|Home | About | Journals | Submit | Contact Us | Français|
Chloramination is often the disinfection regimen of choice for extended drinking water systems. However, this process is prone to instability due to the growth of nitrifying bacteria. This is the first study to use alternative approaches for rapid investigation of chloraminated drinking water system instability in which flow cytometric cell sorting of bacteria with intact membranes (membrane-intact fraction) (BacLight kit) or with active esterases (esterase-active fraction) (carboxyfluorescein diacetate) was combined with 16S rRNA gene-directed PCR and denaturing gradient gel electrophoresis (DGGE). No active bacteria were detected when water left the water treatment plant (WTP), but 12 km downstream the chloramine residual had diminished and the level of active bacteria in the bulk water had increased to more than 1 × 105 bacteria ml−1. The bacterial diversity in the system was represented by six major DGGE bands for the membrane-intact fraction and 10 major DGGE bands for the esterase-active fraction. PCR targeting of the 16S rRNA gene of chemolithotrophic ammonia-oxidizing bacteria (AOB) and subsequent DGGE and DNA sequence analysis revealed the presence of an active Nitrosospira-related species and Nitrosomonas cryotolerans in the system, but no AOB were detected in the associated WTP. The abundance of active AOB was then determined by quantitative real-time PCR (qPCR) targeting the amoA gene; 3.43 × 103 active AOB ml−1 were detected in the membrane-intact fraction, and 1.40 × 104 active AOB ml−1 were detected in the esterase-active fraction. These values were several orders of magnitude greater than the 2.5 AOB ml−1 detected using a routine liquid most-probable-number assay. Culture-independent techniques described here, in combination with existing chemical indicators, should allow the water industry to obtain more comprehensive data with which to make informed decisions regarding remedial action that may be required either prior to or during an instability event.
For extended drinking water systems, chloramine (in particular, monochloramine) is often the preferred disinfectant. Chloramine is less reactive than free chlorine, maintains an extended disinfection residual (45), and produces lower concentrations of disinfection by-products, such as trihalomethanes and haloacetic acids (5, 23, 28). Despite these advantages, the use of chloramine can introduce ammonia into a distribution system, both as excess ammonia from chloramine formation and as released ammonia from chloramine decay. These factors may lead to biological instability in such systems due to biological nitrification by nitrifying bacteria, such as the chemolithotrophic ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB) (34, 35).
Nitrification episodes in chloraminated distribution systems are common, and there is evidence that 63% of medium and large utilities in the United States have experienced episodes of nitrification (45). Cunliffe (6) studied five chloraminated distribution systems in South Australia and found that 64% of the samples contained nitrifying bacteria and that nitrification commonly occurred at the ends of distribution systems. As outlined by Wilczak et al. (45) and Odell et al. (29), several water quality and treatment factors affect nitrification. The principal cause is the presence of free ammonia, and nitrification has been shown to occur even at trace levels (50 μg liter−1). Certain infrastructure conditions, such as long detention times, pipeline biofilm, and absence of sunlight, may also promote nitrification. Nitrification usually occurs in a pH range from 6.5 to 10.0 and when temperatures exceed 15°C (45). While a chloramine residual is effective for preventing growth of most bacteria in drinking water systems, biological nitrification is often reported in the presence of high chloramine residuals (40), such as 5.0 mg liter−1 monochloramine (6).
AOB are involved in the initial phase of biological nitrification, but due to their chemolithotrophic metabolism they are slow growing, with generation times between 8 h and several days (44). This makes traditional culture techniques for detection of these bacteria very time-consuming (22). AOB oxidize ammonia to a hydroxylamine intermediate, which is catalyzed by ammonia monooxygenase (4). This enzyme is comprised of three subunits, and amoA encodes the subunit carrying the active site (37). This gene has been used as a molecular target for primers and probes in studies of AOB communities in soils and sediments and freshwater and marine environments and in studies of wastewater treatment, as reported in numerous papers (4, 10, 16, 30, 32, 33, 43). From the hydroxylamine intermediate, AOB then produce nitrite. Excess nitrite increases chloramine demand, often eliminating the residual, and there is a decrease in dissolved oxygen and often an increase in the number of heterotrophic bacteria (19, 29). Nitrite also has some associated health risks as it can react with secondary amines to produce carcinogenic compounds, such as nitrosamines (19). The disappearance of nitrite from distribution systems is most often attributed to chemical oxidation by chloramine (20), which in turn produces more ammonia and nitrate. However, biological oxidation of nitrite to nitrate may occur if active NOB are present (7, 35), which completes the biological nitrification process.
