Dynamic Metabolic and Transcriptional Profiling of Rhodococcus sp. Strain YYL during the Degradation of Tetrahydrofuran

Abstract

Although tetrahydrofuran-degrading Rhodococcus sp. strain YYL possesses tetrahydrofuran (THF) degradation genes similar to those of other tetrahydrofuran-degrading bacteria, a much higher degradation efficiency has been observed in strain YYL. In this study, nuclear magnetic resonance (NMR)-based metabolomics analyses were performed to explore the metabolic profiling response of strain YYL to exposure to THF. Exposure to THF slightly influenced the metabolome of strain YYL when yeast extract was present in the medium. The metabolic profile of strain YYL over time was also investigated using THF as the sole carbon source to identify the metabolites associated with high-efficiency THF degradation. Lactate, alanine, glutarate, glutamate, glutamine, succinate, lysine, trehalose, trimethylamine-N-oxide (TMAO), NAD⁺, and CTP were significantly altered over time in strain YYL grown in 20 mM THF. Real-time quantitative PCR (RT-qPCR) revealed changes in the transcriptional expression levels of 15 genes involved in THF degradation, suggesting that strain YYL could accumulate several disturbances in osmoregulation (trehalose, glutamate, glutamine, etc.), with reduced glycolysis levels, an accelerated tricarboxylic acid cycle, and enhanced protein synthesis. The findings obtained through ¹H NMR metabolomics analyses and the transcriptional expression of the corresponding genes are complementary for exploring the dynamic metabolic profile in organisms.

INTRODUCTION

Microbial degradation is one of the main mechanisms responsible for the mineralization of pollutants in contaminated environments (1). Many bacterial strains have been isolated from contaminated environments and used to degrade certain pollutants. Knowledge of the genes responsible for the catabolism of pollutants will increase our understanding of the degradation capacities of different strains. However, despite the presence of similar catabolic genes, the degradation of certain pollutants under the optimum culturing conditions is highly variable between bacterial strains. This difference might reflect the fact that pollutant degradation induces basal metabolism changes in bacteria, and certain metabolites might be more relevant for degradation of specific pollutants by the bacterial strain (2, 3). Therefore, functional analyses of the changes in basal metabolism in bacterial strains during pollutant degradation are essential for understanding the mechanism of pollutant degradation.

Metabolomics is a top-down systems biology approach for the metabolic profiling of living organisms and is a tool for the comprehensive and fully quantitative analysis of low-molecular-weight endogenous compounds (sugars, amino acids, fatty acids, etc.) with which metabolic responses to biological interventions or environmental factors are analyzed and modeled (4–6). Characterizing the metabolome via metabolic profiling provides insight into the state of the organism or substance at a particular moment (7). The evaluation of the metabolic profile can be accomplished using a variety of analytical platforms, including gas chromatography-mass spectrometry, liquid chromatography-mass spectrometry, and nuclear magnetic resonance (NMR) spectroscopy. Compared with the two chromatographic platforms, NMR-based analysis requires simple and easy sample preparation, which is beneficial for minimizing changes in the chemical composition and reducing the loss of minor components (8, 9); simultaneously monitors various organic chemical species (10, 11); and shows the potential for efficient high-throughput screening (12). Therefore, the application of an NMR-based platform in metabolomics has been widely reported for studies with a variety of microorganisms, including studies of metabolic flux changes in Escherichia coli during heat (13) and superoxide (14) stresses and the role of genes in Saccharomyces cerevisiae (15, 16). High-throughput NMR-based metabolomics analysis facilitates the characterization of large numbers of individuals. Multivariate statistical analyses, such as principal component analysis (PCA) and orthogonal projection to latent structures discriminant analysis (OPLS-DA), have proven efficient for reducing high-throughput NMR spectral data and visualizing the similarities between different groups (2, 17, 18).

The cyclic aliphatic ether tetrahydrofuran (THF) has been widely used as a common solvent in bulk chemical and pharmaceutical industries and has received much attention due to its release into environments and toxicity to organisms. The high water solubility, vapor pressure, and high production volume pressure of THF make this pollutant easily detectable in groundwater (19). As an inhibitor of cytochrome P450-dependent enzymes, THF induces central nervous system irritation, narcosis, edema, and colonic muscle spasms in animals (20–22). Although THF is not easily mineralized due to its cyclic structure, many microorganisms that utilize THF as their sole carbon and energy source under aerobic conditions have been isolated. These microorganisms primarily belong to Gram-positive actinomycetes of the genera Rhodococcus (23–25) and Pseudonocardia (26, 27) and Gram-negative Pseudomonas species (22, 28). Although the mechanism and pathway of THF oxidation have not been firmly established, THF monooxygenase has been identified as a key enzyme for the initiation of THF degradation in these microorganisms. The THF monooxygenase gene of Rhodococcus sp. strain YYL has been cloned and sequenced and has 96% similarity to that of Pseudonocardia sp. strain K1 (29). Notably, Rhodococcus sp. strain YYL exhibits a high degradation rate, namely, 137.6 mg THF/(h · g [dry weight] strain YYL), which is more than approximately five times greater than the degradation rate of other THF-degrading strains (23). Therefore, it is important to characterize the mechanisms underlying the high-efficiency THF degradation of strain YYL. The variations of the metabolome induced by THF in strain YYL should be relevant to its THF degradation efficiency, and exploring the dynamic metabolic profile in response to THF might reveal the basic mechanism underlying its high degradation efficiency.

