WO2006074194A2 - Engineered phosphite dehydrogenase mutants for nicotinamide cofactor regeneration - Google Patents

Abstract

Phosphite dehydrogenase mutant enzymes provide relaxed cofactor specificity, increased thermostability, increased activity, solubility, and expression over the wild-type enzyme. The mutant enzymes are useful for nicotinamide cofactor regeneration.

Description

ENGINEERED PHOSPHITE DEHYDROGENASE MUTANTS FOR NICOTINAMIDE

COFACTOR REGENERATION

Inventors: Huimin Zhao, Wilfred A. van der Donk, William Metcalf, Ryan Woodyer, and Tyler Johannes

BACKGROUND OF THE INVENTION

[0001] Driven by recent technical advances in genetic engineering and new societal needs, the use of enzymes and microorganisms as catalysts to synthesize chemicals and materials is rapidly expanding. However, many challenges have yet to be fully addressed, such as developmental costs of biocatalysts and the type of chemistry performed. Most biocatalysts currently used in industry (-65%) are hydrolases that do not perform complex chemistry. The primary reason for this lack of use of complicated chemical reactions is that enzymes catalyzing more involved transformations often require one or more costly cofactors, making these reactions industrially impractical when the cofactor is added in a stoichiometric amount.

[0002] Oxidoreductases, for example, can be used for synthesis of chiral compounds, complex carbohydrates, and isotopically labeled compounds, but they often require NADH or NADPH as cofactors. The cost of NADH is about $40/mmol, whereas the price of NADPH is nearly $500/mmol (Sigma 2002 catalog), rendering stoichiometric use of either reduced cofactor at the kilogram scale prohibitively expensive. There is a need, therefore, to develop regeneration systems for NAD(P)(H) that would allow their addition in catalytic amounts, with the goal of making redox bioprocesses industrially feasible. Because approximately 80% of all reductases utilize NAD(P)(H) as a cofactor, probably accounting for over 300 known reactions, regeneration of these cofactors would be particularly advantageous.

[0003] A number of enzymatic, electrochemical, chemical, photochemical, and biological methods have been developed to regenerate cofactors. Advantages of cofactor regeneration in addition to reduced costs, include simplified reaction work up, prevention of product inhibition from the cofactor, and sometimes a favorable influence on the reaction equilibrium. In some uses, the regenerative system drives the synthetic reaction forward, even when the formation of the desired product is less favored under standard conditions. Specific advantages of enzymatic strategies include high selectivity, compatibility with synthetic enzymes, and high turnover numbers. Aspects to be considered when using enzymatic methods include the expense and stability of the enzyme, cost of the substrate for the regenerative enzyme, ease of product purification, catalytic efficiency, K_M for the cofactor, and thermodynamic driving force of the regenerative enzyme.

[0004] Of the enzymatic NADH regeneration systems, a widely used enzyme is formate dehydrogenase

(FDH) from Candida boidini. Phosphite dehydrogenase (PTDH) may have kinetic and practical advantages over FDH in certain applications, e.g. using PTDH as a regeneration system. This enzyme catalyzes the nearly irreversible oxidation of hydrogen phosphonate (phosphite) to phosphate, with the concomitant reduction OfNAD⁺ to NADH. The large change in free energy of this reaction (ΔG° = - 63.3 kJ/mol estimated from redox potentials) and the associated high equilibrium constant (K_eq = 1 x 10¹¹) makes PTDH a promising NADH regenerative enzyme. A particularly interesting application of PTDH is the facile production of isotopically labeled products. Deuterium or tritium labeled water can be used to readily and economically prepare labeled phosphite. Subsequent use of isotopically labeled phosphite during a synthetic reduction using PTDH for NADH regeneration has been shown to efficiently generate labeled products in high isotopic purity.

[0005] NADPH is significantly more expensive than NADH and currently no system for its regeneration is widely used. The most promising enzymatic NADPH regeneration system is a mutant FDH from Pseudomonas sp.101 (mut-Pse FDH) available from Jϋlich Fine Chemicals. However, no mutations are available, the catalytic efficiency is low (1 μM min^'1), and the cost is high. Another alternative is the use of a soluble pyridine nucleotide transhydrogenase which catalyses the transfer of reducing equivalents between NAD⁺ and NADP⁺. However, this route would require addition of both cofactors and a third enzyme to the process. Currently, the high cost of regenerating enzymes and inefficient regeneration makes the production of synthetic products requiring the use of NADPH not very attractive.

[0006] Altering cofactor specificity remains a challenge, because very few examples exist where catalytic efficiency for the disfavored cofactor NADPH has been significantly improved to approximately the activity with the preferred substrate. Even fewer are the examples where specificity becomes relaxed allowing high catalytic efficiency with both NAD(H) and NADP(H). Among this last group are the non-Rossman fold NAD⁺-dependent isocitrate dehydrogenase, glucose-fructose oxidoreductase, glutathione reductase, and aldehyde dehydrogenase. A comparison of the strategies required to achieve efficient use of the non-physiological cofactor in these enzymes indicates that there is no clear recipe for success.

[0007] The primary cost for regenerative biocatalytic processes in addition to cofactors, resides in the biocatalysts themselves. Therefore, in order to make a process economically viable, the regenerative enzyme must be relatively inexpensive in terms of cost per unit, making optimization of enzyme production and stability important. Wild type (WT) PTDH can be heterologously expressed in reasonable yields in E. coli, but improved expression levels would have important economic benefits. Furthermore, although the wild type enzyme is stable at 4⁰C, it undergoes fairly rapid inactivation under relatively mild temperatures.

SUMMARY OF THE DISCLOSURE

[0008] One or more amino acid mutations in wild-type phosphite dehydrogenase improve protein solubility, enzyme activity, relaxed specificity for nicotinamide cofactors, and thermostability. Engineered mutant phosphite dehyrogenases disclosed herein are useful in regenerating NADH, NADPH and also in the production of various products of commercial interest that require NADH and NADPH regeneration. [0009] A mutant phosphite dehydrogenase (PTDH) has increased thermostability and relaxed cofactor specificity for nicotinamade cofactor regeneration as compared to a wild-type phosphite dehydrogenase. The thermostability includes a temperature optima (Topt) from about 42°C to aboout 59°C. Soluble expression in E. coli is about three-fold higher than the wild-type phosphite dehydrogenase. Activity is about two-fold higher than the wild-type phosphite dehydrogenase. An examplary amino acid mutation is at an amino acid position selected from the group consisting of 13, 26, 71, 130, 132, 137, 150, 175, 176, 215, 275, 276, 313, 315, 319, 325, 332, 336, of wild type phosphite deydrogenase with an amino acid sequence as in SEQ ID NO: 1.

[00010] The amino acid mutation is an amino acid substitution selected from the group consisting of D13E, M26I, E175A, E332N, C336D, Q137R, I150F, Q215L, R275Q, L276Q, A319E, V315A, Q132R, V71I, E130K, I313L, A325V, A176R, and E175A.

[00011] The mutant PTDIT with an amino acid sequence of SEQ ID NO: 35, is designated as a

12X PTDH mutant.

[00012] A nucleic acid molecule encoding the phosphite dehydrogenase mutants is described.

[00013] A host cell transformed with the nucleic acid molecule is described.

[00014] An expression vector encoding the mutant PTDH is described.

[00015] The phosphite dehydrogenase mutant is substantially purified.

[00016] The mutant with an amino acid sequence of SEQ ID NO: 36, is designated as a 12X

+A176R mutant PTDH.

[00017] A mutant is characterized by a rate of reaction of about 1.8 times faster than the wild-type phosphite dehydrogenase specific for NAD and about 2.2 times faster than folate dehydrogenase (FDH) specific for NADP.

[00018] A method of generating at least one of NADH and NADPH includes the steps of:

(a) providing a mutant phosphite deydrogenase, with an amino acid mutation selected from the group consisting of mutations at positionsl3, 26, 71, 130, 132, 137, 150, 175, 176, 215, 275, 276, 313, 315, 319, 325, 332, and 336; and

(b) generating at least one of NADH and NADPH by a reduction reaction of at least one OfNAD⁺ and NADP⁺.

[00019] The amino acid mutation includes an amino acid substitution selected from the group consisting of D13E, M26I, E175A, E332N, C336D, Q137R, I150F, Q215L, R275Q, L276Q, A319E, V315A, Q132R, V71I, E130K, I313L, A325V, A176R, and E175A.

[00020] The mutant phosphite deydrogenase includes an amino acid sequence of SEQ ID NO: 35 [00021] Mutant PTDH has increased efficiency for cofactors NAD⁺ and NADP⁺ as compared to non-mutated phosphite dehydrogenase. [00022] Mutations from Glul 75 to Ala 175 and from Alal 76 to Arg 176 in the wild type amino acid sequence are in accordance with FIG. 7. [00023] A mutant phosphite dehydrogenase is characterized by increased acticity and improved soulbility and expression. Use of the mutant phosphite dehydrogenase is to regenerate

NAD⁺, NADP⁺ or both NAD⁺ and NADP⁺. [00024] Single or double or multiple amino acid changes in phosphite dehydrogenase provide relaxed nicotinamide cofactor (NAD⁺ and NADP⁺) specificity, increased catalytic efficiency, increased thermostability; or a combination of all of these improvements. [00025] Phosphite dehydrogenases disclosed herein catalyse the nearly irreversible (K_eq= 1 x lθ") oxidation of hydrogen phosphonate (phosphite) to phosphate with a concomitant reduction OfNAD⁺ to

NADH and/or NADP+ to NADPH. Phosphite dehydrogenases disclosed herein are also useful to regenerate NADH for in vivo biocatalytic processes requiring it as a reducing equivalent and also as a cheap source of specifically deuterated (4R)-[4-²H]-NAD²H. [00026] Directed evolution was used to enhance the industrially desirable properties such as increased solubility and thermostability of phosphite dehydrogenase for NAD(P)H regeneration. Some of the mutations described herein are disclosed in the context of a three dimensional structural model and the resulting changes in kinetics and soluble expression. The engineered phosphite dehydrogenases are useful for industrial NAD(P)H regeneration. [00027] Directed evolution was also used to improve the thermostability of PTDH. About twelve mutations identified through directed evolution were combined by site-directed mutagenesis, resulting in mutant phosphite dehydrogenases with improved thermostability. [00028] Relaxed co-factor specificity refers to the ability of the phosphite dehydrogenases disclosed herein to bind to at least two kinds of nicotinamide co-factors such as NAD⁺ and NADP⁺. [00029] "Improved characteristic" refers to a statistically significant, measurable increase in a characteristic, or an improvement in at least one feature such as kinetics, thermostability, solubility, relaxed specificity in a mutant phosphite dehydrogenase as compared to a wild-type phosphite dehydrogenase. [00030] "Mutation" refers to a change or alteration at the amino acid or at the nucleotide level including insertion, deletion, and substitution of amino acids or nucleotides. "Mutant" refers to a protein or a peptide or a nucleic acid that is different either structurally or functionally from the wild-type counterpart.

ABBREVIATIONS

Computer Application and Network Services (CANS)

Dehydrogenases (DH)

Fast performance liquid chromatography (FPLC) High performance liquid chromatography (HPLC)

Isoelectric focusing (IEF)

Isopropyl-β-D-thiogalactopyranoside (IPTG)

Molecular Operating Environment (MOE)

Nicotinamide adenine dinucleotide (NAD⁺, NADH)

Nicotinamide adenine dinucleotide phosphate (NADP⁺, NADPH)

Nitro blue tetrazolium (NBT),

Nuclear Magnetic Resonance (NMR)

Phenazine methosulfate (PMS)

Phosphite dehydrogenase (PTDH) (PtxD)

Polymerase chain reaction (PCR)

Protein Data Bank (PDB)

Root-mean-square (RMS)

Wild type (WT)

Xylose reductase (XR)

BRIEF DESCRIPTION OF THE DRAWINGS

[00031] FIG. 1 shows sequence alignment of wild type (WT) PTDH (SEQ ID NO: 1) with three NAD⁺- dependent proteins used for homology modeling including glycerate dehydrogenase (IGDH; SEQ ID NO: 13), phosphoglycerate dehydrogenase (IPSD; SEQ ID NO: 14), and D-lactate dehydrogenase (2DLD; SEQ ID NO: 15). Residues under («--») represent the GxxGxGxxG nucleotide binding motif, residue under (•) represent the acidic residue responsible for binding the adenine 2'-hydroxyl group of NAD(H), and residues under (■) represent the catalytic residues.

[00032] FIG. 2 is a modeled structure of PTDH in comparison to the crystal structure of D-lactate dehydrogenase. The NAD⁺-binding domain (Rossmanfold) is on the right of each structure, while the catalytic domain is on the left, forming the active site in the middle. Residues indicated with asterisks represent the acidic residue responsible for binding 2'-hydroxyl of NAD(H) and residues indicated with arrows represent the catalytic residues.

[00033] FIG. 3 shows a modeled cofactor interactions with mutant enzymes. The interaction between residue E 175 of the WT enzyme and the 2'-hydroxyl OfNAD⁺ and the repulsion of this same residue by the 2'-phosphate OfNADP⁺ are apparent. Replacement of this residue with alanine in silico allows both cofactors to form stable interactions with the enzyme.

[00034] FIG. 4 is a model of the double mutant showing that R176 forms both ionic interactions and H- bonding interactions with NADP⁺ while A 175 allows sufficient room for binding of the 2 '-phosphate OfNADP⁺.

[00035] FIG. 5 shows (A) SDS-PAGE analysis of the purified WT and mutant PTDH proteins. (B)

Isoelectric focusing native gel analysis of the same protein samples. The proteins are separated based on pi showing that both single mutants have a higher pi as predicted and that the effect is additive for the double mutant. [00036] FIG. 6 shows thermal inactivation of WT and the double mutant (E175A; A176R) PTDH at 40.5

⁰C. (A) WT PTDH is inactivated with a half-life of 9.6 min, but in the presence of 1 mM NAD⁺ (and not 0.1 mM or 1 mM NADP⁺), it forms a more thermally stable enzyme-substrate complex with a half- life of 23 min; (B) the double mutant PTDH is inactivated with a half-life of 8.8 min. In the presence of both 1 mM NAD⁺ and 0.1 mM NADP⁺ the double mutant forms a thermally stable enzyme substrate complex with half-lives around 19 min. In the presence of 1 mM NADP⁺ the double mutant retains approximately 100% activity over a 15 minute period. [00037] FIG. 7 shows amino acid sequences of (A) PTDH wild-type; (B) E175A mutant; (C) A176R mutant; and (D) E175A, A176R double mutant; designated by SEQ ID NOS: 1-4 respectively. The mutated amino acids, with respect to the wild-type, are shown in bold. [00038] FIG. 8A-B shows amino acid sequence (SEQ ID NO: 5) and a double strand DNA sequence

(SEQ ID NO: 26) of the PTDH "parent". [00039] FIG. 9A-B shows amino acid sequence (SEQ ID NO: 6) and a double strand DNA sequence

(SEQ ID NO: 27) of Q132R mutant. The mutated amino acids in are highlighted in grey with respect to the parent, as in FIG. 8A-B. [00040] FIG. 10A-B shows amino acid sequence (SEQ ID NO: 7) and a double strand DNA sequence

(SEQ ID NO: 28) of Q137R mutant. The mutated amino acids in are highlighted in grey with respect to the parent, as in FIG. 8A-B. [00041] FIG. 1 IA-B shows amino acid sequence (SEQ ID NO: 8) and a double strand DNA sequence

(SEQ ID NO: 29) of I150F mutant. The mutated amino acids in are highlighted in grey with respect to the parent, as in FIG. 8A-B. [00042] FIG. 12A-B shows amino acid sequence (SEQ ID NO: 9) and a double strand DNA sequence

(SEQ ID NO: 30) of Q215L mutant. The mutated amino acids in are highlighted in grey with respect to the parent, as in FIG. 8A-B. [00043] FIG. 13A-B shows amino acid sequence (SEQ ID NO: 10) and a double strand DNA sequence

(SEQ ID NO: 31) of R275Q mutant. The mutated amino acids in are highlighted in grey with respect to the parent, as in FIG. 8A-B. [00044] FIG. 14A-B shows an amino acid sequence (SEQ ID NO: 11) and a double strand DNA sequence

(SEQ ID NO: 32) of PTDH 4x mutant. The mutated amino acids are highlighted in grey with respect to the parent in FIG. 8A-B. [00045] FIG. 15A-B shows an amino acid sequence (SEQ ID NO: 12) and a double strand DNA sequence

(SEQ ID NO: 33) of PTDH 5x mutant. The mutated amino acids are highlighted in grey with respect to the parent in FIG. 8A-B. [00046] FIG. 16 shows an example of a selection based on phosphite toxicity using increasing phosphite concentrations (A). Minimal media plates containing 1, 5, 10, 50, or 100 mM phosphite have different levels of growth of E. coli BW2514ltransformed with the pRW2-Round3 mutant library. Subsequently, 100 mM phosphite was chosen as the concentration for the third round selection. Result of an NBT assay comparing the lysates from the best mutants in each round ten minutes after adding the components of the NBT coupled assay (B).

