Identification and Characterization of a Tetramethylpyrazine Catabolic Pathway in Rhodococcus jostii TMP1

Abstract

At present, there are no published data on catabolic pathways of N-heterocyclic compounds, in which all carbon atoms carry a substituent. We identified the genetic locus and characterized key reactions in the aerobic degradation of tetramethylpyrazine in Rhodococcus jostii strain TMP1. By comparing protein expression profiles, we identified a tetramethylpyrazine-inducible protein of 40 kDa and determined its identity by tandem mass spectrometry (MS-MS) de novo sequencing. Searches against an R. jostii TMP1 genome database allowed the identification of the tetramethylpyrazine-inducible protein-coding gene. The tetramethylpyrazine-inducible gene was located within a 13-kb genome cluster, denominated the tetramethylpyrazine degradation (tpd) locus, that encoded eight proteins involved in tetramethylpyrazine catabolism. The genes from this cluster were cloned and transferred into tetramethylpyrazine-nondegrading Rhodococcus erythropolis strain SQ1. This allowed us to verify the function of the tpd locus, to isolate intermediate metabolites, and to reconstruct the catabolic pathway of tetramethylpyrazine. We report that the degradation of tetramethylpyrazine is a multistep process that includes initial oxidative aromatic-ring cleavage by tetramethylpyrazine oxygenase, TpdAB; subsequent hydrolysis by (Z)-N,N′-(but-2-ene-2,3-diyl)diacetamide hydrolase, TpdC; and further intermediate metabolite reduction by aminoalcohol dehydrogenase, TpdE. Thus, the genes responsible for bacterial degradation of pyrazines have been identified, and intermediate metabolites of tetramethylpyrazine degradation have been isolated for the first time.

INTRODUCTION

Pyrazines, monocyclic aromatic rings with two nitrogen atoms in para position, are a class of compounds that occur almost ubiquitously in nature. Various pyrazines can be synthesized both chemically and biologically, including tetramethylpyrazine (TTMP), which is produced by different bacteria (1, 2) or plants (3, 4). However, there is very little information available on the biodegradation of these N-heterocyclic compounds. While bacterial strains able to use various alkyl-substituted pyrazines as a sole carbon and energy source have been isolated and described (5–9), almost nothing is known about the degradation pathways of alkylpyrazines, including tetramethylpyrazine, in which all carbon atoms carry a substituent.

Under aerobic conditions, alkylated pyrazines are metabolized via oxidative degradation, leading to the hydroxylation of the ring at a free position. Rhodococcus erythropolis DSM 6138 and Arthrobacter sp. strain DSM 6137 can use 2,5-dimethylpyrazine as a source of carbon and energy. The catabolism of 2,5-dimethylpyrazine by these microorganisms gives rise to the intermediate metabolite 2-hydroxy-3,6-dimethylpyrazine, which accumulates in the medium, indicating that ring hydroxylation occurs during the initial steps of degradation (5). However, no enzymes involved in this bioconversion were reported in the patent that describes the aforementioned reactions (5).

The 2,5-dimethylpyrazine is also catabolized by another R. erythropolis strain, DP-45, as reported by Rappert et al. (7). The DP-45 strain also grew on a variety of other alkylpyrazines, including 2,3-dimethylpyrazine, 2,6-dimethylpyrazine, 2-ethyl-5(6)-dimethylpyrazine, 2-ethylpyrazine, 2-methylpyrazine, and 2,3,5-trimethylpyrazine (7). As was the case with strains DSM 6138 and DSM 6137, the degradation of 2,5-dimethylpyrazine by DP-45 was accompanied by the accumulation of the intermediate metabolite 2-hydroxy-3,6-dimethylpyrazine, which then disappeared with the release of ammonium into the medium (7). The hydroxylation of 2,5-dimethylpyrazine was mediated by an inducible enzyme, while the enzyme catalyzing the subsequent ring cleavage was shown to be constitutively expressed (7). Based on inhibition studies, it was proposed that the initial hydroxylation was catalyzed by a flavin monooxygenase or a cytochrome P450 monooxygenase, while the ring cleavage required P450 monooxygenase (7). However, the identities of the enzymes remain unknown. It is known, however, that in contrast to the degradation of pyridines, which are also metabolized via ring hydroxylation, the degradation of 2,5-dimethylpyrazine does not depend on molybdenum-containing enzymes (10).

The degradation of trisubstituted pyrazines was demonstrated to follow the same metabolic pattern as disubstituted pyrazines (6). Mycobacterium sp. strain DM-11 oxidized 2,3-diethyl-5-methylpyrazine to an intermediate compound, 5,6-diethyl-2-hydroxy-3-methylpyrazine, which was further degraded with the release of ammonium into the culture medium (6). Other compounds, including 2,3,5-trimethylpyrazine, were also used by strain DM-11 as a sole carbon, nitrogen, and energy source (6). Exposure to 2,3-diethyl-5-methylpyrazine induced the expression of the enzymes involved in its degradation, but these enzymes have not been identified yet (6).

