Quantification of archaea-driven freshwater nitrification from single cell to ecosystem levels

Deep oligotrophic lakes sustain large populations of the class Nitrososphaeria (Thaumarchaeota) in their hypolimnion. They are thought to be the key ammonia oxidizers in this habitat, but their impact on N-cycling in lakes has rarely been quantified. We followed this archaeal population in one of Europe’s largest lakes, Lake Constance, for two consecutive years using metagenomics and metatranscriptomics combined with stable isotope-based activity measurements. An abundant (8–39% of picoplankton) and transcriptionally active archaeal ecotype dominated the nitrifying community. It represented a freshwater-specific species present in major inland water bodies, for which we propose the name “Candidatus Nitrosopumilus limneticus”. Its biomass corresponded to 12% of carbon stored in phytoplankton over the year´s cycle. Ca. N. limneticus populations incorporated significantly more ammonium than most other microorganisms in the hypolimnion and were driving potential ammonia oxidation rates of 6.0 ± 0.9 nmol l‒1 d‒1, corresponding to potential cell-specific rates of 0.21 ± 0.11 fmol cell–1 d–1. At the ecosystem level, this translates to a maximum capacity of archaea-driven nitrification of 1.76 × 109 g N-ammonia per year or 11% of N-biomass produced annually by phytoplankton. We show that ammonia-oxidizing archaea play an equally important role in the nitrogen cycle of deep oligotrophic lakes as their counterparts in marine ecosystems.

Chlorophyll a was sampled from 22 depths over a gradient of 0−60 m and analyzed spectrophotometrically after extraction in hot ethanol as described previously [2], but without correcting for pheopigments. For RNA extraction, 0.22 µm-filters (142 mm diameter) filters were cut with a sterilized scissor into thirds and extracted as described above except for the following modifications: filters were extracted in extraction buffer (50 mM sodium acetate and 10 mM EDTA, pH 4.2) with 0.025% SLS (Sigma-Aldrich) and phenol-chloroform-isoamylalcohol 25:24:1 (Roti-Aqua-P/C/I 4.5-5.0, Carl Roth GmbH).

DNA and RNA extraction
Washing of the aqueous phase with chloroform-isoamylalcohol 24:1 was done in the presence of 0.1 volume 3 M sodium acetate. RNA was finally precipitated with 1 volume ice-cold isopropanol in the presence of 1 µl glycogen (35 mg ml -1 , RNase-free, VWR), washed as stated above, and eluted in nuclease-free water (MP biomedicals). DNA was digested with the TURBO DNA-free™ kit (Thermo Fisher Scientific) and RNA samples were stored afterwards at -80°C until sequencing.

Determination of AOA abundance by CARD-FISH
Before CARD-FISH, cells on the filter sections were immobilized by embedding in 0.1% low-gelling agarose (Metaphor). CARD-FISH was performed using a specific HRP-labeled oligonucleotide probe for Nitrososphaeria (HRP-labeled Thaum726 [GCTTTCATCCCTCACCGTC] and unlabeled competitors [Thaum726_compA: GCTTTCGTCCCTCACCGTC, Thaum726_compB: GCTTTCATCCCTCACTGTC]) [3,4] as described previously [5]. Negative controls using NonEUB [6] to exclude unspecific signals were performed according to a defined protocol [7]. Briefly, endogenous peroxidases were inactivated by incubation in 0.01 M HCl for 10 min. Cells were permeabilized by HCl (0.1 M HCl for 1 min) and subsequently washed with MilliQ water. Filter pieces were hybridized with HRP probes and the respective competitor probes at 25% formamide concentration at 46°C for up to 3 h. After a 5 min washing step at 48°C and HRP probe equilibration in 1× PBS for 5 to 15 min, signal amplification was performed with OregonGreen488-labeled tyramides at 48°C for 30 min. Cells were counterstained with 4',6-diamidino-2-phenylindole (DAPI, 10 µg ml −1 , 5 min at room temperature). Filter sections were mounted onto glass slides, and embedded in a 4:1 mixture of Citifluor AF1 and Vectashield (Citifluor Ltd, London, UK; Vector Laboratories, Burlingame, CA, USA). Nitrososphaeria and DAPI signals were counted on an Axiophot or Axioplan 2 microscope (Zeiss, Germany).
For metatranscriptome sequencing, messenger RNA (mRNA) was enriched from total RNA extracts by depleting ribosomal RNA with the Ribo-off rRNA Depletion Kit for bacteria (Vazyme, Nanjing, China).
Thereafter, the sequencing library was prepared with the TruSeq® Stranded mRNA Library Prep (Illumina) and sequenced on a NextSeq500 sequencer using 2 × 150 bp. The sequencing depth ranged between 0.7−1.9 × 10 8 reads per metatranscriptome with an average of 1.2 × 10 8 reads (10−27 Gbp, average 17 Gbp). Raw reads were quality filtered and trimmed using trimmomatic v0.38 [22] and the fastx toolkit v0.0.14 (hannonlab.cshl.edu/fastx_toolkit). Residual ribosomal reads were removed using SortMeRNA v2.1b [23]. Curated metatranscriptome reads were mapped against MAGs and contigs of interest using bowtie2 v2.30 [24] to determine the transcription levels of individual genes. Subsequent network analysis of co-transcribed genes of MAG AOA-LC4 was based on genes with transcription values higher than the median transcription of all AOA-LC4 genes (35.35 FPKM). Transcription values were correlated pairwise using Spearman correlation; only significant (FDR-corrected p-value < 0.05) correlations with a correlation coefficient of r S > 0.8 were further processed. For the final network construction, only genes, which correlated in their transcriptional response to at least two of the either amoA, amoB or amoC were taken into account. The network was created with the R package igraph v1.2.5 [25] and refined with cytoscape v3.8.1 [26].
A gene-centric analysis was performed to gain an overview of all ammonia transporter (amt) and urea transporter (dur3) or urea transport system substrate-binding protein (urtA) gene sequences in Lake Constance. Therefore, single assemblies of all 9 metagenomes were annotated with DRAM [27].

