Quantitative response of nitrifying and denitrifying communities to environmental variables in a full-scale membrane bioreactor

Bioresource Technology 169 (2014) 126–133

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Quantitative response of nitrifying and denitrifying communities to environmental variables in a full-scale membrane bioreactor C. Gómez-Silván a,⇑, R. Vílchez-Vargas b,1, J. Arévalo c, M.A. Gómez c, J. González-López a, D.H. Pieper b, B. Rodelas a a b c

Departamento de Microbiología, Facultad de Farmacia, University of Granada, Granada, Spain Microbial Interactions and Processes Research Group, Helmholtz Center for Infection Research, Braunschweig, Germany Departamento de Ingeniería Civil, University of Granada, Granada, Spain

h i g h l i g h t s  A full-scale MBR treating urban wastewater was operated for nine months.  Gene markers specific of AOB, NOB and denitrifiers were quantified by qPCR.  Both abundance (DNA) and transcription level (cDNA) of gene markers were analyzed.  MDS was used to link population abundances to changes in operation parameters.  N-cycle bacterial groups showed varying trends of response to operation parameters.

a r t i c l e

i n f o

Article history: Received 2 May 2014 Received in revised form 24 June 2014 Accepted 25 June 2014 Available online 2 July 2014 Keywords: Urban wastewater treatment Reverse-transcription qPCR AOB NOB Multivariate analysis

a b s t r a c t The abundance and transcription levels of specific gene markers of total bacteria, ammonia-oxidizing Betaproteobacteria, nitrite-oxidizing bacteria (Nitrospira-like) and denitrifiers (N2O-reducers) were analyzed using quantitative PCR (qPCR) and reverse-transcription qPCR during 9 months in a full-scale membrane bioreactor treating urban wastewater. A stable community of N-removal key players was developed; however, the abundance of active populations experienced sharper shifts, demonstrating their fast adaptation to changing conditions. Despite constituting a small percentage of the total bacterial community, the larger abundances of active populations of nitrifiers explained the high N-removal accomplished by the MBR. Multivariate analyses revealed that temperature, accumulation of volatile suspended solids in the sludge, BOD5, NH+4 concentration and C/N ratio of the wastewater contributed significantly (23–38%) to explain changes in the abundance of nitrifiers and denitrifiers. However, each targeted group showed different responses to shifts in these parameters, evidencing the complexity of the balance among them for successful biological N-removal. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction One of the main objectives of biological wastewater treatment (WWT) is the removal of N compounds from water bodies, due to their otherwise harmful effects in water ecosystems, such as eutrophication or ground-water contamination. Pre-denitrification is among the most efficient strategies reported (Le-Clech, 2010).

⇑ Corresponding author. Address: Departamento de Microbiología, Facultad de Farmacia, Campus de Cartuja s/n, 18071 Granada, Spain. Tel.: +34 958 249966; fax: +34 958 246235. E-mail address: [email protected] (C. Gómez-Silván). 1 Present address: Laboratory of Microbial Ecology and Technology, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium. http://dx.doi.org/10.1016/j.biortech.2014.06.089 0960-8524/Ó 2014 Elsevier Ltd. All rights reserved.

Under this mode of operation, the wastewater first enters an anoxic phase, where denitrifiers use NO3 to oxidize the organic matter in a four-step reductive pathway generating N2 as end product. The water then enters an aerated phase where the nitrifiers, ammonia oxidizing bacteria (AOB), and nitrite oxidizing bacteria (NOB), sequentially transform NH+4 into NO3 . Finally, NO3 is supplied into the anoxic phase by partial recirculation of the nitrified effluent. Among the abovementioned bacterial populations, nitrifiers have slow growth rates, complicating the correct development of a community capable of highly efficient N-removal (Le-Clech, 2010). Membrane bioreactors (MBRs) have been established as an alternative technology to the conventional activated sludge (CAS) processes, showing several advantages that make them

