Introduction

Many members of the betaproteobacterial order Rhodocyclales are widespread and abundant in wastewater treatment systems (Juretschko et al., 2002; Etchebehere et al., 2003; Tsuneda et al., 2005; Liu et al., 2006), and have diverse physiological capabilities that are beneficial for the degradation and transformation of pollutants, such as nitrogen, phosphorous, and aromatic compounds (Loy et al., 2005). For example, yet uncultivated ‘Candidatus Accumulibacter’ species are involved in phosphorus removal from wastewater (Hesselmann et al., 1999; Crocetti et al., 2000). In addition, anaerobic nitrogen removal by denitrification in full- and laboratory-scale wastewater treatment systems is predominantly catalyzed by ‘Candidatus Accumulibacter’-, Dechloromonas-, Thauera- and Azoarcus-related bacteria, as recently shown by stable isotope probing and fluorescence in situ hybridization coupled with microautoradiography (MAR-FISH) studies (Kong et al., 2004; Ginige et al., 2005; Thomsen et al., 2007; Morgan-Sagastume et al., 2008). Some members of the Rhodocyclales are endowed with additional catabolic capabilities of potential benefit for wastewater treatment. Certain Thauera and Dechloromonas strains degrade oil derivates such as toluene (Shinoda et al., 2004; Chakraborty et al., 2005), and might thus contribute to reducing the ecological burden of such aromatic waste compounds. However, niche partitioning of different, co-occurring Rhodocyclales species in activated sludge and the implication of this diversity for the functioning and stability of the wastewater treatment process is still poorly understood.

The vast majority of microorganisms that are responsible for central functions in many natural and biotechnological ecosystems (such as wastewater treatment plants) have not yet been cultured. Radioactive- and stable-isotope-substrate-mediated labeling methods for linking the identity of an uncultivated microorganism with its physiological abilities have thus become indispensable for our understanding of microbial ecology and ecosystem function. Stable isotope probing of nucleic acids (Radajewski et al., 2000; Manefield et al., 2002) and MAR-FISH (Lee et al., 1999; Ouverney and Fuhrman, 1999) represent the core of our methods arsenal for structure–function analysis of complex microbial communities (reviewed in Friedrich, 2006; Madsen, 2006; Wagner et al., 2006; Whiteley et al., 2006; Neufeld et al., 2007). Additional methods were recently developed on the basis of advanced secondary ion mass spectrometry (SIMS) and vibrational spectroscopy techniques and their adaptation for measurements in biological systems. The combination of multi-isotope imaging mass spectrometry (also called nanoSIMS) (Behrens et al., 2008; Li et al., 2008; Musat et al., 2008) or Raman microspectroscopy (Huang et al., 2007) with FISH offers unprecedented opportunities for tracing substrate incorporation at the (sub)cellular level (Wagner, 2009). Each of these isotope techniques has its specific advantages and disadvantages (Wagner et al., 2006; Neufeld et al., 2007), but all are limited by the number of samples and/or microbial groups that can be analyzed with justifiable effort. DNA microarrays containing hundreds to thousands of nucleic acid probes are suited for high-throughput analyses. The application of rRNA-targeted oligonucleotide microarrays (PhyloChips) for microbial community analysis (Liu et al., 2001; Loy et al., 2002; Wilson et al., 2002; Zhou, 2003; Bodrossy and Sessitsch, 2004; Lehner et al., 2005; Loy and Bodrossy, 2006; Palmer et al., 2006) has propelled the development of the so-called isotope array technique, which allows the parallel identification and functional screening of numerous phylogenetic groups of microorganisms for their ability to consume a radioactive substrate (Adamczyk et al., 2003). In a first proof-of-principle study, a manually spotted prototype microarray consisting of six rRNA-targeted oligonucleotide probes was employed for isotope array surveys of ammonia-oxidizing bacteria in activated sludge samples (Adamczyk et al., 2003). However, the applicability of PhyloChips containing larger probe sets and produced with automatic spotting devices for the detection of fluorescently and radioactively labeled native rRNA remains to be proven.

The main objectives of this study were to adapt the isotope array technique for PhyloChips containing a larger number of probes and to use it for insights into the substrate-consumption profiles of Rhodocyclales in activated sludge incubated under oxic and anoxic, nitrate-reducing conditions. For this purpose, a recently designed 16S rRNA-targeted microarray, consisting of 79 nested oligonucleotide probes, for most cultivated and uncultivated Rhodocyclales (RHC-PhyloChip) (Loy et al., 2005) was used. Microarray results were complemented by retrieval and phylogenetic analyses of 16S rRNA gene sequences, and by using existing and newly developed Rhodocyclales subgroup-specific probes for quantitative FISH and MAR-FISH analyses.

Materials and methods

Sampling of activated sludge

Activated sludge samples were collected in October 2004 at the municipal wastewater treatment plant, Aalborg East (Denmark, 100 000 person equivalents). Nitrogen and phosphorus in the plant are removed biologically by a Biodenipho configuration in which influent is mixed with activated sludge in anaerobic tanks before entering tanks with alternating denitrifying and nitrifying conditions. Samples for this study were collected in the denitrifying–nitrifying tanks. Samples for isotope labeling and DNA extraction were stored at 5 °C in the dark (for not more than 3 days) and at −20 °C, respectively. For FISH analyses, activated sludge was fixed with paraformaldehyde as described earlier (Daims et al., 2005).

DNA extraction

Genomic DNA was isolated from activated sludge samples by using the Power Soil DNA Kit (MoBio Laboratories Inc., Carlsbad, CA, USA) according to the manufacturer's instructions; with the exception that in the final step DNA was eluted from the column with 2 × 30 μl of double-distilled water.