Since nitrification occurs as a result of several coinciding factors, monitoring of a system for the occurrence of nitrification is performed utilizing a collection of indicators. These indicators include the chemical indicators chloramine residual, oxidized nitrogen (nitrite and nitrate), dissolved oxygen, and the concentration of free and total ammonia. In addition, biological indicators determined by culture-based technologies are commonly assessed, and these include the number of bacteria culturable by routine heterotrophic plate counting (HPC) and the abundance of AOB detected using routine liquid most-probable-number (MPN) assays (6, 19, 29, 45).
Advances in fluorescent dye technology have created new opportunities for assessment of individual physiological activity traits (31). Such dyes have become valuable tools for detection of bacteria in aquatic environments, especially when they are combined with flow cytometry (FCM) for rapid enumeration of active bacteria and cell sorting for subsequent molecular characterization (39). We recently described a new approach for monitoring the efficacy of water treatment. This approach combined flow cytometric cell sorting of active bacteria from various stages of a chlorinated water treatment process with subsequent universal nested 16S rRNA gene-directed PCR of DNA extracts, followed by denaturing gradient gel electrophoresis (DGGE) (13). A combination of flow cytometric cell sorting with subsequent molecular analysis, including DGGE, has also been used for analyzing the taxonomic composition of marine bacterioplankton (9). Based on our previous study it was envisaged that an additional application of this approach could be assessment of distribution system instability by detection and identification of active bacteria recalcitrant to routine HPC culture (13). Such bacteria include the AOB, whose detection by routine MPN assays requires a minimum of 3 weeks of incubation (22), and the results are often obtained long after corrective action has been or needs to be taken.
In this study, flow cytometric cell sorting coupled with PCR-DGGE was used to study a water treatment plant (WTP) and distribution system employing chloramination during an instability event. Two 16S rRNA gene-directed PCR assays were used, one that allows universal detection of eubacteria and one that allows specific detection of chemolithotrophic AOB. The universal bacterial assay was used to characterize changes in active bacteria through water treatment and to detect distribution system instability due to the regrowth of bacteria. The AOB assay was used to rapidly confirm that these organisms were actively present in the distribution system but not in the WTP, and DNA sequencing allowed identification of these bacteria. Liquid MPN culture of AOB has previously been shown to underestimate total AOB numbers (15, 42); however, little is known regarding the accuracy of the MPN technique for estimating the number of physiologically active AOB present during nitrification in chloraminated distribution systems. This was also investigated for the system studied by using quantitative real-time PCR (qPCR) targeting amoA for active bacterial fractions previously sorted by FCM.
The Loxton WTP (Loxton, Riverland, South Australia, Australia) uses a conventional treatment process consisting of coagulation (the primary coagulants include aluminum sulfate and polyelectrolyte), followed by flocculation, sedimentation, rapid gravity filtration using beds of sand and filter coal, and finally chloramination. For 2 days prior to sampling, prechlorination was employed in the first of the rapid mixers before filtration. Ten-liter samples of the following types of water were taken on the same day: (i) raw water from the raw water sampling tap; (ii) filtered water from the filter water sampling tap (no chlorine residual was present following prechlorination); (iii) finished water from the product water sampling tap; and (iv) distribution system water obtained approximately 12 km downstream of the WTP. Neither examination nor sampling of the pipe biofilm was possible in this study. Water chemistry data are shown in Table Table11.
The total bacteria (both active and inactive) were stained with SYTO-9 (final concentration, 2.0 μmol l−1; Molecular Probes, Eugene, OR) for 15 min at room temperature in the dark for subsequent FCM enumeration. The active process of maintaining bacterial membrane integrity was assessed by staining bacteria with a BacLight bacterial viability kit (Molecular Probes) as described previously (12). Bacteria with intracellular esterase activity (esterase-active bacteria) were detected using carboxyfluorescein diacetate (CFDA) (Molecular Probes) as described previously (11). Subsequent FCM enumeration and cell sorting were performed separately for the active bacteria using a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA) as described previously (13).
Total bacteria were isolated from separate 100-ml portions of raw water and filtered water by centrifugation as reported previously (13). Total bacterial DNA in finished water and distribution system bulk water was isolated from 1-liter samples using an UltraClean water DNA kit (MoBio Laboratories Inc., Solana Beach, CA) by following the manufacturer's instructions.
