Community survey of ammonia-oxidizing bacteria in full-scale activated sludge processes with different solids retention time

Authors


Sara Hallin, Department of Microbiology, Swedish University of Agricultural Sciences, Box 7025, S-750 07 Uppsala, Sweden
(e-mail: sara.hallin@mikrob.slu.se).

Abstract

Aims:  To study the effects of different solids retention time (SRT) on the nitrification activity and community composition of ammonia-oxidizing bacteria (AOB) in two full-scale activated sludge processes during a 5-month period.

Methods and Results:  The AOB community composition was analysed using fluorescence in situ hybridization (FISH) and denaturing gradient gel electrophoresis (DGGE), and the identified populations were enumerated by quantitative FISH. Potential nitrification rates were determined in batch tests and the in situ rates were calculated from mass balances of nitrogen in the plants. Increased SRT reduced the nitrification activity, but neither the number per mixed liquor suspended solids nor community composition of AOB were affected. Two dominant AOB populations related to Nitrosomonas europaea and Nitrosomonas oligotropha were identified by FISH, whereas only the latter could be detected by DGGE.

Conclusions:  The effect of a longer SRT on the activity was probably because of physiological changes in the AOB community rather than a change in community composition.

Significance and Impact of the Study:  Physiological alterations of a stable AOB community are possible and may stabilize activated sludge processes. The commonly used FISH probes designed to target all beta-proteobacterial AOB does not detect certain Nitrosomonas oligotropha populations, leading to an underestimation of AOB if a wider set of probes is not used.

Introduction

Biological nitrogen removal from wastewater is dependent on nitrification of ammonia (NH3) to nitrite (NOinline image) and nitrate (NOinline image) followed by denitrification that reduces these products to dinitrogen gas (N2). Nitrification is carried out in two sequential steps by two phylogenetically different groups of lithoautotrophic bacteria; the ammonia-oxidizing bacteria (AOB) and the nitrite-oxidizing bacteria. The number, activity, and community composition of nitrifying bacteria controls the conversion of nitrogen in wastewater, where ammonia oxidation is considered as the rate-limiting step. Ammonia-oxidation activity is dependent on the AOB community structure given that different populations of AOB exhibit different substrate affinities, growth kinetics, and sensitivities to environmental factors. Nitrifying bacteria, and especially the AOB, are sensitive to toxic compounds in the sewage, and to pH and temperature shifts, which often results in process failure (Eighmy and Bishop 1989). Fluctuations in flow, organic load, and solids retention time (SRT) also affect the nitrification process. Even though AOB are fairly well studied, nitrification is difficult to maintain in wastewater treatment systems. This is in part because of the slow growth rate of nitrifying bacteria, which makes nitrogen removal even more challenging in cold climates. A better understanding of the structure and function of AOB communities in wastewater treatment processes is essential for controlling, improving and optimizing process performance.

Since the advent of molecular techniques, the 16S rRNA approach is often used to determine the AOB composition in environmental samples both by direct methods such as in situ hybridization and by indirect methods, such as PCR followed by analysis of PCR-products. The latter includes sequencing of clone libraries or different fingerprinting methods based on electrophoretic separation of PCR-products, e.g. denaturing gradient gel electrophoresis (DGGE; Kowalchuk et al. 1997) followed by sequencing. By combining in situ techniques, such as fluorescence in situ hybridization (FISH), with PCR-based methods the advantages of each can be combined thus giving a more reliable picture of the AOB community in activated sludge. By using nested 16S rRNA-targeted oligonucleotide probe sets it is possible to analyse AOB diversity on different taxonomic levels. Moreover, FISH has not only been successfully used to determine community structure of AOB but also to enumerate and study the spatial distribution of AOB in wastewater treatment processes or other high ammonium loading systems (e.g. Wagner et al. 1995; Mobarry et al. 1996; Schramm et al. 1998).

The aim of this work was to survey the AOB communities in relation to nitrification during a 5-month period in two full-scale activated sludge processes, within the same treatment plant, with different SRTs and mixed liquor suspended solids (MLSS) concentrations. Both processes were operated with fairly long SRTs to ensure stable nitrification as the required aerobic SRT is more than 10 d in cold climates. Both the MLSS concentration and retention time could affect the activity, density, and composition of the biomass in the activated sludge. We assumed that the process with the shorter SRT should have a higher nitrification rate than a system with longer SRT. A change in nitrification rate would either depend on a difference in AOB numbers, AOB community composition or an increased cell-specific ammonia oxidation rate of the same populations. To test this, we monitored nitrification rates, numbers and community composition of AOB in two separate trains operated in different modes, as indicated above, during 22 weeks from January to June at a Swedish municipal wastewater treatment plant (WWTP). We used quantitative FISH to determine the numbers of AOB and to detect certain groups of AOB during the experimental period. A large set of probes targeting the 16S rRNAs of identified AOB groups were used with a nested approach. To our knowledge, FISH has not earlier been employed for surveys of AOB numbers other than for shorter periods. The AOB community composition determined by DGGE analysis of partial 16S rRNA genes (rDNA) was discussed in relation to that obtained by FISH.

Materials and methods

WWTP operation and samples

Henriksdal municipal WWTP, Stockholm, Sweden, was constructed for 640 000 person equivalents and employs a single-sludge predenitrification activated sludge process with seven parallel trains. Each train consists of one basin with a total volume of 29 000 m3, divided in six serial zones, followed by two parallel clarifiers. Zones 4–6 were aerated. Process water was re-circulated from zones 6 to 1 with a constant flow of 1·8 m3 s−1. The average influent flow was 0·42 m3 s−1 with a load of 33 ± 3·8 mg total N l−1, 26 ±3·1 mg NHinline image–N l−1, and 210 ± 19 mg chemical oxygen demand (COD) l−1 during the experimental period.

