Structure-guided alteration of coenzyme specicity of formate dehydrogenase by saturation mutagenesis to enable efcient utilization of NADP+

Aggeliki Andreadeli1, Dimitris Platis1, Vladimir Tishkov2, Vladimir Popov3 and Nikolaos E. Labrou1
1 Laboratory of Enzyme Technology, Department of Agricultural Biotechnology, Agricultural University of Athens, Greece 2 Department of Chemical Enzymology, Faculty of Chemistry, M.V. Lomonosov Moscow State University, Russia 3 Institute of Biochemistry, Russian Academy of Sciences, Moscow, Russia

Keywords coenzyme; coenzyme specicity; enzyme redesign; formate dehydrogenase; saturation mutagenesis Correspondence N. E. Labrou, Laboratory of Enzyme Technology, Department of Agricultural Biotechnology, Agricultural University of Athens, 75 Iera Odos, GR 118 55 Athens, Greece Fax Tel: +30 210 5294308 E-mail: lambrou@aua.gr (Received 18 February 2008, revised 29 May 2008, accepted 2 June 2008) doi:10.1111/j.1742-4658.2008.06533.x

Formate dehydrogenase from Candida boidinii (CboFDH) catalyses the oxidation of formate anion to carbon dioxide with concomitant reduction of NAD+ to NADH. CboFDH is highly specic to NAD+ and virtually fails to catalyze the reaction with NADP+. Based on structural information for CboFDH, the loop region between b-sheet 7 and a-helix 10 in the dinucleotide-binding fold was predicted as a principal determinant of coenzyme specicity. Sequence alignment with other formate dehydrogenases revealed two residues (Asp195 and Tyr196) that could account for the observed coenzyme specicity. Positions 195 and 196 were subjected to two rounds of site-saturation mutagenesis and screening and enabled the identication of a double mutant Asp195Gln Tyr196His, which showed a more than 2 107-fold improvement in overall catalytic efciency with NADP+ and a more than 900-fold decrease in the efciency with NAD+ as cofactors. The results demonstrate that the combined polar interactions and steric factors comprise the main structural determinants responsible for coenzyme specicity. The double mutant Asp195Gln Tyr196His was tested for practical applicability in a cofactor recycling system composed of cytochrome P450 monooxygenase from Bacillus subtilis, (CYP102A2), NADP+, formic acid and x-(p-nitrophenyl)dodecanoic acid (12-pNCA). Using a 1250-fold excess of 12-pNCA over NADP+ the rst order rate constant was determined to be equal to kobs = 0.059 0.004 min)1.

NAD+-dependent formate dehydrogenase (FDH; EC 1.2.1.2) plays an important role in methylotrophic yeasts, catalysing the nal step in the methanol oxidation pathway [1,2]. In general, FDH is a highly conservative enzyme. The primary sequence homology is no < 8085% between enzymes of the same organismal group and 5055% or more between enzymes from different groups. In addition, the catalytic amino acids, as well as the amino acids that contribute to the structural stability, are almost totally conserved (sequence

homology of approximately 95%). Formate dehydrogenase from Candida boidinii (CboFDH) is a dimer with two identical subunits, each possessing an independent active centre [1,3]. Detailed studies of FDH are justied by the considerable biotechnological potential of this enzyme. For example, it is used as a diagnostic enzyme for the determination of formic acid [4] and oxalic acid [5] in solution and in physiological uids. FDH has been used for the development of a NAD+ regeneration

Abbreviations 12-pNCA, x-(p-nitrophenyl)dodecanoic acid; CboFDH, Candida boidinii FDH; CYP102A2, cytochrome P450 monooxygenase from Bacillus subtilis; FDH, formate dehydrogenase; G6PDH, glucose 6-phosphate dehydrogenase; PTDH, phosphite dehydrogenase.

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

3859

Engineering coenzyme specicity of CboFDH

A. Andreadeli et al.

system in organic synthesis for the production of added-value products [611]. According to prescriptions of the Food and Drug Administration, the optical purity of all chiral compounds used as drugs has to be no < 99%. Dehydrogenases can be used to produce optically active compounds from nonchiral ones because they are extremely stereospecic in the transfer of hydride ion between the substrate and coenzyme (optical purity in the range 99.999.99%) [12]. Because of the high price of NADH and especially of NADPH, processes involving only dehydrogenases are economically unprotable. To decrease the cost of the process, a coenzyme regeneration system has been developed [10]. Different approaches have been evaluated (chemical, photochemical, electrochemical, enzymatic) [13] towards the regeneration of coenzymes, with the latter showing the most promise [10]. In this way, a coupled reaction can be used where NAD(P)+ is converted to NAD(P)H by an enzyme towards the production of a commercial product and simultaneously regenerated by another dehydrogenase using NAD(P)H as its coenzyme [69,14]. The reaction catalyzed by FDH is suitable for the system of NADH regeneration. However, NADP+specic FDHs have not yet been found in nature and this prevents such a method being used for the regeneration of NADPH. Therefore, protein engineering of FDH with the aim of transforming coenzyme specicity (NADP+ versus NAD+) is of practical importance. Previous attempts to alter the specicity of FDH from Pseudomonas sp. 101, Saccharomyces cerevisae and Candida methylica towards NADP+ by specic site-directed mutagenesis have yielded promising results [1517]. In the present study, we describe the site-specic evolution of CboFDH in an attempt to alter the coenzyme specicity. In this way, we constructed a library of CboFDH mutants, which were screened for their ability to utilize NADP+ as a coenzyme.

