Inactivation of tyrosine phenol-lyase by Pictet–Spengler reaction and alleviation by T15A mutation on intertwined N-terminal arm Seung-Goo Lee 1 , Seung-Pyo Hong 2 , Do Young Kim 1 , Jae Jun Song 1 , Hyeon-Su Ro 3 and Moon-Hee Sung 2,4 1 Systems Microbiology Research Center, KRIBB, Daejeon, Korea 2 Bioleaders Corporation, Daejeon, Korea 3 Department of Microbiology and Research Institute of Life Science, KyeongSang National University, Chinju, Korea 4 Department of Bio- and Nanochemistry, Kookmin University, Seoul, Korea Tyrosine phenol-lyase (TPL; EC 4.1.99.2) is a carbon- carbon lyase that catalyzes the a,b-elimination and b-replacement of l-tyrosine and its related amino acids, with pyridoxal-5¢-phosphate (PLP) as the cofac- tor [1]. Meanwhile, at high concentrations of ammo- nium pyruvate, the enzyme catalyzes the synthesis of aromatic amino acids from phenolic substrates through the reverse reaction of a,b-elimination [2,3] (Scheme 1). Application of the enzyme for the synthe- sis of 3,4-dihydroxyphenyl-l-alanine (l-DOPA) from catechol has also attracted particular attention [4–6], because l-DOPA is used as a general medicine for the treatment of Parkinson’s disease [7]. Investigations on the metabolic fate of l-DOPA in biological fluids have discovered the formation of con- densation adducts with endogenous aldehydes, like PLP, Keywords cofactor affinity; L-DOPA; N-terminal arm; Pictet–Spengler condensation; tyrosine phenol-lyase Correspondence M H. Sung, Department of Bio- and Nanochemistry, Kookmin University, Seoul 136-702, Korea Fax: +82 2 910 4415 Tel: +82 2 910 4808 ⁄ 5098 E-mail: smoonhee@kookmin.ac.kr (Received 27 August 2006, revised 16 October 2006, accepted 18 October 2006) doi:10.1111/j.1742-4658.2006.05546.x Citrobacter freundii l-tyrosine phenol-lyase (TPL) was inactivated by a Pictet–Spengler reaction between the cofactor and a substrate, 3,4-dihyd- roxyphenyl-l-alanine (l-dopa), in proportion to an increase in the reaction temperature. Random mutagenesis of the tpl gene resulted in the genera- tion of a Thr15 to Ala mutant (T15A), which exhibited a two-fold improved activity towards l-DOPA as the substrate. The Thr15 residue was located on the intertwined N-terminal arm of the TPL structure, and comprised an H-bond network in proximity to the hydrophobic core between the catalytic dimers. The maximum activity of the mutant and native enzymes with l-DOPA was detected at 45 and 40 °C, respectively, which was 15 °C lower than when using l-tyrosine as the substrate. The half-lives at 45 °C were about 16.8 and 6.4 min for the mutant and native enzymes, respectively, in 10 mml-DOPA. On treatment with excess pyrid- oxal-5¢-phosphate (PLP), the l-DOPA-inactivated enzymes recovered over 80% of their original activities, thereby attributing the inactivation to a loss of the cofactor through Pictet–Spengler condensation with l-DOPA. Consistent with the extended half-life, the apparent Michaelis constant of the T15A enzyme for PLP (K m,PLP ) increased slowly when increasing the temperature, while that of the native enzyme showed a sharp increase at temperatures higher than 50 °C, implying that the loss of the cofactor with the Pictet–Spengler reaction was prevented by the tighter binding and smal- ler release of the cofactor in the mutant enzyme. Abbreviations AspAT, aspartate aminotransferase; IPTG, isopropyl thio-b- D-galactoside; LDH, lactate dehydrogenase; L-DOPA, 3,4-dihydroxyphenyl-L-alanine; PLP, pyridoxal-5¢-phosphate; TNA, tryptophan indole-lyase; TPL, tyrosine phenol-lyase. 5564 FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS excreting tetrahydroisoquinolines in the urine of patients after the oral administration of l-DOPA [8,9]. The for- mation of l-DOPA-PLP cyclic adducts has also been detected in the inactivation of l-DOPA decarboxylase by a substrate [10,11], eventually leading to the dissoci- ation of the cofactor. However, despite extensive studies on TPL as a biocatalyst [2–6,12,13], the inhibitory effect of l-DOPA-PLP adduct formation on the enzymatic synthesis of l-DOPA has not yet been addressed. Structural studies on the enzymes from Citrobacter freundii (PDB entries: 1TPL, 2TPL) and Erwinia herbicola (1C7G) have found them to be composed of four identical subunits, each with one molecule of PLP [14,15]. Each subunit of C. freundii TPL is comprised of an N-terminal arm (residues 1–19), small domain, and PLP-binding large domain. The active site is located in a cleft surrounded by one subunit and the large domain of the adjacent subunit, constituting a catalytic dimer. The two dimers are then tightly combined through a hydrophobic cluster at the center of the tetramer and intertwined N-terminal arms (Fig. 1). The above mentioned architecture is conserved in many a-family PLP-enzymes including tryptophan indole-lyase (TNA; PDB entry: 1AX4) and aspartate aminotransferases (AspAT; PDB entry: 1ARI) [16–18]. In porcine cytosolic aspartate aminotransferase (AspAT), the N-terminal arm protruding toward the large domain of the other subunit is essential for both the catalytic activity and thermal stability of the enzyme [19–21]. Similarly, the AspAT of Bacillus circulans shows a weakened cofactor affinity at the truncation of the N-terminal arm, resulting in a monomeric nonfunc- tional conformation [22]. Meanwhile, structural studies of Proteus vulgaris TNA have revealed an intimate correlation between cofactor binding and the interfacial H-bonds formed on the subunit interface [17]. In this study, a random mutagenesis approach to evolve a robust TPL for l-DOPA synthesis resulted in an effective mutation, T15A, located on the N-terminal arm of C. freundii TPL. Biochemical characterization of the native and mutant enzyme proved the mutation on the interface increased the stability of the catalytic capability of the enzyme by preventing cyclic conden- sation between l-DOPA and PLP (Fig. 1). Results Random mutagenesis and structural identification of T15A mutant An error-prone PCR of C. freundii TPL and subse- quent cloning into Escerichia coli XL1-Blue resulted in a mutant library containing 1–5 mutations that were evenly distributed over the entire TPL sequence. About 10 000 colonies from the library were subjected to rapid screening on microtiter plates with l-DOPA as the substrate. To select a highly active mutant from the library, the activity with l-DOPA was divided by the corresponding activity when using l-tyrosine as the substrate, thereby compensating for a variation in the expression levels. When comparing the normalized activities, mutant #44 was identified as the most active, with a two-fold increased activity with l-DOPA. A sequence analysis of #44 exhibited an amino acid change from Thr15 to Ala, while a structural analysis of C. freundii TPL (1TPL, 2TPL) revealed that Thr15 was located on the intertwined N-terminal arm, com- prising an H-bond network between the catalytic dimers within a proximal distance of the hydrophobic core (Fig. 2A). The hydroxyl group of Thr15 was H-bonded to the sidechain of Lys59, and connected to the sidechain of Asp58 via a water molecule, which was also linked to the backbone nitrogen of Thr15 (Fig. 2B). In addition, the sidechain of Thr15 was involved in nonbonded interactions with the Lys59 and Glu308 sidechains from the other catalytic dimer. Fig. 1. Schematic view of Pictet–Spengler reaction and cofactor release from holo-TPL enzymes. The adductive reaction between L-DOPA and pyridoxal-5¢-phosphate (PLP) leads to the depletion of the cofactor in the reaction solution, inactivating the enzyme depending on the cofactor binding affinity. Scheme 1. Synthesis reaction by TPL. S G. Lee et al. T15A mutation effect on C. freundii TPL cofactor stability FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS 5565 In summary, the proximate interaction of Thr15 with the other subunits suggested that the effect of T15A on the catalytic capability was related to changes in the interdomain architectures of the catalytic dimers. Purification, kinetic parameters, and catalytic stability with L-DOPA as substrate The E. coli XL-1 Blue cells bearing the plasmid pHR1001 or pDA44 revealed a thick protein band with a molecular mass of 52 kDa in an SDS ⁄ PAGE analysis after induction with 1 mm isopropyl thio-b- d-galactoside (IPTG). Based on ammonium sulfate precipitation between 50 and 70% saturation, followed by ion exchange and hydrophobic chromatography, the native TPL and T15A mutant were purified to homogeneity with a recovery yield of 45% and 39%, respectively. The purified proteins were preserved in a refrigerator after being reprecipitated in 70% (NH 4 ) 2 SO 4 , then desalted just before use to recover their original specific activities of around 1.2 and 0.64 unitsÆmg )1 , respectively, with l-tyrosine as the sub- strate. The kinetic parameters were determined in triplicate experiments at 30 °C, with 0.05–1 mml-tyrosine or 0.5–12 mml-DOPA as the substrate. The catalytic rate constants (k cat ) for the native and mutant enzymes with l-DOPA were 0.31 s )1 and 0.68 s )1 , respectively (Table 1), while the Michaelis constants with l-tyrosine were determined as 0.24 and 0.22 mm, respectively, indicating a conserved geometry at the binding site, and with l-DOPA were determined to be 3.2 and 4.6 mm, respectively, yet with larger error limits. Inves- tigations of the substrate range of the TPLs revealed that 3-chloro-l-tyrosine, dl-serine, and dl-cysteine also served as substrates to a lesser extent, whereas d-tyrosine, d-DOPA, dl-tryptophane, dl-phenylalan- ine, and dl-alanine were all inert towards the enzymes. The native and mutant enzymes were then investi- gated for their stability and activity at temperatures between 15 and 80 ° C. When heated for 20 min in the standard buffer, both enzymes remained stable up to 55 °C in a 0.1 m potassium phosphate buffer (pH 8.0) (Fig. 3A). The half-inactivation temperatures for the native and mutant enzymes were calculated to be 62.2 and 65.2 °C, respectively, with a four-parameter sig- moid equation using sigmaplot (Systat Software Inc., Richmond, CA, USA). Plus, the inclusion of two sub- strates for the synthesis of l-DOPA (20 mm catechol and 1.0 m ammonium acetate) increased the half-inac- tivation temperatures