Nitrifier genomics and evolution of the nitrogen cycle


  • Martin G. Klotz,

    1. Evolutionary and Genomic Microbiology Laboratory, Department of Biology, University of Louisville, Louisville, KY, USA
    2. Department of Microbiology and Immunology, University of Louisville, Louisville, KY, USA
    3. Center for Genetics and Molecular Medicine, University of Louisville, Louisville, KY, USA
    Search for more papers by this author
  • Lisa Y. Stein

    1. Department of Environmental Sciences, University of California, Riverside, CA, USA
    Search for more papers by this author

  • Editor: Rustam Aminov

Correspondence: Martin G. Klotz, Department of Biology, University of Louisville, 139 Life Science Building, Louisville, KY 40292, USA. Tel.: +1 502.852.7779; fax: +1 502.852.0725; e-mail:


Advances in technology have tremendously increased high throughput whole genome-sequencing efforts, many of which have included prokaryotes that facilitate processes in the extant nitrogen cycle. Molecular genetic and evolutionary analyses of these genomes paired with advances in postgenomics, biochemical and physiological experimentation have enabled scientists to reevaluate existing geochemical and oceanographic data for improved characterization of the extant nitrogen cycle as well as its evolution since the primordial era of planet Earth. Based on the literature and extensive new data relevant to aerobic and anaerobic ammonia oxidation (ANAMMOX), the natural history of the nitrogen-cycle has been redrawn with emphasis on the early roles of incomplete denitrification and ammonification as driving forces for emergence of ANAMMOX as the foundation for a complete nitrogen cycle, and concluding with emergence of nitrification in the oxic era.


The flux of nitrogen through the global biogeochemical nitrogen cycle has undergone dramatic alterations in the past few decades (Galloway & Cowling, 2002). Over half of the fixed nitrogen that annually enters terrestrial ecosystems now has its origins in anthropogenic processes including production of ammonia-based fertilizers via the Haber–Bosch process, cultivation of nitrogen-fixing crops, and the combustion of fossil fuels leading to the release of nitrogen oxides (Nevison & Holland, 1997; Galloway & Cowling, 2002). The majority of processes in the extant global biogeochemical nitrogen cycle are facilitated by bacteria including: (1) N2 fixation, (2) nitrification and (3) denitrification (Fig. 1). A fourth process, anaerobic ammonia oxidation (ANAMMOX), is a more recently described bacterial contribution to the nitrogen cycle (Dalsgaard et al., 2005; Jetten et al., 2005).

Figure 1.

 Key steps in the evolution of the nitrogen cycle. (a) Very early anaerobic nitrogen metabolism, in which NO2 and NO3 are produced by combustion and NH3 is mainly generated from N2 at hydrothermal vent sites: Incomplete respiratory denitrification (MGD-NarGH/NapAB; cd1-NirS) and respiratory nitrite ammonification [pentaheme-NrfAH (NrfABCD)]. See text for references. (b) Early anaerobic nitrogen metabolism: To avoid NH2OH poisoning, invention of hydroxylamine dehydrogenation (HAO/HZO) and reduction (prismane protein), which replenished nitrite and ammonia pools. Resulting emergence of functional HURM and hydrazine hydrolase constitute ANNAMOX process and thus complete recycling of fixed nitrogen. (c) Late anaerobic and early aerobic nitrogen metabolism: Emergence of assimilatory ammonification (siroheme cytochrome c NIR) and N2 fixation to satisfy increased ammonia demand. Emergence of heme-copper and copper redox centers and subsequent diversification of anaerobic and aerobic respiration (Cu-NirK, HCOs, NOR), complete denitrification (Cu–NOS) as well as aerobic methane- and ammonia oxidation (pMMO/AMO) and thus closure of the biotic nitrogen cycle. Elevated ammonia input from anthropogenic sources leads to elevated nitrifier denitrification in today's nitrogen cycle.