PCR amplification of 16S rRNA genes

Amplification of 16S rRNA gene fragments from DNA of the activated sludge sample was performed by using the bacterial primer pairs 616V-630R and 616V-1492R, and the Rhodocyclales subgroup-selective primer pairs A, R or Z, together targeting almost all Rhodocyclales (Loy et al., 2005).

Isotope labeling of activated sludge

Activated sludge was diluted 1:1 with HEPES [N-(2-hydroxyethyl) piperazine-N′-2-ethanesulfonic acid]-buffered liquid mineral medium without ammonia (Aakra et al., 2000) to a final concentration of 2 g suspended solids per liter. Subsequently, the diluted sludge was flushed for 1 hour with CO2-free air to reduce dissolved inorganic carbon. This treatment decreased the concentration of dissolved inorganic carbon from 4 to 1 mM as measured by gas chromatography after acidification of subsamples (Chai et al., 2001). After aeration, 1 mM ammonium was added to all samples to avoid nitrogen limitation. Sludge samples were additionally incubated with 5 mg of allylthiourea per liter to inhibit assimilation of CO2 by autotrophic ammonia-oxidizing bacteria (Bedard and Knowles, 1989). For anoxic incubations, sludge samples were vacuum flushed with N2 immediately before and during the addition of radioactive compounds. Heterotrophic 14CO2 assimilation (Roslev et al., 2004; Hesselsoe et al., 2005) was exploited for radioactive labeling of biomass by adding NaH14CO3 (58 mCi mmol−1, Amersham Pharmacia Biotech, Buckinghamshire, UK) to 10 ml aliquots of diluted sludge in 60 ml glass vials (final concentration was in the range of 60–100 μCi ml−1). Subsequently, different unlabeled electron donors (5 mM propionate, 5 mM butyrate or 10 mg l−1 toluene) were added. Control incubations devoid of added electron donors were included to monitor and quantify background assimilation of 14CO2 in the absence of added organic substrates. Furthermore, additional activated sludge samples were incubated with 0.17 mM [14C]propionate {[1-14C] propionic acid sodium salt (56.7 mCi mmol−1), MP Biomedicals Inc., Solon, OH, USA} (final concentration 10 μCi ml−1) and with 2 mM unlabeled propionate to avoid substrate limitation during the incubation (ratio of 14C-labeled to unlabeled propionate was 1:11.8). All sludge slurries were exposed to atmospheric oxygen or 2 mM nitrate, sealed with gas-tight rubber stoppers, and incubated at room temperature in the dark for 4.5 h (propionate and butyrate) or 24 h (toluene).

Incorporation of 14C into biomass was monitored during the incubations. For this purpose, subsamples (20–100 μl aliquots) were harvested with a syringe, filtered through a 0.2-μm-pore-sized mixed-cellulose filter (Advantec MFS Inc., Pleasanton, CA, USA), acidified, dissolved in scintillation fluid, and quantified by liquid scintillation counting as described earlier (Hesselsoe et al., 2005).

After incubation, 90% (v/v) of the biomass from each vial was harvested by centrifugation, re-suspended in RNAlater (Ambion, Austin, TX, USA), cooled at 4 °C during the transport between laboratories, and finally stored at −20 °C until RNA extraction. The remaining 10% (v/v) of the biomass was fixed with paraformaldehyde for MAR-FISH analysis.

In vitro transcription of 16S rRNA

All experimental steps involving RNA were carried out under RNase-free conditions (Sambrook and Russell, 2001). Fluorescently labeled in vitro-transcribed RNA was used for optimizing RHC-PhyloChip hybridization conditions. For subsequent in vitro transcription, the 16S rRNA gene from activated sludge clone S3 (AF072918) was amplified by PCR using the general bacterial primers T3-BACT8F (5′-AATTAACCCTCACTAAAGGGAGAGTTTGATYMTGGCTC-3′, the T3 promoter sequence is depicted in bold) and 630R (Loy et al., 2002) and subsequently purified by using the QIAquick PCR purification kit (Qiagen, Wien, Austria). Reaction mixtures (30 μl) for in vitro transcription were prepared in T3 RNA polymerase transcription buffer (Fermentas, St Leon-Rot, Germany), contained 0.5–1 μg purified PCR product, 1 mM of each ribonucleotide, 20 U RNasin and 60 U of T3 RNA polymerase (Fermentas), and were incubated at 37 °C for 4 h. In the next step, the DNA template was digested by adding 3 μl of DNase I buffer and 3 U of RNase-free DNase I (Fermentas) to the reaction mixture and incubation at 37 °C for 15 min. The 16S rRNA was subsequently purified using the RNeasy MinElute Clean-up Kit (Qiagen). Quality and concentration of RNA were determined by agarose gel electrophoresis and spectrophotometric analysis (Adamczyk et al., 2003).

RNA extraction

RNA extraction from activated sludge stored in RNAlater was essentially carried out as described earlier (Adamczyk et al., 2003). The extracted RNA was additionally purified using filter cartridges and solutions that were supplied with the RiboPure-Bacteria or RNAqueous RNA extraction kits (Ambion). Agarose gel electrophoresis showed that the filter cartridge purification removed all small RNA fragments without substantial loss of intact small- and large-subunit rRNA. A total of 5–7 μg of RNA was obtained from 1 ml of undiluted activated sludge.