[00047] FIG. 17 shows a comparison of the total lysate activity (white bars) and soluble expression (gray bars) for the best mutant from each round (A). The final mutant has >6-fold improvement in total activity and >3-fold improvement in soluble expression levels. B) SDS-PAGE analysis of PTDH expressed in BL21 (DE3) as a His6-Tag fusion. Lane 1 contains WT PTDH purified by IMAC and is shown as a standard, lane 2 contains the soluble lysate fraction of BL21 (DE3) ρET15b- WT PTDH post expression, and lane 3 contains the soluble lysate fraction of BL21 (DE3) pET15b-Round 6 PTDH mutant post expression. There is approximately 3-fold more soluble PTDH expression in lane 3 than in lane 2.

[00048] FIG. 18 shows a scheme for production of L-tert-leucine from trimethylpyruvate and ammonia by leucine dehydrogenase with NADH regeneration by the phosphite/PTDH system (A). Comparison of equal amounts (by mass) of WT-PTDH (■) and Round 6 mutant PTDH (O) for NADH regeneration in the production of L-ter/-leucine (B). Production of L-tert-leucine was measured at the various time points by HPLC.

[00049] FIG. 19 shows a homology model of the round 6 mutant. Mutated residues are labeled in white font and shown as sticks designated R, while the proposed catalytic residues (His 292, Arg 237, and Glu266) are indicated as G. All of the mutated residues are greater than 9 A removed from the catalytic residues, although several may interact with the bound cofactor.

[00050] FIG. 20 is a schematic illustration of lineage and protein level mutations in the evolved thermostable PTDH variants. Corresponding DNA level mutations are shown in parentheses. Substitution shown in italics represents a mutation that does not increase PTDH thermostability. EP-PCR - error-prone PCR, SDM - site-directed mutagenesis, * - previous directed evolution work to improve activity and solubility. [00051] FIG. 21 illustrates activity-temperature profiles of the first generation parent (•), 4x thermostable mutant (Δ), 7x thermostable mutant (♦), and 12x thermostable mutant (O). [00052] FIG. 22 illustrates thermal inactivation of 12x thermostable PTDH and Candida boidinii FDH at

50°C.

[00053] FIG. 23 is a ribbon representation of the monomeric structure of phosphite dehydrogenase from

Pseudomonas stutzeri. NAD⁺ cofactor is indicated as R, and active site residues are indicated as Y. The thermostable mutations (G) are shown in stick representation. [00054] FIG. 24 illustrates activity-temperature profiles for the parent PTDH and 12x thermostable mutant.

[00055] FIG. 25 demonstrates enhancements of Lysate Activity and Expression Level for PTDH mutants.

[00056] FIG. 26 is a general schematic representation of a membrane reactor.

[00057] FIG. 27 illustrates production of L-fer/-leucine from trimethylpyruvate with regeneration of

NADH using WT PTDH (■), the 12x mutant PTDH (o), and C. boidinii FDH (A). The small-scale batch reactions each contained 300 mM ammonium trimethylpyruvate, 400 mM diammonium phosphite or ammonium formate, 0.4 mM 0.4 NAD⁺, and equal amounts (by mass) of either PTDH or FDH.

Samples were incubated at 40⁰C. [00058] FIG.28 shows the kinetics of production of (R)-phenylethanol from actophenone with regeneration of NADPH using WT PTDH (■), 12x+A176R (O), and NADP- FDH (Δ). The reaction mixture (1 mL) contained 20 mM acetophenone, 0.2 mM NADP⁺, 1 mM MgCl₂, 1.4 U mL^"1 ADH-LB, and equal amounts (by mass) of either PTDH or FDH. [00059] FIG. 29 shows a continuous production of xylitol in an enzyme membrane reactor using the

PTDH/phosphite regeneration system. [00060] FIG. 30 shows a continuous production of (R)-phenylethanol in an en2yme membrane reactor using the PTDH/phosphite regeneration system. [00061] FIG. 31 shows the amino acid sequence of the 12X thermostable PTDH mutant (SEQ ID NO: 35)

(A) and the 12X + A176R mutant PTDH (SEQ ID NO: 36) (B). Mutations are highlighted in bold. [00062] FIG. 32 shows the nucleotide sequence of 12X thermostable PTDH mutant (SEQ ID NO: 37). A single letter amino acid sequence of 12X thermostable PTDH mutant is also shown (SEQ ID NO: 35). [00063] FIG. 33 shows the nucleotide sequence of 12X thermostable + A176R mutant PTDH (SEQ ID

NO: 38). A single letter amino acid sequence of 12X thermostable PTDH mutant is also shown (SEQ ID

NO: 36). The A176R mutation is higlighted in gray.

DETAILED DESCRIPTION OF THE DISCLOSURE

[00064] Amino acid changes or mutations were introduced in the wild-type phosphite dehydrogenase

(WT PTDH) (SEQ ID NO: 1) from Pseudomonas stutzeri to yield a plurality of phosphite dehydrogenase mutants with various improved characteristics. Phosphite dehydrogenases from other sources are also suitable to the extent they share structural similarity and/or functional homology. Some of the mutations and their properties are disclosed in Table 10. Phosphite dehydrogenase mutants disclosed herein have one or more of the following characteristics:

(a) higher catalytic rate (k_cat);

(b) increased efficiency (k_cat/K_m);

(c) higher thermostability;

(d) relaxed cofactor specificity (both the natural cofactor NAD⁺ and cofactor NADP⁺);

(e) increased solubility; and

(f) increased expression

[00065] The mutant phosphite dehydrogenases with improved characteristics of the present disclosure were rationally designed by incorporating site-specific mutations to use both the natural cofactor NAD⁺ and a generally disfavored cofactor NADP⁺ with higher catalytic rate (k_cot), increased efficiency (k_cat/K_m) and higher thermostability. Directed evolution techniques were also used to identify mutations in phosphite dehydrogenases that improve enzyme kinetics, thermostability, solubility, and expression.

[00066] Because a three-dimensional structure of phosphite dehydrogenase is not available, a homology model was developed from three known crystal structures: D_^glycerate DH from Hyphomicrobiwn methylovorum (IGDH), D-3-phosphoglycerate DH from E. coli (IPSD), D-lactate DH from Lactobacillus helveticus (2DLD) and then docked with NAD⁺ and NADP⁺. From this model and relevant sequence alignments, two residues Glul75 and Alal76 were identified for cofactor specificity and were mutated to Alal75 (E175A) and Argl76 (A176R) individually and as a double mutant. Both of the individual cofactor specificity PTDH mutants resulted in significantly better efficiency for both cofactors, and the double mutant increased efficiency for NAD⁺ by approximately 4-fold while increasing efficiency for NADP⁺ approximately 1000-fold compared to the wild-type PTDH. Isoelectric focusing of the cofactor specificity PTDH mutant proteins in a non-denaturing gel showed that the replacement with more basic residues does indeed change the effective pi of the protein. HPLC analysis of the enzymatic products verified that the reaction proceeds to completion using either substrate, and produces only the corresponding reduced cofactor and phosphate. Thermal inactivation studies show that the cofactor specificity PTDH mutants disclosed herein are as stable as the wild-type PTDH and furthermore is protected from thermal inactivation by both cofactors while the wild-type is protected by NAD⁺ only. A PTDH mutant with relaxed cofactor specificity has been engineered that forms a stable enzyme substrate complex with both cofactors. The double mutant phosphite dehydrogenase is used to regenerate either cofactor as well as produce (4R)-[4-²H]-NADP²H and serves as the foundation of rational and irrational design efforts disclosed herein.

[00067] Several improved thermostable phosphite dehydrogenase (PTDH) mutants were obtained using directed evolution. Approximately 3200 clones created using error-prone PCR were screened in the first round, with incubation at 43°C. Amino acid substitutions Q132R, Q137R, I150F, Q215L and R275Q were identified as thermostablizing mutations. Site-directed mutagenesis was used to create combined mutants 4x (Q137R, I150F, Q215L, R275Q) and 5x ( Q132R, Q137R, I150F, Q215L, R275Q). The T₅₀ of the 4x mutant is 13°C higher and its X_m at 45°C is 180 times that of the parent PTDH (FIG. 8). PTDH mutants combining both relaxed cofactor specificity and increased thermostability compared to wild-type PTDH (PtxD), are formed by transferring the thermostabilizing mutations to relaxed cofactor specificity PTDH mutants such as E175A and A176.

[00068] Directed evolution was used to further improve the thermostability of WT-PTDH. After three rounds of random mutagenesis (error prone PCR), and high throughput screening, twelve thermostable amino acid substitutions were identified. These twelve mutations (shown in FIG. 20) were combined by site-directed mutagenesis, resulting in a mutant (designated "12X") whose T₅₀ is 20°C higher and half-life of thermal inactivation at 45⁰C is >7000-fold greater than that of the "parent" PTDH. The engineered 12X-PTDH has a half-life at 5O⁰C that is 2.4-fold greater than the Candida boidinii formate dehydrogenase (FDH), an enzyme widely used for NADH regeneration. The amino acid changes engineered to develop the 12X-thermostable mutant includes the following thermostable mutations: Q132R, Q137R, I150F, Q215L, R275Q, L276Q, A319E, V315A, V71I, E130K, 1313L and A325V as compared to the wild-type PTDH

[00069] Directed evolution was used to enhance some of the industrially desirable properties such as increased solubility and thermostability of phosphite dehydrogenase for NAD(P)H regeneration. A two-tiered sorting method of selection and screening was used in conjunction with random and rational mutagenesis. Following six rounds of directed evolution, soluble expression of PTDH in E. coli was increased more than 3 -fold, while the rate of turnover was increased about 2-fold, effectively lowering the cost of the enzyme by >6-fold. Large-scale production of the soluble expression PTDH mutant enzyme by fermentation resulted in ~6-times higher yield (Units/Liter) than the WT PTDH. The enhancements of PTDH production were independent of expression vector and E. coli strain utilized. The advantage of the soluble expression PTDH mutant over the WT enzyme was demonstrated using the industrially relevant bioconversion of trimethylpyruvate to L-ferMeucine. The mutations described are disclosed in the context of a three dimensional structural model and the resulting changes in kinetics and soluble expression. The engineered phosphite dehydrogenase created is useful for industrial NAD(P)H regeneration.

[00070] Phosphite dehydrogenases disclosed here are used in combination with leucine dehydrogenases

(leuDH) from Bacillus cereus or from any suitable source in the production of L-tert Leucine, including batch reactors. Similarly, phenyl ethanol and xylitol are produced using batch as well as membrane reactors.

[00071] Amino acid mutations identified for the P. stutzeri PTDH disclosed herein are used as templates or foundations for identifying corresponding mutations in PTDH enzymes derived from other sources. For example, through homology modelling methods disclosed herein, structurally and functionally conserved domains are delineated among various PTDH enzymes. Then, relevant mutations disclosed herein can be engineered using site-directed mutagenesis or any suitable method.

[00072] The combinations of a plurality of mutations cannot be simply predicted to function as did the individual mutations because of the underlying structural and functional differences that result from the mutations. Present understanding of protein structure and function alone does not yet guarantee that rationally designed changes will yield the predicted outcomes. In fact, protein engineers frequently have been surprised by the range of effects brought about by single mutations designed to change only one specific and simple property in a protein. As a result, possible changes to the protein sequences have been made by mutagenesis/recombination, and then the functionally improved variants were isolated by selection or screening. Further analysis of these variants revealed several of the improved characteristics disclosed herein. The effects of thermostabilizing mutations are not necessarily independent and cumulative and therefore one cannot predict with certainty that a plurality of the mutations can be combined without a loss of one or more of the properties, for example, engineering thermostable mutations into the mutants with improved activity without losing their thermostabilizing effects, requires inventive efforts. For example, the 12X mutant exhibits highly thermostability and high activity toward both cofactors. EXAMPLES [00073] The following examples are illustrative and do not limit the scope of the various methods and compositions disclosed herein.

EXAMPLE 1

Engineering phosphite dehydrogenase mutants with relaxed cofactor specificity and increased catalytic efficiency

[00074] A homology modeling approach identified two residues, Glul75 and Alal76, in Pseudomonas stutzeri phosphite dehydrogenase (PTDH) for nicotinamide cofactor (NAD⁺ and NADP⁺) specificity. Replacement of these two residues by site-directed mutagenesis with Alal75 and Argl76, both separately and in combination, resulted in PTDH mutants with relaxed cofactor specificity. All three mutants (2 singles and 1 double) exhibited improved catalytic efficiency for both cofactors, with the double mutant (E175 A; A176R), having a 4-fold higher catalytic efficiency for NAD⁺ and an 1000- fold higher efficiency for NADP⁺. The cofactor specificity was changed from 100-fold in favor of NAD⁺ for the wild-type enzyme to 3-fold in favor OfNADP⁺ for the double mutant. Isoelectric focusing of the proteins in a non-denaturing gel showed the replacement with these more basic residues indeed changed the effective pi of the protein. HPLC analysis of the enzymatic products of the double mutant verified that the reaction proceeded to completion using either substrate, and produced only the corresponding reduced cofactor and phosphate. Thermal inactivation studies showed the double mutant was as stable as the wild-type enzyme and was protected from thermal inactivation by both cofactors, while the wild-type enzyme was protected only by NAD⁺. The combined results provide clear evidence that Glul75 and Alal76 are both relevant for nicotinamide cofactor specificity. The rationally designed double mutant is useful for the development of an efficient in vitro NAD(P)H regeneration systems for oxidative biocatalysis.

[00075] A rational design approach was implemeted to improve cofactor specificity for NAD⁺ and

NADP⁺ cofactors. PTDH contains the consensus sequence of a typical "Rossman" type fold including the GXXGXGXXG (SEQ ID NO: 16) motif common among D-hydroxy acid DHs (FIG. 1). Incorporated in this fold is an acidic residue (typically an aspartic acid and in rare cases a glutamic acid), located 18 residues downstream of the glycine motif and usually just after an aromatic residue, hi PTDH, this position (residue 175) is occupied by the less common glutamic acid and the previous residue (His 174) is not the typical aromatic residue (FIG. 1). The Asp/Glu residue provides a portion of substrate specificity for NAD(H) by hydrogen-bonding to one or both of the T- and 3 '- hydroxyls of the adenine ribose, whereas NADP(H) specific dehydrogenases typically have a basic residue nearby this region to interact with the negatively charged 2' -phosphate. This sequence information in combination with a three-dimensional structure, along with the fact that a less common glutamic acid is in the proximity (± 13 residues) of three other acidic residues and the absence of aromatic residue were used. A homology model of PTDH was tested by using it as a template to create PTDH mutants with relaxed cofactor specificity. Two single mutants and a double mutant were generated using site- directed mutagenesis, and their kinetics, thermal stabilities, and reaction products are disclosed. [00076] Using site-directed mutagenesis of two residues, Glul75 and Alal76, the nicotinamide cofactor specificity of PTDH was relaxed while the enzyme activity with both cofactors was enhanced. The charged residues near the 2'-position of NAD⁺ are likely responsible for cofactor selectivity. This results differs from previous reports where activity with one or both cofactors is generally reduced in order to achieve a specificity change.

[00077] The mutations in PTDH affected the Michaelis constants (K_M) of PTDH for NADP⁺ and phosphite (in the presence of NADP⁺), while smaller effects were seen in k_cat with both NAD⁺ and NADP⁺ as substrates. Previously, the activity of WT PTDH with 1 mM phosphite and NADP⁺ (6 mM) was estimated to be about 7% compared to the activity with 1 mM NAD⁺ and 1 mM phosphite. An aspect of the mutant enzymes is that the k_cat with NADP⁺ is nearly 50% of the k_cat with NAD⁺. The reason for this discrepancy is that the concentration of phosphite previously used was well below its K_M (in the presence of NADP⁺). The K_M was not determined for either substrate (NADP⁺ or phosphite).

[00078] Replacing Alal76 with a positively charged residue (Arg) had the largest effect on the K_M of

NADP⁺, but replacing the large negatively charged residue (GIu) with alanine also had a pronounced effect. The synergistic effect of these two mutations was larger than the effect of the two individual mutations. The resulting double mutant uses NADP⁺ with 1000-fold greater efficiency (k_cat/K_Mi NADP) and NAD⁺ with 3.6-fold greater efficiency (k_cJK_{Mj NAD}) than WT. When comparing catalytic efficiency, the specificity for the cofactor changes from about 100-fold in favor OfNAD⁺ for the WT enzyme to about 3-fold in favor OfNADP⁺ for the double mutant. For both cofactors the turnover number (k_cat) was higher in the mutants than in the WT. An increase in the catalytic efficiency upon mutagenesis without adverse effect in some other property such as k_cat or K_M for the second substrate (disfavored substrate) is novel and not generally predictable.

[00079] Thermal stability of the relaxed cofactor specificity PTDH mutants (E175A, A176R) was also investigated. The t_\/2 of thermal inactivation at 40.5 °C was determined for the WT enzyme and the double mutant. Because the half-lives are nearly identical (9.6 and 8.8 min respectively), the mutations have no significant effect on thermal stability. From the results of thermal inactivation in the presence of either cofactor, it is clear that the WT PTDH forms a complex only with NAD⁺, whereas the double mutant forms a complex with both NAD⁺ and NADP⁺ with complete protection occurring with NADP⁺. This provides further evidence that the increase in activity with NADP⁺ is due mostly to enhanced binding OfNADP⁺ to the enzyme without disrupting the binding OfNAD⁺.

