Growth of complete ammonia oxidizers on guanidine

Abstract

Guanidine is a chemically stable nitrogen compound that is excreted in human urine and is widely used in manufacturing of plastics, as a flame retardant and as a component of propellants, and is well known as a protein denaturant in biochemistry^1,2,3. Guanidine occurs widely in nature and is used by several microorganisms as a nitrogen source, but microorganisms growing on guanidine as the only substrate have not yet been identified. Here we show that the complete ammonia oxidizer (comammox) Nitrospira inopinata and probably most other comammox microorganisms can grow on guanidine as the sole source of energy, reductant and nitrogen. Proteomics, enzyme kinetics and the crystal structure of a N. inopinata guanidinase homologue demonstrated that it is a bona fide guanidinase. Incubation experiments with comammox-containing agricultural soil and wastewater treatment plant microbiomes suggested that guanidine serves as substrate for nitrification in the environment. The identification of guanidine as a growth substrate for comammox shows an unexpected niche of these globally important nitrifiers and offers opportunities for their isolation.

Main

Recently, microbial guanidine metabolism has received a lot of attention, as the identification of four different classes of riboswitches (RNA elements that bind metabolites or metal ions as ligands and regulate mRNA expression) selective for guanidine in the genomes of many bacteria in various phyla suggested that guanidine is an important metabolite of microorganisms. However, the microbial pathways for the formation and degradation of this compound and its ecological importance largely remain to be explored^4,5,6,7,8. During the past few years, bacterial guanidine production by at least three pathways has been demonstrated. Some bacteria encode an ethylene-forming enzyme that has an important role in bioethylene production and produces guanidine from arginine and 2-oxoglutarate⁹. Furthermore, during synthesis of the antibiotic naphthyridinomycin by Streptomyces lusitanus, the arginine-4,5-desaturase NapI leads to guanidine formation as a side reaction¹⁰. Moreover, bacteria can transform guanylurea to ammonia and guanidine¹¹. This is particularly important as guanylurea is one of the most common contaminants in nature formed by degradation of the type 2 diabetes drug metformin (a biguanidine and one of the most prescribed drugs globally) and of the fertilizer additive cyanoguanidine (dicyandiamide)^11,12,13. However, additional guanidine-forming mechanisms must exist in bacteria, as guanidine has been detected under nutrient-poor conditions in Escherichia coli lacking the aforementioned pathways⁴. Notably, in plants and algae, guanidine is also produced by homoarginine-6-hydroxylases^10,14.

Our knowledge of guanidine degradation by microorganisms is also still in its infancy, but recent research has revealed two degradation pathways for guanidine that are used by bacteria to not only detoxify guanidine but also to use it as a nitrogen source and, ultimately, to produce three molecules of ammonia from one molecule of guanidine. A widespread sequential decomposition pathway (Fig. 1a) involves a biotin-containing and ATP-dependent guanidine carboxylase, a heteromeric carboxyguanidine deiminase and an allophanate hydrolase to convert guanidine to ammonia and CO₂^15,16. In 2021, a Ni-containing guanidinase, mediating the direct hydrolysis of free guanidine to urea (which is further converted by urease to ammonia and CO₂) and ammonia, was described in a Synechocystis strain (Fig. 1b). This guanidinase has a crucial role in detoxifying guanidine during bioethylene production and enables this cyanobacterium to tap the guanidine pool as a nitrogen source without spending ATP for its degradation^17,18.

Fig. 1: Pathways and genes involved in guanidine degradation.

a, The guanidine carboxylase pathway. AH, allophanate hydrolase; CD, carboxyguanidine deaminase; Gase, guanidinase; GC, guanidine carboxylase; P_i, inorganic phosphate; Uase, urease. b, The guanidinase pathway. c, Arrangements of genes encoding proteins involved in guanidine degradation under the regulation of type I guanidine riboswitches in the comammox microorganism N. inopinata, the AOB Nitrosospira multiformis and two bacterial model organisms for guanidine catabolism. Each of the four genomes also encodes urease (not displayed) at locations that are not regulated by guanidine riboswitches.

Guanidine use by nitrifying microorganisms

Chemolithoautotrophic nitrifying microorganisms catalyse the aerobic oxidation of ammonia through nitrite to nitrate and therefore have an essential role in global biogeochemical nitrogen cycling, lead to massive fertilizer loss in agriculture, contribute to the emission of the potent greenhouse gas and ozone-depleting substance nitrous oxide, and are essential for nutrient removal in wastewater treatment plants (WWTPs)^19,20. In addition to ammonia-oxidizing bacteria (AOB) and archaea (AOA), which convert ammonia to nitrite (that is subsequently oxidized to nitrate by nitrite-oxidizing bacteria), the recently identified complete ammonia oxidizers (comammox) of the genus Nitrospira use ammonia as a substrate and oxidize it through nitrite to nitrate^21,22. In contrast to canonical nitrite oxidizers of the genus Nitrospira^23,24, the recognized range of energy sources used by ammonia-oxidizing microorganisms is very narrow. In addition to ammonia, only urea and cyanate have been experimentally confirmed to support the growth of some ammonia-oxidizing microorganisms by serving as an ammonia source after being converted enzymatically by cytoplasmic ureases and cyanases, respectively^25,26,27, although meta-omic studies of comammox strains in WWTPs suggested a more pronounced metabolic versatility²⁸.

We wondered whether ammonia-oxidizing microorganisms might be able to exploit guanidine as a source of energy, reductant and nitrogen. We therefore screened all available genomes from this guild for genes possibly encoding enzymes involved in the two known pathways that would enable guanidine to be used as an ammonia source (Fig. 1 and Supplementary Table 1). We found genes for guanidine degradation in most betaproteobacterial AOB (135 out of 145 genomes) and comammox strains (76 out of 83 genomes) but none within the gammaproteobacterial AOB or the various members of the AOA. Notably, most ammonia oxidizers encoding enzymes involved in guanidine utilization also possess an amino acid/polyamine/organocation permease (APC superfamily) in the same genomic region, which is predicted to enable the import of guanidine without using ATP (possibly through proton symport; Fig. 1). By contrast, Pseudomonas syringae pv. tomato str. DC3000 and Synechocystis sp. PCC 6803 (the model organisms for the guanidine carboxylase and guanidinase pathways, respectively) encode an ATP-dependent ABC transporter for guanidine transport in the neighbourhood of the genes encoding either guanidine utilization pathway^15,17,18. Furthermore, the betaproteobacterial AOB exclusively possess genes for the ATP-requiring guanidine carboxylase pathway. By contrast, only two comammox genomes encode this pathway, whereas the majority of comammox genomes instead are predicted to be equipped with the more energy-efficient pathway using a Ni-containing guanidinase (Extended Data Fig. 1). Most comammox genomes and metagenome-assembled genomes (MAGs; 75 out of 83) also encode a urease for conversion of urea formed by guanidinase to ammonia. Taken together, ammonia oxidizers equipped for guanidine metabolism do not rely on an ATP-consuming guanidine transporter and comammox microorganisms additionally use an ATP-independent guanidinase. Thus, these ammonia oxidizers and, in particular, comammox, if indeed capable of using guanidine, would be much more energy efficient than the previously characterized Pseudomonas and Synechocystis strains (that use guanidine as a nitrogen source) in converting guanidine to three molecules of ammonia.

With the distribution patterns of guanidine utilization genes in mind, we tested the guanidine metabolization ability using equally dense cultures of five AOB strains that possess the complete gene set of the pathway (one lacking only the urease), and the only described comammox pure culture N. inopinata. Notably, after incubation for 2 weeks with 50 µM guanidine as the sole substrate, only N. inopinata was able to almost fully degrade guanidine as a sole substrate and produce nitrite and nitrate. Three of the AOB strains were able to convert a fraction of the added guanidine to nitrite but only in the concomitant presence of ammonium (Extended Data Fig. 2).

Consistent with a guanidine metabolism, a guanidine-I riboswitch (ykkC-yxkD) is found immediately upstream of the APC superfamily permease and guanidinase genes in N. inopinata and many other comammox microorganisms (Fig. 1c, Extended Data Fig. 1 and Supplementary Tables 2 and 3). In comammox organisms, ABC transporters and, unexpectedly, also Ni chaperones were absent from the genomic regions surrounding the guanidine-I riboswitch. This riboswitch class acts as a transcriptional suppressor, permitting transcription of downstream genes only in the presence of guanidine⁴. Phylogenetic analyses revealed that the comammox guanidinases belong to the ureohydrolase enzyme superfamily (Fig. 2a). Substrates converted by this enzyme family typically contain a guanidine moiety, which is hydrolysed to release urea. In the guanidinase from Synechocystis sp. PCC 6803, specificity to guanidine was dependent on two amino acids, threonine at position 97 and tryptophan at position 305 (ref. ¹⁸). The guanidinase from N. inopinata belongs to the same clade of this superfamily and also possesses threonine and tryptophan at the key sites for guanidine specificity (Fig. 2a,b). Comparison of the comammox guanidinase phylogenetic tree with the comammox ammonia monooxygenase tree revealed a clear separation between the comammox A and B groups, with several subclades of comammox group A also showing similar clustering and approximate branching order in the trees (Fig. 2c). These phylogenetic consistencies, and the very widespread distribution of putative guanidinases among comammox microorganisms, suggest that guanidine use is probably an ancient and persistent trait of comammox organisms. Together with the observation that most comammox strains contain the enzymatic repertoire for guanidine degradation, this strongly indicates that the ability for guanidine degradation confers a selective advantage to comammox strains and has therefore been retained in their genomes during evolution.

Fig. 2: Phylogeny and structure of comammox guanidinases.

a, Phylogeny of the ureohydrolase superfamily. The circles indicate functionally characterized members. Characterized N¹-aminopropylagmatinases were included in the indicated agmatinase clade. Branches in grey do not correspond to any known function or are not monophyletic for function. b, Simplified phylogeny of guanidinases, using biochemically characterized arginase family members as the outgroup (a full tree is shown in Extended Data Fig. 4a). Specific amino acid residues that might be important for guanidine catalysis according to previous studies^15,17,18 (N. inopinata guanidinase positions Thr105, His222, Trp313 and Glu344; Supplementary Table 4) are indicated. c, Comparison of the phylogenetic relationships of comammox guanidinases (amino acids) and ammonia monooxygenases (amoA nucleotides). Tree tips are connected for genes found in the same genome. d, Electrostatic surface representation of guanidinase (scaled from −5 kT/e (red) to +5 kT/e (blue)), with the suggested entry of a tunnel towards the active site highlighted (left). Right, magnification of the region indicated by a box. The tunnel as determined by CAVER⁵⁷ is shown as a green line, and its width is indicated by a dot representation. Active-site residues are shown as sticks, and nickel (Ni) and manganese (Mn) ions are shown as green and purple spheres, respectively. Guanidine is shown as sticks with corresponding van der Waals atomic radii indicated as dots.

