The catabolism of lignin-derived p-methoxylated aromatic compounds by Rhodococcus jostii RHA1

All reagents were of analytical grade unless otherwise noted. Reagents were obtained from Sigma-Aldrich and Fisher Scientific. Media were prepared using water purified on a Barnstead NANOpure UV apparatus to a resistivity of greater than 16 MΩ/cm.

Loblolly pine woodchips were subject to RCF as described previously (45). The resultant oil was methylated by stirring with excess methyl iodide and potassium carbonate in acetonitrile. Following workup, the methylated pine RCF oil was separated into two fractions by vacuum distillation, affording an oil distillate and a solid oligomeric substrate that remained in the stirred and heated flask. The solid was dissolved in ethyl acetate and filtered, and the volatiles were removed by rotary evaporation affording methylated pine RCF oligomers as a brown solid.

Methylated lignin oligomers were oxidized as described previously (7). Briefly, 75 mL Parr batch reactor fit with a glass liner insert was charged with 65 mg of methylated pine RCF oligomers, acetic acid (15 mL), a stir bar, and Mn(OAc)₂•4H₂O (5.2 mg) and Zr(acetylacetonate)₄ (7.8 mg) catalyst. The mixture was pressurized three times with pure nitrogen and subsequently charged with air (29 bar) and dinitrogen (31 bar), achieving 6 bar total pressure of O₂ diluted to 10%. The vessel was heated to 150°C. Once at temperature, it was left to react for 1.5 h and subsequently cooled in an ice bath. Aliquots of 10–12 mL were taken from the reactions, and the acetic acid was removed under a stream of nitrogen, leaving an oil residue, which was used as a growth substrate. Gas chromatography-flame ionization detection (GC-FID) quantification of the monomers in the two different preparations is shown in Fig. S8.

RHA1 was grown in M9 medium supplemented with goodies (M9G) (46) and the indicated carbon substrate for 2 days shaking at 30°C. Growth substrates were solubilized in dimethyl sulfoxide (DMSO) and added to media at appropriate concentrations. The pellet was then harvested and washed twice with M9G and added to M9G containing the experimental carbon substrate unless otherwise indicated. For the ΔpbdA and ΔpcaL growth experiment, cells were precultured in 10 mM benzoate and grown in 150 µL media in a round-bottomed 96-well (Corning) plate shaking continuously in a Tecan Spark microplate reader. For growth on the methylated lignin stream, cells were grown in 5 mM veratrate, and 10-mL cultures were grown in 50 mL flasks. optical density (OD₆₀₀) was measured using a Cary 60 spectrophotometer. The methylated lignin substrate was introduced into M9G at a concentration of 10% vol/vol. Briefly, the methylated lignin product was treated with NaOH to precipitate the metal catalyst and neutralized with HCl. The substrate was filtered to remove particulate matter and sterilized prior to addition into growth media. For the wild-type (WT) growth experiment on veratraldehyde, cells were precultured in 3 mM veratraldehyde and grown in 25 mL M9G containing 3 mM veratraldehyde in 125-mL flasks. OD₆₀₀ was measured using a Biochrom WPA Biowave II UV/visible spectrophotometer. For the ΔpbdA and ΔvanA intermediate accumulation experiment, cells were precultured in 5 mM benzoate and then grown in 25 mL of M9G containing 1 mM benzoate until the cultures had no significant change in OD₆₀₀ (stationary phase), and veratrate was then added to 0.8 mM. OD₆₀₀ was measured using a Biochrom WPA Biowave II UV/visible spectrophotometer. For the Δvdh experiment, the mutant was precultured in 1 mM benzaldehyde while WT and the complement were precultured in 1 mM veratraldehyde. Cells were then grown in 500 µL M9G containing 3 mM veratraldehyde in a 48-well plate shaking continuously at 30°C. OD₆₀₀ was measured using a Tecan Infinite M200 microplate reader. For the ΔpbdB and ΔpobA experiments, cells were precultured in 5 mM benzoate and grown in 500 µL M9G containing 5 mM p-MBA and p-HBA, respectively, in a 48-well plate. OD₆₀₀ was measured using a Tecan Infinite M200 plate reader. CFU/mL was measured by plating dilutions of cell cultures in saline on lysogeny broth (LB) agar plates, and counting colonies after incubation for 2 d at 30°C.