Results and Discussion

Selection of residues for mutagenesis An intriguing phenomenon in enzymology is the discrimination that pyridine nucleotide-dependent enzymes can make between NADP(H) and NAD(H). The only difference between NAD+ and NADP+ lies in the absence or presence of the phosphate group, respectively. Nevertheless, NAD+- and NADP+dependent dehydrogenases exhibit a pronounced specicity for either one of these coenzymes [18,19]. Several attempts have been made to structurally position the
3860

coenzyme-specicity determining region in dehydrogenases [2024]. These studies have shown that the occurrence of negatively charged aspartic or glutamic acid residue (usually occupied by hydrophobic residues in NADP+-dependent dehydrogenases) at the conserved ngerprint sequence GXGXXGX17-18D(E) in the beta-alpha-beta NAD+-binding region is critical for NAD+-dependent dehydrogenases [18,19]. The presence of positively charged amino acid residues, most likely for charge stabilization purposes, at the binding site of 2-phosphate group of NADP+ is critical for NADP+-dependent dehydrogenases [18,19]. Additionally, as previously demonstrated by Clermont et al. [24] in glyceraldehyde 3-phosphate dehydrogenase, conformation factors may also come into play regarding the size of the coenzyme binding pocket with respect to accommodation of the additional 2-phosphate group of NADP+. To identify amino acids that account for the strict NAD+ specicity of CboFDH, its sequence was aligned with the sequences of other FDH enzymes from different organisms (bacteria, plants, yeasts and fungi). The results reveal highly conserved residues in the beta-alpha-beta NAD+-binding region (Fig. 1). For example, Asp195 (numbering according to CboFDH) is a strictly conserved residue and interacts with the 2- and 3-OH groups of adenosine ribose (Fig. 2A), suggesting that this residue is a major determinant of NAD+ specicity. This Asp residue is located 18 residues downstream from the Gly residue at the end of the ngerprint sequence in yeast FDHs. The bacterial and plant sequences have the conserved Asp as the 17th residue downstream from the end of the ngerprint. Position 196 is occupied by a tyrosine residue in CboFDH (Figs 1 and 2A). Tyr196 is only conserved in yeast and fungi enzymes and is replaced by Arg in bacterial and plant enzymes (Fig. 1). This residue is located at the entrance of the coenzyme binding site and probably prevents NADP+ binding either by unfavourable interactions (e.g. charge repulsion through its p-electron cloud) or by sterically blocking 2-phosphate group binding. Therefore, the positions 195 and 196 were subjected to two consecutive site-specic saturation mutagenesis rounds. Site-saturation mutagenesis at position 195 Saturation mutagenesis is used to explore additional pathways and enable rapid diversication in protein traits [25,26]. This method is completely independent of the mutational bias of DNA polymerases and makes possible the creation of a library of mutants

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

A. Andreadeli et al.

Engineering coenzyme specicity of CboFDH

Fig. 1. Aminoacid sequence alignment of NAD+-binding region of FDHs. Abbreviations and NCBI accession codes are in parenthesis. FDH from yeasts: Candida boidinii (FDH_Cbo; accession code: CAA09466), Saccharomyces cerevisiae (FDH_Sce; accession code: NP_015033), Pichia angusta (FDH_Pan; accession code: P33677), Candida albicans (FDH_Cal; accession code: XP_711169). FDH from bacteria: Pseudomonas sp. 101 (FDH_Psp; accession code: P33160), Thiobacillus sp. KNK65MA (FDH_Tsp; accession code: BAC92737). FDH from plants: Arabidopsis thaliana (FDH_Ath; accession code: NP_196982), Solanum tuberosum (FDH_Stu; accession code: Q07511), Oryza sative (FDH_Osa; accession code: NP_001057666), Quercus robur (FDH_Qro; accession code: CAE12168), Hordeum vulgare (FDH_Hvu; accession code: Q9ZRI8). FDH from fungi: Ajellomyces capsulata (FDH_Aca; accession code: AAV67968), Aspergillus nidulans (FDH_Ani; accession code: XP_664129), Magnaporthe grisea (FDH_Mgr; accession code: AAW69358), Gibberella zeae (FDH_Gze; accession code: XP_386303), Neurospora crassa (FDH_Ncr; accession code: XP_961202). Conserved areas are shown shaded. A column is framed if more than 70% of its residues are similar according to physico-chemical properties. The conserved ngerprint sequence is indicated by asterisks and residues subjected to mutagenesis are identied by crosses above them. The alignments were produced using CLUSTALW. The secondary structure of FDH from C. boidinii (PDB code 2FSS) and its numbering is shown above the alignment. Alpha helices and beta strands are represented as helices and arrows, respectively, and beta turns are marked by TT.

A
Tyr196

Asp195

B
His196

Gln195
Fig. 2. Structural representations of the wild-type and double mutant enzyme. (A) Diagram of the modelled wild-type enzyme with NAD+ (dark grey) bound to the nucleotide binding site. (B) Diagram of the modelled double mutant enzyme (Asp195Gln Tyr196His) with NADP+ (dark grey) bound to the nucleotide binding site. The mutated residues (light grey) are shown in a stick representation and labelled.

containing all possible mutations at one or more pre-determined target positions, in order to determine the best-t residue at that position [27].

