Defining Culture Conditions for the Hidden Nitrite-Oxidizing Bacterium Nitrolancea

Nitrification is a key process for N-removal in engineered and natural environments, but recent findings of novel nitrifying microorganisms with surprising features revealed that our knowledge of this functional guild is still incomplete. Especially nitrite oxidation – the second step of nitrification – is catalyzed by a phylogenetically diverse bacterial group, and only recently bacteria of the phylum Chloroflexi have been identified as thermophilic nitrite-oxidizing bacteria (NOB). Among these, Nitrolancea hollandica was isolated from a laboratory-scale nitrifying bioreactor operated at 35°C with a high load of ammonium bicarbonate. However, its distribution remains cryptic as very few closely related environmental 16S rRNA gene sequences have been retrieved so far. In this study, we demonstrate how such thermophilic NOB can be enriched using modified mineral media inoculated with samples from a wastewater side-stream reactor operated at 39.5°C. Distinct cultivation conditions resulted in quick and reproducible high enrichment of two different strains of Nitrolancea, closely related to Nl. hollandica. The same cultivation approach was applied to a complex nitrite-oxidizing pre-enrichment at 42°C inoculated with biomass from a geothermal spring in the Copahue volcano area in Neuquen, Argentina. Here, an additional distinct representative of the genus Nitrolancea was obtained. This novel species had 16S rRNA and nitrite oxidoreductase alpha subunit (nxrA) gene sequence identities to Nl. hollandica of 98.5% and 97.2%, respectively. A genomic average nucleotide identity between the Argentinian strain and Nl. hollandica of 91.9% indicates that it indeed represents a distinct species. All Nitrolancea cultures formed lancet-shaped cells identical to Nl. hollandica and revealed similar physiological features, including the capability to grow at high nitrite concentrations. Growth was optimal at temperatures of 35–37°C and was strongly enhanced by ammonium supplementation. Genomic comparisons revealed that the four Nitrolancea strains share 2399 out of 3387 orthologous gene clusters and encode similar key functions. Our results define general growth conditions that enable the selective enrichment of Nitrolancea from artificial and natural environments. In most natural habitats these NOB apparently are of low abundance and their proliferation depends on the balanced presence of nitrite and ammonium, with an optimal incubation temperature of 37°C.


