Shewanella species are widespread in diverse environments and have major environmental relevance (Fredrickson et al., 2008). They are commonly found in food, freshwater, hydrothermal vents, cold water, deep-sea sediments, and even fields contaminated with arsenic, chromium (VI), and explosives (Viamajala et al., 2004; Saltikov et al., 2005; Zhao et al., 2005; Gao et al., 2006). Shewanella utilize a number of electron acceptors for energy sources, such as nitrate, nitrite, thiosulfate, elemental sulfur, iron oxide and manganese oxide (Fredrickson et al., 2008). Among these electron acceptors, nitrate is considered to be the preferred acceptor for anaerobic respiration by microorganisms due to its relatively high standard redox potential, with a powerful ability to generate energy, and its abundance in aquatic environments (Richardson, 2000).

Briefly, the microbial nitrogen cycle involves biotic chain reactions comprising nitrogen fixation, dissimilatory nitrate reduction to ammonia (DNRA), denitrification, nitrification and anammox (anaerobic ammonium oxidation) (Figure 1). Denitrification has been intensely studied in recent decades and is largely assumed to be the major pathway for nitrate removal (Wallenstein et al., 2006; Kraft et al., 2011). In contrast to denitrification, knowledge on DNRA is mostly lacking.

To the best of our knowledge all Shewanella species are able to reduce nitrate. Most species reduce nitrate to nitrite and then to ammonia via the DNRA pathway, though a few species, such as S. denitrificans and S. amazonesis, carry out respiratory denitrification where nitrate is reduced to dinitrogen via nitrite, nitric oxide and nitrous oxide (Venkateswaran and Dollhopf, 1998; Brettar et al., 2002). Moreover, a notable number of Shewanellae are amenable to genetic manipulation, which makes detailed genetic studies on nitrate reducing pathways feasible. At the time of the writing of this review, 25 complete genomes of Shewanella were available at NCBI.

Many genes of the nitrate respiration system in Shewanella are identified as having multiple copies. The reduction of nitrate to nitrite is catalyzed by one of two distinct enzyme systems: the periplasmic nitrate reductase system (NAP) and the membrane-bound nitrate reductase system (NAR). Most Shewanella have duo-NAP (Nap-α and Nap-β) systems, which have been documented (Simpson et al., 2010b). Additionally, five strains (S. halifaxensis HAW-EB4, S. sediminis HAW-EB3, S. sp. MR-4, S sp. MR-7 and S. woodyi ATC51908) have both NAP and NAR systems, as listed in Table 1. Moreover, many genomes of Shewanella contain several nitrite reductase genes; these will be discussed below.

At the genetic level, horizontal gene transfer (HGT), gene duplication and loss of gene function are common strategies by microorganisms during evolution to adapt to environmental stress. Previous studies have suggested that Shewanella might acquire many new adaptive strategies through step-wise processes: the acquisition of mobile islands, selection of islands carrying ecologically important genes and loss of mobile and ecologically unimportant genes (Konstantinidis et al., 2009). Many genes of the nitrate respiration system in Shewanella, e.g., duo-NAP (Nap-α and Nap-β) systems, are identified as multi-copy (Simpson et al., 2010b), and the additional gene clusters that encode the same enzymes are likely to confer selective advantages to the host rather than be functionally redundant (reviewed and discussed below).

In this review, the current knowledge of the Shewanella nitrate respiration system will be reviewed with a specific focus on physiological functions, regulation mechanisms and evolutionary origin. This knowledge will certainly help in the better understanding of nitrate reduction pathways and the impact of HGT on the evolution of anaerobic respiration.

