Assessing the composition of microbial communities in textile wastewater treatment plants in comparison with municipal wastewater treatment plants

Abstract It is assumed that microbial communities involved in the biological treatment of different wastewaters having a different chemical composition harbor different microbial populations which are specifically adapted to the environmental stresses encountered in these systems. Yet, little is known about the composition of these microbial communities. Therefore, the aim of this study was to assess the microbial community composition over two seasons (winter and summer) in activated sludge from well‐operating textile wastewater treatment plants (WWTPs) in comparison with municipal WWTPs, and to explain observed differences by environmental variables. 454‐pyrosequencing generated 160 archaeal and 1645 bacterial species‐level Operational Taxonomic Units (OTUs), with lower observed richness in activated sludge from textile WWTPs compared to municipal WWTPs. The bacterial phyla Planctomycetes, Chloroflexi, Chlorobi, and Acidobacteria were more abundant in activated sludge samples from textile WWTPs, together with archaeal members of Thaumarchaeota. Nonmetric multidimensional scaling analysis of the microbial communities showed that microbial communities from textile and municipal WWTPs were significantly different, with a seasonal effect on archaea. Nitrifying and denitrifying bacteria as well as phosphate‐accumulation bacteria were more abundant in municipal WWTPs, while sulfate‐reducing bacteria were almost only detected in textile WWTPs. Additionally, microbial communities from textile WWTPs were more dissimilar than those of municipal WWTPs, possibly due to a wider diversity in environmental stresses to which microbial communities in textile WWTPs are subjected to. High salinity, high organic loads, and a higher water temperature were important potential variables driving the microbial community composition in textile WWTPs. This study provides a general view on the composition of microbial communities in activated sludge of textile WWTPs, and may provide novel insights for identifying key players performing important functions in the purification of textile wastewaters.


| INTRODUCTION
Wastewater treatment using activated sludge processes has been commonly practiced to purify municipal and industrial wastewater, mainly because of the high treatment efficiency and low operating cost. Activated sludge processes are based on the ability of microorganisms to utilize organic material as a source of energy and/or as a source of carbon and other minerals for growth (Carneiro, Umbuzeiro, Oliveira, & Zanoni, 2010), thereby playing important roles in the biodegradation of organic materials, transformation of toxic compounds into harmless products, and removal of nutrients such as ammonia, nitrate, sulfate, and phosphate (Gentile, Jessup, Nyman, & Criddle, 2007;Wang et al., 2011). The stable operation of biological wastewater treatment plants (WWTPs) relies upon the occurrence, relative abundance and activity of several microbial populations in the activated sludge performing these processes (Gentile et al., 2007;Wagner & Loy, 2002;Yang et al., 2014). As variations in microbial community composition are often associated with changes in the functional capabilities of the communities, microbial community and functional stability are generally recognized as important factors to efficiently treat wastewater (Wang, Xia, Wen, Yang, & Zhou, 2014).
Microbial communities of activated sludge in WWTPs have been intensively studied over the last decade, especially for WWTPs dealing with municipal wastewater (e.g., López-Vázquez, Hooijmans, Brdjanovic, Gijzen, & van Loosdrecht, 2008;Miura et al., 2007;Sanapareddy et al., 2009;Wang et al., 2010Wang et al., , 2011. In general, trends are observed with members of the phylum Proteobacteria, frequently being the most abundant in municipal WWTPs (accounting for 30%-60% of the total number of sequences), followed by Actinobacteria and Bacteroidetes Ju, Guo, Ye, Xia, & Zhang, 2014;Saunders, Albertsen, Vollertsen, & Nielsen, 2016;Wang, Hu, Xia, Wen, & Ding, 2012;Wei et al., 2015;Ye & Zhang, 2013;Zhang, Shao, & Ye, 2012;Zhao et al., 2014). Moreover, in a recent Illumina MiSeq-based study of 13 municipal WWTPs across Denmark, it was shown that the plants contained a core community of 63 abundant genus-level operational taxonomic units (OTUs), indicating that microbial communities of activated sludge in municipal WWTPs are quite similar across multiple plants (Saunders et al., 2016). However, it is reasonable to assume that communities involved in the biological treatment of more hazardous wastewaters, such as those originating from the textile industry, harbor different microbial populations that are specifically adapted to the environmental stresses encountered in these systems. Textile industry effluents typically contain high concentrations of dyes, dyeing additives, and diverse chemicals, some of which are nonbiodegradable, toxic, mutagenic, or carcinogenic, which pose a major threat to health and environment. Additionally, textile wastewater generally has a low biological oxygen demand/chemical oxygen demand (BOD/COD) ratio (around 20%), a wide range of pH (4)(5)(6)(7)(8)(9)(10)(11)(12), and may contain several inhibitor compounds (hampering effective biological wastewater treatment), active substances, adsorbable organic halogens (e.g., chlorine compounds) (AOX) and high salt concentrations, altogether making textile wastewater difficult to treat (Sandhya & Swaminathan, 2006;Selcuk, 2005;Verma, Dash, & Bhunia, 2012;Wu, Wang, Kong, Liu, & Xia, 2007). Therefore, it can be hypothesized that microbial communities in textile WWTPs are different from those observed in municipal WWTPs. However, so far only little is known about the microbial community composition and their functioning in activated sludge from textile wastewater treatment systems (but see Yang et al., 2014).
The aim of this study was to assess the microbial community composition in activated sludge from textile WWTPs in comparison with municipal WWTPs in Flanders (Belgium), and to explain observed differences by environmental factors. The first objective of this study was to assess differences in the microbial (both bacterial and archaeal) communities using 454 amplicon pyrosequencing and real-time quantitative PCR (qPCR). Secondly, we aimed at determining which environmental factors drive species richness, diversity, and community composition in activated sludge from different WWTPs dealing with different wastewaters.

