2.1. Multi-Step Reactions Including Alcohol Dehydrogenases/BVMO Activity Catalyzed by Whole Cells

Most of the examples of biocatalytic cascades employing BVMOs have been developed together with alcohol dehydrogenases (ADHs, EC 1.1.1.x), also called ketoreductases (KREDs). These enzymes are able to catalyze the reversible reduction of carbonyl compounds into the corresponding alcohols (Scheme 2), in general with high selectivity ^{[46][47][48][49]}. ADHs require the presence of nicotinamide cofactors for performing their activity, being required to employ effective regeneration systems if these enzymes are employed as isolated biocatalysts.

Scheme 2. Alcohol dehydrogenases catalyzed reversible reduction of carbonyl compounds and oxidation of alcohols.

Initial studies on linear cascades combining ADHs and BVMOs were performed at the beginning of the 1990s, using whole cells systems for the direct bio-oxidation of alcohols to the corresponding lactones. Thus, the starting alcohol was oxidized to the corresponding ketones by the ADH at the expense of NAD(P)⁺, whereas the obtained ketone was further oxidized in presence of the BVMO to the desired lactone, employing NAD(P)H which was converted again into NAD(P)⁺. A pioneer example is the oxidation of alcohol 1, shown in Scheme 3, employing fractured cells of Acinetobacter calcoacetitus NCIMB to the corresponding lactone 3, described in 1991 ^[50]. At short reaction times, 1 was converted into ketone 2 in a great extent (80% after 6 h). Longer reaction time led to an increase in the lactone production, reaching a maximum value of 40% after 48 h, whereas concentration of 2 started to decrease after 10 h. A similar pattern was observed in the bio-oxidation of both endo- and exo-bicyclo [2.2.1]heptan-2-ols. Alcohol concentration decreased as the ketone became formed and lactone concentration started to increase during the course of the biocatalytic process. Fractured cells of A. calcoaceticus were also employed in the bio-oxidation of 6-endo-bicyclo [3.2.0]hept-2-en-6-ol 4 to obtain regioisomeric lactones 5 and 6 with complete conversion after 1 h, as shown in Scheme 3. A 15% of a by-product, which authors assumed that was exo-4, was observed in the crude mixture.

Scheme 3. Acinetobacter calcoaceticus cells catalyzed biotransformations of alcohols into lactones.

The combination of BVMOs and ADHs has been further developed for the preparation of ε-caprolactone (ECL, 9, Scheme 4). This compound is a valuable material with several applications in the field of colorants, adhesives and coating materials, as ECL can be easily polymerized yielding biodegradable thermoplastics and elastomeric polymers ^[51][52]. By these reasons, several approaches have been studied for its preparation under sustainable reaction conditions. BVMOs and ADHs have been employed well as whole cells or well as isolated enzymes for the synthesis of ECL starting from cyclohexanol 7, in a process in which this alcohol is oxidized by the ADH to cyclohexanone 8, which is converted into ECL in presence of the BVMO, with an in situ NAD(P)H cofactor regeneration. When employing this approach, the stability of the BVMO is critical for the process, as this biocatalyst is very sensitive to the cyclohexanol and ECL concentration in the reaction medium. For this reason, several attempts have been made in order to overcome this bottleneck.

Scheme 4. Preparation of ECL 9 and its oligomers through a biocatalytic procedure employing whole cells of Lactobacillus brevis ADH and Acinetobacter calcoaceticus CHMO.