C. Gómez-Silván et al. / Bioresource Technology 169 (2014) 126–133

attractive for wastewater reclamation and reuse. MBRs apply membrane filtration for the separation of particulate material, avoiding the need for a secondary clarifier (Judd, 2011). Compared to CAS, MBRs are characterized by a high solids retention time (SRT), and an independence between SRT and hydraulic retention time (HRT). These features strongly influence the biology of the system and improve the development of slow-growing microorganisms (Le-Clech, 2010). The oxidation of NH+4 to NO2 is regarded as the rate-limiting step for the removal of N compounds (Limpiyakorn et al., 2005). In WWT plants, the prevalent ammonia oxidizers are evolutionally related to a monophyletic group consisting of two genera of the Betaproteobacteria, Nitrosomonas and Nitrosospira (Purkhold et al., 2003). The actual contribution to nitrification of ammoniaoxidizing archaea (AOA) in engineered systems is still unclear (Calderón et al., 2013). Among the microorganisms responsible for the second step of nitrification, Nitrobacter-like bacteria were classically acknowledged as the most relevant NOB group in WWT plants, but cultivation-independent molecular techniques revealed Nitrospira spp. as the most diverse, abundant, and ubiquitous NOB group (Lücker et al., 2010; Winkler et al., 2012). Nitrospira have slower growth rates but are more competitive than Nitrobacter when long SRT are set (Yu et al., 2010b), and under lower NO2 and O2 concentrations (Winkler et al., 2012). Despite their decisive roles, AOB and NOB are commonly detected in only relatively low abundance in WWT systems (Geets et al., 2007; Kim et al., 2011; Xia et al., 2012; Zhang et al., 2010). It is well known that the variations in the operational parameters and environmental variables in WWT systems affect the ability of bacteria to biologically transform N compounds (Kim et al., 2011). Adequately linking the changes in quantitative composition of bacteria involved in the N-cycle to such variables may contribute to further improvements in WWT design and operation. Nevertheless, few studies are currently available taking advantage of multivariate analysis methods for this purpose (Huang et al., 2010; Kim et al., 2006). Furthermore, the number of studies quantifying simultaneously AOB, NOB and denitrifying bacteria in full-scale WWT plants is limited (Geets et al., 2007; Kim et al., 2011), and do not include the evaluation of the levels of gene expression. Because of these concerns, in the present work, quantitative PCR (qPCR) and reverse-transcription qPCR (RT-qPCR) were applied to evaluate the abundance (qPCR) and levels of transcription (RTqPCR) of gene markers specific for Bacteria, AOB, NOB, and denitrifiers (N2O-reducers). The targeted bacterial groups were quantified during a time frame of nine months, spanning three seasons of the year (spring, summer, and autumn), in both the aerated and anoxic bioreactors (BRs) of a full-scale MBR operated in pre-denitrification mode, and fed urban wastewater. Throughout the experimental period, multivariate analyses of the data (non-metric multidimensional scaling, MDS, and BIO-ENV) were used to evaluate the effect of fluctuations in the characteristics of the influent wastewater, temperature, and biomass concentration in the BRs on the abundance of the targeted bacterial populations. 2. Methods 2.1. Description of the full-scale MBR experimental plant, operational parameters and performance The experimental MBR plant, operational parameters, and performance were previously described (Gómez-Silván et al., 2013). Briefly, the system consisted of an aerated BR (19.4 m3), an anoxic BR (6.8 m3), and a filtration tank (2 m3) equipped with three ultrafiltration hollow-fiber membrane modules (0.034 lm nominal pore size) made of polyvinylidenefluoride (PVDF) (GE Water &

127

Process Technologies, Fairfield, USA). The MBR was fed urban wastewater taken from the pretreatment of the nearby ‘‘Estación Depuradora Sur’’-WWT plant (Granada, Spain), managed by the company EMASAGRA S.A. It was operated in the pre-denitrification mode with fixed hydraulic retention time (HRT = 38 h) and solids retention time (SRT = 20 days). The inflow and outflow rates were 1 m3/h, and the recirculation rate between the BRs was seven times the influent flow-rate (700%). Air was supplied to the aerated BR with a fine bubble membrane diffuser, automatically keeping the dissolved oxygen (DO) concentration in the range of 0.5–1.6 mg/ L. The filtration tank was also aerated to control membrane fouling and clogging. The temperature inside the activated sludge of the MBR system was measured automatically every second, and activated sludge daily medium internal temperature (IT) was calculated by the software Active Factory v.9.2 (Wonderware, Spain). Concentrations of fixed and volatile suspended solids (iFSS and iVSS), total nitrogen (tN), NH+4, total chemical oxygen demand (COD), and total biological oxygen demand at five days (BOD5) were analyzed in influent and effluent (permeate) water. Fixed and volatile suspended solids (FSS and VSS) concentrations in the activated sludge of both BRs were also measured. The quality of the effluent was in agreement with the EU regulation for discharges to the environment (Directive 91/271/EEC, European Council, 1991) (Table S1). 2.2. Activated sludge sampling for qPCR assays The biological tests in the activated sludge from the MBR plant were initiated when the system reached steady state conditions. The study spanned from April to December, including three different seasons of the year: spring (April and May), summer (June, July, August, and September) and autumn (October, November, and December), which encompassed the standard annual range of temperatures of the city of Granada (Gómez-Silván et al., 2013). Samples (50 ml) of activated sludge from both BRs (aerobic, A, and anoxic, X) were collected in sterile plastic containers once a month. Two replicates were prepared for each DNA and RNA extraction, centrifuging 4 ml of the activated sludge samples (1 min, 14g) in a MiniSpin Plus table centrifuge (Eppendorf, Hamburg, Germany). After discarding the supernatant, the pellets intended for DNA extraction were carried at 4 °C from the experimental plant to the laboratory and stored at 20 °C. The pellets intended for RNA extraction were resuspended in 1 ml of RNAprotectÒ Bacteria Reagent (Qiagen, Hamburg, Germany) and carried unrefrigerated from the experimental plant to the laboratory, then precipitated again by centrifuging (1 min, 14g) in a MiniSpin Plus table centrifuge (Eppendorf, Hamburg, Germany), discarding the supernatant and storing the pellets at 20 °C. 2.3. Nucleic acids extraction and purification Genomic DNA and RNA were extracted from the preserved samples using the FastDNAÒ SPIN kit for Soil and the FastRNAÒ PRO BLUE kit, respectively, following the manufacturer’s indications. The FastPrepÒ24-Instrument (MP-Bio, Santa Ana, CA, USA) was used for all extractions. Traces of DNA in the RNA samples were digested by the rigorous procedure of TURBO DNA-free™ kit (AmbionÒ, Life Technologies Corporation, Carlsbad, CA, USA), following the manufacturer´s indications, and were subsequently purified using the RNA Cleanup protocol from the RNeasy Mini Kit (Qiagen, Hamburg, Germany). The concentrations and quality of DNA and RNA extractions were measured with a BioPhotometer Plus (Eppendorf, Hamburg, Germany).

