Involvement of RpoN in Regulating Bacterial Arsenite Oxidation

Abstract

In this study with the model organism Agrobacterium tumefaciens, we used a combination of lacZ gene fusions, reverse transcriptase PCR (RT-PCR), and deletion and insertional inactivation mutations to show unambiguously that the alternative sigma factor RpoN participates in the regulation of As^III oxidation. A deletion mutation that removed the RpoN binding site from the aioBA promoter and an aacC3 (gentamicin resistance) cassette insertional inactivation of the rpoN coding region eliminated aioBA expression and As^III oxidation, although rpoN expression was not related to cell exposure to As^III. Putative RpoN binding sites were identified throughout the genome and, as examples, included promoters for aioB, phoB1, pstS1, dctA, glnA, glnB, and flgB that were examined by using qualitative RT-PCR and lacZ reporter fusions to assess the relative contribution of RpoN to their transcription. The expressions of aioB and dctA in the wild-type strain were considerably enhanced in cells exposed to As^III, and both genes were silent in the rpoN::aacC3 mutant regardless of As^III. The expression level of glnA was not influenced by As^III but was reduced (but not silent) in the rpoN::aacC3 mutant and further reduced in the mutant under N starvation conditions. The rpoN::aacC3 mutation had no obvious effect on the expression of glnB, pstS1, phoB1, or flgB. These experiments provide definitive evidence to document the requirement of RpoN for As^III oxidation but also illustrate that the presence of a consensus RpoN binding site does not necessarily link the associated gene with regulation by As^III or by this sigma factor.

INTRODUCTION

Evidence of microbial arsenite (As^III) oxidation was first reported nearly a century ago (18). Subsequent progress has been sporadic, with work that identified some organisms capable of As^III oxidation (46, 48, 60) and then a study of a Pseudomonas arsenitoxidans strain reported to grow chemolithoautotrophically with As^III as a sole electron donor (23). Subsequent follow-up characterizations of this organism and this process failed to materialize; however, approximately 2 decades later, Santini et al. (52) described the isolation and initial characterization of a Rhizobium-like bacterium (strain NT-26) that could grow chemolithoautotrophically with As^III as a sole electron donor for energy generation and with CO₂ as a sole carbon source. Soon thereafter, and in part stimulated by the massive arsenic poisoning disaster in Bangladesh (2), a series of studies initiated the characterization of microbial As^III oxidation in natural environments, including geothermal springs (9, 11, 12, 17, 19, 24, 25, 35, 51) and soils (41); in mining-contaminated environments (6, 13, 40); and, most recently, in anoxic photosynthesis (21, 33). Likewise, progress has been made in the understanding of the biochemistry of the As^III oxidase enzyme (1, 14, 37).

Studies examining the genetics and regulatory control of As^III oxidation have advanced, particularly within the past decade. Muller et al. (43) cloned and described the genes from a betaproteobacterium (subsequently identified as Herminiimonas arsenicoxydans [44]) that encode the small (aoxA [now aioB]) and large (aoxB [now aioA]) subunits of the As^III oxidase enzyme. The As^III oxidase structural genes were later cloned from the above-mentioned Rhizobium NT-26 organism (53). The symbols for genes coding for functions associated with As^III oxidation have recently been changed from aox, aro, or aso to aio (36), and we will use this new aio gene nomenclature throughout this report and in the future.

Kashyap et al. (28) used random transposon mutagenesis to identify and characterize a two-component signal transduction pair, aioS and aioR, that is immediately adjacent to the aioBA genes in an Agrobacterium tumefaciens soil isolate, defining what was assessed at the time to be the aio operon. Parallel studies by Kashyap et al. (29) also identified a molybdate transporter and an Na⁺/H⁺ antiporter that were also found to be essential for As^III oxidase oxidation. Later, and using a similar transposon mutagenesis approach, Koechler et al. (31) also identified the aioSR two-component pair and molybdate transporter as being essential for As^III oxidation. While transposon mutation-based experiments (28, 31) indicated the role and importance of the sensor kinase AioS and its putative regulatory partner AioR (a bacterial enhancer binding protein), direct proof of these two proteins working together as part of a putative As^III signal perception and transduction cascade was just recently provided by Sardiwal et al. (54), who demonstrated the autophosphorylation of an AioS component and the AioS-specific phosphorylation of AioR. Recently, our work has expanded this regulatory model to now include a third component, AioX, which is a periplasmic As^III binding protein that is also essential for aioBA expression (39).

