INTRODUCTION
Transaminases (TAs) (EC 2.6.1.x), also called aminotransferases, are versatile enzymes with industrial potential (
1). They catalyze asymmetric amine transfer reactions between an amine and a ketone, aldehyde, or keto-acid and, thus, are key enzymes to produce building blocks for drug discovery and chemical biology. All transaminases reported so far require pyridoxal-5′-phosphate (PLP) as a coenzyme, which serves as a molecular shuttle for ammonia and electrons between the amine donor and the acceptor in a catalytic cycle. First, the amine group from the amine donor binds to the enzyme, and then pyridoxamine-5′-phosphate (PMP) is formed from PLP and the amine donor is released as a keto product. Afterwards, PMP transfers the amine group to the acceptor and PMP is regenerated to PLP, closing the catalytic cycle. Based on their amino acid sequences, transaminases are classified in six groups (classes I to VI) (
1), with class III covering the so-called ω-transaminases (ω-TAs). Within the ω-TAs, the class of amine transaminases (ATAs) has industrial relevance, as they have been used for the preparation of optically pure amines starting from the corresponding ketones (
1,
2).
In an ideal scenario, functional screening with genomics and metagenomics techniques would allow the identification of a new generation of microbial biocatalysts, including ATAs of the class III ω-TAs (
3–6). However, extensive bioprospecting by metagenomics was only rarely successful (
5), despite the growing number of sequences available in public databases (
7). Indeed, only three class III ω-TAs have been identified by metagenomics techniques; however, this was by applying sequence homology-based techniques rather than functional methods (
8). These enzymes showed poor levels of performance with ketones compared with their performance with aldehydes and keto acids. As an example, the conversion of acetone was measured to be less than 0.04% relative to that for 2-oxobutyrate and propionaldehyde (
8), which were the preferred keto acid and aldehyde substrates, respectively. A thermodynamic limitation for the amination of ketones is a common characteristic of ω-TAs (
9). For instance, the
kcat/
Km ratio of the ω-TA from
Ochrobactrum arthropi for acetophenone was only 0.0004% relative to that for pyruvate (
10). Also, ω-TAs from
Parococcus denitrificans and
Chromobacterium violaceum showed from 0.015% to 0.083% relative activities for ketones compared to the values for α-keto acids and aldehydes (
9). This prompted the research to create, by active-site engineering, ω-TA variants displaying improved capacity for the synthesis of chiral amines from bulky ketones. By applying this procedure, a 105-fold activity improvement for the conversion of butyrophenone was achieved for the ω-TA from
O. arthropi, although the relative activity compared to the value for aldehyde was still low (from 1.4 to 11.3% relative activity) (
10).
Finding new ω-TAs displaying high capability for the conversion of ketones, in combination with good stability and preferably stringent (
R) or (
S) selectivity (
11–13), is thus a priority for the synthesization of pharmaceutically valuable chiral amines (
9). However, their discovery is limited, most likely due to a lack of suitable screening methods at large scale. Recently, new assays for high-throughput screening of ATAs in liquid or solid phase were described (
14,
15). By adapting these methods to screen a large collection of clone libraries generated from environmental DNA of diverse origins, we successfully identified 10 genes encoding presumptive ATAs of the class III ω-TA family. These genes were expressed in fusion proteins fused to polyhistidine (His) affinity tags, purified by immobilized metal affinity chromatography, and characterized. The results presented here illustrate the benefits of the methods herein applied to screen for ω-TAs using metagenomics. The extensive analysis of their substrate spectra allowed the identification of those capable of converting bulky ketones and bulky amines, as well as environmentally relevant amines like putrescine. Finally, the application of sequence and 3-dimensional-model analyses shed new lights on the molecular determinants of their substrate specificities and stereochemistry. The present study may help future bioprospecting and engineering programs to identify and design class III ω-TA family proteins converting bulky ketones and bulky amines with stringent (
R) or (
S) stereospecificity.