Two trains at the plant were run in two different modes during 22 weeks of operation between January and June 2001: I, long SRT and II, short SRT (Table 1). The latter was the standard operation at the WWTP and at the start of the experiment one train was switched to operation mode I. The SRT was controlled by removal of excess sludge, which resulted in different MLSS concentrations in the two processes (Table 1). In the following text only the terms long SRT and short SRT will be used for the different processes. After 9 weeks of build-up of MLSS in the system, which was accomplished by a decreased excess sludge flow, the aerobic SRT stabilized at 15·6 ± 3·6 d (n = 12). The aerobic SRT in the train with operation II was 10·7 ± 2·1 d (n = 22) when calculated over the whole period. The difference in SRT was statistically significant during the last 12 weeks (t-test, P < 0·001). During the experimental period the temperature increased from 10 to 17°C, with the major increase between the end of April and June.

Table 1.  Operational modes of the two full-scale experimental activated sludge processses
ParameterI (long SRT) (mean ± SD)II (short SRT) (mean ± SD)
MLSS (mg l−1)3300 ± 3402300 ± 250
MLVSS/MLSS (%)65 ± 164 ± 2
DO (mg l−1)2·02·8
Aerobic SRT (d)15·6 ± 3·610·7 ± 2·1
Nitrogen removal (%)79 ± 676 ± 5
Effluent NOinline image–N (mg l−1)4·9 ± 1·05·7 ± 0·9
Effluent NH4–N (mg l−1)0·5 ± 0·60·7 ± 0·9

Grab samples for AOB studies were taken weekly in the effluent flow of zone 6. Potential nitrification activity was measured directly on fresh samples. For the DNA-based studies and the FISH analysis, 10-ml portions of activated sludge was centrifuged for 10 min at 4500 g. The supernatant was discarded and the pellet was stored at −20°C.

Process monitoring and chemical analysis

In the influent and effluent water MLSS was analysed and total N, NHinline image–N, NOinline image–N, and COD were determined by colorimetric methods, all according to Swedish standards (SIS, Swedish Standards Institute; http://www.sis.se), in 24-h composite samples during weekdays and 48-h composite samples during weekends. Grab samples from the effluent of the aerated basins were analysed for mixed liquor suspended solids (MLSS), MLVSS, and pH according to Swedish standards (http://www.sis.se). On-line instruments also monitored temperature, pH, and MLSS (Cerlic, Solna, Sweden) in the basins, as well as dissolved oxygen (DO) in zones 3–6 (Danfoss, Nordborg, Denmark) and NHinline image–N in the effluent (Contronic, Dr Lange AB, Sköndal, Sweden).

Nitrification rates

The potential nitrification rate was estimated weekly in four replicate samples according to the method developed by Vandkvalitetsinstitutet in Denmark (Anon 1993). Briefly, 5 ml sludge was incubated with 5 ml of 17·8 mmol l−1 (NH4)2SO4, 80 mmol l−1 NaHCO3 and oxygen in excess at 20°C for 1 h on a rotary shaker. The production rate of NOinline image–N + NOinline image–N was measured photometrically using flow injection analysis. MLSS and MLVSS concentrations were determined in all four subsamples. The actual nitrification rates in each experimental train at the WWTP were calculated through nitrogen mass-balances in the composite samples each sampling day. The measured flow, MLVSS, and influent and effluent total nitrogen and ammonium concentrations for each sampling day were used in the calculations. The average ammonia-assimilation values were assumed to be 2 mg N l−1, which was based on experience from the staff at the Henriksdal WWTP. It was also assumed that nitrification occurred in the zones with DO levels higher than 1 mg l−1.

FISH and confocal laser-scanning microscopy

As all samples were collected, centrifuged and frozen as pellets immediately after sampling, no fresh material was available for FISH. To check whether the frozen pellets could be used for FISH, a separate sample of fresh sludge was collected. One portion of the fresh sample was immediately fixed in 4% paraformaldehyde, and another portion was treated exactly like the samples in this study (i.e. centrifuged and frozen after the supernatant had been discarded). After 14 d incubation in the freezer, the frozen pellets were thawed and fixed as described below. The two differently treated samples were then hybridized with the probe EUB338, and co-stained with SYTO9 (n = 7). Random pictures were collected and the relative abundance of EUB338 to SYTO9 was calculated for the two treatments (see below for details). As no significant difference (paired t-test, P = 0·76) between the frozen (60·4 ± 9.1%) and fresh samples (58·1 ± 14.5%) or any visible damage to the frozen cells could be noted in the microscope, FISH was applied to the frozen pellets in this study as described in the following.

Frozen pellets of activated sludge were thawed and fixed in 4% paraformaldehyde, washed with phosphate-buffered saline (PBS) and stored in PBS-ethanol (1 : 1) at −20°C until further use. In situ hybridization with fluorescently labelled rRNA-targeted probes was performed at 46°C for 2 h as described by Manz et al. (1992). Target sequences, hybridization conditions, and references for the probes used in this study are listed in Table 2. All fluorescent probes and unlabelled competitor probes were obtained from Thermo Hybaid (Interactiva Division, Ulm, Germany). Fluorescent probes were 5′-labelled with one of the sulfoindocyanine dyes indocarbocyanine (Cy3) or indodicarbocyanine (Cy5). After the hybridization, all samples were additionally stained for 10 min with a 1 μmol l−1 solution of the nucleic acid stain SYTO9 (LIVE/DEAD®BacLightTM; Molecular Probes Inc., Eugene, OR, USA). To prevent fluorochrome bleaching, all slides were mounted in Citifluor AF1 (Citifluor Ltd, London, UK).