Koechler et al. (31) also isolated rpoN::Tn5::lacZ2 and dnaJ::Tn5::lacZ2 mutants that were defective in As^III oxidation, although those observations were not accompanied by complementation experiments to demonstrate that the loss of function with these mutants was due to the interrupted genes as opposed to polar effects of the transposon on the transcription of adjacent downstream genes. However, ancillary data presented by Koechler et al. (31) provided additional indirect evidence of the rpoN gene product (also referred to as σ⁵⁴, σ^N, and NtrA) being important for aioBA expression. In the current study, with A. tumefaciens strain 5A, we focused on the importance of RpoN, by introducing specific and precise mutations that removed the putative RpoN binding site immediately upstream of the aioBA genes as well as insertionally inactivating rpoN. In addition, we examined the genome for additional putative RpoN-dependent promoters to investigate the role and importance of this sigma factor in As^III-dependent and As^III-independent expression. We provide unambiguous evidence that RpoN is required for As^III oxidation but apparently not for the expressions of other genes associated with a consensus RpoN promoter.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains and plasmids used for this study are listed in Table 1. The A. tumefaciens strains were cultured at 30°C in a defined minimal mannitol medium (MMN) (58) containing mannitol as a carbon and energy source and 50 μM phosphate, with aeration by shaking at 200 rpm. For some experiments, log-phase MMN-grown cells were washed via centrifugation and resuspended in MMN that lacked ammonium to simulate N starvation. Escherichia coli strains Top10 and S17-1 were grown at 37°C in Luria-Bertani (LB) medium. Bacterial growth was monitored via measurements of the culture optical density at 595 nm (OD₅₉₅) by use of a SpectraMax (Molecular Devices, CA) microtiter plate reader. Where indicated, A. tumefaciens growth media were amended with 80 μg ml⁻¹ gentamicin (Gen) and/or 15% sucrose for the selection of double recombinants by using levansucrase selection (see below). Where required for mutant construction, E. coli cells were grown with 20 μg ml⁻¹ Gen. The genome of A. tumefaciens 5A has been sequenced and deposited in the NCBI database (20).

Table 1.

