Anaerobic regulation by an atypical Arc system in Shewanella oneidensis


E-mail; Tel. (+1) 626 395 6790; Fax (+1) 626 683 0621.


Shewanella oneidensis strain MR-1 is well known for its respiratory versatility, yet little is understood about how it regulates genes involved in anaerobic respiration. The Arc two-component system plays an important role in this process in Escherichia coli; therefore, we determined its function in S. oneidensis. arcA from S. oneidensis complements an E. coli arcA mutant, but the Arc regulon in S. oneidensis constitutes a different suite of genes. For example, one of the strongest ArcA-regulated gene clusters in E. coli, sdh, is not regulated by the Arc system in S. oneidensis, and the cyd locus, which is induced by ArcA in E. coli under microaerobic conditions, is repressed by ArcA in S. oneidensis under anaerobic conditions. One locus that we identified as being potentially regulated by ArcA in S. oneidensis contains genes predicted to encode subunits of a dimethyl sulphoxide (DMSO) reductase. We demonstrate that these genes encode a functional DMSO reductase, and that an arcA mutant cannot fully induce their expression and is defective in growing on DMSO under anaerobic conditions. While S. oneidensis lacks a highly conserved full-length ArcB homologue, ArcA is partially activated by a small protein homologous to the histidine phosphotransfer domain of ArcB from E. coli, HptA. This protein alone is unable to compensate for the lack of arcB in E. coli, indicating that another protein is required in addition to HptA to activate ArcA in S. oneidensis.


Surprisingly little is known about the regulation of anaerobic respiration in Shewanella species, given that they can utilize an extraordinarily wide array of anaerobic electron acceptors. These acceptors include: soluble and insoluble forms of iron and manganese [Fe(OH)3, MnO2, Fe(III)-citrate], organic compounds [fumarate, dimethyl sulphoxide (DMSO), trimethylamine-N-oxide (TMAO), humic substances] and inorganic compounds (nitrate, nitrite, thiosulphate and sulphite); in addition, Shewanella species can reduce a variety of toxic metal(loids)s (As, Cr, Se, Tc and U) (Nealson and Scott, 2003). In some cases, the reduction of these compounds can have an important effect on their mobility in the environment, and thus, Shewanella species have received attention in the context of metal bioremediation (Liu et al., 2002). Given the broad spectrum of electron acceptors used by Shewanella species, a sophisticated regulatory network must exist that modulates the expression of genes involved in their reduction. Shewanella oneidensis strain MR-1 is a Gram-negative γ-proteobacterium that can be isolated in many aquatic environments (Venkateswaran et al., 1999) and is an important model organism used to study unusual respiratory processes (Nealson et al., 2002; Lovley et al., 2004). A better understanding of how S. oneidensis regulates the expression of its various respiratory systems will not only improve our appreciation for how bacteria adapt to changing environmental conditions, but has the potential to be useful in the design of appropriate strategies to stimulate its respiratory activity for the purpose of bioremediation.

Although little is known about regulation of anaerobic metabolism in S. oneidensis, it is clearly different from the Escherichia coli paradigm. Mutants lacking CRP (cyclic AMP receptor protein) in S. oneidensis are defective in utilizing several anaerobic electron acceptors, suggesting that CRP plays a role in regulating anaerobic respiration, but it is unclear whether this is a direct or indirect effect (Saffarini et al., 2003). Interestingly, the major regulatory component of anaerobic respiration in E. coil, FNR (fumarate nitrate regulator), appears to have no significant role in regulating this process in S. oneidensis (Maier and Myers, 2001; Beliaev et al., 2002). Another system that partially overlaps the FNR regulon (Liu and De Wulf, 2004), and that is also used for regulating aspects of anaerobic respiration in E. coli, is the Arc (aerobic respiration control) system (Iuchi and Lin, 1988).