HPC was performed in triplicate by the spread plate technique using a 0.1-ml sample (or an appropriate dilution) on either R2A (Oxoid, Heidelberg, Australia) or tryptone soy agar (TSA) (Oxoid). Incubation was performed separately using standard conditions (35°C for 48 h and 20°C for 72 h) in accordance with Australian standard AS/NZS 4276.3.1 (41). Following incubation, colonies were visualized using an illuminated magnified colony counter (magnification, ×2.5). The HPC-culturable fractions of each type of water were obtained for subsequent DGGE profiling (see below) by colony homogenization on the agar surface as reported previously (13).
Enumeration of AOB by routine an MPN assay was performed using an 11-tube assay format (1 1-ml tube, 5 0.1-ml tubes, and 5 0.01-ml tubes; series 3/2) with AOB medium (5 mg liter−1 ammonium sulfate [BDH Chemicals], 0.4 mg liter−1 magnesium sulfate [BDH Chemicals], 20 mg liter−1 potassium dihydrogen orthophosphate [BDH Chemicals], 0.4 mg liter−1 calcium chloride [BDH Chemicals], 0.3 mg liter−1 ferric citrate [Sigma], 0.01 mg liter−1 phenol red). Following inoculation, the tubes were wrapped in aluminum foil, placed in a sealed opaque container, and incubated at 30°C for 28 days. Following incubation, the presence of AOB was assessed by a change in the color of the medium from pink to yellow as a result of acid production. Calculations were performed using a McGrady MPN table. MPN tubes that exhibited positive results were pooled, and cells were concentrated by centrifugation at 2,500 × g for 1 h at 4°C. Following centrifugation, less than 100 bacteria ml−1 (the detection limit of the FCM method) remained in the aspirated supernatant, as determined by SYTO-9 staining and FCM enumeration. Cells were resuspended in 1 ml sterile Milli-Q water, and the total DNA was extracted (see below) for subsequent molecular analyses.
DNAs from the total fraction, active fractions (bacteria with intact membranes [membrane-intact bacteria] and esterase-active bacteria), HPC-culturable fraction, and AOB fraction were extracted for subsequent molecular analyses by rapid freeze-thawing (14). Cell suspensions were subjected to three cycles of boiling at 100°C for 5 min, followed by freezing in liquid nitrogen for 2 min.
Bacterial DNAs from total, active (membrane-intact and esterase-active), and HPC-culturable fractions were amplified by universal nested 16S rRNA gene-directed PCR prior to DGGE analysis. The 27F-1492R primer set was used for the first round of PCR, and this was followed by nested amplification using the 357F-GC-518R primer set, as reported previously (13).
An AOB-specific nested 16S rRNA gene-directed PCR was performed for all fractions from each sample. The first-round PCR (consisting of 30 cycles) was specific for AOB and was performed with primer set CTO as described by Kowalchuk et al. (18). The product of the first-round AOB-specific reaction was used as a template for the nested PCR, which was performed using the 357F-GC-518R universal bacterial primer set as reported previously (13). Use of the 357F-GC-518R primer set for nested amplification following the AOB-specific PCR allowed side-by-side DGGE gel comparisons with the universal eubacterial nested 16S rRNA gene-directed PCR described above.
DNA from Nitrobacter sp. was detected by PCR using the EUB338f-NIT3 primer set, and DNA from Nitrospira sp. was detected by PCR using the EUB338f-Ntspa0685M primer set, as described by Regan et al. (34). Cloning of the PCR products from the Nitrobacter- and Nitrospira-specific reactions followed by DNA sequence analysis (see below) verified that target sequences were amplified (data not shown).
DGGE was performed using a Bio-Rad D-GENE denaturing gel electrophoresis system (Bio-Rad, Hercules, CA) and conditions described previously (13). DGGE reproducibility was confirmed by performing five independent PCR-DGGE analyses with equivalent raw and distribution water samples (data not shown). Cluster analysis was used to investigate the relationships between DGGE profiles with the software program Molecular Evolutionary Genetics Analysis (MEGA, version 2.1), as reported previously (13).