Bacterial strains, plasmids, and primers used in this study

Strain, plasmid, or primer pair	Relevant marker(s) and/or characteristic(s) or sequence	Reference, source, or purpose
Bacterial strains
Agrobacterium tumefaciens
5A	Wild type, soil isolate, AsIII oxidizing	41
5A(ΔPaioB)	aioB promoter region deletion mutant	This study
5A(rpoN::aacC3)	Genr; rpoN mutant by gentamicin cassette interruption	This study
5A(rpoN::aacC3)(pCPP30-rpoN)	Genr Tetr; rpoN::aacC3 complemented strain	This study
5A(PrpoN)	Kmr; 5A with pLSP-PrpoN	This study
5A(PaioB)	Kmr; 5A with pLSP-PaioB	Laboratory stock
5A(PphoB1)	Kmr; 5A with pLSP-PphoB1	Laboratory stock
5A(PpstS1)	Kmr; 5A with pLSP-PpstS1	Laboratory stock
Escherichia coli
S17-1	Pro− Mob+; conjugation donor	57
Top10	High-competency cloning host	Invitrogen
Plasmids
pCR2.1-TOPO	PCR TA cloning vector	Invitrogen
pCPP30	Broad host range; tetA	22
pJQ200sk	Genr traJ oriT sacB; suicide vector	49
pLSP-kt2lacZ	Kmr oriV; lacZ fusion vector used for lacZ fusion constructs	L. Pierson, Texas A&M University
pCR2.1-PaioB	pCR2.1-TOPO with PaioB region	Laboratory stock
pCR2.1-rpoN	pCR2.1-TOPO with rpoN region	This study
pJQ200sk-PaioB	pJQ200sk with deleted PaioB region	This study
pJQ200sk-rpoN::aacC3	pJQ200sk with rpoN gene interrupted with gentamicin cassette	This study
pCPP30::rpoN	pCPP30 containing the complete rpoN gene	This study
pLSP-PrpoN	pLSP-kt2lacZ containing rpoN promoter region	This study
pLSP-PaioB	pLSP-kt2lacZ containing aioB promoter region	Laboratory stock
pLSP-PphoB1	pLSP-kt2lacZ containing phoB1 promoter region	Laboratory stock
pLSP-PpstS1	pLSP-kt2lacZ containing pstS1 promoter region	Laboratory stock
Primer pairs
rpoN-f/rpoN-r (2,124 bp)	5′-GGCGTTCTCATCACCGACCAC-3′/5′-TTTACCAGATACACGCACACTCAT-3′	For construction of rpoN::aacC3
PaioBMu-1f/PaioBMu-1r (540 bp)	5′-GCAGCGACGCCAGTTCCTT-3′/5′-CCCATCCACTAAACTTAAACAGTGTGGGGGTTCGGTTTTC-3′	For deletion of aioB promoter
PaioBMu-2f/PaioBMu-2r (547 bp)	5′-TGTTTAAGTTTAGTGGATGGGCTTCGACATTCAGTGGAGGAG-3′/5′-CGTTGGACAGGCGGCCGTAGATGA-3′	For deletion of aioB promoter
PrpoN-f/PrpoN-r (454 bp)	5′-GGCGTTCTCATCACCGACCAC-3′/5′-GATTGCGCTCCACCTCCTG-3′	For construction of PrpoN::lacZ
B22Gm-f/B22Gm-r (471 bp)	5′-ACAAAGTTAGGTGGCTCAA-3′/5′-TGGGTCGATATCAAAGTGC-3′	Gentamicin cassette-specific primer
P4/P5	5′-GACGTTGCCTATCCCGATGAAGAT-3′/5′-GTTTGTTGATTGGCCAGGTGTAGG-3′	For RT-PCR of aioBa
phoB1-2f/phoB1-2r (361 bp)	5′-TGTTTAAGTTTAGTGGATGGGAGGGGCGCCGGCTATTCAA-3′/5′-CGCTCTAGAGACACCGACGACCTCCCTCAG-3′	For RT-PCR of phoB1
flgBRT-f/flgBRT-r (313 bp)	5′-CAACATCGCCAACGCCAACACA-3′/5′-CCTCTTTACCGTCATCAGCATCAT-3′	For RT-PCR of flgB
nrtART-f/nrtART-r (400 bp)	5′-CCGGCAAGGTGACGCAGAA-3′/5′-TCCCCGGTGGTACAGGCAGTGAA-3′	For RT-PCR of nrtA
glnART-f/glnART-r (356 bp)	5′-AGGCCACTCTTCGGATTG-3′/5′-CATTGATGCCTTCGTGGTTGAT-3′	For RT-PCR of glnA
glnBRT-f/glnBRT-r (227 bp)	5′-CAAGCTCGATGAAGTGAAGG-3′/5′-CGATACGTCCGGTCTGTGC-3′	For RT-PCR of glnB
dctART-f/dctART-r (359 bp)	5′-CGACAAGGCCCATGAGCAGACCA-3′/5′-CAGCGCCGAGGACCACGAACAC-3′	For RT-PCR of dctA

^aSee reference 28.

Construction of A. tumefaciens 5A mutant strains.

Primers used for the construction of A. tumefaciens deletion mutations, the insertional inactivation mutation, as well as the lacZ reporter fusions are provided in Table 1. The deletion of the RpoN binding site was constructed by crossover PCR as described previously (38) and by use of a levansucrase resistance strategy that we described previously (59). Briefly, the 497-bp C-terminal region of aioR was PCR amplified by using primers PaioBMu-1F and PaioBMu-1R (540-bp amplicon that includes 43 bp of the intergenic region). A second PCR amplified the region from 27 bp upstream of the aioB translational start site using primers PaioBMu-2F and PaioBMu-2R. A mixture of these two amplicons was used as the template for a third PCR that included primers PaioBMu-1F and PaioBMu-2R, yielding a product that contained a 67-bp deletion that removed the RpoN binding region of the aioB promoter (Fig. 1B). This product was ligated into pCR2.1-TOPO, yielding plasmid pCR2.1-PaioB. The 1,087-bp fragment was then subcloned into a BamHI- and XbaI-digested pJQ200sk vector, giving pJQ200sk-PaioB. This plasmid was transformed into E. coli S17-1 cells and then mobilized to A. tumefaciens 5A by conjugation. Merodiploids were selected on MMN agar plus Gen (MMN^Gen agar), with double crossovers then being selected on MMN agar containing 15% sucrose, which selects for the loss of the plasmid-borne sacB gene. Gentamicin-sensitive, sucrose-resistant colonies were screened and confirmed by PCR and sequencing, yielding the in-frame ΔPaioB deletion mutant.

