Vp StyA 1 and Vp StyA 2 B of Variovorax paradoxus EPS : rather an aryl alkyl sulfoxidase than a styrene epoxidizing monooxygenase

1 Affiliation 1; Institute of Biosciences, Environmental Microbiology, TU Bergakdemie Freiberg, Leipziger Str. 29, 09599 Freiberg, Germany; dirk-tischler@email.de, ringoschwabe007@gmail.com, lucasbenedikt@aol.com, kristin.friebel@student.tu-freiberg.de, stefan.kaschabek@ioez.tu-freiberg.de, anika.scholtissek@gmail.com, heinet@tu-freiberg.de 2 Affiliation 2; Microbial Biotechnology, Ruhr University Bochum, Universitätsstr. 150, 44780 Bochum, Germany * Correspondence: dirk.tischler@rub.de; Tel.: +49-234-32-22656

SMOs rely on the flavin cofactor flavin adenine dinucleotide (FAD).The reduced flavin is transferred by diffusion or by a direct transfer between both components [14][15][16] monooxygenase tightly binds the reduced cofactor and the oxygen driven catalysis gets initiated [14][15][16][17][18].Here oxygen gets activated by the reduced flavin to a (hydro)peroxy-FAD which can attack the actual substrate.Upon substrate oxygenation a hydroxyl-FAD intermediate is formed and decomposes to oxidized FAD and water [17][18][19].The product is subsequently released and another catalytic cycle can start.
Styrene monooxygenases of type E2 have been shown to be excellent in sulfoxidation with respect to substrate conversion and enantioselectivity whereas the E1 type only shows a high activity at low enantioselectivity [1][2][3][4][5][6][7][8][9][10].However, so far there is only a single E2 type SMO in detail studied and it has a comparable low activity [5,6,9].Thus, it was reasonable to screen for further E2 type SMO candidates which can be applied in biocatalysis.
Powered by our former studies on E2 type SMOs [4][5][6]9,11,20,21] we took the chance to investigate the phylogenetic more distinct system VpStyA1/VpStyA2B of Variovorax paradoxus EPS (accession numbers: ADU39063 and ADU39062).Thus we here present its general activity, capability to act on styrene but also on sulfides (Scheme 1) compared to other monooxygenases and to evaluate its biotechnological applicability.
Scheme 1. Cartoon of styrene monooxygenase activity.The reductase domain of StyA2B reduces FAD (displayed in oxidized form between protein monomers) upon NADH-consumption.Reduced FAD can be used by both monooxygenase units StyA1 (major; right) and StyA2 (minor; left) to activate molecular oxygen and to subsequently oxygenate the substrates (here for example styrene and benzyl methyl sulfide) [5][6][7][8][9][10].Upon substrate oxygenation hydroxyl-FAD is formed and thereof water is eliminated to recycle the FAD in its oxidized form for the next catalytic cycle.

Evolution of VpStyA1 and VpStyA2B from strain EPS
Already during an earlier study, the putative genes encoding for VpStyA1 and VpStyA2B and thus the respective gene cluster of strain EPS had been identified [20].According to a phylogenetic analysis and the surrounding genomic region the proteins were assigned as SMO related to the E2 prototype of Rhodococcus opacus 1CP [5,6,11].Interestingly, the sequence similarity of VpStyA1 and VpStyA2B (74 % identity over 404 amino acids) between both monooxygenase subunits was much higher as among other StyA1/StyA2B couples.Further, they form together a separate branch in a distance tree of monooxygenase components [20,22].As in the meanwhile more of those putative sequences had been released we could refine the phylogenetic study and try to understand the fusion even in more detail (Fig. 1).Respectively, the fusion event was discussed as functionally convergent event which is reinforced by the updated phylogenetic distance tree.However, no activity of these putative SMOs of type E2 originating of Variovorax species has been reported until now!Mining the databases for styA2B-like genes/proteins revealed that they occur mainly among represents a flexible loop between two helices according to the models obtained from homology modelling.This is not surprising as there is no sequence-structure relation available for this part.