128

C. Gómez-Silván et al. / Bioresource Technology 169 (2014) 126–133

2.4. Reverse transcription of RNA to cDNA Reverse transcription reactions were performed with SuperScript™ III Reverse Transcriptase (Invitrogen, Life Technologies Corporation, Carlsbad, CA, USA), following the manufacturer’s directions, and using 150–200 ng RNA in a final volume of 20 ll. The specific primers (Table S2) were supplied by Sigma Aldrich (St. Louis, MO, USA), and the dNTPs by MBL (Córdoba, Spain). The complementary DNA (cDNA) was purified using QIAquickÒ PCR Purification Kit (Qiagen, Hamburg, Germany). 2.5. qPCR assays The evaluation of the quantitative changes of N-cycle bacteria in the MBR was achieved using qPCR/RT-qPCR. The following genes were targeted: total bacterial 16S rDNA, AOB 16S rDNA, nxrB of Nitrospira, and nosZ of denitrifying bacteria. Both the number of copies of the targeted genes and their transcripts were quantified in each sample, using either extracted DNA or cDNA as a template for qPCR and previously described primer sets (Table S2). All quantitative amplifications were performed in triplicate. Quantification of Nitrobacter-like nxrA genes was attempted using the primer set nxrAF1370/nxrAR2843 (Table S2), but the expected amplification products were not detected in any of the MBR samples. First, the qPCR kit MaximaÒ SYBR Green qPCR Master Mix (2) (Fermentas, Thermo Fisher Scientific Inc., Waltham, MA, USA) was tested, using representative samples and the sets of primers to amplify the nrxB and nosZ genes (Table S2). PCR reactions were done following the qPCR kit manufacturer’s indications and PCR conditions described in Table S3. The PCR products were examined in 2% (w/v) ethidium bromide-stained agarose gels (Fig. S1). Due to the generation of unspecific PCR products, the qPCR kit was not used for further sample analysis, and manual protocols were set up. A qPCR procedure was optimized using TrueStart Hot Start DNA polymerase (Fermentas, Thermo Fisher Scientific Inc., Waltham, MA, USA) and SYBR Green I (Sigma Aldrich, St. Louis, MO, USA), in a total volume of 25 ll, using a LightCyclerÒ 480 System (Roche Applied Science, Penzberg, Germany). Primers and conditions for each qPCR reaction are summarized in Tables S2 and S3. The primers and DMSO were supplied by Sigma Aldrich (St. Louis, MO, USA), the dNTPs by MBL (Córdoba, Spain), and BSA by New England Biolabs (Ipswich, MA, USA). Absolute quantifications for DNA samples were achieved by constructing standard curves with serial dilutions of linearized plasmids harboring PCR-amplified inserts of the targeted genes. The pGEMÒ-T Vector System II (Promega, Madison, WI, USA) was used for cloning and transformation into Escherichia coli JM109. For the absolute quantification of RNA samples, non-linearized plasmids were used as a template for in vitro transcription of the

target genes into RNA, using T7-RNA polymerase (Invitrogen, Life Technologies Corporation, Carlsbad, CA, USA), following the indications of the manufacturer. DNA traces were digested as already described, and RNA was retrotranscribed to cDNA, which was serially diluted for calibration curve construction. The concentration and quality of nucleic acids was measured with a BioPhotometer Plus (Eppendorf, Hamburg, Germany). The number of copies of the targeted genes per L of activated sludge was calculated. The standard curves for qPCR, generated with tenfold dilutions (10 1–10 7) of plasmids harboring the target sequences, had a correlation coefficient r2 > 0.99 in all the assays. The ranges of the Ct values of the standard curves were always wider than the range of Ct values of the activated sludge samples. In all the cases, the number of copies detected in the samples was included in the range of the detection limits imposed by the standard curves (Table 1). Also, when amplification was detected in the non-template control or negative control, it yielded a Ct value at least fivefold higher than the Ct value of the highest dilution used to construct the standard curves. All the qPCR amplifications resulted in a single band of the expected size for each set of primers, verified by agarose gels (Fig. S2) and by melting-curve analysis (data not shown). The specificity of the selected primers was confirmed by the sequence analysis of a selection of cloned amplicons used for the construction of the standard curves. 2.6. Statistics IBM SPSS Statics v. 19 (SPSS INC., USA) was used for a preliminary analysis of data distributions and subsequent statistical tests to find differences among the groups of samples. As most of the data sets did not fit the normal distribution, a non-parametric analysis for related samples (Wilcoxon signed-rank test) was conducted for each biological assay and system variable. A 90% of significance level (p < 0.1) was selected. The seasonal distribution patterns of the variables in the MBR were studied by ordinations using non-parametric multidimensional scaling (MDS), aided by the Primer software (PRIMER-E, vs. 6.0, Plymouth, UK). The data sets were transformed to log (X + 1) and normalized. Sample-resemblance matrices were generated using Euclidian distance. Values of stress level on the MDS plots 60.05 indicate that they give an accurate representation of the data distribution (Clarke et al., 2008). The variables included were: influent wastewater concentrations of iVSS and iFSS, NH+4, C/N ratio, BOD5; and activated sludge IT and concentrations of VSS and FSS. The pH of the influent wastewater was highly stable and close to neutral throughout the entire experimental period (average ± SD = 7.4 ± 0.2), and as such, it was not included as a variable. Spearman rank correlations of each variable were calculated and are represented in the plots as vectors that illustrate their directional influence and role in the ordination.

Table 1 Detection limits of qPCRs assays. Bacterial group

Targeted gene

Nucleic acid

Bacteria

V3-16S rRNA

AOBa

16S rRNA

b

a b

NOB

nxrB

Denitrifiers

nosZ

Betaproteobacterial AOB. Nitrospira-like NOB.

