Cofactor Specificity Engineering of Streptococcus mutans NADH Oxidase 2 for NAD(P)+ Regeneration in Biocatalytic Oxidations

Soluble water-forming NAD(P)H oxidases constitute a promising NAD(P)+ regeneration method as they only need oxygen as cosubstrate and produce water as sole byproduct. Moreover, the thermodynamic equilibrium of O2 reduction is a valuable driving force for mostly energetically unfavorable biocatalytic oxidations. Here, we present the generation of an NAD(P)H oxidase with high activity for both cofactors, NADH and NADPH. Starting from the strictly NADH specific water-forming Streptococcus mutans NADH oxidase 2 several rationally designed cofactor binding site mutants were created and kinetic values for NADH and NADPH conversion were determined. Double mutant 193R194H showed comparable high rates and low K m values for NADPH (k cat 20 s-1, K m 6 µM) and NADH (k cat 25 s-1, K m 9 µM) with retention of 70% of wild type activity towards NADH. Moreover, by screening of a SeSaM library S. mutans NADH oxidase 2 variants showing predominantly NADPH activity were found, giving further insight into cofactor binding site architecture. Applicability for cofactor regeneration is shown for coupling with alcohol dehydrogenase from Sphyngobium yanoikuyae for 2-heptanone production.


Introduction
Enzyme catalyzed oxidation reactions have gained increasing interest in biocatalysis recently, reflected also by a number of excellent reviews on this topic published in the last years [ -3]. Oxidoreductases constitute an important group of biocatalysts as they facilitate not only the widely used stereoselective reduction of aldehydes and ketones but also the less well exploited oxidation of alcohols and amines. Oxidoreductases catalyzed oxidations are also used for production of chiral alcohols and amines by deracemization [ ,4-6].
Oxidoreductases, especially aldo-keto-reductases and dehydrogenases, act on the substrate by the transfer of electrons from or to a cofactor, mostly the nicotinamide-based nucleotides NAD(H) and NADP(H). As nicotinamide cofactors are expensive, regeneration of cofactors is necessary for economically feasible biocatalytic processes. While for the regeneration of the reduced cofactors NADH and NADPH several systems (engineered formate dehydrogenase [7,8], phosphite dehydrogenase [9, 0], glucose dehydrogenase [ , 2] plus cosubstrate) are well established and widely used, universal regeneration systems for the oxidized forms NAD + and NADP + are less well developed.
Enzyme based, electrochemical, chemical, and photochemical regeneration methods are known. Coupled substrate or coupled enzyme systems [4,3,4] constitute two possibilities for enzymatic NAD(P) + recycling. In these reaction set-ups the cofactor is regenerated via the reduction of a carbonyl group of a cosubstrate, catalyzed either by the production enzyme itself (coupled substrate) [ 5] or by an additionally added dehydrogenase (coupled enzyme; glutamate dehydrogenase [ 6,7], lactate dehydrogenase [ 8]). Carbonyl cosubstrate reductions by dehydrogenases normally provide little thermodynamic driving force for mostly energetically unfavorable biocatalytic alcohol oxidations. Generally, it is therefore necessary to supply the cosubstrate in excess to achieve high substrate conversion rates. In recent studies several smart concepts have been introduced to reduce the need for cosubstrate. The use of one-way cosubstrates [ 9] or cofactor regeneration as an integral part of a redox neutral multi-enzyme network [20,2 ] was reported.
Several cofactor regeneration systems benefit from the high driving force of molecular oxygen as hydrogen acceptor. One example therefore is a 9, 0 phenantrenequinone/xylose reductase system where the quinone is auto-reoxidized by oxygen [22]. O2 reduction also drives cofactor regeneration via mediators as ABTS or Meldola's blue, which are reoxidized by a laccase under H2O formation [23][24][25]. Instead of using laccase the mediator reoxidation can also be achieved by electrochemical means, albeit at moderate turnover numbers. To overcome the still rather low productivity of electrochemical regeneration processes careful reaction and cell design is necessary [26][27][28]. In pure chemical regeneration processes the chemical agent directly reoxidizes the cofactor without biocatalyst. Often Ruthenium complexes are used as oxidants [ 4]. The direct regeneration of NAD(P) + via FMN was found to be strongly accelerated by lightinduced excitation of FMN [29].
