Methylophaga and Hyphomicrobium can be used as target genera in monitoring saline water methanol-utilizing denitrification

Which bacterial taxonomic groups can be used in monitoring saline water methanol-utilizing denitrification and whether nitrate is transformed into N2 in the process are unclear. Therefore, methylotrophic bacterial communities of two efficiently functioning (nitrate/nitrite reduction was 63–96 %) tropical and cool seawater reactors at a public aquarium were investigated with clone library analysis and 454 pyrosequencing of the 16S rRNA genes. Transformation of nitrate into N2 was confirmed using 15N labeling in incubation of carrier material from the tropical reactor. Combining the data with previous study results, Methylophaga and Hyphomicrobium were determined to be suitable target genera for monitoring the function of saline water methanol-fed denitrification systems. However, monitoring was not possible at the single species level. Interestingly, potential nitrate-reducing methylotrophs within Filomicrobium and closely related Fil I and Fil II clusters were detected in the reactors suggesting that they also contributed to methylotrophic denitrification in the saline environment.


Introduction
Denitrification, step-wise reduction of water-soluble nitrate (NO3 -) via nitrite (NO2 -) to gaseous nitric oxide (NO), nitrous oxide (N2O) and di-nitrogen (N2) by facultative anaerobic heterotrophic bacteria, provides an important biotechnological water treatment process for nitrogen (N) removal.Denitrification is mostly a community process, as many denitrifiers perform only partial denitrification reducing NO3 -to NO2 -or to N2O, and only some bacterial species are capable of the whole denitrification chain from NO3 -to N2 gas [10].Due to the low C:N ratio of the influent water in many N removal systems, an external carbon source, usually methanol, is added to the process.Methanol-utilizing denitrification systems are widely applied in municipal wastewater treatment plants [18].
In addition, these systems are especially important for reducing toxic inorganic N compounds in closed marine facilities, for example, in public aquaria [21] and aquaculture farms [25].
The physicochemical and technical aspects of methanol-utilizing denitrification processes have been comprehensively characterized [18,24,25,26].However, denitrification is usually measured by NOx reduction, that is, the disappearance of NO3 - /NO2 - [23].This indirectly measured denitrification rate denotes the conversion of watersoluble NOx -into gaseous forms, but the proportions of NO, N2O and N2 in the endproduct are not specified.Thus far, only a few studies have been conducted on direct measurements of gaseous end-products [17,27].These studies mostly focused on N2O production [17,27], and the conversion of NO3 -to N2 has only very rarely been measured or even confirmed in water treatment facilities [6].
Optimal control and operation of wastewater treatment processes would also greatly benefit from microbiological data [27,43], such as monitoring the presence and abundance of taxonomic groups crucial for system function [27].The search for potential target taxonomic groups for monitoring saline water methanol-utilizing denitrification processes should be carried out in efficiently functioning systems and should focus on methylotrophic (C1-compound utilizing) organisms as they play a key role in the current processes, by utilizing methanol as an electron donor in denitrification and by transforming methanol into various extracellular organic compounds, which can then be utilized by co-occurring non-methylotrophic denitrifiers [27].Methylotrophic bacteria of the genera Methylophaga and Hyphomicrobium have been shown to dominate the two previously studied saline water methanol-fed denitrification systems: a moving bed biofilm reactor at a seawater aquarium [3,21,22] and a laboratory-scale continuously stirred tank reactor (CSTR) that treat synthetic saline wastewater [32].In addition to these two genera, Azoarcus and Paracoccus were important methylotrophs in CSTR [32].
Further analyses in the aquarium showed the genetic potential of Methylophaga to reduce NO3 -to NO2 -and Hyphomicrobium to complete the denitrification by converting NO2 - into N2 [2,3,42].However, physicochemical and biological variations among bioreactors could lead to differences in the community composition.Thus, other methylotrophs could be important for the function of these systems given that methylotrophy is a quite widely dispersed trait among bacteria [19], and many can also conduct partial or complete denitrification [4,5,8,12,20,45].Therefore, more studies are needed to determine suitable target taxonomic groups for monitoring the function of saline water methanol-fed denitrification processes.
This study investigated NOx -reduction and the bacterial communities of two methanol-utilizing denitrifying bioreactors in a marine fish aquarium operating at two water circulation temperatures.We aimed to confirm the microbiological transformation of NO3 -to N2 using batch incubations and the 15 N tracer technique.Furthermore, we aimed to find suitable target taxonomic groups for monitoring the function of saline water methanol-utilizing denitrification processes.This was achieved via a clone library and 454 pyrosequencing analysis of the 16S rRNA genes and comparison of the results for the two study reactors to those for previously studied systems [3,21,32].We specifically focused on the analysis of taxonomic groups that harbor known methylotrophs.