The process of N2 fixation is carried out by a variety of bacteria that use nitrogenase, in concert with other enzymes and cofactors, to facilitate the highly endergonic breakage of the triple bond in dinitrogen to yield ammonia (Postgate, 1970). Bacterial nitrification proceeds by the sequential oxidation of ammonia to nitrite predominantly by ammonia-oxidizing bacteria (AOB) and of nitrite to nitrate by nitrite-oxidizing bacteria (NOB) (Prosser, 1989). However, there is recent evidence that a number of Crenarchaea, abundant in soils, estuarine and marine environments, are also capable of nitrification (Könneke et al., 2005; Francis et al., 2007) by a different mechanism than that of AOB (Arp et al., 2007). Nitrification, whether facilitated by bacteria or archaea can proceed only in oxic environments or in anoxic environments by select species given an external supply of NO2 (N2O4) (Schmidt et al., 2001). Nitrate and nitrite are substrates for denitrification, the process that recycles fixed nitrogen back to gaseous dinitrogen. Denitrification is facilitated by either anaerobic respiration of bacteria in anoxic environments, i.e. ‘complete’ or ‘canonical’ denitrification (Zumft, 1997; Brandes et al., 2007), or by reductive detoxification of nitrite to nitrous oxide in aerobic environments, i.e. ‘incomplete’ or ‘nitrifier’ denitrification (Lipschultz et al., 1981). ANAMMOX, performed by anaerobic ammonia-oxidizing bacteria (ANAOB), couples the oxidation of ammonia to the reduction of nitrite to produce dinitrogen in anoxic ecosystems. The release of ammonia during degradation of organic matter, as well as assimilatory and respiratory reduction of nitrate (or nitrite) to ammonia, i.e. ammonification, are also biotic contributions to the nitrogen cycle performed by bacteria, fungi and plants (Fig. 1c). Ammonification and nitrification can be regarded as short circuits that bypass the vast dinitrogen reservoir connected in the present cycle by denitrification and nitrogen fixation (Fig. 1c). In addition to these biotic processes, the nitrogen cycle is amended by abiotic processes including ammonia production from N2 (Brandes et al., 1998; Wachtershauser, 2007 and references therein) at hydrothermal vents, the oxidation of N2 to nitrite and nitrate by combustion (Yung & McElroy, 1979; Mancinelli & McKay, 1988; Kasting, 1993; Navarro-González et al., 2001) and mineralization (McLain & Martens, 2005).

Enzymes of ammonia oxidation

Knowledge of the nitrification process and its best-studied facilitators, the AOB and NOB, goes back more than 100 years to the work of Winogradsky (1892). In AOB catabolism, ammonia is first aerobically oxidized to hydroxylamine by ammonia monooxygenase (AMO) followed by the dehydrogenation of hydroxylamine to nitrite by hydroxylamine oxidoreductase (HAO), which is proposed to relay the four extracted electrons to the ubiquinone pool via two interacting cytochromes, c554 and cM552 (Fig. 2; Hooper et al., 2005). Owing to its soluble nature, HAO is the best-studied functional component in the nitrification process followed by cytochrome c554 (Hooper et al., 2005 and references therein). Both proteins have been crystallized and their structures resolved (Igarashi et al., 1997; Iverson et al., 2001). In contrast, AMO, a multimeric transmembrane copper-enzyme, has yet to be functionally isolated, crystallized and its structure solved. The process of AMO reduction by electrons obtained from ubiquinone, which it needs to oxidize ammonia, remains elusive and the study of AMO is by far less progressed compared with the other functional players involved in ammonia oxidation by AOB (Arp et al., 2007).

Whereas the sequence availability of individual genes encoding AMO and HAO led to a surge in information about AOB distribution and abundance through design and use of molecular probes (Purkhold et al., 2000; Kowalchuk & Stephen, 2001 and references therein), aspects of the molecular biology and biochemistry of these organisms aside from carbon assimilation and use of ammonia as an energy source have received little attention. In particular, very little to nothing is known about the regulation of gene expression required for nitrifier denitrification by AOB (Fig. 2). This is surprising given that nitrifier denitrification produces reactive nitrogen species (RNS), competes for electrons with primary bioenergetic processes such as reverse electron flow and the electron transport chain that produces proton motive force (PMF), and may reduce the availability of nitrite, the catabolic substrate to NOB (Fig. 2).

Figure 2.

 Flow of energy and reductant in the nitrification process through pertinent catabolic modules in ammonia-oxidizing (AOB) and nitrite-oxidizing (NOB) bacteria. Solid lines show experimentally verified reactions, dotted lines with question marks indicate lack of experimental verification of reactions assumed to occur (Hooper et al., 2005; Arp et al., 2007). REDOX reactions via dashed lines indicate that multiple nonidentical copies of c552 are required to interact with a diverse set of reaction partners. The circuit for reverse electron flow was omitted. Colored backgrounds were used to highlight the cyclic electron flow module (see also Fig. 3) as well as the hydroxylamine/hydrazine-ubiquinone-redox-module, HURM. Blue-colored boxes indicate copper-dependent enzyme complexes. (c)aa3, cytochrome (c)aa3; bc1, cytochrome bc1 (complex III); NirK, Cu-dependent nitrate reductase; c′-β, cytochrome c′-β; c550, cytochrome c550; c552, cytochrome c552; cM552, cytochrome cM552; c554, cytochrome c554; NXR, nitrite oxidoreductase; P460, cytochrome P460; PMF, proton-motive force; Q/QH2, ubiquinone-ubiquinol pool; sNOR, cNOR, ccNOR, nitric oxide reductase with differing electron acceptor mechanisms. See text for further details.

