Applications of Ene-Reductases in the Synthesis of Flavors and Fragrances

Abstract

Flavors and fragrances (F&F) are interesting organic compounds in chemistry. These compounds are widely used in the food, cosmetic, and medical industries. Enzymatic synthesis exhibits several advantages over natural extraction and chemical preparation, including a high yield, stable quality, mildness, and environmental friendliness. To date, many oxidoreductases and hydrolases have been used to biosynthesize F&F. Ene-reductases (ERs) are a class of biocatalysts that can catalyze the asymmetric reduction of α,β-unsaturated compounds and offer superior specificity and selectivity; therefore, ERs have been increasingly considered an ideal alternative to their chemical counterparts. This review summarizes the research progress on the use of ERs in F&F synthesis over the past 20 years, including the achievements of various scholars, the differences and similarities among the findings, and the discussions of future research trends related to ERs. We hope this review can inspire researchers to promote the development of biotechnology in the F&F industry.

1. Introduction

As early as a thousand years ago, people began to use fragrant plants for sacrifices and funerals. Today, flavors and fragrances (F&F) compounds are widely used in food, feed, tobacco, and cosmetics, among other fields; for example, irritant spices can remove or cover the bad smell of raw meat, milk flavor essence can increase the taste of food, and mint essence can provide mosquito repellent and refreshing properties.¹ According to statistics, the global market size of F&F in 2017–2022 continuously increased and reached approximately 39 billion euros in 2022, which was the highest value in nearly six years and an increase of 4% from 2021. Generally, the methods for F&F production can be divided into two types: natural extraction and chemical synthesis (Figure 1). Natural extraction methods are limited by climate, geography and processing factors, resulting in certain restrictions in terms of variety, quantity, and product quality. In chemical synthesis methods, the production scale can be easily controlled, the products exhibit good stability and the process is economic. However, chemical synthesis often results in environmental pollution and energy consumption. Moreover, individuals are more inclined to purchase products with natural flavors due to doubts and concerns about the safety of aromatic chemicals, even at relatively high prices.

Figure 1.

Methods for obtaining F&F.

Notably, F&F compounds obtained by physical methods (i.e., extraction and adsorption) or fermentation methods (i.e., plant tissue culture, microbial fermentation, and enzyme catalysis) can be considered natural flavors. (Figure 1) Therefore, various enzymes have been applied to produce F&F. For instance, lipase is used to synthesize ester flavors,²⁻⁴ while oxidase is mainly used to synthesize aldehyde and ketone flavors.⁵ On the one hand, enzymes have high catalytic efficiencies, mild reaction conditions, and few side reactions, achieving economic output at a relatively low environmental cost and reducing human dependence on natural resources. This reduction is key to solving major problems related to resources, energy, and security. On the other hand, enzymes have significant advantages in the synthesis of chiral F&F compounds, as different configurations of F&F compounds may provide significantly different olfactory experiences.^6,7 Multiple reviews on F&F biosynthesis have described this topic from different perspectives, such as the biosynthesis of monoterpenoids and sesquiterpenoids, the use of lipases for the synthesis of ester compounds, the biocatalytic oxidation of alcohols into aldehydes, and the enzymatic synthesis of unsaturated fatty acids.^3−5,8 However, almost no reviews have been written concerning the synthetic application of ERs in different F&F.

In the past, research on the synthetic applications of ERs was relatively scarce. However, due to its unique role in C=C reduction, an increasing number of scientists have recently attached importance to its application in asymmetric reduction and published a series of comprehensive studies on this topic. In this review, we summarize the research findings on the use of ERs in F&F synthesis over the past 20 years. According to the characteristics of the products, the results, differences and progress of ER-mediated F&F synthesis are discussed. Moreover, we also noted the limitations of ER applications and the key directions for future development.

2. Ene-Reductase

Ene-reductases (ERs) are versatile biocatalysts that can catalyze the asymmetric reduction of α,β-unsaturated aldehydes and ketones at the expense of nicotinamide cofactors. They have been increasingly considered an ideal alternative to their chemical counterparts for their excellent specificity and selectivity under mild pH and temperature conditions. To date, at least five known enzymes have been used to explore the biocatalytic applicability of C=C reduction, namely, nicotinamide adenine dinucleotide phosphate (NADPH)-dependent quinone-reductase-like ene-reductase (QnoR), enoate reductase (EnoR), old yellow enzyme (OYE), medium-chain dehydrogenase/reductase (MDR), and short-chain dehydrogenase/reductase (SDR)⁹ (Figure 2). Among these enzymes, the classic OYE family has been extensively studied, and its substrate range has extended from the initial α,β-unsaturated aldehydes and ketones to the α,β-unsaturated carboxylic acids, esters,¹⁰ nitriles,¹¹ and C≡C bonds.¹²

Figure 2.

Classifications of ERs.

The C=C hydrogenation of activated olefins containing electron-withdrawing groups catalyzed by ERs adopts the Ping-Pong Bi–Bi reaction mechanism (Figure 3). Net trans-addition of H₂ proceeds by initial reduction of the noncovalently bound FMN cofactor at N5 by NADPH. Hydride is delivered from N5 to the β-C of the alkene substrate on the side facing the reduced FMN with α-C protonation by the side chain of Tyr 196 on the opposite alkene side to complete the reaction.¹³ Reetz’s team confirmed the mechanism through quantum and molecular mechanics calculations.¹⁴

Figure 3.

Structures (OYE1, PDB: 3TX9) and catalytic mechanisms of ERs.

While ERs are best known for C=C reduction, they can be combined with photocatalysis to perform dehalogenation hydrogenation, carbonyl reduction, dehalogenation cyclization, bimolecular hydrogenation addition, and β–C-H functionalization reactions.¹⁵ Nevertheless, compared with cutting-edge and immature unnatural reactions, the natural reactions of ERs, including the synthetic application of pharmaceutical intermediates,¹⁶⁻¹⁸ polymers¹⁹ and especially F&F, are more widely used in industry.

3. Terpenoid Flavor

3.1. Citronellal and Citronellol

Citronellal, i.e., 3,7-dimethyl-6-octenal, is a primary component of plant essential oils (such as eucalyptus oil and citronella oil) and has strong lemon, citronella, and rose scents.²⁰ This molecule contains a chiral carbon atom and thus has (R)- and (S)-configuration. Notably, (R)-citronellal has higher commercial value as it is an important intermediate in the synthesis of l-menthol,²¹ one of the largest spice chemicals in the world, with a value increase of approx 1:20.²²

Asymmetric hydrogenation of citral is the main method for synthesizing optically pure citronellal. (Figure 4) Citral is a fragrant chemical with a strong lemon aroma, containing two stereoisomers: cis [(E)-citral: geranial] and trans [(Z)-citral: neral]. Due to the presence of two carbon–carbon double bonds and one carbon–oxygen double bond within citral molecules, the asymmetric hydrogenation of citral faces challenges in terms of chemical, regio, and stereo selectivity. BASF presented enantioselective reduction of (E)-citral to synthesize (R)-citral, which requires obtaining high-purity (E)-citral in advance from (E/Z)-citral (a mixture of cis and trans isomer).²³ A dual-catalyst system comprising enantiopure 2-diarylmethylpyrrolidines and heterogeneous Pd/BaSO4 was developed for the asymmetric hydrogenation of citral to afford (R)-citronellal with insufficient enantioselectivity [89% enantiomeric excess (ee)].²⁴ Therefore, ERs-mediated catalysis known for its high selectivity becomes a promising alternative to classical organic catalysis (Table 1).

