In quest of the nitrogen oxidizing prokaryotes of the early Earth


  • Siegfried E. Vlaeminck,

    Corresponding author
    1. Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Gent, Belgium
      E-mail; Tel. (+32) 9 264 59 76; Fax (+32) 9 264 62 48.
    Search for more papers by this author
  • Anthony G. Hay,

    1. Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Gent, Belgium
    2. Department of Microbiology, Cornell University, Ithaca, NY 14853, USA
    Search for more papers by this author
  • Loïs Maignien,

    1. Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Gent, Belgium
    Search for more papers by this author
  • Willy Verstraete

    1. Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Gent, Belgium
    Search for more papers by this author

E-mail; Tel. (+32) 9 264 59 76; Fax (+32) 9 264 62 48.


The introduction of nitrite and nitrate to the relatively reduced environment of the early Earth provided impetus for a tremendous diversification of microbial pathways. However, little is known about the first organisms to produce these valuable resources. In this review, the latest microbial discoveries are integrated in the evolution of the nitrogen cycle according to the great ‘NO-ON’ time debate, as we call it. This debate hypothesizes the first oxidation of nitrogen as abiotic and anoxic (‘NO’) versus biological and aerobic (‘ON’). Confronting ancient biogeochemical niches with extant prokaryotic phylogenetics, physiology and morphology, pointed out that the well-described ammonia and nitrite oxidizing Proteobacteria likely did not play a pioneering role in microbial nitrogen oxidation. Instead, we hypothesize ancestral and primordial roles of methanotrophic NC10 bacteria and ammonia oxidizing archaea, respectively, for early nitrite production, and of anammox performing Planctomycetes followed by Nitrospira for early nitrate production. Additional genomic and structural information on the prokaryotic protagonists but also on their phages, together with the continued search for novel key players and processes, should further elucidate nitrogen cycle evolution. Through the ramifications between the biogeochemical cycles, this will improve our understanding on the evolution of terrestrial and perhaps extraterrestrial life.


Although debates on the niche and nature of the origin of life are ongoing, it is commonly accepted that life originated in the Hadean ocean (around 4.5–4 Gyr) and that a last universal common ancestor (LUCA) preceded the diversification of prokaryotic life in the Archaean eon (around 4–2.5 Gyr) (Nisbet and Sleep, 2001). Most extant microbial conversions are considered to have emerged and been globally distributed by 3.5 Gyr ago. During the evolution and proliferation of life, it is clear that the introduction of the oxidized compounds nitrite (NO2-) and nitrate (NO3-) in a relatively reduced environment had tremendous impacts. Nitrite and nitrate are highly bioavailable due to their good solubility and negative charge over a range of pH values. These attributes contributed to nitrate's potential to serve as a possible nitrogen form for growth, a capacity that became widespread among bacteria, and later among eukaryotes (Richardson et al., 2001; Stolz and Basu, 2002). Furthermore, thanks to the high oxidation numbers of nitrogen in nitrite (+3) and nitrate (+5), a variety of new redox couples came into existence. This provided impetus for the development of various heterotrophic and autotrophic dissimilatory pathways that allow oxidized nitrogen species to serve as electron acceptors in the absence of free oxygen (Zumft, 1997; Stolz and Basu, 2002). However, little is known about the first organisms to produce these valuable oxidized nitrogen forms. Here we briefly review oxidative nitrite and nitrate producing processes, prior to discussing the emergence of these processes from an evolutionary perspective according to the two ruling schools of thought. In doing so, for every process, we endeavoured to identify likely ancestors, pioneers and followers, revealing key roles for some recently discovered prokaryotes.

Microbial nitrogen cycling

Figure 1 reviews the different steps and enzymes involved in the extant microbial nitrogen cycle, showing mainly a pivotal role for nitrite, involving 10 different enzyme systems, which have already been described for its direct production and consumption. Ammonificationand assimilation represent two redox neutral steps, whereas nitrogen fixation, denitratation, denitritation, nitrite reduction to ammonium and nitrite reduction by anammox are reductive steps, and nitritation, nitratation and oxidation of ammonium and nitrite by anammox are oxidizing steps. To clearly distinguish between the consecutive steps consuming and producing nitrite and nitrate, the umbrella terms ‘nitrification’, i.e. nitritation followed by nitratation, and ‘denitrification’, i.e. dissimilatory denitratation and subsequent canonical denitritation, have been avoided.

Figure 1.