Table 2.  FISH probes targeting 16S rRNA and the hybridization conditions that were used in this study
Probe Sequence (5′–3′)Target site*SpecificityFormamide (%)†NaCl (mmol l−1)‡Reference
  1. *Escherichia coli 16S rRNA position (Brosius et al. 1981).

  2. †Percentage of formamide in the hybridization buffer.

  3. ‡Concentration of sodium chloride in the washing buffer.

  4. §Wrong sequence is given in the original reference, erratum in Appl Environ Microbiol (1997) 63, 815.

  5. ¶Formamide concentration according to Konuma et al. (2001).

  6. **Used together with an equimolar amount of unlabelled competitor oligonucleotide as indicated in the reference.

  7. ††Referred to as S-*-Nse-1472-a-A-18 in the reference.

  8. ‡‡Referred to as S-*-Ntspa-0712-a-A-21 in the reference.

Nso190CGATCCCCTGCTTTTCTCC189–207Ammonia-oxidizing beta-proteobacteria25¶159Mobarry et al. 1996
Nso1225CGCCATTGTATTACGTGTGA§1225–1244Ammonia-oxidizing beta-proteobacteria3580Mobarry et al. 1996§
Nsv443CCGTGACCGTTTCGTTCCG444–462Nitrosospira spp.30112Mobarry et al. 1996
Nsm156TATTAGCACATCTTTCGAT156–174Various Nitrosomonas spp. and Nitrosococcus mobilis5636Mobarry et al. 1996
NEU**CCCCTCTGCTGCACTCTA653–670Halophilic and halotolerant members of the genus Nitrosomonas4056Wagner et al. 1995
NOLI191**CGATCCCCCACTTTCCTC191–208Various members of the Nitrosomonas oligotropha lineage30112Gieseke et al. 2001
Nmo218CGGCCGCTCCAAAAGCAT218–235Nitrosomonas oligotropha lineage3580Gieseke et al. 2001
Cluster 6a192**CTTTCGATCCCCTACTTTCC192–211Nitrosomonas oligotropha lineage3580Adamczyck et al. 2003
NmIITTAAGACACGTTCCGATGTA120–139Nitrosomonas communis lineage25159Pommerening-Röser et al. 1996
NmIVTCTCACCTCTCAGCGAGCT1004–1023Nitrosomonas cryotolerans lineage3580Pommerening-Röser et al. 1996
NmVTCCTCAGAGACTACGCGG174–191Nitrosococcus mobilis lineage3580Pommerening-Röser et al. 1996
Nse1472††ACCCCAGTCATGACCCCC1472–1489Nitrosomonas europaea5028Juretschko et al. 1998
Ntspa712‡‡CGCCTTCGCCACCGGCCTTCC712–732Phylum Nitrospira3580Daims et al. 2000

Confocal images of FISH- and SYTO9-stained samples were collected with a Bio-Rad Radiance 2000 MP microscope (Bio-Rad, Hemel Hempstead, UK) using the Ar Kr/Ar (488 nm), GHe/Ne (543 nm) and Red Diode (638 nm) lasers and the bundled software LaserSharp 2000. Images for quantification were collected as 8-bit images of 512 × 512 pixels (resolution: 1·65 pixels μm−1) using a Nikon Plan Fluor (Nikon Corporation, Tokyo, Japan) 40×/1·40 oil objective and Kalman filtration (n = 3). Images for micrographs and calibration of cell sizes were collected as 8-bit images of 1024 × 1024 pixels (resolution: 4·97 pixels μm−1) using a Nikon Plan Apo 60×/1·30 oil IR objective and Kalman filtration (n = 7).

Quantification of total cell biovolume

From each sample, a given volume of sludge (5 μl per well) was spotted onto microscope slides with wells (Ø 6 mm), and care was taken to make sure that the sludge was evenly distributed in each well. The total biovolume of each sample was measured by randomly collecting 25 full z-stacks (step size = 1 μm) in all SYTO9-stained samples. Five separately stained subsamples were analysed for each process and sampling date. The z-stacks were exported as a series of TIFF files, and analysed for biovolume (μm3 μm−2) in MATLAB using the program COMSTAT (Heydorn et al. 2000). A graph of the accumulated mean was generated for each sample to make sure that the 25 stacks collected were enough to get a stabilized accumulative mean (with stabilized standard deviation) of the biovolume. In most cases 15–20 stacks would have been enough, but all 25 were collected. As the size of the well and the volume of sludge spotted onto each well was known, the biovolume in the original samples could then be calculated from the measured fluorescing biovolume.

Quantification by FISH

The signal of specific probes in each sample (n = 5) was measured by collecting the probe signal along with the SYTO9 signal in 50 microscope fields, randomly selected in both xy and xz direction. The sections were exported as TIFF files, and analysed using the signal area function in COMSTAT (see above). The relative area of probe signal to SYTO9 signal was calculated and used to determine the corresponding probe biovolume from the total biovolume. Finally, the probe biovolume was divided by a specific cell volume (see below) to estimate the cell numbers.

To measure the mean AOB cell size, highly magnified z-stacks (60× objective and 4× digital zoom, optical resolution d = 0·20 μm) were collected from clusters of AOB targeted with the probes Nso190 and 6a192. The images were exported as series of TIFF files and analysed using the measure tool in Adobe Photoshop 7·0 (Adobe Systems Inc., San Jose, CA, USA). By comparing each section with the adjoining sections, only cells that appeared to be viewed from the side were measured. For each probe a mean cell volume was calculated from the length and width of 100 distinct cells for each probe, randomly picked from several different aggregates in both processes.

