Correlation of Community Dynamics and Process Parameters As a

ARTICLE pubs.acs.org/est

Correlation of Community Dynamics and Process Parameters As a Tool for the Prediction of the Stability of Wastewater Treatment Susanne G€unther,† Christin Koch,‡ Thomas H€ubschmann,† Isolde R€oske,§ Roland Arno M€uller,|| Thomas Bley,^ Hauke Harms,† and Susann M€uller†,* †

Department of Environmental Microbiology, UFZ
Helmholtz Centre for Environmental Research, Leipzig, Germany Department of Bioenergy, UFZ
Helmholtz Centre for Environmental Research, Leipzig, Germany § Institute of Microbiology, University of Technology, Zellescher Weg 20b, Dresden, Germany Centre for Environmental Biotechnology, UFZ
Helmholtz Centre for Environmental Research, Leipzig, Germany ^ Institute of Food Technology and Bioprocess Engineering, University of Technology, Bergstrasse 120, Dresden, Germany

)

‡

bS Supporting Information ABSTRACT: Wastewater treatment often suffers from instabilities and the failure of specific functions such as biological phosphorus removal by polyphosphate accumulating organisms. Since most of the microorganisms involved in water clarification are unknown it is challenging to operate the process accounting for the permanent varying abiotic parameters and the complex composition and unrevealed metabolic capacity of a wastewater microbial community. Fulfilling the demands for water quality irrespective of substrate inflow conditions may emit severe problems if the limited management resources of municipal wastewater treatment plants are regarded. We used flow cytometric analyses of cellular DNA and polyphosphate to create patterns mirroring dynamics in community structure. These patterns were resolved in up to 15 subclusters, the presence and abundances of which correlated with abiotic data. The study used biostatistics to determine the kind and strength of the correlation. Samples investigated were obtained from a primary clarifier and two activated sludge basins. The stability of microbial community structure was found to be high in the basins and low in the primary clarifier. Despite major abiotic changes certain subcommunities were dominantly present (up to 80% stability), whereas others emerged only sporadically (down to 3% stability, both according to equivalence testing). Additionally, subcommunities of diagnostic value were detected showing positive correlation with substrate influxes. For instance blackwater (rs = 0.5) and brewery inflow (both rs = 0.6) were mirrored by increases in cell abundances in subclusters 1 and 6 as well as 4 and 8, respectively. Phosphate accumulation was obviously positively correlated with nitrate (rs = 0.4) and the presence of denitrifying organisms (Rhodacyclaceae). Various other correlations between community structure and abiotic parameters were apparent. The bacterial composition of certain subcommunities was determined by cell sorting and phylogenetic tools like T-RFLP. In essence, we developed a monitoring tool which is quick, cheap and causal in its interpretation. It will make laborious PCR based technique less obligatory as it allows reliable process monitoring and control in wastewater treatment plants.

’ INTRODUCTION Wastewater treatment plants (WWTPs) are mostly managed as black box systems and their functioning is controlled using empirical knowledge of correlations with abiotic process parameters. Their stability is reasoned based on experience with plant design, amount and composition of wastewater, weather conditions etc. Despite all precautions, wastewater treatment is failure prone. The microorganisms responsible for the clarification process, though having been investigated since decades in countless WWTPs,1 still escape control. In the past scientists focused on cultivable organisms1 which however comprise only one to fifteen percent of the whole community.2 Therefore, more recent research focused on cultivation-independent fingerprinting techniques3 combined with the analysis of clone libraries or r 2011 American Chemical Society

the so-called full cycle rRNA analysis including in situ observation.4 These relatively expensive approaches are more informative than cultivation, since they account better for the complex microbiology of WWTPs. However, they do not inform well about abundances of the community members due to selective DNA extraction, amplification and the choice of restriction enzymes (e.g., refs 5 and 6). Special Issue: Ecogenomics: Environmental Received: April 6, 2011 Accepted: August 1, 2011 Revised: June 20, 2011 Published: August 01, 2011 84