Sampling site and sampling for molecular microbiology
The public fish aquarium SEA LIFE Helsinki Finland

Denitrification measurements
The NOx -concentrations inside the reactors, near the reactor outlet (NOx - out) and in the inflow water feeding the reactors (NOx and as the actual NOx -reduction: which was converted into the NOx -reduction rate of the carrier material (µmol N/Lcar/h).
Direct denitrification measurement was performed with bottle incubation in 2010.
Batches of oolitic sand (~36 mL) collected from the middle of the filter bed in AQUAR_T were put in 120 mL glass bottles (11 bottles altogether).The bottles were filled with anoxic reactor water by submerging them in the reactor, and while submerged, they were closed with caps that had butyl rubber stoppers.Each bottle was injected with ~219 µmol (~7 mg) of methanol and ~50 µmol (~0.7 mg) of NO3 --N, which had a 2% 15  High Wycombe, UK), and the microbial activity in the vials was terminated by adding 0.1 mL of ZnCl2 (1 g/mL).The concentration and 15 N content of the N2 gas in the water was measured as in Tiirola et al. [41].Denitrification was calculated as the rate of the total N2 gas accumulation and converted to the N2 production rate of the carrier material (µmol N/Lcar/h).In addition, reduction of NO3 -to N2 gas was verified by the accumulation of excess 15 N-containing N2 gas.The concentration of the excess 15 N-containing N2 gas, [excess 15 N], was calculated for each sample as [ 15 ] = (% 15 sample × [2sample] − % 15 zero × [2sample]) 100 where [N2sample] is the N2 gas concentration in the incubated sample and the at% 15 Nsample and the at% 15 Nzero are the 15 N content (in %) of the N2 gas in the incubated and nonincubated (0 time point) samples, respectively.

Molecular microbiological analyses
DNA was extracted from 0.5-0.6 g of frozen oolitic sand from each sample using glass bead beating and phenol-chlorophorm extraction, which was followed by isopropanol-NaCl precipitation (pH 8) and dissolution of the DNA pellet in TE buffer.
PCR for the clone library analyses of the 16S rRNA genes was performed with primers 27F (5´-AGAGTTTGATCMTGGCTCAG-3´) / 907R (5´-CCGTCAATTCMTTTGAGTTT-3´) as previously described [36] but using the following program: initial denaturation at 95 °C for 5 min and 30 cycles of amplification (94 °C for 30 s, 53 °C for 1 min, 72 °C for 3 min) and final elongation at 72 °C for 15 min.The PCR products of the replicate samples were pooled for subsequent cloning and sequencing, which was carried out as previously described [36].
To get deeper insight into the bacteria inhabiting the methanol-fed denitrification reactors, the bacterial community in AQUAR_T, which used methanol as the sole carbon source, was studied with 454 pyrosequencing.Equal amounts of nucleic acid extracts from replicate samples of AQUAR_T in 2008 were pooled before the PCR reactions.PCR amplification of the 16S rRNA genes using primers 341F (5´-CCTACGGGNGGCWGCAG-3´) / 805R (5´-GACTACHVGGGTATCTAATCC-3´), subsequent purification steps and sequencing were performed as previously described [33].