The extant microbial nitrogen cycle with pathways related to catabolism (dashed arrows) and anabolism (full arrows). Oxidation states of the nitrogen compounds are indicated in pink, and intermediates are shown between brackets. In alphabetical order of the abbreviations, the key enzymes are ammonia monooxygenase (AMO), hydroxylamine oxidoreductase (HAO), hydrazine hydrolase (HH), hydrazine oxidoreductase (HZO), periplasmic nitrate reductase (NAP), membrane-bound nitrate reductase (NAR), cytoplasmic nitrate reductase (NAS), nitrogenase (NIF), siroheme nitrite reductases (NIR and NirB), Cu-containing nitrite reductase (NirK), cytochrome cd1 nitrite reductase (NirS), nitric oxide reductase (NOR), nitrous oxide reductase (NOS), pentaheme cytochrome c nitrite reductase (NrfA), nitrite oxidoreductase (NXR), particulate methane monooxygenase (pMMO) (Stolz and Basu, 2002; Bergmann et al., 2005; Strous et al., 2006; Francis et al., 2007; Semrau et al., 2008; Jetten et al., 2009; Ettwig et al., 2010; Walker et al., 2010). *For archaeal nitritation, this enzyme is unknown and proposed intermediates are NH2OH or HNO. **Ammonia oxidation remains to be demonstrated for methane oxidizing NC10 bacteria and Verrucomicrobia. +Enzyme unknown. ++NAP functions in tandem with NrfA, and NAR with NirB.

In the following paragraphs, the phylogeny and significance of extant nitrogen-oxidizing microorganisms is summarized. In contrast to most reductive nitrogen conversions (putative) nitrogen oxidation capacities are not widespread among prokaryotes (Fig. 2), indicating that these mostly autotrophic conversions are rather specialized and have not been subject to broad horizontal gene transfer.

Figure 2.

Phylogenetic 16S rRNA tree of life, displaying the taxonomic groups relevant to the evolution of oxidative nitrite and nitrate production. ‘The all-species living tree’ project release LTP 100 (September 2009) was used as a backbone for the most up-to-date topology for the bacterial and archaeal domain, as calculated with the RAxML algorithm from about 10950 carefully selected and corrected high-quality sequences (Yarza et al., 2008). The scale bar represents 10% divergence. From the SILVA database (Pruesse et al., 2007), aligned rRNA gene sequences were selected for 1961 Eukaryota (sequence quality, alignment and pintail at least 94/100) and for nitrogen cycle prokaryotes, using the selections of Hatzenpichler and colleagues (2008) and de la Torre and colleagues (2008) for ammonia oxidizing archaea (AOA), Hoffmann and colleagues (2009) and Mohamed and colleagues (2010) for anoxic ammonia-oxidizing bacteria (AnAOB), Hanson and Hanson (1996), Semrau and colleagues (2008) and Ettwig and colleagues (2010) for methane oxidizing bacteria (MOB), and Bock and Wagner (2006) for aerobic ammonia-oxidizing bacteria (AerAOB) and nitrite oxidizing bacteria (NOB). In ARB software (Ludwig et al., 2004), the selected species were added to the tree with the parsimony tool using domain-specific positional variability filter sets from the 102 SILVA release (February 2010). For clarity, taxonomic groups of less interest were not pruned but hidden instead, and the angles on branches display their branching off, visualizing how deeply groups of interest branch. The position of the last universal common ancestor (LUCA) is commonly thought to be between the deepest branching bacterial phylum (Thermotogae) and the node branching towards Archaea and Eukaryota (Baldauf et al., 1996; Woese, 2000).

Reductive pathways

Biological nitrogen fixation is a crucial step in nitrogen cycling, since it is the only biological process that makes nitrogen bioavailable from the abundantly present nitrogen gas in the atmosphere. Note that for extant marine ecosystems, biological nitrogen fixation (121 Tg N year−1) exceeds the abiotic fixation from electrical discharges (1.1 Tg N year−1) by two orders of magnitude (Galloway et al., 2004). The genes encoding nitrogenase, the key enzyme for nitrogen fixation, are very widespread among the bacterial and even archaeal domains (Martinez-Romero, 2006). According to phylogenetic analysis of these genes, LUCA might already have been able to fix nitrogen, but it is also possible that methanogenic archaea were the first to fix nitrogen and that this capacity was then horizontally transferred to bacteria (Raymond et al., 2004).

The ability to use nitrate as a nitrogen source (assimilatory denitratation) is not only common for marine heterotrophs (Allen et al., 2001), but also relatively widespread among many other bacteria (Richardson et al., 2001; Stolz and Basu, 2002). Dissimilatory nitrite reduction to ammonium is still understudied, but the available information indicates that this process is also relatively widespread among bacteria (Mohan et al., 2004; Smith et al., 2007). While nitrite can also be reduced by anammox, this catabolic process is discussed below with the oxidative pathways, since it also involves the oxidation of ammonium and nitrite.

Canonical denitritation reduces nitrite via nitric oxide and nitrous oxide to nitrogen gas (Fig. 1), a trait that occurs mostly in combination with dissimilatory denitratation and which is broadly distributed throughout the bacterial and archaeal domain (Zumft, 1997; Stolz and Basu, 2002). In contrast, a novel, methanotrophic denitritation route was recently described in bacteria from the NC10 candidate division, in which nitrite is reduced to nitric oxide which is subsequently acted upon by an unknown enzyme to nitrogen gas and molecular oxygen (Fig. 1), the latter being used intracellularly for aerobic methane oxidation (Ettwig et al., 2010).

