1. The optimization of reaction conditions is ineffective in achieving the reported LadA activity

LadA belongs to the group C FPMO family [22]. It is the monooxygenase component of a two-enzyme system that requires molecular oxygen (O₂) and the reduction of its flavin mononucleotide (FMN) cofactor. Like other group members, LadA cannot reduce FMN and therefore relies on the activity of an independent enzyme -a FMN reductase- that reduces FMN to FMNH₂ using NAD(P)H as a cofactor. The FMNH₂ is then transferred to the monooxygenase component via protein-protein interaction or free diffusion [23]. Then, the FMNH₂ reacts with O₂ to produce the catalytic intermediate, a flavin hydroperoxy species, responsible for the oxygenation of the organic substrate. Therefore, the monooxygenase and reductase components are essential for the reaction, along with NAD(P)H and FMN as cofactors.

Three previous studies from the same research group have reported the biochemical activity of one LadA enzyme [14,20,21]. The activity was assessed by detecting the hydroxylated product of hexadecane (C16), 1-hexadecanol, using gas chromatography-mass spectrometry (GC-MS). In the study by Feng et al. (2007), the reaction setup included 1 mM hexadecane and 1 mM MgSO₄ in a 50 mM Tris-HCl buffer at pH 7.5, along with 1 mM NADH. The second study by Li et al. (2008), which focused on the structural analysis of LadA, also used 1 mM hexadecane in a 50 mM Tris-HCl buffer at pH 7.5; however, NADH was omitted, and 1 mM FMNH₂ was added instead. Finally, the study by Dong et al. (2012) prepared the reactions with 1 mM FMN and 10 mM NADPH in a 50 mM Tris-HCl buffer at a pH of 8.0. This study also employed a commercial surfactant, Plysurf at 0.015% (m/v), to facilitate hexadecane solubilization. In the three studies, details on the enzyme concentration used were not provided, and none of the studies mentioned adding any source for generating FMNH₂.

We successfully expressed and purified the same LadA protein and attempted to replicate the previously reported assay conditions, as they were ambiguously described and lacked a clear biochemical rationale (S1 Fig). Most notably, the omission of a flavin reductase despite including FMN and NADH.

We then focused on establishing more mechanistically grounded conditions. To this end, we generated FMNH₂ in situ by supplementing the reaction with purified E. coli flavin reductase (Fre), maintaining a 1:3 ratio relative to LadA (Fig 1A).This setup was successfully tested for other group C FPMOs [23–25]. The stability of Fre at elevated temperatures is undocumented; therefore, we assessed monooxygenase activity at both 37°C, the optimal temperature for Fre, and 60°C, the reported optimal temperature for LadA. [2] Unfortunately, the three previous studies on LadA did not report specific protein concentrations. Other group C FPMOs were typically evaluated at concentrations ranging from 0.01 µM to 90 µM. Thus, we added approximately 1 g/L (18 µM) LadA to our assays, which represented the highest possible concentration before protein aggregation occurred [20,26]. [3] The surfactant Plysurf A210G (octyl phenol ethoxylate) is no longer distributed. Consequently, we employed alternative anionic surfactants for hexadecane solubilization. We used HITENOL-AR10 (DSK, Japan), a surfactant produced by the same company that manufactured the Plysurf A210G, which presumably has similar properties (personal communication with DSK). [4] Finally, while the three previous studies added Mg² ⁺ , we omitted it. LadA lacks metal-binding sites, and no metal dependency has been reported for group C FPMOs [20]. Under these new conditions, we evaluated LadA’s long-chain n-alkane monooxygenase activity. Despite extensive optimization, no hydroxylated products or hexadecane consumption were observed (Fig 1B). Detailed conditions and GC-MS analyses are presented in the methods section and supplementary information (S1 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Protein expression, purification, and GC-MS analyses for LadA and Fre reaction.

(A) SDS-PAGE. Lane 1: Molecular weight (MW) marker. Lane 2: purified wild-type N-strepTag LadA (WP_011888513.1, estimated MW = 51.8 kDa). Lane 3: purified wild-type N-strepTag Fre (WP_000209826.1, estimated MW = 27.5 kDa). (B) The gas chromatogram showed no hexadecane (t_R = 12.74 min) conversion after two hours of incubation at 60°C. Reaction conditions were wild-type N-strepTag LadA (18 μM), wild-type N-strepTag Fre (6 μM), hexadecane (1 mM), FMN (1 mM), NADH (1 mM) and HITENOL-AR 0.5% (m/v). kDa - kilodaltons. Additional peaks at t_R = 11.54 min and t_R = 11.84 min correspond to pentadecane (internal standard) and butylated hydroxytoluene (contamination), respectively.

