Microautoradiographic Study of Rhodocyclus-Related Polyphosphate-Accumulating Bacteria in Full-Scale Enhanced Biological Phosphorus Removal Plants

Abstract

The ecophysiology of uncultured Rhodocyclus-related polyphosphate-accumulating organisms (PAO) present in three full-scale enhanced biological phosphorus removal (EBPR) activated sludge plants was studied by using microautoradiography combined with fluorescence in situ hybridization. The investigations showed that these organisms were present in all plants examined and constituted 5 to 10, 10 to 15, and 17 to 22% of the community biomass. The behavior of these bacteria generally was consistent with the biochemical models proposed for PAO, based on studies of lab-scale investigations of enriched and often unknown PAO cultures. Rhodocyclus-related PAO were able to accumulate short-chain substrates, including acetate, propionate, and pyruvate, under anaerobic conditions, but they could not assimilate many other low-molecular-weight compounds, such as ethanol and butyrate. They were able to assimilate two substrates (e.g., acetate and propionate) simultaneously. Leucine and thymidine could not be assimilated as sole substrates and could only be assimilated as cosubstrates with acetate, perhaps serving as N sources. Glucose could not be assimilated by the Rhodocyclus-related PAO, but it was easily fermented in the sludge to products that were subsequently consumed. Glycolysis, and not the tricarboxylic acid cycle, was the source that provided the reducing power needed by the Rhodocyclus-related PAO to form the intracellular polyhydroxyalkanoate storage compounds during anaerobic substrate assimilation. The Rhodocyclus-related PAO were able to take up orthophosphate and accumulate polyphosphate when oxygen, nitrate, or nitrite was present as an electron acceptor. Furthermore, in the presence of acetate growth was sustained by using oxygen, as well as nitrate or nitrite, as an electron acceptor. This strongly indicates that Rhodocyclus-related PAO were able to denitrify and thus played a role in the denitrification occurring in full-scale EBPR plants.

Enhanced biological phosphorus removal (EBPR) has been widely used in wastewater treatment industries to remove phosphate from wastewater in order to protect receiving water bodies against eutrophication. In EBPR processes, activated sludge is cycled together with the influent through sequential anaerobic-aerobic or anaerobic-denitrifying (anaerobic in the presence of nitrate or nitrite) periods to enrich polyphosphate-accumulating organisms (PAO). These organisms have an ability to accumulate excess polyphosphate as they are able to accumulate more polyphosphate than they normally need for physiological growth. Since no PAO exist in pure culture, biochemical models (16, 21) for predicting PAO behavior are based solely on transformations of organic compounds and phosphate in lab-scale EBPR systems with enriched microbial communities. According to these models, PAO can take up volatile fatty acids and sequester them in polyhydroxyalkanoates (PHA) during the anaerobic period. The energy for this comes from hydrolysis of intracellular polyphosphate, and the reducing power may come from glycolysis of intracellular glycogen (16) or from anaerobic utilization of acetate through the tricarboxylic acid (TCA) cycle (18). In the subsequent aerobic or denitrifying phase, the PAO can use the PHA stored anaerobically for growth and for refreshing their polyphosphate and glycogen pools (24).

Identifying and isolating the PAO in pure culture remain a challenge. As determined by culture-dependent methods, bacteria belonging to the genus Acinetobacter were believed to be potential PAO for a long time (7). However, recent studies in which culture-independent 16S rRNA-based molecular techniques, including fluorescence in situ hybridization (FISH), were used revealed that this is not the case (16, 21). Instead, several authors have proposed that Rhodocyclus-related bacteria are important PAO (5, 6, 14). These bacteria could be enriched in lab-scale P-removing systems, where they took up acetate, formed PHA anaerobically, and grew and accumulated polyphosphate aerobically (5, 6, 10, 14). Moreover, they dominated in lab-scale EBPR systems with anaerobic-denitrifying cycling conditions, and it was suggested that they were involved in the N removal observed (26). Rhodocyclus-related bacteria have also been shown to be present in some full-scale EBPR plants, where they are assumed to be important PAO (27).

It is not possible to conclude that the physiology of Rhodocyclus-related PAO (RPAO) can be described in detail by the present biochemical models, as other, unknown PAO may have been present in the lab-scale systems investigated. This statement is supported by several observations made with lab-scale systems, which indicated that other bacteria may be PAO (21). Furthermore, only a few studies have been performed with RPAO present in full-scale EBPR plants, so it is not clear whether the RPAO in such plants behave as the biochemical models predict and whether they play an important role in the P removal observed. Full-scale experience does not always tie in with the present biochemical models, and EBPR breakdown occurring for reasons that are unknown may be due to a poor understanding of EBPR microbiology. Thus, in order to overcome instability in EBPR plants, optimize plant operation, and design novel and efficient processes, it is important to link the true identity of dominant PAO to their function (i.e., ecophysiology and ecology).

In this study, we focused on the ecophysiology of the RPAO in full-scale EBPR plants. These bacteria can be visualized microscopically with species-specific oligonucleotide probes by using FISH (5, 6). By combining FISH with microautoradiography (MAR), which can reveal organic substrate assimilation and phosphorus uptake by bacteria targeted by probes (11), it is possible to answer important questions about the physiology of the organisms in relation to the biochemical models. For example, we do not know the range of organic substrates that the bacteria use or whether glycolysis is important in their anaerobic substrate assimilation. Another important question is whether these organisms can denitrify. These questions were all dealt with in this study by using a range of incubation conditions for activated sludge from three full-scale EBPR plants and by using MAR-FISH.