Growth of N. inopinata on guanidine

To further characterize guanidine use by N. inopinata, and to test whether this comammox organism can grow on guanidine as the sole source of energy, reductant and nitrogen, batch incubation experiments were performed. For quantification of the guanidine concentrations, we adapted an analytical workflow based on derivatization of guanidine with benzoin²⁹. In addition to the previously reported derivatization product, our analysis revealed an additional product that showed a higher signal intensity. We optimized the derivatization protocol to increase the fraction of this product and thereby improve the analytical sensitivity, confirmed the specificity of the derivatization (Supplementary Fig. 1 and Supplementary Table 5) and then used this protocol (with calibration solutions containing 1–100 μM guanidine) for all measurements.

In batch incubation experiments with around 50 μM of guanidine, N. inopinata converted guanidine stoichiometrically to nitrite and nitrate (Fig. 3 and Extended Data Fig. 3f). No abiotic guanidine degradation was detectable with dead N. inopinata biomass (Fig. 3a). During incubation with guanidine, N. inopinata cell numbers increased significantly in comparison to the controls without guanidine (Fig. 3c). Furthermore, N. inopinata assimilated nitrogen and carbon from isotopically labelled guanidine and bicarbonate, clearly demonstrating chemolithoautotrophic growth of N. inopinata on guanidine as a sole substrate (Fig. 3f,g). N. inopinata growth on guanidine was slower compared with growth on ammonium with maximum growth rates (division rates) of 0.076 and 0.632 d⁻¹, respectively (Fig. 3c, Extended Data Fig. 3c and Supplementary Table 6), and nitrite/nitrate production in the presence of both ammonium and guanidine was slightly (but significantly) slower compared with ammonium only (Extended Data Fig. 3g), yet the cell yield (that is, the number of cells formed per mol of combined nitrite and nitrate produced) did not differ between treatments receiving guanidine, ammonium or both substrates (Extended Data Fig. 3i). Notably, when both ammonium and guanidine were supplied, both were used simultaneously, rather than in a diauxic growth pattern (Extended Data Fig. 3d). Guanidine utilization rates per cell did not differ between treatments receiving only guanidine, or both ammonium and guanidine (Extended Data Fig. 3h). These observations imply that guanidine is used by comammox microorganisms whenever it becomes available, consistent with the ‘on/off’ type regulation of guanidinase expression by the guanidine-dependent riboswitch.

Fig. 3: Growth of N. inopinata on guanidine as the sole source of energy, reductant and nitrogen.

a, Biodegradation of guanidine over time. Around 50 μM (150 µM N) isotopically labelled guanidine was added to a washed culture of N. inopinata (after pre-incubation with guanidine and ammonium for 1 month) and incubated for 126 days. The control for abiotic guanidine decay was performed using autoclaved N. inopinata biomass. On days 84 and 112, additional spikes of around 50 μM of guanidine were added to living biomass incubations (dashed grey lines). Ammonium never increased above the level of detection (5 µM) and urea concentrations remained below 2.5 µM (Supplementary Table 6). b, NO₂⁻ and NO₃⁻ production (combined). At the end of the incubation, 78% of the total added guanidine nitrogen was oxidized to NO₂⁻ and NO₃⁻. Nitrogen balances are shown in Extended Data Fig. 3f. c, qPCR analysis of 16S rRNA gene copy numbers. Statistical analysis was performed using Welch two-sample t-tests; ***P = 0.0049, t = 10.348, d.f. = 4, comparing N. inopinata cell numbers incubated with guanidine versus without guanidine at the 126 day timepoint. For a–c, data are mean ± s.d. across five biological replicates. d,e, Representative images of DAPI-stained N. inopinata cells (red, 10 ml culture filtered onto 0.2 µm GTTP filter) after 107 days of incubation with guanidine (d) and without guanidine (e). The same results were observed for all five biological replicates. f,g, Nitrogen (f) and corresponding carbon (g) isotopic enrichment of N. inopinata cells after 107 days of incubation with ¹⁵N-guanidine and ¹³C-bicarbonate as measured using nanoSIMS. The ¹³C-enrichment is lowered by dilution of the ¹³C-bicarbonate with CO₂ from the headspace air, CO₂ formation from pyruvate and breakdown of isotopically unlabelled guanidine in the medium. Data from one biological replicate are shown; a second replicate was measured with the same results. Scale bars, 10 µm (d–g).

Source Data

Protein expression of N. inopinata grown on guanidine was compared to growth on ammonium. Pure cultures were repeatedly fed with either 50 µM guanidine or 150 µM ammonium after substrate depletion. After incubation for 3 weeks, three proteins showed significant differential expression between the guanidine and ammonium treatments: guanidinase (log₁₀[fold change (FC)] = 1.2, adjusted P (P_adj) = 0.001), a putative fatty acid methyltransferase (FAMT, log₁₀[FC] = 1.1, P_adj = 0.036) and a putative RNA-binding protein (RbpB, log₁₀[FC] = 2.5, P_adj = 0.036) (Extended Data Fig. 4b). All three showed higher expression levels in the guanidine incubation than in the ammonium incubation, with guanidinase showing the highest significance (P_adj = 0.001). The APC permease, the putative guanidine transporter of N. inopinata, was detected in only one ammonium-treatment replicate, but was observed in 4 out of 5 guanidine-treatment replicates (Extended Data Fig. 4d). The APC permease and guanidinase are both under the control of the ykkC-yxkD riboswitch (Fig. 1c) and, therefore, both were expected to respond to guanidine in the same manner. However, the APC superfamily permease as an integral membrane protein is difficult to detect using proteomic approaches despite our use of protocols to maximize the recovery of such proteins. Neither FAMT nor RbpB are in the vicinity of a guanidine-dependent riboswitch in N. inopinata. Both were detected as differentially expressed with marginal significance (P_adj = 0.036) and were not pursued further.

Comammox guanidinase characterization

The guanidinase enzyme of N. inopinata was expressed in E. coli, purified to homogeneity and characterized as a homohexamer with a molecular mass of 240 kDa (Extended Data Fig. 5a), resembling the recently discovered guanidinase of Synechocystis sp. PCC6803 (GdmH)¹⁸. Crystal structure analyses of the heterologously expressed N. inopinata guanidinase (resolution of 1.58 Å) revealed that it exhibits the same α-β-α fold of the arginase subfamily as its cyanobacterial homologue¹⁸, but with each subunit (Fig. 2d and Extended Data Fig. 6) containing one nickel and one manganese ion with the overall structure being highly similar to the Synechocystis enzyme with an root mean squared deviation of 0.74 Å over an aligned length of 361 Cα atoms. However, the N. inopinata guanidinase possesses an N-terminal extension, which is lacking in the cyanobacterial enzyme (Extended Data Fig. 6b,c). This extension stabilizes the tertiary and quaternary structure of the guanidinase through extensive interactions, consistent with its notably high thermostability and activity in a wide pH range (Extended Data Fig. 5b,d and Supplementary Table 7). On the other hand, in comparison to the cyanobacterial enzyme, the N. inopinata guanidinase lacks a C-terminal extension (Supplementary Table 8 and Extended Data Fig. 6f).

In the active site of the heterologously expressed N. inopinata guanidinase, one nickel and one manganese ion are coordinated through conserved histidine and aspartic acid residues (Fig. 2d, Extended Data Fig. 6 and Supplementary Table 9). The positions and identity of metal sites were determined by the diffraction anomalous dispersion signal (Extended Data Fig. 6h,i) and corroborated by inductively coupled plasma mass spectrometry (ICP-MS) metal analysis (Extended Data Table 1). Another notable difference between the active sites of the Synechocystis and the N. inopinata enzymes is a tilt of N. inopinata guanidinase Trp313 towards the active-site nickel ion. The tilted Trp313 in N. inopinata guanidinase permits access to the active site from the surface, in contrast to Trp305 of the cyanobacterial enzyme, which blocks it (Extended Data Fig. 6g). The hexamer’s ridges display a negative electrostatic potential, and each cavity marks the start of a 17 Å negatively charged tunnel towards the active site (Fig. 2d and Extended Data Fig. 6e). This is suggestive of electrostatic attraction of the positively charged substrate guanidinium (the predominant form of guanidine under physiological conditions).

Kinetic analysis of the purified guanidinase of N. inopinata revealed a substrate affinity (K_m) of 13.6 ± 0.76 mM for guanidine at pH 7.5 and a temperature of 37 °C (Fig. 4a) when expressed in the presence of 1 mM NiSO₄, highly similar to the guanidinase of the cyanobacterium Synechocystis sp. PCC6803^17,18, although some kinetic properties were dependent on the nickel concentration used during heterologous expression (Extended Data Table 1). Moreover, the maximum reaction rate (V_max), enzyme turnover (K_cat) and enzymatic catalytic efficiency (K_cat/K_m) are all highly similar to the values obtained in a previous study¹⁷ after guanidinase overexpression in Synechocystis, but much lower than the reported values of the latter guanidinase overexpressed along with nickel-loading chaperones in E. coli¹⁸ (Extended Data Table 1). By contrast, the nickel-loading mechanism of N. inopinata is unclear as no such chaperones were found in the genome and, therefore, could not be co-expressed with the guanidinase in E. coli. Thus, it is tempting to speculate that the lower V_max and K_cat values of the heterologously expressed N. inopinata guanidinase reflect that the enzyme was not completely loaded with metals. This is consistent with an inferred average occupancy of 0.33 for nickel ions (0.66 per subunit) in the N. inopinata enzyme and with the ICP-MS data of the purified enzyme, which revealed a stoichiometry of 0.42 nickel ions per subunit (Extended Data Table 1). An alternative explanation for the low affinity for guanidine could be that the enzyme of N. inopinata converts guanidine as a side reaction and is actually specialized on another substrate (which would require in vivo its recognition by the riboswitch, which has a very high substrate specificity and has been shown to not recognize metabolites with a guanidino group³⁰). However, consistent with previous work on the cyanobacterial guanidinase¹⁸, we could not detect urea formation by the comammox guanidinase from the putative alternative guanidine compounds methylguanine, arginine, creatine, guanidino-butyrate and guanidino-propionate. Only for agmatine, a very minor urea production was detected that was much lower than the enzyme activity measured with guanidine as substrate (Extended Data Fig. 5e). This lack of activity for the alternative substrates aligns well with structural analysis, which indicates that the small entry channel can accommodate guanidine only when accounting for dynamic channel ‘breathing’. The calculated molecular volumes for the potential alternative substrates are 127% to 287% larger than that of guanidine³¹.