In contrast to partially substituted pyrazines, TTMP cannot be degraded via initial hydroxylation to form hydroxypyrazine, since each of the TTMP ring carbons carries a substituent. Müller and Rappert (8) suggested that the initial step of TTMP degradation may involve ring cleavage. They used cell extracts from a Rhodococcus opacus strain that can use TTMP as a single carbon, nitrogen, and energy source and were not able to detect any intermediates during the degradation of TTMP (8). However, with the exception of the minireview (8), no experimental data have been published to support these findings. Although the TTMP-degrading bacteria have been isolated, neither enzymes catalyzing TTMP biodegradation nor the corresponding genes have been identified in R. opacus so far.

In this study, we report the TTMP catabolic pathway of Rhodococcus jostii strain TMP1, a strain previously shown to use TTMP as the sole source of carbon and energy (9). The genetic tpdA-tpdE locus encoding the proteins required for the initial steps of TTMP biodegradation was identified, and the corresponding genes were cloned and heterologously expressed in a different Rhodococcus sp. strain, thus enabling it to metabolize TTMP. The identification of the intermediate metabolites (Z)-N,N′-(but-2-ene-2,3-diyl)diacetamide (BDNA) and N-(3-oxobutan-2-yl)acetamide (OBNA) was used to characterize the reactions of aerobic TTMP degradation, which allowed us to describe the catabolic pathway of TTMP metabolism in bacteria for the first time.

MATERIALS AND METHODS

Bacterial strains, plasmids, primers, and standard techniques.

The TTMP-degrading bacterium R. jostii TMP1 was previously isolated from a soil sample (9). R. erythropolis strain SQ1 was chosen as the host strain for the expression of recombinant genes in bioconversion experiments. Escherichia coli strain DH5α was used for cloning experiments. The TpdE protein was overexpressed in E. coli strain BL21(DE3). The bacterial strains, plasmids, and primers used in this study are listed in Table S1 in the supplemental material. Standard techniques were used for DNA manipulations (11).

Bacterial growth medium and conditions.

Rhodococcus strains were grown at 30°C with aeration, E. coli strains were grown at 37°C with aeration. R. jostii TMP1 was cultivated either in nutrient broth (NB) (Oxoid) medium or in minimal medium (5 g/liter NaCl, 1 g/liter K₂HPO₄, 1 g/liter NH₄H₂PO₄, 0.1 g/liter MgSO₄, and 0.2 g/liter yeast extract, pH 7.2) supplemented with either TTMP (0.05%) or pyridine (0.05%). For cell suspension and bioconversion experiments, R. erythropolis SQ1 was grown in 1-liter flasks containing 250 ml of NB medium until the culture reached an optical density at 600 nm (OD₆₀₀) of 1.6 to 2.0. Then, cells were collected by centrifugation, washed twice, and resuspended in 10 mM potassium phosphate, pH 7.2, to achieve 4-fold-higher cell density. E. coli strains transformed with recombinant plasmids were grown in NB medium supplemented with either 50 μg/ml ampicillin or 40 μg/ml kanamycin, as required. R. erythropolis SQ1 transformed with recombinant plasmids was grown in the presence of 60 μg/ml kanamycin.

Construction of plasmids.

Total DNA from R. jostii TMP1 cultivated in NB medium was isolated as described previously (12). To create a genomic library, the DNA was digested with HindIII, and the resulting fragments were inserted into the HindIII site of pUC19. The strategies employed to construct the TTMP-inducible green fluorescent protein (GFP)-encoding plasmid, as well as recombinant plasmids carrying different genes of the tetramethylpyrazine degradation locus, are described in Table S1 in the supplemental material. R. erythropolis SQ1 and E. coli strains were transformed with plasmid DNA by electroporation.

Analysis of the protein expression profile induced by TTMP.

R. jostii TMP1 was cultivated in minimal medium supplemented with either TTMP or pyridine; cells were collected by centrifugation and suspended in 50 mM potassium phosphate, pH 7.2. After the addition of silica beads (0.1-mm diameter; 0.5 g/ml), the cells were disrupted by sonication at 750 W for 10 min using a VC-750 ultrasound processor (Sonics & Materials, Inc.). Cell debris was removed by centrifugation at 16,000 × g for 10 min. Proteins were separated on 14% SDS-PAGE gel and visualized by Coomassie blue staining.

MS-MS analysis.

R. jostii TMP1 was cultivated in minimal medium supplemented with TTMP. Cells were collected, and proteins were separated on an SDS-PAGE gel. The band corresponding to the induced 40-kDa protein was excised and subjected to de novo sequencing based on matrix-assisted laser desorption ionization–time of flight (MALDI-TOF)/TOF mass spectrometry (MS) and subsequent computational analysis at the Proteomics Centre of the Institute of Biochemistry, Vilnius University (Vilnius, Lithuania). The sample was purified as described previously (13). Tryptic digest from the gel slice was analyzed with a 4000 QTrap (AB Sciex, Framingham, MA) mass spectrometer in linear ion trap mode using information-dependent acquisition (IDA) and a dynamic-exclusion protocol. The acquisition method consisted of an IDA scan cycle, including an enhanced-mass scan (EMS) as the survey scan, an enhanced-resolution (ER) scan to confirm the charge state, and six dependent enhanced-product-ion (EPI) scans (tandem mass spectrometry [MS-MS]). With the threshold of the ion intensity at 100,000 cps, the IDA criteria were set to allow the most abundant ions in the EMS to trigger EPI scans. The survey MS scan was set to a mass range from 400 m/z to 1,400 m/z. Dynamic ion exclusion was set to exclude precursor ions after their two occurrences during a 60-s interval. Peak lists were generated using Analyst software 1.4.2 (AB Sciex, Framingham, MA).