Spotting of DNA microarrays

Characteristics of oligonucleotide probes used for the production of the RHC-PhyloChips were described earlier (Loy et al., 2005) and can be viewed at probeBase (http://www.microbial-ecology.net/probebase/) (Loy et al., 2007), including the theoretical binding affinity of each probe to its perfectly matched rRNA, calculated as free energy ΔG12 (Yilmaz and Noguera, 2004). A 384-well flat bottom plate was prepared with 30 μl of 50 pmol μl−1 oligonucleotide solutions in 50% dimethyl sulfoxide. Probes were printed with a single pin on aldehyde group-coated CSS-100 glass slides (CEL Associates, Houston, TX, USA) with an OmniGrid 100 spotter (Genomics Solutions, Ann Arbor, MI, USA) at 50% relative humidity and 22 °C. RHC-PhyloChips for hybridization of labeled PCR products were spotted with a SMP3 pin (tip diameter 100 μm, TeleChem International Inc., Sunnyvale, CA, USA) in triplicate on each slide. In contrast, RHC-PhyloChips for isotope array analysis were spotted with a SMP15XB pin (tip diameter 600 μm, TeleChem) in duplicate on each slide. Spotted slides were further processed as described earlier (Loy et al., 2002). The first and last slides from each spotting run were hybridized with a fluorescently labeled 16S rRNA gene PCR product from Desulfovibrio halophilus (DSM 5663) to confirm the quality of the spotting. Slides were stored in the dark for <3 months before use.

Fluorescence labeling and microarray hybridization

Fluorescence labeling of PCR products and hybridization on microarrays were performed as outlined earlier (Loy et al., 2002). Per RHC-PhyloChip hybridization, two times 300 ng of purified 16S rRNA gene PCR product were random prime labeled in separate reactions and pooled before microarray hybridization to ameliorate labeling biases.

For fluorescence labeling of RNA, a direct chemical labeling procedure was performed using the CyScribe Direct mRNA labeling Kit (Amersham Bioscience, Uppsala, Sweden) as described earlier (Adamczyk et al., 2003). Only one type of dye (Cy3 or Cy5) was used for a coherent set of experiments. Before hybridization, the RNA was fragmented for 30 min at 60 °C in a solution containing 0.2 M Tris-HCl (pH 7.4) and 0.01 ZnSO4. After fragmentation, the solution was immediately chilled on ice, and 0.05 M EDTA was added to deactivate Zn2+ and stop the fragmentation process. RNasin was added to a final concentration of 1 U μl−1. Subsequently, fluorescently labeled and fragmented RNA was divided in aliquots and stored at −20 °C until hybridization. Microarray hybridization of RNA was performed as described earlier (Adamczyk et al., 2003). The quantity of RNA and percentage of formamide in the hybridization reaction was varied during optimization of the hybridization procedure.

Fluorescence scanning

Fluorescence images of the RHC-PhyloChip were recorded by scanning the slides with a GenePix Personal 4100A array scanner (Axon Instruments, Molecular Devices Corporation, Sunnyvale, CA, USA). The fluorescent signals were quantified by using the GenePix Pro software version 5.1 (Axon Instruments). Slides hybridized with fluorescently labeled PCR products were analyzed as described earlier (Loy et al., 2005). When analyzing microarrays hybridized with RNA, all slides in an experiment were scanned at the same laser intensity and the same amplification gain of the photomultiplier tube to ensure slide-to-slide comparability. For subsequent calculation of the cumulative fluorescence intensities, the scanned images did not contain signal-saturated pixels in the probe spot and background area. During image analysis, the circles defining the individual probe spots were adjusted to cover the whole spot area. Before the calculation of the cumulative fluorescence intensity of a probe spot, the mean pixel intensity in the local background was subtracted from the intensity measured for each pixel in the spot.

Radioactivity scanning and calculation of activity values

Radioactivity was recorded with a β-imager (BioSpace Mesures, Paris, France) using a highly sensitive scintillation membrane (Zinsser Analytics, Frankfurt, Germany) and chemoluminescence detection. To obtain an optimal signal-to-noise ratio the slides were incubated for 24 h. The radioactivity for each spot and background was quantified in disintegrations per minute using β-Vision software (BioSpace Mesures). In analogy to the fluorescence image analysis, the circles defining the individual probe spots were adjusted to cover the whole spot area before measuring the background-corrected cumulative radioactivity. The activity value (A-value) of each probe spot was calculated as the ratio of the cumulative radioactive intensity and the cumulative fluorescence intensity (Adamczyk et al., 2003).

Cloning and comparative sequence analysis

The 16S rRNA gene sequences of Rhodocyclales were retrieved from the pristine-activated sludge sample and phylogenetically analyzed as described earlier (Loy et al., 2005). In brief, PCR products obtained with the primer pairs A, R and Z were used for the construction of three separate 16S rRNA gene clone libraries and insert sequences of the isolated plasmids were sequenced using the vector-targeted primers M13F and M13R. Chimeric and Rhodocylales sequences were identified by employing the respective software tools of the Ribosomal Database Project II website (Cole et al., 2005) and by phylogenetic analyses with the ARB program package (Ludwig et al., 2004). Homologous coverage C was calculated according to C=[1−(n1 × N−1)] × 100%, with n1 as number of operational taxonomic units (OTUs) containing only one sequence, and N as total number of clones analyzed. Non-parametric Chao1 richness estimates were calculated by using the program EstimateS (version 7.00, http://viceroy.eeb.uconn.edu/estimates). The presence of fully complementary target sites for RHC-PhyloChip probes in the retrieved 16S rRNA sequences were analyzed by using the ‘Probe Match’ option of ARB and the program CalcOligo 2.03 (Stralis-Pavese et al., 2004).

In silico probe design and evaluation

Novel 16S rRNA-targeted oligonucleotide probes for FISH (Table 1), specific for selected Rhodocyclales-Aalborg East activated sludge clones, were designed by using the ‘Design Probes’ option of ARB and evaluated with probeCheck (Loy et al., 2008). Final probes contained a maximal number of strongly destabilizing mismatching bases that were centrally located in the probe-binding sites of non-target sequences.