Fig. 4: Guanidine oxidation kinetics of purified guanidinase and of N. inopinata cells.

a, Kinetic characterization of the heterologously expressed N. inopinata guanidinase. A Michaelis–Menten model (red line) was fit to triplicate guanidine consumption rates by the guanidinase (pH 7.5, 37 °C) and used to determine the K_m and V_max. b, The whole-cell guanidine oxidation rates of N. inopinata were determined for cells that were pre-induced with guanidine for around 12 h with microsensor measurements (two additional biological replicates are shown in Extended Data Fig. 7a,b). A Michaelis–Menten model (red line) was used to determine the apparent half-saturation constant (K_m(app)), V_max and substrate specific affinity (a^o) based on guanidine oxidation rates.

Source Data

We also determined the whole-cell kinetic properties of N. inopinata using guanidine through substrate-dependent O₂ consumption rates. This analysis reflects more accurately the in vivo ability of this organism to convert guanidine, as it assesses the entire pathway from guanidine uptake and degradation through ammonia and nitrite oxidation to terminal oxidase activity in the actual cellular environment. Guanidine oxidation rates increased with increasing guanidine concentrations in a Michaelis–Menten kinetic profile, which was used to model the apparent substrate affinity (mean K_m(app), 1.34 ± 0.25 μM guanidine; n = 3) and maximum guanidine oxidation rate (mean V_max, 0.16 ± 0.02 μM guanidine per mg protein per h; n = 3) (Fig. 4b and Extended Data Fig. 7a,b). The whole-cell affinity for guanidine was much higher than the affinity of the purified heterologously expressed enzyme. This might, for example, reflect a more efficient metal loading of the enzyme in vivo, differences in post-translational modification in the natural host or crowding effects in the cytoplasm. At guanidine concentrations above around 400 μM, which probably do not occur in environments in which comammox thrive, and which are much higher than the guanidine concentrations used for the growth experiments, additional guanidine resulted in a decreased rate of respiratory activity. Guanidine inhibition at high concentrations was not specifically related to inhibition of the guanidinase (see the enzyme kinetic experiments) and also occurred when guanidine was added to ammonia-oxidizing N. inopinata cells (Extended Data Fig. 7c,d). The ability of N. inopinata to scavenge guanidine from a dilute pool, such as in an environmental setting, can be assessed by its substrate-specific affinity (a^o), incorporating the whole-cell K_m(app) and V_max (ref. ³²). N. inopinata has a mean a^o for guanidine of 21.6  l per g cells per h. In comparison, when pregrown under the same conditions, N. inopinata has a comparable mean a^o of 74.5  l per g cells per h for urea (although a higher a^o for this substrate has been reported recently³³) and an a^o of 528 to 2,262 l per g cells per h for total ammonium (Extended Data Fig. 7e–h). This highlights the hierarchy of substrate acquisition and use by N. inopinata. Although the whole-cell affinity of N. inopinata for guanidine is by two orders of magnitude below its affinity for NH₃, it still resembles or exceeds the whole-cell affinities for NH₃ of several terrestrial AOA from the Nitrososphaerales and Nitrosocaldales phylogenetic lineages and of many AOB³⁴.

Comammox use guanidine in WWTPs

Nitrification is an essential step for the efficient removal of nitrogen compounds in WWTPs. Comammox thrives in some but not all of these systems. As guanidine has been detected in human urine at concentrations between 2 and 20 µM (refs. ^35,36), and we detected guanidine in the influent of a municipal WWTP (concentration, 0.5 µM; Extended Data Table 2), we hypothesized that comammox microorganisms in WWTPs degrade guanidine under competitive conditions and studied how they respond to guanidine pulses. For our experiments, we selected two municipal WWTPs (Ribe and Haderslev, Denmark), in which comammox Nitrospira were abundant nitrifiers (Extended Data Fig. 8a–d). A previous metagenomic analysis had reconstructed an abundant comammox MAG from each of the two WWTPs³⁷. Both comammox MAGs contained the guanidine-I riboswitch upstream of the APC superfamily permease and guanidinase genes (Extended Data Fig. 1). After addition of 50 μM guanidine to starved biomass from these WWTPs, considerable guanidine degradation was observed, whereas a much slower degradation rate occurred in a control experiment with biomass from an Austrian WWTP in which comammox could not be detected by PCR³⁸ (Extended Data Fig. 8e–g). Activated sludge from the Ribe WWTP was additionally incubated with urea or ammonium and one replicate of each substrate amendment experiment was analysed by metatranscriptomics, using the different timepoints as replicates for comparison. At T = 0, baseline transcription of the guanidinase gene was 227 transcripts per million (TPM) in the Ribe sludge. In the Ribe WWTP, only the comammox guanidinase (log₂[FC] = 2.72; P_adj = 1.3 × 10⁻⁸) and APC superfamily permease (log₂[FC] = 2.62; P_adj = 7.5 × 10⁻¹¹) were more highly transcribed in the guanidine-spiked incubation compared with in the ammonia-spiked incubation, demonstrating the functionality of the riboswitch. Three genes were more highly transcribed in the guanidine-spiked incubation compared with in the urea-spiked incubation: guanidinase (log₂[FC] = 2.63; P_adj = 8.2 × 10⁻⁸), APC superfamily permease (log₂[FC] = 2.57; P_adj = 3.6 × 10⁻¹⁰) and a CBS-domain-containing protein with unknown function (MBK9946601.1: log₂[FC] = 2.19; P_adj = 3.9 × 10⁻³). High transcriptional levels of guanidinase and APC superfamily permease in the guanidine-spiked incubation were maintained throughout the experiment compared with the ammonia-spiked and urea-spiked incubations, and incubations with no experimental substrate addition (Fig. 5a), consistent with a high concentration of guanidine (>40 µM) over the time course (Fig. 5c and Extended Data Fig. 8f).

Fig. 5: Metatranscriptomic response of nitrifying activated sludges to guanidine amendment.

a,b, The transcriptional response of the guanidine APC superfamily permease (APCP) and guanidinase to substrate-spiked (guanidine, ammonium, urea) incubations in abundant comammox organisms from the Ribe (a) and Haderslev (b) WWTPs. Metatranscriptomic reads from the Ribe and Haderslev experiments were mapped to the respective Ribe and Haderslev comammox MAGs; transcriptional levels of the genes of interest are shown in TPM for all timepoints. c,d, Corresponding concentrations of guanidine in the replicate used for metatranscriptomics at time of sampling for Ribe (c) and Haderslev (d). Note that transcripts of the comammox guanidinases were also found at the start of the experiment, although the activated sludge samples had been starved during transport and storage, showing that transcription is either also occurring in the WWTP or that the sludge produced guanidine that was immediately consumed.

Source Data

Similar metatranscriptomic experiments were performed with biomass from the Haderslev WWTP, but no ammonia- or urea-spiked incubations were included, and the guanidine-spiked incubation was compared to a no-substrate control. At T = 0, the baseline transcription of the guanidinase gene was 316 TPM in the Haderslev sludge. In total, 373 genes were found to be differentially transcribed, including the guanidinase (log₂[FC] = 1.91; P_adj = 1.3 × 10⁻⁶). The APC superfamily permease transcripts increased at the early timepoints but were not found to have a significant difference between the guanidine-spiked incubation and the no-substrate control over the entire experiment (log₂[FC] = 1.58; P_adj = 0.09). The guanidine concentration in the Haderslev replicate time series used for transcriptomics dropped substantially between 8 and 16 h of incubation; this decrease was accompanied by a decrease in the transcriptional activity of the guanidinase and the permease (Fig. 5b). These results corroborate the findings from the Ribe sludge experiment, in that the guanidine concentration appears to exert strong control over the transcription of the comammox guanidine permease and guanidinase in WWTP communities.

Guanidine metabolism in soil

As we detected guanidine at low concentrations in urine and faecal samples from cows, pigs, chicken and sheep (Extended Data Table 2), guanidine metabolism in a comammox-containing agricultural soil, which had been fertilized with 525 kg total N per ha per year from solid cattle manure, was also investigated. As expected, the soil microbiome showed strong nitrification activity and added ammonium (30 µg N per g dry-weight soil) was rapidly nitrified (<2 days). Soil nitrification (ammonium consumption and nitrate formation) was completely inhibited by acetylene (Fig. 6). Notably, guanidine added to the soil was nitrified to nitrate over 27 days whether provided as the sole nitrogen source (30 µg N per g dry-weight soil) or in combination with ammonium (each ammonium and guanidine, 15 µg N per g dry-weight soil). However, when nitrification was inhibited, guanidine persisted in the soil at a high concentration for more than 27 days. Although these data do not prove guanidine degradation by comammox, they highlight the large contribution of ammonia-oxidizing microorganisms (or potentially other acetylene-sensitive microorganisms) to the observed guanidine degradation (Fig. 6). In all treatments, quantitative PCR (qPCR)-based quantification of AOB, AOA, and comammox clade A and B using the functional marker gene ammonia monooxygenase subunit A did not detect growth of ammonia-oxidizing microorganisms over the course of 27 days (P > 0.05; Supplementary Table 10). This is not surprising, as conditions that support growth of ammonia oxidizers in soil vary widely among soils^39,40,41. Growth of AOB is typically observed at higher nitrogen fertilization treatments than those used in our study^42,43, while the few studies reporting on comammox growth in soil observed it in soils unamended with nitrogen, or after substantially higher urea-N additions than those used in this study⁴¹.

Fig. 6: Soil nitrification activity in the presence and absence of externally added ammonium and guanidine.