Illumina sequencing, contig assembly, and inducible gene locus identification.

R. jostii TMP1 DNA was sequenced using an Illumina GA2 platform (Macrogen, South Korea), and contigs were assembled using CLC-Genomics Workbench software (CLC bio, Denmark). To identify the TTMP-inducible gene, a search of the mass spectrometry-derived data against the R. jostii TMP1 genome was performed.

Gene sequence analysis.

The deduced amino acid sequences of the proteins encoded by the tpd locus were searched against the NCBI database using BLAST (14). Protein functions were assigned based on a sequence similarity search against the NCBI Conserved Domain Database (15). Phylogenetic and molecular evolutionary analyses were conducted using MEGA version 5 (16).

Qualitative and quantitative RT-PCR.

R. jostii TMP1 was cultivated in minimal medium containing glucose (0.1%), TTMP (0.05%), or pyridine (0.05%) as the sole carbon source until the culture reached an OD₆₀₀ of 0.5. Total RNA was isolated using a ZR Soil/Fecal RNA MicroPrep kit (Zymo Research Corporation). Quantitative-PCR (qPCR) amplification was performed using a Rotor-Gene Q 5-plex HRM (Qiagen, Germany). qPCR was conducted in 15 μl of reaction mixture containing 1.5 μM Syto9, 7.5 μl of Verso 1-step QRT-PCR mix, 0.15 μl of Verso Enzyme, 0.75 μl of RT enhancer (all from the Verso 1-step QRT-PCR Kit, Thermo Scientific, USA), 200 nM each primer (see Table S1 in the supplemental material), and 1 μl of the RNA tested. The qPCR was initiated with reverse transcription (RT) at 50°C for 15 min, followed by initial denaturation at 95°C for 15 min, and either a subsequent 45 cycles of 95°C for 15 s, 58°C for 1 min, and 72°C for 10 s (for tpdA, tpdB, tpdC, tpdD, and tpdE) or a subsequent 35 cycles of 95°C for 20 s, 50°C for 1 min, and 72°C for 1 min (for 16S RNA). For qualitative evaluation, endpoint PCR products were analyzed using electrophoresis. For quantitative analysis, fluorescence data were recorded after the annealing step. All experiments were carried out in duplicate. To verify the absence of DNA in the RNA samples, the procedure was performed without the reverse transcriptase step. The threshold cycle (C_T) (threshold value, 0.05) values were obtained using Rotor-Gene Q Series Software 2.1.0 (build 9). Relative target RNA analysis was performed using the 2^−ΔCT algorithm and 16S RNA as a reference for normalization.

Enhanced GFP (EGFP) fluorescence measurement.

R. jostii TMP1 was transformed with either the pART3-gfp or the pART3-5′UTR-gfp plasmid and cultivated in 25 ml of minimal medium supplemented with 0.05% TTMP, 0.05% pyridine, or 0.1% glucose for 48 h until the OD₆₀₀ reached 1.0 to 1.6. Cells were collected by centrifugation, washed three times, and resuspended in 10 mM potassium phosphate, pH 7.2, to yield a final OD₆₀₀ of 10. The suspension (200 μl/well) was transferred to a black 96-well plate (Costar), and fluorescence was measured in a FlexStationII³⁸⁴ fluorimeter (Molecular Devices) at an excitation wavelength (λ_ex) of 485 nm and an emission wavelength (λ_em) of 510 nm. Samples of three independent experiments were measured in triplicate.

Cell suspension experiments.

R. erythropolis SQ1 transformed with pART2 recombinant plasmids carrying different genes of the tpd locus was cultivated as described above. The cell suspension was supplemented with 0.1 mM TTMP or 0.1 mM (Z)-N,N′-(but-2-ene-2,3-diyl)diacetamide and incubated at 30°C for 1 hour. Bacteria were removed by centrifugation at 16,000 × g for 1 min, and the UV absorption spectra of each supernatant were recorded in a PowerWave XS plate reader (BioTek Instruments, Inc.).

Bioconversion of TTMP.

R. erythropolis SQ1 transformed with either pART2-tpdABC or pART2-tpdAB recombinant plasmid was cultivated as described above. Bioconversion reactions were carried out in a total volume of 250 ml at 30°C with shaking at 180 rpm. TTMP and glucose were added to the reaction mixture in portions of 20 mg and 125 mg, respectively, while monitoring the progress of conversion by the changes in the UV absorption spectrum in the 200- to 320-nm range. The reaction was performed for 4 days; the total amounts of TTMP added were 240 mg for pART2-tpdABC and 140 mg for pART2-tpdAB. Accumulation of bioconversion products was monitored by thin-layer chromatography in chloroform-methanol (9:1) using the substrate compound as a reference.

Intermediate metabolite isolation.