https://doi.org/10.1371/journal.pone.0332702.g001

2. One optimized LadA and four homologs are ineffective in yielding the activity

We hypothesized that the lack of activity could be related to the enzyme’s instability under such denaturing reaction conditions. So, we enhanced LadA’s robustness using the Protein Repair One-Stop Shop (PROSS) algorithm. This method incorporates atomistic modeling and phylogenetic data to design sequence variants with enhanced stability and solubility [27]. We assessed the expression, purification, and activity of two PROSS variants: LadA:P2 with eight amino acid substitutions, which did not express, and LadA:P5 with 18 amino acid substitutions, which showed a 3-fold increase in expression (S1 Table). We tested LadA:P5 in the optimized assay conditions. However, it was also inactive (S2 Fig).

Next, we tested whether the activity could be detected in other LadA homologs. We picked four LadA proteins with reported activity: the LadA from G. thermodenitrificans NG80−2, the only homologue with biochemical evidence of activity, and three LadA homologs from G. thermoleovorans B23 (LadB, LadAɑ, and LadAβ) for which the evidence was microbiological. These three homologues enhanced the growth of Pseudomonas fluorescens KOB2Δ1 on n-alkanes. P. fluorescens KOB2Δ1 lacks the AlkB enzyme, hence it grows significantly slower than the wildtype on n-alkanes as a carbon source [28,29]. The four protein sequences were used as queries to search across 215,713 bacterial proteomes in the NCBI RefSeq database (August 2021). This search yielded 660 homologs with >50% global sequence identity and >70% coverage to at least one of the query sequences. We then organized these sequence homologs into a maximum-likelihood cladogram and divided it visually into eight clades (Fig 2A). We generated a consensus sequence (cutoff >50%) for each clade by aligning all corresponding sequences. From each clade, we then selected the closest similar sequence to the consensus sequence (Fig 2B, S2 Table), which is likely to result in increased structural and functional stability of the respective proteins [30]. Further, we verified that the chosen sequences contained the conserved FTNVAF motif previously proposed to be important for interaction with the FMN cofactor and modulation of its oxygen-reactivity [31,32]. The presence of this motif thus suggests similar flavin binding features and catalytic properties to those in the query sequences. We then successfully expressed and purified four out of the eight representative sequences (S3 Fig). None of these proteins, however, demonstrated detectable long-chain n-alkane monooxygenase functionality under our assay conditions (S4 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. LadA sequence homologs.

(A) Cladogram displaying the organization of 660 putative LadA homologs. (B) Pairwise global sequence identity comparison of LadA query sequences and the eight homologs we tested. The red arrows indicate the only biochemically characterized sequence, wild-type LadA (WP_011888513.1). The homologues we managed to express, purify, and perform biochemical tests are shown in bold letters.

https://doi.org/10.1371/journal.pone.0332702.g002

3. Exploring other plausible roles for LadA in hydrocarbon metabolism

The available microbiology and genetic data suggest that LadA activity is indeed related to medium- and long-chain n-alkane metabolism: (1) ladA’s gene transcript was upregulated when culturing G. thermodenitrificans NG80−2 in crude oil, a complex mixture of hydrocarbons, including medium- and long-chain n-alkanes; (2) In a proteomic analysis, the LadA protein from the same organisms was overexpressed when growing in hexadecane as the carbon source compared to sucrose. (3) The heterologous expression of LadA in Pseudomonas fluorescens KOB2Δ1 enhanced its growth on n-alkanes. P. fluorescens KOB2Δ1, lacking the AlkB gene, grows more slowly than the wild type on n-alkanes as the carbon source. The fact that LadA enhanced the growth rate could indicate that this enzyme performs the same function as AlkB but could also be reducing a bottleneck in a downstream metabolic step [12]. Since we did not detect n-alkane hydroxylation, we attempted to test other hypotheses connecting LadA activity to n-alkane or oil-related hydrocarbon metabolism outside the first step in metabolism, the hydroxylation.

A similar observation was recently reported for the enzyme family AlmA, which was initially thought to hydroxylate n-alkanes but was later recognized as acting downstream on the metabolic pathway by converting aliphatic ketones to esters via a Baeyer-Villiger monooxygenation (BVMO) [19]. BVMO activity has been reported for three other group C FPMO members, like LadA. Among them, 2,5-diketocamphane monooxygenase (2,5-DKCMO, no significant sequence identity to LadA) and 3,6-diketocamphane monooxygenase (3,6-DKCMO, 25% sequence identity to LadA), both related to camphor metabolism, and catalyzing the insertion of an oxygen atom adjacent to the carbonyl group of 2,5- and 3,6-diketocamphane, respectively [33]. Additionally, luciferases (LuxAB, 25–37% sequence identity to LadA) from Photobacterium phosphoreum NCIMB 844 demonstrated activity on aliphatic, monocyclic, and bicyclic ketones [34]. Thus, we tested if LadA could show BVMO activity on 2-hexadecanone as the substrate. However, we could not detect ester formation or 2-hexadecanone depletion (S5 Fig).