A very promising NAD(P) + regeneration method is the application of soluble NAD(P)H oxidases (EC .6.3. NOX) from bacteria or archaea which use molecular oxygen as oxidant. This CSBJ Abstract: Soluble water-forming NAD(P)H oxidases constitute a promising NAD(P) + regeneration method as they only need oxygen as cosubstrate and produce water as sole byproduct. Moreover, the thermodynamic equilibrium of O2 reduction is a valuable driving force for mostly energetically unfavorable biocatalytic oxidations. Here, we present the generation of an NAD(P)H oxidase with high activity for both cofactors, NADH and NADPH. Starting from the strictly NADH specific waterforming Streptococcus mutans NADH oxidase 2 several rationally designed cofactor binding site mutants were created and kinetic values for NADH and NADPH conversion were determined. Double mutant 93R 94H showed comparable high rates and low Km values for NADPH (kcat 20 s -, Km 6 µM) and NADH (kcat 25 s -, Km 9 µM) with retention of 70 % of wild type activity towards NADH. Moreover, by screening of a SeSaM library S. mutans NADH oxidase 2 variants showing predominantly NADPH activity were found, giving further insight into cofactor binding site architecture. Applicability for cofactor regeneration is shown for coupling with alcohol dehydrogenase from Sphyngobium yanoikuyae for 2-heptanone production.

Oxidase 2 for NAD(P) + Regeneration in Biocatalytic Oxidations
regeneration method has the advantage of being cheap as no cosubstrate or mediator is needed. Straightforward downstream processing is possible as only hydrogen peroxide or water is formed as byproduct. Moreover, the high redox potential of oxygen results in a high thermodynamic driving force. The electron and hydrogen transfer from NADH to oxygen is catalyzed by known soluble NAD(P)H oxidases via a two electron transfer producing hydrogen peroxide (equation ) or a four electron transfer producing water (equation 2) [30,3 ]. NAD(P)H + O2 + H +  NAD(P) + + H2O2 (1) The four electron transferring oxidases are the preferred choice for cofactor regeneration as they only form water as byproduct. In case of H2O2 production, catalase has to be added to the system to prevent enzyme damage by the peroxide. Water-forming NADH oxidases have been studied from several bacteria as Streptococcus [32][33][34][35] NAD(P)H oxidases belong to the pyridine nucleotide disulfide oxidoreductases (PNDOR) together with, among others, glutathione reductase and CoA-disulfide reductase [46]. NOX enzymes contain a single conserved redox-active cysteine that circulates between the thiol/thiolate and the sulfenic acid state during catalysis. Overoxidation of the cysteine leads to enzyme deactivation. Several NADH oxidases need FADH or DTT addition for optimal performance [36,45,47]. Enzymes with high specific activities (> 50 U/mg) were recently reported from L. sanfranciscensis, L. plantarum, L. rhamnosus and S. pyogenes [34,37,38,48]. A drawback in using NADH oxidases for cofactor regeneration is that almost all water-forming NADH oxidases are specific for NADH. In wild type form only two water forming NOXs and one hydrogen peroxide forming NOX show activity with NADPH (around half / one third of activity with NADH depending on the NOX [38,49,50]. NOX 299 from T. kodakarensis is the only NOX showing higher wild type activity with NADPH than with NADH but disadvantageously it produces high amounts of H2O2 [45]. NOX of L. plantarum was recently successfully mutated to accept NADPH but resulting in a simultaneous decrease in NADH activity [37]. The kcat value for NADH reaction of the variant with highest catalytic efficiency with NADPH was six-fold reduced compared to the wild type. In this study we aimed at developing