INTRODUCTION
As important part of the global nitrogen cycle, nitrification converts reduced nitrogen species into oxidized forms, namely ammonium (−III) to nitrite (+III) and further to nitrate (+V). In wastewater treatment plants, nitrification is essential to remove N-compounds like ammonium and nitrite, and to provide nitrate as electron acceptor for denitrification. In agricultural, but also natural systems, nitrification causes nitrogen loss due to leaching of negatively charged nitrate ions and emission of gaseous N-compounds produced by denitrifying bacteria.
Both steps of nitrification are performed by distinct groups of bacteria or archaea (Koops and Pommerening-Röser, 2005;Spieck and Bock, 2005;Stieglmeier et al., 2014), but the traditional differentiation into ammonia and nitrite oxidizers has been overturned with the discovery of comammox Nitrospira, which can perform the complete oxidation of ammonia to nitrate van Kessel et al., 2015).
Nitrite-oxidizing bacteria (NOB) from six genera belonging to four different phyla have been isolated so far. These are Nitrobacter, Nitrotoga and Nitrococcus within the Alpha-, Betaand Gammaproteobacteria, respectively Alawi et al., 2007), Nitrospira and Nitrospina, which belong to separate phyla (Ehrich et al., 1995;Lücker et al., 2013), and the moderate thermophilic NOB Nitrolancea, which is phylogenetically affiliated with the Chloroflexi (Sorokin et al., , 2014. The latter phylum also contains novel thermophilic nitrite oxidizers that have been highly enriched from hot springs at Yellowstone National Park (Spieck et al., 2020). Indications for further marine NOB have been obtained by metagenomics studies (Lüke et al., 2016;Ngugi et al., 2016;Sun et al., 2019).
Nitrolancea hollandica has been isolated from a lab-scale nitrifying bioreactor receiving a high load of ammonium bicarbonate and operating at 35 • C . It has a high tolerance against nitrite as typical for r-strategists among the NOB (Andrews and Harris, 1986;Nowka et al., 2015), which appears to correlate with the cytoplasmic localization of the key enzyme nitrite oxidoreductase (NXR) (Spieck et al., 1998;Lücker et al., 2010). The genomic organization of the nxr genes in Nitrolancea is very similar to those in Nitrobacter, Nitrococcus  and the recently described Candidatus (Ca.) Nitrocaldera (Spieck et al., 2020). In contrast, NOB containing a periplasmic NXR like Nitrospira (Spieck et al., 1996) are mostly inhibited by high substrate concentrations (Off et al., 2010). This distinct Nitrospira-type NXR is closely related to that of anaerobic ammonium-oxidizing (anammox) bacteria (Lücker et al., 2010), while the cytoplasmic Nitrobacter-type NXR is closely related to dissimilatory nitrate reductases (NAR) of nitrate-reducing microorganisms (Kirstein and Bock, 1993).
The low substrate affinity of Nl. hollandica (K m (nitrite) = 1 mM; Sorokin et al., 2012) posed the question of its natural niche. Nitrite is an intermediate of nitrification, denitrification and nitrate reduction to ammonium. It can accumulate when these processes are imbalanced, an effect that can be exaggerated by abiotic factors like pH, temperature or certain operational modes in reactor systems (Philips et al., 2002).
Temperature is a main factor shaping the community structure of nitrite-oxidizing bacteria and variations of this incubation parameter have resulted in the discovery of novel NOB (Alawi et al., 2007;Lebedeva et al., 2008). Nitrite oxidation at elevated temperature was found to be performed by Nitrospira (Lebedeva et al., 2005(Lebedeva et al., , 2011Courtens et al., 2016), a common NOB in geothermal habitats (Marks et al., 2012;Edwards et al., 2013), or novel NOB of the Chloroflexi phylum (Spieck et al., 2020).
Based on the observed properties, the natural habitat of Nl. hollandica has been proposed to feature high temperatures as well as high ammonium and nitrite concentrations, as present for instance in compost or dung . Nitrolancea-like 16S rRNA gene sequences have only rarely been detected in specialized incubations of water or soil samples (Sorokin et al., 2018). These include soil columns used for aquifer treatment (Friedman et al., 2018), 1,3-dichloropropene fumigated soil , and manure composting (Zhang et al., 2016). A closely related 16S rRNA gene sequence (JN087902) has been detected in a SHARON (single reactor system for high activity ammonium removal over nitrite) reactor in Korea running at elevated temperature, a system with conditions comparable to the source of Nl. hollandica. Furthermore, recent surveys revealed the presence of this NOB in four Danish enhanced biological phosphorus removal (EBPR) systems and an N-removal plant in China (Speirs et al., 2019). Additionally, Nitrolancea-like 16S rRNA gene sequences have been found in reactors used for partial nitrification (Yu et al., 2018(Yu et al., , 2020 and in free nitrous acid (FNA) treated wastewater (Wang et al., 2020). However, so far it is unclear what the lower temperature limit for proliferation of this moderate thermophilic NOB is and if Nitrolancea is present and active in full-scale nitrifying sludge systems especially as activated sludge does not contain ammonium or nitrite in the high concentrations apparently required by Nitrolancea to be competitive.
Apart from engineered systems, 16S rRNA gene sequences distantly related to Nitrolancea were found in warm natural environments, like the Atacama Desert (Orlando et al., 2010, EU603380) and the Kalahari (Elliott et al., 2014). These findings raise the question whether geothermal habitats might be suitable habitats for Nitrolancea, too. Therefore, in this study the geothermal area Las Máquinas in Argentina was investigated for the presence of nitrifying microorganisms. This region belongs to the Copahue-Caviahue system, which is mostly located in the Neuquen Province in Northern Patagonia and is characterized by a high biodiversity (Chiacchiarini et al., 2010;Urbieta et al., 2012Urbieta et al., , 2015aGiaveno et al., 2013;Willis Poratti et al., 2016).
In the present study, three new representatives of Nitrolancea were selectively enriched under comparable conditions, and two pure cultures were obtained. Two of the enrichment cultures originated from a full-scale centrate treatment reactor at a wastewater treatment plant (WWTP) in Hamburg, Germany, and one from the geothermal Las Máquinas area in Argentina. The obtained species were compared to the type strain Nl. hollandica with regard to their genomic, phenotypic and physiological features.

WWTP Hamburg-Dradenau, Germany
On May 22, 2017, a centrate treatment side-stream reactor (3300 m 3 ) at the full-scale WWTP Hamburg-Dradenau was sampled (sample Z2; Supplementary Figure S1A). The centrate originated from centrifuged digested sludge and contained high ammonium (38 mM) and nitrite (44 mM), and low nitrate (9 mM) concentrations, and had a temperature of 39.5 • C and a pH of 6.5-6.9 on the sampling day. In the treatment reactor, this liquid was mixed with activated sludge from the nitrification stage (approx. 30% v/v) and was supplied with 2 mg O 2 l −1 . The HRT was 1 day. The solids content and chemical oxygen demand (COD) amounted to 550 mg l −1 and 1750 mg l −1 , respectively. Subsequent to this "Store and treat" (SAT) procedure, patentregistered by the HSE (Hamburger Stadtentwässerung), the reactor effluent was applied for anammox treatment.