DNA extraction from AOB cultures and activated sludge

Liquid, pure cultures of AOB were grown at room temperature in the dark in an ammonium-containing medium (pH 7·5) as described by MacDonald and Spokes (1980) and Donaldson and Henderson (1989). Cells from Nitrosomonas europaea (NCIMB 11850) and Nitrosospira multiformis (NCIMB 11849) cultures were harvested by centrifugation and DNA extractions were performed using a QIAamp Tissue Kit (QIAGEN GmbH, Hilden, Germany) according to the protocol supplied by the manufacturer. DNA from Nitrosomonas eutropha and Nitrosomonas marina was kindly provided by Dr John R. Stephen at the Netherlands Institute of Ecology, the Netherlands.

DNA was extracted from activated sludge with the FastDNATM SPIN kit for Soil (Qbiogene, Inc., Carlsbad, CA, USA) weekly (Wednesdays). The frozen pellet from 10 ml activated sludge was suspended in 2·93 ml phosphate buffer from the extraction kit. The suspension was divided in three extraction tubes and DNA was extracted and purified according to the manufacturer's instructions. The DNA concentration was determined with PicoGreen® (Molecular Probes) on a FLUOstar spectrometer (BMG LabTechnologies, GmbH, Offenburg, Germany). The manufacturer's instructions were followed except for a threefold increase of the PicoGreen® concentration.

PCR amplification of AOB 16S rRNA genes and DGGE analysis

A nested PCR approach was used to amplify partial 16S rDNA of AOB. In the first PCR, the primers EC9-26 (GAGTTTGATCMTGGCTCA, modified from ‘fD2’ by Weisburg et al. 1991) and P13B (GTGTACTAGGCCCGGGAACGTATTC, Tiveljung et al. 1995) were employed to amplify a 1·4-kbp fragment of bacterial 16S rDNA. The PCR was performed on a PTC-100TM thermal cycler (MJ Research, Inc., Waltham, MA, USA) under the following conditions: 2 min at 94°C followed by 30 cycles of 30 s at 94°C, 30 s at 45°C, 1 min at 72°C, and a final extension of 10 min at 72°C. PCR amplification of the fragments was carried out in 25-μl reactions in thin-walled Eppendorf tubes containing 10–50 ng template DNA, 1·25 U Taq polymerase (Amersham Biosciences, Uppsala, Sweden) with the manufacturer's reaction buffer at 1·5 mmol l−1 MgCl2, 50 μmol l−1 of each primer, and 200 μmol l−1 of each dNTP. A 50-fold dilution of the amplicons was used as templates in the second PCR for amplification of partial rDNA sequences from AOB belonging to the β-subdivision of Proteobacteria with the primers CTO189fA/B-GC and CTO189fC-GC (CGCCGCGCGGCGGGCGGGGCGGGGGCACGGGGGGAGRAAAGCAGGGGATCG and CGCCCGCCGCGCGGCGGGCGGGGCGGGGGCACGGGGGGAGGAAAGTAGGGGATCG) and CTO654r (CTAGCYTTGTAGTTTCAAACGC) developed by Kowalchuk et al. (1997). The reaction mixture was composed as stated above. The second PCR was run with an initial denaturation of the template DNA at 94°C for 3 min followed by 35 cycles of 30 s at 94°C, 30 s at 57°C, and 45 s at 72°C. The reaction was completed after 10 min at 72°C. All primers were purchased from Invitrogen AB (Lidingö, Sweden).

The partial 16S rDNA amplified from samples taken in weeks 5, 6, 8, 10, 14, 15, 22, 24, and 26 was analysed by DGGE on a DCode universal mutation detection system (Bio-Rad Laboratories, Inc., Hercules, CA, USA). 160 × 160 mm Polyacrylamide gels were cast using a 170-9042 Model 475 Gradient Delivery System (Bio-Rad Laboratories). The gels consisted of 7% acrylamide : bisacrylamide (37·5 : 1) and a 30–60% denaturant gradient (100% denaturant was defined as 7 mol l−1 urea and 40% formamide). A quantity of 15 μl of the respective PCR product was loaded on the gels, which were run at 60°C and 130 V for 13 h in 1X TAE (40 mmol l−1 Tris–HCl, 20 mmol l−1 acetic acid, 1 mmol l−1 EDTA, pH 8·3). The DNA fragments were visualized with UV translumination after ethidium bromide staining and the visible bands were cut out from the gels, placed in 160 μl of distilled water, and stored at −70°C until sequencing.

Nucleotide sequencing and sequence analysis

To elute the DNA from the polyacrylamide gels the samples were thawed for 1 h at room temperature, frozen at −70°C for 1 h, and then thawed again at 8°C for 12 h. The eluted DNA was used as template in a PCR-amplification with the CTO-primers without GC-clamp at a concentration of 5 μmol l−1, but otherwise as described above. Prior to sequencing, 45 μl of PCR product was purified with a MicroSpin S-400 HR column (Amersham Pharmacia Biotech, Uppsala, Sweden). The DNA fragments were sequenced in 11 μl reactions by using the DYEnamic ET Terminator Cycle Sequencing kit (Amersham Pharmacia Biotech). The CTO primers without GC-clamp were also used as sequencing primers and the products were separated on an ABI PRISM 377 (Applied Biosystems, Foster City, CA, USA) automated sequencer. Nucleotide sequences were aligned using the CLUSTAL W software (http://www.ebi.ac.uk/clustalw/) and the sequences were compared with the GenBank database using BLASTn (http://www.ncbi.nlm.nih.gov/BLAST/). A phylogenetic analysis was performed with the software TREECON (Van de Peer and De Wachter 1994) applying Jukes and Cantor correction and the neighbour-joining method (Saitou and Nei 1987) in the program.