In this study, we used an NMR-based metabolomics method coupled with multivariate data analysis to systematically investigate the metabolic differences in strain YYL when utilizing THF or yeast extract, individually or simultaneously, as a carbon source. Moreover, real-time quantitative PCR (RT-qPCR) analysis was further employed to measure the transcriptional expression of 15 relevant genes when YYL cells utilized THF as the sole carbon source. These findings will provide an overview of the physiological response and adaptation of Rhodococcus sp. YYL to THF exposure and degradation.

MATERIALS AND METHODS

Strain and culture conditions.

THF-degrading Rhodococcus sp. strain YYL was previously isolated from the activated sludge of a wastewater treatment plant (23). Strain YYL was cultured in 100 ml liquid optimal base mineral medium (BMM) (23). The medium was supplemented with 1 g/liter yeast extract (CY), with 20 mM THF (CT), or with both THF and yeast extract (TY). During cultivation, the agitation ratio was maintained at 140 rpm and the temperature was maintained at 30°C. One Erlenmeyer flask of culture was used for each sample, and eight replicates were prepared for every sampling time point in the three treatment groups.

Sample collection and metabolite extraction.

The sampling time in the experiment was decided according to the growth kinetics of Rhodococcus sp. strain YYL in the three treatment groups (see Fig. S1 in the supplemental material). Three sampling time points were chosen in each treatment: 12 h (exponential-phase growth), 48 h (early stage of stationary-phase growth), and 84 h (late stage of stationary-phase growth) in CY; 12 h (exponential-phase growth), 48 h (exponential-phase growth), and 84 h (stationary-phase growth) in TY; and 48 h (exponential-phase growth), 84 h (exponential-phase growth), and 132 h (stationary-phase growth) in CT. Metabolism was quenched after the flasks were chilled on ice, followed by centrifugation at 6,000 rpm for 5 min at 4°C and the subsequent decanting of the supernatant. The cell pellet was subsequently washed three times with 1 ml of ice-cold phosphate-buffered saline (6,000 rpm, 5 min, 4°C) and homogenized in 600 μl of an ice-cold aqueous methanol solution (methanol–double-distilled H₂O [2:1]). Intracellular metabolites were extracted ultrasonically (i.e., 3 s of sonication with a 3-s break) for 10 min. The supernatants were collected after 10 min of centrifugation at 12,000 rpm and 4°C, and the cell debris was further extracted twice using the same procedure. After combining the three solutions obtained, the extracts were lyophilized after removing the methanol in vacuo. Each bacterial extract was separately reconstituted in 600 μl Na⁺/K⁺ phosphate buffer (K₂HPO₄-NaH₂PO₄, 0.1 M, pH 7.4) containing 100% D₂O and 0.005% sodium 3-trimethylsilyl [2,2,3,3-d₄] propionate (TSP). Following centrifugation, the supernatant (550 μl) of each extract was transferred to a 5-mm NMR tube for NMR analysis.

NMR measurements.

All NMR spectra of bacterial samples were recorded using a Bruker Avance III 600-MHz spectrometer (Bruker BioSpin, Germany), operating at 600.13 MHz and 298 K during acquisition. A standard one-dimensional NMR spectrum was acquired with the first increment of the nuclear Overhauser enhancement spectroscopy pulse sequence (relaxation delay-90°-t₁-90°-t_m-90°-acquisition), with a relaxation delay of 2 s, a mixing time (t_m) of 100 ms, a delay time (t₁) of 6.5 μs, and a 90° pulse length of 10 μs. Water suppression was achieved through presaturation at the water resonance frequency. Sixty-four free induction decays were collected into 32,000 data points for each spectrum, with a spectral width of 20 ppm centered at the water resonance. An exponential window function with a line-broadening factor of 0.5 Hz was applied to free induction decays prior to the Fourier transformation of the data. For NMR signal assignment, a range of two-dimensional NMR spectra for the selected samples was acquired as previously described (2, 30), including ¹H-¹H correlation spectroscopy (COSY), ¹H-¹H total correlation spectroscopy (TOCSY), ¹H-¹³C heteronuclear multiple-bond correlation (HMBC), and ¹H-¹³C heteronuclear single quantum correlation (HSQC) spectra.