Arc is a two-component system in E. coli, comprising the sensor kinase ArcB and the DNA binding response regulator ArcA (Iuchi and Lin, 1995; Lynch and Lin, 1996a). ArcB is localized in the cytoplasmic membrane and undergoes autophosphorylation when conditions become anaerobic by sensing the redox state of the quinone pool (Georgellis et al., 2001; Malpica et al., 2004). Under these conditions, ArcB stimulates changes in gene expression by transferring a phosphate group to ArcA (Kwon et al., 2000), which then becomes competent to bind DNA (Lynch and Lin, 1996a,b; Jeon et al., 2001). Phosphorylated ArcA (ArcA-P) has the capacity to influence transcription of many genes in E. coli, both positively and negatively, which has been the subject of several comprehensive bioinformatic and array-based approaches to predict what genes it regulates (Lynch and Lin, 1996b; McGuire et al., 1999; Liu and De Wulf, 2004). It is now clear that while the Arc system regulates some aspects of respiration, its regulon also partially overlaps with that of FNR, and is more expansive than originally thought (Liu and De Wulf, 2004). Because the Arc system in E. coli regulates gene expression in response to anaerobic conditions, it is a good candidate for possibly regulating anaerobic processes in S. oneidensis.

This work represents the initial steps in characterizing the Arc system in S. oneidensis using genetic, bioinformatic and molecular techniques. Because S. oneidensis has tremendous diversity in the kinds of electron acceptors it can use for respiration, we were particularly interested in identifying respiratory genes potentially regulated by ArcA. We demonstrate that genes required for DMSO respiration are positively regulated under anaerobic conditions by ArcA and HptA, a protein similar to the histidine phosphotransfer domain (HPt) of ArcB in E. coli. These proteins comprise two parts of an atypical Arc system in S. oneidensis that is likely to include at least one additional sensory factor. We also begin to explore general differences between the strategies of anaerobic regulation in E. coli and S. oneidensis.

Results and discussion

ArcA from E. coli and S. oneidensis are functionally equivalent

To begin probing the role of the Arc system in S. oneidensis, we predicted potential ArcA binding sites using a bioinformatics approach. Because no data regarding ArcA from S. oneidensis exist in the literature, we based our approach on the large body of biochemical and phenotypic data from E. coli. Implicit in this approach is the assumption that the ArcA proteins from both systems bind the same, or similar, nucleotide sequence motifs. DNA–DNA hybridization microarray experiments showed that arcA was one of the most highly conserved genes between E. coli and S. oneidensis (Murray et al., 2001). While the amino acid sequence identity between the two proteins (81%) strongly suggested that they would recognize the same sequence (Fig. 1), we performed complementation experiments to demonstrate this in vivo. We cloned the arcA gene and its upstream promoter region from S. oneidensis (pARCA-1, see Table 1) and provided it in trans in an E. coli strain defective in arcA (BW29409). Two phenotypes were tested that have been attributed to defects in arcA: (i) increased sensitivity to toluidine blue dye (Iuchi and Lin, 1988) and (ii) decreased aerobic growth rate (Oshima et al., 2002). Both zones of inhibition caused by the toluidine blue dye (Fig. 2A) and aerobic growth defects (Fig. 2B) were eliminated when arcA from S. oneidensis was provided in the E. coli arcA mutant strain. Additionally, the reciprocal experiment was performed where arcA from E. coli complemented the aerobic growth defect of an arcA mutant in S. oneidensis (data not shown). These complementation data, coupled with the high amino acid identity shared between the ArcA homologues from these two organisms (Fig. 1), strongly suggest that the response regulator is binding the same, or very similar, nucleotide sequences, and therefore validates using a bioinformatics approach to identify genes putatively regulated by ArcA in S. oneidensis.

Figure 1.

Alignment of ArcA homologues from E. coli and S. oneidensis. Alignment generated using the Blast 2 Sequences program at the National Center for Biotechnology Information ( (Tatusova and Madden, 1999). The proteins are 87% similar and 81% identical.

Table 1.  Strains and plasmids used in this study.
StrainRelevant genotype or markers; characteristics and usesSource or reference
E. coli
 DH10βHost for E. coli cloningLife Technology
 WM3064Donor strain for conjugation; ΔdapAW. Metcalf, University of Illinois
 BW25113Parent strain of BW29409 and BW29859Coli Genetic Stock Center
 BW29409ΔarcAColi Genetic Stock Center
 BW29859ΔarcBColi Genetic Stock Center
 JG32BW29409 with pBBR1MCS-3This study
 JG33BW29409 with pARCA-1This study
 JG109BW25113 with pBBR1MCS-3This study
 JG111BW29859 with pBBR1MCS-3This study
 JG112BW29859 with pSO1327-2 (hptA)This study
 JG132BW29859 with pARCBThis study
 MR-1Wild typeLab stock
 JG90arcA::KmThis study
 JG46ΔhptAThis study
 JG82JG46 with pBBR1MCS-3This study
 JG84JG46 with pSO1327-2 (hptA)This study
 JG89MR-1 with pBBR1MCS-3This study
 JG115JG46 with pARCBThis study
 JG130dmsE::KmThis study
 pBBR1MCS-3Conjugatable cloning vectorKovach et al. (1995)
 pARCA-1arcA and upstream promoter region from S. oneidensis cloned into pBBR1MCS-3This study
 pARCBarcB and upstream promoter region from E. coli cloned into pBBR1MCS-3This study
 pSO1327-2hptA and upstream promoter region from S. oneidensis cloned into pBBR1MCS-3This study
 pUC4KKm-resistant cloning vectorLab stock
Figure 2.