[00080] Because NAD⁺ and NADP⁺ differ only by a 2'-phosphate group, it was possible that the mutant enzyme dephosphorylated NADP⁺ to NAD⁺ and the observed activity was due to reduction OfNAD⁺. It was also possible that some NAD⁺ was present in the NADP⁺ used in the experiments. Therefore, HPLC was used to analyze the starting materials and enzymatic products. No NAD⁺ was present in the NADP⁺ within the detection limits of the HPLC. A slight contamination of NADP⁺ in the NAD⁺ was did not affect the functionality of the enzymes, because the enzymes utilized NAD⁺ with low K_M's and the contamination level was only -2%. When examining the reaction products, NADPH was produced from NADP⁺ and NADH from NAD⁺. Further examination of the HPLC data indicated no detectable remaining oxidized cofactor after reaction. These data show that the reaction proceeds essentially to completion under physiological conditions and can provide a potent driving force when coupled to unfavorable reactions.

[00081] A mutant Pseudomonas sp. 101 FDH (mut-Pse FDH), commercially available from Juelich Fine

Chemicals is an useful enzyme for NADP⁺ regeneration. The PTDH double mutant of the present disclosure has a catalytic efficiency with NADP⁺ (k_oat/K_M, NADP) that is about 33 -fold higher than that of mut-Pse FDH. The PTDH double mutant can regenerate both cofactors and has a catalytic efficiency with NAD⁺ (k_cat/K_M, _NAD) that is 39-fold greater than mut-Pse FDH. It is also slightly more active (18%) than WT Pse FDH (NAD⁺-dependent). Additionally, whereas the FDH mutants were assayed near optimal conditions (30 ⁰C), PTDH mutants were assayed at 25 ⁰C. The k_cat of PTDH is reduced at 25 ⁰C in comparison to its activity at 35 ⁰C, and hence the improvement over mut-Pse FDH is underestimated. Approximately 100-fold lower concentration of the second substrate (phosphite versus formate) is required for maximal activity with PTDH than with mut-Pse FDH. Thus, the relaxed cofactor specificity PTDH mutants (E 175 A, A 176R) represent a very useful NADP⁺ regeneration system.

[00082] The relaxation of cofactor specificity of the mutants of the present disclosure was achieved by protein engineering based on structural information derived from homology modeling and sequence similarity with other NAD(P)⁺-dependent dehydrogenases. From the homology model, it was predicted that the double mutant should bind NADP⁺ by electrostatic and hydrogen-bond interactions between Argl76 and the cofactor, whereas Alal75 would not interfere with its binding (FIG. 4). The success of this strategy suggests that the homology model is a good working hypothesis for the structure of PTDH.

[00083] Homology Modeling. A protein sequence BLAST search was performed against the Protein Data

Bank, and four sequences were chosen from the highest scoring results. They were D-glycerate DH from Hyphomicrobium methylovorum (IGDH), D-3-phosphoglycerate DH from E. coli (IPSD), D- lactate DH from Lactobacillus helveticus (2DLD) and NAD-dependent FDH from Pseudomonas sp. 101 (2NAC). These four enzymes represent NAD-specific two domain D-hydroxy acid dehydrogenases, and share between 25% and 30% sequence identity with PTDH. FDH (2NAC) was later excluded from this group because its structure was the most divergent and made the initial structural alignment difficult. The structural model was built as disclosed herein. After the model was completed, it bore a striking resemblance to D-lactate dehydrogenase as seen in FIG. 2, with a RMS difference of 0.55 A in the polypeptide backbone of the two structures. Using ProStat (Insight II) under default parameters the Phi and Psi angles were determined to be 79% within their expected values, comparing favorably to the 74.3%, 80.6% and 85.8% for the analysis of the template PDB structures 2DLD, IPSD, and IGDH respectively. A value of 90% correct self-compatability of amino acids with the modeled structure was obtained when inspected by Profiles3-D (Insight II, default parameters). [00084] Three active site residues (Arg237, Glu266, and His292 in PTDH) are highlighted in both the sequence alignment (FIG. 1) and the structure comparison (FIG. 2). The location of these residues in the structure and the sequence is highly conserved in D-hydroxy acid dehydrogenases (Kochhar et al., 2000). In their typical roles, the histidine acts as an active site base, while the glutamic acid is hydrogen bonded to and raises the pKa of the histidine, thus making it a stronger base. The arginine is likely to be involved in binding the typically negatively charged substrates (D-hydroxy acids). These residues, through several possible mechanisms, are involved in catalysis for PTDH. This is supported by the model showing the close interactions of His292 and Glu266, with Arg237 positioned nearby this dyad. In addition, the hydride-accepting carbon of the modeled NAD⁺ is very close to these residues (within 5.5 A of the nearest heavy atom of His292).

[00085] Two regions in the PTDH are highly variable. The first is the loop directly after active site residue Glu266 containing the sequence 267-DWARADRPR-275 (SEQ ID NO: 17) and the second is the C-terminal region containing approximately the last 15 residues. The homologous regions for the template dehydrogenases are not well structurally conserved, introducing more freedom in modeling these regions. Furthermore, it is not unusual for loops and termini to obtain several conformations that are nearly equal in energy. The significance of the loop region in this model is that it is involved in the dimerization interface of the protein (in both the model and templates) and is located near the active site. The loop region is fairly well conserved in dehydrogenases that can oxidize phosphite, but not in other dehydrogenases. Thus, it is likely that this region containing three arginines is involved in binding phosphite. The flexibility of the C-terminal region may be in part responsible for the difficulties experienced during crystallization efforts. In many of the iterations of model structures, this region is found at or near the NAD⁺ binding site. Interestingly, PTDH ends with Cys336 and it has previously been reported that for NADP⁺-dependent malate dehydrogenase, a C-terminal disulfide bond helps regulate enzyme activity by blocking the NADP⁺ binding site (Issakidis et al., 1994; Krimm et. al., 1999). Thus, it is possible that a similar disulfide is formed under certain conditions in PTDH. There is reduced activity when PTDH is purified in the absence of a thiol-reducing reagent such as DTT.

[00086] Modeling of Mutants with Relaxed Cofactor Specificity. Sequence alignment analysis was performed and was found that PTDH binds NAD⁺ by a Rossman-type fold, characterized by alternating α/β regions with α helixes on either side of a plane of 6 antiparallel β-sheets, and indeed this substructure is present in the model (FIG. 2). Among the various hydrogen bond contacts with NAD⁺ created by the loop regions at the ends of the β-sheets, one particular residue, Glul75, stood out as a possible determinant of cofactor specificity. In the model, it is this residue that forms hydrogen- bonds with the hydroxyls of the adenine ribose OfNAD⁺ (FIG. 3) consistent with the sequence alignment prediction (FIG. 1). Glul75 would sterically and electrostatically repulse the 2'-phosphate OfNADP⁺, resulting in its poor binding by the WT enzyme. Replacing Glul75 with sterically smaller residues such as alanine, glycine, and valine might enhance the energetics OfNADP⁺ binding. [00087] The model of a Glul75Ala mutant is shown in FIG. 3 in which the phosphate group OfNADP⁺ is not repelled, but is allowed to hydrogen-bond with the amide backbone proton. This FIG. also demonstrates that NAD⁺ can still interact with the mutant enzyme in a similar manner as the WT enzyme with the exceptions of the hydrogen-bond contribution from GIu 175 and more steric freedom in the mutant binding site. Unlike many Rossman NAD⁺ binding sites, there is no aromatic residue in the +1 site relative to the acidic residue to sterically exclude a phosphate group. In NADP⁺ dependent dehydrogenases, a basic residue (most commonly an arginine) involved in binding the 2'-phosphate moiety is typically present at this +1 position. In PTDH, a histidine is present at the -1 site with the +1 site was occupied by an alanine (FIG. 3). Therefore to probe potential interactions without steric interference, a double mutant Glul75Ala, Alal76Arg was modeled (FIG. 4). In the modeled binding OfNADP⁺ to this mutant, it was clear that the Arg could engage in electrostatic interactions with the 2'-phosphate OfNADP⁺, while also making hydrogen-bond contacts with the adenine base. It was therefore considered that this mutant would be capable of increasing the catalytic efficiency with NADP⁺ without significantly reducing the catalytic efficiency with NAD⁺.

[00088] Mutant Creation, Expression, and Purification. To explore the activity of the modeled mutants, they were first tested with the cell lysate activity assay as disclosed herein. Three mutations (Ala, GIy, and VaI) at the GIu 175 position were generated using mutagenic primers with a single degenerate codon as disclosed herein. Thus, three different gene products were subcloned into the arabinose inducible pRW2 vector and tested simultaneously. The WM1788 strain of E. coli was used in the cell based assay since it contains aphoBR deletion that suppresses activation of endogenous phosphite oxidation pathways in E. coli resulting in minimized background activity. When the lysates often transformed clones expressing Glul75Ala, Glul75Gly, or Glul75Val mutants were assayed with NADP⁺, four showed significant activity, while the others had activity indistinguishable from background. All ten clones were subsequently sequenced revealing that the four active clones contained the Glul75Ala mutation, while Glul75Val and Glul75Gly mutations were both represented in the sequenced DNA from inactive clones. The same pattern was observed in a NAD⁺-dependent cell lysate activity assay. This suggests that the Glul75Val and Glul75Gly mutations resulted in inactive proteins, possibly as a result of misfolding, insolubility, or some other type of inactivation. Therefore, Glul75Ala was chosen for additional studies. Two additional mutants, Alal76Arg and the double mutant GIu 175AIa-AIa 176Arg were subsequently generated and assayed. These two mutants showed a qualitative increase in activity with NADP⁺ over GIu 175AIa and retained high activity with NAD⁺.

[00089] To further characterize these mutants, proteins were overexpressed for large-scale purification as

His₆-tag fusion proteins. The three mutant genes were inserted into the pET15b expression vector as disclosed herein. Overexpression in E. coli BL21 (DE3) resulted in production of PTDH at levels greater than 20% of total cellular protein. Ni²⁺ affinity purification resulted in approximately 30-50 mg of highly pure protein from 1.5 L of each culture. SDS-PAGE analysis of the proteins showed no contaminating bands with only the expected 38.5 kDa band from the His₆-tagged monomer [FIG. 5(A)]. When the WT protein and the mutant proteins were analyzed based on pi by IEF, a clear distinction could be noticed (FIG. 5(b)). Both Glul75Ala and Alal76Arg had a more basic pi (~6.2) than the WT protein (-5.8) due to the removal of the negatively charged residue (Glul75Ala) and the addition of a positively charged residue (Alal76Arg), respectively. The double mutant resulted in a shift towards more basic pi (-6.6) approximately twice as large as for either single mutant when compared to the WT protein (Fig. 5(B)), due to the introduction of a positive residue and the loss of a negative residue. The proteins were activity stained based on NAD⁺-dependent PTDH activity, thus clearly showing that all mutants were active with the natural substrate.

[00090] Kinetic Analysis. The effect of the mutations on the nicotinamide cofactor preference of PTDH was assessed by comparing the kinetic parameters in the forward reaction (reduction of cofactor). The reverse reaction is too energetically unfavorable to assay by conventional means. The activities of the enzymes were determined as a function of concentration of either cofactor under saturating phosphite concentrations. Then activities were determined as a function of phosphite concentration in the presence of either cofactor at saturating concentration. The results of the kinetic analyses are depicted in Table 1. The turnover number (k_cat) of the WT enzyme is lower than previously described due to the assays being performed at 25 ⁰C rather than at 30 ⁰C and a slight deactivation by introduction of the His₆-tag. The WT enzyme has a clear preference for NAD⁺ over NADP⁺ by about 100-fold when comparing catalytic efficiency (k_cat/K_M, _NAD(P>), primarily as a function of lowered K_M. The effect of the mutations on relaxing this preference by lowering the K_M for NADP⁺ is clear. Glul75Ala lowers the K_M by a factor of about 17, while Alal76Arg lowers the K_M by a factor of about 33 compared to the WT enzyme. The synergistic effect of these two mutations results in a K_M for NADP⁺ approximately 700-fold lower in the double mutant. Unexpectedly, the turnover number improves approximately 35-55% in all cases. Therefore the overall efficiency with NADP⁺ of the double mutant (k_cat/K_M, NADP) is approximately 1000-fold better than the WT enzyme. An additional 90-fold improvement in the K_M for phosphite in the presence OfNADP⁺ was observed in the double mutant over the WT enzyme (K_M,p_hosp_hi_te in the presence OfNAD⁺ remains about the same).

[00091] For each mutant enzyme, an improvement in efficiency (k_cat/K_M, _NAD) was also obtained with

NAD⁺ as the substrate. The K_M for NAD⁺ was reduced for both GIu 175AIa and the double mutant while it was similar to WT for Alal76Arg, suggesting that the Glul75Ala mutation was responsible for reducing the K_M in the double mutant. The turnover number was improved as well, with the highest increase of nearly 46% for Alal76Arg. The increase in k_cat for the double mutant of about 34% coupled with the reduction in K_M for NAD⁺ (2.7-fold) resulted in an approximate 3.6-fold increase in catalytic efficiency (k_ai/K_M, N_AD)- In the presence OfNAD⁺, the K_M of the double mutant for phosphite was not significantly changed.

[00092] HPLC Analysis. The purity of the nicotinamide substrates was analyzed to verify that none of the observed activity was the result of contamination. Samples of the oxidized cofactors NAD(P)⁺ were therefore prepared (Sigma) and analyzed by ion-pair HPLC as described herein. There was no discernable NAD⁺ present in the NADP⁺ sample, which appeared to be greater than 99% pure. However, when analyzing NAD⁺, a small amount of (-2%) OfNADP⁺ was present. In order to verify that NADPH was the respective product OfNADP⁺ reduction by the double mutant PTDH, a small- scale reaction was carried out. When the products were analyzed by HPLC, a single peak (UV 340 nm) was observed that had the same retention time as the authentic NADPH. The same process was carried out for NAD⁺ as the substrate and again a peak was observed with a retention time corresponding to an authentic sample of NADH. A small peak with the retention time of NADPH was also observed corresponding to the reduction of the small amount (~2%) OfNADP⁺ present in the NAD⁺ starting material, providing an internal control.

[00093] Thermal Stability and NAD(P) Protection. WT PTDH proved relatively stable at 37 ⁰C, however at higher temperatures, irreversible thermal inactivation was observed. The WT enzyme gradually lost its activity over a 15-min period at 40.5 ⁰C [FIG. 6(A)] with a half-life (t_1/2) of 9.6 min. The thermal stability of the double mutant was very similar, with a ti_/2 of 8.8 min [FIG. 6(B)]. Pre-incubation of the WT enzyme with 1 mM NAD⁺ protected the enzyme from inactivation, lengthening the t_I/2 to nearly 23 min, while pre-incubation with 1 mM NADP⁺ afforded almost no protection (ti_/2 = 11.1 min) (FIG. 6(A)). Performing the same experiment with the double mutant resulted in complete protection from thermal inactivation by 1 mM NADP⁺, retaining 100% activity after 15 min, and protection with 1 mM NAD⁺ was similar to that of WT (ti_/2=18.9 min) (FIG. 6(B)). Furthermore, when the NADP⁺ concentrations were reduced to 0.1 mM, the WT enzyme was not protected (ti_/2 = 9.1 min), while the double mutant was still significantly protected with a ti_/2 = 19.1 min. The WT enzyme has a higher affinity for NAD⁺, while the double mutant has relaxed cofactor specificity and strongly binds NADP⁺.

EXAMPLE 2 Engineering phosphite dehydrogenase mutants with improved thermostability

[00094] For one set of PTDH mutants, error-prone PCR was used to create a library of PTDHs with an average of 1-2 amino acid substitutions per variant. Approximately 3200 clones were screened for increased enzyme activity and thermostability, with incubation at 43⁰C. Five thermostable variants were identified that had half-lives and T₅₀ values greater than the parent (FIG. 8, Table 2). All five variants had single amino acid substitutions (Q132R, Q137R, I150F, Q215L and R275Q). All five first generation variants showed similar enzymatic activities to the parent, while the K_M ^NAD+ varied slightly. Variant 1150F had a 74% increase in K_M ^Pt"H (54 mM to 99 mM) compared to the parent.

[00095] Sequential site-directed mutagenesis was used to combine thermostable mutations from the first generation variants. 4x and 5 x mutants were created using this method. The 4x mutant contains all the single amino acid substitutions except Q132R. This mutation was excluded based on its proximity to Q137R. The 4x mutant had a T₅₀ that is 13⁰C higher and its t_m at 45°C is 180 times that of the parent PTDH. The 5x mutant had a T₅₀ that is 14⁰C higher; however, its t_1/2 at 45⁰C is only 150 fold better than the parent PTDH. The catalytic efficiency of the 4x mutant is -17% lower than the parent, while the 5x mutant is -35% lower. Both combined mutants had higher K_M ^Pt"H than the parent PTDH. [00096] The thermostability of PTDH was improved by combining error-prone PCR (EP-PCR) and site- directed mutagenesis approaches. Error-prone PCR was used to introduce an average of 1-2 amino acid substitutions per PTDH gene. This low mutation rate made it possible to identify the contribution of individual amino acid substitutions on enzyme thermostability. Three rounds of EP-PCR mutagenesis were used to generate libraries of PTDH variants. After each round of EP-PCR, site-directed mutagenesis was used to combine thermostable mutations onto one template and the resulting template served as the parent for the next round of EP-PCR mutagenesis.