NMR data processing and multivariate data analysis.

All ¹H NMR spectra were manually phased, baseline corrected, and calibrated to TSP at δ 0.00 using the TOPSPIN (v2.0) program (Bruker BioSpin, Germany). The data reduction was performed after dividing the region of δ 9.7 to 0.7 into bins with a width of 0.004 ppm (2.4 Hz) using the AMIX package (v3.8.3; Bruker BioSpin, Germany). The regions at δ 3.4 to 3.3 and δ 4.95 to 4.7 were discarded to remove the effects of methanol and imperfect water suppression resonance. Following the normalization of each spectrum to the total spectral intensity, PCA was performed on the mean-centered NMR data using the SIMCA-P⁺ software package (v11.0; Umetrics, Sweden) to generate an overview of sample clustering, such as distributions and possible outliers. OPLS-DA, a supervised multivariate data analysis tool, was subsequently applied to analyze the ¹H NMR spectral data scaled to unit variance as the x matrix and group information as the y matrix. The OPLS-DA models were successively evaluated using an 8-fold cross-validation (CV) method (31) and CV analysis of variance (ANOVA) (32). To facilitate the interpretation of the OPLS-DA results, the loading indicating altered metabolites in response to THF treatment was back-transformed (33) and plotted with a color-coded correlation coefficient (r) using a Matlab script (Mathworks), with some modifications. The coefficients describe the weighting of variables contributing to the class separation of the model, with a hot color (i.e., red) being more significant than a cold color (i.e., blue). In this study, a cutoff value of 0.67 (|r|, >0.67) for correlation coefficients was used for the statistical significance analysis on the basis of the discrimination significance at the level of P equal to <0.05. In addition, one-way ANOVA was used to test the variances between CY and TY or over time in CT. The analysis was performed using SPSS (v16.0) statistical software.

RNA isolation, cDNA generation, and RT-qPCR.

Total RNA of strain YYL grown in CT was separately extracted with the RNAiso Plus reagent (TaKaRa, Dalian, China) and reverse transcribed into cDNA using a PrimeScript reverse transcriptase reagent kit (TaKaRa).

RT-qPCR reactions were performed in a 20-μl volume containing 10 μl SYBR Premix Ex Taq (TaKaRa), 0.4 μl of each forward and reverse gene-specific primer (10 μM), and 2.0 μl of cDNA (1:10 dilution). Gene-specific primers were designed using Beacon Designer (v7) software (see Table S1 in the supplemental material). The V3 region of 16S rRNA, used as an internal standard, was amplified using the primers F338 (CCTACGGGAGGCAGCAG) and R518 (ATTACCGCGGCTCGTGG). All RT-qPCR experiments were performed on a Mastercycler ep realplex system (Eppendorf, Hamburg, Germany) according to the user's guide. Three independent cDNA samples were assayed, and the 2^−ΔΔCT method (where C_T is the threshold cycle) (34) was used to calculate the gene expression levels for strain YYL. The expression of 15 related genes at 48 h was normalized to 1. Analysis of variance was performed as described above.

RESULTS

Metabolites of strain YYL.

Figure 1 shows some typical ¹H NMR spectra of aqueous extracts of Rhodococcus sp. strain YYL grown in yeast extract (CY), THF (CT), and both yeast extract and THF (TY). The endogenous metabolite resonances were assigned according to their chemical shifts on the basis of data obtained from the literature (35, 36) and/or an online database (http://www.hmdb.ca/). We also confirmed peak assignments according to a series of two-dimensional NMR experiments, including ¹H-¹H COSY, ¹H-¹H TOCSY, ¹H-¹³C HSQC, and ¹H-¹³C HMBC (detailed information is provided in Table S2 in the supplemental material). The spectra of strain YYL showed signals primarily from sugars (trehalose and glucose), amino acids (isoleucine, leucine, valine, threonine, alanine, glutamine, glutamate, aspartate, d-hydroxylysine, lysine, and glycine), nucleoside and nucleotide metabolites (adenosine, NAD⁺, adenosine 2′,3′-cyclic phosphate, CTP, and uridine), amines (trimethylamine, dimethylamine, and trimethylamine-N-oxide [TMAO]), organic acids (glutarate, succinate, fumarate, lactate, formate, acetate, and 3-d-hydroxybutyrate), ethanol, acetone, betaine, and choline. The visual inspection of these spectra revealed that at 84 h strain YYL grown in TY had a lower lactate level as well as a higher trehalose level than YYL grown in CY (Fig. 1). Analysis of the NMR profiles for strain YYL grown in CT showed significant changes of metabolites over time. For instance, an increased level of trehalose was observed at 84 h relative to the level at 48 h, whereas at 132 h the trehalose level was decreased relative to that at 84 h (data not shown). To obtain detailed information on THF-induced metabolic alterations, a multivariate data analysis of these NMR profiles was performed.