Las Máquinas, Argentina
The Copahue volcano (37 • 51 S, 71 • 10.2 W; 2977 m a.s.l), the Rio Agrio (pH 0.5-7) and Caviahue Lake (pH 2.1-3.7) are geographical landmarks that frame a geothermal field, which is composed of five hydrothermal sites: Las Máquinas, Las Maquinitas, Anfiteatro, Termas de Copahue (in Argentina) and Chancho-Co (in Chile). The Las Máquinas area contains many thermal fluid emission systems, consisting of fumaroles, boiling-bubbling water and mud pools with wide ranges of pH (2-7) and temperature (30-90 • C). The chemical composition of the fumaroles, which are mainly fed by meteoric water (Chiodini et al., 2015), showed that the main emitted gas is CO 2 , followed by minor concentrations of N 2 , H 2 , CH 4 , and H 2 S. CO and He concentrations are <1 ppm. Ammonium as a substrate for nitrification was detected in concentrations between 0 and 20 mM at different sampling sites of Las Máquinas, whereas nitrite and nitrate occurred between 0 and 6 µM (unpublished results).
In December 2008 (sample E2) and in March 2011 (sample A4), two mud pools in the geothermal area Las Máquinas, Argentina, were sampled. Both sampling sites were characterized by high sulfate concentrations (Supplementary Table S1). For sample E2 (Supplementary Figure S1B), red to brownish sediment was removed at a site that had a temperature of 40 • C, a pH of 2.5-3.0 and an ammonium content of 0.6 mM. The sampling site of A4 (Supplementary Figure S1C) had a temperature of 45 • C, a pH of 6.0 and ammonium was below the detection limit. Sample aliquots were transported at ambient temperature by courier to the University of Hamburg, Germany. Temperature, pH, and electric conductivity were measured in situ in each sample point using specific electrodes (Thermo Scientific Orion 5-Star Plus benchtop multifunction meter, electrode Orion 9142BN, Waltham, MA, United States). Water samples for metal analysis were filtered using 0.2 µm membrane filters (Cellulose acetate, Sartorius, Göttingen, Germany) and their concentrations (Supplementary Table S1) determined by atomic absorption spectrophotometry (AAS; Merodio and Martínez, 1985). To quantify anions, a standard turbidimetric assay for the determination of sulfate with BaCl 2 (Kolmert et al., 2000) and a titration of chloride ions with mercuric nitrate solution using diphenyl carbazide indicator were used (APHA et al., 1992).

WWTP Hamburg-Dradenau
Two batches of mineral NOB medium (see above) containing 3 mM NaNO 2 and 3 mM NH 4 Cl were inoculated (1% v/v) with the centrate sample Z2 on June 22, 2017, and incubated without agitation at 42 • C and 46 • C, respectively. A transfer was performed on October 23, 2017, and the incubation temperature was lowered to 37 • C. In the following, the cultures grew very well and nitrite concentrations supplied to the medium were continuously increased (Supplementary Table S2).

Copahue -Argentina
A 0.5 ml aliquot of sample E2 (spring water with sediment) was inoculated into a 100 ml Erlenmeyer flask with 50 ml mineral unbuffered AOB medium (Krümmel and Harms, 1982) containing 0.5 mM ammonium as substrate. When ammonium was depleted, it was replenished to 400-500 µM with sterile 5 M NH 4 Cl solution. Ammonium measurements were done with Quantofix test stripes (Macherey-Nagel, Düren, Germany). The pH was manually adjusted to approx. 7.4 with 5% (w/v) NaHCO 3 solution. For sample A4, 1 ml was inoculated in 100 ml mineral medium for NOB with 0.3 mM nitrite as substrate in a 250 ml screw cap Schott bottle. Both cultures were grown at 42 • C under static conditions and evaporation was compensated for with sterile distilled water. Both cultures were regularly transferred with 1% (v/v) inoculum into mineral NOB media containing different concentrations of ammonium and nitrite with or without addition of 0.5 mM bicarbonate, and incubated under different conditions, as detailed in Supplementary Table S2.

Physiological Experiments
Growth of Nitrolancea was investigated in batch cultures, using 100 ml mineral medium in 300 ml Erlenmeyer flasks under static conditions at 37 • C. Cultures were inoculated with 1% (v/v). To test for nitrite consumption as indication for growth, the Griess-Ilosvay spot test was used (Schmidt and Belser, 1994) or nitrite and nitrate were determined qualitatively with analytical test stripes (Merck KGaA, Germany). Nitrite concentrations for physiological experiments were measured by the HPLC technique.

Chemical Analyses
Nitrite and nitrate were determined quantitatively by HPLC via ion pair chromatography on a LiChrospher RP-18 column (125 × 4 mm; Merck) with UV detection in an automated system (Hitachi LaChrom Elite; VWR International GmbH, Darmstadt, Germany). Data acquisition and processing of nitrite and nitrate concentrations was performed with the integrated software EZChrom Elite 3.3.2.

Growth Characteristics
Nitrite consumption, nitrate production, cell numbers and protein content of cultures of A4.5.2 and Z2.3.4 were determined to calculate generation times (d −1 ) as well as protein and cell yield per mmol nitrite. Cells were counted using a Thoma cell chamber (0.02 mm depth), and protein content was quantified by fluorescence using the NanoOrange protein quantitation kit (Thermo Fisher Scientific, Waltham, MA, United States).

Activity Measurements and Oxidation Kinetics
Nitrite-dependent oxygen consumption of cells from culture of A4.5.2 (1 replicate) and culture Z2.3.4 (2 biological replicates) was measured following Nowka et al. (2015), using a microrespiration system (Unisense AS, Denmark) consisting of a 1-channel oxygen sensor amplifier (OXY-Meter), a Clark-type oxygen microsensor (OX-MR), 2 ml glass chambers with glass stoppers, glass coated magnets and a stirring system. Cells were grown at 37 • C in mineral medium as described above, with 10 mM and 35 mM (A4.5.2) or 5 mM and 15 mM (Z2.3.4) ammonium and nitrite, respectively. 12-48 h after nitrite was depleted, cells were added to the chambers, which were sealed and mounted on a stirring rack in a water bath at 37 • C. The oxygen microsensor was inserted through a capillary hole inside the glass stopper. After an initial equilibration period of 15-30 min, the measurements were started by adding nitrite from stock solutions through a second capillary hole using a syringe. Nitrite oxidation kinetics were estimated from oxygen consumption rates at varying defined nitrite concentrations using the R-package "drc" (Ritz et al., 2015), which fits Michaelis-Menten kinetic to the data with the following equation: where V is activity, V max is maximum specific activity (µmol mg protein −1 h −1 ), K m is the half-saturation constant for nitrite oxidation (µM), and [S] is concentration of nitrite (µM). For cell specific V max , the calculated maximum activity was translated to fmol NO 2 − cell −1 h −1 .