PCR products were cloned into Escherichia coli using a TOPO TA cloning kit (Invitrogen, California) by following the manufacturer's instructions. To verify that bands on a DGGE gel were equivalent to bands of the cloned inserts, reamplification of the plasmid inserts was performed with the 357F-GC-518R primer set, and the products were analyzed by DGGE by comparison with the original nested PCR-DGGE profile. When required, the DNA sequences of PCR products were determined using a BigDye version III sequencing kit (Applied Biosystems) and were analyzed with a Perkin-Elmer 3700 capillary DNA sequence analyzer (Applied Biosystems); the sequence analysis was performed as described previously (13).
The qPCR was optimized and used to analyze the bacterial population in the bulk water of the distribution system. For the universal assay, DNA standards were prepared by dilution of a known amount of PCR product previously amplified using the 27F-1492R primer set. For the AOB-specific assay, DNA standards were prepared by dilution of a known amount of PCR product previously amplified from Nitrosospira multiformis ATCC 25196, using the amoA1F-amoA2R primer set (37). Before each experiment, the stock PCR product concentrations were determined spectrophotometrically, and the preparations were diluted accurately to obtain 1.0 × 107 copies μl−1. From these preparations 10-fold serial dilutions were prepared using a CAS-1200 liquid handling system (Corbett Research) with nuclease-free water to obtain a concentration of 1 copy μl−1.
qPCR was performed as follows. The PCR was performed with a Rotor Gene 3000 thermal cycling system (Corbett Research), in which samples and standards were added to aliquots of the master mixture using a CAS-1200 liquid handling system. Each reaction was performed in quadruplicate, and each 20-μl reaction mixture contained each deoxynucleoside triphosphate at a concentration of 200 μM, each primer at a concentration of 1.0 μM, 2.5 mM MgCl2, 1× PCR buffer II, 5% (vol/vol) dimethyl sulfoxide, 3.3 μM SYTO-9 (24), 2.5 U of Ampli Taq Gold DNA polymerase, and 5 μl of either a DNA standard or a sample. For eubacterial qPCR, the thermal cycle consisted of an initial denaturation at 95°C for 10 min to allow activation of the Taq polymerase, followed by 50 cycles of denaturation at 94°C for 30 s, annealing at 50°C for 1 min, and extension at 72°C for 2 min. For the AOB-specific qPCR, the thermal cycle consisted of an initial denaturation at 95°C for 10 min, followed by 50 cycles of denaturation at 94°C for 15 s, annealing at 60°C for 20 s, and extension at 72°C for 40 s. Data were acquired at the end of each extension step and collected in the FAM channel (source wavelength, 470 nm; detector wavelength, 510 nm; gain set to 8). Reaction products were verified by DNA melting curve analysis at temperatures ranging from 65°C to 95°C for the eubacterial qPCR and at temperatures ranging from 75°C to 95°C for the AOB-specific qPCR, with the temperature at each step increasing 1°C every 30 s. Reaction products were also verified by 1.0% (wt/vol) agarose gel electrophoresis (data not shown). In addition, AOB-specific PCR of the total extract sample was repeated without addition of SYTO-9, and DNA sequencing confirmed that the amplified fragment had the expected amoA target sequence (data not shown).
Quantitation was performed using the quantitation analysis feature of the RotorGene software (version 6). This analysis involved construction of a standard curve by plotting the threshold cycle (Ct) at which exponential amplification began for each standard against the concentration of each standard. The Ct of unknown samples was then used to estimate the starting DNA concentration by reference to the standard curve.
Probes EUB (domain Bacteria), ALF (Alphaproteobacteria), BET (Betaproteobacteria), GAM (Gammaproteobacteria), and CF (Bacteroidetes [formally Cytophaga-Flavobacterium group]) were used as described previously (13). In order to determine total bacterial abundance, bacteria hybridized with probes were also stained with 4′,6′-diamidino-2-phenylindole (DAPI) (final concentration, 5 μg ml−1) at room temperature for 5 min prior to microscopy. Cells were examined by epifluorescence microscopy (Olympus BX50 system) at a magnification of ×1,000. The numbers of bacteria labeled with oligonucleotide probes and stained with DAPI in each field were counted, until a minimum of 500 oligonucleotide probe-labeled bacteria were recorded. The results were expressed as the means of triplicate counts (that is, three separate counts of 500 labeled bacteria per sample). For each sample the coefficient of variation for each of the three replicates was less than 7.5%.