In the rst engineering round, amino acid residue at position 195 was subjected to saturation mutagenesis. The generated library of plasmids were transformed in Escherichia coli and the recombinant clones were screened for their ability to utilize NADP+ NAD+ using activity assays. The wild-type and the mutated enzymes were expressed in E. coli as soluble proteins as previously described [3]. The expression levels of the wild-type and mutated enzymes were between 2030% of total E. coli soluble proteins. Four clones displaying the highest NADP+ NAD+ activity ratio were selected for further study. Sequencing analysis conrmed that the four clones contained a single point mutation at position 195. In general, the substitutions were amino acids with relatively small, non-aromatic side-chains: Ser, Ala, Gln and Asn. These mutants along with the wild-type enzyme were puried and subjected to steady-state kinetic analysis and the results are presented in Table 1. Wild-type CboFDH has a clear preference for NAD+ as coenzyme [(kcat Km)NAD+ (kcat Km)NADP+) > 2.3 108] and is almost indifferent towards utilizing NADP+ NADP+ NAD+ )9 [(kcat Km) (kcat Km) ) < 4.26 10 ]. Kinetic analysis of the rst round mutants revealed a noticeable decrease of Km for NADP+ (approximately ten-fold) with concomitant increase of kcat (approxi3861

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

Engineering coenzyme specicity of CboFDH

A. Andreadeli et al.

Table 1. Kinetic parameters for rst round selected mutants. Coenzyme Wild-type NAD+ NADP+ Asp195Ser NAD+ NADP+ Asp195Asn NAD+ NADP+ Asp195Ala NAD+ NADP+ Asp195Gln NAD+ NADP+ Km (mM) kcat (s)1) kcat Km (s)1mM)1) (kcat Km)NADP+ (kcat Km)NAD+ kcat =Km NADP =kcat =Km NAD Mut kcat =Km NADP =kcat =Km NAD WT

0.015 0.01 > 38 1.5 0.05 6.2 0.1 5.01 0.2 13.2 0.3 4.8 0.1 3.3 0.2 0.96 0.06 4.5 0.2

3.7 0.1 4 10)5 0.34 0.02 0.34 0.03 0.21 0.02 0.26 0.02 0.76 0.04 0.052 0.01 0.26 0.02 0.26 0.02

246.7 < 1.05 10)6 0.227 0.055 0.0419 0.0196 0.158 0.0157 0.271 0.058

4.26 10)9

0.242

5.68 107

0.468

10.98 107

0.099

2.32 107

0.214

5.02 107

mately 10 000-fold) for NADP+ compared to the wild-type enzyme. In addition, the Km for NAD+ was dramatically increased (64- to 334-fold) compared to the wild-type enzyme. The mutant Asp195Ala displayed the lowest Km for NADP+ compared to the other mutants. However, this mutant showed the lowest kcat for NADP+ (0.052 s)1) and the highest for NAD+ (0.76 s)1) compared to the other mutants. Therefore, it appears that, despite the fact that Ala might logically be an ideally suited amino acid for position 195, being small and uncharged, bulkier side chains such as of Ser, Asn and Gln result in better catalytic efciencies. It should be noted that, although the mutant Asp195Ser displayed the highest kcat, it showed a moderate afnity for NADP+ compared to the other mutants. The mutant Asp195Gln, which displayed the highest specicity for NADP+, was selected for the second round of saturation mutagenesis. The molecular models of the mutant enzymes were constructed to place the kinetic data in a structural context. Analysis of the modelled structures of the mutant enzymes showed that Gln and Ala leave more room at the 2-group of adenosine ribose, compared to Asn and Ser, in agreement with the Km values for NADP+ (Table 1). In the case of the Asp195Gln mutant, the side-chain of Gln (Fig. 2B) adopts different conformation compared to Asn and Ser and, as a consequence, leaves more room for the accommodation of the 2-phosphate group. In this conformation, its amide side-chain is able to form hydrogen bonds with the 2-phosphate group. This mutation may promote a more productive binding of NADP+ and lead to improved catalytic efciency. Further analysis of the kinetic data showed that the Km for NADP+ follows a linear relationship with
3862

4 WT

2
Hydrophilicity

Asp195Ser Asp195Asn 0 Asp195Gln

Asp195Ala 0 10 20 Km (mM) 30 40

Fig. 3. Structurefunction relationship. Relationship between the Km for NADP+ of the rst round mutants and amino acid hydrophilicity [28]. WT, wild-type.

amino acid hydrophilicity scale [28] with R2 = 0.9674. As shown in Fig. 3, less hydrophobic residues at position 195 have a greater afnity for NADP+. Therefore, besides steric factors, hydrophilicity and polarity are important determinants at position 195. Site-saturation mutagenesis at position 196 In the second engineering round, the amino acid residue at position 196 was subjected to saturation mutagenesis using the mutant Asp195Gln as a parental sequence. The resulted library of recombinant clones was similarly screened for their ability to utilize NADP+ NAD+ using an activity assay and the clones with the highest specic activity for NADP+ were

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

A. Andreadeli et al.

Engineering coenzyme specicity of CboFDH

selected. The expression levels of the mutated enzymes were between 2030% of total E. coli soluble proteins. Four clones that displayed the highest specic activities using NADP+ as substrate were chosen for further study. Sequencing analysis of the selected mutants revealed that Tyr196 was substituted for Ser (in two clones), Pro and His. A low activity clone was also selected and sequenced as a negative control. Sequencing analysis showed that, in this clone, Tyr196 was substituted for Ile. This mutant was not studied further, whereas the three other high activity mutants were puried and subjected to steady-state kinetic analysis (Table 2). The results of the kinetic analysis showed that the double mutant Asp195Gln Tyr196His displayed a remarkable shift in its coenzyme specicity towards NADP+ ([(kcat Km)NADP+ (kcat Km)NAD+]mut (kcat Km)NADP+ (kcat Km)NAD+]wt > 2.2 10)8) compared to the parent mutant Asp195Gln and the wildtype CboFDH. It also exhibited high catalytic activity with NADP+ as coenzyme (14% of the catalytic activity of the wild-type for NAD+). Analysis of the modelled structure showed that, in the case of the Asp195Gln Tyr196His mutant, the side-chain of His196 (Fig. 2) may produce less steric hindrance at the entrance of the coenzyme binding site. In addition, its side-chain is probably able to form amino-aromatic interaction with the adenine aromatic ring. It is interesting to note that, in the case of the Asp195Gln Tyr196Pro, a reversal in specicity towards NAD+ was observed. This is likely due to the presence of a Pro residue. Pro may change the conformation of the preceding residue, Gln195. This conformational change of Gln195 could lead to unproductive binding of NADP+. It is well established [29] that the residue preceding Pro in a sequence changes its conformation because the bulky pyrrolidine ring of Pro restricts the available conformational space. Kinetic analysis of the Asp195Gln Tyr196His mutant using formate as a variable substrate showed
Table 2. Kinetic parameters for second round selected mutants. Coenzyme Km (mM) kcat (s)1) kcat Km (s)1mM)1)