Nucleotide sequence accession number

The partial 16S rDNA sequence Nitrosomonas sp. DGGE band B obtained has been deposited in GenBank under accession number AY525028.

Statistical analysis

All quantitative data from the two processes were compared statistically using paired t-tests at a confidence level of 95%.

Results

Functional performance of activated sludge processes

There was no difference regarding process performance and nitrogen removal in the two different systems (Table  1). During the period both had nearly 80% nitrogen removal efficiency and the average effluent NHinline image–N and NOinline image–N concentration was below 1 and 6 mg l−1 respectively. The potential nitrification, indicating the maximum capacity of the process, exceeded the actual rates on all sampling occasions in both systems (Fig. 1). Both the actual and potential nitrification rates were significantly higher in the short SRT process than in the long SRT process, although the potential rates did not differ as much between the two process configurations as the actual rates. The difference between the processes in potential rate was more pronounced at the end of the experimental period while the difference in actual rates was already noticeable 1 week after the suspended solids started to accumulate.

Figure 1.

Actual (——) and potential (- - - -) nitrification rates in samples from the process with a long (•) and a short (bsl00001) solids retention time. Error bars indicate the standard deviation (n = 4)

Composition of AOB communities as determined by DGGE

The PCR amplification with the AOB-specific CTO primers did not consistently yield detectable PCR products. It was therefore necessary to undertake a nested PCR approach with the universal bacterial primers followed by the CTO primers. The amplified partial 16S rDNA sequences were then analysed by DGGE and sequenced to determine the composition of the ammonia oxidizer communities in the two different activated sludge processes. There was no difference in the community composition determined by DGGE banding pattern either between the two operational strategies or during the experimental period (Fig. 2 and data not shown). Moreover, the triplicate DNA extractions from every sample resulted in the same patterns. In Fig. 2, DGGE patterns of representative samples from weeks 14 and 22 are shown. A pattern of one distinct and one less distinct band was detected in all samples and this pattern was reproducible. In the sludge sample collected during week 8 from the process with short SRT another weak band was present in all three replicates, located below the dominant band (not shown). None of the bands in the samples co-migrated with the Nitrosomonas sp. and Nitrosospira sp. reference strains.

Figure 2.

Negative DGGE image of AOB in samples from the process with (I) long solids retention time and (II) short solids retention time collected during weeks 14 and 22. Arrows indicate the two possible bands that were visible at levels A and B in all samples. Lanes 1–4 show the positions of the four reference strains: 1, Nitrosomonas europaea NCIMB 11850; 2, Nitrosomonas marina; 3, Nitrosomonas eutropha; 4, Nitrosomonas europaea NCIMB 11850/Nitrosospira multiformis NCIMB 11849. The gels were stained with ethidium bromide

Unambiguous identification of ammonia oxidizer populations cannot be based on migration patterns alone. The identity of the two visible bands from levels A and B, as well as the band appearing in week 8, was therefore confirmed by direct sequencing of the excised and re-amplified 16S rDNA DGGE bands. Four bands from each level were randomly picked from each process and sequenced and they all contained the same sequence. The phylogenetic analysis showed that the sequence, denoted DGGE band B, related to cluster 6a of the genus Nitrosomonas. Within this cluster the sequence had the closest relationship to Nitrosomonas oligotropha-like strains.

Composition of AOB community as determined by FISH analysis

Fluorescence in situ hybridization was performed with all probes listed in Table 2. Positive results were obtained with Nso1225, Nso190, Nsm156, Nse1472 and 6a192 in all samples from both processes. Simultaneous hybridization of these probes revealed the existence of two main groups of AOB (Fig. 3). Group I was targeted by both Nso1225 and Nso190. The major part of group I, denoted Ia, was also targeted by the probes Nsm156 and Nse1472, showing that they belonged to the Nitrosomonas europaea/Nitrosococcus mobilis lineage. The remaining bacteria of group I were only targeted by Nso1225 and Nso190, which indicated the presence of another unidentified, although very small, population of AOB (Ib) within this group. Group II consisted only of bacteria belonging to the Nitrosomonas oligotropha lineage within cluster 6a, as shown by the positive signal from probe 6a192. The latter is consistent with the findings from the DGGE analysis. The bacteria of this group did not hybridize at all with Nso190, although they target overlapping regions. Most of the bacteria within group II were targeted only by probe 6a192 (IIa), but within group II there was also another smaller population (IIb) targeted both by 6a192 and Nso1225 (Fig. 3). To conclude, all members of group I could be identified by a single probe (Nso190) and all members of group II by probe 6a192.

Figure 3.

