UNCO

RREC

TED

PROO

F

Environmental Pollution xxx (2018) xxx-xxx

Contents lists available at ScienceDirect

Environmental Pollutionjournal homepage: www.elsevier.com

Proteomics and genetic analyses reveal the effects of arsenite oxidation on metabolicpathways and the roles of AioR in Agrobacterium tumefaciens GW4☆

Kaixiang Shi, Qian Wang1, Xia Fan, Gejiao Wang∗

State Key Laboratory of Agricultural Microbiology, College of Life Science and Technology, Huazhong Agricultural University, Wuhan, 430070, PR China

A R T I C L E I N F O

Article history:Received 27 July 2017Received in revised form 2 January 2018Accepted 2 January 2018Available online xxx

Keywords:Arsenite oxidation regulatorComparative proteomicsPhosphateArsenite resistanceEnergy

A B S T R A C T

A heterotrophic arsenite [As(III)]-oxidizing bacterium Agrobacterium tumefaciens GW4 isolated fromAs(III)-rich groundwater sediment showed high As(III) resistance and could oxidize As(III) to As(V). TheAs(III) oxidation could generate energy and enhance growth, and AioR was the regulator for As(III) oxi-dase. To determine the related metabolic pathways mediated by As(III) oxidation and whether AioR regulatedother cellular responses to As(III), isobaric tags for relative and absolute quantitation (iTRAQ) was performedin four treatments, GW4 (+AsIII)/GW4 (-AsIII), GW4-ΔaioR (+AsIII)/GW4-ΔaioR (-AsIII), GW4-ΔaioR(-AsIII)/GW4 (-AsIII) and GW4-ΔaioR (+AsIII)/GW4 (+AsIII). A total of 41, 71, 82 and 168 differen-tially expressed proteins were identified, respectively. Using electrophoretic mobility shift assay (EMSA) andqRT-PCR, 12 genes/operons were found to interact with AioR. These results indicate that As(III) oxidationalters several cellular processes related to arsenite, such as As resistance (ars operon), phosphate (Pi) metab-olism (pst/pho system), TCA cycle, cell wall/membrane, amino acid metabolism and motility/chemotaxis. Inthe wild type with As(III), TCA cycle flow is perturbed, and As(III) oxidation and fermentation are the mainenergy resources. However, when strain GW4-ΔaioR lost the ability of As(III) oxidation, the TCA cycle isthe main way to generate energy. A regulatory cellular network controlled by AioR is constructed and showsthat AioR is the main regulator for As(III) oxidation, besides, several other functions related to As(III) areregulated by AioR in parallel.

© 2017.

1. Introduction

Arsenic (As) is widespread in the environment and is related tohuman health. The most common As species are arsenite [As(III)]and arsenate [As(V)] (Cai et al., 2009). Of these two species, As(III)is more toxic than As(V) (Zhu et al., 2014). Microbes are the prin-cipal drivers of arsenic transformation, and the oxidation of As(III)to As(V) results in less toxic and less mobile species (Stolz et al.,2006). Some As(III)-oxidizing bacteria have evolved to use the en-ergy generated from As(III) oxidation in nature, especially chemoau-totrophic As(III)-oxidizing bacteria that assimilate inorganic carbonusing As(III) as the electron donor (Rhine et al., 2007; Santini etal., 2000). In addition, some heterotrophic As(III)-oxidizing strains,such as Hydrogenophaga sp. NT-14 and A. tumefaciens GW4 (vandenHoven and Santini, 2004; Wang et al., 2015), could also gain en-ergy from the

? This paper has been recommended for acceptance by Dr. Jorg Rinklebe.∗ Corresponding author.Email address: gejiao@mail.hzau.edu.cn (G. Wang)1 Current address: Department of Land Resources and Environmental Sciences,Montana State University, Bozeman, MT, 59717, USA.

As(III) oxidation reaction, suggesting that some heterotrophic As(III)oxidizers could generate energy from both carbohydrate metabolismand As(III) oxidation; however, the mechanisms of energy generationfor heterotrophic As(III) oxidizers remain unknown.

As and phosphorus are both members of Group 15 on the peri-odic table, indicating they are structural analogs and that As(V) andphosphate may be co-metabolized (Chen et al., 2015). In H. arseni-coxydans ULPAs1, phosphate import ATP-binding protein (PstB1)and putative phosphate uptake regulator (PhoU1) were induced by 3or 4 fold in the presence of As(III) (Weiss et al., 2009). Of 54 or-ganisms containing the As(III) oxidase aioBA genes, 11 genes encod-ing proteins related to various functions associated with phosphorusacquisition were located directly adjacent to, or very nearby, the aiogenes (Kang et al., 2012). These genes were expressed in response tophosphate starvation and included pst and pho genes encoding pro-teins for high-affinity Pi transport or regulatory functions (Li et al.,2013). The pst1/pho1 system was located within As islands, whilethe pst2/pho2 system was localized distantly on the respective chro-mosomes (Li et al., 2013). In A. tumefaciens 5A, phosphate starva-tion regulators PhoB1 and PhoB2 were involved in the expression ofthree-component signal transduction system including the periplasmicAs(III)-binding protein AioX, the sensor kinase AioS, its cognate re-sponse regulator AioR and AioBA (Kang et al., 2012), and the ex-pression of aioBA was proven to be regulated by PhoB in Halomonas

https://doi.org/10.1016/j.envpol.2018.01.0060269-7491/© 2017.

UNCO

RREC

TED

PROO

F

2 Environmental Pollution xxx (2018) xxx-xxx

sp. HAL1 (Chen et al., 2015). These observations revealed that therewas a link between As(III) oxidation and the pst/pho system.

Previously, cellular global responses to As(III) have been ana-lyzed by proteomic and transcriptomic studies. In chemoautotrophicAs(III)-oxidizing strains, the ribulose-1,5-biphosphate carboxylase(RuBisCo) and fructose-1,6-biphosphatase involved with CO2 fixa-tion were up-regulated in the presence of As(III), indicating that Asmetabolism, carbon assimilation, and energy acquisition were linkedin chemoautotrophic As(III)-oxidizing bacteria, and As(III) oxida-tion was a mechanism for generating bioenergy (Bryan et al., 2009;Andres et al., 2013). In the heterotrophic As(III)-oxidizing strainsHerminiimonas arsenicoxydans ULPAs1 and Pseudomonas aerugi-nosa PAO1, the proteins involved in carbohydrate metabolism wereall up-regulated (Bryan et al., 2009; Andres et al., 2013). The in-creased expression of enzymes involved in carbohydrate metabolismrepresent an important means of adaptation during stress such as thatresulting from the exposure of bacteria to heavy metals (Mallik etal., 2012). In addition, the differentially expressed proteins and geneswere mainly involved with As(III) oxidation, As resistance and efflux,stress response, exopolysaccharide synthesis and phosphate mecha-nism (Bryan et al., 2009; Andres et al., 2013). Another common de-fense mechanism against As(III) exposure results from the inductionof the As resistance and efflux (ars operon) (Kashyap et al., 2006).As(V) was reduced by a cytoplasmic As(V) reductase (ArsC) to themore toxic As(III), and then As(III) was extruded by a membrane-as-sociated ArsB or Acr3 efflux pump (Kashyap et al., 2006). In addition,the alteration of membrane permeability has been proven to be anotherAs(III) resistance mechanism of Gram-negative bacteria (Weiss et al.,2009). It leads to the adsorption of As(III) into the lipopolysaccharidelayer of the outer membrane of these bacteria, and strains produce athick capsule of exopolysaccharides, which has been shown to scav-enge arsenic as granules (Weiss et al., 2009). Overall, As(III) oxida-tion in heterotrophic As(III) oxidizers was mainly detoxification andstress response mechanisms.

