Introduction

Azo dyes, whose utilization and discharge into the environment by textile industry increase annually, are currently worldwide pollutants (Jin et al. 2014; Singh and Arora 2011). Most azo dyes are toxic, mutagenic, and carcinogenic, showing potential risks to human health (Hafshejani et al. 2014; Stolz 2001). Different technologies have been developed to remediate azo dye contamination, including adsorption, membrane filtration, photocatalysis, and microbiological or enzymatic decomposition (Forgacs et al. 2004; Willetts and Ashbolt 2000). Among these, microbiological technologies are the most promising due to their low cost, high efficacy, and environmentally friendly nature (Khan et al. 2014b; Silva et al. 2014). Many microorganisms are capable of degrading azo dyes through the azo reduction process mediated by azoreductase, which is expressed in vivo by bacteria and is considered as a key enzyme (Hu 1994; Saroj et al. 2014; Silva et al. 2014; Yang et al. 2013). In most cases, complete azoreduction by bacteria is achieved under anaerobic or microaerophilic conditions, as previously observed in Stenotrophomonas sp. strain BHUSSp X2 (Kumari et al. 2016) and Sphingomonas sp. strain BN6 (Kudlich et al. 1997). However, the products generated by azoreduction under microaerophilic conditions are less toxic than those produced under aerobic conditions (Hafshejani et al. 2014; Khan et al. 2014a). Bacteria within the Shewanella genus, particularly Shewanella oneidensis MR-1, possess the ability to degrade a diverse range of azo dyes under microaerophilic conditions, such as amaranth (Le Laz et al. 2014), methyl red (Yang et al. 2013), methyl orange (MO), and naphthol green B (Cao et al. 2013). However, azo dye degradation by MR-1 is largely inhibited by dissolved oxygen (DO) (Hong et al. 2007). The mechanisms underlying the biodegradation of azo dyes by S. oneidensis MR-1 have not been extensively explored nor have the mechanisms underlying the inhibition by DO, particularly at metabolome and transcriptome levels. Knowledge of gene expression alterations involved in azoreduction, as well as resulting metabolite changes, will help us gain a better understanding of the mechanisms involved in azoreduction under different conditions (Tyagi et al. 2014; Wang et al. 2014; Zhang et al. 2011).

Recently, metabolomics has been widely adopted to generate profiles of endogenous small molecular compounds in living organisms exposed to exogenous stimuli (Rochfort 2005; Wishart 2005). Nuclear magnetic resonance (NMR) is preferred for metabolomic profiling compared with chromatography-mass spectrometry methods due to its simpler sample preparation procedures that result in less disturbance of intracellular metabolites (Lindon et al. 2001). NMR has been utilized by many researchers to investigate the responses of organisms to various environmental stresses, such as heavy metals (Tyagi et al. 2014), organic pollutants (Hong et al. 2014), and extreme temperatures (Ellis et al. 2014; Yanagisawa et al. 2014). Currently, RNA-Seq has been widely employed as a powerful approach to study the transcriptional levels of genes in plants (Pradhan et al. 2014), bacteria (Liu et al. 2014; Mazin et al. 2014), and mammalian cells (Li et al. 2015) due to its higher sensitivity and ability to identify novel genes, transcripts, and single-nucleotide polymorphisms (SNPs).

The primary objectives of this study were to investigate the molecular mechanisms underlying MO biodegradation by strain MR-1 under microaerophilic conditions and to explore the genetic mechanisms involved in azoreduction inhibition caused by DO under aerobic conditions. 1H NMR metabolome and RNA-Seq analyses were performed to reveal bacterial metabolic profiles and gene transcription levels, respectively. The results of this study allow us to gain a systematic understanding of the mechanism underlying azo dye biodegradation by bacteria under different conditions.