DNA RNA DNA RNA DNA RNA DNA RNA

Range of detection (no. copies/ll DNA or RNA sample)

Ct values range Standard 4.0–24.5 1.5–22.0 10.0–27.5 8.5–30.0 6.0–27.5 13.5–29.5 10.0–27.0 13.0–31.5

Samples 11.0–16.0 5.0–17.0 20.0–24.0 13.0–28.0 18.5–27.0 17.0–27.5 20.5–25.0 17.5–30.0

Standard 2

7.3  10 1.8  106 1.5  103 8.5  102 8.4  101 3.0  105 4.3  102 4.0  104

Samples 8

7.3  10 1.8  1012 1.5  108 8.5  108 8.4  107 3.0  1010 4.3  107 4.0  1010

2.0  107 3.5  109 1.4  105 4.6  105 3.7  102 1.6  108 1.4  105 3.0  107

8.0  107 1.6  1012 5.6  106 7.7  108 4.5  104 2.5  1010 1.7  106 1.7  1010

C. Gómez-Silván et al. / Bioresource Technology 169 (2014) 126–133

Two different levels of analyses were performed to study the trends of biological data (numbers of gene or transcript copies of each targeted bacterial group) related to the variables’ changes: – First, Spearman rank correlations were calculated between the biological data and the variables’ ordinations. Those correlations are represented in the plots as vectors. Each vector illustrates the biological data’s directional increase through the ordination. – Second, BIO-ENV analysis was performed to find potential correlations between the biological data and environmental variables. BIO-ENV calculates the correlation coefficients between the variables’ resemblance matrix and the biological matrix, generated by Bray Curtis similarity, using the Spearman rank correlation. Global permutation tests (499 permutations) were conducted to determine the significance of BIO-ENV analyses (Clarke et al., 2008). 3. Results and discussion 3.1. Seasonal changes of the environmental variables In order to characterize the MBR in each studied season, the environmental variables included in the current study were statistically analyzed (Gómez-Silván et al., 2013), and the average ± SD values are summarized in Table S1 and Fig. 1. Briefly, the highest contamination levels in the influent wastewater (BOD5, C/N ratio and concentrations of NH+4 and suspended solids) were measured in spring, the highest IT and lowest influent contamination levels in summer, and the highest accumulation of suspended solids in the BRs and the lowest food/microbial (F/M) ratio (kg BOD5 per kg VSS in the aerated BR per day) were observed in autumn. 3.2. Absolute and relative quantifications of target molecules by qPCR/ RT-qPCR The average DNA- and RNA-based quantifications of all the targeted bacterial groups are shown in Fig. 2A. According to the Wilcoxon signed-rank test, the differences were often significant between both quantifications in samples collected from each BR in each season. The average numbers of copies detected of all the targeted genes were inside the ranges described in previous works performed in CAS-based and MBR WWT plants treating urban and

Fig. 1. Box-and-Whisker plots of the concentration of ammonia (NH+4) in the influent wastewater of the full-scale membrane bioreactor (MBR) during the spring, summer, and autumn of 2009. In plots, upper and lower boundaries of the box denote the 75th and 25th percentiles, upper and lower boundaries of bars are the 90th and 10th percentiles. Average values ± standard deviations are shown below the boxes. Data followed by the same lower-case letter do not significantly differ according to the Wilcoxon signed-rank test (p < 0.10).

129

industrial wastewater (Fig. 2C). In these previous studies, the quantified abundances were in the orders of 1012–1014 16S rDNA copies/L activated sludge for Bacteria, 1010–1012 16S rDNA copies/L activated sludge for AOB, 106–1010 copies of nxrB/L activated sludge for Nitrospira-like NOB, and 105–1012 copies of nosZ/L activated sludge for denitrifiers (Geets et al., 2007; Kim et al., 2011; Xia et al., 2012; Zhang et al., 2010). In order to calculate which fraction of the total bacterial community was represented by each N-cycle bacterial group studied, the number of copies of the targeted genes (DNA samples) was expressed as the percentage of the number of copies of 16S rDNA of Bacteria (Fig. 3). For the correct interpretation of these results, it should be taken into consideration that, in accordance with the available data, the average number of 16S rDNA copies per bacterial cell is 4.2, and the AOB (Nitrosomonas and Nitrosospira) usually harbor only one copy of the rRNA operon (http://rrndb.umms.med.umich.edu/), denitrifiers most often carry a single copy of the nosZ gene (Palmer et al., 2009), and two copies of the nxrB gene are reported in the Candidatus Nitrospira defluvii genome (Lücker et al., 2010). The Nitrospira-like NOB were the less abundant group, accounting on average for 4  10 2% of the total bacterial 16S rDNA copies, while the AOB and denitrifiers accounted for 1.8% and 1.5%, respectively. The relative abundance data of N-cycle bacterial gene copies detected in the present study were also in the ranges previously reported (0.0004–3% for AOB, 2  10 7 13% for Nitrospira-like NOB and 0.1–4.8% for denitrifiers) (Geets et al., 2007; Kim et al., 2011; Xia et al., 2012; Zhang et al., 2010). Overall, the average number of copies detected of all the targeted genes remained rather stable throughout the sampling period (differences lower than one order of magnitude). However, the amounts of transcripts indicated higher longitudinal variations in the abundances of active populations. Comparing seasons, there were significant differences in the abundances of bacterial 16S rRNA and transcripts of nosZ and nxrB, which were lower during spring (Fig. 2A). The RNA/DNA ratios increased gradually throughout the experiment for all the targeted genes, indicating the progressive increase of the abundance of active populations (Fig. 2B). Comparing the two functional genes targeted, higher average transcription rates (10–10,000 fold) were always measured for nxrB (RNA/DNA > 105) than for nosZ (RNA/ DNA = 102–104). Denitrifiers are mostly heterotrophic bacteria, while nitrifiers carry out a low-efficient metabolism. Therefore, the amount of active enzymes required by each group to obtain enough energy for maintenance and growth differs considerably. In most of the assays, the Wilcoxon signed-rank test revealed small although significant differences (lower than one order of magnitude) between the quantifications in each BR (Fig. 2A). The similar abundances detected, in spite of the different aeration conditions, can be explained by the high recirculation rate imposed on the system, which is required for efficient N-removal (Judd, 2011). As an exception, the number of active AOB populations showed larger fluctuations between the BRs. The average number of transcripts in the anoxic BR was 2–36-fold higher compared to the aerated BR (Fig. 2A), leading also to higher RNA/DNA ratios (Fig. 2B). In activated sludge from both CAS and MBR systems, it is known that the cell decay rates of AOB are significantly lower under anoxic than under aerated conditions, although the reasons for this difference are not yet clarified (Manser et al., 2006). Besides, the abundance of active AOB populations in the anoxic zone of a WWT plant was previously described (Milner et al., 2008). This fact was explained by the ability of AOB to use nitrite as electron acceptor under O2-limited conditions, a process known as nitrifier denitrification (Chandran et al., 2011). Homologues of a number of genes of the denitrification pathway are present in several species of Nitrosomonas and Nitrosospira, suggesting its universality in AOB (Chandran et al., 2011). The importance of this process for energy