Moreover, a functionally characterized relative of LadA is the group C FPMO DszA from Rhodococcus erythropolis D1 (35% sequence identity to LadA) that functions as dibenzothiophene sulfone monooxygenase [35]. Since transcriptional analysis of the ladA gene showed overexpression while exposing the bacteria to crude oil containing sulfur compounds, we hypothesized that LadA might be active on dibenzothiophene sulfone, the most abundant sulfur compound in crude oil mixtures [2]. However, we could not detect any LadA activity on dibenzothiophene sulfone under the conditions reported for DszA (S6 Fig). The specific role of LadA in long-chain n-alkane or hydrocarbon metabolism, therefore, remains to be elucidated.

LadA, sequence homologs and DszA expression and purification

The nucleotide sequences of LadA, the PROSS designs, selected LadA homologs and DszA were codon-optimized with custom python scripts for expression in E. coli. The sequence of these proteins can be found in supplementary information, S1 Table. The optimized sequences were synthesized, cloned into the expression vector pET-28a(+) with an N-terminal StepTag added for affinity purification, and sequenced by Twist Biosciences (San Francisco, CA, USA). The constructs were transformed into One Shot™ BL21 Star™ (DE3) chemically competent E. coli cells (Thermo Fisher Scientific), plated on LB-Kan agar (kanamycin 50 mg/mL), and incubated overnight at 37°C.

A positive transformant colony was picked and grown overnight in 5 mL of LB-Kan to an optical density (OD) of 1.0-1.2, then used to inoculate a 1000-mL flask containing 500 mL of LB-Kan to an OD of 0.05. The culture was incubated at 37°C until the OD reached 0.4-0.6, at which point protein overexpression was induced with isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM. The culture was further incubated at 30°C for approximately 16 hours, except for DszA which was incubated at 15°C for 16 hours. Cells were harvested by centrifugation at 15,400 x g for 20 minutes at 4°C. The supernatant was discarded.

The cell pellet was resuspended in 10 mL of 50 mM TRIS-HCl buffer (pH 7.5) supplemented with Benzonase® Nuclease (Merck Millipore, Cat# 70746), EZBlock™ Protease Inhibitor Cocktail (EDTA-Free) (BioVision), and chicken egg white lysozyme (Sigma-Aldrich) (1 mg/mL). The cells were lysed by incubating on ice for 20 minutes, followed by sonication for 7 minutes (5 seconds pulse on, 5 seconds off, 60% of maximum power). Cell debris was removed by centrifugation at 15,400 x g for 20 minutes at 4°C. The supernatant was loaded onto a gravity column containing streptavidin StrepTag® resin (IBA Lifesciences GmbH) for affinity chromatography. The column was washed with 6 column volumes of washing buffer (100 mM Tris-HCl, pH 8, 150 NaCl, 1 mM EDTA) and elution was performed in six steps with 0.5 column volume each of washing buffer supplemented with 25 mM desthiobiotin. The purity of the eluted fractions was evaluated by electrophoresis using a 12.5% polyacrylamide gel. Total protein content was estimated by the coomassie-blue quantification method, through a calibration curve using bovine serum albumin as a standard protein. The typical yield per 500 mL of culture was 1.0–1.5 mg for LadA, 3.0–4.0 mg for LadA:P5, 1.5–2.0 mg for DszA, 0.1–0.5 mg for WP_149412683.1, 0.5–1.0 mg for WP_174015212.1, 0.1–0.5 mg for WP_088236120.1 and 0.8–1.0 mg for WP_023806228.1.

