Development of a Rapid Assay for Determining the Relative ...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 2005, p. 8481–8490 0099-2240/05/$08.00⫹0 doi:10.1128/AEM.71.12.8481–8490.2005 Copyright © 2005, American Society for Microbiology. All Rights Reserved.

Vol. 71, No. 12

Development of a Rapid Assay for Determining the Relative Abundance of Bacteria Arlene K. Rowan,1 Russell J. Davenport,1* Jason R. Snape,2 David Fearnside,3 Michael R. Barer,4 Thomas P. Curtis,1 and Ian M. Head1 School of Civil Engineering and Geosciences, University of Newcastle, Newcastle upon Tyne NE1 7RU,1 Brixham Environmental Laboratory, AstraZeneca Global SHE, Brixham, Devon TQ5 8BA,2 Yorkshire Water Services, Western House, Halifax Road, Bradford, West Yorkshire BD6 2LZ,3 and Department of Microbiology and Immunology, University of Leicester, Leicester LE1 7RH,4 United Kingdom Received 3 June 2005/Accepted 8 September 2005

A sandwich hybridization assay for high-throughput, rapid, simple, and inexpensive quantification of specific microbial populations was evaluated. The assay is based on the hybridization of a target rRNA with differentially labeled capture and detector probes. Betaproteobacterial ammonia-oxidizing bacteria (AOB) were selected as the target group for the study, since they represent a phylogenetically coherent group of organisms that perform a well-defined geochemical function in natural and engineered environments. Reagent concentrations, probe combinations, and washing, blocking, and hybridization conditions were optimized to improve signal and reduce background. The detection limits for the optimized RNA assay were equivalent to approximately 103 to 104 and 104 to 105 bacterial cells, respectively, for E. coli rRNA and RNA extracted from activated sludge, by using probes targeting the majority of bacteria. Furthermore, the RNA assay had good specificity, permitted discrimination of rRNA sequences that differed by a 2-bp mismatch in the probe target region, and could distinguish the sizes of AOB populations in nitrifying and nonnitrifying wastewater treatment plants. often require the use of radioactive isotopes or expensive equipment and consumables. An alternative sensitive and specific method is a sandwich hybridization assay for the detection of rRNA, developed by Wicks and coworkers (46). The RNA assay is based on hybridization of a target rRNA with differentially labeled capture and detector probes. It allows quantification of specific rRNA species from crude preparations of RNA. Furthermore the RNA assay is amenable to a 96-well plate format, opening the opportunity for automated analysis of large numbers of samples, which would allow rapid quantitative high-throughput analyses. Such analyses are required for generating the long time series needed for understanding the ecology and population dynamics of organisms that mediate many important biological processes, such as nitrification in wastewater treatment plants (WWTPs). Nitrification is a two-stage process in which ammonia, derived from degraded organic nitrogenous compounds, is oxidized into nitrite and then into nitrate. It is an essential component of the nitrogen cycle and occurs in many environments, including soils, estuaries, and WWTPs. The process is largely mediated by two groups of microorganisms, the autotrophic betaproteobacterial ammonia-oxidizing bacteria (AOB) and the autotrophic nitrite-oxidizing bacteria, where the AOB are responsible for the rate-limiting step of ammonia oxidation. In the present study, the primary format of the RNA assay developed by Wicks et al. was retained (46), but the assay was extensively modified, developed, and optimized for the specific detection and quantification of betaproteobacterial AOB. Autotrophic AOB are ideal candidates for such an approach, because they form a coherent phylogenetic group that performs a distinct metabolic and process function. To demonstrate the utility of the assay, it was used to measure the

Quantification of specific bacterial populations is key to understanding the mechanisms that underlie many biologically mediated processes, including nitrification. The use of rRNAbased cultivation-independent techniques may be one potentially useful means to do this. Numerous rRNA-based techniques are now commonly employed to quantify microbial communities in environmental samples. These include fluorescence in situ hybridization (FISH) (see, e.g., references 11, 12, 17, 27, 37, and 40), quantitative dot blot hybridization (see, e.g., references 14 and 34), PCR-based methods (e.g., real time PCR [19] and competitive PCR [15, 33]), and microarray technology (see, e.g., references 1, 18, and 38). Each of these methods has particular advantages and disadvantages. For example, FISH allows direct identification and quantification of microorganisms in environmental samples and, because of its ability to measure individuals accurately, represents the “gold standard.” However, FISH has a low sample throughput and is thus too slow for routine analysis. In contrast, PCR-based methods allow high sample throughput, which may be required for statistically valid spatial and temporal analysis. However, PCR-based methods are prone to biases (typically introduced during nucleic acid extraction and amplification) that may make quantification of natural bacterial populations difficult (see, e.g., references 20, 39, 39a, and 43). In addition, relatively pure preparations of DNA are required to avoid problems of inhibition of DNA polymerases. Quantitative dot blots and microarrays, on the other hand,

* Corresponding author. Mailing address: School of Civil Engineering and Geosciences, Cassie Building, University of Newcastle upon Tyne, Newcastle upon Tyne NE1 7RU, United Kingdom. Phone: 44 191 222 5544. Fax: 44 191 222 6502. E-mail: [email protected]. 8481

8482

ROWAN ET AL.