Thus, the bienzymatic system employing whole cells of ADH-BVMO for the preparation of ECL can be coupled with the lipase-catalyzed hydrolysis of this compound in order to obtain oligo-caprolactone. This approach was performed for the first time in 2015 ^[53]. In order to overcome the BVMO deactivation caused by the presence of ECL in the reaction medium, this product subjected to in situ ring-opening oligomerization catalyzed by CALA (lipase from Candida antarctica, Scheme 4). The formed oligo-ECLs were easily removed from the reaction by extraction or precipitation. CALA is able to catalyze the formation of the polymers even in presence of high amounts of water, not showing any hydrolytic activity on ECL at high substrate concentrations (1.0 M). After a few hours in presence of CALA, the lactone was converted into oligomers with a maximum molecular weight of 1200 g/mol, which can be transformed into high-molecular weight polymer. Recombinant E. coli cells of Lactobacillus brevis ADH (LbADH) and the stable mutant C376L/M400I of Acinetobacter calcoaceticus cyclohexanone monooxygenase (AcCHMO), at 100 g wet cells weight/mL concentration, were combined with lipase CALA (10 mg/mL lyophilized) in a one-pot process at different concentrations of cyclohexanol. A decrease in the ECL concentration in these reactions was detected when compared to those in absence of the lipase, thus indicating the beneficiary effect of the conversion of the ECL into oligo-ECL. When both the ADH and the BVMO were employed as separate cells, results were better than expressing both biocatalysts into the same cell-system. LbADH/AcCHMO ratio was optimized 1:10 in order to achieve the highest conversion. Addition of acetone and glucose as co-substrates was also positive for a faster regeneration of the endogenous NADPH in the cell systems. The optimized cascade was performed at preparative scale and, after 48 h, a complete conversion was achieved when starting from 200 mM of 7, yielding a 75% of oligo-ECL and a 25% of ECL. Higher substrate concentrations, 300 mM and 500 mM, afforded lower conversions, 74% and 43% respectively.

In a further development of this trienzymatic system, the same authors described the preparation of methyl-substituted ECL derivatives in a one-pot two-step process ^[54], as shown in Scheme 5.

Scheme 5. Preparation of (S)-4-methylcaprolactone (S)-12 and its oligomers through a biocatalytic procedure employing whole cells of Lactobacillus kefir ADH and Acinetobacter calcoaceticus CHMO mutant QM.

A diastereomeric mixture of 4-methylcyclohexanol (10) was employed as starting substrate; under those conditions, AcCHMO was able to convert 4-methylcyclohexanone 11 into the enantiopure lactone (S)-12 with high yield and complete selectivity. Initial experiments were performed with E. coli cells expressing Lactobacillus kefir ADH (LkADH) and wild type AcCHMO in 1:10 ratio (100 g wet cells weight/mL), but the conversions at all concentrations tested (5–20 mM) were low, achieving only a 34% of the enantiopure lactone as highest value when working at 5 mM. Poor conversion values were caused by the low conversion in the ADH catalyzed oxidation, due to the inhibition of this biocatalyst by the ketone 11 formed. In order to improve this process, a faster second oxidation process, with the aim of the fast removing of 11 from the reaction medium was envisioned. Thus, a more stable mutant of AcCHMO was employed (C376L/M400I/T415C/A463C, the so-called AcCHMO QM) and the ratio of the E. coli cells for both biocatalysts was optimized to 1:1. The supply of pure oxygen to the reaction also resulted in a higher conversion, close to 90% after 24 h at 25 °C. The addition of CALA (10 mg/mL) to the enzymatic system led to a further decrease in (S)-12 concentration, as it was being hydrolyzed into the oligo-(S)-lactone.

The production of ECL starting from cyclohexanol employing E. coli cells of ADH and AcCHMO has been studied by using a computational approach in 2017 ^[55], applying a kinetic model for the study this cascade reaction in both batch and fed-batch synthesis. To this purpose, a fed-batch synthesis was developed in order to circumvent the CHMO inhibition caused by cyclohexanol, which hampered the use of this substrate at concentrations higher than 60 mM when employed in batch. Thus, 7 was added to the reaction medium at different ratios, from 14.2 to 19.0 μmol/min; feed rates higher than 17.5 μmol/min led to the accumulation of 7, therefore reducing the rate formation of ECL due to CHMO inhibition. The amount of ECL in the reaction medium was reduced by hydrolyzing this compound to 6-hydroxyhexanoic acid (13, Scheme 6) in a process catalyzed by Candida antarctica lipase B (CALB). After 6 h, 162 mM of 13 was obtained with a feed rate of 19 μmol/min. The addition of CALB maintained a low ECL concentration in the reaction medium (lower than 10 mM).