Bacteria were removed from the bioconversion reaction mixtures by centrifugation at 4,000 × g for 20 min, and the supernatants were evaporated under reduced pressure. The product of TpdABC was extracted from concentrated aqueous solution with chloroform. The product of TpdAB was isolated by sequential dissolving in acetonitrile and chloroform. The isolated intermediate metabolites were used for structural analysis and for the whole-cell and enzyme experiments.

Metabolite structural analysis.

The structures of the products of TpdABC and TpdAB were determined using ¹H nuclear magnetic resonance (NMR), ¹³C NMR, and MS analyses. ¹H and ¹³C NMR spectra were recorded on a Varian Unity Inova 300 spectrometer (300 and 75 MHz, respectively). The TpdABC product was dissolved in deuterated dimethyl sulfoxide, and the TpdAB product was dissolved in CDCl₃. Spectra were calibrated with respect to the solvent signal (CDCl₃, ¹H δ = 7.26, ¹³C δ = 77.2; DMSO-d₆, ¹H δ = 2.50, ¹³C δ = 39.5). High-resolution MS was performed on a Dual-ESI Q-TOF 6520 mass spectrometer (Agilent Technologies).

TpdE expression and purification.

The tpdE gene was fused with the 3′-polyhistidine sequence of the pET21b(+) expression vector (see Table S1 in the supplemental material). E. coli BL21(DE3) was transformed with the recombinant plasmid pET21-tpdE and cultured aerobically at 30°C in 1-liter conical flasks with 200 ml of brain heart infusion (BHI) medium (Oxoid) supplemented with 50 mg/ml ampicillin. When an OD₆₀₀ of 1.2 was reached, 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added to induce the expression of tpdE, and the culture was incubated for 4 h. Cells were collected by centrifugation; washed with 50 mM potassium phosphate, pH 7.2; resuspended in 8 ml of the same buffer; and disrupted by sonication at 750 W for 5 min using a VC750 ultrasound processor (Sonics & Materials, Inc.). Cell debris was removed by centrifugation at 16,000 × g for 10 min. Cell extracts were loaded onto a HiTrap IMAC FF 5-ml nickel column (GE Healthcare), and proteins were eluted with 50 mM potassium phosphate, 1 M imidazole, pH 7.2. The purity of TpdE was confirmed by electrophoresis on a 14% SDS-PAGE gel.

TpdE activity measurements.

The NADPH-dependent ketoreductase activity of TpdE was determined at 30°C in a Helios gamma UV-visible (UV-Vis) spectrophotometer (Thermo Fisher Scientific) by measuring the decrease in A₃₄₀ resulting from the oxidation of NADPH (ε₃₄₀ = 6,220 M⁻¹ cm⁻¹) after the addition of the substrate OBNA. TpdE activity was assayed in buffer containing 0.2 mM NADPH and 5 μM OBNA. One unit of activity was defined as the amount of the enzyme that catalyzed the oxidation of 1 μmol of NADPH per minute. The optimum temperature for TpdE activity was determined to be in the range of 15 to 50°C in 50 mM potassium phosphate, pH 7.2. The optimal pH for TpdE activity was determined to be in the range of pH 6.0 to 8.5, using potassium phosphate and Tris-HCl buffers. The apparent TpdE K_m values for NADPH and OBNA were determined by varying concentrations of these substrates from 50 to 500 μM and from 0.1 to 10 μM, respectively. Data from at least three independent experiments were combined. For data fitting, GraFit (Erithacus Software Ltd.) software was used.

Nucleotide sequence accession number.

The R. jostii TMP1 genome fragment sequence with the TTMP degradation locus was deposited in GenBank under accession no. HF544504.

RESULTS

We previously reported that R. jostii TMP1 is capable of using TTMP as a source of carbon and energy (9). To elucidate the metabolic pathway of TTMP, the tetramethylpyrazine degradation (tpd) gene locus was identified, and intermediate metabolites of TTMP were determined.

TTMP-inducible protein expression in R. jostii TMP1.

To detect TTMP catabolism-related enzymes, we investigated whether TMP1 cultivation with TTMP could cause the upregulation of proteins that might be involved in the TTMP metabolic pathway. SDS-PAGE analysis of cell extracts revealed several TTMP-inducible protein bands, including a dominant 40-kDa band (Fig. 1A). The expression of inducible 40-kDa protein was TTMP specific rather than shared with N-heterocyclic compounds, since TMP1 cultivation in the presence of pyridine showed a different protein expression profile, and no upregulation of the 40-kDa protein was observed (Fig. 1B). These results suggest that the 40-kDa protein plays an important and specific role in TTMP degradation. Therefore, to elucidate the TTMP metabolic pathway, the 40-kDa protein was selected for further analysis.

Fig 1.

Tetramethylpyrazine-inducible protein. (A) Cultures of R. jostii TMP1 were cultivated for 18 h in NB medium supplemented with different concentrations of tetramethylpyrazine (0.1 to 10 mM). (B) Strain TMP1 was cultivated for 48 h in minimal medium supplemented either with 0.05% TTMP or 0.05% pyridine (PYR) as a single source of carbon. SDS-PAGE gels were stained with Coomassie blue. The arrows indicate the 40-kDa protein. Lanes M, molecular mass ladders (in kDa).

Identification of the tpd gene locus.