Liquid chromatography-mass spectrometry (LC-MS) analyses were performed using an Agilent 1290 Infinity II UHPLC in line with an Agilent 6546 Q-TOF equipped with a dual AJS ESI source operating in positive ionization mode. A sample (2 µL) was injected onto a Zorbax Eclipse Plus C18 Rapid Resolution HD, 2.1 × 100 mm 1.8 µm column and run on a 20 min linear gradient from 5% to 100% solvent B at 0.45 mL min⁻¹. Solvent A was 0.1% formic acid in water, and solvent B was 0.1% formic acid in methanol. MS parameters were as follows: capillary voltage, 3,500 V; nozzle voltage, 500 V; drying gas temp, 300°C; drying gas flow rate, 10 L min⁻¹; sheath gas temperature, 350°C; sheath gas flow rate, 12 L min⁻¹; nebulizer pressure, 45 psi; fragmentor voltage, 100 V. Data were collected using MassHunter Workstation LC/MS Data Acquisition version 10.1 and analyzed using MassHunter Workstation Qualitative Analysis version 10.0. Concentrations were determined by interpolation on a standard curve of 0–200 μM.

GC-FID analyses were performed using an Agilent Technologies 8890 autosampler. GC-MS analyses were performed using an Agilent Technologies 7890A autosampler equipped with a 5975C inert XL MSD with a Triple-Axis Detector. The GC-MS was operated with the same temperatures, injection volumes, and programmed temperature ramps as those used for GC-FID (see below) to enable comparable retention times. A 1-µL injection was used with a split ratio of 10:1. A 30 m × 250 µm × 0.25 µm Agilent Technologies HP-5ms column was used. The inlet temperature was set to 260°C. The oven was initially held at 50°C for 2 min and then ramped at 40°C min⁻¹ until 100°C. The ramp was then changed to 5°C min⁻¹ until 110°C where it was held for 5 min. The method continued to ramp at 5°C min⁻¹ until 160°C where the temperature was held for 2 min. The temperature was then ramped at 25°C min⁻¹ until 300°C where it was held for 2 min, reaching a total duration of 29.85 min. A flame ionization detector was used to quantify the products, and calibration curves were generated for the oxidation products based on peak area ratios between analyte standards and a naphthalene internal standard. The analyte mixture was obtained as described previously (7).

Plasmids and primers used in this study are listed in Table S1 and S2, respectively. DNA was propagated, purified, and manipulated using standard procedures (47). The pbdA, pbdB, and pobA genes were deleted using sacB counterselection as described previously (48) to generate the ΔpbdA, ΔpbdB, and ΔpobA strains, respectively (Table 1). Flanking regions were amplified using the primers in Table S2 and inserted into p18mobsacB using Gibson assembly. The plasmid was conjugated into RHA1, and mutagenesis was performed by selecting for two successive crossover events. The candidate mutants were screened using colony PCR and screening primers in Table S2 (Fig. S3B). The pbdB and pbdA complementation plasmids were constructed via Gibson assembly using PCR amplicons generated using appropriate primers (Table S2) and linearized pTipQC1 plasmid. The vdh and pobA complementation plasmids were similarly constructed via Gibson assembly using linearized pRIME vector (49). Escherichia coli DH5α was used for all plasmid propagation and storage.

For detection of aromatic monomers, culture supernatants were acidified by adding acetic acid to 10% final concentration, centrifuged (16,000×g for 10 min), and filtered through a 0.2 µm polytetrafluoroethylene (PTFE) membrane. Samples were run over a Luna 5 µm C18 (2) 100 Å 150 × 3 mm column (Phenomenex) at 0.7 mL min⁻¹ by a Waters 2695 separation high-performance liquid chromatography (HPLC) module. The samples were eluted with a 16.8 mL linear gradient from 1% methanol plus 0.1% formic acid in water to 100% methanol plus 0.1% formic acid and were monitored at 280 nm with a Waters 2996 photodiode array detector. Concentrations were determined by interpolation on a standard curve of 0 to 3 mM of authentic standard. For acetate quantification, culture supernatants were acidified by adding sulfuric acid to 20% final concentration and centrifuged (16,000×g for 10 min). Samples were run over an Aminex HPX-87H 250 × 4 mm column at 0.6 mL min⁻¹ at 35°C using a Waters 2695 separation HPLC module with an isocratic 8 mM sulfuric acid mobile phase. Acetate was detected at 210 nm with a Waters 2996 photodiode array detector. Concentrations were determined by interpolation on a calibration curve of 0 to 30 mM of authentic standard.