Figure 4.

Synthesis of citronellal and citronellol.

Table 1. Application of ERs in the Synthesis of Citronellal and Citronellol.

entry	whole cell or ene reductase	performance	reference
1	Zymomonas mobiliz, Citrobacter freundii, Candida rugosa, Saccharomyces bayanus, and Saccharomyces cerevisiae	(R), 98% ee	(25)
		(S), >99% ee
2	Candida parapsilosis DSM 70125	(R), >95% ee	(26)
3	OYE1 from Saccharomyces pastorianus, OYE2 from Saccharomyces cerevisiae, OYE3 from Saccharomyces cerevisiae, and NCR from Zymomonas mobiliz	(R), >95% ee	(27)
		(S), >99% ee
4	OYE1, OYE2, OYE, and NCR	(R), 49% conv., 99% ee	(28)
		(S), >99% conv., >95% ee
5	OPR1 and OPR3 from Lycopersion esculentum (tomato)	(S), 95% conv., >95% ee	(30 and 31)
6	PgOPR1–3 from Pelargonium graveolens	(R), ∼95% ee	(32)
7	YqjM from Bacillus subtilis	(S), 70% conv., >95% ee	(31)
8	TOYE from Thermoanaerobacter pseudethanolicus E39	(S), 23% yield, 91% ee	(35)
9	YqiG from Bacillus subtilis str. 168	(S), 12% conv., >99% ee	(36)
10	PETNR from Enterobacter cloacae st. PB2	(S), 56% yield, 87% ee	(37)
11	KYE1 from Kluyveromyces lactis	(R), 68% conv., 86% ee	(40 and 41)
12	PaER from Pichia angusta	(R), >99% conv., 25% ee	(42)
13	EBP1 from Candida albicans	(R), 34% conv., 53% ee	(39)
14	Yers-ER from Yersinia bercovieri	(S), 96% conv., >99% ee	(41)
15	EnR from Gluconobacter oxidans	(S), >99% conv., >99% ee	(43)
16	Ppo-Er1 from Paenibacillus polymyxa	(S), 29 ± 1.4% conv., 94% ee	(44)
17	NEMR from Escherichia coli	(S), 91% conv., >66% ee	(39)
18	MorR from Pseudomonas putida	(S), 55% conv., 41% ee	(39)
19	OYE1 → OYE1 W116F	(S), 38% conv., 19% ee	(13)
		→
		(R), 97% conv., 65% ee
20	XenA → XenA C25G	(R), 8.1% conv., 99.5% ee	(47)
		→
		(R), 33.5% conv., 99.5% ee
21	NCR → NCR W66A/F269A	(S), >99% ee → (R), 56% ee	(48)
22	CgrAlcOx + OYE2 or GluER from Gluconobacter oxidans	(R), 95.1% conv., 95.9% ee	(49)
		(S), 95.3% conv., 99.2% ee
23	OYE2.6 from Pichia pastoris or NemA from Escherichia coli	(R), 67% conv., 98% ee	(52)
		(S), 69% conv., 99% ee
24	OYE2p from Saccharomyces cerevisiae YJM1341	(R), 87.2% yield, 88.8% ee	(21)
25	NCR + ADH-H from Rhodococcus ruber	95% yield	(56)
26	NemR-PS from Providencia stuartii + YsADH from Yokenella sp. WZY002	(S), >99 conv.	(57)

Due to the complexity of the cofactor cycle, researchers mostly use whole cells for biotransformation in the initial stage of utilizing ERs. A series of bacteria, fungi, and yeast were screened for enantio-specific reduction of citral to produce citronellal. Zymomonas mobiliz and Citrobacter freundii produce (S)-citronellal with >99/75% ee, while fungi exhibited opposite stereoselectivity, with Candida rugosa/Saccharomyces bayanus/Saccharomyces cerevisiae yielding 98/97/96% ee (R)-citronellal.²⁵Candida parapsilosis DSM 70125 also exhibits excellent stereoselectivity, providing >95% ee (R)-citronellal.²⁶ Although the stereoselectivities were achieved with whole cells, chemoselectivities for reducing the C=C versus C=O bond were partially concealed by competing prim-alcohol dehydrogenases, which led to the undesired reduction of the carbonyl group. In order to overcome these limitations, Müller expressed four types of ene reductases (OYE1 from Saccharomyces pastorianus, OYE2 and OYE3 from Saccharomyces cerevisiae and NCR from Zymomonas mobiliz) in recombinant Escherichia coli.²⁷ (E/Z)-citral was stereoselectively reduced by NCR to generate (S)-citronellal, resulting in an 8-fold increase in yield (∼45% in this study versus ∼6% in previous research²⁵) and using fewer whole cells (1.5 g/L versus 14 g/L dry cell mass). OYE1 and OYE2 stereoselectively reduced (E)-citral to (R)-citronellal, while for (Z)-citral, racemates were obtained. OYE 3 showed mixed characteristics: it was similar to NCR in the conversion of (Z)-citral and similar to OYE 1 and 2 in the conversion of (E)-citral. Hall published similar work using purified OYE1–3 and NCR and further investigated the effects of different cofactors and cycling systems on yield and stereoselectivity.²⁸ Surprisingly, different cofactors and cyclic systems resulted in a reversal of product stereoselectivity. For example, OYE3 produced (R)-citronellal in moderate stereoselectivities (42% ee) with the NAD⁺/GDH system, whereas the NADP⁺/G6PDH system gave the opposite enantiomer 49% ee (S)-citronellal. This variation can be explained by the isomerization of (E/Z)-citral²⁹ and the reduction preference of OYE3 for (E)-citral.²⁷ The relative rate of competing C=C reduction (fast) of (E)-citral versus (E/Z)-isomerization (slow) determines the overall outcome of the process. Notably, purified ERs reduced the α,β-unsaturated C=C moiety in a highly selective fashion, while the nonactivated C=C bond and the aldehyde moiety remained untouched, in contrast to whole-cell bioreduction. Therefore, subsequent research often used pure or crude enzymes with high target protein content.