to 72.1 and 73.9 °C, respectively, similar to the stabilization of porcine cytosolic AspAT by a-ketoglutarate [20]. The maximum activity for the a,b-elimination of l-tyrosine was observed at 55 and 60 °C for the native and mutant enzymes, respectively (Fig. 3B). However, when l-DOPA was applied as the substrate, the temperatures producing the maximum activity were down-shifted by 15 °C for both the native and mutant enzymes to 40 and 45 °C, respectively (Fig. 3C). A B Fig. 2. Interfacial architectures of catalytic dimers of C. freundii TPL. (A) Overall struc- ture, extracted from prediction of oligomeric states server at EBI (http://pqs.ebi.ac.uk/). Yellow colored molecules represent 3-(4¢- hydroxyphenyl)propionic acid adopted from 2TPL PDB file. (B) Magnified view of red- lined box in overall structure. Green lines represent intimate molecular interactions including hydrogen bond networks in vicinity of Thr15 on intertwined N-terminal arm. H-bond information was extracted from entry code 1TPL of Protein Data Bank. Table 1. Kinetic constants for C. freundii TPL and T15A mutant. Enzymes L-Tyrosine L-DOPA PLP K m (mM) k cat (sec -1 ) k cat ⁄ K m K m (mM) k cat (sec -1 ) k cat ⁄ K m (K m,PLP , lM) Native 0.24 ± 0.1 1.8 ± 0.2 7.5 3.2 ± 0.8 0.31 ± 0.04 0.10 2.0 T15A 0.22 ± 0.03 1.2 ± 0.1 5.5 4.6 ± 1.8 0.68 ± 0.15 0.15 2.5 T15A mutation effect on C. freundii TPL cofactor stability S G. Lee et al. 5566 FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS Inactivation of C. freundii TPL by Pictet–Spengler reaction Cyclic adducts of l-DOPA with endogenous aldehydes have been detected in biological solutions for decades. As such, when the C. freundii TPL (30 lm) was incu- bated with 10 mml-DOPA in a 0.1 m potassium phosphate buffer (pH 7.5) at 30 °C, a time-dependant decrease in the absorbance at 400 nm was detected, resulting in a new absorption peak at 330 nm (Fig. 4A), corresponding to previous literature on the inactivation of l-DOPA decarboxylases when using l-DOPA as the substrate [5,6,23]. When a pseudo first-order kinetic (low initial concentration of the enzyme) was applied for the decolorization rate of C. freundii TPL with l-DOPA, the rate constant (k 1 ) was calculated as 0.012 min )1 using a kinetic equation, log A t A 0 ¼Àk 1 t , where A 0 and A t are the Fig. 3. Effect of temperature on stability and activity of C. freundii TPL and T15A mutant. (A) Stability, (B) activity with L-tyrosine, and (C) activity with L-DOPA as substrates. The stability was evaluated as the remaining activity after the enzymes were incubated in a 100 m M potassium phosphate buffer (pH 8.0) at the indicated tem- peratures for 20 min. The activity with L-tyrosine and L-DOPA as the substrates was measured in the standard reaction mixture for 20 min at different temperatures and the amount of pyruvate pro- duced determined by the salicylaldehyde method. Closed symbols represent native enzyme (d,r) and open symbols represent T15A mutant (s,e). Diamond symbols indicate improved stability in the presence of 20 m M catechol and 1.0 M ammonium acetate. Fig. 4. Pictet–Spengler adduct formation from C. freundii TPL in presence of L-DOPA. (A) Spectral analyses of enzyme solution (30 l M) treated with 10 mML-DOPA in 0.1 M potassium phosphate buffer (pH 7.5) at 30 °C. Inset: Time-dependent decrease in absorb- ance at 400 nm. (B) HPLC analyses of enzyme mixture. After the spectral change was completed, the reaction solution was subjec- ted to centrifugal ultrafiltration (molecular cutoff: 10 000), the fil- trate loaded on a DOWEX 50 W column (pH 3.0), and the eluted solution precipitated with three volumes of isopropanol on ice. The precipitates were dissolved in water, then analyzed by HPLC. The standard was a 2 : 1 mixture of pyridoxal-5¢-phosphate and L-DOPA-PLP adduct synthesized in the authors’ laboratory. S G. Lee et al. T15A mutation effect on C. freundii TPL cofactor stability FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS 5567 absorbance at times 0 and t, respectively (inset in Fig. 4A). After the spectral change was completed, the reaction solution was treated with a cation exchange resin (DOWEX 50 W), analyzed by HPLC, and found to include an l-DOPA-PLP adduct with the same retention time and molecular mass (426 Da) as the Pictet–Spengler type adduct (Fig. 4B) synthesized as described below. Meanwhile, the rate constant (k 1 )of free PLP was 0.12 ± 0.02 min )1 under the pseudo first-order reaction conditions. Consequently, the free cofactor was estimated to be 10-fold more susceptible to adduct formation than the enzyme-bound PLP. The rate constants also increased with the pH and temperature, as previously reported for the reaction between l-DOPA and d -glucose [7]. In particular, the k 1 values increased up to 0.5 ± 0.1 min )1 when 1.0 m ammonium chloride (pH 8.2) was included in the reac- tion solution. As seen in Fig. 3C, the maximum activity of the mutant and native enzymes with l-DOPA as the sub- strate was 15 °C lower than when using l-tyrosine as the substrate, plus both enzymes were similarly stable up to 55 °C. Consequently, because the spectral and