| Study samples
Activated sludge samples (0.5 L) were collected in triplicate from aerobic tanks of five textile WWTPs and five municipal WWTPs.
Additionally, samples were taken from one plant dealing with both textile and municipal wastewater. All WWTPs were located in Flanders (Belgium) and were characterized by a stable operating system, discharging wastewater effluents within legal standards. Sampling was performed in two different seasons, including winter (February 2015) and summer (July 2015). Following sampling, samples were immediately centrifuged at 3500 g using a portable Hettich EBA 20 centrifuge (Hettich Lab Technology, Tuttlingen, Germany) to precipitate the sludge. Approximately 1 g of precipitated sludge was resuspended in 20 ml RNAlater (Life Technologies, Carlsbad, CA, USA) to preserve the nucleic acids present in the samples. At the same time, influent samples (1 L) were collected from each wastewater for chemical analysis.
Samples were transported in an ice-cooled container to the laboratory and stored overnight at 4°C prior to further analysis.

| DNA extraction, PCR amplification, and 454 amplicon pyrosequencing
Following centrifugation of the samples (10 ml), genomic DNA was extracted from 0.15 g precipitated material using the Power Soil DNA isolation kit (MoBio Laboratories Inc., Solana Beach, CA, USA) according to the manufacturer's instructions. Subsequently, DNA extracts from the three samples taken per studied WWTP were pooled and stored at −80°C until further processing.
Amplicon libraries were created using two PCR primer sets, targeting part of the bacterial and archaeal 16S ribosomal RNA (rRNA) genes, including the primer pairs S-D-Bact-0341-b-S-17 / S-D-Bact-0785-a-A-21 (covering the V3-V4 region; amplicon size of approximately 464 bp; Klindworth et al. 2013) and S-D-Arch-0519-a-S-15 / S-D-Arch-1041-a-A-18 (covering the V4-V6 region; amplicon size of approximately 540 bp; Klindworth et al. 2013), respectively. Previous research has shown that these primer pairs were successfully implemented for water research and outperformed other primer pairs for classical and next-generation sequencing-based diversity studies (Klindworth et al. 2013;Connelly, Baer, Cooper, Bronk, & Wawrik, 2014;Richert et al., 2015;Savio et al., 2015;Shan et al., 2015). "Fusion" primers, required for the 454-pyrosequencing process, were designed according to the guidelines for 454 GS-FLX+ Titanium Lib-L sequencing and contained the Roche 454-pyrosequencing adapters and a samplespecific multiplex identifier (MID) sequence in between the adapter and the forward primer for sample-specific sequence tracking (Table S1). Sequences obtained from the 454-pyrosequencing run were assigned to the appropriate sample based on their barcodes and primer sequences, allowing zero discrepancies, and were subsequently trimmed using a custom Python script implemented within the USEARCH v. 7 analysis pipeline (Edgar, 2013). Sequences obtained from both PCR replicates per sample were combined and further trimmed based on a minimum Phred score of 30 (base call accuracy of 99.9%) averaged over a 50 bp moving window. Sequence length was determined