Materials and methods

Strain and culture conditions

The MO-degrading strain S. oneidensis MR-1 (ATCC® 700550™) was granted by Prof. Haichun Gao of Zhejiang University. Strain MR-1 was inoculated into 30 mL of liquid discoloration medium, containing 1.0 g L−1 lactose, 2.0 g L−1 sodium lactate, 1.5 g L−1 KH2PO4, 0.5 g L−1 NaCl, and 0.1 g L−1 NH4Cl, and was cultured in an shake incubator maintained at 30 °C. To unravel the mechanisms involved in metabolic and transcriptional changes caused by MO and DO, strain MR-1 was incubated in 30 mL of liquid discoloration medium under three different conditions: microaerophilic conditions with MO (MAMO), microaerophilic conditions without MO (MA), or aerobic conditions with MO (AMO). Microaerophilic conditions were implemented by maintaining the incubator under static conditions, during which the concentration of DO in the flasks varied between 0.9 and 1.6 mg L−1 (Işik and Sponza 2003). Aerobic conditions were implemented by holding the incubator at a rotation rate of 180 rpm, during which the concentration of DO in the flasks varied between 4.8 and 7.2 mg L−1 (Işik and Sponza 2003). DO concentration was determined as previously reported (Carbo et al. 2015). One 150-mL Erlenmeyer flask with a plastic sealing membrane was used for each sample in the MA and MAMO groups and flasks with gauze were used for the AMO group. Eight replicates under each culture condition were prepared for metabolome analysis, and two replicates under each culture condition were used for transcriptome analysis.

Sample preparation for 1H NMR

Based on our experimental results, strain MR-1 entered exponential phase within 3 h in all three treatment groups (Fig. S1 in the Supplementary Material), and the degradation ratio of MO under MAMO conditions was approximately 80% (Fig. S2 in the Supplementary Material). Therefore, bacterial samples were collected after 3 h of cultivation. The flasks were first chilled on ice to quench metabolism, and then, bacterial cells were harvested. Metabolite extraction was performed as previously described (He et al. 2014). After methanol removal in vacuo, all extracts were lyophilized overnight. Solid extracts were reconstituted in 600 μL of a Na+/K+ phosphate buffer containing 80% D2O, 0.001% sodium 3-trimethylsilyl [2,2,3,3-d4] propionate (TSP), and 0.01% NaN3. After centrifugation at 4 °C for 20 min, the supernatants were transferred into 5-mm NMR tubes for NMR analysis.

1H NMR spectra acquisition

The NMR spectra of 24 samples were recorded at 600.13 MHz on a Bruker Avance III 600-MHz spectrometer with a cryogenic probe (Bruker BioSpin, Karlsruhe, Germany). A standard 1H NMR spectrum was obtained at 298 K by employing the first increment of the NOESY pulse sequence (RD-90°-t 1-90°-t m-90°-acquisition; RD = 2 s, t 1 = 4 μs, t m = 200 ms). The 90° pulse length was set to 11.5 μs. Water suppression was achieved through presaturation at the water resonance frequency. For each spectrum, 128 transients were collected into 32 K data points with a spectral width of 20 ppm. For further resonance assignments, a series of 2D NMR spectra 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) was acquired (Ye et al. 2012).

NMR data processing and multivariate data analysis

An exponential window function with a 1-Hz line broadening factor was added to the free induction decays (FIDs) prior to Fourier transformation. All 1H NMR spectra were manually phased, baseline-corrected, and calibrated to TSP (δ 0.00) using Topspin (V3.0, Bruker BioSpin, Karlsruhe, Germany). Data integration was carried out by dividing the spectrum from 9.7 to 0.5 ppm into discrete regions with a wide bucket of 0.004 ppm (2.4 Hz) using the AMIX package (V3.9.2, Bruker BioSpin, Karlsruhe, Germany). The region from 4.7 to 4.9 ppm was abandoned to eliminate the effects of the residual water signal. Data normalization was performed based on the volumes and optical density (OD) values of the bacteria cultures. Following normalization, multivariate data analysis was conducted using the SIMCA-P+ package (v11.0 and v12.0, Umetrics AB, Umea, Sweden). Principal component analysis (PCA) was performed to identify outliers within the mean-centered NMR data. Orthogonal projection to latent structure discriminant analysis (OPLS-DA) models were subsequently constructed with the normalized 1H NMR spectra data. Models were validated through a sevenfold cross-validation method (Bylesjo et al. 2006) and were further validated by cross-validation (CV)-ANOVA (Eriksson et al. 2008). The quality and reasonability of the model were monitored by parameters Q 2 (predicted ability) and R 2 (interpret ability). For statistical analysis, the cutoff value used in this study was 0.707. A Q 2 value greater than 0.707 and a p value smaller than 0.05 (obtained from CV-ANOVA) indicated that the OPLS-DA model was valid. Loadings indicating the metabolites responsible for differentiation were calculated via back-transformation (Cloarec et al. 2005) and were plotted with a color-coded correlation coefficient (r) using Matlab Scripts (the MathWorks, Natick, USA). The absolute value of the correlation coefficient (|r|) is represented by different colors in the coefficient plot. Warm colors (e.g., red) indicate greater significance of the differentiation contributed by variations between groups than cold colors (e.g., blue).

