Re‐Investigation of Hydration Potential of Rhodococcus Whole‐Cell Biocatalysts towards Michael Acceptors

The implementation of a stereoselective Michael addition with water as substrate is still a major challenge by classical, chemical means. Inspired by nature's ability to carry out this attractive reaction with both high selectivity and efficiency, the interest in hydratases (EC 4.2.1.x) to accomplish a selective water addition is steadily rising. The gram‐positive bacterial genus Rhodococcus is known as biocatalytic powerhouse and has been reported to hydrate various Michael acceptors leading to chiral alcohols. This study aimed at the in‐depth re‐investigation of the hydration potential of Rhodococcus whole‐cells towards Michael acceptors. Here, two concurrent effects responsible for the hydration reaction were found: while the majority of substrates was hydrated in an oxygen‐independent manner by amino‐acid catalysis, an enzyme‐catalysed water addition to (E)‐4‐hydroxy‐3‐methylbut‐2‐enoic acid was proven to be oxygen‐dependent. 18O2‐labelling studies showed that no 18O2 was incorporated in the product. Therefore, a novel O2‐dependent hydratase distinct from all characterised hydratases so far was found.


Introduction
The selective addition of water to (un)-activated double bonds is known to be a very appealing yet chemically challenging reaction. Though water as a reactant provides benefits regarding sustainability and atom-efficiency, it is both a poor electroand nucleophile and therefore often first requires activation by e. g. strong acids. Additionally, unfavourable reaction equilibria often impede a profitable reaction. [1][2][3] While chemists still struggle to find activating conditions without diminishing the stereoselectivity, nature developed a way to circumvent these problems by the use of enzymes. Their unique 3-dimensional structures provide ways to activate the water molecule as well as stabilise transition states during the reaction. [1,4] Therefore, using microbial activities or applying purified enzymes in biocatalytic reactions to achieve a selective water addition to double bonds is nowadays often seen as advantageous. [5][6][7] Hydratases (EC 4.2.1.x) catalyse the water addition to activated as well as isolated double bonds. However, especially enzymes from the primary metabolism exhibit a narrow substrate scope which confines the applicability of these enzymes to their natural substrates. [4] While the use of hydratases to convert natural substrates already offers a huge advantage over classical chemical routes and has already been applied in industry, [8][9][10][11] the identification of hydratases with a broader substrate scope is highly desirable. A biocatalytic hydration of two non-natural, α,β-unsaturated substrates using whole-cells of Rhodococcus rhodochrous ATCC 17895 was first reported in 1998. [12] Further research to develop a straightforward approach to produce βhydroxy carbonyl compounds catalysed by a presumed 'Michael hydratase' followed in 2015. It was described that whole-cells of Rhodococcus additionally add water to small ring-closed organic molecules like c-hexenone or c-pentenone thereby improving future applications of an attractive water addition system. [13] Rhodococcus is a genus of a gram-positive bacteria that has been described to host a variety of enzymes and degradation pathways with high potential for the use in industrial processes. [8,14] Its metabolic diversity often stems from large, linear plasmids that carry the protein-encoding genes for degradative enzyme systems. [15] Abundant horizontal genetransfer events occurring via these linear plasmids as well as a high multiplicity of catabolic enzymes contributed to the catabolic versatility of Rhodococcus genus members. [15][16][17] To study the described water addition in-depth, the identification and isolation of the responsible enzyme were first aimed for. Due to complications in the protein isolation, wholecell systems were further investigated to explore the substrate range. During this process inconsistencies and contradictions within the earlier reports became evident. [12,13] Therefore, the aim of this study was to assess and extend those reports. In addition to a re-evaluation of described substrate structures, the main focus was on the clarification of the reaction type. Labelling studies with D 2 O and 18 O 2 with subsequent high resolution liquid chromatography-mass spectrometry (HRLC-MS) analysis were used to identify whether the net microbial hydration activity involves a true water addition or an oxidative process.