DISCUSSION
In this study, we adapted two high-throughput screening methods to identify 10 class III ω-TA family proteins. The identified ω-TAs originated from bacteria from at least four different and divergent lineages: the
Pseudomonas,
Acidihalobacter, and
Amphritea genera and the
Rhodobacteraceae family. ω-TAs from the
Pseudomonas genus (
31) and
Rhodobacteraceae family (
32,
33) have been reported previously. However, this study examines ω-TAs derived from bacteria of the
Amphritea (TR
8) and
Acidihalobacter (TR
2) genera, which are bacterial groups rarely investigated from an enzymatic point of view (
34,
35). Our results suggest that the class III ω-TA (TR
8) from the bacterium of the
Amphritea genus is thermoactive (up to 65°C) and stable (it retained more than 35% activity in methanol, acetonitrile, and DMSO, each at a concentration of 50% [vol/vol]) and efficiently converts bulky substrates and (
S) amines. The enzyme from the bacterium of the
Acidihalobacter genus was capable of accepting bulky substrates as well as (
R) and (
S) amines. These features clearly suggest that those transaminases and their bacterial origins should be considered for chemical transformations in the future.
Particularly noticeable was the high level of performance with ketones in comparison to the results using aldehydes and keto acids, with specific activities for ketones being up to 173.4% relative to those of the best-accepted aldehyde and keto acid substrates. Such a preference for ketones is rarely observed for other native or engineered ω-TAs (
8–10), which exemplifies the potential of bioprospecting programs to identify new enzymes for the amination of bulky ketones. Molecular determinants for this unusual specificity for bulky ketones were found and suggested, including a hairpin region proximal to the highly conserved Arg414 and residues in its proximity (
Fig. 7) as a major determinant of the preference for bulky ketones and amines. Identifying transaminases containing this hairpin and applying rational or traditional protein engineering in the region may allow class III ω-TAs capable of converting bulky ketones and bulky amines to be designed. The conserved Ser231 was also found to be a major determinant of the preference for amines with longer alkyl substituents (Fig. S7).
The present study also reported 6 class III ω-TAs with a stringent (
S) enantioselectivity. This is a common feature within most transaminases, such as the ω-TAs belonging to PLP type I fold, which are all specific toward the (
S) enantiomer of their substrates. This also accounts for class III ω-TAs (
30). Interestingly, we also reported four ATAs capable of acting toward (
S) and (
R) amines, with the (
S) enantiomer being preferred. Note that both (
S)- and (
R)-selective ω-TAs have been found in distantly related families other than class III ω-TAs, namely those belonging to Pfam class IV transaminases with a PLP type IV fold (
28,
36–40). Compared to the (
S)-selective enzymes, the (
R)-selective counterparts are less abundant and have been less studied. All (
R)-specific class IV ω-TAs preferentially convert aliphatic substrates with high yields and high enantioselectivities [enantiomeric excess (ee) higher than 90% for (
R) enantiomers], but the yields are significantly lower with aromatic substrates. Recently, the introduction of 27 mutations into a fold IV ATA allowed the substrate scope toward bulky substrates to be broadened (
29). Therefore, four of the enzymes herein reported represent examples of the class III ω-TA family converting (
R) amines, although they also convert (
S) amines. Having a class III ω-TA acting toward (
S) and (
R) amines with low selectivity toward the latter may be more of a disadvantage than an advantage, as they cannot give access to highly optically pure amines. Although they are not (
R) selective, the capacity to act efficiently toward (
R) amines can be used as a starting point to apply rational design and protein engineering to design (
R)-selective variants from the naturally occurring class III ω-TAs identified herein and, possibly, others. This study suggests a number of molecular determinants which may help in the rational design of such enzymes. These determinants include a large active-site pocket, the presence of a hairpin region close to the conserved Arg414, and the outward orientation of Arg414 (
Fig. 7). Improving or even reversing the selectivity by single point mutations has also been shown by recent examples in other classes of ω-TA (
11,
13), although engineering has been shown to lead to variants with low selectivity and it is not a universal effect for all substrates (
11).