Nitrogen applications used in the soil microcosm incubations are comparable to those applied in field nitrogen fertilizer additions⁵⁸. The red symbols represent addition of 30 µg guanidine-N per g dry weight (DW) soil; the orange symbols represent addition of 15 µg guanidine-N + 15 µg NH₄⁺-N per g dry-weight soil; the yellow symbols represent addition of 30 µg NH₄⁺-N per g dry-weight soil; and the black symbols represent no N addition. The circles represent treatments with active nitrification, and the crosses represent treatments with nitrification inhibited with 0.02% acetylene. For a–d, data are mean ± s.d. across three biological replicates. a, Net nitrate accumulation with active nitrification. In all treatments, urea and nitrite remained below the detection limit for the entire time course. b, Ammonium consumption with active nitrification. c, Nitrate and nitrite concentrations with nitrification inhibited with acetylene. d, Ammonium concentrations in the presence of acetylene. e, Guanidine concentrations (detection limit of 50 nM in the soil extract, corresponding to 14.1 ng N per g dry-weight soil) at day 0 and day 27 in the absence and presence of acetylene. Data are mean ± s.d. across three biological replicates. The overlaid dots show values from biological triplicates.

Source Data

Discussion

Comammox Nitrospira are to our knowledge the first organisms identified to grow with guanidine as their sole source of energy, reductant and nitrogen. It is well documented that functionally redundant microorganisms can coexist by partitioning low-concentration substrates even though they compete for one dominant substrate^44,45. Thus, it is tempting to speculate that the ability to use the low-concentration substrate guanidine provides an additional niche for comammox organisms in guanidine-containing environments and is probably important for their co-existence with other ammonia-oxidizing microorganisms, with which they compete for the dominant substrate ammonia. Co-existence instead of competitive exclusion enables functional redundancy of ammonia oxidizers, is frequently observed in many environments and probably preserves their overall ecosystem service.

The specific adaptation of comammox to using guanidine for nitrification also opens interesting perspectives for research and applications. Guanidine is the first substrate that has the potential to be selective for any lineage of ammonia oxidizers and has been reported to inhibit the AOB Nitrosomonas europaea at 1 mM concentration^46,47. Thus, mineral cultivation media containing guanidine as the only substrate should be tested for purification of new comammox Nitrospira strains from enrichments and environmental samples. Currently, only one comammox isolate (N. inopinata) and one high enrichment are available and have been partly physiologically characterized^21,48,49,50. In comparison to ammonia oxidation by AOB, complete nitrification by N. inopinata yields significantly less nitrous oxide (N₂O) as a by-product under oxic and hypoxic conditions⁴⁹. Future work is needed to examine whether the long-term application of guanidine-containing nitrogen fertilizers will influence the composition of the ammonia-oxidizer community in agricultural soils and the emissions of the greenhouse gas and ozone-depleting N₂O. Comammox Nitrospira could also have an important role for the removal of the widely used drug metformin in the environment. It is tempting to speculate that the presence of metformin—for example, in wastewater^51,52—opens together with other guanidine sources a niche for comammox in exposed habitats. In this case, the continuous release of metformin into the environment might influence the composition of nitrifying populations.

As we are just beginning to understand the environmental concentrations of guanidine, the importance of this nitrogen-rich compound for the biogeochemical nitrogen cycle cannot yet be quantified. For this purpose, a more sensitive analytical method for measuring in situ guanidine concentrations with high specificity is urgently needed. As shown previously for cyanate, even low standing concentrations of a nitrification substrate can be linked to relatively high production and consumption rates⁵³. Thus, future studies should aim to determine both the steady-state concentrations and the turnover rates of guanidine in pristine and human-affected ecosystems. Moreover, the diversity of natural and synthetic guanidinium compounds and their degradation pathways, such as of many neonicotinoid pesticides⁵⁴ and nitroguanidine, which is used in explosives^55,56, indicates that a multitude of hitherto mostly overlooked, biotic and abiotic guanidine-producing reactions may exist. Insights into the environmental dynamics of guanidine and its derivatives will close yet another gap in our picture of the nitrogen cycle and shed light onto the ecological roles of comammox Nitrospira and their enigmatic functional separation from the canonical nitrifiers.

Methods

Gene collection and functional classification

Supplementary Table 11 contains details for the collection and analyses parameters of each gene family and the following description is generalized. Predicted proteins from all publicly available genomes in GenBank as of 1 July 2022 (429,896 genomes) were screened with hmmsearch⁵⁹ using ‘collection HMMs’ for genes related to guanidine metabolism (Fig. 1). The resulting genes were used as query sequences against a combined HMM set of PFAMs, TIGRFAMS, NCBIFAMs and PANTHERFAMs for a set of acceptable ‘cross-check HMMs’. Genes were further screened using specific e-value and coverage cut-offs. Cross-check HMM names were used to query UniProt and results were filtered for reviewed entries with evidence at the protein level to identify functionally characterized proteins and download them if not already present in the dataset. The portion of the protein sequence that was aligned to the ‘collection HMM’ was extracted and clustered using usearch⁶⁰ with specified -id and -query_cov values to identify centroids. HMM-based alignments of centroid sequences generated from the initial hmmsearch were used in FastTree⁶¹ to generate phylogenetic trees for each gene family of interest. For the APC superfamily permease and the allophanate hydrolase, all proteins that passed the e-value and coverage cut-offs were inferred to possess the expected function. Guanidinases were also required to possess threonine at N. inopinata position 105 (PF00491 HMM position 15), histidine at N. inopinata position 222 (PF00491 HMM position 134) and tryptophan at N. inopinata position 313 (PF00491 HMM position 223)¹⁸. Guanidine carboxylases were required to possess a conserved aspartic acid at K. lactis position 1,584 (TIGR02712 HMM position 956) and were further differentiated from urea carboxylase by having an aspartic acid at K. lactis position 1,330 (TIGR02712 HMM position 701)¹⁵. Carboxyguanidine deiminases were defined using the common ancestor of the two subunits (CgdA and CgdB) in the general tree for PF09347. This common ancestral node then gave rise to CgdA and CgdB as two monophyletic clades defined as such.

Riboswitch collection

Genomes from ammonia-oxidizing microorganisms were screened using infernal (v.1.1.3)⁶² using established RFAM covariance models for guanidine riboswitches I (RF00442), II (RF01068) and III (RF01763) and a model for the recently described guanidine IV riboswitch, which was constructed with infernal using ‘GGAM-1-curated.sto’⁶³. Scaffold IDs, coordinates and orientation were recorded and cross-referenced against gff files to identify downstream genes. A gene was considered to be under the control of a riboswitch if it was in the same orientation as the riboswitch and the 5′ end of the gene was within 1,000 nucleotides of the riboswitch. The inferred operon was then extended downstream until genes could be found with the opposite orientation.

Phylogenetic analyses

Ureohydrolase

HMM-based alignments of centroid sequences defined above were used in FastTree2⁶¹ with the default parameters to generate a phylogenetic tree. The tree was midpoint rooted using the function midpoint() from the phangorn package, and functional clades were defined using the getMRCA() function within the ape package and visualized using the ggtree package in R.

Guanidinase

The most recent common ancestor of all guanidinases (as defined above) was identified in the ureohydrolase tree using the getMRCA() function and the descendant centroids were collected using the Descendants() function, both from the ape package⁶⁴. All HMM alignments of ureohydrolases that were represented by centroids collected using the Descendants() function were additionally required to have covered PF00491 over 90% of its length reclustered using usearch (-id 0.9 -query_cov 0.9). The HMM-aligned portion of this sequence dataset was used to calculate phylogeny with IQ-TREE2⁶⁵, using the best model (LG+I+I+R5), and bipartition support was evaluated using ultrafast bootstraps. Logos for each resulting clade of guanidinases and close relatives were generated using the ggseqlogo package in R.

For co-phylogeny analyses of comammox guanidinases and ammonia monooxygenases, comammox genomes were screened for the presence of amoA and guanidinase genes. As most available genomes were MAGs, it was required that exactly one copy of each gene was identified per genome. This resulted in 54 genomes for analysis. The AlignTranslation and AlignSeqs functions from the Decipher package were used to align amoA nucleotide and guanidinase amino acid sequences, respectively. IQ-TREE2 was used to identify the best models (amoA, TPM3u+F+I+I+R3; guanidinase protein, LG+I+G4), calculate trees and evaluate bipartition support with ultrafast bootstraps. Trees were visualized in R using a combination of the cophylo function from the phytools package and the ggtree package.

