Differential contribution of nitrifying prokaryotes to groundwater nitrification

The ecophysiology of complete ammonia-oxidizing bacteria (CMX) of the genus Nitrospira and their widespread occurrence in groundwater suggests that CMX bacteria have a competitive advantage over ammonia-oxidizing bacteria (AOB) and archaea (AOA) in these environments. However, the specific contribution of their activity to nitrification processes has remained unclear. We aimed to disentangle the contribution of CMX, AOA and AOB to nitrification and to identify the environmental drivers of their niche differentiation at different levels of ammonium and oxygen in oligotrophic carbonate rock aquifers. CMX ammonia monooxygenase sub-unit A (amoA) genes accounted on average for 16 to 75% of the total groundwater amoA genes detected. Nitrification rates were positively correlated to CMX clade A associated phylotypes and AOB affiliated with Nitrosomonas ureae. Short-term incubations amended with the nitrification inhibitors allylthiourea and chlorate suggested that AOB contributed a large fraction to overall ammonia oxidation, while metaproteomics analysis confirmed an active role of CMX in both ammonia and nitrite oxidation. Ecophysiological niche differentiation of CMX clades A and B, AOB and AOA was linked to their requirements for ammonium, oxygen tolerance, and metabolic versatility. Our results demonstrate that despite numerical predominance of CMX, the first step of nitrification in oligotrophic groundwater appears to be primarily governed by AOB. Higher growth yields at lower ammonia turnover rates and energy derived from nitrite oxidation most likely enable CMX to maintain consistently high populations.


N rate incubations and IRMS analysis
Groundwater was sampled between 2018 and 2020 covering nine wells for rate measurements.Nitrification rate measurements of were conducted at a final concentration of 50 µM as ( 15 NH4)2SO4.Samples were incubated at 15°C in the dark for five days without shaking.Subsamples of 10 ml were taken and replaced by equal amount of N2 at the start of the experiment and after 12, 24, 48, 70 and 120 hours, followed by filtration through 0.2 µm filters and storage at -20°C until isotopic ratio mass spectrometry (IRMS) analyses.Nitrification rates were calculated based on the formation of 15 NO2 -+ 15 NO3 -during incubation with 15 NH4 + .15 NO2 -and 15 NO3 -were transformed to N2 for analysis via cadmium reduction followed by a sulfamic acid addition [1,2].The N2 produced ( 15 N 15 N and 14 N 15 N) was analyzed on a gas chromatography IRMS as previously described [3].Single rates were determined from the slope of the linear regression of 15 N produced over time and corrected for the fraction of the NH4 + pool labelled in the initial substrate pool.Rates were assumed to be significantly different from zero, if the linear 15 N increase was significant (p < 0.05) (see Dataset S1).Abiotic 15 N transformation was excluded since filtered controls did not show any significant production of 15 N over the incubation period.The rate detection limit was 2.1 nmol N L -1 d -1 which was estimated from the median of the standard error of the slope from significant rates multiplied by the t value for p = 0.05 [4].This value estimates the magnitude of rates that may potentially escape detection.As the significance of rates is tested individually for each incubation, detectable rates may in some cases be below this detection limit.

Determination of inorganic nitrogen from mesocoms containing nitrification inhibitors
Subsamples for colorimetry of inorganic nitrogen compounds were taken at the onset and every two to three days until the end of incubation to monitor changes of the nitrogen chemistry.NO3 -from control and allylthiourea treated samples was measured using the sodium salicylate method [5].As chlorate impaired the later NO3 -detection method [6], NO3 - from chlorate treated samples was determined via ion chromatography [6].NO2 -and NH4 + were measured as indicated before.

cDNA-synthesis and quantitative PCR reactions
After extraction of total RNA from groundwater filters, 4 µl DNase treated RNA was used for reverse transcription in a total 15 µl reaction volume and 1 µl Random Primers following the manufacturer's protocol.Additional reactions without reverse transcriptase for each sample and a reaction with RT-PCR grade water instead of RNA were used to check for potential contamination with residual DNA in subsequent PCR steps.PCR using bacterial 16S rRNA primer pair 341f/785r [7] was performed to verify the presence of cDNA in the RT reactions and the absence of residual genomic DNA in the negative control reactions.The integrity of PCR products was checked by agarose gel electrophoresis.
Reactions for quantitative PCR were performed in 25 µl total volume with 12.5 µl Brilliant II SYBR Green qPCR Master Mix (Agilent), 0.4 µM of each primer and 5 µl of prediluted DNA or cDNA.Cycler conditions for amoA genes of AOB and AOA, and nxrB genes are described in [8] and [9], respectively.CMX amoA amplification was conducted in 45 cycles of 30 sec at 95°C, 45 sec at 52°C and 1 min at 72°C.Serial dilutions of standards used for gene quantification originated from plasmid DNA containing the respective gene as insert.Standard curves were linear from 5 x 10 8 to 50 copies per reaction with R² > 0.99 and efficiencies ranging from 80 to 95%.
Transcript/ gene ratios of CMX, AOB and AOA were calculated by dividing the amoA gene abundances L -1 by the amoA transcripts L -1 .To estimate the proportion of CMX Nitrospira to canonical Nitrospira, the percentages of amoA and nxrB abundance were calculated from the sum of both gene abundances at each groundwater well.Nitrospira nxrB abundance was divided by 1.5 to account for a higher copy number estimated from nxrB copy numbers from Hainich Nitrospira MAGs.
Amplicon sequencing of amoA genes and sequence analysis PCR for amplicon library construction was performed in 50 µl reactions including 2 µl of DNA, 0.4 µM of each primer, 1 µg/ µl BSA and 2X HotStartTaq Master Mix (Qiagen).PCR conditions for AOB amoA were 35 cycles for 45 sec at 94°C, 30 sec at 57°C and 45 sec + 1 per cycle at 72°C and final 10 min at 72°C.AOA amoA was amplified using 35 cycles for 45 sec at 94°C, 60 sec at 53°C and 60 sec at 72°C, and final 10 min at 72°C.For amplification of CMX amoA the following protocol was used: 40 cycles for 30 sec at 95°C, 30 sec at 53°C and 1 min at 72°C ending with 10 min at 72°C [10].The size and integrity of all PCR products was validated by agarose gel electrophoresis.Paired-end 2 x 300 bp reads were sequenced on an Illumina MiSeq instrument using v3 chemistry [11].
Primer sequences of raw amoA reads were trimmed with Trim Galore (https://github.com/FelixKrueger/TrimGalore).Sequences were quality filtered with DADA2 [12] using filterAndTrim default settings and truncLen of 200, 229 and 260 bp for CMX, AOB and AOA, respectively.Subsequent steps were performed in Mothur [13] including merging of paired reads and dereplication, followed by chimera search using the uchime algorithm implemented in Mothur v.1.46.1 [14].Reads were translated to amino acid sequences with transeq incorporated in Emboss [15].Sequences containing stop codons were excluded from further analysis.The remaining reads were assigned to OTUs using vsearch at 95% for CMX, 95% for AOB [16] and 96% for AOA [17] sequence identity.OTUs with < 10 reads were removed before classification.The taxonomical assignment of amoA sequences was done with BLASTx [18] using a custom amoA database on amino acid level and identification of closest relatives with best hits.AmoA sequences of OTUs accounting for more than 2% of all reads were aligned using Clustal Omega [19].Phylogenetic tree construction of deduced AmoA amino acid sequences was performed using maximum likelihood with the JTT+CAT evolutionary model and 1000 bootstraps in FastTree [20] and trees were visualized in iTOL [21].