[00097] For another set of PTDH mutants, approximately 6000 clones were screened for increased thermostability in the second round of directed evolution with incubation at 57⁰C. Three clones were identified with higher thermostability than the 4x parent (Table 6). Two of the clones contained single amino acid substitutions, while the third clone contained two amino acid substitutions (V315A and D162N). By creating single amino acid substitutions on the 4x template, the V315A substitution was determined to be the thermostabilizing mutation. The mutations L276Q, V315 A, and A319E were each incorporated into the 4x template by site-directed mutagenesis to create a 7x mutant.

[00098] The 7x mutant served as the parent for the third round of directed evolution. Approximately

10000 clones were screened for increased thermostability in the third round with incubation at 62⁰C. Five clones were identified with higher thermostability than the 7x parent. All five clones had single amino acid substitution and one of the clones contained the mutation Q132R which was discovered in both the first and third rounds of screening. All five mutations were each incorporated into the 7x template by site-directed mutagenesis to create the final 12x mutant (Table 6).

[00099] The first generation mutant library was constructed using a mutant PTDH that differs from wild- type PTDH at five amino acid positions (D13E, M26I, E175A, E332N and C336D). These mutations have improved the enzyme solubility and activity, but showed little effect on the enzyme thermostability. At 45°C, the t,_/2 of wild-type PTDH is 1.4 ± 0.3 min whereas the ti_/2 of the first generation parent is 1.2 ± 0.2 min. After creating the first generation mutant library using EP-PCR, approximately 3200 clones were screened for increased thermostability with incubation at 43⁰C. Five variants were identified and confirmed to have longer half-lives than the parent at 45°C (Table 6). Sequencing these variants revealed the following amino acid substitutions: 1-1A9 (Q137R), 1-8D6 (R275Q), 1-11A6 (Q215L), 1-23C7 (Q 132R) and 1-25E8 (1150F) (FIG. 20). Site-directed mutagenesis was used to combine these thermostable mutations onto the parental template to create 4x (Q137R, I150F, Q215L, R275Q) and 5x (Q132R, Q137R, I150F, Q215L, R275Q) mutants.

[000100] The gene encoding the 4x mutant was chosen as a template for generating the second generation mutant library using EP-PCR. The 4x mutant was chosen over the 5x mutant for the second generation parent based on a higher catalytic activity and longer half-life at 45°C. Approximately 6000 clones from the 4x library were screened, with incubation at 57⁰C. Three clones were found to have longer half-lives than the parent at 45°C. After sequencing these three variants, the following amino acid substitutions were identified: 2-2C10 (D162N, V315A), 2-9B6 (A319E) and 2-17C2 (L276Q) (FIG. 20). Clone 2- 2C10 had two amino acid substitutions. Both mutations were incorporated individually onto the 4x template using site-directed mutagenesis to determine the effect of each mutation separately on enzyme thermostability. The substitution V315A had a positive effect on enzyme thermostability, whereas D162N showed a slight decrease in thermostability. Using site-directed mutagenesis, the substitutions L276Q, A319E and V315A were combined on the 4x template to create a 7x mutant.

[000101] The 7x mutant was used as the parent for the third round of random mutagenesis. From this library, ~10,000 clones were screened for increased thermostability at 62°C. Five variants were identified and confirmed to have longer half-lives than the 7x parent. Sequencing these variants revealed the following changes: 3-34G8 (Q132R), 3-57E11 (V71I), 3-89D3 (E130K), 3-110G7 (1313L) and 3-135H2 (A325V) (FIG. 20). It is noteworthy that Q132R appeared twice, in the first and third generations. This gave further evidence that Q132R is indeed a thermostable mutation, and thus it was incorporated into the final template. Site-directed mutagenesis was used to incorporate all third round mutations into the 7x template to create a 12x mutant.

[000102] The kinetics and thermostability data for all the mutants are summarized in Table 7. The parent

PTDH has a half-life of 1.2 minutes at 45°C and its T_opt is 4O⁰C. The T_opt of the 12x mutant is 57°C and its half-life at 45⁰C is 8440 minutes. Thus, directed evolution has been successfully used to increase the ti_/2 of the parent PTDH at 45⁰C more than 7000-fold and shift the T_opt by 17°C.

[000103] Some of the thermostabilizing mutations disclosed herein from directed evolution are introduced into the rationally designed mutants with relaxed cofactor specificity one by one using site-directed mutagenesis. Each variant is tested for its thermostability and activity toward both cofactors. The mutations identified herein through either error prone PCR or through rational design, have been tested for properties such as increased thermostability, increased activity and expression, and relaxed co-factor specificity.

EXAMPLE 3 Thermostabilities of the evolved PTDHs including the 12x mutant

[000104] Table 6 lists the ti_/2 and T50 values of the first generation parent and the evolved thermostable

PTDHs accumulated during three generations of directed evolution. At 45 ⁰C, the t_m of the first generation parent is 1.2 min and its T50 is 39⁰C. The first generation variants have half-lives of thermal inactivation 2-7 fold longer than the first generation parent at 45°C and their T50s are 1.0-3.4⁰C higher. The half-lives of thermal inactivation of the combined site-directed mutants 4x and 5x are -130-170 times that of the first generation parent and their T50s are ~14°C higher.

[000105] A further round of random mutagenesis resulted in three variants that have half-lives of thermal inactivation 360-530 times longer than the first generation parent at 45°C and their T50s are 14.8-15.5°C higher. The recombined product 7x is -1200 times that of the first generation parent and its T50 is increased by 17.9°C. The third generation variants have half-lives of thermal inactivation -1250-2000 times longer than the first generation parent at 45°C and their T50s are 18.5-19.3°C higher. The site- directed 12x mutant has a half-life at 45⁰C that is approximately 7000 times longer than the first generation parent or the wild type enzyme. The T50 of the 12x mutant is 59.3 ⁰C, an increase of 20.3⁰C compared with the first generation parent. The 12X PTDH mutant amino acid sequence is shown in FIG. 30, with the mutations shown in bold.

EXAMPLE 4

Activities of the evolved thermostable PTDHs including the 12X mutant [000106] By error prone PCR, mutant libraries were created and expressed in E. coli. The libraries were sorted based the ability of the harboring clones to grow on phosphite containing media. The selected clones were then further sorted based on activity in the NBT assay. The best mutants were characterized and carried on to the next round. In round 2, saturation mutagenesis was used to optimize the C-terminus of the protein and in round 4, site-directed mutagenesis was used to incorporate a previously identified mutation E175A that relaxes cofactor specificity. The amino acid and genetic basis of each mutant is shown in Tables 3-4.

[000107] All of the mutants showed enhanced activity in the NBT cell lysate assay (FIG. 16B). This activity difference was also apparent by kinetic assay in which reduction of NAD was measured directly by UV absorbance at 340 nm (FIG. 25). The total rate enhancement for the best mutant PTDH was approximately 6-fold. To discern the basis of this activity enhancement, the expression level of the protein was analyzed by SDS-PAGE and densitometry (FIG. 25). It was found that the expression level gradually increased with the rounds of directed evolution resulting in the best mutant having 3 -fold greater soluble expression (FIG. 25). This result suggested that the remaining 2-fold enhancement was a result of increased catalytic rate (k_oat). To test this hypothesis, each mutant gene was subcloned into a His-tag fusion overexpression vector (pET 15b) for purification. The mutant proteins were purified and kinetically characterized. It was found that the expression level enhancements were not vector dependant as a 2.5-fold increase in expression was seen for the best mutant in this vector. The kinetic data is presented in Tables 9 and 10.

[000108] The kinetic parameters of the first generation parent and evolved PTDHs toward the substrates

NAD⁺ and phosphite are listed in Table 5. All five first generation variants show similar k_cat and K_M, NAD⁺ to the first generation parent. First-generation variants 1-8D6 and 1-25E8 exhibit a slight increase in K_M, p_t-_H- The 4x and 5x mutants both have lower activities than the parent, whereas the 4x mutant exhibits about a 2.5-fold increase in K_{M, Pt}._H. All three third generation variants and the 7x mutant have similar kinetic parameters to the 4x mutant. All five third generation variants are slightly more active than the 7x mutant, while their K_M, NAD⁺ remains unchanged and their K_M> pt-π increased slightly. The final 12x mutant retains approximately 75% of the first generation parent's activity and has a slightly higher catalytic efficiency (k_cat / K_M, NAD⁺). The 12x mutant exhibits a K_M, _Pt._H similar to the first generation parent. Kinetic parameters of the 12x mutant toward nicotinamide cofactor NADP⁺ were also determined. The evolved 12x mutant has similar k_cat and K_M> NADP⁺ to the first generation parent (data not shown).

[000109] The activity-temperature profile for the first generation parent and the combined mutants is shown in FIG. 21. For each successive round (Parent→ 4x→ 7x→ 12x), the activity-temperature profile broadens and the activities of PTDH enzymes increase with increasing temperature until the enzyme denatures. The temperature optima, T_opl, of the 12x mutant is 59°C, which is -2O⁰C higher than that of the parent, which is in good agreement with the observed increases in T₅₀ (Table 6).

EXAMPLE 5

Comparison of 12X Thermostable PTDH mutant with Candida boidlnii Formate Dehydrogenase [000110] The half-lives of thermal inactivation of the 12x PTDH mutant and Candida boidinii FDH were measured at 5O⁰C. Both enzymes exhibit first-order inactivation kinetics as shown in FIG. 22. At 5O⁰C, the half-lives of thermal inactivation of the 12x PTDH mutant and FDH are 705 and 299 minutes, respectively. The t^of FDH at 50⁰C is in good agreement with values reported elsewhere for the wild- type and recombinant FDH (ti_/2 >300 min at 50⁰C in 100 mM potassium phosphate buffer, pH 7.5). Slusarczyk and coworkers improved the oxidative stability of the wild-type FDH by replacing two cysteine residues using site-directed mutagenesis but this resulted in several mutants with significantly lower thermostability. Thus the 12x mutant is substantially more thermostable than these variants.

EXAMPLE 6 Evolutionary strategy of thermostable mutants including the 12X PTDH mutant

[000111] The evolutionary strategy includes random mutagenesis and high-throughput screening followed by site-directed mutagenesis to incorporate the best mutations into the parental template for the next round of directed evolution (FIG. 20). By only introducing an average of 1-2 amino acid substitutions per PTDH gene, it was possible to identify the contribution of individual amino acid substitutions on enzyme thermostability. The effects of thermostable mutations are additive as found in many other enzymes. Combined mutants (4x, 5x, 7x and 12x) were significantly more thermostable than any single mutation in a given round. Analysis of the mutations disclosed herein show that the thermal stability of PTDH could be increased significantly without a significant loss of catalytic activity at lower temperatures. The activity-temperature profile for the first generation parent and the combined mutants is shown in FIG. 21. For each successive round (Parent— >^■ 4x→ 7x—> 12x), the activity-temperature profile broadens and the activities of PTDH enzymes increase with increasing temperature until the enzyme denatures.

EXAMPLE 6 Structural Analysis of evolved thermostable PTDH mutants including the 12X mutant

[000112] A three-dimensional homology model of the wild-type PTDH was created as described herein, and the twelve thermostable mutations were mapped into this model (FIG. 23). None of these mutations occurred near the three active site residues (Arg237, Glu266, His292) although one mutation (I150F) did occur in the "Rossman"-type fold involved with NAD(P) binding. A mutation (R275Q) occurred in a loop region directly after the active site Glu266 in the sequence 267-DWARADRPR-275. It has been suggested that the three arginines in this region were involved in binding substrate phosphite. Substituting arginine for glutamine at position 275 does not appear to affect the binding of phosphite according to the kinetic data (Table 5). Therefore, while the other two arginines may still be involved in binding phophite, it appears that Arg275 is not essential to phosphite binding. The mutations A319E and A325V occur in the highly flexible C-terminal region. [000113] Based on this structural model, the molecular basis for improved thermal stability was further probed. The thermostable mutations are all distributed over the surface of the enzyme except for mutations V71I and I150F. This finding underscores the importance of protein surface on stability as shown in many other enzymes with improved thermostability. The two non-surface mutations are located in β-sheets and both are buried in extremely hydrophobic regions within the protein. The isoleucine to phenylalanine substitution (I150F) in β5 is located in the "Rossman"-type fold. Surprisingly, the incorporation of the large hydrophobic phenylalanine residue into the GxxGxGxxG (x represents any type of amino acids) nucleotide binding motif does not seem to affect the enzyme's ability to bind the nicotinamide cofactor NAD⁺. The K_m, NAD⁺ for the first generation parent and the 1150F mutation (1- 25E8) are essentially the same (66 and 75 μM, respectively). Based on the homology model, the substitution of isoleucine for phenylalanine (Il 50F) seems to increase the hydrophobic interaction between residue 150 and residues Metl60, Leul71, Leul89 and Leu205, which enhances the thermostability of the enzyme. The stabilizing effect of V7 II may be attributed to increased hydrophobic interactions with residues Met49, Val57, Leu93, and Phe81.

[000114] Only two mutations (1313L and V315A) are located in α-helices. These substitutions may increase stability by increasing helical propensity. Three mutation sites, E130K, Q132R, and Q137R, are located in the loop region between oc6/β5. Mutations Q132R and Q137R introduce positively charged residues which may increase protein stability by creating a more favorable surface charge distribution. Mutation E 130K is located at the dimer interface and may increase enzyme stability by creating beneficial electrostatic interactions between charged residues of the two subunits (E 13 OK is within ~8A of D272 of the other subunit). The three mutations, Q215L, R275Q, and L276Q, are located in the loop region between βlθ/αl3. It is surprising that substituting an uncharged hydrophilic residue for a non-polar hydrophilic residue (Q215L) at the protein surface increases protein stability. Mutations R275Q substitutes a positively charged reside for an uncharged reside which may increase protein stability by influencing the surface charge distribution. A319E and A325V are both located on the highly flexible C- terminal region. These substitutions most likely increase protein thermostability by anchoring the C- terminal and this would make the protein more resistant to unfolding.

[000115] Thermostability of phosphite dehydrogenase was improved by directed evolution without a loss of catalytic efficiency. The overall stability of PTDH has been increased such that it is possible to run reactor studies to evaluate the performance of PTDH in regenerating NAD⁺ and NADP⁺ for useful synthetic reactions. Engineered PTDH mutants disclosed herein provide a low cost, highly efficient biocatalyst for NAD(P)H regeneration.

EXAMPLE 7

Engineering phosphite dehydrogenase mutants with improved activity and expression [000116] Selection and screening method. A two tiered selection and screening method for enhancing the activity of phosphite dehydrogenase was developed. Using a selection method for phosphorous oxidation pathways, cells with PTDH activity were distinguished from those without. By gradually lowering the concentration of the inducer arabinose, a PTDH expression level just below the lower limit that allowed growth was determined and used for the first round of selection. Since this method did not yield a large increase in total enzymatic activity, but rather resulted in decreases in K_M values and solubility as described herein, the selection was altered for subsequent rounds based on phosphite's toxicity. The rationale for this selection method was that k_cat would be more likely to improve at high concentrations of phosphite, whereas K_M improvements seen in the first round result from low phosphite concentrations. Thus, the minimum amount of phosphite on which the parent gene no longer allowed growth of the host was used to select during subsequent rounds (FIG. 16A). In this case, the E. coli cells must have sufficient PTDH activity to convert a toxic level of phosphite to a non-toxic level, thereby also producing phosphate, an essential nutrient in phosphate depleted media. This selection method was successfully used in rounds 2-6 and the concentration of phosphite at which the mutants could grow increased from 10 mM in round 2 to 250 mM in the final round. The gradual increase in concentration of phosphite required to inhibit growth corresponded with a gradual increase in total activity for the best mutants in the NBT assay (FIG. 16B). Each of the successive best mutants for the 6 rounds of directed evolution showed an improvement in the activity in the lysate as assessed with a coupled assay that reports on the amounts of NADH produced (FIG. 16B).

[000117] Amino Acid Substitutions. The amino acid substitutions for the best mutants of each round are listed in Table 3. In round 1, many proteins of similar total activity had mutations clustered at the C- terminus including positions 332, 336, and 337 (extra stop codon) in addition to K330 being changed to a stop codon (K330*). In order to determine the best mutant to continue with in round 2, EP-PCR was coupled to saturation mutagenesis at codons 332, 336 and 337. The variant protein with the highest total activity had two mutations (E332N and C336D) from saturation mutagenesis and one mutation (M26I) from EP-PCR. This enzyme had better total activity than the K330* mutant of round 1 and was therefore carried on as the parent for EP-PCR in round 3, which revealed D13E mutation. Site-directed mutagenesis was utilized in round 4 to incorporate the E 175 A mutation previously shown to allow NADP utilization and increase turnover rate. The libraries used in rounds 5 and 6 were generated by EP-PCR and the mutations Tl 81 S and A308T were found respectively. In each round besides the second round, only one mutation was found eliminating the possibilities of silent mutations.