Sequence analysis
In the analysis of the clone library sequences, the Mothur program package [38] was used for sequence alignment, chimera-checking (chimera.uchimeexecutable), classification of sequences into operational taxonomic units (OTUs; 97 % identity threshold) and taxonomic classification of the OTUs (using the Ribosomal Database Project database).
Variations in the community structure among the samples were analyzed with hierarchical clustering (UPGMA linkage, Bray-Curtis distances) using PAST version 3.09 [13].
Tags and primer sequences, as well as low-quality sequences (containing ambiguous nucleotides and homopolymers longer than eight nucleotides) were removed from the 454 pyrosequencing library.Sequences (~200 bp) were thereafter processed as described above for the clone library analysis except phylogenetic trees were not constructed.
The 16S rRNA gene sequences of the clone libraries were deposited in the NCBI GenBank (accession numbers KP098736-KP098970, KP098976-KP098984 and KP098989-KP099005).The 454 pyrosequencing data were deposited in the NCBI SRA database (SRX646347).

Functioning of denitrifying reactors
The operator-adjusted Wf and Metf were lower in 2010 than in 2008 in both reactors during the 1.5 to 2 month period before bacterial sampling (Table 1).The average hourly NOx -load in the inflow decreased in AQUAR_T and increased in AQUAR_C from 2008 to 2010 (Table 1).The average molar ratio of the added methanol to the N load methanol:NOx --Ninflow (mol:mol) decreased from 2008 to 2010 in both reactors (Table 1).
The ratio was higher in AQUAR_C than in AQUAR_T in 2008 and vice versa in 2010 (Table 1).1).
Bottle incubation tests confirmed that N2 was produced and NO3 -was transformed into N2 by microbes in AQUAR_T (Fig. 2).The estimated N2 production rate of the carrier material was 56.4 µmol N/Lcar/h.

Variation in microbial community structure
The clone library analysis of the 16S rRNA genes indicated that the variation in the bacterial community structure was higher between the reactors than between the years (Table 2; Online Resource 1).Variation between the years in the community structure was considerably higher in the cooler reactor AQUAR_C, which was first fed with a mixture of methanol and saccharose, than in the warmer reactor AQUAR_T (Table 2; Online Resource 1).
Different taxonomic groups were assigned to a putative methylotrophic function based on the previous literature.The methylotrophic groups detected in the reactors were the genera Methylophaga, Hyphomicrobium, Filomicrobium, as well as clusters Fil I and Fil II, which were more closely related to Filomicrobium than to Hyphomicrobium (Table 2; Figs. 3 and 4).Betaproteobacterial methylotrophs were not detected.In the phylogenetic tree, the OTUs assigned to the genus Hyphomicrobium were positioned between the previously assigned Hyphomicrobium clusters I and II [35] (Fig. 3).Most of the OTUs and sequences assigned to Methylophaga clustered close to M.
nitratireducenticrescens, whereas one of the OTUs had its closest relative in M.
The relative abundance of the putative methylotrophs was considerably higher in the warmer reactor AQUAR_T than in the cooler reactor AQUAR_C in both years (Table 2).In addition, methylotrophs were more abundant in 2010 than in 2008 in AQUAR_C whereas the opposite took place in AQUAR_T (Table 2).The relative abundance of Methylophaga was considerably higher in the warmer reactor AQUAR_T than in the cooler reactor AQUAR_C (Table 2).In AQUAR_C, Methylophaga were absent in 2008 and present at very low numbers in 2010.Correspondingly, there was a higher abundance of Methylophaga in 2010 than in 2008 in AQUAR_T (Table 2).Methylophaga in AQUAR_T had their closest relative in M. nitratireducenticrescens, whereas those in AQUAR_C were most closely related to M. thiooxydans (Fig. 4).Of the family Hyphomicrobiaceae, Hyphomicrobium were much more abundant in the warmer reactor AQUAR_T than in the cooler reactor AQUAR_C.There was a higher abundance of Hyphomicrobium in 2010 than in 2008 in AQUAR_C and vice versa in AQUAR_T (Table 2).In contrast, the bacteria of the Fil I cluster were much more abundant in AQUAR_C than in AQUAR_T (Table 2).In AQUAR_T, Fil I bacteria were absent in 2008 and present at very low abundance in 2010.Correspondingly, there was a higher abundance of Fil I bacteria in 2010 than in 2008 in AQUAR_C.Fil II cluster bacteria were present in low abundance in both reactors and were absent in AQUAR_T in 2010 (Table 2).
Filomicrobium cluster bacteria were present in low abundance and only in AQUAR_T (Table 2; Fig. 3).
The relative abundance of other bacteria also varied between reactors and years and was generally higher in AQUAR_C than in AQUAR_T (Table 2).Within Gammaproteobacteria, there was a considerably large group of unclassified 16S rRNA gene sequences that probably represented clades without cultured representatives (Table 2).Of Alphaproteobacteria (other than methylotrophs), Rhodobacteraceae and Phyllobacteriaceae as well as non-methylotrophic Hyphomicrobiaceae within Maritalea and Mar I clusters were detected (Table 2; Fig. 3).According to BLAST searches [1] of the NCBI nr-database, the detected Rhodobacteraceae were most closely related to the non-methylotrophic genus Roseovarius.Of the Proteobacteria, also Deltaproteobacteria were present.Other detected phyla were Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Deferribacteres, Planctomycetes and Spirochaetes (Table 2).
There were differences in the results of the clone library analysis and 454 pyrosequencing (Table 2).Most importantly, 454 pyrosequencing had lower resolution to detect Alphaproteobacteria or resolve different genera within Hyphomicrobiaceae and resulted in a higher proportion of unclassified bacterial sequences than the clone library analysis (Table 2).Since high taxonomic resolution was necessary to identify clusters especially within Hyphomicrobiaceae, the focus in this study was on the clone library results.