an NAD(P)H oxidase which is universally applicable for regeneration of NADH as well as NADPH. As starting point NADH specific water-forming S. mutans NADH oxidase 2 (SmNOX) was chosen. SmNOX was chosen as it was well characterized to be stable, highly active, not dependent on FADH or DTT addition [5 ], and had already been expressed in E. coli before [32]. Moreover, a crystal structure of a closely related enzyme from S. pyogenes was available which enabled us to model the SmNOX structure. In a thorough mutation study of the cofactor binding site a SmNOX mutant with matched activities with NADH and NADPH was generated. Mutants with increased NADPH/NADH activity ratios were identified by SmNOX library screening. The conversion of 2-heptanol to 2-heptanone with NADPH regeneration by engineered SmNOX was shown. Methods E. coli TOP 0F' was originally bought from Invitrogen (Carlsbad, CA, USA), E. coli BL2 -Gold (DE3) was from Stratagene (La Jolla, CA, USA). NAD(P)H was from Roche Diagnostics GmbH (Mannheim, Germany) or Roth (Karlsruhe, Germany). Materials for cloning were from Fermentas (St. Leon-Roth, Germany, now Thermo-Fisher Scientific) if not stated otherwise. All other chemicals were purchased from Sigma-Aldrich, Fluka (St. Luis, MO, USA) or Roth (Karlsruhe, Germany) if not stated otherwise.
Homology modeling for Streptococcus mutans NOX 2 was based on an X-ray structure of NADH oxidase from Streptococcus pyogenes (template: 2BC0A, 2.00 Å). The sequence identity between target and template was 77.5 %. The homology model was created with the automated protein structure homology-modeling server  Table . PCR reaction mixtures (50 µL) contained template plasmid (28 pM Additionally, a plasmid pMSsN2 was transformed into an E. coli BL2 (DE3) giving a strain overexpressing Lactobacillus sanfranciscensis NOX (LsNOX). pMSsN2 is identical to pMSsN except that it carries the gene coding for LsNOX (synthetic variant, ordered at DNA2.0, protein sequence: GI: 862874 ) instead of the SmNOX. Precultures of all resulting E. coli strains were cultivated in LB media (50 mL) containing Ampicillin ( 00 mg/L) in baffled shake flasks (300 mL) at 37 °C and 30 rpm overnight. Main cultures were inoculated to an OD of 0.05 in the described medium (250 mL in L baffled flasks) and cultivated at 37 °C and 30 rpm. NOX production was induced at OD 0.8-by addition of IPTG ( mM). Cells were harvested after an overnight induction period (25 °C, 0 rpm) by centrifugation ( 5 minutes, 5000 rcf, Avanti J-20 XP centrifuge, Beckman Coulter, Krefeld, Germany, rotor JA-0). Cell pellets were diluted in potassium phosphate buffer (50 mM, pH 7.0) to a final volume of 25 mL. Cell breakage was achieved by ultrasonication with a Branson sonifier 250 (Branson ultrasonic corporation, Danbury, CT, USA) for 6 minutes at 50 W with continuous cooling, pulsed with one 700 ms pulse per second with a cm diameter tip. Cell free lysates were prepared by collecting the supernatant of centrifugation at 36000 rcf (rotor JA-25.50) for 45 minutes and concentrating it to half the volume via Vivaspin 20 centrifugal concentration tubes with 30 kDa molecular weight cutoff (Sartorius, Göttingen, Germany). Cell free extracts from an E. coli strain expressing S. yanoikuyae ADH (SyADH) from the pEamTA based plasmid pEam_SyADH was prepared following the same protocol.
The protein content was determined with bichinonic acid protein assay (BCA) kit (Thermo Scientific, Waltham, MA, USA) using BSA as standard. For SDS-PAGE NuPAGE® 4-2 % Bis-Tris Gels, .0 mm, from Invitrogen, (Carlsbad, CA, USA) were used with a NuPAGE MOPS SDS Running Buffer for Bis-Tris Gels. All strains were stored as glycerol stocks at -80 °C. Cell free extracts were stored in aliquots at -20 °C.