Guanidine quantification

The following chemicals were purchased from Sigma-Aldrich: guanidine hydrochloride (≥99%, G3272), benzoin (≥99%, 8.01776), potassium hydroxide (≥85%, 1.05033), ethanol (≥99.8%, 02851), formic acid (≥98%, 5.43804), β-mercaptoethanol (≥99%, 8.05740), l-arginine (≥99.5%, 11009) and sodium sulfite (≥98%, 239321). Hydrogen chloride solution (32%, 20254.321) and acetonitrile (≥99.9%, 20060.320) were purchased from VWR. 2-Methoxyethanol (≥99.5%, 10582945) was purchased from Thermo Fisher Scientific. MilliQ water was obtained from a water purification system (0.071 µS cm⁻¹; Elga Veolia, PURELAB Chorus). The derivatization protocol for guanidine was adapted from a previous study²⁹. In brief, 150 µl of an aqueous solution potentially containing guanidine was cooled to 0 °C in a 0.5 ml plastic tube (Eppendorf, Protein LoBind, 0030108434) and spiked with 75 µl of a benzoin solution in ethanol (4 mM), 75 µl of an aqueous solution containing both β-mercaptoethanol (0.1 M) and sodium sulfite (0.2 M), and 150 µl of an aqueous solution of potassium hydroxide (1.6 M). The resulting solution was mixed, heated in a bath of boiling water for 10 min and cooled in an ice bath for 2 min. Subsequently, 25 µl of an aqueous solution of hydrogen chloride (4.8 M) was added. The resulting solution was mixed and transferred to a 1.5 ml plastic tube (Eppendorf, 0030120086) and centrifuged at 10,000g for 2 min. Before analysis, the supernatant was diluted to obtain analyte concentrations in the optimal quantification range of the analytical instrument (that is, 0.05–5 μM). The predominant derivatization product (proposed structure in Supplementary Fig. 1c) was analysed using liquid chromatography (Agilent 1290 Infinity II) coupled to triple quadrupole mass spectrometry (Agilent, 6470) with a retention time of 3.73 min. We used the InfinityLab Poroshell 120 Bonus-RP (Agilent, 2.7 µm, 2.1 × 150 mm) column for separation, an injection volume of 2 µl, a flow rate of 0.4 ml min⁻¹, a column compartment temperature of 40 °C and the following eluents: aqueous (A): MilliQ water with 0.1% (v/v) formic acid; organic (B): acetonitrile with 0.1% (v/v) formic acid. The eluent gradient was as follows: 0–1.5 min, 5% B; 1.5–4 min, 5–61% B; 4–4.5 min, 61–95% B; 4.5–7 min, 95% B; 7–8 min, 95-5% B; 8–10 min, 5% B. The source parameters were set as follows: positive mode electrospray ionization; gas temperature, 250 °C; gas flow, 10 l min⁻¹; nebulizer, 45 psi; sheath gas temperature, 280 °C; sheath gas flow, 11 l min⁻¹; capillary voltage, 3.5 kV; nozzle voltage, 0.5 kV. The following product ions of the derivatization product (m/z of parent ion: 252.2) were monitored: m/z: 182.1 (quantifier) and m/z: 104.1 (qualifier). The resulting chromatographs were integrated using MassHunter (Agilent, v.10.1). For absolute quantification, we used a series of guanidine solutions with a concentration range after dilution between 0.05 and 5 µM. For pure culture medium, activated sludge and soil extracts, calibration solutions were prepared in the respective matrix. Animal urine and faeces were quantified with calibration solutions in water. Accurate quantification for animal samples was confirmed by spiking 20 µM guanidine to an animal faeces sample with a recovery of >83%. Animal manure samples were freeze-dried, and subsamples were dispersed in 2 M KCl solution (1 ml per 100 mg sample) and bead-beated for 15 min in a Lysing matrix A tube (MPBiomedicals), then centrifuged at 20,000g for 15 min. Wastewater treatment plant influent was quantified by standard addition. Calibration solutions were derivatized and analysed in the same way as and in parallel to the respective samples. For Orbitrap (high-resolution) MS analyses, we used liquid chromatography coupled to the Thermo QExactive mass spectrometer with the following parameters: positive electrospray ionization; capillary temperature, 275 °C; sheath gas, 15; aux gas, 10; sweep gas, 1; S-lens RF, 50.0; resolution, 140,000 (MS full-scan), 17,500 (MS/MS); NCE (stepped), 10,20,30. For growth experiment samples containing heavy-isotope-labelled guanidine, the total guanidine concentrations were inferred by assuming the measured isotopically unlabelled guanidine concentrations to correspond to 90% (we used 10% isotopically labelled guanidine).

Physiology experiments with N. inopinata and ammonia-oxidizing bacteria

The cells were grown in medium containing 54.4 mg l⁻¹ KH₂PO₄, 74.4 mg l⁻¹ KCl, 49.3 mg l⁻¹ MgSO₄·7 H₂O, 584 mg l⁻¹ NaCl, 147 mg l⁻¹ CaCl₂, 34.4 μg l⁻¹ MnSO₄·1H₂O, 50 μg l⁻¹ H₃BO₃, 70 μg l⁻¹ ZnCl₂, 72.6 μg l⁻¹ Na₂MoO₄·2 H₂O, 1 mg l⁻¹ FeSO₄·7 H₂O, 20.0 μg l⁻¹ CuCl₂·2 H₂O, 80 μg l⁻¹ CoCl₂·6 H₂O, 3 μg l⁻¹ Na₂SeO₃·5H₂O, 4 μg l⁻¹ Na₂WO₄·2H₂O, 24 μg l⁻¹ NiCl₂·6 H₂O and 0.5 mM pyruvate. The medium was buffered by addition of 4 mM HEPES, with the pH set to 8. For regular culture maintenance, cultures were kept in closed Schott bottles at 37 °C without shaking in the dark. When indicated, guanidine hydrochloride was added from a filter-sterilized 0.1 M stock solution to a final concentration of 50 μM.

For comparing guanidine utilization by pure cultures of N. inopinata and AOB, all strains were induced in 1 l batch cultures for 6 weeks with 0.5 mM ammonium and 1 µM guanidine fed weekly. Subsequently, the same amount of biomass per culture as determined using the Pierce BCA protein quantification kit (Thermo Fisher Scientific; calculated final concentration in the incubation, 10 µg ml⁻¹) was collected, washed and resuspended in fresh medium in equal volumes and transferred to 96-well, flat bottom culture plates (Greiner Bio-One). In these plates the following incubations were done with either 50 µM guanidine only; 50 µM guanidine plus 150 µM ammonium; or 150 µM ammonium only for 14 days at 28 °C in the dark and without agitation (optimal growth conditions for the ammonia oxidizing organisms used, while 9 °C colder than the optimum for N. inopinata).

For growth experiments, N. inopinata pure culture cells pregrown on 10 µM guanidine and 0.5 mM ammonium (with weekly refeedings) for 1 month were collected by centrifugation (4,500g, 30 min), washed with N-free medium three times and resuspended in fresh medium. Aliquots of 200 ml were distributed into 250 ml serum bottles. Aliquots used as dead controls were autoclaved (120 °C, 20 min) before substrate additions. The following N substrates were added (always 150 µM N) to five replicate bottles each: (A) ¹⁵N-guanidine (10% ¹⁵N-guanidine hydrobromide, 90% guanidine hydrochloride); (B) guanidine (as guanidine hydrochloride); (C) ¹⁵N-guanidine (10% ¹⁵N-guanidine hydrobromide, 90% guanidine hydrochloride) and ammonium (each 150 µM N); (D) ammonium only; (E) no N addition (starved control); (F) dead (autoclaved) control with ¹⁵N-guanidine (10% ¹⁵N-guanidine hydrobromide, 90% guanidine hydrochloride). Moreover, all bottles received ¹³C-bicarbonate additions (¹³C-NaHCO₃; 1 mM final concentration, 99% ¹³C) to detect chemolithoautotrophic growth and 0.5 mM sodium pyruvate as a reactive-oxygen-species scavenger⁶⁶. Serum bottles were closed with sterile, HCl-cleaned blue butyl rubber stoppers (Chemglass) and incubated at 37 °C in the dark without agitation. Samples of 2 ml for the determination of cell numbers (using qPCR) and of N-compound concentrations were taken with sterile syringes and needles and replaced with air every 1 to 14 days (frequent sampling in the beginning of the experiment, more spaced-out sampling after incubations containing ammonium were ended) over a time course of 126 days (12 days for treatments containing ammonium). Substrates were replenished after depletion. After 107 days of incubation, 10 ml samples were removed from treatments A, B, E and F, fixed with 3% formaldehyde (final concentration) for 30 min at room temperature, filtered onto polycarbonate filters (0.2 µm pore size, GTTP, 40 nm gold sputtered), washed with sterile 1× PBS, dried and stored frozen until further use. Cells were visualized after staining with 4′,6-diamidino-2-phenylindole (DAPI, 10 µg ml⁻¹, 5 min at room temperature) using a confocal laser-scanning microscope (inverted Leica TCS SP8X CLSM equipped with a 405 nm UV diode). At the end of the growth experiments, the absence of heterotrophic contaminants was confirmed by inoculation into heterotrophic growth medium (LB and TSY).

Ammonium, urea, nitrite and nitrate concentrations were measured by colorimetric protocols published previously⁶⁷. In brief, combined ammonia and ammonium concentrations were determined using the indophenol blue method. Nitrite concentrations were measured spectrophotometrically using the Griess method after reacting with sulfanilamide and N-1-naphthyl-ethylenediamine dihydrochloride. Nitrate was measured by the same method after reduction to nitrite with vanadium chloride. Urea concentrations were measured using the thiosemicarbazide-diacetylmonoxime method⁶⁸, according to a previous study¹⁸.

For quantification of N. inopinata cell numbers, qPCR was performed using the primers 515F/806R, targeting the V4 region of the 16S rRNA gene as described previously^69,70. Standards were generated from purified PCR products generated from N. inopinata genomic DNA as template. The standards were quantified according to the Qubit dsDNA HS Assay Kit instructions. Standards containing 10⁹ gene copies per µl were aliquoted and stored frozen at −20 °C until further use. Each standard aliquot was used and defrosted only once to freshly prepare tenfold serial dilutions (10⁸–10² gene copies per µl). The qPCR assays were performed as follows: the frozen culture aliquots were four times freeze-thawed for cell disruption. A total of 0.25 µM of each primer was used in a mixture of 10 µl SYBR Green Supermix (Bio-Rad), 2 µl cell lysate or standard, and water in a final volume of 22 µl per reaction. The qPCR cycler (C1000-CFX96, Bio-Rad) settings were as follows: 95 °C for 15 min; 40 cycles of 95 °C for 30 s, 50 °C for 1 min and 72 °C for 45 s (plate read); and finishing with 72 °C for 2 min and a melting curve performance from 40 °C to 95 °C with an increase of 0.5 °C every 5 s. The efficiencies of the standard curves had an average of 86% and an R² of 0.999. Growth rates (division rates) were calculated as follows:

$$v({{\rm{d}}}^{-1})={\log }_{2}({N}_{i+1}/{N}_{i})/t$$

(1)

where v is the rate of division (d⁻¹), N is the qPCR determined cell number at timepoint i + 1 and i, and t is the time interval between time point i + 1 and i in days.