Oxidative pathways

Aerobic oxidation of ammonium is the only known way to oxidatively produce nitrite (Fig. 1), which can be the substrate for many further oxidation or reduction reactions. Extant nitritation is performed by aerobic ammonia-oxidizing bacteria (AerAOB) belonging to the β- and γ-Proteobacteria (Koops and Pommerening-Roser, 2001), and by ammonia-oxidizing archaea (AOA) affiliated to the marine groups 1.1a and 1.1b of the Crenarchaeota (Francis et al., 2007) and recently proposed to represent a new phylum, i.e. the Thaumarchaeota (Brochier-Armanet et al., 2008a; Spang et al., 2010). Apart from their oxidative capacities, AerAOB have also been reported to encode for two of the three steps of canonical denitritation (Fig. 1), i.e. Cu-containing nitrite reductase (NirK) and nitric oxide reductase (NOR). As such, some AerAOB can combine oxidative and reductive capacities to convert ammonium to nitrogen gas under autotrophic oxygen-limited conditions (Kuai and Verstraete, 1998). In comparison, in the two annotated AOA genomes NirK was also encoded, whereas matches for NOR genes were either weak or absent (Hallam et al., 2006; Walker et al., 2010).

The ammonia monooxygenase (AMO) of AerAOB is closely related to the particulate methane monooxygenase (pMMO) of aerobic methane-oxidizing bacteria (MOB) belonging to the α- and γ-Proteobacteria, enabling MOB to oxidize ammonia as well, albeit at much lower rates (Hanson and Hanson, 1996; Nyerges and Stein, 2009). Furthermore, genes encoding hydroxylamine oxidoreductase (HAO), the second enzyme involved in nitritation, have also been retrieved in proteobacterial MOB (Bergmann et al., 2005), showing high similarity between the proteobacterial AerAOB and MOB nitritation pathways. It should be noted, however, that one γ-MOB species was found incapable of ammonia and hydroxylamine oxidation (Nyerges and Stein, 2009). Recently, two novel types of MOB were described. Members of the Verrucomicrobia can oxidize methane aerobically at low pH values (Dunfield et al., 2007; Pol et al., 2007; Semrau et al., 2008) and members of the NC10 candidate division can perform methanotrophy under anoxic conditions, coupled to denitritation (Ettwig et al., 2010). So far, however, it is not known if the pMMOs in these bacteria have a low enough enzyme specificity to attack ammonia as is the case for their proteobacterial counterparts. In contrast to the MOB, other heterotrophs performing nitritation were not included in this review since their biochemical reactions are not well known (Verstraete and Alexander, 1973; Castignetti and Hollocher, 1984; Robertson and Kuenen, 1990; Hooper et al., 1997). Overall, heterotrophs presumably do not gain energy from nitritation (Robertson and Kuenen, 1990; Klotz and Stein, 2008), and AerAOB and AOA are considered to be the protagonists in extant global nitrite production (Kowalchuk and Stephen, 2001; Francis et al., 2007).

For the generation of nitrate in the environment, nitratation is the best studied process and requires the presence of oxygen. Extant nitratation is catalysed by nitrite oxidoreductase (NXR) and is exclusively a bacterial trait performed by autotrophic nitrite-oxidizing bacteria (NOB) belonging to the α-, γ- and δ-Proteobacteria as well as to the Nitrospirae phylum (Koops and Pommerening-Roser, 2001). Until recently, the genomes of only three α-proteobacterial Nitrobacter species (Starkenburg et al., 2008) and one γ-proteobacterial Nitrococcus species (GenBank Accession No. NZ_AAOF00000000) were available. However, the lately published genome of a ‘Candidatus Nitrospira’ species significantly increased our understanding of nitratation diversity (Lücker et al., 2010). Surpisingly, the Nitrobacter and Nitrospira genomes also encode the denitritation enzyme NirK, suggesting that these NOB can perform uncomplete denitritation.

An often overlooked source of nitrate is anammox, a unique bacterial trait performed by autotrophic anoxic ammonia-oxidizing bacteria (AnAOB) belonging to the Planctomycetes phylum (Jetten et al., 2009). During this anoxic process ammonium oxidation is coupled to nitrite reduction, although 20% of the latter is also oxidized to nitrate anoxically (Strous et al., 1998), by a putative membrane-bound nitrate reductase (NAR) (Strous et al., 2006), with the released electrons being used for the reduction of carbon dioxide:


Despite the lack of pure AnAOB cultures, it is very unlikely that anammox nitrate production is an artefact, since the nitrate production stoichiometry was confirmed from highly enriched cultures and from physically purified cells, consisting for 97.6 and 99.6% out of AnAOB respectively (Strous et al., 1999; van der Star et al., 2008).

Since extant nitrogen gas production on a global scale is estimated to be 175–450 Tg N year−1 (Codispoti et al., 2001), for which anammox is likely 30–50% responsible (Devol, 2003; Dalsgaard et al., 2005), anammox nitrate production amounts to 7–29 Tg NO3--N yr−1 (Eq. 1). Since oxygen minimum zones occupy 8% of the extant oceans (Paulmier and Ruiz-Pino, 2009), anammox nitrate production can, especially on a local scale, represent a considerable nitrate input for the environment. In the Black Sea for instance, aerobic and anoxic ammonium oxidation rates are of the same order of magnitude, and AerAOB/AOA and AnAOB abundance shows the same spatial profile (Coolen et al., 2007; Lam et al., 2007). This suggests that a significant part of the nitrite produced during nitritation is consumed by anammox, rather than by nitratation. The inclusion of the anammox process and its role in nitrate production in global nitrogen models should therefore be considered.

