Linear enzyme cascade for the production of (–)-iso-isopulegol

1.1 Biocatalytic cascades

In their broadest sense, cascade reactions are defined as the combination of several chemical reactions in one pot [1]. The main advantages of cascade reactions lie in the elimination of intermediate isolation and purification steps, thus allowing for a reduction of costs, production time and waste compared to classical step-by-step synthesis. Using a cascade strategy may also lead to an overall increase in yield, as unstable intermediates are directly consumed by the next step and are not accumulated in the reaction mixture [2].

In biocatalytic cascades, at least one reaction step is catalysed by a biocatalyst in the form of either isolated enzymes, immobilised enzymes, crude cell extracts or whole microbial cells [3], [4]. Nature frequently uses biocatalytic cascades as an efficient system to allow for life and growth, thus being the best teacher for the set-up of artificial analogues. The design of man-made in vivo and in vitro biocatalysis cascade reactions is additionally facilitated by the fact that most enzymes work in aqueous buffer systems with similar pH and temperature profiles, thus simplifying the establishment of in situ approaches. Due to the decreasing costs for gene synthesis and the development of improved enzyme engineering as well as high-throughput technologies [5], a plethora of artificial biocatalytic cascade reactions have been developed in the last two decades [2], [3], [4], [6], [7].

Biocatalytic cascades can be classified by reaction concepts into linear, orthogonal, parallel and cyclic cascades (Figure 1) [8]. Linear cascade reactions are the most prevalent. Here, one substrate is converted via one or more intermediates to the product. An example for a linear biocatalytic cascade reaction is the production of (R)- or (S)-1-phenyl ethyl amine from ethyl benzene via oxyfunctionalisation by a peroxygenase followed by amination catalysed by a transaminase [9]. In case of parallel cascade reactions, two transformations are coupled via their respective co-factor recycling to generate two products from two different starting materials in a redox neutral fashion. In contrast to parallel cascades, orthogonal cascade reactions employ additional enzymes for co-factor or co-substrate regeneration or side-product elimination. Cyclic cascade reactions are in most cases chemo-enzymatic reactions including a chemical racemisation [8]. Classification into one of the four concepts is not always absolutely clear, in which case the classification depends on the most important part of the cascade reaction.

Figure 1:

Different types of biocatalytic cascade reactions: (A) linear cascade, (B) parallel cascade, (C) orthogonal cascade, and (D) cyclic cascade.

1.2 Ene reductase

Members of the ene reductase (ER) enzyme class are effective biocatalysts for stereo-selective trans-hydrogenation of activated alkenes [10], [11], [12] complementing the respective chemical cis-hydrogenations via chiral rhodium or ruthenium phosphines (Knowles and Noyori, Nobel Prize in Chemistry 2001) [13], [14]. ERs are NAD(P)H-dependent oxidoreductases and can create up to two stereogenic centres in a single step by catalysing the stereo- and enantio-selective reduction of α,β-unsaturated ketones, aldehydes, nitro alkenes and carboxylic acids [10]. ERs can be divided into several families: old yellow enzymes (OYEs), the Fe-S cluster-depending enoate reductases, medium-chain dehydrogenases and short-chain dehydrogenases. Thanks to their broad substrate scope and favourable physico-chemical properties, members of the OYE family are currently most often used for synthetic purposes [12]. The first enzyme of this family was discovered by Warburg and Christian in bottom-fermented yeast and named “yellow enzyme” [15]. Subsequently, several other members of the yellow enzyme family were found, and the first ER was redefined as “OYE-I” [16]. Since then, various ERs that belong to the OYE family have been described from yeasts, bacteria, plants, fungi, and eukaryotes [17], [18].

The catalytic mechanism that all enzymes of the OYE family follow is well understood [19]. The enzymes use a bi-bi ping-pong mechanism that can be divided into the reductive half-reaction, in which flavin mononucleotide is reduced through hydride transfer from NAD(P)H, and the oxidative half-reaction, in which the hydride is transferred from the reduced flavin to the Cβ-atom of the activated alkene. To complete the reaction, the missing proton for the Cα is provided by a tyrosine residue [19], [20]. This specific mechanism typically leads to an anti-addition hydrogenation [21]. For the OYE enzyme family, it has been found that their natural stereo-specificity is nearly always the same [22]. However, the stereo-preference can be changed by point mutations making available stereo-complementary pairs of ERs [23].