Taken together, this study reported examples of class III ω-TAs efficiently converting not only bulky ketones with stringent (S) stereochemistry but also one converting bulky ketones and bulky amines with a large alkyl substituent and a number converting bulky ketones and both (R) and (S) amines. Four enzymes additionally retained significant activity up to 60 to 65°C, and five were stable in concentrations of up to 50% (vol/vol) of organic solvents. Altogether, the amine transaminases herein reported display biochemical properties that make them attractive candidates for a variety of chemical conversions and suggest future actions to design (R)-selective class III ω-TAs.
The characterization of the 10 class III ω-TAs herein described also allows their participation in polyamine catabolism, namely that of putrescine, a ubiquitous and important biological molecule (
41), to be increased. We found that class III ω-TAs that contained the highly conserved Arg414 in an outward conformation (TR
2, TR
6, TR
9, and TR
10) (
Fig. 7) were capable of degrading putrescine via the formation of 4-aminobutanal (
Fig. 3). In the opposite case, class III ω-TAs in which Arg414 was in the inward conformation (TR
3 to TR
5, TR
7, and TR
8) (
Fig. 7) were not able to degrade putrescine. TR
1 was also capable of degrading putrescine. Its sequence differs from those of TR
2 to TR
10 by the absence of a hairpin region proximal to Arg414 (
Fig. 7), causing an inward configuration of this residue, which is slightly differently oriented than those in TR
3 to TR
5, TR
7, and TR
8. Therefore, the orientation of the highly conserved Arg414 may be used not only as an indicator of the capacity of class III ω-TAs to degrade bulky ketones, bulky amines, and (
R) and (
S) amines but also as an indicator of the capacity to degrade putrescine. Note that in previously reported class III ω-TAs, the conserved Arg414 has also been shown to adopt different conformations, inward or outward (
30). This orientation has been implicated in the recognition of carboxylate groups of keto acids and in determining the size of the large pocket (
27). However, no environmental implication was suggested for the different orientations of the conserved Arg414. In this study, we found that the different orientations of the conserved Arg414 cannot,
per se, be directly linked to the distinct capacity to convert keto acids over ketones, as was previously suggested (
27,
30), since enzymes in which Arg414 is similarly oriented show marked differences in substrate preference and their capacity to use keto acids (
Fig. 1). Rather, we found that the orientation can be linked to the capacity to degrade environmentally important biological polyamines (
41), such as putrescine, as shown in this study (
Fig. 3).
MATERIALS AND METHODS
General experimental procedures.
All chemicals used for enzymatic tests were of the purest grade available and were purchased from Fluka-Aldrich-Sigma Chemical Co. (St. Louis, MO, USA). E. coli strains MC1061, a generous gift from Eric Geertsma, and DH5α were used for expressing TR1 to TR10.
Naive screens.
The fosmid libraries used in the present study derived from 9 geographically distinct marine samples. They include samples from Port of Milazzo and Port of Messina (Sicily, Italy) (
16,
18,
42), the Ancona harbor (Ancona, Italy), with uric acid and ammonium amendments (
17), the Priolo Gargallo harbor (Syracuse, Italy) (
18), the Arenzano harbor (Ligurian Sea, Genoa, Italy) (
18), an acidic beach pool on Vulcano Island (Italy) (
18,
42), the El Max site (Alexandria, Egypt) (
18), Bizerte Lagoon (Tunisia) (
18), and the Gulf of Aqaba (Red Sea, Jordan) (
18). In all cases, DNA extraction and preparation of pCCFOS1 fosmid libraries were performed as described elsewhere (
16,
18,
42). A genomic library of
P. oleovorans strain DSM 1045 was constructed as described previously (
43), with minor modifications. Briefly, genomic DNA of
P. oleovorans DSM 1045 was isolated and fragmented by sonication, and appropriately sized fragments were then collected by gel extraction and end repaired (
44). 5′-End phosphates were removed by using alkaline phosphatase (FastAP; Thermo Scientific) followed by DNA precipitation. 3′-End adenine overhangs were added by using
Taq polymerase and cloned into the pCR-XL-TOPO vector according to the manufacturer’s recommendations (TOPO XL PCR cloning kit; Invitrogen). The recombinant plasmids were then transformed into