The great ‘NO-ON’ time debate

A key event in the Archaean eon was the appearance of oxygenic photosynthesis, which evolved from anoxygenic photosynthesis and became the first source of free oxygen in a relatively reduced environment (Kasting, 1993). Although Cyanobacteria first produced oxygen possibly as early as 3.8 Gyr ago, local consumption by abiotic and biotic processes prevented the escape of oxygen from the ocean to the atmosphere until the start of the Proterozoic eon (around 2.5–0.5 Gyr), the so-called ‘Great Oxidation Event’ (Buick, 2008). Hence, over a significant time frame in the Archaean eon, the ocean water column had a large range of redox gradients, with oxidizing conditions in parts of the top layer, and reducing conditions elsewhere (Kasting, 1993). Similar conditions in extant oxygen-minimum zones like the Black Sea suggest that these may be a useful model for this early ocean, despite differences in the levels of atmospheric oxygen.

There has been much discussion over the last 35 years (Egami, 1973; Broda, 1975), about the timing of nitrogen oxidation relative to the appearance of free oxygen. To date, however, this great ‘NO-ON’ time debate as we call it, remains unresolved (Ducluzeau et al., 2009; Godfrey and Falkowski, 2009). Uncertainty about the composition of the primordial atmosphere lies at the base of this debate. The amount of solar radiation received by the early Earth from its younger Sun was much less than it is today (Kasting, 1993). Hence, the primordial ocean would have completely frozen over unless there was an insulating layer containing a sufficient concentration of greenhouse gases. Specifically, the amount of carbon dioxide present in the atmosphere is an item of special contention since, as discussed below, it would have been a major factor governing chemical nitrogen oxidation.

Overall, the basic evolutionary concept of the ‘NO-ON’ debate is that the unprecedented accumulation of a nitrogen compound would, by virtue of its mere presence, have provided an impetus to evolve enzymes, yielding products or reactions being beneficial to the host. In the nitrite/nitrate then oxygen (‘NO’) view, significant abiotic sources of nitric oxide (NO) led to the existence of nitrite and nitrate before the advent of oxygen, providing the first incentive for the development of anoxic processes that consume these oxidized nitrogen species (Egami, 1973). As shown in detail below, high atmospheric levels of carbon dioxide were expected in the ‘NO’ school of thought. In contrast, according to the oxygen then nitrite/nitrate (‘ON’) view, the first significant introduction of nitrite and nitrate in the environment occurred aerobically and biologically, and hence, nitrite and nitrate consuming processes could only have evolved after oxygenic photosynthesis (Broda, 1975). In this school, the atmosphere consisted of lower levels of carbon dioxide in the presence of methane.

The ‘NO’ school

Egami (1973) can be considered as the founder of the ‘NO’ school, advocating the presence of nitrite and nitrate on Earth before oxygen. Later models and lab-scale simulations confirmed the plausibility of this school in an atmosphere rich in carbon dioxide. Carbon dioxide was a likely candidate gas to insulate the early Earth, requiring an atmospheric partial pressure of at least 0.1 bar in the period from 4.5–3.5 Gyr (Kasting, 1993). If the carbon dioxide level exceeded 50%, electrical discharges would have led to the abiotic production of nitric oxide from carbon dioxide and nitrogen gas (Navarro-Gonzalez et al., 2001). High temperature sources for such discharges would have included volcanic gases, volcanically induced lightning, thunderstorm lightning and meteorite impacts. These combined events could have rendered an abiotic nitric oxide production of 3 Tg N year−1 in the most optimistic scenario (Ducluzeau et al., 2009). In addition to small amounts of nitrous oxide, further reaction of nitric oxide with water vapour in the presence of UV light would have yielded mainly nitrate and nitrite, in a ratio of approximately 4/1 (Mancinelli and McKay, 1988; Summers and Khare, 2007). In the ‘NO’ school of thought, the anoxic bioprocesses consuming nitrite or nitrate would therefore have emerged before the oxygen requiring processes nitritation and nitratation (Fig. 3; Egami, 1973; Klotz and Stein, 2008; Ducluzeau et al., 2009). It should be noted that nitrite, and to a minor extent nitrate, could also have been reduced abiotically with Fe2+ (Summers and Chang, 1993; Ottley et al., 1997), if the early Archaean ocean had a temperature above 25°C and a pH above 7.3. The high carbon dioxide partial pressure required to achieve abiotic nitric oxide formation would likely have rendered the early ocean relatively acidic however, rendering Fe2+ dependent reductions less likely. Furthermore, at high carbon dioxide levels, Fe2+ precipitates as a carbonate, requiring high temperatures (80°C) to obtain detectable nitrate reduction (Summers and Chang, 1993). Note that significant abiotic nitrate reduction to ammonium is also possible in the presence of green rust (Hansen et al., 1996), but the presence of such iron hydroxides prior to the presence of free oxygen is unclear.

