Process Performance
Prior to this experimental phase, the reactor had been operating for nearly 19months, and the last fivemonths using a consistent regime that included threeisochronous aerated and non-aeratedintervals during the react phases: total nitrogen (TN) removal efficiencywasaround 84.5% [16].During the reported phase, the frequency of redox switching increased on a monthly basis from 3 to 4, 6, 8, 10, 16 and 25. Throughout these sixmonths of operation, process performance deteriorated slightlywith removal rates varying from 86±4% TNin early phase and reaching 75±4% at highest switching frequency (increasing oxygen loading recovered 85% efficiency).No residual nitrite was ever detected in the effluent and the ratio of nitrate produced per unit of ammonium removed(RNatTot[17])remained below 0.13, pointing at AnAOB as the only generators of NO3- [18].Furtherdetails are reported elsewhere [16].
Community analysis via nxrand 16S rRNA genetargeted qPCR [16]and 16S rRNA gene amplicon sequencing (Fig S1) indicated extremely low abundances of taxa containing typical nitrite oxidizing bacteria (predominantly in smallest aggregates (< 90 micrometer))Nitrobacter spp. and Nitrospira spp., with a high abundance of taxa comprising AnAOB (Brocadia spp.) and AOP (Nitrosomonas spp.). AOB fractions increased with redox switching frequency, but only in the largest aggregates (>600 µm)[16].
Quality of MAGs
Whole-community DNA sequencing from seven samples, taken at monthly intervals, from the PNA reactor, generated an average of 2.7 ± 0.5 Gbp high-quality, paired-end sequence data per sample and 55 metagenome-assembled genomes (MAGs) (average completeness and contamination of 81.5% and 1.7%, respectively (Table S1)) were retrieved from the co-assembled contigs. On average 79% (± 1) of the community metagenome could be assigned to the recovered MAGs indicating a solid coverage of the community by the retrieved MAGs.Temporal dynamics of the community, as inferred from relative MAG abundance, was limited, indicating resilience to the changing aeration conditions (Fig. 1).
Overall Community Composition
MAGs were classified as autotrophs if they contained the key genes of any of the previously described carbon fixation pathways(the Calvin-Benson-Bassham (CBB) cycle, the 3-hydroxypropionate cycle, the 3-hydroxypropionate-4-hydroxybutyrate cycle, the reductive tricarboxylic acid (rTCA) cycle or the Wood-Ljungdahl pathway). The community was dominated by non-autotrophic (46 MAGs, ca. 57% abundance) over autotrophic MAG types (9 MAGs, ca. 43% abundance) even though no organic carbon was fed to the system (Table S2; Fig 1). The autotrophic MAGs comprised three MAGs each in the AOB(reductive pentose phosphate (CBB)) and AnAOB (Wood-Ljungdahl (WL)) guilds, respectively, and threeadditional MAGs PRO3, PRO5 and CFX7carry the CBB and WL pathways, respectively.
The Ammonia Oxidizing Guilds Of the core functional guilds, the AOB guild comprises threedifferent MAGs; while AOB2 had an AAI of 97% with Nitrosomonas europaea, AOB1and AOB3 are sufficiently divergent from known genomes to be separate species with Nitrosomonas eutropha and Nitrosomonas europaea as closest relatives (Fig S2). Although differential binning was unable to assign the amo and hao operons to the appropriate MAGs (see explanation in Fig S3), all AOB MAGs have the genes for the ammonia monooxygenase complex (amoCABDE) and hydroxylamine oxidase (hao), and typical Copper resistance/homeostasis (copCD) genes. Genes for NOx reduction were also present: The AOBMAGs contain a Copper-containing nitrite reductase (nirK) and cytochrome bc-type complex cNOR (respiratory nitric oxide reductase, norBC) (Table S1).
The AnAOB comprises threedifferent MAGs: AMX1 had an AAI of 99% with Candidatus Brocadia fulgida; AMX3had an AAI of 94% with Candidatus Brocadia sp. UTAMX1; AMX2 constitutes a new species, with Candidatus Brocadia sp. UTAMX2 as closest relative (Fig S2). UTAMX1 and UTAMX 2 were reported as dominant AnAOB in a similar study [19].