Discussion
Both study reactors removed N very efficiently.The relative NOx -reduction was even slightly higher than measured at other methanol-fed sea water denitrification bioreactors that reduced 65-90 % [22] and up to 88 % [24] of NOx -.In contrast, the average actual NOx -reduction rates were within the lower end of the range (300-9000 µmol N/Lcar/h) measured at other systems [21,23,24].The differences in the actual NOx -reduction rates are mostly due to 3-10 times higher N loads (as expressed per carrier volume) in the previously studied reactors [23,24].However, the differences can be also partially attributed to the use of deaeration systems to remove O2 in the inflow and the amendment of the trace metal solution in the previously studied reactors, which enhanced denitrification [21,23].To our knowledge, this is the first study of saline water methanolutilizing denitrification systems that confirmed NO3 -is reduced to N2.However, the N2 production measured in the batch tests was only 6-7 % of the actual NOx -reduction rates in the AQUAR_T reactor.This difference is very likely due to the lack of shaking during incubation, which reduced the contact between NO3 -and the bacteria on the carrier material.However, the possible formation of other gases, N2O [17] and NO, during the incubation periods can also explain a small part of this discrepancy.
Combining the results from the two study reactors and the two previously studied systems [3,21,32] shows that putative methylotrophic bacteria belonging to Methylophaga and Hyphomicrobium are very common in efficiently functioning saline water methanol-fed denitrification systems.This result indicates that these bacteria are crucial for the function of the process.Thus, these two genera can serve as targets when monitoring the function of saline-water methanol-utilizing denitrification systems.
However, there were considerable species-level differences between the systems.
Hyphomicrobium nitrativorans, which dominated in a previously studied system [3,21], were not found at all in the AQUAR reactors, and Methylophaga nitratireducenticrescens, which inhabited a previously studied reactor [3] and AQUAR_T, were not found in AQUAR_C.Thus, no single Methylophaga or Hyphomicrobium species can be determined for use as a general target species for monitoring the function of saline water methanol-fed denitrification systems.
Interestingly, this study is also the first to show that members of Filomicrobium Previous studies also suggest that Filomicrobium participate in denitrification by dissimilatory reduction of NO3 -to NO2 - [14,45].Thus, in addition to Methylophaga [2,3], Filomicrobium, Fil I and Fil II bacterial clusters could couple methylotrophy with the reduction of NO3 -to NO2 -and supply it for denitrifiers capable of further denitrification steps [3].However, previously studied strains of Methylophaga thiooxydans [7] and M.
nitratireducenticrescens [42] also had the NO2 -reductase gene, although it was truncated in M. nitratireducenticrescens [42].In addition, M. nitratireducenticrescens had genes for NO and N2O -reduction [42].This suggests that Methylophaga species coupling methylotrophy to NO2 -, NO and N2O reduction could exist in nature and in saline water methanol-fed denitrification systems.Since only a few Filomicrobium strains have been characterized for their metabolic potential [14,45], it is possible that some species within Filomicrobium, Fil I and Fil II clusters are also capable of this.
Thorough comparative analyses of the factors that affect the growth and activity of Filomicrobium and Hyphomicrobium species do not exist.Therefore, many possible factors might have affected the genera-and species-level variation of Hyphomicrobiaceae between the AQUAR reactors and previously studied systems [3,21].In a comparison of denitrifying Hyphomicrobium species, NO3 -significantly affected the growth and activity of Hyphomicrobium species with H. nitrativorans growing and denitrifying at higher NO3 - concentrations than the other studied species, H. zavarzinii and H. denitrificans [30].