130

C. Gómez-Silván et al. / Bioresource Technology 169 (2014) 126–133

Fig. 2. (A) Box-and-Whisker plots of the number of copies (in logarithmic scale) of targeted genes and transcripts per liter of activated sludge, quantified by qPCR in the fullscale membrane bioreactor (MBR) during spring, summer, and autumn of 2009. Data in the same column followed by the same lower-case letter do not significantly differ according to the Wilcoxon signed-rank test (p < 0.10). (B) Box-and-Whisker plots of RNA/DNA ratios of targeted bacterial groups (copies of transcripts/copies of genes) in the same samples. Data followed by the same lower-case letter do not significantly differ according to the Wilcoxon signed-rank test (p < 0.10). In all the plots, upper and lower bounds of the box denote the 75th and 25th percentiles, upper and lower bounds of bars are the 90th and 10th percentiles. Average values ± standard deviations are shown below the boxes. A: aerated BR; X: anoxic BR; Sp: spring; Sm: summer; Au: autumn. (C) Average ± standard deviation of the number of copies of target genes or transcripts quantified in the MBR, expressed per liter of activated sludge.

generation by AOB under O2-limiting conditions has been proposed (Yu et al., 2010a). The transcription levels of the nosZ gene were frequently similar under both aerated and anoxic conditions (Fig. 2A), indicating active denitrification in the aerated BR. The expression of nosZ under suboxic and oxic conditions was also evidenced in marine environments, and was attributed to the generation of O2-depleted microzones inside bacterial aggregates (Wyman et al., 2013). Indeed, in MBRs there is a bimodal distribution of floc size with micro and macroflocs. Within macroflocs (60–240 lm), the oxygen can only diffuse through the first 100–150 lm, creating anoxic microenvironments in the cores of the cell aggregates. The proportion of both sizes of flocs varies with the operational conditions, the SRT being one of the most influential parameters increasing the proportion of macroflocs (Judd, 2011). Evaluation of the flocs’ diameter was beyond the purpose of this study; however, the high SRT used (20 days) is well known to favor the development of a high proportion of macroflocs with anoxic cores (Judd, 2011). Additionally, bacteria able to perform aerobic denitrification (co-respiration with O2, NO2 or NO3 ) had been isolated in WWT

plants in which the aeration conditions were switched during operation (Frette et al., 1997). However, the contribution of this process to N-removal in WWT remains poorly explored. As in the case of the nosZ gene, there were small although significant differences in the transcription levels of nxrB between the BRs (Fig. 2A), indicating that Nitrospira-like NOB were active under both aerated and anoxic conditions. Nitrospira are regarded as K-strategists, showing a high affinity for O2, which allows them to sustain respiration in low-oxygen environments (Downing and Nerenberg, 2008). In fact, the enrichment of functional Nitrospira populations in activated sludge long-term kept at low DO (0.16 mg/L) has been demonstrated recently (Liu and Wang, 2013). 3.3. Linking the abundance of targeted genes and their transcription levels with environmental variables in the MBR system The MDS plots generated for each of the two BRs (aerated and anoxic) are shown in Fig. 4. Overall, the variables’ ordinations explained 23–38% of the biological data trends (BIO-ENV = 0.230– 0.377), with the exception of RNA-derived quantifications in the

C. Gómez-Silván et al. / Bioresource Technology 169 (2014) 126–133

131

Fig. 3. Relative abundance, expressed as percentage of total bacterial community, of the bacteria, nitrifiers and denitrifiers measured in DNA samples of the full-scale membrane bioreactor (MBR) during the spring, summer, and autumn of 2009. A: aerated BR; X: anoxic BR.

