More e ﬃ cient redox biocatalysis by utilising 1,4-butanediol as a ‘ smart cosubstrate ’†

1,4-Butanediol is shown to be an e ﬃ cient cosubstrate to promote NAD(P)H-dependent redox biocatalysis. The thermodynamically and kinetically inert lactone coproduct makes the regeneration reaction irreversible. Thereby not only the molar surplus of co-substrate is dramatically reduced but also faster reaction rates are obtained.

1,4-Butanediol is shown to be an efficient cosubstrate to promote NAD(P)H-dependent redox biocatalysis.The thermodynamically and kinetically inert lactone coproduct makes the regeneration reaction irreversible.Thereby not only the molar surplus of cosubstrate is dramatically reduced but also faster reaction rates are obtained.
Enzymes are amongst the catalysts of choice if selectivity is desired.This is particularly true for oxidoreductases catalysing preparatively important reactions ranging from reduction of CvO, CvC and other functional groups to specific oxyfunctionalisation reactions such as hydroxylation, epoxidation or Baeyer-Villiger reactions. 1Most of these reactions depend on the supply with reducing equivalents, delivered to the enzymes through the nicotinamide cofactors (NAD(P)H).For economic and practical reasons, NAD(P)H has to be applied in catalytic amounts combined with a suitable in situ regeneration system. 1,2After more than 2 decades of intensive research 3 this 'cofactor challenge' is generally considered to be solved.Amongst a variety of different regeneration systems, ADHmediated oxidation of simple alcohols such as ethanol or isopropanol represents one of the most common NAD(P)H regeneration systems (Scheme 1).This approach is most elegant if the regenerating ADH is also the production enzyme mediating an (enantio)selective reduction reaction. 4This 'substrate coupled' approach represents a biocatalytic version of the wellknown Meerwein-Ponndorf-Verley (MPV) reduction. 5owever, as is common amongst all MPV reductions, the reversibility and poor thermodynamic driving force of the reaction necessitates (unless elaborate coproduct removal is applied) significant molar surpluses of the cosubstrate.From an environmental point of view this is not desirable as the significant waste generated (Scheme 1) has to be dealt with.
Inspired by recent work of Lavandera et al. 6 we hypothesised that thermodynamically stable coproducts may represent a facile way to shift the equilibrium of ADH-catalysed MPV reductions and thereby reduce the molar surplus of the cosubstrate used.In that respect α,ω-diols such as 1,4-butanediol (1,4-BD) appeared most promising.First, 1,4-BD can be oxidised twice thereby doubling the yield of reducing equivalents liberated from the cosubstrate and, secondly, the resulting γ-butyrolactone (GBL) represents a thermodynamically stable and kinetically inert coproduct (Scheme 1).Hence, the waste generated for >95% conversion would be reduced approximately 40-fold (see ESI † for further details).Overall, a 'smart cosubstrate' approach solving the above-mentioned limitations was envisaged.
In a first set of experiments we screened a range of commercially available ADHs for activity towards 1,4-BD.Out of the 12 ADHs evaluated 8 showed significant activity (Fig. 1).Both NAD-and NADP-dependent ADHs showed significant activity towards 1,4-BD (Fig. 1).Amongst them, the well-known ADH from horse liver (HLADH) showed the highest activity and therefore was chosen as a biocatalyst for further investigation.Interestingly, HLADH has been known for decades to accept a broad range of 1,4-diols yielding enantiopure lactones but a 'smart cosubstrate' application of this reaction has not been proposed yet. 7he kinetic parameters of 1,4-BD (Fig. S1 †) as well as EtOH and i PrOH were determined (Table S1 †) showing that HLADH exhibits a reasonable apparent K M value of 23 mM towards 1,4-BD (together with a mild substrate inhibition, K i,app = 1.3 M).
Next we compared the performance of 1,4-BD to isopropanol ( i PrOH) as a sacrificial electron donor in the HLADHdriven reduction of cinnamaldehyde (Fig. 2).HLADH has been reported as a suitable enzyme for cinnamyl alcohol/cinnamaldehyde substrate/product coupling among the other enzyme preparations. 8ven when a 5-fold molar excess of i PrOH was applied, initial rate and maximal conversion fell back significantly behind the results obtained with 0.5 equiv. of 1,4-BD.Similar observations were also made using ethanol as a cosubstrate and/or using further substrates.† It is worth mentioning that using 1,4-BD as a cosubstrate always gave GBL in the expected 1 : 2 molar ratio to the product formed; the intermediate hydroxyaldehyde or its corresponding hemiacetal was not observed.