kinetic studies on the decolorization of TPL suggested that the compromised activity was closely related to the loss of the coenzyme via adduct formation, an experiment on the stability of the enzyme-bound cofac- tors was performed in a 1.0 m ammonium chloride solution (pH 8.2) with 10 mml-DOPA. During incu- bation at 45 °C, the enzyme mixtures were withdrawn intermittently, 100-fold diluted in an assay solution, and examined for their remaining activity using 1 mm l-tyrosine as the substrate. The remaining activity of the mutant and native enzymes decreased according to the incubation time, down to 30% and 6% of the ini- tial activity (dotted lines in Fig. 5) with a half-life of 16.8 and 6.4 min, respectively. However, when the same samples were assayed in the presence of excess PLP (200 lm), both enzymes recovered over 80% of their original activity (solid lines in Fig. 5), indicating that the inactivation could be attributed to the loss of the cofactor through a con- densation reaction with l-DOPA. Stabilization of cofactor binding by T15A mutation The extended lifetime of T15A in 10 mml-DOPA sug- gested that the intertwined N-terminal architecture, where Thr15 is located, was closely related with the cofactor binding affinity of C. freundii TPL. To verify the effect of the T15A mutation on the cofactor affin- ity, the apparent Michaelis constants for PLP (K m,PLP ) with the native and mutant enzymes were investigated at temperatures ranging from 30 to 60 °C. As shown in Fig. 6A, the K m,PLP for C. freundii TPL increased slowly below 45 °C, accompanied by an increase in the catalytic rate constant (k cat ). However, above 50 °C, the binding constants increased very sharply, while the k cat remained at a similar level (Fig. 6B). An increase in the K m,PLP was also detected with the T15A mutant, yet significantly slower than that with the native enzyme (Fig. 6A). As such, the cofactor release from the active site was increased rel- ative to the temperature, likely accelerating the adduct formation with l-DOPA, yet this was significantly relieved by the T15A mutation located on the inter- twined N-terminal arm. Finally, the effect of the T15A mutation on l-DOPA synthesis was investigated in a reaction solution (10 mL) including 0.65 m ammonium chloride (pH 8.5), 50 mm sodium pyruvate, 50 mm catechol, 0.1 mm PLP, 0.1% sodium sulfite, and 15 units of the enzyme. In addition, because alcoholic additives have been shown to be beneficial for the synthesis of l-DOPA by C. freundii TPL [3], 10% ethanol was also included in the reaction solution. When the synthesis reaction was carried out for 2.5 h at 45 °C, the concentration of l-DOPA increased rapidly up to the maximum level within an hour, then slightly decreased (Fig. 7), prob- ably because of the adduct formation between l-DOPA and remaining pyruvate [2,5]. However, the upward curve flattened much earlier with C. freundii TPL, consequently the l-DOPA productivity of T15A was at least two-fold higher than that with C. freundii Fig. 5. Inactivation of TPL enzymes by L-DOPA and its reactivation by PLP. Timecourse profiles of inactivation and activity recovery in presence of excess pyridoxal 5¢-phosphate. An enzyme mixture containing 0.01 unitsÆml )1 of TPL, 0.1 mM PLP, and 10 mML-DOPA in a 100 m M potassium phosphate buffer (pH 8.0) was incubated at 45 °C for different times, and the remaining activity determined in the presence of excess PLP (d,s) or without PLP (r,e). Closed symbols represent native enzyme and open symbols represent T15A mutant. T15A mutation effect on C. freundii TPL cofactor stability S G. Lee et al. 5568 FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS TPL, consistent with the robustness of T15A at eleva- ted temperatures (Figs 5 and 6C). No oxidation of l-DOPA was detected while the solutions were flushed with nitrogen gas. Discussion The enzymatic synthesis of l-DOPA using E. herbicola TPL is more successful at a low temperature range from 15 to 24 °C [5,24]. Likewise, with C. freundii TPL, the synthesis was facilitated at 18 °C [2], although the enzyme activity was about 20% of the maximal activity (Fig. 3C). The compromised produc- tivity at high temperatures has been attributed to the formation of byproducts and the oxidative deterior- ation of catechol or pyruvate during the reaction, all of which are accelerated by the temperature [2,5]. In this study, it was postulated that the lowered pro- ductivity of C. freundii TPL at elevated temperatures partly resulted from a decolorization reaction in the enzyme mixture, which eventually led to the depletion of the cofactor PLP, accompanied by the inactivation of the enzyme. HPLC and 1 H NMR analyses of the purified adduct revealed that the inactivation resulted from Pictet–Spengler type condensation between l-DOPA and PLP. Notwithstanding previous reports on the inactivation of PLP enzymes, aromatic decarb- oxylases, by l-DOPA [10,23], this is the first time the rapid inactivation of TPL has been explained based on a Pictet–Spengler reaction. Consistent with the observation that a Pictet–Spen- gler reaction is accelerated relative to