Microscopic Investigations
For electron microscopy, cells were collected by centrifugation (13,000 × g for 15 min), fixed with 2.5% (v/v) glutaraldehyde as well as 2% (w/v) osmium tetroxide, and embedded in a mixture of Spurr and acetone as described previously (Spieck and Lipski, 2011). Thin sections were stained with 2% (w/v) uranyl acetate and 2% (w/v) lead citrate. Microscopic examination was carried out with a transmission electron microscope (Zeiss model Leo 906E with a CCD camera model 794). For the visualization of whole cells, 3 µl concentrated biomass was pipetted on EM grids (300 mesh, Stork Veco B.V, Netherlands) and stained with 2% (w/v) uranyl acetate. 4 -6 -diamidino-2phenylindole (DAPI) stained cells were observed using a confocal laser scanning microscope LSM 800 (Zeiss, Jena, Germany) and Plan-Apochromat 63x and 100 × 1.4 oil objectives.

DNA Extraction
DNA for PCR and Illumina sequencing was extracted with the PowerSoil DNA isolation Kit (MO BIO Laboratories, Inc., Carlsbad, CA, United States) according to manufacturer's instructions with slight modifications except for cultures E2 and A4.1: Before cell disruption via vortexing (step 5), Proteinase K (1 mg ml −1 ), lysozyme (4 mg ml −1 ) and RNase A (1 mg ml −1 ) were added. The vortexing step was extended to 30 min at 37 • C.

Amplicon Sequencing
For selected cultures, 10 ng of genomic DNA were used for 16S rRNA gene amplicon sequencing at MR DNA (Molecular Research LP, Shallowater, TX, United States). For the cultures E2 and A4.1, amplicons were generated using the primers 341F and 785R (Herlemann et al., 2011), followed by 454 pyrosequencing (300 bp, 3000 reads per sample). For all other cultures the primers 515F and 806R (Caporaso et al., 2011) were used with subsequent Illumina MiSeq sequencing (2 × 300 bp, 20000 reads per sample). Operational taxonomic units (OTUs) were obtained from the preprocessed sequences using the de novo OTU picking workflow in qiime (v1.9.1) (Caporaso et al., 2010), and summarized and visualized by the QIIME summarize_taxa_through_plots.py script. Illumina sequences were processed, classified and summarized by MR DNA (Dowd et al., 2008). In short, paired-end sequences were joined and depleted of barcodes, followed by removal of sequences <150 bp or with ambiguous base calls. Sequences were denoised, OTUs generated and chimeras removed. OTUs were defined by clustering at 3% divergence (97% similarity).

Metagenome Assembly and Binning
Adapter removal, contaminant filtering and quality trimming of the Illumina HiSeq and MiSeq paired-end sequencing reads was performed using BBDUK (BBTOOLS version 37.76). Terminal base calls with a quality score < Q17 were trimmed and the reads with a mean quality score of ≤Q20 were discarded. The processed HiSeq reads with a length of ≥70 bp were then corrected for sequencing errors using BayesHammer (Nikolenko et al., 2013). Additional filtering parameters were applied to the MiSeq reads (k = 23, mink = 11, hdist = 1, ktrim = r, tbo = t, maxns = 0, tossjunk = t) and only reads with a length of ≥150 bp were kept. The corrected reads were assembled using metaSPAdes v3.11.1 (Nurk et al., 2017) with default settings. For the samples sequenced with HiSeq, the reads for all samples were co-assembled with iterative assemblies using kmer sizes of 21, 33, 55, 77, 99, and 127, whereas for the MiSeq reads, the assemblies were performed separately for each sample with kmer sizes of 21, 33, and 55. Trimmed HiSeq reads were mapped back to the metagenome separately for each sample using Burrows-Wheeler Aligner (BWA v0.7.17) (Li and Durbin, 2009), employing the "mem" algorithm. For samples sequenced with MiSeq, reads were mapped back to the metagenomes using the minimap2 (v2.16-r922) preset for genomic short-read mapping (Li, 2018). The sequence mapping files were handled and converted as needed using SAMtools 1.6 . The MiSeq metagenomes were manually binned using anvi'o (v5, Eren et al., 2015) and taxonomically classified using the 'classify workflow' of GTDB-Tk (version 0.3.2, Parks et al., 2018). Subsequently, the Nitrolancealike bins were annotated using Prokka (version 1.12-beta, Seemann, 2014) with the RefSeq bacterial nr protein database as a reference and the Kyoto Encyclopedia of Genes and Genomes (KEGG). The completeness and redundancy of the bins was assessed using the checkM 'lineage' workflow (Parks et al., 2015).

Phylogenetic Analyses
The average nucleotide identity (ANI) between the recovered Nitrolancea genomes and the genome of Nl. hollandica was calculated using orthoANI (Lee et al., 2016). Sequences (16S rRNA and nxrA) were aligned using the multiple sequence alignment web server T-Coffee (Notredame et al., 2000). Subsequently, maximum-likelihood trees were calculated using Mega-X (Kumar et al., 2018) with 1000 bootstraps.