Sample preparation for RNA-Seq analysis

Individual cultures of strain MR-1 were rapidly transferred into 50-mL centrifuge tubes after 3 h of cultivation, followed by centrifugation at 8000 rpm (4 °C) for 10 min; supernatants were subsequently discarded. Total RNA extraction was carried out using an RNAiso Plus Kit (Takara, Dalian, China). RNA concentrations were determined using Nanodrop (Thermo Fisher Scientific, Waltham, USA), and RNA quality was checked by non-denaturing agarose gel electrophoresis. To obtain qualified complementary DNA (cDNA), the concentrations of RNA are at least 300 ng μL−1 and the lanes of 16S and 23S on the gel must be clear and intact. Following DNase I treatment, the GeneRead ribosomal RNA (rRNA) Depletion Kit (QIAGEN, Frankfurt, Germany) was used to remove rRNAs. mRNA was fragmented into short fragments in fragmentation buffer (Thermo Fisher Scientific, Waltham, USA). The cleaved RNA fragments were used as templates to amplify the first-strand cDNA (First-Strand cDNA Synthesis Mix Kit, TransGen, Beijing, China). The second strand synthesis was conducted using DNA Polymerase I and RNase H. The resulting fragments were resolved in elution buffer for end reparation, single-nucleotide adenine addition, and ligation of adapters. Following quantification and qualification of the cDNA library by using Agilent 2100 Bioanalyzer (Agilent, Santa Clara, USA) and ABI StepOnePlus Real-Time PCR System (Thermo Fisher Scientific, Waltham, USA), the library was sequenced using Illumina HiSeq™ 2000 (Illumina, San Diego, USA).

Transcriptome data processing and analysis

Raw reads produced by Illumina HiSeq™ 2000 underwent quality control (see detailed information in the Supplementary Material) and were then filtered into clean reads, which were aligned to the reference genome of S. oneidensis MR-1 (GenBank accession: NC_004347). The alignment data were utilized to calculate the distribution of reads on reference genes and the mapping ratio. To screen differentially expressed genes (DEGs), PCA and correlation analysis were performed to achieve deep data processing. The Noiseq software package, which can identify positive changes and false-positive changes, was applied to reveal the DEGs between AMO and MAMO and between MAMO and MA as previously described (Tarazona et al. 2011). DEGs were defined based on the criteria of M ≥ 2 (or log2 ratio ≥1) and p ≥ 0.8, of which M indicates the fold change and p indicates the probability. Clean data obtained from RNA-Seq have been deposited in the NCBI Sequence Read Archive database (accession: SRA486119).

RT-qPCR verification of transcriptome data

To verify the validity of the transcriptome data, the expression levels of 20 genes under both MAMO and AMO conditions were verified by RT-qPCR. The reaction system was 20 μL, containing 10 μL of SYBR Premix Ex Taq (Takara, Dalian, China) 1 μL of primers, 1 μL of cDNA, and 8 μL of ddH2O. Specific primers (Table S1 in the Supplementary Material) were designed using the NCBI-Primer BLAST tool (see detailed information in the Supplementary Material). The 16S rRNA gene was employed as the internal standard and was amplified using the primers F1369 (CGGTGAATACGTTCYCGG) and R1492 (GGWTACCTTGTTACGACTT). The log2 2−ΔΔCT method was applied to calculate gene expression levels. Correlation analysis of data obtained from RNA-Seq and RT-qPCR is displayed in the Supplementary Material (Fig. S3).