Figure 3.

Representation of the ‘NO-ON’ time debate on a geological timeline showing a different appearance order of the nitrogen cycle processes consuming or producing nitrite or nitrate for both schools of thought. For simplicity, only the substrates and products for the nitrite and nitrate generating processes are depicted. Note that the cessation of abiotic nitric oxide (NO) formation in the ‘NO’ school could also have occurred after the onset of oxygenic photosynthesis, as discussed in the text (see The ‘NO’ school). LUCA, last universal common ancestor.

As time passed, atmospheric carbon dioxide levels would have gradually decreased due to lower production from decreasing meteorite impact rates and higher consumption from the weathering of the growing continents (Kasting and Siefert, 2001). With insufficient carbon dioxide levels, abiotic nitric oxide production would have eventually ceased, possibly limiting bioavailable nitrogen. This may have acted as a stimulus for the emergence of biological nitrogen fixation (Navarro-Gonzalez et al., 2001) in the event that LUCA was not capable of biological nitrogen fixation. However, given the large range of possible Archaean carbon dioxide levels (Kasting, 1993), it is not possible to infer whether abiotic nitric oxide production ceased before or after the onset of oxygenic photosynthesis.

The ‘ON’ school

Broda (1975) was the first to hypothesize the basics of the ‘ON’ school: nitrite and nitrate could only be formed biologically after the onset of biogenic oxygen production. In the ‘ON’ school of thought, carbon dioxide levels were not high enough to obtain significant nitric oxide formation. Models confirmed that at lower carbon dioxide levels, methane could have been an important greenhouse gas in the early atmosphere (Pavlov et al., 2000). Significant methane sources could have included hydrothermal vents and, after the emergence of LUCA, methanogenic archaea, which are generally considered to be among the earliest prokaryotes (Nisbet and Sleep, 2001).

Only after the emergence of Cyanobacteria and their introduction of free oxygen in the Archaean ocean (≤ 3.8 Gyr), could aerobic ammonia oxidation to nitrite emerge. This created the impetus for the development of processes oxidizing nitrite to nitrate (nitratation and anammox) and reducing nitrite, and later nitrate, to ammonium or nitrogen gas (denitritation, anammox and later denitratation) (Fig. 3; Broda, 1975; Falkowski, 1997; Godfrey and Falkowski, 2009). In contrast to the ‘NO’ school, the order in which these processes emerged would have had major implications for Archaean biological nitrogen availability. Indeed, in the ‘ON’ school dissolved nitrogen compounds could only be converted to gaseous compounds after the onset of biogenic oxygen production (Fig. 3). As a result, the first closure of the nitrogen cycle occurred relatively late, thereby limiting the nitrogen availability in the Archaean ocean. Note that the formation of nitrogen gas was not necessarily a complete loss to all microbes, as some use the gas to alter their buoyant density, thereby allowing aggregated cells to travel through the water column (Vlaeminck et al., 2007), and profit from the access to additional resources thanks to the induced advective flow (De Schryver et al., 2008).

Evolution of biological nitrogen oxidation

Emergence of oxidative biological nitrite production

Candidate processes for early microbial ammonia oxidation to nitrite are proteobacterial nitritation (AerAOB), archaeal nitritation (AOA) and aspecific methane oxidation under anoxic (NC10 MOB) or aerobic (other MOB) conditions. Except for the NC10 MOB, the organisms performing nitritation or methanotrophy need free oxygen from their environment, and so they could only emerge after the biogenic production of oxygen. Consistent with this, the proteobacterial AerAOB and MOB as well as the verrucomicrobial MOB branch off later from the tree of life than the Cyanobacteria (Fig. 2). Furthermore, free oxygen was also required to liberate a key element for nitritation. Indeed, the catabolism of AerAOB, AOA and MOB requires copper (Hanson and Hanson, 1996; Arp and Stein, 2003; Walker et al., 2010), whose release into the early ocean is only thought to have occurred once oxygen was present (Lewis and Landing, 1992; Klotz and Stein, 2008).

For proteobacterial nitritation, the relatedness of AerAOB with β- and γ-proteobacterial anoxygenic phototrophs at the 16S rRNA level and the similarity of the cytoplasmatic membrane arrangements indicate that AerAOB might have inherited their cell plan from photosynthesizing Proteobacteria (Teske et al., 1994). More recent bioinformatic analyses of proteobacterial AerAOB and MOB has suggested that the genes encoding nitritation might have been transferred from the γ-proteobacterial MOB to the AerAOB of the same subphylum, and later to the β-proteobacterial AerAOB (Klotz et al., 2008). This implies that the well-studied β-proteobacterial AerAOB are likely relatively young, consistent with their 16S position on the tree of life (Fig. 2).

