Methylophilaceae and Hyphomicrobium as target taxonomic groups in monitoring the function of methanol-fed denitrification biofilters in municipal wastewater treatment plants

Molecular monitoring of bacterial communities can explain and predict the stability of bioprocesses in varying physicochemical conditions. To study methanol-fed denitrification biofilters of municipal wastewater treatment plants, bacterial communities of two full-scale biofilters were compared through fingerprinting and sequencing of the 16S rRNA genes. Additionally, 16S rRNA gene fingerprinting was used for 10-week temporal monitoring of the bacterial community in one of the biofilters. Combining the data with previous study results, the family Methylophilaceae and genus Hyphomicrobium were determined as suitable target groups for monitoring. An increase in the relative abundance of Hyphomicrobium-related biomarkers occurred simultaneously with increases in water flow, NO x − load, and methanol addition, as well as a higher denitrification rate, although the dominating biomarkers linked to Methylophilaceae showed an opposite pattern. The results indicate that during increased loading, stability of the bioprocess is maintained by selection of more efficient denitrifier populations, and this progress can be analyzed using simple molecular fingerprinting.


29
Denitrification is an essential biotechnological process in municipal wastewater 30 treatment plants (WWTPs) for reducing the nitrogen (N) load to recipient waters. This 31 step-wise reduction of water-soluble nitrate (NO3 -) via nitrite (NO2 -) to gaseous nitric 32 oxide (NO), nitrous oxide (N2O), and di-nitrogen (N2) is catalyzed by facultative 33 anaerobic heterotrophic bacteria. Denitrification is a community process, as many 34 denitrifiers perform only a portion of the reduction steps, reducing NO3to NO2or to 35 N2O, and only some bacterial species are capable of the whole denitrification chain 36 from NO3to N2 gas [8]. Due to the unfavorably low carbon-to-nitrogen (C:N) ratio of 37 the water in many N removal systems, an additional organic C and energy source, 38 usually methanol, is used in the process. In WWTPs, methanol-fed denitrification is 39 often accomplished by filtration of the wastewater through a support material in 40 biofilters [17]. 41 The physicochemical and technical aspects of the methanol-utilizing 42 denitrification processes have been comprehensively characterized [17,20]. However, 43 the optimal control and operation of the processes would also benefit greatly from 44 microbiological data [22,39], such as the identity and potential controlling factors of 45 the taxonomic groups crucial for the system function, which could be used in process 46 monitoring [22]. Methylotrophs play a key role in methanol-fed denitrification systems, 47 biofilters; 2) whether variations in physicochemical conditions affect the bacterial 79 community structure; and 3) whether methylotrophs and non-methylotrophs as well as 80 4) different taxonomic groups of methylotrophs respond differently to these variations. 81

83
Microbiological sampling 84 85 Samples were collected from the methanol-fed denitrification filters of two municipal 86 wastewater treatment plants: the Viikinmäki wastewater treatment plant in Helsinki, 87 Finland (WWTPA), and the Salo wastewater treatment plant in Salo, Finland 88 (WWTPB) ( Table 1). WWTPA is a large plant with one of the largest denitrification 89 filter systems in the world, whereas WWTPB is a small-sized plant (Table 1). and O2in and NOxout, PO4 3out, and SSout were calculated for the whole denitrification 119 system. The NOxload (µmol s -1 ) in the inflow (LNOxin) and outflow (LNOxout) water 120 was calculated from Wf and NOxin or NOxout. Denitrification in the filters was 121 calculated either as relative (%) or actual (µmol s -1 ) NOxreduction as follows: 122

191
Performance of the denitrification biofilters 192 193 As is typical for WWTPs in Northern countries in autumn, Wf increased and T 194 decreased during the study period in both filter systems ( compared to WWTPA could be due to possible differences in the total N concentrations 200 feeding the WWTPs, the nitrification efficiency between WWTPA and WWTPB, or the 201 lack of a pre-denitrification system in WWTPB (Table 1). In the filters, Metf is 202 controlled by a feedback loop that controls the NO3-N concentration inside the filter 203 cells [7]. As a result, Metf followed LNOxin tightly, and they both controlled the actual 204 NOxreduction rate (µmol/s) in the systems (Fig. 1   Samples of the sheared biomass in the backwash water were used in comparing 232 the methylotrophic communities between WWTPA and WWTPB. The relative 233 abundance of putative methylotrophs was much higher in WWTPB than in WWTPA 234 ( Hyphomicrobiaceae had a much higher relative abundance in WWTPA than in 239 WWTPB, whereas the opposite was observed for Methylophilaceae (Table 2). In 240 contrast to the backwash sample, the carrier material of WWTPB did not harbor 241 Paracoccus or Methyloversatilis but rather Bradyrhizobium. The carrier material of 242 WWTPB also had a higher and lower relative abundance of Hyphomicrobiacea and 243 Methylophilacea, respectively, than the backwash material of WWTPB (Table 2). Methylotenera with no cultured representatives so far. The fourth group included two 260 rare OTUs that were not closely affiliated to known Methylophilaceae genera (Table 2,  Methylococcaceae, Acinetobacter, and Flavobacterium in WWTPA (Table 2). 454-276 pyrosequencing also resulted in a higher proportion of unclassified bacterial sequences 277 than the clone library analysis (Table 2). 278 The abundant non-methylotrophic bacterial groups (≥ 5 % of 16S rRNA 279 sequences in any of the libraries) included Acidobacteria, Actinobacteria, Bacteroidetes 280 (other than Flavobacterium), Chloroflexi, Comamonadaceae, Deltaproteobacteria, 281 Planctomycetes, and Rhodocyclaceae (other than Methyloversatilis) ( Table 2). 282