Since ERs have demonstrated their chiral synthesis potential in the biotransformation of citronellal, researchers identified a series of ERs from different sources and tested their reduction ability toward citral. ERs in plants were first discovered from Lycopersion esculentum (tomato), known as 12-oxophydioate reductase (OPR). OPR1 and OPR3 from Lycopersion esculentum (tomato) reduced (E)-citral to give (S)-citronellal in greater than 95% ee.^30,31 PgOPR1–3 from Pelargonium graveolens showed obvious preference for (Z)-citral reduction, similar to NCR,²⁷ and produced (S)-citronellal and (R)-citronellal at a molar ratio of ∼95:5.³² In addition, a multisubstrate reductase from Plantago major belonging to the SDR superfamily can also catalyze the reduction of citral, but the stereoselectivity of the product has not been detected.³³ YqjM from Bacillus subtilis, the most prominent member of OYE family members from thermophilic microorganisms, reduced (E/Z)-citral to yield (S)-citronellal in >95% ee.³¹ However, the enzyme kinetic determination indicated that aliphatic enals are moderate substrates of YqjM with low k_cat/k_m values (0.02 mM^–1 s^–1 for citral).³⁴ A thermophilic “ene” reductase TOYE isolated from Thermoanaerobacter pseudethanolicus E39 reduced (E/Z)-citral to 91% ee (S)-citronellal with low yield (23%) after 24 h of reaction.³⁵ YqiG from Bacillus subtilis str. 168, most related to XenA (an atypical thermoplastic like OYE from Pseudomonas putida) with 37.9% identity, catalyzed (E/Z)-citral to (S)-citronellal with a low conversion (12%) but high ee value (>99%).³⁶ PETNR from the anaerobic microorganism Enterobacter cloacae st. PB2 was first isolated via its ability to degrade a variety of high explosives such as trinitrotoluene (TNT), while most other ERs cannot.³⁷ It also reduced citral (E/Z = 65/35) to obtain (S)-citronellal in moderate yield (56%) and ee value (87%).^38,39 Further, (R)-citronellal can be prepared using KYE1 from Kluyveromyces lactis (68% conversion and 86% ee),^40,41 PaER from Pichia angusta (>99% conversion and 25% ee)⁴² and EBP1 from Candida albicans (34% conversion and 53% ee).³⁹ On the contrary, (S)-citronellal can be synthesized using Yers-ER from Yersinia bercovieri (96% conversion and >99% ee),⁴¹ EnR from Gluconobacter oxidans (>99% conversion and >99% ee),⁴³ Ppo-Er1 from Paenibacillus polymyxa (29 ± 1.4% conversion and 94% ee),⁴⁴ NEMR from Escherichia coli (91% conversion and 66% ee)³⁹ and MorR from Pseudomonas putida (55% conversion and 41% ee).³⁹ GYE1 and GYE2 from Gluconobacter oxidans M5 specifically reduced (Z)-citral without reducing (E)-citral and the stereoselectivity of the products was not investigated.⁴⁵

In addition to identifying new ERs for citral reduction, using protein modification engineering with strong purposiveness to construct mutants is also a means to obtain high-yield and high-enantioselectivity biocatalysts. The Stewart team discovered the key amino acid W116 in OYE1 that controls stereoselective reduction.¹³ The mutant W116F (65% ee (R)-citronella) exhibited opposite stereoselectivity to the wild-type (19% ee (S)-citronella) when catalyzing the reduction of (Z)-citral, and the conversion was significantly improved (38 → 97% conversion), while for (E)-citral, it remained the same as the wild-type. Moreover, XenA increased the yield of (R)-citronellal by four times (8.1 to 33.5%) through genetic mutations, while maintaining high enantioselectivity (99.5% ee). Inspired by previous research,⁴⁶ Bommarius’s team mutated the key amino acid C25 in XenA, resulting in a C25G mutant that increased the conversion by four times (8.1 → 33.5%) and maintained high enantioselectivity of the product (99.5% ee (R)-citronella).⁴⁷ Kress replaced amino acids of NCR within 5 Å around the docked citral with smaller (glycine or alanine) and larger (phenylalanine or tryptophan) hydrophobic residues and found that W66 was the key amino acid for stereoselective reduction of (E)-citral.⁴⁸ Combining iterative saturation mutagenesis, mutant W66A/F269A was constructed to flip the stereo configuration of the product from >99% ee (S)-citronellal to 56% ee (R)-citronellal while maintaining the catalytic activity of the wild-type. It is worth noting that high conversion does not necessarily equate to high yield, and substrate degradation and undetectable byproducts may also lead to complete substrate conversion. Therefore, we cannot definitively conclude that we have found an ideal ER that can obtain a single configuration of citronellal from mixed citral. However, it has been found that starting from a single configuration of citral can obtain products with better enantioselectivity. Therefore, Ribeaucourt proposed directly generating a single configuration of citral for in situ reduction, different from the separation of a single configuration from mixed citral.⁴⁹ CgrAlcOx (a copper radical oxidase from Colletotrichum graminicola M1.001) was used to provide (E)-citral, which was then reduced to (R)-citronellal (95.1% yield and 95.9% ee) and (S)-citronellal (95.3% yield and 99.2% ee) by OYE2 and GluER (from Gluconobacter oxidans), respectively. This one-pot bienzymatic cascade overcomes the above problem of mixed (E/Z)-citral reduction and produces the alternative enantiomer by switching the ERs, which provides a new idea of dual enzyme cascade for the stereoselective synthesis of citral.

In previous studies, whole cells and purified ERs have been tested as catalysts for citral reduction to citronellal on small scales with excellent conversion and stereoselectivity. To further expand the possibility of its industrial application, one of the key factors is the tolerance of enzymes to high substrate concentrations. 5–20 mM is a commonly used substrate concentration for ERs catalytic reduction,^{25−27,30,31,35,41,43,47} which falls far short of the high-concentration enzyme catalysis reported in literature (>1 M substrate).^50,51 Bougioukou used a pure (>98%) geranial feed strategy (i.e., detecting substrate consumption and supplementing substrates promptly through coupled glucose oxidation) at a substrate concentration of 150 mM, and 5.75 h reaction time to yield 67% (R)-citronellal (98% ee).⁵² The author attempted to prepare OYE2.6 (from Pichia pastoris) cross-linking enzyme aggregates (CLEAs) to stabilize biocatalysts, but the catalytic activity was significantly reduced. Therefore, crude extracts containing OYE 2.6 treated with ammonium sulfate were used for catalysis. Starting from neral, the strategy was also used to obtain (S)-citronellal in 69% yield with 99% ee after a 4 h reaction, using cells overexpressing NemA (from Escherichia coli). Zheng used purified OYE2p (from Saccharomyces cerevisiae YJM1341) to prepare 174.3 mM (R)-citronellal from 200 mM (E/Z)-citral in 87.2% yield and 88.8% ee after 52 h reaction, which is currently the highest known substrate concentration for synthesizing citral catalyzed by ERs.²¹

In fact, citronellol is also a valuable perfume compound and flavoring agent, presenting a strong rose-like odor, with a low threshold of 50 parts per billion. The odor characteristics of (S)-citronellol are considered superior to (R)-citronellol. (S)-citronellol exhibits a fresh, light, and clean rose petal-like odor, while the aroma of (R)-citronellol is certainly similar to the (S)-isomer, but somewhat greasy and irritating.⁵³ At the same time, (S)-citronellol serves as an important intermediate for the synthesis of optically active odor molecules, e.g., (−)-cis-rose oxide⁵⁴ (Figure 4). However, due to the inertness of the allyl alcohol double bond, geraniol or nerol are almost not included in the catalytic substrate range of ERs. So far, only one work has reported that HbOPR from rubber tree and OYE2 catalyze the reduction of geraniol to citronellol with unclear yield and stereoselectivity of the products.⁵⁵ The dual enzyme system combining ERs-mediated double bond reduction and alcohol dehydrogenases (ADHs)-mediated carbonyl reduction is a common method for synthesizing citronellol. For example, Tauber constructed a tandem system combining ADH-A (from Rhodococcus ruber) and NCR to obtain >95% citronellol (ee value was not determined) from (E/Z)-citral, where ADH-A plays a role in both reducing carbonyl groups and cofactor cycling.⁵⁶ Jia used Escherichia coli coexpressing NemR-PS (from Providencia stuartii), YsADH (from Yokenella sp. WZY002), and BmGDH_M6 (from Bacillus megaterium) to catalyze the biotransformation of (E/Z)-citral to (S)-citronellol.⁵⁷ Due to limited cell capacity, the expression level of BmGDH_M6 is reduced in coexpressed bacteria, resulting in insufficient activity to meet the requirements of two reductases. After 24 h of reaction, feeding additional wet cells expressing BmGDH_M6 achieved a complete substrate conversion of up to 400 mM.