MATERIALS AND METHODS

Activated sludge and description of full-scale EBPR plants.

Activated sludge samples were obtained from three full-scale EBPR wastewater treatment plants (WWTP) in Denmark: Aalborg East (AAE) WWTP, Egaa WWTP, and Skagen WWTP. All of these plants have a Biodenipho configuration (20), which includes an anaerobic tank and another tank with alternating denitrifying and nitrifying aerobic conditions. The Skagen WWTP mainly treats industrial wastewater from the fish industry and processes an amount corresponding to a population of 280,000 persons. Treating mainly domestic wastewater, the AAE and Egaa WWTP serve communities with populations of 100,000 each. In these two plants, small amounts of ferric iron (FeCl₃) are added to precipitate orthophosphate that is not removed by the biological removal of phosphorus. During the sampling period of this study, the average total P content in the influents of these three plants varied from 7 to 10 mg/liter, while the filtrate P content in the effluent was always less than 0.5 mg/liter. Activated sludge samples were taken from the oxic (nitrifying) tank and transferred to the laboratory within a couple of hours. All experiments in this study were carried out from May to October 2003. The experiments were repeated three times for the AAE and Skagen WWTP during this period, while one investigation was conducted at the Egaa WWTP.

MAR-FISH.

MAR-FISH was carried out by using the procedure described by Nielsen et al. (17) and Lee et al. (11), with a slight modification. Briefly, activated sludge was incubated in 9- or 60-ml serum bottles with labeled and unlabeled substrates under aerobic conditions and under anaerobic conditions with or without nitrate or nitrite. All anaerobic preparations were carefully flushed with O₂-free N₂. The samples that were incubated were fixed by addition of freshly prepared paraformaldehyde in phosphate-buffered saline to a final concentration of 4% and allowed to stand for 3 h at 4°C (1), and they were subsequently washed in citrate buffer (pH 2.0; for ³³P_i incubations only) and distilled water. The samples were gently homogenized by rubbing two glass slides with a 40-μl sample against each other before they were transferred to gelatin-coated cover glasses (24 by 60 mm), allowed to dry, and hybridized with oligonucleotide probes (Thermo Electron Corporation, Dreieich, Germany) labeled with fluorescent dyes (Cy3 and FLUOS). Hybridization was carried out as described by Amann (1). The cover glasses were carefully dipped in prewarmed (43°C) LM-1 emulsion (Amersham Bioscience) and exposed at 4°C for 3 days (or in some cases 6 or 12 days); this was followed by development (0.5 to 4 min) in Kodak L-19 developer as described by Lee et al. (11). Microscopic examinations of MAR-FISH samples were carried out by using an LSM510 Meta scanning confocal microscope (Carl Zeiss, Oberkochen, Germany) and an epifluorescence microscope (Axioskop 2 Plus; Zeiss) equipped with a charge-couple device camera (CoolSNAP HQ; Photometrics, Oberkochen, Germany). The percentages of the RPAO assimilating [³H]acetate and ³³P_i (MAR positive) were estimated by evaluating at least 200 FISH-positive bacteria (for [³H]acetate) or microcolonies (for ³³P_i) observed after 3, 6, and 12 days of exposure. Bacteria that took up an amount of ³³P_i that gave a clear MAR signal after 3 days of exposure at 4°C were referred to as PAO to distinguish excessive bacterial uptake from the normal physiological requirement.

In experiments in which anaerobic conditions were applied, an anaerobic removal step was introduced if nitrate or nitrite was present in the sample (tested with test stickers from Merck, Darmstadt, Germany), until none of the compounds could be detected. Furthermore, an anaerobic 1-h pretreatment without addition of any substrate was included to remove any residual O₂ before the anaerobic incubation was started. For aerobic incubation, substrates and/or other chemicals were added after the activated sludge was vigorously shaken for 20 min to ensure aeration. During all preincubations and incubations, the serum bottles were shaken (250 rpm) at 20 ± 1°C. The biomass concentration used in all incubations was 1 g of mixed liquor suspended solids (MLSS) per liter unless indicated otherwise. This biomass concentration was prepared by diluting activated sludge (4 to 5 g of MLSS per liter) with nitrate- and nitrite-free filtered effluent water from the same WWTP.

The incubation conditions used for investigating ³³P_i uptake by RPAO under different electron acceptor conditions are shown in Table 1. Prior to the incubations with ³³P_i, sludge samples were incubated anaerobically with 2 mM acetate for 2 h to build up the internal PHA storage compound. The unconsumed acetate was removed by washing the samples three times with the filtered effluent water, before labeled ³³P_i and nonlabeled P_i was added to a final concentration of 0.3 mM at the very beginning of the 3- or 6-h incubation period with either oxygen, nitrate, or nitrite present as the electron acceptor. When nitrate or nitrite was used as the electron acceptor, the biomass concentration used was 0.5 g of MLSS per liter in order to prevent the electron acceptors from being depleted during the incubation. The concentration of nitrate or nitrite was kept low to ensure that no toxic effects of nitrite took place by adding 0.3 mM nitrate or nitrite at time zero and again every 1.5 h. Nitrate and nitrite contents were measured in parallel incubations with unlabeled P_i by using a similar test sticker, as described above.