the reaction temperature [8], the optimal temperature for enzyme activity in the presence of l-DOPA was 15 °C lower than that with l-tyrosine as the substrate (Fig. 3B,C). The inactivation profile of the enzyme with 10 mm l-DOPA (Fig. 5) also agreed with the optimal tem- perature results. Meanwhile, the incubation of TPL with d-DOPA, a stereoisomer of l-DOPA that does Fig. 6. Effect of temperature on kinetic constants for C. freundii TPL and T15A mutant. (A) Apparent binding constant (K m,PLP ) for PLP, (B) catalytic rate constant (k cat ), and (C) ratio of k cat ⁄ K m , PLP . The kinetic constants were determined from double reciprocal plots of the reaction rate versus the PLP concentration at different tem- peratures. Closed symbols represent K m , PLP values for C. freundii TPL, while open symbols represent K m , PLP values for T15A mutant. Fig. 7. Synthesis of L-DOPA by C. freundii TPL and T15A mutant. The synthetic reaction was carried out using partially purified enzymes in a solution (10 mL) containing 0.65 M ammonium chlor- ide (pH 8.5), 50 m M sodium pyruvate, 50 mM catechol, 0.1 mM PLP, 0.1% sodium sulfite, 10% ethanol and 15 units of enzyme. The reaction bottle was flushed with nitrogen gas, tightly sealed with a rubber stopper, and incubated at 45 °C. Samples were with- drawn using a syringe in a stream of nitrogen gas to prevent oxida- tion of the ingredients. Closed symbols represent native enzyme and open symbols represent T15A mutant. S G. Lee et al. T15A mutation effect on C. freundii TPL cofactor stability FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS 5569 not serve as a substrate for the enzyme reaction, pro- duced a similar effect to l-DOPA, indicating that the adduct-forming reaction was independent of the enzyme reaction and a chemical reaction between l-DOPA and the free PLP released from the active site. The release of PLP from the enzyme was acceler- ated at an elevated temperature, as shown by the K m,PLP versus temperature profile of the native enzyme (Fig. 6A). The enzyme-bound PLP reacted with the l-DOPA in the reaction mixture to form an l-DOPA- PLP adduct at a rate of 0.012 min )1 , as shown by the inset in Fig. 4A, which was 10 times slower than the rate with the free PLP and l-DOPA (Fig. 1). The removal of PLP by release and the subsequent Pictet– Spengler reaction may have been responsible for the rapid decrease in the k cat ⁄ K m,PLP value of the native enzyme at temperatures above 45 °C (Fig. 6C). Note that the k cat ⁄ K m,PLP value was the catalytic rate in the presence of a limited concentration of PLP. In contrast, the K m,PLP -value for T15A was less sensitive to the temperature (Fig. 6A), suggesting a tight binding of the cofactor at the enzyme active site. Therefore, the mutation on the intertwined N-terminal arm stabilized the cofactor binding affin- ity, thereby improving the catalytic properties at elevated temperatures (Fig. 7), as indicated by the higher stability of the k cat ⁄ K m,PLP value for the T15A enzyme (Fig. 6C). Citrobacter freundii TPL has a 50% sequence iden- tity with the tryptophanase from P. vulgaris, which degrades tryptophan to indole, ammonia, and pyruvate [14,25]. The secondary, tertiary, and quaternary struc- tures are also highly conserved, plus a hydrophobic cluster and intertwined N-terminal arms are formed on the intersubunit interface, contributing to its stability. The network of hydrogen bonds and salt bridges formed upon the binding of PLP is known to influence the quaternary structure of tryptophanases [17]. There- fore, when considering the common structural features of a-family PLP enzymes [26,27], the T15A mutation on the N-terminal arm may have increased the rigidity of the cofactor binding architecture of C. freundii TPL through adjusting the quaternary interfaces. One poss- ible communication between the N-terminal arm and the active site is through Tyr71, which belongs to the adjacent subunit of the catalytic dimer. Tyr71 is known to be essential for activity, as a general acid catalyst for the elimination of the leaving group from a quinonoid intermediate, and also for PLP binding [28]. The PLP binding constant for the Y71F mutant of C. freundii TPL was estimated to be 1 mm, while the wildtype TPL showed a binding constant of 0.6 lm based on spectrophotometric titration. Consistently, the equivalent Tyr70 in aspartate aminotransferase also has a PLP binding function [29]. Thus, this study demonstrated that the deterioration of the cofactor through a Pictet–Spengler reaction with l-DOPA appeared to be a significant interference with the biotechnological production of l-DOPA when using C. freundii TPL. The T15A mutation improved the cofactor binding affinity at high temperatures, along with the apparent turnover rate when using l-DOPA as the substrate, through an interfacial inter- action between the N-terminal arm and the cleft active site. However, l-DOPA synthesis at a high temperature also increases the adduct formation between l-DOPA and a substrate pyruvate [2,5], eventually decreasing the l-DOPA concentration during a prolonged reaction at a high temperature, as