arcA from S. oneidensis can complement defects of an E. coli arcA mutant.
A. Soft agar overlay containing BW29409 (arcA) with indicated plasmids was made over LB agar plates and toluidine blue was spotted in the centre of the plate (see Experimental procedures). Pictures are a close-up of the dye spot after ∼20 h of growth at 37°C. The white bar indicates the zone of inhibition detected in strains containing pBBR1MCS-3. No zone was detected in strains containing pARCA-1.
B. Aerobic growth curves in LB of E. coli strains: wild type with pBBR1MCS-3 (▪), arcA with pBBR1MCS-3 (▴) and arcA with pARCA-1 (▾). Error bars represent standard deviation for triplicate cultures.

Identification of a locus involved in anaerobic respiration of DMSO

A comprehensive paper on ArcA binding sites and regulation in E. coli was recently published (Liu and De Wulf, 2004). We adapted the binding site matrix used by these authors (see Experimental procedures) and applied the matrix to the genome and megaplasmid of S. oneidensis strain MR-1. Results of this analysis are available online as supplementary data ( While a detailed bioinformatics study of ArcA-P binding sites is currently underway, preliminary investigation of these results identified several potential target genes and operons that could be regulated by this protein. In several instances ArcA-P binding sites are predicted in the intergenic region between two divergently transcribed gene clusters, making it difficult to anticipate which cluster (or both) is potentially regulated. It is striking that cydAB is the only gene cluster identified by our bioinformatic search in S. oneidensis that has been shown to be regulated by the Arc system in E. coli (see Supplementary material and Discussion below). Of the genes that appear to be regulated by ArcA-P in S. oneidensis, we focused on a putative operon that contains genes predicted to encode subunits of a DMSO reductase. While it is known that S. oneidensis can use DMSO as an anaerobic electron acceptor (Nealson and Scott, 2003), the genes encoding this activity have not been identified. Potential binding sites were found starting at 276 bp and 365 bp upstream of the start codon for SO1427, the first gene of this operon (Fig. 3A). Two genes located near the end of this gene cluster were predicted to encode DMSO reductase subunits A and B (SO1429 and SO1430 respectively). Six genes (SO1427–1432) appear to be co-ordinately regulated and maximally induced under anaerobic conditions in the presence of DMSO (Fig. 3B, data not shown), suggesting that they may be involved in DMSO utilization, and that they may also be part of an operon. Expression of the first four genes was approximately twofold higher when grown with DMSO as the sole electron acceptor compared with when the strain was grown on fumarate. These results suggest an additional level of regulation beyond anaerobiasis: either expression of these genes is further induced in the presence of DMSO or fumarate partially represses their expression. Additional expression studies with S. oneidensis MR-1 grown in the presence of both DMSO and fumarate are consistent with the latter possibility (see below).

Figure 3.

Induction of dms gene expression in MR-1 grown with different terminal electron acceptors using Q-RT-PCR.
A. Diagram of dms gene cluster in S. oneidensis. Lines indicate putative ArcA-P binding sites at 276 bp and 365 bp upstream of dmsE start codon.
B. Wild-type S. oneidensis strain MR-1 was grown to early log phase with oxygen, DMSO or fumarate as the sole electron acceptor. Relative expression was measured using Q-RT-PCR, then normalized to expression under aerobic (oxygen) conditions to calculate fold induction of dmsEFAB. Expression of dmsGH was not quantified in this experiment.