Table 1 rRNA-targeted oligonucleotide probes used for FISH
Full size table

Quantitative FISH and MAR-FISH

The abundance of selected Rhodocyclales groups in the activated sludge sample was determined by FISH with confocal laser scanning microscopy and subsequent image analysis by using the DAIME program version 1.1 (Daims et al., 2006). Ribosomal RNA-targeted oligonucleotide probes that were used for FISH are depicted in Table 1. Optimal hybridization conditions for newly developed probes were determined as outlined earlier (Juretschko et al., 2002). Further details on the probes can be found at probeBase (Loy et al., 2007). In addition, the incorporation of radioactive substrate into activated sludge bacteria was monitored on a single-cell level by using MAR-FISH (Lee et al., 1999).

Nucleotide sequence accession numbers

The sequences determined in this study were deposited at the EMBL Nucleotide Sequence Database under accession numbers AM268340 to AM268373.

Results

Rhodocyclales diversity in Aalborg East activated sludge

The presence and abundance of members of the order Rhodocyclales in a nitrifying–denitrifying, phosphorus-removing activated sludge were analyzed by applying different molecular methods. Initially, 16S rRNA genes were amplified from activated sludge DNA by using general bacterial and Rhodocyclales group-selective primers. PCR products were obtained with all primers, except with the Zoogloea-selective Z primer pair. A PCR product was only obtained with the Z primer pair by applying a nested PCR approach in which 1 μl of the bacterial 16S rRNA gene PCR product served as template. The different PCR products were fluorescently labeled in separate reactions and subsequently hybridized individually with RHC-PhyloChips. Based on positive probe signals, a composite microarray fingerprint of the Rhodocyclales community present in the activated sludge was created by merging the separate microarray hybridization patterns obtained with the Rhodocyclales subgroup selective and the common bacterial 16S rRNA gene PCR products (Loy et al., 2005). This merged microarray pattern indicated the presence of bacteria belonging or related to the Sterolibacterium lineage, the ‘Candidatus Accumulibacter’ cluster, and the genera Quadricoccus, Thauera and Zoogloea (Figure 1).

Figure 1
figure 1

Presence of distinct Rhodocyclales groups in the Aalborg East activated sludge as interpreted from 16S rRNA gene-PCR-based RHC-PhyloChip analysis. Positive probes (with identical and hierarchical specificities) of the composite RHC-PhyloChip hybridization pattern used for interpretation of the results according to the multiple probe concept (Loy and Bodrossy, 2006) are depicted in bold.

Full size image

In the following step, microarray results were evaluated and the phylogeny of Rhodocyclales present in the Aalborg East wastewater treatment plant was further resolved by PCR amplification of their 16S rRNA genes using the subgroup-selective primers, cloning, and subsequent sequencing. Only sequences affiliated with the order Rhodocylales were further analyzed. After removal of chimeras, the remaining 34 sequences were grouped in 16 OTUs using an arbitrary 16S rRNA sequence similarity threshold of 97%. Calculation of the homologous coverage indicated that 77% of the OTU diversity in the combined gene libraries was sampled. Based on Chao1 analysis, 27 OTUs of Rhodocyclales were indicated. Phylogenetic analysis of the clone sequences confirmed presence of members of the Sterolibacterium lineage and the genera ‘Candidatus Accumulibacter’, Thauera and Zoogloea, as indicated by the RHC-PhyloChip fingerprint (Figure 2). In addition, some of the sequences affiliated with these taxa contained the fully complementary target sites for those probes that were positive in the microarray analyses (Figure 3). The positive signal of RHC-PhyloChip probe QUACO135, indicating the presence of Quadricoccus australiensis and related bacteria (Figure 1), most likely derive from bacteria represented by clones R10 and R11, which were retrieved with primer pair R (Figure 2). Although the probe-binding site of QUACO135 is located outside the primer pair R-based 16S rRNA gene fragment (Loy et al., 2005), the closest relative of clones R10 and R11 is the clone sequence PHOS-HE23 (accession number AF314420), which is perfectly matching probe QUACO135. In summary, the comparative sequence analysis showed the presence of a diverse Rhodocylales community, including members of potentially novel Rhodocyclales lineages for which no closely related isolates are available (Figure 2) and no specific probes were included on the RHC-PhyloChip.

Figure 2
figure 2

Phylogenetic affiliation and function of Rhodocyclales in Aalborg East activated sludge. 16S rRNA gene consensus tree based on maximum-likelihood analysis (AxML) performed with a 50% conservation filter for the Betaproteobacteria. The tree shows the affiliation of clone sequences (boldface type) retrieved by using Rhodocyclales subgroup-selective primer pairs A (A clones), R (R clones), and Z (Z clones) for PCR. In addition, the incubation conditions under which the different Aalborg East sludge microorganisms were active according to isotope array and MAR-FISH analyses are shown. Abbreviations: O, oxic incubation; N, anoxic incubation with nitrate; P, propionate; B, butyrate; T, toluene. A hashmark indicates that activity was confirmed by MAR-FISH, whereas an asterisk indicates that activity was only detected by MAR-FISH. Parentheses show affiliation to the different Rhodocyclales lineages as outlined earlier (Loy et al., 2005); dotted line indicates that a sequence cluster could not be unambiguously affiliated with the respective lineage. The bar indicates 10% estimated sequence divergence. Polytomic nodes connect branches for which a relative order could not be determined unambiguously by applying neighbor-joining, maximum-parsimony and maximum-likelihood treeing methods. Percent parsimony bootstrap values are shown by filled (⩾90%) and open (75–89%) squares. Branches without squares had bootstrap values of <75%.