observed in Fig. 7. Thus, despite the increased catalytic efficiency and stability of the T15A mutant, l-DOPA synthesis at a high temperature should be further scrutinized to minimize the adduct formation between l-DOPA and pyruvate. For example, a continuous limited supply of pyruvate into the reaction solution could be used to maintain the pyruvate concentration at a minimal level, thereby decreasing the adduct formation rate. In addition, based on the effect of alcohols [3], the reaction ingre- dients could also be optimized to increase the l-DOPA synthesis and relieve the adduct formation. Consequently, with its enhanced l-DOPA synthesis activity and stability, the T15A enzyme of this work could be used for the development of a new bioconver- sion strategy for the efficient production of l-DOPA at high temperatures, where it can catalyze the reaction more actively. Experimental procedures Materials The PLP was purchased from Sigma (St Louis, MO, USA) and the l-DOPA purchased from Boehringer Mannheim (Mannheim, Germany). The restriction endonucleases and T4 DNA ligase were purchased from New England Biolabs (Beverly, MA, USA) and the Taq DNA polymerase from PerkinElmer (Branchburg, NJ, USA). The oligonucleotides were synthesized at Bioneer Co. (Daejeon, Korea) and the DNA sequencing performed at Solgent Co. (Daejon, Korea). The l-DOPA-PLP adduct was synthesized by mixing l-DOPA (0.32 g) and PLP (0.2 g) in a 50 mm sodium phosphate buffer (80 mL, pH 8.0) at 45 °C for 30 min. The reaction product was purified on a DOWEX 50 W column (pH 3.0, Sigma) and the eluted solution precipitated with isopropanol (240 mL) on ice for 2 h. The precipitates were then washed on a sintered glass filter with acetone and T15A mutation effect on C. freundii TPL cofactor stability S G. Lee et al. 5570 FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS stored in a deep freezer after vacuum-drying. The molecular mass of the adduct was 426 Da on a ESI-MS spectrometer, and the chemical shift values in D 2 O determined by 300 MHz 1 H NMR experiments were as follows: d 2.36 (3 H, s, H-2¢), 3.2 (2 H, m, H-b), 4.0 (1 H, m, H-a), 4.91 (2 H, d, H-5¢), 5.77 (1 H, s, H-4¢), 6.20 (1H, s, H-5¢¢), 6.69 (1 H, s, H-2¢¢), and 7.74 (1 H, s, H-6). The chemical struc- ture of the adduct was identified as shown in Fig. 1. All other chemicals used were chemical reagent grade. Random mutagenesis and screening on microtiter plates The plasmid pHR1001 harboring the C. freundii tpl gene (gene bank accession no. DQ907529) [3] was used as the template for an error-prone PCR with the following prim- ers: 5¢-AATTATCCGGCAGAACCCTT-3¢ (forward) and 5¢-GATC AAGCTTTTAGATATAGTCAAAGCGTGC-3¢ (reverse, underlined HindIII). The thermal cycling was per- formed using a DNA Thermal Cycler (PerkinElmer): 5 min at 95 °C, a subsequent 25 cycles of 1 min at 95 °C, 2 min at 50 °C, 3 min at 72 °C, and a final extension of 7 min at 72 °C. The amplified PCR products were digested with Hin- dIII to yield a 1.37 kb DNA fragment. The plasmid pTrc99A was then digested with NcoI, blunt-ended by Klenow treatment, and digested with HindIII. The resulting plasmid was ligated with the HindIII-treated PCR product by blunt-cohesive ligation at 16 °C with a T4 DNA ligase. E. coli XL1 Blue cells were then transformed with the ligate by electroporation and spread on LB-ampicillin plates. After being incubated overnight at 37 °C, the evolved colonies were transferred by toothpick to fresh LB-ampicillin plates. The mutant library was inoculated into an LB-ampicillin- IPTG medium (500 lL) contained in a deep 96-well plate, and cultivated in a wellplate culture system, Megagrow TM (Bioneer Co.). The cultivated cells were centrifuged at 5000 g for 20 min with a wellplate centrifuge Union 5KR TM , rotor type TM96-65 (Hanil Sci. Ind., Inchon, Korea), washed in a 50 mm Tris ⁄ HCl buffer (pH 8.0), and treated with 200 lL Cellytic TM B (Sigma) for 1 h at 37 °C. The cell lysate (100 lL) was then transferred into 96-well PCR plates and mixed with the same amount of substrate solutions, including 10 mml-DOPA or 1 mml-tyrosine, and 20 lm PLP in a 50 mm potassium phosphate buffer (pH 8.0). After being incubated at 37 °C for 20 min, the reaction solutions were heated for 3 min at 95 °C, centri- fuged at 5000 g for 20 min with Union 5KR TM centrifuge to remove any insoluble aggregates, and analyzed for pyruvate formation using the salicylaldehyde method [25] to compare the enzyme activities towards l-DOPA and l-tyrosine. Expression and purification Escerichia coli XL-1 Blue cells harboring pHR1001 or pDA44 were cultivated at 37 °C for 16 h in 1 litre of an LB medium containing 100 lgÆmL )1 ampicillin. Protein expres- sion was induced by the addition of 1 mm IPTG when the absorbance at 600 nm reached 0.5. The harvested cells were then disrupted by sonification in a standard buffer, inclu- ding 0.01% 2-mercaptoethanol, 0.05 mm PLP, and 50 mm Tris ⁄ HCl (pH 8.0). The centrifugation supernatant was col- lected, and subjected to ammonium sulfate fractionation between 50% and 70% saturation. The enzyme dissolved in the standard buffer was then loaded on