Denitrification is a respiratory process in which nitrite is first reduced to nitric oxide by nitrite reductase (NIR), nitric oxide reductase (NOR) then catalyzes nitric oxide reduction to nitrous oxide, and dinitrogen is synthesized from the reduction of nitrous oxide by nitrous oxide reductase (NOS). This process occurs in three separate reactions, and a proton-motive force is independently generated by each reaction. S. denitrificans possesses all sets of genes involved in denitrification in its genome (Genebank accession: NC_007954.1), and its capability of denitrification was confirmed by physiological analysis (Brettar et al., 2002). In addition to S. denitrificans, two other Shewanella species are potentially capable of denitrification. Although no growth on nitrate or nitrite was observed, S. loihica possesses the entire set of genes for denitrification (Gao et al., 2006). S. amazonesis was reported to denitrify nitrate to dinitrogen (Venkateswaran and Dollhopf, 1998); however, the genome of S. amazonesis lacks the gene encoding the NOS enzyme, which catalyzes the last step of denitrification, reducing nitrous oxide to dinitrogen. As the enzymatic mechanisms of dinitrogen synthesis in S. amazonesis have not yet been established, S. amazonesis is not a typical denitrifying bacterium. In addition, the nitrite reductase gene nrfA within DNRA pathway has been discovered in the genomes of S. amazonesis and S. loihica. Advanced studies on the function of the nrf gene in these two potential denitrifying bacteria will be interesting because to the best of our known, no strain reported so far is able to perform both denitrification and DNRA.

In DNRA, nitrite is reduced directly to ammonia by nitrite reductase (NrfA or NirB) without any intermediate; energy is also obtained by generating an electrochemical proton potential across the membrane. The pentaheme cytochrome NrfA is a membrane-associated respiratory enzyme that couples to membrane-associated formate-oxidizing enzymes via quinones to generate the membrane potential (Simon, 2002). NirB is an iron-sulfur protein that carries FAD and a siroheme cofactor (Jackson et al., 1981). The function of NirB was assigned primarily to the detoxification of nitrite produced by cytoplasmically oriented membrane-bound nitrate reductase (NarGHI) (Cole, 1996). However, the NirB enzyme may function as a nitrite-scavenger rather than a respiratory enzyme because energy is not conserved during nitrite reduction by NADH (Simon, 2002). For example, in E. coli, the nrfA operon is only induced under low-nitrate concentrations, whereas the NirB nitrite reductase is optimally synthesized only when nitrate or nitrite is in excess (Wang and Gunsalus, 2000). Most shewanellae carry out DNRA because their genomes have either the nrfA or nirB gene (or both) and lack genes for denitrification.

Notably, there are some indications that nitrous oxide is also generated as a byproduct of DNRA (Cruz-García et al., 2007; Vermeiren et al., 2009), though the mechanisms by which non-denitrifying bacteria generate nitrous oxide have not been revealed. One reasonable interpretation is that nitrous oxide is generated by nitrate-dependent iron reduction because the abiotic reduction of nitrite to nitrous oxide was observed in S. putrefaciens when both nitrate and iron (III) were available (Coby and Picardal, 2005).

Although the subsequent reactions involve different enzymes and intermediates, the first steps of DNRA and denitrification are identical and can be carried out by the same enzyme: periplasmic nitrate reductase (NAP) or membrane-bound nitrate reductase (NAR). All the available Shewanella genomes contain NAP, and only 5 strains out of 23 (S. halifaxensis HAW-EB4, S. sediminis HAW-EB3, S. sp. MR-4, S sp. MR-7 and S. woodyi ATC51908) possess both the NAP and NAR systems (Table 1). Although not well understood, NAR and NAP may also be of physiological significance in addition to nitrate reduction. In Shewanella species, the understanding of the NAP system is more comprehensive and therefore is selected for discussion in greater detail in this review.

Nap genes are widely distributed in gram-negative bacteria, and our understanding on the NAP system of Shewanella is mainly derived from research in model organisms such as E. coli. The current understanding of the subunits involved in the NAP system of the Shewanella genus is that the NAP system comprises seven compounds: NapA-H (except for NapF). NapA is the catalytic subunit where nitrate is reduced to nitrite, whereas NapB serves as the direct electron transporter, without a catalytic function. NapC is a quinol dehydrogenase protein, transporting electrons from both ubiquinol and menaquinol to terminal reductase NapAB (Simon and Kern, 2008). When NapC is missing, protein CymA performs the function of NapC, which will be discussed later. The electron transport model of NapABC and CymA is shown in Figure 3. NapGH couples electron transfer from the quinol pool via NapB to NapA in E. coli (Brondijk et al., 2002, 2004). However, it seems functionally redundant to retain NapGH in the few species of Shewanella that also possess NapC and/or CymA. It was suggested that NapGH might mediate the alternative pathway for nitrate reduction in addition to NapC or CymA (Gao et al., 2009), though the physiological significance of this alternative pathway is unclear.

