Introduction

Anaerobic treatment systems offer many advantages such as biogas production in the form of methane and low sludge production. However, the effluents of anaerobic treatment systems still contain dissolved methane and ammonium that need to be removed. While there are several processes that can be used to remove ammonium from the anaerobic effluents, currently there are no established biological methods to remove dissolved methane, resulting in the release of this potent greenhouse gas into the atmosphere (van Kessel et al. 2018; ** to contigs were reassembled in CLC (word size 30, bubble size 5000). The completeness of the draft genomes and contamination were assessed by CheckM (Parks et al. 2015). BLAST was used to search for key genes in the ‘Ca. Methylomirabilis’ and anammox genomes. The contigs containing marker genes were manually curated. The curated contigs were checked using the visualization and annotation tool Artemis (Rutherford et al. 2000).

Results

Performance Lieshout WWTP

The carrousel of the WWTP Lieshout is a low-loaded system that receives on average 4700 m3 effluent per day from 3 UASB reactors. The total nitrogen supply to the carrousel is about 216 kg N per day and the average ammonium and nitrate concentrations of the UASB effluent are 5.4 and 0.9 mg N L−1, respectively. After treatment in the carrousel, the ammonium and nitrate concentrations are 0.7 and 1.2 mg N L−1, indicating presence of both nitrifying and denitrifying microorganisms. Therefore, the potential rates for ammonium, nitrite, and nitrate conversions were measured in batch tests (Table 3). The 4.7 mg N L−1 ammonium removal in the carrousel results in a removal rate of about 0.33 mmol NH4+ L−1 day−1 and is in good agreement with the potential ammonium-oxidation (0.26–0.3 mmol NH4+ L−1 day−1) measured in the batch tests. The nitrite-oxidation and nitrate reduction potential exceed these values 2- to 5-fold, indicating that aerobic ammonium oxidation might be the limiting step in the nitrogen removal in WWTP Lieshout. N-damo and anammox activity could not be detected within 36 h of incubation, possibly because of high back ground denitrification rates (Table 3).

Table 3 Nitrifying and denitrifying potential of Lieshout sludge
Full size table

Molecular survey Lieshout sludge

DNA of the Lieshout sludge was extracted and sequenced using Ion Torrent Technology. Extraction of 16S rRNA gene sequences (Fig. 1a) of the total metagenome yielded 2801 reads, the majority belonging to Proteobacteria (38%), Bacteroidetes (13%), Chloroflexi (12%), and Firmicutes (10%). Planctomycetes comprised 6% of the 16S rRNA reads, of which many corresponded to the order of Phycisphaerales. In the metagenome, no 16S rRNA reads belonging to anammox bacteria were retrieved, while 23 (0.8%) reads were affiliated to Nitrosomonas and 25 (0.9%) to aerobic methanotrophs (Methylococcocaea). Despite the absence of anammox reads in the metagenome, 16S rRNA and hydrazine synthase (hzsA) genes could readily be amplified using DNA from the Lieshout carrousel sludge as a template. The majority of anammox clones were affiliated to ‘Ca. Brocadia’ (Supplementary Fig. S1). Also, the pmoA gene of ‘Ca. Methylomirabilis’ could be amplified from the sludge DNA, and the diversity of these N-damo bacteria (Supplementary Fig. S2) was similar to previous studies based on nested 16S rRNA gene amplifications (Luesken et al. 2011c). The amplification of the mcrA gene showed a large diversity of methanogens (Supplementary Fig. S3) probably seeded into the carrousel from the UASB effluent. One of the 16 analyzed mcrA clones was affiliated to N-damo ‘Ca. Methanoperedens’ archaea.

Fig. 1
figure 1

Percentages of reads containing parts of 16S rRNA genes for different phyla extracted from the metagenomes of a the original Lieshout WWTP sludge and b the enrichment bioreactor metagenome after 14 months of operation, by map** on the Silva database