to a Resource Q ion exchange (Pharmacia, Uppsala, Sweden), washed with the standard buffer, and eluted using a KCl gradient from 0 to 0.5 m. Most of the active fractions were then pooled, adjusted to include 1.7 m (NH 4 ) 2 SO 4 , and loaded on to a Phenyl Superose (Pharmacia). The elution from the hydro- phobic column was performed using a reverse gradient of (NH 4 ) 2 SO 4 from 1.7 m to 0 m, then the active fractions were dialyzed against a 100 mm Tris ⁄ HCl buffer (pH 8.0) containing 0.2 m KCl, reprecipitated in 70% saturated (NH 4 ) 2 SO 4 , and stored in a refrigerator. All the column procedures were carried out using an AKTA system (Amer- sham Bioscience, Uppsala, Sweden) at room temperature. Determination of kinetic parameters and cofactor binding affinity The kinetic constants for l-DOPA and l-tyrosine as sub- strates were determined using a lactate dehydrogenase (LDH)-coupled assay of the pyruvate formation rate. The reaction was started by the addition of 0.05–1.0 mml-tyro- sine or 0.5–12 mml-DOPA as the substrate, and the decrease in A 340 monitored at 30 °C using a spectrophoto- meter, Ultrospec3000 (Pharmacia Biotech, Uppsala, Sweden), equipped with a Peltier cuvette-heating system. The pyruvate formation rate was calculated using the extinction coefficient of NADH (6200 m )1 Æcm )1 ) from the slope between 0.5 and 5.0 min, after the early perturbation of the absorbance was settled. The apparent binding constants of PLP to the enzymes were presumed as the concentration of PLP for half the maximal activity of the enzyme. The assay mixture with dif- ferent PLP concentrations (0.5–200 lm) and 2.5 mml-tyro- sine was equilibrated to different temperatures for 5 min in a thermo-controlled spectrophotometer, and the enzyme activity measured using an LDH coupling assay, as des- cribed above. The apparent binding constants for PLP (K m,PLP ) were calculated from a double reciprocal plot of the reaction rate (v) versus the PLP concentration: m V max ¼ ½PLP K m;PLP þ½PLP , where V max is the maximum reaction rate at saturating PLP concentrations. All the kinetic experiments were performed in triplicate. Enzyme assay and analysis The a,b-elimination activity of TPL was calculated from the pyruvate formation rate determined by a coupling assay S G. Lee et al. T15A mutation effect on C. freundii TPL cofactor stability FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS 5571 with LDH (Roche Diagnostics, Bazel, Switzerland) or using the salicylaldehyde method [25]. The standard reaction mix- ture contained 10 mml-DOPA or 1 mml-tyrosine as the substrate, 50 l m PLP, 0.2 mm NADH, 10 lgÆmL )1 LDH, and TPL in a 0.1 m potassium phosphate buffer (pH 8.0). One unit of enzyme was defined as the activity to catalyze the formation of 1 lmol of pyruvate per min at 30 °C. The protein concentration was determined using a Bradford rea- gent (Bio-Rad, Hercules, CA, USA) with bovine serum albumin as the standard. The analysis of the l-DOPA-PLP adduct was performed on a HPLC system (Young-in Co., Seoul, Korea) equipped with an ODS18 column (Shimazu, Kyoto, Japan) and UV- detector (295 nm). The elution was carried out using a co- solvent consisting of a 50 mm potassium phosphate buffer with 2 mm sodium dodecylsulfate (pH 3.0), methanol, and acetonitrile (volumetric ratio ¼55 : 40 : 5) at a flow rate of 0.6 mLÆmin )1 . Acknowledgements This project was supported by a grant from the Clea- ner Production Program 10007946 of NCPC, the KRIBB Research Initiative Program, and the 2006 research fund of Kookmin University, Korea. References 1 Kumagai H, Yamada H, Matsui H, Ohkishi H & Ogata K (1970) Tyrosine phenol lyase. I. purification, crystallization, and properties. J Biol Chem 245, 1767–1772. 2 Yamada H & Kumagai H (1975) Synthesis of 1-tyro- sine-related amino acids by beta-tyrosinase. Adv Appl Microbiol 19, 249–288. 3 Lee SG, Ro HS, Hong SP, Lee KJ, Wang JW, Tae DN, Uhm KN, Bang SG, Kim YJ & Sung MH (1996) Pro- duction of 3,4-dihydroxyphenyl-l-alanine by using the b-tyrosinase of Citrobacter freundii overexpressed in recombinant Escherichia coli. Kor J Appl Microbiol Biotechnol 24, 44–49. 4 Ajinomoto Co. Inc. (1993) Method for producing l-3,4- dihydroxyphenylalanine by Erwinia having tyrosinase activity. US Patent 05338672. 5 Enei H, Nakazawa H, Okumura S & Yamada H (1973) Synthesis of 1-tyrosine and 3,4-dihydroxyphenyl-l-ala- nine from pyruvic acid, ammonia, and phenol or pyro- catechol. Agr Biol Chem 37, 725–735. 6 Foor F, Morin N & Bostian KA (1993) Production of 1-dihydroxyphenylalanine in Escherichia coli with the tyrosine phenol-lyase gene cloned from Erwinia herbi- cola. Appl Environ Microbiol 59, 3070–3075. 7 Dougan D, Wade D & Mearric PK (1975) Effects of l-DOPA metabolites at a dopamine receptor suggest a basis for ‘on-off’ effect in Parkinson’s disease. Nature 254, 70. 8 Manini P, Ischia M & Prota G (2001) Pictet-Spengler condensation of the antiparkinsonian drug l-DOPA with d-glyceraldehyde. Opposite kinetic effects of Fe 3+ and Cu 2+ ions and possible implications for the origin of therapeutic side effects. Bioorg Med Chem 9, 923–929. 