All known enzymes of the OYE family have a typical TIM-barrel as basic structure [24]. The enzymes of the OYE family are further subdivided into two main subclasses: the classical OYEs and the thermophilic-like OYEs. The two classes can be distinguished by sequence homology and structural features [10]. One of the first classified thermophilic-like enzymes is YqjM from Bacillus subtilis [25], [26]. This well-characterised enzyme shows a broad substrate scope from aliphatic aldehydes to unsaturated esters [27].

1.3 Squalene hopene cyclases

Terpenes comprise the most structurally diverse family of natural products, currently numbering >80,000 members in a greater family that also includes steroids and carotenoids [28]. Thus, terpene cyclases, the enzymes that catalyse the formation of these polycyclic compounds in a regio- and stereo selective fashion, are an attractive source for generating compound diversity. Terpene cyclases are classified according to the chemical strategy for initial carbocation formation: class I terpene cyclases use a trinuclear metal cluster to break a labile allylic diphosphate ester bond yielding an allylic cation and inorganic pyrophosphate, whereas a class II terpene cyclase relies on a general acid in the active centre to protonate the terminal ispoprene unit [29]. Due to their reaction mechanism, enzymes belonging to class I are restricted to substrates containing diphosphates, while the class II enzymes may have a broader substrate scope.

Squalene hopene cyclases (SHCs) are class II terpene cyclases and initiate cyclisation of their natural substrate squalene, a C₃₀ triterpene, via the activity of an unusually acidic aspartic acid residue acting as a natural Brønsted acid. Controlled by the hydrophobic pocket of the enzymes, the polycyclisation reaction proceeds in a predetermined manner and a final deprotonation or water addition leads to the formation of the products hopene or hopanol [30]. Interestingly, SHCs show substantial substrate promiscuity and are also capable of catalysing the cyclisation of small (C₁₀ to C₂₀) terpenoids and terpenoid-like molecules [31], [32], [33].

The best-characterised SHC, AacSHC, was isolated from Alicyclobacillus acidocaldarius and is the only SHC for which a crystal structure could be elucidated [34], [35]. The dimeric thermostable enzyme is capable to catalyse the cyclisation of small compounds such as geranylacetone (C₁₃) but also accepts C₃₅ non-natural polyprenoids [29].

Siedenburg et al. showed for citronellal that the inherent catalytic properties of SHCs are more generally utilizable. Using wild-type enzymes or variants with only one amino acid exchange, SHCs were able to convert citronellal to the different isopulegols via a Prins reaction [32], [36], [37]. More specifically, an AacSHC mutant (AacSHC_I261A) was identified that converts 2 mM (S)-citronellal to the single diasteromer (–)-iso-isopulegol with 8% conversion after 40 h at 30 °C [38], [39].

1.4 Biocatalytic cascade reaction toward (–)-iso-isopulegol

To accomplish our goal to transform (E/Z)-citral (1) selectively to (–)-iso-isopulegol (see Figure 2), an important intermediate in the synthesis of isomenthol derivatives [40], we chose an enzymatic cascade approach by coupling the well-known ER YqjM from B. subtilis [26] with the SHC variant AacSHC_I261A [38]. The ER YqjM converts 1 with >95% ee and good activity into the desired S-product 2 [25], [27], [41], while AacSHC I126A has been reported to exhibit precise conformational control of the cyclisation of (S)-citronellal forming exclusively (–)-iso-isopulegol (selectivity >95%) [38].

Figure 2:

Enzyme cascade capable of converting citral (1) via citronellal (2) to (–)-iso-isopulegol (3).