In the raw water, membrane-intact and esterase-active bacteria represented 90.3% and 84.0% of the total bacteria, respectively (Fig. (Fig.1,1, panel i). HPC counts were significantly less, by 3 to 4 logs. The numbers of bacteria in each fraction were less in the filtered water (Fig. (Fig.1,1, panel ii), but the decrease in the physiologically active bacteria was more pronounced in the finished water, in which the numbers of bacteria were below the limit of detection for the FCM activity assays (Fig. (Fig.1,1, panel iii). In addition, no bacteria were detected by routine HPC (Fig. (Fig.1,1, panel iii). Following 12 km of travel through the distribution system (chloramine residual, <0.1 mg liter−1), bacterial regrowth or release of pipeline biofilm was evident as the numbers of active bacteria increased to 1.32 × 105 cells ml−1 and 2.22 × 105 cells ml−1 in the bulk water of the distribution system for the membrane-intact and esterase-active fractions, respectively (Fig. (Fig.1,1, panel iv). Of particular note was the finding that routine HPC failed to detect any culturable bacteria in the bulk water of the distribution system. In addition, a total of 250 AOB per 100 ml (equivalent to 2.5 cells ml−1) were detected by liquid MPN culture. MPN culture failed to detect any AOB in the samples taken from the WTP.
Betaproteobacteria, which accounted for 16% of the total bacteria stained with DAPI, were the most predominant phylogenetic group detected in raw water (Fig. (Fig.2,2, panel i) (P < 0.05). This group remained dominant in filtered water, but the level decreased significantly to 10% of the total DAPI-stained cells (Fig. (Fig.2,2, panel ii) (P < 0.05). No significant change in community composition was detected for the raw and filtered water samples for any of the other phylogenetic groups tested (P > 0.05).
In finished water, chloramination resulted in the absence of any bacteria with detectable hybridization to any of the FISH probes. However, in the bulk water of the distribution system probe EUB successfully hybridized to 52% of the 4.32 × 105 bacteria ml−1 stained with DAPI (Fig. (Fig.2,2, panel iii). Betaproteobacteria (which include the chemolithotrophic AOB) were the predominant phylogenetic group detected and accounted for 20% of the organisms (Fig. (Fig.2,2, panel iii), which was a significantly higher percentage than the percentage of Betaproteobacteria reported for the raw and filtered water samples (P < 0.05). The Alphaproteobacteria (which include Nitrobacter sp.) were the second most abundant group and accounted for 10% of the bacteria detected in the bulk water of the distribution system. This level was significantly greater than the level of Alphaproteobacteria determined for the raw and filtered water samples (P < 0.05).
Universal DGGE analysis of the raw water revealed a diverse population of bacteria (Fig. (Fig.3).3). There was significant variation in diversity between the total and active fractions (Fig. (Fig.4,4, panel i), most likely due to the presence of inactive bacteria in the total fraction; however, there was little difference between the membrane-intact and esterase-active fractions. Strong bootstrap support (99%) revealed a large difference in DGGE profiles between the bacteria detected in the total and active fractions and the bacteria culturable by the routine HPC method (Fig. (Fig.4,4, panel i). This was most likely due to the large number of active bacteria that were not detected by HPC culture. The physiological states of some of these bands are listed as “active not cultured” in Table Table22.
There was a significant change in bacterial diversity from the raw water to the filtered water (Fig. (Fig.4,4, panel ii). The change at the species level was in contrast to the absence of any major shift at the higher subclass level, as determined by the FISH analyses (Fig. (Fig.2).2). There was an additional change in bacterial diversity between the organisms from the filtered water and the inactive bacteria in the total fraction of the finished water (Fig. (Fig.33 and and4,4, panel ii). Cluster analysis also suggested that there was very little similarity between the total bacterial fraction in the raw water entering the plant and the highly diverse but inactive bacterial population in the finished water leaving the plant (Fig. (Fig.4,4, panel ii).
Despite similar band numbers, cluster analysis revealed a significant change in community composition between the total but inactive bacterial fraction leaving the WTP and the total fraction in the bulk water of the distribution system (Fig. (Fig.4,4, panel ii). In the bulk water of the distribution system the diversity of membrane-intact and esterase-active bacteria was represented by 6 and 10 major DGGE bands, respectively (Fig. (Fig.3).3). Sequence analysis of the major bands revealed members of the AOB (Table (Table2).2). Of particular note was the detection of an active Nitrosospira-related species (band 29; 98% similarity) and an active bacterium closely related to Nitrosomonas cryotolerans (band 30; 98% similarity). Sequencing of DGGE bands in the raw, filtered, and finished water samples did not detect these members of the AOB in the WTP. Active NOB belonging to Nitrobacter-related species (Fig. (Fig.3,3, bands 23 and 32) were also detected in the physiologically active fractions of the distribution system water sample.