Scheme 6. Preparation of ECL 9 and its oligomers through a biocatalytic procedure employing whole cells of Lactobacillus brevis ADH and Acinetobacter calcoaceticus CHMO M15.

The production of ECL has been optimized by performing the co-expression of the BVMO and the ADH in E. coli, employing a Duet^TM vector ^[56], in order to obtain a higher efficiency for the enzymatic cascade. The pRSFDuet plasmid form Novagen was selected as dual expression vector, cloning the AcCHMO QM gene into the first multiple cloning site and the LkADH gene into the second multiple cloning site. With the aim of balancing the expression levels of the cascade enzymes, the ADH gene was subjected to engineering of the ribosome binding site (RBS), employing the native RBS sequence as well as two point mutations. In order to compare the results, both types of enzymes were expressed separately into E. coli cells, performing the biotransformations with an optimized ratio of CHMO: ADH cells 5:1 (100 g wet cells weight/mL) and in presence of glucose and acetone and co-substrates, as higher conversions were obtained than in absence of these compounds. This result can be explained by the difficult diffusion of the NADPH formed in the E. coli cells of the LkADH to the AcCHMO QM cells. After 16 h, it was possible to achieve a 90% conversion from 20 mM of cyclohexanol employing wild type RBS, while much lower conversions were measured with the other two RBS mutations, indicating that a decreased LkADH expression is required to obtain a similar enzyme ratio for optimal conversion. When the same process was carried out with the E. coli cells containing the double gene, a complete conversion was obtained after 16 h for all the three RBS preparations in the absence of co-substrates. In presence of glucose and acetone, complete conversion was also observed for the CHMO co-expressed with one of the RBS mutations of the ADH after 16 h, whereas after 2 h, a 54% conversion was reached. This system was the most efficient for the co-expression of both biocatalysts, suggesting that the ratios of expressed AcCHMO QM and LkADH were similar in this system. When higher substrate concentrations were employed, lower conversion values were achieved, mainly debt to the inhibitory effect of both 7 and 9 on AcCHMO.

Very recently, the conversion of cyclohexanol to ECL by combining LkADH with two stable mutants of CHMO has been studied in order to obtain a suitable process at 200 mM scale. Different parameters that affect the process were analyzed and optimized in order to achieve the most productive process for the preparation of the final lactone ^[57]. The reaction was carried out in a stirred-tank reactor with maintained temperature and pH, bubbling a mixture of oxygen with synthetic air and pumping cyclohexanol at a constant rate proportional to the whole cell cascade activity. As both 7 and 8 present a high vapor pressure, they can be stripped out of the system by the off gas. The key component of the complete system, as previously stated, is CHMO, due to its low stability and the requirement of a proper oxygen supply. This last parameter was optimized by performing a bubble aeration technique. BVMO stability was improved by assessing two CHMO mutants; thus, AcCHMO QM showed a 40% higher long-term stability, whereas AcCHMO M15, containing eight mutations, has a higher oxidative stability combined with an improved thermostability. These two variants were compared employing E. coli cells co-expressing ADH and the CHMO at pH 7.5 and 30 °C, with 200 mM cyclohexanol fed at 4.5 mM/h. When using AcCHMO QM, the formation of ECL achieved a value of 45 mM after 13 h, but at longer reaction times this concentration decreased to 39 mM, probably debt to the ECL autohydrolysis to yield 13. The same experiment was performed with AcCHMO M15, reaching a maximum concentration of ECL after 18 h (79 mM). Longer reactions times resulted in substrate accumulation in the reactor with no further formation of the desired product. AcCHMO M15 showed longer activity, but it was not possible to achieve complete conversion. The stability of both CHMO mutants under the process conditions was analyzed; when employing AcCHMO QM, the biocatalyst retained its complete activity for one hour and it decreased to 85% after 8 h. After this reaction time, there was an important drop in the biocatalyst activity. On the other hand, AcCHMO M15 maintained its activity (98% of the initial one) even after 18 h. Thus, when working with AcCHMO QM, fresh cells were added after 8 h, whereas no extra addition was required for AcCHMO M15. Further experiments showed that an increase in the cells loading in the reactor led to a decrease in the volumetric mass transfer, thus resulting in a lower efficiency. The complete optimization of the biocatalytic process allowed scientists to achieve conversions higher than 98% for both CHMO mutants, being possible to obtain a higher final product concentration (21.1 g/L), space time yield (1.1 g/L h) and isolated ECL amount (9.1 g) employing AcCHMO M15, due to its higher operational stability. These values represent a 100% increase regarding the non-optimized processes, indicating the development of a more efficient process for the preparation of ECL.