In a previous study, we isolated the highly As(III) resistant [min-imal inhibitory concentration (MIC) = 25mM] and As(III)-oxidizingbacterium A. tumefaciens GW4 from As-enriched groundwater sedi-ments (Fan et al., 2008). Unlike most of the heterotrophic As(III)-oxi-dizing bacteria using As(III) oxidation as a detoxification process, theAs(III) oxidation of strain GW4 enhanced the bacterial growth, andthe strain showed positive chemotaxis toward As(III) (Wang et al.,2015; Shi et al., 2017). However, the mutant strain GW4-ΔaioR failedto demonstrate increased growth, and its As(III) oxidation and As(III)chemotaxis phenotypes were both disrupted (Shi et al., 2017). It ap-peared that the effect of As(III) oxidation in strain GW4 was differ-ent from that in all the well-recognized heterotrophic and chemoau-totrophic As(III)-oxidizing strains. Genes aioBA are regulated byAioXSR (Liu et al., 2012; Li et al., 2013). It has been revealed thatAioR was the key factor to regulate bacterial As(III)-oxidation inA. tumefaciens 5A (Liu et al., 2012). However, AioR was also ex-pressed at a low level without As(III) in A. tumefaciens GW4 (Fig.S1). The consensus DNA-binding sequenceGT[TC][AC][CG][GCT][AG][AG]A[ACT][CGA][GCT][GTA]AAChas been documented for the regulator AioR (Shi et al., 2017), and theAioR putative binding sites have been found at 49 locations on theRhizobium sp. NT-26 chromosome (Andres et al., 2013). Based on thephenotypic characteristics of strain GW4, we speculated that As(III)may be involved with several different metabolism pathways, andAioR may regulate other cellular functions besides As(III) oxida-tion.

Thus, in this study, we developed A. tumefaciens GW4 as a modelto understand the alteration of global metabolism pathways with

As(III) oxidation, and the regulatory roles of AioR. Using isobarictags for relative and absolute quantitation (iTRAQ) proteomics com-bined with the electrophoretic mobility shift assay (EMSA) and genetranscription analyses, we found that As(III) oxidation is related toseveral metabolic pathways, and AioR is the main regulator of As(III)oxidation; however, it is also related to several cellular processes, es-pecially with phosphate metabolism and As(III) resistance.

2. Materials and methods

2.1. Strain and culture condition

A. tumefaciens GW4 was grown at 28°C in a defined minimalmannitol medium MMNH4 containing 55mM mannitol as the primarycarbon source and modified to contain 0.1mM phosphate. As noted,1.0mM NaAsO2 [As(III)] was added to the medium. When required,50μgmL−1 of kanamycin (Kan), 50μgmL−1 chloramphenicol of (Cm)or 100μgmL−1 of ampicillin (Amp) was added (Shi et al., 2017;Somerville and Kahn, 1983). Strains GW4-ΔaioR and GW4-ΔaioR-Chave been constructed in our previous study (Shi et al., 2017).

2.2. Protein preparation

Four experimental groups GW4 (+AsIII)/GW4 (-AsIII),GW4-ΔaioR (+AsIII)/GW4-ΔaioR (-AsIII), GW4-ΔaioR (-AsIII)/GW4 (-AsIII) and GW4-ΔaioR (+AsIII)/GW4 (+AsIII) were de-signed. Total protein was extracted from the controls, the As(III)treated strain GW4 and GW4-ΔaioR cells. The cells were collectedby centrifugation and freeze-dried, and then 200μL of L3 buffer (thecomposition was undisclosed by Gene-create Company) and 800μLpre-cooling acetone (containing 10mM final concentration of dithio-threitol) were added. Then, samples were centrifuged at 13,000 rpmfor 20min to remove supernatant. The same step was performed onemore time, and the sediment was dissolved into 100μL L3 buffer.The concentration of protein was measured using the Bradford method(Martinez-Esteso et al., 2014).

2.3. iTRAQ labeling and strong cation exchange

Proteins (100 mg) from each sample were blended with 500μL50mM NH4HCO3 and 2 μg Trypsin for 8–16 h at 37 °C. Tryptic pep-tides were then added with isometric 0.1% formic acid to acidifica-tion dispose. Strata –X C18 column was activated by 1 mL methanoland balanced by 1 mL 0.1% formic acid. The acidulating Tryptic pep-tides was added into the strata –X C18 column thrice, and then 0.1%formic acid +5% acetonitrile was used to clean the column twice.At last, 1 mL 0.1% formic acid and 80% acetonitrile were addedinto the column to collect the solution, and then the sample was per-formed by vacuum drying treatment. Dry powder peptides was di-luted with 0.5 M TEAB to 20 μL, and sample was labeled with 8-plex(Martinez-Esteso et al., 2014).

2.4. Protein identification by MS/MS

The instrument was AB SCIEX nanoLC-MS/MS (Triple TOF5600 plus). The analytical column was AB SCIEX column (inside ra-dius 75μm, fill 3μm, 120Å ChromXP C18, length 10cm). The noz-zle was NEW objective (inside radius 20μm, diameter 10μm). Thecapture column was eksigent Chromxp Trap Column (3μm C18-CL,120Å, 350μm × 0.5mm) (Martinez-Esteso et al., 2014).

UNCO

RREC

TED

PROO

F

Environmental Pollution xxx (2018) xxx-xxx 3

2.5. iTRAQ proteomics identification and quantification

The MS data were processed through Proteinpilot™ V4.0.8085and searched against a Uniprot Agrobacterium tumefaciens data set.The Proteinpilots were set as the following values viz., Type of search(iTRAQ 8plex), Enzyme (Trypsin), Cys Alkylation (Iodoacetamide),Instrument (TripleTOF 5600), Bias Correction (TRUE), BackgroundCorrection (TRUE), ID focus (Biological modifications), Search Ef-fort (Thorough ID), Protein Mass (Unrestricted) and Database(Uniprot Agrobacterium tumefaciens). The processes of protein prepa-ration, iTRAQ labeling and strong cation exchange, protein identifi-cation by MS/MS and iTRAQ protein identification and quantifica-tion were all performed by Gene-create Company (Martinez-Estesoet al., 2014). The differentially expressed proteins were categorizedinto functional COG classes using three websites—WebMGA, NCBIand KEGG (Table S1). At least two peptides were required for proteinidentification, and a total of 3103 proteins were identified. The sig-nificance values were up-regulation of ≥1.5, down-regulation of ≤ -1.5and p value of ≤0.01.

To comprehensively analyze the changes in carbohydrate metab-olism in strain GW4, additional 2D-gel proteomics analysis andqRT-PCR were also measured. 2D gel experiments were performedusing the same sampling time as that of iTRAQ and the detail analyti-cal methods was described by Li et al. (2015).

2.6. Prediction of AioR putative binding sites and EMSA

The putative AioR binding motif was predicted by MEME onlinesoftware (http://meme-suite.org/tools/meme). Using the putative AioRbinding motif, the AioR putative binding sites in the genomes of A.tumefaciens GW4 (AWGV00000000) were predicted by FIMO on-line software (http://meme-suite.org/tools/fimo), with the parameter ofmatch value of 0.0001 (Li et al., 2013; Shi et al., 2017).