relative amounts of AOB 16S rRNA in two WWTPs, one exhibiting good nitrification performance and the other exhibiting poor nitrification. MATERIALS AND METHODS Organisms, growth media, and conditions. Nitrosomonas europaea NCIMB 11850T was cultured, using the medium and methods described by Watson and Mandel (44). Cultures were incubated at 30°C in the dark. The growth of AOB was monitored by following the change in pH caused by the oxidation of ammonia to nitrite by means of a pH indicator (phenol red) present in the growth medium. When the medium changed color from pink to yellow, filter-sterilized sodium bicarbonate was added to neutralize the growth medium. Cells were harvested after three to five rounds of neutralization. Ralstonia eutropha (DSM 531T ) was cultured in nutrient broth at 30°C. Cells from R. eutropha were harvested in the logarithmic phase. Harvested cells were suspended in 1 ml of TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and stored at ⫺20°C prior to RNA extraction. Wastewater treatment plant samples. Samples of activated sludge were collected from two wastewater treatment plants in the United Kingdom. One of the reactors was known to be nitrifying effectively at the time of sampling (average 99.7% ammonia removal), whereas the other reactor was suffering from poor nitrification (average 23% ammonia removal). Activated-sludge samples were taken from each reactor, preserved immediately in 50% ethanol, and stored at ⫺20°C (26). FISH was performed on these samples as previously described to independently validate the relative abundance of AOB in these samples (11). RNA extraction. Total RNA was extracted from cultures and activated-sludge samples using a FastRNA extraction kit (blue matrix; Q-biogene, Fisher Scientific UK) and a Ribolyser (Hybaid Ltd., Middlesex, United Kingdom) according to the manufacturers’ instructions. Six or more extracts were pooled, and the concentration and purity of RNA were determined by measuring the absorbances at 260 nm and 280 nm. The amount of RNA was expressed as the number of 16S rRNA molecules. The percentage of 16S rRNA in RNA extracts (5S, 16S, and 23S rRNA, tRNA, and mRNA) was estimated to be 27% (31). In preparations of purified 16S-plus-23S rRNA from Escherichia coli MRE600 (Roche Diagnostics, United Kingdom), the 16S rRNA was estimated to account for 34% of the total RNA (31). All the usual precautions were utilized to prevent degradation of RNA (46). RNA assay. The RNA assay is a three-step sandwich hybridization assay (hybridization, capture, and detection) that uses a pair of labeled oligonucleotide probes, one to capture the rRNA and another to detect the captured target rRNA (Fig. 1). Helper probes were included in certain assays to improve the accessibility of some target sites on the rRNA. The assay was extensively modified from the procedure developed by Wicks and coworkers (46). Two main modifications were made to adapt the assay for use with AOB and environmental samples. Solid-phase capture of the probe-target hybrids on streptavidin-coated 96-well plates was employed in the original assay to capture target rRNA (45, 46). In the modified assay, capture using suspensions of paramagnetic beads was used to improve the kinetics of the reaction and reduce hybridization time. Furthermore, the detection method employed by Wicks et al. (46), which required extensive purification of the reporter substrate needed for the final detection step, was changed from a direct to an indirect detection method. In the original assay, the probe was conjugated to an enzyme that directly catalyzed the detection reaction, while in the modified method, a digoxigenin (DIG)-labeled probe was indirectly detected using an anti-digoxigenin antibody conjugated to an enzyme that catalyzed the final detection reaction. Furthermore, the assay was optimized extensively, the specificity of the assay was more rigorously tested, and its utility for measuring the abundance of a specific bacterial populations in complex environmental samples was demonstrated. Hybridization. RNA samples were denatured at 95°C for 10 min. The denatured RNA was added to hybridization buffer (0.9 M NaCl, 0.2 M Tris [pH 7.6], 0.05% [wt/vol] N-laurylsarcosine) containing 1012 molecules each of the capture probe (S-D-Bact-0320-a-S-40-B) (45) (Table 1) and the detector probe (S-D-Bact-1524-a-S-17-D) (Table 1) (45) to give a total volume of 10 ␮l in each well of a 96-well plate. Hybridization reactions were mixed by vortexing and incubated at 50°C for 30 min (Fig. 1, Hybridization). While the probes were hybridizing, streptavidin-coated magnetic beads were prepared (Promega, Madison, Wis.). Preparation of beads. The concentration of strepavidin-coated magnetic beads used in the assay was measured by absorbance at 600 nm. Beads were added to each reaction mixture to give the equivalent of 500 absorbance units (AU) per reaction. The appropriate amount of strepavidin-coated magnetic beads was added to the wells of a sterile 96-well plate (white opaque plate; NUNC, Fisher

APPL. ENVIRON. MICROBIOL. Scientific UK). The beads were separated from the bulk liquid by capturing them with a plate magnet (Promega, Madison, Wis.), and the supernatant was removed. A 100-␮l volume of blocking reagent 1 (2⫻ Denhardt’s reagent, 0.1 mg/ml sonicated salmon testis DNA in antibody buffer [AB] [1⫻ SSC {0.15 M NaCl plus 0.015 M sodium citrate}, 20 mM Tris {pH 7.6}, and 0.1% {wt/vol} Tween 20]) was added to each well of the 96-well plate. The beads were incubated with the blocking solution for 5 min at room temperature. The beads were captured, using the plate magnet, and the blocking reagent was replaced with 100 ␮l of AB. The blocked bead preparation was stored at 4°C until required. Capture. Following hybridization, the reaction products were chilled on ice. Hybridization reaction products were transferred to 96-well plates containing the strepavidin-coated magnetic beads and incubated with gentle mixing on an orbital shaker at room temperature for 90 min (Fig. 1, Capture). The beads were captured using the magnet, the supernatant was removed, and the beads were washed twice with 100 ␮l of AB with gentle mixing using an orbital shaker. The beads were recaptured, using the magnet, and the supernatant was removed. AB (100 ␮l) was added to the beads while they were on the magnet, and the beads were incubated for 3 min. The AB was removed, and the beads were resuspended in 100 ␮l of blocking reagent 2 (2⫻ Denhardt’s reagent prepared in AB) and mixed gently for 5 min. Detection. The beads were recaptured, and the blocking reagent was replaced with 100 ␮l anti-DIG antibody labeled with alkaline phosphatase (1/500 dilution; Roche Diagnostics, Germany). The samples were incubated, with gentle mixing, for 30 min. After incubation, the beads were recaptured and washed three times with 100 ␮l of AB (two washes on the magnet, i.e., without mixing, and one wash with mixing). The beads were captured and resuspended in 50 ␮l of AB and 100 ␮l of 0.4 mM CDP-Star Chemiluminescent Substrate with Sapphire-II Enhancer (Applied Biosystems, United Kingdom). The 96-well plate containing the assay reaction mixtures was placed in a luminometer [MicroLumat LB 96 P; Berthold Technologies (U.K.) Ltd.], and the light produced per second per well was measured every 3 min for a total of 1 h. Data were reported as relative light units (RLU) for the total integrated light output (Fig. 1, Detection). The entire procedure takes about 1 day to complete. Wastewater treatment samples (approximately 4 to 6 ml) and culture preparations were analyzed in triplicate; positive and negative controls were always included. The concentration of RNA in extracts from these samples was determined spectrophotometrically, and the number of 16S rRNA molecules was calculated. The same number of 16S rRNA molecules from each sample was used for the assay. Furthermore, a parallel assay was always conducted using a probe set targeting the 16S rRNA of most bacteria to ensure that equivalent amounts of bacterial 16S rRNA were present in all assays. Purified 16S and 23S rRNA from Escherichia coli MRE600 (Roche Diagnostics, United Kingdom) was used as a positive control. A range of negative controls was included in the assay. These controls were used to test for different sources of error. A magnetic-beadonly control (BO) (no probes or RNA, only magnetic beads treated in the same way as the samples) was used to test for background signal; a probe-only control (PO) (probes and magnetic beads treated in the same way as the samples), a nonbacterial probe (NonBact338 or NONEUB [26]) control (sample hybridized with a NonBact338 detector probe), and a yeast (Saccharomyces cerevisiae) RNA control (i.e., yeast RNA instead of sample RNA) were used to test for nonspecific binding. During optimization of the assay, slight modifications of the procedure described above were made. These are detailed in Results and Discussion. Probe design. Probes targeting 16S rRNA from the majority of bacteria were designed (Table 1) and checked for self-complementarity using PRIMROSE v. 1.1.7 (4) and the RDP database (10). Many of the probes used in this study were modifications of existing probes and primers designed to target all bacteria. Statistical analysis. All quantitative data were checked for normality and, where appropriate, transformed using values determined by the Box-Cox transformation (7). Analysis of variance (Tukey’s pairwise comparisons) and comparison of means (parametric [t test] or nonparametric [Mann-Whitney test]) were carried out using Minitab v. 11 (Minitab Inc.).