dx.doi.org/10.1021/es2010682 | Environ. Sci. Technol. 2012, 46, 84–92

Environmental Science & Technology

ARTICLE

and conserved in fixation buffer. Approximately 109 cells mL
1 were fixed for following cytometric analyses and found to be stable for about 9 days. For cell sorting only the 10% sodium azide PBS solution was used for fixation to avoid disturbances during the following PCR procedure (see below). Therefore, such samples can be stored for only up to 3 days without change in community structure. Staining Procedures. DNA was stained using the fluorescent dye DAPI as described previously.10 Cellular poly-P was stained with the antibiotic tetracycline hydrochloride (Tc, Roth, Germany). Tc was applied as a 2 mg mL
1 stock solution in bidistilled water (4.16 mM) to give a final concentration of 0.225 mM in the cell suspension. Stock solutions were prepared freshly to avoid Tc aging.10 Vital cells were washed in PBS (see above) and diluted to a cell density of approximately 0.035 (dλ700 nm = 0.5 cm) and centrifuged again. The pellet was treated with 50 μL of the Tc stock solution for 30 min in the dark at room temperature (RT). Prior measurement 1 mL PBS was added and the sample measured by flow cytometry immediately. Controls were performed as described in G€unther et al. (2009, ref 10). To verify reliable staining the cells were subjected to microscopy and image analysis. Fluorescence spectra of poly-P bound and unbound Tc and the controls are shown in Supporting Information Figure SI 1. Up to 250 000 cells were analyzed and the dominant subcommunities presented in blue or yellow color. Flow Cytometry. Analyses were carried out using a MoFlo cell sorter (DakoCytomation, U.S.). Excitation of 400 mW at 488 nm was used to analyze forward scatter (FSC) which is related to cell size and side scatter (SSC) which is related to cell granularity. SSC was used as a trigger signal to discriminate bacterial cells from electronic noise. Different from the procedure described in ref 10 the Tc-fluorescence was recorded after excitation at 488 nm by using a 520/20 nm band-pass filter. DAPI was excited by 100 mW of ML-UV (333
365 nm). Tuning of the device, filter settings and PMT choices are described in Kleinsteuber et al. (2006, ref 3). Bacterial cells were sorted using the most accurate sort mode (single and one drop mode: highest purity 99%) at a rate not higher than 2500 cells per second. Cell sorting was performed using the four-way-sort-option at high speed (12 m s
1). The cells were sorted into sterile 1.5 mL plastic tubes. To obtain sufficient DNA for the following DNA preparation and analyses, up to 106 cells were sorted. The sort gates (=ellipses) were placed around particularly abundant subcommunities defined by similarity in fluorescence and light scatter properties. To determine the abundances of cells within these subcommunities (=gates or subsets of cells) specific cytometric programs were used (Summit 3.1, DakoCytomation, Fort Collins, CO). DNA Preparation. DNA was extracted from Tc and DAPI stained, cytometrically sorted cells. The cells were harvested from the sheath buffer by centrifugation (25 min, 20 000g at 6 C) and stored at
20 C. DNA extraction was performed with a chelex based extraction method.11 Between 0.22 and 1.0  106 sorted cells were suspended in 70 μL of 10% (wt/vol) chelex solution (Chelex 100 Resin, Bio-Rad), vortexed for 10 s at highest speed and then incubated for 45 min at 95 C. Afterward, the samples were again vortexed for 10 s and then centrifuged at 7000g for 5 min at 4 C. The supernatant was carefully transferred to a new, prechilled tube and directly used for PCR analysis or stored at
20 C. For the unsorted cell fractions 0.8 or 1 mL sheath buffer with stained cells was collected. The DNA was extracted as described above with the only difference that

Direct quantitative monitoring of arising and disappearing populations would be an attractive way to survey a WWTP’s performance and stability. We approached this goal using flow cytometry as a high-throughput method of single-cell analysis. As application of fluorescent oligomer techniques (like FISH, CARD-FISH or TSA-FISH) is problematic for flow cytometric measurement of samples from natural environments (see discussion in ref 7) simpler approaches were used in this study. Optical characteristics of single cells having diagnostic value are diffraction (forward scatter, FSC), refraction (side scatter, SSC) and fluorescence caused by specific staining of polyphosphate (poly-P) and DNA. Earlier studies have shown that cytometrically derived patterns can be correlated with quantitative phylogenetic community compositions.3,8 Here we demonstrate that time-resolved cytometric pattern monitoring can be used to judge the functional stability of WWTPs. To this aim, the derived patterns were correlated to abiotic parameters (e.g., nitrogen regime) and external boundary conditions (weather, wastewater composition). The genetic analyses gave additional confidence in the methods as they helped to underpin the derived correlations mechanistically. Enhanced biological phosphorus removal (EBPR) was chosen as an example since it is of increasing interest for WWTPs because the product of this process is deployable in many ways and profitable.9