To identify the TTMP-inducible protein, the 40-kDa band was excised from an SDS-PAGE gel and analyzed by MS-MS sequencing. The candidate peptides were searched against the genomic Illumina database, resulting in the identification of a contig that contained the locus of the 40-kDa protein-coding gene. To verify the sequence of the tpd locus, a genomic library was constructed and a clone containing a sequence of interest was identified by PCR screening using primers specific to the 40-kDa protein-coding gene (tpdA_F and tpdA_R [see Table S1 in the supplemental material]). Sequencing of the identified clone confirmed the nucleotide sequence of a 7-kb region of the tpd locus. The remaining sequence of the tpd locus was verified by direct sequencing of overlapping PCR products amplified from total DNA using primers designed according to the Illumina contig sequence (primer pairs seq_F and seq_R [see Table S1 in the supplemental material]). Thus, a 13-kb region that contained the genes required for the initial steps of TTMP degradation was sequenced. Sequence analysis of the tpd locus revealed eight open reading frames (ORFs) that were denominated tpdA, tpdB, tpdC, tpdR, tpdD, tpdE, ORF1, and ORF2 (Fig. 2). The 40-kDa inducible protein was encoded by the tpdA gene. Genes tpdA, tpdB, and tpdC were arranged in tandem on the same strand and contained very short to no intergenic spaces, suggesting that the genes are organized into the operon tpdABC (Fig. 2). To confirm that these three genes were cotranscribed, RT-PCR analysis was performed using primers designed to amplify the intergenic regions (AB_F and AB_R; BC_F and BC_R [see Table S1 in the supplemental material]). Endpoint PCR revealed that gene pairs tpdA-tpdB and tpdB-tpdC were transcribed as a contiguous transcript, confirming the tpdABC operon (Fig. 3A). The transcription of tpdABC was specifically induced by TTMP, but not by glucose or pyridine (Fig. 3A).

Fig 2.

Gene locus of R. jostii TMP1 involved in the metabolism of tetramethylpyrazine. The markings at the top denote separate fragments that were sequenced to verify the tpd locus: the clone from the HindIII genomic library (pUC19-H1) and four fragments obtained by PCR (Seq1 to Seq4). The gray arrows represent genes that are known to be involved in TTMP degradation; the white arrows represent other ORFs in the locus. *, the tpdA gene encodes a 40-kDa tetramethylpyrazine-inducible protein.

Fig 3.

Induction of tpd locus expression by tetramethylpyrazine in R. jostii TMP1. Strain TMP1 was cultivated in liquid minimal medium supplemented with either 0.05% TTMP, 0.1% glucose (GLC), or 0.05% PYR as a single source of carbon. (A) RT-PCR analysis using primers designed to amplify the intergenic regions between tpdA-tpdB and tpdB-tpdC. (B) TMP1 was transformed with pART3-gfp plasmid containing the tpdA upstream region inserted as a promoter (pART-5′UTR-gfp). Bacterial EGFP fluorescence was measured in a plate reader (λ_ex = 485 nm; λ_em = 510 nm); the data are presented as averages of three independent experiments plus the standard deviation. RFU, relative fluorescence units. (C) Quantitative RT-PCR analysis of the transcription of tpdA genes to tpdE genes. The data are presented as relative RNA amounts calculated from the threshold cycles using the threshold cycle of 16S RNA as a reference; averages of duplicate runs plus standard deviations are presented.

To verify that the expression of the tpdABC operon was regulated by TTMP, the fragment of the upstream region of the tpdA gene was amplified by PCR and fused to an EGFP gene from the pART3-gfp vector. Strain TMP1, transformed with the pART3-5′UTR-gfp plasmid, expressed EGFP when cultivated on TTMP, as determined by the developed bacterial fluorescence (Fig. 3B). The level of EGFP fluorescence increased 10-fold in TTMP-induced bacteria carrying pART3-5′UTR-gfp compared to that in glucose-induced bacteria. In contrast, no EGFP fluorescence above the background level was detected upon induction with pyridine. This confirmed that the upstream region of tpdA contained a promoter that was specifically activated upon exposure to TTMP.

To investigate whether the expression of the tpdD and tpdE genes was also dependent on TTMP, real-time RT-PCR analysis was performed. The results revealed that the expression of each of the tpdA-tpdE genes was specifically induced when R. jostii TMP1 was cultivated in the presence of TTMP (Fig. 3C), suggesting that proteins TpdD and TpdE participate in TTMP catabolism.

TTMP oxidation and subsequent hydrolysis by TpdABC proteins.

Sequent analysis revealed that TpdA shows high sequence similarity to luciferase-like monooxygenases (see Table S2 and Fig. S1 in the supplemental material). In addition, TpdA contains conserved domains characteristic of the cl07892 superfamily of flavin-utilizing monooxygenases and shares several common domains with coenzyme F420-dependent flavin oxidoreductases (Table 2; see Table S3 in the supplemental material).

Table 2.