Fusion proteins have been initially employed to enhance the soluble expression of proteins or to improve the enzyme purification, but in the last years some examples have appeared in which these systems have been used with synthetic purposes ^[58]. In 2015, the preparation of an enzyme fusion formed by the ADH from Micrococcus luteus NCTC2665 and the BVMO from Pseudomonas putida KT2440 (3.0 g dry cells/L), able to catalyze the double oxidation of long-chain unsaturated secondary alcohols (14a–c) to the corresponding esters (16a–c) through the corresponding ketones 15a–c, was reported ^[59], as shown in Scheme 7. These products can be hydrolyzed into valuable ω-hydroxycarboxylic acids and n-alkanoic acids, valuable compounds. The design of the fusion enzyme showed the highest expression level when using a glycine-rich linker, formed by 12 aminoacids, between the two biocatalysts. When the bio-oxidations were carried out with the fused biocatalyst expressed in E. coli, higher conversions were achieved for all the substrates when compared with the independent ADH and BVMO expressed in E. coli. Thus, esters 16a–c can be recovered with conversions higher than 75%, with activities higher than 22 µmol/g dry cells min. This positive effect of the fusion enzyme in the cascade activity can be debt to a higher functional expression of the BVMO and/or to a better mass transport efficiency, due to the presence of both active centers at close positions.

Scheme 7. Preparation of ECL 9 and its oligomers through a biocatalytic procedure employing whole cells of Lactobacillus brevis ADH and Acinetobacter calcoaceticus CHMO M15.

2.2. Multi-Step Reactions Including ERED/BVMO Activity Catalyzed by Whole Cells

Ene-reductases (EREDs) from the old yellow enzyme (OYE) family are flavin-dependent enzymes that catalyze the chemo- and stereo-selective asymmetric reduction of electronically activated carbon–carbon double bonds ^{[25][27][60][61][62]}, as depicted in Scheme 8.

Scheme 8. Ene-reductases for the reduction of C=C.

Several examples can be found in literature illustrating the use of whole cells-catalyzed cascades coupling BVMO and ERED activities. For instance, Silva et al. ^[63] reported the preparation of ECL 9 starting from either cyclohexenone 17, cyclohexanone 8 or cyclohexanol 7, using whole cells of Brazilian Geotrichum candidum CCT 1205 (Scheme 9).

Scheme 9. Preparation of ε-caprolactone 9, using whole cells of Geotrichum candidum CCT 1205.

When using cyclohexanone 8 as substrate, these authors reported quantitative yields of the desired ε-caprolactone 9 after only 5 h, using a higher amount (3.0 g versus 1.0 g) of whole cells (otherwise, the cascade accumulated 7 and 8). Using 7 or 8 as substrates, the reaction was even faster (4 h) than using only 10 g of cells. These results are better than those previously reported by Mihovilovic et al. ^[64], leading to 52% of 9 in 48 h, using BVMO expressed in E. coli. Recently, Silva et al. ^[65] have shown that it is possible to immobilize the whole cells from Geotrichum candidum CCT 1205 in modified silica (n SiO₂–Cl, SiO₂–NH₂ and SiO₂–SH supports), without altering their catalytic performance and also allowing the stabilization and reuse of the cells.