2.1 In vivo enzyme cascade

First, synthetic or isolated genes encoding the selected enzymes for the biocatalytic cascade were cloned into vector pET28b+ before enzymes were expressed individually in Escherichia coli BL21 (DE3) to determine best cultivation and reaction conditions. The reactivity optimum of YqjM and AacSHC_I261A in different buffer systems was established by carrying out whole-cell biocatalysis reactions in the presence of either 50 mM sodium phosphate buffer (pH 7.5), 50 mM sodium citrate buffer (pH 6.0) or 50 mM sodium citrate buffer (pH 6.0) containing 0.2% Triton X-100. Analysis of the reaction mixtures after 24 h showed that YqjM exhibited the best conversion from 1 to 2 using sodium citrate buffer containing 0.2% Triton X-100 at pH 6.0 (see Table 1). This result is coherent with published pH profiles. It has been reported, for example, that YqjM exhibits the highest activity for ketoisophorone at pH 6.0 [27]. Notably, even when using low concentrations of cell mass (20 mg/mL), we detected production of the alcohol side product 2s. In parallel to the synthetic biocatalysis reactions, we therefore determined the background reaction of endogenous E. coli enzymes by using non-transformed E. coli BL21 (DE3) cells as negative control. In this case, we observed very low (S)-citronellal 2 production by the E. coli ER N-ethylmaleimide reductase [42] but significant formation of side product alcohol 1s, presumably due to the activity of endogenous alcohol dehydrogenases (ADHs) [43], [44]. Further control reactions included incubation of (S)-citronellal 2 and (–)-iso-isopulegol 3 with YqjM-containing cells, leading to the formation of side product 2s or no reaction, respectively.

The whole-cell biotransformations of (S)-citronellal 2 employing 200 mg/mL cells expressing AacSHC_I261A exhibited activity only in combination with citrate buffer independent of the presence of Triton X-100. The detected conversions were low (<1%), as after 6 h, the substrate 2 was already quantitatively converted to the unwanted side product 2s. Further control experiments in which (E/Z)-citral 1 was incubated with AacSHC_I261A-expressing cells revealed the predominant formation of side product 1s and low amount of alcohol 2s.

As expected, initial attempts of in vivo reactions, in which 20 mg/mL YqjM and 200 mg/mL AacSHC_I261A-expressing cells were co-incubated with substrate 1, confirmed the presence of active endogenous E. coli ADHs, as only side product 2s was detected.

2.2 In vitro enzyme cascade

In an attempt to crudely purify the enzyme solutions and inactivate the problematic endogenous ADHs, we applied heat shock to both cell suspensions for 30 min at 60 °C. As expected, the thermostable AacSHC_I261A was still active after prolonged incubation at 60 °C [45]; however, the ER YqjM did not retain its catalytic activity. Moreover, we also found that some of the endogenous ADHs were still active after the heat treatment. As an alternative approach to reduce the side reactivity of the native ADHs, the pH of the biocatalysis reaction was lowered to pH 4.0. While the AacSHC variant retained its activity, the biocatalytic conversion rates by YqjM dropped dramatically and no product 3 was detected. Having unsuccessfully explored several optimisation strategies in the context of a whole-cell biocatalysis cascade, we decided to set up a biocatalytic cascade based on semi-purified enzyme.

As AacSHC_I261A is a membrane-bound protein, separation of the SHC from contaminating soluble proteins is straightforward. Using the semi-purified AacSHC fraction in a resolubilised form with 2% Triton X-100 (see Figure 3), the intermediate substrate 2 could be converted to product 3 with a final conversion of 7.3% after 24 h. In a next step, we employed semi-purified AacSHC_I261A in combination with 10 mg/mL whole cells containing YqjM, thus reducing the total whole-cell load in the biocatalysis reactions and consequently also the quantity of contaminating ADHs. Using this strategy, the resulting cascade generated 3.4% of product 3 after 24 h; however, we also again detected significant side product 2s formation (66%) (see Table 2).

Figure 3:

Reducing SDS-PAGE showing the expression levels of enzymes YqjM and AacSHC with 12% acrylamide using a Precision Plus Protein™ dual-colour standard marker: (A) cultivation samples normalised to OD7; (B) samples after the semi-purification of the enzymes.