[000118] Total Activity and Soluble Expression Enhancement. One drawback to the lysate assay is that only total activity can be determined and therefore it is unclear what parameters (i.e. expression level, solubility or kinetics) have caused the increase in total activity. Therefore the best mutant from each round was characterized in greater detail to dissect the contributions to the improvements. First, in order to assign an accurate numeric improvement in total activity, lysates were assayed by analyzing production of NADH at 340 nm in the presence of saturating concentrations of phosphite and NAD. To determine what role enhanced soluble expression played in the increased activity, SDS-PAGE and densitometry were utilized. FIG. 17A shows the average increase in relative activity for each of the mutants (white bars) next to the average relative amounts of soluble expressed PTDH (grey bars) showing an approximate 6.2-fold improvement in the total activity of the final mutant over the WT enzyme with the largest increases in the first three rounds. The first round mutant only had a small enhancement in total activity mostly as a result of increased expression level and not of enhanced catalytic activity, prompting the change in selection methodology mentioned above. Using the method that selects for optimal phosphate detoxification rates, the next two rounds of directed evolution resulted in the greatest enhancement in soluble expression and total activity, which were raised 2.7-fold and 4.8-fold, respectively, over the WT PTDH. The level of soluble expression was improved approximately 3.4-fold over the six rounds of directed evolution; about 2-fold less than the total activity enhancement. This difference between the enhancement in expression and total activity suggests that in addition to enhanced expression, the turnover rate was also improved. To confirm this, the mutant enzymes were expressed as His6-tag fusions and purified by IMAC. The soluble overexpression of these constructs was analyzed by SDS-PAGE to verify that the improvements were not strain or vector dependent, but rather PTDH sequence dependent. FIG. 17B shows the clear difference in the overexpression of the His6-WT PTDH in comparison to the His6-tagged mutant from round 6, which maintained the 3 -fold difference in BL21(DE3) that was previously found with expression from pRW2 in E. coll BW25141. Therefore, the expression improvement was not vector or E. coli strain dependent and resulted in excellent expression of the final mutant PTDH.

[000119] Kinetics of Purified Mutant PTDHs. The kinetic parameters of each purified enzyme were determined as described herein and are listed with NAD as the cofactor at the top of Table 2. The k_cat of the best mutant of round 1 was not significantly altered, which is in good agreement with FIG. 17A, suggesting that the enhancement comes from increased soluble expression. However, in rounds 2-6 a gradual increase in k_cat was obtained with a 2-fold higher turnover rate for the final mutant over that of the WT PTDH. Overall, the round 1 mutant had lower K_M values for both substrates, while the incorporation of the E 175 A mutation in round 4 caused the K_M values to increase. The final mutant had K_M values that were essentially unchanged from those of the WT enzyme, resulting in catalytic efficiencies that were 2- fold better for both substrates.

[000120] The kinetic parameters were also determined with NADP as the cofactor as presented in Table 5

(bottom) along with those of the isolated E175A mutation in WT-PTDH. With NADP the k_cat is also enhanced in the mutants, however, to a lesser degree than just the E175A mutation. As expected, a major decrease occurred in the K_M value for NADP upon incorporation of the El 75 A mutation in round 4, which also lowered the K_M for phosphite. This mutation seems to work synergistically with other mutations as the final mutant has K_M values that are approximately 2-fold lower than those obtained with the E 175 A mutation alone. The final mutant therefore has catalytic efficiencies for both NADP and phosphite (with NADP) that are nearly 2 orders of magnitude higher than WT PTDH.

[000121] Saturating concentrations of NAD and deuterated phosphite were utilized to determine a primary kinetic isotope effect on k_cat of 2.1 for the final mutant, which is the same as the previously reported isotope effect for WT PTDH. Therefore, hydride transfer is rate limiting to the same degree in the final mutant as with the WT enzyme, suggesting no major change in the isotope sensitive step nor in steps that may mask the overall isotope effect. [000122] Enzyme Production. In order to determine the maximum yield of the round 6 mutant in comparison to the WT PTDH, non-tagged overexpression constructs based on pET26b were prepared and transformed into BL21(DE3). The cells were then grown, induced, and lysed using a 10 L fermentor as described herein. E. coli BL21(DE3) cells harboring the round 6 mutant grew to a higher overall density, yielding 26% more cell wet mass (101 g) than the WT (80 g). The total activity of the lysates for the WT was ~70 U/g wet cell mass (600 U/L fermentation), while for the round 6 enzyme it improved to -360 U/g wet cell mass (3700 U/L fermentation). This represents a ~5-fold improvement, which corresponds well to the total lysate activity improvement in FIG. 17A with pRW2 in BW25141.

[000123] Regeneration of NADH for L-tert-Leucine Production. The potential of the final PTDH mutant for cofactor regeneration was probed by coupling it with leucine dehydrogenase (LeuDH) for the production of L-tert-leucine. In this scheme (FIG. 18A), PTDH converts NAD to NADH during conversion of phosphite to phosphate, which doubles as the buffer. The produced NADH is utilized by LeuDH to convert trimethylpyruvate and ammonia to L-fert-leucine. This reaction was performed under conditions in which the conversion of NAD to NADH by PTDH was limiting for the overall rate of the reaction. When equal amounts (by mass) of WT PTDH and the mutant PTDH from round 6 were used in separate reactions, the reactions proceeded much faster with the mutant. The reaction rate at the linear portion of the reaction was determined using the first four data points of L-fert-leucine production for each enzyme. The rate of the WT enzyme in this linear region was ~11%/hour, whereas the conversion rate for the final mutant was about twice as fast at ~22%/hour. This correlates well with the determined kcat for the final mutant.

[000124] The phosphite oxidation activity of phosphite dehydrogenase from Pseudomonas stutzeri heterologously expressed in E. coli was successfully enhanced. Seven mutations were produced that each contributed to enhanced total activity as a result of increased rate of turnover and/or increased soluble expression. FIG. 19 displays a homology model of PTDH with the locations of the seven mutations (in red) in relation to a bound NAD molecule and the catalytic residues (in green). Upon inspecting this structural model, it is immediately evident that none of the mutations is near the three proposed catalytic residues. The closest mutation is E332N, which is still over 9 A removed from the nearest catalytic residue (R237). Several mutated residues are near the NAD cofactor binding pocket including N332, D336, and A175.

[000125] In the first round, the preponderance of mutations were found near the C-terminus of the protein, with the best performing mutant lacking the last seven residues resulting in reduced K_M values (Table 3) and slightly enhanced expression. This suggests that the C-terminal region of the protein interferes with substrate binding, a hypothesis that is supported by the homology model in which the C-terminus appears to cap the NAD binding site (FIG. 19). Previously, the C-terminal cysteine was proposed to have a possible regulatory role in an analogous fashion to malate DH where disulfide bond formation involving the C-terminal Cys blocks NADP binding. Here, mutations that remove or change this residue in PTDH result in higher activity, which may be related to eliminating disulfide bond formation. [000126] Saturation mutagenesis of the C-terminal region in the second round resulted in E332N and

C336D combined with the M26I mutation from error-prone PCR. This mutant exhibited a slightly higher turnover frequency, mostly as a result of the removal of the C-terminal cysteine, possibly coupled to minor shifts in NAD binding due to contacts of the new N332 and D336 residues. The additional enhanced soluble expression of this mutant likely results from the M26I mutation, which is distant from the active site (~ 35 A) and appears to be solvent exposed. Since the mutation itself does not greatly change the hydrophilicity of this residue, packing factors may be more relevant for the improved soluble expression levels. The D13E mutant found in the third round is also solvent exposed, far from the active site, and is a rather conservative mutation that results in increased expression level without large changes in kcat (Table 5). E175A was incorporated in the fourth round and opens the pocket where the adenine ribose phosphate of NADP would bind, allowing this mutant to utilize both NAD and NADP efficiently (Table 5). This mutation was previously characterized in more detail with the WT PTDH enzyme as a template and here the effects of this mutation are similar, except that the K_M values of NAD(P) for the round 4 enzyme are much smaller than for the previously reported E175A-WT PTDH (Table 5).

[000127] The final two mutations identified were Tl 81 S and A308T and again both are far from the active site. These two mutations affect predominantly the k_cat of the enzyme. It is possible that A308T causes a minor change in the folding or position of the α-helix where it is located, which subsequently alters the position of the proposed catalytic base H292 just upstream of that α-helix. It is not clear why Tl 81 S would affect k_cat, as it is not close in sequence or location to any catalytic residues. Clearly, none of these mutations could have been predicted a priori and some of them are difficult to explain functionally, thus highlighting the power of combinatorial approaches such as directed evolution.

[000128] Although a turnover frequency enhancement of 2-fold is a significant improvement for protein engineering of a native activity, even greater improvements were anticipated based upon the large free energy of the PTDH catalyzed reaction (ΔG⁰¹ = -63.3 kJ/mol) and its relatively low turnover frequency in the native enzyme. However, no further improvements could be obtained by EP-PCR after the 2-fold increase in k_cat. Several other D-hydroxyacid dehydrogenases belonging to the same family as PTDH have much higher k_cal values, such as D-lactate DH (k_cat ~ 18,000 min-1) and glycerate DH (kcat = 2200 min-1). However, PTDH and FDH (also a D-hydroxyacid DH), which both utilize small substrates and have a considerable thermodynamic driving force (FDH ΔG⁰' = -33.2 kJ/mol, PTDH ΔG⁰' = -63.3 kJ/mol), both have much lower turnover frequencies. This is not the result of a slow physical step limiting the rate of catalysis since for FDH, hydride transfer is completely rate limiting and for PTDH, hydride transfer is at least partially if not fully rate limiting.

[000129] The improvements in k_cat render PTDH very competitive in terms of turnover frequency with the three most commonly used FDH enzymes from Candida hoidinii, C. methylica, and an NADP specific mutant from Pseudomonas sp.101, which have k_cat values of 240 min-1, 84 min-1, and 300 min-1, respectively, with NAD as the cofactor. The final PTDH mutant has a higher k_cat value (340 min-1) than the above-mentioned FDH enzymes. Additionally, the FDH turnover numbers are reported at 30⁰C, with that of PTDH measured at 25 ⁰C at which regenerative enzymatic reactions are preferentially carried out. The K_M value for the sacrificial substrate is also much lower for the final PTDH mutant (48 μM phosphite) than for any of the FDH enzymes (5-9 mM formate). Thus, the catalytic efficiency (kcat/KM) in terms of the sacrificial substrate for the final PTDH mutant is >150-fold higher than any of the FDH enzymes. As for NADPH regeneration, the kcat for the final PTDH mutant is comparable at 114 min-1 to the NADP specific mutant FDH from Pseudomonas (150 min-1), but KM values of the PTDH mutant are significantly lower for both NADP and the sacrificial substrate. Therefore, in terms of turnover rate and catalytic efficiency for regenerative processes the final PTDH mutant appears to be superior to commonly utilized FDH enzymes. Additional advantages of the PTDH mutant over the FDH system include the ease of producing and using deuterium or tritium labeled phosphite to make labeled products and the fact that sodium azide does not inhibit PTDH as it does FDH, allowing it to be included as an antimicrobial for long-term enzymatic reactions.

[000130] The increases in the soluble expression and turnover rate are significant in terms of industrial application. The primary cost of cofactor regeneration systems lies with production of the enzymes. The production yield of the final mutant biocatalyst was successfully increased 5-fold compared to the WT enzyme, thus reducing the cost of a PTDH regenerative process by approximately the same amount. The mutant enzyme can be produced at the level of 360 U/g wet cells (E. coli) which is very competitive with FDH production yields.

[000131] Production of L-terMeucine from trimethylpyruvate is an ongoing industrial transformation carried out by Degussa, which utilizes the formate/FDH regeneration system to regenerate the reduced cofactor. As shown in FIG. 18, the phosphite/PTDH system has potential for this transformation. In ongoing studies, this process is being optimized under membrane reactor conditions for the best space/time yield in order to compare it directly to productivity with the formate/FDH system. Additionally, a variety of other interesting enzymatic reactions requiring NADH or NADPH regeneration are under investigation, such as production of xylitol by xylose reductase and preparation of chiral alcohol building blocks by alcohol dehydrogenase. Furthermore, the longevity of PTDH has been improved by enhancing its thermal stability. With the enhancements in expression and activity of PTDH described herein, several steps have been taken towards the generation of a industrially viable NAD(P)H regeneration system.

EXAMPLE 8

Production of L-tert- leucine with the 12x thermostable PTDH mutant coupled with leucine dehydrogenase from Bacillus cereus (LeuDH) [000132] Small-scale batch reactions were conducted to demonstrate the improved stability and effectiveness of the 12x thermostable PTDH mutant. The 12x thermostable PTDH mutant was coupled with leucine dehydrogenase from Bacillus cereus (LeuDH) for the production of L-te/-Meucine. The 12x mutant was compared to wild-type (WT) PTDH and commercially available formate dehydrogenase (FDH) from Candida boidinii. FIG. 27 shows the production of L-fert-leucine over 300 minutes with the three different regeneration enzymes. The WT PTDH precipitates rapidly within the first 30 minutes of the reaction whereas the 12x mutant retains its stability and reaches 100% conversion. The 12x mutant also has a ~2-fold faster reaction rate than FDH under these conditions. EXAMPLE 9 Engineering of 12x+A176R mutant PTDH

[000133] Site-directed mutagenesis was used to incorporate mutation A176R into the 12x template. The kinetic parameters of the wild-type and mutant PTDHs toward the substrates NAD(P)⁺ and phosphite are listed in Table 8. Mutant 12x+A176R shows a 8.4-fold increase in catalytic efficiency (k_cat / K_MNADP⁺) for the NADP⁺ cofactor compared to the 12x mutant. Mutant 12x+A176R exhibits a 3.4- fold decrease in half-life at 50⁰C compared to the 12x mutant. At 50⁰C, the t_1/2of 12x+A176R is 207 ± 11 min whereas the ti_/2of the 12x mutant is 705 ± 5 min.

[000134] The effectiveness of the NADP-specific PTDH mutant (12x+A176R) was verified by coupling it with alcohol dehydrogenase from Lactobacillus brevis (ADH-LB) for the small-scale production of (R)-phenylethanol and with xylose reductase from Neurospora crassa for the continuous production of xylitol in an enzyme membrane reactor.

EXAMPLE 10 Production of (R)-phenylethanoI using 12x+A176R mutant PTDH

[000135] Small-scale batch reactions containing 20 mM acetophenone were carried out using wild-type

PTDH, the 12x+A176R PTDH mutant, and commercially available NADP-specific FDH mutant (mut- Pse FDH). FIG. 28 shows the time course of production of (R)-phenylethanol with NADPH regeneration. The rate of reaction for the 12x+A176R mutant PTDH mutant is about 1.8 and 2.2 times faster than WT PTDH and NADP-specific FDH, respectively. Furthermore, the 12x+A176R mutant PTDH reaches 100% conversion after 40 minutes, whereas the WT PTDH and NADP-specific FDH have yet to achieve full conversion after 100 minutes. For the 12x+A176R PTDH mutant, the (R)- phenylethanol productivity was 88 g L^'1 d^'1 with a total turnover number of 100 for NADPH under non-optimized conditions.

EXAMPLE 11 Continuous production of xylitol using the 12X+A176R mutant PTDH

[000136] The stability and effectiveness of the 12x+A176R mutant PTDH was demonstrated in a continuously operated enzyme membrane reactor (EMR) along with xylose reductase (XR). The conversion of D-xylose to xylitol was chosen as a model to evaluate the performance of the PTDH/phosphite regeneration system. Several batch reactions were carried out to determine optimal reaction conditions for the reactor. Small-scale regeneration reactions carried out at an enzyme ratio of 3:2 (PTDH:XR), pH 6.9, 0.35 mM NADP⁺, and a xylose to sodium phosphite ratio of 0.8 yielded the highest production of xylitol per hour. The continuous production of xylitol was performed in a 10-mL stainless-steel reactor. Table 9 shows the conditions and results for the continuous production of xylitol. The reactor was continuously operated for 180 hours and a substrate flow rate of 2.4 mL/h was used, resulting in a residence time of 4.2 hours. FIG. 29 shows the results for the production of xylitol in the EMR. Since there are no side reactions in the system described herein, yield and conversion are identical. The deactivation of the enzymes under these reactor conditions is approximately 2.8% per day. The conversion gradually decreased as time elapsed due to this deactivation. After 160 hours, 25% of both enzymes were injected into the reactor to compensate for enzyme deactivation and the conversion increased from 60% to 73%. An average space-time yield of 190 g L^'1 d^'1 was achieved during the 180 hours of operation. This indicates that the main reaction was efficiently coupled to the enzymatic regeneration of the cofactor.