RESULTS AND DISCUSSION In the present study, a sandwich hybridization RNA assay was developed with the aim of creating a rapid, simple, highthroughput quantification technique for AOB in WWTPs. The assay was optimized to obtain high levels of sensitivity and specificity. Poor sensitivity is often due to a low signal-to-noise

VOL. 71, 2005

RNA ASSAY TO DETERMINE BACTERIAL ABUNDANCE

8483

FIG. 1. Schematic diagram of the sandwich hybridization assay. DIG, digoxigenin; B, biotin, AP, alkaline phosphatase-conjugated antidigoxigenin; RNA, extracted RNA; CDP STAR, chemiluminescent substrate.

ratio. Hence, a reduction in background signal (noise) and an improvement in signal intensity should improve sensitivity. Optimization of sensitivity. (i) Reduction of background (noise). Background noise is produced by nonspecific hybridization, or binding of the probes, RNA, and/or anti-DIG antibody to other components in the assay (e.g., the magnetic beads, or the wells of the 96-well plate) (30). To reduce nonspecific hybridization, the washing, blocking, and hybridization steps were optimized. (ii) Washing. Other researchers have found that increasing the number and length of hybridization washes leads to a reduction in signal intensity without improvement in specificity (see, e.g., references 8 and 24). Indeed, we found that increas-

ing the washing steps (length and number of washes) appeared to have no effect on the background signal. However, an increase in the frequency of washes resulted in a greater variation in the signal intensity (e.g., coefficients of variation were 7.4% for 5 washes and 47% for 15 washes), most likely due to a loss of magnetic beads. (iii) Blocking. The utilization of blocking agents during hybridizations is commonplace (see, e.g., references 3 and 46). Blocking is employed to reduce nonspecific binding by the addition of a mixture of polymeric molecules. These “sticky” polymeric molecules block potential sites for nonspecific binding by, for example, occupying the site through electrostatic interactions prior to the addition of the assay components. A

a Nomenclature based on reference 2. Probe names used previously are given in parentheses. Letters at ends of probe names correspond to label: B, biotin; D, digoxigenin; H, helper (no label); C, competitor (no label). All probes were obtained from Thermo Hybaid, Germany.

Nonbacterial probe; nonspecific 26 None 5⬘ ACT CCT ACG GGA GGC AGC 3⬘ Digoxigenin (5⬘ labeled)

Capture probe; AOB specific 25 ´ A TCC CCT RCT TTY CTC C 3⬘ 5⬘ CG

S-F-bAOB-0190-a-S-19-B (CTO18f) S-D-NBact-0320-b-A-18-H (NonBact)

Biotin

Competitor probe; AOB specific This study

16S rRNA of Ralstonia eutropha 16S rRNA of ␤-AOB 5⬘ CAC CAT TGT ATG ACG TGT GA 3⬘ No label

28 16S rRNA of ␤-AOB 5⬘ CGC CAT TGT ATT ACG TGT GA 3⬘ Digoxigenin (5⬘ labeled)

S-F-bAOB-1224-a-S-20-D (Nso1225) S-S-Ral-1224-a-S-20-C

This study 16S rRNA of bacteria 5⬘ GTA TTA CCG CGG CTG CTG 3⬘ Digoxigenin (5⬘ labeled) S-D-Bact-0518-a-S-18-D

This study 16S rRNA of bacteria 5⬘ TCT GGV CCG TRT CTC AGT 3⬘ No label S-D-Bact-0320-b-S-18-H

This study 16S rRNA of bacteria 5⬘ TCT GGV CCG TRT CTC AGT 3⬘ Digoxigenin (3⬘ labeled) S-D-Bact-0320-b-S-18-D

This study 16S rRNA of bacteria 5⬘ CAC TGC TGC CTC CCG TA 3⬘ Biotin (5⬘ labeled)

45 16S rRNA of bacteria Digoxigenin (3⬘ labeled)

S-D-Bact-1524-a-S-17-D (DIG 1541) S-D-Bact-0343-b-S-17-B

45 16S rRNA of bacteria

5⬘ CAC TGC TGC CTC CCG TAG GAG TCT GGA CCG TGT CTC AGT T 3⬘ 5⬘ AAG GAG GT G ATC CAG CC 3⬘ Biotin (5⬘ labeled) S-D-Bact-0320-a-S-40-B (B3)

Reference Target

TABLE 1. Oligonucleotide probes used in this study

Sequence Label Probe name

a

APPL. ENVIRON. MICROBIOL.

Capture probe; Bacteria-specific original probe used (distant) Detector probe; Bacteria-specific original probe used (distant) Capture probe; Bacteria specific (proximate) Detector probe; Bacteria specific (proximate) Helper probe; Bacteria specific (proximate) Detector probe; Bacteria specific (proximate) Detector probe; AOB specific

ROWAN ET AL.

Function

8484

FIG. 2. Background signal from a range of controls relative to signal produced with low concentrations of rRNA. NB, NonBact probe control (nonsense probe used with all other reagents); Yeast, broadspecificity control. BO and PO are described in Materials and Methods. Relative light units represent the total integrated light output (light produced per second per well measured every 3 min for 1 h). Error bars, 95% confidence intervals.

blocking reagent was added in the hybridization reaction to reduce nonspecific probe binding. Optimal blocking of the beads was achieved using salmon testis DNA and Denhardt’s solution prior to the capture step. Additionally, in a modification of the method of Wicks et al. (46), we blocked with Denhardt’s solution alone after the capture step. Subsequently, salmon testis DNA and Denhardt’s solution were also included in the hybridization step, which is normal practice in many hybridization protocols (see, e.g., references 3 and 46). The signals from the bead-only and probe-only controls were reduced by the addition of a blocking reagent to the hybridization reaction (signal from PO without blocking, 2.11 ⫻ 109 ⫾ 0.021 ⫻ 109 RLU; signal from PO with blocking, 1.9 ⫻ 109 ⫾ 0.032 ⫻ 109 RLU [P ⬍⬍ 0.05]). (iv) Negative controls. In contrast to similar studies, we used a broad range of negative controls to ensure that the signal obtained during the assay was due to the specific interaction of the probes with the target rRNA. The negative controls included an oligonucleotide probe that should not hybridize to the target rRNA (NonBact338, commonly used in FISH [26]), BO (mentioned above), PO (mentioned above), and broadspecificity negative-control RNA (e.g., yeast RNA). The different negative-control procedures gave different levels of background signal (Fig. 2). The highest background was generally observed for the probe-only control and for control hybridizations containing nontarget rRNA (S. cerevisiae). The capture and detector probes were examined for complementarity, and a maximum match of 4 bp was observed (in the middles of the probe sequences). Hence, the probes are unlikely to hybridize with each other under the assay conditions. Nevertheless, the signal from probe-only controls was frequently higher than the signal from reaction mixtures containing low concentrations of target rRNA (108 to 109 molecules), which otherwise showed a smooth decrease in signal toward background with the bead-only control (see, e.g., Fig. 4A). The presence of any RNA (whether target or nontarget) in the reaction mixture apparently blocked non-