3.2. Dihydrocarvone and Dihydrocarveol

Dihydrocarvone, also known as 2-methyl-5-(prop-1-en-2-yl)cyclohex-2-en-1-one, mainly exists in caraway (Carum carvi) and dill (Anethum graveolens) seed oil, emitting a herbaceous and spearmint-like odor.^58,59 Meanwhile, dihydrocarvone is also an effective inhibitor of bacteria and filamentous fungi, a natural insect repellent, and a chiral raw material, which plays an important role in the synthesis of complex chiral natural products (e.g, striatenic acid, pechueloic acid).⁵⁸ Owing to the existence of two stereogenic centers in the structure, dihydrocarvone has four stereoisomers, namely, (2R,5R), (2R,5S), (2S,5R), and (2S,5S) (Figure 5). In previous studies, metal catalysis,⁶⁰⁻⁶² acid catalysis,⁶³⁻⁶⁵ and electrooxidation⁶⁶ synthesized dihydrocarvone starting from carvone or limonene without considering the stereoselectivity of the product.

Figure 5.

Synthesis of dihydrocarvone and dihydrocarveol.

Biocatalysis typically displays specific stereoreferences, therefore multiple studies have reported the use of whole-cell or isolated enzyme catalysts for dihydrocarvone synthesis (Table 2). In the early stages, fungi (such as yeast and mushrooms),⁶⁷⁻⁷² plant cells (such as moss and marine microalgae)^73,74 and bacteria⁷⁵ were carried out for the biotransformation from (5R)- or (5S)-carvone to dihydro products. However, the conversion and stereoselectivity show significant differences due to different biocatalysts. For example, Hanseniaspora guilliermondii CSIR Y894 T catalyzed the reduction of (5R)-carvone to (2R,5R)-dihydrocarvone (>95% conversion after 80 min reaction), while Geotrichum candidum G 400 and Yarrowia lipolytica KBP 3364 mainly yielded (2R,5R)-dihydrocarvone, which was rapidly reduced to (1S,2R,5R)-dihydrocarveol by endogenous cellular CRs (>99% conversion after 75 min reaction). Notably, some whole-cell catalysts did not exhibit interference from endogenous CRs and achieved high conversion and high stereoselectivity in a short reaction time, making bioreduction attractive from an industrial perspective. Recombinant E. coli or isolated ERs typically have higher target protein content, resulting in lower background interference and higher conversion. So far, various new ERs have been identified to exhibit the ability to reduce carvone, including KYE 1,⁴⁰ Yers-ER,⁴⁰ Gox0502 (from Gluconobacter oxidans 621H),⁷⁶ Gox2684 (from Gluconobacter oxidans 621H),⁷⁶ Dbr2 (from Artemisia annua),⁷⁷ PETNR,^38,39 TOYE,³⁵ LacER (from Lactobacillus casei str. Zhang),⁷⁸ SYE-4 (from Shewanella oneidensis),⁷⁹ XenA (from Pseudomonas putida ATCC 17453),⁸⁰ XenB (from Pseudomonas putida ATCC 17453),⁸⁰ NemA (from Pseudomonas putida ATCC 17453),⁸⁰ YqiG,³⁶ OYE2p,²¹ Ppo-Er1,⁴⁴ and PaER (from Pichia angusta).⁴²

Table 2. Application of ERs in the Synthesis of Dihydrocarvone and Dihydrocarveol.

entry	whole cell or ene reductase	performance	reference
1	yeasts belonged to the genera Arxula, Brettanomyces, Bullera, Candida, Debaryomyces, Dekkera, Eremothecium, Exophiala, Geotrichum, Hanseniaspora, Hormonema, Kloeckera, Kluyveromyces, Lipomyces	(2R,5R), >99% conv.	(67)
	Metschnikowia, Pachytichospora, Pichia, Rhodotorula, Saccharomyces, Schwanniomyces, Sporodiobolus, Torulaspora, Trichosporon, Yarrowia, Zygozyma	(2R,5S), 50–60% conv.
2	Diplogelasinospora groovesii, G. butleri, and S. octosporus	(2R,5R), <10% yield	(68)
		(2R,5S), 50–60% yield
3	yeasts belonged to the genera Candida, Cryptococcus, Hanseniaspora, Kluyveromyces, Pichia, and Saccharomyces	(2R,5S), 39.74 ± 2.60% yield	(59)
		(2S,5S), 44.36 ± 5.44% yield
4	Marchantia polymorpha, Marchantia plicata, and Riccia fluitans	(2R,5R), 28% yield	(73)
5	Chlorella minutissima, Nannochloris atomus, Dunaliella parva, Porphyridium purpureum, and Isochrysis galbana	(2R,5S), 53% yield	(74)
6	Pseudomonas proteolytica FM18Mci1 and Bacillus sp. FM18civ1	(2R,5S), 81 ± 6% yield	(75)
		(2R,5R), 94 ± 7% yield
7	KYE 1and Yers-ER	0.73 and 2.54 U/mg	(40)
8	Gox0502 and Gox2684 from Gluconobacter oxidans 621H	8.0 and 2.5 U/mg	(76)
9	Dbr2 (from Artemisia annua)	(2R,5S), 83% de	(77)
10	PETNR	(2R,5R), 95% yield, 88% de	(38 and 39)
		(2R,5S), 82% yield, 95% de
11	TOYE	(2R,5R), 61% yield, 95% de	(35)
		(2R,5S), 77% yield, 85% de
12	LacER (from Lactobacillus casei str. Zhang)	(2R,5R), 99% conv., 98% de	(78)
13	SYE-4 (from Shewanella oneidensis)	(2R,5R), 59% isolated yield, 97% de	(79)
		(2R,5S), 53% isolated yield, 95% de
14	XenA from Pseudomonas putida ATCC 17453	(2R,5R), 73% conv., >99% de	(80)
		(2R,5S), 62% conv., >99% de
15	XenB from Pseudomonas putida ATCC 17453	(2R,5R), 75% conv., >99% de	(80)
		(2R,5S), 60% conv., >99% de
16	YqiG	(2R,5R), 21% conv., 86.3% de	(36)
17	OYE2p	(2R,5R), 74% de	(21)
		(2R,5S), 98% de
18	Ppo-Er1	(2R,5R), >99 ± 2.1% conv., 98% de	(44)
19	PaER (from Pichia angusta)	(2R,5R), 68% de	(42)
		(2R,5S), 72% de
20	Absidia glauca ATCC 22752, Cunninghamella echinulata ATCC 9244, Penicillium claviforme MR 376, Pseudomonas putida NRRL-13	(2R,5R), 81.57% yield	(81)
		(2R,5S), 87.67% yield
21	Pseudomonas putida and Acinetobacter lwoffi	(2R,5R), 95% conv.	(82)
		(2R,5S), 84% conv.
22	nonconventional yeasts (NCYs) belonged to the genera Candida, Cryptococcus, Debaryomyces, Hanseniaspora, Kazachstania, Kluyveromyces, Lindnera, Nakaseomyces, Vanderwaltozyma, and Wickerhamomyces	(2S,5R), 62% yield, 95% de	(84)
23	RmER → RmER C25D/I66T	93 → 96% conv.	(88)
24	TsER → TsER C25D/I67T	(2R,5S) → (2S,5S), >90% de	(88)
25	OYE1 → OYE1 F296S/W116A	(2R,5R), >99% conv., >99% de	(89)
		(2R,5S), 59% conv., 83% de
		→
		(2S,5R), 3% conv., 28% de
		(2S,5S), 27% conv., 96% de
26	OYE1 → OYE1 W116I	(2R,5S), 48% conv., 93% de	(13)
		→
		(2S,5S), >98% conv., 88% de
27	OYE1 → OYE1 W116A or W116V	(2R,5R), >90% conv., >95% de	(90)
		→
		(2S,5R), >70% conv., >50% de
28	OYE3 → OYE3 W116A	(2R,5R), >75% de → (2S,5R), 8% de	(91)
29	LacER + CMCR (from Candida magnolia)	(1S,2R,5R), 93% conv., >99% de	(95)
30	LacER + SSCR (from Sporobolomyces salmonicolor)	(1S,2R,5S), 87% conv., 99% de	(95)
31	PENTR + LfSDR1 V186W	(1S,2R,5R), 48% isolated yield	(96)
		93% de
		(1S,2R,5S), 40% isolated yield
		92% de
32	OYE1W116A + LfSDR1 V186W	(1S,2S,5S), 52% isolated yield	(96)
		91% de