E. coli TOP10 cells by electroporation.
Clones were scored for their ability to perform transamination reactions by adapting a colorimetric assay (
14). Briefly, the method is based on the use of 2-(4-nitrophenyl)ethan-1-amine as an amine donor that when converted into the corresponding aldehyde and subsequently deprotonated would give a highly conjugated structure with absorbance in the UV region and an orange/red precipitate. Fosmid clones were plated onto large (22.5- by 22.5-cm) petri plates with Luria-Bertani (LB) agar containing chloramphenicol (12.5 μg ml
−1) and induction solution (Epicentre Biotechnologies, Madison, WI, USA) in an amount recommended by the supplier to induce a high fosmid copy number. After overnight incubation at 37°C, the plates were overlaid with 40 ml of a solution of K
2HPO
4 buffer, pH 7.5 (100 mM), containing 0.4% (wt/vol) agar, to which the following chemicals were added immediately prior to use: PLP (10 mg, or 1.0 mM final concentration), the amine donor 2-(4-nitrophenyl)ethan-1-amine (202.64 mg, or 25 mM final concentration), and benzaldehyde (106.12 mg, or 25 mM final concentration) as the aldehyde acceptor. Note that the method can be adapted to any other ketone or aldehyde. Positive colonies in agar plates change to an orange/red color in 20 to 30 min when the colonies are overlaid with the screening solution. Screens were also performed with
o-xylylenediamine hydrochloride as the amine donor by adapting a colorimetric assay (
15). Briefly, clones were plated on LB agar containing 12.5 μg ml
−1 chloramphenicol (for screening the clone library from environmental sources) or 50 μg ml
−1 kanamycin (for screening the
P. oleovorans clone library). After overnight incubation at 37°C, clones were transferred to Whatman paper. A drop of reaction solution containing 5 mM
o-xylylenediamine hydrochloride and 2.02 mM PLP in 100 mM K
2HPO
4 buffer, pH 8.0, was placed in the lid of a petri dish and covered with the colony-bearing Whatman paper. After the petri dish was sealed to prevent drying of the reaction solution, the plate was incubated at 30°C overnight. Positive clones with transaminase activity were identified by black coloration.
Positive clones containing presumptive transaminases were selected, and their DNA inserts were sequenced using a MiSeq sequencing system (Illumina, San Diego, CA, USA) with a 2 × 150-bp sequencing kit. Upon completion of sequencing, the reads were quality filtered and assembled to generate nonredundant metasequences, and genes were predicted and annotated as described previously (
45). The sequences of the inserts of the plasmids containing TR
9 and TR
10 genes were obtained from
P. oleovorans genome data (
46) after terminal sequencing of the plasmid insert (LGC Genomics GmbH, Berlin, Germany).
Gene expression.
Two expression platforms were used. Codon-optimized synthetic versions of the TR
1 and TR
3 to TR
8 candidate genes were synthesized by GenScript (Hong Kong) and delivered in a customized pUC plasmid. These constructs were dissolved in sterile water upon arrival and used as delivery plasmid for subcloning by fragment exchange (FX) into expression vector pBXCH or pBXNH3 using
E. coli MC1061 as a host (
24,
47). Candidate genes for TR
2, TR
9, and TR
10 were amplified from clonal DNA using gene-specific primers containing overhangs with restriction sites (NdeI/XhoI for TR
2 and NdeI/HindIII for TR
9 and TR
10) and cloned into expression vector pRhokHi-2 using
E. coli MC1061 as the host (
25).
Recombinant protein purification.