ANAOB, on the other hand, were discovered less than two decades ago (Dalsgaard et al., 2005; Jetten et al., 2005) such that the molecular mechanisms and genetic inventory involved in ANAMMOX (Fig. 3) are just beginning to take shape (Strous et al., 2006). Although AOB and ANAOB both oxidize ammonia as their primary source for energy and reductant, they interconnect the pools of fixed and gaseous nitrogen in different metabolic contexts. Whereas aerobic AOB and NOB oxidize ammonia to nitrate collaboratively and independently of denitrification (Fig. 2), ANAOB combine nitrification (ammonia oxidation) and denitrification (nitrite reduction, dinitrogen production) activities in a single process (Fig. 3; Kartal et al., 2007).

Figure 3.

 Comparison of cyclic electron flow modules, HURM (hydroxylamine/hydrazine-ubiquinone-redox-module) and the flux of nitrogen in (a) ANAOB (Kuenenia stuttgartiensis) and (b) AOB (Nitrosococcus oceani). Dashed lines represent external source and sink for NO2 and NO, respectively, in the proposed NOx cycle of AMO under anoxic conditions (Schmidt et al., 2001); dotted lines indicate lack of experimental verification of reactions assumed to occur (Hooper et al., 2005; Strous et al., 2006; Arp et al., 2007; Kartal et al., 2007). Broken lines indicate alternative start points of linear electron flow pathways. bc1, cytochrome bc1 (complex III); HZH, hydrazine hydrolase; HZO, hydrazine oxidoreductase; NirS, Fe-cytochrome cd1 nitrite reductase; PMF, proton-motive force; Q/QH2, ubiquinone-ubiquinol pool. See text for further details.

Evolutionary relatedness of ammonia- and methane oxidation

Nitrification, denitrification and ANAMMOX have always been understood per se as different processes performed by different organisms with different evolutionary histories. Because the mechanism of ANAMMOX has been worked out only recently (Strous et al., 2006; Kartal et al., 2007), little attention has been given to its evolution. Further, the great diversity of denitrifying organisms has prevented formulation of a plausible model for the evolution of denitrification. In contrast, many have speculated on the evolution of nitrification mostly by focusing on AMO (Teske et al., 1994; Holmes et al., 1995; Klotz & Norton, 1998; Purkhold et al., 2000; Norton et al., 2002). It was found that AMO is homologous with particulate methane monooxygenase (pMMO), a copper enzyme like AMO that carries out an analogous function in methane-oxidizing bacteria (MOB, Hanson & Hanson, 1996; Murrell & Holmes, 1996). While the first published hypothesis on the evolutionary relatedness of AMO and pMMO relied on an AmoA sequence from the γ-AOB Nitrosococcus oceani that, in hindsight, turned out to be a PmoA of a contaminating MOB in the culture, later comparisons of available AmoA and PmoA sequences and enzyme properties confirmed that AMO and pMMO are, indeed, homologous enzymes (Lontoh et al., 2000; Purkhold et al., 2000; Norton et al., 2002).

Although it has been known for over a decade that MOB contribute both to nitrification (Hanson & Hanson, 1996) and nitrifier denitrification (Sutka et al., 2003 and references therein), genes involved in this process, aside from MMO, were only recently identified (Ward et al., 2004; Bergmann et al., 2005). Hydroxylamine oxidation in Methylococcus capsulatus Bath was initially thought to be performed solely by cytochrome P460, a monoheme cytochrome capable of oxidizing hydroxylamine and nitric oxide in both AOB (Fig. 2) and MOB (Bergmann et al., 1998; Sutka et al., 2003; Hooper et al., 2005). Recent findings, however, assigned cytochromes P460 and c′-β a NO-detoxification role (Fig. 2; Elmore et al., 2007), an important function to nitrifier denitrification. Analysis of the M. capsulatus Bath genome sequence has now established residence of HAO in this methanotroph in addition to cytochrome P460 (Ward et al., 2004; Bergmann et al., 2005). This was an important finding because it cemented the importance of HAO as a central player in the ammonia-oxidizing pathway of both AOB and MOB. Although the processes of chemolithotrophic ammonia-catabolism and chemoorganotrophic methanotrophy are evolutionarily linked by common descent of the nitrification module (AMO/pMMO, HAO and electron-transferring cytochromes), they represent totally different catabolic lifestyles, which is a conundrum to some and a beautiful demonstration of prokaryotic diversity created by the modular evolution of catabolism (Spirin et al., 2006 and references therein) to others including these authors.