Initial rate data of NAD(P)H oxidation were acquired measuring the decrease in NAD(P)H absorption at 340 nm (= 6220 Mcm -) in potassium phosphate buffer (50 mM, pH 7.0) at 25 °C. The tempered buffer was vortexed for saturation with oxygen before mixing with the enzyme and cofactor solution. Absorption measurements were performed on a Spectramax Plus 384 (Molecular Devices, Sunnyvale, CA, USA) or on a Synergy MX (Biotek, Winooski, VT, USA) in UV-star micro titer plates (Greiner, Kremsmünster, Austria). The total reaction volume was 200 µL, reactions were started by addition of NAD(P)H. Data collection was started after 5 seconds of mixing. For activity measurements with FAD, enzyme preparations first were pre-incubated in 0 fold concentration in 50 mM KPi containing 25 µM or 250 µM FAD and then were diluted : 0 in the same buffer for initial rate detection after addition of NADH. In case of DTT addition the assay buffer contained 5 mM of DTT.
Apparent kinetic parameters were obtained from initial rate measurements at air saturation oxygen level (250 M) with eight cofactor concentrations varying over a concentration range of 5 to 0 times the apparent Km or to a maximum NAD(P)H concentration of mM. Enzymes were applied as crude lysates in dilutions chosen to give rates between 0.00 and 0.05 ΔAbs/min and rates were constant for ≥ minute. Appropriate controls containing crude lysate without overexpressed NOX verified that blank rates were insignificant for all conditions used. Results from initial rate measurements were fitted to a Michaelis-Menten type equation (equation 3) using unweighted least-squares regression analysis performed with Sigmaplot program version (Systat Software Inc.). v = vmax*A/(Km+A) (3) v is the initial rate, vmax is the apparent maximum rate (U/mg total protein in cell free extract), A the cofactor concentration and Km the apparent Michaelis constant for NAD(P)H at air saturation oxygen levels.
Enzyme assay reaction mixtures with mM of NAD(P)H were fully converted with purified SmNOX wild type or variant 93R 94H. The assay solution or glucose standard solutions were diluted + with an o-dianisidine / glucose oxidase / horse radish peroxidase mixture [54] and absorption was measured at 460 nm.
SmNOX wild type gene and variant 93R 94H were recloned in pEHISTEV vector [55] via EcoRV/HindIII restriction sites. A pEHISTEV version was used in which a second EcoRV restriction site was eliminated by introduction of a silent mutation. Plasmid preparations checked for correct sequence were transformed into E. coli BL2 (DE3) Star. Expression and cell free extract preparation were done as described above but in LB/Kanamycin medium. The cells were harvested by centrifugation (5000 g, 5 minutes). The pellet was resuspended in 50 mM KPi pH 7.0 and disrupted by ultrasonication. The cell free extract was applied to a 5 mL Ni-Sepharose 6 Fast Flow column (GE Healthcare, Chalfont St Giles, UK). The tagged enzymes were obtained by a one-step purification using the buffers recommended in the manual. After purification, the enzyme buffer was exchanged to 50 mM KPi and enzymes were concentrated to protein concentrations above 5 mg/mL by Vivaspin 20 tubes with 10 kDa molecular weight cutoff (Sartorius) before storage at -20 °C. The 6xHis tag was cleaved off from mg NOX by incubation with tobacco etch virus protease (TEV protease) in a reaction mixture containing 0.2 mM EDTA, and mM DTT in 50 mM Tris/HCl buffer, pH 8.0 by overnight incubation at 4 °C. 0 µg TEV protease were used per mg NOX. The mixture was applied on the Ni-Sepharose 6 Fast Flow column and washed through with 5 mL of 30 mM sodium phosphate buffer containing 0.3 M NaCl and 20 mM imidazole, pH 7.5. The flow through was collected. Buffer exchange to 50 mM KPi samples and concentration to > mg/mL protein was done in Vivaspin 20 centrifugal concentrators (Satorius AG, Göttingen, Germany), aliquots of concentrated NOX solutions were stored at -20 °C.