an unexpected dramatic effect on Km for this substrate. The formate binding site is located away from the region under study, between Arg258 and Asn119 [3]. The mutant enzyme exhibits increased the Michaelis constant for formate (Km = 139.6 19.3 mm with NAD+ as cofactor; Km = 80.2 9.5 mm with NADP+ as cofactor). This can be explained by the ordered Bi Bi kinetic mechanism catalysed by CboFDH. CboFDH obeys an ordered kinetic mechanism, which requires the coenzyme to be the rst substrate to bind, with formate being second [30]. The mutations may produce a non-optimal conformation of the proteincoenzyme complex and, as a consequence, may impair formate binding. The same phenomenon has also been observed in the case of SceFDH [16] and PseFDH [17]. Similar mutagenesis experiments have resulted in quite different changes in the kinetic behaviour of different enzymes, depending on the initial Km and kcat values for NAD+ and NADP+, as well as, the 3D conformation of the active center and the catalytic mechanism of the enzyme. For example, three approaches of engineering coenzyme specicity of FDHs from different sources have been reported [10,1416]. NADP+dependent FDH was rst prepared for the enzyme from Pseudomonas sp. 101 [14]. The (kcat Km)NADP+ value (16.6 s)1mm)1) of this mutant FDH was only three fold less than the (kcat Km)NAD+ value for the wildtype enzyme [10]. Gul-Karaguler et al. [15] described a mutant Asp195Ser of FDH from C. methylica reacting with both coenzymes but still showing a preference (> 40-fold) for NAD+ over NADP+. The double substitution Asp196Ala Tyr197Arg in FDH from bakers yeast resulted in enzyme with 1.46% of the catalytic activity of the wild-type enzyme (with NAD+ as coenzyme) [16] which is approximately ten-fold lower than that achieved in the present study. The (kcat Km)NADP+ value for bakers yeast enzyme is approximately 0.03 s)1mm)1, which is 8.7-fold lower than that reported in the present study.

(kcat Km)NADP+ (kcat Km)NAD+

kcat =Km NADP =kcat =Km NAD Mut kcat =Km NADP =kcat =Km NAD WT 16.5 107

Asp195Gln Tyr196Ser NAD+ 5.1 NADP+ 6.2 Asp195Gln Tyr196Pro NAD+ 0.13 NADP+ 3.7 Asp195Gln Tyr196His NAD+ 1.8 NADP+ 1.7

0.06 0.3 0.01 0.2 0.09 0.08

0.40 0.03 0.34 0.02 0.87 0.04 0.34 0.03 0.49 0.03 0.44 0.03

0.078 0.055 6.69 0.092 0.27 0.26

0.705

0.0138

3.2 107

0.96

22.5 107

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

3863

Engineering coenzyme specicity of CboFDH

A. Andreadeli et al.

pH dependence of Vmax in the Asp195Gln/ Tyr196His mutant FDH The effect of pH on Vmax was investigated to dene the acidbase properties of the double mutant and enable the identication of specic functional groups of the enzyme in the catalytic reaction. Figure 4 shows the pH-logVmax prole of the double mutant. This mutant remained catalytically active in the pH range 5.58.5. The rate of formate oxidation using NADP+ as cofactor reached its maximum value at pH 7.5 and retained appreciable activity (> 35%) over a broad range of pH (between 68). This is of particular importance because this enzyme may be coupled efciently in coenzyme recycling systems with several enzymes that show activity between pH 6 and 8. The pH prole shows two inection points corresponding to pKa values of 5.9 0.1 and 8.2 0.1. The results from previous investigations on the wildtype enzyme [3,18] have also demonstrated two inection points with a pKa equal to 5.95 0.2 and 10.16 0.2, respectively. The rst pKa in the acidic pH range was interpreted as suggesting that the active site consists of a carboxylate ionizable group, which plays an important role in the proper orientation and polarization of the NAD+ nicotinamide moiety. Asp308 and Asp282 have been assigned the pKa in the acidic pH range as observed in the Pseudomonas and C. boidinii FDHs, respectively [3,31]. An important observation is that the pKa in the acidic pH range is not perturbed in the case of the Asp195Gln Tyr196His double mutant enzyme. This is probably expected because Asp282 is located away from the region under study.
1.2 1 0.8 logVmax 0.6 0.4 0.2 0