Full size image

Enrichment of N-damo and anammox bacteria

As anammox and N-damo bacteria were present in relatively low abundance, the excess biomass in the sludge needed to be mineralized as a strategy to reduce enrichment times. The first 3 days after inoculation of an anaerobic SBR, air was introduced into the system to convert easily degradable carbon and biomass. After the dissolved oxygen concentration started to increase, the air supply was turned off, and the SBR was fed with methane, ammonium, nitrate, and nitrite under anoxic conditions. After 100 days, the N-load to the reactor could gradually be increased to 5 mM NH4+, 8 mM NO2−, and 5 mM NO3− resulting in nitrite and ammonium consumption rates of 1.25 mmol L−1 day−1 and 1 mmol L−1 day−1, respectively (Fig. 2). When the reactor was run in batch mode without ammonium present, nitrite consumption (0.12 mmol h−1) at the expense of methane could readily be observed (Fig. 3a). In the absence of methane, anammox bacteria converted nitrite (0.5 mmol h−1) with ammonium as electron donor (Fig. 3b), indicating that anammox converted about 80% of the nitrite fed to the reactor while N-damo was responsible for the remaining 20%.

Fig. 2
figure 2

Conversion of ammonium and nitrite in the enrichment of N-damo and anammox from WWTP Lieshout

Full size image
Fig. 3
figure 3

a Conversion of methane and nitrite by N-damo. b Conversion of ammonium and nitrite by anammox in the Lieshout N-damo and anammox enrichment culture after roughly 42 months of operation

Full size image

The fate of 15N-labeled substrates was measured in a reactor batch test. According to expectation (Ettwig et al. 2010), labeled 15N-nitrite was predominantly converted to 30N2 gas (Fig. 4a) by the N-damo bacteria. In the anammox test, most of the 15N label of ammonium was converted in to 29N2 following the described stoichiometry (Strous et al. 1999b). In both incubation, some 28N2 from unlabeled nitrate was produced presumably via background denitrification, and therefore exact mass balanced could not be made.

Fig. 4
figure 4

a Labeled dinitrogen gas production from 15N nitrite with methane as electron donor by N-damo, without presence of ammonium. b Labeled dinitrogen gas production from 15N ammonium with nitrite by anammox, without methane present, in the Lieshout N-damo and anammox enrichment culture after 42 months of operation

Full size image

In accordance with the nitrogen conversions in reactor batch tests, ‘Ca. Methylomirabilis’ and anammox bacteria were the dominant members of the community as estimated from FISH microscopy (Fig. 5) and metagenomic analysis after 14 months of operation.

Fig. 5
figure 5

FISH microscopy of the N-damo and anammox enrichment. a After 200 days, anammox bacteria were shown to be present (amx820, pink) whereas N-damo bacteria could not be visualized by FISH (damo1027, green). b After 408 days of enrichment, both anammox bacteria (amx820, blue) and N-damo bacteria (damo1027, pink) were clearly present. Scale bar represents 20 μm

Full size image

Metagenomic analysis and classification based on the 16S rRNA gene

After 4.5 months of enrichment, genomic DNA was extracted and sequenced by Ion Torrent Technology to investigate enrichment of anaerobic ammonia and methane-oxidizing microorganisms. Not much changes were observed compared with the original sludge and we decided to continue the enrichment. After 14 months of enrichment, DNA was extracted once more to analyze the 16S rRNA gene diversity and to potentially assemble draft genomes. From this metagenome, 1996 16S rRNA reads were extracted. The phylogenetic classifications for groups with an abundance of greater than 1% of the total number of 16S rRNA gene reads are shown in Fig. 1b. More than 60% of the 16S rRNA gene reads belonged to three dominant groups: Brocadiaceae (21%), N-damo ‘Ca. Methylomirabilis’ bacteria (32%), and Phycisphaera-like Planctomycetes (10%), confirming the results of FISH microscopy.

The Ion Torrent reads were assembled into contigs, and contigs were binned by differential coverage and GC content. The binning resulted in 6 high-quality draft genomes (above 92% completeness; Table 4). Two draft genomes (Bins 1 and 2) represented distinct ‘Ca. Methylomirabilis’ species, closely related to ‘Ca. M. oxyfera’ (Bin2) and ‘Ca. M. lanthanidiphila’ (Bin1). Three bins showed high similarity to ‘Ca. Brocadia’ anammox species and further refinement was hampered by microdiversity differences. Interestingly, one draft genome (Bin6) represented an unknown genus, most closely related to the Phycisphaera species within the phylum Planctomycetes.