The recently discovered MOB in the Verrucomicrobia and the NC10 candidate division might bring clarity on the origin of pMMO. These two phylogenetic groups possibly harbour the oldest MOB, since they branch off early from the 16S tree of life (Fig. 2). Furthermore, the Verrucomicrobia contain purportedly ancient structural features, as discussed more in detail below (see Emergence of biological nitrate production). The ability of NC10 MOB to produce their own oxygen from nitric oxide under anoxic conditions potentially has important evolutionary consequences. In the ‘ON’ school, this oxygenic and methanotrophic denitritation did not play a key role, since it could have only evolved after oxygenic photosynthesis and the subsequent first biological nitrite production. In strong contrast, however, the discovery of Ettwig and colleagues (2010) changes the traditional view of the ‘NO’ school, rendering the aerobic oxidation of methane, and possibly of ammonium and other reduced compounds, possible before the onset of oxygenic photosynthesis. It is hoped that future experiments with these newly discovered MOB confirm the pMMO aspecificity for ammonia oxidation and lend credence to an ancestral role of the NC10 pMMO in microbial nitrite production.

Concerning archaeal nitritation, phylogenetic/genomic, enzymatic and physiological arguments are consistent with a pioneering role for AOA in microbial nitrite production. First, the 16S rRNA tree of life indicates that the AOA branch off relatively early, although their relative emergence with respect to the Cyanobacteria cannot be resolved from the tree, given the range of the possible LUCA positions (Fig. 2). Molecular clock analyses of the sequences of 16S rRNA, 23S rRNA and proteins are consistent with the emergence of the Crenarchaeota prior to the advent of oxygenic photosynthesis by as little as 0.1–0.4 Gyr (Blank, 2009) or more than 1.0 Gyr (Battistuzzi et al., 2004). The distinct phylogenetic and genomic branching of AOA early in the archaeal domain even corroborated the suggestion that AOA represent a novel phylum, i.e. the Thaumarchaeota (Brochier-Armanet et al., 2008a; Spang et al., 2010). Second, the ancestral status of the AOA is further supported by the shared presence of a DNA topoisomerase and an unsplit RNA polymerase subunit in AOA and Eukaryota, in contrast to other Archaea and Bacteria (Kwapisz et al., 2008; Brochier-Armanet et al., 2008b). Also, the archaeal AMO genes are quite distantly related from the bacterial AMO and pMMO genes (Ettwig et al., 2010). Third, if extant physiology is any indication of historical physiology, conditions conducive to AOA growth likely arose shortly after the onset of oxygen production and only gradually changed over the next millions of years to conditions which favour AerAOB. Specifically, AOA have a hypothesized preference for lower dissolved oxygen concentrations and thus would have been better suited to a microaerobic ocean (Francis et al., 2007; Erguder et al., 2009). Furthermore, prior to the availability of free oxygen, phosphate is considered to have been a limiting nutrient (Bjerrum and Canfield, 2002). Since AOA are thought to have low phosphate requirements (Erguder et al., 2009), they would have been better suited to those conditions than AerAOB which have relatively high phosphate needs (e.g. Purchase, 1974; Hue and Adams, 1984; Nordeidet et al., 1994; Zhang et al., 2009). The aforementioned points clearly show that the AOA should not be overlooked as potential early ammonia oxidizers. In contrast to proteobacterial AerAOB and MOB, no HAO was retrieved in AOA (Hallam et al., 2006; Walker et al., 2010). The latter study recently revealed major differences in AerAOB and AOA catabolism, including the possible involvement of nitroxyl (HNO) as intermediate in archaeal nitritation, instead of hydroxylamine (NH2OH). Future elucidation of the other involved enzyme will likely shed more light on the evolutionary origin of the genetic module encoding archaeal nitritation, as will the physiological characterization of more AOA isolates.

Emergence of biological nitrate production

Based solely on the phylogenetic distances of their specialized hosts from LUCA, it is difficult to infer whether anammox or nitratation evolved first (Fig. 2). Both sides of the ‘NO-ON’ time debate agree that nitratation could only emerge after oxygenic photosynthesis had provided the required oxygen. From the 16S tree of life, Nitrospira appears to be the most ancient NOB group whereas Nitrobacter is likely the youngest group to perform nitratation (Fig. 2). Similar to the proteobacterial AerAOB, the close 16S rRNA relationship between the α- and γ-proteobacterial NOB, Nitrobacter and Nitrococcus, and anoxygenic phototrophic Proteobacteria, together with the shared presence of intracytoplasmatic membrane arrangements, is suggestive of shared inheritance with anoxygenic photosynthesizing cells (Teske et al., 1994). In contrast, the genera Nitrospina and Nitrospira are at the 16S level not closely related to phototrophs and, consistent with this observation, their cells do not harbour peculiar membrane arrangements (Bock and Wagner, 2006). Nitrobacter is the only NOB genus without obligately halophilic species (Bock and Wagner, 2006), but since the colonization of land probably preceded the emergence of oxygenic photosynthesis (Battistuzzi et al., 2004), no conclusions can be drawn from that.