Preprints
However, it fits to earlier made observations in which the start of the reductase domain was investigated [9,19,20,23].Respectively, this proposed linker region need to be investigated.
Adjacent to the monooxygenase domain and the proposed linker region the reductase domain of the fusion proteins follows.Interestingly, the reductase sequence misses in all cases a few amino acids (range: 3 to 14) when compared to the StyB reductases of E1 type SMOs (Fig. 1) [5,7].The mentioned linker region was now target for a subsequent site-directed mutagenesis investigation.In all cases the original sequence (408-AREAV) was replaced or even extended.The following mutants have successfully been prepared and verified by sequencing: 408-TIVVV, 408-AAAAA, 408-HHHHH, 408-WYHHH, 408-WYHHHHH, and 408-GQWCSQY.In order to validate the made assumptions, the wildtype protein VpStyA2B and the mutant proteins were produced and assayed.
The chosen linker sequences of mutants to be produced base on the following rationale.Linker with A-and V-rich sequences were chosen to be flexible or H-rich to be more rigid.Whereas, the WY-Hvariants should represent bulkier and thus even more restricted linkers.The 408-GQWCSQY motif represents a terminal part of a StyA-protein originating of pseudomonads (Fig. 1) and was picked as a comparison with respect to earlier made fusion proteins [19].

Molecular genetic work and enzyme production
The cloning of both genes, VpstyA1 and VpstyA2B, as well as the generation of mutants was successfully accomplished which had been verified by sequencing the inserts of gene expression plasmids as described in experimental section (See supporting information, Table S1) and by a simple activity assay.Since E. coli BL21 derivatives allow the formation of indole from tryptophan during growth on complex medium as used herein, the activity of styrene monooxygenases and related enzymes can be verified by indole transformation to indigo [4][5][6]10,11,19].And indeed, all clones obtained (E. coli BL21 (DE3) pLysS derivatives harboring the wildtype or mutant genes in a pETvector) produced indigo during cultivation even without being induced for overexpression of target proteins.Clones which showed highest rate in indigo formation had been selected, propagated and stored as glycerol stocks for later protein production and characterization efforts.
Gene expression and protein purification was performed as described elsewhere [9].The individual clones were cultivated either in a 3 L-fermenter or in 1 L-flasks in LB medium with appropriate antibiotics and 0.1 mM IPTG as inductor as described previously.In both cases (VpStyA1 and VpStyA2B), enzyme production was successfully achieved with a yield of 2 to 4 mgVpStyA2B and up to 9 mgVpStyA1 protein per liter broth, respectively.This is in congruence to other studies.Subsequently, the proteins produced and purified were characterized.

Reductase activity of VpStyA2B and mutants
The fusion protein VpStyA2B of strain EPS was supposed to be rather a reductase of the complete SMO system [6,7], and was for those reasons in analogy to the enzyme of strain 1CP characterized.
The protein VpStyA2B was successfully produced (verified by SDS-PAGE, not shown).The fraction obtained from Ni-chelate chromatography was slightly yellow which is an indicator for a bound flavin.This was analyzed as described for other SMOs and FAD was clearly determined by means of RP-HPLC as well as by spectroscopic methods and the use of authentic standards.A FAD saturation of 4.9 to 20.8 mol-% was calculated for the protein applied.This is in congruence to results obtained earlier for RoStyA2B [6,17].This fraction was immediately assayed for activity and then concentrated and stored at -20 °C in a suitable storage buffer until characterization.Indeed, it could use NADH as source for reducing equivalents in order to reduce flavins.NADH cannot be replaced by NADPH.But, in case of the flavins FAD, FMN and riboflavin can be acceptors of reducing equivalents (Table 1).However, a clear preference could not be determined with respect to catalytic efficiency, which was for all here employed flavins between 0.57 and 0.88 s -1 µM -1 , respectively.
Table 1.NADH:flavin oxidoreductase activity of VpStyA2B (MW = 66.32 kDa which was calculated from the amino acid sequence including the N-terminal tag). 2 Either electron donor or acceptor was present in excess and the data obtained of triplicates were analyzed assuming Michaelis-Menten kinetics.
The reductase activity and catalytic efficiency of VpStyA2B is up to 10-times higher than reported for the Rhodococcus enzyme RoStyA2B [5].Thus, differences on amino acid level (57 % identity) are reflected within the biochemical properties of StyA2B-like SMOs.Still, the activity of the fusion protein is by an order of magnitude lower if compared to most reductases of two-component systems [22][23][24].However, it was reported recently that the N-terminal fusion to the monooxygenase drastically decreases the oxidoreductase activity in RoStyA2B [23].In addition, an artificial fusion protein was constructed from two-component relatives of Pseudomonas [19].Herein, the coupling was improved and the catalytic mechanism changed.It was also shown that the N-terminal region of the reductase influences the binding and affinity for the substrate [19,22].Therefore, it is likely that the same is true for VpStyA2B.However, it is likely that this effect of the fusion to StyA2 is different in dependence of the linker region in the Variovorax enzyme.
In contrast to the monooxygenase part (StyA2), promiscuity towards the flavin cosubstrate is in accordance with most characterized SMO reductases.And so far, only one representative from strain 1CP (RoStyB) is reported to be selective for FAD [9,[22][23][24].
However, this is unlikely as two additional purification steps were applied for the VpStyA2B preparation and majority of the Ni 2+ should be removed.Interestingly, thioanisole decreases the reductase activity, which was not observed for SMO-reductases before.In order to investigate the fusion region of this protein affords to determine the respective region had been accomplished on sequence level (see above) and served for the direction in a mutagenesis study.The six mutants of VpStyA2B obtained were separately produced and purified as described above and in analogy to the wildtype characterized for their NADH:FAD oxidoreductase activities (Table 2).During expression also these clones produced significant amounts of indigo indicating a functional expression of respective SMO-variants. 1 Concentrations of NADH and FAD in order to determine kinetic properties were chosen as for the wildtype VpStyA2B given in Table 1.As previously triplicates were used to get the values and standard deviations were in all cases less than 15 % and thus comparable to Table 1.
The mutants were all active and it was possible to determine kinetic parameters as for the wildtype oxidoreductase.Respectively, the wildtype was most active with NADH and all the mutants showed a lower activity.However, the affinity for NADH seemed not to be altered as the Km values were all in the same high range.This picture changed somewhat for the series collected for FAD.The general activities were similar to those for NADH-variation and also the order from the most active to the worst representative were same.But, the affinity for FAD was drastically different among the variants which was obvious from the Km values determined.Thus, the best and most efficient variant was 408-AAAAA with respect to its oxidoreductase activity.Respectively, the catalytic efficiency of FAD-reduction is about 5.4 s -1 μM -1 which is about 8.4-times more efficient than the wildtype.This is due to the tighter FAD binding expressed by the 6.6-times lower Km value for FAD.Thus means the proposed linker region, which was mutated, has a strong effect on the FAD binding and turnover.This is in congruence to earlier studies which could demonstrate that the Nterminal part of NAD(P)H:flavin oxidoreductases is important for the flavin binding and thus for catalysis [19,23].Further, it indicates that the selected region indeed can be the linker as it is functional relevant for the reductase domain.