Unfortunately, cinnamaldehyde proved to be a poor model substrate due to significant product inhibition.† Already in the presence of 0.2 equiv. of alcohol (approx.corresponding to 20% conversion), the initial reduction rate decreased by more than 75% (Fig. S2 †).Nevertheless, 1,4-BD enabled significantly higher conversions than i PrOH or EtOH, which we attribute to the higher thermodynamic driving force exerted by the irreversible regeneration half-reaction.
To circumvent inhibition issues, we drew our attention to the reduction of α-arylpropionaldehydes (Profen aldehydes) enabling full conversion of the racemic starting material in a reductive dynamic kinetic resolution approach (RDKR). 8Obviously, product inhibition in this case is less pronounced.We therefore evaluated the 'smart cosubstrate' approach to promote the synthesis of enantiopure Profen alcohols.Gratifyingly, we found indeed that only 0.5 equiv. of 1,4-BD as a 'smart cosubstrate' was necessary to achieve complete conversion of the racemic starting material (Table 1).With comparable    amounts of EtOH or i PrOH, conversions of 24 and 14% were observed, respectively.These numbers are significantly lower than the expected equilibrium conversion assuming an equilibrium constant of 1.Most probably, thermodynamic reasons account for this as an MPV reduction between aldehydes and secondary alcohols should be overall thermodynamically uphill. 9This probably also is the reason for the huge molar excess of cosubstrate utilised in previous studies (Table 1). 8Interestingly, the optical purity of the final product decreased with increasing enzyme concentration (Table 1).In the presence of 10 times more biocatalyst the optical purity of the product dropped from 95% ee to 56% ee.We attribute this to the comparably slow in situ racemisation rate of the starting material under the given reaction conditions combined with an imperfect enantiodiscrimination of HLADH. 10 Hence, at high HLADH concentrations, the enzymatic reduction activity outperforms the racemisation rate leading to decreased optical purity of product.
Overall, the 'smart cosubstrate' approach appears to be a promising alternative to the established methods of substrate coupled biocatalytic reduction of alcohols.Further exploration of the scope is currently ongoing in our laboratories.
Encouraged by these promising results we also became interested in whether the 'smart cosubstrate' method might be generally applicable to promote NADH-dependent redox reactions such as (1) an enoate reductase-catalysed reduction of conjugated CvC-double bonds and (2) a monooxygenase-catalysed oxyfunctionalisation reaction.It is worth mentioning that these reactions are thermodynamically favoured and hence irreversible.As a model reaction for the reduction of conjugated CvC-double bonds we chose the enantioselective reduction of ketoisophorone using the enoate reductase from Thermus scotoductus SA-01 (TsER, Fig. 3). 11ompared to EtOH as a cosubstrate 1,4-BD gave almost 3 times higher initial rates thus comparing well with the initial rates obtained for the biocatalytic MPV reduction.The product (R)-levodione was produced in high optical purity (ee > 95%) in both cases but slowly racemised over time; 12 after 24 h the ee-value had dropped to 86% underlining the necessity for fast reaction kinetics as achieved with the 'smart cosubstrate' approach.Hence, we conclude that the 'smart cosubstrate' approach is advantageous in terms of rates and amount of waste product even for a thermodynamically favourable reaction such as the alcohol-oxidation promoted reduction of conjugated CvC-bonds. 13he second model reaction chosen was the regioselective hydroxylation of 3-hydroxybenzoate yielding 2,5-dihydroxybenzoate as catalysed by 3-hydroxybenzoate-6-hydroxylase (3HB6H) from Rhodococcus jostii RHA1 (Fig. 4). 14gain, 1,4-BD proved to be the most efficient cosubstrate enabling initial product formation rates 7.5 and 24.5 times higher than with ethanol and isopropanol, respectively.As a result, full conversion of the starting material was achieved in the time frame of the experiment only using 1,4-BD.Interestingly, conversions obtained using EtOH were somewhat higher than the expected maximal 50%.Possibly, an endogenous E. coli aldehyde dehydrogenase present in the commercial enzyme preparation may further oxidise the acetaldehyde coproduct thereby providing more NADH equivalents.Encouraged by these results, we further scaled up the reaction to the semi-preparative scale (7 g L −1 with full conversion into 2,5-dihydroxybenzoate within 8 h, Fig. S3 †).It should be mentioned here that the catalytic performance of the nicotinamide cofactor under the reaction conditions chosen here still is far from economic feasibility.However, we are convinced that after further optimization and upscaling, high total turnover numbers for the nicotinamide cofactors can be achieved.