FIG 1.

Typical 600-MHz ¹H NMR spectra of aqueous extracts of Rhodococcus sp. YYL grown in CY, TY, and CT. The spectral regions were vertically expanded 2 and 8 times from δ 0.7 to 3.3 and 32 times from δ 5.7 to 9.5 compared with the spectra of the region from δ 3.4 to 5.6. Key to the peaks: 1, isoleucine; 2, leucine; 3, valine; 4, ethanol; 5, 3-d-hydroxybutyrate; 6, lactate; 7, threonine; 8, alanine; 9, acetate; 10, acetone; 11, glutarate; 12, glutamate; 13, glutamine; 14, succinate; 15, aspartate; 16, dimethylamine; 17, trimethylamine; 18, d-hydroxylysine; 19, lysine; 20, choline; 21, betaine; 22, trimethylamine-N-oxide; 23, trehalose; 24, glycine; 25, β-glucose; 26, α-glucose; 27, fumarate; 28, uridine; 29, adenosine 2′,3′-cyclic phosphate; 30, adenosine; 31, formate; 32, NAD⁺; 33, CTP; 34, residual methanol.

THF-related metabolomic changes.

PCA of the normalized NMR profiling of strain YYL was conducted (see Fig. S2 in the supplemental material). The plot of the PCA scores obtained from the NMR data showed intercellular metabolomic differences (dashed line in Fig. S2 in the supplemental material) in strain YYL, dominated by the carbon resource intervention in the first two principal components (PC1 and PC2), which cumulatively explained 85.4% of the total variances. A clear difference between the CY and CT groups was observed, reflecting differences in the carbon resource; the TY group represented the transition of the metabolome in strain YYL using yeast extract and THF as cocarbon sources. Moreover, the change trajectory illustrated relatively similar metabolic profiles between the CY and TY groups at 12 h and 48 h, but a clear difference in the metabolic profiles of the CY and TY groups was observed at 84 h. Notably, in the plot of PCA scores, the metabolic profile of strain YYL grown in TY at 84 h clustered with that of strain YYL grown in CT at 48 h.

Pairwise comparative OPLS-DA was further conducted with the NMR data for strain YYL from the TY and CY groups obtained at various time points and the NMR data obtained at different time points for the CT group. The quality of these OPLS-DA models was validated through the Q² (predicted classification) values and P values obtained from CV analysis of variance (ANOVA) (Table 1). The dominant metabolites for differentiation from selected models are shown in the OPLS-DA coefficient plots (Fig. 2). The values of the correlation coefficients for the metabolites whose levels were significantly altered are listed in Table 1. Compared with the levels for strain YYL in the CY group, a significant decrease in the lactate level and a marked increase in the glutamine level in strain YYL in the TY group were observed at 12 and at 48 h, respectively. More significantly altered levels of metabolites, including a significant elevation of glutamate and trehalose levels and a significant depletion of lactate, alanine, glutarate, lysine, and TMAO levels, were observed at 84 h for strain YYL grown in TY compared with the values for strain YYL grown in CY. The THF-associated dynamic concentration variations for some metabolites representing different pathways were further determined using one-way ANOVA (see Fig. S3 in the supplemental material). Alanine in strain YYL from both the CY and TY groups exhibited similar trends of decreased levels over time, with a significant depletion in TY compared with the amount in CY at 84 h (P < 0.01). Glutarate and lysine in strain YYL in CY presented a trend of decreased levels, followed by a trend of increased levels, but maintained a substantial depletion during the entire process when strain YYL was grown in TY. In contrast, glutamate and trehalose exhibited steady increases in levels in both of these groups, with a marked increase in TY compared with that in CY at 48 h (P < 0.05) and 84 h (P < 0.01).

TABLE 1.