Confocal laser scanning micrographs showing three of the AOB populations detected by FISH in activated sludge samples from the process with long solids retention time. Scale bar = 10 μm. (a) Large aggregate of the most abundant AOB population (Ia), belonging to the Nitrosomonas europaea/Nitrosococcus mobilis lineage, hybridized with Nso190. (b) Two aggregates of the dominating Nitrosomonas oligotropha-like population that was detected only by 6a192. (c) An aggregate containing two Nitrosomonas oligotropha-like populations. The most abundant population (IIa) appears as red (targeted by 6a192), and the less abundant population (IIb) as yellow (targeted by 6a192 and Nso1225). The latter appeared exclusively in mixed aggregates

Group I was significantly more abundant than group II in the samples from weeks 6, 8 and 26 in the process with short SRT, and in the samples from weeks 14 and 26 in the process with long SRT (Fig. 4). No significant difference was found between the two groups in all other samples but there was a tendency for group I to be the most numerous in all samples. Group I constituted between 1·5 and 3·6% of the total amount of bacteria. Group II, was less abundant and constituted 0·3–1·9% of the total amount of bacteria in both processes. No signal was obtained with any of the other probes, which includes both Nmo218 and NOLI191 that target bacteria within the Nitrosomonas oligotropha lineage and Nsv443 that target AOB of the genus Nitrosospira. To get a rough estimate of the complete nitrifying population, a few samples from both processes were also hybridized with the probe Ntspa712, targeting nitrite-oxidizing bacteria of the phylum Nitrospira. The amount of signal from this probe was 5–7% of the total amount of bacteria in all samples and no difference between the two processes was observed (data not shown).

Figure 4.

Total bacterial biomass stained by SYTO9 (○) and number of group I (•) and group II (bsl00001) AOB cells determined with probes Nso190 and Cluster 6a192 respectively using FISH analysis. Error bars indicate the standard deviation. Process with (a) long and (b) short solids retention time

Morphology and size of AOB cells and aggregates

A small difference in shape was noted between the two AOB groups detected by FISH. Both groups consisted of short rod-shaped bacteria of roughly the same size, but the cells of group I were longer and more slender than the cells of group II. The size and morphology of all observed AOB cells were in accordance with that reported in the literature for isolated AOB cells of the genus Nitrosomonas and especially Nitrosomonas europaea, Nitrosomonas oligotropha and Nitrosomonas ureae (Koops et al. 2003). Almost all of the AOB cells were associated in aggregates. No difference in shape or cell size was noted between the two operational strategies, but the cells of group I generally appeared in larger aggregates (mean size 15–30 μm; maximum size >100 μm; Fig. 3a) than the cells of group II (mean size 5–10 μm; maximum size 35–40 μm; Fig. 3b,c). The less abundant population of group II (IIb) always appeared in mixed aggregates together with the dominant population IIa, whereas the latter also appeared in separate aggregates (Fig. 3). The two populations of group I exclusively appeared in separate aggregates.

Size of ammonia-oxidizing community and total bacterial biomass

As mentioned above, the most universal AOB probe Nso1225 did not target all AOB in the samples, and could therefore not be used to calculate the total amount of AOB alone. However, the hybridization results showed that all AOB that could be detected always hybridized with either Nso190 (group I) or 6a192 (group II), and that there was no overlap between the two. The total AOB numbers were therefore calculated by adding the hybridization signals from the probes Nso190 and 6a192. The AOB numbers were not significantly different between the two operational strategies during the period, but a significant increase in AOB numbers from the start to the end of the period could be seen in both operational strategies (Fig. 4). In the process operated with long SRT, the amount of group I AOB increased more than twofold from weeks 6 to 8, and then again from weeks 22 to 26 resulting in a population size five times larger than during week 6. Group II showed no increase during the period, with the exception of the sample from week 22, which was significantly higher than in week 6. In the short SRT process, the amount of group I AOB initially increased in a similar manner, with a twofold increase from weeks 6 to 14, but no increase could be seen at the end of the period. In contrast, a significant increase of group II could be seen, which increased four times between weeks 8 and 14.

The amount of bacterial biomass, which was calculated as the volume of SYTO9-positive cells per gram MLVSS, did not differ between the two operation modes at any time (Fig. 4). However, both strategies showed a significant two- to threefold increase of biomass in the beginning of the period, from weeks 6 to 8. Another significant increase followed from weeks 8 to 22, resulting in a threefold total increase of biomass from weeks 6 to 26 in the process with long SRT. In the one with short SRT, the highest biomass value was reached between weeks 14 and 22, followed by a significant decrease of the biomass by week 26. This resulted in a less than twofold total increase of biomass from weeks 6 to 26.

Discussion

Nitrification activity and abundance of AOB

Both modes of operation, i.e. long or short SRT, resulted in efficient nitrogen removal processes with approximately 80% nitrogen removal and low NHinline image–N discharges. As expected, the process with short SRT had a more biologically active sludge as was seen from the nitrification rates expressed on an MLVSS basis. The difference in the actual rates, based on nitrogen mass balances, was immediate when the operational mode in one of the trains was shifted to increase the SRT at the start of the experiment. The actual rate in the short SRT process gradually increased over the period while the other process fluctuated at a low and constant rate (Fig. 1). The ratio between the actual nitrification rates of the two processes was 0·69, which is the same as the ratio between their SRTs (10·7/15·6 = 0·69). This seems to confirm the relation between the SRT and nitrification activity. The lower DO levels in the process with long SRT may have suppressed the activity and the higher organic matter content could have resulted in nitrifying bacteria competing with heterotrophs for ammonia (Hanaki et al. 1990). Nevertheless, the ammonium disappeared efficiently in both processes and the difference in DO probably did not affect the total nitrification in the plant.

The potential rates, which were not affected by the environmental conditions at the time of sampling, demonstrate the maximum capacity of the AOB communities in the treatment plant. The potential rate was only slightly higher in the process run with a short SRT than in the one with a long SRT but a large difference between the operational modes was observed during the last 5–6 weeks (Fig. 1). We interpret the different potential rates that were seen at the end of the experimental period as an indication of physiological differences between the AOB communities in the two processes. As our data showed that that the AOB numbers were similar in the two processes on all sampling occasions (Fig. 4) and that no detectable community shifts had occurred, we suggest that the effect of a longer SRT on the activity mainly was because of physiological alterations of the original AOB populations.