All recombinant proteins were expressed with His tags and purified as follows. Briefly, selected
E. coli clones that were found to convert the screening substrates were grown at 37°C on solid LB agar medium supplemented with the appropriate antibiotics (100 μg ml
−1 ampicillin for pBXCH or pBXNH3 or 30 μg ml
−1 kanamycin for pRhokHi-2). Single colonies were picked and used to inoculate 10 ml of LB broth supplemented with the appropriate antibiotic in a 0.25-liter flask. The cultures were then incubated at 37°C and 200 rpm overnight. Afterwards, 10 ml of this culture was used to inoculate 0.5 liter of LB medium plus antibiotic, which was then incubated to an optical density at 600 nm (OD
600) of approximately 0.7 (ranging from 0.55 to 0.75) at 37°C. The expression of TR
1 and TR
3 to TR
8 was induced by adding
l-arabinose to a final concentration of 0.1%, followed by incubation for 16 h at 16°C. TR
2, TR
9, and TR
10 were constitutively expressed using the same conditions (no inductor needed). In all cases, the cells were harvested by centrifugation at 5,000 ×
g for 15 min to yield a pellet of 2 to 3 g (wet weight). The wet cell pellet was frozen at −86°C overnight, thawed, and resuspended in 15 ml of 40 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.0. Lysonase bioprocessing reagent (Novagen, Darmstadt, Germany) was added (4 μl g
−1 wet cells) and cells incubated for 60 min on ice with rotation. The cell suspension was sonicated for a total of 5 min and centrifuged at 15,000 ×
g for 15 min at 4°C, and the supernatant was retained. The soluble His-tagged proteins were purified at 4°C after binding to Ni-NTA His-Bind resin (Sigma Chemical Co., St. Louis, MO, USA), followed by extensive dialysis of the protein solutions against 100 mM K
2HPO
4 buffer, pH 7.5, by ultrafiltration through low-adsorption, hydrophilic, 10,000 (10K)-nominal-molecular-weight-cutoff membranes (regenerated cellulose, Amicon) and storage at 4°C. The proteins were further purified by gel filtration as described previously (
48). Purity was assessed as >98% using SDS-PAGE analysis (
49) in a Mini-PROTEAN electrophoresis system (Bio-Rad). The protein concentration was determined according to the Bradford method with bovine serum albumin as the standard (
50).
Enzyme assays for determinations of acceptor substrates.
Transaminase activity was assayed using 2-(4-nitrophenyl)ethan-1-amine and structurally diverse keto acids, aldehydes, and ketones in 96-well plates as previously described with some modifications (
14). Note that K
2HPO
4 buffer was used, following recommendations described elsewhere (
11,
14,
28). Briefly, assay reactions were conducted as follows. Prior to the assay, a solution of 25 mM amine donor 2-(4-nitrophenyl)ethan-1-amine and 1.0 mM cofactor PLP was first prepared in 100 mM K
2HPO
4 buffer, pH 7.5 (40 ml). A 400 mM acceptor (keto acid, aldehyde, or ketone) stock solution was prepared in acetonitrile or buffer, depending on solubility. Reaction assays, in 96-well microtiter plates, were started by adding 2.5 μl of a protein solution (stock solution, 10.0 mg/ml in 100 mM K
2HPO
4 buffer, pH 7.5) to an assay mixture containing 185 μl of PLP–2-(4-nitrophenyl)ethan-1-amine solution and 12.5 μl of acceptor stock solution. The final volume of the assay mixture was 200 μl, and the amine donor and acceptor concentrations were 25 mM each. All measurements were performed in triplicates at 40°C in a microplate reader at 600 nm (Synergy HT multimode microplate reader; BioTek) in continuous mode for a total time of 180 min. Specific activities (in U g
−1 protein) were determined. One unit (U) of enzyme activity was defined as the amount of protein required to transform 1 μmol of substrate in 1 min under the assay conditions using a reaction product extinction coefficient (aldehyde 4 in reference
14) of 537 M
−1 cm
−1 at 600 nm, as determined experimentally. All values were corrected for nonenzymatic transformation (background rate).
Enzyme assays for determination of amine substrates, including enantiopure amines.