Functional annotations of deduced Tpd proteins

Protein	Size (amino acids/kDa)	Putative function	Superfamily designation
			Superfamily	Accession no.	E value	Conserved domaina
TpdA	387/42.9	Flavin-utilizing monooxygenase	Flavin-utilizing monooxygenases	cl07892	5.3e−16	Not described
TpdB	136/15.2	Unknown	NTF2-like	cl09109	7.4e−09	Not described
TpdC	484/51.4	Amidase	Amidase	cl11426	2.3e−63	Not described
TpdR	787/86.2	LuxR family transcriptional regulator	LuxR_C-like	cl17315	1.4e−14	Sequence-specific DNA binding domains
			AAAb	cl17189	1.2e−4	Conserved nucleotide phosphate-binding motif
TpdD	179/19.3	Flavin reductase	Flavin reductases	cl00801	5.7e−26	FMN-binding domain found in NAD(P)H-flavin oxidoreductases
TpdE	260/26.9	Short-chain dehydrogenase/reductase	NADB Rossmann	cl09931	2.4e−73	Rossmann-fold NADB domain
ORF1	747/81.2	Large subunit of N,N-dimethylformamidase
ORF2	123/13.6	Peptidase	GAT_1	cl00020	8.9e−16	Not described

^aThe conserved domains of putative proteins encoded by the tpd locus were analyzed against the Conserved Domain Database at the NCBI website.

^bAAA, ATPases associated with a wide variety of cellular activities.

Flavin monooxygenases usually act as two-component enzyme systems: a larger component (an oxidase), which uses the reduced flavin nucleotides to hydroxylate substrates, and a smaller component (flavin reductase), which uses NAD(P)H to reduce FAD or FMN (17). Sequence analysis predicted that a flavin reductase of this type may be encoded by the tpdD gene (Table 2; see Tables S2 and S3 in the supplemental material).

A deduced tpdB gene product belongs to the nuclear transport factor 2 (NTF2)-like superfamily, which contains proteins that, despite many common structural details, diverge greatly in their functions, including ketosteroid isomerases and the beta subunit of the ring-hydroxylating dioxygenases (Table 2; see Fig. S2 in the supplemental material).

The third gene of the tpdABC operon, tpdC, encodes the protein that belongs to the superfamily of amidases (Table 2; see Table S3 in the supplemental material) and shares very high sequence homology with omega-octalactam hydrolase (see Table S2 in the supplemental material), whose function has been confirmed at the protein level (18).

To characterize Tpd proteins involved in TTMP degradation and to investigate the associated metabolic reactions, pART2 expression vectors containing different genes of the tpdABC operon and their combinations were constructed. The involvement of individual genes in the metabolism of TTMP was assessed in R. erythropolis strain SQ1, which does not metabolize TTMP. The transformation of strain SQ1 with the pART2 vector led to effective protein expression, as confirmed by EGFP fluorescence developed in the cells transformed with pART2-gfp (data not shown). TTMP metabolism in resting cell suspensions was evaluated by UV absorption spectroscopy. As seen in Fig. 4, while SQ1 cells transformed with either tpdABC or tpdAB were capable of metabolizing TTMP, those harboring either tpdA, tpdB, or tpdC did not have the ability to metabolize the compound. Furthermore, TTMP degradation by TpdABC leads to the formation of metabolites different from those found for TpdAB, as was evaluated by UV absorption spectroscopy (Fig. 4) and thin-layer chromatography (data not shown).

Fig 4.

Metabolism of tetramethylpyrazine in R. erythropolis SQ1 transformed with recombinant pART2 plasmids, each carrying a different combination of genes from the tpd locus. Cultures of R. erythropolis SQ1 were incubated in potassium phosphate buffer supplemented with 0.1 mM tetramethylpyrazine for 1 h, and the UV absorption spectra were recorded. The solid lines represent the initial spectrum of tetramethylpyrazine; the dotted lines indicate the final spectra of the bioconversion products.

TpdABC- and TpdAB-generated metabolites were extracted from the bioconversion media, and their structures were determined by ¹H NMR, ¹³C NMR, and MS analyses. The ¹H NMR spectrum of the TpdABC product showed five peaks in the ¹H NMR spectrum [δ 6.37 (br s, 1H, NH), 4.58 (p, J = 7.1 Hz, 1H, CH), 2.21 (s, 3H, COCH₃), 1.99 (s, 3H, NHCOCH₃), 1.36 (d, J = 7.1 Hz, 3H, CH₃)] and six peaks in the ¹³C NMR spectrum (δ 207.1, 169.7, 54.8, 26.8, 23.4, and 17.8), identifying the TpdABC product as OBNA. These NMR spectra were in full agreement with previously published results (19). Meanwhile, the product of TpdAB showed three peaks with δ 8.67 (s, 1H, NH), 1.88 (s, 3H, CH₃), and 1.82 (s, 3H, CH₃), while the ¹³C NMR spectrum contained four peaks with δ 167.7, 121.2, 23.2, and 16.4. High-resolution MS analysis confirmed the molecular formula of the compound as C₈H₁₅N₂O₂ (calculated [M+H]⁺ mass was 171.11335, found mass was 171.1134). These NMR spectra revealed that TpdAB metabolized TTMP to BDNA.