The high pKa in the Vmax-pH prole (pKa = 10.16) that was observed in the wild-type enzyme is believed to have a conformational origin and has been assigned to a cooperative event, which leads to protein reorganization [3,31]. In the case of Pseudomonas enzyme, these changes included the movement of the catalytic domain around a hinge region by an angle of 7.5 and some secondary structure reorganization involving formation of a C-terminal helix a9 [32]. The shift of this pKa that was observed in the Asp195Gln Tyr196His double mutant enzyme (pKa = 8.2 0.1) probably indicates that the conformational changes of the protein occur near this region of the protein or, alternatively, that this region plays an important role in the structural reorganization of the protein during catalysis. Unfortunately, there is no crystal structure available of the enzymeNAD+ formate complex to enable these conformation changes to be precisely assigned to specic parts of the protein structure. Cofactor recycling system with cytochrome P450 monooxygenase and Asp195Gln/Tyr196His mutant FDH Cytochromes P450 constitute a large family of hemoproteins that catalyze the NADPH-dependent monooxygenation of a diversity of hydrophobic substrates [33,34]. They catalyze the reductive scission of molecular oxygen, with one atom of oxygen being reduced to water and the other used to hydroxylate the substrate. CYP102A2 is a catalytically self-sufcient cytoplasmic enzyme from B. subtilis, containing both a monooxygenase and a reductase domain on a single polypeptide chain [33]. P450 biotransformations require a constant supply of NADPH. To render this enzyme more suitable for industrial applications, CYP102A2 was combined with Asp195Gln Tyr196His mutant to generate a cofactor recycling system (Fig. 5A). The oxidation of x-(p-nitrophenyl)dodecanoic acid (12-pNCA) was chosen as a model reaction [35,36] because it allows the simple photometrical measurement of p-nitrophenolate that is produced during the reaction (Fig. 5B). Figure 6 shows the conversion of 0.025 mm 12-pNCA (1250-fold excess over NADP+) by CYP102A2 monitored at A410. To ensure that only NADPH is recycled from NADP+ (by the action of mutant FDH), no external NADPH was added to the reaction mixture. Using 1250-fold excess of 12-pNCA over NADP+, the rst order rate constant was determined equal to kobs = 0.059 0.004 min)1. The reaction reached an approximately 20% yield within only

7 pH

Fig. 4. Dependence of enzyme activity on pH. pH-logVmax prole of the double mutant Asp195Gln Tyr196His of FDH. Assays were carried out at 37 C in 0.2 M potassium phosphate buffer.

3864

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

A. Andreadeli et al.

Engineering coenzyme specicity of CboFDH

Fig. 5. Cofactor recycling system. (A) Diagram of the cofactor recycling system for CYP102A2 with mutant FDH. (B) Reaction scheme of the conversion of 12-pNCA by CYP102A2. The hydroxylation of 12-pNCA results in the formation of a semiacetale that dissociates to an oxo-carboxylic acid and the yellow p-nitrophenolate. The formation of p-nitrophenolate can be directly monitored at 410 nm.

wild-type PTDH has a preference for NAD+ over NADP+ by approximately 100-fold, a mutant PTDH was created by rational design with relaxed specicity toward both nicotinamide cofactors [39]. The mutant PTDH has a 3.5-fold higher catalytic efciency toward NADP+ (kcat Km, NADP+) than the double mutant FDH with a comparable turnover number.

Conclusions
The molecular basis of coenzyme specicity remains an issue of fundamental interest in enzymology. There is no universal approach for changing coenzyme specicity in dehydrogenases and the experimental strategy strongly depends on the particular features of the active center of each individual enzyme. In the present study, we have described the redesign of coenzyme specicity of CboFDH using two rounds of site-saturation mutagenesis and screening. The approach proved to be very effective for this particular evolutionary pathway and enabled the identication of a double mutant Asp195Gln Tyr196His that showed more than 2 107-fold improvement in overall catalytic efciency. The mutant enzyme was used in a cofactor recycling system composed of CYP102A2, 12pNCA, NADP+ and formic acid. Although the mutant enzyme shows high Km for NADP+ and formate, it ts the best requirements for a NADPH regeneration: (a) enzyme cost is low and availability is high; (b) the reaction of formate oxidation to carbon dioxide is irreversible, allowing the use of thermodynamic pressure to obtain a high yield of the desired product; (c) there is facile removal of nal product (CO2); and (d) there is high catalytic activity over a wide pH range.
3865

Fig. 6. Time course of enzymatic conversion of 12-pNCA (0.025 mM) by CYPA2102 monitored at 410 nm. No NADPH was supplied externally and the reduction equivalents were generated by 0.02 lM NADP+ and 0.5 units of Asp195Gln Tyr196His mutant.

the rst 40 min, with a total turnover number 0.3 103. It is noteworthy that practical application of the double mutant FDH needs further investigation and careful optimization of the reaction conditions. A general strategy for regenerating NADPH is based on glucose 6-phosphate dehydrogenase (G6PDH). G6PDH from Leuconostoc mesenteroides accepts both NAD+ and NADP+, whereas yeast-G6PDH accepts only NADP+. Although the kcat and catalytic efciency (kcat Km) of G6PDH are very high (20 200 min)1 and 2530 min)1lm)1, respectively) [37] compared to FDH, a major disadvantage of this method is the high cost of glucose 6-phosphate. Another particularly promising new method for NADPH regeneration uses a newly discovered enzyme, phosphite dehydrogenase (PTDH) [38]. This enzyme catalyzes the nearly irreversible oxidation of phosphite to phosphate. Although the

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

Engineering coenzyme specicity of CboFDH

A. Andreadeli et al.

Experimental procedures
Materials
NAD+, NADH, crystalline bovine serum albumin (fraction V) were obtained from Sigma-Aldrich (St Louis, MO, USA). Pfu DNA polymerase was obtained from Stratagene (La Jolla, CA, USA). Salts and buffers were of analytical grade and purchased from Sigma-Aldrich. All other molecular biology reagents were obtained from Promega (Southampton, UK). 12-pNCA was a much appreciated gift from U. Schwaneberg (Jacobs University, Bremen, Germany).