Although the previous example illustrates the use of non-engineered cells, it is becoming more usual the use of genetically modified cells. Hence, Liu and Li described an enantioselective reduction−oxidation−hydrolysis cascade catalyzed by engineered whole cells for the synthesis of (R)-2-alkyl-δ-lactones 22–23 (useful flavor and fragrance materials) starting from the corresponding 2-alkylidenecyclopentanones 18–19 (Scheme 10) ^[66].

Scheme 10. Preparation of enantiopure (R)-2-alkyl-δ-lactones by using a whole-cell-catalyzed cascade involving BVMO activity.

In this example, the cascade catalysis was started with 50 mL of cell suspension of Acinetobacter sp. RS1 in Tris buffer (cell density 12 g cdw/L) containing 40 mg of alkylidene ketones 18 or 19 and 20 mg/mL glucose. After 3 h of reaction to allow the ERED-mediated accumulation of alkyl ketones 20 and 21 (R-configuration), 150 mL of Tris buffer containing engineered cells of E. coli (CHMO−GDH) (10 g cdw/L) was added to start the Baeyer–Villiger oxidation. After simultaneous oxidation and hydrolysis for 1.5 h, extraction with ethyl acetate and purification by flash chromatography, compounds (R)-22 (56% yield, 98% ee) or (R)-23 (41% yield, 97% ee) could be obtained. The presence of a hydrolytic activity inside Acinetobacter sp. RS1 cells was responsible for the hydrolysis of the intermediate (S)-22 or (S)-23 leading to δ-hydroxyacids 24 or 25, obtained with low enantioselectivity (E = 8–11).

The different enantiomers of carvolactones (7-methyl-4-(prop-1-en-2-yl)oxepan-2-ones) are gaining importance as monomers for the preparation of polymeric thermoplastic elastomers (shape-memory polymers) and pressure-sensitive adhesive components ^[67][68][69]. These carvolactones can be obtained starting from biogenic carvones after a C=C reduction and a Baeyer–Villiger oxidation. To this purpose, Iqbal et al. ^[70] have reported the concurrent redox cascade to transform (-) carvone and (+)-carvone [(R) and (S)-26] into the correspondent lactones, as depicted in Scheme 11.

Crude cell extracts (5.0 mg in 200 µL) from E. coli BL21 (DE3) containing the corresponding EREDs or BVMOs were used. Initially, each biotransformation step was carried out independently, starting with bioreductions of enantiomerically pure carvones (R) or (S)-26 and subsequent bio-oxidation with different BVMOs to furnish different stereoisomers of carvolactones 28 and 29. Finally, the reduction-oxidation steps were performed in a concurrent one-pot cascade. Both for the individual steps and the cascade, NADP⁺, glucose -6-phosphate and glucose -6-phosphate dehydrogenase were used for cofactor recycling. The EREDs that expressed E. coli BL21 (DE3) were SYE-4 (from Shewanella oneidensis ^[71]), OPR1 and OPR3 (12-oxophytodienoate reductase from Lycopersicon esculentum, tomato ^[72]), YqjM (from Bacillus subtilis ^[73]) and the variant W116I (replacement of tryptophan-116 by isoleucine in the OYE from Saccharomyces pastorianus ^[74]). On the other hand, the BVMOs were either cyclohexanone monooxygenases (CHMOs: AcCHMO ^[75], CHMO_Brevi1 from Brevibacterium sp. ^[76] or cyclopentanone monooxygenases (CPMOs-type enzymes: CHMO_Brevi2 from Brevibacterium sp. ^[76] and CPMO_Coma from Comamonas sp. ^[77]). Regarding the reaction stereochemistry of the bioreduction, as depicted in Scheme 5, when starting from (-)-carvone (R)-26 all the EREDs catalyzed the trans C=C bioreduction to afford (+)-trans-dihydrocarvone (2R,5R)-27. Opposite, if (+)-carvone (S)-26 was the initial substrate, most of the EREDs catalyzed the cis C=C bioreduction to afford (+)-cis-dihydrocarvone (2R,5S)-27, while the trans bioreduction, leading to (-)-trans-dihydrocarvone (2S,5S)-27, was observed only when employing W116I. Analyzing the regioselectivity of the BVMOs, depending on the absolute configuration of 27, some of the enzymes were leading to the “normal” lactones 28 (classical migration of the Criegge intermediate leading to the oxygen insertion in the most substituted alpha carbon ^[78]) or the “abnormal” ones 29, in good yields and enantiopurity.