9 Sandler M, Carter SB, Hunter KR & Stern GM (1973) Tetrahydroisoquinoline alkaloids: in vivo metabolites of l-DOPA in man. Nature 241, 439–443. 10 Bertoldi M & Voltattorn CBI (2000) Reaction of dopa decarboxylase with 1-aromatic amino acids under aero- bic and anaerobic conditions. Biochem J 352, 533–538. 11 Nishino J, Hayashi H, Ishii S & Kagamiyama H (1997) An anomalous side reaction of the Lys303 mutant aro- matic 1-amino acid decarboxylase unravels the role of the residue in catalysis. J Biochem 121, 604–611. 12 Faleev NG, Ruvinov SB, Demidkina TV, Myagkikh I, Gololobov M, Bakhmutov VI & Belikov VM (1988) Tyrosine phenol-lyase from Citrobacter intermedius: fac- tors controlling substrate specificity. Eur J Biochem 177, 395–401. 13 Faleev NG, Spirina SN, Ivoilov VS, Demidkina TV & Phillips RS (1996) The catalytic mechanism of tyrosine phenol-lyase from Erwinia herbicola: the effect of sub- strate structure on pH-dependence of kinetic parameters in the reactions with ring-substituted tyrosines. Z Natur- forsch [C] 51, 363–370. 14 Antson AA, Demidkina TV, Gollnick P, Dauter Z, von Tersch RL, Long J, Berezhnoy SN, Phillips RS, Har- utyunyan EH & Wilson KS (1993) Three-dimensional structure of tyrosine phenol-lyase. Biochemistry 32, 4195–4206. 15 Sundararaju B, Antson AA, Phillips RS, Demidkina TV, Barbolina MV, Gollnick P, Dodson GG & Wilson KS (1997) The crystal structure of Citrobacter freundii tyrosine phenol-lyase complexed with 3-(4-hydroxyphe- nyl) propionic acid, together with site-directed mutagen- esis and kinetic analysis, demonstrates that arginine 381 is required for substrate specificity. Biochemistry 36, 6502–6510. 16 Alexander FW, Sandmeier E, Mehta PK & Christen P (1994) Evolutionary relationships among pyridoxal-5¢- phosphate-dependent enzymes. regio-specific a, b, and ! families. Eur J Biochem 219, 953–960. 17 Isupov MN, Antson AA, Dodson EJ, Dodson GG, Dementieva IS, Zakomirdina LN, Wilson KS, Dauter Z, Lebedev AA & Harutyunyan EH (1998) Crystal structure of tryptophanase. J Mol Biol 276, 603–623. 18 Iwamori S, Yoshino S, Ishiwata K & Makiguchi N (1991) Structure of tyrosine phenol-lyase genes from Citrobacter freundii and structural comparison with tryptophanase from Escherichia coli. J Ferment Bioeng 72, 147–151. T15A mutation effect on C. freundii TPL cofactor stability S G. Lee et al. 5572 FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS 19 Ford GC, Eichele G & Jansonius JN (1980) Three- dimensional structure of a pyridoxal-phosphate-depen- dent enzyme, mitochondrial aspartate aminotransferase. Proc Natl Acad Sci USA 77, 2559–2563. 20 Fukumoto Y, Tanase S, Nagashima F, Ueda S, Ikegami K & Morino Y (1991) Structural and functional role of the amino-terminal region of porcine cytosolic aspartate aminotransferase. J Biol Chem 266, 4187–4193. 21 Sandmeier E & Christen P (1980) Mitochondrial aspar- tate aminotransferase 27 ⁄ 32-410. Partially active enzyme derivative produced by limited proteolytic cleavage of native enzyme. J Biol Chem 255, 10284–10289. 22 Kravchuk Z, Tsybovsky Y, Koivulehto M, Vlasov A, Chumanevich A, Battchikova N, Martsev S & Korpela T (2001) Truncated aspartate aminotransfer- ase from alkalophilic Bacillus circulans with deletion of N-terminal 32 amino acids is a non-functional monomer in a partially structured state. Protein Eng 14, 279–285. 23 Bertoldi M, Frigeri P, Paci M & Voltattorni CB (1999) Reaction specificity of native and nicked 3,4-dihydroxy- phenylalanine decarboxylase. J Biol Chem 274, 5514– 5521. 24 Koyanagi T, Katayama T, Suzuki H, Nakazawa H, Yokozeki K & Kumagai H (2005) Effective production of 3,4-dihydroxyphenyl-l-alanine (l-DOPA) with Erwi- nia herbicola cells carrying a mutant transcriptional regulator TyrR. J Biotechnol 115, 303–306. 25 Berntsson S (1955) Spectrophotometric determination of pyruvic acid by the salicylaldehyde method. Anal Biochem 27, 1659–1660. 26 Phillips RS, Demidkina TV & Faleev NG (2003) Struc- ture and mechanism of tryptophan indole-lyase and tyro- sine phenol-lyase. Biochim Biophys Acta 1647, 167–172. 27 Phillips RS, Chen HY, Shim D, Lima S, Khadijeh K & Sundararaju B (2004) Role of lysine-256 in Citrobacter freundii tyrosine phenol-lyase in monovalent cation acti- vation. Biochemistry 43, 14412–14419. 28 Chen HY, Demidkina TV & Phillips RS (1995) Site- directed mutagenesis of tyrosine-71 to phenylalanine in Citrobacter freundii tyrosine phenol-lyase: evidence for dual roles of tyrosine-71 as a general acid catalyst in the reaction mechanism and in cofactor binding. Biochemis- try 34, 12276–12283. 29 Toney MD & Kirsch JF (1991) Tyrosine 70 fine-tunes the catalytic efficiency of aspartate aminotransferase. Biochemistry 30, 7456–7461. S G. Lee et al. T15A mutation effect on C. freundii TPL cofactor stability FEBS Journal 273 (2006) 5564–5573 ª 2006 The Authors Journal compilation ª 2006 FEBS 5573 . Inactivation of tyrosine phenol-lyase by Pictet–Spengler reaction and alleviation by T15A mutation on intertwined N-terminal arm Seung-Goo. the loss of the cofactor through a con- densation reaction with l-DOPA. Stabilization of cofactor binding by T15A mutation The extended lifetime of T15A in