Table 4 Overview of bin quality, completeness, and characteristics
Full size table

The ‘Ca. Methylomirabilis‘ draft genomes (Bins 1 and 2) contained the diagnostic pmoCAB, xoxF, nirS, and putative NO dismutase genes. The pmoA marker genes encoding a subunit of the membrane-bound methane monooxygenase had 99.2% and 97.1% identity at the protein level to ‘Ca. M. oxyfera’ and ‘Ca. M. lanthanidiphila,’ respectively. We also identified a gene encoding the lanthanide-dependent methanol dehydrogenase (xoxF) in both genomes with 93–98% identity at the protein level to ‘Ca. Methylomirabilis’ species. Genes encoding the canonical calcium-dependent methanol dehydrogenase were absent in both genomes. Genes encoding the putative nitric oxide dismutase (nod) showed 85–90% amino acid identity to ‘Ca. Methylomirabilis’ species. The identified cd1 cytochrome nitrite reductase (nirS) had 92–99% identity to other ‘Ca. Methylomirabilis’ species.

The genomes assembled in Bins 3, 5, and 8 were determined to belong to the Planctomycetes phylum by the CheckM taxonomy assessment and most closely related to the anammox bacterium ‘Ca. Brocadia fulgida.’ This points to microdiversity of this species with bin3 as the most dominant (Supplementary Fig. S4). The genes encoding the crucial proteins involved in the anammox process were identified in the bins. These proteins include the hydrazine synthase, the hydrazine dehydrogenase, and multiple hydroxylamine oxidoreductases. The phylogenetic functional marker genes hzsA (Harhangi et al. 2012) encoding the alpha subunit of the hydrazine synthase had 89–93% identity on protein level to ‘Ca. Brocadia’ species.

The genome present in Bin6 was the only bin with complete 16S rRNA and 23S rRNA genes on one of the contigs. A BlastN search with the 16S rRNA gene revealed 99.3% identity with clones from anammox reactors treating wastewater (NCBI, unpublished sequences). The closest cultured representative with 92.4% identity was Phycisphaerae bacterium RAS2 isolated from a fish tank biofilter enrichment (Wiegand et al. 2020). The Prokka annotated genome does not reveal the potential for an autotrophic life style.

Discussion

Nitrate- and nitrite-dependent anaerobic oxidation of methane (N-damo) was discovered a decade ago, but the ecophysiological characterization of the microorganisms involved has been hindered by the slow growth of the responsible organisms (Raghoebarsing et al. 2006; Ettwig et al. 2010; Haroon et al. 2013). The N-damo microorganisms ‘Candidatus Methanoperedens nitroreducens’ archaea and ‘Ca. Methylomirabilis’ bacteria have been detected in various freshwater sediments (Welte et al. 2016). In this study, we started an enrichment culture fed with methane, ammonium, nitrate, and nitrite using carrousel sludge from a WWTP treating brewery effluent, as the inoculum. Based on activity measurements, FISH microscopy, and metagenome analyses after 14 months, the enrichment was dominated by ‘Ca. Methylomirabilis’ species and anammox bacteria. Previous enrichments were fed with mixtures of nitrite and nitrate without ammonium, to ensure that anammox bacteria would not outcompete ‘Ca. Methylomirabilis’ bacteria for nitrite (Luesken et al. 2011b; Shi et al. 2013; Hu et al. 2015; Ding et al. 2017; Vaksmaa et al. 2017b). Although the inoculum sludge had no detectable anammox or N-damo activity within 36 h, and only one 16S rRNA gene read resembling anammox could be detected in the initial metagenome, the diagnostic genes hzsA and pmoA could be readily amplified using DNA extracted from the sludge as a template. This indicated that the Lieshout carrousel sludge does harbor indigenous populations of N-damo and anammox bacteria and could serve as start-up material when large-scale installation need to be seeded (van Kessel et al. 2018). Indeed, after a start-up phase of about 100 days, an increasing consumption of nitrite, methane, and ammonium could be observed, and ultimately anammox and N-damo bacteria were the key players of the microbial community. The start-up phase was relatively short, as previous enrichments obtained from a minerotrophic peatland only showed significant methane oxidation rates after 9 months and the original enrichment of N-damo took more than a year before activity could be measured (Raghoebarsing et al. 2006; Zhu et al. 2012). In addition to substrate preference and availability, temperature has been implicated as a decisive factor in the outcome of N-damo enrichments. In enrichments started from wastewater treatment sludge and lake sediments, a co-enrichment of ‘Ca. Methylomirabilis’ bacteria and ‘Ca. Methanoperedens’ archaea was obtained at 35 °C, whereas at 22 °C, only N-damo bacteria were enriched (Hu et al. 2009). Furthermore, ‘Ca. Methanoperedens’ might be more sensitive to oxygen exposure (Guerrero-Cruz et al. 2018) than anammox and ‘Ca. Methylomirabilis’ bacteria (Luesken et al. 2011a, 2012). As we used air in the first 3 days of enrichment to degrade excess sludge, we may have strongly inhibited the few ‘Ca. Methanoperedens’ cells present and prevented their subsequent enrichment.