Fig 1.

Physical arrangement and locations of genes involved in this study, and PCR-based evidence of mutation constructs. (A) Relative positions of the putative RpoN binding sites associated with the aioB gene and located between the phoB1 and pstS1 coding regions. (B) Sequence of the aioB promoter region that was deleted. The 3′ end of aioR and the 5′ end of aioB are boxed, and the ribosome binding site for aioB is highlighted in dark gray. The consensus RpoN binding site is highlighted with boldface uppercase type, and the region deleted (ΔPaioB mutant) is highlighted in light gray. In addition, PCR evidence to document the 67-bp deletion created in the aioB promoter region is provided. (C) Position of rpoN in strain 5A relative to the flanking genes, the approximate insertion point of the aacC3 cassette (2.5 kb) (inverted black arrowhead) in the rpoN coding region (1.48 kb), the region of DNA that was PCR cloned and used for complementation of the rpoN::aacC3 mutant, and the promoter region of rpoN used for the rpoN::lacZ reporter construct (dashed box). Also, PCR evidence documents the insertion of aacC3 in the rpoN gene (4.6-kb construct). HP, hypothetical protein; WT, wild type; MW, molecular weight marker.

Primer pair rpoN-f/rpoN-r was used to construct an insertion mutation in the rpoN coding region. The 2.1-kb amplified fragment was ligated and cloned into pCR2.1-TOPO, creating pCR2.1-rpoN. The rpoN gene harbored by pCR2.1-rpoN was then inactivated by the insertion of a gentamicin cassette (aacC3) at a ClaI site at bp 1378. The gentamicin cassette was prepared by the digestion of Tn5-B22 with BamHI and HindIII, polished with Klenow fragment (Promega) to create blunt ends, and ligated into the Klenow fragment-treated ClaI site. The rpoN::aacC3 construct (4.6 kb) was then cloned into the BamHI and XbaI cloning sites of pJQ200sk, creating pJQ200sk-rpoN::aacC3, which was then transformed into S17-1 cells, followed by conjugation into strain 5A. Mutants wherein the wild-type allele was replaced by this construct were generated by using the above-described levansucrase selection strategy. For the complementation of the rpoN::aacC3 insertion mutant, the same rpoN reading frame and upstream promoter region were PCR cloned by using primer pair rpoN-f/rpoN-r. The rpoN fragment from pCR2.1-rpoN was subcloned into the EcoRI cloning site of the pCPP30 vector, creating pCPP30-rpoN, which was transformed into S17-1 cells and then introduced into the rpoN::aacC3 mutant by conjugation.

Arsenic chemical species analysis.

As^III oxidation was assayed in the wild type and the rpoN::aacC3 mutant by using HPLC-ICP-MS (high-performance liquid chromatography–inductively coupled plasma mass spectrometry) (Agilent 7500) to separate (ammonium carbonate mobile phase at a flow rate of 1 ml min⁻¹) and quantify As^III and As^V in culture supernatants. For calibration, standard solutions of 25, 50, 75, and 100 μM combined As^III and As^V were employed. Cell-free culture fluids (0.22-μm filtration of centrifugation supernatants; Millipore) from cells grown in low-phosphate MMN containing 100 μM As^III were harvested at time zero and at the log and stationary phases.

Promoter activity analysis.

The aioB, rpoN, phoB1, and pstS1 promoter regions of the A. tumefaciens 5A genome were amplified by PCR using primers (Table 1) that incorporated EcoRI and BamHI restriction sites that allowed for directional cloning into lacZ reporter plasmid pLSP-kt2lacZ (Table 1). The resulting constructs were designated pLSP-PaioB, pLSP-PrpoN, pLSP-PphoB1, and pLSP-PpstS1, respectively, and were used for the analysis of promoter activity after the introduction of each construct into A. tumefaciens wild-type strain 5A via conjugation using E. coli strain S17-1. The lacZ reporter assays used cells grown in 50 ml of MMN with kanamycin under various conditions (with or without As^III) and incubated at 30°C on a rotary shaker at 200 rpm. After the various incubation times, 1-ml aliquots from each flask were collected and used for assays of β-galactosidase activity according to a method described previously by Miller (42).