FDH assays and kinetic analysis

Enzyme assays were performed using a Hitachi U-2000 double-beam spectrophotometer (Hitachi Corp., Tokyo, Japan) carrying a thermostated cell holder, as previously described method [2]. One unit of enzyme activity is dened as the amount that catalyses the conversion of 1 lmol of NAD(P)+ to NAD(P)H per minute. Enzyme activity was determined at 340 nm using a molar extinction coefcient of 6.2 103 m)1cm)1. Steady-state kinetic measurements were performed in 0.1 m potassium phosphate buffer (pH 7.5), by varying the concentration of one of the corresponding substrates, at the saturating concentration of the second substrate, as described by Blanchard and Cleland [40]. All initial velocities were determined in triplicate. The kinetic parameters kcat and Km were calculated by nonlinear regression analysis of experimental steady-state data using the GraFit (Erithacus Software Ltd, Horley, UK) program. Measurements of the pH-dependence of enzyme activity was carried out at 37 C in 0.2 m potassium phosphate buffer (pH 5.58.5). The molar extinction coefcient of NADPH was corrected for pH effects as previously described [41]. Turnover numbers were calculated on the basis of one active site per 42 kDa subunit.

Determination of protein concentration

Protein concentration was determined by the Bradford method [42] using crystalline bovine serum albumin (fraction V) as standard.

For saturation mutagenesis at position 195, the following mutagenic oligonucleotide primers were used (mismatched bases are underlined): 5-GAA TAA TTA TAC TAC NNN TAT CAA GC-3 and 5-GCT TGA TA NNN G TAG TAT AAT AAT TC-3 (N = A, T, G and C). The conditions used were: 95 C for 2 min, 30 cycles of 95 C for 2 min; 50 C for 2 min; 72 C for 8 min, and a nal extension of 72 C for 10 min. For saturation mutagenesis at position 196, the following mutagenic oligonucleotide primers were used (mismatched bases are underlined): 5-CTA CCA A NNN CA AGC TTT ACC AAA AG-3 and 5-CTT TTG GTA AAG CTT G NNN TT GGT AG-3 (N = A, T, G and C). The conditions used were: 95 C for 2 min, 30 cycles of (95 C for 2 min; 47 C for 2 min; and 72 C for 8 min), and a nal extension of 72 C for 10 min. After completion of PCR, the product (30 ng DNA) was digested with DpnI, (10 units) for 4 h at 37 C. Following digestion, the PCR product was used to directly transform E. coli BL21(DE3) cells. Protein expression was carried out as described previously [3]. Bacteria were grown overnight in LB agar plates containing 100 lgmL)1 ampicillin at 37 C. Single colonies were grown overnight in 5 mL of LB containing 100 lgmL)1 ampicillin at 37 C in a shaker incubator. The overnight culture was added to 100 mL of LB containing 100 lgmL)1 ampicillin and grown in a shaker incubator at 37 C until D600 of 0.60.8 was reached, at which time the culture was induced by the addition of 0.1 mm isopropyl thio-b-d-galactoside. After an incubation time of 6 h at 25 C, the cells were harvested by centrifugation (8000 g for 20 min at 4 C) and stored at )20 C until used. To identify enzyme forms with new specicity, cells (0.15 g) were suspended in 300 lL of potassium phosphate buffer (0.1 m, pH 7.4) and the suspension was sonicated on ice (total 30 s, breaking every 10 s). The extract was claried by centrifugation (13 000 g for 5 min at 4 C). Screening was carried out using activity assays, as described in above, employing NADP+- and NAD+ as coenzymes and formic acid as organic acid substrate. Expression levels were determined by a densitometric scan using SDS PAGE.

Purication of recombinant FDH by afnity chromatography

Cells were lysed as described above and the supernatant was dialyzed overnight at 4 C against 500 volumes of 10 mm Mes NaOH (pH 6) for the wild-type enzyme [3] or against 500 volumes of 10 mm potassium phosphate buffer (pH 7.5) for the mutant enzymes. The dialyzate was applied on Cibacron Blue 3GA-Sepharose afnity column (0.5 mL) (SigmaAldrich, St Louis, MO, USA). Non-adsorbed proteins were removed with 10 mL of equilibration buffer. Bound proteins were nally eluted in equilibration buffer containing 0.15 mm NAD+, 1.5 mm Na2SO3 (four fractions of 1 mL).

Saturation mutagenesis and library creation

Saturation mutagenesis at positions 195 and 196 was performed using the PCR-overlap extension method [43]. The reactions were carried out in a total volume of 50 lL containing 10 and 2 pmol of each primer for the rst and second mutagenesis round, respectively; 5 and 2.5 ng plasmid DNA (plasmid pFDH2) [3] for the rst and second mutagenesis round, respectively; 0.2 mm of each dNTP, 5 lL 10 Pfu buffer and 2 units of Pfu DNA polymerase.

3866

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

A. Andreadeli et al.

Engineering coenzyme specicity of CboFDH

Molecular modelling
The crystal structure of ligand-free (open conformation) CboFDH has recently been determined (protein databank code 2FSS) [44]. However, because the enzyme undergoes a large conformational change upon cofactor and formate binding, it became necessary to produce a model of the ligand-bound ternary complex using homology building. The model structure of CboFDH-NAD+-azide was constructed via the Automated Protein Modelling Server SWISS-MODEL (http://www.expasy.org) [45] using as templates the structures with protein databank codes: 2NAD (Pseudomonas sp. 101 FDH), 2NAC (Pseudomonas sp. 101 FDH), 2GUG (Pseudomonas sp. 101 FDH), 2GSD (Moraxella sp. C2 FDH), and 2GO1 (Pseudomonas sp 101 FDH), onto which the CboFDH sequence was threaded. The molecular modelling program what if [46] was used to predict the conformation of the mutant enzymes. The conformation of NADP+ bound to glyoxylate reductase from Pyrococcus horikoshii OT3 (protein databank code 2dbq, a homologue enzyme to CboFDH) [47] was used to model the CboFDH-NADP+ complex using the icm-pro program [48]. Nonbonded interactions were analysed by moltalk (http://i.moltalk.org). For inspection of models and crystal structures, the programs pymol [49] and icm-pro [48] were used. Sequences homologous to CboFDH were sought in the National Center for Biotechnology Information database using blast [50] and psi-blast [51]. The resulting sequence set was aligned with clustalw [52]. espript (http://espript.ibcp.fr//ESPript/cgibin/ESPript.cgi) was used for alignment visualization and manipulation.