Significantly altered metabolites in Rhodococcus sp. strain YYL in response to THF treatment

Metabolite	ra
	TY vs CY			CT
	12 h	48 h	84 h	84 h vs 48 h	132 h vs 84 h
Lactate	−0.85	—	−0.92	−0.89	0.83
Alanine	—	—	−0.86	−0.95	0.77
Glutarate	—	—	−0.94	−0.72	—
Glutamate	—	—	0.81	−0.88	—
Glutamine	—	0.94	—	−0.89	−0.77
Succinate	—	—	—	−0.84	—
Lysine	—	—	−0.93	—	—
TMAO	—	—	−0.78	—	0.87
Trehalose	—	—	0.98	0.69	−0.75
NAD	—	—	—	−0.79	—
CTP	—	—	—	−0.88	—

^aPositive and negative signs for the correlation coefficients indicate positive and negative correlations in the concentrations, respectively. P was equal to 0.05, 7 degrees of freedom was used, and r equal to 0.67 was used as the corresponding cutoff value of the correlation coefficient for the statistical significance on the basis of the discrimination significance. For TY versus CY at 12, 48, and 84 h, R²X = 0.549, Q² = 0.948, and P = 0.0062; R²X = 0.549, Q² = 0.948, and P = 0.0072; and R²X = 0.549, Q² = 0.948, and P = 4.86 × 10⁻⁷, respectively. For CT at 84 h versus 48 h and 132 h versus 84 h, R²X = 0.482, Q² = 0.934, and P = 1.92 × 10⁻⁶ and R²X = 0.690, Q² = 0.931, P = 1.38 × 10⁻⁵, respectively. —, |r| was less than the cutoff value. The R²X values describe the goodness of fit of the models with the NMR data, and the Q² values describe the predictive capacity of the model.

FIG 2.

Cross-validated OPLS-DA scores (a to c) and corresponding coefficient plots (d to f) derived from the ¹H NMR spectra of the extracts of Rhodococcus sp. YYL. (a, d) Comparison of results with growth in CY (black stars) and TY (red dots) at 84 h; (b, e) comparison of data at 48 h (black stars) and 84 h (red dots) with growth in CT; (c, f) comparison of data at 84 h (black stars) and 132 h (red dots) with growth in CT. In the scales on the right, red indicates important discriminatory metabolites, whereas blue indicates no significance in discrimination. Metabolite keys are shown in Table 1. t[2]O, cross-validated score value; t[1]P, model score value.

When strain YYL utilized THF as the sole carbon source, the content of metabolites varied with the bacterial growth stages and THF degradation rates. As shown in the OPLS-DA coefficient plot (Fig. 2; see Fig. S3 in the supplemental material), strain YYL grown in CT for 84 h had a significantly higher level of trehalose (P < 0.01) and remarkably lower levels of lactate, alanine, glutamate, glutamine, succinate, NAD⁺, and CTP compared with those at 48 h (P < 0.01). At 132 h, strain YYL had dramatically higher levels of lactate, alanine, and TMAO and markedly lower levels of glutamine and trehalose compared with those at 84 h (P < 0.01).

RT-qPCR analysis of gene expression in strain YYL.

The transcriptional levels of 15 related genes involved in THF degradation, sugar and amino acid synthesis, the central genetic processes, and oxidative stress in strain YYL grown in CT were assessed over time (48, 84, and 132 h) by RT-qPCR using paired primers (see Table S1 in the supplemental material). Our observations showed that the expression of thm, encoding THF-degrading monooxygenase, decreased with decreasing THF concentrations over time, with a significant change in the level of thm expression at 84 h compared with the level at 48 h being observed when the strain was grown in CT (P < 0.01) (Fig. 3; see Fig. S1 in the supplemental material).

FIG 3.

Relative transcript levels (RTLs) of selected metabolism genes over time with growth in CT. One-way ANOVA analysis was used to analyze variances between 48 and 84 h with growth in CT and between 84 and 132 h with growth in CT. **, P < 0.01; *, P < 0.05. Genes in panel a: gyrB, DNA gyrase subunit B gene; rpoB, RNA polymerase sigma gene; pcrA, ATP-dependent DNA helicase gene; rhiB, ATP-dependent RNA helicase gene; pto, serine/threonine protein kinase gene; cyp, cytochrome P450 monooxygenase gene; sod, superoxide dismutase gene; thm, tetrahydrofuran monooxygenase gene. Genes in panel b: pfk, phosphofructokinase gene; gpd, glyceraldehyde-3-phosphate dehydrogenase gene; treS, trehalose synthase gene; otsA, trehalose-6-phosphatase phosphatase gene; mdh, malate dehydrogenase gene; sdh, succinate dehydrogenase gene; gltA, citrate synthase gene.