for each primer pair assessed, with the average being 250 bp. Sequences with ambiguous base calls or homopolymers with a length of more than eight sequences were rejected, as were chimeric sequences detected by UCHIME 4.2 chimera detection (de novo algorithm) (Edgar, Haas, Clemente, Quince, & Knight, 2011). Remaining sequences were aligned and grouped into species-level OTUs based on a 3% sequence dissimilarity cut-off using the UPARSE algorithm implemented in USEARCH (Edgar, 2013). To minimize the risk of retaining sequences from sequencing errors, "global" singletons (i.e., OTUs representing only a single unique sequence in the entire dataset) were removed after UPARSE clustering (Brown et al., 2015;Waud, Busschaert, Ruyters, Jacquemyn, & Lievens, 2014). Due to uneven sequencing depth and correlation between number of sequence reads and number of OTUs per sample (data not shown), the number of sequences was rarefied to 1,000 sequences per sample for both bacteria and archaea. OTU representative sequences were assigned taxonomic identities using the "classify.seqs" command in Mothur (v. 1.36.1) (Schloss et al., 2009)

| Real-time quantitative PCR
To confirm and further assess the occurrence and distribution of two bacterial OTUs that could be specifically attributed to textile or municipal WWTPs (based on the 454 data), a qPCR analysis was performed.
To this end, specific primers were designed for OTU217 (Planctomyces sp.) and OTU23 (Rhodoferax sp.), respectively (Table S2). Specificity of the primers was evaluated using the BLAST algorithm against GenBank, and further evaluated against the 454 datasets obtained in this study. Furthermore, qPCR analyses were performed for two bacterial genes involved in nitrogen removal, including the amoA and nirK gene, encoding a functional nitrifying (ammonium monooxygenase alpha subunit) and denitrifying enzyme (copper-containing nitrite reductase) (for primers see Table S2) (Geets et al., 2007). Analyses were performed on an ABI StepOnePlus real-time PCR system. Each reaction contained 1.0 μl 10× diluted genomic DNA, 0.5 μl of each primer (20 μM stock), 10.0 μl 2× iTaq universal SYBR Green supermix, and 8.0 μl nuclease-free water. The qPCR run consisted of the same thermal profile as described above except for the annealing temperature which was 64°C (for OTU23 and OTU217) or 59° (for the amoA and nirK genes). At the end of each qPCR run, a melting curve analysis was performed as described above. Quantification was performed using a standard curve based on known concentrations of DNA standard dilutions from 10 7 copies μl −1 down to 10 2 copies μl −1 . All qPCR analyses were conducted in duplicate.

| Chemical analyses
In order to determine the environmental conditions to which the different microbial communities have been exposed, a number of chemical analyses were performed on the influent samples. Analyses were performed using Nanocolor test tubes and a Nanocolor 500D photometer (Macherey-Nagel, Düren, Germany) according to manufacturer's instructions, and included measurement of ammonium (NH 4

| Data analysis
Rarefaction curves were generated for each sample using the Vegan