Genome Comparison
The orthologous gene clusters were identified using the OrthoVenn 2 web service tool (Xu et al., 2019) with an E-Value of 1e-2 and an inflation value of 1.5.

Selective Enrichment of Nitrolancea
The enrichment process of Nitrolancea from WWTP centrate and geothermal mud pool samples was monitored by specific PCR, fluorescence in situ hybridization (FISH ; Supplementary  Table S3), and 16S rRNA gene amplicon and metagenomic sequencing. Additional details on the enrichment process can be found in the Supplementary Material.
As shown by amplicon sequencing, the microbial community of the original centrate treatment reactor sample Z2 was dominated by Proteobacteria (56%), Bacteriodetes (22%) and Firmicutes (10%). Chloroflexi were present at low amounts (2%; Supplementary Figure S2), but none of these sequences were affiliated with the genus Nitrolancea. Members of the nitrifying genera Nitrosomonas and Nitrosovibrio were detected (2.1 and 0.7%, respectively), but no known nitrite oxidizer could be identified.
Over a period of only 4 months, Nitrolancea could be selectively enriched from 1.4 to 24% (Supplementary Table S2). Different isolation attempts were performed to increase the relative abundance of Nitrolancea-like bacteria in the nitriteoxidizing cultures. For Z2, incubations with increasing substrate concentrations (up to 70 mM nitrite) revealed a very high enrichment of lancet-shaped cells (Figure 1a) and combination with the use of an optical tweezers system or serial dilution series increased the enrichment further (Figure 1b). Two consecutive dilution series with growth up to the dilution 10 −8 and 10 −6 , respectively, lead to a pure culture of Nitrolancea strain Z based on purity tests using four different complex media.
Unlike the centrate treatment reactor sample, the primary nitrite-oxidizing enrichment of Las Máquinas (A4.1) grown at 42 • C was dominated by members of the phylum Chloroflexi (45%), followed by Deinococcus-Thermus (28%) and Proteobacteria (22%) according to 16S rRNA amplicon sequencing (Supplementary Figure S2). This enrichment had been supplied with 0.3 mM of nitrite without ammonium addition. Nitrolancea abundances increased to 14.1% after the culture had been repeatedly replenished with nitrite over a period of 4 years and increased further to 44% after using high substrate concentrations (30 mM) and adding ammonium (Supplementary Table S2). Manual sorting of cells from this enrichment with the optical tweezers system generated six follow-up cultures that, however, all still contained several accompanying heterotrophic bacteria (not shown). Meanwhile, the parallel culture A4.5.2, which received 40 mM nitrite in addition to the ammonium and bicarbonate also present in A4.5.1, was nearly pure as shown by inoculation of four different complex media and FISH (Figure 1c) and only tiny colonies grew on half-strength R2A plates. These heterotrophic bacteria were lost upon another serial dilution to the step 10 −7 (A4.5.3) and the strain deemed isolated. The novel Nitrolancea strain was provisionally designated as Ca. Nitrolancea copahuensis based on 16S rRNA gene and whole-genome dissimilarity to Nl. hollandica (see below).
Notably, in an ammonia-oxidizing culture that had been inoculated with an earlier sample from Las Maquinas (E2) with 0.5 mM ammonium, but without nitrite addition, Nitrolancealike NOB were also co-enriched with unknown ammoniaoxidizing microorganisms. This culture was dominated by Chloroflexi (34%), Proteobacteria (40%) and Actinobacteria (21%), while Nitrolancea-like bacteria constituted nearly 20% of the microbial community after 3.5 years of incubation at 42 • C. Contrastingly, no Nitrolancea-like sequences could be detected in a parallel nitrite-oxidizing culture of E2, which was grown without ammonium addition at 42 • C (not shown).

Morphology and Ultrastructure
Light and transmission electron microscopy of the nitriteoxidizing enrichment cultures that had been inoculated with biomass from the Hamburg-Dradenau centrate treatment reactor and from Las Máquinas revealed Nitrolancea-like cells with characteristic tapered ends (Figure 2) similar to those of Nl. hollandica , but smaller in size ( Table 1). The lancet shape was less pronounced in A4 cultures, where most cells had an ovoid shape ( Figure 2F). Nitrolancea-like bacteria in all cultures appeared as single cells, or formed pairs and short chains. In late nitrite-oxidizing cultures, cells of Nitrolancea aggregated to roundish flocs, whereas in primary enrichments of cultures E2 and A4 tiny yellow to orange particles accumulated.
The Nitrolancea-like cells in both incubation series had the typical Gram-positive type complex cell wall, including a regular proteinaceous surface layer (Figure 2C), which is a common component of archaeal and bacterial cell walls (Rachel et al., 1997). This protein layer is easily disrupted during the embedding process (Figures 2D,F), resulting in loss of the typical cell morphology. As described for Nl. hollandica, the novel Nitrolancea strains did not possess carboxysomes or intracytoplasmic membrane systems. However, in some cells membrane invaginations as described by Sorokin et al. (2014) were observed at the cell poles (not shown).