To identify the putative AioR binding site in gene regulatory re-gion, these fragments containing the putative AioR binding site wereamplified using the primers shown in Table S2. After the preliminaryscreening, fragments of kdgD, ubiE, proC, anmK, PIMT, phoU2 genesand the other four genes shown in Fig. S2 were fused into T-vector.The primer T7 was labeled with fluorophore 6-carboxy-fluorescein(FAM) when needed, and the above ten fragments were amplified us-ing the primers T7 and SP6. EMSA was carried out with a 0.5pmol la-beled probe and increasing concentrations of AioR (from 0 to 8pmol).All reaction mixtures were incubated at 28°C for 30min in bindingbuffer [20 mM Tris-HCl, pH 7.0; 50mM NaCl; 1mM dithiothreitol(DTT); 10mM MgCl2; 100μgmL−1 bovine serum albumin (BSA)]and then loaded onto a 6% native PAGE. After 3h of running at 80Vin 1× TGE buffer, gels were exposed to a phosphor imaging system(Fujifilm FLA-5100) (Chen et al., 2015; Shi et al., 2017).

2.7. Quantitative RT-PCR analysis

Overnight cultures (OD600 = 0.7–0.8) were inoculated into 100mLMMNH4 medium with or without the addition of 1mM As(III) andincubated at 28°C with 100 rpm shaking. Samples used for RNA iso-lation were taken after 16h cultivation (mid log phase). Total RNAwas extracted by Trizol (Invitrogen) and incubated with RNase-freeDNase I (Takara) at 37°C to remove the genomic DNA, which wasthen terminated by addition of 50mM EDTA at 65°C for 10min.After determining the concentration of RNA by spectrophotometry(NanoDrop 2000, Thermo), 300ng total RNA was reverse transcribedinto cDNA with RevertAid First Strand cDNA Synthesis

Kit (Thermo). The resulting cDNA was diluted 10-fold for real-timeRT-PCR analysis using SYBR® Green Realtime PCR Master Mix(Toyobo) with primers listed in Table S2. Quantitative RT-PCR wasperformed by ABI VIIA7 in 0.1mL Fast Optical 96-well ReactionPlate (ABI). Each reaction was replicated three times to estimate er-ror. Gene expression was normalized by the 2-ΔΔCT analysis with aniQ5 Real-Time PCR Detection System (Bio-Rad, USA). The ANOVAanalysis was performed with Excel 2013 (Wang et al., 2012; Shi et al.,2017).

3. Results and discussion

3.1. Comparative proteomics analysis

Four experimental groups, GW4 (+AsIII)/GW4 (-AsIII),GW4-ΔaioR (+AsIII)/GW4-ΔaioR (-AsIII), GW4-ΔaioR (-AsIII)/GW4 (-AsIII) and GW4-ΔaioR (+AsIII)/GW4 (+AsIII), were de-signed. Using iTRAQ analysis, 41 proteins showed differences ingroup GW4 (+AsIII)/GW4 (-AsIII) (fold change ≥1.5 or ≤-1.5, pvalue ≤0 .01) (Fig. 1A). Twelve proteins were up-regulated, and 29proteins were down-regulated under As(III) stress (Fig. 1A). Forgroup GW4-ΔaioR (+AsIII)/GW4-ΔaioR (-AsIII), 71 proteins wereidentified—57 proteins were up-regulated, and 14 proteins weredown-regulated in the presence of As(III) (Fig. 1B). For groupGW4-ΔaioR (-AsIII)/GW4 (-AsIII), 82 proteins were identified—39proteins were up-regulated, and 43 proteins were down-regulated (Fig.1C). For group GW4-ΔaioR (+AsIII)/GW4 (+AsIII), 168 proteinswere identified—113 proteins were up-regulated, and 55 proteinswere down-regulated (Fig. 1D). Proteins involved in As(III) oxidation,As(III) resistance, pst/pho system, carbohydrate metabolism, cell wall/membrane and motility/chemotaxis showed main differences amongthe four treatments (Fig. 1E–H). These iTRAQ results in combina-tion of 2D gel proteomics, Real-time RT-PCR and EMSA are ana-lyzed in detail to show the effects of As(III) and the roles of AioR (see3.4–3.8).

3.2. Identification of AioR binding sites on strain GW4

Comparative proteomics analysis showed that AioR was involvedin the regulation of As(III) oxidation, As(III) resistance/efflux, thepst1/pho1 system and the pst2/pho2 system (Fig. 1 and Table S1).AioR is a transcriptional regulator, and its consensus DNA bind-ing sequenceGT[TC][AC][CG][GCT][AG][AG]A[ACT][CGA][GCT][GTA]AAChas been documented (Shi et al., 2017). Using the FIMO website,54 AioR putative binding sites were found on the A. tumefaciensGW4 chromosome (Table S3). EMSA was then performed to examinethe interaction between AioR and the 54 AioR putative binding re-gions. As shown in Fig. 2, the following six genes were provento interact with AioR: (A) 5-dehydro-4-deoxyglucarate dehydratase(KdgD)—KdgD uses 5-dehydro-4-deoxy-D-glucarate as a substrate toproduce 2,5-dioxopentanoate, and then enters the TCA cycle (Jeffcoatet al., 1969); (B) ubiquinone biosynthesis protein (UbiE)—UbiE is in-volved in ubiquinone biosynthesis to participate in the electron trans-port chain (Lee et al., 1997); (C) pyrroline-5-carboxylate reductase(ProC)—ProC catalyzes the third step of proline synthesis (Misenerand Walker, 2001); (D) anhydro-N-acetylmuramic acid kinase(AnmK)—AnmK is involved in the recycling of peptidoglycan inbacteria (Dai et al., 2015); (E) protein-L-isoaspartate O-methyltrans-ferase (PIMT)—PIMT is involved in the repair of damaged proteins;(F) phosphate transport system regulatory protein (PhoU2)—PhoU2,belonging to the pst2/pho2 system (far from the As(III) oxidationgenes), is involved in the uptake of phosphate, when strains undergophosphate starvation (Kang et al., 2012). Addi

UNCO

RREC

TED

PROO

F

4 Environmental Pollution xxx (2018) xxx-xxx

Fig. 1. iTRAQ-based proteomics analysis of global perturbation of A. tumefaciens GW4 cellular metabolism in response to As(III) exposure. A represents the differentially expressedproteins of GW4 (+AsIII)/GW4 (-AsIII). B represents the differentially expressed proteins of GW4-ΔaioR (+AsIII)/GW4-ΔaioR (-AsIII). C represents the differentially expressedproteins of GW4-ΔaioR (-AsIII)/GW4 (-AsIII). D represents the differentially expressed proteins of GW4-ΔaioR (+AsIII)/GW4 (+AsIII). Gray vertical lines denote the 1.5 and−1.5 fold change designated cut-off. Horizontal gray lines denote a statistical significance p-value ≤0.01. (E–H) Classification of proteins in COG categories. Red bars representthe up-regulated proteins and green bars represent the down-regulated proteins. (For interpretation of the references to colour in this figure legend, the reader is referred to the Webversion of this article.)

tionally, four other functionally unknown genes were also proven tobe regulated by AioR (Fig. S2).

In our comparative proteomics data, PhoU2 (No. 315) was iden-tified to be down-regulated by 7-fold in group GW4 (+AsIII)/GW4(-AsIII) and up-regulated by 13.6-fold in group GW4-ΔaioR (+AsIII)/GW4 (+AsIII) (Table S1). Although ProC and PIMT were not iden-tified in our comparative proteomics data, glutamate 5-kinase (ProB)(No. 699), catalyzing the second step of proline synthesis, and an-other copy of PIMT (No. 801) were both down-regulated by 3.0 or2.3-fold in group GW4-ΔaioR (+AsIII)/GW4 (+AsIII), respectively(Table S1). Hence, there was a certain degree of coincidence betweenthe comparative proteomics data and EMSA results. Although the pos-sible AioR-binding sites have been predicted in the As(III)-oxidizingstrain NT-26 (Andres et al., 2013), this was the first experimental evi-dence to identify the possible AioR-binding sites in an As(III) oxidiz-ing bacterium.

