvo, T.P. Ko, F.N. Musayev, S. Raboni, V. Schirch, M.K. Safo, Active site structure and stereospecificity of Escherichia coli pyridoxine-5′-phosphate oxidase, J. Mol. Biol. 315 (2002) 385–397. [7] M.L. di Salvo, R. Contestabile, M.K. Safo, Vitamin B(6) salvage enzymes: mechanism, structure and regulation, Biochim. Biophys. Acta 1814 (2011) 1597–1608. [8] F.N. Musayev, M.L. Di Salvo, M.A. Saavedra, R. Contestabile, M.S. Ghatge, A. Haynes, V. Schirch, M.K. Safo, Molecular basis of reduced pyridoxine 5′-phosphate oxidase catalytic activity in neonatal epileptic encephalopathy disorder, J. Biol. Chem. 284 (2009) 30949–30956. [9] T. Mukherjee, J. Hanes, I. Tews, S.E. Ealick, T.P. Begley, Pyridoxal phosphate: biosynthesis and catabolism, Biochim. Biophys. Acta 1814 (2011) 1585–1596. [10] Y. Yang, H.C. Tsui, T.K. Man, M.E. Winkler, Identification and function of the pdxY gene, which encodes a novel pyridoxal kinase involved in the salvage pathway of pyridoxal 5′-phosphate biosynthesis in Escherichia coli K-12, J. Bacteriol. 180 (1998) 1814–1821. [11] T. Mizote, H. Nakayama, Purification and properties of hydroxymethylpyrimidine kinase from Escherichia coli, Biochim. Biophys. Acta 991 (1989) 109–113. [12] T. Mizote, M. Tsuda, D.D. Smith, H. Nakayama, T. Nakazawa, Cloning and characterization of the thiD/J gene of Escherichia coli encoding a thiamin-synthesizing bifunctional enzyme, hydroxymethylpyrimidine kinase/phosphomethylpyrimidine kinase, Microbiology 145 (Pt 2) (1999) 495–501. [13] M.L. di Salvo, S. Hunt, V. Schirch, Expression, purification, and kinetic constants for human and Escherichia coli pyridoxal kinases, Protein Expr. Purif. 36 (2004) 300–306. [14] R.S. White, W.B. Dempsey, Purification and properties of vitamin B6 kinase from Escherichia coli B, Biochemistry 9 (1970) 4057–4064. [15] Y. Furukawa, R. Yamada, A. Iwashima, Inactivation of microbial pyridoxal kinase by pyridoxal, Acta Vitaminol. Enzymol. 3 (1981) 145–156.

7

[16] M.S. Ghatge, R. Contestabile, M.L. di Salvo, J.V. Desai, A.K. Gandhi, C.M. Camara, R. Florio, I.N. Gonzalez, A. Parroni, V. Schirch, M.K. Safo, Pyridoxal 5′-phosphate is a slow tight binding inhibitor of E. coli pyridoxal kinase, PLoS ONE 7 (2012) e41680. [17] E.A. Peterson, H.A. Sober, Preparation of crystalline phosphorylated derivatives of vitamin B6, J. Am. Chem. Soc. 76 (1954) 161–169. [18] S. Sussmane, J. Koontz, A fluorometric assay for pyridoxal kinase applicable to crude cell extracts, Anal. Biochem. 225 (1995) 109–112. [19] M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein–dye binding, Anal. Biochem. 72 (1976) 248–254. [20] C.M. Metzler, D.E. Metzler, Quantitative description of absorption spectra of a pyridoxal phosphate-dependent enzyme using lognormal distribution curves, Anal. Biochem. 166 (1987) 313–327. [21] M.B. Nodwell, M.F. Koch, F. Alte, S. Schneider, S.A. Sieber, A subfamily of bacterial ribokinases utilizes a hemithioacetal for pyridoxal phosphate salvage, J. Am. Chem. Soc. 136 (2014) 4992–4999. [22] M.K. Safo, F.N. Musayev, M.L. di Salvo, S. Hunt, J.B. Claude, V. Schirch, Crystal structure of pyridoxal kinase from the Escherichia coli pdxK gene: implications for the classification of pyridoxal kinases, J. Bacteriol. 188 (2006) 4542–4552. [23] M.H. Li, F. Kwok, W.R. Chang, C.K. Lau, J.P. Zhang, S.C. Lo, T. Jiang, D.C. Liang, Crystal structure of brain pyridoxal kinase, a novel member of the ribokinase superfamily, J. Biol. Chem. 277 (2002) 46385–46390. [24] Y. Zhang, M. Dougherty, D.M. Downs, S.E. Ealick, Crystal structure of an aminoimidazole riboside kinase from Salmonella enterica: implications for the evolution of the ribokinase superfamily, Structure 12 (2004) 1809–1821. [25] A.K. Gandhi, M.S. Ghatge, F.N. Musayev, A. Sease, S.O. Aboagye, M.L. di Salvo, V. Schirch, M.K. Safo, Kinetic and structural studies of the role of the active site residue Asp235 of human pyridoxal kinase, Biochem. Biophys. Res. Commun. 381 (2009) 12–15. [26] M.K. Safo, F.N. Musayev, S. Hunt, M.L. di Salvo, N. Scarsdale, V. Schirch, Crystal structure of the PdxY Protein from Escherichia coli, J. Bacteriol. 186 (2004) 8074–8082. [27] T.F. Fu, M. di Salvo, V. Schirch, Distribution of B6 vitamers in Escherichia coli as determined by enzymatic assay, Anal. Biochem. 298 (2001) 314–321. [28] M.L. di Salvo, M.K. Safo, F.N. Musayev, F. Bossa, V. Schirch, Structure and mechanism of Escherichia coli pyridoxine 5′-phosphate oxidase, Biochim. Biophys. Acta 1647 (2003) 76–82. [29] M.K. Safo, F.N. Musayev, M.L. di Salvo, V. Schirch, X-ray structure of Escherichia coli pyridoxine 5′-phosphate oxidase complexed with pyridoxal 5′-phosphate at 2.0 A resolution, J. Mol. Biol. 310 (2001) 817–826. [30] E.S. Yang, V. Schirch, Tight binding of pyridoxal 5′-phosphate to recombinant Escherichia coli pyridoxine 5′-phosphate oxidase, Arch. Biochem. Biophys. 377 (2000) 109–114.

Please cite this article as: M.L. di Salvo, et al., On the mechanism of Escherichia coli pyridoxal kinase inhibition by pyridoxal and pyridoxal 5′phosphate, Biochim. Biophys. Acta (2015), http://dx.doi.org/10.1016/j.bbapap.2015.01.013