One of the apparent differences between NAP-α and NAP-β is the arrangement of their gene clusters. Briefly, NAP-α (napEDABC) could be defined as a napC-dependent system, with NAP-β (napDAGHB) being napC independent. In S. oneidensis, which lacks NapC, the NapC homolog CymA mediates electrons from the quinone pool to both NapAB-β and NrfA (Cruz-García et al., 2007; Gao et al., 2009). In S. piezotolerans, which possesses both NAP-α and NAP-β, CymA is involved in both the reduction of nitrate to nitrite and nitrite ammonification (Chen et al., 2011). Furthermore, the different gene arrangements indicate the structural distinction between NAP-α and NAP-β proteins, which will be discussed below.

NAP could be divided into two types: monomeric and heterodimeric NAP. Both NAP-α and NAP-β are components of heterodimeric NAP, which is considered more evolved than monomeric NAP. In heterodimeric NAP, all NapA monomers form a heterodimeric complex with NapB. In contrast, monomeric NapA does not form a complex with the redox partner NapB; instead, a small tetraheme cytochrome C (~10 kDa), called NapM, may be the direct electron donor for NapA (Marietou et al., 2005). The monomeric class of NapA is considered to be the evolutionary link between monomeric NAS (assimilatory nitrate reductase) and heterodimeric NAP and thus a more ancient protein compared with heterodimeric NAP (Stolz and Basu, 2002; Jepson et al., 2006).

The existence of multiple nitrate reductases within Shewanella is not as functionally redundant as it appears. By investigating the NAP system of Shewanella and other nitrate reducers, we can find some clues to the functional importance of duo-NAP, especially when the host is under environmental stress.

The distinct physiological functions of the NAP system indicate that NAP-α and NAP-β may have multiple regulation mechanisms. Studies with S. oneidensis suggest that anaerobic respiration systems containing NAP are regulated by the electron transport regulator (EtrA) and cyclic AMP receptor protein (CRP), which control anaerobic respiratory gene expression individually (Maier and Myers, 2001; Beliaev et al., 2002; Saffarini et al., 2003).

EtrA is a global transcriptional regulator involved in the regulation of aerobic and anaerobic metabolism, and it shares 73.6% amino acid sequence similarity with the oxygen-sensing regulator Fnr in E. coli (Saffarini and Nealson, 1993). In mutant strains of S. oneidensis with etrA gene deletion, the expression of nap genes (nap-β) was significantly repressed (Cruz-García et al., 2011). CRP is a well-known regulator of carbon metabolism (Botsford and Harman, 1992), but CRP is essential for anaerobic growth in S. oneidensis, and a crp gene-deleted strain is defective in nitrate reduction (Saffarini et al., 2003; Gao et al., 2008). According to previous studies, CRP lacks an apparent redox-sensing domain, and it is still unclear which CRP is activated under anaerobic conditions (Fredrickson et al., 2008). Moreover, the nap, nar, nrf, and nir gene operons are further controlled by the NarQ-P two-component system. NarQ senses the presence of nitrate/nitrite, the response regulator NarP is then phosphorylated; finally, phospho-NarP activates the transcription of nap gene operons (Darwin and Stewart, 1995). Other environmental factors that may also influence duo-NAP system regulation include the carbon source, temperature, and oxygen concentration.

As discussed above, intensive studies have focused on the regulating and operating mechanisms of the duo-NAP system. However, little is known about the origin of this system; for example, how did the duo-NAP system of the Shewanella genus arise during evolution? Based on current understanding, the duo-NAP system is more likely to have resulted from HGT as opposed to gene duplication. Indeed, on the phylogenetic tree constructed with NapA protein sequences, NapA-β of Shewanella forms a relatively independent branch, while NapA-α are more closely related to the NapA of Vibrio species and Photobacterium profundum, indicating that the duo-NAP system is the result of HGT rather than gene duplication (Chen et al., 2011).