To test whether these genes are involved in DMSO utilization, the first gene in the putative DMSO utilization operon (SO1427) was replaced with the Km resistance gene from pUC4K. This strain was unable to utilize DMSO as an anaerobic electron acceptor, but had no defect utilizing fumarate (Fig. 4A and B). S. oneidensis grows faster anaerobically with fumarate as the sole electron acceptor than DMSO, but does not achieve the same density under the conditions tested. As a dmsE in-frame deletion has only a partial defect in DMSO utilization (data not shown), the dmsE::Km insertion is likely polar on the downstream genes in this cluster. Based on both phenotypic and expression data, we will refer to the SO1427–1432 gene cluster as dmsEFABGH[note that SO1429 was annotated as dmaA (, but should be dmsA, based on similarity to the gene of the same name from E. coli and other organisms]. A second set of DMSO reductase-related subunits were annotated in the SO4357–4360 gene cluster; however, these genes were not induced either anaerobically or in the presence of DMSO, and expression remained at levels similar to aerobic expression of dmsEAB (data not shown). Together, this suggests that the second gene cluster is not required for DMSO reduction under the conditions tested.

Figure 4.

Analysis of anaerobic growth of arcA and dmsE mutants in S. oneidensis.
A and B. A polar insertion in dmsE eliminates the ability of S. oneidensis to respire DMSO anaerobically. MR-1 (▪) and JG130 (▾, dmsE::Km) grown anaerobically with fumarate (A) or DMSO (B) as the sole electron acceptor.
C and D. arcA mutants in S. oneidensis are defective using DMSO as an anaerobic electron acceptor. (C) Anaerobic growth with fumarate as the sole electron acceptor. (D) Anaerobic growth with DMSO as the sole electron acceptor. In both experiments, MR-1 is indicated by (▪) and JG90 (arcA) by (▾).
Error bars represent standard deviation for triplicate cultures in all experiments.

ArcA is required for induction of the dms gene cluster and for DMSO respiration

ArcA is known to regulate genes both positively and negatively in E. coli. If ArcA were required to induce expression of the dms operon, an arcA null strain should be unable to grow anaerobically with DMSO as the sole electron acceptor. To test this, we generated a strain of S. oneidensis lacking arcA, and found that these mutants have a severe growth defect anaerobically using DMSO as the sole electron acceptor compared with the wild-type parent strain (Fig. 4C). arcA mutant strains had no defect growing with fumarate (Fig. 4D), TMAO or nitrate, and had no change in the rate of iron mineral reduction (data not shown), but had a similar aerobic growth defect relative to the E. coli arcA strain (see below).

To address whether the DMSO growth defect of arcA mutants resulted from lack of induction of the dms operon, we assayed expression of three dms genes from wild-type and arcA mutant strains grown aerobically or anaerobically in the presence of both fumarate and DMSO. Both anaerobic electron acceptors were used to allow growth of the arcA mutant strain. Expression of dmsE, dmsA and dmsB was induced approximately sevenfold, 25-fold and 23-fold (respectively) in wild-type MR-1 when grown anaerobically with DMSO and fumarate compared with aerobically grown cells (Fig. 5). These ratios were similar to the ratios observed with MR-1 grown with only fumarate (Fig. 3B), suggesting that the dms genes tested are repressed in the presence of fumarate. Induction of these genes was severely decreased in arcA mutants grown anaerobically, demonstrating that ArcA is required for their induction (Fig. 5). A slight induction of these genes (approximately twofold) is still observed in the arcA mutant, suggesting that an additional factor is also influencing their expression anaerobically, possibly CRP. S. oneidensis mutants lacking CRP are known to have lower levels of DMSO reductase activity compared with wild type (Saffarini et al., 2003).

Figure 5.

arcA and hptA mutants in S. oneidensis cannot fully induce expression of dmsE, dmsA or dmsB under anaerobic conditions. MR-1, arcA (JG90) and hptA (JG46) were grown to early log phase aerobically, or anaerobically with both DMSO and fumarate as electron acceptors. Q-RT-PCR was used to quantify expression. All data were normalized to expression of recA. Relative expression was determined using an equation described in Experimental procedures. Error bars represent standard deviation for duplicate cultures.

Based on the results presented thus far, we conclude that ArcA is active in anaerobically growing cells, much like its counterpart in E. coli. However, we predict that the proteins required to activate ArcA in S. oneidensis are likely to be significantly different from those used by E. coli due to the absence of a highly conserved full-length homologue of arcB. The following section details our initial steps in elucidating the mechanism of ArcA activation in S. oneidensis.