Inferring metabolic capacities of nitrifying prokaryotes from metagenome assembled genomes
In addition to sorted CMX Nitrospira MAGs, other nitrifier MAGs from Overholt et al. [22] which employed either amoA for AOB and AOA and nxrA or nxrB for NOB affiliated genomes were included in the analysis.Genes encoding for key pathways involved in nitrogen and alternative energy metabolism and uptake, as well as genes for carbon fixation and cell defense mechanisms against reactive oxygen species were examined.

Protein extraction and mass spectrometric analysis
For metaproteomic analysis, tryptic peptides were dissolved in 0.1% formic acid (v:v) and subjected to LC-MS/MS analysis on a Q Exactive HF instrument (Thermo Fisher Scientific, Waltham, MA, USA) in LC chip coupling mode, equipped with a TriVersa NanoMate source (Advion Ltd., Ithaca, NY, USA).Raw mass spectral data was analysed with the Sequest HT search algorithm in Proteome Discoverer (v1.4.1.14,Thermo Fisher Scientific, Waltham, MA, USA).The following parameters were used: enzyme specificity was set to trypsin with two missed cleavages allowed, carbamidomethylation (cysteine) was set as static modification and oxidation (methionine) as dynamic modification, and peptide ion and MS/MS tolerances were set to 10 ppm and 0.05 Da, respectively.Peptides were considered identified upon scoring a q-value < 1% based on a decoy database and obtaining a peptide rank of 1.

Statistical analysis
Statistical analysis was performed in R 4.12 (R Core Team, 2020) using packages FSA [24], ComplexHeatmap [25] and vegan [26].Shapiro-Wilk test was applied to test for normal distribution of data before conducting either parametric or non-parametric tests.Significant differences of groundwater hydrochemical properties and gene abundances between groundwater wells were determined using Kruskal-Wallis rank sum test and Dunn's multiple comparison post-hoc test with package FSA [24].Correlations between hydrochemical properties, total amoA abundances and absolute amoA read abundances of OTUs were calculated using Spearman's rank correlation and heatmap construction with ComplexHeatmap package [25].Ordination analyses were performed with vegan [26].For redundancy analysis, the absolute amoA read abundances of OTUs were standardized using Hellinger method [27] and hydrochemical parameters were log-transformed before performing rda function from vegan.Nonmetric multidimensional scaling using Bray-Curtis distance matrix was performed to show dissimilarity of ammonia oxidizing communities between groundwater from different wells.Anosim function from vegan was used to verify the significance of observed differentiation.

Supplemental Datasets
Dataset S1: Calculation of groundwater nitrification rates and inhibitor approaches using linear regression and determination of limit of detection.

Figure S 1
Figure S 1 Display of A) amoA transcripts L -1 , B) amoA transcript/ gene ratios and C) CMX amoA and Nitrospira-like nxrB genes L -1 across groundwater sites (n ≥ 3).Dots represent single sample measurements (n.a.= no RNA sample due to low sample volume), the crossbar shows the mean of all samples and the error bars display the variation from the median.Outliers are shown beyond the error bars.Lowercase letter code indicates significant differences based on Dunn's multiple comparison test (* = p < 0.05, ** = p < 0.01, ns = not significant).D) Correlation between CMX amoA genes and Nitrospira-like NOB nxrB genes across all groundwater wells based on Spearman's rank correlation.

Dataset S2 :
Abundances of amoA, nxrB genes and transcripts across the groundwater wells and incubation experiments determined by qPCR.Dataset S3: Taxonomic classification of ammonia oxidizer OTUs generated by amoA targeted amplicon sequencing of groundwater samples.Dataset S4: Specifications of Hainich nitrifier MAGs and reference genomes.