Urea assimilation and oxidation supports the activity of a phylogenetically diverse microbial community in the dark ocean

14 Urea is hypothesized to be an important source of nitrogen and chemical energy to 15 microorganisms in the deep sea; however, direct evidence for urea use below the epipelagic 16 ocean is lacking. Here, we explore urea utilization from 50 to 4000 meters depth in the 17 northeastern Pacific Ocean using metagenomics, nitrification rates, and single-cell stable-18 isotope-uptake measurements with nanoscale secondary ion mass spectrometry (nanoSIMS). We 19 find that the majority (>60%) of active cells across all samples assimilated urea-derived N, and 20 that cell-specific nitrogen-incorporation rates from urea were higher than that from ammonium. 21 Both urea concentrations and assimilation rates relative to ammonium generally increased below 22 the euphotic zone. We detected ammonia- and urea-based nitrification at all depths at one of two 23 sites analyzed, demonstrating their potential to support chemoautotrophy in the mesopelagic and 24 bathypelagic regions. Using newly generated metagenomes we find that the ureC gene, encoding 25 the catalytic subunit of urease, is found within 39% of deep-sea cells in this region, including the 26 Nitrosophaerota (likely for nitrification) as well as thirteen other phyla such as Proteobacteria, 27 Verrucomicrobia, Plantomycetota, Nitrospinota, and Chloroflexota (likely for assimilation). 28 Analysis of public metagenomes revealed ureC within 10-46% of deep-sea cells around the 29 world, with higher prevalance below the photic zone, suggesting urea is widely available to the 30 deep-sea microbiome globally. Our results demonstrate that urea is a nitrogen source to abundant 31 and diverse microorganisms in the dark ocean, as well as a significant contributor to deep-sea 32 nitrification and therefore fuel for chemoautotrophy. 33 34

urea; nitrogen; nitrification; chemoautotrophy; ureC; metagenomics, nanoSIMS; 35 deep sea; mesopelagic; bathypelagic 36 37 38 39 40 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 2

41 Nitrogen (N) is an essential nutrient for all living organisms [1] , however, bioaccesible N 42 can be a scarce and therefore limiting element in marine environments [2] . Ammonium and 43 nitrate are among the most important forms of nitrogen in the oceans. While ammonium is 44 assimilable by most microorganisms, nitrate must be enzymatically reduced to ammonium before 45 assimilation, incurring an energetic cost and excluding organisms without this machinery [3, 4] . 46 Ammonium is therefore typically preferred, and is generally scarce (low nM range)[5] while 47 nitrate concentrations can be orders of magnitude higher, especially at depth[6, 7]. Some 48 microorganisms also use inorganic nitrogen as electron acceptors or donors in respiratory 49 processes, increasing the demand for nitrogen in the environment. For example, ammonia can be 50 oxidized to nitrite by chemoautotrophic ammonia-oxidizing archaea (AOA; i.e., Nitrososphaeria, 51 syn., Thaumarchaeota) and ammonia-oxidizing bacteria (AOB) [8, 9] . Nitrogen use in general, 52 and ammonium use in particular, connects closely with carbon cycling, as its availability can 53 influence rates of both heterotrophic[10] and photo/chemo-autotrophic activity [11–13] . 54 Nitrogen dynamics have been studied extensively in the euphotic zone (e.g., [5, 14] ), but 55 less is known about nitrogen cycling in the deep sea, a region increasingly recognized as hosting 56 a diverse, active, and influential microbiome ( [7, 15, 16] ). Urea, a form of organic nitrogen 57 which can be cleaved enzymatically to create two molecules of ammonia, has been proposed as a 58 key substrate for both anabolism and nitrification in the deep sea [17, 18]. As a source of energy 59 for chemoautotrophy, urea-based nitrification could support organic matter production at depth, 60 thereby ameliorating current discrepancies in the oceanic carbon cycle[19]. However, 61 experimental evidence regarding the abundance [21, 22] and use of urea in the meso- and 62 bathypelagic is still rare or lacking, respectively. Nitrososphaeria-affiliated ureC genes and 63 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 3 transcripts (encoding urease) have been detected in the epipelagic[17], mesopelagic [19, 23–25] 64 and bathypelagic[26, 27], suggesting the ability of nitrifying archaea to utilize this substrate 65 through the entire water column. Supporting this, urea-based nitrification has been measured at 66 the ocean surface[17], at the base of the epipelagic (at 150 m [28, 29] ), and as deep as the mid-67 mesopelagic (300 m [25] )—notably at rates comparable to those for ammonia. Similarly, urea 68 assimilation is extensive in the surface ocean[30], and has been implicated in the mesopelagic 69 based on the observation of urea degradation at rates exceeding calculated N demand for 70 nitrification [31]. However, direct measurements of urea assimilation or oxidation have not been 71 made in the lower mesopelagic or bathypelagic, nor has the prevalence or phylogenetic diversity 72 of organisms containing ureC in the aphotic ocean been determined. Therefore, whether the 73 ability to cleave urea is common or rare in the deep sea, taxonomically or numerically, is still 74 unknown, and leaves the accessibility of this potentially large source of nitrogen and energy 75 unconstrained. 76 In this work, we assessed the role of urea in sustaining microbial biomass production and 77 nitrification from 50 to 4000 m water depth in the northeast Pacific Ocean. We start with an 78 investigation of urea concentrations with depth at six sites across a 300 km transect. At two of 79 these sites, one at the base of the continental slope (“Slope Site”) and one at the far end of the 80 transect (“Open Ocean Site”), we use incubation experiments with 13C15N-urea and single-cell 81 analysis by nanoscale secondary ion mass spectrometry (nanoSIMS) to determine the proportion 82 of cells assimilating urea-derived nitrogen, and at what rates. We use these same incubations to 83 determine urea- and ammonia- based nitrification rates to assess their role in microbial 84 catabolism throughout the water column. Indeed, although genomic evidence for ammonia-based 85 nitrification at depth is convincing[32, 33], even ammonia-based nitrification has not been 86 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 4 experimentally confirmed below the mid-mesopelagic. We furthermore generate thirteen deeply-87 sequenced metagenomes throughout the Slope and Open Ocean sites, and together with public 88 metagenomes from around the world assess the distribution of the ureC gene and the potential 89 role of specific taxa in the utilization and recirculation of urea. Finally, we use the combined 90 ammonium- and urea-based nitrification rates to estimate deep-sea carbon fixation rates, and 91 compare these to estimated rates of sinking particulate organic carbon to estimate the 92 significance of nitrification-based chemoautotrophy at these sites. Together, our lines of inquiry 93 demonstrate the use of urea-derived nitrogen in both microbial anabolism and catabolism in the 94 deep northeastern Pacific Ocean, with implications for nitrogen and carbon cycling globally. 95 96 97