Fre expression and purification

Flavin reductase (NCBI ID: WP_000209826.1) was prepared as standard molecular biology methods [24,32]. Briefly, the DNA sequence for Fre was amplified by PCR using E. coli K12 genomic DNA as a template with oligonucleotide primers Fre_F and Fre_R (S3 Table). The backbone was amplified by PCR from an empty pET28a(+) vector containing a StrepTag on the N-terminal end with primers pET28_F and pET28_R (S3 Table). The reactions were performed using Q5 polymerase according to the manufacturer’s instructions (New England Biolabs). Then, the two amplified fragments were purified with QIAquick PCR Purification Kit (QIAGEN Labs, US) and assembled into a single construct with NEBuilder HiFi DNA Assembly reaction as indicated by the manufacturer (New England Biolabs). The construct was transformed into NEB 5-alpha competent E. coli, a derivative of DH5α (New England Biolabs,), plated on LB-Kan agar (kanamycin 50 mg/mL), and incubated overnight at 37°C. The functional plasmids were extracted with QIAprep Spin Miniprep Kit (QIAGEN Labs) from the overnight culture of the positive colonies and further extracted for sequence corroboration. The whole-plasmid sequencing was performed by Plasmidsaurus using Oxford Nanopore Technology with custom analysis and annotation. Protein expression and purification was carried out as earlier stated. Modifications to the protocol were the induction conditions, where IPTG was added to a final concentration of 0.25 mM, and cultures were incubated at 16°C for 4 hours. A typical yield ranges from 0.5–1.0 mg of pure protein per 300 mL of culture.

Enzymatic activity assessment with hexadecane, 2-hexadecanone and dibenzothiophene sulfone

Reactions were prepared by mixing purified LadA and homologs (1–18 µM) with purified Fre (0.3-6 µM) in 50 mM Tris-HCl buffer, pH 7.5, containing 1 mM FMN, 1 mM hexadecane or 0.5 mM 2-hexadecanone, and 0.5% (w/v) HITENOL-AR. The total reaction volume was  mL, and the mixture was contained in a closed 4-mL amber glass vial equipped with a magnetic stirrer. Reactions were initiated by adding NADH to a final concentration of 1 mM and incubated at either 37°C or 60°C, in a water bath with stirring at 200 rpm.

Reactions for dibenzothiophene sulfone monooxygenase were prepared by mixing purified LadA:P5 or DszA (50 µM) with purified Fre (0.2 µM) in 50 mM Tris-HCl buffer, pH 7.5, containing 1 mM FMN and 0.5 mM dibenzothiophene sulfone. The total reaction volume was 3 mL, and the mixture was contained in a closed 4-mL amber glass vial equipped with a magnetic stirrer. Reactions were initiated by adding NADH to a final concentration of 1 mM and kept at 37°C, in a water bath with stirring at 200 rpm.

Every reaction was run for at least two technical replicates using two independently purified protein batches.

GC-MS analysis

The reactions were analyzed at two time points: the initial stage (time zero, t₀) and two hours after the addition of NADH to the reaction medium. One milliliter of reaction mixture was sampled, and hydrophobic compounds were extracted three times with hexane (3 × 1 mL). Pentadecane was used as an internal standard. The combined organic layers were dried at room temperature. The resulting residue was dissolved in 1 mL of hexane, and 1 µL of this solution was injected into a gas chromatography-mass spectrometry (GC-MS) system (Agilent 7890B GC system coupled with a 5977B MSD). The separation was achieved using an HP-5ms column (30 m × 0.25 mm i.d., 0.25 µm film thickness), with helium as the carrier gas at a flow rate of 1.0 mL/min. The oven temperature was programmed from 70°C (hold for 0.5 min) to 280°C at a rate of 10°C/min, with a final hold time of 3 minutes. Mass spectra were recorded in electron impact (EI) mode at 70 eV. Compound identification was carried out by comparing mass spectra with spectral libraries and confirmed by matching retention times with standards.

HPLC analysis

The analysis of the reactions was performed by HPLC as previously described, using a Thermo Fisher Vanquish Core 700 bar HPLC system equipped with a diode array detector (DAD) and a Luna C18 column (2.1 × 150 mm, 2.2 µm) [39].

LadA homologous sequences selection

To identify homologous sequences of LadA, we queried the NCBI RefSeq database (August 2021) using pBLAST. The search aimed to retrieve sequences with significant similarity to the LadA enzyme from Geobacillus thermodenitrificans NG80−2 and the three homologs from G. thermoleovorans B23. Retrieved sequences were filtered based on 70% sequence coverage and 50% identity thresholds to ensure the inclusion of relevant homologs.

The retrieved sequences were aligned using MAFFT with default parameters to ensure accurate multiple sequence alignment. A cladogram was subsequently generated using IQ-TREE, employing default settings. This approach allowed us to visually group homologous sequences into distinct clades for downstream analysis.

To select representative sequences from each clade, a consensus sequence approach was applied. First, a consensus sequence was constructed for all sequences within a clade using alignment data. This consensus sequence was then used as a query for a BLAST search against the sequences in the same clade. The sequence with the highest identity to the consensus was selected as the representative sequence for that clade.