References
1 Popov VO & Lamzin VS (1994) NAD(+)-dependent formate dehydrogenase. Biochem J 301, 625643. 2 Schute H, Flossdorf J, Sahm H & Kula MR (1976) Purication and properties of formaldehyde dehydrogenase and formate dehydrogenase from Candida boidinii. Eur J Biochem 62, 151160. 3 Labrou NE & Rigden DJ (2001) Active-site characterization of Candida boidinii formate dehydrogenase. Biochem J 354, 455463. 4 Schaller KH & Triebig G (1994) Methods in Enzymatic Analysis. VCH, Weinteim. 5 Hatch M, Bourke E & Costello J (1977) New enzymic method for serum oxalate determination. Clin Chem 23, 7678. 6 Hummel W & Kula MR (1989) Dehydrogenases for the synthesis of chiral compounds. Eur J Biochem 184, 113. 7 Hummel W (1999) Large-scale applications of NAD(P)dependent oxidoreductases: recent developments. Trends Biotechnol 17, 487492. 8 Ohsuima T, Wandey C, Kula MR & Soda K (1985) Improvement for L-leucine production in a continuously operated enzyme membrane reactor. Biotechnol Bioeng 27, 16161618. 9 Shaked Z & Whitesides GM (1980) Enzyme-catalyzed organic synthesis: NADH regeneration by using formate dehydrogenase. J Am Chem Soc 102, 7104 7105. 10 Tishkov VI & Popov VO (2006) Protein engineering of formate dehydrogenase. Biomol Eng 23, 89110. 11 van der Donk WA & Zhao H (2003) Recent developments in pyridine nucleotide regeneration. Curr Opin Biotechnol 14, 421426. 12 Bentley R (1970) Molecular Asymmetry in Biology, pp. 199. Academic Press, Orlando, FL. 13 Nakamura K, Aizawa M & Miyawaki O (1988) ElectroEnzymology Coenzyme Regeneration. Springer-Verlag, Berlin and Heidelberg. 14 Tishkov VI, Galkin AG, Fedorchuk VV, Savitsky PA, Rojkova AM, Gieren H & Kula MR (1999) Pilot scale production and isolation of recombinant NAD+- and NADP+-specic formate dehydrogenases. Biotechnol Bioeng 64, 187193. 15 Gul-Karaguler N, Sessions RB, Clarke AR & Holbrook J (2001) A single mutation in the NAD-specic formate dehydrogenase from Candida methylica allows the enzyme to use NADP. Biotechnol Lett 23, 283287. 16 Serov AE, Popova AS, Fedorchuk VV & Tishkov VI (2002) Engineering of coenzyme specicity of formate dehydrogenase from Saccharomyces cerevisiae. Biochem J 367, 841847. 17 Tishkov VI & Popov VO (2004) Catalytic mechanism and application of formate dehydrogenase. Biochemistry (Mosc) 69, 12521267.

Conversion of 12-pNCA in a coenzyme recycling system

Conversion of 0.025 mm 12-pNCA by cytochrome P450 monooxygenase from Bacillus subtilis (CYPA2102) was carried out by following the kinetics of p-nitrophenolate production at 410 nm [53]. The molar extinction coefcient of p-nitrophenol under the assay conditions was 7.8 103 m)1cm)1. No NADPH was supplied externally and the reduction equivalents were generated by 0.02 lm NADP+ (1250-fold excess of 12-pNCA over NADP+), 0.5 m formate, 0.5 units of Asp195Gln Tyr196His mutant FDH and 0.5 units of CYPA2102. The reaction was carried out in 100 mm Tris HCl buffer (pH 7.5) at 37 C. CYPA2102 was expressed and puried as described previously [33].

Acknowledgements
This work was partially supported by the NATO LST.CLG.979818 grant. We wish to thank Mr Y. Mavridis and Mr N. Soulos (Agricultural University of Athens, Greece) for their technical assistance.