3.3. Identification of the co-transcription and induction of kdgD,ubiE, proC, anmK, PIMT and phoU2 genes

To further investigate the regulation of AioR in six functionallyknown genes, RT-PCR and qRT-PCR assays were employed to testthe co-transcription and expression levels of these genes. The RT-PCRassays showed the co-transcription of fabG–kdgD (Fig. 3A), indicat-ing that the AioR binding site was within the operon. With the pres-ence of As(III), the transcription of kdgD was weakly decreased by1.4-fold in strain GW4 and strain GW4-ΔaioR-C, but the transcrip-tion levels were recovered in the mutant strain GW4-ΔaioR (Fig.3A). Thus, AioR repressed the expression of the kdgD gene. RT-PCRshowed that the ubiE, proC, anmK and PIMT genes all belonged toindependent operons (Fig. 3B–E). These genes were all induced byAs(III) by 2.2, 2.5, 4.1 and 2.5-fold, respectively, but the transcrip

tion levels were significantly decreased in the mutant strainGW4-ΔaioR and were recovered in the complementary strainGW4-ΔaioR-C (Fig. 3B–D). The above results indicated that AioRregulated the expression of the ubiE, proC, anmK and PIMT genes.RT-PCR showed the co-transcription ofpstC2-pstA2-pstB2-phoU2-phoB2 (Fig. 3F), indicating that the AioRbinding site was within the operon. The transcription of the phoU2gene was decreased by 2.5-fold with the presence of As(III), a find-ing that was consistent with the proteomics data (Table S1) and wasrecovered in the strain GW4-ΔaioR (Fig. 3F). Additionally, AioRrepressed the expression of the pstC2-pstA2-pstB2-phoU2-phoB2operon. Thus, the data provide the first evidence that AioR can func-tion as both activator and repressor depending on its binding location.

3.4. AioR regulates As(III) oxidation and As(III) chemotaxis

Proteomics experiments showed that AioR, AioA and molyb-dopterin biosynthesis protein A (MoeA) involved in As(III) oxidationwere all up-regulated with a fold change ranging from 13.7 to 16.6 inthe presence of As(III) (Table S1). This was in agreement with the re-sults obtained in A. tumefaciens 5A and Rhizobium sp. NT-26 (Andreset al., 2013; Kashyap et al., 2006). In group GW4-ΔaioR (+AsIII)/GW4 (+AsIII), AioR, AioA, MoeA and cytochrome c-550 (CytC550)were down-regulated by 8.9, 10.4, 7.0 and 6.6, respectively (TableS1). The literature (vanden Hoven and Santini, 2004) and our previ-ous work indicated that bacterial As(III) oxidation is not only a detox-ification mechanism but also an energy resource (Wang et al., 2015),and we have proven that AioR could regulate the expression of theaioBA gene (Shi et al., 2017). Thus, AioR could regulate strain GW4As(III) oxidation to detoxify As(III) and generate energy (Shi et al.,2017; Wang et al., 2015).

UNCO

RREC

TED

PROO

F

Environmental Pollution xxx (2018) xxx-xxx 5

Fig. 2. EMSA assays. The FAM-labeled kdgD probe (A), ubiE probe (B), proC probe (C), anmK probe (D), PIMT probe (E) and phoU2 probe (F) each interacted with AioR protein.The amounts of the DNA probes and AioR are shown in the above tables of each panel.

Fig. 3. Influence of As(III) and AioR on the expression of the kdgD (A), ubiE (B), proC (C), anmK (D), PIMT (E) and phoU2 (F) genes. The locations of the arrows are the AioRputative binding sites. The gene clusters in the dashed boxes are co-transcribed (data not shown). The data are shown as the means of three replicates, with the error bars illustratingone standard deviation.

UNCO

RREC

TED

PROO

F

6 Environmental Pollution xxx (2018) xxx-xxx

In our previous study, we demonstrated that AioR could bind thepredicted sequence in the regulatory region of the mcp gene, regu-lating the expression of mcp, and was involved in the regulation ofchemotaxis toward As(III) in A. tumefaciens GW4 (Fig. 4) (Shi et al.,2017). Our proteomics data showed that the Mcp (No. 1783) involvedin As(III) chemotaxis was down-regulated, and another five unknownMcp proteins were up-regulated with the deletion of the aioR gene(Table S1). Microorganisms have evolved numerous abilities to re-spond and adapt to nutrient scarcity and toxic materials. Chemotaxisenables microorganisms to migrate toward attractants or away fromrepellents. The different methyl-accepting chemotaxis proteins (Mcp)have been characterized to sense different ligands and, as such, arecritical elements of the different chemotaxis responses (Porter et al.,2011). We inferred that, due to the loss of As(III) chemotaxis in strainGW4-ΔaioR, it began to activate other Mcp proteins to swim towardother substances.

Interestingly, with the deletion of the aioR gene, oxidoreductase(AioE) within the ars operon (arsR1-arsC1-arsC2-acr3-aioE) wasdown-regulated by 27.3-fold in the absence of As(III) and 6.9 in thepresence of As(III) (Table S1). AioE has been proven to act as a novelAs(III) oxidation electron transporter associated with NADH gener-ation, inferring that electrons may transfer from AioBA to CytC550via AioE (Wang et al., 2017). As mentioned above, we have proventhat AioR could regulate the expression of the mcp gene by bindingwith its promoter (Fig. 4A and B) (Shi et al., 2017). Further genomeanalysis showed that this AioR binding site was also in the ars operonregulatory region (Fig. 4A and B). According to the proteomics andqRT-PCR results (Table S1 and Fig. 4C), we inferred that AioR alsoregulated the expression of ars operon. To our knowledge, this wasthe first experimental evidence to prove that the ars operon was reg-ulated by AioR in addition to its autologous transcriptional regulatorArsR (Kang et al., 2016).

3.5. AioR activates the pst1/pho1 system and represses the pst2/pho2system

There are two copies of the pst/pho system in strain GW4. Thepst1/pho1 system is localized within As islands, while the pst2/pho2

system was localized distantly on the respective chromosomes (Wanget al., 2015). Remarkably, the most significantly disturbed proteinswere involved in the pst1/pho1 system and pst2/pho2 system (TableS1). Periplasmic phosphate-binding protein (PstS1), phosphate trans-port ATP-binding protein (PstB1) and phosphate transport system reg-ulatory protein (PhoU1) were all up-regulated, while proteins involvedin the pst2/pho2 system were all down-regulated by 2.6–40.6-foldby As(III) in the wild type (Table S1). Interestingly, with the dele-tion of AioR, the pst1/pho1 system was down-regulated, and the pst2/pho2 system was up-regulated in the absence or presence of As(III)(Table S1). EMSA and qRT-PCR assays showed that AioR couldbind to the front of the phoU2 gene to suppress the expression ofthe pstC2-pstA2-pstB2-phoU2-phoB2 operon (Figs. 2F and 3F). PhoUand PhoB were both involved in the regulation of the pst/pho sys-tem (Lubin et al., 2015). We inferred that AioR bound to the front ofphoU2 and phoB2 genes to repress the pst2/pho2 system. AioR alsoparticipated in the up-regulation of the pst1/pho1 system with or with-out As(III).