For visualization of stable N and C isotope assimilation into N. inopinata cells from the supplied ¹⁵N-guanidine and ¹³C-bicarbonate, gold-sputtered filters containing cells from two replicate bottles (Treatment A, replicate A1 and A2) and a natural abundance (NA) control were glued onto antimony-doped silicon wafers (7.1 × 7.1 × 0.11 mm, Active Business Company) using superglue (Loctide). NanoSIMS measurements were performed on the NanoSIMS 50L instrument (Cameca) at the Large-Instrument Facility for Environmental and Isotope Mass Spectrometry at the University of Vienna. Before image acquisition, each analysis area was preconditioned by sequence of high and extreme low ion impact energy (EXLIE) Cs⁺ depositions as follows: high energy (16 keV) at 50 pA beam current to a fluence of 5 × 10¹⁴ ions cm⁻²; EXLIE (50 eV) at 400 pA beam current to a fluence of 5 × 10¹⁶ ions cm⁻²; high energy to an additional fluence of 2.5 × 10¹⁴ ions cm⁻². Data were acquired as multilayer image stacks by sequential scanning with a finely focused primary Cs⁺ ion beam (approximately 80 nm probe size at a 2 pA beam current) over 45 ×  45 μm² areas with 512 × 512 pixel image resolution. The primary ion beam dwell time varied between 1 ms (A1, 74 planes; NA, 50 planes) and 5 ms (A2, 21 planes) per pixel per cycle. The detectors of the multicollection assembly were positioned to enable parallel detection of ¹²C₂⁻, ¹²C¹³C⁻, ¹²C¹⁴N⁻, ¹²C¹⁵N⁻, ³¹P⁻ and ³²S⁻ secondary ions. Image data analysis was performed using the OpenMIMS ImageJ plugin (OpenMIMS v.3.0.5, ImageJ v.1.54f), where the acquired datasets were aligned, deadtime and QSA corrected, processed (for example, accumulation, stable isotope ratio calculation) and exported for visualization of ¹³C and ¹⁵N enrichment (as ¹³C and ¹⁵N atom%).

N. inopinata shotgun proteomics

For protein analysis, biomass was dissolved in lysis buffer (8 M urea, 2 M thiourea, 1 mM PMSF). Protein extraction was done by incubation at 95 °C, while shaking at 1,400 rpm for 5 min. Subsequently, the samples were treated for 3 min in an ultrasonication water bath (Elmasonic S30 H). To the cell suspension, 6.75 µl 25 mM 1,4 dithiothreitol (in 20 mM ammonium bicarbonate) was added and incubated for 1 h at 60 °C and 1,400 rpm shaking. Next, 150 µl 10 mM iodoacetamide (in 20 mM ammonium bicarbonate) was added and incubated for 30 min at 37 °C with 1,400 rpm shaking in the dark. Finally, 200 µl of 20 mM ammonium bicarbonate was added and the protein lysates were proteolytically cleaved overnight at 37 °C with trypsin (2.5 µl of 0.1 µg µl⁻¹ trypsin, Promega). The cleavage was stopped by adding 50 µl 10% formic acid. The peptide lysates were desalted using ZipTip μC18 tips (Merck Millipore). The peptide lysates were resuspended in 15 µl 0.1% formic acid and analysed using nanoliquid chromatography–MS (UltiMate 3000 RSLCnano, Dionex, Thermo Fisher Scientific). MS analyses of eluted peptide lysates were performed on the Q Exactive HF mass spectrometer (Thermo Fisher Scientific) coupled with a TriVersa NanoMate (Advion). Peptide lysates were injected onto a trapping column (Acclaim PepMap 100 C18, 3 μm, nanoViper, 75 μm × 2 cm, Thermo Fisher Scientific) with 5 μl min⁻¹ by using 98% water/2% acetonitrile with 0.5% trifluoroacetic acid, and separated on an analytical column (Acclaim PepMap 100 C18, 3 μm, nanoViper, 75 μm × 25 cm, Thermo Fisher Scientific) at a flow rate of 300 nl min⁻¹. Mobile phase was 0.1% formic acid in water (A) and 80% acetonitrile/0.08% formic acid in water (B). Full MS spectra (350–1,550 m/z) were acquired in the Orbitrap at a resolution of 120,000 with automatic gain control target value of 3 × 10⁶ ions.

Acquired LC–MS data were analysed with the Proteome Discoverer (v.2.5, Thermo Fischer Scientific) using SEQUEST HT and INFERYS Rescoring. Protein identification was performed using a database constructed from predicted proteins of N. inopinata downloaded from MicroScope⁷¹ and common contaminating proteins. Searches were conducted with the following parameters: trypsin as enzyme specificity and two missed cleavages allowed. A peptide ion tolerance of 10 ppm and an MS/MS tolerance of 0.02 Da were used. As modifications, oxidation (methionine) and carbamidomethylation (cysteine) were selected. Peptides that scored q > 1% based on a decoy database and with a peptide rank of 1 were considered identified. Differential expression of proteins was evaluated using the DEqMS⁷². Normalized spectral abundance factors were also calculated for visualization purposes only.

Heterologous expression and purification of N. inopinata guanidinase

The guanidinase gene of N. inopinata was amplified with self-designed, specific PCR primers which already contained the vector-specific linker overhangs for Gibson cloning (5′-CTGGAAGTTCTGTTCCAGGGGCCCATGGCGAAAAAGAGAACGTACC-3′ and 5′-CCCCAGAACATCAGGTTAATGGCGTCAGCGTTTCTTTCGATTGCC-3′), using high-fidelity Phusion Plus PCR Master Mix (Thermo Fisher Scientific). The purified product was cloned into the pCoofy4 (pETM44; His6-MBP) expression vector by using the GeneArt Gibson Assembly EX Cloning Kit (Invitrogen) according to the manufacturer’s protocol. The sequence of the insert was verified by Sanger sequencing.

Cultures were grown at 37 °C in auto-induction ZYP-5052 medium⁷³ supplemented with 0.5 μM, 20 μM, or 1 mM NiSO₄ for 5 h before cooling down at 4 °C for 15 min, followed by overnight expression at 20 °C. Cells were lysed in the presence of a protease inhibitor cocktail in 50 mM HEPES, 200 mM NaCl, 5% glycerol, pH 7.4 using a cell disruptor (Constant Systems) and centrifuged at 4 °C and 45,000g for 30 min. Guanidinase fused N-terminally to a His-MBP-tag was purified by affinity chromatography using MBPTrap HP columns (Cytiva). Subsequently, the His-MBP-tag was cleaved overnight with HRV-3C protease added at a mass ratio of protease to protein of 1:50. Guanidinase was further purified by MBPTrap HP columns (Cytiva), followed by size-exclusion chromatography on the HiLoad Superdex 200 26/600pg column (Cytiva) equilibrated with 20 mM HEPES, 200 mM NaCl, 5% glycerol, pH 7.4. For the 20 μM nickel in expression batch, this nickel concentration was maintained in all buffers during purification.

The sample was concentrated to around 10 mg ml⁻¹ by ultrafiltration by using Vivaspin centrifugal concentrators (Sartorius) and flash-frozen and stored at −80 °C. Protein identity and purity were analysed using SDS–PAGE.

Size-exclusion chromatography combined with multiangle light scattering

SEC-MALS was performed using a Superdex 200 increase 10/300 GL (Cytiva) operated at 20 °C on the 1260 Infinity HPLC system (Agilent Technologies) coupled to a miniDawn Treos MALS detector (Wyatt Technology). The samples were injected (80 μl at 1 mg ml⁻¹) onto a column extensively equilibrated with 20 mM HEPES, 150 mM NaCl, pH 7.4. Measurement was performed using BSA as a control. The protein concentration was measured with a RI-101 refractive index detector (Shodex) and the average molecular mass was calculated using the program Astra (Wyatt Technology). The first-order fit Zimm formalism was used for analysis of light-scattering data as a data process procedure in Astra, and a generic protein dn/dc value of 0.185 ml g⁻¹ was used for guanidinase and BSA.

Protein T _m determination

The Prometheus NT.48 instrument (NanoTemper Technologies) was used to determine the melting temperatures (T_m). Before measurements, samples of the guanidinase expressed with 0.5 µM Ni²⁺ were centrifuged for 10 min at 16,000g at 4 °C to remove any large aggregates. To identify the buffer/pH, at which the T_m of the protein was the highest, the protein was diluted using a DSF-buffer/pH screen containing different buffers and pH values⁷⁴. The capillaries were filled with 10 μl of sample and placed onto the sample holder. A temperature gradient of 1 °C min⁻¹ from 20 to 95 °C was applied and the intrinsic protein fluorescence at 330 and 350 nm was recorded. Data were processed using MoltenProt⁷⁵, where the melting temperatures from the curves were estimated using the two-state reversible unfolding model.

MS for heterologous expression experiments

Protein identity and purity were verified by intact protein mass spectrometry. A total of 40 ng of the sample was injected into a column on the LC–MS system: Dionex nano HPLC, Waters XBridge C4, flow rate 250 μl min⁻¹ step gradient 12–40–80% ACN Synapt G2Si, resolution mode. Reconstruction of average mass was done with MaxEnt1software⁷⁶.

Metal determination by ICP-MS

To quantify Ni²⁺ and Mn²⁺ concentrations of the purified guanidinase, the samples were acid-digested and measured using ICP-MS. For acid digestion, HCl 30% (Supelco Suprapur, 100318, Merck), HNO₃ 65% (3-fold subboiled, provided in analytically pure quality; 1.00441.1000, Merck) were used. H₂O₂ (31%, ROTIPURAN Ultra, HN69.1) was purchased from Carl Roth. Deionized water was produced with 0.075 µS cm⁻¹ using an Elga Veolia, PURELAB Chorus 3 RO. 180 µl of the sample was pipetted into 7 ml PFA vials (Savillex), corresponding to a total sample amount of between 2 and 2.5 mg. Subsequently, 0.5 ml HCl and 1.5 ml HNO₃ were added. After closing the vials gas tight, they were heated to 120 °C on a hot plate (Savillex). The temperature was kept constant for 12 h. After the samples had cooled down to room temperature, a total of 500 µl of H₂O₂ was added in 50 µl steps. Vials were closed again and heated at 120 °C for 12 h. Subsequently, vials were opened and the samples were brought to dryness at 120 °C. After cooling, the digestions were dissolved in 2 ml HNO₃ and brought again to dryness at 140 °C. Finally, the digestions were dissolved in 1 ml HNO₃ and 2 ml deionized water. Vials were closed and heated again at 120 °C for 12 h to ensure complete dissolution. The digestions were then quantitatively transferred to 15 ml centrifuge tubes (polypropylene, metal free) and filled up to 10 ml with deionized water. Twofold dilutions of the digestions were measured with an Agilent 7900 Single Quad ICP-MS instrument (Agilent Technologies) in no-gas mode. The operation parameters for the plasma were set to the following values: RF power: 1,550 W; RF matching, 1.80 V; sample depth, 10 mm; nebulizer gas flow, 0.8 l min⁻¹; makeup/dilution gas, 0.4 l min⁻¹. The parameters for data acquisition were as follows: acquisition mode, spectrum; sweeps/ replicate, 80; replicates, 3; integration time/mass, 0.1 sec. External calibration standards with an element concentration of 0.025 to 25 µg l⁻¹ were used for quantification. The limit of quantification values achieved for Mn²⁺ and Ni²⁺ were ≤0.17 µg l⁻¹ and ≤0.61 µg l⁻¹, respectively. The limit of detection (LOD) was ≤0.05 µg l⁻¹ for Mn²⁺ and ≤0.18 µg l⁻¹ for Ni²⁺. The measured concentrations of the diluted digestions ranged from 1.9 to 21.6 µg l⁻¹ for Mn²⁺ and from <LOD to 22.8 µg l⁻¹ for Ni²⁺. To exclude any contamination by the buffer used, this was also digested and measured. Here the concentrations ranged from <LOD to 0.07 µg l⁻¹ for Mn²⁺ and from <LOD to 0.29 µg l⁻¹ for Ni²⁺. Given that the concentrations of metals in all of the analysed samples were either significantly above the limit of quantification or below the LOD, the presence of these metals in the buffer solution was deemed not to have a relevant impact on the overall results.