Full size image
Figure 3
figure 3

Number of mismatching bases in the RHC-PhyloChip probe target sites of Aalborg East clone sequences. All probes that were positive in the 16S rRNA gene PCR product-based composite RHC-PhyloChip hybridization fingerprint are shown, except of probes STEBA1426 and ZOGLO1416 that could not be analyzed because the respective target sites are only partly present on the amplified gene fragments. Clone sequences are phylogenetically ordered according to Figure 2. NS, probe-binding site is outside the amplified fragment. P, probe-binding site (partially) overlaps with the primer-binding site.

Full size image

Based on the microarray and sequencing data, suitable rRNA-targeted probes (Table 1) were selected or newly developed and applied for quantitative FISH analysis. Relative abundances of the different probe-defined Rhodocyclales groups ranged between <1% and 2.6% of the bacterial biovolume detectable with the EUB338 probe set (Table 2).

Table 2 Relative abundance and substrate uptake of different Rhodocyclales groups in Aalborg East activated sludge as determined by quantitative FISH and MAR-FISH, respectively
Full size table

Isotope labeling of RNA

The efficiency of the heterotrophic 14CO2-assimilation labeling strategy (Hesselsoe et al., 2005) was initially evaluated for labeling of RNA from propionate-utilizing organisms in Aalborg East activated sludge. Anoxic and oxic incubations with unlabeled propionate plus 14CO2 yielded considerably higher levels of radioactivity in the extracted RNA than incubations with [14C]propionate and unlabeled propionate (Table 3). Hence, 14CO2 was used as the only isotope source for incubations of activated sludge with butyrate and toluene. In addition, control incubations without additional substrates were performed for all conditions and showed considerably less incorporation of isotope in biomass and purified RNA (Table 3).

Table 3 Labeling efficiency of biomass and RNA after incubation of activated sludge with 14C-labeled substrates under different conditions
Full size table

Optimization of the RHC-PhyloChip for isotope array analysis

Earlier optimization of the RHC-PhyloChip only focused on hybridization with random-prime-labeled 16S rRNA gene PCR products (Loy et al., 2005). A published protocol for hybridization of non-fragmented, native RNA (Adamczyk et al., 2003) was initially adapted for use with fragmented RNA. For this purpose, separate RHC-PhyloChips were hybridized with 0.75 μg of in vitro transcribed, fluorescently labeled and fragmented 16S rRNA from the reference clone S3 (accession number AF072918) under different stringencies, that is, at 5%, 10%, 15%, 20%, 25% and 30% formamide in the hybridization buffer. As expected, the specificity index (the difference in mean cumulative fluorescence signal intensity between perfectly matched and mismatched probes) increased by a factor of ∼2 with increasing formamide concentration, whereas the signal intensity constantly decreased (Figure 4). To maintain a relatively high level of sensitivity (that is, high signal intensity of a probe with its perfectly matched RNA), 10% formamide in the hybridization buffer was used for all subsequent hybridizations with native RNA as a compromise between specificity (that is, the tendency of some probes to cross-react with mismatching target RNA) and sensitivity. The multiple probe concept, which formed the basis during the design of the RHC-PhyloChip (Loy et al., 2005), alleviates problems that are due to unspecific hybridizations of individual probes at a lower stringency (Loy and Bodrossy, 2006).

Figure 4
figure 4

Signal intensities of probes forming detectable, perfectly matched and mismatched hybrids with fragmented, fluorescently labeled 16S rRNA from reference clone S3 (AF072918) at different formamide concentrations in the hybridization buffer. (a) normalized signal intensities of three replicate spots are shown for each perfectly matching probe (closed circles) and each probe with one, two or three mismatch(es) (open circles). Cumulative fluorescence intensity of each probe was normalized with the highest intensity of this probe observed at the different stringencies. (b) (left Y-axis); mean cumulative signal intensity of all perfectly matched probes (closed circles) and of all probes with one to three mismatches (open circles) are indicated. (right Y-axis); open diamonds indicate the specificity index at a given stringency, that is, the ratio of the mean signal intensity of all perfectly matched probes versus the mean signal intensity of all probes with one to three mismatches, for each formamide concentration.

Full size image

Less than 0.1% of the added isotope was generally incorporated into community RNA during the isotopic labeling incubations (Table 3). To determine the optimal amount of 14C-labeled, fragmented RNA per hybridization, separate RHC-PhyloChips were hybridized with 0.3–10 μg of Aalborg East activated sludge RNA (Figure 5). The cumulative radioactivity per probe spot on the RHC-PhyloChip increased with increasing RNA amount and no signal saturation was observed within the RNA concentration range tested. In contrast, use of higher RNA amounts improves the detection limit of the isotope array approach. In contrast, hybridization under probe-saturated, target-excess conditions (that is, conditions under which most of the probe molecules on a spot have bound a target molecule) impairs reliable quantification of probe signals (Schena, 2003). Six (from the propionate, butyrate, and the corresponding control incubations) or 10 μg of RNA (from the toluene and the corresponding control incubations) were thus used for each further isotope array hybridization.

Figure 5
figure 5

Mean radioactivity of RHC-PhyloChip probes after hybridization with different amounts of 14C-labeled RNA. Fragmented, community RNA was prepared from oxic incubations of activated sludge with 14CO2 and (a) propionate, (b) butyrate or (c) toluene.