TABLE 1.

Effects of different electron acceptors on the uptake of ³³P_i by RPAO in the three different WWTP investigated: incubation conditions and results based on MAR investigations^a

Incubation		Fraction of RPAO involved in 33Pi uptake (%)
Electron acceptor	Time (h)	AAE WWTP	Egaa WWTP	Skagen WWTP
O2	3	27 ± 8b	56 ± 13	86 ± 9
O2 (+ acetate)	3	0	0	0
NO3−	3	31 ± 11	47 ± 9	76 ± 5
NO3− (+ acetate)	3	0	0	0
NO3−	6	25 ± 7	NDc	ND
NO2−	3	31 ± 15	ND	89 ± 9
NO2− (+ acetate)	3	0	0	0
NO2−	6	36 ± 9	ND	ND
Not addedd	3	0	0	0
Not added (+ acetate)	3	0	0	0

^aIncubation with labeled ³³P_i was carried out after 2 h of incubation with unlabeled acetate under anaerobic conditions.

^bThe values are averages ± standard deviations for six counts based on at least two different samples. For each count at least 80 microcolonies were examined.

^cND, not determined.

^dNot added indicates anaerobic conditions.

The conditions used for investigation of substrate assimilation by RPAO and the related experimental conditions are shown in Tables 2 and 3. The organic substrates were added either as single substrates (Table 2) or as two substrates simultaneously (Table 3). The investigations were carried out with oxygen present or under anaerobic conditions (no electron acceptor added).

TABLE 2.

Uptake of organic substrates under anaerobic conditions or with oxygen as the electron acceptor by RPAO in the three different WWTP investigated: incubation conditions and results based on MAR investigations^a

Unlabeled substrate	Radioactive tracer	Electron acceptor	Substrate uptakea by RPAOb
			AAE WWTP	Egaa WWTP	Skagen WWTP
Formic acid	[1-14C]formic acid	Not added or O2c	−	−	−
Acetate	[3H]acetate (Na)	Not added or O2	+	+	+
Acetate	[3H]acetate (Na)	Not added (+IO)d	−	−	−
Propionate	[1-14C]propionate (Na)	Not added or O2	+	+	+
Butyrate	[2,3-3H]butyric acid	Not added or O2	−	−	−
Pyruvate	[1-14C]pyruvate (Na)	Not added or O2	+	+	+
Oleic acid	[9,10(n)-3H]oleic acid	Not added or O2	−	−	−
Aspartic acid	d-[2,3-3H]aspartic acid	Not added or O2	−	ND	ND
Glutamic acid	l-[G-3H]glutamic acid	Not added or O2	+	ND	ND
Leucine	[4,5-3H]leucine	Not added or O2	−	−	−
Glycine	[2-3H]glycine	Not added or O2	−	−	−
Thymidine	[methyl-3H]thymidine	Not added or O2	−	−	−
Glucose	d-[2-3H]glucose	Not added	+	+	+
Glucose	d-[2-3H]glucose	Not added (+IO)	−	−	−
Glucose	d-[2-3H]glucose	O2	−	−	−
Ethanol	[1-3H]ethanol	Not added or O2	−	−	−

^aThe final concentration of all substrates used was 1.5 mM, except for oleic acid (1.0 mM).

^b+, >95% of RPAO took up radioactive substrate; −, RPAO did not take up radioactive substrate; ND, not determined.

^cNot added indicates anaerobic conditions.

^dIO, iodoacetate added at a final concentration of 200 mg/liter.

TABLE 3.

Assimilation of two organic substrates simultaneously by RPAO in different plants, as determined by MAR

Unlabeled substrates	Radioactive tracer	Electron acceptor	Substrate uptake by RPAOa
			AAE WWTP	Egaa WWTP	Skagen WWTP
Acetate (1.5 mM) + propionate (1.5 mM)	[3H]acetate (Na)	Not addedb	+	+	+
Acetate (3.0 mM) + propionate (0.5 mM)	[1-14C]propionate (Na)	Not added	+	+	+
Acetate (1.5 mM) + butyrate (1.5 mM)	[1-3H]butyric acid	Not added or O2	−	−	−
Acetate (1.5 mM) + thymidine (1.5 mM)	[methyl-3H]thymidine	Not added or O2	+	+	+
Acetate (1.5 mM) + leucine (1.5 mM)	[4,5-3H]leucine	Not added or O2	+	+	+

^a+, radioactive substrate taken up; −, radioactive substrate not taken up.

^bNot added indicates anaerobic conditions.