Transaminase activity was assayed using benzaldehyde as the acceptor and structurally diverse amines in 96-well plates. Any other aldehyde, ketone, or keto acid may be used instead of benzaldehyde. Prior to the assay, a solution of 25 mM benzaldehyde as the acceptor and 1.0 mM PLP as the cofactor was first prepared in 100 mM K
2HPO
4 buffer, pH 7.5 (40 ml). Then, a stock solution of each amine was prepared in acetonitrile or buffer, depending on the solubility. Reaction assays, in 96-well microtiter plates, were started by adding 2.5 μl of a protein solution (stock solution, 10.0 mg/ml in 100 mM K
2HPO
4 buffer, pH 7.5) to an assay mixture containing 185 μl of PLP-benzaldehyde solution and 12.5 μl of amine stock solution. The final volume of the assay mixture was 200 μl, and the amine donor and acceptor concentrations were 25 mM each. Reactions were allowed to proceed for 60 min at 40°C, during which time the amount of benzaldehyde remaining (not reacting with the amines) was determined every 5 min by adding 12.5 μl of a stock solution of the amine 2-(4-nitrophenyl)ethan-1-amine (400 mM in 100 mM K
2HPO
4 buffer, pH 7.5). After adding 2-(4-nitrophenyl)ethan-1-amine, the reaction was allowed to proceed for 10 min and absorbance due to the appearance of orange/red color was recorded continuously in a microplate reader at 600 nm (Synergy HT multimode microplate reader; BioTek). Lower absorbance values imply higher consumption of benzaldehyde and, thus, of the corresponding amines used as donors. Enzyme activity under the assay conditions was expressed as the amount of enzyme required to transform 1 μmol of substrate in 1 min under the assay conditions using a reaction product extinction coefficient (aldehyde 4 in reference
14) of 537 M
−1 cm
−1 at 600 nm. All values were corrected for nonenzymatic transformation (background rate) and for the results from a control reaction mixture containing benzaldehyde but not amines [no transfer reaction occurs, so all of the benzaldehyde reacts with 2-(4-nitrophenyl)ethan-1-amine].
Mass spectrometry.
Conventional mass spectrometry analyses were performed on a hybrid quadrupole time-of-flight (Q-TOF) analyzer (QSTAR Pulsar I; AB Sciex, Framingham, MA, USA). Reaction samples were analyzed by direct infusion and ionized by electrospray ionization-mass spectrometry (ESI-MS) with methanol as the mobile phase in positive reflector mode. High-resolution mass spectrometry (HR-MS) analysis was carried out by flow injection analysis combined with electrospray ionization-mass spectrometry (FIA-ESI-MS) on an Agilent G6530A accurate-mass Q-TOF liquid chromatography-mass spectrometry (LC-MS) system (Agilent Technologies, Santa Clara, CA, USA). The sample was directly infused and ionized by ESI in negative reflector mode. Ionization was enhanced by JetStream technology, and the mobile phase was 99.9:0.1 (vol/vol) H2O–formic acid. Data were processed with MassHunter Data Acquisition B.05.01 and MassHunter Qualitative Analysis B.07.00 software (Agilent Technologies).
GC analysis for determination of chiral selectivity.