Results

1H NMR spectra of intracellular extracts from strain MR-1

Typical 1H NMR spectra for the intracellular extracts obtained from the three treatment groups are shown in Fig. 1. Metabolite resonances were assigned based on 2D NMR incorporating both 1H and 13C data (Table S2 in the Supplementary Material) and were further verified according to the literature (Fan 1996; Fan and Lane 2008). A total of 25 metabolites were identified, including mannose, lipoprotein lipids, aliphatic organic acids (lactate, acetate, succinate and formate), amino acids (alanine, glutamate, and allothreonine), nucleosides and nucleotide metabolites (uracil, uridine, cytidine, adenosine, guanosine, adenosine 2′,3′-cyclic phosphate, and NAD+), amines (putrescine, acetamide, ethanolamine, O-phosphoethanolamine, and propan-2-amine), alcohols (methanol and isopropanol), acetoin, and isoamyl acetate .

Fig. 1
figure 1

Typical 600 Hz 1H NMR spectra of extracts from S. oneidensis MR-1 grown under MA (a), MAMO (b), and AMO (c) conditions in 30 mL of liquid discoloration medium (n = 8). Resonance assignments are listed in Table S1 in the Supplementary Material

Full size image

Metabolite changes caused by MO and DO

To gain further insight into differential metabolites by strain MR-1 under different culture conditions, a multivariate data analysis was performed. OPLS-DA comparisons between AMO and MAMO as well as MAMO and MA were carried out. The scores and corresponding coefficient plots for dominant metabolites are shown in Fig. 2. In total, the concentrations of nine metabolites significantly increased (p < 0.05) under AMO conditions compared with those concentrations under MAMO conditions (Table 1). Among these nine metabolites, NAD+, an important coenzyme, showed the most positive change with a correlation coefficient (|r|) of 1.49. The concentrations of putrescine and lipoprotein lipids also significantly increased (p < 0.05) under aerobic conditions (with |r| values of 1.32 and 1.26, respectively). However, metabolite concentrations under MAMO conditions exhibited no significant changes (p > 0.05) compared with those under MA conditions (data not shown). These changes are most likely explained by the recovery of MR-1 under MAMO conditions, accompanied by the biodegradation of MO over time.

Fig. 2
figure 2

OPLS-DA score (a) and coefficient-coded loading plot (b) obtained from 1H NMR data for S. oneidensis MR-1 aqueous extracts under AMO (black squares) and MAMO (red dots) conditions. t[2]O cross-validated score value, t[1]P model score value. In the loading plot, red indicates significantly increased (p < 0.05) metabolites and blue indicates no significant changes (p > 0.05). The metabolite key is displayed in Table 1 (Color figure online)

Full size image
Table 1 Significantly changed metabolites in S. oneidensis MR-1 growing under AMO conditions compared to MAMO conditions
Full size table

DEGs due to MO exposure

Over 4000 genes were detected and analyzed in each sample, within which 145 expressed genes were upregulated and 26 genes were downregulated under MAMO compared to MA conditions (Fig. 3a). DEGs related to certain crucial metabolic processes under MAMO conditions are shown in Fig. 4a. In the presence of MO, the expression levels of gene azoR (encoding azoreductase) underwent a significant increase (log2 ratio (MAMO/MA) = 4.26, p > 0.8). Gene SO1755 (encoding phosphoglucomutase) and yqhD (encoding alcohol dehydrogenase) were upregulated, indicating improved glucose metabolism. Transcriptional levels of gene acnD (encoding aconitase), which is associated with the tricarboxylic acid (TCA) cycle, also increased. In terms of amino acid metabolism, genes involved in the biodegradation of lysine (liuE), leucine (liuB/D), phenylalanine (hppD, phhA, hmgA), isoleucine (ivdA/C/E), valine (ivdB/F), and histidine (hutU) were upregulated. However, the expression of argA and argB, which are responsible for arginine biosynthesis, declined. The expression levels of argF (encoding ornithine carbamoyltransferase), which is involved in the urea cycle, were also reduced. Notably, the expression levels of yceJ, encoding cytochrome (Cyt) b561, and bfd, which encodes (2Fe-2S)-binding protein, were enhanced, implying an incremental requirement for the redox process of in strain MR-1 under MAMO conditions.