Nitrogen cycle evolution on the anoxic earth

A major milestone on the path to understanding the evolution of the nitrogen cycle was the recent availability of genomes from ecotypically different AOB [freshwater sediment/soil, Nitrosomonas europaea (Chain et al., 2003); marine, Nitrosococcus oceani (Klotz et al., 2006); sewage, Nitrosomonas eutropha (Stein et al., 2007); soil, Nitrosospira multiformis (J.M. Norton et al., unpublished data)] as well as that of an ANAOB, Kuenenia stuttgartiensis (Strous et al., 2006). These genomes made it possible to distill common catabolic denominators and differences in aerobic and ANAMMOX. The genome inventories combined with existing literature on the genetics and physiology of AOB and ANAOB allowed identification of individual catabolic modules including the definitive catabolic core to extract and recycle electrons from ammonia (Figs 2 and 3), a finding that has sparked efforts to reformulate the evolutionary history of the nitrogen cycle.

Diverse speculations on the emergence of both biotic and abiotic components of the biogeochemical nitrogen cycle, based on geochemical and oceanographic data (i.e. chemical composition and isotopic signatures in sediments) over geological time scales, have been presented (Mancinelli & McKay, 1988; Falkowski, 1997; Brandes et al., 1998; Navarro-González et al., 2001; Raymond et al., 2004; Canfield et al., 2006; Wachtershauser, 2007 and references therein). Some authors interpreted these data as evidence for a late emergence of denitrification; even after nitrification became established in the post-Neoproterozoic era during the successive oxygenation of the world's oceans (Falkowski, 1997). Furthermore, some authors proposed that N2 fixation was a very early evolutionary development urgently needed because of the rapid depletion of the primordial fixed nitrogen pool (Navarro-González et al., 2001). While iron and molybdenum, key trace metals for nitrogenase function, were likely abundant in the Archaean era (Anbar & Knoll, 2002; Canfield et al., 2006 and references therein), the encoding gene complexity and high cost of the N2 fixation process were likely prohibitive to early evolution and selection of this function (Mancinelli & McKay, 1988). Ongoing N2 fixation, even with the less efficient Fe-nitrogenase (rather than Fe–Mo nitrogenase) and at a slower pace than in modern times, would have led to a remarkable decrease of N2 in the atmosphere in the absence of sufficient recycling by denitrification or ANAMMOX. Although the wide taxonomic distribution of nif genes in extant bacteria and archaea has been a frequent argument for early emergence of N2 fixation (Falkowski, 1997; Navarro-González et al., 2001), nif genes are clustered and hence prone to lateral transfer (Raymond et al., 2004). Finally, the lack of N2-fixing organelles in eukaryotes provides an additional argument against an early evolution of the N2 fixation inventory in prokaryotes (McKay & Navarro-González, 2001). Therefore, the following postulates are in congruence with the hypothesis of a late emergence of N2 fixation and early emergence of denitrification (Mancinelli & McKay, 1988; Capone, 2000; Capone et al., 2006; Canfield et al., 2006), occurring soon after the stabilization of the independent lineages (domains) of cellular life (Koonin & Martin, 2005; Wachtershauser, 2007 and references therein) and when nitrogen flux was still largely controlled by abiotic activities (Yung & McElroy, 1979; Mancinelli & McKay, 1988).

In the modern nitrogen cycle (Fig. 1c), only denitrification in a wider sense is suited to return fixed nitrogen and sustain the atmospheric N2 pool (Capone et al., 2006); however, an often overlooked point is that not all enzymes that function in the extant nitrogen cycle were likely around during the anoxic Archaean and reducing Proterozoic eras because their metal cofactors were not widely available. Based on pertinent isotope signatures, enzymes with nickel (e.g. hydrogenase), iron (e.g. sulfur–iron and cytochrome c proteins) and molybdenum (e.g. formate dehydrogenase and nitrate reductase) cofactors were likely functional in the Archaean, while enzymes with Class B transition metals such as copper, zinc and cadmium were not. Class B transition metals likely became even scarcer in the Proterozoic era, in which the sulfidic nature of the oceans locked metals like copper into biounavailable sulfidic minerals (Lewis & Landing, 1992; Canfield, 1998; Anbar & Knoll, 2002; Poulton et al., 2004).