98 Sample collection 99 Seawater was collected in the northeast Pacific Ocean, off the coast of San Francisco 100 north of Monterey Bay (Figure 1, Fig. S1, Table S1),, onboard the R/V Oceanus in March 2017. 101 Seawater was sampled with Niskin bottles at six sites along a 300 km transect (OC1, OC2, OC3, 102 OC4, OC5, and OC6). Samples were collected at 50 m (all sites), 100 m (5 sites), 500 m (5 103 sites), 1000 m (4 sites), 2000 m (4 sites), 3000 m (4 sites), and 4000 m (1 site), as the water 104 depth allowed. Physicochemical water properties of temperature, conductivity, pressure, and 105 fluorescence were determined with a CTD (SeaBird, USA). 106 107 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 5 Quantification of nitrogen species 108 Water samples for the analysis of nitrogen species were collected from different Niskin 109 bottles in duplicate. Samples were filtered through 0.2 μm polycarbonate filters (Isopore) and 110 stored at -80 ºC until analysis on shore. Urea concentration was determined following the 111 colorimetric method in Revilla et al. (2005) [34] , using 12-hour incubation times at room 112 temperature using replicates. The detection limit was calculated to be 50 nM. Ammonium, 113 nitrate, and nitrite concentrations were previously reported in Arandia-Gorostidi et at. (2023)[7] . 114 115 Seawater incubations with stable isotopes 116 Seawater samples for incubations with stable isotope-labelled substrates were collected at 117 the sampling sites with a maximum depth of 3000 m (OC3, the ‘Slope Site’) and 4500 m (OC6, 118 the ‘Open Ocean Site’). Two sets of incubations were carried out; the first one, previously 119 described in Arandia-Gorostidi et al. (2023) to describe overall microbial activity[7] , used 50 nM 120 of 15N-labeled ammonium chloride (99% 15N, Cambridge Isotope Laboratories, USA), and a 121 second set, newly reported in this study, used 50 nM of 13C15N-labeled urea (99% 13C and 98% 122 15N, Cambridge Isotope Laboratories, USA). Both sets of incubations were conducted as 123 described in Arandia-Gorostidi et al. (2023) [7] . Briefly, seawater samples were incubated in the 124 dark in polycarbonate bottles at 10.5ºC (for water samples between 50 /i1 m and 150/i1 m depth) or 125 4 ºC (water samples from 500 m to 4000 /i1 m). Subsamples from each incubation were fixed 126 using 3% formaldehyde at 0 and 72 hours, and filtered onto polycarbonate filters (25 mm 127 diameter, 0.2 µm pore size; GTTP type, Millipore). Filtered, fixed cells were washed with PBS, 128 1:1 PBS:EtOH and EtOH before storage at -80 °C for nanoSIMS analysis. Additionally, a 129 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 6 portion (15 mL) of the incubated seawater was filtered through polycarbonate filters (25 mm 130 diameter, 0.2 µm pore size; GTTP type, Millipore) into 50 mL Falcon tubes at 0, 24 and 72 131 hours. The filtrate was stored at -80 ºC until nitrification analysis. 132 133 Single-cell isotope uptake by nanoSIMS 134 Single-cell uptake rates for 13C and 15N were analyzed by nanoscale secondary ion mass 135 spectrometry (nanoSIMS) using a NanoSIMS 50L (CAMECA, Gennevilliers, France) housed in 136 the Stanford Nano Facility. Analysis conditions are described in the Supplemental Material. 137 Between 61 and 128 cells were analyzed per sample. The isotope images were analyzed using 138 LANS software[35] , resulting in the quantitative analysis of isotopic ratios of 13C-12C-/12C2 - and 139 12C15N-/12C14N-. To determine cell-specific isotope ratios, the 12C14N- channel was used to 140 manually draw regions of interest (ROIs) with outlines just inside the cells. Cells were 141 considered isotopically enriched and therefore consumers of a particular substrate if their isotope 142 ratio was greater than 2 standard deviations above the mean isotope ratio of the 0 h cells from 143 each site[7, 36] . The isotope-based growth (Ka), relative to the initial N and C content, and the 144 single-cell assimilation rates in fg cell -1 h -1 were calculated following the equations in 145 Stryhanyuk et al. (2018) [37] . The integrated rates for the epi-, meso-, and bathypelagic regions 146 were calculated as described in Arandia-Gorostidi et al. (2023) [7] ; multiplying cell density at 147 each depth with the cell-specific assimilation rates (in fg cell -1 h-1) and the total volume of each 148 region. Statistical differences between the assimilation rates of each substrate were calculated 149 using the Wilcoxon test in R (R version 4.1.3). 150 151 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 7 Nitrification rates 152 Nitrification rates were determined from the rate of production of 15N-labeled NOx (NO3 - 153 + NO2 -) in incubations with 15N-ammonium and 15N-urea at both sites between 150 m and 4000 154 m depths. 50 m samples could not be analyzed due to the low concentration of NOx at this depth. 155 The 15N/14N ratio of NOx was determined by isotope ratio mass spectrometry using the 156 denitrifier method[38, 39] in the Stanford Stable Isotope Lab and calibrated using parallel 157 analyses of nitrate isotope reference materials USGS32, USGS34, and USGS35[40]. The 158 nitrification rates were determined using a linear fit of 15N-NOx over time in each 159 incubation[41]. 160 161 DNA extraction and metagenomic sequencing 162 Samples for metagenomic analysis were collected at all depths at the Slope Site and the 163 Open Ocean Site. Seawater was filtered through 0.2 /i1 µm Sterivex filter units (Millipore, 164 Germany) and flash-frozen in liquid N 2 immediately after collection. DNA was extracted using 165 the AllPrep DNA/RNA kit (Qiagen, Valencia, CA, USA), following the manufacturer’s protocol. 166 Prior to sequencing, the DNA concentration was measured using the Quant-iT PicoGreen 167 dsDNA Reagent (Invitrogen, Carlsbad, CA, USA). Metagenome sequencing was performed 168 using the NovaSeq S4 PE150 platform at UC Davis sequencing facility (California, USA) with a 169 target sequencing depth of ~45 gigabasepairs per sample. 170 171 Metagenomic analysis and binning 172 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 8 Paired-end reads were pre-processed with bbduk to remove adapters and to trim low-173 quality sequences. Trimmed reads were assembled individually using MEGAHIT (v1.2.9 [42, 174 43] ). To analyze the distribution and diversity of ureC genes within the OC1703 assemblies, as 175 well as those from the public GEOTRACES, TARA Oceans, and Malaspina datasets, we first 176 created a gold standard list of functional ureC proteins in cultured marine microorganisms 177 capable of urea degradation (see Table S7). Next, we used this list to identify ureC-encoding 178 contigs using a modification of the PPIT [44] R package. The abundance of ureC in each sample 179 was calculated by mapping trimmed metagenomic reads against identified ureC-containing 180 contigs with bbmap[45] and quantifying these reads (in terms of reads per kilobase million 181 mapped, or RPKM) with the samtools package[46]. The abundance of recA, amoA, and nitrate-182 related genes were calculated using the same approach, though with a modified initial annotation 183 step (Supp. Info). To estimate the propotion of cells containing ureC, we normalized its relative 184 abundance to that of recA, taking into account the average ureC gene copy number per genome 185 observed in the set of metagenome-assembled genomes (MAGs) resolved using methodology 186 described below (1.1 ureC genes/MAG). Additionally, we analyzed ureC genetic diversity using 187 a BLAST-based approach that assigns a putative taxonomy to contigs based on the consensus of 188 all genes it encodes (Supp. Info.). The relative abundance of each ureC-encoding contig for 189 which taxonomy could be assigned was computed by dividing its sequencing coverage by the 190 total coverage of all ureC -encoding contigs in that sample ( ≥ 50% breadth). We also subjected 191 the newly-generated OC1703 assemblies to metagenomic binning, forming a set of medium to 192 high quality set of metagenome-assembled genomes (MAGs) ( ≥ 50% completeness and ≤ 5% 193 redundancy as determined by CheckM) for downstream analysis (Supp. Info.).