Enantioselectivity was evaluated by kinetic resolution of (R) and (S) amines. Prior to the kinetic resolution assay, a solution of 25 mM benzaldehyde as the acceptor and 1.0 mM PLP as a cofactor was first prepared in 100 mM K2HPO4 buffer, pH 7.5 (40 ml). A stock solution of a racemic mixture of (R)- and (S)-aminononane at a concentration of 400 mM each in acetonitrile was prepared. Reaction assays, in 5.0-ml Eppendorf tubes, were started by adding 25 μl of a protein solution (stock solution, 10.0 mg/ml in 100 mM K2HPO4 buffer, pH 7.5) to an assay mixture containing 1,850 μl of PLP-benzaldehyde solution and 125 μl of racemic (R)-/(S)-aminononane stock solution. The final volume of the assay mixture was 2,000 μl, and the concentrations of (R)- and (S)-aminononane and benzaldehyde were 25 mM each. Reactions were allowed to proceed at 40°C for 60 min. Next, the reaction mixture was filtered through an adsorptive, hydrophilic, 3K-nominal-molecular-weight-cutoff membrane (regenerated cellulose, Amicon) to remove the enzymes. Then, 10 μl of a stock solution of (R)-2-aminohexane was added as an internal standard to take into consideration biases due to the extraction procedure, as follows. To 0.2 ml of the reaction mixture, 0.2 ml ethyl acetate was added, and after vigorous vortexing, the solvent used for GC-MS analysis. The GC system (Agilent Technologies 7890A) consisted of an autosampler (Agilent Technologies 7693) and an inert MSD (mass selective detection) instrument with quadrupole (Agilent Technologies 5975). A total of 2 μl of the sample was injected through a CP-Chirasil-Dex CB GC column (25 m in length, 0.25-mm internal diameter, 0.25-μm film) (J&W GC columns; Agilent). The flow rate of the helium carrier gas, the split ratio, and the temperature gradient were optimized for each of the chiral mixes. After each injection, the column was cleaned up during 2 min at 200°C using a 1.5-ml/min flow rate. The detection of each chiral compound [(R)- and (S)-aminononane] was performed in single-ion-monitoring (SIM) mode in order to maximize the sensitivity. The elution order and the target ions were previously validated with a mixture of standards and the NIST 14 library. The semiquantification of (R)- and (S)-aminononanes was performed using MassHunter Qualitative Analysis software (B.06.00; Agilent), reporting the area for the individual peaks in arbitrary units, on the basis of which enantiomeric excess and conversion were calculated.
Stereochemistry was also calculated by following the asymmetric synthesis of (R)- and (S)-aminononane. Asymmetric synthesis assays were performed at 40°C for 180 min in 100 mM K2HPO4 buffer, pH 7.5, containing 1 mM PLP and 25 μg pure protein. The reaction mixture contained 25 mM 2-nonane as the acceptor and 25 mM 2-(4-nitrophenyl)ethan-1-amine as the amine donor. The conversion was measured by detection of the formed amines (R)-aminononane and (S)-aminononane by GC after extraction of the reaction products as described above. The percent-enantiomeric-excess values for products were analyzed by GC. Note that conversions were not optimized.
Enzyme assays for determinations of optimal parameters for activity.
Temperatures between 25 and 70°C were tested to determine the conditions under which each protein displayed maximal activity in 100 mM K2HPO4 buffer, pH 7.5. Assays of activity in the presence of a water-miscible solvent were performed by adding methanol, acetonitrile, dimethyl sulfoxide, dimethyl acetamide, isopropanol, or acetone at concentrations from 5% to 50% (vol/vol). Benzaldehyde and 2-(4-nitrophenyl)ethan-1-amine were used as the acceptor and donor, respectively. The standard assay conditions to determine optimal temperatures and activities in the presence of solvents were as for determinations of acceptor substrates (see above).
Homology modeling and docking simulations.
Homology models were developed using Prime software from Schrödinger. Prime uses BLAST (with the BLOSUM62 matrix) for homology search and alignment and refines the results using the Pfam database and pairwise alignment with ClustalW. Docking simulations of (
S)-(+)-2-aminononane and (
R)-(−)-2-aminononane with the structural models created for each of the class III ω-TAs were carried out using the Protein Energy Landscape Exploration software, which offers one of the best modeling alternatives to map protein-ligand dynamics and induced fit, as described previously (
48). The substrate was initially positioned in the active site with the nitrogen atom of the substrate toward the Lys catalytic base. The substrate conformation was set to be fully flexible in the docking simulations, whereas the protein conformation was not allowed to change.
Genetic enzymology analysis.