Proteins involved in these two pst/pho systems showed many dif-ferent characteristics—e.g., PstS1 could transfer phosphate and As(V)into cells, but PstS2 may only transfer phosphate (Wang et al., 2015).In H. arsenicoxydans ULPAs1, PstB1 and PhoU1 encoded by thegenes localized in the As island was induced by 3- or 4-fold, whereasPhoU2 was repressed by As(III) (Weiss et al., 2009). Hence, the dif-ferent characteristics between the pst1/pho1 system and pst2/pho2 sys-tem may widely exist in As(III)-oxidizing bacteria. We speculated thatAioR suppressed the expression of the pst2/pho2 system and indirectlyup-regulated the pst1/pho1 system to transfer As(V) and phosphateeconomically.

3.6. AioR regulates multiple As(III) resistance metabolism

3.6.1. AioR regulates the As(III) resistance ars operon to effluxAs(III)

Strain GW4 showed high As(III) resistance (MIC = 25mM). Asmentioned above, the ars operon was induced by As(III), and AioRcould regulate the ars operon to participate in As resistance metabo-lism to avoid excessive As(III) in the cytoplasm (Fig. 4). Proteomicsexperiments also showed that As(III) reductase (ArsC1), ArsC2, Ar

Fig. 4. AioR could regulate the ars operon. (A) The gene cluster of ars operon and mcp gene. The dashed boxes means these genes are co-transcription. (B) FAM-labeled ars operonprobe interacted with AioR protein. The amounts of DNA probes and AioR were shown in the above table of each panel. (C) Influence of As(III) and AioR on the expression ofarsR1.

UNCO

RREC

TED

PROO

F

Environmental Pollution xxx (2018) xxx-xxx 7

sC3, ArsR1, As resistance protein (ArsH) and arsenical-resistance pro-tein (Acr3) were all down-regulated with a fold change ranging from12.1 to 48.3-fold in group GW4-ΔaioR (-AsIII)/GW4 (-AsIII) and4.0–16.4-fold in group GW4-ΔaioR (+AsIII)/GW4 (+AsIII) (TableS1). The induction of the ars operon was the common defense mecha-nism against As(III) exposure (Weiss et al., 2009).

3.6.2. As(III) induces the cell wall/membrane metabolismEMSA results showed that AnmK was up-regulated with the ad-

dition of As(III) (Fig. 3D), and it was regulated by AioR (Fig. 2D),indicating that AioR was involved in the recycling of peptidoglycan.In our proteomics results, two proteins, glycerophosphodiester phos-phodiesterase (GDE1) and dolichol monophosphate mannose synthase(YcjR), were identified to be down-regulated (Table S1). GDE1 be-longed to the family of hydrolases and could hydrolyze glycerophos-phodiester to alcohol and sn-glycerol 3-phosphate. However, whenAioR was deleted, nine proteins in group GW4-ΔaioR (+AsIII)/GW4-ΔaioR (-AsIII) were up-regulated by 1.8–3.6-fold, and 16 pro-teins in group GW4-ΔaioR (+AsIII)/GW4 (+AsIII) were up-regulatedby 1.6–21.5-fold (Table S1).

In H. arsenicoxydans ULPAs1, two proteins probably involved inexopolysaccharide (EPS) production were accumulated, in agreementwith previous data demonstrating that EPS synthesis was induced inresponse to arsenic, and probably played a major role in resistance tothis toxic element by scavenging nanoparticle inside the EPS matrix(Weiss et al., 2009). However, proteins involved in the cell wall/mem-brane were down-regulated by As(III) in A. tumefaciens GW4, andproteins were up-regulated when AioR was deleted in the presence ofAs(III) (Table S1). These observations were consistent with the resultsthat As(III) could be considered an attractant for strain GW4 and a re-pellent for mutant GW4-ΔaioR. When GW4-ΔaioR lost the ability ofAs(III) oxidation, proteins involved in the cell wall/membrane metab-olism were induced to prevent the entrance of As(III), indicating thatAs(III) oxidation was related to cell wall/membrane metabolism, andAioR was also involved in the detoxification mechanism of the cellwall/membrane.

3.6.3. AioR regulates proline protection and PIMT repairThe results of qRT-PCR and EMSA assays showed that proC and

PIMT were also regulated by AioR (Figs. 2 and 3). These observationsrevealed that AioR may be involved in the accumulation of prolineand PIMT in the presence of As(III). In Leptospirillum ferriphilumML-40, glutamate-1-semialdehyde amino transferase involved in thesynthesis of proline was also up-regulated by 3-fold (Li et al., 2010).In addition, in C. arsenoxydans ULPAs1 and H. arsenicoxydans UL-PAs1, PIMT was shown to be up-regulated (Carapito et al., 2006;Weiss et al., 2009). Proline is important in all organisms, not onlyfor its role in protein biosynthesis but also due to its implication ina range of stress responses, including the resistance to osmotic, cold,and desiccation stress in both insects and plants (Misener and Walker,2001). PIMT is involved in the repair of damaged proteins (Reissnerand Aswad, 2003). We speculated that AioR could regulate As(III) re-sistance mediated by proline and PIMT (Fig. 6). These complicatedresistance mechanisms may endow strain GW4 with strong As(III) re-sistance.

As(III) toxicity was thought to be due predominantly to its abilityto covalently bind protein sulfhydryl groups, causing oxidative stress(Parvatiyar et al., 2005). It is interesting that no stress response pro-teins were detected in our proteomics data (Table S1). In our previous2D gel results, one stress response protein, catalase KatA, was identi-fied to be down-regulated by 3.6-fold in the presence of 1mM As(III)(Table S1). To date, when the strains were grown with As(III), al

most all the proteomics results showed an induction of stress re-sponse proteins, such as peroxiredoxin–alkyl hydroperoxide reductasesub-unit F in F. acidarmanus Fer1 (Baker-Austin et al., 2007), super-oxide dismutase [Mn] in Klebsiella pneumoniae MR4 (Daware et al.,2012), superoxide dismutase in Exiguobacterium sp. S17 (Belfiore etal., 2013) and superoxide dismutase in Rhizobium sp. NT-26 (Andreset al., 2013). Hence, we speculated that AioR was the key regulator toregulate As resistance and efflux, cell wall/membrane metabolism andthe synthesis of proline and PIMT to endow strain GW4 with strongAs(III) resistance.

3.7. Energy acquisition from As(III) oxidation and carbonmetabolism

3.7.1. Alteration of carbon metabolism pathways by As(III) oxidationAmong the proteins involved in glycolysis and fermentation me-

tabolism, with the addition of As(III), the amounts of citrate lyase(CitE), acetoacetyl-CoA synthetase (AcsA), aldehyde dehydrogenase(ALDH), phosphoenolpyruvate carboxykinase (PckA), pyruvate ki-nase (Pyk) and alcohol dehydrogenase (ADH) were increased by 7.8,4.5, 4.4, 3.1, 2.5 and 2.1-fold, respectively (Fig. 5A and Table S1).The proteins involved in the TCA cycle, aconitate hydratase (AcnA),succinate dehydrogenase (SdhA) and fumarate hydratase (FumC),were decreased by 3.3, 2.4 and 1.5-fold, respectively (Fig. 5A andTable S1). EMSA and qRT-PCR results showed that AioR couldrepress the expression of kdgD (Figs. 2A and 3A), consistent withthe TCA cycle being decreased in the presence of As(III). However,when AioR was deleted, proteins involved in the TCA cycle, aconitatehydratase (AcnA), dihydrolipoamide succinyltransferase (SucB) andsuccinyl-CoA synthetase subunit beta (SucD) were up-regulated in thepresence of As(III) (Fig. 5B and Table S1).