Because copper was not bioavailable to serve as a redox cofactor in catalysis, present day nitrous oxide reductase (NOS), a copper enzyme found in all extant canonical denitrifiers that produce dinitrogen (Zumft, 1997; Brandes et al., 2007), either was (1) preceded by a functional noncopper NOS lost from or not yet recognized in current genome inventories, (2) evolved from a noncopper NOS, or (3) was a de novo invention of the oxic era (Klotz, NSF Microbial Genome Sequencing Program workshop 2007, San Diego, CA). The authors favor the last hypothesis since early denitrifiers would have had a sufficient NO3 pool to support anaerobic respiration (Fig. 1a) and incomplete denitrification lacking NOS activity argues against an efficient early N2 fixation process. Early emergence of both N2 fixation and incomplete denitrification would have caused marked depletion of the atmospheric N2 pool for which there are no supporting data (Capone & Knapp, 2007).

In addition, the authors support the hypothesis of Mancinelli & McKay (1988) that ammonification was a key process providing fixed-nitrogen to early microbiota in the early nitrogen cycle rather than N2 fixation (Fig. 1a). Respiratory nitrite ammonification, the electrogenic reduction of nitrite to ammonia via formate or H2, is facilitated by pentaheme cytochrome c nitrite reductases (NrfABCD/NrfAH) found in a wide range of obligate and facultative chemolithotrophic bacteria (Simon, 2002). With abundant nitrate, NrfAH could receive nitrite from the respiratory, membrane-associated, Mo-containing, nitrate reductase (NarGHJI) or with scarce nitrate from the Mo-containing periplasmic nitrate reductase (NapFDAGHBC) (Lin & Stewart, 1998; Gunsalus & Wang, 2000; Simon, 2002; McLain & Martens, 2005). The NrfAH complex, whose structure has been solved from several Proteobacteria (Einsle et al., 2002; Rodrigues et al., 2006), functions independently of oxygen and is dependent on the availability of heme-iron; requirements likely met in the Archaean era.

In addition to nitrite reduction, extant NrfA (EC also reduces nitric oxide and hydroxylamine to ammonia without the release of RNS intermediates (Simon, 2002 and references therein). This may not have applied to early NrfA, however, which may have leaked catalytic intermediates including hydroxylamine (Fig. 1b). Because hydroxylamine is a potent mutagen, its production would have provided functional pressure for the evolution of two unrelated, oxygen-independent, catalytic complexes capable of scavenging hydroxylamine, and hence, the evolution of the ANAMMOX process (Fig. 1b). The first hydroxylamine-scavenging complex is the ammonia-forming hydroxylamine reductase (EC, a soluble iron–sulfur protein known as prismane or hybrid-cluster protein as it contains a combination of a 4Fe–4S cluster and a novel 4Fe–2S–2O cluster (Pino et al., 2006). Genes encoding prismane protein are found in the genomes of many extant strict and facultative anaerobic bacteria and archaea. The second complex is the hydroxylamine/hydrazine oxidoreductase, an octaheme cytochrome c dehydrogenase (Hooper et al., 2005), found in both AOB and ANAOB.

The hydroxylamine/hydrazine oxidoreductase is called HAO in AOB where it oxidizes hydroxylamine to nitrite (Hooper et al., 2005), and HZO in ANAOB where it oxidizes hydrazine to N2 (Schalk et al., 2000). Biochemically, HAO is currently misclassified as both an oxygen-dependent enzyme (EC and a ‘hydroxylamine (acceptor) oxidoreductase (EC’ Because HAO/HZO interact with cytochromes c as acceptors and both are hydroxylamine/hydrazine (donor) oxidoreductases (Fig. 3), a reclassification as ‘EC 1.7.2./’ is in order. Electron flow from hydroxylamine to ubiquinol in AOB involves both a cyclic path, to provide reductant to AMO, and a linear path for the generation of PMF (Fig. 3; Arp et al., 2007). Comparison of the cyclic electron flow in AOB and ANAOB (Strous et al., 2006) revealed a striking similarity and a central positioning of HAO/HZO (Fig. 3). HAO and HZO are functional analogs as both dehydrogenate hydroxylamine and hydrazine and deliver the extracted electrons to ubiquinol via cognizant cytochromes c. Thus, HAO and HZO are interchangeable, and define the hydroxylamine/hydrazine-ubiquinol redox module (HURM) of AOB and ANAOB (Figs 2 and 3; Klotz, NSF Microbial Genome Sequencing Program workshop 2007, San Diego, CA).