Phylogenetic analyses
Phylogenomic analyses of the nitrifying MAGs were performed on the basis of concatenated amino acid alignments of 122 translated archaeal or 120 bacterial single copy genes [

Phylogenetic analysis of bacterial ammonia oxidizers
Phylogenomic maximum likelihood tree construction revealed that MAG AOB-LC263 formed a stable cluster with other freshwater MAGs, which represented a sister clade to bona fide Nitrosospira species (Fig. S3). This was corroborated by phylogenetic analysis of its amoA gene (Fig. S5). Closest relatives of AOB-LC263 were MAGs retrieved from Lake Baikal, the Great Lakes, and Lake Biwa. Based on the currently proposed species and genus delineation thresholds of ca. <95% ANI and <65% AAI, respectively [37], AOB-LC263 would represent a new species and genus within the Nitrosomonadaceae (Fig. S4). The phylogenetic affiliation of contigs AOB-LC199628 and AOB-LC368213 could only be assessed based on their amoA genes. Contig AOB-LC199628 clustered in a stable clade consisting of environmental sequences that was distinct from the AOB-LC263 and Nitrosospira clusters. Its closest cultured relative was Nitrosospira sp. Np39-19 as based on 78.3% amoA nucleotide identity. Contig AOB-LC368213 clustered within sequences affiliated with Nitrosomonas species with its closest cultured relative being Nitrosomonas ureae Nm10 with 89.9% amoA nucleotide identity (Fig. S5)

Phylogenetic analysis of nitrite-oxidizing bacteria and comammox bacteria
Phylogenomic maximum likelihood tree construction placed the two MAGs NOB-LC29 and NOB-LC32 into Nitrospira lineage II but outside the intra-lineage comammox clades A and B (Fig. S6). This was corroborated by phylogenetic analysis of their nxrB genes (Fig. S8). The two MAGs shared an ANI and AAI of 85% and 85%, respectively, indicating that they represent two separate species within the genus Nitrospira (Fig. S7). Phylogenomic tree construction of MAG COM-LC224 placed it into comammox clade B within Nitrospira lineage II (Fig. S6), which was corroborated by phylogenetic placement of its single genes nxrB and amoA ( Fig. S8 and S9). Interestingly, COM-LC224 showed <65% AAI to both type and Candidatus species within the genus Nitrospira (Fig. S7), but at the same time exhibited AAI values of >65% with freshwater Nitrospira like NOB-LC29 and NOB-LC32. Without further data, its affiliation at the taxonomic rank of a genus is currently inconclusive.

30.
Parks Supplementary Tables   Table S1. Overview of metagenome-assembled genomes (MAGs) and contigs related to the nitrifying community in the hypolimnion of Lake Constance.