During glycolysis, phosphofructokinase (pfk) and glyceraldehyde-3-phosphate dehydrogenase (gpd) regulate two significant steps (from fructose-6-phosphate to fructose-1,6-bisphosphate and from glyceraldehyde-3-phosphate to 1,3-bisphosphoglycerate, respectively). The expression of pfk was upregulated 5-fold (P < 0.01) at 84 h compared with the level of regulation at 48 h and 2-fold downregulated (P < 0.01) at 132 h compared with the level of regulation at 84 h, while gpd expression was gradually downregulated over time. Two genes involved in the citrate cycle, mdh, encoding malate dehydrogenase, and gltA, encoding citrate synthase, showed significant downregulation with decreasing THF concentration (P < 0.01 between 48 and 84 h for mdh, P < 0.01 over time for gltA). However, sdh, encoding succinate dehydrogenase, exhibited the highest transcription level at 84 h, when strain YYL was in the initial stationary phase, and this was accompanied by the highest THF degradation rate (1.92 mM THF/h · g [dry weight] strain YYL) (P < 0.01). Trehalose and glucose were the only two sugars identified in the NMR analysis, and their relevant genes were also analyzed. The gene otsA, encoding trehalose-6-phosphatase, responsible for converting UDP-glucose to trehalose, showed a much higher level of expression during THF degradation when strain YYL was in the initial stationary phase (84 h) than when it was in the exponential phase (48 h) (P < 0.01), and expression of the gene treS, encoding trehalose synthase for converting maltose to trehalose, showed no significant change in regulation at 84 h compared with that at 48 h, but it was significantly downregulated at 132 h compared with the level of regulation at 84 h (Fig. 3).

The changes in the dynamic metabolic profile of strain YYL in response to THF may reflect not only the metabolite concentration but also changes in other processes, such as DNA replication, RNA transcription, and the oxidative stress response system. In this study, DNA repair helicase (pcrA) and DNA gyrase (gyrB) were examined to determine changes in DNA replication, and it was found that both of these genes were slightly upregulated at 84 h compared with their levels of regulation at 48 and 132 h. Genes encoding RNA polymerase sigma factor (rpoB) and ATP-dependent RNA helicase (rhiB) were examined to explore the influence of THF on RNA synthesis over time in CT. The expression of rpoB was consistent with that of pcrA and gyrB, while rhiB expression decreased with increasing THF degradation (Fig. 3; see Fig. S1 in the supplemental material). The differences between rpoB and rhiB might be due to the fact that the rpoB gene is responsible for RNA synthesis in converting DNA into RNA and the rhiB gene is responsible for RNA synthesis in RNA self-replication. The serine/threonine protein kinase gene (pto), regulating protein phosphokinase, showed a greater increase with decreasing THF concentration over time in CT (Fig. 3; see Fig. S1 in the supplemental material). Cytochrome P450 monooxygenase (cyp), which catalyzes the oxidation of a wide range of endogenous compounds in biosynthetic and biodegradation pathways, as well as xenobiotics, such as drugs and environmental contaminants (37), showed an expected increase in transcription level with decreasing THF concentration, with a marked increase at 132 h compared with the transcription level at 84 h (P < 0.01) (Fig. 3; see Fig. S1 in the supplemental material). A higher concentration of THF had more inhibitory effects on cytochrome P450-dependent monooxygenase in strain YYL, so when the THF concentration decreased over time in CT, the transcription of cyp increased. Superoxide dismutase (sod), a significant antioxidative enzyme in strain YYL, showed decreased transcription with decreasing THF concentration (P < 0.05) (Fig. 3; see Fig. S1 in the supplemental material). The THF concentration decreased due to strain YYL degradation, and the toxicity of THF to strain YYL was also alleviated. Hence, in strain YYL, expression of the activated antioxidant systems (sod) decreased due to weakened stress.

DISCUSSION

In the present study, we examined the metabolite changes in strain YYL in response to THF exposure and further analyzed the changes in metabolites and gene transcription levels in strain YYL utilizing THF as the sole carbon source. Significant changes in metabolites in strain YYL grown in TY compared with those in strain YYL grown in CY occurred only at 84 h, reflecting the exhaustion of yeast extract and the utilization of THF as the primary carbon source in mineral medium. These results demonstrate that the metabolome of strain YYL shows slight variations in response to THF exposure when THF is utilized as a cocarbon source. Furthermore, remarkable alterations of gene expression, including inhibition of cytochrome P450-dependent monooxygenase synthesis, transcription of degradation genes, and induction of activated antioxidant systems to reduce toxicity, were found over time in strain YYL using THF as the sole carbon source. Changes were also observed in a number of other genes and systemwide metabolic networks involved in glycolysis, the tricarboxylic acid (TCA) cycle, osmoregulation, and amino acid and nucleotide metabolism (Fig. 4).

FIG 4.