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

3867

Engineering coenzyme specicity of CboFDH

A. Andreadeli et al.

18 Carugo O & Argos P (1997) NADP-dependent enzymes. I: Conserved stereochemistry of cofactor binding. Proteins 28, 1028. 19 Carugo O & Argos P (1997) NADP-dependent enzymes. II: Evolution of the mono- and dinucleotide binding domains. Proteins 28, 2940. 20 Bocanegra JA, Scrutton NS & Perham RN (1993) Creation of an NADP-dependent pyruvate dehydrogenase multienzyme complex by protein engineering. Biochemistry 32, 27372740. 21 Chen R, Greer A & Dean AM (1996) Redesigning secondary structure to invert coenzyme specicity in isopropylmalate dehydrogenase. Proc Natl Acad Sci USA 93, 1217112176. 22 Chen Z, Lee WR & Chang SH (1991) Role of aspartic acid 38 in the cofactor specicity of Drosophila alcohol dehydrogenase. Eur J Biochem 202, 263267. 23 Labrou NE, Rigden DJ & Clonis YD (2000) Characterization of NAD+ binding site of Candida boidinii formate dehydrogenase by afnity labelling and site-directed mutagenesis. Eur J Biochem 267, 66576664. 24 Clermont S, Corbier C, Mely Y, Gerard D, Wonacott A & Branlant G (1993) Determinants of coenzyme specicity in glyceraldehyde-3-phosphate dehydrogenase: role of the acidic residue in the ngerprint region of the nucleotide binding fold. Biochemistry 32, 1017810184. 25 Miyazaki K & Arnold FH (1999) Exploring nonnatural evolutionary pathways by saturation mutagenesis: rapid improvement of protein function. J Mol Evol 49, 716 720. 26 Miyazaki K, Takenouchi M, Kondo H, Noro N, Suzuki M & Tsuda S (2006) Thermal stabilization of Bacillus subtilis family-11 xylanase by directed evolution. J Biol Chem 281, 1023610242. 27 Bosley AD & Ostermeier M (2005) Mathematical expressions useful in the construction, description and evaluation of protein libraries. Biomol Eng 22, 5761. 28 Hopp TP & Woods KR (1981) Prediction of protein antigenic determinants from amino acid sequences. Proc Natl Acad Sci U S A 78, 38243828. 29 MacArthur MW & Thornton JM (1991) Inuence of proline residues on protein conformation. J Mol Biol 218, 397412. 30 Kato N, Sahm H & Wagner F (1979) Steady-state kinetics of formaldehyde dehydrogenase and formate dehydrogenase from a methanol-utilizing yeast, Candida boidinii. Biochim Biophys Acta 566, 1220. 31 Mesentsev AV, Lamzin VS, Tishkov VI, Ustinnikova TB & Popov VO (1997) Effect of pH on kinetic parameters of NAD+-dependent formate dehydrogenase. Biochem J 321, 475480. 32 Lamzin VS, Dauter Z, Popov VO, Harutyunyan EH & Wilson KS (1994) High resolution structures of holo

33

34

35

36

37

38

39

40

41

42

43

44

45

and apo formate dehydrogenase. J Mol Biol 236, 759 785. Axarli I, Prigipaki A & Labrou NE (2005) Engineering the substrate specicity of cytochrome P450 CYP102A2 by directed evolution: production of an efcient enzyme for bioconversion of ne chemicals. Biomol Eng 22, 8188. Otey CR, Silberg JJ, Voigt CA, Endelman JB, Bandara G & Arnold FH (2004) Functional evolution and structural conservation in chimeric cytochromes p450: calibrating a structure-guided approach. Chem Biol 11, 309318. Schwaneberg U, Sprauer A, Schmidt-Dannert C & Schmid RD (1999) P450 monooxygenase in biotechnology. I. Single-step, large-scale purication method for cytochrome P450 BM-3 by anionexchange chromatography. J Chromatogr A 848, 149159. Schwaneberg U, Schmidt-Dannert C, Schmitt J & Schmid RD (1999) A continuous spectrophotometric assay for P450 BM-3, a fatty acid hydroxylating enzyme, and its mutant F87A. Anal Biochem 269, 359366. Cosgrove MS, Naylor C, Paludan S, Adams MJ & Levy HR (1998) On the mechanism of the reaction catalyzed by glucose 6-phosphate dehydrogenase. Biochemistry 37, 27592767. Costas AM, White AK & Metcalf WW (2001) Purication and characterization of a novel phosphorus-oxidizing enzyme from Pseudomonas stutzeri WM88. J Biol Chem 276, 1742917436. Johannes TW, Woodyer RD & Zhao H (2007) Efcient regeneration of NADPH using an engineered phosphite dehydrogenase. Biotechnol Bioeng 96, 1826. Blanchard JS & Cleland WW (1980) Kinetic and chemical mechanisms of yeast formate dehydrogenase. Biochemistry 19, 35433550. Ziegenhorn J, Senn M & Bucher T (1976) Molar absorptivities of beta-NADH and beta-NADPH. Clin Chem 22, 151160. Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72, 248254. Aiyar A, Xiang Y & Leis J (1996) Site-directed mutagenesis using overlap extension PCR. Methods Mol Biol 57, 177191. Schirwitz K, Schmidt A & Lamzin VS (2007) High-resolution structures of formate dehydrogenase from Candida boidinii. Protein Sci 16, 11461156. Guex N & Peitsch MC (2007) SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18, 27142723.

3868

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

A. Andreadeli et al.

Engineering coenzyme specicity of CboFDH

46 Vriend G (1990) WHAT IF: a molecular modeling and drug design program. J Mol Graph 8, 5256. 47 Shinoda T, Arai K & Taguchi H (2007) A highly specic glyoxylate reductase derived from a formate dehydrogenase. Biochem Biophys Res Commun 355, 782787. 48 Abagyan RA, Totrov MM & Kuszetsov DN (1994) ICM a new method for protein modelling and design. Applications to docking and structure prediction from the distorted native conformation. J Comp Chem 15, 488506. 49 DeLano WL (2002) The PyMOL Molecular Graphics System. DeLano Scientic, San Carlos, CA. 50 Altschul SF, Gish W, Miller W, Myers EW & Lipman DJ (1990) Basic local alignment search tool. J Mol Biol 215, 403410.

51 Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W & Lipman DJ (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25, 33893402. 52 Thompson JD, Higgins DG & Gibson TJ (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specic gap penalties and weight matrix choice. Nucleic Acids Res 22, 46734680. 53 Steffen G, Maurer C, Schulze H, Schmid RD & Urlacher VB (2003) Immobilisation of P450 BM-3 and an NADP+ cofactor recycling system: towards a technical application of heme-containing monooxygenases in ne chemical synthesis. Adv Synth Catal 345, 802810.

FEBS Journal 275 (2008) 38593869 2008 The Authors Journal compilation 2008 FEBS

3869