Next, the intermediate metabolite BDNA was used as a substrate for strain SQ1 transformed with recombinant plasmids, each carrying either individual genes of the tpdABC operon or different combinations of the tpdABC genes. TpdABC, TpdBC, and TpdC were able to metabolize BDNA, indicating that TpdC alone was sufficient to catalyze the conversion of BDNA (Fig. 5). The results of bioconversion experiments (summarized in Table 1) demonstrated that the catabolism of TTMP is a multistep process starting with TTMP oxidation by TpdAB to produce BDNA, which is then hydrolyzed by TpdC to form OBNA. BDNA hydrolysis by TpdC led to the formation of an optically active OBNA that rotated plane-polarized light with a specific rotation at 20°C and 586 nm ([a]_D²⁰) of +102° (1.16 g in 100 ml CHCl₃), indicating that the reaction is stereospecific.

Fig 5.

Metabolism of (Z)-N,N′-(but-2-ene-2,3-diyl)diacetamide in R. erythropolis SQ1 transformed with recombinant pART2 plasmids, each carrying a different combination of genes from the tpd locus. Cultures of R. erythropolis SQ1 were incubated in potassium phosphate buffer supplemented with 0.1 mM (Z)-N,N′-(but-2-ene-2,3-diyl)diacetamide for 1 h, and the UV absorption spectra were recorded. The solid lines represent the initial spectrum of (Z)-N,N′-(but-2-ene-2,3-diyl)diacetamide; the dotted lines represent the final spectra of the bioconversion products.

Table 1.

Products of resting-cell reactions of R. erythropolis SQ1 cells transformed with tpd locus genes^a

Plasmid	Expressed gene(s)	Product on substrateb:
		TTMP	BDNA
pART2		ND	ND
pART2-tpdA	tpdA	ND	NA
pART2-tpdB	tpdB	ND	NA
pART2-tpdC	tpdC	ND	OBNA
pART2-tpdAB	tpdA, tpdB	BDNA	NA
pART2-tpdBC	tpdB, tpdC	ND	OBNA
pART2-tpdABC	tpdA, tpdB, tpdC	OBNA	OBNA

^aThe cultures of transformed bacteria were incubated in the presence of either 0.1 mM TTMP or 0.1 mM BDNA. Substrate degradation was observed by UV absorption spectroscopy. The reaction products BDNA and OBNA were determined by either ¹H NMR, ¹³C NMR, and MS analysis (TTMP conversion) or by UV spectrum comparison and thin-layer chromatography (BDNA conversion).

^bND, bioconversion not detected; NA, bioconversion not analyzed.

OBNA reduction by TpdE.

TpdE shares high homology with classical short-chain dehydrogenases/reductases that have a Rossmann fold NAD(P)H/NAD(P)⁺ binding (NADB) domain (Table 2; see Table S3 in the supplemental material). This domain is found in numerous dehydrogenases and other redox enzymes that can use a wide variety of substrates, including alcohols, glucose, and steroids (20). OBNA, which contains a keto group, might be a substrate for short-chain dehydrogenases/reductases, since most bacterial short-chain dehydrogenases/reductases are known to be alcohol dehydrogenases that can catalyze bidirectional oxidoreduction.

To demonstrate that TpdE is indeed a short-chain dehydrogenase/reductase, as predicted by the sequence analysis, His₆-tagged TpdE protein was overproduced in E. coli and purified by affinity chromatography with a yield of 2,600 U per liter of culture. A purified protein migrated as an ∼30-kDa band, which agreed with the expected size of the recombinant His₆-TpdE protein (27.9 kDa) (see Fig. S3 in the supplemental material). The specific ketoreductase activity of TpdE, with OBNA as a substrate and NADPH as an electron donor, was 26 U/mg. The optimum pH of TpdE was 7.2, and the optimum temperature was 35°C. The apparent K_m for NADPH calculated from steady-state analysis was 379 ± 73 μM in the presence of 6 mM OBNA, and the apparent K_m for OBNA was 5.2 ± 1.5 μM in the presence of 500 μM NADPH, with k_cats of 7.7 s⁻¹ and 2.5 s⁻¹, respectively. The enzyme has an absolute requirement for NADPH, yielding no detectable OBNA reduction in the presence of NADH.

Hypothetical proteins encoded by ORF1 and ORF2, which were found upstream of tpdE, may contribute to further degradation of TTMP. ORF1 encodes the putative large subunit of N,N-dimethylformamidase, while a hypothetical protein encoded by ORF2 is homologous to peptidases (Table 2; see Tables S2 and S3 in the supplemental material).

DISCUSSION

To determine in detail the degradation pathway of TTMP in R. jostii TMP1, we isolated and characterized the genes involved in TTMP metabolism. Based on the identity of the TTMP-inducible protein, we identified the tetramethylpyrazine degradation locus and described the enzymes encoded in this locus.

TTMP degradation was demonstrated to be a multistep process, involving oxidative aromatic ring cleavage by the TpdA-TpdB complex and subsequent hydrolysis catalyzed by TpdC, followed by keto group reduction by TpdE (Fig. 6). Flavin monooxygenases, such as TpdA, are capable of oxidizing various substrates, including some N-heterocyclic compounds (17, 21). To oxidize TTMP, TpdA requires TpdB in a manner similar to that observed in bacterial luciferases, which act as a complex of α and β subunits. However, while luciferase α and β subunits are homologous to each other (17), the TpdB protein is considerably smaller than TpdA and shares no sequence homology with luciferase-like monooxygenases.