Whereas the evolutionary history of hydroxylamine reductase is still elusive (Cabello et al., 2004; Pino et al., 2006), sequence and phylogenetic analyses link the evolution of HAO to pentaheme cytochrome c nitrite reductase, NrfA (Bergmann et al., 2005). Recent analysis of HZO with other putative octaheme cytochrome c oxidoreductases revealed that HAO and HZO are homologs evolving from NrfA via N-terminally expanded octaheme cytochrome c nitrate reductases found in a few sulfur-dependent Deltaproteobacteria (Geobacter, Pelobacter) and the γ-proteobacterial genus Thioalkalivibrio, purple sulfur bacteria like the γ-AOB Nitrosococcus (M.G. Klotz et al., unpublished data). This evolutionary decent of HAO/HZO with a common root in octaheme cytochrome c oxidoreductases of sulfur-dependent nitrite-respiratory bacteria is remarkable as it reflects the selection of inventory capable of hydroxylamine detoxification formed as a byproduct of nitrite reduction by NrfA and represents an evolutionary path supportive of an early incomplete denitrification pathway depicted in Fig. 1a and b.

Although likely not suitable to sustain global primary production on the early Earth (Canfield et al., 2006), it is proposed that ANAMMOX evolved shortly after the incomplete denitrification pathway with the inclusion of Fe-dependent cytochrome cd1 nitrite reductase (NirS) and a unique protein, hydrazine hydrolase (Strous et al., 2006). According to this model, ANAMMOX likely provided the first complete recycling of fixed nitrogen to the dinitrogen pool (Dalsgaard et al., 2005) and fulfilled this role until the emergence of copper enzymes like Cu-NOS; NirK, the copper-dependent version of nitrite reductase (Cantera & Stein, 2007b); and the many members of the heme-copper oxidase (HCO) superfamily (Garcia-Horsman et al., 1994). At the same time, early evolution of HURM by ANAMMOX-performing microorganisms also provided the opportunity for later utilization of hydroxylamine as a reductant and energy source in oxic environments by AOB, leading to the evolution of nitrification and closure of the nitrogen cycle (Fig. 1c, Klotz, NSF Microbial Genome Sequencing Program workshop 2007, San Diego, CA).

The closed nitrogen cycle: emergence of copper enzymes and nitrification

Aerobic ammonia oxidation likely evolved into an efficient catabolic process only after a large enough pool of reduced inorganic nitrogen became available to sustain it (Fig. 1c). Aerobic ammonia oxidation also likely coincided with or quickly succeeded the evolution of copper-containing enzymes, which necessarily required the oxygenation of the Earth's oceans to release bio-available copper. The most vital copper-containing enzymes in AOB are the O2-dependent AMO and oxygen-reducing terminal HCOs at the beginning and the end, respectively, of the electron transport chain (Fig. 2). Availability of copper likely also promoted the evolution of a large amount of multicopper blue proteins (Nakamura & Go, 2005) including Cu-NirK (Cantera & Stein, 2007b). Likewise, the emergence of Cu-NOS and its lateral distribution in many prokaryotes, but not AOB and ANAOB, led to the highly efficient complete denitrification process, as it is known today.

AMO and pMMO are homologous and the authors support de novo emergence of their ancestor, a promiscuous ammonia-methane monooxygenase, in α-proteobacterial MOB and its holophyletic evolution or early lateral dispersal to recipients in the γ-proteobacterial orders Chromatiales (Nitrosococcus) and Methylococcales (all γ-MOB). Once in place as a modular extension of existing catabolic units, AMO (to extend HURM) and pMMO (to extend methanol catabolism) were subject to niche-dependent adaptation towards higher affinity for the substrate best suited to the ammonia- or methane-catabolic lifestyle in γ-proteobacterial AOB or MOB (Arp et al., 2007). The authors further support lateral transfer of the ammonia-oxidizing inventory to the ancestor of the β-proteobacterial family of the Nitrosomonadaceae from γ-AOB. This model is supported by recent comparative analyses of nitrifier genomes (Arp et al., 2007) and by the enzymic properties of AMO of Nitrosococcus oceani, which has nearly equal affinity for methane and ammonia (Lontoh et al., 2000).

The noted inability of AOB to catabolize alternative natural energy sources combined with their strict dependence on O2 suggests that γ-AOB and β-AOB may have evolved by genome economization and reductive evolution (i.e. loss of uptake and processing capacity for other energy sources including organics) and that this process of genome reduction likely occurred in concert with their adaptation to specific environmental niches (Arp et al., 2007; Stein et al., 2007). Furthermore, adaptation to specific niches appears to have created specialized inventories for each AOB ecotype as reported from the genomes of the marine AOB, Nitrosococcus oceani (Klotz et al., 2006), and the sewage AOB, Nitrosomonas eutropha (Stein et al., 2007). The inventory unique to the soil AOB, Nitrosospira multiformis, is included in a forthcoming report.