Thus, the considerably lower N loads could explain the presence and dominance of Hyphomicrobium species other than H. nitrativorans in the AQUAR reactors.Similarly, Filomicrobium, Fil I and Fil II bacteria could have been favored in the lower N load conditions that prevail in the AQUAR system.Furthermore, O2 constantly enters AQUAR reactors due to the lack of a preceding deaeration step but is rapidly consumed there based on the anoxic conditions ([O2] < 0.1 mg/L, below the detection limit of the O2 probe) that prevail inside the reactors [9].Thus, variation in the O2 availability between the systems can also explain the differences in the bacterial communities.Part of the Hyphomicrobium and Filomicrobium OTUs in the AQUAR reactors may have been favored by O2.
Analogously, as in aerobic methane oxidation coupled with denitrification (AME-D) [46], these bacteria could have contributed to the overall denitrification performance by consuming O2 and by aerobically converting methanol to organic substrates utilizable by non-methylotrophic denitrifiers.However, in addition to variating NO3 -and O2, differences in the inocula (the original bacterial community that colonized the reactors) and in the carrier materials between the reactors as well as the amendment of the trace metal solution and the higher addition of methanol in the previously studied systems [3,21,22] could have also affected the differences in the bacterial communities.
The lower relative abundance of methylotrophs in the cooler reactor AQUAR_C than in AQUAR_T can be first explained by the saccharose addition in 2008 leading to over-competition of non-methylotrophic organisms in AQUAR_C.In 2010, the difference can be attributed to the lower availability of methanol (lower Metf and methanol:NOx - inflow) in AQUAR_C.Variation in the availability of methanol can also explain the lower abundance of Hyphomicrobium and Methylophaga and the higher abundance of Fil I bacteria in AQUAR_C than in AQUAR_T.However, differences in temperature could also play a role here since the lowest limits of the growth temperature ranges of Methylophaga (M.nitratireducenticrescens, 15-37 °C) [42] and Hyphomicrobium (H. nitrativorans, 15-35 °C) [29] are higher than that of Filomicrobium (e.g., F. insigne, 4-45 °C) [45], which indicates that Filomicrobium-related Fil I bacteria grew better in the cooler AQUAR_C reactor.The growth temperature range of M. thiooxydans, the closest cultured relative of Methylophaga in AQUAR_C, is not known.However, the absence of M. thiooxydans in the previously studied sea water reactor with a slightly lower temperature, 16-18 °C [3], than in AQUAR_C, 18-19 °C, suggests that temperature variations do not explain the differences in the Methylophaga species between AQUAR_C and AQUAR_T.M. thiooxydans can grow on a wider range of carbon substrates (methanol, fructose, monomethylamine) than M. nitratireducenticrescens (methanol) [42].
Thus, differences in the Methylophaga species between the reactors could be explained by M. thiooxydans surviving better with lower methanol availability.
Many other detected phyla, that is, Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Deferribacteres, Planctomycetes and Proteobacteria (other than the detected methylotrophs), contain species capable of either partial or complete denitrification [11,27,37,39,44] suggesting that they contribute to the overall denitrification performance of the AQUAR reactors and utilize the organic compounds produced by methylotrophs.
However, recent cultivation studies have shown that methylotrophy is present within Bacteroidetes [28] and Actinobacteria [15].In addition, DNA stable isotopic probing (SIP) analyses, although not able to distinguish between direct C1 utilization and crossfeeding, raise speculation about the existence of unclassified methylotrophic Gammaproteobacteria [31], as well as methylotrophic members within Bacteroidetes, Deferribacteres [16] and Acidobacteria [34].Further studies are needed to show whether members of these groups couple methylotrophy with denitrification in saline water systems and could also be considered potential target taxonomic groups for monitoring purposes.