Reverse transcriptase PCR analysis of gene expression.

Cells were incubated in MMN with or without 100 μM As^III for 4 h. The total RNA of the cells (OD₅₉₅ of ∼0.2 to 0.3) was extracted and purified with an RNeasy kit (Qiagen) in accordance with the manufacturer's instructions. DNA contamination was removed by DNase I treatment and cleanup with a Turbo DNA-free kit (Ambion). The mRNA was then reverse transcribed into cDNA by using avian myeloblastosis virus (AMV) reverse transcriptase (RT) (Promega) with gene-specific primers (Table 1), according to the manufacturer's instructions, and with 500 ng RNA as the template in each RT reaction. PCR programs were standard, with primer-specific annealing temperatures, but all programs included final extension steps performed at 68°C for 5 min.

RESULTS

RpoN is required for As^III oxidation.

Kashyap et al. (28) previously described an aio gene cluster comprised of aioS-aioR-aioB-aioA-cyt-moeA and provided RT-PCR evidence for these genes being expressed in a monocistronic fashion. Those authors also described the isolation of an aioR mutant having a Tn5-B22 transposon inserted into the 3′ region of the coding frame and which was interpreted as having polar effects on the transcription of the As^III oxidase-encoding genes aioBA. However, the intergenic region separating aioR and aioB (Fig. 1) was noted to be significant, perhaps accommodating another independently regulated promoter. This was consistent with a previous suggestion by Sardiwal et al. (54) and an earlier personal communication with J. M. Santini that a σ⁵⁴-dependent promoter was directly upstream of the Rhizobium NT-26 aioBA genes. An analysis of this region in A. tumefaciens strain 5A yielded a similar conclusion (Fig. 1A), and thus, experiments were designed to assess this possibility. Specifically, a deletion mutation was precisely constructed in the region directly upstream of aioB which included the entire predicted RpoN binding site but which excluded any of the flanking coding frames for aioR or aioB and likewise excluded the putative ribosome binding site for aioB (Fig. 1B); this mutant is referred to as the ΔPaioB mutant. In parallel, the wild-type rpoN allele in strain 5A was replaced with an insertionally inactivated rpoN allele (rpoN::aacC3), creating an RpoN⁻ mutant (Fig. 1C). Both mutations yielded organisms incapable of oxidizing As^III (Fig. 2A and B), but the complementation of the rpoN::aacC3 mutant with a defined region of DNA that included only the rpoN reading frame and the upstream promoter region (Fig. 1C) reversed the defect back to wild-type As^III oxidation (Fig. 2A). Further analyses of the ΔPaioB deletion mutant illustrated that the expression of aioBA was silenced (Fig. 2C). An analysis of aioB::lacZ and rpoN::lacZ reporters illustrated the expected As^III-dependent expression of aioBA, whereas rpoN expression appeared to be constitutive and not influenced by As^III (Fig. 3).

Fig 2.

As^III oxidation phenotype of A. tumefaciens wild-type strain 5A and mutants described in this study. (A) Qualitative assessment of As^III oxidation using the AgNO₃ staining method for wild-type strain 5A, the ΔPaioB mutant, the rpoN::aacC3 mutant, and the rescued rpoN::aacC3 mutant carrying pCPP30::rpoN. The brown precipitate indicates the presence of As^V. (B) HPLC-ICP-MS-based measurement of aqueous As^III and As^V in As^III oxidation assays with wild-type strain 5A and the ΔPaioB mutant. Error bars represent the standard deviations of data from two replicate cultures. (C) RT-PCR monitoring of aioBA expression in the wild type and the ΔPaioB mutant in the log and stationary phases of cells grown in the presence of 100 μM As^III.

Fig 3.

Reporter gene assays to track the expressions of aioB and rpoN as a function of time after the addition (or not) of 100 μM As^III. β-Galactosidase activity is presented as Miller units (MU), and error bars represent the standard deviations of data from two replicate cultures.