The conditions used for investigation of RPAO's acetate assimilation and possible growth under different electron acceptor conditions are shown in Table 4, which also shows the related experimental conditions. In these experiments, uptake of labeled acetate after prolonged uptake of unlabeled acetate (6 or 9 h) indicated that the bacteria were able to grow and not only store acetate under the conditions used. Therefore, preincubation steps with unlabeled substrate were included before the incubation with labeled substrate under various electron acceptor conditions. Acetate (2.0 mM) was added to start the preincubation, and more acetate was added to obtain a final concentration of 0.5 mM after 3 h (for 6-h preincubations) and after 6 h (for 9-h preincubations). For preincubation and incubation under aerobic conditions, 0.2 g of MLSS per liter was used to prevent the added acetate from being used up. Acetate was added to a final concentration of 2.0 mM to start the preincubation. After 3 h (for 6-h preincubations) or after 6 h (for 9-h preincubations) more acetate was added to obtain a final concentration of 1.0 mM. Parallel incubations with unlabeled acetate were always carried out in order to ensure that acetate was not depleted (which was determined by high-performance liquid chromatography [HPLC] as described below). In some experiments, the glycolytic pathway was inhibited by using iodoacetate (13) at a final concentration of 200 mg/liter. In all incubations with P_i and organic substrates, the amount of labeled substrate used was 10 μCi/ml. As a control for chemography, sludge was pasteurized at 70°C for 10 min just before incubation with labeled substrates. No uptake was ever observed in the controls with any of the radiochemicals used in this study. With ³³P_i an increase in the background due to precipitates was detected, which made it difficult to observe uptake of ³³P_i by single cells. Therefore, only cells present in microcolonies were assessed.

TABLE 4.

Effect of preincubation (0 to 9 h) with unlabeled acetate on the assimilation of labeled acetate by RPAO under different electron acceptor conditions as determined by MAR

Preincubation		Incubation		Uptake by RPAOa
Electron acceptor	Time (h)	Electron acceptor	Time (h)	AAE WWTP	Egaa WWTP	Skagen WWTP
O2	3	O2	2	+	+	+
O2	6	O2	2	+	+	+
O2	9	O2	2	+	+	+
Not addedb	0	Not added	2	+	+	+
Not added	3	Not added	2	+ (<20% ± 8%)	−	−
Not added	6	Not added	2	−	−	−
Not added	6	NO3−	2	+ (27% ± 12%)	+ (38% ± 6%)	+ (83% ± 12%)
Not added	6	NO2−	2	+ (35% ± 11%)	ND	ND
NO3−	6	NO3−	2	+ (31% ± 6%)	ND	ND
NO2−	6	NO2−	2	+ (22% ± 9%)	ND	ND

^a+, >95% of RPAO assimilated labeled acetate; −, RPAO did not assimilate labeled acetate. The values in parentheses are the percentages of RPAO taking up labeled acetate (means ± standard deviations) estimated from six counts for at least two samples. For each count 50 single cells were examined. ND, not determined.

^bNot added indicates anaerobic conditions.

The radiochemicals sodium [³H]acetate, d-[2-³H]glucose, [9,10(n)-³H]oleic acid, [4,5-³H]leucine, [methyl-³H]thymidine, and ³³P_i were purchased from Amersham Biosciences (Amersham Bioscience, Hilleroed, Denmark); sodium [1-¹⁴C]pyruvate and [1-³H]ethanol were obtained from American Radiolabeled Chemical, Inc. (Bio Nuclear AB, Bromma, Sweden); and sodium [1-¹⁴C]propionate and [2,3-³H]butyric acid were obtained from ICN Biochemicals, Inc. (Bie og Berntsen, Roedovre, Denmark).

FISH and enumeration of RPAO.

Fixation and FISH probing of paraformaldehyde-fixed activated sludge samples were also carried out as described previously (1). Oligonucleotide probes EUB338, EUB338II, EUB338III, BET42a, ALF968, GAM42a, HGC69a, CF319a, ARCH915, PAO651, and RHC439 were used. The specificities of these probes and the formamide concentrations used are described in probeBase (15). The percentages of RPAO based on the levels of Bacteria present in the three EBPR plants were estimated by measuring the percentage of area fluorescing with probe PAO651 to the percentage of area fluorescing with EUB338mix (EUB338, EUB338II, and EUB338III) on the same images taken with the LSM510 Meta microscope by using the MetaVue software (version 6.4; Universal Imaging Corp., Downingtown, Pa.). At least 10 different microscopic fields with proper biomass distribution were analyzed for each enumeration.

Glucose fermentation and chemical analysis.

In order to investigate the fermentation abilities of the activated sludges, glucose fermentation and production of short-chain fatty acids by activated sludges from the three plants were studied by using replicates in 120-ml serum bottles. Sixty milliliters of mixed liquor with a dry matter concentration of 2.0 g of MLSS per liter was flushed with O₂-free N₂, glucose (1.5 mM) was added, and the preparation was shaken at 20°C on a rotary shaker for 2 h. A Dionex HPLC (Radiometer, Copenhagen, Denmark) equipped with an Ionpac column [DIONEX AS11-HC4mm (19-32) P/N 52960] was used to measure short-chain fatty acids in the filtrate (obtained with Millipore Millex HV filters [pore size, 0.45 μm]). The HPLC operating conditions used were those recommended by the supplier of the column.

RESULTS

Operation of full-scale plants and FISH probing for RPAO.