aerated BR (BIO-ENV = 0.088, Fig. 4E). The later MDS plot will not be further discussed in this section, as the contribution of variables to explain the biological responses was statistically negligible. In the remaining MDS plots, the variables were highly and consistently correlated to the biological data trends, although markedly different patterns were found for each target group (Fig. 4). As a general rule, the abundances of Bacteria and denitrifiers displayed similar trends of response to most of the included variables, while AOB responses were often opposed. This tendency was particularly clear in the anoxic BR, where the contribution of environmental variables to explain changes in the abundance of the targeted groups was the highest (Fig 4D and F). Increasing IT was linked to a higher abundance of total and active populations of Bacteria, denitrifiers and Nitrospira-like NOB in the anoxic BR (r = 0.55–0.9) (Fig. 4B and D; B and F). In contrast, a strong negative correlation (r 6 0.7) was observed between IT and the abundance of total AOB in both BRs (Fig. 4A and C; B and D). The influence of IT on denitrifying populations was in accordance with the previous knowledge of the denitrification process. Although denitrification takes place under a wide range of temperatures (5–50 °C), both N2O production and reduction are reported to increase with increasing temperature (Holtan-Hartwig et al., 2002). As opposed to denitrification, nitrification is strongly influenced by temperature, particularly ammonia oxidation (Bitton, 2010). In a long-term study analyzing twelve municipal WWT plants, Limpiyakorn et al. (2005) found that both the abundance and the activity of AOB were largely influenced by seasonal temperature variations, but in different ways. While ammonia oxidizing rates were higher in summer (27–31 °C), the size of the total AOB community significantly increased at lower temperatures in autumn (19–26 °C) and even winter (14–22 °C) in most of the WWT plants. Negative correlations between temperature and the abundance of AOB were observed in soils as well (Adair and Schwartz, 2008). Nitrospira spp. are known to efficiently oxidize nitrite under a wide range of temperatures (Alawi et al., 2009), while AOB’s activity begins to decline at 15 °C (Kim et al., 2006). Control of temperature in full-scale systems is unfeasible due to its high cost. During the sampling period in the studied MBR, IT (15.3–28.8 °C) was

constantly over the critical limit for nitrification. Accordingly, there was no correlation between the abundance of the active AOB populations and IT (Fig. 4B and F), demonstrating that under the tested climate conditions, temperature shifts were not crucial for ammonia oxidation. In the anoxic BR, most of the targeted bacterial groups showed positive quantitative responses to the availability of organic substrates in the influent water. In particular, high positive correlations were displayed between BOD5 and iVSS, and the abundance of active populations of Bacteria (r = 0.9 and r = 0.5, respectively), AOB (r = 0.8 and r = 0.99, respectively) and Nitrospira-like NOB (r = 0.99 and r = 0.7, respectively) (Fig. 4B and F). Nitrifiers are classically addressed as autotrophic organisms; however, Nitrosomonas europaea and Nitrosomonas eutropha use several organic compounds as an energy source under anoxic conditions (Schmidt, 2009). Additionally, both Nitrosomonas and Nitrospira spp. assimilate pyruvate as C source, although the ability to also use this organic substrate for the generation of energy has not been demonstrated in Nitrospira spp. (Lücker et al., 2010; Schmidt, 2009). The accumulation of biomass inside the MBR (VSS) displayed a strong negative influence on the abundance of the whole bacterial community and the denitrifiers (r < 0.9) in both BRs (Fig. 4A and C; B and D). Although the retention of suspended solids in MBRs allows for the accumulation of higher bacterial cell numbers in the sludge, the increasing concentration of biomass lowers the microbial metabolic activities, and the growth rates decrease when the substrates become limited (Gómez-Silván et al., 2013). The MBR was operated under a low F/M ratio (Table S1); thus, the increase of VSS in the BRs correlated negatively with the abundance of heterotrophs. On the contrary, VSS promoted the abundance of AOB (r = 0.9, Fig. 4B and D) and their level of rRNA transcription (r = 0.7, Fig. 4B and F) under anoxic conditions. AOB are able to grow autotrophically, being thus unaffected by low F/ M ratios and favored by the long SRT. The NH+4 concentration in the influent water displayed high and consistent correlations with the responses of all the bacterial groups quantified in the study. The effect of this variable on the abundance of total and active populations of Bacteria, Nitrospiralike NOB and denitrifiers was negative, particularly in the anoxic

132

C. Gómez-Silván et al. / Bioresource Technology 169 (2014) 126–133

Fig. 4. Non-metric multidimensional scaling (MDS) plots illustrating the distributions of the sludge samples in each bioreactor (aerated and anoxic), according to the relative similarity of the operational parameters of the full-scale membrane bioreactor (MBR) during the spring, summer, and autumn of 2009. The vectors represent the direction along the samples of each operational parameter or each bacterial group’s quantification based in DNA and RNA. IT: internal temperature; BOD5: biological oxygen demand in five days; C/N: carbon/nitrogen ratio, expressed as chemical oxygen demand/total nitrogen, COD/tN; NH+4: ammonia concentration; iVSS and iFSS: volatile and fix suspended solids in the influent wastewater; VSS and FSS: volatile and fix suspended solid in the bioreactors. The low stress level (stress < 0.05) of the MDS plots indicate excellent fit of the variables’ data into two dimensional spaces. BIO-ENV values indicate the level of explanation of the biological data trends given by the variables’ ordination.

BR (r 6 0.5) (Fig. 4 B and D; B and F). AOB populations showed a behavior opposite to the rest of the targeted groups, as the abundances of AOB 16S rDNA and rRNA copies were always positively correlated to NH+4 concentration (r > 0.7). The NH+4 concentration defines the growth rate of AOB, since they follow Monod kinetics (Bitton, 2010), and the positive correlation between the size and activity of their communities and NH+4 input was observed in previous studies (Avrahami et al., 2002; Zhang et al., 2013). However, to the best of the authors’ knowledge, there are no previous reports of negative correlations between the increase of NH+4 concentration and the abundance of total bacteria or denitrifiers in WWT plants. Studies conducted in forest soils concluded that under limited organic C substrate availability (starvation conditions), heterotrophs reduce their demand for N, and autotrophic nitrifiers become very competitive for NH+4 (Hart et al., 1994). A significant reduction in the abundance of copies of the nirS, nirK and nosZ genes was also observed in soils in response to increased NH+4 content (Zhang et al., 2013). In natural water columns and soils, the structure of the denitrifying communities is largely influenced by NH+4 concentration (Avrahami et al., 2002). Due to the vast phylogenetic variability of denitrifiers, changes in their population diversity are likely related to shifts in their abundance or activity. In both BRs, the abundances of total populations of Bacteria, denitrifying and Nitrospira-like NOB were linked to lower C/N