VOL. 71, 2005


8485

FIG. 3. Response from a twofold dilution series of target 16S rRNA molecules (E. coli 16S and 23S rRNA) using the RNA assay with two different probe sets. Open squares, capture probe S-D-Bact-0320-a-S-40-B and detector probe S-D-Bact-1524-a-S-17-D (1,221 bases apart; probes developed by Wicks [45]). Closed squares, capture probe S-D-Bact-0343-b-S-17-B and detector probe S-D-Bact-0518-a-S-18-D (159 bases apart), with unlabeled helper probe S-D-Bact-0320-b-S-18-H (probes developed in this study). Yeast, broad-specificity control. For BO and PO, see Materials and Methods. Relative light units represent the total integrated light output (light produced per second per well measured every 3 min for 1 h). Error bars, 95% confidence intervals.

specific binding of the probe to the paramagnetic beads, leading to a lower background signal. However, it must be noted that the PO control signal was inconsistent and highly variable, and the assay would never be used with samples that contain no RNA. Nucleic acid preparations may also contain trace amounts of naturally occurring biotin, or there may be nonspecific interactions between the biotin-labeled capture probe and the DIG-labeled detection probe, or between the DIG-labeled detection probe and the magnetic beads, when RNA is not present. This could give rise to the background signal seen in the PO negative control. However, the probe-only control may be inappropriate for the actual conditions in the RNA assay reaction, since any biological sample analyzed is likely to contain at least some RNA. It is interesting, however, that even with salmon testis DNA in the assay mix, some form of nonspecific binding of the detector probe to the beads occurred. This is all the more surprising given that the presence of nontarget RNA (yeast RNA in this case) markedly reduced the nonspecific binding of the detector probe (Fig. 2). This suggests that the blocking properties of RNA are quite different from those of DNA. Generally, the negative controls gave lower levels of signal than positive-control samples; however, appropriate replicate controls should always be included in order to discriminate the background signal from a genuine positive signal (21). (v) Enhancement of the signal. To increase the intensity of the signal, the concentrations of the assay components (e.g., beads, probes, anti-DIG, and CDP-Star) and the combinations of probes used were optimized. The optimized concentrations for each of the reagents were the equivalent of 500 AU (A600) for the magnetic beads, 1 ⫻ 1012 molecules of each probe, 1/500 dilution of the anti-digoxigenin–alkaline phosphatase preparation (Roche Diagnostics, Germany), and 100 ␮l of CDP-Star Chemiluminescent Substrate with Sapphire-II Enhancer (0.4 mM; Applied Biosystems, United Kingdom). The

lengths of the hybridization, capture, and detection stages in the assay have been optimized previously (45, 46), and the optimal conditions were confirmed in this study (optimized times, 30 min, 90 min, and 30 min for hybridization, capture, and detection, respectively). However, the most significant factor in increasing signal intensity was a change in the combination of probes used (Fig. 3). The probes initially employed were those used by Wicks et al. (46). The target sites for these probes were distant from each other on the rRNA molecule (capture probe S-D-Bact-0320-a-S-40-B [target region, bases 320 to 359 by E. coli numbering] and detector probe S-D-Bact1524-a-S-17-D [target region, bases 1525 to 1541 by E. coli numbering] [Table 1]). Indeed, the target site for probe S-DBact-1524-a-S-17-D is at the very end of the rRNA molecule, making the target region more prone to degradation by exoribonucleases. Additionally, the target sequences for probes S-D-Bact-0320-a-S-40-B and S-D-Bact-1524-a-S-17-D are not universally conserved. Furthermore, it has been reported previously that an important factor for achieving positive hybridization in microarrays using two probes is the proximity of the oligonucleotide probe target sites on the rRNA molecule (38). Therefore, capture, detector, and helper probes that targeted all bacteria were designed: some targeted adjacent locations on the 16S rRNA molecule, and others targeted regions distant from each other. Different combinations of these probes were examined (Table 1). When smaller detector, capture, or helper probes were used in combination, targeting positions in close proximity, an almost 10-fold increase in the signal intensity was obtained (Fig. 3). This was achieved either by using detector and capture probes directly adjacent to one another (the biotin-labeled capture probe S-D-Bact-0343-b-S-17-B and the DIG-labeled detector probe S-D-Bact-0320-b-S-18-D) or by using an alternative probe combination where an unlabeled helper probe hybridized between the capture and detector probe positions

8486

ROWAN ET AL.

(capture probe S-D-Bact-0343-b-S-17-B, detector probe S-DBact-0518-a-S-18-D, and unlabeled helper probe S-D-Bact0320-b-S-18-H). The increase in the signal intensity may be due to the fact that internal positions, which are less likely to be exposed to degradation by exonucleases, were targeted, as well as to the faster kinetics of hybridization of smaller probes. In a previous study using microarrays to analyze rRNA sequences, Small and coworkers (38) found that probes in close proximity (1 to 3 bp) produced a more specific and reliable signal. These workers suggested that this was due to secondarystructure interactions. Probes in close proximity are thought to open up the region of hybridization on the rRNA molecule, thus increasing the accessibility of the target site. Small et al. (38) suggested the inclusion of chaperone probes (cf. helper probes [16]) to prevent “incorrect” secondary structure. Helper probes should be designed to anneal at a similar or higher temperature than the capture and detector probes and to have similar or broader specificity than the capture and detector probes (42). However, in a similar study it was found that probes located in immediate proximity to one another resulted in a high degree of binding of probes to nontarget RNA sequences (9). This phenomenon was thought to be due to “base-stacking effects” (9). To reduce base-stacking effects, these workers used probes separated by 10 to 14 bp. In this study, we obtained similar signals with probes in close proximity (5-bp separation) and probes that were distant from each other (⬎100-bp separation), with no apparent increase in nonspecific binding to nontarget RNA. However, due to the close proximity of the set of adjacent probes used in this study, the S-D-Bact-0343-b-S-17-B capture probe and S-D-Bact-0320-b-S-18-D detector probe were labeled at opposite ends (the 5⬘ and 3⬘ ends, respectively) to minimize any steric effects between the label molecules. Optimization of hybridization conditions. To achieve optimal specificity and reduce the background signal, the hybridization conditions were optimized using probes targeting the majority of bacterial 16S rRNA sequences. This optimization procedure included the use of different temperatures, different formamide concentrations, and different temperature-ramping profiles. Hybridization was conducted at a range of temperatures (4 to 60°C) and formamide concentrations (0 to 60%), and the signal intensities were comparable under all conditions tested (approximately 7 ⫻ 109 RLU for 5 ⫻ 1011 molecules of RNA under all conditions), with little or no effect on specificity or background. Standard probe optimization curves (see, e.g., reference 40), analogous to nucleic acid temperature melting curves, were not observed under these conditions. However, ramping of the temperature from 95°C (denaturation temperature) to the hybridization temperature (50°C) appeared to improve sensitivity, producing a significantly higher signal (P ⬍ 0.05), significantly lower background (P ⬍ 0.05), and increased specificity (Table 2). Other researchers have also found, when quantifying RNA, that different temperatures and formamide concentrations have little or no effect on the hybridization signal. For example, Small et al. (38) found that the presence of formamide had no effect on signal enhancement in microarray assays but was crucial for specific hybridization. In contrast, Peplies and coworkers (32) found that the inclusion of formamide in microarray analyses led to fewer true-positive signals (due to lower signal in the presence of formamide) and more false-negative signals (target molecules which should produce