To investigate the ability of RHA1 to transform a lignin stream derived from autoxidative C–C cleavage of methylated pine RCF oligomers, we first evaluated the growth of the strain on minimal media supplemented with this substrate at a 10% vol/vol concentration. Under these conditions, RHA1 grew robustly as measured by OD₆₀₀ (Fig. 2; Fig. S1) and CFU (Fig. S2A and B). The results were replicated with two separately oxidized lignin substrates that contained final concentrations in the growth media of up to 400 µM veratrate, 68 µM veratraldehyde, and 19 µM p-MBA (Table 2). Acetate, derived from the acetic acid solvent of the autoxidation step, was present at up to 10 mM. To determine whether RHA1 catabolized the components of the methylated lignin stream, we measured the concentration of constituent compounds during growth. As shown in Figure 2, RHA1 consumed all of the veratrate, veratraldehyde, p-MBA, and acetate within 96 h of inoculation. Although growth was monophasic, consumption of compounds was successive and overlapping. The aromatic acids veratrate and p-MBA were consumed first, followed by acetate and, lastly, veratraldehyde.

As RHA1 efficiently catabolized the p-methoxylated benzoates, p-MBA, and veratrate, in the methylated lignin stream, we interrogated the genetic determinants of the catabolic pathway. Of particular interest was the p-O-demethylation step, which is essential for further catabolism and for which no enzymes have previously been identified in RHA1 to our knowledge. As several P450s catalyze the O-demethylation of LDACs (41, 50), we searched the P450 contingent of RHA1 for candidate enzymes. The RHA1 gene RS14355 encodes a P450 in the CYP199A subfamily, designated CYP199A3. CYP199A3 shares at least 53% amino acid sequence identity with the other members of the subfamily and shares the highest sequence identity with CYP199A25 (74%) (Table 2) (12–14, 51). As CYP199A25 catalyzes the O-demethylation of p-substituted benzoates, we hypothesized that the P450 encoded by RS14355, annotated here as pbdA, exhibits a similar function. RS14360, annotated as pbdB, is located immediately upstream of pbdA in a putative operon (Fig. S3A) and encodes a predicted PDR-type reductase (Table 3). The putative operon also includes RS14365 directly upstream of pbdB. RS14365 encodes a TetR/AcrR family transcriptional regulator and was therefore annotated as pbdR (Table 3) (52).

To determine the role of pbdAB in the catabolism of p-MBA and veratrate, we constructed deletion mutant strains of each gene and evaluated their growth on these compounds. The ΔpbdA strain did not grow on p-MBA or veratrate (Fig. 3A and C), but grew as WT RHA1 on the O-demethylated products, p-HBA and vanillate, respectively (Fig. 3B and D). There was a small increase in OD₆₀₀ when we tested the growth of the ΔpbdA strain on p-MBA and p-HBA (Fig. 3A and B). However, endpoint CFU/mL excluded the possibility that this represented genuine growth (Fig. S4). A complement strain harboring the pTip plasmid expressing pbdA fully rescued wild-type growth on p-MBA and veratrate (Fig. 3A and C). Similarly, the ΔpbdB strain did not grow on p-MBA (Fig. 4), but grew on benzoate (Fig. S5), which is metabolized through the catechol branch of the β-ketoadipate pathway. The wild-type phenotype was rescued by complementing the ΔpbdB strain with the pTip plasmid expressing pbdB. Together, these results indicate that both PbdA and PbdB are necessary for p-MBA catabolism. Moreover, ΔpbdA grew robustly on p-HBA and vanillate, the p-O-demethylated counterparts of p-MBA and veratrate, consistent with PbdAB catalyzing O-demethylation specifically at the para position.

To determine how p-MBA and veratrate are further catabolized to central metabolites, we tested the ability of the previously constructed ΔpcaL mutant to grow on p-MBA and veratrate, as well as their p-O-demethylated counterparts, p-HBA and vanillate. PcaL is a key enzyme in the β-ketoadipate pathway where it functions both as a γ-carboxy-muconolactone decarboxylase and β-ketoadipate enol-lactone hydrolase, and is the point of convergence of the protocatechuate and catechol arms of the pathway (Fig. S6B) (42). As expected, the ΔpcaL mutant did not grow on any of the four tested aromatic substrates, but grew robustly on acetate, which is not metabolized through the β-ketoadipate pathway (Fig. S6A). As described for the ΔpbdA mutant, the inability of the ΔpcaL mutant to grow on the aromatic compounds was confirmed by monitoring CFU (Fig. S4). These data indicate that p-MBA and veratrate are catabolized through the β-ketoadipate pathway.