An examination of the strain 5A genome for genes/operons known in other bacteria to be controlled by RpoN found putative RpoN binding sites associated with 16 genes coding for a variety of functions and which displayed the well-defined GG-(N10)-GC consensus RpoN binding site (Fig. 4). For all of these genes, the putative RpoN binding sites contained a perfectly conserved thymine at the 5′ end and a near-perfect consensus addition of a thymine at the 3′ end (Fig. 4). These genes code for functions associated with the phosphate stress response (phoB1 and pstS1), flagellum construction (flgB), nitrogen metabolism (nrtA, glnA, and glnB), poly-β-hydroxybutyrate synthesis (phbB), and others (Fig. 4).

Fig 4.

Alignment of the RpoN binding sites in the promoter regions from a sample of genes of the genome of strain 5A. Nucleotides highly conserved are highlighted in boldface type, and the consensus sequence depicting the relative conservation of each predicted position is shown at the bottom. CoA, coenzyme A.

The importance of RpoN for the expression of a random sample of these genes was examined by qualitative RT-PCRs using RNA extracted from cells that were exposed to As^III or that were starved for nitrogen; the latter analysis was included because of the known association of RpoN with the regulation of N metabolism (e.g., see reference 16). Consistent with the experiments summarized in Fig. 2C, aioB expression was significantly upregulated in As^III-exposed wild-type cells but was not detectable in the rpoN::aacC3 mutant regardless of the incubation conditions (Fig. 5A). Surprisingly, dctA was also upregulated by As^III and, like aioB, was silent in the rpoN::aacC3 mutant. The glnA gene required RpoN for full expression, but it was found to be expressed under all incubation conditions and in both strains although at reduced levels in the rpoN::aacC3 mutant and at further reduced levels in the rpoN::aacC3 mutant incubated under N starvation conditions (Fig. 5A). Examples of genes where transcription appeared to be unaffected by the loss of functional RpoN included phoB1, glnB, and flgB (Fig. 5A). RT-PCR analysis showed that phoB1 is upregulated by As^III (Fig. 5B). A quantitative lacZ reporter gene analysis of phoB1 and the adjacent (but divergently transcribed) pstS1 gene showed that both genes were dramatically induced by As^III; however, neither gene was affected by the rpoN::aacC3 mutation (Fig. 5B). A significantly more expansive characterization of the As^III and phosphate coregulation of the aio, pst, and pho genes will be presented elsewhere (Y. S. Kang, B. Bothner, C. Rensing, and T. R. McDermott, submitted for publication).

Fig 5.

Influence of the rpoN::aacC3 mutation on expressions of genes associated with an RpoN binding site. (A) Qualitative standardized RT-PCR expression analysis of genes (in the wild-type strain or the rpoN::aacC3 mutant) associated with an RpoN binding site. Cells were grown in the absence (−) or in the presence (+) of As^III or under conditions with or without N. For each gene, RT-PCR amplicon volumes loaded into the gel were standardized. (B) β-Galactosidase activity (in Miller units) derived from the transcription of the phoB1::lacZ and pstS1::lacZ reporters in the wild type and the rpoN::aacC3 mutant and as a function of cells grown in the presence or absence of 100 μM As^III.