EXAMPLE 12 Continuous Production of (R)-phenylethanoI using the 12x+A176R PTDH mutant The continuous production of (R)-phenylethanol in a 10 mL stainless steel enzyme membrane reactor was achieved using the 12x+A176R PTDH mutant. FIG. 30 shows the time course of production of (R)-phenylethanol over 80 hours with NADPH regeneration. The conversion reached 88% and the (R)-phenylethanol productivity was 12 g L^'1 d^'1.

TABLE l

Table 1 : Kinetic Parameters for Recombinant WT Phosphite Dehydrogenase and Mutants using NADP⁺ and NAD⁺ as Substrates

NAD⁺ NADP⁺

Enzyme K_M(IDH NAD⁺) KAW (l/mM*min) K_M(mU Pt-H) K_M(mM; NADP⁺) Ui/s) (l/mM*min) K_M(mM, Pt-H)

WT 53 ±9.0 2.93 ±0.14 3.3 47 ±6.0 2510±410 1.41 ±0.08 3.37E-02 1880 ±325

O E175A 16±0.8 3.50 ±0.05 13.1 23 ±2.9 144±14 2.18 ±0.07 0.91 138 ±25 A176R 60 ±7.0 4.28 ±0.08 4.3 156 ±60 77 ±8.4 2.18 ±0.07 1.7 140 ±20 E175A,A176R 20 ±1.3 3.94±0.08 11.8 61± 13 3.5 ±0.5 1.90±0.08 32.5 21 ±2.7

*A11 assays were performed at 25 ' ^DCpH 7.25 in 5OmMMOPS

TABLE 2

Table 1: Kinetic and thermostability parameters for the parent phosphite dehydorqenase, sinqle mutants and combined mutants.^a

PTDH variant kcat K_M kcat/Ku.NAD KM ti/2 Fold Improvement T₅₀

(min-^η) (μM, NAD) (μM^"1 min^"1) (μM, Pt-H) (min, 45°C) (ti_/2 Mutant / Ua Parent) CO

Parent 262 ± 7.0 75 ±18 3.4 57 ± 4.0 1.1 ±0.3 1 39.0 ±0.1

Single Mutants Q132R 238 ± 21 60 + 14 4.0 45 ± 3.0 2.3 + 0.1 2.1 40.0 ± 0.3 Q137R 285 ± 25 66 + 1.0 4.0 48 ± 5.0 3.8 ±0.8 3.5 41.9 ±0.2 1150F 262 ±15 75 ±30 3.5 99 ±33 7.0 ±1.6 6.4 42.2 ± 0.8 Q215L 278 ±13 64 ±16 4.5 58 ±1.0 8.7 ±0.8 7.9 42.5 ± 0.9 R275Q 244 ±16 70 ±11 3.3 78 ±16 4.6 ± 0.4 4.2 40.7 ± 0.1

4x Mutant Q137R/I150F/ 218 + 16 74 + 18 3.0 144 + 38 200 ±8 182 52.4 ± 0.2 Q215L/R275Q

5x Mutant Q132R/Q137R/I150F, 170 + 3.0 46 + 1.0 3.7 75 ±18 161 ±10 146 53.4 ± 0.2 Q215L/R275Q ^a All assays were performed at 25°C, f )H 7.25, in 50 mM MOPS.

TABLE 3: Sequence information for thermostable PTDH mutants

Round ' Acid Change Codon Chanqe

None WT Sequence

A212V GCC to GTC

K330* AAG to TAG

M26I ATG to ATA

E332N GAG to AAT

C336D TGT to GAC

D13E GAA to GAT

M26I ATG to ATA

E332N GAG to AAT

C336D TGT to GAC

D13E GAA to GAT

M26I ATG to ATA

E175A GAG to GCG

E332N GAG to AAT

C336D TGT to GAC

D13E GAA to GAT

M26I ATG to ATA

E175A GAG to GCG

T181S ACA to TCA

E332N GAG to AAT

C336D TGT to GAC

D13E GAA to GAT

M26I ATG to ATA

E175A GAG to GCG

T181S ACA to TCA

A308T GCA to ACA

E332N GAG to AAT

C336D TGT to GAC

TABLE 4: Mutations discovered by directed evolution

Round Amino Acid Change Codon Change

WT None None

1 K330* AAG to TAG

₂a

M26I ATG to ATA

E332N GAG to AAT

C336D TGT to GAC

3 D13E GAA to GAT

M26I ATG to ATA

E332N GAG to AAT

C336D TGT to GAC

4^b D13E GAA to GAT

M26I ATG to ATA

E175A GAG to GCG

E332N GAG to AAT

C336D TGT to GAC

5 D13E GAA to GAT

M26I ATG to ATA

E175A GAG to GCG

T181S ACA to TCA

E332N GAG to AAT

C336D TGT to GAC

6 D13E GAA to GAT

M26I ATG to ATA

E175A GAG to GCG

T181S ACA to TCA

A308T GCA to ACA

E332N GAG to AAT

C336D TGT to GAC ^aRound 2 mutant was created by EP-PCR coupled to saturation mutagenesis of WT at positions found in Round 1 ^bRound 4 mutant was created by site- directed mutagenesis

TABLE 5: Kinetic parameters for phosphite dehydrogenase mutants with NAD and NADP

NAD

KM, NAD NAD K, M₁Pt kca/KM.Pt

Enzyme k_cat (min^"1) (μM) (μM^"1 mirf¹) (μM) (μM^"1 mirf¹)

WT 176 ±8 53 ±9 3.3 47 ±6 3.7

Rnd 1 168 ±6 37 ±5 4.5 14±3 12

Rnd2 217 ±9 32 ±3 6.8 27 ±2 6.8

Rnd 3 220 ±20 34 ±8 6.5 30 ±1 7.3

Rnd 4 240 ±20 41 ±6 5.9 82 ±2 2.9

Rnd 5 260 ±5 39 ±1 6.7 84 ±2 3.1

Rnd 6 340 ±30 45 ±1 7.6 48 ±2 7.1

NADP

Rnd 3 105 ±9 1070 ±50 0.098 1900 ±600 0.055

Rnd 4 110 ± 1 65 ±8 1.7 87 ±2 1.3

Rnd 5 112± 10 57 ±9 2.0 62 ±9 1.8

Rnd 6 114±5 56 ±5 2.0 70 ±10 1.6

E175A 130 ±0.4 140 ±10 0.91 138 ±25 0.94

*AII assays were performed at 25⁰C pH 7.25 in 50 mM MOPS

TABLE 6: Thermostability data for the parent and mutant PTDHs shown in FIG. 20.

TABLE 7: Kinetic parameters of the parent and evolved PTDH variants in 50 mM MOPS (pH 7.25) at 25⁰C.

Table 8: Kinetic parameters of the wild-type and mutant PTDHs in 50 mM MOPS (pH 7.25) at 25⁰C.

NAD* NADP* enzyme

κ_M KM IWKM NADP K_M

(min^'1) (μM, NAD) (μM^"1 min^"1) (μM, R-H) (min^'1) (μM, NADP) (μM^'1 min^"1) (μM, Pt-H)

WT 176 53 3 3 47 85 2510 0034 ¹⁸⁸⁰ Table 9

E175A, A176R 236 20 11.8 61 114 4 28.5 21

12x 195 40 4.9 46 80 49 1.6 75

12X+A176R 152 22 6.9 34 81 6 13 5 26

Average of all statistically relevant data (standard deviation <10%)

Conditions and results for the continuous production of xylitol in an enzyme membrane reactor.

Feed Concentrations

Xylose 30O mM

Sodium phosphite 375 mM

NADP⁺ 0.35 mM

Sodium azide 0.025% (w/v)

Reactor Conditions

XR 2.0 mg mL^'1

PTDH 3.0 mg mL^'1

BSA 1.0 mg mL^'1

Reactor volume 1O mL pH 6.9

Temperature 25°C

Residence time 4.2 h

Mean conversion 72%

Space-time yield 190 g L^'1 d^'1

Total turnover number 617

Enzyme deactivation 2.8 % d^'1

Table 10: List of various mutations, their designations, and some of their properties for engineered phosphite dehydrogenase (PTDH) mutants.

MATERIALS AND METHODS

[000138] Materials for Relaxed Specificity Mutants. Escherichia coli BL21 (DE3) and pET- 15b were purchased from Novagen (Madison, WI). E. coli WM1788 and plasmid pLA2 were provided by the inventors (Woodyer et al., 2003). The plasmid pRW2 was created from the pLA2 vector by digestion with Nde I and Pci I to remove the majority oflacZ, followed by directional cloning of the PTDH gene digested with the same enzymes. Cloned Pfu turbo polymerase was obtained from Stratagene (La Jolla, CA) and Tag polymerase was obtained from Promega (Madison, WI). PCR grade dNTPs were obtained from Roche Applied Sciences (Indianapolis, IN). DNA modifying enzymes Nde I, Pci I, Dpn I, Bam HI and T4 DNA ligase and their corresponding buffers were purchased from New England Biolabs (NEB) (Beverly, MA). D-glucose was purchased from Fisher Scientific (Pittsburgh, PA), while L-(+)-arabinose and tetrabutylammonium hydrogen sulfate were purchased from Fluka (St. Louis, MO). Ampicillin, kanamycin, isopropyl-β-D-thiogalactopyranoside (IPTG), nitro blue tetrazolium (NBT), phenazine methosulfate (PMS), NAD⁺, NADP⁺, NADH, and NADPH were purchased from Sigma (St. Louis, MO). Phosphorous acid was obtained from Aldrich (Milwaukee, WI) and sodium phosphite from Riedel-de Haenel (Seelze, Germany). Other required salts and reagents were purchased from either Fisher or Sigma-Aldrich. QIAprep spin plasmid mini-prep kit, QIAEX II gel purification kit, and QIAquick PCR purification kit were purchased from Qiagen (Valencia, CA). Various oligonucleotide primers were obtained from Integrated DNA Technologies (Coralville, IA). Isoelectric focusing gels (pH 3-9), buffers, SDS-PAGE gels (12%) and protein size markers were purchased from Bio-Rad (Hercules, CA).

[000139] Materials For Thermastable Mutant. Escherichia coli WM1788 and plasmid pLA2 (Woodyer et al, 2003) and modified plasmid pRW2 containing the mutant E175A gene was obtained as disclosed by Woodyer (2003). Taq DNA polymerase was obtained from Promega (Madison, WI) and cloned PfuTurbo DNA polymerase was obtained from Stratagene (La Jolla, CA). The DNA-modifying enzymes Ndel, Pcil, BamHI, and T4 DNA ligase were purchased from New England Biolabs (NEB) (Beverly, MA). PCR grade dNTPs and DNaseI were obtained from Roche Applied Sciences (Indianapolis, IN).

[000140] Homology modeling. The following structures were downloaded from the Protein Data Bank

(PDB) database (PDB accession code): glycerate dehydrogenase (IGDH) (Goldberg et al., 1994), phosphoglycerate dehydrogenase (IPSD) (Schuller et al., 1995), and D-lactate dehydrogenase (2DLD). Insight II software (Insight II, version 2000, Accelrys Inc., San Diego, CA) was used to align these three structures by conserved structural regions to achieve the lowest root-mean-square (RMS) score. The amino acid sequence of PTDH was then manually aligned by sequence with the structural alignment, taking great care to make sure the aligned sequences represented homologous structural regions. This alignment was then used as input for the automated MODELER module within Insight II using default parameters with moderate refinement of the structure and loop regions. Of approximately thirty structural models created, the best model was selected based on visual inspection for obvious flaws, the score from the Profiles 3-D function, and the ProStat inspection of psi and phi angles. NAD⁺ from the 2DLD crystal structure was manually docked using Molecular Operating Environment (MOE, Chemical Computing Group Inc., Montreal, Canada) into the created model and then the whole structure was subjected to energy minimization to relieve steric and torsional artifacts from the modeling and docking processes. To create mutant enzymes in MOE, a rotamer search was performed with the mutated residue implemented in the homology model of the wild type (WT) enzyme. The lowest energy conformation was selected and energy minimized with the bound cofactor. All Insight II and MOE calculations were performed in the University of Illinois' School of Chemical Sciences' Computer Application and Network Services (CANS) in the VizLab laboratory.

[000141] Site-directed Mutagenesis for Relaxed Specificity Mutants. An overlap extension PCR (OE-

PCR) method was utilized to introduce site specific mutations using purified pRW2-PTDH-wild-type enzyme as the template. Two oligonucleotide primers flanking the gene were used in combination with the following mutagenic primers (underlined codons encode desired amino acid substitutions): E175A/G/V forward (5'-CTG CAG TAC CAC GBG GCG AAG GCT CTG-3' B=T,C,G) (SEQ ID NO: 18), E175A/G/V reverse (5'-CAG AGC CTT CGC CVC GTG GTA CTG CAG-3' V=A,C,G) (SEQ ID NO: 19), A176R forward (5'-CAG TAC CAC GAG CGG AAG GCT CTG GAT-3') (SEQ ID NO: 20), A176R reverse (5'- ATC CAG AGC CTT CCG CTC GTG GTA CTG-3') (SEQ ID NO: 21), double mutant forward (5'-CTG CAG TAC CAC GCG CGG AAG GCT CTG GAT AC-3') (SEQ ID NO: 22), double mutant reverse (5'-GT ATC CAT AGC CTT CCG CGC GTG GTA CTG CAG-3 ') (SEQ ID NO: 23). For the construction of each mutant, two separate PCR reactions were carried out, each containing one flanking primer and one mutagenic primer. The two PCR products were purified from the agarose gel after DNA electrophoresis, treated with Dpn I to remove methylated template, and then elongated by OE-PCR and amplified with the two flanking primers. Products of the correct size were purified from the gel, digested with Pci I and Nde I, and ligated into the Pci l-Nde I digested pRW2 vector. E. coli WM 1788 was then transformed with the ligation mixture and grown on agar plates containing 50 μg/mL kanamycin. Several colonies were picked and clones were first analyzed by cell extract activity assay as described herein. Cultures of the clones with desired activity were grown again and the subsequently isolated plasmids were sequenced in both directions at the Biotechnology Center of the University of Illinois using the Big Dye™ Terminator sequencing method and an ABI PRISM^® 3700 sequencer (Applied Biosystems, Foster City, CA). The genes containing the desired mutations were then subcloned into the pET15b expression vector as a N-terminal His₆-Tag fusion using Nde I and BamΑ I restriction sites. Following subcloning, the mutant genes were again sequenced to eliminate the chance of PCR- introduced random mutations being incorporated into the final DNA construct. The plasmids containing the correct mutant genes were then used to transform E. coli BL21 (DE3) and colonies selected by ampicillin resistance were used for protein expression and purification.

[000142] Cell Extract Activity Assay. A solution of 100 mM Tris HCl pH 7.4 with 0.13% (w/v) gelatin and a 10 x assay solution consisting of 1 mg/mL NBT, 0.5 mg/mL PMS, 15 mM NAD⁺ or 60 mM NADP⁺, and 40 mM phosphite were prepared. Directly prior to the assay, the latter mixture was diluted ten-fold in the Tris-HCl buffer. Cell lysates from arabinose induced E. coli WM1788 cells containing pRW2-PTDH were prepared by lysozyme incubation and freeze-thaw. Clarified cell extract (50 μL) was aliquoted into a 96-well plate followed by rapid addition of assay mix (150 μL) to each well using a multichannel pipetter. The initial rates of reaction and timed endpoints were observed by measuring the OD₅₈O in a Spectramax 340PC microplate reader (Molecular Devices, Sunnyvale, CA).

[000143] Overexpression and Purification of PTDH. The buffers used for protein purification included start buffer A (SBA) (0.5 M NaCl, 20% glycerol, and 20 mM Tris, pH 7.6), start buffer B (SBB) (same as A but with 10 mM imidazole) and elute buffer (EB) (0.5 M imidazole, 0.5 M NaCl, 20% glycerol, and 20 mM Tris, pH 7.6). The transformants with pET15b derived vectors were grown in LB medium containing 100 μg/mL ampicillin at 37 °C with good aeration (shaking at 250 RPM). Upon reaching the log phase (OD₆₀O ~ 0.6) cells were induced with IPTG (final concentration 0.3 mM) and incubated at 25 ⁰C for 8 h. Cells were harvested by centrifugation at 5,000xg, 4 ⁰C, for 15 min and then resuspended in 3 mL/(g cell pellet) start buffer containing 0.6 mg/g lysozyme and stored at -80 ⁰C. The frozen cell suspension was thawed at room temperature and lysed by sonication using a Vibra-cell™ sonicator (Newtown, CT) with amplitude set at 40%, and with a pulse sequence of 5 s on, 9.9 s off, for about 8-10 min. Cells were centrifuged at 20,000Xg at 4 °C for 10 min and the supernatant containing the crude extract was filtered through a 0.45 μm filter to remove any particles. The clarified supernatant was purified by FPLC, with a flow rate of 6 mL/min and fraction size of 8 mL. A POROS MC20 column (7.9 mL bed volume) (Boehringer Mannheim) was charged and equilibrated according to the manufacturer's protocol. The following method was used for purification of PTDH (with His₆-Tag) from a ~20-60 mL of clarified supernatant (from ~5-15g cell paste): 1) load sample through pump, 100 mL, 2) wash column with 100 mL SBB, 3) elute with a linear gradient of 100 mL 100% SBB to 100% EB in 16.7 min, and 4) wash with 100 mL EB. The elute fractions were monitored at λ= 280 nm. PTDH (with His₆-Tag) typically eluted from the column halfway through the gradient (40% EB). The protein was concentrated using a Millipore Amicon 8400 stirred ultrafiltration cell with a YMlO membrane at 4 ⁰C, washed twice with 75 mL of 50 mM MOPS buffer (pH 7.25 containing 1 mM DTT and 200 mM NaCl) and concentrated again. The enzyme was then stored as concentrated as possible (usually > 2 mg/ml) in 200 μL aliquots at -80 °C, in a solution of Amicon wash buffer containing 20% glycerol.