To further reduce sources of ADH, we extended the semi-purification approach also to YqjM. As the available YqjM construct did not have any affinity tags, semi-purification was carried out using anion exchange chromatography followed by buffer exchange via size exclusion chromatography. The cascade reactions with both enzymes in a semi-purified form led to product 3 formation of 6.2% after 24 h and 10.8% after 62 h, while generating <5% of side product 2s. As reported, the use of AacSHC_I261A exclusively led to the formation of (–)-iso-isopulegol [38], [39]. This overall yield corresponds closely to results obtained with the semi-purified AacSHC in combination with intermediate substrate citronellal 2, leading us to believe that the next optimisation steps would have to be carried out on the level of the AacSHC variant. Another possible optimisation will be to increase the temperature to 60 °C after the conversion of 1 to 2 is completed. Going beyond these strategies, background ADH activity could be reduced by the use of a protease-mediated purification strategy, provided both target enzymes are protease resistant [46]. In cascades, fusion proteins [47], [48] have also been considered to facilitate direct transport of intermediates to the following enzyme. However, in the context of the here-described cascade, such a strategy would not only require the fusion of a membrane protein to a soluble protein but would likely pose additional difficulties as the two enzymes are characterised by differing kinetic parameters.

4.1 Chemicals and materials

All chemicals were purchased from Merck KGaA (Darmstadt, Germany), Chemie Brunschwig AG (Basel, Switzerland), VWR (Hannover, Germany) or Carl Roth (Karlsruhe, Germany), and were used without further purification unless otherwise specified. NADPH was obtained from OYC Europe (Rotterdam, Netherlands) and also used without further purification. Phusion polymerase was purchased from Thermo Fisher Scientific (Waltham, MA, USA). DpnI was provided by New England Biolabs (Ipswich, MA, USA), and primers were ordered from Microsynth AG (Balgach, Switzerland). The Mono Q columns and the desalting columns were ordered from GE Healthcare (Uppsala, Sweden).

4.2 Bacterial strains and plasmids

Escherichia coli BL21 (DE3) [fhuA2 [lon] ompT gal (λ DE3) [dcm] ΔhsdS] was purchased from New England Biolabs (Ipswich, MA, USA). Escherichia coli XL1-Blue [recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F ́ proAB lacI^qZ∆M15 Tn10 (Tet^r)] was obtained from Agilent Technologies (Santa Clara, CA, USA). The plasmid pet28_AacSHC was ordered from Twist Bioscience (San Francisco, CA, USA) and contained the codon-optimised gene of AacSHC (AAA75452.1) without any tags, cloned into the pet28b+ vector system.

4.3 Cloning

The ER gene yqjm was amplified from genomic DNA of B. subtilis sp. (strain CCOS014 from Culture Collection of Switzerland, Wädenswil, Switzerland) with the following primers: forward 5′-GAGTGCGGCCGCAAGCTTTTACCAGCCTCTTTCG-3′ and reverse 5′-CTTTAAGAAGGAGATATACCATGGCCAGAAAATTATTTAC-3′. The empty pet28b(+) vector was amplified via forward primers 5′-CGAAAGAGGCTGGTAAAAGCTTGCGGCCGCACTC-3′ and 5′-GTAAATAATTTTCTGGCCATGGTATATCTCCTTCTTAAAG-3′. PCR products were purified using PCR Clean Up Kit from Machery Nagel (Düren, Germany), combined and transformed into E. coli XL1 Blue via the FastCloning method [49]. The correctness of the resulting plasmid pet28_YqjM was verified by sequencing with MWG Eurofins (Ebersberg, Germany), and correct construct was subsequently transformed into E. coli BL21 (DE3). The AacSHC mutant I261A was created by amplification from pet28_AacSHC with the forward primer 5′-GTAGTTGGGGTGGTGCGCAGCCTCCG-3′ and T7-reverse primer. The PCR product was used as MegaPrimer in a MEGAWHOP PCR [50] using the original pet28_AacSHC as template to introduce the desired mutation. After a 2-h DpnI digest at 37 °C, the PCR product was transformed into E. coli XL1 Blue cells. The resulting plasmid, pet28_AacSHC_I261A, was sequenced and a correct construct was transformed into E. coli BL21 (DE3).