A sequence saturation mutagenesis library [56] of SmNOX gene with random mutations was bought at SeSaM Biotech (Aachen, Germany). The library was based on mutant SmNOX 94H200K. The library was cloned into pMS470 vector and transformed into E. coli TOP 0F'. Transformants were picked into 60 µL LB/Ampicillin media in 384 well plates, grown overnight at 37 °C and 60 % of humidity and stored as 5 % glycerol stocks at -80 °C. Cultivation of the expression library was done in 96 well plate format. Preculture plates with 50 µL of LB/Ampicillin media per well were inoculated from glycerol stock plates and cultivated at 37 °C at 60 % humidity for at least 2 hours. Main culture plates with V-shaped bottom contained 80 µL of LB/Ampicillin media and were inoculated from the preculture plates. After 8 hours of growth at 37 °C and 60 % humidity SmNOX2 expression was induced by addition of 20 µL of a 0.5 mM IPTG solution in LB/Ampicillin media. The plates were kept at 28 °C and 60 % humidity for 6 hours. Cells were harvested by 5 minutes of centrifugation at 2500 g. Supernatant was decanted and the cell pellets were frozen at -20 °C for at least two hours. Screening assays were carried out in 96 well plates. After thawing cell lysis was accomplished by addition of 00 µL lysis buffer (50 mM KPi, pH 7.0, mg/mL lysozyme) and an h incubation at 28 °C at 600 rpm. Cell debris was separated by centrifugation at 2500 g for 5 min at 4 °C. The supernatant was diluted + with 50 mM KPi, pH 7.0 and used for screening assays. 40 µL of 50 mM KPi, pH 7.0, were added to 0 µL of diluted supernatant in two plates in parallel. Reactions were started by addition of 50 µL of a 0.8 mM NADH or NADPH solution. Initial rates of NADH and NADPH conversion were measured by detection of decrease in absorption at 340 nm over three minutes. Activity with NADH and NADPH was compared for each well. 2800 clones were screened, 480 thereof were chosen for a re-screen and the best 40 thereof were measured in a re-re-screen. Best variants were finally cultivated in shake flasks. From best variants plasmid DNA was isolated with Gene JetTM Plasmid Miniprep Kit (Fermentas, St. Leon-Roth, Germany) and sent for sequencing.
Conversion experiments were set up in .5 mL reaction tubes. The reaction mixture contained NADP + ( 00 µM) and 2-heptanol ( 0 mM) in potassium phosphate buffer (50 mM, pH 7.0) in a total volume of 500 µL. Sphingobium yanoikuyae ADH (SyADH) was applied as crude E. coli lysate and SmNOX 93R 94H was applied as purified enzyme in amounts to give U/mL. After 2 hours at 25 °C and 600 rpm 00 µL of n-butanol (50 mM) was added as internal standard for GC analysis and the mixture was extracted with ethyl acetate (500 µL). Substrate conversion was determined by GCanalysis on a Varian CP7503 gas chromatograph equipped with an FID detector (300 °C) and a Phenomenex ZB-FFAP column (30 m x 0.32 mm; 0.25 µM) with a Restek Hydroguard MT precolumn (5 m x 0.32 mm). H2 was used as carrier gas (2.7 mL/min). The following temperature program was used: 65 °C to 0 °C, 9 °C/min; 0 °C to 60 °C, 25 °C/min. The retention time for 2-heptanol was 2.48 min and for 2-heptanone 3.44 min.