SO1327 (htpA) is required for maximal expression of the dms operon

The locus SO1327 is predicted to encode a small protein (121 amino acids) with a significant degree of similarity (58%) to the last ∼120 amino acids of ArcB from E. coli(Fig. 6). This domain in ArcB is known as the histidine phosphotransfer, or HPt domain (Ishige et al., 1994; Kato et al., 1997), and is required for transferring phosphate groups to the ArcA response regulator in E. coli (Kwon et al., 2000). Significantly, the histidine residue known to be phosphorylated in the HPt domain of ArcB (His-717) is conserved in SO1327 (His-62) and is required for complementation in S. oneidensis (data not shown). Based on this similarity, we named the SO1327 locus in S. oneidensis hptA (histidine phosphotransfer). The function of the HPt domain from ArcB in E. coli has also been defined in vitro (Georgellis et al., 1997), and a potential role for transferring phosphate groups to other response regulators has been suggested (Yaku et al., 1997; Matsushika and Mizuno, 1998). Because this domain has been shown to function on its own in vitro, we hypothesized that hptA encodes a protein that could function in a similar fashion, activating ArcA via phosphate transfer. To test this model, we generated an in-frame deletion of hptA in S. oneidensis.

Figure 6.

Sequence comparison between ArcB from E. coli and HptA from S. oneidensis.
A. Domain alignment between ArcB (after Kwon et al., 2000) and HptA. Phosphorelay residues are indicated in the appropriate domains.
B. Alignment of HptA and the HPt domain of ArcB. Alignment generated using blast at NCBI ( Conserved histidine residue (His-717 in ArcB, His-62 in HptA) is shown in bold.

If HptA is involved in the activation of ArcA, we would expect the hptA null mutant to have phenotypes similar to an arcA mutant, and this is what we observed. Mutants lacking hptA have an aerobic growth defect similar to arcA mutants in S. oneidensis(Fig. 7A) and have no growth defect anaerobically with fumarate (Fig. 7B). Additionally, hptA mutants have a slight defect utilizing DMSO as an anaerobic electron acceptor (Fig. 7C). Consistent with this phenotype, expression of the three dms genes tested (dmsEAB) was still induced in the hptA mutant background, compared with MR-1, but to a lesser extent (Fig. 5). Expression of these genes was three to four times higher in the hptA background compared with arcA, but expression of dmsA and dmsB were 2.5- and 4-fold lower than wild type. These results suggest that ArcA is partially active in the hptA mutant background.

Figure 7.

Growth curves showing various phenotypes attributed to arcA and hptA mutants in S. oneidensis. Panels are as indicated: (A) aerobic growth,  (B)  anaerobic  growth  with  fumarate  as  the  sole  electron  acceptor,  (C)  anaerobic  growth  with  DMSO  as  the  sole  electron  acceptor. MR-1 is indicated by (▪), arcA by (▾) and hptA by (▴). Error bars represent standard deviation for triplicate cultures.

To begin to test other putative ArcA-regulated genes, and to ask whether hptA mutants also had a partial effect on their transcription, we assayed expression of several genes in wild-type, arcA and hptA backgrounds. Table 2 shows four additional genes regulated by ArcA in S. oneidensis. Note that ArcA acts as a repressor of their transcription given that their ratios are 3- to 18-fold higher in the arcA mutant background, under anaerobic conditions (Table 2). This contrasts with the regulation of dmsEAB, where ArcA is required for activation (Fig. 5), but is consistent with the fact that ArcA can act both positively and negatively to direct transcription in E. coli (Lynch and Lin, 1996a). As observed with the regulation of the dmsEAB genes, hptA mutants had only a partial effect on transcription at four additional loci compared with arcA mutants (Table 2), consistent with the model that ArcA can be partially activated in a hptA mutant background.

Table 2.  Other genes regulated by ArcA and HptA, identified by bioinformatics approach, and verified using Q-RT-PCR.
Gene testedBinding energyaExpression differencebAnnotationcComments
  • a

    . Binding energy predicted using weight matrix as previously described (Brown and Callan, 2004).

  • b

    . Expression difference, relative to MR-1, as described in Experimental procedures. Strains were grown anaerobically in the presence of fumarate and DMSO.