In addition to the universal nested PCR-DGGE analyses, an AOB-specific nested PCR-DGGE was performed for all fractions from each sample. This analysis confirmed that active AOB were present in the bulk water of the distribution system (Fig. (Fig.5),5), but no AOB were detected in the WTP. Comigration and sequencing confirmed that bands 29 and 30 in Fig. Fig.33 represented members of the AOB, while all other bands that were present on the universal DGGE gel were absent (Fig. (Fig.5).5). DNA was also extracted from positive AOB MPN culture tubes and was subjected to AOB-specific nested PCR-DGGE analysis. The results revealed that the species related to N. cryotolerans (Fig. (Fig.5,5, band 30) was detected by the MPN culture technique, while the Nitrosospira-related species (Fig. (Fig.5,5, band 29) was not detected. The MPN technique also revealed an additional species not observed in the distribution system extracts in either the universal or AOB-specific DGGE patterns (Fig. (Fig.5,5, band MPN1). Sequence analysis showed that this organism had a sequence that was 97% similar to that of Nitrosomonas oligotropha.
NOB-specific PCR and DNA sequencing further confirmed the presence of an active Nitrobacter-related species in each fraction of the distribution system sample. Nitrospira-specific PCR and DNA sequencing also revealed the presence of an active Nitrospira-related species (98%) in each fraction, whereas the universal eubacterial nested 16S rRNA gene-directed PCR-DGGE approach did not reveal this bacterium.
An additional analysis was performed with the distribution system sample to compare the number of AOB detected by the routine MPN culture technique and the number of physiologically active AOB. Amplification products of the 27F and 1492R standards had a major melting peak between 80.1 and 80.3°C (data not shown). Quantitative analysis of the qPCR data for the standards (Fig. (Fig.6,6, panel i) allowed construction of a standard curve (R2 = 0.99928), in which the Ct standard deviation did not exceed 0.28 for any of the dilution replicates and the efficiency of amplification was equivalent to 0.77. Amplification products of the samples had melting peaks identical to those of the standards (Fig. (Fig.6,6, panel ii). Using the standard curve, the number of 27F-1492R copies in the total extract was calculated to be 5.58 × 105 ± 1.72 × 104 copies ml−1. The number of 27F-1492R targets in the total fraction was compared to the 3.18 × 105 total bacteria ml−1 determined by SYTO-9 staining and FCM analysis (Fig. (Fig.1,1, panel iv), and the results suggested that each bacterium had on average 1.75 copies of the 16S rRNA gene per genome (assuming 100% DNA extraction efficiency).
Using the standard curve (R2 = 0.99928), the membrane-intact and esterase-active fractions were shown to have average numbers of 27F-1492R targets equivalent to 1.24 × 105 ± 1.22 × 104 copies ml−1 and 1.68 × 105 ± 1.90 × 104 copies ml−1, respectively. Unlike the total bacterial fraction results, the qPCR results for the active fractions could not be compared directly to the number of active bacteria determined by FCM (Fig. (Fig.1,1, panel iv). The numbers of bacteria in the active fractions were biased by the “single-cell” FCM sort mode used. This mode is designed to provide high purity, and as a result it did not collect all of the active cells that were enumerated during sorting.
qPCR was also performed for each of the fractions to determine the numbers of AOB. Following amplification of the standards, melting curve analysis revealed a major melting peak for the standards between 86.8 and 87.0°C (data not shown). Quantitative analysis of the qPCR data for the standards (Fig. (Fig.6,6, panel iii) allowed construction of a standard curve (R2 = 0.99843), in which the Ct standard deviation did not exceed 0.25 for any of the dilution replicates and the efficiency of amplification was equivalent to 0.78. Amplification products of the samples had melting peaks identical to those of the standards (Fig. (Fig.6,6, panel iv). As determined with the standard curve, the average sample concentration of amoA targets in the total extract was 3.14 × 104 ± 1.47 × 102 copies ml−1. On a gene copy basis, the total number of amoA gene copies (AOB specific) detected represented 5.6% of the total 27F-1492R gene copies (universal eubacterial PCR) in the total bacterial fraction of the distribution system. Assuming that amoA heterogeneity varied in a way similar to that of the 16S rRNA gene, this revealed that the total AOB in the bulk water of the distribution system accounted for 5.6% of the entire bacterial population, although this analysis gave no indication of what percentage of AOB in the distribution system was active.