The AOB numbers and the total bacterial biomass increased in the two systems during the experimental period. This was probably an effect of the temperature increase during the season. FISH results showed that the ammonia oxidizers constituted 3–5% of the total biomass in the two processes. The proportion of AOB increased to some extent in both of them, but it was only statistically significant in the long SRT process. This was probably also a temperature effect. The proportion of nitrifying bacteria, which include both AOB and nitrite-oxidizing bacteria, in activated sludge is assumed to be 2–5% (Randall et al. 1992) and the latest ‘Activated Sludge Model’ estimated the nitrifying bacteria to be 2–3% of the total biomass (Koch et al. 2001). Taking into account the roughly estimated value of 5–7% nitrite-oxidizing bacteria in this study, our results suggest a somewhat higher proportion of nitrifying bacteria (10%). However, AOB values ranging from 0·0033% (Dionisi et al. 2002a) to 15% (Wagner et al. 1995) have been reported, which may reflect differences in the relative ammonia-oxidizer community size between the treatment plants.

As the increased activity in the process with short SRT, as compared with the one with long SRT, could not be explained by a difference in AOB numbers, the specific activity (k0) was calculated using the actual and potential nitrification rates presented in Fig. 1 (Table 3). The results show that in general the short SRT process had a higher specific activity, except for in week 14. The specific activity was highest in the beginning of the experiment in both processes, due to the low AOB numbers, and eventually decreased. However, as the AOB numbers have a high standard deviation, the differences in specific activity must be regarded as tendencies only. Moreover, the k0 estimations may be biased by FISH detection of dormant AOB cells, as cells can retain high ribosome content during periods of low physiological activity (Flärdh et al. 1992; Wagner et al. 1995; Morgenroth et al. 2000). The specific activity of pure cultures of AOB is in the range of 1·2–23 fmol cell−1 h−1 and species belonging to the genus Nitrosomonas have higher k0-values than those within the genus Nitrosospira (Belser 1979; Laanbroek and Gerards 1993). Daims et al. (2001) determined a slightly lower in situ activity than we did when FISH was used for AOB enumeration and Wagner et al. (1995) reported values as low as 0·22 fmol N cell−1 h−1 with the same method. Harms et al. (2003), who used a real-time PCR TaqMan assay to enumerate AOB, reported a mean k0 value of 7·7 fmol N cell−1 h−1 and calculations based on enumeration with cPCR of Nitrosomonas oligotropha-like AOB were in the range of 3·5–56 fmol cell−1 h−1 in the same samples. In activated sludge processes the k0 values are subject to fluctuating environmental conditions and can therefore vary widely.

Table 3.  Specific activity (k0) of AOB cells determined by quantification of FISH signal from the Nso190 and Cluster 6a192 probe in samples from the process with a long SRT (I) and a short SRT (II)
WeekActual specific activity* (fmol N cell−1 h−1)Potential specific activity† (fmol N cell−1 h−1)
IIIIII
  1. *Actual nitrification rates from nitrogen mass balances.

  2. †Potential nitrification rates from batch experiments.

 613153134
 8591418
1474179
22571415
263558

Only a few studies on AOB enumeration over longer periods in activated sludge systems have been reported and all of these were performed with PCR-based methods (Dionisi et al. 2002a,b; Harms et al. 2003). Traditionally, FISH has mostly been used for relative measurements of bacterial populations in biofilms or activated sludge samples, as the complex distribution of the cells in these environments makes the cells hard to quantify. One way to circumvent this problem is to spike the samples with a known amount of Escherichia coli cells, and use the signal from these cells to calculate the absolute numbers of other cells in the sample (Daims et al. 2001). However, this method did not work in the present study, as the E. coli cells would not distribute evenly in the samples due to the large, impenetrable sludge flocs. Instead we developed another way to quantify the FISH signal based on measurements of biovolume, relative area and specific cell volumes. The numbers of AOB obtained with this method (Fig. 3) were within the reasonable range that has previously been determined in activated sludge processes (Wagner et al. 1995; Kowalchuk et al. 1999; Daims et al. 2001; Dionisi et al. 2002a; Harms et al. 2003).

We used a large set of probes to target the AOB population (Table 2). Generally, the probe Nso1225 is considered to be the most universal, targeting almost all beta-proteobacterial AOB. Nevertheless, in this study we were not able to detect all AOB present with Nso1225. In fact, none of the probes we used initially could detect any bacteria belonging to the Nitrosomonas oligotropha lineage, which had been identified by DGGE. No signal was obtained from the probes NOLI191 and Nmo218, both targeting bacteria within this lineage. However, after the recent publication of the probe 6a192 (Adamczyck et al. 2003), the lost population was finally detected. Co-staining of the samples with both Nso1225 and 6a192 showed that Nso1225 was able to detect a very small population of bacteria within the Nitrosomonas oligotropha lineage (Fig. 3). This result strongly emphasizes the need for further development of FISH probes for AOB, as the most commonly used ones Nso190 and Nso1225 in some cases clearly underestimate the total number of AOB, thereby giving a biased picture of the microbial ecology of the activated sludge systems.