Protein crystallization

For initial screening, guanidinase expressed with 0.5 μM Ni²⁺ was concentrated to 12.3 mg ml⁻¹ using the Amicon ultra centrifugal filter unit with 30 kDa MWCO and crystallized in MRC two-well crystallization plates with 50 µl of mother liquor set up using the TTPLabtech Mosquito pipetting robot system using the drop ratios 150 nl:200 nl and 200 nl:200 nl (protein:reservoir). Initial screens were performed using JCSG + HT, Index Screen, Morpheus Screen, PACT Premier screen and Crystal Screen at room temperature. Several hits were obtained from Crystal Screen and the condition F3, containing 0.5 M (NH₄)₂SO₄, 0.1 M Na₃ citrate pH 5.6 and 1 M Li₂SO₄ was used as a template for optimization screening by varying the (NH₄)₂SO₄ and Li₂SO₄ concentrations. The best crystals were obtained at 1 M (NH₄)₂SO₄ and 0.5 M to 0.7 M Li₂SO₄.

X-ray data collection, model building and refinement

Crystals were cryo-protected using 20% glycerol, flash-frozen in liquid nitrogen and diffraction datasets collected at beamline ID30B at the European Synchrotron Radiation Facility (ESRF, France) under cryogenic conditions. The collected datasets were processed with XDS and converted to the mtz file format using XDSCONV⁷⁷. The phase problem was solved with Phaser-MR⁷⁸, using its AlphaFold⁷⁹ prediction as a search model. The structure was further refined in iterative cycles of the manual model building using COOT⁸⁰ and maximum-likelihood refinement using the PHENIX software suite⁸¹. The final stages of refinement used the automated addition of hydrogens, and TLS refinement with one TLS group per chain. The models were validated with MolProbity⁸² and PDBREDO⁸³. Figures were created using PyMOL (The PyMOL Molecular Graphics System, v.2.0, Schrödinger) (Supplementary Table 12). Anomalous datasets were collected at ID30B at a wavelength of 1.8929 Å, close to the manganese anomalous scattering absorption edge, and at a wavelength of 1.4825 Å, close to the nickel anomalous scattering absorption edge. The anomalous datasets were processed as described above and the obtained mtz files were refined using the finalized model obtained from the native dataset. The anomalous maps obtained from refinement were averaged using phenix.ncs_average supplying the refined pdb structure file and the corresponding anomalous map in ccp4 file format. Averaged anomalous maps were visualized using PyMol.

Substrate-dependent oxygen uptake measurements

Whole-cell substrate oxidation kinetics were determined from oxygen-uptake measurements as previously described^34,48,84. Here oxygen-uptake measurements were performed using a microrespirometry (MR) system equipped with a four-channel MicroOptode meter (Opto-F4 UniAmp) and O₂ MicroOptodes. Real-time O₂ concentration monitoring was supported through SensorTrace Rate software (Unisense).

N. inopinata biomass was cultivated in batch cultures in the same growth medium as described above and ammonium (1 mM) or urea (0.5 mM) as sole substrates. Ammonium and guanidine were also used as co-substrates and here ammonium grown cultures (1 mM) were supplemented with guanidine (10–20 µM) around 12 h before MR experiments to induce the expression of the guanidine transporter and the guanidinase. In all cases, active N. inopinata biomass was collected (3,000g, 6 min, 20 °C) from substrate replete cultures, washed and resuspended in identical but substrate-free medium, and incubated in a recirculating water bath (>30 min, 37 °C). Samples were taken for chemical analysis to ensure the absence of detectable ammonium, nitrite, nitrate and urea before MR experiments. All chemical species were determined photometrically as described above.

MR experiments were conducted in a glass MR chamber (~2 ml) containing a glass-coated magnetic stir bar, on an MR2 stirring rack (350 rpm), in a recirculating water bath (37 °C). MR chambers were overfilled with concentrated biomass to ensure the absence of a gaseous headspace, closed with an MR injection lid and submerged in the water bath. An O₂ MicroOptode was inserted into each MR chamber and left to equilibrate (~1 h), before a stable background signal was determined (15–30 min). The background rate of oxygen depletion was subtracted from all subsequent rate determinations in each MR chamber. A Hamilton syringe (10 μl; Hamilton) was used for all substrate (ammonium, urea, guanidine) injections. Both single- and multiple-trace oxygen uptake measurements were performed.

For single-trace measurements, a single-substrate injection was performed, and the oxygen uptake was recorded until complete substrate depletion in the presence of excess O₂ (>30 μM O₂). The single-injection scheme was used to determine the molar ratio of urea and guanidine consumed per O₂. The whole-cell kinetics of N. inopinata with urea and guanidine as substrates, respectively, were performed with single-injection traces. Here a single injection of urea (~20 μM) or guanidine (~20 μM) into the MR chamber was performed. Moreover, the whole-cell kinetics of total ammonium oxidation in N. inopinata precultivated with urea (0.5 mM) or ammonium plus guanidine (1 mM and ~20 μM) was determined with single-trace measurements. Here a single injection of NH₄Cl (~25 μM) was performed. In all cases, the experiments were halted after complete substrate depletion in the presence of excess O₂ (>30 μM O₂). Nitrate was the only detectable end product in all MR chambers used for whole-cell urea and guanidine kinetic calculations.

Multiple-trace measurements were used to determine the inhibitory effect of guanidine on the rate of maximum ammonium oxidation in N. inopinata. The maximum rate of ammonium oxidation was achieved with an initial injection of NH₄Cl (250–500 μM). Subsequently, several injections of varying guanidine concentrations were performed, and discrete slopes of oxygen depletion were calculated after each injection (~2–5 min).

In all cases, for both single and multiple injections, MR chamber contents were immediately centrifuged (19,000g, 15 min, 20 °C) after the measurements and the cell pellets and supernatant were stored separately for protein and chemical analysis, respectively (−20 °C). For protein analysis, the total protein content was determined photometrically using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). The chemical analyses (ammonium, nitrite, nitrate and urea) were performed as described above.

In vitro enzyme activity assay of guanidinase

Guanidine degradation by the heterologously expressed and purified guanidinase (in the presence of different Ni²⁺ concentrations; see above) was measured at 37 °C, pH 7.5, in a buffer containing 20 mM Tris-HCl and 50 mM NaCl by measuring urea production over 25 min of incubation. The measurements of the enzyme expressed in the presence of 1 mM Ni²⁺ were done in the presence of 1 mM Ni²⁺. Kinetics were calculated from measurements at 50, 100, 250, 500, 1,000, 2,500, 5,000, 10,000, 25,000, 50,000 and 100,000 μM guanidine starting concentrations. For screening alternative substrate use, the guanidinase expressed in the presence of 1 mM Ni²⁺ was used. Then, 10 mM of methylguanidine, agmatine, arginine, creatine, guanidinobutyrate and guanidinopropionate each were incubated with the purified guanidinase enzyme or BSA at 37 °C for 30 min in three or six replicates. Guanidinase pH dependence was screened at 37 °C with incubations at pH 5.5, 6, 6.5, 7, 7.5, 8, 8.5, 9, 9.5, 10, 10.5 and 11 (set by addition of HCl or NaOH). Temperature dependence was screened at pH 7.5 with incubations at 14, 20, 28, 37, 46, 50, 55, 60, 65, 70, 80 and 90 °C. These incubations were done in triplicates.

Calculation of cellular substrate oxidation kinetic properties

The cellular kinetic properties of total ammonium, urea and guanidine oxidation were calculated from single-trace substrate-dependent oxygen uptake measurements. The substrate oxidation rates were calculated from oxygen uptake measurements using a substrate-to-oxygen consumption ratio. For total ammonium oxidation, a substrate-to-oxygen ratio of 1:2 was used. Single-trace experiments were used to confirm the substrate-to-oxygen ratio for urea (3.9 ± 0.31, n = 3) and guanidine (6.17 ± 0.24, n = 4) oxidation. Thus, for total urea oxidation and total guanidine oxidation, substrate-to-oxygen ratios of 1:4 and 1:6 were used, respectively. All substrate oxidation rates were normalized to total cellular protein in each MR chamber. In the case of total ammonium oxidation, the K_m(app) for unprotonated NH₃ was calculated based on the K_m(app) for total ammonium, incubation temperature, pH and salinity⁸⁵.

The cellular kinetic properties of total ammonium, urea and guanidine oxidation were determined with a Michaelis–Menten model fit to the data using equation (2) where V is the reaction rate (μM per mg protein per h), V_max is the maximum reaction rate (μM per mg protein per h), S is the total substrate concentration (μM), and K_m(app) is the reaction half saturation concentration (μM). An unconstrained nonlinear least-squares regression analysis was used to estimate the K_m(app) and V_max values^86,87.

$$V=\left({V}_{\max }\times \left[S\right]\right)\times {\left({K}_{{\rm{m}}\left({\rm{app}}\right)}+\left[S\right]\right)}^{-1}$$

(2)

The reaction half-inhibition concentration for total ammonium oxidation (K_i, μM), inhibition by guanidine, was also determined. The K_i was determined graphically with a Dixon plot analysis⁸⁸. Inverse total ammonium oxidation rates were plotted against total guanidine concentration. Total ammonium oxidation rates resulting in a linear trend were used for these analyses. Linear best fit trendlines from each biological replicate were used to determine intersection focal points and estimate K_i values. Furthermore, a linear regression of the percentage of the total ammonium oxidation rate at varying guanidine concentrations was used to determine the K_i.