Until recently, the α-proteobacterial Nitrobacter was the only NOB group for which the genes encoding nitratation had been fully described and annotated. Early reports already showed the reversibility of the Nitrobacter NXR reaction (Sundermeyer-Klinger et al., 1984). The relatedness of Nitrobacter nitratation and denitratation was confirmed at the molecular level by a remarkable similarity between NXR in Nitrobacter and the NAR-type enzymes involved in dissimilatory denitratation (Kirstein and Bock, 1993; Starkenburg et al., 2008). Furthermore, the similarity between the genes encoding NirK in Nitrobacter and the AerAOB Nitrosomonas led to the hypothesis that these nitrogen metabolism genes were acquired through horizontal gene transfer between species sharing the same ecological niche (Starkenburg et al., 2008). A late emergence of Nitrobacter is further indicated by the relatively young 16S positions of Nitrobacter and Nitrosomonas (Fig. 2), and the likely emergence of Nitrosomonas as the youngest AerAOB group (see Emergence of oxidative biological nitrite production). Furthermore, if extant and historical physiology are related to some extent, Nitrobacter likely emerged after Nitrosomonas when the ambient oxygen levels were higher, since Nitrobacter shows a lower affinity for oxygen than Nitrosomonas (Laanbroek and Gerards, 1993).

The recent release of genomic data on a Nitrospira species allowed an important step forward in our understanding of NOB catabolic and anabolic diversity (Lücker et al., 2010). NXR of Nitrospira was found to be only distantly related to those of Nitrobacter and Nitrococcus and to the denitratation enzyme NAR. Furthermore, the location and electron transport chain of Nitrospira NXR and the Nitrospira carbon fixation pathway differed fundamentally from the other NOB. Overall, these arguments suggest that nitratation independently evolved multiple times (Lücker et al., 2010). Interestingly and in contrast to Nitrobacter, Nitrospira was detected as dominant NOB under oxygen-limited conditions (Schramm et al., 1999; 2000; Gieseke et al., 2003; Vlaeminck et al., 2010), and even prefers low oxygen levels (Park and Noguera, 2008; Off et al., 2010). Furthermore, the Nitrospira carbon fixation pathway is more common to anaerobes, and Nitrospira does not share protection mechanisms to reactive oxygen species with most aerobes (Lücker et al., 2010). Hence, following the advent of free oxygen production, the anaerobic/anoxic ocean with microaerobic zones would have been conducive to Nitrospira, and make a likely candidate as pioneering NOB, in agreement with its early position on the tree of life (Fig. 2).

Complete aerobic ammonia oxidation to nitrate (comammox) has never been found in one organism, although this process is thought to have energetic advantages in certain environments (Costa et al., 2006). The authors considered that the evolution of sequential nitritation and nitratation was favoured above comammox, given the more efficient division of the metabolic labour in the former. Alternatively, the absence of comammox could also be related to the biogeochemical history of the Earth. Indeed, if abiotic nitrite production in the ‘NO’ school ceased only after the advent of oxygenic photosynthesis, nitrite might have been more abundant than ammonium, selecting for the emergence of nitratation prior to nitritation. Once both processes were diversified and in place, there would have been no impetus for comammox to evolve. The plausibility of the such sequential emergence of denitritation, nitratation and nitritation might be addressed with a revisited NirK comparison, combining analyses on AerAOB and the NOB Nitrobacter (Cantera and Stein, 2007) with those on denitriting bacteria (Jones et al., 2008), and extending with newly available data from the NOB Nitrococcus (GenBank Accession No. NZ_AAOF00000000), the NOB Nitrospira (Lücker et al., 2010) and AOA (Hallam et al., 2006; Walker et al., 2010).

Nitrospira was shown to be a likely candidate to pioneer aerobic nitrate formation, and may even have received its enzymatic repertoire for nitrite oxidation from the AnAOB through horizontal gene transfer. Indeed, the high similarity of a small set of proteins encoding nitrite oxidation (NXR and NAR) and putative electron transport/respiration components clearly indicates horizontal gene transfer (Lücker et al., 2010). Since the genes encoding this set are tightly grouped in the AnAOB and widely spread in Nitrospira, it is likely that Nitrospira incorporated the organized gene set from the AnAOB in a rather fragmentary way in its genome. This suggests that anammox nitrate production predated nitratation and that AnAOB were the first nitrate producers on Earth. Both 16S phylogeny, metabolism and cell structure indeed suggest an ancient origin for the Planctomycetes phylum, in which the AnAOB represent an early branch (Fig. 2). Planctomycetes along with the closely related Verrucomicrobia, Chlamydiae, Poribacteria and two other phyla, consitute the so-called PVC superphylum in the bacterial domain (Wagner and Horn, 2006). In the Planctomycetes, three apparently ancient features have been described. First, Planctomycetes along with Chlamydiae, are the only bacteria with proteins instead of peptidoglycan as a major cell wall constituent (Lindsay et al., 2001). Second, the genes of C1 transfer reactions, which are the basis of the carbon metabolism of the putatively ancient methanogenic Euryarchaeota and of all known MOB (Pol et al., 2007; Ettwig et al., 2010), have been found in distantly related Planctomycetes (Chistoserdova et al., 2004), revealing a potentially ancestral role of the latter. Third, Planctomycetes and Poribacteria cells show a special cell plan with intracellular membranes dividing the cytoplasm into different compartments (Fuerst, 2005), a feature that was recently also discovered in the sister phylum of the Verrucomicrobia (Lee et al., 2009). Furthermore, Santarella-Mellwig and colleagues (2010) have recently shown a link between the endomembrane systems of the PVC superphylum and that of eukaryotic cells. Future availability of more genomes from PVC bacteria should provide insight into the unique features of these organisms and show if these traits share a common phylogenetic origin (LUCA), if they emerged independently in Bacteria and Eukaryota, or if they were transferred horizontally.

