Growth of complete ammonia oxidizers on guanidine

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:

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.

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.

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.