Based on the differentially expressed proteins, the metabolic path-ways in strain GW4 were analyzed. In this study, we used D-manni-tol as the carbon source, which may be used in the glycolytic pathwayand catabolized to acetyl-CoA via phosphoenolpyruvate and pyruvate,based on genome analysis (Fig. 5). In the presence of As(III), the de-creased AcnA, FumC and SdhA suggested that As(III) exposure per-turbed carbon flow into and within the TCA cycle (Fig. 5A). Mean-while, the up-regulation of CitE implied that the citrate generated fromacetyl-CoA may be catabolized to oxaloacetate and acetate (Fig. 5A).Catalyzed by the increased PckA, oxaloacetate may be then trans-formed to phosphoenolpyruvate (Fig. 5A). However, the up-regula-tion of Pyk suggested that phosphoenolpyruvate can produce pyruvateand acetyl-CoA via the glycolytic pathway (Fig. 5). This pathway, cat-alyzed by the three significant up-regulated proteins, Pyk, CitE andPckA, may accumulate acetate and generate NADH (Fig. 5). In ad-dition, acetate accumulated from citrate via the induced CitE may besubsequently transformed to acetaldehyde and ethanol catalyzed bythe up-regulated ALDH and ADH (Fig. 5A). The up-regulated ADHand ALDH may be used to reduce the toxicity of acetate to bacterialcells. However, when strain GW4-aioR lost the ability of As(III) ox-idation, the TCA cycle was recovered to generate energy to againstAs(III) stress.

3.7.2. As(III) oxidation contributes to energy generationIn EMSA results, we have proven that AioR could regulate the ex-

pression of the ubiE gene, which was involved in ubiquinone biosyn-thesis (Figs. 2B and 3B). Ubiquinone is a component of the electrontransport chain and participates in aerobic cellular respiration, whichgenerates energy in the form of ATP (Lee et al., 1997). In groupGW4-ΔaioR (+AsIII)/GW4 (+AsIII), cytochrome c oxidase cbb3-typesubunit III (CoxC), cytochrome oxidase Cu insertion factor (Sco),cbb3-type cytochrome oxidase (CcoO), membrane-bound pro

UNCO

RREC

TED

PROO

F

8 Environmental Pollution xxx (2018) xxx-xxx

Fig. 5. The metabolic pathways for glycolysis and TCA cycle in strains GW4 and GW4-ΔaioR. Green arrows represent the down-regulated proteins; red arrows represents the up-reg-ulated proteins and red R means the regulation of AioR. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

ton-translocating pyrophosphatase (HppA) and 2-polyprenylphenol6-hydroxylase (UbiB) were up-regulated by 3.5, 3.1, 2.6, 2.5 and2.2-fold, respectively (Table S1). CytC550 was reported to be respon-sible for transferring electrons from As(III) oxidation (Wang et al.,2017), and it was down-regulated when AioR was deleted (TableS1). We speculated that, due to strain GW4-ΔaioR losing the abil-ity of As(III) oxidation, the TCA cycle was recovered, and then othercytochromes such as CoxC, Sco and CcoO were used to transformNADH to ATP instead of CytC550 (Table S1).

In other heterotrophic As(III)-oxidizing strains, the expression lev-els of FumC and succinyl-CoA synthetase SucC, which were involvedin the TCA cycle, were both up-regulated (Carapito et al., 2006;Weiss et al., 2009), indicating that these heterotrophic As(III)-oxi-dizing strains could use As(III)-oxidation to counter the toxicity ofAs(III). However, in our case of strain GW4, with the addition ofAs(III), the TCA cycle was perturbed, and the strain grew better andproduced more NADH and ATP (Wang et al., 2015). These can beexplained as follows: a) NADH could be produced through carbonmetabolism via the pathway containing Pyk, CitE and PckA becausethese three genes/or proteins were up-regulated (Fig. 5A); b) Be-cause AioE was related to As(III) oxidation in strain GW4, it mayalso participate in the production of NADH based on the literature,and CytC550 could contribute to the generation of ATP (Wang etal., 2017). Consistent with this hypothesis, we found the decrease ofNADH in the aioE-disrupted strain and the decrease of ATP in the

cytC550-disrupted strain in our recent study (Wang et al., 2017); and c)As(III) oxidation is an exergonic reaction; generally, bacterial As(III)oxidation could produce 256 KJ/Rx under aerobic conditions (Santiniet al., 2000), approximately equal to 8.5 ATP. In another heterotrophicAs(III)-oxidizing strain, Hydrogenophaga sp. NT-14, As(III) oxida-tion could yield energy for bacterial growth (vanden Hoven andSantini, 2004). In our case, although strain GW4 was also a het-erotrophic As(III)-oxidizer, the As(III) oxidation may also directlycontribute to the energy generation. All these data indicate that As(III)oxidation was the main source to improve the production of NADHand ATP or contribute to the energy generation directly in some het-erotrophic As(III) oxidizers.

3.8. Other metabolisms related to AioR

EMSA results also showed that AioR could regulate four otherfunctionally unknown proteins (Fig. S2): (A) protein of unknownfunction; (B) 4-oxalocrotonate tautomerase, involved in the degrada-tion of xylene, but strain GW4 has lost the ability of xylene degrada-tion; (C) lipoprotein, which may also participate in bacterial As(III)resistance; (D) NAD(P)H nitroreductase, which may participate inthe redox reaction with the presence of As(III). Seven proteins in-volved in transcription and translation were all down-regulated by1.8–12.6-fold in strain GW4-ΔaioR (Table S1). AioR was proven tobe a transcription regulator, and the decreased expression of these

UNCO

RREC

TED

PROO

F

Environmental Pollution xxx (2018) xxx-xxx 9

Fig. 6. The regulatory cellular networks controlled by AioR. 1. Energy generation. AioR regulates the expression of AioBA to oxidize As(III) (1.1), and regulates the expressionof AioE to transfer the electron from AioBA to AioE with the generation of NADH (1.2). AioR could also regulate the expression of UbiE to maintain the electron transport withinoxidative phosphorylation complex (1.2). AioR suppress the processes of glycolysis and TCA cycle by regulating KdgD (1.3). 2. As(III) resistance. AioR also regulates the expres-sion of Ars, AnmK, ProC and PIMT to participate in As efflux (2.1), recycling of peptidoglycan (2.2), synthesis of proline (2.3) and PIMT repair system (2.4) to enhance As(III)resistance. Besides, AioR regulates PhoU2-PhoB2 in phosphate metabolism and regulates Mcp in As(III) chemotaxis.

proteins may be due to the reduction of protein synthesis in strainGW4-ΔaioR. A porin was remarkably up-regulated by 19.4-fold inthe presence of As(III) (Table S1) and may be involved in the up-take or excretion of a certain substance. Three proteins noted as glyc-erol 3-phosphate ABC transporter (UgpB) were all down-regulatedwith As(III) and were up-regulated when AioR was deleted (TableS1). Additionally, an iron ABC transporter substrate-binding proteinwas up-regulated by 13.2-fold in the strain GW4-ΔaioR (Table S1). Invitro, natural remediation can occur as a result of the concurrent ox-idation of Fe(II) and As(III), leading to the co-precipitation of both(Karimian et al., 2017). In vivo, the addition of Fe(II) significantly de-creased H2O2 and the malondialdehyde (MDA) content in cells (Nathet al., 2014), indicating that the metabolism of As and Fe were associ-ated. In addition, the regulator of the iron-sulfur cluster, IscR, has beenproven to positively contribute to antimonite resistance and oxidationin Comamonas testosteroni S44 (Liu et al., 2015). We speculated thatFe(II) is highly uptaken and utilized to synthesize the iron-sulfur clus-ter for dependence, when strain GW4-ΔaioR has lost the As(III) oxi-dation ability.