Conclusions
The applicability of the 'smart cosubstrate' approach was demonstrated on range of NAD(P)H-dependent redox reactions ranging from enantioselective reduction of carbonyl-or CvC-bonds to specific aromatic hydroxylation.1,4-BD can be oxidised twice forming a thermodynamically and kinetically stable lactone coproduct.Especially compared to the 'traditional' substrate-coupled regeneration approach, the smart cosubstrate concept excels by a significant reduction in cosubstrate needed to achieve quantitative conversion, significantly reducing waste product formation.Furthermore, the coproduct (γ-butyrolactone) serves as an activated precursor for the synthesis of biodegradable polyesters. 168][19][20][21][22] Being intrinsically favourable reactions, the equilibrium issue here, if existing at all, is less pronounced.Nevertheless, apart from the electrochemical methods and hydrogenations, the 'smart cosubstrate' approach ranges amongst the least waste-intensive methods shown in Table 2.
We believe that the 'smart cosubstrate' approach is a robust and versatile concept to promote a broad range of NAD(P)Hdependent reactions.Ongoing research in our laboratories focuses on broadening the 'smart cosubstrate' scope (e.g.yielding enantiopure lactone products) and optimisation and application of this approach.

Experimental
HLADH, isoform E, recombinantly expressed in Escherichia coli is commercially available from evocatal GmbH (Düsseldorf, Germany).TsER was produced according to a literature procedure 11d by recombinant expression in E. coli followed by heatpurification.3HB6H was produced by recombinant expression in E. coli following a literature procedure. 14All chemicals were purchased from Sigma Aldrich (Zwijndrecht, The Netherlands) in the highest purity available and used as received.A detailed description of the experimental procedures as well as the analytical protocols can be obtained from the ESI.† The authors thank Deutsche Bundesstiftung Umwelt (DBU) for financial support of the project (AZ 13261).Serena Gargiulo, Remco van Oosten and Maarten Gorseling are acknowledged for fruitful discussions and technical support.Heidelberg, 2005, p. 225.  3 H.Chenault and G. Whitesides, Appl.Biochem.Biotechnol.,  1987, 14, 147-197.4 (a) M. Wolberg, M. V. Filho, S. Bode, P. Geilenkirchen, R. Feldmann, A. Liese, W. Hummel and M. Muller, Bioprocess Biosyst.Eng., 2008, 31, 183-191; (b)

Table 2
Comparison of the waste generated by various NAD(P)H regeneration systems ADH: alcohol dehydrogenase; AldDH: aldehyde dehydrogenase; PDH: phosphite dehydrogenase; GDH: glucose dehydrogenase.a Theoretical value according to the reaction equation.Depending on the way the electrical current was generated.c Double oxidation of the cosubstrate.d A phosphite buffer is transformed into a phosphate buffer.