[47] 194 195 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 9 Estimation of gravitational particulate organic carbon flux 196 The gravitational Particulate Organic Carbon flux (POCflux) at 100 m depth was estimated 197 by multiplying the net primary production by the carbon export efficiency ( e-eff) calculated 198 according to the following Equation 1[48]: 199 e-eff = 0.23 x e(-0.08 x SST) (1) 200 where SST is the Sea Surface Temperature. Satellite-derived net primary production was 201 downloaded from Ocean productivity site 202 (http://sites.science.oregonstate.edu/ocean.productivity/index.php) as 8 days file format treated 203 from the VGPM algorithm[49]. 204 205 Statistical Analyses 206 To test the differences in isotopic assimilation (Figure 2C), the Wilcoxon signed-rank test was 207 used to compare the median assimilation rates between the 15N-urea and 15N-ammonium 208 incubations at each depth, due to the data distribution not following a normal distribution. 209 Differences were considered significant if p-value<0.05). To test the differences in nitrification 210 rates between the urea and ammonium incubations, we used a t-test to compare the mean rates of 211 each incubation at a given depth. A comparison of the change of ureC relative abundance with 212 depth was performed by ANOVA analysis with the Tukey’s honest significant difference test 213 correction. All the statistical tests were performed in R (R version 4.3.1). 214 215

216 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 10 Quantification of nitrogen species 217 Urea was detected in all samples, ranging from 21 nM at 50 m water depth at the Open 218 Ocean Site (OC6, 281 km from shore) to 1.1 µM at 50 m water depth at OC1, the most coastal 219 site (14 km from shore) (Figure 1). Urea concentrations were highest below the euphotic zone, 220 and generally higher than those of ammonium. The physicochemical analysis of the sampling 221 sites and the concentration of ammonium and nitrate are described in Arandia-Gorostidi et al. 222 (2023)[7]. 223 224 Single-cell urea uptake with depth 225 Microbial assimilation of urea-derived N was detected in all depths investigated at both 226 the Slope and Open Ocean sites except one (Open Ocean site, 3000 m water depth) (Figure 2). 227 The proportion of cells assimilating urea-derived N as well as the magnitude of incorporation 228 (per cell and the average for all cells assimilating it) was determined and compared to the values 229 previously determined for ammonium assimilation in the same samples [7] . In the Open Ocean 230 site, the highest proportion of cells incorporating urea-derived 15N was found at 50 m (90%; 231 Figure 2A), and was almost exactly the same as for ammonium (89%). The proportion of cells 232 incorporating urea-derived N generally decreased with depth, similar to the trend for ammonium, 233 with an exception at 2000 m where the proportion spiked to 53%. The trend in the Slope Site was 234 different from the Open Ocean site, with the lowest proportion of cells incorporating urea-235 derived N at 50 m (26%) and the highest at 150 m (78%). The proportions in the meso- and the 236 bathypelagic region remained relatively high (37% in average) at the Slope Site, higher than at 237 the same depths at the Open Ocean site (18% on average). The trends in the average N-238 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 11 assimilation rates from urea followed the same pattern as the trends in the portion of cells 239 assimilating urea (Figure 2C). Overall, the Slope Site had a higher proportion of cells 240 assimilating urea-N and at higher rates than at the Open Ocean Site. 241 On average, urea-N was assimilated at higher single-cell rates than ammonium. For the 242 cells assimilating these substrates, the mean single-cell assimilation rates in the Slope Site were 243 statistically significantly higher for urea than ammonium at 1000 m, 2000 m, and 3000 m depth 244 (Wilcoxon test, p-value <0.05), with an average of 83% higher assimilation rates for urea. The 245 difference between urea and ammonium uptake was even more pronounced at the Open Ocean 246 Site, with statistically significantly higher mean incorporation rates of urea at 50 m, 150 m, 2000 247 m, and 4000 m depth (Wilcoxon test, p-value <0.05), and an average assimilation rate an order of 248 magnitude higher for urea than ammonium. 249 Assimilation of urea-derived 13C was also detected (Fig. S2). In the Slope site, cells 250 assimilating urea-derived 13C were only detected in the epipelagic region and at 3000 m depth 251 (up to ~20% of cells), while in the open ocean site, cells assimilating urea-derived 13C were 252 detected throughout the water column (~25% at 500 m depth and ~20% at 2000 m and 4000 m 253 depth). At the Open Ocean site, we found that cell-specific 13C-urea assimilation was lower than 254 that for 15N in the epipelagic region, but that 13C-urea assimilation surpassed that for 15N in the 255 meso and bathypelagic regions. In general, while single-cell rates of urea-derived 15N 256 assimilation showed variable trends with depth, those for urea-derived 13C increased with depth. 257 258 Ammonia- and urea-based nitrification 259 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 12 Rates of ammonium- and urea-based nitrification were determined for all depths in the 260 Slope and Open Ocean sites below 50 m (Figure 3). We detected the production of 15N-NOx in 261 incubations with 15N-urea and 15N-ammonium at all depths in the Slope Site. Rates of 262 nitrification were highest at 150 m (16.1 and 11.8 nmol N l -1 for ammonium and urea, 263 respectively), dropped by an order of magnitude by 500 m, and then were relatively consistent 264 with depth with a slight uptick at 3000 m. At the Open Ocean site we detected ammonium-based 265 nitrification at 150 and 500 m, and urea-based nitrification at 150 m, but not below. Rates of 266 urea-based nitrification were similar to that for ammonia, and statistically indistinguishable at all 267 depths (t-test>0.1 for all depths). 268 269 Relative abundance and distributions of ureC and amoA 270 We sequenced and assembled metagenomes (average 57.8Gb/metagenome) from all 271 depths of the two sites where isotope incubations were conducted (Table S2), and we detected 272 ureC at all depths investigated (Figure 4). The relative abundance of ureC genes was lowest at 273 50 m (0.11 ureC/recA ratio) and highest at 150 m (0.76 ureC/recA ratio) at the Slope site (Figure 274 4), generally mirroring the trend in proportion of cells assimilating urea at this site. The 275 prevalence of ureC was more consistent with depth at the Open Ocean Site, with the maximum 276 ratio (0.5 ureC/recA) found at 500 m, 1000 m, and 3000 m depth (Figure 4). We compared the 277 relative abundance of ureC genes within the entire microbial population to that of amoA—the 278 gene encoding subunit A of ammonia monooxygenase, essential for archaeal nitrification—to 279 determine a minimum portion of ureC genes found outside of Nitrososphaeria. Trends in ureC 280 and amoA prevalence were similar with depth, although ureC was consistently twice as abundant 281 as amoA (0.45 ureC/recA and 0.23 amoA/recA on average; Figure 4). This indicates that even if 282 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 13 all amoA-encoding organisms contained ureC, half of the community potential to cleave urea is 283 found in organisms outside of that group, i.e., with different catabolisms. 