Tipping balance of the extant nitrogen cycle: anthropogenic nitrogen and nitrifier denitrification

Ever increasing and disproportionally high inputs of anthropogenically produced nitrogen have greatly stimulated nitrifier denitrification activities by AOB (Fig. 1c) and other aerobic bacteria in terrestrial, freshwater and marine ecosystems, leading to concomitant increases in atmospheric nitrous oxide (Colliver & Stephenson, 2000; Houghton et al., 2001; Wrage et al., 2001; Beaumont et al., 2002; Stein & Yung, 2003). Stimulation of nitrifier denitrification is particularly acute in marine environments where fertilizer runoff into large rivers and estuaries and aquaculture operations have led to increased productivity, and hence, large oxygen-depleted zones (Dore & Karl, 1996; Shailaja et al., 2006). The globally distributed marine AOB (Ward & O'Mullan, 2002) are considered a major source of global N2O production as the oceans cover 71% of Earth's surface and regulate weather, climate, and composition of the atmosphere (Nevison & Holland, 1997). Furthermore, methane consumption rates, particularly in upland soils, tend to decrease with increased N-inputs as methanotrophs cometabolize ammonia at the expense of methane (Robertson et al., 2000) and are inhibited by nitrite produced by AOB (King & Schnell, 1994; Bodelier & Laanbroek, 2004). In addition, more recent studies showed that interactions between ammonia and methane flux change methanotroph community structure and dynamics (Mohanty et al., 2006), and preliminary results suggest that methanotrophs lacking HURM (Figs 2 and 3) are particularly vulnerable to hydroxylamine rather than to nitrite toxicity (L.Y. Stein and M.G. Klotz, unpublished results). Taken together, this information suggests that mitigation of present and future imbalances in the global nitrogen cycle will be required to reduce microbial production of greenhouse gases and slow global warming.

The authors are only beginning to define the genetic inventory involved in nitrifier denitrification in the AOB, some of which may have originated in AOB as these organisms adapted to toxic NOx intermediates produced during ammonia-oxidizing metabolism. For example, most AOB encode cytochromes P460 (cytL) and c′-β (cytS), both of which mediate NO toxicity in other bacteria (Elmore et al., 2007 and references therein). Nitrosocyanin (ncyA) has thus far been found only in AOB genomes, and is hence believed necessary for the obligate chemolithotrophy of these bacteria (Arp et al., 2007 and references therein). Interestingly, NcyA synthesis was up-regulated in NO- and NO2-exposed Nitrosomonas europaea biofilms (Schmidt et al., 2004), and its structure suggests that it can bind and reduce NO (Arciero et al., 2002). Also, the ncyA gene of Nitrosomonas eutropha is preceded by a conserved FNR binding motif (L.Y. Stein, unpublished results), a regulatory protein involved in response to NO or low oxygen (Van Spanning et al., 1995; Cruz-Ramos et al., 2002). Physiological investigations of nitrifier denitrification in Nitrosomonas spp. have implicated involvement of the Cu-NirK nitrite reductase and the NorB nitric oxide reductase (Beaumont et al., 2002, 2004, 2005; Schmidt et al., 2004; Cantera & Stein, 2007a) and cultivated AOB from all ecotypes encode both nirK and norCBQD in their genomes (Casciotti & Ward, 2005; Cantera & Stein, 2007a; Garbeva et al., 2007). Both NirK and NorB were shown to be involved in nitrifier denitrification under aerobic and anaerobic conditions in the β-AOB, Nitrosomonas europaea (Schmidt et al., 2004). Furthermore, the four-member nirK gene clusters of Nitrosomonas and Nitrobacter genera are preceded by the nsrR gene, which encodes a Rrf2 family transcriptional regulator, and are separated from nsrR by a conserved NsrR-binding motif (Beaumont et al., 2004; Cantera & Stein, 2007b). In Nitrosomonas europaea, nirK expression is up-regulated under high nitrite concentrations by derepression of NsrR (Beaumont et al., 2004). Although putative nsrR genes are present in the genomes of Nitrosospira multiformis and Nitrosococcus oceani, they are not near the nirK genes nor are the nirK genes preceded by NsrR-binding motifs; therefore, NsrR-specific regulation of nirK appears to be only in nitrifiers adapted to ammonia-rich environments (Cantera & Stein, 2007b). Interestingly, neither nirK- nor norB-knockout mutants of Nitrosomonas europaea had completely diminished N2O production (Beaumont et al., 2002; Schmidt et al., 2004), and the nirK mutant actually produced more N2O than wild-type Nitrosomonas europaea (Cantera & Stein, 2007a). Thus, additional nitrite and nitric oxide reductases apparently contribute to nitrifier denitrification. A global transcriptome study of NirK-deficient Nitrosomonas europaea revealed selective up-regulation of genes in a cluster encoding subunits I and II of HCO (coxAB) and a gene encoding SenC/ScoI (Cho et al., 2006). It was recently proposed that this particular CoxBA complex, found only in AOB and a handful of genomes from bacterial sulfur oxidizers, is actually a NO reductase now termed sNOR and encoded by norYS (Stein et al., 2007; J. Hemp et al., manuscript in preparation). In Nitrosomonas europaea, the norYS-senC gene cluster is preceded by a conserved binding motif for FNR, indicating its regulation by NO or low O2. Although other AOB also encode fnr genes, it is yet unknown whether FNR mediates NOx toxicity and nitrifier denitrification in the AOB, as an FNR mutant of Nitrosomonas europaea did not affect NirK or NorB expression (Beaumont et al., 2002; Schmidt et al., 2004). Finally, Nitrosomonas eutropha, an AOB that occupies only high N-load niches, is the only known AOB to encode a fixNOP gene cluster, which encodes a high-affinity cbb3 cytochrome c oxidase (Stein et al., 2007). While, still speculative in part, all this inventory is currently under investigation for involvement in nitrifier denitrification and mediation of NOx toxicity in AOB representative of different ecotypes.