Results and Discussion
Water-forming Streptococcus mutans NADH oxidase 2 (SmNOX) is a monomeric 50 kDa enzyme which is NADH specific [5 ]. We intended to establish SmNOX as universal NAD(P) + regeneration system by engineering the NADH specific wild type towards the effective usage of both cofactors, NADH and NADPH. Ideally, the created variant should have comparable characteristics for both cofactors to simplify application for cofactor regeneration in industrial processes with varying oxidizing enzymes. In NOX enzymes the nicotinamide cofactor is bound in the well described Rossmann fold manner [57]. In Rossmann fold enzymes often an acidic residue, typically an aspartate at the C-terminus of the second -strand of the alternating αα-regions plays a key role in NAD(H) binding [58] by forming hydrogen bonds to the 2'-OH and 3'-OH of the adenine ribose. In contrast, NADP(H) specific Rossmann fold enzymes typically miss this acidic residue and instead carry a basic residue at the following amino acid position. Positive charges in the cofactor binding site facilitate the binding of the negatively charged phosphate group present in NADP(H) but not in NAD(H). The relevance of the described positions for cofactor specificity has first been shown in a mutation study of glutathione reductase by Scrutton et al. [58]. NADH oxidase from Lactobacillus sanfranciscensis (LsNOX) exhibits a limited NADPH activity in parallel to NADH activity [38]. An alignment of the SmNOX sequence with the glutathione reductase, B. anthracis coenzyme Adisulfide reductase and LsNOX sequence (Figure )   fold the Km value to equation 3. Results are listed in Table 2. All mutants were cultivated under identical conditions and SDS-PAGE analysis showed a comparable expression level for all variants (see Figure 3).
Ratios of NADPH and NADH activities and efficiencies measured from one cell free extract in parallel clearly indicated cofactor specificity changes between wild type and the variants. As expected, wild type SmNOX showed only marginal activity with NADPH, while LsNOX wild type showed NADPH oxidation, as reported [38]. In SmNOX Mut and Mut2 (D 92A and D 92N) the negatively charged aspartate, which is known to be a key residue for NADH binding, is missing. In absence of the negative charge the activity with NADH was slightly reduced and higher NADH Km values were detected. Remarkably, only by the exchange of the aspartate to non-polar or positively charged residues NADPH conversion rates increased to levels comparable to NADH rates, albeit with 0 fold higher Km values.
The introduction of a positively charged residue next to the aspartate in Mut3 and Mut4 (V 93R, V 93H) without removal of the aspartate also enabled activity with NADPH although to a lower extent than the aspartate removal. NADH activity was not reduced in Mut3 and Mut4 compared to wild type SmNOX containing extracts.
The introduction of a positive charge was even more effective at position 200. NADH activity stayed unchanged compared to wild type for a Mut6 (G200K) (Figure ). The results from the SmNOX mutation study indicate that these positively charged residues might be responsible for making TkNOX the only known bacterial wild type NOX showing higher activity with NADPH than with NADH.
Since Scrutton et al. [58] reported the first NAD(P)H specificity engineering of an enzyme, a vast number of enzymes have been mutated to alter cofactor specificities. The outcome of these mutation studies provided evidence that the effects obtained by site directed mutagenesis of positions known to be relevant for cofactor specificity, especially the aspartate or glutamate at the end of the second -sheet, vary tremendously. Only in very few cases the catalytic efficiency of reactions with the originally disfavored cofactor could be increased to values of the same order of magnitude as for reactions with the originally used cofactor in the unmutated enzyme [ 0,[62][63][64]. For mutants in which the conserved aspartate or glutamate was exchanged to smaller non-polar residues, the efficiency with NADP + ranged from /4000 to /3 of the efficiency of the unmutated enzymes with NAD + [ 0,65,66]. Especially rare are examples of increased efficiency with one cofactor while keeping also the efficiency with the other cofactor high [ 0,63]. In the outstanding case of B. subtilis lactate dehydrogenase the mutation of valine, the residue equivalent to SmNOX V 93, into arginine led to a 40 fold increased NADPH kcat value. The increase in NADPH kcat did not lead to a decrease but even to a four-fold increase in NADH kcat. Km values stayed at wild type NADH level for both cofactors for the variant. In NADH oxidase from L. plantarum [37] the introduction of positively charged residues next to the conserved aspartate enabled kcat values for NADPH of up to 69 % of wild type level with NADH but decreased kcat for NADH to 58 % in variant G 78R and to 6 % in G 78V/L 79R. Interestingly, the single mutation G 78R led to a low NADH Km of 6 µM compared to 50 µM in the wild type form. NADPH Km of the same variant was 490 µM and could be decreased to 9 µM in the G 78V/L 79R double mutant.