Fig 6.

Proposed tetramethylpyrazine catabolic pathway in R. jostii TMP1. 1, tetramethylpyrazine; 2, BDNA; 3, OBNA; 4, N-(3-hydroxybutan-2-yl)acetamide; 5, 5-3-amino-2-butanol. The dashed arrows indicate hypothetical reactions. TpdAB, TTMP oxygenase; TpdD, flavin reductase; TpdC, BDNA hydrolase; TpdE, aminoalcohol dehydrogenase.

TpdB shares sequence homology with proteins belonging to the families of ketosteroid isomerases and the beta-subunit of the ring-hydroxylating dioxygenases (15). Since TTMP oxidation might be facilitated by the functions of the enzymes of both of these families, the function of TpdB cannot be inferred from sequence similarity alone. Ketosteroid isomerases catalyze double-bond migration reactions of steroid substrates (22); therefore, such a function of TpdB might facilitate TTMP conversion through the destabilization of the aromatic ring. The enzyme PhzA/B, a homologue of TpdB, has been shown to participate in the condensation of two amino ketone molecules (23). This condensation is accompanied by PhzA/B-mediated rearrangement of double bonds, yielding a phenazine precursor that is structurally similar to tetraalkylpyrazine (23). The analogous rearrangement of double bonds by TpdB may contribute to the destabilization of TTMP, facilitating its oxidation by TpdA.

Even though the product of TpdAB-catalyzed TTMP oxidation resembles that of dioxygenases (24), similar products may also be formed via the cleavage of the aromatic ring by flavin monooxygenases, such as 2-methyl-3-hydroxypyridine-5-carboxylic acid oxygenase or 5-pyridoxic acid oxygenase (21). These monooxygenases open the aromatic ring of the substrate and incorporate only one oxygen atom from O₂, whereas another atom comes from an H₂O molecule. The exact mechanism of aromatic ring oxidation catalyzed by the TTMP oxygenase TpdAB requires further investigation, such as oxygen tracer experiments.

Flavin monooxygenases homologous to TpdA act as part of a two-component enzyme system that also includes flavin reductase, such as that encoded by tpdD. Therefore, the oxidation of TTMP should require both TpdA and TpdD acting together. However, in R. erythropolis SQ1, tpdD was not required to oxidize TTMP (Fig. 4). This is not surprising, since Rhodococcus strains encode many flavin reductases (as seen from published genomes [25, 26]) that may have substituted tpdD.

Oxidized TTMP is further metabolized via a hydrolysis reaction catalyzed by TpdC (Fig. 6). ¹H NMR spectra (in CDCl₃) (data not shown) revealed that BDNA may exist in different tautomeric forms. Since BDNA hydrolysis yields optically active OBNA, rather than a racemate, we believe that the imine form of BDNA, rather than amine, is a true substrate for TpdC. Nevertheless, further analysis is required to elucidate the precise mechanism of TpdC-mediated cleavage of OBNA.

Subsequent OBNA reduction is catalyzed by the amino alcohol dehydrogenase TpdE (Fig. 6), which uses NADPH as an electron donor and presumably converts OBNA into amino alcohol. The detailed biochemical characterization of TpdE is currently in progress.

Further steps of TTMP metabolism may involve the enzymes encoded by ORF1 and ORF2. Both ORF1 and ORF2 may recognize the amidic bond in the TpdE product and may hydrolyze it to produce 3-amino-2-butanol (Fig. 6).

In R. jostii TMP1, TTMP degradation is an inducible process involving the induction of TpdA expression (Fig. 1 and 3). A putative regulator of the tpd locus is encoded by tpdR. The analysis of the deduced amino acid sequence of TpdR revealed that the protein contains a number of domains that are characteristic of transcription regulators belonging to the LuxR family. However, to ascertain the ability of TpdR to regulate the tpd locus, further research is required.

In summary, here, we identified for the first time the genetic locus responsible for bacterial degradation of pyrazines, namely, tetramethylpyrazine. Our data revealed that, in accordance with previous reports on TTMP metabolism by Rhodococcus sp. strains (8), the initial step of TTMP degradation in R. jostii TMP1 is ring oxidation and cleavage. However, in contrast to the data published by Müller and Rappert (8) suggesting that TTMP is oxidized by a cytochrome P450-type enzyme, the TTMP oxidase characterized here is a putative flavoenzyme.

While identification of the TTMP catabolism pathway will undoubtedly advance the field of bacterial degradation of pyrazines, one must bear in mind that the reactions described here may be specific only to TTMP and similar pyrazines that carry a substituent at each carbon of the ring. Usually, the degradation of semisubstituted pyrazines starts with hydroxylation of the pyrazine ring at a free ring position (2, 8) and therefore employs different enzymes than those identified in this study.

Identification of the genes responsible for bacterial degradation of TTMP and isolation of its intermediate metabolites allowed us for the first time to describe the catabolic pathway of alkylpyrazine metabolism in bacteria, providing fundamental knowledge about the biodegradation of these N-heterocyclic compounds and revealing various enzymes that may be employed in the future for selective and specific bioconversion reactions.