Monooxygenase activity of VpStyA1, VpStyA2B and VpStyA2B-mutants
Already during the gene cloning and expression experiments the formation of indigo was observed and used to identify best protein producers as mentioned above.This was later verified by a plate assay as described in experimental section in order to get a view on the strains producing either a highly active SMO or a lot of this SMO (Table 3).The wildtype proteins VpStyA1 and VpStyA2B and especially the mutant 408-AAAAA produced most indigo.The acceptance of indole as a substrate and the formation of indigo fits to a very recent finding that related E2 type monooxygenases were assigned to indole degradation and act basically as indole monooxygenases (IMOs) [12,13]. 1 Mutants are made of VpStyA2B as described in methods and were in analogy to the wildtype investigated. 2The electron transfer yield (ETY) was calculated from the NADH-consumption vs. epoxidation rate. 3In case of StyA1 an additional NADH:FAD oxidoreductase was needed which could also assist StyA2Blike proteins [6,23]. 4The plate screening was performed with clones expressing respective genes on an agar plate and the indigo formation was followed online by a camera.

n.a. = no activity measureable
The wildtype monooxygenases VpStyA1 and VpStyA2B show a comparable epoxidase activity as other SMOs (range 0.02 to 2.1 U mg -1 ) (Table 3) [5][6][7][8][9][10]24].VpStyA1 is about 2.9-times and 2.1-times more active than VpStyA2B and the Rhodococcus counterpart RoStyA1, respectively.And it was confirmed that VpStyA1 represents the major monooxygenase activity of the system as it was found for the prototype RoStyA1/RoStyA2B [6].This was expected but also indicates that the high sequence similarity of the Variovorax monooxygenases does not change the catalytic properties to a large extent.
However, it is interesting that VpStyA2B is about 8.4-times more active on styrene as RoStyA2B.The latter one could be boosted by additional FAD-reductase, but, VpStyA2B not.Later another batch of VpStyA2B was produced for biotransformations.It showed only a specific epoxidase activity of 27 mU mg -1 which is about 6-times lower.In this case surplus of reductase activity by an additional FAD-reductase could boost the activity 2.3-times.This indicates that VpStyA2B can be saturated by reduced FAD to a certain maximum as it was found for the Rhodococcus counterpart [6].Another difference between both StyA2B-proteins is the FAD-reductive power which is about 4.3-times higher in case of VpStyA2B with a kcat of 22.3 s -1 vs. that of RoStyA2B of 5.2 s -1 .This might be the reason that VpStyA2B can reach a maximum epoxidase activity even without an additional FAD-reductase (Table 1).Furthermore, the quality of the protein batches prepared can differ with respect to FAD saturation as well as activity which is known for related enzymes [5,6,9,17].However, we could not identify the fact which led to this different specific activity as the only differences during enzyme production were the fermentation volume and a final polishing step (gel filtration).The protein batches had otherwise a similar concentration and purity after purification which was also consistent except for the volumes applied.This needs further investigation.
The mutants obtained were in analogy to the wildtype enzymes characterized for their capability to convert styrene into styrene oxide.The wildtype VpStyA2B showed an activity of 159 mU mg -1 .
The VpStyA2B-variants were prepared similarly and assayed immediately in order to allow a direct comparison.The following relative styrene epoxidase activities were determined: 84.9 % for 408-TIVVV, 163.5 % for 408-AAAAA and less than 2 % in case of the other variants (408-HHHHH, 408- WYHHH, 408-WYHHHHH, and 408-GQWCSQY).This clearly indicates the region mutated has significant effects also on the epoxidase activity of StyA2B-proteins even if it seemed to be only the C-terminal part of the respective monooxygenase domain.Further, from the nature of this proposed linker sequence it can be reasoned that smaller hydrophobic residues yield higher activities compared to larger or more polar residues.This is indeed true for the FAD-reductase as well as the corresponding styrene epoxidase activity.And again the mutant 408-AAAAA showed a very promising catalytic behavior which was better than the wildtype in terms of activity and efficiency (Table 3).It was followed by variant 408-TIVVV which was even more active with surplus of reduced FAD formed by an additional FAD-reductase.It achieved almost the epoxidase activity of VpStyA1.