The EBPR plants investigated have all performed efficient biological N and P removal for several years. They have had stable operation, and the effluent concentrations of total P and N are usually less than 0.5 and 6 mg/liter, respectively. Two oligonucleotide probes targeting RPAO with different specificities were used for the initial biomass FISH probing. Coccoid cells in clusters responding to probes PAO651 and RHC439 were detected in sludge from all three plants. Most of the bacterial cells responding to probe PAO651 (>95%) also fluoresced with probe RHC439. All the cells hybridizing with these two probes also responded to the BET42a probe and EUB338mix targeting Betaproteobacteria and Bacteria, respectively, confirming that they were typical RPAO (5). Bacterial cells hybridizing with probe PAO651 were abundant in all three treatment plants. During the period of investigation, such cells constituted 5 to 10% of all the bacterial cells responding to EUB338mix in the AAE WWTP, 10 to 15% of all the bacterial cells responding to EUB338mix in the Egaa WWTP, and 17 to 22% of all the bacterial cells responding to EUB338mix in the Skagen WWTP. In this study, bacteria that were positive with probe PAO651 were investigated further by performing MAR-FISH experiments, and they were referred to as RPAO.

³³P_i uptake under different electron acceptor conditions.

MAR-FISH investigations showed that the RPAO present in the three full-scale EBPR plants were the dominant population that assimilated ³³P_i under aerobic conditions after 2 h of anaerobic preincubation with unlabeled acetate. An example of positive FISH-labeled, MAR-positive RPAO is shown in Fig. 1A and B. The RPAO had similar patterns for uptake of labeled ³³P_i in the three plants (Table 1). In addition to RPAO, only a few other microcolonies took up ³³P_i under aerobic conditions. The latter organisms belonged to the Alphaproteobacteria and the Gammaproteobacteria, which responded to probes ALF968 and GAM42a, respectively. Not all RPAO took up ³³P_i; the highest percentage was found at the Skagen WWTP (86% ± 9% [average ± standard deviation]), and the lowest percentage was found at the AAE WWTP (27% ± 8%). These values for MAR-positive RPAO were based on several experiments for each treatment plant during the 6-month investigation period. In order to investigate whether a fraction of the RPAO were inactive or took up only very small amounts of ³³P_i, the time of exposure of the fixed MAR samples was increased from the normal 3 days to 6 and 12 days to detect lower levels of incorporated radioactivity. No significant changes in the percentages of RPAO that took up ³³P_i were detected for any of the treatment plants (data not shown), indicating that there was an inactive fraction of RPAO in terms of P_i uptake. Uptake of ³³P_i also took place with nitrate or nitrite as the electron acceptor (Table 1). A fraction similar to the fraction under aerobic conditions was able to take up ³³P_i during a 3-h incubation with nitrate or nitrite, while no uptake was observed under anaerobic conditions. Prolongation of the denitrifying incubation with nitrate or nitrite from 3 to 6 h did not significantly change the percentage of RPAO taking up ³³P_i, supporting the observation from the aerobic incubations that a fraction of the RPAO was inactive in ³³P_i uptake. When acetate was present together with oxygen, nitrate, or nitrite as the electron acceptor, no uptake of ³³P_i by the RPAO was ever observed, indicating that RPAO did not accumulate polyphosphate when an external substrate was present for growth or storage in the presence of these electron acceptors (see below).

FIG. 1.

Images of activated sludge samples with RPAO after FISH and MAR. (A and C) FISH images showing bacteria hybridized with the bacterial probe EUBmix (FLUOS labeled, green) and the RPAO probe PAO651 (Cy3 labeled, red). Yellow microcolonies and cells (red and green) are RPAO. (B and D) Bright-field images of MAR-positive cells taking up radiolabeled substrate. (A and B) Microcolonies of RPAO taking up ³³P_i under aerobic conditions. Arrows indicate FISH- and MAR-positive microcolonies. (C and D) Microcolonies of RPAO assimilating acetate under anaerobic conditions. Bar = 10 μm.

Assimilation of organic substrates.

The RPAO present in the three EBPR plants were able to assimilate only a small part of the organic substrate tested (Table 2), but in all cases a very high fraction of the RPAO was MAR positive (>95%) under both aerobic and anaerobic conditions. This shows that most RPAO were active in terms of organic substrate assimilation. All RPAO had the same pattern of assimilation of different labeled organic substrates during the 2-h incubation period. They assimilated acetate, propionate, pyruvate, and glutamic acid (tested only in the AAE WWTP) (Fig. 1C and D). Formate, butyrate, oleic acid, ethanol, leucine, glycine, aspartic acid (tested only in the AAE WWTP), and thymidine were not consumed. Other bacteria in the sludges were able to take up these substrates under all conditions tested, thus serving as positive controls.