Sequence similarity networks (SSNs) were generated by using the Enzyme Function Initiative’s Enzyme Similarity Tool (EFI-EST) (
51). All-vs-all BLAST was performed against the first 500 BLAST hits in UniProt (2018_06) for each query sequence (option A). A negative LogE value was applied for the initial network generation. The network consists of the nodes representing protein clusters with >60% sequence identity. The network was visualized in Cytoscape version 3.3.0 (
52), using the organic layout in which the lengths of the edges correlate with the dissimilarity of the connected sequences represented by the nodes. The subgroupings in each major cluster were visualized by gradually increasing the stringency of the LogE filter of the networks. The final network was used to build genome neighborhood networks (GNNs) in EFI’s Genome Neighborhood Network Tool (EFI-GNT). Initial default values of a 10-ORF window and 20% cooccurrence were chosen, although the values were eventually narrowed to a 5-ORF window and 20% cooccurrence. The GNT database uses the updated UniPro 2018_06 and ENA 136 versions. Accession numbers within relevant Pfam nodes were extracted and used for building a new SSN using option D. The published functional data were used to determine the consensus function and substrate preference of each subfamily of protein sequences.
Accession number(s).
The sequences were named based on the code TR, which refers to
transaminase, followed by an arbitrary number representing 1 of the 10 enzymes analyzed. Sequences encoding the enzymes designated TR
1 to TR
8 were deposited at the NCBI public database under accession numbers
MF158200,
MH588437,
MF158202,
MF158203,
MF158204,
MF158205,
MF158206, and
MF158207. Sequences encoding TR
9 and TR
10 are available as part of the genome sequence of
P. oleovorans (accession numbers
NZ_NIUB01000001 and
NZ_NIUB01000017).
ACKNOWLEDGMENTS
This project has received funding from the European Union’s Horizon 2020 research and innovation program (Blue Growth: Unlocking the Potential of Seas and Oceans) through Project INMARE under grant agreement no. 634486 and ERA-IB 5 METACAT. This work was further funded by grants no. PCIN-2014-107 (within the ERA NET IB2 grant no. ERA-IB-14-030—MetaCat) and PCIN-2017-078 (within the Marine Biotechnology ERA-NET [ERA-MBT], funded under the European Commission’s Seventh Framework Program, 2013 to 2017, grant agreement no. 604814) and BIO2014-54494-R and BIO2017-85522-R from the Ministerio de Ciencia, Innovación y Universidades, formerly Ministerio de Economía, Industria y Competitividad. The present investigation was also funded by grant no. BB/M029085/1 from the UK Biotechnology and Biological Sciences Research Council (BBSRC). P.N.G. and R.B. acknowledge the support of the Supercomputing Wales project, which is partly funded by the European Regional Development Fund (ERDF) via the Welsh Government. O.V.G. and P.N.G. acknowledge the support of the Centre of Environmental Biotechnology Project funded by the European Regional Development Fund (ERDF) through the Welsh Government. We gratefully acknowledge financial support provided by the European Regional Development Fund (ERDF). C.C. thanks the Spanish Ministry of Economy, Industry and Competitiveness for a Ph.D. fellowship (grant no. BES-2015-073829).
We acknowledge Bayer AG for kindly providing some of the amines tested. We also acknowledge María J. Vicente at the Servicio Interdepartamental de Investigación (SIDI) from the Autonomous University of Madrid for the ESI-MS analyses.
C.C., N.K., J.N.-F., D.A., M.M.-M., A.B., and M.F. contributed to protein purification and characterization. C.C. and M.F. contributed three-dimensional modeling. C.G., T.N.C., H.T., O.V.G., M.M.Y., P.N.G., and K.-E.J. contributed to sample collection and library construction. R.B., A.G.-M., G.E.K.B., and M.F. performed sequence, phylogenetic, and genomic analyses. N.K., A.G.-M., A.B., and G.E.K.B. contributed to gene cloning, synthesis and expression. D.R. and C.B. carried out gas chromatography analyses. R.K. contributed suggestions for activity tests and provided substrates. M.F., K.-E.J., and P.N.G. conceived the work. M.F. wrote the initial draft of the manuscript.