All AnAOB MAGs harbor the hydrazine dehydrogenase (hdh), hydrazine synthase (hzsABC) and nitrate oxidoreductase (nxrAB) genes (Table S2). A gene encoding nitrite reductase (nirK), the typical enzyme converting NO2- to NO was present in AMX1, but absent from both AMX2 and AMX3 (Table S2). Besides these genes, multiple copies of hao-like genes were present in all AMX MAGs: ten in AMX1 and AMX2 and six in AMX3 (Fig S4). Phylogenetic analysis indicates at least eighthao clusters congruent with published hao. Both AMX1 and AMX2 harbor threehomologous copies of a hao-like gene previously associated with hydroxylamine oxidation to nitric oxide [19], while AMX3 contains one copy. On the other hand, each of the AMX MAGs have threehomologous copies of a hao-like gene hypothesized to be involved in nitrite reduction to either nitric oxide or hydroxylamine [19] (Fig S4).No amoA-like sequences were found that could be assigned to non-autotrophic MAGs; however some haoA-like sequences were assigned to non-AOB MAGs (PRO3, PRO5, PRO11); PRO3 has two haoA gene paralogues – and both have as closest relative a gene found in Lautropia SCN 69-89, previously identified as abundant in nitration/anammox communities[20] and suggested as nitrite-denitrifier.
About half of the recovered MAGs (28/55) harbor genes homologous to nxr/nar (Table S2) comprising both the cytoplasmic and periplasmic NxrA/NarG encoding operons (Fig S5). All of those genes were phylogenetically distinct from those in previously characterized nitrite oxidizing bacteria, however, seven MAGs (PRO3, PRO6, PRO11, PRO12, CFX 1, CFX9, ARM1) encode for a periplasmic Nxr thatbelong to the NOB and AnAOB phyletic group (Figure S5). Whether these nxr-like genes encode for proteins involved in nitrite oxidation or nitrate reduction remains unknown. Except for PRO3, no MAG contains both anxr operon and a carbon fixation pathway, suggesting the absence of canonical NOBs, but possibility for non-autotrophic nitrite oxidation as noted for an ARM genome retrieved from a anammox microbiome[21].
Even though the recovered MAGs represented approximately 80% of the whole metagenome, we also screened the metagenomic reads to detect possible canonical nxr missed during the assembly or binning process. Some reads mapped against canonical nxr, especially from Nitrobacter spp., although the number was much lower compared to the reads mapping to AnAOB nxr (3.0 ± 0.8reads per million (RPM) vs 100.2 ± 24.6 RPM). An even lower number of reads mapped to nxr from other known NOB (Nitrolancea spp.: 0.7 ± 0.2 RPM; Nitrospira spp.: 0.3 ± 0.2 RPM; and 0 RPMtoCandidatusNitrotogaspp. and Nitrospinaespp.) (Table S3).
The heterotrophic guilds – NOx respiration
As more than half of the metagenome was heterotrophic (Fig 1, Table S2) it was examined in further detail, especially for its respiratory abilities towards nitrogen oxides (Fig 2).
Almost half of the MAGs (23/55) encode genes for respiratory nitrate reductase (narGHIJ, 28/55), one encodes theperiplasmic nitrate reductase gene (napAB), some carry thegenes for dissimilatory nitrite reduction to ammonia (DNRA) via the periplasmic nitrite reductase (nrfHA; 12/55) or the cytoplasmic nitrite reductase (nirBD; 2/55) (Table S2), although complete DNRA (carriage of nitrate reductases gene in addition to nrfHA or nirBD)) was found in just5 MAGs (ACD2, CFX8, CFX11, CLB2 and CLB3) (Table S2). If the above identified nxr-like genes (in ARM1 and CFX1) encode a nitrate-reductase, the number of MAGsthat can reduce nitrate increases to 25/55.