To understand why most Shewanella species have acquired NAP-α, we should take a close look at NapC and CymA. Both NapC and CymA are tetraheme c-type cytochromes belonging to the NapC/NirT family of proteins. The distribution of NapC/NirT proteins is relatively less widespread (they are found only in several subgroups of proteobacterial), which suggests that they may be evolutionarily more recent (Sharma et al., 2010). Members of this protein family play a role in the mediation of electrons in several anaerobic respiration pathways: NrfH and NirT for nitrite reduction, TorC for TMAO reduction, DorC for DMSO reduction, and FccC for methacrylate reduction (Simon and Kern, 2008). The proposed electron transport network of CymA and other specific transfers is illustrated in Figure 3. Sequences among the members of this family have a high degree of similarity, and they are suggested to have a common ancestor prior to recruitment into their specific anaerobic respiration system (Bergmann et al., 2005). This common ancestor is likely to function in the most ancient microbial metabolism, such as iron (III) reduction because iron reduction is widely agreed as one of the most ancient microbial metabolisms on the early Earth (Vargas et al., 1998). Alternatively, the ancestor may be a universal component of many electron transport pathways. If this is correct, CymA may have developed various functions at the time, and CymA may have also inherited multi-function capacities from the common ancestor. In addition, E. coli NapC has been reported to show ferric reductase activity (Gescher et al., 2008), suggesting that the specified electron transport proteins still share a commonality with their ancestor.

Based on multiple lines of evidence, we speculate that the origin of the duo-NAP system is the consequence of the co-evolution of microorganisms and environments. At the very beginning of its origin, Shewanella may have used metal as its sole electron acceptor. During evolution on Earth when extensive substrates became available, CymA may have developed its enzyme functions specifically to utilize alternative electron acceptors. By making use of the multifunctional protein CymA, the organism benefits by not increasing the burden of an expanding genome when selection pressure limits genome size (Huberts and Van Der Klei, 2010). However, for a protein that has multiple related functions, the additional roles might result in a low specificity of the active site (Jeffery, 1999). This model of respiration may have become less competitive as the environment changed. To cope with this deficiency, Shewanella, had to recruit specific electron-transport proteins (such as NapC, PsrC, SirC, and TorC, see Figure 3) by HGT. Nitrate is the most preferred electron acceptor in the natural environment for Shewanella because the cells can synthesize more energy for growth. Therefore, to primarily use nitrate, it is conceivable that Shewanella recruited the specific electron transport proteins NapB and NapC and routed electrons to NapA only.

Moreover, HGT was analyzed in several Shewanella baltica strains (Caro-Quintero et al., 2011), and the results indicate that the exchange rate of the genes, including nap, is rapid and that this gene-exchange contributes to successful adaptation to specific physicochemical conditions.

According to genetic and physiological studies, most Shewanella species are ammonifiers, whereas only a few are denitrifiers. Genes for nitrate reductase (NAP and NAR) and nitrite reductase (NRF and NIR) are identified as multi-copies in the genomes of most shewanellae. The duo-NAP system (NAP-α and NAP-β) was probably derived through horizontal gene transfer and environmental adaptation, with NAP-α having apparent advantages over NAP-β with regard to nitrate scavenging. Moreover, environmental factors, such as carbon source, temperature and oxygen levels, may influence the regulation of the duo-NAP system of Shewanella via different mechanisms. Based on available evidence, a model for the evolutionary course of the anaerobic respiration system in Shewanella is proposed here (Figure 4). At the early point of their emergence, Shewanella species may have used CymA as a common electron transfer that served multiple pathways of respiration; afterward, the specific transport protein NapC was acquired via gene transfer, which provided competitive advantages to the host under stressed conditions.

Recently, DNAR was suggested to play an important role in N metabolism (Kraft et al., 2011), which has been long ignored in previous investigations, as it was generally assumed that most of the nitrate in nature is respired via denitrification. It would be important and challenging to quantify the ecological contributions, diversity, and controlling factors of different nitrate respiration processes (DNAR and denitrification) in different environments. Shewanellae are believed to be important players in environmental N cycling, though additional detail and quantitative evaluation on their contribution are still lacking. Possessing multiple genes in nitrogen metabolism pathways is believed to confer Shewanella with environmental adaptation flexibility and advantages, though these assumptions need to be tested in the future. In addition, the mechanisms by which the duo NAP system benefits the host need to be elucidated. Understanding the regulation and evolution of the NAP systems in Shewanella would help in the understanding of the adaptation and evolution of this environmentally important bacterial genus and their application in bioremediation.