The introduction of emerging technologies (such as headspace-solid phase microextraction, In situ substrate feeding and product removal and protein modification engineering) often brings new vitality to the biotransformation catalyzed by ERs. Solid-phase microextraction (SPME) is based on the use of fused silica fiber coated with stationary phase to absorb and enrich the analyte in the sample, which can concentrate the analyte while extracting.⁸¹ Demirci and his colleagues introduced headspace SPME in conjunction with GC/MS analysis into the detection of volatile metabolites (such as dihydrocarvone, dihydrocarveol and carveol) formed by microbial transformation of carvone.⁸² Subsequently, Cramarossa⁸³ and Goretti⁸⁴ also utilized this technique to compare the activity and stereoselectivity of bacteria and nonconventional yeasts (NCYs) in catalyzing (5R)- and (5S)-carvone. In situ substrate feeding and product removal (SFPR) refers to the use of adsorbent resins to simultaneously supply substrates and remove products.⁸⁵ Weister-Potz’s team introduced in situ SFPR strategies to form biphasic reaction modes (using hydrophobic adsorbent resin XAD4) to reduce the toxicity of organic compounds in the aqueous phase to biocatalysts.⁸⁶ They achieved the conversion of 300 mM (5R)-carvone to 290.4 mM (2R,5R)-dihydrocarvone (96.5% diastereomeric excess (de), 32.3 mM L^–1 h^–1 space-time yield) after 9 h reaction time by coexpressing FDH_3M (a triple mutant of formate dehydrogenase from Mycobacterium vaccae) required for cofactor cycling and NostocER1 (from the cyanobacterium Nostoc sp. PCC 7120) in recombinant E. coli. Furthermore, their group utilized protein modification engineering to alter the cofactor preference of NostocER1 from NADPH to NADH, as the intracellular concentration of NAD(H) is up to 20-fold higher than the concentration of NADP(H) in exponentially growing E. coli cells.⁸⁷ The modification idea is to exchange loop regions of NostocER1 that participate in cosubstrate binding for the respective regions of two native donor OYEs with high activity or affinity for NADH. E. coli coexpressing NostocER1 mutant (loop 1,5,2a) and FDH_3M achieved a conversion of 99.4% within 6 h under the same conditions as previous work and the spatiotemporal yield increased by 1.8 times (57.4 mM L^–1 h^–1). A similar “scaffold sampling” strategy has also been used for mutations in other ERs, resulting in excellent mutants with better catalytic activity or stereoselective reversal.^88,89 Stewart’s team mentioned earlier also used OYE1 mutant W116F and W116I for carvone reduction.¹³ Among them, W116I achieved a stereoconfiguration reversal and conversion increase of (5S)-carvone reduction product, from 48% (2R,5S)-dihydrocarvone (93% de) to >98% (2S,5S)-dihydrocarvone (88% de). The docking result shows that replacing Trp 116 with Ile has opened additional active site space that allows the isopropenyl substituent of (5S)-carvone to extend into the area previously occupied by the indole ring of Trp 116, which caused “flipped” substrate binding orientation rather than “normal” substrate binding orientation. Subsequently, this group demonstrated the two substrate binding orientations using X-ray crystallography.⁹⁰ The data showed that when W116I crystallized with (5S)-carvone, the electron density of the substrate’s isopropenyl moiety was mainly concentrated near the side-chain of Ile, while when using wild-type crystallization, the electron density was very close to the side-chain of Phe 296, consistent with previous conjectures. Based on previous research, this team continued to investigate whether the mutation at the W116 position of OYE3 affects the stereoselectivity of the product, as OYE3 shares 80% sequence identity with OYE1.⁹¹ Only OYE3-W116A achieved stereoselective reversal of (5S)-carvone reduction but many OYE1-W116 mutants achieved stereoselective reversal of both (5R)- and (5S)-carvone, indicating that OYE 3 has a strong preference for carvone binding orientation, which cannot be overridden by changes to Trp 116. Lutz’s team applied a new protein engineering strategy called circular permutation (CP) to enhance the function of OYE1.⁹² The principle is to covalently connect the protein’s original amino and carboxyl termini through a peptide linker. Then the new termini is introduced to other parts of the protein structure through peptide bond cleavage. A series of CP mutants were identified and the catalytic performance for (5S)-carvone reduction was increased up to 13-fold. Further mechanistic studies have shown that loop β6 plays an important role in exposing active sites, which forms a lid that isolates FMN and substrate in the active site from the environment.⁹³ The substitution of loop β6 with a tether increased the environmental exposure of the active site, resulting in faster exchange of substrates and products without affecting catalytic activity. In addition, CP strategy combined with traditional random mutations showed additive or synergistic effects.

Dihydrocarveol is a valuable perfume ingredient with notes of spearmint and pepper, which is currently used in cosmetics, perfume, and household products, such as detergents and detergents.⁹⁴ It is a more complex chiral compound with three chiral centers and eight stereoisomers, making it difficult to precisely control stereoselectivity through chemical methods (Figure 5). Researchers have prepared eight optically pure stereoisomers of dihydrocarveol by concatenating ERs (LacER,⁹⁵ PETNR,⁹⁶ OYE1W116A⁹⁶) and CRs, with 40–93% yield and 91 → 99% de. However, research on dihydrovarveol preparation is limited as it is often seen as a byproduct of dihydrocarvone synthesis (Table 2).