3.9. AioR regulatory networks

Based on the above results, we show a graph of the AioR regula-tory networks to provide a better understanding of how strain GW4could survive in an As(III)-enriched environment (Fig. 6). The growthof A. tumefaciens GW4 was enhanced with As(III), and the concen-trations of ATP and NADH were significantly increased (Wang et al.,2015). AioR is one of the key proteins to influence the energy re-sources from As(III) oxidation, glycolysis, fermentation and the TCAcycle. With As(III), AioR regulates the expression of AioBA to oxi-dize As(III) and regulates the expression of AioE to transfer the elec-tron from AioBA to AioE with the generation of NADH, and then toCytC550 with the generation of ATP (Wang et al., 2017). AioR mayalso regulate the expression of UbiE to maintain the electron transport

within the oxidative phosphorylation complex. Additionally, strainGW4 suppresses the processes of the TCA cycle by regulating KdgDto better utilize As(III) oxidation and fermentation mechanisms as itsmain energy resource.

In addition, we speculate that AioR could regulate the Ars to effluxthe cytoplasmic As(III) to avoid its toxicity and/or regulate anmK toparticipate in cell wall/membrane metabolism to prevent the entranceof As(III). Additionally, proline is important in a range of stress re-sponses, including resistance to osmotic, cold, and desiccation stress,in both insects and plants (Misener and Walker, 2001). PIMT is alsoinvolved in the repair of damaged proteins (Reissner and Aswad,2003). Hence, AioR could also regulate the synthesis of proline andPIMT to avoid cell damage from As(III) (Fig. 6).

In A. tumefaciens GW4, the MMNH4 culture medium containeda low phosphate concentration (100 μM), and PstS1 could bind phos-phate or As(V) (Wang et al., 2015). AioR could suppress the expres-sion of the pst2/pho2 system and indirectly regulate the pst1/pho1 sys-tem to transfer phosphate and As(V) economically. In addition, AioRcould regulate the expression of mcp and was involved in the regula-tion of chemotaxis towards As(III) (Shi et al., 2017) (Fig. 6).

4. Conclusion

A. tumefaciens GW4 represents a new type of heterotrophicAs(III)-oxidizing bacteria that can use As(III) to enhance growth,which is different from most of the known heterotrophic As(III)-oxi-dizing bacteria using As(III) oxidation as a detoxification process. Ourresults show that cellular responses to As(III) are delicate, and AioRmainly drives As(III) oxidation. Additionally, AioR could regulatemultiple genes involved in As(III) resistance and energy acquisition.AioR suppresses the expression of the pst2/pho2 system and up-reg-ulates the pst1/pho1 system to transport As and phosphate econom-ically. The regulatory cellular networks controlled by AioR provide

UNCO

RREC

TED

PROO

F

10 Environmental Pollution xxx (2018) xxx-xxx

the high As(III) resistance and better adaptation for strain GW4 to livein the As-rich sediment environment.

Acknowledgements

The present study was supported by the National Natural ScienceFoundation of China (31670108 and 31500088).

Appendix A. Supplementary data

Supplementary data related to this article can be found at https://doi.org/10.1016/j.envpol.2018.01.006.

References

Andres, J., Arsene-Ploetze, F., Barbe, V., Brochier-Armanet, C., Cleiss-Arnold, J.,Coppee, J.Y., Dillies, M.A., Geist, L., Joublin, A., Koechler, S., Lassalle, F., Mar-chal, M., Medigue, C., Muller, D., Nesme, X., Plewniak, F., Proux, C.,Ramirez-Bahena, M.H., Schenowitz, C., Sismeiro, O., Vallenet, D., Santini, J.M.,Bertin, P.N., 2013. Life in an arsenic-containing gold mine: genome and physiol-ogy of the autotrophic arsenite-oxidizing bacterium rhizobium sp. NT-26. GenomeBiol. Evol. 5, 934–953.

Baker-Austin, C., Dopson, M., Wexler, M., Sawers, R.G., Stemmler, A., Rosen, B.P.,Bond, P.L., 2007. Extreme arsenic resistance by the acidophilic archaeon 'Ferro-plasma acidarmanus' Fer1. Extremophiles 11, 425–434.

Belfiore, C., Ordonez, O.F., Farias, M.E., 2013. Proteomic approach of adaptive re-sponse to arsenic stress in Exiguobacterium sp. S17, an extremophile strain iso-lated from a high-altitude Andean Lake stromatolite. Extremophiles 17, 421–431.

Bryan, C.G., Marchal, M., Battaglia-Brunet, F., Kugler, V., Lemaitre-Guillier, C.,Lievremont, D., Bertin, P.N., Arsene-Ploetze, F., 2009. Carbon and arsenic metab-olism in Thiomonas strains: differences revealed diverse adaptation processes.BMC Microbiol. 9, 127.

Cai, L., Liu, G., Rensing, C., Wang, G., 2009. Genes involved in arsenic transforma-tion and resistance associated with different levels of arsenic-contaminated soils.BMC Microbiol. 9, 4.

Carapito, C., Muller, D., Turlin, E., Koechler, S., Danchin, A., Van Dorsselaer, A.,Leize-Wagner, E., Bertin, P.N., Lett, M.C., 2006. Identification of genes and pro-teins involved in the pleiotropic response to arsenic stress in Caenibacter arsenoxy-dans, a metalloresistant beta-proteobacterium with an unsequenced genome.Biochimie 88, 595–606.

Chen, F., Cao, Y., Wei, S., Li, Y., Li, X., Wang, Q., Wang, G., 2015. Regulation of ar-senite oxidation by the phosphate two-component system PhoBR in Halomonas sp.HAL1. Front. Microbiol. 6, 923.

Dai, J., Qu, H., Yu, Z., Yang, J., Zhang, H., 2015. Computational analysis ofAnmK-like kinase: new insights into the cell wall metabolism of fungi. J. Theor.Biol. 379, 59–65.

Daware, V., Kesavan, S., Patil, R., Natu, A., Kumar, A., Kulkarni, M., Gade, W., 2012.Effects of arsenite stress on growth and proteome of Klebsiella pneumoniae. J.Biotechnol. 158, 8–16.

Fan, H., Su, C., Wang, Y., Yao, J., Zhao, K., Wang, Y., Wang, G., 2008. Sedimentaryarsenite-oxidizing and arsenate-reducing bacteria associated with high arsenicgroundwater from Shanyin, Northwestern China. J. Appl. Microbiol. 105,529–539.

Jeffcoat, R., Hassall, H., Dagley, S., 1969. Purification and properties of D-4-de-oxy-5-oxoglucarate hydro-lyase (decarboxylating). Biochem. J. 115, 977–983.

Kang, Y.S., Brame, K., Jetter, J., Bothner, B.B., Wang, G., Thiyagarajan, S., McDer-mott, T.R., 2016. Regulatory activities of four ArsR proteins in Agrobacteriumtumefaciens 5A. Appl. Environ. Microbiol. 82, 3471–3480.