283
Temporal variation in the bacterial community in the WWTPA biofilter 284 285 The bacterial community structure changed over time (non-metric multidimensional 286 scaling analysis, Fig. 1), along with a temporal change in several operational parameters 287 (Fig. 1). The fluctuations in the community structure were correlated with variations in 288 Wf (Mantel's test, r = 0.36, p < 0.05, n = 10), LNOxin (r = 0.61, p < 0.05, n = 10), Metf (r 289 = 0.55, p < 0.05, n = 10), and T (r = 0.59, p < 0.05, n = 10). In addition, the community 290 structure correlated with the actual NOxreduction rate (r = 0.62, p < 0.05). 291 To study the variation of the methylotrophic taxa in WWTPA, the phylogenetic 292 classification was linked to the LH-PCR peaks in silico using the length and taxonomic 293 data obtained from the clone library analyses (Online Resource 3). All the clone library 294 sequences with a size of 466 bp in the area amplifiable by LH-PCR primers belonged to 295 OTU 16 within the Hyphomicrobium II cluster, and all the sequences of genus 296 Hyphomicrobium had the size of this peak (see Fig. 2). The sequences assigned to 297 Methylophilaceae were found only within peaks 521 bp and 524 bp, and they dominated 298 only within peak 521 bp (73 %), which was also the largest peak in the LH-PCR Methylophilaceae, respectively. Furthermore, the sum of LH-PCR peaks 466 bp and 305 521 bp were used as a general biomarker for methylotrophs, whereas the sum of all 306 peaks excluding methylotrophic peaks 466 bp, 521 bp, and 524 bp (see above) were 307 used as a biomarker for non-methylotrophs. 308 During the study period, there was a negative correlation between the relative 309 abundances of Hyphomicrobium and Methylophilaceae (r = -0.91, p < 0.001) (Fig. 4). 310 The relative abundance of Hyphomicrobium increased as Metf, Wf, and LNOxin 311 increased (Metf: r = 0.74, p < 0.05; Wf, ρ = 0.67, p < 0.05; LNOxin, r = 0.80, p < 0.05, n 312 = 10) (Figs. 1 & 4), while the opposite took place with Methylophilaceae (Metf: r = -313 0.74, p < 0.05; Wf, ρ = -0.66, p < 0.05; LNOxin, r = -0.77, p < 0.05, n = 10). The relative 314 abundance of Methylophilaceae also increased as T increased (r = 0.67, p < 0.05, n = 315 10), while there was no correlation between T and Hyphomicrobium (r = -0.62, p = 316 0.06, n = 10) (Fig. 4). The relative abundance of total methylotrophs decreased as Metf 317 Capable of aerobic denitrification, Paracoccus tolerates O2 better than 383 Hyphomicrobium, which thrive in anoxic conditions, and thus Paracoccus were favored 384 in the surface zones of the biofilm in a previously studied full-scale biofilter (a sand 385 filter) [21]. This is in accordance with our results on the higher and lower relative 386 abundance of Paracoccus and Hyphomicrobium, respectively, in the sheared biomass of 387 the backwash water (representing more aerobic surface biofilm) than in the carrier 388 material (representing deeper anoxic biofilm) in WWTPB. Similarly, the lower O2 load 389 (as expressed per carrier volume) could explain the higher abundance of 390 Hyphomicrobium and the absence of Paracoccus in WWTPA. Since some 391 Methylotenera strains are aerobic [3,14] or perform aerobic denitrification [25], the 392 higher abundance of Methylophilaceae in the sheared biomass than in the carrier 393 material could also be due to differences in O2 availability. However, it could also be 394 due to differences in NOxand methanol availability, which is expected to be higher in 395 the biofilm surface. The results indicate that Cluster Met I, which was the sole 396 Methylophilaceae group in the sheared biomass of WWTPB, was especially favored by 397 the higher availability of O2, NOx -, and/or methanol. Therefore, the lower O2, NOx -, and Their presence in WWTPB but not in WWTPA might also reflect higher temporal 410 variation in the availability of methanol or higher and temporally more variable 411 availability of other C sources (present in feed water or produced from methanol) in 412

WWTPB. 413
In accordance with the results from the comparison of the biofilters, many 414 possible physicochemical factors might have affected the temporal variation in the 415 bacterial community structure within the WWTPA biofilter. The overall bacterial 416 community structure changed due to variations in the availability of electron acceptors 417 (NOx -) and donors (methanol) as well as in temperature, which has also previously been 418 shown to affect denitrifying communities [9, 40]. In addition, changes in the water flow, 419 which act through changing the HRT and surface load, possibly affected the community 420 structure. However, due to the covariation among these factors (Fig. 1)

693
The peaks assigned to Hyphomicrobium and Methylophilaceae at WWTPA are marked by arrows 694