Fig. 3
figure 3

Scatter plot of all expressed genes under MAMO compared to MA (a) and AMO (b) compared to MAMO conditions (n = 2). Orange triangles indicate significantly upregulated genes, and blue squares indicate significantly downregulated genes (M ≥ 2, p ≥ 0.8). Gray dots indicate no significant changes (M < 2) (Color figure online)

Full size image
Fig. 4
figure 4

Partial DEGs under MAMO (a) and AMO (b, c) conditions. Log2 ratio values are greater than 1, indicating M ≥ 2. Positive values indicate upregulation, and negative values indicate downregulation. The probability (p) is greater than 0.8

Full size image

DEGs caused by a high concentration of DO

Compared with MAMO conditions, 312 genes were upregulated and 759 genes were downregulated (Fig. 3b) under AMO conditions. The DEGs taking part in certain crucial metabolic processes are shown in Fig. 4b, c. Transcriptional levels of azoR significantly declined under AMO conditions (log2 ratio (AMO/MAMO) = −1.9, p > 0.8). The expression levels of glyceraldehyde 3-phosphate dehydrogenase (gapA) were downregulated (Fig. 4b), whereas gene gapB, which encodes the same product as gapA, was upregulated (Fig. 4b), suggesting balanced glycolysis. Downregulation of genes acnD and sucD (encoding succinyl-CoA synthase) resulted in a weakened TCA cycling rate. However, upregulation of genes encoding argininosuccinate synthase (argG) and encoding argininosuccinate lyase (argH) indicated elevated activity of the urea cycle. Genes fadE2 and fadA/I, related to fatty acid oxidation, were notably upregulated. Amino acid biodegradation rate may be greatly reduced under AMO conditions, which is reflected in the downregulation of ivdA/B/C and liuA/B/D/E. In addition, amino acid biosynthesis may be improved under AMO conditions, suggested by upregulation of gltB/D, asnB, argA/B, thrA/B/C, leuA/B/C/D, trpD/E/G, hisB/C/G, and ilvD/G/M, which are responsible for the biosynthesis of glutamate, asparagine, arginine, threonine, leucine, tryptophan, histidine, and isoleucine, respectively. Expression of the genes encoding DNA primase (dnaG), DNA replication protein (repA), and DNA mismatch repair protein (mutL) greatly increased, which signified improved DNA replication processes. Significant reductions (M ≥ 2, p ≥ 0.8) in the genes responsible for electron transfer were observed as well, including yceJ, frdB (fumarate reductase), nuo (NADH dehydrogenase), omcA (Cyt c), fdx (2Fe-2S ferredoxin), and fdhT (Cyt c oxidase).

Discussion

Effects of MO exposure on bacterial metabolism

Changes in glycolysis in various bacteria have been reported in response to diverse environmental stresses. For example, alterations have been shown in Escherichia coli (Payen et al. 2016; Rui et al. 2010) and Staphylococcus aureus (Lechnera et al. 2014). In this study, increased expression of phosphoglucomutase (SO1755), which is responsible for converting glucose-1-phosphate (G1P) into glucose-6-phosphate (G6P), was clear evidence of improved glycolysis activity. Moreover, the increased levels of aldehyde dehydrogenase (exaC, Fig. 4a) and alcohol dehydrogenase (yqhD, Fig. 4a) reflected accelerated glycolysis as well (Fig. 5). Based on these observations, we deduced that MO (even at 0.1 g L−1) stress induces MR-1 to consume greater amount of glucose to supply energy.