In the present enrichment, the cell numbers estimated by FISH correlated well with the metagenome sequencing results for ‘Ca. Methylomirabilis,’ with an abundance of about 30–40%. Anammox seemed somewhat underrepresented in the metagenome (20%) as activity measurements indicated that 80% of the nitrite conversion was accounted for by anammox activity. A similar observation was made previously where anammox converted about 70% of nitrite while FISH counts of anammox were less than 50% of the community (Luesken et al. 2011a). In our enrichment, there was always ammonium-limiting to ensure that ‘Ca. Methylomirabilis’ was not outcompeted by anammox (Hu et al. 2015). Incubations performed with the whole bioreactor revealed average ammonium oxidation and nitrite reduction after 14 months of 1 mmol day−1 L−1 and 1.25 mmol day−1 L−1, respectively which is a little bit lower compared with other enrichments (Luesken et al. 2011b; Hu et al. 2009; Vaksmaa et al. 2017b). With 15N isotope labeling experiments, we could show that in the absence of ammonium, a good methane-dependent nitrite conversion took place, and that the dominant product was the expected 30N2 gas (Ettwig et al. 2010). In the absence of methane, 15N-labeled ammonium was combined with 14N nitrite to form 29N2 by anammox bacteria in accordance with the anammox reaction equation (Strous et al. 1999b). As not all unlabeled nitrate could be removed from the reactor, some background denitrification and N2O production was observed as well, as has been reported for other N-damo enrichments (Vaksmaa et al. 2017b).

Metagenome analysis revealed that ‘Ca. Methylomirabilis,’ ‘Ca. Brocadia,’ and a Phycisphaera-like bacterium were the dominant members of the community after 14 months of enrichments. The diagnostic genes for nitrite-driven anaerobic ammonium and methane oxidation were identified in the genomes assembled from the metagenome sequences. To elucidate the role of the Phycisphaera-like bacterium, further research will be needed. Recently, members of the family Phycisphaeraceae have been reported to be potentially involved in anaerobic ammonium oxidation coupled to the reduction of sulfate (Sulfammox) or Fe(III) reduction (Feammox) in marine sediments (Rios-Del Toro et al. 2018a). Furthermore, representatives of this family were dominant in marine sediments from the eastern tropical North Pacific coast (Rios-Del Toro et al. 2018b). In this study, anaerobic ammonium oxidation linked to the microbial reduction of natural organic matter (NOM) fueled nitrogen loss. Other community members in our reactor system, mostly less than 5% of the population, belonged to various groups that have been reported before to be present in anaerobic methanotrophic enrichments (Vaksmaa et al. 2017b). In the enrichment, betaproteobacterial Rhodocyclaceae accounted for 6% and Comamonadaceae for 4% of the genome reads. They have both been implicated to perform denitrification, which may explain the observed back ground nitrate reduction, which was higher than expected based on the methane oxidation rate alone. Further significant numbers of Chloroflexi were present. They are obligate anaerobes that have previously been found in both anaerobic methanotrophic (Ettwig et al. 2009; Siniscalchi et al. 2015) and methanogenic enrichment cultures (Yamada et al. 2005; Gray et al. 2011; Liang et al. 2015). These Chloroflexi may degrade organic acids to formate, acetate, and hydrogen and feed the other community members (Hug et al. 2013).

In summary, we have enriched a N-damo and anammox coculture from the Lieshout WWTP carrousel sludge. Metagenome analysis, FISH microscopy, and activity test were in good agreement with each other that anammox contributed most to the nitrite conversion under ammonium limitation. The newly enriched co-culture will be used in future studies to unravel the ecophysiological properties of N-damo and anammox under oxygen limitation and investigate their potential role in more sustainable wastewater treatment systems.