284 285 Identification and relative abundance of ureC-containing contigs 286 A subset of ureC -encoding contigs ( ≥ 3000 bp, non-eukaryotic, Table S4) were then 287 analyzed to determine the taxonomic composition of organisms with the genetic potential to 288 cleave urea. These ureC-containing contigs were associated with diverse phylogenetic groups 289 (fourteen distinct phyla) and showed a consistent shift with depth between sites (Figure 5A, 290 Table S3). In the 50 m sample, ureC-containing contigs were associated primarily with 291 Proteobacteria at both the Open Ocean and Slope Sites (73.5% and 78.4% of the ureC-containing 292 community, respectively), but were also associated with Cyanobacteria (17.7% and 10.4%), 293 Nitrosophaerota (10.7%, only in the Slope Site), and Verrucomicrobia (7% and 0.5%). While 294 Cyanobacteria were only detected at 50 m, members of the Nitrosophaerota comprised an 295 increasingly large fraction of the ureC-containing community with depth (>50% in most 296 samples), together with increases in members of the Planctomycetota (up to 14.0%), 297 Verrucomicrobiota (up to 12.5%), Nitrospinota (up to 7.2%), and Myxococcota (up to 1.6%). 298 299 Taxonomic identification and investigation of MAGs containing ureC 300 To complement our contig-based analyses, we generated 109 unique ureC -containing 301 MAGs with >50% completeness and <5% redundancy (Table S5). Taxonomic identification of 302 these ureC-containing MAGs indicated they were from twelve distinct phyla (Figure 5B), 303 capturing most groups identified in the contig-based approach. Similarly, MAG-based abundance 304 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 14 analyses roughly recapitulated the distribution pattern of taxonomic groups with depth, with 305 Proteobacterial ureC-containing MAGs more relatively abundant at the surface and those of the 306 Nitrosophaerota and others becoming more relatively abundant at depth. Only MAGs belonging 307 to the Alphaproteobacterium TMED109 clade and the Chloroflexota UBA1151 were more 308 relatively abundant in the epipelagic region than in the deepest regions. Overall, a higher relative 309 abundance of ureC-containing MAG groups was found in the bathypelagic (2.56% of all mapped 310 reads in the bathypelagic versus 0.76% in the epipelagic). MAGs within the Nitrososphaerales 311 order (Phylum Nitrososphaerota) and the Pseudomonadales order (Phylum Proteobacteria) were 312 the most relatively abundant ureC-containing groups in the bathypelagic region (with a coverage 313 of 0.60% and 0.59% respectively). Other groups, including the Verrucomicrobiales and the 314 Myxococcota UBA9160 orders, were exclusively found below the photic zone. 315 316 UreC prevalence in global datasets and comparison to other genes 317 To assess the generality of our findings across the global ocean, we determined the 318 relative abundance of ureC genes in epipelagic, mesopelagic, and bathypelagic depths from 319 different ocean basins using the publicly available Tara Ocean, GEOTRACES and Malaspina 320 databases (Figure 6, Table S6). Based on comparison with recA, we estimate that 10-46% of 321 cells in the global deep sea contain ureC (median 36%), consistent with our findings in the North 322 Pacific Ocean. Relative abundances of ureC increased with depth in the GEOTRACES data from 323 the South Pacific[50], Tara Oceans data in the Arctic and South Atlantic Ocean[51, 52], and 324 Malaspina data in the North Atlantic, South Atlantic, North Pacific, South Pacific, and Indian 325 Oceans[33]. The only exception was found in sites from the northern Indian Ocean where higher 326 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 15 ureC abundance was observed at epipelagic depths than mesopelagic depths (no bathypelagic 327 metagenomes are available). 328 We also compared the distribution of ureC with genes involved in the metabolism of 329 nitrate, the most abundant source of nitrogen in the deep sea, in order to compare the potential 330 importance of both substrates. Similar to ureC , genes such as nirA or nasA (key genes in 331 assimilatory nitrate reductase), nirB and nirD (dissimilatory nitrate reduction), nirK and nirS 332 (denitrification), as well as nxrA and nxrB (nitrite oxidoreductase) increased with depth in all 333 analyzed datasets (OC1703, GEOTRACES and Malaspina) (Figure S3). However, ureC was 334 consistently more abundant than these other genes (Figure S3). This difference is particularly 335 notable in the Malaspina dataset, which is the dataset that best reflects the gene distribution in 336 the world oceans, with an order of magnitude more ureC than nasA. 337 338 Comparison of organic carbon sources to the deep sea 339 In order to assess the significance of deep-sea nitrification to the marine carbon cycle we 340 estimate and compare two sources of organic carbon to the deep sea: the particulate organic 341 carbon (POC) consisting of photosynthetic detritus sinking from the euphotic zone 342 (‘gravitational POC’) and the inorganic carbon fixed via nitrification-based chemoautotrophy in 343 the deep sea. We estimated POC export fluxes to be 26 and 128 mg C/m 2/day at the base of the 344 euphotic zone (100 m) at the open ocean and coastal sites, respectively. In parallel, assuming a 345 DIC fixation yield of 0.09 mol C fixed/mol N oxidized[53], we converted the rates of urea- and 346 ammonium-based nitrification measured here into carbon fixation rates and integrated them over 347 the dark water column (100-3000 m or 100-4000 m, depending on the site). We found rates of 348 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 16 1.4 and 8.8 mg C/m 2/day of autotrophic carbon fixation for the open ocean and slope sites 349 respectively, corresponding to 5 and 7% of the gravitational POC fluxes entering the dark ocean. 350 351

352 Urea is increasingly recognized as a source of nitrogen for cell growth [30] [20, 54] and 353 nitrification[17, 18] in the sunlit ocean. In the euphotic zone, nitrogen from urea is assimilated by 354 phylogenetically diverse taxa, including Cyanobacteria, Proteobacteria, and Thermoproteota 355 (e.g., Nitrosophaeria) [14, 18, 20] [19, 28, 55] , at rates exceeding those for nitrate, leucine, 356 glutamate[14], and even ammonium[56], and it is also oxidized by nitrifying Nitrosophaeria to 357 support chemoautotrophy[57, 58]. Our observations in the aphotic zone of the northwest Pacific 358 Ocean indicate that the significance of this molecule not only extends to the aphotic zone—359 where an equally broad yet predominantly different set of organisms cleave it—but indicate that 360 its role may be even more central to ecosystem functioning there than in surface waters. Our data 361 show that its availability—assessed as a function of both its concentration and the prevalence of 362 the genetic capacity to access it—increases with de pth, as does the microbial preference for it as 363 a nitrogen source over ammonia. Additionally, while rates of urea-based nitrification are lower at 364 depth than at the surface, they are comparable to that of ammonia in both realms, and likely play 365 an outsized role in microbial community dynamics at depth by supporting the production of 366 organic matter in a more energy- and carbon-limited system than at the surface. 