3.3. Isopulegon and Isopulegol

(−)-Isopulegol is an important intermediate in producing (−)-menthol,^97,98 marketed as Coolact P for its notable cooling sensation⁹⁹ (Figure 6). The ene cyclization of (+)-citronellal to (−)-isopulegol catalyzed by Lewis acids proceeds in high yields and selectivities.^97,100,101 However, the major drawback of Lewis acid catalysts is the environmentally unfriendly processes. Scrutton utilized IPR (from Mentha piperita) to catalyze the biotransformation of isopiperitenone into cis-isopulegon with 91% yield and 19% de, which can be tandem with CRs to generate (−)-isopulegol.¹⁰² Peters constructed an enzymatic cascade approach coupling YqjM with AacSCH (a squalene hopene cyclase from Alicyclobacillus acidocaldarius) mutant which converts citral stereoselectively to (−)-iso-isopulegol with 10.8% conversion after 62 h¹⁰³ (Table 3).

Figure 6.

Synthesis of isopulegon and isopulegol.

Table 3. Application of ERs in the Synthesis of Isopulegon, Isopulegol, Dihydroperillaldehyde and Dihydro-β-ionone.

entry	whole cell or ene reductase	performance	reference
1	IPR from Mentha piperita	cis-isopulegone, 91% yield, 19% de	(102)
		(R)-pulegone, 28% yield, >99% ee
2	YqjM + AacSHC-I261A from Alicyclobacillus acidocaldarius	(−)-iso-isopulegol, 10.8% yield	(103)
3	OYE1 → OYE1-W116	(1S,4S)-DHPA, 42% conv., 79% de	(13)
		→
		(1R,4R)-DHPA, 20% conv., 52% de
4	DBR1 from Artemisia annua	dihydro-β-ionone, 93.8% conv.	(110)
5	PhCCD1 from Petunia hybrid + DBR1	dihydro-β-ionone, 85.8% conv.	(112)

3.4. Dihydroperillaldehyde

1,2-Dihydroperillaldehyde (DHPA) was found in the essential oil of Enhydra fluctuans,¹⁰⁴ a species common in South Asia, where it is used in folk medicine and as a condiment in food.¹⁰⁵ It also has been patented as a flavoring agent in 2008¹⁰⁶ (Figure 7). Notably, (1S,4S)-DHPA is suitable at a dilution level of 0.2 ppm to impart an effective watery character to fruit formulations, especially in watermelon aroma since the main part of the flavor sold on the market is lacking this feature and shows an undesirable melon side note.¹⁰⁵ Baker’s yeast,¹⁰⁵ NCY (Kazachstania naganishii DBVPG 7133)¹⁰⁷ and Saccharomyces cerevisiae NBRC 2260¹⁰⁸ were used for the C=C reduction of (S)-perillaldehyde. Unfortunately, due to the endogenous carbonyl reductases in whole cells, there was significant generation of alcohol byproducts, with little accumulation of target dihydro aldehydes. Zheng used purified OYE2p to catalyze (S)-perillaldehyde reduction with 2.71 U/mg specific activity.¹⁰⁹ Surprisingly, it also exhibited catalytic activity toward the inert double bonds of perillyl alcohol with low specific activity (0.18 U/mg). The OYE1-W116 mutant mentioned earlier was also used to screen for perillaldehyde reduction.¹³ For (S)-perillaldehyde, W116F increased de value from 76 to 96% while maintaining >98% conversion. For (R)-perillaldehyde, W116I partially reversed the stereo configuration of the product (79% de (1S,4S)-DHPA → 52% de (1R,4R)-DHPA) (Table 3).

Figure 7.

Synthesis of dihydroperillaldehyde.

3.5. Dihydro-β-ionone

Dihydro-β-ionone is a main aroma compound with a mellow, sweet, and fresh cedar scent in Osmanthus oil, which is widely used in foodstuffs, beverages, perfumes, and cosmetics.^110,111 According to statistics, the global market production of dihydro-β-ionone is approximately 10–100 tons per year.¹¹¹ Due to its widespread presence in roses, boronia fower, Osmanthus fragrans Lour, and juniper needles, it is mainly obtained by chemical extraction from plants.¹¹¹ To obtain stable sources of dihydro-β-ionone, green and efficient biocatalysis has been introduced into its preparation (Figure 8). DBR1 (from Artemisia annua) has been identified as a good biocatalyst for the reduction synthesis of dihydro-β-ionone, with up to 93.8% conversion.¹¹⁰ Furthermore, Zhao’s team constructed a dual enzyme system consisting of PhCCD1 (carotenoid cleavage dioxygenase from Petunia hybrid) and DBR1 for dihydro-β-ionone production in vitro.¹¹² By optimizing reaction conditions such as enzyme ratio, pH, and temperature, the maximum yield of dihydro-β-ionone was increased by five times, reaching 85.8% (13.34 mg/L). Recently, Wei’s team identified a new ene reductase KaDBR (from Kazachstania exigua HSC6) and provided kinetic parameters for β-ionone reduction.¹¹¹ The optimal reaction temperature for this enzyme is 60 °C, which provides possible application of biocatalysts under harsh industrial reaction conditions (Table 3).

Figure 8.

Synthesis of dihydro-β-ionone.

4. Aromatic Flavor

4.1. Lilial

3-(4-(tert-Butyl)phenyl)-2-methylpropanal, trade name Lilial or Lysmeral, is an aroma chemical with a sweet lily like odor. It is presented as a framing ingredient in many everyday cosmetics and household products and also has mosquito-repellent effects.^113,114 However, its use in cosmetics products has been prohibited in the EU due to its classification as a reproductive toxicant.¹¹⁴ The molecular structure of Lilial contains a chiral center, resulting in two isomers [(R)-Lilial and (S)-Lilial] (Figure 9). (R) configuration emits a stronger and more aggressive odor and the top scene is a little more watery, while (S) configuration has a softer and less expressive odor.¹¹⁵ Stueckler screened seven ERs (YqjM, OPR1, OPR3, NCR, and OYE1–3) for Lilial biosynthesis, four of which exhibited excellent enantioselectivity (NCR and OYE1–3, 83 → 95% ee (S)-Lilial).¹¹⁶ OPR1 showed opposite stereoselectivity with low ee values (17% (R)-Lilial). However, YqjM and OPR3 have no catalytic activity toward Lilial precursor (Table 4).

Figure 9.

Synthesis of Lilial.

Table 4. Application of ERs in the Synthesis of Aromatic Flavor.

entry	whole cell or ene reductase	performance	reference
1	OPR1, NCR, and OYE1–3	(R)-Lilial, 3% conv., 17% ee	(116)
		(S)-Lilial, 32% conv., > 95% ee
2	YqjM, OPR1, OPR3, NCR, and OYE1–3	(R)-Helial, 78% conv., 6% ee	(116)
		(S)-Helial, >99% conv., 97% ee
3	KYE1, XenA, YersER, and OYE2p	dihydrocinnamaldehyde, 2.91 U/mg	(40)
4	NcCAR from Neurospora crassa + OYE1	dihydrocinnamaldehyde, ∼80% yield	(123)
5	OYE3 + PLADH from Parvibaculum lavamentivorans or READH from Rhodococcus erythropolis	(2R,3S)-3-methyl-4-phenylbutan-2-ol	(127)
		76% yield, 99% de
		(2S,3S)-3-methyl-4-phenylbutan-2-ol
		79% yield, 99% de
6	OYE1 → OYE1 W116Q	(S)-3-methyl-4-phenylbutan-2-one	(89)
		56% ee→
		(R)-3-methyl-4-phenylbutan-2-one, ∼20% ee
7	Mucor subtilissimus CBS 735.70	raspberry ketone, 66.4% yield	(132)