Catabolism of veratrate through the protocatechuate arm of the β-ketoadipate pathway requires O-demethylations at its para and meta positions. To determine the order of the para- and meta-O-demethylation reactions catalyzed by PbdAB and VanAB, respectively, we grew the ΔpbdA and ΔvanA mutants to stationary phase in media supplemented with benzoate, and then we spiked in veratrate and monitored the metabolite profiles. Neither the ΔpbdA mutant nor the previously constructed ΔvanA mutant (40) exhibited additional growth on veratrate (Fig. 5). When veratrate was spiked into the growth medium of the ΔpbdA strain, isovanillate accumulated in the cell supernatant (Fig. 5B). Conversely, when veratrate was added to the ΔvanA strain, vanillate accumulated in the cell supernatant (Fig. 5C). These results support the hypothesis that PbdA catalyzes the p-O-demethylation of veratrate and VanA catalyzes the m-O-demethylation of veratrate (Fig. 5D). Both appear to be highly specific, as the p-O-demethylated product was not detected for ΔvanA and the m-O-demethylated product was not detected for ΔpbdA. Moreover, both enzymes can tolerate a methoxy substituent ortho to the one being removed. However, only vanillate was transiently detected when veratrate was added to wild-type cells (Fig. 5A).

In many genera including Rhodococcus, the hydroxylation of p-HBA to protocatechuate is catalyzed by para-hydroxybenzoate-3-hydroxylase (PHBH), a flavin-containing oxygenase, also known as PobA (56). The RHA1 gene RS12415 encodes an enzyme that is predicted to be a PHBH (Table 3) (57). To investigate the role of this enzyme in p-MBA metabolism in RHA1, we constructed a ΔpobA deletion mutant. The mutant did not grow on p-HBA (Fig. 6) but grew on benzoate (Fig. S7). Complementation of the ΔpobA strain using an integration vector bearing pobA rescued growth (Fig. 6). These results indicate that PobA is necessary for catabolism of p-HBA, catalyzing its hydroxylation and the second step of p-MBA metabolism.

Because RHA1 completely consumed the veratraldehyde in the methylated lignin stream, we evaluated the growth of RHA1 on veratraldehyde. RHA1 grew robustly on veratraldehyde and, during this growth, transiently accumulated veratrate, isovanillate, and vanillate (Fig. 7A through C). We hypothesized that Vdh, the aromatic aldehyde dehydrogenase that transforms vanillin to vanillate, also oxidizes veratraldehyde. Consistent with this hypothesis, the previously constructed Δvdh strain of RHA1 (40) did not grow on 2 mM veratraldehyde. Moreover, complementing the mutant with an integrated copy of vdh rescued growth after 90 h (Fig. 8). These results indicate that Vdh is necessary for veratraldehyde catabolism in RHA1.

Using both pine and poplar substrates, Palumbo et al. recently developed a catalytic oxidation method to cleave the C–C bonds in lignin dimers and oligomers that had undergone methyl protection of the phenolic hydroxyl groups (7), thereby enabling aromatic monomer yields from lignin that exceed those accessible from C–O bond cleavage alone. Given the need for phenol stabilization in autoxidation catalysis, the output of this catalytic oxidation process is a slate of methylated aromatic compounds. In the present study, we showed that RHA1 grows on the oxidized pine lignin stream produced using the same approach, and that RHA1 efficiently catabolizes the three major aromatic monomers found in this stream—p-MBA, veratrate, and veratraldehyde—as well as acetate. Targeted gene deletion and metabolite analysis yielded a comprehensive metabolic network for veratraldehyde, veratrate, and p-MBA catabolism in RHA1 (Fig. 9). In this network, PbdAB catalyzes the O-demethylation of p-MBA to p-HBA, which is hydroxylated by PobA to form protocatechuate. Vdh, previously characterized as a vanillin dehydrogenase, also catalyzes the oxidation of veratraldehyde to veratrate. The VanAB RO and PbdAB P450 systems both catalyze the O-demethylation of veratrate at physiologically relevant rates, as revealed by measuring accumulation of metabolites. VanAB and PbdAB also catalyze the O-demethylation of vanillate and isovanillate, respectively, to protocatechuate. Thus, the catabolism of the three major aromatic monomers converges at protocatechuate, which feeds into central carbon metabolism via the β-ketoadipate pathway.