Glucose could not be consumed by the RPAO. When labeled glucose was added as the sole substrate, the RPAO were MAR positive under anaerobic conditions but not under aerobic conditions (Table 2). This suggests that the RPAO were unable to assimilate glucose directly and that the observed uptake under anaerobic conditions was indirectly due to assimilation of labeled fermentation products produced by other heterotrophic bacteria. To confirm this hypothesis, the sludge (2 g of MLSS per liter) was incubated with glucose under anaerobic conditions for 2 h, and the amounts of the fermentation products were measured. Significant amounts of soluble acetate (40 to 470 μM), propionate (10 to 240 μM), lactate (50 to 260 μM), and butyrate (30 to 150 μM) were detected in all three WWTP, supporting our hypothesis. Furthermore, to confirm that RPAO were unable to assimilate glucose, iodoacetate was added to the anaerobic glucose incubations (Table 2) to block glycolysis by all the heterotrophic bacteria in order to inhibit glucose fermentation. As expected, no assimilation of glucose by RPAO was observed. However, blockage of glycolysis could also inhibit the anaerobic substrate assimilation by blocking the intracellular source from which the organisms obtained reducing power for PHA formation (degradation of glycogen). To confirm this, iodoacetate was also added to anaerobic incubations with labeled acetate present (Table 2). No acetate assimilation was observed, confirming that RPAO depended on degradation of glycogen to generate reducing power for PHA buildup and that they were unable to assimilate glucose.

The ability of RPAO to assimilate two substrates simultaneously under anaerobic conditions was also investigated (Table 3). Acetate and propionate were taken up simultaneously, even when one of the compounds was present at a much higher concentration than the other. When propionate was tested, there was a potential risk that it would be fermented to acetate and thus lead to a wrong conclusion. Therefore, different concentrations of acetate and propionate were added (3.0 and 0.5 mM, respectively), so if any labeled acetate was produced from labeled propionate, it would be diluted in the large unlabeled acetate pool and not result in strong uptake of labeled acetate and thereby provide only a weak MAR signal. A strong MAR signal was observed, confirming that propionate was indeed taken up simultaneously with acetate, and this was confirmed by the fact that no acetate was detected from possible fermentation after incubation of the activated sludge with propionate anaerobically for 2 h (data not shown). Butyrate could not be assimilated as a sole substrate, and the presence of acetate did not promote any co-uptake of butyrate. In contrast, the amino acid leucine and thymidine (a component necessary for DNA biosynthesis), which could not be taken up as sole substrates, could be assimilated when acetate was present. This indicated that the latter compounds were not used as primary substrates for growth.

Uptake, storage, and growth under different electron acceptor conditions.

In order to investigate whether the assimilated acetate was used by the RPAO for storage and/or growth under anaerobic or aerobic conditions, we performed a series of experiments in which there were various periods of preincubation with unlabeled acetate before the labeled acetate was added (Table 4). Under anaerobic conditions, the storage capacity of RPAO was saturated after 2 to 3 h, as only a few RPAO could assimilate acetate after 3 h of preincubation and none could do so after 6 h. This indicated that no growth took place under anaerobic conditions. Under aerobic conditions, all RPAO took up labeled acetate after 3, 6, and 9 h of preincubation, indicating that growth took place because the storage capacity most likely would have been saturated after 2 to 3 h, as it was under anaerobic conditions. However, whether storage and growth took place simultaneously could not be assessed. The possible growth of RPAO with nitrate or nitrite was investigated in other experiments (Table 4). After the storage capacity was saturated after 6 h of preincubation with unlabeled acetate under anaerobic conditions, labeled acetate was added in the presence of either nitrate or nitrite. In both cases most RPAO were able to take up labeled acetate in the Skagen WWTP, indicating that growth took place under these electron acceptor conditions. In the two other treatment plants fewer RPAO were able to grow by using nitrate and nitrite as electron acceptors. Prolonged exposure under these electron acceptor conditions by including a preincubation with nitrate or nitrite together with unlabeled acetate for 6 h (Table 4) did not increase the percentage of RPAO taking up acetate in the presence of these electron acceptors. This indicates that a fraction of the RPAO was unable to use nitrate and nitrite as electron acceptors.

DISCUSSION

In this study, we investigated the ecophysiology of probe-defined RPAO, which are believed to be some of the most abundant PAO in lab-scale reactors, as well as full-scale WWTP (21). This study revealed that RPAO were abundant in the full-scale WWTP investigated and in some cases accounted for up to 22% of the total population. This finding emphasizes that these bacteria are very important members of the microbial community in EBPR plants and that detailed knowledge of the ecophysiology and ecology of these uncultured bacteria is essential for understanding and controlling the EBPR process. Importantly, the probe-defined RPAO behaved very similarly in all three WWTP investigated, indicating that the physiologies were very similar and thus providing a high level of reliability. Below we discuss the results obtained mainly in relation to the existing biochemical models, as reviewed by several authors (16, 21, 24).

Assimilation of organic substrates.

The biochemical models proposed for PAO are all based on the uptake of acetate under anaerobic conditions for storage as PHA. In these models it is also assumed that several other substrates (e.g., propionate and butyrate) can be used by PAO, but conflicting results have been obtained (21). The MAR-FISH results obtained in this study revealed that RPAO could assimilate acetate, propionate, pyruvate, and glutamic acid, while several other compounds, including formate, ethanol, butyrate, and several other amino acids, were not utilized. This means that only relatively few of the low-molecular-weight compounds assumed to be present in domestic wastewater or likely fermentation products are important substrates for RPAO.

RPAO were able to assimilate two organic substrates simultaneously. This was tested by first adding one unlabeled compound together with another labeled compound and then performing the opposite experiment, and in both cases the RPAO were MAR positive. It is not known how common the simultaneous uptake of several substrates is for activated sludge bacteria, but this physiological trait is presumably an advantage in the EBPR process, in which several organic compounds might be available during the anaerobic period.