To assess the percentage of active AOB in the distribution system, active fractions were also analyzed by qPCR targeting the amoA gene. The average numbers of amoA copies in the membrane-intact and esterase-active fractions were 3.22 × 103 ± 5.53 × 102 copies ml−1 and 1.06 × 104 ± 1.63 × 103 copies ml−1, respectively. Using these values and those obtained with the 27F-1492R assay and assuming that there was no preferential selection of bacterial species with the FCM cell sorting technique, we calculated that 2.6% of the planktonic bacteria that had maintained membrane integrity and 6.3% of the esterase-active bacteria in the bulk water of the distribution system were AOB.
Since the concentration of bacteria with intact membranes in the bulk water of the distribution system was determined to be 1.32 × 105 bacteria ml−1 by FCM (Fig. (Fig.1,1, panel iv), the approximate concentration of active AOB in the bulk water of the distribution system whose membranes remained intact was therefore 2.6% of this value, or 3.43 × 103 active AOB ml−1. Also, the concentration of bacteria with intracellular esterase activity in the distribution system was determined to be 2.22 × 105 bacteria ml−1 by FCM analysis (Fig. (Fig.1,1, panel iv), and the concentration of esterase-active AOB in the distribution was 6.3% of this value, or 1.40 × 104 active AOB ml−1.
Successful detection of AOB in the distribution system using the universal nested PCR-DGGE approach required sequence analysis of many DGGE bands, a somewhat laborious task. A more rapid method was the use of an AOB-specific PCR-DGGE technique that enabled profiling of the active AOB community directly (Fig. (Fig.5).5). This culture-independent approach also circumvented problems with that culture variability that is believed to exist for genera of AOB (1), as demonstrated in this study by the lack of growth of the active Nitrosospira-related species during MPN incubation. This may have been due to the Nitrosospira-related species being in an active but nonculturable or sublethally injured state, to the extended culture time beyond the culture time of the MPN assay, or to limited growth as a result of competition with faster-growing AOB. The latter possibility was likely as Belser and Schmidt (3) determined that mean generation times for Nitrosospira spp. are on the order of 20 h, which is significantly longer than the mean generation times for Nitrosomonas spp. (8 to 14 h).
A limitation of the FCM cell sorting PCR-DGGE approach was revealed when the MPN technique revealed an additional species that was 97% similar to N. oligotropha, which was not detected in the sorted fractions by the culture-independent techniques. PCR-DGGE has a reported limit of sensitivity when the abundance approaches as little as 1% of the total population (25), and the relatively low abundance of this organism compared to the other AOB in the system may have accounted for its absence in the active fractions of the DGGE profiles. In addition, depending on the environmental strain of N. oligotropha present, its detection may have been hindered by a mismatch of up to 2 bp with the forward AOB PCR primer and at least a 1-bp mismatch with the reverse primer (17), although this did not prevent PCR-DGGE detection of this organism in the AOB MPN culture medium. In contrast, both AOB PCR primers exhibited 100% homology to Nitrosospira sp. and N. cryotolerans (17). Additional factors, such as sampling error, preferential DNA extraction, inefficient or selective DNA amplification, or insufficient gel staining (25), may also have been significant in the failure of the culture-independent approach to detect N. oligotropha in the sorted fractions. The detection of the N. oligotropha-related species by the MPN assay also suggested that this organism had growth requirements that were more suited to the MPN medium than those of the apparently more abundant Nitrosospira-related species not detected by the MPN technique.