[000144] Protein Characterization. Protein concentration was determined by the Bradford method

(1976) using bovine serum albumin as a standard. The purity of the protein was analyzed by SDS-PAGE. SDS-PAGE gels were stained with coomassie brilliant blue. The net pi of the purified mutants and wild type proteins was determined by non-denaturing isoelectric focusing (IEF) (Hara et. al., 1982). The native IEF gel was subsequently activity stained by the same substrate mixture described herein for cell extract activity assay, allowing visualization of the protein by NBT precipitation.

[000145] Kinetic Analysis. Initial rates were determined by monitoring the increase in absorbance, corresponding to the production OfNAD(P)H (ε_NAD(P)H ⁼ 6.22 mM^cm^"1 at 340 nm). AH initial rate assays were carried out at 25 ⁰C using a Varian Gary 100 Bio UV- Visible spectrophotometer. The reaction was initiated by addition of 1.5-3.5 μg of PTDH. Concentrations OfNAD⁺ stock solutions were determined by UV- Visible spectroscopy (εNAD⁺ = 18 mM^"'cm^"' at 260 nm). Phosphite concentrations were determined enzymatically by measuring the amount of NADH produced after all phosphite had been oxidized. Michaelis-Menten constants V_max and K_M were determined by a series of assays in which five varying concentrations of one substrate were used in the presence of saturating concentrations of the second substrate. The data was then converted to specific activity and fitted with the Michaelis-Menten equation. The WT and double mutants were also analyzed by a sequential matrix of 25 assays. This kinetic data was analyzed with a modified version of Cleland's program (1979). V_max and K_M for both phosphite and NAD(P)⁺, were obtained by fitting the data to a sequential ordered mechanism with NAD(P)⁺ binding first, where v is the initial velocity, V is the maximum velocity, K_A and K_B are the Michaelis-Menten constants for NAD(P)⁺ and phosphite respectively, A and B are the concentrations of NAD(P)⁺ and phosphite respectively, and K_ia is the dissociation constant for A (NAD(P)⁺) (eq. 1). All assays were performed in duplicate and each series of duplicates was performed a minimum of two times. Data presented in Table 1 represents an average of all statistically relevant data, v = VAB/(KiaK_B + K_AB + K_BA + AB) (eq. 1)

[000146] Thermal Inactivation. Thermal inactivation was studied by incubating either WT or the double mutant at 40.5 ⁰C in 50 mM MOPS (pH 7.25) at a protein concentration of approximately 200 ng/μL. The samples were pre-incubated on ice for 5 min in the presence of 0.1 mM NADP⁺, 1 mM NAD⁺, or no cofactor, and then placed in the water bath. At various time points 10 μL of the protein sample was used to initiate the reaction of 0.5 mM OfNAD⁺ and 0.5 mM phosphite. Plotting the data as activity versus time followed by fitting to an exponential curve was performed to determine the half- lives of thermal inactivation.

[000147] HPLC Analysis of Reaction Products. The purity of the nicotinamide cofactor substrates and reaction products was assessed by HPLC. The separation OfNAD⁺, NADP⁺, NADH, and NADPH was carried out. An Agilent 1100 series solvent selector, pump, column and detector modules were utilized with a Zorbax 150 mm x 3.0 mm C-18 (3.5 μm) column and a flow rate of 0.5 mL/min. Instead of 6 mM tetrabutylammonium phosphate, 5 mM tetrabutylammonium sulfate was used in the mobile phase. The total run time was increased to 20 min by the addition of a 5-min isocratic elution at the end of the gradient. Sample volumes for each pure substrate were 20 μL at a concentration of 1 mM in 50 mM MOPS (pH 7.25). Reaction products were prepared by mixing equal parts of 1 mM of the NAD(P)⁺ with 5 mM phosphite, adding approximately 1 μg of enzyme, and allowing the reaction to proceed for 20 min at 30 ⁰C. These samples were then treated the same as other samples, tracking the UV absorbance at both 260 ran (^_3x NAD(P)⁺) and 340 nm (λ_max NAD(P)H).

[000148] Random Mutagenesis and Library Creation from the "parent" enzyme. A mutant PTDH isolated served as the "parent" enzyme. The "parent" PTDH differs from wild type PTDH by five mutations (D13E, M26I, E175A, E332N and C336D). These mutations help increase enzyme solubility and enhance activity. Random mutagenesis was carried out by error-prone PCR. Plasmid pRW2 containing the parent gene was used as the template for the first generation mutagenesis. For the 1.0-kb PTDH-parent target gene, 0.20 mM MnCl₂ was required to obtain the desired level of mutagenesis (-1-2 amino acid substitutions). Forward (5'-TTTTTGGATGGAGGAATT CATATG-3') (SEQ ID NO: 24) and reverse (5'-CGGGAAGACGTACGGGGTATΛG47U:r-3') (SEQ ID NO: 25) primers were designed to amplify the gene. Restriction enzyme recognition sites, Ndel in the forward primer and Pcil in the reverse primer, are shown in italics. PCR-mutated genes were digested with Ndel and Pcil and ligated into a high copy shuttle vector. Ligation reactions (10 μl total volume) contained ~50 ng inserts, ~50 ng vector, IX T4 DNA ligase buffer and 0.5 U T4 DNA ligase and were incubated at 16°C for 16 h. The resulting plasmids were transformed into freshly prepared electrocompotent WM1788 cells, which were plated on Luria-Bertani agar plates containing 50 μg/ml kanamycin.

[000149] Thermostability Screening. Colonies were grown in 96-well plates containing 100 μL of LB media and 50 mg/ml kanamycin. The plates were incubated at 37°C for 5 hours, and then the cultures were induced by adding 10 mM arabinose final concentration and incubating at 30⁰C overnight. Cells were lysed by adding lysozyme (1 mg/ml) and Dnase I (4 UAmI) followed by a freeze-thaw. The plates were centrifuged at 4000 rpm for 15 min at 4⁰C and 50 μL of clarified supernatant was transferred to two fresh plates. One plate was placed into a machined aluminum block holder that had been pre-incubated in an oven set at a specific temperature. After 10 min incubation at the elevated temperature, the plate was allowed to cool at room temperature. Initial and residual activities were determined by adding NBT assay solution and monitoring the change in absorbance at 580 nm for 5 min in a Spectramax 340PC microplate reader (Molecular Devices, Sunnyvale, CA). Thermostable mutants were identified by comparing residual activity to initial activity (R_Aβ._A).

[000150] DNA Sequencing and Analysis. Plasmid DNA from E. coli WM1788 was isolated using

QIAprep spin plasmid mini-prep kits. Sequencing reactions consisted of 100-200 ng of template DNA, 10 pmol each primer, sequencing buffer and the BigDye reagent. Reactions were carried out for 25 cycles of 96⁰C for 30 s, 5O⁰C for 15 s, 6O⁰C for 4 min in a PTC-200 Peltier thermal cycler from MJ Research. Prepared samples were submitted to the Biotechnology Center at the University of Illinois for sequencing on an ABI PRISM 3700 sequencer (Applied Biosystems, Foster City, CA).

[000151] Site-Directed Mutagenesis for Thermostable Mutants. A modified Megaprimer PCR method was used to introduce site-specific mutations using purified pRW2-parent as the template (Sarkar and Somner, 1990). For the construction of the combined 4x and 5x mutants, sequential PCR reactions were used to introduce each mutation. The 4x mutant contains the all single thermostable mutations except Q132R. The 5x mutant contains all single thermostable mutations. The genes were subcloned into pET15b.

[000152] Half-lives of Thermal Inactivation. Purified enzymes (0.2 mg/ml) were incubated in an MJ

Research (Water/own, MA) PTC-200 thermocylcer to study enzyme inactivation. Timed aliquots were taken at specific time points and placed on ice before assaying. Half-lives of thermal inactivation were calculated using t_υ2 = In2 / k_inact where k_inaot is the inactivation rate constant obtained from the slope by plotting log (residual activity / initial activity) versus time. Purified enzymes (0.2 mg/ml) were incubated for 20 min at various fixed elevated temperatures. After incubation, samples were placed on ice for 15 min before being assayed. Residual activity was determined and expressed as a percentage of the initial activity. [000153] Production of PTDH in a bio reactor. PTDH mutant enzymes can be produced in a large-scale bioreactor using standard techniques in microbiological fermentation and downstream processing. For example, a batch reactor containing suitable growth media for bacterial can be operated to grow the bacterial cells (harboring a plasmid that encodes a PTDH enzyme) to appropriate growth density for further downstream processing. Other cultures such as yeast can also be used and other modes of bioreactors such as continuous stirred reactor can also be used to produce and purify the enzyme in a large scale. Appropriate selection markers, oxygen concentration, agitation speeds, nutrient supplements can be optimized using techniques known in the art.

[000154] The standard downstream processing steps usually include harvesting cells by continuous centrifugation or cross-flow filtration. For intracellular products, cells are lysed by a French press, mill, sonication, or detergent and the cell debris is removed via crossflow filtration. Crude purification of the protein is generally performed via ammonium sulfate precipitation followed by chromatography (gel permeation, ion exchange, hydrophobic interaction, hydrophilic interaction, and/or metal affinity) and desalting with a dialysis membrane. The purified product is concentrated under vacuum with or without centrifugation and followed by freeze-drying if necessary. Concentration of the protein and activity of the enzyme can be performed using standard assays known to those of ordinary skill in the art.

[000155] Perform membrane reactor analysis on the phosphite/PTDH system and the formate/FDH system, respectively.

[000156] A membrane bioreactor to evaluate the catalytic perfoπnance of the wild type PTDH enzyme, the engineered PTDH variants, and the FDH enzyme, respectively is used. To save time and minimize the variations from reactor setup, a lab-scale enzyme membrane reactor has been purchased from Julich Fine Chemical. Pn the case of using NAD⁺ as a cofactor, both enzymatic systems are coupled to the production of L-tert-Leucine from trimethylpyruvate using L-Leucine dehydrogenase. The product formation and substrate depletion is monitored by high-pressure liquid chromatography (HPLC). The total turnover number and stability of each system are determined. Data for the FDH system is consistent with those reported in the literature, which will be used as a benchmark for the development of a proposed phosphite/PtxD system. In the case of using NADP⁺ as a cofactor, the engineered PtxD variants are coupled with recently discovered xylose reductase to convert xylose and glucose into xylitol and sorbitol, respectively. Similarly, the total turnover number and stability of each system will be determined. In both cases, the cofactors are tethered to polyethyleneglycol (PEG, MW=20,000) to increase their sizes as did in the existing FDH-based cofactor regeneration system.

[000157] Mutants with improved activity and expression (the 12X mutant generation). The phoBR(-

) phenotype of this strain prevents phosphate starvation response, ΔaraBAD prevents them from metabolizing arabinose, and the integrated pir gene allows replication of pir dependent plasmids. The plasmid pRW2 was created from pLA2 containing a pir dependent replication origin that allowed it to be replicated in low copy number in cells containing the pir gene and it utilizes an arabinose promoter making it suitable for with E. coli BW25141. E. coli BL21 (DE3), pET15b and pET26b were all purchased from Novagen (Madison, WI). Nitro blue tetrazolium (NBT), phenazine methosulfate (PMS), ampicillin, kanamycin, L-fert-leucine, β-D-thiogalactopyranoside (IPTG), NAD and NADP were purchased from Sigma (St. Louis, MO). Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs (Beverly, MA). Cloned PfuTurbo DNA polymerase was obtained from Stratagene (La JoIIa, CA) while Taq polymerase was obtained from Promega (Madison, WI). Sodium phosphite was obtained from Riedel de Haen (Seelze, Germany), ammonium phosphite from City Chemicals (West Haven, CT), and phosphorous acid from Aldrich (St. Louis, MO). Kits for plasmid purification, gel and PCR purification of DNA were obtained from Qiagen (Valencia, CA). DNA oligonucleotide primers were obtained from Integrated DNA technologies (Coralville, IA). Trimethylpyruvic acid (85%) was generously donated by Cams Chemical Company (Lasalle, IL) and was then precipitated by neutralization with ammonia, and subsequently recrystallized by dissolving in minimal water and adding acetone:diethyl ether (5:1). Crystals were collected and washed with diethyl ether using vacuum filtration and dried overnight. Leucine dehydrogenase was purchased from Jϋlich Fine Chemicals (Jϋlich, Germany).

[000158] Mutagenesis and Library Construction for 12X mutant. In all rounds except the 4th, genetic diversity was incorporated by error-prone PCR. The fidelity of Taq polymerase was modified by unbalanced dNTP concentrations in the presence of Mn2+. The PTDH gene was amplified using pRW2 specific flanking primers forward (5'-TTT TTG GAT GGA GGA ATT CAT ATG-3' (SEQ ID NO: 24), Ndel restriction site is underlined) and reverse (5'-CGG GAA GAC GTA CGG GGT ATA CAT GT-3' (SEQ ID NO: 25), Pcil restriction site is underlined). Li round 2, a saturation mutagenesis reverse primer was used (5'-CGT ACG GGG TAT ACA TGT TTA TCN VNN TGC GGC AGG VNN GGC CTT GGG C-3' V=A,C,G, (SEQ ID NO: 34) Pcil restriction site is underlined and degenerate regions in bold) to add additional diversity to the C-terminus of PTDH. A typical PCR reaction mixture contained 50 ng of pRW2 plasmid template, 1 x Promega Taq buffer, 7 mM MgC12, 0.15 MnC12, 0.2 mM each dATP and dGTP, 1 mM each dCTP and dTTP, 0.5 μM of each primer and 5 units of Taq DNA polymerase (Promega) in a 100 μL reaction volume. The reaction was cycled 20 times through typical melting, annealing and extension temperatures of 95, 55, and 72 °C respectively. In the 4th round the E175A mutation was introduced by site directed mutagenesis. The PCR products were purified using a QIAquick PCR purification kit (Qiagen) and digested with Pcil, Ndel, and Dpnl. The resulting products were purified again, and ligated with Ndel and Pcil digested pRW2. Following overnight ligation at 16 ⁰C, the 10 μL reactions containing -150 ng of DNA were precipitated by addition 50 μL water and 500 μL 1-butanol and resuspended in 10 μL water.

[000159] Selection and Screening of the 12X mutant. Aliquots of electrocompetent E. coli BW25141

(50 μL) were transformed with 2.5 μL of the ligated library DNA. These cells were recovered in SOC media for 45 min with shaking at 250 rpm, centrifuged at 1000 xg, and washed with 1 mL of water and resuspended to a final volume of 100 μL. A small portion of these cells was plated on LB + kanamycin (50 μg/mL) solid media to check library size and the remainder was plated on solid PTDH selection media. The library sizes ranged from 2 x 105 to 2 x 106 members. Solid PTDH selection media consisted of 15 g of agarose which was washed several times to remove residual phosphate and autoclaved in 1 L millipore water. This served as the base for 0.4% glucose-MOPS minimal media lacking phosphate to which varying concentrations of phosphite (0.5-250 mM) and arabinose (0.5-10 mM) were added for selection. After 60 hours growth at 37⁰C, individual colonies were picked from the PTDH selective media and grown in 96 well plates in liquid LB plus kanamycin. The clones were then subjected to a secondary screen based on the rate of NBT reduction. The most active clones in the NBT assay were selected, grown and induced in 5 mL LB plus kanamycin, lysed, and reassayed by the NBT method. The best mutants at the end of each round served as the templates for the next round of mutagenesis.

[000160] 12X Mutant Analysis. PTDH genes from the most active mutants of each round were sequenced at the Biotechnology Center of the University of Illinois using the Big DyeTM Terminator sequencing method and an ABI PRISM® 3700 sequencer (Applied Biosystems, Foster City, CA). To accurately determine the fold improvement of the best mutant for each round, lysates prepared as above were assayed in 10 mM NAD, 10 mM phosphite, and 50 mM MOPS buffer, pH 7.25 by measuring the increase in absorbance at 340 nm corresponding to production of NADH. These lysates were also analyzed by SDS-PAGE followed by coomassie blue staining and densitometry analysis using a Bio-Rad Gel-Doc with Quantity One software to determine the relative expression of the mutant PTDH enzymes. Only protein bands on the same gel were compared and this experiment was repeated in triplicate to obtain accurate results. The genes of the best mutants were subcloned into pET15b. These constructs were then used to transform BL21 DE3 (Novagen), which were used to overexpress the encoded His6-tagged PTDH proteins that were purified by immobilized metal affinity chromatography. Protein concentration was determined by Bradford assay and by using the experimentally determined extinction coefficient of 30 mM-1 cm-1. Mutations were visualized in silico utilizing the homology model. The mutations were incorporated into this model using the rotamer explorer operation in Molecular Operating Environment followed by energy minimization using the MMF94s forcefield.