APPL. ENVIRON. MICROBIOL. TABLE 2. Effects of temperature ramping on the sensitivity and broad specificity of the assay Signal intensity (108 RLU) a Assay condition

With temp ramping

Without temp ramping

1 ⫻ 1012 molecules of rRNA 1 ⫻ 1010 molecules of rRNA Probe-only control Yeast rRNA negative control

94.2 ⫾ 6.07 13.1 ⫾ 2.16 1.45 ⫾ 0.275 1.5 ⫾ 0.675

31.6 ⫾ 1.69 2.67 ⫾ 0.122 3.87 ⫾ 0.527 2.67 ⫾ 0.33

a

Values are reported as means ⫾ standard deviations for three replicates.

signal but are prevented due to highly stringent conditions). Furthermore, in some studies optimal hybridization was achieved at room temperature (25°C) (8, 38). In this study, only the implementation of a temperatureramping protocol appeared to have a positive effect on the signal intensity and specificity when a bacterial rRNA-specific probe set was used. This may be due to the structure of the 16S rRNA molecule ex situ (rRNA molecules in situ form 3-dimensional interactions with ribosomal proteins and rRNA). rRNA extracted from cells by using the protocol described above are free of ribosomal proteins and their interactions. However, complex RNA-RNA interactions may still occur. Nonetheless, to improve target site accessibility, chemical or temperature denaturants (e.g., ethanol [5]) or helper probes (see, e.g., reference 38) (discussed above) could be used. The temperature-ramping protocol used here may serve the same purpose. By gradual reduction of the temperature from the denaturation temperature, the rRNA molecules may remain relatively open so that the probes can gain access and bind to specific target sites on the rRNA at their optimal hybridization temperature, while the possibility of nonspecific binding is reduced. At lower temperatures, when probes are more likely to bind nonspecifically, the accessibility of the site will be reduced as secondary structures re-form, and probes will be unable to bind to those sites. Assay sensitivity. The sensitivity of the assay was determined using the optimized conditions with bacterial probes (capture probe S-D-Bact-0343-b-S-17-B, detector probe S-D-Bact-0518-aS-18-D, and helper probe S-D-Bact-0320-b-S-18-H [Table 1]) and dilutions of E. coli rRNA (Roche; twofold dilutions, ranging from approximately 1011 to 107 16S rRNA molecules) (Fig. 4A) or activated-sludge RNA (twofold dilutions, ranging from approximately 1012 to 109 16S rRNA molecules) (Fig. 4B). The detection limit for the assay was defined as the minimum number of target molecules required to produce a mean signal significantly greater than background control signals. The detection limit with purified E. coli rRNA was lower than that with RNA from activated-sludge samples. For the standard E. coli rRNA, 5 ⫻ 108 molecules of 16S rRNA were detected above background; for the activated-sludge sample, 1 ⫻ 109 16S rRNA molecules were detected. These values correspond to approximately 103 to 104 E. coli cells and 104 to 105 bacterial cells, respectively, based on the assumptions that an actively growing bacterium contains 104 to 105 ribosomes (31) and that the rRNA was released with high efficiency during extraction. This level of sensitivity of the assay is similar to, or higher than, that with other quantification methods (for

VOL. 71, 2005


8487

FIG. 4. Response from a twofold dilution series of target 16S rRNA molecules from E. coli (A) and nitrifying activated sludge (B) using the RNA assay with a probe set consisting of probes S-D-Bact-0343-b-S-17-B, S-D-Bact-0518-a-S-18-D, and S-D-Bact-0320-b-S-18-H. Open diamonds, BO (see Materials and Methods). Relative light units represent the total integrated light output (light produced per second per well measured every 3 min for 1 h). Error bars, 95% confidence intervals.

microarrays, 109 to 1010 16S rRNA molecules [38]; for FISH, approximately 103 cells ml⫺1 in activated sludge [13, 41]). Furthermore, this level of detection is within the range of the average concentration of AOB cells in wastewater treatment plants (106 to 108 cells ml⫺1 [11, 19, 36]). Hence, the assay technique should be suitable for quantification of ammoniaoxidizing bacteria in WWTPs. However, the measured sensitivity of the assay was half an order of magnitude lower for the activated-sludge samples than for purified E. coli rRNA. The reason for this may be the more complex nature of the activated-sludge samples, which may lead to problems with probe binding and RNA extraction. RNA extracted from activatedsludge samples was obtained using a crude extraction procedure with little purification. A more intensive cleanup procedure of the extracted RNA may lower the detection limit. However, extensive purification may also lead to a loss of target RNA molecules. Moreover, the RNA extracted from the activated-sludge samples contains RNA from bacteria, archaea, and eukaryotes. Therefore, the concentrations of bacterial RNA, determined spectrophotometrically, for the activatedsludge samples were probably conservative overestimates, which would otherwise lead to detection limits lower than those determined above. Specificity of the assay. The specificity of the bacterial probes with respect to distantly related organisms was initially examined by hybridization with excess yeast rRNA (according to reference 46). The yeast rRNA background control was within the range of other background controls (Fig. 2). The potential of the assay to differentiate between rRNA molecules with very similar sequences at the probe target sites was examined by comparing the results obtained, using an AOB-specific detector probe set (detector probe S-F-bAOB1224-a-S-20-D, competitor probe S-S-Ral-1224-a-S-20-C, capture probe S-D-Bact-0343-b-S-17-B, and helper probe S-DBact-0320-b-S-18-H [Table 1]), with RNAs extracted from cultures of Nitrosomonas europaea NCIMB 11850T (AOB; pos-