Fig. 5
figure 5

Alterations in metabolic pathways of S. oneidensis MR-1 under AMO and MAMO conditions. Genes, abbreviations, and symbols: fad includes fadA, fadI, and fadE2. thr includes thrB and thrC. ilv includes ilvG, ilvM, and ilvD. leu includes leuA, leuB, leuC, and leuD. glt includes gltD and gltB. G1P glucose-1-phosphate, G6P glucose-6-phosphate, AzoR azoreductase, CoA ubiquinone, Cyt cytochrome, Cyt cR cytochrome c reductase, Cyt cO cytochrome c oxidase; red triangles, significant upregulation (M ≥ 2, p ≥ 0.8); green inverse triangles, significant downregulation; red squares, significantly increased metabolites (p > 0.05) (Color figure online)

Full size image

Bacteria generally regulate amino acid metabolism under adverse conditions (Seo et al. 2013). In this study, increased amino acid biodegradation was observed at the transcriptional level. However, a slight but not significant decrease (p > 0.05) in the concentration of amino acids produced by MR-1 occurred in response to MO exposure (Table S3 in the Supplementary Material). This may occur because biodegradation in response to MO over time allows MR-1 to gradually return to a normal growth state (Figs. S1 and S2 in the Supplementary Material). Rapid amino acid biodegradation inevitably produces sufficient acetyl-CoA (lysine, phenylalanine and leucine) and succinyl-CoA (methionine, isoleucine and valine), which are shunted into the TCA cycle (Fig. 5) and improve this process.

The urea cycle and the TCA cycle play important roles in nitrogen metabolism and carbon metabolism, respectively, and these cycles are frequently coupled via a shared source of fumarate (Allen et al. 2011). In this study, acceleration of the TCA cycle (acnD) and deceleration of the urea cycle (argF) were noted. The TCA cycle is commonly accelerated during the degradation of hazardous organic pollutants (Seo et al. 2013). Obvious changes of TCA cycle in the present study were in accordance with observations obtained in Rhodococcus sp. strain YYL during tetrahydrofuran degradation (He et al. 2014). Alterations in the TCA cycle may directly or indirectly result in the upregulation of aldehyde dehydrogenase and acetyl-CoA synthetase (acs, which converts acetate into acetyl-CoA, Figs. 4a and 5). Deceleration of the urea cycle slows down fumarate utilization. This reduction may also be responsible for decreased arginine biosynthesis (argA and argB).

In addition to increased azoreductase (azoR) levels, the presence of MO also induced an elevation in cytochrome b561 (yceJ)- and (2Fe-2S)-binding protein (bfd) levels, which are primarily associated with complexes of the respiratory electron transfer chain. Azo dye degradation was previously assumed to an extracellular electron transfer process involving multicomponents located on membrane (Brige et al. 2008; Hong et al. 2007). Thus, azoreduction may be closely related to the respiratory chain.

Effects of a high concentration of DO on bacterial metabolism

A prominent finding in this study was the acceleration of fatty acid oxidation caused by a high concentration of DO. Long-chain fatty acid oxidation is the primary pathway that supplies energy for bacteria and certain protists (Kim et al. 1989). Therefore, the notable enhancement of acyl-CoA dehydrogenase (fadE2) and β-ketoacyl-CoA thiolase (fadI/A) was potentially triggered by high energy demands for MR-1 to maintain normal growth. Additionally, the obvious accumulation of lipoprotein lipids, alanines, NAD+, and putrescine was observed under AMO conditions. Lipoprotein lipids are important components of the cell membrane. In bacteria, under certain circumstances, alanine serves as a growth factor to improve growth levels (Snell and Guirard 1943). NAD+ may be involved not only in glycolysis and the TCA cycle to produce NADH, but also in DNA replication and DNA mismatch repair processes (Fouquerel and Sobol 2014). Putrescine, whose increase may be attributable to the upregulation of ornithine decarboxylase (speC, Fig. 4b), plays an essential role in bacterial DNA, RNA, and protein biosynthesis as well as cell division (Ando et al. 1994). Since biomass of facultative anaerobe is usually greater when they are incubated under aerobic conditions than under microaerophilic conditions (Romano et al. 2015), increases in these metabolites may be primarily related to bacterial growth because more biomass was produced under AMO conditions (Fig. S1 in the Supplementary Material). Moreover, the increased demand for putrescine accompanying bacterial growth may enhance the urea cycle under AMO conditions (Fig. 5).