This might allow to draw some conclusions as it might be possible to generate more efficient and catalytic active self-sufficient fusion proteins; here SMO-like proteins.But, as the mutagenesis of the C-terminal part of the StyA2-part improved catalytic properties it might be also possible to alter StyA1-or StyA-like proteins at their C-terminus in order to improve their catalytic behavior.
However, structural investigations become necessary in order to elucidate the nature of substrate and cofactor binding to really understand why the mutants show a distinct activity and efficiency!

Biotransformation of sulfides
For the biotransformation experiments, another fresh batch of each enzyme (VpStyA1 and VpStyA2B) was produced and the initial specific activity of styrene epoxidation was determined to 27 mU mg -1 VpStyA2B, 63 mU mg -1 VpStyA2B and surplus RoStyBart as well as 121 mU mg -1 VpStyA1 and surplus RoStyBart, respectively [23].The lower specific activities of these batches might be due to remaining impurities in the protein preparation as for these experiments after the Ni-chelate chromatography the additional gel filtration was omitted.Another important point is that the addition of extra reducing power in form of RoStyBart supports the monooxygenase performance, as it was discussed above and which is in congruence with earlier studies.For comparison, in the 2 h biotransformation assay with styrene as substrate, observed activities with surplus RoStyBart were 0.002 U mg -1 for VpStyA2B and 0.029 U mg -1 for VpStyA1.
As target compounds sulfoxides had been chosen and thus kinetics and later on biotransformations were done with selected model sulfides.Kinetic parameters were determined as described above for styrene in order to allow a comparison.Maximum activity on the substrate benzyl methyl sulfide (BMS) from such kinetic studies was determined to 3.113 U mg -1 VpStyA1 and 1.616 U mg -1 VpStyA2B.This is about 25 to 26-times faster as the conversion of styrene.And taking the maximal epoxidase activity of VpStyA1 (Table 3) into account, a specific activity of about 10 U mg -1 could be possible.This high sulfoxidase activity of the SMO system of Variovorax paradoxus EPS is impressive and thus more sulfides were tested in a series of 2 h biotransformations (Table 4).
VpStyA1 reaches observed activities of up to 0.281 U mg -1 and BMS is the best substrate.These activities are much higher if compared to styrene as substrate.VpStyA1 produces almost exclusively the (S)-sulfoxide products.VpStyA2B has lost about 97 % of its initial activity by up to 0.044 U mg -1 , but also preferably produces the (S)-enantiomer.However, due to the mentioned loss of activity the exact ee-determination failed.
However, both monooxygenases lost at least one order of magnitude of activity and it has to be elucidated what causes the decrease.It might be possible that there is product inhibition or that the enzymes are not stable over time under the assay conditions.Especially, the more complex fusion protein seems to be less stable in cell-free approaches.Also it need to be mentioned that in all cases no other products had been observed which could indicate any overoxidation of sulfoxides to sulfones.This is an important feature to produce chiral sulfoxides.
When comparing the sulfide biotransformations it gets obvious that they are for each enzyme in the same order of magnitude.And in both cases a significant preference for the substrate benzyl methyl sulfide was observed.Here the S-atom is one positon more far from the aryl system as in case of the other sulfides applied.This indicates that the activated oxygen of the hydroperoxy-FAD intermediate in the active site of the monooxygenase is more closely located to this position of the substrate and allows a faster attack (Scheme 2) [17,18].This hypothesis needs to be verified by structural and kinetic investigations.But, so far there is no experimentally proven SMO structure inclusive substrate and cofactor available.distal position of the S-atom is likely closer to the C4a-hydroperoxyflavin, as BMS is preferred over sulfides, were the S-atom is adjacent to the aryl group.