Physiology
As also known for Nl. hollandica, cells of both Nitrolancea enrichments derived from the samples Z2 and A4 were able to grow lithoautotrophically with nitrite as energy source and CO 2 as carbon source. Nitrite was oxidized aerobically and stoichiometrically to nitrate with a minimum generation time of 53 h (culture Z2.3.4; Figure 3A) or 43 h (culture A4.5.2; Figure 4A).
In the beginning of the cultivation process, the Z2.2 culture was tested for its ammonium requirements. Growth was slow in ammonium-free medium and was clearly accelerated by the addition of 0.5 -3 mM ammonium (not shown). Consequently, all mineral NOB media were prepared with the addition of ammonium. Fascinatingly, NH 4 Cl concentrations up to 500 mM could be tolerated, although consumption of 3 mM nitrite took 4 months. Addition of 250 mM ammonium caused slight inhibition only (Figure 3C). In accordance with Nl. hollandica, the cultures derived from the centrate were able to grow in the presence of very high nitrite concentrations and completely oxidized up to 150 mM nitrite within several months. Growth occurred between 22 and 43 • C and with a delay at 19 and 45 • C. Nitrite was oxidized optimally at 37 • C ( Figure 3D). Additionally, the influence of aeration and bicarbonate supplementation on growth and activity were tested. The highest nitrite oxidation rates were obtained when cells were incubated in Erlenmeyer flasks without agitation and were further accelerated by addition of 0.5 mM NaHCO 3 ( Figure 3B).
The nitrite oxidation rates of the Nitrolancea cultures obtained from Las Máquinas (A4.4) increased with increasing substrate concentrations (1-10 mM NaNO 2 ) without a prolonged lagphase ( Figure 4B). The nitrite tolerance level was 40 mM. Nitrite was also oxidized slowly in medium without ammonium, but optimal growth depended on the addition of up to 50 mM NH 4 Cl ( Figure 4C). The ammonium tolerance limit was 500 mM. Growth of the Argentinian Nitrolancea strain was observed between 22 and 45 • C, delayed at 19 • C, and with an optimum growth temperature of 35-37 • C (Figure 4D). A comparison of key parameters between Nl. hollandica and the strains investigated here is shown in Table 1.
Nitrite oxidation kinetics were determined by microsensorbased oxygen consumption measurements in the different Nitrolancea cultures (Supplementary Figure S3). Both representatives revealed moderate substrate affinities with apparent half-saturation constants (K m ) of 0.12-0.45 mM nitrite for Nl. hollandica strain Z and 0.22 mM for Nl. copahuensis (Supplementary Table S4). The maximum specific activity (V max ) was relatively low (11.1-26.8 µmol nitrite mg  protein −1 h −1 ). Growth yields ranged from 0.17 to 0.88 mg protein mmol nitrite −1 .

Genomic Analyses
Biomass of cultures Z2.2, Z2.3.2, and A4.4 was used for metagenomic sequencing ( Table 2). The draft genomes of the three resulting Nitrolancea strains described here (Z2, Z4, A2) could be assembled into 48 to 150 contigs. These high-quality draft genomes ranged in size from 3.19 to 3.45 Mbp and had G+C contents between 62.7 and 62.9 mol%. The number of predicted coding sequences was between 3023 and 3235, including 40 to 46 tRNAs. After assembly and binning of the Nitrolancea strains, comparative genomics revealed that two highly similar strains (Z2 and Z4) were enriched from the centrate treatment reactor sample, while the strain A2 obtained from sample A4 was clearly distinct (see below). The comparison of these metagenomeassembled genomes (MAGs) with Nl. hollandica revealed 2399 conserved clusters out of in total 3387 orthologous gene clusters within the four Nitrolancea strains (Figure 5). The type strain Nl. hollandica exhibits 14 unique gene clusters that were absent in the genomes of the three novel strains, which, on the contrary, shared 131 of orthologous clusters. Additionally, the Argentinian strain A2 contained 6 unique gene clusters, while the two genomes obtained from the centrate enrichment cultures in total had 109 clusters not present in the other Nitrolancea strains.
Phylogenetic and sequence similarity-based analysis of the 16S rRNA genes of the novel Nitrolancea strains (Supplementary  Table S5, Figure 6) revealed a close affiliation with the previously described type strain Nl. hollandica (Sorokin et al., 2014). Ca. Nitrolancea copahuensis had a 16S rRNA gene sequence identity of 98.5% to Nl. hollandica, whereas the two centrate strains were almost identical to the type strain (99.9 and 100% identity). Comparison of the NxrA sequences of the novel strains and Nl. hollandica revealed high amino acid identities (Supplementary Table S5, Supplementary Figure S4). Furthermore, the draft genome sequences of the centrate strains and Nl. hollandica had high average nucleotide identity (ANI) values of 99.2 and 99.3%, respectively, which was only 91.9% for the geothermal strain (Supplementary Table S6). The two draft genomes from the centrate enrichments were almost identical and had an ANI value of 99.96%. Thus, based on 16S rRNA gene and ANI similarity, Z2 and Z4 belong to the species Nl. hollandica, while the Argentinian strain A2 had 16S rRNA and ANI identity values below the species threshold of 98.7-99.0% (Stackebrandt and Ebers, 2006) and 95% (Rodriguez-R and Konstantinidis, 2016), respectively. Consequently, this strain represents a separate species within the genus Nitrolancea.
The genomic data was furthermore mined to reconstruct the core metabolic pathways of the three novel Nitrolancea strains ( Table 3). Genes encoding all proteins of the Calvin cycle for CO 2 fixation, but not for carboxysomes have been identified. The potential for nitrite and formate oxidation was also present and the genomes contained additional non-operonal nxrA copies, as has been observed for Nl. hollandica . The A2 and Z2 genomes contained only one orphan nxrA gene, while the Z4 genome encoded 3 additional NxrA copies. All other NXR subunits were not duplicated and encoded within a highly conserved operon (Supplementary Figure S5). Furthermore, genes for a NO-forming nitrite reductase (narK) and the cytoplasmic assimilatory nitrate reductase (nasAB) were detected in the three analyzed genomes, but not for assimilatory nitrite reductases (nirA or nirBD) or dissimilatory nitrite reduction to ammonium (nrfA).
Additionally, all Nitrolancea genomes encoded an aerobictype CO dehydrogenase, cyanase and an [NiFe]-hydrogenase. Based on the genomic data, all strains have the potential to grow heterotrophically and to use thiosulfate for energy conservation. So far, none of these phenotypes could be observed in growth experiments with Nl. hollandica . Within the sulfur metabolism, the assimilatory sulfate reduction pathway involving sulfate adenylyltransferase (sat), adenylyl-sulfate kinase (cysC), phosphoadenosine phosphosulfate reductase (cysH) and ferredoxin-dependent sulfite reductase (sir) could be identified in the three analyzed genomes, along with genes encoding thiosulfate sulfurtransferases (glpE, sseA) and a putative sulfite oxidase (sorA).