Full size image

Isotope array analysis of Rhodocyclales in activated sludge

A separate RHC-PhyloChip was hybridized with fluorescently labeled and fragmented RNA from each activated sludge subsample that was incubated with 14CO2 and allylthiourea under different conditions (Table 3). An activity and substrate-utilization profile of the different Rhodocyclales groups in the activated sludge was created (Table 4) based on the presence–absence of radioactive probe signals after isotope array analysis (Supplementary online Tables) and the retrieved 16S rRNA gene sequences (Figures 2 and 3). Radioactive signals of microarray probe QUACO135 in all hybridizations indicated that bacteria represented by clones R10 and R11 were active under all conditions tested. In contrast, the isotope array analyses further suggested that all other Rhodocyclales groups, for which specific probes were present on the RHC-PhyloChip, displayed more specialized substrate-incorporation behaviors. For example, radioactive 16S rRNA from members of the genus Zoogloea and from a putative novel Rhodocyclales lineage (represented by clones R1, R7 and R9) was detectable after oxic incubation with butyrate and propionate, but not with toluene. Under nitrate-reducing conditions, only hybridization of RNA that was extracted from the incubation with butyrate yielded radioactive signals with a genus Zoogloea-specific probe and the probe RHO828, fully matching clones R1, R7 and R9 (Figures 2 and 3).

Table 4 Phylogenetic and functional profile of Rhodocyclales in the Aalborg East activated sludge as inferred from presence/absence of radioactive probe signals after RHC-Isotope Array and 16S rRNA gene clone library analyses
Full size table

By assuming an unbiased fluorescence-labeling step and that each probe hybridized only to its perfectly matched target organisms, the A-values were used to infer changes in heterotrophic 14CO2-assimilation activity of probe-defined populations. Addition of electron donors stimulated incorporation of 14C into rRNA in all experiments, that is, yielded higher A-values in comparison with the control samples (Figure 6). For example, addition of each of the three substrates under oxic conditions caused a five- to sixfold increase in the A-value for probe QUACO135. An even higher increase was observed for probe QUACO135 under denitrifying conditions in the presence of propionate and butyrate, but toluene seemed to stimulate heterotrophic 14CO2 assimilation only moderately (Figure 6).

Figure 6
figure 6

Relative A-values of selected RHC-PhyloChip probes after isotope array analysis of activated sludge. The A-value of each probe was divided by the mean A-value of the universal probes (UNIV1389a, UNIV1389b and UNIV1389c) recorded on reference slides that were hybridized with a corresponding control sample (that is, a sample that was incubated with 14CO2 under the same conditions without addition of electron donor).

Full size image

MAR-FISH analysis of Rhodocyclales in activated sludge

For independent confirmation of the isotope array results, substrate incorporation was also analyzed at the single-cell level for selected incubations and Rhodocyclales groups by MAR-FISH (Table 2). MAR-FISH results obtained with probes AT1458, PAO651, RHC439 and ZOGLO1416 could be directly compared with isotope array results because corresponding probes were present on the RHC-PhyloChip. With one exception, similar results were obtained with both techniques. Although MAR-FISH with probe ZOGLO1416 clearly showed that 14CO2 assimilation of Zoogloea species was enhanced during anoxic incubation with propionate and nitrate, no radioactivity was recorded on the isotope array with the same probe. This suggests that the amount and specific radioactivity of Zoogloea 16S rRNA was below the detection limit of probe ZOGLO1416, when it was applied in the microarray format. In addition, newly developed probes with a narrower target range for the Aalborg East activated sludge clones were used for MAR-FISH analysis at a higher phylogenetic resolution than offered by the current RHC-PhyloChip probe set. Application of probes ZOGLO441 and ZOGLO454, differentiating between two Zoogloea clone groups in the activated sludge, indicated that both the probe-defined populations reduced nitrate anaerobically with propionate. In contrast, only ZOGLO441-targeted cells were MAR positive in anoxic incubations with nitrate and butyrate.

The MAR-positive signals were also observed for cells detected by the new RHC445–RHC827 probe combination (targeting clones R10 and R11 that are presumably also targeted by RHC-PhyloChip probe QUACO135) in all samples analyzed (Table 2). Compared with the control, considerably larger proportions of MAR-positive cells were observed in incubations with substrates, clearly indicating that the RHC445–RHC827 probe-defined population metabolized the offered substrates.

Discussion

Methodological and conceptual considerations for isotope array analysis

Additional modifications of the original isotope array protocol (Adamczyk et al., 2003) were introduced in this study and their implications for inferring biologically meaningful data are highlighted in the following section. General strengths and limitations of the isotope array approach have been discussed earlier and are thus not explicitly mentioned (Adamczyk et al., 2003; Wagner et al., 2006; Neufeld et al., 2007).