Attempts to reconstruct the evolutionary history of the biotic nitrogen cycle revealed the important molecular starting points of an incomplete denitrification process and ammonification, all based on Fe–Mo-, Fe–S- and Fe-heme-containing enzymes. The emergence of HURM, molecularly rooted in these processes and based on Fe-heme cytochrome c biochemistry, paved the way for both a complete recycling of reduced fixed nitrogen to the atmospheric N2 pool (ANAMMOX) as well as the evolution of nitrification once the world's oceans became oxygenated. The bioavailability of copper was a critical determinant in both completing the canonical denitrification pathway (NOS) and in constructing the nitrification pathway. The proposed evolutionary scenario also supports a late emergence of N2 fixation as the main process for making fixed nitrogen available after the oxygen-enabled explosion in biodiversity created a demand that exceeded its availability.


Comprehensive and comparative assessments of extant inventories, including their genomic environments, need to be combined with unequivocal mechanisms to capture molecular echoes of the past. Specifically, the complete extant metabolic inventory of the nitrogen cycle (i.e. Nir, Nar, Nor, Nos, Nrf) needs to be structurally and functionally defined and scrutinized by molecular evolutionary inference to establish molecular lineages of decent and distribution. These parallel approaches will help identify original pools of donors and recipients involved in horizontal transfer of the inventory required to support and retain niche-dependent metabolic functions (see relevant literature on lateral gene transfer and gene fitness, i.e. Lawrence & Hendrickson, 2005). Such analyses might clarify whether successful gene transfers were niche-specific (e.g. creating ecotypic guilds of recipients) or function-specific (e.g. for respiratory, assimilative, or detoxification purposes) whereby the function-specific amelioration likely expanded existing metabolic modules. Investigations of regulatory strategies that control expression of the inventory in extant organisms (i.e. by effectors like ammonia, nitrite or nitric oxide, and by DNA-binding proteins such as FNR, NnrR, or NsrR) would also clarify function(s) of the transferred genes.

A particularly urgent task at hand is the discovery of inventory in AOA and its comparative analysis with the respective inventory in AOB. For instance, initial analyses of available AOA genome sequences revealed the absence of cytochrome proteins and HURM, the essential components of ammonia-oxidation in AOB and ANAOB (see above text and Figs 2 and 3). Thus, alternative models for archaeal ammonia-oxidation remain to be established. These analyses will contribute to both a more complete picture of the evolution of the nitrogen cycle as well as to assess the contributions of extant bacteria and archaea in global nitrogen cycling.


The authors thank all colleagues in the nitrification network ( for sharing data and discussions, to Dr James Hemp (University of Illinois-Urbana) for critical reading of the manuscript, and to three anonymous reviewers for constructive advice. M.G.K. was supported, in part, by incentive funds provided by the UofL-EVPR office, the KY Science and Engineering Foundation (KSEF-787-RDE-007), and the National Science Foundation (EF-0412129). L.Y.S. was supported by funds from the Agricultural Experiment Station of UCR.