According to a previous study, increased allothreonine reflects drastic amino acid synthesis and cell growth (Jhee et al. 2002). Thus, the accumulation of allothreonine (Table 1) and the upregulation of genes responsible for amino acid biosynthesis suggest that strain MR-1 under AMO conditions was starved for amino acids. Moreover, improved amino acid biosynthesis may be responsible for reductions in the TCA cycle (acnD and sucD) due to the consumption of 2-ketoglutarate, oxaloacetate, and pyruvate (Fig. 5). For instance, the accumulation of acetoin (Table 1) under AMO conditions was accompanied by the mass depletion of pyruvate (Xiao and Lu 2014); accelerated arginine biosynthesis (argA and argB) may also result in enhancement of the urea cycle. Amino acid metabolism, the TCA cycle, and the urea cycle generally interact with each other in bacteria during cell growth (Tremaroli et al. 2009).

In addition to the drop in azoreductase, another prominent finding was the decreased expression levels of Cyt b561 (yceJ), NADH dehydrogenase (nuo), 2Fe-2S ferredoxin (fdx), Cyt c (omcA), and Cyt c oxidase (fdh) under AMO conditions. All of these proteins are involved in the respiratory electron transfer chain. Thus, higher DO appears to heavily impact the respiratory electron transfer chain in strain MR-1, which may be responsible for the inhibition of MO reduction.

Proposed mechanism underlying MO biodegradation and inhibition by DO

Azoreductase in S. oneidensis MR-1 is FMN-dependent and preferred by NADH as an electron donor (Yang et al. 2013). Therefore, based on our observations in this study, we made the following assumptions: (1) azoreductase anchored on the plasma membrane is likely combined with complex I, and (2) azoreduction is a short electron process that overlaps with the initial section of the respiratory electron transfer chain (Fig. 6). During MO biodegradation, dioxygen may compete with MO as an electron acceptor. Under MAMO conditions, the enhancement of the TCA cycle provides more NADH to the electron transfer chain (Fig. 5), which, together with the increase in Cyt b561, enhances the overall electron transfer from NADH to both MO and dioxygen. Meanwhile, the concentration of DO under MAMO conditions was low enough and did not compete with MO as a final electron acceptor. However, under AMO conditions, the decreases in NADH dehydrogenase and the TCA cycle inevitably diminished overall electron transfer from NADH to MO and higher concentration of DO under AMO conditions may lead to greater electron flux to dioxygen rather than to MO. Furthermore, the large amounts of FADH2 produced by fatty acid oxidation under AMO conditions directly participate in electron transfer to generate adenosine triphosphate (ATP), potentially accelerating the flux of electrons from coenzyme Q (CoQ) to dioxygen as well (Figs. 5 and 6). The decrease in FMN reductase (fre, converting FMNH2 to FMN, Fig. 4c) under AMO conditions may also result in reduced azoreduction.

Fig. 6
figure 6

A hypothetical pathway for azoreduction coupled with complex I (NADH-Q) of the respiratory chain in S. oneidensis MR-1. CoQ coenzyme Q or ubiquinone, AzoR azoreductase, MO methyl orange, Suc succinate, Fum fumarate, Cyt cytochrome, ATP Syn ATP synthase complex, OM outer membrane, CM cytoplasmic membrane

Full size image

In summary, the findings in this study revealed the following: (1) an integrated relationship between crucial metabolic processes and MO biodegradation, and (2) in terms of MO degradation, strain MR-1 showed a tendency to maintain growth in the presence of high concentrations of DO.