  • c

    . Annotation from Heidelberg et al. (2002), available through

SO20995.60 7.22.7Quinone-reactive Ni/Fe hydrogenase, small subunit 
SO14795.7617.76.3Hypothetical protein 
SO21995.83 3.12.4Hypothetical proteinSingle gene
cydB6.98 6.23.7Cytochrome d ubiquinol oxidase, subunit IIDownstream of cydA (SO3286)

hptA from S. oneidensis is not functionally equivalent to arcB from E. coli

To test whether hptA alone is sufficient to activate ArcA in E. coli, we provided a complementing clone (pSO1327-2) in trans to an E. coli arcB deletion strain (BW29859). This construct had no effect on the aerobic growth defect of the arcB strain, whereas a plasmid containing the arcB gene from E. coli fully rescued the growth defect (Fig. 8A). This result suggests the presence of an additional factor in S. oneidensis that is required to activate HptA, which can then go on to activate ArcA. Experiments designed to identify these additional components of the ArcA activation pathway are currently underway. We also carried out the reciprocal experiment, providing arcB from E. coli to the hptA mutant strain and testing aerobic growth (Fig. 8B). The aerobic growth defect of the hptA mutant strain was partially reversed by both arcB from E. coli and hptA from S. oneidensis. This result is not surprising, given that we have established that ArcA from S. oneidensis can interact with ArcB from E. coli (Fig. 2). This result argues either that E. coli ArcB is completely replacing the endogenous S. oneidensis sensor system, or that the HPt domain of ArcB is interacting with the sensory system in a functional way. It is not clear why providing hptA in trans does not fully complement the growth defect of hptA mutants in S. oneidensis, but one possible explanation could be a toxic gene-dosage effect. This idea is supported by the fact that expressing hptA constitutively (using the lac promoter present on pBBR1MCS-3) in an hptA mutant background causes the strain to grow even more slowly than hptA with the empty vector (data not shown). Only its own promoter drives expression of hptA from pSO1327-2, the plasmid used for experiments in Fig. 8.

Figure 8.

Cross-complementation experiments using hptA from S. oneidensis and arcB from E. coli.
A. hptA from S. oneidensis cannot complement the aerobic growth defect of an E. coli arcB mutant. Strains are wild-type E. coli with pBBR1MCS-3 (▪), arcB with pBBR1MCS-3 (□), arcB with pSO1327-2 (containing hptA, ▾) and arcB with pARCB (•).
B. arcB from E. coli partially rescues the aerobic growth defect of a S. oneidensis hptA mutant. Strains are MR-1 with pBBR1MCS-3 (▪), hptA with pBBR1MCS-3 (▴), hptA with pARCB (○) and hptA with pSO1327-2 (•).
In both experiments, error bars represent standard deviation for triplicate cultures.

A conceptual model for ArcA activation is presented in Fig. 9. Because hptA mutants appear to only be partially defective in activating ArcA (Fig. 5, Table 2) and hptA could not complement an E. coli arcB mutant (Fig. 8), an additional factor (Y) must also be involved in activating ArcA directly. A second factor (X) is probably involved in activating HptA itself. It is also possible that the functions of both X and Y are encoded by the same protein. The domains used by ArcB in E. coli to sense and transduce signal to the HPt domain are absent from HptA (Kwon et al., 2000; Murray et al., 2001); however, several hypothetical proteins from S. oneidensis are predicted to contain canonical two-component transmitter domains, and are good candidates for roles in either HptA and/or ArcA activation. A specific mechanism for deactivation of HptA could also be utilized by S. oneidensis, which contains a gene (SO3082) predicted to encode a SixA (signal inhibitory factor-X) homologue. SixA has been shown to dephosphorylate the HPt domain of ArcB in E. coli (Ogino et al., 1998). Based on work with the HPt domain of ArcB in E. coli (Yaku et al., 1997; Matsushika and Mizuno, 1998), it is also possible that HptA is involved in signalling of other two-component response regulators such as OmpR, CheY or others.

Figure 9.

Model of ArcA activation in S. oneidensis. See text for discussion.