’ EXPERIMENTAL SECTION Study Area. The WWTP Eilenburg in Saxony, Germany, was chosen as a municipal plant typical for Germany. The WWTP has a service area of about 49 000 (with midrated load of 37 584) inhabitants including several business parks with relatively constant wastewater discharges. From Monday to Friday a local brewery (midrated 10 000 m3 month
1 and amounts altering between 1 and 791 m3 day
1) and a mineral water plant (midrated 12 000 m3 month
1) also drain their wastewater into the WWTP. Phosphorus removal is primarily biological, however, if high phosphate concentrations occur (1.5 mg L
1), chemical precipitation with Fe (III) is initiated. Meteorological and abiotic conditions of the WWTP area are summarized in Supporting Information Table SI 1a,b. After rough mechanical treatment, the wastewater flows through a primary clarifier (PC) where it stays 30 min before its purification in about 4 h within two parallel activated sludge tanks (AT 1 and 2) set to sequential anaerobic and aerobic phases altering in a 90 min mode for intermittent nitrification/denitrification processes but with an aeration regime which is inverse. Subsequently, the purified effluent is directed into the secondary clarifier (for 3 h) whereas the activated sludge is reused or channeled into the digestion tank. Chemical Analysis. Chemical analyses were performed according to the German DIN guidelines (Supporting Information Table SI 1a,b). Sampling. Cells from wastewater were harvested from the primary clarifier (PC) and the two parallel activated sludge tanks (AT 1 and 2). Samples were taken several times a week, fixed with fixation buffer (pH 7.0, vol/vol) containing 5 mM BaCl2 (BaCl2  2H2O; Laborchemie Apolda, Germany), 5 mM NiCl2 (NiCl2  6H2 O; Merck, Germany), and 10% sodium azide (Merck, Germany) right at the sampling site and kept on ice until final fixation. For final fixation the samples were sonicated for 5 min, washed thrice to remove any disturbing substances (cell debris, organic material) for 10 min at 3200g, sonicated again for 10 min 85

dx.doi.org/10.1021/es2010682 |Environ. Sci. Technol. 2012, 46, 84–92

Environmental Science & Technology

ARTICLE

150 μL of 10% (wt/vol) chelex solution was used. Sheath buffer (0.8 or 1 mL) without any introduced sample was included as control at every sorting day and handled in the same way as sorted cells. 16S rRNA Gene Clone Library. DNA from the sample day 54 was used for bacterial 16S rRNA gene fragment amplification and construction of a clone library. The PCR reaction was performed in a 25 μL reaction volume containing 12.5 μL Taq PCR master mix (Qiagen, Germany), 5 pmol of each of the universal primers 27F and 1492R (Microsynth, Switzerland) and 1 μL of the template DNA (details in ref 10). T-RFLP Analyses. T-RFLP analyses were performed for PCRamplified bacterial 16S rRNA gene fragments (details in 10) using restriction endonucleases HaeIII and RasI (New England Biolabs, Germany). Nucleotide Sequence Accession Numbers. The 115 partial sequences of the 16S rRNA gene determined in this study were deposited in the GenBank database under accession numbers JN000704 to JN000818. Stability Analysis
Equivalence Tests. Here the criterion for the stability of the subcommunities is the equivalence of the DNA-Scatter data sets created at different sampling times. To prove the equivalence of two data sets an equivalence test needs to be applied (for detailed information see ref 12). Since highly diverse activated sludge was measured the maximally acceptable tolerance for the tolerance interval was set to 25%. The values for the stability of each subcommunity were determined day by day with the cell abundances of day 1 as reference point. If the proportion of the cells of the respective subcommunity/gate is within the 25% range then the subcommunity at this day is regarded as stable and the number of stable subcommunity values (days) is expressed as percentage of all days. Correlation Analyses. Correlation analyses were done using the Spearman’s rank-order correlation coefficient (rS) and evaluated with Kendall’s rank correlation coefficient tau (τ, not shown, for further information see 13,14). Both factors were calculated using the program PAST (Paleontological STatistics, Version 2.02, distributed by Øyvind Hammer at the Natural History Museum, University of Oslo; 15). Statistical significance is tested by PAST for n > 10 (n: number of data sets/sample days) with a two tailed Student t test with n-2 degrees of freedom at a significance level of 5% (confidence interval 0.05). For n < 10 the program switches automatically to an exact test, which compares the observed correlation results to the values obtained from all possible permutations of the first parameter.15 Test results were additionally verified using the Monte Carlo permutation test with 10 000 random replicates. Both coefficients vary between
1 (inverse ranks) and 1 (identical ranks). If rS or τ values are >0.8 or