4.2. Helional

3-(Benzo[d][1,3]dioxol-5-yl)-2-methylpropanal, marketed as Helional or Tropional, is a high added-value fragrance existing in many perfumes for its green, floral (cyclamen), and marine fresh note.¹¹⁷ It also has a chiral center at the same position as lilac aldehyde, with two configurations [(R)-Helional and (S)-Helional] (Figure 10). In the synthesis process of Helional, traditional chemical methods usually use H₂/catalyst to achieve hydrogenation of olefin precursors¹¹⁸ or alcoholysis reaction to remove ester groups of alkylated precursors.¹¹⁹ The former’s problems lie in the safety and poor selectivity of H₂ reduction, while the latter uses expensive substrates and difficult-to-handle catalysts in intermediates preparation. The introduction of biocatalysis, known for its gentleness, environmental friendliness, and sustainability, is beneficial in overcoming the above difficulties. OYE2 stands out for its excellent high activity (>99% conversion) and high enantioselectivity (97% ee (S)-Helional).¹¹⁶ YqjM and OPR1 yielded products with opposite configurations with low enantioselectivity (13 and 6% ee (R)-Helional, respectively)¹¹⁶ (Table 4).

Figure 10.

Synthesis of Helional.

4.3. Dihydrocinnamaldehyde and Its Derivatives

3-Phenolpropionaldehyde, also known as dihydrocinnamaldehyde, is an aroma chemical with hyacinth-like flavor, which is widely used in floral essence and tobacco flavor¹²⁰ (Figure 11). Many studies described the application of metal catalysts (such as Pd, Ni, and Zr) in the hydrogenation of cinnamaldehyd and have been summarized to form a review.¹²¹ KYE1,⁴⁰ XenA,⁴⁰ YersER⁴⁰ and OYE2p¹⁰⁹ were tested for catalytic activity in HCAL synthesis, among which YersER showed the best specific activity (2.91 U/mg). In addition, some NCYs have been identified to have the ability to reduce α-methyl-cinnamaldehyde, all of which generate excessively reduced alcohols as major products.^107,122 Unlike NCYs as catalysts, recombinant E. coli has lower endogenous CRs interference. Rudroff’s team prepared dihydrocinnamaldehyde from cinnamic acid using recombinant E. coli coexpressing NcCAR (carboxylic acid reductase from Neurospora crassa) and OYE1 with ∼80% dihydrocinnamaldehyde accumulation.¹²³ Taking inspiration from previous literature,¹²⁴ an in situ product separation (ISPR) strategy (using isooctane to form a two-phase system) was introduced to reduce the generation of alcohol byproducts to a greater extent (Table 4).

Figure 11.

Synthesis of dihydrocinnamaldehyde and its derivatives.

3-Phenylpropan-1-ol, i.e. hydrocinnamyl alcohol, which can be purified from the leaves of Cinnamomum species, has a sweet-spicy odor and is mainly applied in perfumery and personal care products (Figure 11). CaER (from Clostridium acetobutylicum) participated in the reconstruction microbial cell factory for 3-phenylpropanol synthesis, producing 847.97 mg/L of 3-phenylpropanol under optimal fermentation conditions¹²⁵ (Table 4).

3-Methyl-4-phenylbutan-2-ol, trade name Muguesia, is a commercial odorant, with floral, muguet, rose, and minty.¹²⁶ The molecular structure contains two adjacent chiral centers (at the C2 and C3 positions of the side chain), where the chirality at the C3 position of the side chain plays a crucial role in odor (Figure 11). Among them, (3R)-stereoisomers were described as weak and completely devoid of odor, while the (3S)-stereoisomers were found to be the effective odor vectors.¹²⁷ OYE3 catalyzed the reduction of (E)-3-methyl-4-phenylbut-3-en-2-one to optically pure (S)-3-methyl-4-phenylbutan-2-one (98% ee), subsequently reduced by PLADH (alcohol dehydrogenase from Parvibaculum lavamentivorans) or READH (alcohol dehydrogenase from Rhodococcus erythropolis) with different stereoselectivity to produce (2R,3S)- or (2S,3S)-3-methyl-4-phenylbutan-2-ol, respectively.¹²⁷ Monti’s team further screened more ERs and found that all enzymes generate products with the same (S) configuration.⁸⁹ Focusing on the hot amino acids in OYE1, saturation mutations were performed on Trp 116, and it was found that W116Q partially reversed the stereoselectivity of the product to (R) configuration (∼20% ee).⁸⁹ The target flavor Muguesia can be obtained by concatenating with CRs (Table 4).

4.4. Raspberry Ketone

4-(4-Hydroxyphenyl)butan-2-one, also called raspberry ketone (RK), has a raspberry aroma and a fruity sweetness, mainly found in red raspberry¹²⁸ (Figure 12). RK is widely used as a fragrance ingredient in perfumery, food, and tobacco products, and even as an over-the-counter medication for weight loss, making it of great economic value.¹²⁹ Natural RK is extremely expensive (about $3000/kg), about 52 times that of synthetic RK ($58/kg).¹²⁹ Although many studies have reported efficient chemical synthesis methods for raspberry ketones,¹³⁰ we still pursue green biotransformation approaches as they retain the title of natural flavor. About 30 years ago, Fronza discovered that yeast extraction catalyzed the C=C reduction of (E)-4-(4-hydroxyphenyl)but-3-en-2-one to produce raspberry ketone.¹³¹ The isotope labeling results showed that the hydrogen of β-C comes from NADPH, while the hydrogen of α-C comes from the solvent. Subsequently, 14 fungi and bacteria were screened for raspberry ketone preparation.¹³² Among them, Mucor subtilissimus CBS 735.70 seemed to perform the best because it produced a considerable amount of raspberry ketone in a short period (14 h) with very few alcohol byproducts (dihydro product: dihydro alcohol ≈ 90:10) (Table 4).

Figure 12.

Synthesis of raspberry ketone.

5. Aliphatic Flavor

5.1. Decanal

Decanal has a sweet, waxy, and floral fragrance and is used to blend citrus, orange, lemon, and other edible fruits flavors (Figure 13). In previous studies, LacER was shown to catalyze the C=C reduction of 2-decenal with a relative catalytic activity of 90.1% (the catalytic activity for 2-hexen-1-al reduction is 100%).⁷⁸ Papadopoulou identified 10 ERs, among which Pbr-ER from Pseudomonas brassicacearum had the best catalytic activity, with 94.2% conversion of 40 g/L decanal after 24 h reaction¹³³ (Table 5).

Figure 13.

Synthesis of decanal.

