We have purified and characterized a variant NADH-oxidase from Thermus thermophilus HB27. This enzyme oxidizes NADH to NAD+, utilising oxygen as an electron acceptor and a flavin cofactor as an electron mediator, to yield hydrogen peroxide as a product. The amino acid sequence of this NOX enzyme presents 98.5% identity to a similar enzyme isolated from T. thermophilus HB8. However, the new NOX variant presented 5.9-fold higher catalytic efficiency than its counterpart from the strain HB8, a difference that is likely related to the three divergent residues found at positions 166, 174 and 194 (Additional file 3 Figure S2). Noteworthy, the reported sequence for NOX also differs from the sequence published for the HB27 strain by a single amino acid change at position 194 (H194Y), suggesting either an annotation error at the original sequence  or an adaptation of the strain to the laboratory conditions. Because samples of the HB27 strain obtained from other laboratories did not contain such mutation, it is tempting to speculate that higher NADH-oxidase activity might enhance bacterial fitness in the very rich TB medium routinely used in the laboratory, where an excess of reducing power (increased NADH/NAD+ ratio) is likely to occur. In any case, we have found a potential target on NOX primary sequence to further optimize its catalytic performance by protein engineering.
Steady-state kinetic parameters were calculated for purified NOX with three different cofactors, NADH, FMN and FAD. Analyses of the results, revealed one main difference between this NOX and that from the HB8 strain, the catalytic efficiency of NOX towards NADH was higher than what it was reported for that one from the HB8 strain. Currently, mutagenesis experiments are being undertaken in order to shed light on the specific role of the three residues that differ between both proteins on the catalytic activity.
We also described practical thermal hyperactivation of this enzyme under limited conditions of flavin cofactor. This insight suggests a more active NOX conformation at high temperatures, resulting in higher affinity for the flavin cofactor. This beneficial conformation and associated good specific activity can be fixed and retained in further downstream enzyme applications even at low temperature. The nature of this thermally induced conformational change is not clear at present. We suggest that it could be due to enhanced binding of the flavin cofactor, since it was only observed under conditions in which the flavin cofactor is limiting. Similar hyperactivation effects after high temperature incubation have been described for other thermophilic proteins also likely due to conformational reordering of particular regions of the protein missfolded during expression at low temperatures [[48–50]]. Whatever the reason, this feature gives the NOX described here a high potential for industrial use in NADH recycling in redox reactions. In this sense, we have shown its use coupled to a dehydrogenase for alcohol oxidations at mild temperatures without a requirement for exogenous addition of flavin cofactor, making the process much more cost-effective.
Thinking on these applications, we immobilized the NOX in a broad number of surfaces. In the literature, different immobilization protocols have been used for the immobilization of NADH oxidases especially in the biosensors field [[51–53]]. From our experiments, covalent attachment of NOX to an agarose support activated with glyoxyl groups resulted, to the best of our knowledge, in the most thermostable NADH-oxidase preparation reported so far. This Gx-NOX derivative shows at the same time irreversible binding of the enzyme and a likely homogeneous protein orientation because preferential covalent binding occurs through the lysine richest regions . These properties related to the immobilization chemistry are the basis for the successfully application to a large number of enzymes, all of which achieved a high stabilization factors .
In addition to the high stability of the insoluble biocatalyst, NOX can also be successfully reactivated once it has been fully or partially inactivated. Here, we have presented the first report of protein reactivation for immobilized NOX on highly stabilized derivatives. Moreover, these derivatives can be efficiently reactivated several times by incubation on phosphate buffer after inactivation cycles in 60 vol % dioxane. An explanation for this could be based on the fact that organic solvents drive enzyme to local distortion rather than a global unfolding . These local distortions may be located on the surface of the enzyme because organic solvents may strip the external water layer essential for enzyme activity [[55–57]] (easier to recover when organic solvent is removed) or might occur at a higher scale by unfolding internal domains of the proteins because penetration of the solvent to the hydrophobic core (more difficult to revert). As multipoint covalent immobilization limits severe unfolding , the enzyme can be easily reactivated simply by elimination of the organic solvent. The same behavior was found for other enzymes like lipases  and even for other thermophilic oxidoreductases such as glutamate dehydrogenase from Thermus thermophilus HB7 . In this last example, recovery of active quaternary structure was achieved, making a breakthrough in solid-phase enzyme reactivation protocols for multimeric enzymes.
Finally, this highly stable and heterogeneous biocatalyst was applied as cofactor recycling partner of a main alcohol dehydrogenase from the same thermophilic source, to kinetically resolve pharmaceutically relevant compounds such as 1-phenylethanol. Expectedly, the selectivity of the main alcohol dehydrogenase was extremely high  and cofactor recycling by NOX allowed reaching the maximum yield using only 6.5% mol of cofactor, which means that one mol of cofactor was recycled up to 10 times per reaction cycle under the conditions studied here. Moreover, since this heterogeneous biocatalyst is highly stable and can be re-used for many cycles, the total turnover of such catalyst can be increased. In addition, and still to be tested, coupling of a catalase to the system should improve the global stability of the enzymes involved in the process. The catalase would be able to, in-situ, eliminate the hydrogen peroxide formed as byproduct by NOX. As a result there would be decreased accumulation of such compound capable of inactivating the protein biocatalysts, which would lead to further optimization of both conversion rate and final yields.