Altered metabolic pathways in Rhodococcus sp. strain YYL over the time course of growth in CT. Descriptions of the genes, abbreviations, and symbols: gyrB, DNA gyrase subunit B gene; rpoB, RNA polymerase sigma gene; pcrA, ATP-dependent DNA helicase gene; rhiB, ATP-dependent RNA helicase gene; pto, serine/threonine protein kinase gene; mdh, malate dehydrogenase gene; sdh, succinate dehydrogenase gene; gltA, citrate synthase gene; thm, tetrahydrofuran monooxygenase gene; cyp, cytochrome P450 monooxygenase gene; sod, superoxide dismutase gene; pfk, phosphofructokinase gene; gpd, glyceraldehyde-3-phosphate dehydrogenase gene; treS, trehalose synthase gene; otsA, trehalose-6-phosphatase phosphatase gene; F6P, fructose-6-phosphate; FBP, fructose-1,6-bisphosphate; DHAP, dihydroxyacetone phosphate; 1,3-BPG, 1,3-bisphosphoglycerate; acetyl-CoA, acetyl coenzyme A; blue, compounds detected from NMR data; purple, gene names; dotted arrows, proposed metabolic pathways; red triangles, significant increase; inverted green triangles, significant decrease; black squares, no significant variations. The two symbols adjacent to genes or metabolites represent the variations over time with growth in CT, with the first symbol representing the variation between 48 h and 84 h and the second one representing the variation between 84 h and 132 h. Glucose represents α-glucose and β-glucose combined.

Relation of glycolysis cycle to energy metabolism.

An increased rate of metabolism of glucose in response to exposure to hazardous organic pollutants is observed in most bacterial strains (2, 38). The present study also revealed decreased glucose (α-glucose and β-glucose combined) concentrations due to the increased metabolic rate in strain YYL in response to THF exposure; however, glucose accumulated over time during THF degradation with growth in CT (see Fig. S3 in the supplemental material). Glycolysis is well-known to be the potential main pathway of glucose consumption. Two significant genes involved in glycolysis, those encoding phosphofructokinase (pfk) and glyceraldehyde-3-phosphate dehydrogenase (gpd), showed marked changes with growth in CT over time. The trend for a variation in the expression of pfk differing from that of gpd may be due to anaerobic metabolism in strain YYL, which resulted in the production of lactate or ethanol in the glycolysis cycle (see Fig. S3 in the supplemental material). Lactate was observed to be a metabolite significantly altered over time with growth in TY compared with growth in CY or CT (Table 1). THF exposure reduced lactate accumulation in strain YYL, and THF degradation also resulted in the depletion of lactate from 48 h to 84 h with growth in CT (see Fig. S3 in the supplemental material). Thus, it can be inferred that THF is used by strain YYL as an extra carbon resource and inhibits lactate fermentation in the cytoplasm.

Relation of TCA cycle to THF degradation pathway.

THF can be transformed into 4-hydroxybutyraldehyde through the oxidation or hydration pathway, and then 4-hydroxybutyraldehyde is converted to succinate via succinate semialdehyde in Rhodococcus species (25). Therefore, succinate is the key intermediate linking the TCA cycle with the THF degradation pathway. THF exposure showed almost no effect on the succinate concentration; however, over time the succinate concentration was significantly altered in response to THF when strain YYL was grown in CT. The reason might be that intracellular THF degradation enhances succinate accumulation in strain YYL. This observation is in fair agreement with the observations made in Pseudomonas sp. strain HF-1 during its degradation of nicotine (2). Succinate accumulation would induce the transcription of succinate dehydrogenase (sdh), which catalyzes the conversion of succinate to fumarate, at a high level in strain YYL (Fig. 3), resulting in changes in the level of fumarate (P > 0.05; data not shown) with growth in CT over time. Two other genes involving in the TCA cycle, the citrate synthase gene (gltA) and the malate dehydrogenase gene (mdh), showed higher transcription levels due to the higher THF concentration. The different trends in the changes in transcription levels between sdh and mdh might be because sdh expression mainly depends on succinate accumulation and mdh expression mainly depends on the THF concentration. The results presented above reveal that acceleration of the TCA cycle is beneficial for THF degradation. The TCA cycle is usually observed to be accelerated during the process of degradation of hazardous organic pollutants (39).

Osmoadaptation-associated metabolism.