In another example, Oberleitner et al. ^[79] reported the preparation of carvolactone (4S,7R)-28 starting from (R)-limonene ((R)-30) directly extracted from orange peel by means of a biocatalytic cascade shown in Scheme 12.

Scheme 12. Preparation of carvolactone (4S,7R)-28, using a whole cells-catalyzed cascade involving a BVMO.

To this purpose, two different types of cells were used: (i) an engineered Pseudomona putida S12 expressing cumene dioxygenase (CumDO) and (ii) E. coli BL21 (DE3) cells possessing different oxidative activities. These last ones were previously reported as being capable of catalyzing the conversion from carveol (1R,5S)-31 into (4S,7R)-28 ^[80], as they contain the expression of an alcohol dehydrogenase (RR-ADH from Rhodococcus ruber), an ERED (XenB from Pseudomonas sp.) and a BVMO (AcCHMO). Through the mixed culture approach, combining the two bacterial strains (different concentrations tested) in one pot, 47% of (4S,7R)-28 was produced after 20 h. They used a sequential approach, where hydroxylation of (R)-30 by CumDO was performed first, and E. coli BL21(DE3) resting cells were added to the reaction vessel after 10 h, observing full conversion to (4S,7R)-28 in 20 h. Then, these authors explored the direct use of waste product orange peel (biomass loading of about 3% (w/v), obtaining 3.2 mg of (4S,7R)-28 per g orange peel. Anyhow, by lowering the orange peel amount to 1.5% (w/v) and using the mixed-culture sequential combination, it was possible to furnish 6.3 mg of carvolactone per g orange peel (29% of (4S,7R)-28 from (R)-30 over four biocatalytic steps, 73% per step), only relying on orange peel as the substrate reservoir in aqueous buffer without any additives.

Recently, Avalos et al. ^[81] have described a similar process, using engineered E. coli cells, already designed to follow the Mentha spicata route leading to (-)-carvone (R)-26, in which the addition of a ERED and a BVMO (10 µM) activity finally furnished (4R,7R)-28, also known as (+)-dihydrocarvide (+)-DHCD, as shown in Scheme 13.

Scheme 13. Preparation of (+)-DHCD, using engineered whole-cells-catalyzed cascade.

The initial step into the M. spicata biosynthesis of (-)-carvone (R)-26 is the hydroxylation of (S)-limonene (S)-30 to furnish carveol (1S,5R)-31 catalyzed by limonene-6-hydroxylase (L6H, a P450 enzyme) coupled to SmCPR, a cytochrome P450 reductase from Salvia miltiorrhiza. Based on this system, authors constructed the rest of the enzymatic system by adding genes from ADH (RR-ADH was the best option), ERED (PETNR, pentaerythriol tetranitrate reductase from Enterobacter cloacae PB2 ^[82][83]) and BVMO (CHMO_3M, a triple mutant from Rhodococcus sp. Phi1 CHMO ^[84]). Under optimal conditions, the complete biosynthesis of (+)-DHCD from glucose in E. coli was reported at 6.6 mg/L.