Table 5. Application of ERs in the Synthesis of Aliphatic Flavor.

entry	whole cell or ene reductase	performance	reference
1	Pbr-ER from Pseudomonas brassicacearum	decanal, 94.2% conv.	(133)
2	OYE1 + ADH-T from Thermoanaerobacter or ADH-LK from Lactobacillus kefir	γ-butyrolactone, 50–90% conv., 98 to >99% ee	(134)
3	OYE2 + READH from Rhodococcus erythropolis or KRED + TFA	γ-butyrolactone, 78–83% conv., 98–99% ee, 94–98% de	(127)
4	NCR or OPR1 + ADH-LB from Lactobacillus brevis or ADH-T or ADH-R from Ralstonia sp.	γ-butyrolactone, 9–69% conv., 82 to >99% ee	(136)
5	OYE3 + EVO030 or EVO270 purchased from Evocatal GmbH	2-methyl-3-substituted tetrahydrofuran precursors, 89.9–99.6% conv., 76 to >99% ee	(138)

5.2. γ-Butyrolactone

Lactone compounds often have excellent flavors, such as sweetness, fruit flavor, and milk aroma, while alleviating unpleasant flavors caused by high concentrations of fatty acids, such as sourness and bitterness (Figure 14). γ-Butyrolactone plays an important role in forming cheese flavor, and it imparts caramel, sweet, and coconut flavors to nut oil.

Figure 14.

Synthesis of γ-butyrolactone.

A dual enzyme catalytic system constructed by Pietruszka’s team produces two stereoisomers of γ-butyrolactone with high yield (80–90%) and high enantioselectivity (>99% ee) by successively reducing double bonds with ERs and carbonyl groups with CRs.¹³⁴ Based on this, this group subsequently designed a chemoenzymatic synthesis of γ-butyrolactone which introduced Wittig or Horner–Wadsworth–Emmons reaction to produce olefin intermediates.¹³⁵ Remarkably, γ-butyrolactone was obtained from only the ethyl ester substrates, while the tert-butyl ester substrates could not undergo the final self-cyclization reaction, and only C=C reduction products were obtained. Another notable point was that the chemical and enzymatic methods used in this study were independent and required the separation of intermediates before enzymatic synthesis. Brenna applied a dual enzyme catalytic system for the synthesis of 4,5-dimethylbutyrolactone, a molecule that can impart a unique fruit flavor to tobacco. Unlike previous studies, he chose unconventional enoate substrates to achieve high yields of lactone (78–83%) while maintaining excellent selectivity.¹²⁷ In addition, some researchers extended the carbon chain skeleton of the substrate ethyl enoate, using ADHs connected to ERs in series to obtain various alkyl-substituted γ-butyrolactones with good selectivity and moderate yield, expanding the application scope of the bienzyme system in the preparation of γ-butyrolactone.¹³⁶ However, this approach is limited by the requirement of a strong acid to protonate reduction products to undergo cyclization. Recently, Bai’s team achieved continuous flow preparation of multisubstituted chiral butyrolactones by combining SsER (from Swingsia samuiensi) and SsCR (from Scheffersomyces stipitis CBS 6054) using a three-dimensional microfluidic continuous reactor, resulting in a 3–7-fold increase in spatiotemporal yield.¹³⁷ This method provides a successful case of using cutting-edge microfluidic technology to improve enzyme catalytic efficiency and has motivated researchers to combine additional cutting-edge technologies with enzyme catalysis (Table 5).

5.3. 2,3-Disubstituted Tetrahydrofuran

2-Methyltetrahydrofuran-3-thioacetate is a molecule with a roasted meat aroma. Significantly, (2S,3R)-2-methyltetrahydrofuran-3-thioacetate has the most pleasant flavor, and other configurations can bring unexpected sulfur flavors to food¹³⁷ (Figure 15). Brenna’s team constructed a dual enzyme system of OYE3 and ADH to prepare chiral bromo alcohols with high yield (99%) and excellent enantioselectivity (99% ee).¹³⁸ Then, (2S,3R)-2-methyltetrahydrofuran-3-thioacetate was synthesized from the corresponding chiral bromo alcohol by a chemical method, which proved the application potential of the two-enzyme catalytic system in the molecular synthesis of useful compounds (Table 5).

Figure 15.

Synthesis of 2,3-disubstituted tetrahydrofuran.

6. Summary and Outlook

Whether pursuing green synthesis or high commercial value, biotransformation is the trend of F&F synthesis. In this review, we summarized the application examples of ERs in F&F synthesis over the past 20 years, including the identification of new enzymes, the introduction of cutting-edge technologies and the construction of multienzyme systems. The above work demonstrates that ERs have been attractive tools for chiral high-value-added product synthesis due to their green, efficient, and highly enantioselective advantages.

However, the productivity of these reactions and the scale on which they have been performed are not feasible for truly large-scale industrial synthesis, which is limited by poor industrial stability. In addition, high product levels and economically feasible production processes are key to achieving large-scale industrial preparation processes. Thus, to extend the application of ERs in environmentally benign synthetic processes, we should focus on the following research directions:

• (1)Protein engineering is a fundamental method for improving reaction levels and enzyme stability.¹³⁹ Combining directed evolution with the structural orientation of the target product can increase the reaction rate or reverse the stereostructure, resulting in increasingly efficient and potentially industrially engineered enzymes. Conversely, the key structural domains of enzymes with high temperature or organic solvent resistance can be grafted into the structures of ERs through structural grafting, which may improve the thermal stability or organic solvent resistance of the ER and effectively meet industrial application conditions. In addition, the new enzymes created by protein engineering may catalyze some unnatural reactions, breaking single reaction types and expanding the substrate spectra of ERs.
• (2)Immobilized enzyme engineering can enhance the industrial application stability and economic feasibility of ERs.¹⁴⁰ To date, researchers have used magnetic nanoparticles¹⁴¹ and hydrogels¹⁴² to immobilize ERs, successfully improving the thermal stability of the enzyme and achieving recycling. However, there is still relatively little research on ER immobilization, and the material functions are limited and lack innovation, leaving extensive exploration space.
• (3)Cofactor cycle engineering may simplify the reaction system. Due to the essential role of the cofactor NADPH in ER-mediated catalysis, researchers have been committed to finding suitable methods for avoiding the use of expensive cofactors. Using a glucose dehydrogenase (GDH)/glucose/NADP⁺ cycle system is the most common method and has significant complexity. This system includes substrates, ERs, cofactors, circulating enzymes and sacrificial substrates, increasing the difficulty of postprocessing. Therefore, simplifying the cofactor cycle system is a key issue for the large-scale application of ERs. The coexpression of ERs and cofactors of circulating enzymes can partially simplify the reaction system by forming a dual enzyme aggregate.¹⁴³ Moreover, materials with multiple effects can be used for cofactor cycling and enzyme immobilization, further reducing the difficulty of postprocessing.

In addition, the multienzyme cascade reaction is a green synthesis route that has a natural quality. We can mimic the biosynthetic pathways of natural products and design reasonable in vitro or in vivo biological cascade reaction routes, which is beneficial for reducing chemical reaction steps and improving production efficiency.¹⁴⁴ Emerging technologies, such as photocatalysis,¹⁴⁵ electrocatalysis,¹⁴⁶ and continuous flow technology,¹⁴⁷ can be combined with biocatalysis to create new techniques, providing additional opportunities for the industrial application of biotechnology. An increasing amount of related technological research has driven the development of enzyme catalysis in the field of asymmetric synthesis. The heterologous or in vitro total biosynthesis of compounds, such as essential compounds, natural products, and complex drugs, is becoming a reality. In the foreseeable future, the industrial application of ERs will be further developed and improved, and methods for the biocatalytic synthesis of F&F will become increasingly efficient.

This work was financially supported by the Opening Fund of Harmful Components and Tar Reduction in Cigarette Key Laboratory of Sichuan Province (jykf202209).

The authors declare no competing financial interest.