ratios (r 6 0.55, r 6 0.7 and r < 0.5, respectively), contrary to the trends observed for AOB (r = 0.2–0.99) (Fig. 4A and C; B and D). Kumar et al. (2012) studied in detail the relationships between the C/N ratio and the efficiency of the nitrification–denitrification process in an MBR. While denitrification started to be appreciable at C/N = 3, the maximum nitrification rates were detected at C/N = 2.5, and the efficiency started dropping off as the ratio increased. However, the relative abundance of AOB decreased only one order of magnitude when the C/N ratio increased from 1 to 9, indicating a weaker influence on the abundance of AOB populations than on their metabolic activity. In MBRs combining aerated and anoxic phases, a C/N ratio of 5–15 is required to achieve P70% of total N-removal by simultaneous nitrification and denitrification (Kumar et al., 2012). The present study results are in accordance, as the C/N ratio in the MBR influent was inside the abovementioned range throughout the study, and the average removal of total N (Table S1) was close to the 75–80% limit predicted for predenitrification-based systems (Tchobanoglous et al., 2003). 4. Conclusions The abundances of the bacterial groups targeted by qPCR/ RT-qPCR were explained in a significant proportion (23–38%) by the shifts of environmental variables. AOB abundance correlated positively with NH+4 concentration, C/N ratio and the accumulation

C. Gómez-Silván et al. / Bioresource Technology 169 (2014) 126–133

of VSS in the BRs. In contrast, active populations of Bacteria, denitrifiers, and Nitrospira-like NOB showed opposing trends of response to these variables. These results provide evidence for the complexity of the interactions among the variables influencing MBR systems, contributing relevant data for the improvement of their design and performance. Acknowledgements The authors wish to thank Junta de Andalucía (project reference NET 324936/1) and the Spanish Ministerio de Ciencia e Innovación (MICINN) (project reference CTM2010-17609/TECNO) for supporting this research in collaboration with Fondo Europeo de Desarrollo Regional (FEDER). MICINN and Spanish Ministerio de Educación are also acknowledged for personal grants to C.G.S. (FPI program and Short-term stays, EEBB, program) and J.A. (FPU program), respectively. The authors also thank EMASAGRA for providing their facilities at Estación Depuradora Sur (Churriana, Granada, Spain). Special thanks to Jorge Pérez for his support throughout the study, to Agnes Waliczek for her important contribution with cloning assays, to Amélia Camarinha Silva for her valuable personal and professional support, and to Proof-Reading-Service.com for thorough English language revision. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biortech.2014.06. 089. References Adair, K.L., Schwartz, E., 2008. Evidence that ammonia-oxidizing archaea are more abundant than ammonia-oxidizing bacteria in semiarid soils of Northern Arizona, USA. Microb. Ecol. 56, 420–426. Alawi, M., Off, S., Kaya, M., Spieck, E., 2009. Temperature influences the population structure of nitrite-oxidizing bacteria in activated sludge. Environ. Microbiol. Rep. 1, 184–190. Avrahami, S., Conrad, R., Braker, G., 2002. Effect of soil ammonium concentration on N2O release and on the community structure of ammonia oxidizers and denitrifiers. Appl. Environ. Microbiol. 68, 5685–5692. Bitton, G., 2010. Wastewater Microbiology, fourth ed. Wiley-Lis, John Wiley and Sons, NJ, USA. Calderón, K., González-Martínez, A., Gómez-Silván, C., Osorio, F., Rodelas, B., González-López, J., 2013. Archaeal diversity in biofilm technologies applied to treat urban and industrial wastewater: recent advances and future prospects. Int. J. Mol. Sci. 14, 18572–18598. Chandran, K., Stein, L.Y., Klotz, M.G., van Loosdrecht, M.C., 2011. Nitrous oxide production by lithotrophic ammonia-oxidizing bacteria and implications for engineered nitrogen-removal systems. Biochem. Soc. Trans. 39, 1832–1837. Clarke, K.R., Somerfield, P.J., Gorley, R.N., 2008. Testing of null hypotheses in exploratory community analyses: similarity profiles and biota-environment linkage. J. Exp. Mar. Biol. Ecol. 366, 56–69. Downing, L.S., Nerenberg, R., 2008. Effect of oxygen gradients on the activity and microbial community structure of a nitrifying, membrane-aerated biofilm. Biotechnol. Bioeng. 101, 1193–1204. European Council, 1991. Urban Wastewater Treatment (UWWT) Directive, 91/271/ EEC. Frette, L., Gejlsbjerg, B., Westermann, P., 1997. Aerobic denitrifiers isolated from an alternating activated sludge system. FEMS Microbiol. Ecol. 24, 363–370. Geets, J., de Cooman, M., Wittebolle, L., Heylen, K., Vanparys, B., De Vos, P., Verstraete, W., Boon, N., 2007. Real-time PCR assay for the simultaneous quantification of nitrifying and denitrifying bacteria in activated sludge. Appl. Microbiol. Biotechnol. 75, 211–221.