Kang, Y.S., Heinemann, J., Bothner, B., Rensing, C., McDermott, T.R., 2012. Inte-grated co-regulation of bacterial arsenic and phosphorus metabolisms. Environ.Microbiol. 14, 3097–3109.

Karimian, N., Johnston, S.G., Burton, E.D., 2017. Antimony and arsenic behavior dur-ing Fe(II)-Induced transformation of jarosite. Environ. Sci. Technol. 51,4259–4268.

Kashyap, D.R., Botero, L.M., Franck, W.L., Hassett, D.J., McDermott, T.R., 2006.Complex regulation of arsenite oxidation in Agrobacterium tumefaciens. J. Bacte-riol. 188, 1081–1088.

Lee, P.T., Hsu, A.Y., Ha, H.T., Clarke, C.F., 1997. A C-methyltransferase involved inboth ubiquinone and menaquinone biosynthesis: isolation and identification of theEscherichia coli ubiE gene. J. Bacteriol. 179, 1748–1754.

Li, B., Lin, J., Mi, S., Lin, J., 2010. Arsenic resistance operon structure in Leptospiril-lum ferriphilum and proteomic response to arsenic stress. Bioresour. Technol. 101,9811–9814.

Li, H., Li, M., Huang, Y., Rensing, C., Wang, G., 2013. In silico analysis of bacterialarsenic islands reveals remarkable synteny and functional relatedness between ar-senate and phosphate. Front. Microbiol. 4, 347.

Li, J., Wang, Q., Li, M., Yang, B., Shi, M., Guo, W., McDermott, T.R., Rensing, C.,Wang, G., 2015. Proteomics and genetics for identification of a bacterial anti-monite oxidase in Agrobacterium tumefaciens. Environ. Sci. Technol. 49,5980–5989.

Liu, G., Liu, M., Kim, E.H., Maaty, W.S., Bothner, B., Lei, B., Rensing, C., Wang, G.,McDermott, T.R., 2012. A periplasmic arsenite-binding protein involved in regu-lating arsenite oxidation. Environ. Microbiol. 14, 1624–1634.

Liu, H., Zhuang, W., Zhang, S., Rensing, C., Huang, J., Li, J., Wang, G., 2015. Globalregulator IscR positively contributes to antimonite resistance and oxidation in Co-mamonas testosteroni S44. Front. Mol. Biosci. 2, 70.

Lubin, E.A., Henry, J.T., Fiebig, A., Crosson, S., Laub, M.T., 2015. Identification ofthe PhoB regulon and role of PhoU in the phosphate starvation response ofCaulobacter crescentus. J. Bacteriol. 198, 187–200.

Mallik, S., Virdi, J.S., Johri, A.K., 2012. Proteomic analysis of arsenite - mediatedmultiple antibiotic resistance in Yersinia enterocolitica biovar 1A. J. Basic Micro-biol. 52, 306–313.

Martinez-Esteso, M.J., Casado-Vela, J., Selles-Marchart, S., Pedreno, M.A., Bru-Mar-tinez, R., 2014. Differential plant proteome analysis by isobaric tags for relativeand absolute quantitation (iTRAQ). Meth. Mol. Biol. 1072, 155–169.

Misener, S., Walker, V., 2001. Complementation cloning and characterization of thepyrroline 5-carboxylate reductase gene from Drosophila melanogaster. Biochem.Genet. 39, 15–31.

Nath, S., Panda, P., Mishra, S., Dey, M., Choudhury, S., Sahoo, L., Panda, S.K., 2014.Arsenic stress in rice: redox consequences and regulation by iron. Plant Physiol.Biochem. 80, 203–210.

Parvatiyar, K., Alsabbagh, E.M., Ochsner, U.A., Stegemeyer, M.A., Smulian, A.G.,Hwang, S.H., Jackson, C.R., McDermott, T.R., Hassett, D.J., 2005. Global analy-sis of cellular factors and responses involved in Pseudomonas aeruginosa resis-tance to arsenite. J. Bacteriol. 187, 4853–4864.

Porter, S.L., Wadhams, G.H., Armitage, J.P., 2011. Signal processing in complexchemotaxis pathways. Nat. Rev. Microbiol. 9, 153–165.

Reissner, K.J., Aswad, D.W., 2003. Deamidation and isoaspartate formation in pro-teins: unwanted alterations or surreptitious signals?. Cell. Mol. Life Sci. 60,1281–1295.

Rhine, E.D., Ni Chadhain, S.M., Zylstra, G.J., Young, L.Y., 2007. The arsenite oxidasegenes (aroAB) in novel chemoautotrophic arsenite oxidizers. Biochem. Biophys.Res. Commun. 354, 662–667.

Santini, J.M., Sly, L.I., Schnagl, R.D., Macy, J.M., 2000. A new chemolithoautotrophicarsenite-oxidizing bacterium isolated from a gold mine: phylogenetic, physiologi-cal, and preliminary biochemical studies. Appl. Environ. Microbiol. 66, 92–97.

Shi, K., Fan, X., Qiao, Z., Han, Y., McDermott, T.R., Wang, Q., Wang, G., 2017. Ar-senite oxidation regulator AioR regulates bacterial chemotaxis towards arsenite inAgrobacterium tumefaciens GW4. Sci. Rep. 7, 43252.

Somerville, J.E., Kahn, M.L., 1983. Cloning of the glutamine synthetase I gene fromRhizobium meliloti. J. Bacteriol. 156, 168–176.

Stolz, J.F., Basu, P., Santini, J.M., Oremland, R.S., 2006. Arsenic and selenium in mi-crobial metabolism. Annu. Rev. Microbiol. 60, 107–130.

vanden Hoven, R.N., Santini, J.M., 2004. Arsenite oxidation by the heterotroph Hy-drogenophaga sp. str. NT-14: the arsenite oxidase and its physiological electronacceptor. Biochim. Biophys. Acta 1656, 148–155.

Wang, Q., Han, Y., Shi, K., Fan, X., Wang, L., Li, M., Wang, G., 2017. An oxidore-ductase AioE is responsible for bacterial arsenite oxidation and resistance. Sci.Rep. 7, 41536.

Wang, Q., Lei, Y., Xu, X., Wang, G., Chen, L.L., 2012. Theoretical prediction and ex-perimental verification of protein-coding genes in plant pathogen genomeAgrobacterium tumefaciens strain C58. PLoS One 7, e43176.

UNCO

RREC

TED

PROO

F

Environmental Pollution xxx (2018) xxx-xxx 11

Wang, Q., Qin, D., Zhang, S., Wang, L., Li, J., Rensing, C., McDermott, T.R., Wang,G., 2015. Fate of arsenate following arsenite oxidation in Agrobacterium tumefa-ciens GW4. Environ. Microbiol. 17, 1926–1940.

Weiss, S., Carapito, C., Cleiss, J., Koechler, S., Turlin, E., Coppee, J.Y., Heymann, M.,Kugler, V., Stauffert, M., Cruveiller, S., Medigue, C., Van Dorsselaer, A., Bertin,P.N., Arsene-Ploetze, F., 2009. Enhanced structural and functional genome eluci-dation of the arsenite-oxidizing strain Herminiimonas arsenicoxydans by pro-teomics data. Biochimie 91, 192–203.

Zhu, Y.G., Yoshinaga, M., Zhao, F.J., Rosen, B.P., 2014. Earth abides arsenic bio-transformations. Annu. Rev. Earth Planet Sci. 42, 443–467.

E T O C B L U R B

This is the first report showing the effects of As(III) oxidation and AioR onmetabolic pathways in heterotrophic As(III)-oxidizing bacteria using As(III)to generate energy.