367 Urea-derived nitrogen is widely and extensively assimilated by microrganisms in the 368 aphotic zone. We find that on average 25% of cells in the meso- and bathypelagic assimilate 369 urea-derived nitrogen. Assuming that ammonium uptake reflects cellular activity[7] and then 370 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 17 comparing the subset of cells assimilating urea-N to that assimilating ammonium-N, we find that 371 60% of active cells in the meso- and bathypelagic assimilate urea-derived nitrogen. Although 372 cross-feeding of 15N-labelled substrates can cause these proportions to be greater than the 373 number of cells directly consuming urea, the paired metagenomic data is roughly consistent with 374 these values in the deep-sea; we estimate that an average of 39% of cells in the meso- and 375 bathypelagic at these sites contain a ureC gene, putatively allowing lineages from at least 376 fourteen distinct phyla to cleave urea directly (as described in more detail below). Additionally, 377 cells assimilating urea-derived N did so at higher rates than cells assimilating ammonium in 378 almost all samples investigated. This demonstrates the significance of this nitrogen source to the 379 cells and also decreases the likelihood that cells assimilated the urea-derived N as recycled 380 ammonium. Regardless of what proportion of the assimilation was directly from urea versus 381 recycled substrates, the widespread and high rates of consumption of urea-derived N indicates 382 that the large reservoir of urea-nitrogen in the deep sea—on average an order or magnitude more 383 abundant than ammonium—is available to most cells. 384 Nitrate remains the largest pool of fixed nitrogen in the deep sea, averaging over two 385 orders of magnitude more abundant than urea at our site. Our observations of urea assimilation 386 occurred in the presence of these high concentrations of nitrate, suggesting a preference for urea 387 over nitrate. Indeed, we found that ureC genes were more abundant than those related to 388 assimilatory nitrate reduction (such as nasA) within our study sites, as well as a broad 389 distribution of publically available deep sea datasets (Fig. S3). Previous work has also reported 390 relatively low detection of nasA in the Malaspina global deep-sea metagenomic dataset [6, 7] . 391 Preference for urea is likely related to the higher energy requirements of the assimilatory 392 reduction of nitrate [59][60], a difference that might be particularly relevant in the energy-poor 393 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 18 aphotic zone. While gene abundances are useful indicators of potential activity, and how well a 394 given ability is distributed across a community, direct comparisons of the uptake of nitrate and 395 urea in the deepest regions of the oceans would be beneficial to directly compare the proportions 396 of cells capable of assimilating each, and with what preference. Notably, the meso- and 397 bathypelagic regions accounted for nearly half of the total pelagic urea assimilation in the Slope 398 Site—more than it contributed to either ammonium or amino acid assimilation [7] —indicating 399 that urea is a more important nitrogen source in the deep sea relative to the surface than for either 400 ammonium or amino acids. 401 Urea also represents a major potential substrate for nitrification by members of the 402 Nitrosophaeria phylum [17–19, 25] . Remarkably, urea-based nitrification can also happen in the 403 presence of ammonium[18] , suggesting that urea is not only an alternative for ammonium when 404 it is scarce, but can also represent a primary substrate for nitrification. Furthermore, a recent 405 study shows that some ammonia-oxidizing bacteria repress the use of extracellular ammonium in 406 the presence of ammonium derived from urea hydrolysis in the cytoplasm[56]. While previous 407 studies have highlighted the significance of urea-driven nitrification in the epipelagic[17, 18, 61] 408 and upper mesopelagic region (up to 300 m depth [25] ), as well as demonstrated the potential of 409 deep marine microbes to oxidize ammonia through genomic analysis [19] , urea-driven 410 nitrification has not been directly measured in the lower mesopelagic and bathypelagic regions. 411 The detection of urea-based nitrification at all depths of our Slope site suggests that deep-sea 412 nitrifiers can indeed use urea as a substrate (Figure 2). Oxidation of ammonia after urea 413 hydrolosis by other community members is also possible, but regardless, this confirms urea-414 derived nitrogen is readily available to microbes for nitrification. Direct oxidation of urea by 415 deep-sea nitrifiers is additionally supported by the metagenomic analysis, which showed not only 416 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 19 that ureC genes were found within Nitrosophaeria MAGs, but that over half of read coverage of 417 ureC-containing contigs in the meso- and bathypelagic was affiliated with Nitrososphaeria 418 (Figure 5A). The rates of urea-based nitrification were statistically indistinguishable from those 419 for ammonium at all depths, consistent with the previous work at 300 m depth[25], indicating a 420 potentially significant role for urea in deep-sea nitrification. Rates of both ammonia- and urea-421 based nitrification decreased with depth and distance from shore (Figure 2), consistent with the 422 trends we observed in overall anabolic activity previously at this site [7]. 423 Using metagenomics, we determined both the distribution of ureC in microbial 424 communities at our study site and in globally-sourced datasets, and also phylogenetically 425 identified the taxa containing ureC. We detected ureC genes throughout the water column, and 426 found that their prevelance—the pr oportion of microbial cells possessing it in a given sample—427 reached a maximum in the aphotic zone in both our study site and the other global datasets we 428 analyzed. Overall, we see that about a third of the cells in the dark ocean (average 39% in our 429 dataset, and average of 30% in the global datasets) contain ureC. Both the contig- and MAG-430 based analyses identified diverse taxa containing ureC genes at our site, with fourteen distinct 431 phyla identified by the former and twelve by the latter. While the MAG-based phylogenetic 432 identification of ureC -containing genomes is likely more robust than the contig-based 433 identifications, due to the greater sequence length available for consideration, the contig-based 434 approach provides a more comprehensive overveiew of the community, including taxa that may 435 systematically evade genomic binning. Notably, twelve of the fourteen phyla identified in the 436 contig-based analysis were also identified with the MAG-based analysis. The groups identified 437 as containing ureC in the 50 m samples are generally consistent with previous work in the 438 euphotic zone, especially in the identification of Gammaproteobacteria and Prochlorococcus[14]. 