The ability of VanA and PbdA to catalyze the respective meta- and para-O-demethylation of veratrate is consistent with the available biochemical studies of homologous enzymes. VanAB from C. testosteroni and Streptomyces sp. NL15-2K catalyzed the meta-O-demethylation of veratrate at 18% and 56% efficiency with respect to vanillate, respectively (9, 10). With respect to homologs of PbdA, CYP199A4 and CYP199A25 catalyze the transformation of veratrate, although the closely related CYP199A2 apparently does not (13, 14). Specificity has only been reported for CYP199A4, which catalyzes the p-O-demethylation of veratrate at approximately 50% of the rate of p-MBA (14). Two observations suggest that p-O-demethylation is the preferred first step in the catabolism of veratrate. First, the ΔvanA strain accumulated vanillate faster than the ΔpbdA strain accumulated isovanillate (Fig. 5B and C). Second, only vanillate was detected during growth of wild-type RHA1 on veratrate (Fig. 5A). Interestingly, isovanillate and vanillate were detected at near-equivalent concentrations during growth on veratraldehyde. The reason for this is unclear, although transcriptomic studies of RHA1 grown on veratrate and veratraldehyde might illuminate whether this reflects an underlying regulatory mechanism. Alternatively, the detected metabolites might reflect the substrate specificities of the respective O-demethylase systems. Recently, Palumbo et al. engineered P. putida KT2440 (KT2440) to convert veratrate and veratraldehyde from methylated lignin streams to muconate (7). As KT2440 has no native p-O-demethylase active on veratrate, the p-O-demethylase CYP199A4-HaPuR-HaPux system from R. palustris HaA2 was integrated into the genome. However, this strain accumulated isovanillate, arising from the VanAB-catalyzed meta-O-demethylation of veratrate. By contrast, incorporation of the meta-O-demethylase system from C. testosteroni strain BR6020, IvaAB, yielded a strain that did not accumulate isovanillate. These results indicate that the CYP199A4-HaPuR-HaPux system in an engineered biocatalyst has lower activity on isovanillate compared to the activity of the native PbdAB in RHA1. The lower activity of the system may be due to either enzyme specificity or differential expression.

The aromatic ring hydroxylation catalyzed by PobA_RHA1 can also be rate-limiting in the biotransformation of LDACs (22). For example, hydroxylation is a bottleneck in the biocatalytic transformation of p-HBA by KT2440 due to the specificity of the PHBH in this strain for NADPH (58, 59). By contrast, PobA_RHA1 is a member of the NADH-obligate PHBH clade (55). In the present work, p-HBA was not detected during growth of RHA1 on the methylated lignin stream. However, the concentration of substrate in these studies was relatively low. Therefore, it is possible that the hydroxylation of p-HBA might be rate-limiting under other conditions, particularly as other steps, such as O-demethylation, were optimized by strain engineering. Ultimately, further studies into the hydroxylation of p-HBA in RHA1 are required as it and related strains are developed as biocatalysts tailored to utilize p-HBA and its chemical precursors such as p-MBA and p-coumaric acid.

The involvement of Vdh, previously characterized for its activity on vanillin, in the dehydrogenation of veratraldehyde is consistent with biochemical studies of aromatic aldehyde dehydrogenases. More specifically, homologs of Vdh catalyze the dehydrogenation of veratraldehyde with up to 87% of the efficiency of vanillin (60–62). However, the enzymes characterized to date have broad specificity, and some have relatively low activity on veratraldehyde. Strikingly, RHA1 grew on concentrations of veratraldehyde up to 5 mM, which is much higher than the maximal concentration of 1 mM vanillin that RHA1 tolerates (40). It is unclear whether Vdh_RHA1 contributes to this toxicity differential. In this respect, it would be insightful to determine the specificity of Vdh for the two compounds. Vanillin is a known antimicrobial agent, the mechanism of which has been attributed to aldehyde-induced DNA, protein damage, and membrane destabilization (63–65). Conversely, veratraldehyde toxicity is poorly described; it is possible that the higher tolerance of RHA1 for this compound reflects the different properties of the two chemicals.

The discovery and characterization of O-demethylation systems facilitate their optimization for the development of biocatalysts for lignin valorization. For example, as a two-component system, PbdAB may be advantageous for biocatalytic applications over the three-component CYP199A systems characterized to date. Moreover, characterizing the substrate range of VanAB and PbdAB, when complemented with structural and biochemical experiments, allows for rational engineering and optimization of these enzymes for biocatalysis. Indeed, previous studies on P450 demethylases have established that they are exceptionally amenable to engineering (50, 66–69). Similarly, new insights into RO structure and mechanism have revealed the engineering potential of these enzymes (70, 71). Such advances suggest the possibility of engineering enzymes like VanA and PbdA to transform the tri-methoxylated compounds generated from hardwood biomass fractionation with methyl protection (7, 72). More generally, elucidating the enzymes that catalyze the key reactions in catabolism of the aromatic monomers in methylated lignin streams expands the genetic and enzymatic toolkit for engineering novel biocatalysts. Furthermore, the robust and efficient catabolism of these compounds by RHA1 highlights the potential of rhodococcal biocatalysts for valorizing lignin.