Metabolically active heterotrophic microorganisms assimilate CO2 through various carboxylation reactions (Roslev et al., 2004) catalyzed by enzymes, such as pyruvate carboxylase, phosphoenolpyruvate carboxylase and coenzyme A carboxylase (Barker, 1941; Dijkhuizen and Harder, 1985). In analogy to the HetCO2-MAR variant of MAR-FISH (Hesselsoe et al., 2005), the CO2-assimilating ability of heterotrophic microorganisms was employed for radioactive labeling of community members metabolizing a substrate in the presence of radioactive 14CO2. In this approach, consumption of non-radioactive substrates is measured indirectly through 14CO2 incorporation. It is therefore necessary to perform 14CO2 control experiments without the addition of organic substrates. Subsequent differential display analysis of the isotope array data of the control and the actual experiment distinguishes between background CO2 assimilation (for example, driven by metabolization of cellular storage compounds) and CO2 assimilation that is additionally induced by the added substrate. Furthermore, the relative, substrate-induced increase in CO2-assimilating activity of a probe-defined population can be assessed through the A-value (Figure 6). Although additional control experiments must be performed, this indirect 14CO2-based labeling strategy has several advantages over the alternative isotope-labeling approach for analyzing the metabolization of an organic substrate, that is, the incubation of a sample with the radioactive organic substrate, but without 14CO2 (Hesselsoe et al., 2005). First, 14CO2 is available with maximal specific radioactivity (>99% of C atoms are 14C) and is relatively inexpensive. Therefore, it can be applied economically in high quantities. Second, the variety of commercially available 14C-labeled organic substrates, especially of large molecules, is limited and their specific radioactivity (%14C content) is often low. Homogeneously isotope-labeled complex substrates are often not commercially available at all or very expensive. In our test experiments, community RNA was significantly more strongly labeled after incubation with 14CO2 and unlabeled propionate than after incubation with [14C]propionate and unlabeled propionate (Table 3). However, the labeling efficiency of the latter incubation might be improved by adjusting the ratio between the concentrations of radioactive and non-radioactive substrate (1:11.8 for unlabeled/14C-labeled propionate in this study) and the %14C content of the added substrate (∼33% of C atoms were 14C in the [14C]propionate used in this study). We suggest that the heterotrophic CO2-assimilation labeling strategy offers additional flexibility in the design of isotope array experiments and, at least in cases where large organic molecules are only available with one or a few 14C atoms, also might improve the detection limit of the isotope array.

The sensitivity of the isotope array generally depends on the abundance (that is, absolute and relative rRNA content) and on the specific substrate-assimilation rate (that is, the rate at which substrate-derived 14C is incorporated into in vivo synthesized rRNA) of a probe-defined target population (Adamczyk et al., 2003). The detection limit for different probe-defined populations at a given hybridization condition will further vary owing to the different thermodynamic properties of the probes. For instance, members of the genus Zoogloea, comprising only about 1% of the bacterial biovolume in the Aalborg East activated sludge, are targeted by the genus-specific microarray probes ZOGLO828 and ZOGLO1416 with a theoretical free binding energy (ΔG12) of −17.94 and −20.27 kcal mol−1, respectively. The difference in free binding energy could explain why only probe ZOGLO1416 showed radioactivity after isotope array analysis of the sample exposed to butyrate under denitrifying conditions (Figure 6 and Table 4). However, the efficiency of a microarray probe to hybridize with its target (N.B.: the signal intensity of the resulting probe-target hybrid is often used as proxy for this efficiency) is strongly dependent on the experimental conditions, such as target preparation and hybridization procedures, factors which are currently not adequately addressed by available algorithms for calculating the free energy of a probe-target hybridization event. Theoretical ΔG values will thus inevitably not always correlate with the actual probe signal intensities (Pozhitkov et al., 2006). Hence, an alternative explanation for the observed discrepancy in signal intensities for probes ZOGLO828 and ZOGLO1416 is that they differ in their coverage of target organisms and/or tendency to cross-hybridize with non-target organisms.

For hybridization signals above the detection threshold, the A-value could theoretically compensate for the different binding efficiencies of different probes. However, quantification using the A-value is susceptible to systematic biases. The direct fluorescence labeling of RNA method used in this study attaches Cy-dye molecules covalently to guanine residues. Thus, organisms with a guanine-rich 16S rRNA could contain more dye molecules per RNA molecule, underestimating their A-value-based substrate-utilization rate compared with guanine-poorer organisms. It is noteworthy that this bias is less profound for microarray probes targeting a phylogenetically narrow range of microorganisms with no or little variation in their 16S rRNA guanine contents. Accuracy of the A-value further depends on the specificity of the respective microarray probe. Although these sources of errors generally hamper comparisons among different probe-defined populations, the A-value still offers the possibility to monitor relative changes in substrate incorporation activity of a given probe-defined population under different environmental conditions.

In summary, the described modifications of the isotope array protocol, including the adaptation of a heterotrophic 14CO2-assimilation-based isotope labeling, improved the sensitivity of the technique and enabled detection of native, radioactive 16S rRNA extracted from low-abundant (relative abundance of ∼1% of the total bacterial biovolume as analyzed by quantitative FISH), but active members of a complex-activated sludge community.

Diversity and ecophysiology of Rhodocyclales in Aalborg East activated sludge

In this study, we showed the diversity and some potential nutritional niches of bacteria belonging to the order Rhodocyclales in a nitrogen- and phosphorus-removing, full-scale wastewater treatment plant. Identification and quantification of members of the Rhodocyclales followed the same RHC-PhyloChip-centric strategy as outlined in an earlier study (Loy et al., 2005). For functional analysis of Rhodocyclales using the isotope array approach, the RHC-PhyloChip was newly adapted for hybridization of native community rRNA.

The Aalborg East activated sludge contained an estimated number of 27 different Rhodocyclales OTUs, which belonged to at least seven different lineages (Figure 2). The presence of a high Rhodocyclales richness in denitrifying wastewater treatment systems is consistent with earlier observations (Juretschko et al., 2002; Ginige et al., 2005; Liu et al., 2006; Thomsen et al., 2007; Morgan-Sagastume et al., 2008). In contrast to these studies, the relative abundance of selected Rhodocyclales groups did not exceed 2.6% of the total bacterial biovolume (Table 2). However, combined interpretation of isotope array and MAR-FISH data showed that all Rhodocyclales groups, for which specific probes were applied, were active and metabolized (some of) the offered electron donors in the presence of oxygen and/or under anoxic nitrate-reducing conditions (Figures 2 and 6, Tables 2 and 4). Aside from the involvement in nitrogen transformation, one can further speculate about the ecophysiology of the different Rhodocyclales groups on the basis of their activity pattern recorded for the different incubation conditions.