439 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 20 The deep-sea analysis revealed that some taxonomic groups with members known to use urea at 440 t h e s u r f a c e a l s o h a v e m e m b e r s w i t h t h e g e n e t i c c a p a c i t y t o d o s o a t d e p t h , i n c l u d i n g 441 Nitrososphaerales, Verrucomicrobiota, and Myxococcota, as well as members of several groups 442 not before reported to utilize urea, including SAR202 and Alphaproteobacteria TMED109. We 443 interpret the presence of ureC genes in taxa not known to oxidize ammonia, and in MAGs 444 without an amoA gene, as evidence of potential urea use for nitrogen acquisition. When found 445 together with amoA (i.e., within the Nitrosphaeria), it may be used for both nitrogen acquisition 446 for biomass and for nitrification. While our metagenomic analysis is consistent with a large role 447 for urea in nitrifying organisms in the deep sea – evidenced by the large fraction of ureC genes 448 within the Nitrosphaeria – our work also highlights the wide diversity of organisms capable of 449 cleaving it. And, as not all nitrifiers contain urease (e.g., Nitrosopelagicus brevis CN25, Santoro 450 et al., 2015, Nitrosopumilus maritimus[23]), there may be an important relationship between 451 heterotrophic urea degraders and the chemoautotrophic nitrifiers, with ammonia shared in one 452 direction and organic carbon in the other. 453 The implications of deep-sea nitrification on the marine carbon cycle are considerable. It 454 is often assumed that the main—and essentially only—source of organic carbon to the dark 455 ocean is gravitational POC[62]. However, the persistent imbalance between known supply and 456 demand of organic matter in the deep sea highlight the inadequacies of current knowledge[63]. 457 Our observations indicate that urea- and ammonium-based nitrification could together provide 5 458 and 7% of the estimated gravitational POC entering the top of the mesopelagic at the Open 459 Ocean and Slope Sites, respectively. These are significant values considering gravitational POC 460 fluxes decrease logarithmically with depth in the water column[64, 65]. Additionally, while 461 organic carbon generated at depth is generally labile, gravitational POC is increasingly 462 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 21 recalcitrant and biologically inaccessible with depth, highlighting the potential importance of 463 even small amounts of endogenously produced organic carbon[66]. To fully assess the role of 464 chemoautotrophy in helping balancing the carbon budget more studies are required, including 465 direct measurements of carbon fixation ideally at in situ pressures and concurrent measurement 466 of sinking POC flux. However, our estimates provide new evidence that deep-sea 467 chemoautotrophy, which is often overlooked in models of the biological carbon pump, is not 468 negligible, and could be of paramount importance. 469 In summary, our study reveals a large reservoir of urea-N in the deep sea (Figure 1), 470 widespread genetic potential for urea utilization in the meso- and bathypelagic (Figures 4 and 5), 471 and direct evidence for both extensive assimilation of urea-derived nitrogen (Figure 2) and the 472 persistence of both urea- and ammonia-based nitrification throughout the epi-, meso-, and 473 bathypelagic (Figure 3). While additional direct measurements are necessary to confirm our 474

globally, we contend that urea use is likely widespread throughout the global deep sea on 475 the basis of the generally physicochemically representative nature of our study site and the high 476 proportions of ureC-encoding microorganisms throughout the global metagenomic datasets 477 analyzed here. These results address long-standing hypotheses about the potential for urea to fuel 478 nitrification in deep waters, and indicate the potential for chemoauototrophy at depth to 479 significantly impact the marine carbon budget. 480 481 Data Availability 482 Read data and metagenome-assembled genomes (MAGs) analyzed in this study are available 483 through NCBI at PRJNA1054206. 484 485 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 22

486 We thank Julian Fortney, and Nicolette Meyer, as well as the captain and crew of R/V Oceanus 487 cruise OC1703A, for assisting with field sampling. This work was primarily supported through a 488 Simons Foundation Early Career Investigator Award to AED (507798 ) and a National Science 489 Foundation CAREER Award to AED (2143035). Cruise OC1703A was supported by NSF 490 Award 1634297 to AED. The nanoSIMS analyses were performed at the Stanford Nano Shared 491 Facilities (SNSF), which is partially supported by the National Science Foundation under award 492 ECCS-2026822. We thank Christie Jilly-Rehak and Chuck Hitzman for assistance with the 493 nanoSIMS analyses. We thank Pascale Anabelle Baya-Ardyna for assistance with the 494 nitrification measurements. NAG was supported by the ‘Severo Ochoa Centre of Excellence’ 495 accreditation (CEX2019-000928-S) funded by AEI 10.13039/501100011033, and the Beatriu de 496 Pinós program (2020-BP-00179) during the writing of this manuscript. ALJ was supported by 497 the Stanford Science Fellows program and the National Science Foundation Postdoctoral 498 Research Fellowship in Ocean Sciences. RSRS was supported by the Stanford Graduate 499 Fellowship Program. 500 501 Conflicts of Interest 502 None to declare. 503 References. 504 1. Kirchman DL, Wheeler PA. Uptake of ammonium and nitra te by he tero tro phic ba cteria an d 505 phytoplankt on in the sub-A rctic Pacific. Deep Sea Res 1 Oceanogr Res Pap 1998; 45: 347–365. 506 2. Bristow LA, Moh r W, Ahm erkamp S, Kuypers MMM . Nut rien ts tha t limit growth i n the ocea n. 507 Current Biology 2017; 27: R474–R478 . 508 3. Middelburg J , Nieuw enhuize J. Ni trog en uptake by het ero trop hic bacte ria and ph ytoplankton in 509 the nit rat e-rich Thames est uary. Mar Ecol Prog Ser 2000; 203 : 13–21. 510 4. Fouilland E, Goss elin M, Rivkin RB, Vasse ur C, Mostajir B. Nitr ogen upt ake by het e rotr ophic 511 bacteri a and phytopl ankton in A rctic surf ace waters . J Plankton Res 2007; 29: 369 –376. 512 5. Klawonn I, Bonaglia S, Whit ehouse M J, Li ttmann S, Tienken D, Kuypers MMM, e t al. Unt angling 513 hidden nut rien t dynamics: rapid ammoni um cycling and single-cell ammonium ass imilation in 514 marine plankt on communities . ISME Journal 2019; 1960–1974. 515 6. Sarmient o JL, Simeon J, Gn anadesika n A, Grube r N, Key RM, Schli tze r R. Deep oc e an 516 biogeochemist ry of silicic acid and nitrat e. Global Biogeochem Cycles 2007; 21. 517 7. Arandia- Goros tidi N, Par ada AE, Dekas A E. Single-cell view of deep-sea microbial activity and 518 intracommuni ty hete rogen eity. ISME Journal 2023; 17: 59–69. 519 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 23 8. Kowalchuk GA, Ste phen J R. Ammonia- Ox idizing Bacter ia: A M odel for Mol ecular Microbial 520 Ecology. Annu Rev Microbiol 2001; 55 : 485–529. 521 9. Francis CA, Robe rts KJ, B eman JM , Sant o ro AE, Oakl ey BB. Ubiqui ty and diversity of ammonia-522 oxidizing archa ea in wat er columns and s ediments of th e ocean . Proceedings of the National 523 Academy of Sciences 2005; 102 : 14683–1 4688. 524 10. Tupas L, Koike I. Amino acid and ammoni um utilizatio n by hete rot rophic marin e b acteri a grown 525 in enriched se awate r. Limnol Oceanogr 1 990; 35: 1145–1155. 526 11. Dortch Q. The int erac tion be tween amm onium and nitr ate up take in phyt oplankt on. Mar Ecol 527 Prog Ser 1990; 61: 183–201. 528 12. Pahlow M. Linking chlorophyll-nutrien t d ynamics to the RedÞeld N:C rat io with a model of 529 optimal phytopl ankton grow th. Mar Ecol Prog Ser 2005; 287 : 33–43. 530 13. Tolar BB, Ross M J, Wallsgr ove NJ , Liu Q, Aluwihare LI, Popp B N, e t al. Cont ributi o n of ammonia 531 oxida tion to ch emoaut otr ophy in Anta rct ic coastal wate rs. ISME J 2016; 10: 2605– 2619. 