Trehalose is a nonreducing disaccharide and is the main compatible solute protecting cells against environmental stresses in bacteria (40, 41), Saccharomyces cerevisiae (42, 43), plants (44), and invertebrates (45). THF exposure caused a slight increase in trehalose levels in strain YYL, and the obvious accumulation of trehalose was exhibited in strain YYL during THF degradation. It has been suggested that four pathways, including OtsA/B (from UDP-glucose to glucose-6-phosphate to form trehalose-6-phosphate and UDP), TreY/Z (from glycogen to trehalose), TreS (from maltose to trehalose), and TreT (from ADP-glucose to trehalose) (46, 47), are involved in trehalose synthesis in bacteria. Among these, two pathways (OtsA/B and TreS) were detected in strain YYL on the basis of whole-genome scanning (data not shown). The level of otsA transcription was higher (P < 0.01) at 84 h than that at 48 h; however, the level of transcription of the treS gene showed a slight difference between 84 and 48 h. Trehalose biosynthesis in bacteria depends on bacterial physiological requirements under the given environmental stresses (48). Interestingly, two patterns of trehalose synthesis are flexibly used by strain YYL during THF degradation, and these may be beneficial to the balance between sugar metabolism and glycolysis.

Amino acids also play a significant role in osmotic regulation in response to environmental stress in bacteria (13, 49). Amino acids protect cells in many ways, including acting as osmolytes, which affect cellular water retention, thereby increasing the thermostability of proteins (50, 51). Glutamate and glutamine were shown to be compatible solutes in response to stress. For example, the levels of glutamate and glutamine accumulation were consistent with intermediate salinity (1.0 to 1.5 M NaCl) in the cell external environment (49). In the present study, THF exposure induced the marked accumulation of glutamate in strain YYL grown in TY, whereas THF degradation resulted in the significant depletion of both glutamate and glutamine over time in strain YYL grown in CY. These observations indicate that these two amino acids function as compatible solutes during THF exposure.

Moreover, it is particularly noteworthy that at 132 h a decreased level of TMAO was observed in strain YYL grown in TY and an increased level of TMAO was observed in strain YYL grown in CY. TMAO is known to be a compatible osmolyte for proteins (52, 53). Such a protective role was shown as TMAO accumulation, which has been observed in marine cyanobacteria inhabiting hypersaline environments (54). However, opposite the previous observations, in this study THF induced a significant decrease in TMAO together with marked THF depletion, which indicates that TMAO may be not used by strain YYL as a primary osmolyte in the presence of THF. However, TMAO was accumulated by strain YYL to resist resource exhaustion when this bacterium entered into decline phase. It is interesting to notice the different alterations of TMAO from trehalose as well, for which a reasonable explanation could be that strain YYL has multiple compatible solute systems in response to different stresses. The presence of different osmoadaptation mechanisms was also found in the Halobacteriales (55). A bacterium could also asymmetrically respond to different stresses (56).

Amino acid metabolism and nucleotide metabolism.

In this study, slight variations in the levels of most amino acids were induced in strain YYL in response to THF exposure with growth in TY compared with the levels induced in response to THF exposure with growth in CY, and significant variations only in the levels of alanine, glutamate, and glutamine were observed during THF degradation with growth in CT (Table 1). However, the total amino acid concentration in strain YYL utilizing THF was less with growth in CT than that in CY or TY (data not shown). The bacterium is usually found to decrease the amino acid concentration under stringent growth conditions (39).

Nucleotides are considered of great interest, as these molecules are involved in nucleic acid synthesis, energy production, protein modification, and antisense oligonucleotide synthesis (57, 58). The effect of THF exposure on the nucleotide concentration in strain YYL was slightly greater during growth in TY than during growth in CY. A reduction in nucleotide biosynthesis is typically observed in response to various stress conditions (59). The levels of NAD⁺ and CTP were both significantly reduced at 84 h (1.92 mM THF/h · g [dry weight] strain YYL) compared with those at 48 h (1.82 mM THF/h · g [dry weight] strain YYL) with growth in CT. RT-qPCR results showed that the levels of pcrA and rpoB transcription were slightly enhanced with increased THF degradation in CT (P > 0.05). According to the results obtained in the present study, the reason for nucleotide variations might be bacterial cell growth and not THF exposure.

In summary, slight variations in the strain YYL metabolome between growth in CY and that in TY revealed that THF had little toxicity to strain YYL. However, the metabolic profile exhibited relatively large fluctuations during THF degradation in CT. In addition, RT-qPCR was further used to explore the expression of the genes involved in the significantly changed metabolite profiles, as shown in the NMR data for strains grown in CT. These results suggest that the accumulation of trehalose, enhancement of the conversion of succinate to malate in the TCA, and acceleration of protein and nucleotide synthesis might be involved in THF degradation. Detailed information on the metabolomics and expression of selected genes could be useful for understanding not only THF degradation/catabolism mechanisms but also the effects of THF exposure on bacterial cells.