Conclusions
This study confirmed that NO3 was transformed into N2 in a functioning saline water methanol-fed denitrification system.Combining the results of this study with those of previous studies show that Methylophaga and Hyphomicrobium can serve as target genera in monitoring the function of saline water methanol-utilizing denitrification systems.
However, species-level differences among the systems, stemming from differences in the physicochemical and biological conditions, prevent the determination of individual Methylophaga or Hyphomicrobium species to be used as general target species in process monitoring.Interestingly, other bacteria with potential methylotrophic and NO3 --reducing metabolism, within Filomicrobium and Fil I and Fil II clusters, were found, suggesting they also contributed to saline water methylotrophic denitrification.Whether bacteria within Methylophaga, Filomicrobium, Fil I and Fil II clusters can only reduce NO3 -to NO2 -or reduce NO2 -, NO and N2O, and whether other detected bacteria are able to couple methylotrophy with denitrification in the saline water denitrification bioreactors, is still to be resolved.These studies should employ high-throughput culturing methods to isolate target organisms, as well as metagenomics and metatranscriptomics.

Conflict of interest
The authors declare that they have no conflict of interest.

(
https://www.visitsealife.com/helsinki/) is divided into two water recirculation systems (warm tropical, AQUAR_T, and cold North Atlantic, AQUAR_C) and has a total water volume of 420 m 3 with 27-34 ppt salinity.Both circulation systems have their own fluidized-bed type denitrification reactors that receive NO3 --rich water from the nitrification stage.The volume of each reactor is ~330 L of which ~90 L is the fluidized carrier material bed (oolitic sand, 100 kg, density = 1.5 kg/L).The temperature inside the denitrification reactors of AQUAR_T and AQUAR_C was 23-24 °C and 18-19 °C, respectively.Two samples of oolitic sand for molecular microbiological analyses were collected twice, on 10 November 2008 and 8 September 2010, from the AQUAR_T and AQUAR_C reactors, in sterile 50 mL plastic containers and stored at -20 °C before processing within 1 to 2 months.The reactors utilized methanol as their carbon source except AQUAR_C in 2008 when a mixture of methanol and saccharose was used until the sampling time point after which only methanol was used.
sp. and Filomicrobium-related Fil I and Fil II clusters can be abundant in efficiently functioning saline water methanol-fed denitrification systems.In addition to results from cultivation-based[45] and genomic[14] studies of Filomicrobium strains, further confirmation of the bacteria's methylotrophic metabolism was provided by the concurrent increase in the relative abundance of Hyphomicrobium, Methylophaga and Fil I bacteria in AQUAR_C from 2008 to 2010 when the C source changed from saccharose+methanol to solely methanol.In 2008, the methylotrophs were probably over-competed by more efficiently growing non-methylotrophs, for example, in Bacteroidetes, Rhodobacteraceae and Maritalea.Furthermore, the concurrent decrease in Hyphomicrobium and Fil II bacteria in AQUAR_T from 2008 to 2010 could due to the decrease in the availability of methanol (a decrease in Metf and methanol:NOx-Ninflow before sampling).

Table 2
Bacterial community composition (% of sequences) in denitrifying reactors of sea water aquarium (AQUAR_T and AQUAR_C) based on 16S rRNA 635 gene clone libraries and 454 pyrosequencing.Methanol was used as the sole external carbon source except for AQUAR_C in 2008, when a mixture of