qPCR was used in this study to determine the total numbers of copies of 16S rRNA genes of all active bacteria and amoA of active AOB in the distribution system. qPCR with DNA extracts of active fractions sorted by FCM ensured that there was no amplification of DNA from inactive bacteria. It has been widely reported that PCR targeting the genes coding for 16S rRNA may lead to overestimation of the number of bacteria in a community because some bacteria contain multiple 16S rRNA genes in a single genome (8, 26). This was observed in this study; an average of 1.75 copies of 16S rRNA genes per bacterium were detected in the distribution system sample. The amo operon has been reported to have heterogeneity in AOB similar to that of the gene encoding 16S rRNA in bacterial species (27). Therefore, amoA gene copy numbers could not be directly correlated with the actual number of AOB cells. Instead, amoA copy numbers were expressed as a proportion of 16S rRNA gene copy numbers, from which the number of active AOB in the distribution system was determined to be several orders of magnitude greater than the number of AOB determined using the routine MPN technique. Undoubtedly, the species belonging to Nitrosospira (Fig. (Fig.5,5, band 29) made a significant contribution to the disparity, as this organism was shown to be not detected by the MPN technique. However, the qualitative nature of the PCR-DGGE analyses used here meant that N. cryotolerans may also have contributed to the overall disparity. Underestimates of AOB numbers similar to those reported in this study have been reported for other environments. Völsch et al. (42), studying the AOB in sewage plants by FCM, hypothesized that the concentration of AOB determined by the MPN method was probably 20- to 2,000-fold too low. In addition, Konuma et al. (15) obtained MPN results that were on the order of 1 to 2 log less than the results of antibody, FISH, and dot blot hybridization methods for AOB enumeration. Such underestimates of AOB abundance when MPN techniques are used have largely been attributed to the presence of unculturable AOB, clumping of cells, or culture media that allow the growth of only certain AOB species (2, 15).
Active NOB, including both Nitrobacter and Nitrospira species, were likely to have a role in the biological oxidation of nitrite in the system, resulting in full nitrification. Both species of NOB most likely contributed to the high level of Alphaproteobacteria (10% of DAPI-stained cells; P < 0.05) found in the distribution system by FISH, as probe ALF has been shown to hybridize with species of Nitrospira (38), which are members of a phylum independent of the Alphaproteobacteria. The increased abundance of NOB in the distribution system compared to the abundance in the other types of water tested was consistent with the observations of Martiny et al. (21), who reported that Nitrospira was one of the most dominant phylotypes detected in a model drinking water distribution system. Until recently, the potential for NOB growth in chloraminated systems had only been mentioned (45), and it was not thoroughly investigated until the recent research of Regan et al. (35) and Lipponen et al. (19). As Regan et al. (35) explained, the lack of studies of the growth of active NOB in chloraminated distribution systems most likely has been due to the fact that the primary fate of nitrite that concerns drinking water utilities is reduction of the chloramine residual. Results of our study support the views of Regan et al. (35), who suggested that the presence of NOB in chloraminated distribution systems points to a needed reevaluation of the role of these organisms in nitrite oxidation compared to chemical oxidation by chloramine.
Nitrification such as that found in the distribution system in this study is believed to indirectly provide a source of soluble organic products (36), and increases in the levels of heterotrophic bacteria during nitrification have been well documented (6, 19, 29, 40, 45, 46). Such increases result in routine HPC values that typically vary from 10 to 10,000 CFU ml−1, and the changes appear to be system specific (29). Therefore, the absence of heterotrophic bacteria detected by routine HPC in the bulk water of the distribution system examined here appears to be unique. This casts doubt on the use of HPC data as an indicator of distribution system integrity following a loss of chloramine residual caused by AOB. The failure to detect any bacteria by routine HPC may have been due to many different factors, such as organisms that were in an active but nonculturable or sublethally injured state, vital signaling molecules for quorum sensing that were diluted during inoculation, substrate-accelerated cell death that occurred due to the presence of certain substrates in the HPC medium, incubation times that were not long enough for colony formation, or bacteria that had nonheterotrophic metabolism (13).
In summary, this study demonstrated for the first time how the flow cytometric cell sorting PCR-DGGE approach can be used to detect the bacteriological causes of instability in chloraminated drinking water systems. In this study, AOB were identified as nuisance organisms, and this community was profiled. In addition, this study demonstrated how active AOB abundance could be rapidly determined by combining flow cytometric cell sorting with AOB-specific qPCR. This technique revealed a disparity of several orders of magnitude between active AOB abundance and the number of bacteria detected by the laborious routine MPN culture technique currently used by many water utilities. It is envisaged that application of the culture-independent approaches described here, in conjunction with existing chemical indicators (chloramine residual, oxidized nitrogen, dissolved oxygen, free and total ammonia), will allow the water industry to obtain more comprehensive data with which to make informed decisions regarding remedial action that may be required either prior to or during an instability event.
We thank Riverland Water and United Utilities Australia for allowing access to the Loxton WTP.
This work was supported by the Cooperative Research Centre for Water Quality and Treatment, by the Australian Water Quality Centre, and by the University of South Australia.