In an attempt to reveal the ancestry of anammox, Klotz and colleagues (2008) performed comparative sequence analysis on a key cytochrome c protein family related to HAO, hydrazine oxidoreductase (HZO) and pentaheme cytochrome c nitrite reductase (NrfA). Interpreted according to the ‘NO’ school, this study hypothesized that both anammox and bacterial nitritation were derived from dissimilatory nitrite reduction to ammonium. Note that anammox can consume nitrite while being physically separated from its production source (van der Star et al., 2007), rendering anammox plausible in the absence of a neighbouring nitrite producer, as can be expected in the ‘NO’ school. Regardless of the outcome of the great ‘NO-ON’ time debate, anammox is a likely candidate process for the first ever nitrate production. However, in the ‘NO’ school of thought, abiotic nitrate production would still have been the major nitrate source, as shown in the following best-case estimation for anammox nitrate production. Assuming that a part of the abiotically produced nitric oxide was oxidized to 20% nitrite and 80% nitrate (see The ‘NO’ school) and that AnAOB had access to all this nitrite with minor competition from abiotic reduction or biological denitritation, the anammox stoichiometry (Eq. 1) suggests that 5% (= 4/84) of the overall nitrate production could have been biological. Additionally, if the nitrite produced from denitratation was also consumed by anammox, up to 24% (= 20/84) of the total nitrate production could have been due to anammox.

Phages and prophages

Besides genomic and structural information relevant to the metabolism of nitrogen cycle prokaryotes, data on their parasites can also shed light on the evolutionary establishment of microbial nitrogen oxidation. Phages are obligatory viral parasites to prokaryotes and outnumber prokaryotes in aqueous ecosystems, where they infect significant fractions of the microbial community (Weinbauer, 2004). Furthermore, the genome of most culturable prokaryotes contains complete or defective prophages, i.e. integrated phage genes (Ackermann, 2007). The evolutionary relevance of phages derives from two findings. First, although most prokaryotes have more than one specific phage, the host range of a phage is often quite narrow (Weinbauer, 2004). Second, recent findings show that all viruses share a common ancestor which likely emerged before LUCA, preceding the diversification of cellular life (Bamford et al., 2005).

For the microbial groups relevant to this review (Fig. 2), direct observation of infected cells was limited to type II MOB and β-proteobacterial AerAOB (Ackermann, 2007; Vlaeminck et al., 2010). For other groups, indirect phage infection has been shown by prophages identified in the genomes of the type X MOB Methylococcus, the AerAOB Nitrosococcus, the NOB Nitrobacter and the AOA Nitrosopumilus (Ward et al., 2004; Klotz et al., 2006; Starkenburg et al., 2008; Walker et al., 2010). Despite the widespread occurrence of phages over the prokaryotic phyla (Ackermann, 2007), to date, their detection in the nitrogen cycle is very limited because direct observation is understudied and few full prokaryotic genomes are available. It is therefore expected that future genomic and structural characterization of (pro)phages will shed new light on the evolution of microbial nitrogen oxidation.


Nature has developed multiple mechanisms to generate nitrite and nitrate. It is likely that their accumulation on the early Earth provided emerging life forms with additional redox couples that allowed for the exploitation of new niches and helped to spur diversification. Specifically, the presence of these compounds permitted the development of various extant dissimilatory and assimilatory pathways, which play a critical role in today's nitrogen cycle. This review has, for the first time, discussed possible ancestral and primordial roles of NC10 MOB and AOA, respectively, for early nitrite production, and of AnAOB followed by the NOB Nitrospira for early nitrate production. Although nitrite and nitrate obviously were and are major multifacetted driving forces in biological evolution, the true picture of nitrogen cycle evolution remains murky. The availability of additional genomic and structural information on the prokaryotic protagonists and on their phages, as well as the continued search for novel nitrogen cycling processes will lead to greater clarity and provide insights that may resolve the ongoing ‘NO-ON’ time debate. The ‘ON’ school would be particularly supported by demonstrating that denitratation was the last process to emerge, or that nitritation evolved later than either anammox, nitrite reduction to ammonium or denitritation (Fig. 3). In contrast, more definitive evidence for the reverse would be a strong argument in favour of the ‘ON’ school.


S.E.V. was supported as a postdoctoral fellow from the Research Foundation Flanders (FWO-Vlaanderen), L.M. was recipient of a PhD grant from the Institute for the Promotion of Innovation by Science and Technology in Flanders (IWT-Vlaanderen, number SB-53575) and A.G.H was funded as a visiting foreign researcher by the Special Research Fund from Ghent University (BOF-UGent, number VBO017, 01T01709 B/10902/02). The authors gratefully thank Haydée De Clippeleir, Nico Boon and Peter De Schryver for the inspiring scientific discussions.