The specific substrate affinity (a^o; litres per g wet cells per h) of ammonium, urea and guanidine oxidation was calculated using equation (3). The factor of 5.7 g wet cell weight per g of protein was used^32,48,89.

$${a}^{^\circ }=({V}_{\max }\times {5.7}^{-1})\times {{K}_{{\rm{m}}({\rm{app}})}}^{-1}$$

(3)

WWTP community structure analyses

The Ribe WWTP (GPS coordinates: 55.33, 8.74) has biological N and P removal (enhanced biological phosphorus removal) and treats municipal wastewater with 20% industrial contribution (organic loading) corresponding to a total of 25,000 person equivalents. It is designed with recirculation and has return sludge sidestream hydrolysis. It does not have primary settling. Suspended solids around the time of sampling were ~3.1 g l⁻¹. The Haderslev WWTP (GPS coordinates: 55.25, 9.51) has biological N and P removal and treats municipal wastewater with 5% industrial contribution corresponding to a total of 100,000 person equivalents. It is designed with alternating conditions and includes side stream hydrolysis. It does not have primary settling. Suspended solids around the time of sampling were ~3.2 g l⁻¹. The Klosterneuburg WWTP (GPS coordinates: 48.29, 16.34) treats municipal wastewater corresponding to a total of 50,000 person equivalents with a two-stage, biological hybrid process. Suspended solids around the time of sampling were ~4.4 g l⁻¹.

For characterizing the community structure, amplicon sequencing of the V1 to V3 regions of bacterial 16S rRNA genes was performed on samples from the Ribe and Haderslev WWTPs from the MiDAS BioBank collection. Applied PCR primers were 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 534R (5′-ATTACCGCGGCTGCTGG-3′) with barcodes and Illumina adapters (IDT). PCR reactions (25 μl) were run in duplicate for each sample, using 1× PCRBIO Ultra Mix (PCR Biosystems), 400 nM of both the forward and reverse primer, and 10 ng template DNA. The PCR conditions were 95 °C for 2 min; followed by 20 cycles of 95 °C for 20 s, 56 °C for 30 s and 72 °C for 60 s; and a final elongation step at 72 °C for 5 min. The PCR products were purified using 0.8× CleanNGS beads and eluted in 25 µl nuclease-free water. The amplicon libraries were pooled separately in equimolar concentrations, diluted to 4 nM and paired-end sequenced (2 × 300 bp) on the Illumina MiSeq sequencer using v3 chemistry (Illumina). A 20% phage PhiX control library was added to mitigate low-diversity library effects. The forward and reverse sequence reads were merged using the software usearch⁶⁰ with the -fastq_mergepairs command, filtered to remove phiX sequences using usearch -filter_phix and quality filtered using usearch -fastq_filter with parameter -fastq_maxee set to 1.0. Dereplication was performed by usearch -fastx_uniques with the option -sizeout, and amplicon sequence variants (ASVs) were resolved using the usearch -unoise3 command. An ASV table was created by mapping the quality-filtered reads to the ASVs using the usearch -otutab command with the -zotus and -strand plus options. Taxonomy was assigned to ASVs using the usearch -sintax command with the parameters -strand both and -sintax_cutoff 0.8. The absence of comammox organisms in the sample from Klosterneuburg used for guanidine degradation measurements was confirmed by PCR using comammox clade A and clade B specific primer sets³⁸. Ribe and Haderslev sample DNA in the same concentration were used as positive controls.

Substrate incubation experiments with biomass from WWTPs

Activated sludge samples were collected from the aerated tanks of the Ribe and Haderslev WWTPs on 22 October 2021. Four litres of sludge from each WWTP were scooped into large sterile plastic bottles. The samples were transported to the laboratory on the same day, and were stored in the dark at ambient temperature, that is, ranging from 4 to 10 °C, until the incubations were started. The incubations with sludge from Ribe were started on the same day as collection, and the incubation with samples from Haderslev were started on the day after collection. Before each incubation, the sludge was diluted approximately 1:4 as follows: the sludge was allowed to completely settle (1 h), then 1.5 l of the clear supernatant was gently collected to a new sterile flask without disturbing the flocs and, finally, 0.5 l of the remaining sludge was fully resuspended and added to the 1.5 l of supernatant. Well-mixed aliquots of 100 ml of the diluted sludge were then distributed to 200 ml sterile glass microcosms and covered with aluminium foil to enable gas exchange with the atmosphere. Substrates were added to the following final concentrations: guanidine, 50 µM; ammonia, 150 µM; and urea, 75 µM. These different concentrations were chosen to account for the number of amino groups among the molecules. No substrate controls were also included. The samples were incubated at 23 °C with shaking at 100 rpm. All substrate and control treatments were performed in triplicate. Microcosms were subsampled immediately before and after initial substrate additions at T0. Additional subsamples from the Ribe incubation series were taken at 3, 6, 12 and 24 h. Additional subsamples from the Haderslev incubation series were taken at 2.5, 4, 8, 16, 24 and 48 h. Subsamples for metatranscriptomics were immediately flash-frozen with liquid-N₂ and stored at −80 °C until processing. Parallel samples (1 ml) for chemical analyses were centrifuged at 12,000g for 5 min, and the supernatant was taken and frozen immediately at −80 °C.

RNA extraction and purification

Total nucleic acids were extracted from activated sludge samples (500 µl), which were thawed on ice and centrifuged (5 min, maximum speed, 4 °C), using the RNeasy PowerMicrobiome Kit (Qiagen) according to the manufacturer’s instructions with the addition of phenol:chloroform:isoamyl alcohol (25:25:1) and β-mercaptoethanol (10 μl ml⁻¹ final concentration). Bead beating (40 s at 6 m s⁻¹, four times with 2 min interval on ice) on the Fastprep FP120 (MP Biomedicals) system was performed for cell lysis instead of vortexing to improve lysis of bacteria with rigid cell walls. The total nucleic acid extracts were subjected to DNase treatment to remove DNA contaminants using the TURBO DNA-free kit (Invitrogen), and further cleaned up and concentrated with RNAclean XP beads (Beckman Coulter) before rRNA depletion. The integrity and quality of the purified total RNA were assessed on a Tapestation 2200 (Agilent) with the Agilent RNA ScreenTape (Agilent) system, and the concentration was measured using the Qubit RNA BR Assay Kit (Thermo Fisher Scientific). The average RNA integrity number was above 7.0 for all of the samples.

rRNA depletion, library preparation and sequencing

Total RNA was rRNA-depleted using the NEBNext rRNA Depletion Kit for Bacteria (New England Biolabs) with 100–300 ng total RNA as input. The NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs) was used to prepare cDNA sequencing libraries according to the manufacturer’s instructions. The libraries were pooled in equimolar concentration and 2.0 nM was sequenced on an S4 flow cell on the NovaSeq 6000 platform (Illumina) using the v1.5 300 cycle kit (Illumina, 20012863).

Identification of differentially transcribed genes

rRNA-depleted reads were adapter-screened, quality-filtered and mapped to published MAGs using bbmap v.38.92. Adapter removal and quality filtering was conducted using bbduk (ktrim=r k=21 mink=11 hdist=2 minlen=119 qtrim=r trimq=15). Metatranscriptomic reads from WWTP Ribe were mapped to genome accession GCA_016722055.1 and reads from WWTP Haderslev were mapped to genome accession GCA_016712165.1, which were the dominant comammox MAGs in the respective WWTP³⁷. Both mappings were carried out using bbmap (minid=0.98 idfilter=0.98 ambiguous=toss pairedonly=t killbadpairs=t mappedonly=t bamscript=bs.sh) to produce bamfiles. Counts for each gene were calculated using bedtools coverage (-counts) using BAM files from bbmap and GFF files downloaded from GenBank for each genome. Counts for each coding gene were examined for potential outliers, which identified MBK8278324.1 and MBK9947797.1 as potentially misannotated small RNAs and were removed from subsequent calculations. Differential transcription was evaluated by treating different timepoints as replicates and comparing treatments as factors using DESeq2⁹⁰. TPM was calculated and used for visualization purposes only.

Soil incubations

Soil was collected from a long-term fertilization experiment managed by the Austrian Agency for Health and Food Safety located at the Ritzlhof field experiment (48° 11′ 17.9′′ N 14° 15′ 16.5′′ E) in May 2023. The soil is classified as a Cambisol and has been fertilized since 1991 with solid cattle manure at an application rate of 525 kg N per ha per year⁹¹. Soil incubations were conducted in 125 ml Wheaton bottles capped with grey butyl stoppers. In brief, 30 g soil was added to each replicate (n = 3) bottle, and amended with 820 µl water, ammonium, guanidine or ammonium + guanidine for a final concentration of 30 µg N per g dry-weight soil. Soils were incubated at 23 °C and sampled at 0, 1, 2, 3, 5, 7, 12 and 27 days. Acetylene (0.02%, v/v) was used to inhibit all lithotrophic ammonia oxidation. Acetylene was supplied by adding 0.3 ml of 10% acetylene gas to sealed bottles. Bottles were opened every 1–3 days, and acetylene was resupplied. For chemical analyses, around 2 g soil was extracted in water and 2 M KCl and extracts were frozen at −20 °C until analysis. Nitrate and nitrite were quantified in water extracts and ammonium, urea, and guanidine were quantified in KCl extracts as described above. Approximately 1 g soil was sampled for molecular analysis and was frozen at −80 °C until analysis. DNA extracts were performed using the ZymoBIOMICS DNA/RNA Miniprep Kit according to the manufacturer’s instructions. AOB, AOA and comammox clade A and B amoA qPCRs were carried out as previously described^38,92,93.

Statistics

Statistical analysis on chemical, protein and qPCR data from physiological experiments and WWTP sample incubations were performed using two-tailed t-tests in SigmaPlot v.14.5 and R. No statistical methods were used to predetermine sample size, and blinding and randomization of samples were not used.

Inclusion and ethics statement

All collaborators on this study fulfil the criteria for authorship required by Nature journals, they have been included as authors as their work was essential in designing and performing the study. The roles and responsibilities were agreed among collaborators ahead of the research. No living animals or animal-derived material were used in this study, except dropped animal manure and urine. Animals were not forced to excrete.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.