Attempted Isolation of Michael Hydratase
Whole-cells of Rhodococcus rhodochrous ATCC 17895 were described to catalyse the Michael-addition of water to several substrates. [12,13] Therefore, the isolation of the responsible enzyme was aimed for, but unfortunately, no sequence data was available nor any comparable protein has been described earlier. Subcellular fractionation of Rhodococcus cells led to the conclusion that the desired protein is membrane-associated. However, numerous attempts to solubilise the protein using classical detergents as well as a nanodisc approach using styrene-maleic acid (SMA) copolymers [18,19] did not lead to an isolated enzyme activity (data not shown). On the contrary, the attempted solubilisation of the membrane protein led to a complete loss of activity which could be a consequence of protein instability or the requirement of different subunits or components for the hydratase activity. As advanced comparative genomic analyses as well as membrane proteomics did not lead to the successful identification of the responsible enzyme, isolation efforts were not further pursued.
Instead, whole-cells as well as isolated membranes from a number of Rhodococcus strains were used to investigate the substrate range. Whole-cells were shown to also catalyse a number of unwanted side-reactions due to the presence of alcohol dehydrogenases and ene-reductases which also act on the small organic molecules (Scheme 1). [20] In comparison, isolated membranes offered a way to decrease the number of side-products formed, because soluble enzymes that may catalyse competing reactions were removed earlier (data not shown). Unfortunately, isolated membranes lack stability, demand a time-consuming preparation and resulted in a lower recovered activity. [21] For these reasons it was decided to use whole-cells for further substrate investigations.
In these reports, the synthesis of the biotransformation substrate was carried out as a two-step procedure involving a Wittig-reaction followed by a basic hydrolysis. [12,13] After repeating this protocol, however, 1-and 2-dimensional NMR spectroscopy of the product revealed no ring-closed structure 7, but the open-chain (E)-4-hydroxy-3-methylbut-2-enoic acid (5) which was in fact used as the substrate for bioconversions with Rhodococcus cells (Scheme S1, Figures S1-S4). Attempts to obtain the corresponding open-chain (Z)-isomer through basic hydrolysis of commercially available 4-methyl-furan(5H)-one (7) were unsuccessful.
Following the revision of the substrate structure, the Mhy activity was (re)-evaluated on a number of other Michael acceptors that have been reported previously (Table S1). The Scheme 1. Biotransformation of c-hexenone with Rhodococcus whole-cells leads to desired water addition product and unwanted side-reactions.

Oxygen-Dependency
Whole-cell optimisation experiments with substrate 5 surprisingly showed an improved product yield with an increasing headspace to reaction-mixture ratio (Figure 1). These findings indicate an oxygen-dependency which was not reported earlier.
On the contrary, previous reports stated that both the reaction of (E)-4-hydroxy-3-methylbut-2-enoic acid (5) and c-hexenone (1) were as efficiently catalysed under N 2 -atmosphere as under air. With this experiment the possibility of an oxidative process was previously excluded. [13] To resolve these apparently conflicting findings, the oxygen-dependency of the water addition reaction was re-investigated. Here, numerous experiments using substrate 5 under N 2 -atmosphere did not lead to any product formation while c-hexenone (1) was still converted and gave similar results as in aerobic trials ( Figure 1). The contradicting results of water addition activity under anaerobic conditions depending on the substrate class led to the question whether both substrate classes are converted by the same reaction mechanism and biocatalyst. Consequently, control reactions with whole-cells of Escherichia coli TOP 10 containing an empty pBADHisA expression vector were carried out. No product formation in the case of (E)-4-hydroxy-3-methylbut-2-enoic acid (5) was observed while c-hexenone (1) showed comparable product formations with the E. coli control system.
These results indicate the presence of two different effects acting on the two substrate classes. One possible explanation for the latter observation is a water addition catalysed by amino acids. In the past, amino acids were already shown to add water to Michael acceptors like c-hexenone or c-pentenone without any oxygen requirement. L-lysine obtained the highest formation of rac-3-hydroxy-c-hexanone with a yield of 21 %. [24] It is therefore likely that the amino acids present in both E. coli and Rhodococcus catalyse the reaction on c-hexenone regardless of the presence or absence of oxygen in our system as well.

Oxidation or Water Addition?
To exclude the possibility of an oxidative process instead of a water addition reaction, a number of labelling experiments were carried out. The conversion of 1 and 5 was catalysed both by cells of R. rhodochrous ATCC 17895 and R. pyridinivorans DSM 20415 under 18 O 2 atmosphere in H 2 O as well as under air in both D 2 O and H 2 O. The results were subsequently analysed with high resolution liquid chromatography-mass spectrometry (HRLC-MS). In case of an oxidative process, the product will show a higher mass due to the incorporated 18 O-atom (20) (Scheme 3). On the other hand, a water addition will lead to the incorporation of two deuterium-atoms (21) of which one is readily exchangeable thus 22 will be detected. A subsequent chemical elimination step will reveal whether the deuterated water was added in syn- (7) or anti-fashion (23) as the chemical E2-elimination selectively takes place in anti-fashion.
Biotransformations with both cells and both substrates 1 and 5 did not incorporate any labelled 18  To review the course of the water addition (syn-versus antiaddition) purified deuterated compound 22 was obtained in a large-scale biotransformation with Rhodococcus rhodochrous ATCC 17895 whole-cells. A subsequent chemical anti-elimination was performed yielding exclusively the undeuterated product 7 (Figures S12-S14). This experiment thus confirms earlier results, [13] but due to the revised substrate structure which is an open-chain rather than a ring-closed molecule, the water addition catalysed by Rhodococcus whole-cells consequently occurs in syn-fashion.
Biotransformations carried out under nitrogen atmosphere proved that the presence of oxygen is required for the conversion of 5 to take place. This was further sustained by the fact that more lactone 6 was produced under pure 18 O 2atmosphere than under air-conditions with less oxygen present ( Figure S10). Additional experiments with D 2 O under nitrogen atmosphere revealed no product formation.
The difference to the conversion of c-hexenone (1), however, is the remaining activity in the absence of oxygen as well as comparable activities achieved with E. coli cells. The obtained results from the labelling study further confirm the earlier theory that this conversion is an amino-acid catalysed process for ring-closed α,β-unsaturated carbonyl compounds 1, 18 and 19. This leaves compound 5 and its ethyl-derivative (E)-4-hydroxy-3-ethylbut-2-enoic acid (16) [12] to be the sole substrates requiring oxygen for a water addition. Due to high enantiomeric excess values of the final lactones which were shown in earlier studies, [13] the high substrate specificity and the fact that the activity is associated with the membrane fraction, we propose an enzyme-catalysed process rather than a metalassisted or amino-acid catalysed water addition.
The described O 2 -dependent behaviour of a presumed hydratase is, however, surprising and atypical compared to other identified hydratases as they are either known to be oxygen-independent or negatively affected by trace amounts of oxygen. The latter enzymes often hold metal-containing cofactors or cysteine residues that lead to the sensitivity. [2] One example of this group is the linalool (de)hydratase-isomerase which catalyses the water addition and isomerisation to unactivated monoterpenes like (S)-(+)-linalool to yield βmyrcene and geraniol. [2,25,26] Interestingly, Rhodococcus erythropolis MLT1 cells were shown to catalyse a similar reaction converting β-myrcene to geraniol. [2,27] In this whole-cell system, however, oxygen is required for the reaction and anaerobic conditions lead to no product formation. [27] This phenomenon is similar to our findings and, to the best of our knowledge, the only other report of an oxygen-dependent formal hydration reaction. The substrates, however, differ as our substrates belong to the group of activated Michael acceptors while in βmyrcene an unactivated double bond is hydrated. Nonetheless, it is remarkable that this phenomenon was observed only twice and only in Rhodococcus cells.