Synthesis of sulfoxides
Commercial sulfides served as base to produce sulfoxides which had been applied as standards for analytical methods.The chemical sulfoxidation was achieved according to the protocol published earlier [25] as it provided high yields of desired products.First a solution of 7 mmol sulfide in 100 ml methanol and 20 ml water plus 10 ml titanium (III) chloride (16 % aqueous solution) was prepared.
Then dropwise a hydrogen peroxide solution (3.2 ml 30 % aqueous in 15 ml methanol) was added while constantly stirring at ambient temperature (about 20 °C).Latest after 25 min the substrate was completely converted and the reaction had been stopped by adding 50 ml water.Sulfoxides formed had been extracted by chloroform (three times) and subsequently dried over anhydrous magnesium sulfate.Chloroform was further removed under reduced pressure.Success of reactions and purities of products had been determined as described elsewhere [26].

Nucleotides, sequence analysis and molecular modelling
Sequence analyses based on a previously performed investigation [5].Thus accession numbers of the E2-type SMO components originating of Variovorax paradoxus EPS were already available: ADU39063 (VpStyA1) and ADU39062 (VpStyA2B).These were used for a BLASTP search in order to identify related or even homolog proteins.This allowed to generate an amino acid sequence alignment in analogy to earlier reports [5,20].This served as a template for the subsequent calculation of a dendrogram and the modelling.As templates for homology modelling the two available structures (PDB StyA: 3IHM, and PDB PheA2: 1RZ0) were employed [18,27]

Bacterial strains and cultivation
Escherichia coli strains DH5α and BL21 (DE3) pLysS were cultivated for cloning and expression as described elsewhere [5,6,9,31].A list of genes, primers and plasmids for this work is presented in the Supplemental Information (Table S1).
Indigo formation was observed during the cultivation and gene expression experiments.A plate screening in order to monitor indigo formation was set up as follows.The expression clones were transferred on a M9 mineral media agar-plate in a predefined grid, followed by a 22 h incubation at 30 °C [32].The solid media was supplemented with antibiotics as described previously and 0.5 mM isopropyl-β-D-thiogalactopyranosid for induction of protein expression [5,6,9,31].Indigo formation was started by spraying the plates with an indole-substrate solution (5 Mm Tris-HCl, pH 7.5, 50 mM indole, 20 % DMSO).The plates were then transferred onto a white light-table in a darkened box equipped with a camera.Pictures of the plates were shot in a 30 s interval.The color formation of each colony was monitored and normalized upon colony size to get a time-resolved intensity profile for each clone.

Molecular genetic work
The DNA sequences of VpstyA1 (Accession number: MF781076) and VpstyA2B (MF781075) were optimized for the codon usage and GC content of Acinetobacter baylyi ADP1 as described previously [9,33].The genes were purchased in a pEX-A vector system from Eurofins MWG (Ebersberg) with 5'-NdeI and 3'-NotI restriction sites allowing for subcloning into pET16bP [9].Site-directed mutagenesis of the linker area was done by using the GeneMorph II EZClone Domain Mutagenesis Kit (Agilent . The Preprints (www.preprints.org)| NOT PEER-REVIEWED | Posted: 25 January 2018 doi:10.20944/preprints201801.0245.v1© 2018 by the author(s).Distributed under a Creative Commons CC BY license.

Scheme 2 .
Scheme 2. Illustration of the putative binding and positioning of benzyl methyl sulfide towards the hydroperoxyflavin in the active site of SMOs from V. paradoxus EPS.Kinetic data indicate that the

Figure 1 ,
Figure 1, respectively.As tools have been employed: MEGA7 for the alignment, Modeller program version 9.19 for the comparative homology modelling as well as GenDoc and PyMol V1.1r1 for alignment and 3D-structure visualization [9,28-30].

Table 4 .
2 h Biotransformation of sulfides.Activities are given in U mg -1 and represent the observed activities determined after 2 h (no initial rates).Conversions of 2 mM substrate are given.