In summary, also in comparison to other enzymes, SmNOX turned out to be an excellent choice for the generation of an NAD(P)H oxidase with comparable kinetic characteristics for both nicotinamide cofactors without drastic loss of activity or increase in Km compared to the wild type enzyme.
Around 3000 variants of a sequence saturation mutagenesis library (bought at SeSaM Biotech) built on SmNOX 94H200K were screened for increase in NADPH/NADH activity ratios. We chose a starting point for the library without the well-studied mutation 93R as we rather aimed at finding new promising mutation combinations that were so far unknown to give high activities with NADPH.
No variant with significantly increased NADPH activity without loss in NADH activity compared to SmNOX 94H200K could be detected in the library screening. However, three mutants were identified with clearly increased NADPH/NADH activity ratio albeit with concomitant decrease in NADH activity. Figure 4 shows NADH and NADPH activity levels of crude lysates of mutant 94H200K202N, 94H200K202C and 94H200K340V388T compared to variant SmNOX 94H200K. A 94H200K340V variant without 388T mutation was constructed to check for the influence of the mutation at position 340 without mutation at 94H200K340V showed the same NADH/NADPH activity ratio as variant 94H200K340V388T. NADH activity was around 60 % of NADPH activity in both cases. We conclude that mutation of position 340 has a high impact on cofactor specificity while mutation 388T probably has no influence on cofactor binding. Position 202 is located at the end of the loop which connects cofactor binding site strand B and helix αB and starts after aspartate 92. The modeled SmNOX structure indicates that residue 202 is positioned too far away from bound NAD(P)H in order to allow direct interaction with the cofactor (Figure 2). Strikingly, position 340 is located at the beginning of a -sheet in close vicinity to the before described loop end. The SmNOX wild type model indicates a potential hydrogen bond between Y202 and N340. In LsNOX the equivalent positions are occupied by a tyrosine and a valine, which cannot form the hydrogen bond. We speculate that a presumably higher flexibility of the loop caused by removal of the hydrogen bond hampers the activity with the more structurally demanding NADPH less than activity with NADH.
SmNOX wild type and variant 93R 94H were recloned in vector pEHISTEV for expression with an N-terminal 6xHis tag. The enzymes were purified to apparent electrophoretic homogeneity by Ni-affinity chromatography as demonstrated in Figure 5. After purification the tag was cleaved off by treatment with TEV protease. Apparent kinetic values for NAD(P)H oxidation were determined in air saturated buffer as described in the methods section. kcat and Km values for SmNOX wild type and mutant 93R 94H are shown in Table 3. The SmNOX wild type kcat value corresponded to a specific activity of 44 U/mg. Higuchi et al. [5 ] reported a specific activity of water-forming S. mutans NOX 2 of 00 U/mg. The lower specific activity measured here was not unexpected due to a lower assay temperature than in the previous study. However, this lower activity could also be caused by enzyme deactivation during the tag cleavage procedure which included an overnight incubation at 4 °C. With SmNOX mutant 93R 94H now an efficient NADH oxidase is available which has very similar kinetic values for oxidation of NADH as well as NADPH. . Cell free extract (CFE) activities of S. mutans NADH oxidase 2 variants found in a sequence saturation mutagenesis library by screening for improved NADPH/NADH activity ratios. The library was built on SmNOX 194H200K and initial rates were measured employing 300 µM NADH or NADPH.