| Archaeal and bacterial community composition
Bacterial and archaeal communities were profiled for a total of 22  (Ju et al., 2014;Zhang et al., 2014). Further, as also noticed in previous studies on microbial communities in activated sludge Saunders et al., 2016;Wang et al., 2012;Wei et al., 2015;Yang et al., 2014;Zhang et al., 2012;Zhao et al., 2014) (Nielsen, Kragelund, Seviour, & Nielsen, 2009;Rossetti, Tomei, Nielsen, & Tandoi, 2005). These filamentous bacteria were present (albeit at low densities) in both municipal (eight OTUs representing these four genera) and textile (four OTUs belonging to the genera Anaerolinea, Microthrix, and Thiothrix) WWTPs, but were generally more abundantly present in municipal WWTP samples (data not shown). Furthermore, nitrifying and denitrifying bacteria as well as phosphate-accumulating bacteria showed a higher relative read abundance in municipal WWTPs (Fig. 2). This was confirmed by a qPCR analysis targeting the bacterial amoA and nirK genes, the first being involved in nitrification, the second in denitrification: samples from municipal WWTPs were significantly higher in amoA abundance during winter (p = 6.75E-03) and summer (p = 8.34E-05) as opposed to textile WWTP samples. Also, nirK abundance was higher in municipal samples, albeit not significantly (p = .518). These findings suggest that removal of ammonium, nitrate, and phosphate is likely more efficient in municipal WWTPs. Indeed, effluent measurements of the different municipal wastewaters showed an enhanced removal of phosphate and ammonium for the investigated municipal WWTPs in comparison with the investigated textile WWTPs (data not shown).
In contrast, sulfate-reducing bacteria were almost solely found in textile WWTPs (Fig. 2). Notably, a great number of OTUs (40)

July February
were predominantly found in the textile-activated sludge samples.
Additionally, some Leucobacter species have already been used in microbial consortia to degrade disperse and reactive dyes (Franciscon et al., 2010(Franciscon et al., , 2015. NMDS ordination of the community composition, inferred from the archaeal (p = .033; R² = 0.280) and bacterial (p = .001; R² = 0.663) OTU relative abundance, revealed that there was a significant difference (Goodness-of-Fit) between activated sludge samples from textile and municipal WWTPs (Fig. 3). Furthermore, samples from municipal WWTPs were much more similar than those from textile WWTPs (Fig. 3). Interestingly, NMDS ordination plotted the samples from the plant dealing with both municipal and textile wastewater (TR1) in between samples from municipal WWTPs on one hand and samples from textile WWTPs on the other hand (Fig. 3). Significant differences were found in community composition of archaeal communities sampled in February and July (p = .034; R² = 0.179), but not for bacteria (p = .694; R² = 0.014).
Strikingly, only one archaeal and one bacterial OTU was shared by all samples investigated. These included an OTU corresponding to Methanosaeta sp. (Euryarchaeota) and an OTU corresponding to an unidentified member of Proteobacteria, respectively. The first one covered approximately 23% of all archaeal sequences, while the second covered about 3% of all bacterial sequences. A core microbial community, consisting of five archaeal and 30 bacterial OTUs could be identified for municipal WWTPs that made up 33.0% and 19.3% of the total archaeal and bacterial sequences, respectively. Two of these bacterial genera, Dechloromonas and candidatus Epiflobacter, were also identified as one of the 23 to the genus-level identified core genera in a recent study about activated sludge microbial communities in 13 Danish WWTPs (in total, 63 abundant genus-level F I G U R E 2 Read abundance of bacterial genera performing essential functions in activated sludge processes (nonexhaustive list), including nitrification (a), denitrification (b), sulfate reduction (c), and phosphate accumulation (d), in samples from textile and municipal wastewater treatment plants (WWTPs) (data combined for February and July; 22 samples). The number of Operational Taxonomic Units (OTUs) belonging to the genus is reported between brackets

(d)
OTUs were identified based on a 6% 16S rRNA gene sequence dissimilarity cut-off) (Saunders et al., 2016), suggesting a ubiquitous occurrence of these genera in municipal WWTPs. In contrast, only one archaeal and one bacterial OTU was shared by all textile WWTPs, suggesting that microbial communities in textile WWTPs are driven by diverse factors. ISA, allowing to identify one more given species/ OTUs to serve as an indicator of a particular ecosystem, revealed the presence of two and six archaeal OTUs that could be attributed to textile and municipal WWTPs, respectively. For the bacteria, ISA revealed 10 and 34 indicator OTUs, respectively (Table S5). In order to confirm and generalize these results, all 22 samples investigated as well as six additional sludge samples from three textile WWTPs (sampled in February and July) and 10 additional samples from five municipal WWTPs (sampled in February and July) were subjected to qPCR analysis targeting two randomly selected indicator bacteria. These included OTU23, representing a member of the genus Rhodoferax (Proteobacteria), which was found as an indicator for municipal activated sludge, and OTU217, member of the genus Planctomyces (Planctomycetes), which was found as an indicator for activated sludge from textile WWTPs. OTU23 was found in all municipal WWTP samples analyzed, while it was absent in the textile WWTP samples. Additionally, OTU23 was found in activated sludge from the plant purifying both municipal and textile wastewater. OTU217 was found at five textile WWTPs, both in February and July (10 positive samples on a total of 16), while it was not detected in any sample from the municipal WWTPs (Table S6).