Assimilation of the amino acids aspartic acid and glutamic acid by enriched unknown PAO has been reported for lab-scale EBPR systems (19). In our study, we found that RPAO could assimilate glutamic acid but not aspartic acid under anaerobic as well as aerobic conditions. It is not clear whether the assimilated glutamic acid was stored anaerobically as PHA, as suggested for acetate, or whether it was stored as a polymer consisting of γ-aminobutyric acid and an unknown amino acid, as has been suggested previously (19). Another interesting observation is the fact that thymidine and the amino acid leucine were not assimilated as sole substrates under anaerobic conditions but were assimilated only as cosubstrates in the presence of acetate. These compounds could potentially be used as N sources, or thymidine could be involved in DNA biosynthesis during growth, but as no growth took place under anaerobic conditions, the detailed mechanism for uptake and storage is not clear. However, it can be concluded that RPAO could assimilate and store not only carbon sources but possibly also other nutrients in the anaerobic phase, which may provide a competitive advantage over other bacteria without this capability. It will be interesting to investigate whether this physiological trait is common in other PAO and in glycogen-accumulating organisms (GAO), all of which are aerobes and able to take up carbon sources anaerobically and grow under subsequent aerobic conditions (21). Importantly, the acetate-dependent thymidine and leucine uptake by RPAO also suggests that caution is necessary when incorporation of labeled thymidine or leucine is used to determine the status of cell activity (22).

Assimilation of glucose by RPAO.

The possible ability of PAO to assimilate glucose has attracted a lot of attention as sugar compounds can arrive in significant amounts with wastewater at EBPR plants and occasionally affect the activity of the PAO (16). In some lab-scale activated sludge systems, glucose has successfully been used as the sole organic substrate to support EBPR (3, 8), and in other systems, glucose has been reported to cause EBPR breakdown by selecting GAO instead of PAO. GAO use intracellular glycogen instead of polyphosphate as an energy source to assimilate organic substrates anaerobically (4). However, as the microbial communities in the lab-scale reactors have not been analyzed by molecular tools, the identities of the PAO are not known, and it is not known whether glucose or fermentation products are consumed. In our study, it was clear that the RPAO were unable to assimilate glucose directly but could readily assimilate consumable fermentation products, such as acetate. This is in agreement with the finding that RPAO enriched in a sequencing batch reactor fed with acetate as the sole organic substrate were unable to take up glucose under anaerobic conditions (10). The reason why no glucose assimilation was observed was most probably a lack of fermenting bacteria, as the reactor was only fed acetate, thus washing out the fermenting populations. This indicates that the selection of either RPAO or GAO is dependent on the structure and function of microbial populations involved in glucose consumption. If the fermenters are active, adding glucose may promote EBPR by providing more fermentation products to RPAO; if they are not, glucose could harm the EBPR by selecting the GAO. In full-scale plants, the latter scenario is probably rare as most wastewaters contain a range of compounds that can be fermented.

Importance of glycolysis for anaerobic substrate assimilation by RPAO.

According to the biochemical model (16, 24), PAO need reducing power to sequester the assimilated acetate into PHA under anaerobic conditions. Two possible sources have been proposed. One source is derived from degradation of intracellularly stored glycogen through the glycolytic pathway (16), and the other is the TCA cycle, in which acetate is anaerobically metabolized (18). In this study, we found that iodoacetate, which blocks the glycolytic pathway (13), completely prevented RPAO from anaerobic acetate assimilation. This suggests that anaerobic substrate assimilation and PHA formation are totally dependent on glycogen degradation through glycolysis and that this is the only pathway used to produce reducing power, which supports the first biochemical model (16). However, since the identity of the PAO in the activated sludge used by Pereira et al. (18) is not clear, the possibility of involvement of the TCA cycle of other bacteria cannot be ruled out.

Growth and storage.

The biochemical models predict that PAO have a certain uptake and storage capacity for acetate under anaerobic conditions (24). This was fully confirmed to be the case for the RPAO, in which storage took place for a maximum of approximately 3 h. Under aerobic conditions, continuous uptake of acetate for at least 9 h took place, confirming that the RPAO were growing under these conditions and thus were aerobic bacteria. However, under these conditions with a surplus of external substrate under aerobic conditions, no uptake of orthophosphate took place. This is also in accordance with the model (24) which states that in the presence of acetate PAO do not take up orthophosphate but may use polyphosphate as an energy source to assimilate acetate and form PHA.

Activity of RPAO under denitrifying conditions.

Uptake of orthophosphate with nitrate or nitrite as the electron acceptor has been observed in full-scale plants and in lab-scale EBPR reactors (2). Recently, it was suggested that RPAO are able to conduct denitrification because a highly enriched culture in a lab-scale reactor had high denitrification rates and was able to accumulate orthophosphate with both oxygen and nitrate as electron acceptors (26). Our results generally support these observations. In this study, we found that most RPAO (86% ± 9%) at the Skagen WWTP were able to take up labeled orthophosphate not only with oxygen or nitrate but also with nitrite as the electron acceptor. Furthermore, the same percentages of RPAO were able to assimilate acetate with oxygen, nitrate, and nitrite as electron acceptors, and thus the data strongly suggest that these RPAO were PAO conducting denitrification. There were no indications that nitrite was reduced to ammonium by a possible dissimilatoric nitrite reduction. If nitrite was added to sludge samples, no ammonium accumulation could be detected (data not shown), as would be expected with the high number of RPAO in the sludges if they performed the process. Whether the RPAO conducted full denitrification to N₂ or whether the product was N₂O was not investigated.