Of the MAGs with capacity for nitrate reduction, there is only oneMAG that encodes thecomplete set of genes for respiration of all reduced nitrogen oxides (the proteobacterial MAG PRO3, which carries, in addition tonarGHI, also nirS, norBC, and nosZ of the clade I type). Two of the narGH encoding MAGs (ACT2 at 0.23% and CFX15at 0.34% relative metagenome abundance)have no additional genes related to NOx respiration.
The two most abundant heterotrophic MAG, CFX2 (at 8.75% relative abundance) and CLB1 (at5.20%relative abundance), have, in addition, to a narGHoperon, solely a nirK and a nosZ (class II) gene, respectively, clearly indicating incomplete denitrification pathways.
While 23 MAGs encoded genes for respiratory nitrate reduction, 27 MAGs encoded genes for respiratory nitrite reduction(8 nirS, 19 nirK). Only 7 of these MAGs carried both nir and nar genes. 8 MAGs encoded genes for nitric oxide reduction (3 with norZ, 5 via norBC). Most of these also carried nir genes (7/8), but only half (4/8) carried both nir and nar genes. Only one MAG carried a class I nosZ gene (PRO 3). As stated above, PRO3 was the only MAG with all genes for a complete denitrification pathway. On the other hand, many MAGs encodeda class II nosZ gene (20/55), revealing a very high genomic potential for high-affinity N2O reduction. None of these MAGs carried genes for NO reduction (i.e. the norBC or norZ genes), while 7 of these MAGs carried genes for nitrite reduction (nirS or nirK, CFX6, PRO1, BCT2, PRO5, CFX9, BCT5, BCT6) or for nitrate reduction (narGHI, IGN1, IGN2, PRO5, BCG1, CFX8, BCT11, BCD2), respectively.
While MAGs with the capability of NO3- to NO2- reduction and N2O to N2 were abundant,MAGs with incomplete denitrification pathways prevailed (Figs 2 and 3). MAGs that were characterized as NO2-to N2O reducers consisted of AOB, consistentwith process observations [22]. In addition, a large number of MAGs with potential to exclusively reduce NO2-to NO or with potential for NO2-to NO, and N2O to N2 reduction were recovered. Several MAGs with potential to consume NO were retrieved, but especially the AnAOB MAGs AMX2 and AMX3 stand out as they did not encode the potential for NO2- reduction. The abundance of MAGs with genetic potential for NO production(and not consumption) suggest that NO is exchanged in the community; AMX2 and AMX3 are the obvious NO consumers, suggesting growth of AnAOB on NO (not NO2-) as electron acceptor as recentlydocumented[23]. While there are MAGs with the exclusive ability to reduce N2O, most of them also encode (nxr or nrf) genes allowing for NO3- to NO2- orNO2- to NH4+ reduction (Table S1).
Auxotrophy across MAGs
The heterotrophs in anammox or nitritation-anammox systems can either provide growth factors to [14] or depend on growth factors from [15] the autotrophic community members. Hence, we examined the genomic potential for synthesis of AA and B-vitaminsacross the MAGs. Of all the recovered MAGs, only one (PRO4) was fully prototrophic for its AA; all other MAGs were at least auxotrophic in one and upto all (PAT1) amino acids (Fig 4). The dominant autotrophs (AOB) were nearly prototrophic, all three missing the cysteine, plus the alanine (AOB2) and leucine (AOB1)biosynthesis pathway, respectively. Similarly, among the AnAOB, AMX1 and AMX2 only missed the methionine biosynthetic pathway, while AMX3, in addition, lacked complete histidine and proline biosynthetic pathways. Similarly, none of the MAGs was totally prototrophic for its B-vitamin synthesis, with some MAGs (PAT1, PAT3, PRO10, PLA1, VER1, VER2, VER3 GMM2) were completely devoid of this genetic potential (Fig 4). The potential for cobalamin biosynthesis (Vit B12) was only retrieved in the AnAOB MAGs AMX1 and AMX2. Correlations between the degree of prototrophy and MAG completeness or MAG abundance were low: both nearly complete MAGs (> 95%) and very rare MAGs (< 5%) ranged in AA prototrophy from 25 to 100%; yet the most abundant MAG (AMX2) was the most prototrophic (Fig S6)