532 14. Zimmerman AE, Podowski JC, Gallagher GE, Coleman ML, Waldb aue r JR. Tr acking nitrogen 533 allocation to pro teom e biosynthesis in a marine microbial communi ty. Nat Microbiol 2023; 8 : 534 498–509. 535 15. Acinas SG, Sánch ez P, Salaza r G, Corn ejo-Castillo FM, Seb astián M , Logares R, e t al . Deep ocean 536 metagenom es provide insight in to th e metabol ic archit ectur e of bathypel agic microbial 537 communities. Commun Biol 2021; 4 : 604. 538 16. Sánchez P, Sebas tián M, Pernic e M, Rod r íguez-Martí nez R, Pesan t S, Agustí S , et al . Marin e 539 Picoplankton Me tagenom es from Eleven Vertical Profiles Ob tain ed by the Mal aspi na Expedi tion 540 in the Tropical an d Subtr opical Oce ans. bioRxiv 2023; 2023.02 .06.526790. 541 17. Tolar BB, W allsgrove N J, Popp BN, Holli b augh JT. Oxi dati on of urea-de rived nit rog en by 542 thaumarch aeo ta-domina ted marin e nit ri fying communities. Environ Microbiol 20 17; 19: 4838–543 4850. 544 18. Kitzinger K, Padilla CC, Marchant HK, Hac h PF, Herbold CW, Kidane AT, et al . Cyanate and u rea 545 are subst rat es for nitrifica tion by Thaumarchaeo ta in th e marin e environmen t. Nat Microbiol 546 2019; 4 : 234–243. 547 19. Alonso-Sáez , Wall er a. S, Mend e DR, Bakker K, Farnelid H , Yager PL, et al. R ole for urea in 548 nitrificatio n by polar marin e Archa ea. Proceedings of the National Academy of Sciences 2012; 549 109 : 17989–17994. 550 20. Connelly TL, Baer SE, Cooper JT, Bronk DA. Ure a upt ake and carb on fixati on by marine pel agic 551 bacteri a and arch aea du ring the A rctic su mmer and winter s easons. Appl Environ Microbiol 2014; 552 80: 6013–6022. 553 21. Remsen CC. the Distribu tion of Ur ea in Coastal and Oceanic W ate rs. Limnol Oceanogr 1971; 16: 554 732–740. 555 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 24 22. Lomas MW, Trice TM, Patricia M . Glib ert DAB, McCarthy JJ . Tempor al and Spa tial Dynamics of 556 Urea Up take and R egen era tion Ra tes an d Concentra tions in Chesape ake Bay. 200 2; 25: 469–482. 557 23. Swan BK, Chaffin MD, Martinez-G arcia M , Morrison H G, Field EK, Poult on NJ , et al . Genomic and 558 Metab olic Diversity of Marine Group I Th aumarchae ota in the Mes opelagic of Two Subtropic al 559 Gyres. PLoS One 2014; 9 : e95380. 560 24. Mao X, Chen J, van Oos terh out C, Zhang H, Liu G, Zhuang Y, et al. Diversity, p reval ence, and 561 expr ession of cyanase genes (cynS) in planktonic marine microo rganisms. ISME Journal 2022; 16: 562 602–605. 563 25. Santor o AE, Sai to MA, Goe pfert TJ , Lamborg CH, Dupont CL, DiTullio GR. Thaumar chaeal eco type 564 distribut ions across th e equa tori al Pacific Ocean and t heir po ten tial ro les in nitrific ation and 565 sinking flux attenu ation . Limnol Oceanogr 2017; 62: 1984–2003. 566 26. Yakimov MM, Cono V La, Smedile F, DeLuca TH, Juáre z S, Ciordia S, e t al. Con trib ution of 567 crenarcha eal au tot rophic ammonia o xidi zers to the da rk primary produc tion in Tyrrheni an deep 568 waters (Central Medi ter rane an Sea). ISME J 2011; 5 : 945–961. 569 27. Acinas SG, Sánch ez P, Salaza r G, Corn ejo-Castillo FM, Seb astián M , Logares R, e t al . Deep ocean 570 metagenom es provide insight in to th e metabol ic archit ectur e of bathypel agic microbial 571 communities. Commun Biol 2021; 4 : 604. 572 28. Parada AE, Mayali X, W eber PK, Woll ard J, Sant oro AE, Fu hrman J A, e t al. Const rai ning the 573 composition and qua nti ty of organic mat ter used by abun dant ma rine Thauma rchaeot a. Environ 574 Microbiol 2022; 1–16. 575 29. Laperrie re SM , Mora ndo M, Capon e DG, Gunde rson T, Smith JM , Sant oro AE . Nit ri fication and 576 nitrous o xide dynamics in the So uth ern California Bight . Limnol Oceanogr 2021; 66 : 1099–1112. 577 30. Solomon C, Collier J, B erg G, Glibe rt P. Ro le of urea in microbial me tabolism in aqu atic systems: a 578 biochemical and molecula r review. Aquatic Microbial Ecology 2010; 59: 67–88 . 579 31. Cho BC, Azam F. Urea d ecompositi on by bacteri a in the So uth ern California Bigh t and its 580 implications for th e mesopelagic ni troge n cycle. Mar Ecol Prog Ser 1995; 122 : 21– 26. 581 32. Jiang X, Dang H, Jiao N . Ubiquity and Diversity of Het ero trophic Bac ter ial nasA Ge nes in Diverse 582 Marine Environm ents. PLoS One 2015; 10 : e0117473. 583 33. Sánchez P, Coutinho FH, S ebasti án M, Pe rnice MC, Rodrígue z-Mar tínez R, S alaza r G, et al. Ma rine 584 picoplankton me tagenom es and MA Gs from eleven vertica l profiles ob tained by t he Malaspin a 585 Expediti on. Sci Data 2024; 11: 154. 586 34. Revilla M, Ale xand er J , Glibe rt PM. U rea analysis in coastal wate rs: comparis on of enzymatic and 587 direct me thods. Limnol Oceanogr Methods 2005; 3 : 290–299. 588 35. Polerecky L, Adam B, Milucka J, Mus at N, Vagner T, Kuypers MMM. Look@Nan oSI MS--a tool for 589 the analysis of nanoS IMS da ta in environ mental microbiol ogy. Environ Microbiol 2 012; 14: 1009–590 23. 591 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 25 36. Dekas AE, Parada AE, Mayali X, Fuhrm an JA, Wol lard J , Web er PK, et al . Charact eri zing 592 Chemoauto trophy and H ete rot rophy in Marine Archae a and Bac teri a With Singl e -Cell Multi-593 isotope NanoS IP. Front Microbiol 2019; 10 . 594 37. Stryhanyuk H, Calabres e F, Kümmel S, M usat F, Richnow HH, Musa t N. Calcula tio n of Single Cell 595 Assimilation Ra tes Fr om SIP-NanoSI MS-Derived Iso tope R atios : A Comprehe nsive Approach . 596 Front Microbiol 2018; 9 : 1–15 . 597 38. Sigman DM, Casciotti KL, Andreani M, Ba rford C, Galan ter M , Böhlke JK. A Bac teri al Meth od for 598 the Ni troge n Isot opic Analysis of Nitr ate i n Seawate r and Fr eshwate r. Anal Chem 2 001; 73: 4145–599 4153. 600 39. McIlvin MR, Casciotti KL. Technical Upd at es to th e Bact erial M etho d for Nit rat e Iso topic Analyses. 601 Anal Chem 2011; 83: 1850–1856. 602 40. Böhlke JK, Mr oczkowski SJ, Coplen TB. O xygen isotop es in nitra te : new refer ence materi als for 18 603 O: 17 O: 16 O measuremen ts and obs ervat ions on nitra te-wat er equi libra tion. Rapid 604 Communications in Mass Spectrometry 2 003; 17: 1835–1846. 605 41. Santor o AE, Buchwald C, Knapp AN , Ber e lson WM, Capone DG , Casciotti KL. Ni trifi cation and 606 Nitro us Oxid e Production in the Offshor e Wate rs of the East ern Tropical S outh Pa cific. Global 607 Biogeochem Cycles 2021; 35. 608 42. Li D, Liu C-M, Luo R, Sadakane K, Lam T- W. MEG AHIT: an ul tra-fast single-nod e s olution for la rge 609 and complex me tagenomics assembly via succinct de Bruijn graph. Bioinformatics 2015; 31: 610 1674–1676. 611 43. Li D, Luo R, Liu C-M, Leung C-M, Ting H-F, Sadakane K, e t al. ME GAHIT v1.0: A fast and scalable 612 metagenom e assemble r driven by advan ced methodo logies and community prac t ices. Methods 613 2016; 102 : 3–11. 614 44. Kapili BJ, Dekas AE. PPIT: an R package fo r inferring microbial t axo nomy from nifH sequences . 615 Bioinformatics 2021; 37: 2289–2298. 616 45. Bushnell B. B BMap : A Fast , Accura te, Spli ce-Aware Aligne r. No. LBNL-7065E. Ernest Orlando 617 Lawrence Berkeley National Laboratory, Berkeley, CA. 2014. 618 46. Danecek P, Bonfield JK, Liddle J , Marshall J, Ohan V, Pollard M O, e t al. Twelve year s of SAMtools 619 and BCFtools. Gigascience 2021; 10. 620 47. Suzek BE, Wang Y, Huang H, McG arvey PB, Wu CH. UniRef clust ers: a compr ehens ive and scalable 621 alter native for impr oving sequence simil arity search es. Bioinformatics 2015; 31: 926–932. 