4.4 Enzyme expression

Bacterial cultures were incubated in baffled Erlenmeyer flasks in a New Brunswick™ Innova^® 42 orbital shaker at 200 rpm at 37 °C. Bacteria on agar plates were incubated in a HERATherm Thermo Scientific incubator in air. All materials and biotransformation media were sterilised by autoclaving at 121 °C for 20 min. Aqueous stock solutions were sterilised by filtration through 0.20-mm syringe filters. Agar plates were prepared with LB medium supplemented with 1.5% (w/v) agar and antibiotics. Expression of YqjM in E. coli BL21 (DE3) was performed by inoculation of LB medium (400 mL) supplemented with kanamycin (100 μg/mL) with an overnight culture (4 mL, 1:100). The YqjM culture was incubated at 37 °C at 180 rpm until an OD600 of 0.6–0.8 was reached. The next expression was induced by supplying isopropyl-β-D-thiogalactopyranoside to a final concentration of 0.1 mM, and incubation was continued at 25 °C for 16 h. AacSHC_I261A in E. coli BL21 (DE3) was cultivated identically except for the use of terrific broth as cultivation medium. The cells were harvested by centrifugation for 10 min at 4000×g and stored at −20 °C.

4.5 Partial purification of the enzymes

The partial purification of AacSHC_I261A was performed according to the protocol of Bastian et al. [38]. Thawed cells of a 400 mL cultivation were resuspended in 15 mL cell disruption buffer [0.2 M sodium citrate, DNAse 0.1 mg/mL, pH 6.0 (dH₂O)]. Cell disruption was performed by a single passage through a French pressure cell at 2000 psi. Cell debris was collected by centrifugation at 38,500×g for 40 min at 10 °C, and the supernatant was discarded. The cell debris was resuspended in 5 mL wash buffer (0.05 M sodium citrate, pH 6.0) and again collected by centrifugation at 38,500×g for 40 min at 10 °C. The enzyme was solubilised overnight (16 h) at 4 °C with 5 mL solubilisation buffer (0.05 M sodium citrate, 2% Triton X-100, pH 6.0). Afterwards, the mixture was centrifuged (38,500×g for 40 min at 10 °C) and the supernatant was used for the biotransformations (partially purified enzyme). For the partial purification of YqjM, the protocol of Fitzpatrick et al. [26] was adapted: thawed cells of a 400 mL cultivation were resuspended in 15 mL buffer A (50 mM Tris-HCl, pH 7.5). Cell disruption was performed by a single passage through a French pressure cell at 2000 psi, afterwards the cell debris was removed by centrifugation at 8000×g for 20 min at 4 °C, and the supernatant was subjected to ion exchange chromatography on a Mono Q column (2.5×10 cm) freshly equilibrated in buffer A. All steps were done on an Äkta Pure device with a 4 mL/min flow rate. After washing with six column volumes of buffer A, bound protein was eluted with a linear gradient of 50 mL each of buffer A and buffer B (same as buffer A but containing 500 mM KCl). The yellow fractions containing YqjM were pooled and concentrated, and the buffer was changed to 0.05 M sodium citrate buffer (pH 6.0) by using three coupled 5 mL HiTrap desalting columns. The protein contents of the crude cell extract as well as that of the partial purified fractions were determined by using the BCA kit. Standard curves were recorded with bovine serum albumin at 0.02–2.0 mg/mL. Samples were measured in triplicate at suitable dilutions with a Tecan Spark device.

4.6 Biotransformations

For biocatalysis, whole-cell suspensions or semi-purified enzyme was used. The appropriate volume of the enzyme solution and 5 mM substrate (stock 1 M in DMF) supplemented with 6 mM NADPH were combined in a sodium citrate buffer (0.05 M, pH 6.0) with 0.2% Triton. A total of 1 or 0.5 mL reaction volume was incubated in closed gas chromatography (GC) vials at 30 °C and 850 rpm. All biocatalysis reactions were done in triplicates, and samples of 500 μL were taken after 8, 24 or 62 h. To determine the background reaction, control experiments with buffer only or E. coli cells harbouring no overexpressed protein were performed.

4.7 GC analysis

Samples (500 μL) were extracted two times with 400 and 200 μL ethyl acetate containing 1 g/L octanol as internal standard and dried over sodium sulphate. The extract was subjected to GC-flame ionisation detection analysis. The analysis was done using an Agilent GC 7890B with a DB 5 (30 m×0.25 mm×0.25 μm) column. The separation run was performed at constant pressure of 25 psi with a starting temperature of 75 °C. Next, a temperature ramp of 2 °C/min was carried out. At 110 °C, the heating rate was increased to 90 °C/min up to 300 °C, which was held for 2 min. The method was adapted from Ref. [38]. The conversion calculations were done using the standard curves of the substrate, intermediate product and product.