The bacteria were able to take up orthophosphate in the presence of nitrate or nitrite with no induction time, which is different from the observations obtained with a lab-scale reactor by Zeng et al. (26). These authors found that the RPAO in their reactor operated in an anaerobic-aerobic cycling mode needed approximately 5 h before denitrification or uptake of P_i started after a shift from anaerobic conditions to denitrifying conditions with nitrate as the electron acceptor. This was explained by a lag time for production of denitrifying enzymes, and there was no lag time after an additional anaerobic-denitrifying cycle. In the full-scale systems investigated here, activated sludge from the anaerobic tank is subsequently exposed to several (three to five) alternating denitrifying and aerobic periods that are each 1 to 2 h long before the activated sludge is exposed to anaerobic conditions again. This means that RPAO from the full-scale systems adapted to quick changes in the electron acceptor conditions and thus possibly constitutively expressed the relevant enzymes. However, in two of the plants, the AAE and Egaa WWTP, it was observed that not all RPAO were active during denitrification, both in terms of uptake of labeled acetate after prolonged incubation with unlabeled acetate (indicating denitrification) and during uptake of ³³P_i when nitrate was present as an electron acceptor. Whether a fraction of the bacteria did not express the relevant enzyme systems or whether the results were due to the presence of two different subpopulations of RPAO, a denitrifying subpopulation and a nondenitrifying subpopulation, is not clear. It is known from many full-scale EBPR plants that the ability to take up P_i under denitrifying conditions varies a lot (26), so besides the possible lag phase and the presence of other species of PAO, it is also possible that there are populations with different denitrifying capabilities among the RPAO.

So far, the dominant denitrifiers in nutrient removal plants treating municipal wastewater are mostly unknown. Recently, we found that bacteria belonging to the genus Aquaspirillum in the beta-proteobacteria seem to be important denitrifiers in some full-scale nutrient removal plants (23). In industrial wastewater treatment plants, bacteria belonging to the genera Azoarcus, Thauera, and Zoogloea (all belonging to Betaproteobacteria) are abundant and are believed to be the dominant denitrifiers (9, 25). Thus, this study showed that the RPAO are potentially important denitrifiers in full-scale EBPR plants, and future studies should show whether they are the dominant denitrifiers.

Physiological status and P_i accumulation.

We found that the RPAO present in the full-scale EBPR plants differed in the ability to accumulate P_i. At the Skagen WWTP, 86% ± 9% of the RPAO took up orthophosphate under aerobic conditions, while only 27% ± 8% of the RPAO at the AAE WWTP and 56% ± 13% of the RPAO at the Egaa WWTP were actively involved in P_i uptake. This is in agreement with a previous study in which Lee et al. (12) found that a fraction of RPAO present in pilot-scale plants did not accumulate P_i. The explanation for this might be found in the wastewater characteristics and operation modes of the three plants. The plant treating mainly industrial wastewater (Skagen WWTP) had a surprisingly high number of RPAO, and most of them were active in the uptake of P_i, probably mainly due to the higher content of easily degradable organic matter compared to the content at the two other plants, which treat mainly municipal wastewater. A higher C/P ratio in the wastewater is known to promote good EBPR activity (10). Another operational difference between the plants is that addition of ferric iron takes place at the AAE and Egaa treatment plants, but not at the Skagen WWTP. A certain (low) dose of ferric iron for removal of orthophosphate is added at several EBPR plants in order to ensure a low effluent P concentration. However, if too much ferric iron is added, P_i uptake is limited, which can eventually cause a failure of the EBPR activity. The presence of ferric iron may reduce the amount of orthophosphate available to be taken up by RPAO under aerobic or denitrifying conditions, thereby preventing uptake of sufficient organic matter under subsequent anaerobic conditions. After several cycles, the physiological status may change for a fraction of the RPAO, so although it was possible to see uptake of labeled acetate in most RPAO cells, the amount of labeled orthophosphate taken up may have been so low that it was not detected by the MAR method. These results also show that it is difficult to assess the EBPR activity in a WWTP by using information about the quantity of RPAO, as a significant proportion may be inactive in the excessive P_i uptake.

Are RPAO putative PAO?

In this study, the ecophysiological traits of RPAO present in full-scale EBPR plants were systematically characterized by MAR-FISH. This study confirmed that the ecophysiology of these organisms largely agrees with the biochemical model proposed on the basis of lab-scale studies and that the organisms are able to denitrify. The results strongly suggest that RPAO are the putative PAO functioning in full-scale EBPR processes and that, as abundant bacteria, may be the dominant PAO and perhaps also the dominant denitrifiers in full-scale WWTP. However, during this study we also found that bacteria other than RPAO were involved in P removal, and further studies are needed to reveal the identity, ecophysiology, and significance of these bacteria in full-scale EBPR plants.