Comparison of Arc regulon between E. coli and S. oneidensis

Mutations in both arcA and arcB were initially identified in E. coli by fusing the promoter region of the sdh gene cluster (encoding succinate dehydrogenase) to lacZ (Iuchi and Lin, 1988; Iuchi et al., 1989). Mutants defective in either locus are unable to repress transcription of the sdh genes under anaerobic conditions in E. coli. A putative binding site for ArcA-P was not identified upstream of the S. oneidensis sdh gene cluster (SO1927–1929) using a binding energy cut-off of 7.34 (see Supplementary material), suggesting that this locus is not regulated by ArcA in S. oneidensis. We tested this by examining expression of the first gene of the sdh gene cluster, sdhC (SO1927), with quantitative reverse transcription polymerase chain reaction (Q-RT-PCR). Wild-type S. oneidensis represses sdhC transcription 4.5-fold anaerobically (relative expression of 1100 aerobically versus 245 anaerobically), while arcA mutants repress this gene 5.2-fold anaerobically (relative expression of 1300 aerobically versus 250 anaerobically). These results demonstrate that the sdh gene cluster is not regulated by the Arc system in S. oneidensis. Because expression of this gene was still repressed in an arcA mutant strain, a different mechanism for anaerobic regulation of the sdh gene cluster must be used by S. oneidensis. Microarray experiments suggest EtrA, the homologue of FNR in S. oneidnesis, negatively regulates the last gene in the sdh gene cluster, sdhB (Beliaev et al., 2002).

The only gene cluster identified in our bioinformatics search previously shown to be regulated by the Arc system in E. coli was cydAB, encoding cytochrome d oxidase. This enzyme is used under low oxygen concentrations because it has a higher affinity for oxygen than cytochrome o oxidase (Anraku and Gennis, 1987). ArcA is required for activating these genes in E. coli under microaerobic conditions (Iuchi et al., 1990; Cotter and Gunsalus, 1992), while under anaerobic conditions transcription of cydAB is repressed by FNR (Tseng et al., 1996; Cotter et al., 1997). Unlike the E. coli mechanism, ArcA itself functions to repress their transcription anaerobically in S. oneidensis (Table 2). This result suggests that either S. oneidensis utilizes cytochrome d oxidase more predominantly for aerobic respiration than E. coli or another system exists to maximize its expression during microaerobic conditions. The difference between how cydAB is regulated in S. oneidensis and E. coli further demonstrates the divergence of the Arc regulon, and possibly divergent strategies for aerobic respiration.

It is of interest to note that, at least in the case of regulation of sdh and cyd, the use of ArcA and FNR (EtrA in S. oneidensis) as repressors under anaerobic conditions has been reversed between E. coli and S. oneidensis. In E. coli, ArcA and FNR repress transcription of sdh and cyd gene clusters, respectively, whereas the opposite is true in S. oneidensis. This raises questions regarding the plasticity of regulatory proteins and how organisms can evolve and adapt them to regulate different suites of genes.

Concluding remarks

Shewanella oneidensis has a remarkably diverse array of anaerobic respiratory mechanisms at its disposal. The regulation of these processes is significantly different from in E. coli, where some of these systems have long been studied. The Arc system in S. oneidensis is similar to E. coli in that it uses a response regulator that is identical in function between the two organisms; however, the components involved in the activation of ArcA are different, involving the small protein HptA (and probably more components) that appears to have histidine phosphotransfer activity in place of a traditional E. coli ArcB sensor kinase. The results from a bioinformatic search for ArcA-P-regulated genes in S. oneidensis suggest that the Arc regulon in this organism is entirely different from the regulon in E. coli, with the exception of one gene cluster, cydAB. Although we have shown that the Arc system is required for respiration on DMSO, it does not appear to play a role in the regulation of genes involved in other forms of anaerobic respiration under the conditions tested. This implies that S. oneidensis uses novel mechanisms to regulate the expression of genes involved in mineral respiration, and that unbiased genetic approaches will aid in elucidating their components. This study highlights both parallels and significant differences between how anaerobic respiration is regulated in these two organisms, and provides two specific examples (cydAB and sdhCAB) where the mechanism of anaerobic repression of genes has been reversed between E. coli and S. oneidensis, yet achieves the same physiological affect.