AOB community analysis

The community analysis with both the in situ hybridization and the PCR-dependent approach demonstrated that populations related to Nitrosomonas were responsible for ammonia oxidation at the Henriksdal WWTP and that Nitrosospira-like AOB were not detected. All DGGE analyses of PCR-amplified 16S rDNA from AOB resulted in one AOB sequence. The phylogenetic analysis showed that it falls into cluster 6a where the Nitrosomonas oligotropha and Nitrosomonas ureae lineages are found and that it is closely related to Nitrosomonas oligotropha Nm 45 (Koops et al. 1991). The nearest neighbour to the 410 bp sequence was an uncultured Nitrosomonas sp. (Clone 26Ft, GenBank acc. no. AF527015) that was recently found in a wastewater treatment reactor (Rowan et al. 2003). The population detected by DGGE was also detected by the FISH-probe 6a192, which targets bacteria within the Nitrosomonas oligotropha lineage. As a small part of the 6a192-positive cells also hybridized with Nso1225, there were probably at least two different populations present within the Nitrosomonas oligotropha lineage (IIa and IIb). In addition, another abundant AOB group, denoted Ia was identified by FISH with the probes Nse1472 and Nsm156, which suggests that at least one AOB population belonging to the Nitrosomonas europaea/Nitrosococcus mobilis lineage (cluster 7) was present. This population could not be detected by DGGE. Finally, the FISH results detected the presence of other unidentified members of the genus Nitrosomonas, here referred to as Ib, in the sludge samples, as parts of the community that hybridized to the probes Nso190 and Nso1225 did not hybridize with any of the more specific probes. The CTO primers that were used for PCR are not universal for all beta-proteobacterial AOB and they have mismatches with some sequences in cluster 7, which holds the Nitrosomonas europaea/Nitrosococcus mobilis lineage identified by FISH in this study (Purkhold et al. 2000). This could explain the differences in the results from FISH and DGGE analysis. In addition, the FISH probes NOLI191 and Nmo218 that target the Nitrosomonas oligotropha lineage have mismatches with some of the isolated bacteria from cluster 6a (Koops et al. 2003). This is probably the reason why these probes could not detect the populations that were found by the 6a192 probe and by DGGE analysis.

The finding that members of the genus Nitrosomonas dominated the activated sludge processes was not surprising as they are known to thrive in nitrogen-rich environments such as activated sludge processes. Our results are in agreement with the findings of Dionisi et al. (2002a) who detected nothing but members of Nitrosomonas, with a dominance of Nitrosomonas oligotropha, in a 1-year study of a municipal WWTP. Nevertheless, almost all recognized lineages of beta-proteobacterial AOB can be found in WWTPs and from surveys in wastewater treatment processes it has been suggested that different plants support different populations. Members of the Nitrosomonas europaea/Nitrosococcus mobilis and the Nitrosomonas marina clusters have been most frequently detected (Purkhold et al. 2000; Wagner et al. 2002). These more recently published results are all in agreement with older findings (e.g. Mobarry et al. 1996; Wagner et al. 1996).

The relatively long SRT that both processes were operated with in this study most likely selected for two AOB populations that dominated in both processes, although two others were found in low abundance. Despite the different environmental conditions for the AOB in the two processes, the total selective pressure in the two experimental trains was not strong enough to induce a population shift. The number of different populations can differ significantly between WWTPs and sometimes plants even harbor AOB monocultures (Juretschko et al. 1998; Purkhold et al. 2000). WWTPs with long SRTs, such as in this study, have been reported to have less variable biomass characteristics (Henze et al. 2002). Daims et al. (2001) suggested that the level of AOB diversity relates to the operational stability of the process and there have been indications that plants with a low diversity of a given functional group are more prone to process failure than plants showing a higher diversity of the same bacterial group (Wagner et al. 2002). A selection for AOB populations that are more robust to environmental disturbances can also reduce the vulnerability of the plant, and this would be advantageous for process control. However, diversity determined from 16S rRNA gene analysis does not always reflect true genetic diversity. Hitherto unidentified AOB may be present in the system although they are not detected and other genes, which affect AOB properties and ecosystem functioning, can probably differ between closely related AOB appearing identical in partial 16S rRNA sequences. Jaspers and Overmann (2004) recently showed that identical 16SrRNA gene sequences were found in Brevundimonas alba with highly divergent genomes and ecophysiologies. A better understanding of the links between the true genetic diversity of the functionally important AOB and the stability of the process they carry out is necessary to be able to protect the plants from process deterioration.

Concluding remarks

The MLSS concentration and SRT affected the nitrification activity but not the numbers or the community composition of AOB. Hence, long-term application of a specific operational strategy may result in a change in the physiological state of the bacterial populations, without inducing a change in community structure. Our results support the concept of sludge population optimization, which Yuan and Blackall (2002) argued should be an aim for the design and operation of a treatment plant.

The FISH results in this study clearly indicate that neither of the commonly used probes Nso190 and Nso1225 can be used for correct enumeration of total AOB numbers in activated sludge when certain populations of the Nitrosomonas oligotropha lineage are present in the process. Reliable numbers of AOB are important for monitoring and modelling of activated sludge processes. Therefore, more universal AOB FISH probes and other quantitative methods must be further developed.

Acknowledgements

Financial support for this work was provided by VA-FORSK, Sweden, Stockholm Water Ltd, Sweden, The Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (Formas), contract no. 25·2/2001–2677 and The Swedish Research Council, contract no. 621-2001-1664. We thank Maria Rothman and the rest of the staff at the Henriksdal WWTP, especially Michael Medoc, for running the experimental lines at the plant as well as providing us with samples and process data. Thanks are also extended to Maria Erikson and Annika Åberg for help with DNA extractions and to Anna Hermansson who cultivated the AOB and extracted DNA from them.

Ancillary