DISCUSSION
The first isolate of Nitrolancea was only described in 2012 . Growth depended on the use of a modified NOB medium (Spieck and Lipski, 2011), as Nl. hollandica grew optimally in mineral medium containing high substrate concentrations of 20 to 50 mM KNO 2 and supplementation of 5 mM NH 4 HCO 3 . Unexpectedly for a nitrite-oxidizing organism, Nl. hollandica depends on ammonium as N-source and accordingly its genome did not encode any enzymes for nitrite reduction to ammonium . This absence of assimilatory nitrite reduction pathways was recently also observed in other thermophilic NOB within the phylum Chloroflexi (Spieck et al., 2020), and some Nitrotoga species are apparently dependent on ammonium addition, too (Ishii et al., 2017;Wegen et al., 2019). In contrast, the early Nitrolancea enrichments obtained in this study were able to oxidize nitrite also without ammonium addition, although no assimilatory nitrite reductase genes were detected in their genomes. However, it is likely that these Nitrolancea strains were provided with ammonium or organic N-sources by the accompanying heterotrophs present in the cultures, as the ability for growth without ammonium addition was lost upon increased purity of the cultures.
The optimal growth conditions for the novel Nitrolancea strains were assessed by testing different incubation temperatures, as well as nitrite and ammonium concentrations. Although Nitrolancea has been described as NOB, that requires high substrate concentrations, early enrichments from Las Máquinas were done at very low nitrite availability (Supplementary Table S2). Nevertheless, once an active culture was obtained, an increase of substrate concentrations was observed to be beneficial for the selective enrichment of Nitrolancea-like NOB. High ammonium concentrations can also enhance proliferation of Nitrolancea, as was observed in culture A4.4. However, we cannot exclude that this effect was caused by growth stimulation of accompanying bacteria. After initial cultivation at 42 • C, reducing the incubation temperatures to the apparent optimum of 37 • C led to a reliable enrichment of Nitrolancea-like cells from both sample locations (Supplementary Table S2). Intriguingly, this cultivation strategy resulted in the successful enrichment of two novel Nl. hollandica strains from the centrate treatment reactor, even though no Nitrolancea-like 16S rRNA gene sequences were detected in the original sample. This finding indicates that these NOB are more widely distributed than previously assumed, albeit at very low numbers. The minimum temperature for growth for Nitrolancea derived from the centrate reactor as well as from the geothermal mud pool was 19 • C, but related 16S rRNA gene sequences were also found in a bioreactor system operated at 12 • C (Yu et al., 2018). Therefore, the environmental distribution of these NOB seems not to be restricted to thermophilic conditions.  Fast and reproducible initial enrichment of Nitrolancea spp. was achieved by designing a standard medium consisting of a mineral salt solution supplemented with 3 mM NaNO 2 , 3 mM NH 4 Cl and, as observed in later cultivation approaches, 0.5 mM NaHCO 3 . The positive effect of bicarbonate probably is due to resulting elevated CO 2 concentrations within the cell, which might reduce the reaction of ribulose 1,5-bisphosphate carboxylase/oxygenase (RuBisCO) with the competing substrate O 2 and thus increase carbon fixation efficiency in the absence of other CO 2 -concentrating mechanisms like carboxysomes (Price et al., 2008). In this study, a preference for a reduced oxygen supply (Figure 3B) of N. hollandica strain Z was found and culturing was done without agitation. Accordingly, Nitrolancea-like sequences have been detected in a reactor treating synthetic low-strength wastewater at low dissolved oxygen contents (Yu et al., 2020).
The proliferation of the different Nitrolancea strains at nitrite concentrations between 3 and 40 and even 70 mM, respectively, matches the extremely high half-saturation constant (K m ) for nitrite determined for Nl. hollandica . The K m of this strain was found to be 1 mM nitrite, which corresponds to the lowest substrate affinity of any NOB determined so far and is in line with the high substrate concentrations present in the bioreactor Nl. hollandica was isolated from. The Nitrolancea cultures investigated here revealed a higher affinity for nitrite, although there was a great variance between values for strain Z and only one measurement was done for Nl. copahuensis. Nevertheless, our preliminary results suggest that the K m value of Nitrolancea is in the range of the different Nitrobacter species (Nowka et al., 2015). This higher substrate affinity fits with the presence of Nitrolancea in the geothermal area of Las Máquinas, where nitrite is rarely detected as usual for most natural systems. Nevertheless, high substrate concentrations are necessary for optimal growth of Nitrolancea and nitrite was oxidized slowly when present in low concentrations (Figures 4B,D).