[000161] Steady-state Kinetic Measurements 12X mutant. Kinetic measurements were taken and substrate concentrations were measured as described herein. At least 6 concentrations of each substrate were used varying from below the KM value to at least 5-times higher than the KM value while the other substrate was kept at a constant saturating concentration. The Michaelis-Menton equation was fitted to the data using non-linear least squares regression analysis in Origin 5.0 to determine kcat and KM values. To determine the primary kinetic isotope effect on kcat, deuterated phosphite was prepared and the activity was measured with saturating concentrations of labeled and unlabeled phosphite (20 mM) and NAD (5 mM).

[000162] Fermentation and measure of cell lysate activity of 12X mutant. Non-tagged constructs were created by ligating a Ndel and BamHI digested WT and round 6 PTDH gene into a similarly prepared pET26b vector. These vectors were used to transform BL21 (DE3) and their sequence confirmed by automated DNA sequencing. Fermentation was carried out at the Ohio State University's Fermentation Facility. Briefly, a starter culture was grown overnight in terrific broth plus kanamycin and transferred to the fermentor containing 10 L of terrific broth with kanamycin. The cells were grown to an OD600 of 0.6 and cooled to 25 ⁰C at which time induction of protein expression was started with the addition of 0.5 mM IPTG. After 8 hours of protein expression the cells were collected by centrifugation, weighed, and frozen at -80 ⁰C. Equal amounts of cell lysate prepared by treatment with lysozyme and sonication were analyzed to determine PTDH activity as described in the kinetic measurements section.

[000163] Production of L-tert-Lencine using the 12X mutant. Small-scale regeneration reactions containing 100 mM ammonium trimethylpyruvate, 200 mM diammonium phosphite, 0.4 mM NAD, 5.26 U/mL of leucine DH, and 57.5 μg/mL WT PTDH (0.265 U/mL) or round 6 PTDH (0.508 U/mL). The reactions were mixed gently and incubated at 25 ⁰C. At fixed time intervals, samples were removed from the reaction and immediately frozen at -80 ⁰C. The frozen samples were thawed immediately prior to HPLC analysis. A Shimadzu HPLC equipped with an evaporative light scattering detector was used to quantify the amount of fe/'Meucine in each sample following separation on a Alltech C-18 prevail column with an isocratic elution of 94.5% water, 4.5% acetonitrile, and 1% acetic acid. The peak area oϊtert- leucine in each sample was converted to concentration by a standard curve prepared with five known concentrations of authentic L-terMeucine. The steady state rates for the reactions were determined by fitting the first four data points to a line by linear regression analysis.

[000164] T₅₀. Values of T₅₀, the temperature required to reduce initial enzyme activity by 50% after a fixed incubation period, were determined. Briefly, purified enzymes (0.2 mg/mL) were incubated for 20 min at various fixed elevated temperatures. After incubation, samples were placed on ice for 15 min before being assayed using saturating substrate conditions. Residual activity was determined and expressed as a percentage of the initial activity.

[000165] T_opt, r_Opt was determined by incubating purified enzymes (0.2 mg/mL) with 1 mM phosphite, 0.5 mM NAD in 50 mM MOPS (pH 7.25) at increasing temperatures for 20 minutes, after which the enzyme activity was determined by monitoring the absorbance increase at 340 nm.

DOCUMENTS

These publications are incorporated by reference to the extent they disclose material relevant to the present application.

Banta, S., and Anderson, S. (2002) J. MoI. Evol. 55, 623-31.

Banta, S., Swanson, B. A., Wu, S., Jarnagin, A., and Anderson, S. (2002) Biochemistry 41, 6226- 36.

Berrios-Rivera, S. J., Bennett, G. N., and San, K. Y. (2002) Metab. Eng. 4, 217-29.

Bocanegra, J. A.,^' Scrutton, N. S., and Perham, R. N. (1993) Biochemistry 32, 2737-40.

Bommarius, A. S., and Drauz, K. (1994) Bioorg. Med. Chem. 2, 617-26.

Boonstra, B., Rathbone, D. A., French, C. E., Walker, E. H., and Bruce, N. C. (2000) Appl. Environ. Microbiol. 66, 5161-6.

Bradford, M. M. (1976) Anal. Biochem. 72, 248-54.

Carugo, O., and Argos, P. (1997) Proteins 28, 10-28.

Carugo, O., and Argos, P. (1997) Proteins 28, 29-40.

Chen, R., Greer, A., and Dean, A. M. (1995) Proc. Natl. Acad. ScL U. S. A. 92, 11666-70.

Chen, R., Greer, A., and Dean, A. M. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 12171-6.

Chenault, H. K., and Whitesides, G. M. (1987) Appl. Biochem. Biotechnol. 14, 147-97.

Cleland, W. W. (1979) Methods Enzymol. 63, 103-38.

Corbier, C, Clermont, S., Billard, P., Skarzynski, T., Branlant, C, Wonacott, A., and Branlant, G. (1990) Biochemistry 29, 7101-6.

Costas, A. M., White, A. K., and Metcalf, W. W. (2001) J. Biol. Chem. 276, 17429-36.

Danielson, U. H., Jiang, F., Hansson, L. O., and Mannervik, B. (1999) Biochemistry 38, 9254-63.

Faber, K. (1997) Biotransformations In Organic Chemistry: A Textbook, 3rd ed., Springer- Verlag, Berlin, Germany.

Feeney, R., Clarke, A. R., and Holbrook, J. J. (1990) Biochem. Biophys. Res. Commun. 166, 667- 72.

Galkin, A., Kulakova, L., Ohshima, T., Esaki, N., and Soda, K. (1997) Protein. Eng. 10, 687-90.

Goldberg, J. D., Yoshida, T., and Brick, P. (1994) J. MoI. Biol. 236, 1123-40.

Grimshaw, C. E., Matthews, D. A., Varughese, K. I., Skinner, M., Xuong, N. H., Bray, T., Hoch, J., and Whiteley, J. M. (1992) J. Biol. Chem. 267, 15334-9.

Haldimann, A., and Wanner, B. L. (2001) J. Bacteriol. 183, 6384-93.

Hara, A., Deyashiki, Y., Nakagawa, M., Nakayama, T., and Sawada, H. (1982) J. Biochem. 92, 1753-62.

Holderman, A. and Wanner B.L. (2001) I of Bad. 183, 6384-6393.

Holmberg, N., Ryde, U., and Bulow, L. (1999) Protein. Eng. 12, 851-6.

Huang, Y. W., Pineau, I., Chang, H. J., Azzi, A., Bellemare, V., Laberge, S., and Lin, S. X. (2001) MoI. Endocrinol. 15, 2010-20. Hurley, J. H., Chen, R., and Dean, A. M. (1996) Biochemistry 35, 5670-8.

Issakidis, E., Saarinen, M., Decottignies, P., Jacquot, J. P., Cretin, C, Gadal, P., and Miginiac- Maslow, M. (1994) J. Biol. Chem. 269, 3511-7.

Kochhar, S., Lamzin, V. S., Razeto, A., Delley, M., Hottinger, H., and Germond, J. E. (2000) Eur. J. Biochem. 267, 1633-9.

Koeller, K. M., and Wong, C. H. (2001) Nature 409, 232-40.

Krimm, L, Goyer, A., Issakidis-Bourguet, E., Miginiac-Maslow, M., and Lancelin, J. M. (1999) J. Biol. Chem. 274, 34539-42.

Laemmli, U. K. (1970) Nature 227, 680-5.

Lauvergeat, V., Kennedy, K., Feuillet, C, McKie, J. H., Gorrichon, L., Baltas, M., Boudet, A. M., Grima-Pettenati, J., and Douglas, K. T. (1995) Biochemistry 34, 12426-34.

Leonida, M. D. (2001) Curr. Med. Chem. 8, 345-69.

Liese, A., and Filho, M. V. (1999) Curr. Opin. Biotechnol. 10, 595-603.

McCoy, M. (2001) C.& E.N. 79, 37-43.

Micheli, V., Simmonds, H. A., Bari, M., and Pompucci, G. (1993) Clin. Chim. Acta. 220, 1-17.

Nakanishi, M., Matsuura, K., Kaibe, H., Tanaka, N., Nonaka, T., Mitsui, Y., and Hara, A. (1997) J. Biol. Chem. 272, 2218-22.

Nishiyama, M., Birktoft, J. J., and Beppu, T. (1993) J. Biol. Chem. 268, 4656-60.

Rossmann, M. G., Moras, D., and Olsen, K. W. (1974) Nature 250, 194-9.

Sarkar, G., and Sommer, S. S. (1990) Biotechniques 8, 404-407.

Schepens, L, Johansson, K., Decottignies, P., Gillibert, M., Hirasawa, M., Knaff, D. B., and Miginiac-Maslow, M. (2000) J. Biol. Chem. 275, 20996-1001.

Schmid, A., Dordick, J. S., Hauer, B., Kiener, A., Wubbolts, M., and Witholt, B. (2001) Nature 409, 258-68.

Schuller, D. J., Grant, G. A., and Banaszak, L. J. (1995) Nat. Struct. Biol. 2, 69-76.

Scrutton, N. S., Berry, A., and Perham, R. N. (1990) Nature 343, 38-43.

Serov, A. E., Popova, A. S., Fedorchuk, V. V., and Tishkov, V. I. (2002) Biochem. J. 367, 841-7.

Steen, I. H., Lien, T., Madsen, M. S., and Birkeland, N. K. (2002) Arch. Microbiol. 178, 297-300.

Tishkov, V. I., Galkin, A. G., Fedorchuk, V. V., Savitsky, P. A., Rojkova, A. M., Gieren, H., and KuIa, M. R. (1999) Biotechnol. Bioeng. 64, 187-93. van der Donk, W. A., and Zhao, H. (2003) Curr. Opin. Biotechnol., in press.

Vrtis, J. M., White, A. K., Metcalf, W. W., and van der Donk, W. A. (2001) J. Am. Chem. Soc. 123, 2672-3.

Vrtis, J. M.; White, A. K.; Metcalf, W. W.; Van der Donk, W. A. Angew. Chem., Intl. Ed. 2002, 41, 3257

Vrtis, J. M., White, A., Metcalf, W. W, van der Donk, W.A. (2002) Angew. Chem. Int. Ed. Engl. 41, 3257-3259.

Wang, H., Lei, B., and Tu, S. C. (2000) Biochemistry 39, 7813-9. Wichmann, R., et al. (1981). Biotechnol Bioeng, 67(6): p. 791-804.

Wiegert, T., Sahm, H., and Sprenger, G. A. (1997) J. Biol. Chem. 272, 13126-33.

Wierenga, R. K., De Maeyer, M. C. H., and HoI, W. G. J. (1985) Biochemistry 24, 1346-1357.

Woodyer, R., van der Donk, W. A., and Zhao, H. (2003) Biochemistry 42, 11604-11614.

Yaoi, T., Miyazaki, K., Oshima, T., Komukai, Y., and Go, M. (1996) J. Biochem. 119, 1014-8.

Zaks, A. (2001) Curr. Opin. Chem. Biol. 5, 130-6.

Zhang, L., Ahvazi, B., Szittner, R., Vrielink, A., and Meighen, E. (1999) Biochemistry 38, 11440- 7.

Zhao, H., Chockalingam, K., and Chen, Z. (2002) Curr. Opin. Biotechnol. 13, 104-10.

Zhao, H., Moore, J. C, Volkov, A. A., and Arnold, F. H. (1999) in Manual of Industrial Microbiology and Biotechnology (Demain, A. L., and Davies, J. E., Eds.) pp 597-604, ASM Press, Washington, D.C.

Claims

WE CLAIM:

[C 1] A mutant phosphite dehydrogenase (PTDH) with an increased thermostability and relaxed cofactor specificity for nicotinamade cofactor regeneration as compared to a wild-type phosphite dehydrogenase,

[C2] The mutant phosphite dehydrogenase of claim 1 , wherein the thermostability comprises a temperature optima (r_opt) from about 42⁰C to aboout 59°C.

[C3] The mutant phosphite dehydrogenase of claim 1, wherein soluble expression in E. coli is about three-fold higher than the wild-type phosphite dehydrogenase.

[C4] The mutant phosphite dehydrogenase of claim 1 , wherein activity is about two-fold higher than the wild-type phosphite dehydrogenase.

[C5] The mutant phosphite dehydrogenase of claim 1 , further defined as having an amino acid mutation at an amino acid position selected from the group consisting of 13, 26, 71, 130, 132, 137, 150, 175, 176, 215, 275, 276, 313, 315, 319, 325, 332, and 336 of wild type phosphite deydrogenase with an amino acid sequence as in SEQ ID NO: 1.

[C6] The mutant phosphite dehydrogenase of claim 5, wherein the amino acid mutation is an amino acid substitution selected from the group consisting of D13E, M26I, E175A, E332N, C336D, Q137R, I150F, Q215L, R275Q, L276Q, A319E, V315A, Q132R, V71I, E130K, I313L, A325V, A176R, and E175A.

[C7] The mutant phosphite dehydrogenase of claim 1 comprising an amino acid sequence of SEQ ID

NO: 35, designated as a 12X PTDH mutant.

[C8] A nucleic acid molecule encoding the mutant phosphite dehydrogenase of claim 6.

[C9] A nucleic acid molecule encoding the mutant phosphite dehydrogenase of claim 7.

[C 10] A host cell transformed with the nucleic acid molecule of claim 8 or 9.

[C 11] An expression vector encoding the nucleic acid molecule of claim 8 or 9.

[C 12] Substantially purified mutant phosphite dehydrogenase of claim 1.

[C 13] The mutant phosphite dehydrogenase mutant of claim 7 further comprising an amino acid mutation at position 176.

[C 14] The mutant phosphite dehydrogenase mutant of claim 13, wherein the amino acid mutation is

A176R.

[C15] The mutant phosphite dehydrogenase mutant of claim 14 comprising an amino acid sequence of

SEQ ID NO: 36, designated as a 12X +A176R mutant PTDH.

[C16] The mutant phosphite dehydrogenase mutant of claim 1 further comprising an amino acid mutation at position 175.

[C17] The mutant phosphite dehydrogenase mutant of claim 16, wherein the amino acid mutation is

E175A.

[C 18] The mutant phosphite dehydrogenase mutant of claim 14, wherein the mutant comprises a rate of reaction of about 1.8 times faster than the wild-type phosphite dehydrogenase specific for NAD and about 2.2 times faster than folate dehydrogenase (FDH) specific for NADP.

[C 19] A method of generating at least one of NADH and NADPH, comprising:

(c) providing a mutant phosphite deydrogenase, wherein the mutant has an amino acid mutation selected from the group consisting of mutations at positions 13, 26, 71, 130, 132, 137, 150, 175, 176, 215, 275, 276, 313, 315, 319, 325, 332, and 336 and;

(d) generating at least one of NADH and NADPH by a reduction reaction of at least one of NAD⁺ and NADP⁺.

[C20] The method of claim 19, wherein the amino acid mutation comprises an amino acid substitution selected from the group consisting of D13E, M26I, E175A, E332N, C336D, Q137R, I150F, Q215L,

R275Q, L276Q, A319E, V315A, Q132R, V71I, E130K, I313L, A325V, A176R, and E175A. [C21] The method of claim 20, wherein the mutant phosphite deydrogenase comprises an amino acid sequence of SEQ ID NO: 35. [C22] The method of claim 20, wherein the mutant phosphite deydrogenase comprises an amino acid sequence of SEQ ID NO: 36. [C23] The phosphite dehydrogenase of claim 1, further defined as having increased efficiency for cofactors NAD⁺ and NADP⁺ as compared to non-mutated phosphite dehydrogenase. [C24] The phosphite dehydrogenase of claim 2 comprising mutations from Glul75 to Ala 175 and from

Alal76 to Arg 176 in the wild type amino acid sequence in accordance with FIG. 7. [C25] The phosphite dehydrogenase of claim 2 comprising a mutation from Glul75 to Ala 175 in a wild type amino acid sequence in accordance with FIG. 7. [C26] The phosphite dehydrogenase of claim 2 with a mutation from Alal76 to Argl76 in a wild type amino acid sequence in accordance with FIG. 7. [C27] The phosphite dehydrogenase of claim 6 selected from the group consisting of enzymes with a mutation designated Q132R (SEQ E) NO: 6); Q137R (SEQ ID NO: 7); I150F (SEQ ID NO: 8); Q 215 L

(SEQ ID NO: 9); R 275 Q (SEQ ID NO: 10); Q137R/I150/F, q215L/R275Q (SEQ ID NO: 11); and

Q132R/Q 137R/I150F/Q215L/R275Q1 (SEQ ID NO: 12). [C 28] The phosphite dehydrogenase mutant of claim 1 characterized by increased activity and improved soulbility and expression. [C29] Use of the phosphite dehydrogenase of claim 1 to regenerate NAD⁺, NADP⁺ or both NAD⁺ and

NADP⁺.