Conclusions
Michael additions with the unreactive water-molecule as substrate are highly attractive yet chemically very challenging reactions. Rhodococcus cells were shown to have microbial activity towards a number of Michael acceptors leading to βhydroxy carbonyl compounds. This study serves to expand the knowledge about these water addition reactions and a presumed 'Michael hydratase' was investigated in detail.
Though trials to isolate or identify the responsible hydratase failed, whole-cell and membrane systems were employed to examine the substrate scope. Here, previously described substrates were re-visited and corrected regarding their chemical structure using 1-and 2D-NMR analysis. Experiments under nitrogen atmosphere revealed two different effects being responsible for the water addition depending on the substrate used. The main group of substrates was proposed to be aminoacid catalysed as the same hydration activity was found in the presence and absence of oxygen as well as with E. coli cells. (E)-4-hydroxy-3-methylbut-2-enoic acid, however, was shown to be oxygen-dependently converted only by Rhodococcus cells and presumed to be enzyme-catalysed. Complete absence of oxygen led to no hydration. Labelling studies with D 2 O and 18 O 2 exposed a true water addition in syn-fashion and excluded any oxidative process. The described microbial hydration activity therefore remains highly attractive yet still elusive. A complete understanding of the oxygen dependency, a probable reaction mechanism, the finding of its natural substrate as well as an expansion of its substrate scope for future applications will only be possible upon identification of the responsible membrane protein (-complex) and successful heterologous expression in a suitable host system.

Experimental Section Chemicals
Unless stated otherwise, all commercial chemicals were purchased from Sigma-Aldrich (Schnelldorf, Germany) and used without further purification. Petroleum ether was purchased from VWR International (Amsterdam, The Netherlands) and distilled before utilisation. Hydroxyacetone was purchased from Alfa Aesar (Kandel, Germany) and rac-2-methyl-1,2,4-triol was purchased from Chemspace (Riga, Latvia).

Bacterial Strains and Microorganisms
R. pyridinivorans DSM 20415 was purchased from the German Collection of Microorganisms and Cell Culture (Leibniz Institute DMSZ) while R. rhodochrous ATCC 17895 was bought from the American Type Culture Collection (Manassas, Virginia, US).

Microbiological Protocols
The organisms were maintained at 4°C on a nutrient agar plate (3 g beef extract, 5 g peptone, 15 g agar dissolved 1 L de-ionised water and autoclaved at 121°C) and were regularly sub-cultured. The growth medium used consisted of 1 L de-ionised water with 6.59 g glucose, 9.2 g peptone, 1.84 g yeast extract, 7 mM K 2 HPO 4 (1.2 g), 3 mM KH 2 PO 4 (0.4 g) with a final pH of 6.8 and was autoclaved at 110°C. A preculture was inoculated with a single colony and grown overnight at 28°C. The preculture (2 mL) was used to inoculate 1 L of culture medium grown in 5 L flasks. The culture was incubated for 72 hours at 28°C with 180 rpm orbital shaking. The cells were harvested by centrifugation (17.696 xg, 15 min, 4°C). The cells were washed with potassium phosphate buffer (100 mM, pH 6.2) and stored at À 20°C.
Whole-cells of Rhodococcus were resuspended in 100 mM KPi buffer (pH 6.2) to a final cell content of 100 mg/mL. Substrates (10 mM) were added to the reaction volume of 500 μL and incubated 24 hours at 28°C at 1000 rpm. Small scale samples were extracted