Comparing the genomes and NxrA gene sequences from the different enrichments, it has become obvious that distinct representatives of Nitrolancea grew depending on the respective medium (for instance 3 mM vs. 60 mM nitrite for strains Z2 and Z4). Similarly, the representatives of Nitrolancea enriched from the Las Máquinas samples differed in the initial enrichments E2 and A4.4 in AOB and NOB medium, respectively (Supplementary Figure S4). The nxrA gene within the metagenome 40 of the culture E2 is non-operonal, whereas in close vicinity to the nxrA gene of the metagenome 35 a cytochrome c was found. Therefore, the initial samples must have contained a microdiversity of Nitrolancea strains, which selectively proliferate once their special demands were met.
Nitrolancea-like 16S rRNA gene sequences have only occasionally been detected in full-scale nitrifying activated sludge systems so far (Speirs et al., 2019). The Nitrolancea-typical marker fatty acid 12-methyl octadecanoic acid (18:0 12methyl)  was measured in low amounts (0.2% of total fatty acids) in activated sludge of the Hamburg-Dradenau WWTP by Kruse et al. (2013). The abundance of 12-methyl octadecanoic acid increased slightly to 0.7% after incubations with nitrite at 28 • C and 32 • C, which confirms the preference of Nitrolancea for elevated temperatures. However, similar glycolipids are also characteristic for other thermophilic Chloroflexi, like for instance Thermomicrobium, which has also been detected in WWTPs based on 16S rRNA gene sequences (Spieck et al., 2020). Therefore, it is still speculative if Nitrolancea also occurs in activated sludge, which was used as inoculum for the centrate reactor.
The geothermal area in Las Máquinas apparently offers beneficial conditions for Nitrolancea-like NOB, as revealed by our 16S rRNA amplicon data (Supplementary Figure S2, Supplementary Table S2) although closely related sequences have not been detected in this area previously (Urbieta et al., 2015a). Elevated ammonium or nitrite concentrations could not be measured at the sampling sites, thus posing the question if alternative substrates might have enabled proliferation of Nitrolancea. Some NOB are metabolically versatile and can use simple organic carbon compounds like formate, hydrogen, and even ammonia as additional energy sources (Koch et al., , 2015Daims et al., 2015;van Kessel et al., 2015). Furthermore, different NOB possess the genetic potential to use various sulfur compounds (Starkenburg et al., 2008;Lücker et al., 2013;Boddicker and Mosier, 2018;Kitzinger et al., 2018;Palomo et al., 2018) and the oxidation of sulfide has been demonstrated for Nitrococcus mobilis (Füssel et al., 2017). In the Las Máquinas area, a highly active sulfur cycle is  present (Giaveno et al., 2013;Urbieta et al., 2014) and the ability of Nitrolancea species to oxidize thiosulfate should be tested in further studies. The ability for cyanate degradation might provide an alternative source to generate ammonia for N-assimilation.
In conclusion, we physiologically and genomically characterized three novel representatives affiliated with the genus Nitrolancea, which we enriched from a centrate treatment reactor at the WWTP Hamburg-Dradenau and the geothermal area Las Máquinas in Neuquen-Argentina. One of these strains represents a new species within the genus Nitrolancea. We provisionally refer to this strain as, Candidatus Nitrolancea copahuensis' (co.pa.hu.en'sis. N.L.fem.adj. copahuensis pertaining to the Copahue region in North-Patagonia, Argentina).
The strain is a mesophilic, aerobic, chemolithoautotrophic bacterium, which stoichiometrically converts nitrite to nitrate and uses carbon dioxide as carbon source. Forms short rods with tapered ends; Gram-positive. Requires ammonium supplementation and tolerates high nitrite and ammonium concentrations. Optimum growth temperature is 35-37 • C. DNA G+C content is 62.7 mol%. Originated from a hot spring in the geothermal field Las Máquinas, Neuquen-Argentina.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available in the European Nucleotide Archive (ENA) under accession numbers ERS4399870, ERS4399871, ERS4399872, ERS4647270-74, ERS4648393-96, and LR794103-5. Strain availability will be performed upon request.

ACKNOWLEDGMENTS
We thank the Government of the Province of Neuquen for allowing sampling in the protected natural area Copahue geothermal park and Edgardo Donati and Wolfgang Sand for providing samples from Las Máquinas. HSE is appreciated for the support in getting samples from Hamburg-Dradenau. Elke Woelken is acknowledged for excellent technical help in electron microscopy and Theo van Alen for sequencing. We are grateful that Katja Wendt, Kristina Detzel, Katrin Petersen as well as Laura Nissen did initial cultivation and laboratory work. We also thank Malik Alawi, Lia Burkhardt, and Kerstin Reumann for sequencing assistance.