DISCUSSION

The studies summarized in this report strengthen our understanding of transcriptional controls that regulate the expressions of key genes involved in As^III oxidation. The alternative sigma factor RpoN is essential for aioBA expression and As^III oxidation (Fig. 2), a conclusion supported by two independent lines of evidence. First, an insertional interruption of the rpoN coding region silenced aioBA expression and resulted in a null As^III oxidation phenotype. However, this could be reversed to the wild-type status with a defined DNA fragment containing the rpoN coding region and minimal flanking DNA. Second, the introduction of a precise deletion in a sequence that exhibited the well-known GG-(N10)-GC signature RpoN binding site (4) directly upstream of the aioB gene (but leaving the aioB translational start site intact) also resulted in the silencing of aioBA expression and As^III oxidation. A recent random transposon-based mutation study with the organism Herminiimonas arsenicoxydans identified an rpoN::Tn5 mutant as being defective in As^III oxidation (31). However, the necessary complementation experiments were not performed to confirm that the phenotype was actually due to the interrupted rpoN gene as opposed to transposon polar effects on the expression of rpoX (an rpoS-like sigma factor [61]) and other genes immediately downstream of what Duquesne et al. (13) depicted previously as an H. arsenicoxydans operon that contains rpoN (31). The precisely targeted mutation and complementation experiments in the present study provide definitive evidence that the rpoN gene product is essential for aioBA expression and, hence, As^III oxidation. This observation also allows a more definitive understanding of the As^III oxidase-negative phenotype of the aioR::Tn5-B22 mutant described previously (28). Specifically, the Tn5-B22 insertion in the extreme 3′ region of aioR in the aioS-aioR-aioB-aioA operon likely disrupted the DNA binding component of the regulator AioR and is consistent with AioR participating as a bacterial enhancer binding protein in conjunction with RpoN to facilitate the induction of aioBA. Kashyap et al. (28) also reported evidence that aioSR are cotranscribed with aioBA in what appears to be an operon arrangement. In this specific context, it is not presently clear why the loss of the RpoN binding site would then interrupt aioBA expression in strain 5A.

The requirement for RpoN as an essential regulator of As^III oxidation is not necessarily surprising given the wide variety of cellular functions that RpoN regulates (reviewed in references 34 and 56). Indeed, we found that the consensus GG-(N10)-GC RpoN binding-site signature associated with aioB was also found elsewhere in the genome (Fig. 4), being associated with a wide range of functions previously noted for other bacteria (34, 50, 56). Given such a range of functions, it is also not surprising that the expression of rpoN appears to be constitutive in A. tumefaciens strain 5A and is not influenced by a specific solute or ion such as As^III (Fig. 3), which is in agreement with previous studies (7, 26).

What was surprising, however, was the observation of a C₄-dicarboxylate transporter (dctA) (Fig. 4) upregulated by As^III (Fig. 5A). It is not immediately apparent why, or how, such a transporter might be involved (or not) in As^III oxidation or metabolism. Previously reported microarray studies also identified the As^III induction of a C₄-dicarboxylate transporter in H. arsenicoxydans (8). The Dct system in strain 5A is a secondary transporter operating without a periplasmic solute binding protein, whereas the As^III-inducible dctPMQ dicarboxylate transporter in H. arsenicoxydans is annotated as a tripartite ATP-independent periplasmic (TRAP)-type transporter (reviewed in reference 45). While these two C₄-dicarboxylate transporters are structurally different, their function is the same. Such shared regulatory and functional similarities would seem not to be a simple coincidence and suggest a potential involvement in some facet of arsenic metabolism. Rat and plant mitochondrial C₄-dicarboxylate transporters have been shown to facilitate the exchange of malate and phosphate (15, 27), with arsenate robustly substituting for phosphate in the shuttle (15). In the context of a prokaryote cell, such a mechanism could result in As^V removal from the cell and is an issue that we are examining in more detail.

The other instance where RpoN appeared to be important involved glnA. Not surprisingly, As^III had no influence on glnA expression in the wild-type or mutant strains. Also, glnA expression in the wild-type strain was observed under all conditions, although the level of expression varied depending on the incubation conditions. The glnA expression level was noticeably reduced in the rpoN::aacC3 mutant under N-replete conditions, similar to what was found previously for Xylella fastidiosa (10), and was even further reduced when the mutant was starved for N (Fig. 5A).

In contrast to aioB and dctA, genes such as phoB1, pstS1, glnB, and flgB were examples where an association with a consensus RpoN binding site did not correlate with an RpoN requirement (Fig. 5), at least under the experimental conditions considered. Previous predictions suggested an RpoN involvement wherever the GG-(N10)-GC motif was present (30, 47), but others have documented observations similar to those reported here. For example, in Pseudomonas putida, the presence of the GG-(N10)-GC motif does not always imply RpoN-associated regulation (55), and the lack of functional RpoN in this organism was shown previously to be essentially irrelevant to bacterial environmental responses normally associated with RpoN-controlled regulator functions (5). Furthermore, there is evidence of RpoN potentially interacting with a repressor and, hence, participating in a negative regulatory sense (3, 32) or that the spacing between the GC doublet and the transcriptional start site might not always be optimal (55). These types of explanations likely apply to A. tumefaciens strain 5A as well.