‘Candidatus Accumulibacter’ species and members of the Dechloromonas–Ferribacterium–Quadricoccus–Azonexus-related group (represented by clones R10 and R11) were the only metabolically active groups in the control incubations, under substrate depletion (Table 4). Background 14CO2 assimilation by these organisms could result from mobilization of intracellular compounds that are stored for bridging the unavoidable phases of nutrient limitation (Hesselsoe et al., 2005). The capacity for anaerobically storing poly-β-hydroxy alkanoates puts polyphosphate-accumulating ‘Candidatus Accumulibacter’ species in a selective advantage during enhanced biological phosphorus removal (Blackall et al., 2002). Therefore, one might argue that bacteria represented by clones R10 and R11 are also linked to this process. Whether these bacteria truly have the beneficial polyphosphate-accumulating phenotype or are glycogen non-polyphosphate-accumulating organisms, which also synthesize poly-β-hydroxy alkanoates anaerobically but are detrimental for the enhanced biological phosphorus removal process, remains to be elucidated (Kong et al., 2004, 2006). The FISH probe set RHC445–RHC827, developed in this study (Table 1), is a molecular tool for proving this hypothesis in future MAR-FISH experiments.

Our results further support earlier findings that some polyphosphate-accumulating organisms are also capable of reducing nitrate in wastewater treatment systems (Zeng et al., 2003; Kong et al., 2004; Thomsen et al., 2007; Morgan-Sagastume et al., 2008). Both, the ‘Candidatus Accumulibacter’ species and the bacteria of the Dechloromonas–Ferribacterium–Quadricoccus–Azonexus-related group, showed moderately to strongly enhanced biosynthesis when they were exposed to propionate, butyrate or toluene under nitrate-reducing conditions (Figure 6, Table 2). It remains to be shown whether these organisms are truly denitrifiers or only catalyze the reduction of nitrate to nitrite.

Aromatic hydrocarbons, such as toluene, can occasionally occur in municipal wastewater through contamination with oil, gasoline and industrial waste (Eriksson et al., 2003; Dincer and Muezzinoglu, 2008). Toluene was selected as one of the substrates analyzed in this study because (i) the capability for aerobic and anaerobic aromatic compound degradation is widespread among strains from different Rhodocycales lineages, such as Thauera sp. DNT-1 (Shinoda et al., 2004) or Dechloromonas sp. RCB (Chakraborty et al., 2005) and (ii) pre-experiments have shown that the Aalborg East activated sludge has the capacity for anaerobic and aerobic toluene degradations (data not shown). Apart from being involved in nitrogen- and phosphorus removal, we hypothesized that wastewater Rhodocyclales are also responsible for biodegradation of aromatic hydrocarbons in wastewater. Consistently, several probe-defined populations showed increased 14CO2 assimilation activity on exposure to toluene in the isotope array and MAR-FISH experiments (Figure 6, Table 2). This finding clearly highlights the aromatic compound decontamination potential offered by the Rhodocyclales communities in wastewater treatment plants.

It is likely that microorganisms for which no specific probes were applied, such as members of the Sterolibacterium lineage (Tarlera and Denner, 2003) and the putative novel Rhodocyclales lineage (represented by clones A10, A11 and A22) (Figure 2), and other microorganisms (Morgan-Sagastume et al., 2008), also contributed to anaerobic nitrate reduction in this wastewater treatment plant. Nevertheless, our data show that a high functional redundancy, regarding the ability for anaerobic nitrate reduction (and thus potentially for denitrification), existed within the diverse Rhodocyclales community in the activated sludge (Table 4, Figures 2 and 6). Substrate-utilization profiles of the different Rhodocyclales groups further indicate that they (i) occupied partially overlapping nutritional niches in the wastewater ecosystem and (ii) might differ in their adaptation to fluctuations in the organic compound composition of wastewater. A rather generalist substrate-consumption behavior was exhibited by ‘Candidatus Accumulibacter’ species and members of the Dechloromonas–Ferribacterium–Quadricoccus–Azonexus-related group (represented by clones R10 and R11), whereas other Rhodocyclales groups such as Thauera and Zoogloea seemed to be more specialized. It is self-evident that the implementation of additional substrates, incubation conditions and parameters such as substrate affinities and incorporation rates in the isotope array testing regime would show even more detailed information on the extent of niche overlap and substrate specialization of Rhodocyclales in this activated sludge.

In conclusion, by application of the isotope array in concert with other complementary techniques, we showed the functional versatility and redundancy of a diverse Rhodocyclales community in activated sludge from a full-scale wastewater treatment plant, lending further weight to the importance of members of this taxon for removal of environmental pollutants. We further anticipate that the future of the isotope array techniques lies within its adaptation for the analysis of stable isotopes by using Raman microspectroscopy or nanoSIMS. Technical limitations of these instruments, such as the long image acquisition time of the nanoSIMS for analyzing large areas of interest (Musat et al., 2008), currently still hinder their application for microarray analysis. However, the high spatial resolution of both techniques would theoretically allow exploiting high-density microarrays consisting of thousands of small probe spots (<50 μm in diameter). This option is currently not available for the radioactive isotope array approach because of the low spatial resolution of β-imagers. The high sensitivity of the nanoSIMS (Wagner, 2009) even gives hope for simultaneous analysis of gene expression and substrate utilization by functional gene array-based detection (He et al., 2007) of stable isotope-labeled mRNA. Highly parallel analysis of physiological properties of many phylogenetically different community members is a looked-for option in microbial ecology. The isotope array method thus has much promise to unfolding the complex nutritional networks of uncultivated microorganisms living in multiorganism assemblages.