itive control) and Ralstonia eutropha DSM 17697T (non-AOB: negative control with 2 mismatches at the probe target site). No non-AOB cultured isolates that have a 1-bp mismatch at the probe target site were available. Hence, R. eutropha, which has 2 mismatches at the probe target site (C instead of A at position 1232 and U instead of C at position 1243), was used as a negative control. To ensure that the amounts of 16S rRNA for the positiveand negative-control assays were comparable, the signal from the 16S rRNA was measured using the general bacterial probe set (capture probe S-D-Bact-0343-b-S-17-B and detector probe S-D-Bact-0320-b-S-18-D). This control measured the total concentration of bacterial 16S rRNA in the cultures, assuming that the probes hybridized with equal efficiency to the different rRNA species. The same concentration of RNA (3.5 ⫻ 1011 molecules, determined spectrophotometrically) in the positiveand negative-control samples gave similar signals by using the bacterial probe set in the RNA assay (e.g., N. europaea bacterial-probe signal, 9.9 ⫻ 108 ⫾ 0.46 ⫻ 108 RLU; R. eutropha bacterial-probe signal, 1.42 ⫻ 109 ⫾ 0.654 ⫻ 109 RLU). However, the AOB-specific detector probe set produced a statistically greater signal in the assay with RNA extracted from the N. europaea culture (3.56 ⫻ 108 ⫾ 0.3 ⫻ 108 RLU) than in that with RNA from the R. eutropha culture (2.24 ⫻ 108 ⫾ 0.17 ⫻ 108 RLU [Fig. 5] [P ⬍⬍ 0.001]). The signal intensity of the negative-control assay containing yeast RNA was similar to that for the nontarget rRNA that had only 2 mismatches with the specific detector probe. This indicates that the assay has a good level of specificity in the presence of nontarget RNA (Fig. 5 and 6). However, the signal produced by the AOB-specific probe set (3.56 ⫻ 108 ⫾ 0.3 ⫻ 108 RLU) was less than the signal generated with the bacterial probe set (9.9 ⫻ 108 ⫾ 0.46 ⫻ 108 RLU), possibly indicating that the target site for the AOBspecific detector probe was relatively inaccessible. To overcome this problem, helper probes (16, 32, 38) could be in-

8488

ROWAN ET AL.

FIG. 5. Specificity of the RNA assay with RNAs from positive and negative controls. AOB, Nitrosomonas europaea NCIMB 11850T (positive control; 5 ⫻ 1012 molecules); NonAOB, Ralstonia eutropha DSM 17697T (negative-control culture; 5 ⫻ 1012 molecules); Yeast, negative control containing yeast RNA. Probes used were S-F-bAOB-1224-aS-20-D, S-S-Ral-1224-a-S-20-C, S-D-Bact-0343-b-S-17-B, and S-DBact-0320-b-S-18-H. Relative light units represent the total integrated light output (light produced per second per well measured every 3 min for 1 h). Error bars, 95% confidence intervals.

cluded to open this target region, or, alternatively, probes that target more-accessible regions of the rRNA molecule could be designed. The AOB-specific probes employed in this study have been used successfully in numerous FISH-based studies (see, e.g., references 11, 12, 17, and 37). No difference in signal has been reported between AOB-specific probes and bacterial probes in previous studies, although quantitative comparative analysis of the signal intensities produced from bacterial and AOB-specific probes of the same cells was not undertaken in these studies. It is also possible that rRNA may behave differently in ex situ versus in situ (e.g., FISH) hybridization with respect to the relationship between secondary structure and accessibility to probes. The specificity of the assay could be further increased by using both capture and detector probes that target AOB rRNA sequences. When two AOB-specific probes (capture probe S-F-bAOB-0190-a-S-22-B and detector probe S-F-bAOB1224-a-S-20-D, with competitor probe S-S-Ral-1224-a-S-20-C [Table 1]) were used with rRNA from pure cultures, a more intense signal was obtained than in an analysis using a probe set where only the detector probe was specific for the target sequence (detector probe S-F-bAOB-1224-a-S-20-D, competitor probe S-S-Ral-1224-a-S-20-C, capture probe S-D-Bact0343-b-S-17-B, and helper probe S-D-Bact-0320-b-S-18-H [Table 1]), but the results were more variable (data not shown). Use of the assay to discriminate between the abundances of AOB in WWTPs with different levels of nitrification. To determine the effectiveness of the assay for quantitative discrimination of AOB populations in WWTPs exhibiting differences in nitrification, RNA extracts from activated sludge (approximately 4 to 6 ml) from two different WWTPs were analyzed using the assay. The same total number of rRNA molecules was used for both samples in the assay. WWTP A exhibited


FIG. 6. Specificity of the RNA assay with RNA from nitrifying and nonnitrifying activated-sludge samples and a negative control. WWTP A, nitrifying wastewater treatment sample; WWTP B, nonnitrifying wastewater treatment sample; Yeast, negative control containing yeast RNA. Probes used were S-F-bAOB-1224-a-S-20-D, S-S-Ral-1224-a-S20-C, S-D-Bact-0343-b-S-17-B, and S-D-Bact-0320-b-S-18-H. Relative light units represent the total integrated light output (light produced per second per well measured every 3 min for 1 h). Error bars, 95% confidence intervals.

99.7% ammonia removal, and the second plant, WWTP B, had poor nitrification performance (average 23% ammonia removal). A slightly, but statistically significantly, higher signal was obtained with RNA extracted from the actively nitrifying wastewater treatment reactor (WWTP A) than with RNA from the nonnitrifying reactor (WWTP B) when the bacterium-specific capture and AOB-specific detector probe set was used (Fig. 6) (P ⬍ 0.1). The signal from WWTP B was comparable to that of the negative-control assay containing yeast RNA. FISH analysis of the same samples revealed that WWTP A contained an abundant population of AOB, whereas AOB could not be detected in WWTP B by using FISH. The RNA assay that we have developed was able to clearly discriminate between the sizes of AOB populations in nitrifying and nonnitrifying activated sludge; however, there are limitations associated with any assay based on the detection of RNA. These relate to the efficiency of RNA extraction (e.g., RNA purification, cell lysis, and stability of RNA), sensitivity and probe specificity issues, and/or target site accessibility. However, perhaps the most important limitation of RNA assays is the potential difficulty in correlating rRNA abundance with cell numbers, since the number of ribosomes may differ between different bacterial species and may also differ with the growth rate or physiological status of an organism (22, 23, 31). Therefore, measurements of rRNA may reflect the activity of the organisms rather than the numbers of organisms. Nonetheless, high cellular ribosome levels (measured as fluorescence by FISH) have been observed in AOB cells that were not actively nitrifying (37), were metabolically inhibited (6, 40), or were starved (29). Furthermore, there is evidence to suggest that the abundance of AOB as measured by dot blot hybridization of extracted RNA (35) and FISH (11) correlates with process- and model-based predictions of the sizes of AOB populations (35). Therefore, for AOB, at least under some circumstances, the rRNA content may accurately reflect cell numbers.