622 48. Henson SA, Sa nders R, Madsen E, Mo rris PJ, Le Moigne F, Quar tly GD. A r educed e stimate of th e 623 strength of the oce an’s biological carb on pump. Geophys Res Lett 2011; 38: n/a-n/a. 624 49. Behre nfeld MJ , Falkowski PG. Photosynt hetic ra tes de rived from satelli te-bas ed chlorophyll 625 concentr ation . Limnol Oceanogr 1997; 42 : 1–20. 626 50. Biller SJ , Beru be PM, Dooley K, Williams M, Satinsky BM, Hackl T, et a l. Ma rine mi crobial 627 metagenom es sampled across spac e and time. Sci Data 2018; 5 : 180176. 628 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 26 51. Tully BJ, Grah am ED, Heidelbe rg JF. The r econstruc tion of 2,631 draft me tagenom e-assembled 629 genomes from the global oc eans. Sci Da t a 2018; 5 : 1–8. 630 52. Salazar G, Paoli L, Albe rti A , Huer ta-Cepa s J, Ruscheweyh H-J, Cuenca M, et al . Ge ne Expressio n 631 Changes and Community Turnover Differentially Shap e th e Globa l Ocean M et atr a nscriptome . Cell 632 2019; 179 : 1068-1083.e21. 633 53. Bayer B, McB eain K, Carlson CA, San t or o AE. Carbon content, carbon fix at ion yield and dissolved 634 organic carbon r eleas e from diverse mari ne nitrifie rs. L i m no l O c e an og r 2023; 68: 84–96. 635 54. Bradley PB, Sand erson MP, N ejstgaard JC, Sazhin AF, Frisch er ME, Killbe rg-Thores on LM, et al . 636 Nitrog en uptak e by phytoplankt on and b acteri a during an induced Pha eocystis po uchetii bloom, 637 measured using size frac tiona tion and flo w cytometric sorting . Aq uati c Micr obia l Ecolo gy 2010; 638 61: 89–104. 639 55. Pachiadaki MG, Sin tes E, Be rgauer K, Br o wn JM, Record NR, Swan BK, e t al. M ajor role of nitri te-640 oxidizing bacteria in da rk ocean car bon fixat io n. 2017; 1051 : 1046–1051. 641 56. Qin W, Wei SP, Zheng Y, Choi E, Li X, Johnston J, et al . Ammonia-oxidi zing bacte ri a and archa ea 642 exhibi t differential nitrogen sourc e prefe rences. Nat M icro biol 2024. 643 57. Qin W, Amin SA, M artens-Habben a W, Walker CB, Ur akawa H, Devol AH, e t al. Marine ammoni a-644 oxidizing archa eal isola tes display obliga t e mixot rophy and wide eco typic variatio n. Proceedi ngs 645 of the Na tio nal A ca demy of Sciences 201 4; 111 : 12504–12509. 646 58. Bayer B, Vojvoda J, Offre P, Alves R JE, Elisabeth N H, Ga rcia J AL, et al . Physiological and genomic 647 characte riza tion of two novel marin e tha umarchaeal s trains indica tes niche differ entia tion. ISME 648 J 2016; 10: 1051–1063. 649 59. Mobley HL, Hausinger RP. Micr obial ur ea ses: significance, regula tion, an d molecul ar 650 characte riza tion. Micro biol Rev 1989 ; 53: 85–108. 651 60. Gates AJ, Luqu e-Almagro VM, G oddar d A D, Ferguson SJ, Rold án MD, Richardso n DJ. A composite 652 biochemical system for bacte rial nit ra te and nitri te assimila tion as e xemplified by Paracoc cus 653 denitrifi cans . Bioc hemi cal Jour nal 2011; 435 : 743–753. 654 61. Shiozaki T, Hashihama F, Endo H, Ijichi M, Takeda N, Mak abe A, et al . Assimilatio n and oxida tion 655 of urea-derived ni troge n in the summer Arctic Oce an. Limn ol Oce ano gr 2021; 66: 4159–4170. 656 62. Baumas C, Fuchs R, Gar el M, Poggiale J-C, Memery L, Le Moigne FAC, e t al. R econs tructing th e 657 ocean’s mesop elagic zone car bon budge t : sensitivity and es timation of parame ter s associated 658 with prokaryotic r eminer aliza tion. Bi oge oscien ces 2023; 20: 4165–4182. 659 63. Burd AB, H ansell DA, St einbe rg DK, Ande rson TR, Aríst egui J, B alta r F, et a l. Assess ing the 660 appare nt imbalanc e betw een geoch emical and biochemical indica tors of meso- a nd bathypelagic 661 biological activity: Wh at th e @$ /i1! is wrong with present calcul ations of carb on b udgets? Deep 662 Sea Researc h Part II : Topi cal Stu dies in O cean ogr aph y 2010; 57: 1557–1571. 663 64. Martin JH, Knaue r GA , Karl DM, Bro enko w WW. VERTEX: carbon cycling in the northe ast Pacific. 664 Deep Sea Researc h Part A Oce an ogra phi c Researc h Papers 1987; 34: 267–285. 665 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 27 65. Le Quéré C, And rew RM, Fri edlingste in P, Sitch S, Hauck J, Pongratz J, e t al. Global Carbon Budget 666 2018. Earth Syst Sci Dat a 2018; 10: 2141 –2194. 667 66. Herndl G J, Re inthal er T. Micr obial cont ro l of the dark end of th e biological pump. Nat ure 668 Publishing Gro up 2013; 6 . 669 670 671 Figure legends 672 Figure 1. Concentrations of urea, ammonium, and nitrate for each site and depth. Note that the 673 scale for nitrate is different from the other nutrients. Error bars indicate standard deviation of 674 triplicate measurements. Empty dots in ammonium indicate a concentration below the detection 675 limit. Ammonium and nitrate data are re-plotted from Arandia-Gorostidi et al., 2023. Grey area 676 indicates sea-floor depth. 677 Figure 2: Single-cell assimilation of urea-derived N, compared to ammonium. (A) The depth 678 profile plots show the proportion (%) of cells that assimilated nitrogen from urea or ammonium 679 at each depth of the Open Ocean and Slope Sites. (B) NanoSIMS images of cells showing 680 assimilation of 15N from urea and ammonium in incubations of seawater from the Slope Site. The 681 color scale indicates the 12C15N-/12C14N- ratio for the analyzed areas. (C) Boxplots represent the 682 N-based growth (Ka) in logarithmic scales for individual cells from both urea- and ammonium-683 amended incubations at each depth of the Open Ocean Site (left) and Slope Site (right). Purple 684 asterisks indicate that the difference between ammonium and urea rates are significant (ANOVA, 685 p-value<0.05). 686 Figure 3. Urea- and ammonium-based ammonia oxidation rates at each depth of the Slope Site 687 and Open Ocean Site. While at the Slope site ammonia oxidation was detected at all depths, for 688 the Open Ocean Site no significant ammonia oxidation was detected below 150m depth (except 689 for ammonia incubations at 500m depth). Error bars represent standard deviation of duplicate 690 measurements. 691 Figure 4: Vertical profiles showing gene abundance of ureC and amoA genes relative to a 692 housekeeping gene ( recA) in the metagenomes at the Open Ocean Site (left) and Slope Site 693 (right). 694 Figure 5. Diversity and distribution of taxa containing ureC . (A) Taxonomic analysis of the 695 ureC-containing contigs classified at the phylum level for each water depth, each site separately. 696 (B) Relative abundance of ureC-containing MAGs within each or der (expressed as coverage of 697 total reads) with water depth, Slope and Open Ocean sites combined. Phylum affiliation is 698 indicated. The number of MAGs within each order is indicated in parentheses. 699 Figure 6. Box plots showing abundance of ureC genes relative to total mapped reads in the 700 epipelagic (0-200 mbsl), mesopelagic (200-1000 mbsl), and bathypelagic (1000-4500 mbsl) 701 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint 28 regions of different ocean regions. The North Pacific Ocean is shown in the OC1703 and 702 Malaspina datasets, Southwest Pacific Ocean in the GEOTRACES dataset, North Atlantic in the 703 Malaspina dataset, South Atlantic and Indian Oceans in the Malaspina and Tara Oceans datasets 704 and the Arctic Ocean in the Tara Oceans datasets. Gene abundances are displayed as the ratio 705 between ureC and recA (both gene coverages calculated as RPKM). For the Tara Oceans, only 706 samples for the epipelagic and mesopelagic regions were available. 707 708 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 27, 2024. ; https://doi.org/10.1101/2024.07.26.605319doi: bioRxiv preprint