133

Gómez-Silván, C., Arévalo, J., Pérez, J., González-López, J., Rodelas, B., 2013. Linking hydrolytic activities to variables influencing a submerged membrane bioreactor (MBR) treating urban wastewater under real operating conditions. Water Res. 47, 66–78. Hart, S.C., Nason, G.E., Myrold, D.D., Perry, D.A., 1994. Dynamics of gross nitrogen transformations in an old-growth forest: the carbon connection. Ecology 75, 880–891. Holtan-Hartwig, L., Dörsch, P., Bakken, L.R., 2002. Low temperature control of soil denitrifying communities: kinetics of N2O production and reduction. Soil Biol. Biochem. 34, 1797–1806. Huang, Z., Gedalanga, P.B., Asvapathanagul, P., Olson, B.H., 2010. Influence of physicochemical and operational parameters on Nitrobacter and Nitrospira communities in an aerobic activated sludge bioreactor. Water Res. 44, 4351– 4358. Judd, S., 2011. Principles and applications of membrane bioreactors in water and wastewater treatment, . The MBR Book, second ed. Elsevier Science, Oxford, UK. Kim, D.-J., Lee, D.-I., Keller, J., 2006. Effect of temperature and free ammonia on nitrification and nitrite accumulation in landfill leachate and analysis of its nitrifying bacterial community by FISH. Bioresour. Technol. 97, 459–468. Kim, Y.M., Cho, H.U., Lee, D.S., Park, D., Park, J.M., 2011. Influence of operational parameters on nitrogen removal efficiency and microbial communities in a fullscale activated sludge process. Water Res. 45, 5785–5795. Kumar, M., Lee, P.-Y., Fukusihma, T., Whang, L.-M., Lin, J.-G., 2012. Effect of supplementary carbon addition in the treatment of low C/N high-technology industrial wastewater by MBR. Bioresour. Technol. 113, 148–153. Le-Clech, P., 2010. Membrane bioreactors and their uses in wastewater treatments. Appl. Microbiol. Biotechnol. 88, 1253–1260. Limpiyakorn, T., Shinohara, Y., Kurisu, F., Yagi, O., 2005. Communities of ammoniaoxidizing bacteria in activated sludge of various sewage treatment plants in Tokyo. FEMS Microbiol. Ecol. 54, 205–217. Liu, G., Wang, J., 2013. Long-term low DO enriches and shifts nitrifier community in activated sludge. Environ. Sci. Technol. 47, 5109–5117. Lücker, S., Wagner, M., Maixner, F., Pelletier, E., Koch, H., Vacherie, B., Rattei, T., Sinninghe Damsté, J.S., Spieck, E., Le Paslier, D., Daims, H., 2010. A Nitrospira metagenome illuminates the physiology and evolution of globally important nitrite-oxidizing bacteria. Proc. Natl. Acad. Sci. U.S.A. 107, 13479–13484. Manser, R., Gujer, W., Siegrist, H., 2006. Decay processes of nitrifying bacteria in biological wastewater treatment systems. Water Res. 40, 2416–2426. Milner, M.G., Curtis, T.P., Davenport, R.J., 2008. Presence and activity of ammoniaoxidising bacteria detected amongst the overall bacterial diversity along a physico-chemical gradient of a nitrifying wastewater treatment plant. Water Res. 42, 2863–2872. Palmer, K., Drake, H.L., Horn, M.A., 2009. Genome-derived criteria for assigning environmental narG and nosZ sequences to operational taxonomic units of nitrate reducers. Appl. Environ. Microbiol. 75, 5170–5174. Purkhold, U., Wagner, M., Timmermann, G., Pommerening-Röser, A., Koops, H.-P., 2003. 16S rRNA and amoA-based phylogeny of 12 novel betaproteobacterial ammonia-oxidizing isolates: extension of the dataset and proposal of a new lineage within the nitrosomonads. Int. J. Syst. Evol. Microbiol. 53, 1485–1494. Schmidt, I., 2009. Chemoorganoheterotrophic growth of Nitrosomonas europaea and Nitrosomonas eutropha. Curr. Microbiol. 59, 130–138. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. Wastewater Engineering: Treatment and Reuse. Metcalf & Eddy. Inc., McGraw-Hill, NY. Winkler, M.K., Bassin, J.P., Kleerebezem, R., Sorokin, D.Y., van Loosdrecht, M.C., 2012. Unravelling the reasons for disproportion in the ratio of AOB and NOB in aerobic granular sludge. Appl. Microbiol. Biotechnol. 94, 1657–1666. Wyman, M., Hodgson, S., Bird, C., 2013. Denitrifying Alphaproteobacteria from the Arabian sea that express nosZ, the gene encoding nitrous oxide reductase, in oxic and suboxic waters. Appl. Environ. Microbiol. 79, 2670–2681. Xia, S., Jia, R., Feng, F., Xie, K., Li, H., Jing, D., Xu, X., 2012. Effect of solids retention time on antibiotics removal performance and microbial communities in an A/OMBR process. Bioresour. Technol. 106, 36–43. Yu, R., Kampschreur, M.J., van Loosdrecht, M.C.M., Chandran, K., 2010a. Mechanisms and specific directionality of autotrophic nitrous oxide and nitric oxide generation during transient anoxia. Environ. Sci. Technol. 44, 1313–1319. Yu, T., Qi, R., Li, D., Zhang, Y., Yang, M., 2010b. Nitrifier characteristics in submerged membrane bioreactors under different sludge retention times. Water Res. 44, 2823–2830. Zhang, B., Sun, B., Ji, M., Liu, H., Liu, X., 2010. Quantification and comparison of ammonia-oxidizing bacterial communities in MBRs treating various types of wastewater. Bioresour. Technol. 101, 3054–3059. Zhang, X., Liu, W., Schloter, M., Zhang, G., Chen, Q., Huang, J., Li, L., Elser, J.J., Han, X., 2013. Response of the abundance of key soil microbial nitrogen-cycling genes to multi-factorial global changes. PLoS ONE 8, e76500.