| Environmental factors explaining differences in microbial communities
In order to determine environmental factors potentially explaining the differences in microbial communities in activated sludge from textile WWTPs and municipal WWTPs, several environmental variables were measured on the influent wastewater (Table 2). Samples from textile WWTPs had significantly higher salt levels and were higher in temperature (Table 2 and 3). Further, textile wastewater was found to contain a significantly higher organic load, as shown by the high COD and BOD values as opposed to municipal wastewater, whereas the DO level was significantly lower (Table 2 and 3), supporting previous findings (Verma et al., 2012). Also AOX values were slightly, but not significantly higher for textile wastewater. Little or no differences were found for NH 4 + , NO 2 -, NO 3 -, TP, TN, and pH (Table 2 and 3). Wastewater from the plant treating both municipal and textile wastewater was characterized by values situated between textile and municipal wastewater (Table 2). Notably, for some plants, differences were observed for particular influent characteristics (e.g., NH 4 + , TP, and COD) between the two sampling periods ( Table 2), suggesting that these companies treat wastewaters with a variable composition.
Archaeal community composition significantly (p < .05) varied with temperature, COD, BOD, conductivity, pH, and TN (Table 4), whereas the bacterial communities varied with temperature, conductivity, pH, DO, COD, BOD, and NH 4 + (Table 4). When fitting the environmental variables on the NMDS ordination plot of the microbial communities (Fig. 3), temperature, conductivity, COD, BOD, and DO differentiated both archaeal and bacterial communities from textile and municipal WWTPs, with an increasing gradient toward samples from textile WWTPs except for DO, which shows an increasing gradient toward municipal activated sludge samples. For archaea, also TN and pH significantly divided samples from both groups. Fitting F I G U R E 3 Nonmetric multidimensional scaling (NMDS) ordination (with environmental fit) of the archaeal (a; stress value = 0.153) and bacterial communities (b; stress value = 0.082) in activated sludge samples of textile (green squares) and municipal wastewater purification systems (red circles), as well as from one plant treating both municipal and textile wastewater ("combined"; blue triangles). Arrows represent environmental variables with significant correlations (Table 4). Length and orientation of the arrow is proportional to the direction and the amount of correlation between the ordination and the environmental variables   (Griffiths, Ritz, Ebblewhite, & Dobson, 1998;Liu, Yang, Gong, & Su, 2008;Pietikäinen, Pettersson, & Bååth, 2005;Rietz & Haynes, 2003;Siggins, Enright, & O'Flaherty, 2011;Wakelin et al., 2012;Wang et al., 2012). In addition to these general parameters, other variables specifically linked to the textile dyeing industry (e.g., the dyes used, chemical additives etc.) are also likely to be involved in the T A B L E 4 Results of the permutation test of the nonmetric multidimensional scaling coordinates (NMDS 1 and NMDS 2) testing for significant relationships between activated sludge samples from textile and municipal wastewater treatment plants (WWTPs) and influent chemical variables mechanisms behind the assembly of these microbial communities.
Further research is needed to unravel their importance in driving the assembly of textile WWTP communities. Future research building on our results could also aim at the identification of key players in the community that may be exploited for enhanced purification of textile wastewaters. In this regard, the phyla Planctomycetes (Bacteria) and Thaumarchaeota (Archaea), both abundantly present in activated sludge from textile WWTPs and possibly performing important functions in the purification of textile wastewaters, may provide promising candidates.