VOL. 71, 2005


Conclusions. To accurately monitor microbial populations, especially those involved in nutrient removal processes such as nitrification, a rapid and simple quantification method is required. Current cultivation-independent methods for quantifying bacteria are generally slow (FISH), technically complex (PCR, microarrays, FISH), and often expensive (PCR, microarrays, FISH). Thus, the development of simpler and faster technologies, such as real-time PCR (see, e.g., reference 15) or, alternatively, RNA hybridization assays, is welcome. These methods, perhaps calibrated against FISH, will be essential if quantitative analysis of specific bacteria in activated sludge or other environments is to be adopted by routine analytical laboratories, or for generating very long time series, which will be essential if we are to determine natural variation in the sizes of crucial bacterial populations. Situations outside this normal variation may signal impending process failure. In this study we have developed, to proof of principle, a moderately sensitive, specific, and rapid (about 1 day) assay for determining the relative abundance of ammonia-oxidizing bacteria in WWTPs in order to assess nitrification performance. The general trends in populations that could be provided by the assay for relative abundance measurements may suffice for this type of monitoring. The format of the assay is applicable to quantification of other microorganisms in many environments and has the potential to be automated. ACKNOWLEDGMENTS We are grateful to AstraZeneca Global SHE and Yorkshire Water Services for funding the project and supplying samples. R.J.D. also acknowledges funding by Engineering and Physical Science Research grants GR/N25855 and GR/R82159/01 toward the project. REFERENCES 1. Adamczyk, J., M. Hesselsoe, N. Iversen, M. Horn, A. Lehner, P. H. Nielsen, M. Schloter, P. Roslev, and M. Wagner. 2003. The isotope array, a new tool that employs substrate-mediated labeling of rRNA for determination of microbial community structure and function. Appl. Environ. Microbiol. 69: 6875–6887. 2. Alm, E., D. B. Oerther, N. Larsen, D. A Stahl, and L. Raskin. 1996. The oligonucleotide probe database. Appl. Environ. Microbiol. 62:3557–3559. 3. Amann, R. I., L. Krumholz, and D. A. Stahl. 1990. Fluorescent-oligonucleotide probing of whole cells for determinative, phylogenetic, and environmental studies in microbiology. J. Bacteriol. 172:762–770. 4. Ashelford, K. E., A. J. Weightman, and J. C. Fry. 2002. PRIMROSE: a computer program for generating and estimating the phylogenetic range of 16S rRNA oligonucleotide probes and primers in conjunction with the RDP-II database. Nucleic Acids Res. 30:3481–3489. 5. Behrens, S., B. M. Fuchs, F. Mueller, and R. Amann. 2003. Is the in situ accessibility of the 16S rRNA of Escherichia coli for Cy3-labeled oligonucleotide probes predicted by a three-dimensional structure model of the 30S ribosomal subunit? Appl. Environ. Microbiol. 69:4935–4941. 6. Bouchez, T., D. Patureau, P. Dabert, S. Juretschko, J. Dore, P. Delgenes, R. Moletta, and M. Wagner. 2000. Ecological study of a bioaugmentation failure. Environ. Microbiol. 2:179–190. 7. Box, G. E. P., and D. R. Cox. 1964. An analysis of transformations. J. R. Stat. Soc. 26:211–243. 8. Chandler, D. P., and A. E. Jarrell. 2003. Enhanced nucleic acid capture and flow cytometry detection with peptide nucleic acid probes and tunablesurface microparticles. Anal. Biochem. 312:182–190. 9. Chandler, D. P., G. J. Newton, J. A. Small, and D. S. Daly. 2003. Sequence versus structure for the direct detection of 16S rRNA on planar oligonucleotide microarrays. Appl. Environ. Microbiol. 69:2950–2958. 10. Cole, J. R., B. Chai, T. L. Marsh, R. J. Farris, Q. Wang, S. A. Kulam, S. Chandra, D. M. Mcgarrell, T. M. Schmidt, G. M. Garrity, and J. M. Tiedje. 2003. The Ribosomal Database Project (RDP-II): previewing a new autoaligner that allows regular updates and the new prokaryotic taxonomy. Nucleic Acids Res. 31:442–443. 11. Coskuner, G., S. J. Ballinger, R. J. Davenport, R. L. Pickering, R. R. Solera, I. M. Head, and T. P. Curtis. 2005. Agreement between theory and measurement in the quantification of ammonia oxidizing bacteria. Appl. Environ. Microbiol. 71:6325–6334.