Experimental procedures

Strains, plasmids, media and growth conditions

The strains and plasmids used in this study are listed in Table 1. S. oneidensis strain MR-1 and E. coli were routinely grown in Luria–Bertani (LB) medium (Miller, 1972) at 30°C (for S. oneidensis) and 37°C (for E. coli). Anaerobic growth was assayed in LB supplemented with 20 mM lactate and either 220 mM DMSO, 20 mM fumarate, or both 110 mM DMSO and 10 mM fumarate. To make culture tubes anaerobic, they were stoppered, sealed and flushed with nitrogen gas for several minutes with periodic shaking after the method of Balch and Wolfe (1976). Optical density of cultures was monitored at 600 nm. Plasmids were conjugated from E. coli WM3064 into S. oneidensis as described (Saltikov and Newman, 2003), and were maintained with the following antibiotic concentrations: kanamycin 50 µg ml−1, gentamicin 20 µg ml−1, tetracycline 15 µg ml−1 (E. coli) and 5 µg ml−1 (S. oneidensis). Toluidine blue sensitivity was tested by mixing cells with soft agar, overlaying an LB plate and spotting 5 µl of a 1% solution of Toluidine blue.

Computational search for ArcA binding sites

We identified the genomic positions of 18 of the 19 ArcA binding sites for E. coli studied by Liu and De Wulf (2004), and extended the known binding sites from 15 bp by 5 bp on either side. From these sites, we constructed a 25 bp position weight matrix (PWM) that exhibited a marked increase in preference for non-coding regions in E. coli over the 15 bp PWM used by Liu and DeWulf (C.T. Brown and C.G. Callan, unpublished). We used this final PWM to search the S. oneidensis MR-1 chromosome and megaplasmid sequence (NCBI Accession No. NC_004347.1 and NC_004349.1) for putative ArcA binding sites (see Supplementary material, TableS1). The extended sites, the PWM, and the results of the E. coli and S. oneidensis binding site searches are available online at The software (pyscangenes) used has been previously published and is available as open source at (Brown and Callan, 2004).

Molecular biological methods

Standard molecular techniques were used for this study (Maniatis et al., 1982). All plasmid inserts were fully sequenced to ensure no mutations had developed during the amplification and cloning process. The primers used in this study (Integrated DNA Technologies) are listed in Table 3. The broad range, conjugatable plasmid pBBR1MCS-3 (conferring tetracycline resistance) was used for all complementation studies (Kovach et al., 1995).

Table 3.  List of primers.
Complementation primer pairs
Q-RT-PCR primer pairs
Deletion primers

Generation of mutations in S. oneidensis

Targeted gene knockouts were performed as previously described (Saltikov and Newman, 2003), and verified by PCR and phenotypic analysis.


RNA was extracted from 0.5 ml of cells mid-log (OD600∼0.3–0.4) cultures that had been rapidly pelleted and frozen at −80°C using an RNAEasy kit (Invitrogen), according to manufacturer's instructions and included the optional DNase treatment step. Extracted RNA was used as template for a Taqman (ABI Biosciences) random-primed reverse-transcriptase reaction following the manufacturer's protocol, generating cDNA. The cDNA was used as template for quantitative PCR (Real Time 7300 PCR Machine, Applied Biosystems) using the Sybr Green detection system (Applied Biosystems). Samples were assayed in at least duplicate. Signal was standardized to both recA and envZ using the following equation: relative expression = 2 [(40 − CTsample) − (40 − CTstandard)] × 1000, where CT (cycle threshold) was determined automatically by the Real Time 7300 PCR software (Applied Biosystems) and 40 was the total number of cycles. Numbers reported are standardized to recA expression, but the same trends are found when standardized to envZ expression. Fold difference in expression was consistent between different standardization genes. Primers (Integrated DNA Technologies) used for Q-RT-PCR are presented in Table 3.


We thank Doug Lies, Curtis Callan and Arash Komeli for helpful discussions. J.A.G. was supported through a Texaco Postdoctoral Scholarship awarded through the Caltech Division of Geological and Planetary Sciences. C.T.B. was supported by National Institutes of Health Grant GM61005 to E.H. Davidson. We thank E.H. Davidson and R.A. Cameron of the Beckman Institute Center for Computational Regulatory Genomics at Caltech for access to their computational resources (supported by National Institutes of Health Grant RR15044). Grant support from the Office of Naval Research, the Luce Foundation and the Packard Foundation to D.K.N is gratefully acknowledged.

Supplementary material

The following material is available from

Table S1. Predicted ArcA binding sites in Shewanella oneidensis.