8489

12. Daims, H., U. Purkhold, L. Bjerrum, E. Arnold, P. A. Wilderer, and M. Wagner. 2001. Nitrification in sequencing biofilm batch reactors: lessons from molecular approaches. Water Sci. Technol. 43(3):9–18. 13. Davenport, R. J., and T. P. Curtis. 2002. Are filamentous mycolata important in foaming? Water Sci. Technol. 46(1-2):529–533. 14. de los Reyes, F. L., D. B. Oerther, M. A. de los Reyes, M. Hernandez, and L. Raskin. 1998. Characterization of filamentous foaming in activated sludge systems using oligonucleotide hybridisation probes and antibody probes. Water Sci. Technol. 37(4-5):485–493. 15. Dionisi, H. M., A. C. Layton, G. Harms, I. R. Gregory, K. G. Robinson, and G. S. Sayler. 2002. Quantification of Nitrosomonas oligotropha-like ammonia-oxidizing bacteria and Nitrospira spp. from full-scale wastewater treatment plants by competitive PCR. Appl. Environ. Microbiol. 68:245–253. 16. Fuchs, B. M., F. O. Glockner, J. Wulf, and R. Amann. 2000. Unlabeled helper oligonucleotides increase the in situ accessibility to 16S rRNA of fluorescently labeled oligonucleotide probes. Appl. Environ. Microbiol. 66: 3603–3607. 17. Gieseke, A., U. Purkhold, M. Wanger, R. Amann, and A. Schramm. 2001. Community structure and activity dynamics of nitrifying bacteria in a phosphate-removing biofilm. Appl. Environ. Microbiol. 67:1351–1362. 18. Guschin, D. Y., B. K. Mobarry, D. Proudnikov, D. A. Stahl, B. E. Rittmann, and A. D. Mirzabekov. 1997. Oligonucleotide microchips as genosensors for determinative and environmental studies in microbiology. Appl. Environ. Microbiol. 63:2397–2402. 19. Harms, G., A. C. Layton, H. M. Dionisi, I. R. Gregory, V. M. Garrett, S. A. Hawkins, K. G. Robinson, and G. S. Sayler. 2003. Real-time PCR quantification of nitrifying bacteria in a municipal wastewater treatment plant. Environ. Sci. Technol. 37:343–351. 20. Head, I. M., J. R. Saunders, and R. W. Pickup. 1998. Microbial evolution, diversity, and ecology: a decade of ribosomal RNA analysis of uncultivated organisms. Microb. Ecol. 35:1–21. 21. Hugenholtz, P., G. W. Tyson, and L. L. Blackall. 2002. Design and evaluation of 16S rRNA-targeted oligonucleotide probes for fluorescence in situ hybridization. Methods Mol. Biol. 179:29–42. 22. Kerkhof, L., and B. B. Ward. 1993. Comparison of nucleic acid hybridization and fluorometry for measurement of the relationship between RNA/DNA ratio and growth rate in a marine bacterium. Appl. Environ. Microbiol. 59:1303–1309. 23. Kerkhof, L., and P. Kemp. 1999. Small ribosomal RNA content in a marine Proteobacteria during non-steady-state growth. FEMS Microbiol. Ecol. 30: 253–260. 24. Koizumi, Y., J. R. Kelly, T. Nakagawa, H. Urakawa, S. El-Fantroussi, S. Al-Muzaini, M. Fukai, Y. Urushigawa, and D. A. Stahl. 2002. Parallel characterization of anaerobic toluene- and ethylbenezene-degrading microbial consortia by PCR-denaturing gradient gel electrophoresis, RNA-DNA membrane hybridization, and DNA microarray technology. Appl. Environ. Microbiol. 68:3215–3225. 25. Kowalchuk, G. A., J. R. Stephen, W. Deboer, J. I. Prosser, T. M. Embley, and J. W. Woldendorp. 1997. Analysis of ammonia-oxidizing bacteria of the beta subdivision of the class Proteobacteria in coastal sand dunes by denaturing gradient gel electrophoresis and sequencing of PCR-amplified 16S ribosomal DNA fragments. Appl. Environ. Microbiol. 63:1489–1497. 26. Manz, W., R. W. Amann, M. Ludwig, M. Wagner, and K. H. Schleifer. 1992. Phylogenetic oligodeoxynucleotide probes for the major subclasses of Proteobacteria: problems and solutions. Syst. Appl. Microbiol. 15:593–600. 27. Manz, W., M. Wagner, R. Amann, and K.-H. Schleifer. 1994. In situ characterisation of the microbial consortia active in two wastewater treatment plants. Water Res. 28:1715–1723. 28. Mobarry, B. K., M. Wagner, V. Urbain, B. E. Rittman, and D. A. Stahl. 1996. Phylogenetic probes for analyzing abundance and spatial organization of nitrifying bacteria. Appl. Environ. Microbiol. 62:2156–2162. 29. Morgenroth, E., A. Obermayer, E. Arnold, A. Bru ¨hl, M. Wagner, and P. A. Wilderer. 2000. Effect of long term idle periods on the performance of sequencing batch reactors. Water Sci. Technol. 41(1):105–113. 30. Nagata, Y., H. Yokoto, O. Kosuda, K. Yokoo, K. Takemura, and T. Kikuchi. 1985. Quantification of picogram levels of specific DNA immobilized in microtiter wells. FEBS Lett. 183:379–382. 31. Neidhardt, F. C., J. L. Ingraham, and M. Schaechter. 1990. Physiology of the bacterial cell. A molecular approach. Sinauer, Sunderland, Mass. 32. Peplies, J., F. O. Glo ¨ckner, and R. Amann. 2003. Optimization strategies for DNA microarray-based detection of bacteria with 16S rRNA-targeting oligonucleotides. Appl. Environ. Microbiol. 69:1397–1407. 33. Phillips, C. J., E. A. Paul, and J. I. Prosser. 2000. Quantitative analysis of ammonia-oxidising bacteria using competitive PCR. FEMS Microbiol. 32: 167–175. 34. Raskin, L., L. K. Poulsen, D. R. Noguera, B. E. Rittmann, and D. A. Stahl. 1994. Quantification of methanogenic groups in anaerobic biological reactors by oligonucleotide probe hybridization. Appl. Environ. Microbiol. 60: 1241–1248. 35. Rittmann, B. E., C. S. Laspidou, J. Flax, D. A. Stahl, V. Urbain, H. Harduin, J. J. Van Der Waarde, B. Geurkink, M. J. C. Henssen, H. Brouwer, A. Klapwijk, and A. Wetterauwm. 1999. Molecular and modelling analyses of

8490

ROWAN ET AL.

the structure and function of nitrifying activated sludge. Water Sci. Technol. 39:51–59. 36. Robinson, K. G., H. M. Dionisi, G. Harms, A. C. Layton, I. R. Gregory, and G. S. Sayler. 2003. Molecular assessment of ammonia- and nitrite-oxidising bacteria in full-scale activated sludge wastewater treatment plants. Water Sci. Technol. 48:119–126. 37. Schramm, A., D. De Beer, J. C. van den Heuval, S. Ottengraf, and R. Amann. 1999. Microscale distribution of populations and activities of Nitrosospira and Nitrospira spp. along a microscale gradient in a nitrifying bioreactor: quantification by in situ hybridization and use of microsensors. Appl. Environ. Microbiol. 65:3690–3696. 38. Small, J., D. R. Call, F. J. Brockman, T. M. Straub, and D. P. Chandler. 2001. Direct detection of 16S rRNA in soil extracts by using oligonucleotide microarrays. Appl. Environ. Microbiol. 67:4708–4716. 39. Suzuki, M. T., and S. J. Giovannoni. 1996. Bias caused by template annealing in the amplification of mixtures of 16S rRNA genes by PCR. Appl. Environ. Microbiol. 62:625–630. 39a.von Wintzingerode, F., U. B. Gobel, and E. Stackebrandt. 1997. Determination of microbial diversity in environmental samples: pitfalls of PCR-based rRNA analysis. FEMS Microbiol. Rev. 21:213–229. 40. Wagner, M., G. Rath, R. Amann, H.-P. Koops, and K.-H. Schleifer. 1995. In


41.

42.

43.

44.

45. 46.

situ identification of ammonia-oxidising bacteria. Syst. Appl. Microbiol. 18: 251–264. Wagner, M., A. Loy, N. Nogueira, U. Purkhold, N. Lee, and H. Daims. 2002. Microbial community composition and function in wastewater treatment plants. Antonie Leeuwenhoek 81:665–680. Wagner, M., M. Horn, and D. Daims. 2003. Fluorescent in situ hybridisation for the identification and characterisation of prokaryotes. Curr. Opin. Microbiol. 6:302–309. Wang, G. G. Y., and Y. Wang. 1996. The frequency of chimeric molecules as a consequence of PCR co-amplification of 16S rRNA genes from different bacterial species. Microbiology 142:1107–1114. Watson, S. W., and M. Mandel. 1971. Comparison of the morphology and deoxyribonucleic acid composition of 27 strains of nitrifying bacteria. J. Bacteriol. 107:194–201. Wicks, B. 1997. Ph.D. thesis. University of Newcastle upon Tyne, Newcastle upon Tyne, United Kingdom. Wicks, B., D. B. Cook, M. R. Barer, A. G. O’Donnell, and C. H. Self. 1998. A sandwich hybridization assay employing enzyme amplification for determination of specific ribosomal RNA from unpurified cell lysates. Anal. Biochem. 259:258–264.